{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.776925318053802,
  "eval_steps": 5,
  "global_step": 3000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0,
      "eval_PRM Accuracy": 0.3867924528301887,
      "eval_PRM F1": 0.4144144144144144,
      "eval_PRM F1 AUC": 0.5298585646935569,
      "eval_PRM F1 AUC (fixed)": 0.33813514929282346,
      "eval_PRM F1 Neg": 0.3564356435643564,
      "eval_PRM NPV": 0.23076923076923078,
      "eval_PRM Precision": 0.8214285714285714,
      "eval_PRM Recall": 0.27710843373493976,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.8678227066993713,
      "eval_runtime": 16.2347,
      "eval_samples_per_second": 2.71,
      "eval_steps_per_second": 0.123,
      "step": 0
    },
    {
      "epoch": 0.000258975106017934,
      "grad_norm": 63.037703867152466,
      "learning_rate": 3.2299741602067187e-09,
      "loss": 0.9478,
      "step": 1
    },
    {
      "epoch": 0.000517950212035868,
      "grad_norm": 39.472536090782,
      "learning_rate": 6.4599483204134375e-09,
      "loss": 0.9176,
      "step": 2
    },
    {
      "epoch": 0.0007769253180538021,
      "grad_norm": 40.61452474226669,
      "learning_rate": 9.689922480620155e-09,
      "loss": 0.994,
      "step": 3
    },
    {
      "epoch": 0.001035900424071736,
      "grad_norm": 36.95892027782838,
      "learning_rate": 1.2919896640826875e-08,
      "loss": 0.9114,
      "step": 4
    },
    {
      "epoch": 0.00129487553008967,
      "grad_norm": 26.036252649177914,
      "learning_rate": 1.6149870801033593e-08,
      "loss": 0.9668,
      "step": 5
    },
    {
      "epoch": 0.00129487553008967,
      "eval_PRM Accuracy": 0.4056603773584906,
      "eval_PRM F1": 0.4424778761061947,
      "eval_PRM F1 AUC": 0.5419067574646412,
      "eval_PRM F1 AUC (fixed)": 0.3386589837611315,
      "eval_PRM F1 Neg": 0.36363636363636365,
      "eval_PRM NPV": 0.23684210526315788,
      "eval_PRM Precision": 0.8333333333333334,
      "eval_PRM Recall": 0.30120481927710846,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.8697418570518494,
      "eval_runtime": 14.7356,
      "eval_samples_per_second": 2.986,
      "eval_steps_per_second": 0.136,
      "step": 5
    },
    {
      "epoch": 0.0015538506361076042,
      "grad_norm": 40.337292444544296,
      "learning_rate": 1.937984496124031e-08,
      "loss": 1.0354,
      "step": 6
    },
    {
      "epoch": 0.0018128257421255382,
      "grad_norm": 51.91063484757722,
      "learning_rate": 2.260981912144703e-08,
      "loss": 0.9882,
      "step": 7
    },
    {
      "epoch": 0.002071800848143472,
      "grad_norm": 51.663899635887994,
      "learning_rate": 2.583979328165375e-08,
      "loss": 0.9176,
      "step": 8
    },
    {
      "epoch": 0.0023307759541614063,
      "grad_norm": 21.81186522663788,
      "learning_rate": 2.9069767441860468e-08,
      "loss": 0.8714,
      "step": 9
    },
    {
      "epoch": 0.00258975106017934,
      "grad_norm": 25.886467614209714,
      "learning_rate": 3.2299741602067186e-08,
      "loss": 0.7792,
      "step": 10
    },
    {
      "epoch": 0.00258975106017934,
      "eval_PRM Accuracy": 0.4056603773584906,
      "eval_PRM F1": 0.43243243243243246,
      "eval_PRM F1 AUC": 0.5576217915138816,
      "eval_PRM F1 AUC (fixed)": 0.3397066526977475,
      "eval_PRM F1 Neg": 0.37623762376237624,
      "eval_PRM NPV": 0.24358974358974358,
      "eval_PRM Precision": 0.8571428571428571,
      "eval_PRM Recall": 0.2891566265060241,
      "eval_PRM Specificty": 0.8260869565217391,
      "eval_loss": 0.8677586913108826,
      "eval_runtime": 14.9411,
      "eval_samples_per_second": 2.945,
      "eval_steps_per_second": 0.134,
      "step": 10
    },
    {
      "epoch": 0.0028487261661972745,
      "grad_norm": 44.33097404486639,
      "learning_rate": 3.552971576227391e-08,
      "loss": 0.9529,
      "step": 11
    },
    {
      "epoch": 0.0031077012722152083,
      "grad_norm": 39.28825527565396,
      "learning_rate": 3.875968992248062e-08,
      "loss": 0.9612,
      "step": 12
    },
    {
      "epoch": 0.003366676378233142,
      "grad_norm": 26.936193809280923,
      "learning_rate": 4.198966408268734e-08,
      "loss": 0.9125,
      "step": 13
    },
    {
      "epoch": 0.0036256514842510764,
      "grad_norm": 63.91618838754899,
      "learning_rate": 4.521963824289406e-08,
      "loss": 1.0074,
      "step": 14
    },
    {
      "epoch": 0.0038846265902690103,
      "grad_norm": 46.647418475437426,
      "learning_rate": 4.844961240310078e-08,
      "loss": 0.9009,
      "step": 15
    },
    {
      "epoch": 0.0038846265902690103,
      "eval_PRM Accuracy": 0.4056603773584906,
      "eval_PRM F1": 0.4424778761061947,
      "eval_PRM F1 AUC": 0.5419067574646412,
      "eval_PRM F1 AUC (fixed)": 0.3407543216343636,
      "eval_PRM F1 Neg": 0.36363636363636365,
      "eval_PRM NPV": 0.23684210526315788,
      "eval_PRM Precision": 0.8333333333333334,
      "eval_PRM Recall": 0.30120481927710846,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.8698608875274658,
      "eval_runtime": 10.4028,
      "eval_samples_per_second": 4.23,
      "eval_steps_per_second": 0.192,
      "step": 15
    },
    {
      "epoch": 0.004143601696286944,
      "grad_norm": 57.58762296760476,
      "learning_rate": 5.16795865633075e-08,
      "loss": 0.9692,
      "step": 16
    },
    {
      "epoch": 0.004402576802304879,
      "grad_norm": 30.661267401226237,
      "learning_rate": 5.4909560723514214e-08,
      "loss": 0.8378,
      "step": 17
    },
    {
      "epoch": 0.004661551908322813,
      "grad_norm": 41.39269160524219,
      "learning_rate": 5.8139534883720935e-08,
      "loss": 0.9658,
      "step": 18
    },
    {
      "epoch": 0.0049205270143407465,
      "grad_norm": 42.52726716449715,
      "learning_rate": 6.136950904392765e-08,
      "loss": 0.9987,
      "step": 19
    },
    {
      "epoch": 0.00517950212035868,
      "grad_norm": 49.82620333190475,
      "learning_rate": 6.459948320413437e-08,
      "loss": 0.9629,
      "step": 20
    },
    {
      "epoch": 0.00517950212035868,
      "eval_PRM Accuracy": 0.4056603773584906,
      "eval_PRM F1": 0.4424778761061947,
      "eval_PRM F1 AUC": 0.5419067574646412,
      "eval_PRM F1 AUC (fixed)": 0.33944473546359355,
      "eval_PRM F1 Neg": 0.36363636363636365,
      "eval_PRM NPV": 0.23684210526315788,
      "eval_PRM Precision": 0.8333333333333334,
      "eval_PRM Recall": 0.30120481927710846,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.8663817048072815,
      "eval_runtime": 10.5034,
      "eval_samples_per_second": 4.189,
      "eval_steps_per_second": 0.19,
      "step": 20
    },
    {
      "epoch": 0.005438477226376614,
      "grad_norm": 40.625187370837494,
      "learning_rate": 6.782945736434109e-08,
      "loss": 0.9758,
      "step": 21
    },
    {
      "epoch": 0.005697452332394549,
      "grad_norm": 29.05974126113703,
      "learning_rate": 7.105943152454781e-08,
      "loss": 0.871,
      "step": 22
    },
    {
      "epoch": 0.005956427438412483,
      "grad_norm": 30.951805478184244,
      "learning_rate": 7.428940568475452e-08,
      "loss": 0.8993,
      "step": 23
    },
    {
      "epoch": 0.006215402544430417,
      "grad_norm": 40.05628218268796,
      "learning_rate": 7.751937984496124e-08,
      "loss": 0.939,
      "step": 24
    },
    {
      "epoch": 0.0064743776504483505,
      "grad_norm": 64.28427259734805,
      "learning_rate": 8.074935400516796e-08,
      "loss": 1.0051,
      "step": 25
    },
    {
      "epoch": 0.0064743776504483505,
      "eval_PRM Accuracy": 0.41509433962264153,
      "eval_PRM F1": 0.45614035087719296,
      "eval_PRM F1 AUC": 0.5479308538501834,
      "eval_PRM F1 AUC (fixed)": 0.3399685699319015,
      "eval_PRM F1 Neg": 0.3673469387755102,
      "eval_PRM NPV": 0.24,
      "eval_PRM Precision": 0.8387096774193549,
      "eval_PRM Recall": 0.3132530120481928,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.8646280169487,
      "eval_runtime": 14.5121,
      "eval_samples_per_second": 3.032,
      "eval_steps_per_second": 0.138,
      "step": 25
    },
    {
      "epoch": 0.006733352756466284,
      "grad_norm": 56.89452099484511,
      "learning_rate": 8.397932816537469e-08,
      "loss": 0.986,
      "step": 26
    },
    {
      "epoch": 0.006992327862484219,
      "grad_norm": 49.624626549694966,
      "learning_rate": 8.72093023255814e-08,
      "loss": 0.9552,
      "step": 27
    },
    {
      "epoch": 0.007251302968502153,
      "grad_norm": 40.268884202852206,
      "learning_rate": 9.043927648578811e-08,
      "loss": 0.8838,
      "step": 28
    },
    {
      "epoch": 0.007510278074520087,
      "grad_norm": 35.58022077420101,
      "learning_rate": 9.366925064599485e-08,
      "loss": 0.8722,
      "step": 29
    },
    {
      "epoch": 0.007769253180538021,
      "grad_norm": 42.379835980673334,
      "learning_rate": 9.689922480620156e-08,
      "loss": 0.9022,
      "step": 30
    },
    {
      "epoch": 0.007769253180538021,
      "eval_PRM Accuracy": 0.4056603773584906,
      "eval_PRM F1": 0.4424778761061947,
      "eval_PRM F1 AUC": 0.5419067574646412,
      "eval_PRM F1 AUC (fixed)": 0.33970665269774747,
      "eval_PRM F1 Neg": 0.36363636363636365,
      "eval_PRM NPV": 0.23684210526315788,
      "eval_PRM Precision": 0.8333333333333334,
      "eval_PRM Recall": 0.30120481927710846,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.8620710372924805,
      "eval_runtime": 14.5493,
      "eval_samples_per_second": 3.024,
      "eval_steps_per_second": 0.137,
      "step": 30
    },
    {
      "epoch": 0.008028228286555954,
      "grad_norm": 34.42647540348086,
      "learning_rate": 1.0012919896640826e-07,
      "loss": 0.9226,
      "step": 31
    },
    {
      "epoch": 0.008287203392573888,
      "grad_norm": 16.754621860181707,
      "learning_rate": 1.03359173126615e-07,
      "loss": 1.0476,
      "step": 32
    },
    {
      "epoch": 0.008546178498591822,
      "grad_norm": 41.44570425860884,
      "learning_rate": 1.0658914728682171e-07,
      "loss": 0.906,
      "step": 33
    },
    {
      "epoch": 0.008805153604609758,
      "grad_norm": 38.57423040321388,
      "learning_rate": 1.0981912144702843e-07,
      "loss": 0.9152,
      "step": 34
    },
    {
      "epoch": 0.009064128710627692,
      "grad_norm": 53.23182535449974,
      "learning_rate": 1.1304909560723515e-07,
      "loss": 1.0426,
      "step": 35
    },
    {
      "epoch": 0.009064128710627692,
      "eval_PRM Accuracy": 0.41509433962264153,
      "eval_PRM F1": 0.45614035087719296,
      "eval_PRM F1 AUC": 0.5479308538501834,
      "eval_PRM F1 AUC (fixed)": 0.3397066526977475,
      "eval_PRM F1 Neg": 0.3673469387755102,
      "eval_PRM NPV": 0.24,
      "eval_PRM Precision": 0.8387096774193549,
      "eval_PRM Recall": 0.3132530120481928,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.8567813634872437,
      "eval_runtime": 15.0601,
      "eval_samples_per_second": 2.922,
      "eval_steps_per_second": 0.133,
      "step": 35
    },
    {
      "epoch": 0.009323103816645625,
      "grad_norm": 50.17753754565222,
      "learning_rate": 1.1627906976744187e-07,
      "loss": 0.9266,
      "step": 36
    },
    {
      "epoch": 0.00958207892266356,
      "grad_norm": 43.70938983864694,
      "learning_rate": 1.1950904392764858e-07,
      "loss": 0.8187,
      "step": 37
    },
    {
      "epoch": 0.009841054028681493,
      "grad_norm": 44.16135701565119,
      "learning_rate": 1.227390180878553e-07,
      "loss": 0.9802,
      "step": 38
    },
    {
      "epoch": 0.010100029134699427,
      "grad_norm": 45.38369838408391,
      "learning_rate": 1.2596899224806202e-07,
      "loss": 0.8363,
      "step": 39
    },
    {
      "epoch": 0.01035900424071736,
      "grad_norm": 52.82723187942695,
      "learning_rate": 1.2919896640826874e-07,
      "loss": 1.0043,
      "step": 40
    },
    {
      "epoch": 0.01035900424071736,
      "eval_PRM Accuracy": 0.4339622641509434,
      "eval_PRM F1": 0.4827586206896552,
      "eval_PRM F1 AUC": 0.5599790466212677,
      "eval_PRM F1 AUC (fixed)": 0.33708748035620745,
      "eval_PRM F1 Neg": 0.375,
      "eval_PRM NPV": 0.2465753424657534,
      "eval_PRM Precision": 0.8484848484848485,
      "eval_PRM Recall": 0.3373493975903614,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.8464098572731018,
      "eval_runtime": 14.8041,
      "eval_samples_per_second": 2.972,
      "eval_steps_per_second": 0.135,
      "step": 40
    },
    {
      "epoch": 0.010617979346735295,
      "grad_norm": 35.02370077516764,
      "learning_rate": 1.3242894056847546e-07,
      "loss": 0.8568,
      "step": 41
    },
    {
      "epoch": 0.010876954452753228,
      "grad_norm": 38.713476056558825,
      "learning_rate": 1.3565891472868218e-07,
      "loss": 0.9738,
      "step": 42
    },
    {
      "epoch": 0.011135929558771162,
      "grad_norm": 48.99054930472666,
      "learning_rate": 1.3888888888888888e-07,
      "loss": 0.8334,
      "step": 43
    },
    {
      "epoch": 0.011394904664789098,
      "grad_norm": 35.69913234062826,
      "learning_rate": 1.4211886304909563e-07,
      "loss": 0.842,
      "step": 44
    },
    {
      "epoch": 0.011653879770807032,
      "grad_norm": 47.80559671444046,
      "learning_rate": 1.4534883720930235e-07,
      "loss": 0.9162,
      "step": 45
    },
    {
      "epoch": 0.011653879770807032,
      "eval_PRM Accuracy": 0.4339622641509434,
      "eval_PRM F1": 0.4827586206896552,
      "eval_PRM F1 AUC": 0.5599790466212677,
      "eval_PRM F1 AUC (fixed)": 0.3428496595075956,
      "eval_PRM F1 Neg": 0.375,
      "eval_PRM NPV": 0.2465753424657534,
      "eval_PRM Precision": 0.8484848484848485,
      "eval_PRM Recall": 0.3373493975903614,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.8389459252357483,
      "eval_runtime": 14.6941,
      "eval_samples_per_second": 2.994,
      "eval_steps_per_second": 0.136,
      "step": 45
    },
    {
      "epoch": 0.011912854876824966,
      "grad_norm": 38.95935999442876,
      "learning_rate": 1.4857881136950904e-07,
      "loss": 1.0287,
      "step": 46
    },
    {
      "epoch": 0.0121718299828429,
      "grad_norm": 45.84894726759758,
      "learning_rate": 1.518087855297158e-07,
      "loss": 0.9409,
      "step": 47
    },
    {
      "epoch": 0.012430805088860833,
      "grad_norm": 49.66077506241443,
      "learning_rate": 1.5503875968992249e-07,
      "loss": 0.882,
      "step": 48
    },
    {
      "epoch": 0.012689780194878767,
      "grad_norm": 26.814130429582953,
      "learning_rate": 1.582687338501292e-07,
      "loss": 0.7798,
      "step": 49
    },
    {
      "epoch": 0.012948755300896701,
      "grad_norm": 32.7672953300368,
      "learning_rate": 1.6149870801033593e-07,
      "loss": 0.8485,
      "step": 50
    },
    {
      "epoch": 0.012948755300896701,
      "eval_PRM Accuracy": 0.4339622641509434,
      "eval_PRM F1": 0.4915254237288136,
      "eval_PRM F1 AUC": 0.5442640125720272,
      "eval_PRM F1 AUC (fixed)": 0.34154007333682557,
      "eval_PRM F1 Neg": 0.3617021276595745,
      "eval_PRM NPV": 0.23943661971830985,
      "eval_PRM Precision": 0.8285714285714286,
      "eval_PRM Recall": 0.3493975903614458,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.8221114873886108,
      "eval_runtime": 15.0333,
      "eval_samples_per_second": 2.927,
      "eval_steps_per_second": 0.133,
      "step": 50
    },
    {
      "epoch": 0.013207730406914635,
      "grad_norm": 40.046840263198526,
      "learning_rate": 1.6472868217054268e-07,
      "loss": 0.9715,
      "step": 51
    },
    {
      "epoch": 0.013466705512932569,
      "grad_norm": 27.3304257780402,
      "learning_rate": 1.6795865633074937e-07,
      "loss": 0.7938,
      "step": 52
    },
    {
      "epoch": 0.013725680618950504,
      "grad_norm": 52.78649060661054,
      "learning_rate": 1.711886304909561e-07,
      "loss": 0.9807,
      "step": 53
    },
    {
      "epoch": 0.013984655724968438,
      "grad_norm": 31.60652199020792,
      "learning_rate": 1.744186046511628e-07,
      "loss": 0.8377,
      "step": 54
    },
    {
      "epoch": 0.014243630830986372,
      "grad_norm": 36.740966951868984,
      "learning_rate": 1.776485788113695e-07,
      "loss": 0.9034,
      "step": 55
    },
    {
      "epoch": 0.014243630830986372,
      "eval_PRM Accuracy": 0.4528301886792453,
      "eval_PRM F1": 0.5396825396825397,
      "eval_PRM F1 AUC": 0.5091671031953903,
      "eval_PRM F1 AUC (fixed)": 0.3428496595075956,
      "eval_PRM F1 Neg": 0.32558139534883723,
      "eval_PRM NPV": 0.2222222222222222,
      "eval_PRM Precision": 0.7906976744186046,
      "eval_PRM Recall": 0.40963855421686746,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.7895786762237549,
      "eval_runtime": 14.5999,
      "eval_samples_per_second": 3.014,
      "eval_steps_per_second": 0.137,
      "step": 55
    },
    {
      "epoch": 0.014502605937004306,
      "grad_norm": 13.797714830369891,
      "learning_rate": 1.8087855297157623e-07,
      "loss": 0.7442,
      "step": 56
    },
    {
      "epoch": 0.01476158104302224,
      "grad_norm": 35.405763235862786,
      "learning_rate": 1.8410852713178298e-07,
      "loss": 0.7909,
      "step": 57
    },
    {
      "epoch": 0.015020556149040173,
      "grad_norm": 40.58250255476464,
      "learning_rate": 1.873385012919897e-07,
      "loss": 0.8553,
      "step": 58
    },
    {
      "epoch": 0.015279531255058107,
      "grad_norm": 42.238417195704244,
      "learning_rate": 1.905684754521964e-07,
      "loss": 0.8204,
      "step": 59
    },
    {
      "epoch": 0.015538506361076041,
      "grad_norm": 42.32301478324044,
      "learning_rate": 1.9379844961240311e-07,
      "loss": 0.9246,
      "step": 60
    },
    {
      "epoch": 0.015538506361076041,
      "eval_PRM Accuracy": 0.4528301886792453,
      "eval_PRM F1": 0.5396825396825397,
      "eval_PRM F1 AUC": 0.5091671031953903,
      "eval_PRM F1 AUC (fixed)": 0.3391828182294395,
      "eval_PRM F1 Neg": 0.32558139534883723,
      "eval_PRM NPV": 0.2222222222222222,
      "eval_PRM Precision": 0.7906976744186046,
      "eval_PRM Recall": 0.40963855421686746,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.7745277285575867,
      "eval_runtime": 14.6553,
      "eval_samples_per_second": 3.002,
      "eval_steps_per_second": 0.136,
      "step": 60
    },
    {
      "epoch": 0.015797481467093977,
      "grad_norm": 16.34880758084372,
      "learning_rate": 1.9702842377260983e-07,
      "loss": 0.8876,
      "step": 61
    },
    {
      "epoch": 0.01605645657311191,
      "grad_norm": 33.018050747698034,
      "learning_rate": 2.0025839793281653e-07,
      "loss": 0.7665,
      "step": 62
    },
    {
      "epoch": 0.016315431679129844,
      "grad_norm": 35.941705491024585,
      "learning_rate": 2.0348837209302328e-07,
      "loss": 0.9996,
      "step": 63
    },
    {
      "epoch": 0.016574406785147777,
      "grad_norm": 21.384319021829366,
      "learning_rate": 2.0671834625323e-07,
      "loss": 0.7215,
      "step": 64
    },
    {
      "epoch": 0.016833381891165712,
      "grad_norm": 14.176127306038236,
      "learning_rate": 2.0994832041343672e-07,
      "loss": 0.7689,
      "step": 65
    },
    {
      "epoch": 0.016833381891165712,
      "eval_PRM Accuracy": 0.46226415094339623,
      "eval_PRM F1": 0.5511811023622047,
      "eval_PRM F1 AUC": 0.5151911995809324,
      "eval_PRM F1 AUC (fixed)": 0.3459926663174437,
      "eval_PRM F1 Neg": 0.32941176470588235,
      "eval_PRM NPV": 0.22580645161290322,
      "eval_PRM Precision": 0.7954545454545454,
      "eval_PRM Recall": 0.42168674698795183,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.7576757073402405,
      "eval_runtime": 14.7429,
      "eval_samples_per_second": 2.984,
      "eval_steps_per_second": 0.136,
      "step": 65
    },
    {
      "epoch": 0.017092356997183644,
      "grad_norm": 27.825752091392918,
      "learning_rate": 2.1317829457364341e-07,
      "loss": 0.8875,
      "step": 66
    },
    {
      "epoch": 0.01735133210320158,
      "grad_norm": 14.650186494847837,
      "learning_rate": 2.1640826873385014e-07,
      "loss": 0.8708,
      "step": 67
    },
    {
      "epoch": 0.017610307209219515,
      "grad_norm": 24.186659078983645,
      "learning_rate": 2.1963824289405686e-07,
      "loss": 0.8956,
      "step": 68
    },
    {
      "epoch": 0.017869282315237447,
      "grad_norm": 13.805001585257308,
      "learning_rate": 2.228682170542636e-07,
      "loss": 0.8644,
      "step": 69
    },
    {
      "epoch": 0.018128257421255383,
      "grad_norm": 15.468143300619719,
      "learning_rate": 2.260981912144703e-07,
      "loss": 0.9425,
      "step": 70
    },
    {
      "epoch": 0.018128257421255383,
      "eval_PRM Accuracy": 0.5094339622641509,
      "eval_PRM F1": 0.6060606060606061,
      "eval_PRM F1 AUC": 0.5453116815086433,
      "eval_PRM F1 AUC (fixed)": 0.34363541121005764,
      "eval_PRM F1 Neg": 0.35,
      "eval_PRM NPV": 0.24561403508771928,
      "eval_PRM Precision": 0.8163265306122449,
      "eval_PRM Recall": 0.4819277108433735,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.730787456035614,
      "eval_runtime": 14.2896,
      "eval_samples_per_second": 3.079,
      "eval_steps_per_second": 0.14,
      "step": 70
    },
    {
      "epoch": 0.018387232527273315,
      "grad_norm": 29.660136915528717,
      "learning_rate": 2.2932816537467702e-07,
      "loss": 0.8083,
      "step": 71
    },
    {
      "epoch": 0.01864620763329125,
      "grad_norm": 22.16011134465685,
      "learning_rate": 2.3255813953488374e-07,
      "loss": 0.8421,
      "step": 72
    },
    {
      "epoch": 0.018905182739309183,
      "grad_norm": 14.487608897102827,
      "learning_rate": 2.3578811369509044e-07,
      "loss": 0.8739,
      "step": 73
    },
    {
      "epoch": 0.01916415784532712,
      "grad_norm": 18.961269564998528,
      "learning_rate": 2.3901808785529716e-07,
      "loss": 0.7423,
      "step": 74
    },
    {
      "epoch": 0.01942313295134505,
      "grad_norm": 19.796017096354078,
      "learning_rate": 2.4224806201550393e-07,
      "loss": 0.8343,
      "step": 75
    },
    {
      "epoch": 0.01942313295134505,
      "eval_PRM Accuracy": 0.5754716981132075,
      "eval_PRM F1": 0.6808510638297872,
      "eval_PRM F1 AUC": 0.5717653221581981,
      "eval_PRM F1 AUC (fixed)": 0.3530644316396019,
      "eval_PRM F1 Neg": 0.36619718309859156,
      "eval_PRM NPV": 0.2708333333333333,
      "eval_PRM Precision": 0.8275862068965517,
      "eval_PRM Recall": 0.5783132530120482,
      "eval_PRM Specificty": 0.5652173913043478,
      "eval_loss": 0.6725817322731018,
      "eval_runtime": 14.6895,
      "eval_samples_per_second": 2.995,
      "eval_steps_per_second": 0.136,
      "step": 75
    },
    {
      "epoch": 0.019682108057362986,
      "grad_norm": 18.187683674164898,
      "learning_rate": 2.454780361757106e-07,
      "loss": 0.9895,
      "step": 76
    },
    {
      "epoch": 0.01994108316338092,
      "grad_norm": 13.1508358242096,
      "learning_rate": 2.487080103359173e-07,
      "loss": 0.8959,
      "step": 77
    },
    {
      "epoch": 0.020200058269398854,
      "grad_norm": 12.8284219993238,
      "learning_rate": 2.5193798449612404e-07,
      "loss": 0.9145,
      "step": 78
    },
    {
      "epoch": 0.02045903337541679,
      "grad_norm": 12.116461823137387,
      "learning_rate": 2.5516795865633076e-07,
      "loss": 0.7406,
      "step": 79
    },
    {
      "epoch": 0.02071800848143472,
      "grad_norm": 19.994908230899348,
      "learning_rate": 2.583979328165375e-07,
      "loss": 0.6423,
      "step": 80
    },
    {
      "epoch": 0.02071800848143472,
      "eval_PRM Accuracy": 0.6037735849056604,
      "eval_PRM F1": 0.7123287671232876,
      "eval_PRM F1 AUC": 0.574122577265584,
      "eval_PRM F1 AUC (fixed)": 0.35620743844945,
      "eval_PRM F1 Neg": 0.36363636363636365,
      "eval_PRM NPV": 0.27906976744186046,
      "eval_PRM Precision": 0.8253968253968254,
      "eval_PRM Recall": 0.6265060240963856,
      "eval_PRM Specificty": 0.5217391304347826,
      "eval_loss": 0.6553806066513062,
      "eval_runtime": 14.8024,
      "eval_samples_per_second": 2.972,
      "eval_steps_per_second": 0.135,
      "step": 80
    },
    {
      "epoch": 0.020976983587452657,
      "grad_norm": 21.174111131813078,
      "learning_rate": 2.616279069767442e-07,
      "loss": 0.8049,
      "step": 81
    },
    {
      "epoch": 0.02123595869347059,
      "grad_norm": 24.28580125692702,
      "learning_rate": 2.6485788113695093e-07,
      "loss": 0.6907,
      "step": 82
    },
    {
      "epoch": 0.021494933799488525,
      "grad_norm": 12.136668635564263,
      "learning_rate": 2.6808785529715765e-07,
      "loss": 0.7757,
      "step": 83
    },
    {
      "epoch": 0.021753908905506457,
      "grad_norm": 24.677938731940024,
      "learning_rate": 2.7131782945736437e-07,
      "loss": 0.6658,
      "step": 84
    },
    {
      "epoch": 0.022012884011524392,
      "grad_norm": 14.33727589348448,
      "learning_rate": 2.745478036175711e-07,
      "loss": 0.8024,
      "step": 85
    },
    {
      "epoch": 0.022012884011524392,
      "eval_PRM Accuracy": 0.6226415094339622,
      "eval_PRM F1": 0.726027397260274,
      "eval_PRM F1 AUC": 0.6018858040859089,
      "eval_PRM F1 AUC (fixed)": 0.35908852802514407,
      "eval_PRM F1 Neg": 0.3939393939393939,
      "eval_PRM NPV": 0.3023255813953488,
      "eval_PRM Precision": 0.8412698412698413,
      "eval_PRM Recall": 0.6385542168674698,
      "eval_PRM Specificty": 0.5652173913043478,
      "eval_loss": 0.6360267400741577,
      "eval_runtime": 14.5509,
      "eval_samples_per_second": 3.024,
      "eval_steps_per_second": 0.137,
      "step": 85
    },
    {
      "epoch": 0.022271859117542325,
      "grad_norm": 11.527980390816484,
      "learning_rate": 2.7777777777777776e-07,
      "loss": 0.6023,
      "step": 86
    },
    {
      "epoch": 0.02253083422356026,
      "grad_norm": 16.674272355584908,
      "learning_rate": 2.8100775193798453e-07,
      "loss": 0.6889,
      "step": 87
    },
    {
      "epoch": 0.022789809329578196,
      "grad_norm": 10.66583950521891,
      "learning_rate": 2.8423772609819125e-07,
      "loss": 0.7801,
      "step": 88
    },
    {
      "epoch": 0.023048784435596128,
      "grad_norm": 9.835671755671285,
      "learning_rate": 2.87467700258398e-07,
      "loss": 0.7201,
      "step": 89
    },
    {
      "epoch": 0.023307759541614063,
      "grad_norm": 26.661182986870347,
      "learning_rate": 2.906976744186047e-07,
      "loss": 0.7963,
      "step": 90
    },
    {
      "epoch": 0.023307759541614063,
      "eval_PRM Accuracy": 0.6320754716981132,
      "eval_PRM F1": 0.738255033557047,
      "eval_PRM F1 AUC": 0.5921948664222106,
      "eval_PRM F1 AUC (fixed)": 0.36223153483499215,
      "eval_PRM F1 Neg": 0.38095238095238093,
      "eval_PRM NPV": 0.3,
      "eval_PRM Precision": 0.8333333333333334,
      "eval_PRM Recall": 0.6626506024096386,
      "eval_PRM Specificty": 0.5217391304347826,
      "eval_loss": 0.620624303817749,
      "eval_runtime": 15.2162,
      "eval_samples_per_second": 2.892,
      "eval_steps_per_second": 0.131,
      "step": 90
    },
    {
      "epoch": 0.023566734647631996,
      "grad_norm": 8.061268722268196,
      "learning_rate": 2.9392764857881137e-07,
      "loss": 0.608,
      "step": 91
    },
    {
      "epoch": 0.02382570975364993,
      "grad_norm": 13.284851373348685,
      "learning_rate": 2.971576227390181e-07,
      "loss": 0.8424,
      "step": 92
    },
    {
      "epoch": 0.024084684859667863,
      "grad_norm": 10.094119910567679,
      "learning_rate": 3.0038759689922486e-07,
      "loss": 0.7432,
      "step": 93
    },
    {
      "epoch": 0.0243436599656858,
      "grad_norm": 14.804320906675862,
      "learning_rate": 3.036175710594316e-07,
      "loss": 0.7599,
      "step": 94
    },
    {
      "epoch": 0.02460263507170373,
      "grad_norm": 10.414202643458607,
      "learning_rate": 3.0684754521963825e-07,
      "loss": 0.5895,
      "step": 95
    },
    {
      "epoch": 0.02460263507170373,
      "eval_PRM Accuracy": 0.6415094339622641,
      "eval_PRM F1": 0.7466666666666667,
      "eval_PRM F1 AUC": 0.5982189628077528,
      "eval_PRM F1 AUC (fixed)": 0.36170770036668415,
      "eval_PRM F1 Neg": 0.3870967741935484,
      "eval_PRM NPV": 0.3076923076923077,
      "eval_PRM Precision": 0.835820895522388,
      "eval_PRM Recall": 0.6746987951807228,
      "eval_PRM Specificty": 0.5217391304347826,
      "eval_loss": 0.6061849594116211,
      "eval_runtime": 15.0049,
      "eval_samples_per_second": 2.932,
      "eval_steps_per_second": 0.133,
      "step": 95
    },
    {
      "epoch": 0.024861610177721667,
      "grad_norm": 10.672625979890972,
      "learning_rate": 3.1007751937984497e-07,
      "loss": 0.7713,
      "step": 96
    },
    {
      "epoch": 0.025120585283739602,
      "grad_norm": 12.251684326095432,
      "learning_rate": 3.1330749354005175e-07,
      "loss": 0.781,
      "step": 97
    },
    {
      "epoch": 0.025379560389757534,
      "grad_norm": 13.91913851389255,
      "learning_rate": 3.165374677002584e-07,
      "loss": 0.7431,
      "step": 98
    },
    {
      "epoch": 0.02563853549577547,
      "grad_norm": 9.507497299775913,
      "learning_rate": 3.1976744186046514e-07,
      "loss": 0.6107,
      "step": 99
    },
    {
      "epoch": 0.025897510601793402,
      "grad_norm": 34.88561213639425,
      "learning_rate": 3.2299741602067186e-07,
      "loss": 0.8904,
      "step": 100
    },
    {
      "epoch": 0.025897510601793402,
      "eval_PRM Accuracy": 0.6415094339622641,
      "eval_PRM F1": 0.7466666666666667,
      "eval_PRM F1 AUC": 0.5982189628077528,
      "eval_PRM F1 AUC (fixed)": 0.36406495547407025,
      "eval_PRM F1 Neg": 0.3870967741935484,
      "eval_PRM NPV": 0.3076923076923077,
      "eval_PRM Precision": 0.835820895522388,
      "eval_PRM Recall": 0.6746987951807228,
      "eval_PRM Specificty": 0.5217391304347826,
      "eval_loss": 0.5902833938598633,
      "eval_runtime": 14.8013,
      "eval_samples_per_second": 2.973,
      "eval_steps_per_second": 0.135,
      "step": 100
    },
    {
      "epoch": 0.026156485707811337,
      "grad_norm": 12.625263414672842,
      "learning_rate": 3.262273901808786e-07,
      "loss": 0.7125,
      "step": 101
    },
    {
      "epoch": 0.02641546081382927,
      "grad_norm": 21.834409355265553,
      "learning_rate": 3.2945736434108535e-07,
      "loss": 0.8078,
      "step": 102
    },
    {
      "epoch": 0.026674435919847205,
      "grad_norm": 16.68863184867011,
      "learning_rate": 3.32687338501292e-07,
      "loss": 0.6124,
      "step": 103
    },
    {
      "epoch": 0.026933411025865137,
      "grad_norm": 6.89532276240644,
      "learning_rate": 3.3591731266149874e-07,
      "loss": 0.5681,
      "step": 104
    },
    {
      "epoch": 0.027192386131883073,
      "grad_norm": 9.05599285768429,
      "learning_rate": 3.391472868217054e-07,
      "loss": 0.5845,
      "step": 105
    },
    {
      "epoch": 0.027192386131883073,
      "eval_PRM Accuracy": 0.660377358490566,
      "eval_PRM F1": 0.7631578947368421,
      "eval_PRM F1 AUC": 0.610267155578837,
      "eval_PRM F1 AUC (fixed)": 0.3672079622839183,
      "eval_PRM F1 Neg": 0.4,
      "eval_PRM NPV": 0.32432432432432434,
      "eval_PRM Precision": 0.8405797101449275,
      "eval_PRM Recall": 0.6987951807228916,
      "eval_PRM Specificty": 0.5217391304347826,
      "eval_loss": 0.5670130252838135,
      "eval_runtime": 14.5479,
      "eval_samples_per_second": 3.024,
      "eval_steps_per_second": 0.137,
      "step": 105
    },
    {
      "epoch": 0.02745136123790101,
      "grad_norm": 29.200812937204702,
      "learning_rate": 3.423772609819122e-07,
      "loss": 0.8313,
      "step": 106
    },
    {
      "epoch": 0.02771033634391894,
      "grad_norm": 7.770870043903371,
      "learning_rate": 3.4560723514211885e-07,
      "loss": 0.573,
      "step": 107
    },
    {
      "epoch": 0.027969311449936876,
      "grad_norm": 21.63547347590896,
      "learning_rate": 3.488372093023256e-07,
      "loss": 0.7248,
      "step": 108
    },
    {
      "epoch": 0.028228286555954808,
      "grad_norm": 13.313977325056683,
      "learning_rate": 3.5206718346253235e-07,
      "loss": 0.6245,
      "step": 109
    },
    {
      "epoch": 0.028487261661972744,
      "grad_norm": 10.424747237877718,
      "learning_rate": 3.55297157622739e-07,
      "loss": 0.6744,
      "step": 110
    },
    {
      "epoch": 0.028487261661972744,
      "eval_PRM Accuracy": 0.6698113207547169,
      "eval_PRM F1": 0.7712418300653595,
      "eval_PRM F1 AUC": 0.6162912519643793,
      "eval_PRM F1 AUC (fixed)": 0.3716605552645364,
      "eval_PRM F1 Neg": 0.4067796610169492,
      "eval_PRM NPV": 0.3333333333333333,
      "eval_PRM Precision": 0.8428571428571429,
      "eval_PRM Recall": 0.7108433734939759,
      "eval_PRM Specificty": 0.5217391304347826,
      "eval_loss": 0.55446857213974,
      "eval_runtime": 15.0221,
      "eval_samples_per_second": 2.929,
      "eval_steps_per_second": 0.133,
      "step": 110
    },
    {
      "epoch": 0.028746236767990676,
      "grad_norm": 19.990118643252117,
      "learning_rate": 3.585271317829458e-07,
      "loss": 0.7346,
      "step": 111
    },
    {
      "epoch": 0.02900521187400861,
      "grad_norm": 11.664641699650158,
      "learning_rate": 3.6175710594315246e-07,
      "loss": 0.6889,
      "step": 112
    },
    {
      "epoch": 0.029264186980026544,
      "grad_norm": 26.966194495147256,
      "learning_rate": 3.649870801033592e-07,
      "loss": 0.678,
      "step": 113
    },
    {
      "epoch": 0.02952316208604448,
      "grad_norm": 17.964853935178834,
      "learning_rate": 3.6821705426356595e-07,
      "loss": 0.7198,
      "step": 114
    },
    {
      "epoch": 0.02978213719206241,
      "grad_norm": 14.85706395717968,
      "learning_rate": 3.714470284237726e-07,
      "loss": 0.6366,
      "step": 115
    },
    {
      "epoch": 0.02978213719206241,
      "eval_PRM Accuracy": 0.6792452830188679,
      "eval_PRM F1": 0.7733333333333333,
      "eval_PRM F1 AUC": 0.6537454164484023,
      "eval_PRM F1 AUC (fixed)": 0.3740178103719225,
      "eval_PRM F1 Neg": 0.45161290322580644,
      "eval_PRM NPV": 0.358974358974359,
      "eval_PRM Precision": 0.8656716417910447,
      "eval_PRM Recall": 0.6987951807228916,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.5416440963745117,
      "eval_runtime": 14.3887,
      "eval_samples_per_second": 3.058,
      "eval_steps_per_second": 0.139,
      "step": 115
    },
    {
      "epoch": 0.030041112298080347,
      "grad_norm": 16.61418454996819,
      "learning_rate": 3.746770025839794e-07,
      "loss": 0.725,
      "step": 116
    },
    {
      "epoch": 0.030300087404098282,
      "grad_norm": 9.534591885083701,
      "learning_rate": 3.7790697674418606e-07,
      "loss": 0.6432,
      "step": 117
    },
    {
      "epoch": 0.030559062510116215,
      "grad_norm": 7.169368696427665,
      "learning_rate": 3.811369509043928e-07,
      "loss": 0.6227,
      "step": 118
    },
    {
      "epoch": 0.03081803761613415,
      "grad_norm": 10.08722788125257,
      "learning_rate": 3.843669250645995e-07,
      "loss": 0.7123,
      "step": 119
    },
    {
      "epoch": 0.031077012722152082,
      "grad_norm": 8.13895725897613,
      "learning_rate": 3.8759689922480623e-07,
      "loss": 0.6191,
      "step": 120
    },
    {
      "epoch": 0.031077012722152082,
      "eval_PRM Accuracy": 0.6698113207547169,
      "eval_PRM F1": 0.7586206896551724,
      "eval_PRM F1 AUC": 0.679151388161341,
      "eval_PRM F1 AUC (fixed)": 0.3795180722891566,
      "eval_PRM F1 Neg": 0.47761194029850745,
      "eval_PRM NPV": 0.36363636363636365,
      "eval_PRM Precision": 0.8870967741935484,
      "eval_PRM Recall": 0.6626506024096386,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.5411401391029358,
      "eval_runtime": 14.5257,
      "eval_samples_per_second": 3.029,
      "eval_steps_per_second": 0.138,
      "step": 120
    },
    {
      "epoch": 0.031335987828170014,
      "grad_norm": 6.215018883416692,
      "learning_rate": 3.90826873385013e-07,
      "loss": 0.5748,
      "step": 121
    },
    {
      "epoch": 0.03159496293418795,
      "grad_norm": 8.35797483869244,
      "learning_rate": 3.9405684754521967e-07,
      "loss": 0.5726,
      "step": 122
    },
    {
      "epoch": 0.031853938040205886,
      "grad_norm": 8.05244834758129,
      "learning_rate": 3.972868217054264e-07,
      "loss": 0.6634,
      "step": 123
    },
    {
      "epoch": 0.03211291314622382,
      "grad_norm": 6.625684614301109,
      "learning_rate": 4.0051679586563306e-07,
      "loss": 0.6036,
      "step": 124
    },
    {
      "epoch": 0.03237188825224176,
      "grad_norm": 9.791133812266091,
      "learning_rate": 4.0374677002583983e-07,
      "loss": 0.5487,
      "step": 125
    },
    {
      "epoch": 0.03237188825224176,
      "eval_PRM Accuracy": 0.6415094339622641,
      "eval_PRM F1": 0.7285714285714285,
      "eval_PRM F1 AUC": 0.676794133053955,
      "eval_PRM F1 AUC (fixed)": 0.38344683080146674,
      "eval_PRM F1 Neg": 0.4722222222222222,
      "eval_PRM NPV": 0.3469387755102041,
      "eval_PRM Precision": 0.8947368421052632,
      "eval_PRM Recall": 0.6144578313253012,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.5412638187408447,
      "eval_runtime": 14.6931,
      "eval_samples_per_second": 2.995,
      "eval_steps_per_second": 0.136,
      "step": 125
    },
    {
      "epoch": 0.03263086335825969,
      "grad_norm": 6.757491658349913,
      "learning_rate": 4.0697674418604655e-07,
      "loss": 0.5949,
      "step": 126
    },
    {
      "epoch": 0.03288983846427762,
      "grad_norm": 26.94630756049828,
      "learning_rate": 4.102067183462533e-07,
      "loss": 0.7258,
      "step": 127
    },
    {
      "epoch": 0.03314881357029555,
      "grad_norm": 14.68796120813943,
      "learning_rate": 4.1343669250646e-07,
      "loss": 0.6737,
      "step": 128
    },
    {
      "epoch": 0.03340778867631349,
      "grad_norm": 13.84775550814014,
      "learning_rate": 4.1666666666666667e-07,
      "loss": 0.5792,
      "step": 129
    },
    {
      "epoch": 0.033666763782331424,
      "grad_norm": 15.319435558173073,
      "learning_rate": 4.1989664082687344e-07,
      "loss": 0.7407,
      "step": 130
    },
    {
      "epoch": 0.033666763782331424,
      "eval_PRM Accuracy": 0.660377358490566,
      "eval_PRM F1": 0.7428571428571429,
      "eval_PRM F1 AUC": 0.7045573598742797,
      "eval_PRM F1 AUC (fixed)": 0.3805657412257727,
      "eval_PRM F1 Neg": 0.5,
      "eval_PRM NPV": 0.3673469387755102,
      "eval_PRM Precision": 0.9122807017543859,
      "eval_PRM Recall": 0.6265060240963856,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.5402398705482483,
      "eval_runtime": 14.4767,
      "eval_samples_per_second": 3.039,
      "eval_steps_per_second": 0.138,
      "step": 130
    },
    {
      "epoch": 0.033925738888349356,
      "grad_norm": 11.219710300102047,
      "learning_rate": 4.2312661498708016e-07,
      "loss": 0.6967,
      "step": 131
    },
    {
      "epoch": 0.03418471399436729,
      "grad_norm": 9.028415286023842,
      "learning_rate": 4.2635658914728683e-07,
      "loss": 0.5366,
      "step": 132
    },
    {
      "epoch": 0.03444368910038523,
      "grad_norm": 12.524874702176467,
      "learning_rate": 4.295865633074936e-07,
      "loss": 0.6446,
      "step": 133
    },
    {
      "epoch": 0.03470266420640316,
      "grad_norm": 13.610218061273281,
      "learning_rate": 4.3281653746770027e-07,
      "loss": 0.6463,
      "step": 134
    },
    {
      "epoch": 0.03496163931242109,
      "grad_norm": 8.571461600608345,
      "learning_rate": 4.3604651162790705e-07,
      "loss": 0.6441,
      "step": 135
    },
    {
      "epoch": 0.03496163931242109,
      "eval_PRM Accuracy": 0.6698113207547169,
      "eval_PRM F1": 0.75177304964539,
      "eval_PRM F1 AUC": 0.7105814562598219,
      "eval_PRM F1 AUC (fixed)": 0.38580408590885285,
      "eval_PRM F1 Neg": 0.5070422535211268,
      "eval_PRM NPV": 0.375,
      "eval_PRM Precision": 0.9137931034482759,
      "eval_PRM Recall": 0.6385542168674698,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.5414556860923767,
      "eval_runtime": 14.4384,
      "eval_samples_per_second": 3.047,
      "eval_steps_per_second": 0.139,
      "step": 135
    },
    {
      "epoch": 0.03522061441843903,
      "grad_norm": 7.16674151720117,
      "learning_rate": 4.392764857881137e-07,
      "loss": 0.6491,
      "step": 136
    },
    {
      "epoch": 0.03547958952445696,
      "grad_norm": 7.1083764687848605,
      "learning_rate": 4.4250645994832044e-07,
      "loss": 0.5751,
      "step": 137
    },
    {
      "epoch": 0.035738564630474895,
      "grad_norm": 9.698448189769538,
      "learning_rate": 4.457364341085272e-07,
      "loss": 0.5885,
      "step": 138
    },
    {
      "epoch": 0.03599753973649283,
      "grad_norm": 6.798621458927422,
      "learning_rate": 4.489664082687339e-07,
      "loss": 0.6095,
      "step": 139
    },
    {
      "epoch": 0.036256514842510766,
      "grad_norm": 7.661301814457179,
      "learning_rate": 4.521963824289406e-07,
      "loss": 0.6317,
      "step": 140
    },
    {
      "epoch": 0.036256514842510766,
      "eval_PRM Accuracy": 0.660377358490566,
      "eval_PRM F1": 0.7428571428571429,
      "eval_PRM F1 AUC": 0.7045573598742797,
      "eval_PRM F1 AUC (fixed)": 0.38789942378208486,
      "eval_PRM F1 Neg": 0.5,
      "eval_PRM NPV": 0.3673469387755102,
      "eval_PRM Precision": 0.9122807017543859,
      "eval_PRM Recall": 0.6265060240963856,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.5477471351623535,
      "eval_runtime": 14.1224,
      "eval_samples_per_second": 3.116,
      "eval_steps_per_second": 0.142,
      "step": 140
    },
    {
      "epoch": 0.0365154899485287,
      "grad_norm": 13.365479316439192,
      "learning_rate": 4.554263565891473e-07,
      "loss": 0.6813,
      "step": 141
    },
    {
      "epoch": 0.03677446505454663,
      "grad_norm": 10.854235911252928,
      "learning_rate": 4.5865633074935404e-07,
      "loss": 0.6343,
      "step": 142
    },
    {
      "epoch": 0.03703344016056456,
      "grad_norm": 6.478102222374132,
      "learning_rate": 4.618863049095608e-07,
      "loss": 0.5942,
      "step": 143
    },
    {
      "epoch": 0.0372924152665825,
      "grad_norm": 7.359112000214807,
      "learning_rate": 4.651162790697675e-07,
      "loss": 0.6848,
      "step": 144
    },
    {
      "epoch": 0.037551390372600434,
      "grad_norm": 13.133076829254982,
      "learning_rate": 4.683462532299742e-07,
      "loss": 0.6556,
      "step": 145
    },
    {
      "epoch": 0.037551390372600434,
      "eval_PRM Accuracy": 0.6698113207547169,
      "eval_PRM F1": 0.75177304964539,
      "eval_PRM F1 AUC": 0.7105814562598219,
      "eval_PRM F1 AUC (fixed)": 0.3899947616553169,
      "eval_PRM F1 Neg": 0.5070422535211268,
      "eval_PRM NPV": 0.375,
      "eval_PRM Precision": 0.9137931034482759,
      "eval_PRM Recall": 0.6385542168674698,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.5437543392181396,
      "eval_runtime": 14.4066,
      "eval_samples_per_second": 3.054,
      "eval_steps_per_second": 0.139,
      "step": 145
    },
    {
      "epoch": 0.037810365478618366,
      "grad_norm": 7.676893672467976,
      "learning_rate": 4.7157622739018087e-07,
      "loss": 0.6555,
      "step": 146
    },
    {
      "epoch": 0.038069340584636305,
      "grad_norm": 8.025843991999064,
      "learning_rate": 4.7480620155038765e-07,
      "loss": 0.5676,
      "step": 147
    },
    {
      "epoch": 0.03832831569065424,
      "grad_norm": 9.894553920917076,
      "learning_rate": 4.780361757105943e-07,
      "loss": 0.563,
      "step": 148
    },
    {
      "epoch": 0.03858729079667217,
      "grad_norm": 6.883678153936593,
      "learning_rate": 4.812661498708011e-07,
      "loss": 0.5804,
      "step": 149
    },
    {
      "epoch": 0.0388462659026901,
      "grad_norm": 8.261843621096295,
      "learning_rate": 4.844961240310079e-07,
      "loss": 0.6092,
      "step": 150
    },
    {
      "epoch": 0.0388462659026901,
      "eval_PRM Accuracy": 0.6792452830188679,
      "eval_PRM F1": 0.7605633802816901,
      "eval_PRM F1 AUC": 0.7166055526453641,
      "eval_PRM F1 AUC (fixed)": 0.38999476165531693,
      "eval_PRM F1 Neg": 0.5142857142857142,
      "eval_PRM NPV": 0.3829787234042553,
      "eval_PRM Precision": 0.9152542372881356,
      "eval_PRM Recall": 0.6506024096385542,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.540428102016449,
      "eval_runtime": 14.7992,
      "eval_samples_per_second": 2.973,
      "eval_steps_per_second": 0.135,
      "step": 150
    },
    {
      "epoch": 0.03910524100870804,
      "grad_norm": 7.354522673332907,
      "learning_rate": 4.877260981912145e-07,
      "loss": 0.5619,
      "step": 151
    },
    {
      "epoch": 0.03936421611472597,
      "grad_norm": 10.294306829596843,
      "learning_rate": 4.909560723514212e-07,
      "loss": 0.6696,
      "step": 152
    },
    {
      "epoch": 0.039623191220743904,
      "grad_norm": 8.641933608314154,
      "learning_rate": 4.94186046511628e-07,
      "loss": 0.5843,
      "step": 153
    },
    {
      "epoch": 0.03988216632676184,
      "grad_norm": 9.039792206894193,
      "learning_rate": 4.974160206718346e-07,
      "loss": 0.6636,
      "step": 154
    },
    {
      "epoch": 0.040141141432779776,
      "grad_norm": 7.8634665219535975,
      "learning_rate": 5.006459948320414e-07,
      "loss": 0.5328,
      "step": 155
    },
    {
      "epoch": 0.040141141432779776,
      "eval_PRM Accuracy": 0.6981132075471698,
      "eval_PRM F1": 0.7746478873239436,
      "eval_PRM F1 AUC": 0.7443687794656888,
      "eval_PRM F1 AUC (fixed)": 0.3988999476165532,
      "eval_PRM F1 Neg": 0.5428571428571428,
      "eval_PRM NPV": 0.40425531914893614,
      "eval_PRM Precision": 0.9322033898305084,
      "eval_PRM Recall": 0.6626506024096386,
      "eval_PRM Specificty": 0.8260869565217391,
      "eval_loss": 0.534451425075531,
      "eval_runtime": 14.5832,
      "eval_samples_per_second": 3.017,
      "eval_steps_per_second": 0.137,
      "step": 155
    },
    {
      "epoch": 0.04040011653879771,
      "grad_norm": 9.34303127109423,
      "learning_rate": 5.038759689922481e-07,
      "loss": 0.6202,
      "step": 156
    },
    {
      "epoch": 0.04065909164481564,
      "grad_norm": 6.195118698281523,
      "learning_rate": 5.071059431524549e-07,
      "loss": 0.572,
      "step": 157
    },
    {
      "epoch": 0.04091806675083358,
      "grad_norm": 8.108721137426325,
      "learning_rate": 5.103359173126615e-07,
      "loss": 0.7191,
      "step": 158
    },
    {
      "epoch": 0.04117704185685151,
      "grad_norm": 12.409323930943694,
      "learning_rate": 5.135658914728683e-07,
      "loss": 0.4859,
      "step": 159
    },
    {
      "epoch": 0.04143601696286944,
      "grad_norm": 13.610909760223025,
      "learning_rate": 5.16795865633075e-07,
      "loss": 0.5121,
      "step": 160
    },
    {
      "epoch": 0.04143601696286944,
      "eval_PRM Accuracy": 0.7075471698113207,
      "eval_PRM F1": 0.7891156462585034,
      "eval_PRM F1 AUC": 0.7189628077527501,
      "eval_PRM F1 AUC (fixed)": 0.39392352016762705,
      "eval_PRM F1 Neg": 0.5230769230769231,
      "eval_PRM NPV": 0.40476190476190477,
      "eval_PRM Precision": 0.90625,
      "eval_PRM Recall": 0.6987951807228916,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.5245890617370605,
      "eval_runtime": 14.5943,
      "eval_samples_per_second": 3.015,
      "eval_steps_per_second": 0.137,
      "step": 160
    },
    {
      "epoch": 0.041694992068887375,
      "grad_norm": 6.347252875639677,
      "learning_rate": 5.200258397932817e-07,
      "loss": 0.5312,
      "step": 161
    },
    {
      "epoch": 0.041953967174905314,
      "grad_norm": 7.171273813798448,
      "learning_rate": 5.232558139534884e-07,
      "loss": 0.5634,
      "step": 162
    },
    {
      "epoch": 0.042212942280923246,
      "grad_norm": 14.883956010412117,
      "learning_rate": 5.264857881136951e-07,
      "loss": 0.5834,
      "step": 163
    },
    {
      "epoch": 0.04247191738694118,
      "grad_norm": 5.869889554466555,
      "learning_rate": 5.297157622739019e-07,
      "loss": 0.5194,
      "step": 164
    },
    {
      "epoch": 0.04273089249295912,
      "grad_norm": 11.870697609022255,
      "learning_rate": 5.329457364341085e-07,
      "loss": 0.5417,
      "step": 165
    },
    {
      "epoch": 0.04273089249295912,
      "eval_PRM Accuracy": 0.7264150943396226,
      "eval_PRM F1": 0.8053691275167785,
      "eval_PRM F1 AUC": 0.7310110005238345,
      "eval_PRM F1 AUC (fixed)": 0.3994237820848612,
      "eval_PRM F1 Neg": 0.5396825396825397,
      "eval_PRM NPV": 0.425,
      "eval_PRM Precision": 0.9090909090909091,
      "eval_PRM Recall": 0.7228915662650602,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.514342188835144,
      "eval_runtime": 14.3157,
      "eval_samples_per_second": 3.074,
      "eval_steps_per_second": 0.14,
      "step": 165
    },
    {
      "epoch": 0.04298986759897705,
      "grad_norm": 5.691938211646347,
      "learning_rate": 5.361757105943153e-07,
      "loss": 0.6143,
      "step": 166
    },
    {
      "epoch": 0.04324884270499498,
      "grad_norm": 9.493665131614394,
      "learning_rate": 5.394056847545221e-07,
      "loss": 0.5261,
      "step": 167
    },
    {
      "epoch": 0.043507817811012914,
      "grad_norm": 6.1221521139428,
      "learning_rate": 5.426356589147287e-07,
      "loss": 0.5844,
      "step": 168
    },
    {
      "epoch": 0.04376679291703085,
      "grad_norm": 5.228074975287801,
      "learning_rate": 5.458656330749355e-07,
      "loss": 0.571,
      "step": 169
    },
    {
      "epoch": 0.044025768023048785,
      "grad_norm": 14.423948863883327,
      "learning_rate": 5.490956072351422e-07,
      "loss": 0.6379,
      "step": 170
    },
    {
      "epoch": 0.044025768023048785,
      "eval_PRM Accuracy": 0.7264150943396226,
      "eval_PRM F1": 0.8104575163398693,
      "eval_PRM F1 AUC": 0.6995809324253536,
      "eval_PRM F1 AUC (fixed)": 0.39837611314824517,
      "eval_PRM F1 Neg": 0.5084745762711864,
      "eval_PRM NPV": 0.4166666666666667,
      "eval_PRM Precision": 0.8857142857142857,
      "eval_PRM Recall": 0.7469879518072289,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.5093567967414856,
      "eval_runtime": 14.4938,
      "eval_samples_per_second": 3.036,
      "eval_steps_per_second": 0.138,
      "step": 170
    },
    {
      "epoch": 0.04428474312906672,
      "grad_norm": 10.344568168628063,
      "learning_rate": 5.523255813953489e-07,
      "loss": 0.5654,
      "step": 171
    },
    {
      "epoch": 0.04454371823508465,
      "grad_norm": 8.542836484798716,
      "learning_rate": 5.555555555555555e-07,
      "loss": 0.5971,
      "step": 172
    },
    {
      "epoch": 0.04480269334110259,
      "grad_norm": 8.19809400190519,
      "learning_rate": 5.587855297157623e-07,
      "loss": 0.5235,
      "step": 173
    },
    {
      "epoch": 0.04506166844712052,
      "grad_norm": 6.077323198910218,
      "learning_rate": 5.620155038759691e-07,
      "loss": 0.5564,
      "step": 174
    },
    {
      "epoch": 0.04532064355313845,
      "grad_norm": 5.288812927319927,
      "learning_rate": 5.652454780361757e-07,
      "loss": 0.5437,
      "step": 175
    },
    {
      "epoch": 0.04532064355313845,
      "eval_PRM Accuracy": 0.7075471698113207,
      "eval_PRM F1": 0.7947019867549668,
      "eval_PRM F1 AUC": 0.6875327396542692,
      "eval_PRM F1 AUC (fixed)": 0.4044002095337874,
      "eval_PRM F1 Neg": 0.4918032786885246,
      "eval_PRM NPV": 0.39473684210526316,
      "eval_PRM Precision": 0.8823529411764706,
      "eval_PRM Recall": 0.7228915662650602,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.5160964727401733,
      "eval_runtime": 14.5801,
      "eval_samples_per_second": 3.018,
      "eval_steps_per_second": 0.137,
      "step": 175
    },
    {
      "epoch": 0.04557961865915639,
      "grad_norm": 5.935722932915326,
      "learning_rate": 5.684754521963825e-07,
      "loss": 0.4989,
      "step": 176
    },
    {
      "epoch": 0.045838593765174324,
      "grad_norm": 10.165376980754218,
      "learning_rate": 5.717054263565892e-07,
      "loss": 0.4974,
      "step": 177
    },
    {
      "epoch": 0.046097568871192256,
      "grad_norm": 8.844537517788227,
      "learning_rate": 5.74935400516796e-07,
      "loss": 0.5733,
      "step": 178
    },
    {
      "epoch": 0.04635654397721019,
      "grad_norm": 6.136922569828234,
      "learning_rate": 5.781653746770026e-07,
      "loss": 0.5724,
      "step": 179
    },
    {
      "epoch": 0.04661551908322813,
      "grad_norm": 6.093958466635018,
      "learning_rate": 5.813953488372094e-07,
      "loss": 0.543,
      "step": 180
    },
    {
      "epoch": 0.04661551908322813,
      "eval_PRM Accuracy": 0.7075471698113207,
      "eval_PRM F1": 0.7973856209150327,
      "eval_PRM F1 AUC": 0.671817705605029,
      "eval_PRM F1 AUC (fixed)": 0.40518596123624934,
      "eval_PRM F1 Neg": 0.4745762711864407,
      "eval_PRM NPV": 0.3888888888888889,
      "eval_PRM Precision": 0.8714285714285714,
      "eval_PRM Recall": 0.7349397590361446,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.5174723863601685,
      "eval_runtime": 14.5846,
      "eval_samples_per_second": 3.017,
      "eval_steps_per_second": 0.137,
      "step": 180
    },
    {
      "epoch": 0.04687449418924606,
      "grad_norm": 10.776808527067512,
      "learning_rate": 5.846253229974161e-07,
      "loss": 0.5772,
      "step": 181
    },
    {
      "epoch": 0.04713346929526399,
      "grad_norm": 10.033903169098872,
      "learning_rate": 5.878552971576227e-07,
      "loss": 0.4632,
      "step": 182
    },
    {
      "epoch": 0.04739244440128193,
      "grad_norm": 6.481504308087625,
      "learning_rate": 5.910852713178295e-07,
      "loss": 0.516,
      "step": 183
    },
    {
      "epoch": 0.04765141950729986,
      "grad_norm": 6.015599645256489,
      "learning_rate": 5.943152454780362e-07,
      "loss": 0.5046,
      "step": 184
    },
    {
      "epoch": 0.047910394613317794,
      "grad_norm": 13.76113181680478,
      "learning_rate": 5.97545219638243e-07,
      "loss": 0.5823,
      "step": 185
    },
    {
      "epoch": 0.047910394613317794,
      "eval_PRM Accuracy": 0.7075471698113207,
      "eval_PRM F1": 0.7973856209150327,
      "eval_PRM F1 AUC": 0.671817705605029,
      "eval_PRM F1 AUC (fixed)": 0.4062336301728654,
      "eval_PRM F1 Neg": 0.4745762711864407,
      "eval_PRM NPV": 0.3888888888888889,
      "eval_PRM Precision": 0.8714285714285714,
      "eval_PRM Recall": 0.7349397590361446,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.5205953121185303,
      "eval_runtime": 14.829,
      "eval_samples_per_second": 2.967,
      "eval_steps_per_second": 0.135,
      "step": 185
    },
    {
      "epoch": 0.048169369719335726,
      "grad_norm": 6.725385680199521,
      "learning_rate": 6.007751937984497e-07,
      "loss": 0.4939,
      "step": 186
    },
    {
      "epoch": 0.048428344825353666,
      "grad_norm": 10.561746795952903,
      "learning_rate": 6.040051679586564e-07,
      "loss": 0.5041,
      "step": 187
    },
    {
      "epoch": 0.0486873199313716,
      "grad_norm": 9.334572719128575,
      "learning_rate": 6.072351421188632e-07,
      "loss": 0.5234,
      "step": 188
    },
    {
      "epoch": 0.04894629503738953,
      "grad_norm": 5.443979412272866,
      "learning_rate": 6.104651162790698e-07,
      "loss": 0.4918,
      "step": 189
    },
    {
      "epoch": 0.04920527014340746,
      "grad_norm": 5.453809365018693,
      "learning_rate": 6.136950904392765e-07,
      "loss": 0.5384,
      "step": 190
    },
    {
      "epoch": 0.04920527014340746,
      "eval_PRM Accuracy": 0.7358490566037735,
      "eval_PRM F1": 0.8205128205128205,
      "eval_PRM F1 AUC": 0.6898899947616554,
      "eval_PRM F1 AUC (fixed)": 0.40859088528025145,
      "eval_PRM F1 Neg": 0.5,
      "eval_PRM NPV": 0.42424242424242425,
      "eval_PRM Precision": 0.8767123287671232,
      "eval_PRM Recall": 0.7710843373493976,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.5035573840141296,
      "eval_runtime": 14.772,
      "eval_samples_per_second": 2.979,
      "eval_steps_per_second": 0.135,
      "step": 190
    },
    {
      "epoch": 0.0494642452494254,
      "grad_norm": 16.244963846443802,
      "learning_rate": 6.169250645994833e-07,
      "loss": 0.6508,
      "step": 191
    },
    {
      "epoch": 0.04972322035544333,
      "grad_norm": 8.932185819469234,
      "learning_rate": 6.201550387596899e-07,
      "loss": 0.6221,
      "step": 192
    },
    {
      "epoch": 0.049982195461461265,
      "grad_norm": 4.408328610073295,
      "learning_rate": 6.233850129198967e-07,
      "loss": 0.4826,
      "step": 193
    },
    {
      "epoch": 0.050241170567479204,
      "grad_norm": 13.406641686428832,
      "learning_rate": 6.266149870801035e-07,
      "loss": 0.6162,
      "step": 194
    },
    {
      "epoch": 0.050500145673497136,
      "grad_norm": 4.892120858093281,
      "learning_rate": 6.298449612403102e-07,
      "loss": 0.4512,
      "step": 195
    },
    {
      "epoch": 0.050500145673497136,
      "eval_PRM Accuracy": 0.7452830188679245,
      "eval_PRM F1": 0.8235294117647058,
      "eval_PRM F1 AUC": 0.7273441592456783,
      "eval_PRM F1 AUC (fixed)": 0.41382922996333155,
      "eval_PRM F1 Neg": 0.5423728813559322,
      "eval_PRM NPV": 0.4444444444444444,
      "eval_PRM Precision": 0.9,
      "eval_PRM Recall": 0.7590361445783133,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.5021147131919861,
      "eval_runtime": 14.9083,
      "eval_samples_per_second": 2.951,
      "eval_steps_per_second": 0.134,
      "step": 195
    },
    {
      "epoch": 0.05075912077951507,
      "grad_norm": 5.6387526600892315,
      "learning_rate": 6.330749354005168e-07,
      "loss": 0.4776,
      "step": 196
    },
    {
      "epoch": 0.051018095885533,
      "grad_norm": 7.286720230552998,
      "learning_rate": 6.363049095607236e-07,
      "loss": 0.5219,
      "step": 197
    },
    {
      "epoch": 0.05127707099155094,
      "grad_norm": 4.997205287307194,
      "learning_rate": 6.395348837209303e-07,
      "loss": 0.5234,
      "step": 198
    },
    {
      "epoch": 0.05153604609756887,
      "grad_norm": 6.166767867082422,
      "learning_rate": 6.427648578811369e-07,
      "loss": 0.4966,
      "step": 199
    },
    {
      "epoch": 0.051795021203586804,
      "grad_norm": 5.2138937231508695,
      "learning_rate": 6.459948320413437e-07,
      "loss": 0.5545,
      "step": 200
    },
    {
      "epoch": 0.051795021203586804,
      "eval_PRM Accuracy": 0.7452830188679245,
      "eval_PRM F1": 0.8235294117647058,
      "eval_PRM F1 AUC": 0.7273441592456783,
      "eval_PRM F1 AUC (fixed)": 0.4232582503928759,
      "eval_PRM F1 Neg": 0.5423728813559322,
      "eval_PRM NPV": 0.4444444444444444,
      "eval_PRM Precision": 0.9,
      "eval_PRM Recall": 0.7590361445783133,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.49177101254463196,
      "eval_runtime": 15.1242,
      "eval_samples_per_second": 2.909,
      "eval_steps_per_second": 0.132,
      "step": 200
    },
    {
      "epoch": 0.052053996309604736,
      "grad_norm": 13.586487332304372,
      "learning_rate": 6.492248062015505e-07,
      "loss": 0.5246,
      "step": 201
    },
    {
      "epoch": 0.052312971415622675,
      "grad_norm": 6.063157983759811,
      "learning_rate": 6.524547803617572e-07,
      "loss": 0.4992,
      "step": 202
    },
    {
      "epoch": 0.05257194652164061,
      "grad_norm": 6.388364255784616,
      "learning_rate": 6.556847545219638e-07,
      "loss": 0.4724,
      "step": 203
    },
    {
      "epoch": 0.05283092162765854,
      "grad_norm": 11.43907528969819,
      "learning_rate": 6.589147286821707e-07,
      "loss": 0.5736,
      "step": 204
    },
    {
      "epoch": 0.05308989673367648,
      "grad_norm": 16.818113355638936,
      "learning_rate": 6.621447028423774e-07,
      "loss": 0.6059,
      "step": 205
    },
    {
      "epoch": 0.05308989673367648,
      "eval_PRM Accuracy": 0.7452830188679245,
      "eval_PRM F1": 0.8235294117647058,
      "eval_PRM F1 AUC": 0.7273441592456783,
      "eval_PRM F1 AUC (fixed)": 0.42744892613933994,
      "eval_PRM F1 Neg": 0.5423728813559322,
      "eval_PRM NPV": 0.4444444444444444,
      "eval_PRM Precision": 0.9,
      "eval_PRM Recall": 0.7590361445783133,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.4973340928554535,
      "eval_runtime": 14.7257,
      "eval_samples_per_second": 2.988,
      "eval_steps_per_second": 0.136,
      "step": 205
    },
    {
      "epoch": 0.05334887183969441,
      "grad_norm": 9.323152018153321,
      "learning_rate": 6.65374677002584e-07,
      "loss": 0.4315,
      "step": 206
    },
    {
      "epoch": 0.05360784694571234,
      "grad_norm": 6.5863056309687416,
      "learning_rate": 6.686046511627907e-07,
      "loss": 0.5517,
      "step": 207
    },
    {
      "epoch": 0.053866822051730275,
      "grad_norm": 5.274715923239226,
      "learning_rate": 6.718346253229975e-07,
      "loss": 0.4108,
      "step": 208
    },
    {
      "epoch": 0.054125797157748214,
      "grad_norm": 11.927133654509566,
      "learning_rate": 6.750645994832042e-07,
      "loss": 0.5993,
      "step": 209
    },
    {
      "epoch": 0.054384772263766146,
      "grad_norm": 6.0307980409502155,
      "learning_rate": 6.782945736434108e-07,
      "loss": 0.5751,
      "step": 210
    },
    {
      "epoch": 0.054384772263766146,
      "eval_PRM Accuracy": 0.7547169811320755,
      "eval_PRM F1": 0.8289473684210527,
      "eval_PRM F1 AUC": 0.749083289680461,
      "eval_PRM F1 AUC (fixed)": 0.43766369827134627,
      "eval_PRM F1 Neg": 0.5666666666666667,
      "eval_PRM NPV": 0.4594594594594595,
      "eval_PRM Precision": 0.9130434782608695,
      "eval_PRM Recall": 0.7590361445783133,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.49228909611701965,
      "eval_runtime": 14.7251,
      "eval_samples_per_second": 2.988,
      "eval_steps_per_second": 0.136,
      "step": 210
    },
    {
      "epoch": 0.05464374736978408,
      "grad_norm": 12.144312992076157,
      "learning_rate": 6.815245478036177e-07,
      "loss": 0.4427,
      "step": 211
    },
    {
      "epoch": 0.05490272247580202,
      "grad_norm": 5.780920751157059,
      "learning_rate": 6.847545219638244e-07,
      "loss": 0.6016,
      "step": 212
    },
    {
      "epoch": 0.05516169758181995,
      "grad_norm": 5.163882391479198,
      "learning_rate": 6.87984496124031e-07,
      "loss": 0.5456,
      "step": 213
    },
    {
      "epoch": 0.05542067268783788,
      "grad_norm": 12.425837231569478,
      "learning_rate": 6.912144702842377e-07,
      "loss": 0.4962,
      "step": 214
    },
    {
      "epoch": 0.05567964779385581,
      "grad_norm": 4.927140435616092,
      "learning_rate": 6.944444444444446e-07,
      "loss": 0.4614,
      "step": 215
    },
    {
      "epoch": 0.05567964779385581,
      "eval_PRM Accuracy": 0.7547169811320755,
      "eval_PRM F1": 0.8289473684210527,
      "eval_PRM F1 AUC": 0.749083289680461,
      "eval_PRM F1 AUC (fixed)": 0.43871136720796233,
      "eval_PRM F1 Neg": 0.5666666666666667,
      "eval_PRM NPV": 0.4594594594594595,
      "eval_PRM Precision": 0.9130434782608695,
      "eval_PRM Recall": 0.7590361445783133,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.48628750443458557,
      "eval_runtime": 14.8785,
      "eval_samples_per_second": 2.957,
      "eval_steps_per_second": 0.134,
      "step": 215
    },
    {
      "epoch": 0.05593862289987375,
      "grad_norm": 6.655940409281741,
      "learning_rate": 6.976744186046513e-07,
      "loss": 0.4614,
      "step": 216
    },
    {
      "epoch": 0.056197598005891684,
      "grad_norm": 8.717195609708812,
      "learning_rate": 7.009043927648579e-07,
      "loss": 0.6091,
      "step": 217
    },
    {
      "epoch": 0.056456573111909616,
      "grad_norm": 9.066839424841193,
      "learning_rate": 7.041343669250647e-07,
      "loss": 0.5783,
      "step": 218
    },
    {
      "epoch": 0.05671554821792755,
      "grad_norm": 6.088716178459927,
      "learning_rate": 7.073643410852714e-07,
      "loss": 0.4702,
      "step": 219
    },
    {
      "epoch": 0.05697452332394549,
      "grad_norm": 6.681353857769403,
      "learning_rate": 7.10594315245478e-07,
      "loss": 0.483,
      "step": 220
    },
    {
      "epoch": 0.05697452332394549,
      "eval_PRM Accuracy": 0.7452830188679245,
      "eval_PRM F1": 0.8235294117647058,
      "eval_PRM F1 AUC": 0.7273441592456783,
      "eval_PRM F1 AUC (fixed)": 0.44473546359350447,
      "eval_PRM F1 Neg": 0.5423728813559322,
      "eval_PRM NPV": 0.4444444444444444,
      "eval_PRM Precision": 0.9,
      "eval_PRM Recall": 0.7590361445783133,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.4895632565021515,
      "eval_runtime": 14.7295,
      "eval_samples_per_second": 2.987,
      "eval_steps_per_second": 0.136,
      "step": 220
    },
    {
      "epoch": 0.05723349842996342,
      "grad_norm": 5.913098681014986,
      "learning_rate": 7.138242894056849e-07,
      "loss": 0.5357,
      "step": 221
    },
    {
      "epoch": 0.05749247353598135,
      "grad_norm": 5.158407140201309,
      "learning_rate": 7.170542635658916e-07,
      "loss": 0.4635,
      "step": 222
    },
    {
      "epoch": 0.05775144864199929,
      "grad_norm": 6.47016455998915,
      "learning_rate": 7.202842377260982e-07,
      "loss": 0.4952,
      "step": 223
    },
    {
      "epoch": 0.05801042374801722,
      "grad_norm": 10.23651185109097,
      "learning_rate": 7.235142118863049e-07,
      "loss": 0.5544,
      "step": 224
    },
    {
      "epoch": 0.058269398854035155,
      "grad_norm": 4.578430497378247,
      "learning_rate": 7.267441860465117e-07,
      "loss": 0.5148,
      "step": 225
    },
    {
      "epoch": 0.058269398854035155,
      "eval_PRM Accuracy": 0.7547169811320755,
      "eval_PRM F1": 0.8289473684210527,
      "eval_PRM F1 AUC": 0.749083289680461,
      "eval_PRM F1 AUC (fixed)": 0.4499738082765845,
      "eval_PRM F1 Neg": 0.5666666666666667,
      "eval_PRM NPV": 0.4594594594594595,
      "eval_PRM Precision": 0.9130434782608695,
      "eval_PRM Recall": 0.7590361445783133,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.4968224763870239,
      "eval_runtime": 15.0589,
      "eval_samples_per_second": 2.922,
      "eval_steps_per_second": 0.133,
      "step": 225
    },
    {
      "epoch": 0.05852837396005309,
      "grad_norm": 12.668163398816281,
      "learning_rate": 7.299741602067184e-07,
      "loss": 0.4264,
      "step": 226
    },
    {
      "epoch": 0.058787349066071026,
      "grad_norm": 4.73940430219435,
      "learning_rate": 7.332041343669251e-07,
      "loss": 0.5648,
      "step": 227
    },
    {
      "epoch": 0.05904632417208896,
      "grad_norm": 5.39766175412595,
      "learning_rate": 7.364341085271319e-07,
      "loss": 0.4793,
      "step": 228
    },
    {
      "epoch": 0.05930529927810689,
      "grad_norm": 11.661839654745418,
      "learning_rate": 7.396640826873386e-07,
      "loss": 0.5498,
      "step": 229
    },
    {
      "epoch": 0.05956427438412482,
      "grad_norm": 6.043419075157401,
      "learning_rate": 7.428940568475452e-07,
      "loss": 0.4511,
      "step": 230
    },
    {
      "epoch": 0.05956427438412482,
      "eval_PRM Accuracy": 0.7547169811320755,
      "eval_PRM F1": 0.8266666666666667,
      "eval_PRM F1 AUC": 0.7647983237297015,
      "eval_PRM F1 AUC (fixed)": 0.4562598218962808,
      "eval_PRM F1 Neg": 0.5806451612903226,
      "eval_PRM NPV": 0.46153846153846156,
      "eval_PRM Precision": 0.9253731343283582,
      "eval_PRM Recall": 0.7469879518072289,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.49978208541870117,
      "eval_runtime": 14.4473,
      "eval_samples_per_second": 3.046,
      "eval_steps_per_second": 0.138,
      "step": 230
    },
    {
      "epoch": 0.05982324949014276,
      "grad_norm": 7.968824617613928,
      "learning_rate": 7.461240310077519e-07,
      "loss": 0.5601,
      "step": 231
    },
    {
      "epoch": 0.060082224596160694,
      "grad_norm": 4.1475482933198595,
      "learning_rate": 7.493540051679588e-07,
      "loss": 0.508,
      "step": 232
    },
    {
      "epoch": 0.060341199702178626,
      "grad_norm": 10.162275739039046,
      "learning_rate": 7.525839793281655e-07,
      "loss": 0.4295,
      "step": 233
    },
    {
      "epoch": 0.060600174808196565,
      "grad_norm": 17.411414443373726,
      "learning_rate": 7.558139534883721e-07,
      "loss": 0.5595,
      "step": 234
    },
    {
      "epoch": 0.0608591499142145,
      "grad_norm": 6.93839153430805,
      "learning_rate": 7.590439276485789e-07,
      "loss": 0.5302,
      "step": 235
    },
    {
      "epoch": 0.0608591499142145,
      "eval_PRM Accuracy": 0.7264150943396226,
      "eval_PRM F1": 0.8027210884353742,
      "eval_PRM F1 AUC": 0.746726034573075,
      "eval_PRM F1 AUC (fixed)": 0.45468831849135677,
      "eval_PRM F1 Neg": 0.5538461538461539,
      "eval_PRM NPV": 0.42857142857142855,
      "eval_PRM Precision": 0.921875,
      "eval_PRM Recall": 0.7108433734939759,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.5159228444099426,
      "eval_runtime": 14.8959,
      "eval_samples_per_second": 2.954,
      "eval_steps_per_second": 0.134,
      "step": 235
    },
    {
      "epoch": 0.06111812502023243,
      "grad_norm": 14.524608563458866,
      "learning_rate": 7.622739018087856e-07,
      "loss": 0.5027,
      "step": 236
    },
    {
      "epoch": 0.06137710012625036,
      "grad_norm": 12.416401593553305,
      "learning_rate": 7.655038759689922e-07,
      "loss": 0.5593,
      "step": 237
    },
    {
      "epoch": 0.0616360752322683,
      "grad_norm": 14.295373057361312,
      "learning_rate": 7.68733850129199e-07,
      "loss": 0.6659,
      "step": 238
    },
    {
      "epoch": 0.06189505033828623,
      "grad_norm": 5.522965374953522,
      "learning_rate": 7.719638242894058e-07,
      "loss": 0.5088,
      "step": 239
    },
    {
      "epoch": 0.062154025444304165,
      "grad_norm": 5.111014366876256,
      "learning_rate": 7.751937984496125e-07,
      "loss": 0.4997,
      "step": 240
    },
    {
      "epoch": 0.062154025444304165,
      "eval_PRM Accuracy": 0.7264150943396226,
      "eval_PRM F1": 0.8053691275167785,
      "eval_PRM F1 AUC": 0.7310110005238345,
      "eval_PRM F1 AUC (fixed)": 0.46071241487689885,
      "eval_PRM F1 Neg": 0.5396825396825397,
      "eval_PRM NPV": 0.425,
      "eval_PRM Precision": 0.9090909090909091,
      "eval_PRM Recall": 0.7228915662650602,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.5017386078834534,
      "eval_runtime": 14.7115,
      "eval_samples_per_second": 2.991,
      "eval_steps_per_second": 0.136,
      "step": 240
    },
    {
      "epoch": 0.062413000550322104,
      "grad_norm": 4.557931809240471,
      "learning_rate": 7.784237726098191e-07,
      "loss": 0.4011,
      "step": 241
    },
    {
      "epoch": 0.06267197565634003,
      "grad_norm": 14.003216936602616,
      "learning_rate": 7.81653746770026e-07,
      "loss": 0.3932,
      "step": 242
    },
    {
      "epoch": 0.06293095076235797,
      "grad_norm": 7.843679582137715,
      "learning_rate": 7.848837209302327e-07,
      "loss": 0.4951,
      "step": 243
    },
    {
      "epoch": 0.0631899258683759,
      "grad_norm": 7.845569085491128,
      "learning_rate": 7.881136950904393e-07,
      "loss": 0.4397,
      "step": 244
    },
    {
      "epoch": 0.06344890097439383,
      "grad_norm": 5.61831702884945,
      "learning_rate": 7.913436692506461e-07,
      "loss": 0.4852,
      "step": 245
    },
    {
      "epoch": 0.06344890097439383,
      "eval_PRM Accuracy": 0.7547169811320755,
      "eval_PRM F1": 0.8311688311688312,
      "eval_PRM F1 AUC": 0.7333682556312205,
      "eval_PRM F1 AUC (fixed)": 0.46176008381351497,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.45714285714285713,
      "eval_PRM Precision": 0.9014084507042254,
      "eval_PRM Recall": 0.7710843373493976,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.47543448209762573,
      "eval_runtime": 14.8812,
      "eval_samples_per_second": 2.957,
      "eval_steps_per_second": 0.134,
      "step": 245
    },
    {
      "epoch": 0.06370787608041177,
      "grad_norm": 4.346504560592329,
      "learning_rate": 7.945736434108528e-07,
      "loss": 0.454,
      "step": 246
    },
    {
      "epoch": 0.06396685118642971,
      "grad_norm": 5.69092130458769,
      "learning_rate": 7.978036175710595e-07,
      "loss": 0.4296,
      "step": 247
    },
    {
      "epoch": 0.06422582629244764,
      "grad_norm": 5.621082028360281,
      "learning_rate": 8.010335917312661e-07,
      "loss": 0.4341,
      "step": 248
    },
    {
      "epoch": 0.06448480139846557,
      "grad_norm": 19.347757597432366,
      "learning_rate": 8.04263565891473e-07,
      "loss": 0.4821,
      "step": 249
    },
    {
      "epoch": 0.06474377650448351,
      "grad_norm": 17.86909915880335,
      "learning_rate": 8.074935400516797e-07,
      "loss": 0.6175,
      "step": 250
    },
    {
      "epoch": 0.06474377650448351,
      "eval_PRM Accuracy": 0.7735849056603774,
      "eval_PRM F1": 0.8461538461538461,
      "eval_PRM F1 AUC": 0.7454164484023049,
      "eval_PRM F1 AUC (fixed)": 0.4855945521215297,
      "eval_PRM F1 Neg": 0.5714285714285714,
      "eval_PRM NPV": 0.48484848484848486,
      "eval_PRM Precision": 0.9041095890410958,
      "eval_PRM Recall": 0.7951807228915663,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.44122856855392456,
      "eval_runtime": 14.5299,
      "eval_samples_per_second": 3.028,
      "eval_steps_per_second": 0.138,
      "step": 250
    },
    {
      "epoch": 0.06500275161050144,
      "grad_norm": 10.984232298038023,
      "learning_rate": 8.107235142118863e-07,
      "loss": 0.4992,
      "step": 251
    },
    {
      "epoch": 0.06526172671651938,
      "grad_norm": 16.24839802704603,
      "learning_rate": 8.139534883720931e-07,
      "loss": 0.542,
      "step": 252
    },
    {
      "epoch": 0.0655207018225373,
      "grad_norm": 4.4357811033412045,
      "learning_rate": 8.171834625322998e-07,
      "loss": 0.3877,
      "step": 253
    },
    {
      "epoch": 0.06577967692855524,
      "grad_norm": 4.016940858651593,
      "learning_rate": 8.204134366925066e-07,
      "loss": 0.4171,
      "step": 254
    },
    {
      "epoch": 0.06603865203457318,
      "grad_norm": 3.563876436826327,
      "learning_rate": 8.236434108527132e-07,
      "loss": 0.3769,
      "step": 255
    },
    {
      "epoch": 0.06603865203457318,
      "eval_PRM Accuracy": 0.7830188679245284,
      "eval_PRM F1": 0.8496732026143791,
      "eval_PRM F1 AUC": 0.782870612886328,
      "eval_PRM F1 AUC (fixed)": 0.4963331587218439,
      "eval_PRM F1 Neg": 0.6101694915254238,
      "eval_PRM NPV": 0.5,
      "eval_PRM Precision": 0.9285714285714286,
      "eval_PRM Recall": 0.7831325301204819,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.44621098041534424,
      "eval_runtime": 14.8363,
      "eval_samples_per_second": 2.966,
      "eval_steps_per_second": 0.135,
      "step": 255
    },
    {
      "epoch": 0.0662976271405911,
      "grad_norm": 10.242691540094073,
      "learning_rate": 8.2687338501292e-07,
      "loss": 0.4985,
      "step": 256
    },
    {
      "epoch": 0.06655660224660905,
      "grad_norm": 6.350176841521547,
      "learning_rate": 8.301033591731267e-07,
      "loss": 0.4529,
      "step": 257
    },
    {
      "epoch": 0.06681557735262698,
      "grad_norm": 5.5896844097563845,
      "learning_rate": 8.333333333333333e-07,
      "loss": 0.4475,
      "step": 258
    },
    {
      "epoch": 0.06707455245864491,
      "grad_norm": 8.47310401816812,
      "learning_rate": 8.365633074935402e-07,
      "loss": 0.5157,
      "step": 259
    },
    {
      "epoch": 0.06733352756466285,
      "grad_norm": 4.916322297452303,
      "learning_rate": 8.397932816537469e-07,
      "loss": 0.4844,
      "step": 260
    },
    {
      "epoch": 0.06733352756466285,
      "eval_PRM Accuracy": 0.7452830188679245,
      "eval_PRM F1": 0.8187919463087249,
      "eval_PRM F1 AUC": 0.7587742273441593,
      "eval_PRM F1 AUC (fixed)": 0.5141435306443164,
      "eval_PRM F1 Neg": 0.5714285714285714,
      "eval_PRM NPV": 0.45,
      "eval_PRM Precision": 0.9242424242424242,
      "eval_PRM Recall": 0.7349397590361446,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.4647907018661499,
      "eval_runtime": 14.8252,
      "eval_samples_per_second": 2.968,
      "eval_steps_per_second": 0.135,
      "step": 260
    },
    {
      "epoch": 0.06759250267068079,
      "grad_norm": 5.819653797435914,
      "learning_rate": 8.430232558139535e-07,
      "loss": 0.5159,
      "step": 261
    },
    {
      "epoch": 0.06785147777669871,
      "grad_norm": 4.869616721137864,
      "learning_rate": 8.462532299741603e-07,
      "loss": 0.5286,
      "step": 262
    },
    {
      "epoch": 0.06811045288271665,
      "grad_norm": 6.158491506026584,
      "learning_rate": 8.49483204134367e-07,
      "loss": 0.4261,
      "step": 263
    },
    {
      "epoch": 0.06836942798873458,
      "grad_norm": 7.8691608657818355,
      "learning_rate": 8.527131782945737e-07,
      "loss": 0.4181,
      "step": 264
    },
    {
      "epoch": 0.06862840309475252,
      "grad_norm": 4.668507407839531,
      "learning_rate": 8.559431524547804e-07,
      "loss": 0.4744,
      "step": 265
    },
    {
      "epoch": 0.06862840309475252,
      "eval_PRM Accuracy": 0.7641509433962265,
      "eval_PRM F1": 0.8322147651006712,
      "eval_PRM F1 AUC": 0.7865374541644841,
      "eval_PRM F1 AUC (fixed)": 0.5146673651126245,
      "eval_PRM F1 Neg": 0.6031746031746031,
      "eval_PRM NPV": 0.475,
      "eval_PRM Precision": 0.9393939393939394,
      "eval_PRM Recall": 0.7469879518072289,
      "eval_PRM Specificty": 0.8260869565217391,
      "eval_loss": 0.4597613513469696,
      "eval_runtime": 14.7087,
      "eval_samples_per_second": 2.991,
      "eval_steps_per_second": 0.136,
      "step": 265
    },
    {
      "epoch": 0.06888737820077045,
      "grad_norm": 5.975907751002809,
      "learning_rate": 8.591731266149872e-07,
      "loss": 0.4263,
      "step": 266
    },
    {
      "epoch": 0.06914635330678838,
      "grad_norm": 9.117203887331963,
      "learning_rate": 8.624031007751939e-07,
      "loss": 0.4882,
      "step": 267
    },
    {
      "epoch": 0.06940532841280632,
      "grad_norm": 9.706182790356385,
      "learning_rate": 8.656330749354005e-07,
      "loss": 0.5266,
      "step": 268
    },
    {
      "epoch": 0.06966430351882426,
      "grad_norm": 5.136582860683114,
      "learning_rate": 8.688630490956073e-07,
      "loss": 0.407,
      "step": 269
    },
    {
      "epoch": 0.06992327862484218,
      "grad_norm": 8.3622633548138,
      "learning_rate": 8.720930232558141e-07,
      "loss": 0.436,
      "step": 270
    },
    {
      "epoch": 0.06992327862484218,
      "eval_PRM Accuracy": 0.7735849056603774,
      "eval_PRM F1": 0.8441558441558441,
      "eval_PRM F1 AUC": 0.7611314824515453,
      "eval_PRM F1 AUC (fixed)": 0.5154531168150864,
      "eval_PRM F1 Neg": 0.5862068965517241,
      "eval_PRM NPV": 0.4857142857142857,
      "eval_PRM Precision": 0.9154929577464789,
      "eval_PRM Recall": 0.7831325301204819,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.4233591854572296,
      "eval_runtime": 14.4694,
      "eval_samples_per_second": 3.041,
      "eval_steps_per_second": 0.138,
      "step": 270
    },
    {
      "epoch": 0.07018225373086012,
      "grad_norm": 10.168784870346583,
      "learning_rate": 8.753229974160208e-07,
      "loss": 0.513,
      "step": 271
    },
    {
      "epoch": 0.07044122883687806,
      "grad_norm": 4.326897308284535,
      "learning_rate": 8.785529715762274e-07,
      "loss": 0.4433,
      "step": 272
    },
    {
      "epoch": 0.07070020394289599,
      "grad_norm": 5.672475397941136,
      "learning_rate": 8.817829457364342e-07,
      "loss": 0.4609,
      "step": 273
    },
    {
      "epoch": 0.07095917904891393,
      "grad_norm": 17.141990085978755,
      "learning_rate": 8.850129198966409e-07,
      "loss": 0.6045,
      "step": 274
    },
    {
      "epoch": 0.07121815415493185,
      "grad_norm": 5.220531040231862,
      "learning_rate": 8.882428940568475e-07,
      "loss": 0.4902,
      "step": 275
    },
    {
      "epoch": 0.07121815415493185,
      "eval_PRM Accuracy": 0.7830188679245284,
      "eval_PRM F1": 0.8496732026143791,
      "eval_PRM F1 AUC": 0.782870612886328,
      "eval_PRM F1 AUC (fixed)": 0.5125720272393923,
      "eval_PRM F1 Neg": 0.6101694915254238,
      "eval_PRM NPV": 0.5,
      "eval_PRM Precision": 0.9285714285714286,
      "eval_PRM Recall": 0.7831325301204819,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.43242713809013367,
      "eval_runtime": 14.5138,
      "eval_samples_per_second": 3.032,
      "eval_steps_per_second": 0.138,
      "step": 275
    },
    {
      "epoch": 0.07147712926094979,
      "grad_norm": 3.5766465490704893,
      "learning_rate": 8.914728682170544e-07,
      "loss": 0.416,
      "step": 276
    },
    {
      "epoch": 0.07173610436696773,
      "grad_norm": 10.224622754440063,
      "learning_rate": 8.947028423772611e-07,
      "loss": 0.4704,
      "step": 277
    },
    {
      "epoch": 0.07199507947298565,
      "grad_norm": 7.746233043103798,
      "learning_rate": 8.979328165374678e-07,
      "loss": 0.4716,
      "step": 278
    },
    {
      "epoch": 0.0722540545790036,
      "grad_norm": 5.999186571327644,
      "learning_rate": 9.011627906976744e-07,
      "loss": 0.46,
      "step": 279
    },
    {
      "epoch": 0.07251302968502153,
      "grad_norm": 4.481831549526554,
      "learning_rate": 9.043927648578812e-07,
      "loss": 0.4955,
      "step": 280
    },
    {
      "epoch": 0.07251302968502153,
      "eval_PRM Accuracy": 0.7547169811320755,
      "eval_PRM F1": 0.8266666666666667,
      "eval_PRM F1 AUC": 0.7647983237297015,
      "eval_PRM F1 AUC (fixed)": 0.504190675746464,
      "eval_PRM F1 Neg": 0.5806451612903226,
      "eval_PRM NPV": 0.46153846153846156,
      "eval_PRM Precision": 0.9253731343283582,
      "eval_PRM Recall": 0.7469879518072289,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.46966707706451416,
      "eval_runtime": 14.5021,
      "eval_samples_per_second": 3.034,
      "eval_steps_per_second": 0.138,
      "step": 280
    },
    {
      "epoch": 0.07277200479103946,
      "grad_norm": 15.916954907020221,
      "learning_rate": 9.07622739018088e-07,
      "loss": 0.4277,
      "step": 281
    },
    {
      "epoch": 0.0730309798970574,
      "grad_norm": 10.711836972773476,
      "learning_rate": 9.108527131782946e-07,
      "loss": 0.4199,
      "step": 282
    },
    {
      "epoch": 0.07328995500307534,
      "grad_norm": 17.339949092453057,
      "learning_rate": 9.140826873385014e-07,
      "loss": 0.5185,
      "step": 283
    },
    {
      "epoch": 0.07354893010909326,
      "grad_norm": 6.559593493304679,
      "learning_rate": 9.173126614987081e-07,
      "loss": 0.4751,
      "step": 284
    },
    {
      "epoch": 0.0738079052151112,
      "grad_norm": 12.211437994570204,
      "learning_rate": 9.205426356589148e-07,
      "loss": 0.3782,
      "step": 285
    },
    {
      "epoch": 0.0738079052151112,
      "eval_PRM Accuracy": 0.7641509433962265,
      "eval_PRM F1": 0.8344370860927153,
      "eval_PRM F1 AUC": 0.7708224201152437,
      "eval_PRM F1 AUC (fixed)": 0.5107386066003143,
      "eval_PRM F1 Neg": 0.5901639344262295,
      "eval_PRM NPV": 0.47368421052631576,
      "eval_PRM Precision": 0.9264705882352942,
      "eval_PRM Recall": 0.7590361445783133,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.4466630816459656,
      "eval_runtime": 14.8677,
      "eval_samples_per_second": 2.959,
      "eval_steps_per_second": 0.135,
      "step": 285
    },
    {
      "epoch": 0.07406688032112912,
      "grad_norm": 9.505648050872507,
      "learning_rate": 9.237726098191216e-07,
      "loss": 0.3501,
      "step": 286
    },
    {
      "epoch": 0.07432585542714706,
      "grad_norm": 20.876440461993184,
      "learning_rate": 9.270025839793283e-07,
      "loss": 0.5186,
      "step": 287
    },
    {
      "epoch": 0.074584830533165,
      "grad_norm": 4.231394956008767,
      "learning_rate": 9.30232558139535e-07,
      "loss": 0.3812,
      "step": 288
    },
    {
      "epoch": 0.07484380563918293,
      "grad_norm": 7.798047063861872,
      "learning_rate": 9.334625322997416e-07,
      "loss": 0.4313,
      "step": 289
    },
    {
      "epoch": 0.07510278074520087,
      "grad_norm": 10.845228628781953,
      "learning_rate": 9.366925064599484e-07,
      "loss": 0.4542,
      "step": 290
    },
    {
      "epoch": 0.07510278074520087,
      "eval_PRM Accuracy": 0.7924528301886793,
      "eval_PRM F1": 0.8607594936708861,
      "eval_PRM F1 AUC": 0.7574646411733892,
      "eval_PRM F1 AUC (fixed)": 0.5178103719224725,
      "eval_PRM F1 Neg": 0.5925925925925926,
      "eval_PRM NPV": 0.5161290322580645,
      "eval_PRM Precision": 0.9066666666666666,
      "eval_PRM Recall": 0.8192771084337349,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.4175150692462921,
      "eval_runtime": 14.2443,
      "eval_samples_per_second": 3.089,
      "eval_steps_per_second": 0.14,
      "step": 290
    },
    {
      "epoch": 0.0753617558512188,
      "grad_norm": 12.878911968596602,
      "learning_rate": 9.399224806201551e-07,
      "loss": 0.5125,
      "step": 291
    },
    {
      "epoch": 0.07562073095723673,
      "grad_norm": 10.632764116022619,
      "learning_rate": 9.431524547803617e-07,
      "loss": 0.413,
      "step": 292
    },
    {
      "epoch": 0.07587970606325467,
      "grad_norm": 8.314195563054465,
      "learning_rate": 9.463824289405686e-07,
      "loss": 0.4092,
      "step": 293
    },
    {
      "epoch": 0.07613868116927261,
      "grad_norm": 4.533822120001036,
      "learning_rate": 9.496124031007753e-07,
      "loss": 0.45,
      "step": 294
    },
    {
      "epoch": 0.07639765627529053,
      "grad_norm": 4.147278873212908,
      "learning_rate": 9.52842377260982e-07,
      "loss": 0.4271,
      "step": 295
    },
    {
      "epoch": 0.07639765627529053,
      "eval_PRM Accuracy": 0.7547169811320755,
      "eval_PRM F1": 0.8266666666666667,
      "eval_PRM F1 AUC": 0.7647983237297015,
      "eval_PRM F1 AUC (fixed)": 0.5240963855421688,
      "eval_PRM F1 Neg": 0.5806451612903226,
      "eval_PRM NPV": 0.46153846153846156,
      "eval_PRM Precision": 0.9253731343283582,
      "eval_PRM Recall": 0.7469879518072289,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.4515579342842102,
      "eval_runtime": 14.6113,
      "eval_samples_per_second": 3.011,
      "eval_steps_per_second": 0.137,
      "step": 295
    },
    {
      "epoch": 0.07665663138130847,
      "grad_norm": 12.839286179768832,
      "learning_rate": 9.560723514211886e-07,
      "loss": 0.4118,
      "step": 296
    },
    {
      "epoch": 0.0769156064873264,
      "grad_norm": 14.42556307757587,
      "learning_rate": 9.593023255813955e-07,
      "loss": 0.5255,
      "step": 297
    },
    {
      "epoch": 0.07717458159334434,
      "grad_norm": 3.6706140600620834,
      "learning_rate": 9.625322997416022e-07,
      "loss": 0.4228,
      "step": 298
    },
    {
      "epoch": 0.07743355669936228,
      "grad_norm": 7.9469588293703595,
      "learning_rate": 9.657622739018088e-07,
      "loss": 0.4946,
      "step": 299
    },
    {
      "epoch": 0.0776925318053802,
      "grad_norm": 8.228295223889006,
      "learning_rate": 9.689922480620157e-07,
      "loss": 0.4881,
      "step": 300
    },
    {
      "epoch": 0.0776925318053802,
      "eval_PRM Accuracy": 0.7075471698113207,
      "eval_PRM F1": 0.7832167832167832,
      "eval_PRM F1 AUC": 0.7503928758512309,
      "eval_PRM F1 AUC (fixed)": 0.5301204819277108,
      "eval_PRM F1 Neg": 0.5507246376811594,
      "eval_PRM NPV": 0.41304347826086957,
      "eval_PRM Precision": 0.9333333333333333,
      "eval_PRM Recall": 0.6746987951807228,
      "eval_PRM Specificty": 0.8260869565217391,
      "eval_loss": 0.4778639078140259,
      "eval_runtime": 14.3702,
      "eval_samples_per_second": 3.062,
      "eval_steps_per_second": 0.139,
      "step": 300
    },
    {
      "epoch": 0.07795150691139814,
      "grad_norm": 7.801989344153817,
      "learning_rate": 9.722222222222224e-07,
      "loss": 0.4925,
      "step": 301
    },
    {
      "epoch": 0.07821048201741608,
      "grad_norm": 6.194112700855093,
      "learning_rate": 9.75452196382429e-07,
      "loss": 0.5078,
      "step": 302
    },
    {
      "epoch": 0.078469457123434,
      "grad_norm": 4.334517501586195,
      "learning_rate": 9.786821705426357e-07,
      "loss": 0.4482,
      "step": 303
    },
    {
      "epoch": 0.07872843222945194,
      "grad_norm": 7.02828276789377,
      "learning_rate": 9.819121447028424e-07,
      "loss": 0.4523,
      "step": 304
    },
    {
      "epoch": 0.07898740733546988,
      "grad_norm": 4.699628027384838,
      "learning_rate": 9.85142118863049e-07,
      "loss": 0.3978,
      "step": 305
    },
    {
      "epoch": 0.07898740733546988,
      "eval_PRM Accuracy": 0.7641509433962265,
      "eval_PRM F1": 0.8344370860927153,
      "eval_PRM F1 AUC": 0.7708224201152437,
      "eval_PRM F1 AUC (fixed)": 0.5319539025667889,
      "eval_PRM F1 Neg": 0.5901639344262295,
      "eval_PRM NPV": 0.47368421052631576,
      "eval_PRM Precision": 0.9264705882352942,
      "eval_PRM Recall": 0.7590361445783133,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.4269060790538788,
      "eval_runtime": 14.6875,
      "eval_samples_per_second": 2.996,
      "eval_steps_per_second": 0.136,
      "step": 305
    },
    {
      "epoch": 0.07924638244148781,
      "grad_norm": 4.731440582096809,
      "learning_rate": 9.88372093023256e-07,
      "loss": 0.3662,
      "step": 306
    },
    {
      "epoch": 0.07950535754750575,
      "grad_norm": 5.599732171367987,
      "learning_rate": 9.916020671834626e-07,
      "loss": 0.456,
      "step": 307
    },
    {
      "epoch": 0.07976433265352369,
      "grad_norm": 5.52753719513111,
      "learning_rate": 9.948320413436693e-07,
      "loss": 0.3806,
      "step": 308
    },
    {
      "epoch": 0.08002330775954161,
      "grad_norm": 11.708582980152471,
      "learning_rate": 9.98062015503876e-07,
      "loss": 0.4499,
      "step": 309
    },
    {
      "epoch": 0.08028228286555955,
      "grad_norm": 4.372614035219098,
      "learning_rate": 1.0012919896640828e-06,
      "loss": 0.4207,
      "step": 310
    },
    {
      "epoch": 0.08028228286555955,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.8734177215189873,
      "eval_PRM F1 AUC": 0.785227867993714,
      "eval_PRM F1 AUC (fixed)": 0.5395495023572552,
      "eval_PRM F1 Neg": 0.6296296296296297,
      "eval_PRM NPV": 0.5483870967741935,
      "eval_PRM Precision": 0.92,
      "eval_PRM Recall": 0.8313253012048193,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.3785098195075989,
      "eval_runtime": 14.9751,
      "eval_samples_per_second": 2.938,
      "eval_steps_per_second": 0.134,
      "step": 310
    },
    {
      "epoch": 0.08054125797157748,
      "grad_norm": 14.412298324521696,
      "learning_rate": 1.0045219638242895e-06,
      "loss": 0.527,
      "step": 311
    },
    {
      "epoch": 0.08080023307759542,
      "grad_norm": 10.519064993089941,
      "learning_rate": 1.0077519379844962e-06,
      "loss": 0.425,
      "step": 312
    },
    {
      "epoch": 0.08105920818361335,
      "grad_norm": 21.1615871003155,
      "learning_rate": 1.0109819121447028e-06,
      "loss": 0.5047,
      "step": 313
    },
    {
      "epoch": 0.08131818328963128,
      "grad_norm": 12.442814820549918,
      "learning_rate": 1.0142118863049097e-06,
      "loss": 0.5891,
      "step": 314
    },
    {
      "epoch": 0.08157715839564922,
      "grad_norm": 17.06252344116241,
      "learning_rate": 1.0174418604651164e-06,
      "loss": 0.597,
      "step": 315
    },
    {
      "epoch": 0.08157715839564922,
      "eval_PRM Accuracy": 0.7924528301886793,
      "eval_PRM F1": 0.8552631578947368,
      "eval_PRM F1 AUC": 0.8046097433211105,
      "eval_PRM F1 AUC (fixed)": 0.5468831849135674,
      "eval_PRM F1 Neg": 0.6333333333333333,
      "eval_PRM NPV": 0.5135135135135135,
      "eval_PRM Precision": 0.9420289855072463,
      "eval_PRM Recall": 0.7831325301204819,
      "eval_PRM Specificty": 0.8260869565217391,
      "eval_loss": 0.4063412845134735,
      "eval_runtime": 14.5989,
      "eval_samples_per_second": 3.014,
      "eval_steps_per_second": 0.137,
      "step": 315
    },
    {
      "epoch": 0.08183613350166716,
      "grad_norm": 6.2636774461754285,
      "learning_rate": 1.020671834625323e-06,
      "loss": 0.4252,
      "step": 316
    },
    {
      "epoch": 0.08209510860768508,
      "grad_norm": 6.054523107112382,
      "learning_rate": 1.02390180878553e-06,
      "loss": 0.5367,
      "step": 317
    },
    {
      "epoch": 0.08235408371370302,
      "grad_norm": 9.718566201674209,
      "learning_rate": 1.0271317829457366e-06,
      "loss": 0.3912,
      "step": 318
    },
    {
      "epoch": 0.08261305881972096,
      "grad_norm": 4.987727785104741,
      "learning_rate": 1.0303617571059433e-06,
      "loss": 0.4392,
      "step": 319
    },
    {
      "epoch": 0.08287203392573889,
      "grad_norm": 6.884006523120188,
      "learning_rate": 1.03359173126615e-06,
      "loss": 0.3608,
      "step": 320
    },
    {
      "epoch": 0.08287203392573889,
      "eval_PRM Accuracy": 0.7735849056603774,
      "eval_PRM F1": 0.84,
      "eval_PRM F1 AUC": 0.7925615505500263,
      "eval_PRM F1 AUC (fixed)": 0.5471451021477213,
      "eval_PRM F1 Neg": 0.6129032258064516,
      "eval_PRM NPV": 0.48717948717948717,
      "eval_PRM Precision": 0.9402985074626866,
      "eval_PRM Recall": 0.7590361445783133,
      "eval_PRM Specificty": 0.8260869565217391,
      "eval_loss": 0.4275074303150177,
      "eval_runtime": 14.8425,
      "eval_samples_per_second": 2.964,
      "eval_steps_per_second": 0.135,
      "step": 320
    },
    {
      "epoch": 0.08313100903175683,
      "grad_norm": 5.318854637746786,
      "learning_rate": 1.0368217054263566e-06,
      "loss": 0.4384,
      "step": 321
    },
    {
      "epoch": 0.08338998413777475,
      "grad_norm": 11.435651480653691,
      "learning_rate": 1.0400516795865635e-06,
      "loss": 0.3841,
      "step": 322
    },
    {
      "epoch": 0.08364895924379269,
      "grad_norm": 4.849662752105375,
      "learning_rate": 1.0432816537467702e-06,
      "loss": 0.433,
      "step": 323
    },
    {
      "epoch": 0.08390793434981063,
      "grad_norm": 4.105703022458649,
      "learning_rate": 1.0465116279069768e-06,
      "loss": 0.427,
      "step": 324
    },
    {
      "epoch": 0.08416690945582855,
      "grad_norm": 7.043811548213904,
      "learning_rate": 1.0497416020671835e-06,
      "loss": 0.4356,
      "step": 325
    },
    {
      "epoch": 0.08416690945582855,
      "eval_PRM Accuracy": 0.8018867924528302,
      "eval_PRM F1": 0.8662420382165605,
      "eval_PRM F1 AUC": 0.7792037716081718,
      "eval_PRM F1 AUC (fixed)": 0.5586694604504977,
      "eval_PRM F1 Neg": 0.6181818181818182,
      "eval_PRM NPV": 0.53125,
      "eval_PRM Precision": 0.918918918918919,
      "eval_PRM Recall": 0.8192771084337349,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.3831585943698883,
      "eval_runtime": 14.7099,
      "eval_samples_per_second": 2.991,
      "eval_steps_per_second": 0.136,
      "step": 325
    },
    {
      "epoch": 0.08442588456184649,
      "grad_norm": 5.753783230308081,
      "learning_rate": 1.0529715762273902e-06,
      "loss": 0.4569,
      "step": 326
    },
    {
      "epoch": 0.08468485966786443,
      "grad_norm": 7.813829276934702,
      "learning_rate": 1.0562015503875968e-06,
      "loss": 0.4649,
      "step": 327
    },
    {
      "epoch": 0.08494383477388236,
      "grad_norm": 16.4994165028651,
      "learning_rate": 1.0594315245478037e-06,
      "loss": 0.5808,
      "step": 328
    },
    {
      "epoch": 0.0852028098799003,
      "grad_norm": 15.370926504978964,
      "learning_rate": 1.0626614987080104e-06,
      "loss": 0.5716,
      "step": 329
    },
    {
      "epoch": 0.08546178498591823,
      "grad_norm": 7.9551510457294805,
      "learning_rate": 1.065891472868217e-06,
      "loss": 0.4672,
      "step": 330
    },
    {
      "epoch": 0.08546178498591823,
      "eval_PRM Accuracy": 0.8018867924528302,
      "eval_PRM F1": 0.8662420382165605,
      "eval_PRM F1 AUC": 0.7792037716081718,
      "eval_PRM F1 AUC (fixed)": 0.5576217915138817,
      "eval_PRM F1 Neg": 0.6181818181818182,
      "eval_PRM NPV": 0.53125,
      "eval_PRM Precision": 0.918918918918919,
      "eval_PRM Recall": 0.8192771084337349,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.3691127598285675,
      "eval_runtime": 14.6243,
      "eval_samples_per_second": 3.009,
      "eval_steps_per_second": 0.137,
      "step": 330
    },
    {
      "epoch": 0.08572076009193616,
      "grad_norm": 4.515299251404578,
      "learning_rate": 1.069121447028424e-06,
      "loss": 0.4701,
      "step": 331
    },
    {
      "epoch": 0.0859797351979541,
      "grad_norm": 5.9679777493199095,
      "learning_rate": 1.0723514211886306e-06,
      "loss": 0.466,
      "step": 332
    },
    {
      "epoch": 0.08623871030397202,
      "grad_norm": 4.152376826263897,
      "learning_rate": 1.0755813953488373e-06,
      "loss": 0.4128,
      "step": 333
    },
    {
      "epoch": 0.08649768540998996,
      "grad_norm": 9.31379252266598,
      "learning_rate": 1.0788113695090441e-06,
      "loss": 0.401,
      "step": 334
    },
    {
      "epoch": 0.0867566605160079,
      "grad_norm": 13.247200202082919,
      "learning_rate": 1.0820413436692508e-06,
      "loss": 0.5408,
      "step": 335
    },
    {
      "epoch": 0.0867566605160079,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.8717948717948718,
      "eval_PRM F1 AUC": 0.8009429020429545,
      "eval_PRM F1 AUC (fixed)": 0.5500261917234154,
      "eval_PRM F1 Neg": 0.6428571428571429,
      "eval_PRM NPV": 0.5454545454545454,
      "eval_PRM Precision": 0.9315068493150684,
      "eval_PRM Recall": 0.8192771084337349,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.39321768283843994,
      "eval_runtime": 14.484,
      "eval_samples_per_second": 3.038,
      "eval_steps_per_second": 0.138,
      "step": 335
    },
    {
      "epoch": 0.08701563562202583,
      "grad_norm": 5.385907638555454,
      "learning_rate": 1.0852713178294575e-06,
      "loss": 0.3754,
      "step": 336
    },
    {
      "epoch": 0.08727461072804377,
      "grad_norm": 7.009224169412602,
      "learning_rate": 1.0885012919896641e-06,
      "loss": 0.4683,
      "step": 337
    },
    {
      "epoch": 0.0875335858340617,
      "grad_norm": 7.664329467972336,
      "learning_rate": 1.091731266149871e-06,
      "loss": 0.4783,
      "step": 338
    },
    {
      "epoch": 0.08779256094007963,
      "grad_norm": 4.197296615684727,
      "learning_rate": 1.0949612403100777e-06,
      "loss": 0.4905,
      "step": 339
    },
    {
      "epoch": 0.08805153604609757,
      "grad_norm": 10.035421137511468,
      "learning_rate": 1.0981912144702844e-06,
      "loss": 0.3913,
      "step": 340
    },
    {
      "epoch": 0.08805153604609757,
      "eval_PRM Accuracy": 0.7547169811320755,
      "eval_PRM F1": 0.8243243243243243,
      "eval_PRM F1 AUC": 0.7805133577789418,
      "eval_PRM F1 AUC (fixed)": 0.5413829229963332,
      "eval_PRM F1 Neg": 0.59375,
      "eval_PRM NPV": 0.4634146341463415,
      "eval_PRM Precision": 0.9384615384615385,
      "eval_PRM Recall": 0.7349397590361446,
      "eval_PRM Specificty": 0.8260869565217391,
      "eval_loss": 0.4457910656929016,
      "eval_runtime": 14.5536,
      "eval_samples_per_second": 3.023,
      "eval_steps_per_second": 0.137,
      "step": 340
    },
    {
      "epoch": 0.08831051115211551,
      "grad_norm": 7.566916747027618,
      "learning_rate": 1.101421188630491e-06,
      "loss": 0.4404,
      "step": 341
    },
    {
      "epoch": 0.08856948625813343,
      "grad_norm": 7.718429079088188,
      "learning_rate": 1.1046511627906977e-06,
      "loss": 0.4188,
      "step": 342
    },
    {
      "epoch": 0.08882846136415137,
      "grad_norm": 13.14516943643952,
      "learning_rate": 1.1078811369509044e-06,
      "loss": 0.3584,
      "step": 343
    },
    {
      "epoch": 0.0890874364701693,
      "grad_norm": 5.774045581523776,
      "learning_rate": 1.111111111111111e-06,
      "loss": 0.406,
      "step": 344
    },
    {
      "epoch": 0.08934641157618724,
      "grad_norm": 3.599053746712549,
      "learning_rate": 1.114341085271318e-06,
      "loss": 0.3606,
      "step": 345
    },
    {
      "epoch": 0.08934641157618724,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.8717948717948718,
      "eval_PRM F1 AUC": 0.8009429020429545,
      "eval_PRM F1 AUC (fixed)": 0.5474070193818753,
      "eval_PRM F1 Neg": 0.6428571428571429,
      "eval_PRM NPV": 0.5454545454545454,
      "eval_PRM Precision": 0.9315068493150684,
      "eval_PRM Recall": 0.8192771084337349,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.40695610642433167,
      "eval_runtime": 14.8039,
      "eval_samples_per_second": 2.972,
      "eval_steps_per_second": 0.135,
      "step": 345
    },
    {
      "epoch": 0.08960538668220518,
      "grad_norm": 6.351538227956327,
      "learning_rate": 1.1175710594315246e-06,
      "loss": 0.4081,
      "step": 346
    },
    {
      "epoch": 0.0898643617882231,
      "grad_norm": 4.552702510203715,
      "learning_rate": 1.1208010335917313e-06,
      "loss": 0.4791,
      "step": 347
    },
    {
      "epoch": 0.09012333689424104,
      "grad_norm": 6.400764438436327,
      "learning_rate": 1.1240310077519381e-06,
      "loss": 0.4699,
      "step": 348
    },
    {
      "epoch": 0.09038231200025898,
      "grad_norm": 6.6839400360275265,
      "learning_rate": 1.1272609819121448e-06,
      "loss": 0.5084,
      "step": 349
    },
    {
      "epoch": 0.0906412871062769,
      "grad_norm": 6.068084345164869,
      "learning_rate": 1.1304909560723515e-06,
      "loss": 0.4268,
      "step": 350
    },
    {
      "epoch": 0.0906412871062769,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.8717948717948718,
      "eval_PRM F1 AUC": 0.8009429020429545,
      "eval_PRM F1 AUC (fixed)": 0.569408067050812,
      "eval_PRM F1 Neg": 0.6428571428571429,
      "eval_PRM NPV": 0.5454545454545454,
      "eval_PRM Precision": 0.9315068493150684,
      "eval_PRM Recall": 0.8192771084337349,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.38017958402633667,
      "eval_runtime": 14.8134,
      "eval_samples_per_second": 2.97,
      "eval_steps_per_second": 0.135,
      "step": 350
    },
    {
      "epoch": 0.09090026221229484,
      "grad_norm": 6.000182731927381,
      "learning_rate": 1.1337209302325581e-06,
      "loss": 0.4375,
      "step": 351
    },
    {
      "epoch": 0.09115923731831278,
      "grad_norm": 6.664169382629427,
      "learning_rate": 1.136950904392765e-06,
      "loss": 0.4991,
      "step": 352
    },
    {
      "epoch": 0.09141821242433071,
      "grad_norm": 7.306217251000588,
      "learning_rate": 1.1401808785529717e-06,
      "loss": 0.4768,
      "step": 353
    },
    {
      "epoch": 0.09167718753034865,
      "grad_norm": 7.9308828901201345,
      "learning_rate": 1.1434108527131784e-06,
      "loss": 0.4159,
      "step": 354
    },
    {
      "epoch": 0.09193616263636657,
      "grad_norm": 5.17121110105702,
      "learning_rate": 1.1466408268733852e-06,
      "loss": 0.416,
      "step": 355
    },
    {
      "epoch": 0.09193616263636657,
      "eval_PRM Accuracy": 0.8018867924528302,
      "eval_PRM F1": 0.8627450980392157,
      "eval_PRM F1 AUC": 0.8106338397066527,
      "eval_PRM F1 AUC (fixed)": 0.5573598742797277,
      "eval_PRM F1 Neg": 0.6440677966101694,
      "eval_PRM NPV": 0.5277777777777778,
      "eval_PRM Precision": 0.9428571428571428,
      "eval_PRM Recall": 0.7951807228915663,
      "eval_PRM Specificty": 0.8260869565217391,
      "eval_loss": 0.3932633399963379,
      "eval_runtime": 14.5079,
      "eval_samples_per_second": 3.033,
      "eval_steps_per_second": 0.138,
      "step": 355
    },
    {
      "epoch": 0.09219513774238451,
      "grad_norm": 8.013449320152551,
      "learning_rate": 1.149870801033592e-06,
      "loss": 0.3701,
      "step": 356
    },
    {
      "epoch": 0.09245411284840245,
      "grad_norm": 5.882074020186261,
      "learning_rate": 1.1531007751937986e-06,
      "loss": 0.4908,
      "step": 357
    },
    {
      "epoch": 0.09271308795442038,
      "grad_norm": 3.9901380175520953,
      "learning_rate": 1.1563307493540052e-06,
      "loss": 0.4386,
      "step": 358
    },
    {
      "epoch": 0.09297206306043831,
      "grad_norm": 6.734838796869919,
      "learning_rate": 1.159560723514212e-06,
      "loss": 0.4259,
      "step": 359
    },
    {
      "epoch": 0.09323103816645625,
      "grad_norm": 4.774912365406276,
      "learning_rate": 1.1627906976744188e-06,
      "loss": 0.4545,
      "step": 360
    },
    {
      "epoch": 0.09323103816645625,
      "eval_PRM Accuracy": 0.8301886792452831,
      "eval_PRM F1": 0.8860759493670886,
      "eval_PRM F1 AUC": 0.8129910948140389,
      "eval_PRM F1 AUC (fixed)": 0.5518596123624935,
      "eval_PRM F1 Neg": 0.6666666666666666,
      "eval_PRM NPV": 0.5806451612903226,
      "eval_PRM Precision": 0.9333333333333333,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.37586334347724915,
      "eval_runtime": 14.6961,
      "eval_samples_per_second": 2.994,
      "eval_steps_per_second": 0.136,
      "step": 360
    },
    {
      "epoch": 0.09349001327247418,
      "grad_norm": 4.447339694833538,
      "learning_rate": 1.1660206718346255e-06,
      "loss": 0.419,
      "step": 361
    },
    {
      "epoch": 0.09374898837849212,
      "grad_norm": 7.095783943084598,
      "learning_rate": 1.1692506459948321e-06,
      "loss": 0.427,
      "step": 362
    },
    {
      "epoch": 0.09400796348451006,
      "grad_norm": 4.085236885353116,
      "learning_rate": 1.1724806201550388e-06,
      "loss": 0.3795,
      "step": 363
    },
    {
      "epoch": 0.09426693859052798,
      "grad_norm": 8.894911794969753,
      "learning_rate": 1.1757105943152455e-06,
      "loss": 0.3978,
      "step": 364
    },
    {
      "epoch": 0.09452591369654592,
      "grad_norm": 13.919588842134319,
      "learning_rate": 1.1789405684754523e-06,
      "loss": 0.4439,
      "step": 365
    },
    {
      "epoch": 0.09452591369654592,
      "eval_PRM Accuracy": 0.8018867924528302,
      "eval_PRM F1": 0.8695652173913043,
      "eval_PRM F1 AUC": 0.747773703509691,
      "eval_PRM F1 AUC (fixed)": 0.5531691985332635,
      "eval_PRM F1 Neg": 0.5882352941176471,
      "eval_PRM NPV": 0.5357142857142857,
      "eval_PRM Precision": 0.8974358974358975,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.3621164560317993,
      "eval_runtime": 14.1653,
      "eval_samples_per_second": 3.106,
      "eval_steps_per_second": 0.141,
      "step": 365
    },
    {
      "epoch": 0.09478488880256386,
      "grad_norm": 7.367527828756034,
      "learning_rate": 1.182170542635659e-06,
      "loss": 0.3933,
      "step": 366
    },
    {
      "epoch": 0.09504386390858179,
      "grad_norm": 4.401857963264339,
      "learning_rate": 1.1854005167958657e-06,
      "loss": 0.372,
      "step": 367
    },
    {
      "epoch": 0.09530283901459972,
      "grad_norm": 4.066198280850839,
      "learning_rate": 1.1886304909560723e-06,
      "loss": 0.4406,
      "step": 368
    },
    {
      "epoch": 0.09556181412061765,
      "grad_norm": 3.046639350111974,
      "learning_rate": 1.1918604651162792e-06,
      "loss": 0.3952,
      "step": 369
    },
    {
      "epoch": 0.09582078922663559,
      "grad_norm": 3.6302489391357318,
      "learning_rate": 1.195090439276486e-06,
      "loss": 0.3457,
      "step": 370
    },
    {
      "epoch": 0.09582078922663559,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8789808917197452,
      "eval_PRM F1 AUC": 0.8069669984284967,
      "eval_PRM F1 AUC (fixed)": 0.5586694604504976,
      "eval_PRM F1 Neg": 0.6545454545454545,
      "eval_PRM NPV": 0.5625,
      "eval_PRM Precision": 0.9324324324324325,
      "eval_PRM Recall": 0.8313253012048193,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.3780423104763031,
      "eval_runtime": 14.4931,
      "eval_samples_per_second": 3.036,
      "eval_steps_per_second": 0.138,
      "step": 370
    },
    {
      "epoch": 0.09607976433265353,
      "grad_norm": 6.25623119390206,
      "learning_rate": 1.1983204134366926e-06,
      "loss": 0.4254,
      "step": 371
    },
    {
      "epoch": 0.09633873943867145,
      "grad_norm": 6.9752169729123,
      "learning_rate": 1.2015503875968994e-06,
      "loss": 0.5261,
      "step": 372
    },
    {
      "epoch": 0.09659771454468939,
      "grad_norm": 5.294474073160396,
      "learning_rate": 1.2047803617571061e-06,
      "loss": 0.374,
      "step": 373
    },
    {
      "epoch": 0.09685668965070733,
      "grad_norm": 4.659531419947484,
      "learning_rate": 1.2080103359173128e-06,
      "loss": 0.3569,
      "step": 374
    },
    {
      "epoch": 0.09711566475672526,
      "grad_norm": 4.039346702256107,
      "learning_rate": 1.2112403100775194e-06,
      "loss": 0.4512,
      "step": 375
    },
    {
      "epoch": 0.09711566475672526,
      "eval_PRM Accuracy": 0.8301886792452831,
      "eval_PRM F1": 0.8846153846153846,
      "eval_PRM F1 AUC": 0.8287061288632793,
      "eval_PRM F1 AUC (fixed)": 0.5746464117338921,
      "eval_PRM F1 Neg": 0.6785714285714286,
      "eval_PRM NPV": 0.5757575757575758,
      "eval_PRM Precision": 0.9452054794520548,
      "eval_PRM Recall": 0.8313253012048193,
      "eval_PRM Specificty": 0.8260869565217391,
      "eval_loss": 0.37546131014823914,
      "eval_runtime": 14.8182,
      "eval_samples_per_second": 2.969,
      "eval_steps_per_second": 0.135,
      "step": 375
    },
    {
      "epoch": 0.0973746398627432,
      "grad_norm": 6.1017493311919635,
      "learning_rate": 1.2144702842377263e-06,
      "loss": 0.4177,
      "step": 376
    },
    {
      "epoch": 0.09763361496876113,
      "grad_norm": 10.032807377159754,
      "learning_rate": 1.217700258397933e-06,
      "loss": 0.4064,
      "step": 377
    },
    {
      "epoch": 0.09789259007477906,
      "grad_norm": 9.509320958414998,
      "learning_rate": 1.2209302325581397e-06,
      "loss": 0.4294,
      "step": 378
    },
    {
      "epoch": 0.098151565180797,
      "grad_norm": 6.893477527872258,
      "learning_rate": 1.2241602067183463e-06,
      "loss": 0.3717,
      "step": 379
    },
    {
      "epoch": 0.09841054028681492,
      "grad_norm": 4.619892022039742,
      "learning_rate": 1.227390180878553e-06,
      "loss": 0.4551,
      "step": 380
    },
    {
      "epoch": 0.09841054028681492,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.8930817610062893,
      "eval_PRM F1 AUC": 0.819015191199581,
      "eval_PRM F1 AUC (fixed)": 0.5914091147197484,
      "eval_PRM F1 Neg": 0.6792452830188679,
      "eval_PRM NPV": 0.6,
      "eval_PRM Precision": 0.9342105263157895,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.35004815459251404,
      "eval_runtime": 14.5841,
      "eval_samples_per_second": 3.017,
      "eval_steps_per_second": 0.137,
      "step": 380
    },
    {
      "epoch": 0.09866951539283286,
      "grad_norm": 8.222153879654302,
      "learning_rate": 1.2306201550387597e-06,
      "loss": 0.4807,
      "step": 381
    },
    {
      "epoch": 0.0989284904988508,
      "grad_norm": 6.9014811801050895,
      "learning_rate": 1.2338501291989665e-06,
      "loss": 0.4988,
      "step": 382
    },
    {
      "epoch": 0.09918746560486873,
      "grad_norm": 6.180778661225844,
      "learning_rate": 1.2370801033591732e-06,
      "loss": 0.3045,
      "step": 383
    },
    {
      "epoch": 0.09944644071088667,
      "grad_norm": 9.700681223535515,
      "learning_rate": 1.2403100775193799e-06,
      "loss": 0.458,
      "step": 384
    },
    {
      "epoch": 0.0997054158169046,
      "grad_norm": 4.728499860184501,
      "learning_rate": 1.2435400516795866e-06,
      "loss": 0.4022,
      "step": 385
    },
    {
      "epoch": 0.0997054158169046,
      "eval_PRM Accuracy": 0.8301886792452831,
      "eval_PRM F1": 0.8860759493670886,
      "eval_PRM F1 AUC": 0.8129910948140389,
      "eval_PRM F1 AUC (fixed)": 0.606862231534835,
      "eval_PRM F1 Neg": 0.6666666666666666,
      "eval_PRM NPV": 0.5806451612903226,
      "eval_PRM Precision": 0.9333333333333333,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.34537962079048157,
      "eval_runtime": 14.7382,
      "eval_samples_per_second": 2.985,
      "eval_steps_per_second": 0.136,
      "step": 385
    },
    {
      "epoch": 0.09996439092292253,
      "grad_norm": 8.353583602250334,
      "learning_rate": 1.2467700258397934e-06,
      "loss": 0.4462,
      "step": 386
    },
    {
      "epoch": 0.10022336602894047,
      "grad_norm": 4.259457118742542,
      "learning_rate": 1.25e-06,
      "loss": 0.4462,
      "step": 387
    },
    {
      "epoch": 0.10048234113495841,
      "grad_norm": 6.045747029072608,
      "learning_rate": 1.2499997444416303e-06,
      "loss": 0.316,
      "step": 388
    },
    {
      "epoch": 0.10074131624097633,
      "grad_norm": 6.549109829669429,
      "learning_rate": 1.2499989777667297e-06,
      "loss": 0.3509,
      "step": 389
    },
    {
      "epoch": 0.10100029134699427,
      "grad_norm": 5.862650772896159,
      "learning_rate": 1.2499976999759252e-06,
      "loss": 0.4487,
      "step": 390
    },
    {
      "epoch": 0.10100029134699427,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.89171974522293,
      "eval_PRM F1 AUC": 0.8347302252488213,
      "eval_PRM F1 AUC (fixed)": 0.6191723415400734,
      "eval_PRM F1 Neg": 0.6909090909090909,
      "eval_PRM NPV": 0.59375,
      "eval_PRM Precision": 0.9459459459459459,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.8260869565217391,
      "eval_loss": 0.3483627140522003,
      "eval_runtime": 14.609,
      "eval_samples_per_second": 3.012,
      "eval_steps_per_second": 0.137,
      "step": 390
    },
    {
      "epoch": 0.1012592664530122,
      "grad_norm": 4.633090184138419,
      "learning_rate": 1.249995911070262e-06,
      "loss": 0.4869,
      "step": 391
    },
    {
      "epoch": 0.10151824155903014,
      "grad_norm": 14.179051341269778,
      "learning_rate": 1.2499936110512028e-06,
      "loss": 0.4242,
      "step": 392
    },
    {
      "epoch": 0.10177721666504808,
      "grad_norm": 3.6200503143835547,
      "learning_rate": 1.2499907999206288e-06,
      "loss": 0.3389,
      "step": 393
    },
    {
      "epoch": 0.102036191771066,
      "grad_norm": 8.94604690308416,
      "learning_rate": 1.2499874776808384e-06,
      "loss": 0.4031,
      "step": 394
    },
    {
      "epoch": 0.10229516687708394,
      "grad_norm": 6.1031012803457365,
      "learning_rate": 1.2499836443345493e-06,
      "loss": 0.4491,
      "step": 395
    },
    {
      "epoch": 0.10229516687708394,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.8701298701298701,
      "eval_PRM F1 AUC": 0.8166579360921948,
      "eval_PRM F1 AUC (fixed)": 0.6191723415400733,
      "eval_PRM F1 Neg": 0.6551724137931034,
      "eval_PRM NPV": 0.5428571428571428,
      "eval_PRM Precision": 0.9436619718309859,
      "eval_PRM Recall": 0.8072289156626506,
      "eval_PRM Specificty": 0.8260869565217391,
      "eval_loss": 0.3896908760070801,
      "eval_runtime": 14.657,
      "eval_samples_per_second": 3.002,
      "eval_steps_per_second": 0.136,
      "step": 395
    },
    {
      "epoch": 0.10255414198310188,
      "grad_norm": 11.484707576135863,
      "learning_rate": 1.2499792998848956e-06,
      "loss": 0.3512,
      "step": 396
    },
    {
      "epoch": 0.1028131170891198,
      "grad_norm": 8.423723647150908,
      "learning_rate": 1.2499744443354305e-06,
      "loss": 0.5061,
      "step": 397
    },
    {
      "epoch": 0.10307209219513774,
      "grad_norm": 3.9133153276637427,
      "learning_rate": 1.2499690776901246e-06,
      "loss": 0.3537,
      "step": 398
    },
    {
      "epoch": 0.10333106730115568,
      "grad_norm": 5.0512139528828355,
      "learning_rate": 1.2499631999533668e-06,
      "loss": 0.5046,
      "step": 399
    },
    {
      "epoch": 0.10359004240717361,
      "grad_norm": 5.887848469989917,
      "learning_rate": 1.2499568111299641e-06,
      "loss": 0.5188,
      "step": 400
    },
    {
      "epoch": 0.10359004240717361,
      "eval_PRM Accuracy": 0.8301886792452831,
      "eval_PRM F1": 0.8860759493670886,
      "eval_PRM F1 AUC": 0.8129910948140389,
      "eval_PRM F1 AUC (fixed)": 0.6097433211105291,
      "eval_PRM F1 Neg": 0.6666666666666666,
      "eval_PRM NPV": 0.5806451612903226,
      "eval_PRM Precision": 0.9333333333333333,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.37192729115486145,
      "eval_runtime": 14.852,
      "eval_samples_per_second": 2.963,
      "eval_steps_per_second": 0.135,
      "step": 400
    },
    {
      "epoch": 0.10384901751319155,
      "grad_norm": 6.188275540711092,
      "learning_rate": 1.2499499112251407e-06,
      "loss": 0.4381,
      "step": 401
    },
    {
      "epoch": 0.10410799261920947,
      "grad_norm": 4.422397712181315,
      "learning_rate": 1.2499425002445396e-06,
      "loss": 0.3981,
      "step": 402
    },
    {
      "epoch": 0.10436696772522741,
      "grad_norm": 3.848781758597374,
      "learning_rate": 1.2499345781942212e-06,
      "loss": 0.4092,
      "step": 403
    },
    {
      "epoch": 0.10462594283124535,
      "grad_norm": 11.553444118243934,
      "learning_rate": 1.2499261450806644e-06,
      "loss": 0.424,
      "step": 404
    },
    {
      "epoch": 0.10488491793726328,
      "grad_norm": 5.996576278633823,
      "learning_rate": 1.249917200910765e-06,
      "loss": 0.4393,
      "step": 405
    },
    {
      "epoch": 0.10488491793726328,
      "eval_PRM Accuracy": 0.8301886792452831,
      "eval_PRM F1": 0.8846153846153846,
      "eval_PRM F1 AUC": 0.8287061288632793,
      "eval_PRM F1 AUC (fixed)": 0.5955997904662127,
      "eval_PRM F1 Neg": 0.6785714285714286,
      "eval_PRM NPV": 0.5757575757575758,
      "eval_PRM Precision": 0.9452054794520548,
      "eval_PRM Recall": 0.8313253012048193,
      "eval_PRM Specificty": 0.8260869565217391,
      "eval_loss": 0.37841737270355225,
      "eval_runtime": 14.4713,
      "eval_samples_per_second": 3.041,
      "eval_steps_per_second": 0.138,
      "step": 405
    },
    {
      "epoch": 0.10514389304328121,
      "grad_norm": 9.190741668299102,
      "learning_rate": 1.249907745691838e-06,
      "loss": 0.4367,
      "step": 406
    },
    {
      "epoch": 0.10540286814929915,
      "grad_norm": 5.8752181526111515,
      "learning_rate": 1.2498977794316158e-06,
      "loss": 0.5232,
      "step": 407
    },
    {
      "epoch": 0.10566184325531708,
      "grad_norm": 4.1488194676706005,
      "learning_rate": 1.2498873021382483e-06,
      "loss": 0.4059,
      "step": 408
    },
    {
      "epoch": 0.10592081836133502,
      "grad_norm": 5.550763774298918,
      "learning_rate": 1.2498763138203038e-06,
      "loss": 0.4206,
      "step": 409
    },
    {
      "epoch": 0.10617979346735296,
      "grad_norm": 5.264681618753274,
      "learning_rate": 1.2498648144867686e-06,
      "loss": 0.3559,
      "step": 410
    },
    {
      "epoch": 0.10617979346735296,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.8930817610062893,
      "eval_PRM F1 AUC": 0.819015191199581,
      "eval_PRM F1 AUC (fixed)": 0.5835515976951283,
      "eval_PRM F1 Neg": 0.6792452830188679,
      "eval_PRM NPV": 0.6,
      "eval_PRM Precision": 0.9342105263157895,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.36815035343170166,
      "eval_runtime": 14.7219,
      "eval_samples_per_second": 2.989,
      "eval_steps_per_second": 0.136,
      "step": 410
    },
    {
      "epoch": 0.10643876857337088,
      "grad_norm": 13.688204458169853,
      "learning_rate": 1.2498528041470464e-06,
      "loss": 0.4144,
      "step": 411
    },
    {
      "epoch": 0.10669774367938882,
      "grad_norm": 8.91627110732376,
      "learning_rate": 1.2498402828109594e-06,
      "loss": 0.3403,
      "step": 412
    },
    {
      "epoch": 0.10695671878540676,
      "grad_norm": 8.443031410437765,
      "learning_rate": 1.2498272504887472e-06,
      "loss": 0.414,
      "step": 413
    },
    {
      "epoch": 0.10721569389142468,
      "grad_norm": 12.324495442513667,
      "learning_rate": 1.2498137071910677e-06,
      "loss": 0.4573,
      "step": 414
    },
    {
      "epoch": 0.10747466899744262,
      "grad_norm": 4.4998378067906994,
      "learning_rate": 1.249799652928996e-06,
      "loss": 0.3958,
      "step": 415
    },
    {
      "epoch": 0.10747466899744262,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.89171974522293,
      "eval_PRM F1 AUC": 0.8347302252488213,
      "eval_PRM F1 AUC (fixed)": 0.5845992666317443,
      "eval_PRM F1 Neg": 0.6909090909090909,
      "eval_PRM NPV": 0.59375,
      "eval_PRM Precision": 0.9459459459459459,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.8260869565217391,
      "eval_loss": 0.36931851506233215,
      "eval_runtime": 14.5493,
      "eval_samples_per_second": 3.024,
      "eval_steps_per_second": 0.137,
      "step": 415
    },
    {
      "epoch": 0.10773364410346055,
      "grad_norm": 5.59070218598532,
      "learning_rate": 1.2497850877140258e-06,
      "loss": 0.4822,
      "step": 416
    },
    {
      "epoch": 0.10799261920947849,
      "grad_norm": 8.968937575600412,
      "learning_rate": 1.2497700115580683e-06,
      "loss": 0.483,
      "step": 417
    },
    {
      "epoch": 0.10825159431549643,
      "grad_norm": 5.078418619404218,
      "learning_rate": 1.2497544244734524e-06,
      "loss": 0.4509,
      "step": 418
    },
    {
      "epoch": 0.10851056942151435,
      "grad_norm": 12.05259697828282,
      "learning_rate": 1.2497383264729254e-06,
      "loss": 0.49,
      "step": 419
    },
    {
      "epoch": 0.10876954452753229,
      "grad_norm": 5.62697187532568,
      "learning_rate": 1.2497217175696517e-06,
      "loss": 0.5459,
      "step": 420
    },
    {
      "epoch": 0.10876954452753229,
      "eval_PRM Accuracy": 0.7735849056603774,
      "eval_PRM F1": 0.8378378378378378,
      "eval_PRM F1 AUC": 0.8082765845992665,
      "eval_PRM F1 AUC (fixed)": 0.5971712938711368,
      "eval_PRM F1 Neg": 0.625,
      "eval_PRM NPV": 0.4878048780487805,
      "eval_PRM Precision": 0.9538461538461539,
      "eval_PRM Recall": 0.7469879518072289,
      "eval_PRM Specificty": 0.8695652173913043,
      "eval_loss": 0.38043850660324097,
      "eval_runtime": 14.9479,
      "eval_samples_per_second": 2.944,
      "eval_steps_per_second": 0.134,
      "step": 420
    },
    {
      "epoch": 0.10902851963355023,
      "grad_norm": 9.10491260940364,
      "learning_rate": 1.2497045977772138e-06,
      "loss": 0.4919,
      "step": 421
    },
    {
      "epoch": 0.10928749473956816,
      "grad_norm": 9.647920546907573,
      "learning_rate": 1.2496869671096123e-06,
      "loss": 0.4094,
      "step": 422
    },
    {
      "epoch": 0.1095464698455861,
      "grad_norm": 10.094138514190103,
      "learning_rate": 1.2496688255812653e-06,
      "loss": 0.35,
      "step": 423
    },
    {
      "epoch": 0.10980544495160403,
      "grad_norm": 5.931036453890864,
      "learning_rate": 1.2496501732070086e-06,
      "loss": 0.3879,
      "step": 424
    },
    {
      "epoch": 0.11006442005762196,
      "grad_norm": 3.8019022867593213,
      "learning_rate": 1.2496310100020957e-06,
      "loss": 0.3735,
      "step": 425
    },
    {
      "epoch": 0.11006442005762196,
      "eval_PRM Accuracy": 0.8490566037735849,
      "eval_PRM F1": 0.9,
      "eval_PRM F1 AUC": 0.8250392875851232,
      "eval_PRM F1 AUC (fixed)": 0.6160293347302253,
      "eval_PRM F1 Neg": 0.6923076923076923,
      "eval_PRM NPV": 0.6206896551724138,
      "eval_PRM Precision": 0.935064935064935,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.328229695558548,
      "eval_runtime": 14.6799,
      "eval_samples_per_second": 2.997,
      "eval_steps_per_second": 0.136,
      "step": 425
    },
    {
      "epoch": 0.1103233951636399,
      "grad_norm": 5.07041457856754,
      "learning_rate": 1.2496113359821982e-06,
      "loss": 0.3717,
      "step": 426
    },
    {
      "epoch": 0.11058237026965782,
      "grad_norm": 3.9324565423627926,
      "learning_rate": 1.2495911511634052e-06,
      "loss": 0.3909,
      "step": 427
    },
    {
      "epoch": 0.11084134537567576,
      "grad_norm": 4.646597382905398,
      "learning_rate": 1.2495704555622238e-06,
      "loss": 0.3355,
      "step": 428
    },
    {
      "epoch": 0.1111003204816937,
      "grad_norm": 9.385735386008209,
      "learning_rate": 1.2495492491955783e-06,
      "loss": 0.4172,
      "step": 429
    },
    {
      "epoch": 0.11135929558771163,
      "grad_norm": 4.973918058742094,
      "learning_rate": 1.2495275320808112e-06,
      "loss": 0.4354,
      "step": 430
    },
    {
      "epoch": 0.11135929558771163,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.8957055214723927,
      "eval_PRM F1 AUC": 0.7875851231011,
      "eval_PRM F1 AUC (fixed)": 0.6296490309062337,
      "eval_PRM F1 Neg": 0.6530612244897959,
      "eval_PRM NPV": 0.6153846153846154,
      "eval_PRM Precision": 0.9125,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.31378573179244995,
      "eval_runtime": 14.6402,
      "eval_samples_per_second": 3.005,
      "eval_steps_per_second": 0.137,
      "step": 430
    },
    {
      "epoch": 0.11161827069372957,
      "grad_norm": 17.05390807031748,
      "learning_rate": 1.2495053042356823e-06,
      "loss": 0.4862,
      "step": 431
    },
    {
      "epoch": 0.1118772457997475,
      "grad_norm": 7.012072282963652,
      "learning_rate": 1.2494825656783692e-06,
      "loss": 0.4364,
      "step": 432
    },
    {
      "epoch": 0.11213622090576543,
      "grad_norm": 3.7158958642428055,
      "learning_rate": 1.2494593164274673e-06,
      "loss": 0.3721,
      "step": 433
    },
    {
      "epoch": 0.11239519601178337,
      "grad_norm": 11.095370577712787,
      "learning_rate": 1.2494355565019897e-06,
      "loss": 0.4619,
      "step": 434
    },
    {
      "epoch": 0.11265417111780131,
      "grad_norm": 8.423672434963388,
      "learning_rate": 1.2494112859213667e-06,
      "loss": 0.4653,
      "step": 435
    },
    {
      "epoch": 0.11265417111780131,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8789808917197452,
      "eval_PRM F1 AUC": 0.8069669984284967,
      "eval_PRM F1 AUC (fixed)": 0.6272917757988475,
      "eval_PRM F1 Neg": 0.6545454545454545,
      "eval_PRM NPV": 0.5625,
      "eval_PRM Precision": 0.9324324324324325,
      "eval_PRM Recall": 0.8313253012048193,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.33313944935798645,
      "eval_runtime": 14.7779,
      "eval_samples_per_second": 2.977,
      "eval_steps_per_second": 0.135,
      "step": 435
    },
    {
      "epoch": 0.11291314622381923,
      "grad_norm": 6.314062135089114,
      "learning_rate": 1.2493865047054465e-06,
      "loss": 0.3681,
      "step": 436
    },
    {
      "epoch": 0.11317212132983717,
      "grad_norm": 5.757644590025226,
      "learning_rate": 1.2493612128744948e-06,
      "loss": 0.5514,
      "step": 437
    },
    {
      "epoch": 0.1134310964358551,
      "grad_norm": 5.523908619118926,
      "learning_rate": 1.249335410449195e-06,
      "loss": 0.4419,
      "step": 438
    },
    {
      "epoch": 0.11369007154187304,
      "grad_norm": 5.943773174013052,
      "learning_rate": 1.2493090974506483e-06,
      "loss": 0.4776,
      "step": 439
    },
    {
      "epoch": 0.11394904664789098,
      "grad_norm": 20.29662028796124,
      "learning_rate": 1.2492822739003727e-06,
      "loss": 0.5249,
      "step": 440
    },
    {
      "epoch": 0.11394904664789098,
      "eval_PRM Accuracy": 0.7452830188679245,
      "eval_PRM F1": 0.8163265306122449,
      "eval_PRM F1 AUC": 0.7744892613933997,
      "eval_PRM F1 AUC (fixed)": 0.6115767417496071,
      "eval_PRM F1 Neg": 0.5846153846153846,
      "eval_PRM NPV": 0.4523809523809524,
      "eval_PRM Precision": 0.9375,
      "eval_PRM Recall": 0.7228915662650602,
      "eval_PRM Specificty": 0.8260869565217391,
      "eval_loss": 0.3871985375881195,
      "eval_runtime": 14.7587,
      "eval_samples_per_second": 2.981,
      "eval_steps_per_second": 0.136,
      "step": 440
    },
    {
      "epoch": 0.1142080217539089,
      "grad_norm": 8.993042445023589,
      "learning_rate": 1.2492549398203041e-06,
      "loss": 0.4704,
      "step": 441
    },
    {
      "epoch": 0.11446699685992684,
      "grad_norm": 12.199717818133639,
      "learning_rate": 1.2492270952327962e-06,
      "loss": 0.3264,
      "step": 442
    },
    {
      "epoch": 0.11472597196594478,
      "grad_norm": 8.978520690223112,
      "learning_rate": 1.2491987401606198e-06,
      "loss": 0.4283,
      "step": 443
    },
    {
      "epoch": 0.1149849470719627,
      "grad_norm": 3.809395479659997,
      "learning_rate": 1.2491698746269634e-06,
      "loss": 0.4611,
      "step": 444
    },
    {
      "epoch": 0.11524392217798064,
      "grad_norm": 3.3687122852485687,
      "learning_rate": 1.2491404986554328e-06,
      "loss": 0.385,
      "step": 445
    },
    {
      "epoch": 0.11524392217798064,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.89171974522293,
      "eval_PRM F1 AUC": 0.8347302252488213,
      "eval_PRM F1 AUC (fixed)": 0.6162912519643792,
      "eval_PRM F1 Neg": 0.6909090909090909,
      "eval_PRM NPV": 0.59375,
      "eval_PRM Precision": 0.9459459459459459,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.8260869565217391,
      "eval_loss": 0.3389197587966919,
      "eval_runtime": 14.7519,
      "eval_samples_per_second": 2.983,
      "eval_steps_per_second": 0.136,
      "step": 445
    },
    {
      "epoch": 0.11550289728399858,
      "grad_norm": 5.513267646704666,
      "learning_rate": 1.2491106122700512e-06,
      "loss": 0.373,
      "step": 446
    },
    {
      "epoch": 0.11576187239001651,
      "grad_norm": 4.586155749253553,
      "learning_rate": 1.2490802154952595e-06,
      "loss": 0.3982,
      "step": 447
    },
    {
      "epoch": 0.11602084749603445,
      "grad_norm": 4.643300680941585,
      "learning_rate": 1.2490493083559156e-06,
      "loss": 0.3947,
      "step": 448
    },
    {
      "epoch": 0.11627982260205237,
      "grad_norm": 12.093335727107991,
      "learning_rate": 1.249017890877295e-06,
      "loss": 0.475,
      "step": 449
    },
    {
      "epoch": 0.11653879770807031,
      "grad_norm": 8.246043522729842,
      "learning_rate": 1.2489859630850904e-06,
      "loss": 0.4301,
      "step": 450
    },
    {
      "epoch": 0.11653879770807031,
      "eval_PRM Accuracy": 0.8490566037735849,
      "eval_PRM F1": 0.9012345679012346,
      "eval_PRM F1 AUC": 0.8093242535358826,
      "eval_PRM F1 AUC (fixed)": 0.6278156102671556,
      "eval_PRM F1 Neg": 0.68,
      "eval_PRM NPV": 0.6296296296296297,
      "eval_PRM Precision": 0.9240506329113924,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.32794228196144104,
      "eval_runtime": 14.7823,
      "eval_samples_per_second": 2.977,
      "eval_steps_per_second": 0.135,
      "step": 450
    },
    {
      "epoch": 0.11679777281408825,
      "grad_norm": 6.077070805952264,
      "learning_rate": 1.2489535250054122e-06,
      "loss": 0.4336,
      "step": 451
    },
    {
      "epoch": 0.11705674792010617,
      "grad_norm": 4.370192483038455,
      "learning_rate": 1.2489205766647878e-06,
      "loss": 0.379,
      "step": 452
    },
    {
      "epoch": 0.11731572302612411,
      "grad_norm": 3.0466331122024153,
      "learning_rate": 1.2488871180901615e-06,
      "loss": 0.3485,
      "step": 453
    },
    {
      "epoch": 0.11757469813214205,
      "grad_norm": 3.505329612931744,
      "learning_rate": 1.2488531493088958e-06,
      "loss": 0.3886,
      "step": 454
    },
    {
      "epoch": 0.11783367323815998,
      "grad_norm": 4.757786960622849,
      "learning_rate": 1.2488186703487692e-06,
      "loss": 0.3205,
      "step": 455
    },
    {
      "epoch": 0.11783367323815998,
      "eval_PRM Accuracy": 0.8301886792452831,
      "eval_PRM F1": 0.8860759493670886,
      "eval_PRM F1 AUC": 0.8129910948140389,
      "eval_PRM F1 AUC (fixed)": 0.6220534311157674,
      "eval_PRM F1 Neg": 0.6666666666666666,
      "eval_PRM NPV": 0.5806451612903226,
      "eval_PRM Precision": 0.9333333333333333,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.33780306577682495,
      "eval_runtime": 14.8292,
      "eval_samples_per_second": 2.967,
      "eval_steps_per_second": 0.135,
      "step": 455
    },
    {
      "epoch": 0.11809264834417792,
      "grad_norm": 4.590493377069403,
      "learning_rate": 1.248783681237979e-06,
      "loss": 0.3612,
      "step": 456
    },
    {
      "epoch": 0.11835162345019586,
      "grad_norm": 3.476272900504724,
      "learning_rate": 1.2487481820051384e-06,
      "loss": 0.382,
      "step": 457
    },
    {
      "epoch": 0.11861059855621378,
      "grad_norm": 3.1106107478564233,
      "learning_rate": 1.2487121726792784e-06,
      "loss": 0.3405,
      "step": 458
    },
    {
      "epoch": 0.11886957366223172,
      "grad_norm": 4.674172376087834,
      "learning_rate": 1.2486756532898465e-06,
      "loss": 0.3694,
      "step": 459
    },
    {
      "epoch": 0.11912854876824965,
      "grad_norm": 4.708991145064097,
      "learning_rate": 1.2486386238667083e-06,
      "loss": 0.3287,
      "step": 460
    },
    {
      "epoch": 0.11912854876824965,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.8930817610062893,
      "eval_PRM F1 AUC": 0.819015191199581,
      "eval_PRM F1 AUC (fixed)": 0.63986380303824,
      "eval_PRM F1 Neg": 0.6792452830188679,
      "eval_PRM NPV": 0.6,
      "eval_PRM Precision": 0.9342105263157895,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.32568544149398804,
      "eval_runtime": 14.6822,
      "eval_samples_per_second": 2.997,
      "eval_steps_per_second": 0.136,
      "step": 460
    },
    {
      "epoch": 0.11938752387426758,
      "grad_norm": 3.2665701190298364,
      "learning_rate": 1.2486010844401458e-06,
      "loss": 0.309,
      "step": 461
    },
    {
      "epoch": 0.11964649898028552,
      "grad_norm": 14.608377136123803,
      "learning_rate": 1.2485630350408577e-06,
      "loss": 0.4372,
      "step": 462
    },
    {
      "epoch": 0.11990547408630345,
      "grad_norm": 4.692121241159457,
      "learning_rate": 1.2485244756999615e-06,
      "loss": 0.3702,
      "step": 463
    },
    {
      "epoch": 0.12016444919232139,
      "grad_norm": 3.5397062939111357,
      "learning_rate": 1.2484854064489895e-06,
      "loss": 0.3609,
      "step": 464
    },
    {
      "epoch": 0.12042342429833933,
      "grad_norm": 3.4262687284743394,
      "learning_rate": 1.2484458273198923e-06,
      "loss": 0.3629,
      "step": 465
    },
    {
      "epoch": 0.12042342429833933,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8789808917197452,
      "eval_PRM F1 AUC": 0.8069669984284967,
      "eval_PRM F1 AUC (fixed)": 0.6490309062336301,
      "eval_PRM F1 Neg": 0.6545454545454545,
      "eval_PRM NPV": 0.5625,
      "eval_PRM Precision": 0.9324324324324325,
      "eval_PRM Recall": 0.8313253012048193,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.3342059850692749,
      "eval_runtime": 14.3984,
      "eval_samples_per_second": 3.056,
      "eval_steps_per_second": 0.139,
      "step": 465
    },
    {
      "epoch": 0.12068239940435725,
      "grad_norm": 6.765235304929425,
      "learning_rate": 1.2484057383450372e-06,
      "loss": 0.5116,
      "step": 466
    },
    {
      "epoch": 0.12094137451037519,
      "grad_norm": 4.3624619071051765,
      "learning_rate": 1.2483651395572086e-06,
      "loss": 0.4132,
      "step": 467
    },
    {
      "epoch": 0.12120034961639313,
      "grad_norm": 8.833253754878113,
      "learning_rate": 1.2483240309896073e-06,
      "loss": 0.4045,
      "step": 468
    },
    {
      "epoch": 0.12145932472241106,
      "grad_norm": 4.98960724246876,
      "learning_rate": 1.2482824126758517e-06,
      "loss": 0.3841,
      "step": 469
    },
    {
      "epoch": 0.121718299828429,
      "grad_norm": 4.3471531348502275,
      "learning_rate": 1.2482402846499764e-06,
      "loss": 0.3716,
      "step": 470
    },
    {
      "epoch": 0.121718299828429,
      "eval_PRM Accuracy": 0.8301886792452831,
      "eval_PRM F1": 0.8860759493670886,
      "eval_PRM F1 AUC": 0.8129910948140389,
      "eval_PRM F1 AUC (fixed)": 0.6485070717653222,
      "eval_PRM F1 Neg": 0.6666666666666666,
      "eval_PRM NPV": 0.5806451612903226,
      "eval_PRM Precision": 0.9333333333333333,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.32934725284576416,
      "eval_runtime": 14.705,
      "eval_samples_per_second": 2.992,
      "eval_steps_per_second": 0.136,
      "step": 470
    },
    {
      "epoch": 0.12197727493444693,
      "grad_norm": 4.861294081639487,
      "learning_rate": 1.2481976469464334e-06,
      "loss": 0.4188,
      "step": 471
    },
    {
      "epoch": 0.12223625004046486,
      "grad_norm": 7.2929021973775345,
      "learning_rate": 1.2481544996000913e-06,
      "loss": 0.3866,
      "step": 472
    },
    {
      "epoch": 0.1224952251464828,
      "grad_norm": 5.221068095367759,
      "learning_rate": 1.2481108426462352e-06,
      "loss": 0.3554,
      "step": 473
    },
    {
      "epoch": 0.12275420025250072,
      "grad_norm": 3.392837388495582,
      "learning_rate": 1.248066676120567e-06,
      "loss": 0.3725,
      "step": 474
    },
    {
      "epoch": 0.12301317535851866,
      "grad_norm": 3.7543275834934877,
      "learning_rate": 1.248022000059206e-06,
      "loss": 0.3458,
      "step": 475
    },
    {
      "epoch": 0.12301317535851866,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.8944099378881988,
      "eval_PRM F1 AUC": 0.8033001571503404,
      "eval_PRM F1 AUC (fixed)": 0.6437925615505501,
      "eval_PRM F1 Neg": 0.6666666666666666,
      "eval_PRM NPV": 0.6071428571428571,
      "eval_PRM Precision": 0.9230769230769231,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.31160077452659607,
      "eval_runtime": 14.6465,
      "eval_samples_per_second": 3.004,
      "eval_steps_per_second": 0.137,
      "step": 475
    },
    {
      "epoch": 0.1232721504645366,
      "grad_norm": 21.593384607104042,
      "learning_rate": 1.2479768144986874e-06,
      "loss": 0.564,
      "step": 476
    },
    {
      "epoch": 0.12353112557055453,
      "grad_norm": 6.930229379845048,
      "learning_rate": 1.2479311194759636e-06,
      "loss": 0.4011,
      "step": 477
    },
    {
      "epoch": 0.12379010067657246,
      "grad_norm": 5.781015235162338,
      "learning_rate": 1.2478849150284029e-06,
      "loss": 0.2612,
      "step": 478
    },
    {
      "epoch": 0.1240490757825904,
      "grad_norm": 7.084988924381968,
      "learning_rate": 1.247838201193791e-06,
      "loss": 0.435,
      "step": 479
    },
    {
      "epoch": 0.12430805088860833,
      "grad_norm": 8.18854984265065,
      "learning_rate": 1.24779097801033e-06,
      "loss": 0.4731,
      "step": 480
    },
    {
      "epoch": 0.12430805088860833,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.8930817610062893,
      "eval_PRM F1 AUC": 0.819015191199581,
      "eval_PRM F1 AUC (fixed)": 0.6348873755893137,
      "eval_PRM F1 Neg": 0.6792452830188679,
      "eval_PRM NPV": 0.6,
      "eval_PRM Precision": 0.9342105263157895,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.3187015652656555,
      "eval_runtime": 14.4311,
      "eval_samples_per_second": 3.049,
      "eval_steps_per_second": 0.139,
      "step": 480
    },
    {
      "epoch": 0.12456702599462627,
      "grad_norm": 4.355038554690252,
      "learning_rate": 1.2477432455166382e-06,
      "loss": 0.3797,
      "step": 481
    },
    {
      "epoch": 0.12482600110064421,
      "grad_norm": 5.838501143643666,
      "learning_rate": 1.2476950037517505e-06,
      "loss": 0.4502,
      "step": 482
    },
    {
      "epoch": 0.12508497620666215,
      "grad_norm": 4.824847173062183,
      "learning_rate": 1.2476462527551186e-06,
      "loss": 0.4074,
      "step": 483
    },
    {
      "epoch": 0.12534395131268006,
      "grad_norm": 10.703781791565072,
      "learning_rate": 1.2475969925666104e-06,
      "loss": 0.3866,
      "step": 484
    },
    {
      "epoch": 0.125602926418698,
      "grad_norm": 11.792915909373495,
      "learning_rate": 1.2475472232265102e-06,
      "loss": 0.4068,
      "step": 485
    },
    {
      "epoch": 0.125602926418698,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8758169934640523,
      "eval_PRM F1 AUC": 0.8383970665269774,
      "eval_PRM F1 AUC (fixed)": 0.6272917757988475,
      "eval_PRM F1 Neg": 0.6779661016949152,
      "eval_PRM NPV": 0.5555555555555556,
      "eval_PRM Precision": 0.9571428571428572,
      "eval_PRM Recall": 0.8072289156626506,
      "eval_PRM Specificty": 0.8695652173913043,
      "eval_loss": 0.34957724809646606,
      "eval_runtime": 14.9385,
      "eval_samples_per_second": 2.945,
      "eval_steps_per_second": 0.134,
      "step": 485
    },
    {
      "epoch": 0.12586190152471594,
      "grad_norm": 12.342288855982682,
      "learning_rate": 1.2474969447755184e-06,
      "loss": 0.4546,
      "step": 486
    },
    {
      "epoch": 0.12612087663073387,
      "grad_norm": 7.4421966783654865,
      "learning_rate": 1.2474461572547525e-06,
      "loss": 0.4749,
      "step": 487
    },
    {
      "epoch": 0.1263798517367518,
      "grad_norm": 3.933100369825377,
      "learning_rate": 1.2473948607057456e-06,
      "loss": 0.3958,
      "step": 488
    },
    {
      "epoch": 0.12663882684276973,
      "grad_norm": 7.290452928627164,
      "learning_rate": 1.2473430551704477e-06,
      "loss": 0.4805,
      "step": 489
    },
    {
      "epoch": 0.12689780194878766,
      "grad_norm": 3.4175667460178616,
      "learning_rate": 1.2472907406912242e-06,
      "loss": 0.3287,
      "step": 490
    },
    {
      "epoch": 0.12689780194878766,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.8930817610062893,
      "eval_PRM F1 AUC": 0.819015191199581,
      "eval_PRM F1 AUC (fixed)": 0.6346254583551597,
      "eval_PRM F1 Neg": 0.6792452830188679,
      "eval_PRM NPV": 0.6,
      "eval_PRM Precision": 0.9342105263157895,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.3145117461681366,
      "eval_runtime": 14.5492,
      "eval_samples_per_second": 3.024,
      "eval_steps_per_second": 0.137,
      "step": 490
    },
    {
      "epoch": 0.1271567770548056,
      "grad_norm": 4.335883611097396,
      "learning_rate": 1.2472379173108574e-06,
      "loss": 0.2983,
      "step": 491
    },
    {
      "epoch": 0.12741575216082354,
      "grad_norm": 6.626921553086188,
      "learning_rate": 1.2471845850725456e-06,
      "loss": 0.4223,
      "step": 492
    },
    {
      "epoch": 0.12767472726684148,
      "grad_norm": 3.180075529152741,
      "learning_rate": 1.247130744019903e-06,
      "loss": 0.3306,
      "step": 493
    },
    {
      "epoch": 0.12793370237285942,
      "grad_norm": 8.566289040757376,
      "learning_rate": 1.2470763941969604e-06,
      "loss": 0.4057,
      "step": 494
    },
    {
      "epoch": 0.12819267747887733,
      "grad_norm": 19.349772589892186,
      "learning_rate": 1.2470215356481644e-06,
      "loss": 0.553,
      "step": 495
    },
    {
      "epoch": 0.12819267747887733,
      "eval_PRM Accuracy": 0.8490566037735849,
      "eval_PRM F1": 0.9012345679012346,
      "eval_PRM F1 AUC": 0.8093242535358826,
      "eval_PRM F1 AUC (fixed)": 0.6369827134625459,
      "eval_PRM F1 Neg": 0.68,
      "eval_PRM NPV": 0.6296296296296297,
      "eval_PRM Precision": 0.9240506329113924,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.3201056122779846,
      "eval_runtime": 14.5466,
      "eval_samples_per_second": 3.025,
      "eval_steps_per_second": 0.137,
      "step": 495
    },
    {
      "epoch": 0.12845165258489527,
      "grad_norm": 3.4846284567808343,
      "learning_rate": 1.246966168418377e-06,
      "loss": 0.3323,
      "step": 496
    },
    {
      "epoch": 0.1287106276909132,
      "grad_norm": 3.0603555272843113,
      "learning_rate": 1.2469102925528776e-06,
      "loss": 0.3422,
      "step": 497
    },
    {
      "epoch": 0.12896960279693115,
      "grad_norm": 4.009331855682547,
      "learning_rate": 1.2468539080973603e-06,
      "loss": 0.3721,
      "step": 498
    },
    {
      "epoch": 0.1292285779029491,
      "grad_norm": 6.272480918620614,
      "learning_rate": 1.2467970150979356e-06,
      "loss": 0.427,
      "step": 499
    },
    {
      "epoch": 0.12948755300896703,
      "grad_norm": 13.427629672952166,
      "learning_rate": 1.2467396136011298e-06,
      "loss": 0.4693,
      "step": 500
    },
    {
      "epoch": 0.12948755300896703,
      "eval_PRM Accuracy": 0.8301886792452831,
      "eval_PRM F1": 0.8860759493670886,
      "eval_PRM F1 AUC": 0.8129910948140389,
      "eval_PRM F1 AUC (fixed)": 0.6327920377160817,
      "eval_PRM F1 Neg": 0.6666666666666666,
      "eval_PRM NPV": 0.5806451612903226,
      "eval_PRM Precision": 0.9333333333333333,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.33357107639312744,
      "eval_runtime": 14.4212,
      "eval_samples_per_second": 3.051,
      "eval_steps_per_second": 0.139,
      "step": 500
    },
    {
      "epoch": 0.12974652811498494,
      "grad_norm": 5.434972663555558,
      "learning_rate": 1.246681703653885e-06,
      "loss": 0.3905,
      "step": 501
    },
    {
      "epoch": 0.13000550322100288,
      "grad_norm": 4.58952122757678,
      "learning_rate": 1.2466232853035596e-06,
      "loss": 0.5206,
      "step": 502
    },
    {
      "epoch": 0.13026447832702082,
      "grad_norm": 7.372300675268678,
      "learning_rate": 1.246564358597927e-06,
      "loss": 0.4235,
      "step": 503
    },
    {
      "epoch": 0.13052345343303876,
      "grad_norm": 4.142484155230843,
      "learning_rate": 1.2465049235851768e-06,
      "loss": 0.4143,
      "step": 504
    },
    {
      "epoch": 0.1307824285390567,
      "grad_norm": 10.4043112649764,
      "learning_rate": 1.2464449803139142e-06,
      "loss": 0.4165,
      "step": 505
    },
    {
      "epoch": 0.1307824285390567,
      "eval_PRM Accuracy": 0.8018867924528302,
      "eval_PRM F1": 0.8609271523178808,
      "eval_PRM F1 AUC": 0.826348873755893,
      "eval_PRM F1 AUC (fixed)": 0.6356731272917758,
      "eval_PRM F1 Neg": 0.6557377049180327,
      "eval_PRM NPV": 0.5263157894736842,
      "eval_PRM Precision": 0.9558823529411765,
      "eval_PRM Recall": 0.7831325301204819,
      "eval_PRM Specificty": 0.8695652173913043,
      "eval_loss": 0.3677997589111328,
      "eval_runtime": 15.0664,
      "eval_samples_per_second": 2.92,
      "eval_steps_per_second": 0.133,
      "step": 505
    },
    {
      "epoch": 0.1310414036450746,
      "grad_norm": 9.922312215454863,
      "learning_rate": 1.2463845288331596e-06,
      "loss": 0.3574,
      "step": 506
    },
    {
      "epoch": 0.13130037875109254,
      "grad_norm": 6.670265269248358,
      "learning_rate": 1.2463235691923503e-06,
      "loss": 0.4276,
      "step": 507
    },
    {
      "epoch": 0.13155935385711048,
      "grad_norm": 7.307285016571106,
      "learning_rate": 1.2462621014413373e-06,
      "loss": 0.3718,
      "step": 508
    },
    {
      "epoch": 0.13181832896312842,
      "grad_norm": 7.0173057930362095,
      "learning_rate": 1.2462001256303888e-06,
      "loss": 0.3169,
      "step": 509
    },
    {
      "epoch": 0.13207730406914636,
      "grad_norm": 5.3637100732393135,
      "learning_rate": 1.2461376418101877e-06,
      "loss": 0.3718,
      "step": 510
    },
    {
      "epoch": 0.13207730406914636,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.89171974522293,
      "eval_PRM F1 AUC": 0.8347302252488213,
      "eval_PRM F1 AUC (fixed)": 0.644578313253012,
      "eval_PRM F1 Neg": 0.6909090909090909,
      "eval_PRM NPV": 0.59375,
      "eval_PRM Precision": 0.9459459459459459,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.8260869565217391,
      "eval_loss": 0.323575884103775,
      "eval_runtime": 14.7572,
      "eval_samples_per_second": 2.982,
      "eval_steps_per_second": 0.136,
      "step": 510
    },
    {
      "epoch": 0.1323362791751643,
      "grad_norm": 3.949656761693001,
      "learning_rate": 1.246074650031832e-06,
      "loss": 0.3234,
      "step": 511
    },
    {
      "epoch": 0.1325952542811822,
      "grad_norm": 4.248192544379391,
      "learning_rate": 1.2460111503468363e-06,
      "loss": 0.4262,
      "step": 512
    },
    {
      "epoch": 0.13285422938720015,
      "grad_norm": 3.891048084630212,
      "learning_rate": 1.245947142807129e-06,
      "loss": 0.3222,
      "step": 513
    },
    {
      "epoch": 0.1331132044932181,
      "grad_norm": 4.148693831751682,
      "learning_rate": 1.2458826274650552e-06,
      "loss": 0.3477,
      "step": 514
    },
    {
      "epoch": 0.13337217959923603,
      "grad_norm": 6.540096957578557,
      "learning_rate": 1.2458176043733744e-06,
      "loss": 0.3139,
      "step": 515
    },
    {
      "epoch": 0.13337217959923603,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.8930817610062893,
      "eval_PRM F1 AUC": 0.819015191199581,
      "eval_PRM F1 AUC (fixed)": 0.6547930853850183,
      "eval_PRM F1 Neg": 0.6792452830188679,
      "eval_PRM NPV": 0.6,
      "eval_PRM Precision": 0.9342105263157895,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.3057812452316284,
      "eval_runtime": 14.6202,
      "eval_samples_per_second": 3.01,
      "eval_steps_per_second": 0.137,
      "step": 515
    },
    {
      "epoch": 0.13363115470525397,
      "grad_norm": 4.703165811412571,
      "learning_rate": 1.2457520735852618e-06,
      "loss": 0.4059,
      "step": 516
    },
    {
      "epoch": 0.13389012981127188,
      "grad_norm": 6.028053885312962,
      "learning_rate": 1.2456860351543074e-06,
      "loss": 0.4418,
      "step": 517
    },
    {
      "epoch": 0.13414910491728982,
      "grad_norm": 3.816982454189665,
      "learning_rate": 1.2456194891345167e-06,
      "loss": 0.3853,
      "step": 518
    },
    {
      "epoch": 0.13440808002330776,
      "grad_norm": 4.1845349649740395,
      "learning_rate": 1.2455524355803102e-06,
      "loss": 0.3931,
      "step": 519
    },
    {
      "epoch": 0.1346670551293257,
      "grad_norm": 5.069175129972723,
      "learning_rate": 1.2454848745465234e-06,
      "loss": 0.3706,
      "step": 520
    },
    {
      "epoch": 0.1346670551293257,
      "eval_PRM Accuracy": 0.8490566037735849,
      "eval_PRM F1": 0.8974358974358975,
      "eval_PRM F1 AUC": 0.8564693556836039,
      "eval_PRM F1 AUC (fixed)": 0.6519119958093242,
      "eval_PRM F1 Neg": 0.7142857142857143,
      "eval_PRM NPV": 0.6060606060606061,
      "eval_PRM Precision": 0.958904109589041,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.8695652173913043,
      "eval_loss": 0.31878462433815,
      "eval_runtime": 14.9989,
      "eval_samples_per_second": 2.934,
      "eval_steps_per_second": 0.133,
      "step": 520
    },
    {
      "epoch": 0.13492603023534364,
      "grad_norm": 5.554389828528387,
      "learning_rate": 1.2454168060884065e-06,
      "loss": 0.3495,
      "step": 521
    },
    {
      "epoch": 0.13518500534136157,
      "grad_norm": 4.269474285741495,
      "learning_rate": 1.2453482302616254e-06,
      "loss": 0.351,
      "step": 522
    },
    {
      "epoch": 0.1354439804473795,
      "grad_norm": 4.285143074157683,
      "learning_rate": 1.2452791471222605e-06,
      "loss": 0.4821,
      "step": 523
    },
    {
      "epoch": 0.13570295555339743,
      "grad_norm": 4.800867083848419,
      "learning_rate": 1.2452095567268069e-06,
      "loss": 0.3048,
      "step": 524
    },
    {
      "epoch": 0.13596193065941536,
      "grad_norm": 5.37470151118888,
      "learning_rate": 1.2451394591321747e-06,
      "loss": 0.4761,
      "step": 525
    },
    {
      "epoch": 0.13596193065941536,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.8930817610062893,
      "eval_PRM F1 AUC": 0.819015191199581,
      "eval_PRM F1 AUC (fixed)": 0.6521739130434783,
      "eval_PRM F1 Neg": 0.6792452830188679,
      "eval_PRM NPV": 0.6,
      "eval_PRM Precision": 0.9342105263157895,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.3053739368915558,
      "eval_runtime": 14.8675,
      "eval_samples_per_second": 2.959,
      "eval_steps_per_second": 0.135,
      "step": 525
    },
    {
      "epoch": 0.1362209057654333,
      "grad_norm": 4.76701924346624,
      "learning_rate": 1.2450688543956887e-06,
      "loss": 0.416,
      "step": 526
    },
    {
      "epoch": 0.13647988087145124,
      "grad_norm": 4.626643528344459,
      "learning_rate": 1.2449977425750886e-06,
      "loss": 0.3598,
      "step": 527
    },
    {
      "epoch": 0.13673885597746915,
      "grad_norm": 8.703126246641308,
      "learning_rate": 1.244926123728529e-06,
      "loss": 0.3911,
      "step": 528
    },
    {
      "epoch": 0.1369978310834871,
      "grad_norm": 15.454895749771413,
      "learning_rate": 1.2448539979145786e-06,
      "loss": 0.4538,
      "step": 529
    },
    {
      "epoch": 0.13725680618950503,
      "grad_norm": 3.4674946395586135,
      "learning_rate": 1.2447813651922208e-06,
      "loss": 0.34,
      "step": 530
    },
    {
      "epoch": 0.13725680618950503,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.8944099378881988,
      "eval_PRM F1 AUC": 0.8033001571503404,
      "eval_PRM F1 AUC (fixed)": 0.6511262441068622,
      "eval_PRM F1 Neg": 0.6666666666666666,
      "eval_PRM NPV": 0.6071428571428571,
      "eval_PRM Precision": 0.9230769230769231,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.3079970180988312,
      "eval_runtime": 14.6555,
      "eval_samples_per_second": 3.002,
      "eval_steps_per_second": 0.136,
      "step": 530
    },
    {
      "epoch": 0.13751578129552297,
      "grad_norm": 9.901461765314394,
      "learning_rate": 1.2447082256208538e-06,
      "loss": 0.4309,
      "step": 531
    },
    {
      "epoch": 0.1377747564015409,
      "grad_norm": 8.819595922470516,
      "learning_rate": 1.2446345792602901e-06,
      "loss": 0.3939,
      "step": 532
    },
    {
      "epoch": 0.13803373150755885,
      "grad_norm": 7.240407887637497,
      "learning_rate": 1.244560426170757e-06,
      "loss": 0.426,
      "step": 533
    },
    {
      "epoch": 0.13829270661357676,
      "grad_norm": 3.304557932787079,
      "learning_rate": 1.2444857664128957e-06,
      "loss": 0.3499,
      "step": 534
    },
    {
      "epoch": 0.1385516817195947,
      "grad_norm": 4.779209710817938,
      "learning_rate": 1.2444106000477619e-06,
      "loss": 0.3743,
      "step": 535
    },
    {
      "epoch": 0.1385516817195947,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.89171974522293,
      "eval_PRM F1 AUC": 0.8347302252488213,
      "eval_PRM F1 AUC (fixed)": 0.6550550026191724,
      "eval_PRM F1 Neg": 0.6909090909090909,
      "eval_PRM NPV": 0.59375,
      "eval_PRM Precision": 0.9459459459459459,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.8260869565217391,
      "eval_loss": 0.331030011177063,
      "eval_runtime": 14.3834,
      "eval_samples_per_second": 3.059,
      "eval_steps_per_second": 0.139,
      "step": 535
    },
    {
      "epoch": 0.13881065682561264,
      "grad_norm": 5.984315015375788,
      "learning_rate": 1.2443349271368257e-06,
      "loss": 0.2657,
      "step": 536
    },
    {
      "epoch": 0.13906963193163058,
      "grad_norm": 8.492618114157318,
      "learning_rate": 1.2442587477419717e-06,
      "loss": 0.3658,
      "step": 537
    },
    {
      "epoch": 0.13932860703764852,
      "grad_norm": 9.882539638982529,
      "learning_rate": 1.244182061925498e-06,
      "loss": 0.4839,
      "step": 538
    },
    {
      "epoch": 0.13958758214366643,
      "grad_norm": 4.006257788106991,
      "learning_rate": 1.2441048697501174e-06,
      "loss": 0.4156,
      "step": 539
    },
    {
      "epoch": 0.13984655724968437,
      "grad_norm": 6.065415435767625,
      "learning_rate": 1.2440271712789568e-06,
      "loss": 0.4337,
      "step": 540
    },
    {
      "epoch": 0.13984655724968437,
      "eval_PRM Accuracy": 0.8490566037735849,
      "eval_PRM F1": 0.8987341772151899,
      "eval_PRM F1 AUC": 0.8407543216343635,
      "eval_PRM F1 AUC (fixed)": 0.6665793609219487,
      "eval_PRM F1 Neg": 0.7037037037037037,
      "eval_PRM NPV": 0.6129032258064516,
      "eval_PRM Precision": 0.9466666666666667,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.8260869565217391,
      "eval_loss": 0.31409063935279846,
      "eval_runtime": 14.4095,
      "eval_samples_per_second": 3.054,
      "eval_steps_per_second": 0.139,
      "step": 540
    },
    {
      "epoch": 0.1401055323557023,
      "grad_norm": 8.055749515946466,
      "learning_rate": 1.2439489665755566e-06,
      "loss": 0.4103,
      "step": 541
    },
    {
      "epoch": 0.14036450746172024,
      "grad_norm": 3.159244262365573,
      "learning_rate": 1.243870255703872e-06,
      "loss": 0.2871,
      "step": 542
    },
    {
      "epoch": 0.14062348256773818,
      "grad_norm": 3.3484268018384755,
      "learning_rate": 1.2437910387282715e-06,
      "loss": 0.3617,
      "step": 543
    },
    {
      "epoch": 0.14088245767375612,
      "grad_norm": 8.017493622586152,
      "learning_rate": 1.2437113157135375e-06,
      "loss": 0.344,
      "step": 544
    },
    {
      "epoch": 0.14114143277977403,
      "grad_norm": 8.661634985456647,
      "learning_rate": 1.2436310867248668e-06,
      "loss": 0.4113,
      "step": 545
    },
    {
      "epoch": 0.14114143277977403,
      "eval_PRM Accuracy": 0.8490566037735849,
      "eval_PRM F1": 0.8987341772151899,
      "eval_PRM F1 AUC": 0.8407543216343635,
      "eval_PRM F1 AUC (fixed)": 0.6678889470927187,
      "eval_PRM F1 Neg": 0.7037037037037037,
      "eval_PRM NPV": 0.6129032258064516,
      "eval_PRM Precision": 0.9466666666666667,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.8260869565217391,
      "eval_loss": 0.31337836384773254,
      "eval_runtime": 14.6079,
      "eval_samples_per_second": 3.012,
      "eval_steps_per_second": 0.137,
      "step": 545
    },
    {
      "epoch": 0.14140040788579197,
      "grad_norm": 4.4569376746391445,
      "learning_rate": 1.2435503518278695e-06,
      "loss": 0.3847,
      "step": 546
    },
    {
      "epoch": 0.1416593829918099,
      "grad_norm": 3.5855721585468143,
      "learning_rate": 1.2434691110885695e-06,
      "loss": 0.3407,
      "step": 547
    },
    {
      "epoch": 0.14191835809782785,
      "grad_norm": 6.692845918255269,
      "learning_rate": 1.2433873645734042e-06,
      "loss": 0.5017,
      "step": 548
    },
    {
      "epoch": 0.1421773332038458,
      "grad_norm": 4.714823671241315,
      "learning_rate": 1.243305112349225e-06,
      "loss": 0.377,
      "step": 549
    },
    {
      "epoch": 0.1424363083098637,
      "grad_norm": 10.08925546130851,
      "learning_rate": 1.2432223544832969e-06,
      "loss": 0.4127,
      "step": 550
    },
    {
      "epoch": 0.1424363083098637,
      "eval_PRM Accuracy": 0.8490566037735849,
      "eval_PRM F1": 0.9,
      "eval_PRM F1 AUC": 0.8250392875851232,
      "eval_PRM F1 AUC (fixed)": 0.6571503404924044,
      "eval_PRM F1 Neg": 0.6923076923076923,
      "eval_PRM NPV": 0.6206896551724138,
      "eval_PRM Precision": 0.935064935064935,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.30864760279655457,
      "eval_runtime": 14.6728,
      "eval_samples_per_second": 2.999,
      "eval_steps_per_second": 0.136,
      "step": 550
    },
    {
      "epoch": 0.14269528341588164,
      "grad_norm": 9.255856783180022,
      "learning_rate": 1.2431390910432978e-06,
      "loss": 0.4187,
      "step": 551
    },
    {
      "epoch": 0.14295425852189958,
      "grad_norm": 3.203425763387509,
      "learning_rate": 1.2430553220973196e-06,
      "loss": 0.344,
      "step": 552
    },
    {
      "epoch": 0.14321323362791752,
      "grad_norm": 7.163800749072193,
      "learning_rate": 1.2429710477138676e-06,
      "loss": 0.3816,
      "step": 553
    },
    {
      "epoch": 0.14347220873393546,
      "grad_norm": 3.4812956449664845,
      "learning_rate": 1.24288626796186e-06,
      "loss": 0.3907,
      "step": 554
    },
    {
      "epoch": 0.1437311838399534,
      "grad_norm": 7.383612072825209,
      "learning_rate": 1.2428009829106288e-06,
      "loss": 0.438,
      "step": 555
    },
    {
      "epoch": 0.1437311838399534,
      "eval_PRM Accuracy": 0.8490566037735849,
      "eval_PRM F1": 0.8974358974358975,
      "eval_PRM F1 AUC": 0.8564693556836039,
      "eval_PRM F1 AUC (fixed)": 0.6568884232582504,
      "eval_PRM F1 Neg": 0.7142857142857143,
      "eval_PRM NPV": 0.6060606060606061,
      "eval_PRM Precision": 0.958904109589041,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.8695652173913043,
      "eval_loss": 0.3244212865829468,
      "eval_runtime": 14.2781,
      "eval_samples_per_second": 3.082,
      "eval_steps_per_second": 0.14,
      "step": 555
    },
    {
      "epoch": 0.1439901589459713,
      "grad_norm": 5.631807558276998,
      "learning_rate": 1.2427151926299187e-06,
      "loss": 0.3592,
      "step": 556
    },
    {
      "epoch": 0.14424913405198925,
      "grad_norm": 5.0220728120088145,
      "learning_rate": 1.2426288971898881e-06,
      "loss": 0.4078,
      "step": 557
    },
    {
      "epoch": 0.1445081091580072,
      "grad_norm": 9.058383349897746,
      "learning_rate": 1.2425420966611081e-06,
      "loss": 0.3206,
      "step": 558
    },
    {
      "epoch": 0.14476708426402513,
      "grad_norm": 7.986430368212883,
      "learning_rate": 1.2424547911145635e-06,
      "loss": 0.3563,
      "step": 559
    },
    {
      "epoch": 0.14502605937004306,
      "grad_norm": 3.5609507470899473,
      "learning_rate": 1.2423669806216509e-06,
      "loss": 0.4017,
      "step": 560
    },
    {
      "epoch": 0.14502605937004306,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.89171974522293,
      "eval_PRM F1 AUC": 0.8347302252488213,
      "eval_PRM F1 AUC (fixed)": 0.6705081194342588,
      "eval_PRM F1 Neg": 0.6909090909090909,
      "eval_PRM NPV": 0.59375,
      "eval_PRM Precision": 0.9459459459459459,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.8260869565217391,
      "eval_loss": 0.31759682297706604,
      "eval_runtime": 14.5424,
      "eval_samples_per_second": 3.026,
      "eval_steps_per_second": 0.138,
      "step": 560
    },
    {
      "epoch": 0.14528503447606098,
      "grad_norm": 9.331894617265462,
      "learning_rate": 1.242278665254181e-06,
      "loss": 0.3495,
      "step": 561
    },
    {
      "epoch": 0.14554400958207891,
      "grad_norm": 4.2029122364941855,
      "learning_rate": 1.2421898450843768e-06,
      "loss": 0.4254,
      "step": 562
    },
    {
      "epoch": 0.14580298468809685,
      "grad_norm": 3.7627145658128334,
      "learning_rate": 1.2421005201848744e-06,
      "loss": 0.3512,
      "step": 563
    },
    {
      "epoch": 0.1460619597941148,
      "grad_norm": 7.780938416484376,
      "learning_rate": 1.2420106906287224e-06,
      "loss": 0.3139,
      "step": 564
    },
    {
      "epoch": 0.14632093490013273,
      "grad_norm": 5.945589615268719,
      "learning_rate": 1.2419203564893822e-06,
      "loss": 0.4031,
      "step": 565
    },
    {
      "epoch": 0.14632093490013273,
      "eval_PRM Accuracy": 0.8584905660377359,
      "eval_PRM F1": 0.9079754601226994,
      "eval_PRM F1 AUC": 0.8153483499214248,
      "eval_PRM F1 AUC (fixed)": 0.6846516500785752,
      "eval_PRM F1 Neg": 0.6938775510204082,
      "eval_PRM NPV": 0.6538461538461539,
      "eval_PRM Precision": 0.925,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.300191730260849,
      "eval_runtime": 14.3339,
      "eval_samples_per_second": 3.07,
      "eval_steps_per_second": 0.14,
      "step": 565
    },
    {
      "epoch": 0.14657991000615067,
      "grad_norm": 6.2263899172585955,
      "learning_rate": 1.241829517840728e-06,
      "loss": 0.3507,
      "step": 566
    },
    {
      "epoch": 0.14683888511216858,
      "grad_norm": 6.387650973352577,
      "learning_rate": 1.2417381747570464e-06,
      "loss": 0.3567,
      "step": 567
    },
    {
      "epoch": 0.14709786021818652,
      "grad_norm": 6.671960513858668,
      "learning_rate": 1.2416463273130364e-06,
      "loss": 0.3629,
      "step": 568
    },
    {
      "epoch": 0.14735683532420446,
      "grad_norm": 7.090385235466389,
      "learning_rate": 1.24155397558381e-06,
      "loss": 0.4472,
      "step": 569
    },
    {
      "epoch": 0.1476158104302224,
      "grad_norm": 5.756477153036491,
      "learning_rate": 1.2414611196448906e-06,
      "loss": 0.4065,
      "step": 570
    },
    {
      "epoch": 0.1476158104302224,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.89171974522293,
      "eval_PRM F1 AUC": 0.8347302252488213,
      "eval_PRM F1 AUC (fixed)": 0.680460974332111,
      "eval_PRM F1 Neg": 0.6909090909090909,
      "eval_PRM NPV": 0.59375,
      "eval_PRM Precision": 0.9459459459459459,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.8260869565217391,
      "eval_loss": 0.3101547360420227,
      "eval_runtime": 14.3796,
      "eval_samples_per_second": 3.06,
      "eval_steps_per_second": 0.139,
      "step": 570
    },
    {
      "epoch": 0.14787478553624034,
      "grad_norm": 9.003905398536311,
      "learning_rate": 1.2413677595722154e-06,
      "loss": 0.3899,
      "step": 571
    },
    {
      "epoch": 0.14813376064225825,
      "grad_norm": 4.523238842150813,
      "learning_rate": 1.241273895442132e-06,
      "loss": 0.3687,
      "step": 572
    },
    {
      "epoch": 0.1483927357482762,
      "grad_norm": 4.1866996609752904,
      "learning_rate": 1.2411795273314025e-06,
      "loss": 0.33,
      "step": 573
    },
    {
      "epoch": 0.14865171085429413,
      "grad_norm": 8.369351825162653,
      "learning_rate": 1.2410846553171988e-06,
      "loss": 0.4078,
      "step": 574
    },
    {
      "epoch": 0.14891068596031207,
      "grad_norm": 9.110580389178104,
      "learning_rate": 1.2409892794771066e-06,
      "loss": 0.4616,
      "step": 575
    },
    {
      "epoch": 0.14891068596031207,
      "eval_PRM Accuracy": 0.8301886792452831,
      "eval_PRM F1": 0.8846153846153846,
      "eval_PRM F1 AUC": 0.8287061288632793,
      "eval_PRM F1 AUC (fixed)": 0.6781037192247249,
      "eval_PRM F1 Neg": 0.6785714285714286,
      "eval_PRM NPV": 0.5757575757575758,
      "eval_PRM Precision": 0.9452054794520548,
      "eval_PRM Recall": 0.8313253012048193,
      "eval_PRM Specificty": 0.8260869565217391,
      "eval_loss": 0.3210476040840149,
      "eval_runtime": 14.8749,
      "eval_samples_per_second": 2.958,
      "eval_steps_per_second": 0.134,
      "step": 575
    },
    {
      "epoch": 0.14916966106633,
      "grad_norm": 10.564572190452354,
      "learning_rate": 1.2408933998891226e-06,
      "loss": 0.3938,
      "step": 576
    },
    {
      "epoch": 0.14942863617234795,
      "grad_norm": 15.194471093181686,
      "learning_rate": 1.2407970166316562e-06,
      "loss": 0.4729,
      "step": 577
    },
    {
      "epoch": 0.14968761127836586,
      "grad_norm": 9.4388656054538,
      "learning_rate": 1.2407001297835281e-06,
      "loss": 0.4167,
      "step": 578
    },
    {
      "epoch": 0.1499465863843838,
      "grad_norm": 3.577291285694216,
      "learning_rate": 1.2406027394239712e-06,
      "loss": 0.3604,
      "step": 579
    },
    {
      "epoch": 0.15020556149040173,
      "grad_norm": 6.76903188686745,
      "learning_rate": 1.2405048456326302e-06,
      "loss": 0.3794,
      "step": 580
    },
    {
      "epoch": 0.15020556149040173,
      "eval_PRM Accuracy": 0.8490566037735849,
      "eval_PRM F1": 0.9012345679012346,
      "eval_PRM F1 AUC": 0.8093242535358826,
      "eval_PRM F1 AUC (fixed)": 0.6901519119958093,
      "eval_PRM F1 Neg": 0.68,
      "eval_PRM NPV": 0.6296296296296297,
      "eval_PRM Precision": 0.9240506329113924,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.30129262804985046,
      "eval_runtime": 14.3739,
      "eval_samples_per_second": 3.061,
      "eval_steps_per_second": 0.139,
      "step": 580
    },
    {
      "epoch": 0.15046453659641967,
      "grad_norm": 7.344346715445107,
      "learning_rate": 1.240406448489561e-06,
      "loss": 0.4737,
      "step": 581
    },
    {
      "epoch": 0.1507235117024376,
      "grad_norm": 7.796548314532856,
      "learning_rate": 1.2403075480752318e-06,
      "loss": 0.3963,
      "step": 582
    },
    {
      "epoch": 0.15098248680845552,
      "grad_norm": 4.3140902460127695,
      "learning_rate": 1.2402081444705218e-06,
      "loss": 0.4505,
      "step": 583
    },
    {
      "epoch": 0.15124146191447346,
      "grad_norm": 10.426328309205195,
      "learning_rate": 1.2401082377567222e-06,
      "loss": 0.4682,
      "step": 584
    },
    {
      "epoch": 0.1515004370204914,
      "grad_norm": 3.1885828757437666,
      "learning_rate": 1.2400078280155352e-06,
      "loss": 0.2957,
      "step": 585
    },
    {
      "epoch": 0.1515004370204914,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.8930817610062893,
      "eval_PRM F1 AUC": 0.819015191199581,
      "eval_PRM F1 AUC (fixed)": 0.6809848088004191,
      "eval_PRM F1 Neg": 0.6792452830188679,
      "eval_PRM NPV": 0.6,
      "eval_PRM Precision": 0.9342105263157895,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.31476715207099915,
      "eval_runtime": 14.6253,
      "eval_samples_per_second": 3.008,
      "eval_steps_per_second": 0.137,
      "step": 585
    },
    {
      "epoch": 0.15175941212650934,
      "grad_norm": 3.149213053009891,
      "learning_rate": 1.2399069153290748e-06,
      "loss": 0.2859,
      "step": 586
    },
    {
      "epoch": 0.15201838723252728,
      "grad_norm": 4.484121045797467,
      "learning_rate": 1.2398054997798657e-06,
      "loss": 0.3807,
      "step": 587
    },
    {
      "epoch": 0.15227736233854522,
      "grad_norm": 4.509486231542979,
      "learning_rate": 1.2397035814508447e-06,
      "loss": 0.4663,
      "step": 588
    },
    {
      "epoch": 0.15253633744456313,
      "grad_norm": 3.8777329311615483,
      "learning_rate": 1.2396011604253585e-06,
      "loss": 0.3775,
      "step": 589
    },
    {
      "epoch": 0.15279531255058107,
      "grad_norm": 7.543389414522049,
      "learning_rate": 1.2394982367871663e-06,
      "loss": 0.2858,
      "step": 590
    },
    {
      "epoch": 0.15279531255058107,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.8903225806451613,
      "eval_PRM F1 AUC": 0.8504452592980618,
      "eval_PRM F1 AUC (fixed)": 0.6733892090099528,
      "eval_PRM F1 Neg": 0.7017543859649122,
      "eval_PRM NPV": 0.5882352941176471,
      "eval_PRM Precision": 0.9583333333333334,
      "eval_PRM Recall": 0.8313253012048193,
      "eval_PRM Specificty": 0.8695652173913043,
      "eval_loss": 0.33046090602874756,
      "eval_runtime": 14.6589,
      "eval_samples_per_second": 3.002,
      "eval_steps_per_second": 0.136,
      "step": 590
    },
    {
      "epoch": 0.153054287656599,
      "grad_norm": 5.42693700134366,
      "learning_rate": 1.2393948106204376e-06,
      "loss": 0.3109,
      "step": 591
    },
    {
      "epoch": 0.15331326276261695,
      "grad_norm": 7.0826540601798404,
      "learning_rate": 1.2392908820097526e-06,
      "loss": 0.434,
      "step": 592
    },
    {
      "epoch": 0.1535722378686349,
      "grad_norm": 4.604182378707424,
      "learning_rate": 1.2391864510401034e-06,
      "loss": 0.3272,
      "step": 593
    },
    {
      "epoch": 0.1538312129746528,
      "grad_norm": 5.012793342377323,
      "learning_rate": 1.2390815177968915e-06,
      "loss": 0.3389,
      "step": 594
    },
    {
      "epoch": 0.15409018808067074,
      "grad_norm": 4.471973602793316,
      "learning_rate": 1.2389760823659305e-06,
      "loss": 0.3548,
      "step": 595
    },
    {
      "epoch": 0.15409018808067074,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.8930817610062893,
      "eval_PRM F1 AUC": 0.819015191199581,
      "eval_PRM F1 AUC (fixed)": 0.6838658983761131,
      "eval_PRM F1 Neg": 0.6792452830188679,
      "eval_PRM NPV": 0.6,
      "eval_PRM Precision": 0.9342105263157895,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.31398123502731323,
      "eval_runtime": 14.9235,
      "eval_samples_per_second": 2.948,
      "eval_steps_per_second": 0.134,
      "step": 595
    },
    {
      "epoch": 0.15434916318668868,
      "grad_norm": 5.292384586964645,
      "learning_rate": 1.2388701448334438e-06,
      "loss": 0.3323,
      "step": 596
    },
    {
      "epoch": 0.15460813829270662,
      "grad_norm": 7.288360954102471,
      "learning_rate": 1.2387637052860657e-06,
      "loss": 0.4074,
      "step": 597
    },
    {
      "epoch": 0.15486711339872455,
      "grad_norm": 4.833374417379775,
      "learning_rate": 1.2386567638108412e-06,
      "loss": 0.3843,
      "step": 598
    },
    {
      "epoch": 0.1551260885047425,
      "grad_norm": 10.124157768027903,
      "learning_rate": 1.238549320495226e-06,
      "loss": 0.4332,
      "step": 599
    },
    {
      "epoch": 0.1553850636107604,
      "grad_norm": 8.37696636780662,
      "learning_rate": 1.238441375427085e-06,
      "loss": 0.38,
      "step": 600
    },
    {
      "epoch": 0.1553850636107604,
      "eval_PRM Accuracy": 0.8490566037735849,
      "eval_PRM F1": 0.9012345679012346,
      "eval_PRM F1 AUC": 0.8093242535358826,
      "eval_PRM F1 AUC (fixed)": 0.6849135673127291,
      "eval_PRM F1 Neg": 0.68,
      "eval_PRM NPV": 0.6296296296296297,
      "eval_PRM Precision": 0.9240506329113924,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.3145059645175934,
      "eval_runtime": 14.6733,
      "eval_samples_per_second": 2.999,
      "eval_steps_per_second": 0.136,
      "step": 600
    },
    {
      "epoch": 0.15564403871677834,
      "grad_norm": 10.573652158959245,
      "learning_rate": 1.2383329286946946e-06,
      "loss": 0.3895,
      "step": 601
    },
    {
      "epoch": 0.15590301382279628,
      "grad_norm": 6.215297625313593,
      "learning_rate": 1.2382239803867413e-06,
      "loss": 0.3188,
      "step": 602
    },
    {
      "epoch": 0.15616198892881422,
      "grad_norm": 3.8865032915789315,
      "learning_rate": 1.2381145305923215e-06,
      "loss": 0.3603,
      "step": 603
    },
    {
      "epoch": 0.15642096403483216,
      "grad_norm": 4.305704623668269,
      "learning_rate": 1.2380045794009416e-06,
      "loss": 0.415,
      "step": 604
    },
    {
      "epoch": 0.1566799391408501,
      "grad_norm": 7.777263912530598,
      "learning_rate": 1.2378941269025184e-06,
      "loss": 0.3643,
      "step": 605
    },
    {
      "epoch": 0.1566799391408501,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.8903225806451613,
      "eval_PRM F1 AUC": 0.8504452592980618,
      "eval_PRM F1 AUC (fixed)": 0.6843897328444212,
      "eval_PRM F1 Neg": 0.7017543859649122,
      "eval_PRM NPV": 0.5882352941176471,
      "eval_PRM Precision": 0.9583333333333334,
      "eval_PRM Recall": 0.8313253012048193,
      "eval_PRM Specificty": 0.8695652173913043,
      "eval_loss": 0.33305254578590393,
      "eval_runtime": 14.4625,
      "eval_samples_per_second": 3.042,
      "eval_steps_per_second": 0.138,
      "step": 605
    },
    {
      "epoch": 0.156938914246868,
      "grad_norm": 13.549319660584498,
      "learning_rate": 1.2377831731873786e-06,
      "loss": 0.3611,
      "step": 606
    },
    {
      "epoch": 0.15719788935288595,
      "grad_norm": 5.115617377808031,
      "learning_rate": 1.2376717183462583e-06,
      "loss": 0.4097,
      "step": 607
    },
    {
      "epoch": 0.1574568644589039,
      "grad_norm": 8.064002200115613,
      "learning_rate": 1.2375597624703042e-06,
      "loss": 0.3544,
      "step": 608
    },
    {
      "epoch": 0.15771583956492183,
      "grad_norm": 8.743740116901071,
      "learning_rate": 1.2374473056510723e-06,
      "loss": 0.3979,
      "step": 609
    },
    {
      "epoch": 0.15797481467093977,
      "grad_norm": 9.05779305197767,
      "learning_rate": 1.237334347980528e-06,
      "loss": 0.3519,
      "step": 610
    },
    {
      "epoch": 0.15797481467093977,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.89171974522293,
      "eval_PRM F1 AUC": 0.8347302252488213,
      "eval_PRM F1 AUC (fixed)": 0.6898899947616554,
      "eval_PRM F1 Neg": 0.6909090909090909,
      "eval_PRM NPV": 0.59375,
      "eval_PRM Precision": 0.9459459459459459,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.8260869565217391,
      "eval_loss": 0.30802619457244873,
      "eval_runtime": 15.0217,
      "eval_samples_per_second": 2.929,
      "eval_steps_per_second": 0.133,
      "step": 610
    },
    {
      "epoch": 0.15823378977695768,
      "grad_norm": 5.861247610988427,
      "learning_rate": 1.2372208895510469e-06,
      "loss": 0.3434,
      "step": 611
    },
    {
      "epoch": 0.15849276488297562,
      "grad_norm": 3.670151363461034,
      "learning_rate": 1.2371069304554133e-06,
      "loss": 0.3702,
      "step": 612
    },
    {
      "epoch": 0.15875173998899356,
      "grad_norm": 8.084637417212509,
      "learning_rate": 1.2369924707868222e-06,
      "loss": 0.4298,
      "step": 613
    },
    {
      "epoch": 0.1590107150950115,
      "grad_norm": 5.687659548763739,
      "learning_rate": 1.236877510638877e-06,
      "loss": 0.3255,
      "step": 614
    },
    {
      "epoch": 0.15926969020102943,
      "grad_norm": 7.056523326388455,
      "learning_rate": 1.23676205010559e-06,
      "loss": 0.3883,
      "step": 615
    },
    {
      "epoch": 0.15926969020102943,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.8944099378881988,
      "eval_PRM F1 AUC": 0.8033001571503404,
      "eval_PRM F1 AUC (fixed)": 0.6964379256155055,
      "eval_PRM F1 Neg": 0.6666666666666666,
      "eval_PRM NPV": 0.6071428571428571,
      "eval_PRM Precision": 0.9230769230769231,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.2987489402294159,
      "eval_runtime": 14.7965,
      "eval_samples_per_second": 2.974,
      "eval_steps_per_second": 0.135,
      "step": 615
    },
    {
      "epoch": 0.15952866530704737,
      "grad_norm": 8.052724316864127,
      "learning_rate": 1.2366460892813839e-06,
      "loss": 0.4153,
      "step": 616
    },
    {
      "epoch": 0.15978764041306528,
      "grad_norm": 11.44407409138463,
      "learning_rate": 1.2365296282610897e-06,
      "loss": 0.4042,
      "step": 617
    },
    {
      "epoch": 0.16004661551908322,
      "grad_norm": 14.598720324528122,
      "learning_rate": 1.2364126671399477e-06,
      "loss": 0.4369,
      "step": 618
    },
    {
      "epoch": 0.16030559062510116,
      "grad_norm": 5.482057017022813,
      "learning_rate": 1.236295206013607e-06,
      "loss": 0.3319,
      "step": 619
    },
    {
      "epoch": 0.1605645657311191,
      "grad_norm": 3.3221121701486727,
      "learning_rate": 1.2361772449781263e-06,
      "loss": 0.3924,
      "step": 620
    },
    {
      "epoch": 0.1605645657311191,
      "eval_PRM Accuracy": 0.8301886792452831,
      "eval_PRM F1": 0.8846153846153846,
      "eval_PRM F1 AUC": 0.8287061288632793,
      "eval_PRM F1 AUC (fixed)": 0.6783656364588789,
      "eval_PRM F1 Neg": 0.6785714285714286,
      "eval_PRM NPV": 0.5757575757575758,
      "eval_PRM Precision": 0.9452054794520548,
      "eval_PRM Recall": 0.8313253012048193,
      "eval_PRM Specificty": 0.8260869565217391,
      "eval_loss": 0.3130224049091339,
      "eval_runtime": 14.3415,
      "eval_samples_per_second": 3.068,
      "eval_steps_per_second": 0.139,
      "step": 620
    },
    {
      "epoch": 0.16082354083713704,
      "grad_norm": 3.3351070027555,
      "learning_rate": 1.236058784129972e-06,
      "loss": 0.3507,
      "step": 621
    },
    {
      "epoch": 0.16108251594315495,
      "grad_norm": 5.057842166618868,
      "learning_rate": 1.23593982356602e-06,
      "loss": 0.344,
      "step": 622
    },
    {
      "epoch": 0.1613414910491729,
      "grad_norm": 9.181148751364352,
      "learning_rate": 1.2358203633835549e-06,
      "loss": 0.4648,
      "step": 623
    },
    {
      "epoch": 0.16160046615519083,
      "grad_norm": 6.083979457492019,
      "learning_rate": 1.2357004036802695e-06,
      "loss": 0.2599,
      "step": 624
    },
    {
      "epoch": 0.16185944126120877,
      "grad_norm": 4.382520565652978,
      "learning_rate": 1.2355799445542649e-06,
      "loss": 0.3058,
      "step": 625
    },
    {
      "epoch": 0.16185944126120877,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.8903225806451613,
      "eval_PRM F1 AUC": 0.8504452592980618,
      "eval_PRM F1 AUC (fixed)": 0.6846516500785752,
      "eval_PRM F1 Neg": 0.7017543859649122,
      "eval_PRM NPV": 0.5882352941176471,
      "eval_PRM Precision": 0.9583333333333334,
      "eval_PRM Recall": 0.8313253012048193,
      "eval_PRM Specificty": 0.8695652173913043,
      "eval_loss": 0.3220345973968506,
      "eval_runtime": 14.5696,
      "eval_samples_per_second": 3.02,
      "eval_steps_per_second": 0.137,
      "step": 625
    },
    {
      "epoch": 0.1621184163672267,
      "grad_norm": 4.1437604506815955,
      "learning_rate": 1.2354589861040514e-06,
      "loss": 0.3002,
      "step": 626
    },
    {
      "epoch": 0.16237739147324465,
      "grad_norm": 4.374843106284846,
      "learning_rate": 1.2353375284285475e-06,
      "loss": 0.3704,
      "step": 627
    },
    {
      "epoch": 0.16263636657926256,
      "grad_norm": 5.608222695272279,
      "learning_rate": 1.235215571627079e-06,
      "loss": 0.344,
      "step": 628
    },
    {
      "epoch": 0.1628953416852805,
      "grad_norm": 4.8582080854128895,
      "learning_rate": 1.2350931157993807e-06,
      "loss": 0.3336,
      "step": 629
    },
    {
      "epoch": 0.16315431679129844,
      "grad_norm": 12.482439714954587,
      "learning_rate": 1.2349701610455959e-06,
      "loss": 0.5153,
      "step": 630
    },
    {
      "epoch": 0.16315431679129844,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.8734177215189873,
      "eval_PRM F1 AUC": 0.785227867993714,
      "eval_PRM F1 AUC (fixed)": 0.6856993190151912,
      "eval_PRM F1 Neg": 0.6296296296296297,
      "eval_PRM NPV": 0.5483870967741935,
      "eval_PRM Precision": 0.92,
      "eval_PRM Recall": 0.8313253012048193,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.31282979249954224,
      "eval_runtime": 14.476,
      "eval_samples_per_second": 3.04,
      "eval_steps_per_second": 0.138,
      "step": 630
    },
    {
      "epoch": 0.16341329189731638,
      "grad_norm": 2.8269173986313767,
      "learning_rate": 1.2348467074662746e-06,
      "loss": 0.3002,
      "step": 631
    },
    {
      "epoch": 0.16367226700333432,
      "grad_norm": 9.367758085039979,
      "learning_rate": 1.234722755162376e-06,
      "loss": 0.4783,
      "step": 632
    },
    {
      "epoch": 0.16393124210935223,
      "grad_norm": 4.360443477858389,
      "learning_rate": 1.2345983042352668e-06,
      "loss": 0.3304,
      "step": 633
    },
    {
      "epoch": 0.16419021721537017,
      "grad_norm": 4.814617876325908,
      "learning_rate": 1.2344733547867211e-06,
      "loss": 0.3975,
      "step": 634
    },
    {
      "epoch": 0.1644491923213881,
      "grad_norm": 8.736415057185651,
      "learning_rate": 1.2343479069189208e-06,
      "loss": 0.3967,
      "step": 635
    },
    {
      "epoch": 0.1644491923213881,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8789808917197452,
      "eval_PRM F1 AUC": 0.8069669984284967,
      "eval_PRM F1 AUC (fixed)": 0.6841278156102671,
      "eval_PRM F1 Neg": 0.6545454545454545,
      "eval_PRM NPV": 0.5625,
      "eval_PRM Precision": 0.9324324324324325,
      "eval_PRM Recall": 0.8313253012048193,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.32189351320266724,
      "eval_runtime": 14.5716,
      "eval_samples_per_second": 3.02,
      "eval_steps_per_second": 0.137,
      "step": 635
    },
    {
      "epoch": 0.16470816742740604,
      "grad_norm": 8.16922264657535,
      "learning_rate": 1.2342219607344553e-06,
      "loss": 0.3905,
      "step": 636
    },
    {
      "epoch": 0.16496714253342398,
      "grad_norm": 4.102239363110395,
      "learning_rate": 1.2340955163363223e-06,
      "loss": 0.3864,
      "step": 637
    },
    {
      "epoch": 0.16522611763944192,
      "grad_norm": 6.666667743598508,
      "learning_rate": 1.2339685738279262e-06,
      "loss": 0.3439,
      "step": 638
    },
    {
      "epoch": 0.16548509274545983,
      "grad_norm": 6.359302938917453,
      "learning_rate": 1.2338411333130786e-06,
      "loss": 0.3799,
      "step": 639
    },
    {
      "epoch": 0.16574406785147777,
      "grad_norm": 3.4027973170905943,
      "learning_rate": 1.2337131948959988e-06,
      "loss": 0.3117,
      "step": 640
    },
    {
      "epoch": 0.16574406785147777,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.8734177215189873,
      "eval_PRM F1 AUC": 0.785227867993714,
      "eval_PRM F1 AUC (fixed)": 0.6901519119958093,
      "eval_PRM F1 Neg": 0.6296296296296297,
      "eval_PRM NPV": 0.5483870967741935,
      "eval_PRM Precision": 0.92,
      "eval_PRM Recall": 0.8313253012048193,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.3099837303161621,
      "eval_runtime": 14.876,
      "eval_samples_per_second": 2.958,
      "eval_steps_per_second": 0.134,
      "step": 640
    },
    {
      "epoch": 0.1660030429574957,
      "grad_norm": 4.065907555813363,
      "learning_rate": 1.2335847586813134e-06,
      "loss": 0.359,
      "step": 641
    },
    {
      "epoch": 0.16626201806351365,
      "grad_norm": 9.506205075827122,
      "learning_rate": 1.2334558247740555e-06,
      "loss": 0.3711,
      "step": 642
    },
    {
      "epoch": 0.1665209931695316,
      "grad_norm": 5.585007233460523,
      "learning_rate": 1.2333263932796658e-06,
      "loss": 0.3794,
      "step": 643
    },
    {
      "epoch": 0.1667799682755495,
      "grad_norm": 6.2824853591678895,
      "learning_rate": 1.2331964643039914e-06,
      "loss": 0.3819,
      "step": 644
    },
    {
      "epoch": 0.16703894338156744,
      "grad_norm": 3.0945268049557955,
      "learning_rate": 1.233066037953287e-06,
      "loss": 0.2563,
      "step": 645
    },
    {
      "epoch": 0.16703894338156744,
      "eval_PRM Accuracy": 0.8301886792452831,
      "eval_PRM F1": 0.8860759493670886,
      "eval_PRM F1 AUC": 0.8129910948140389,
      "eval_PRM F1 AUC (fixed)": 0.7008905185961236,
      "eval_PRM F1 Neg": 0.6666666666666666,
      "eval_PRM NPV": 0.5806451612903226,
      "eval_PRM Precision": 0.9333333333333333,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.2902960777282715,
      "eval_runtime": 14.3898,
      "eval_samples_per_second": 3.058,
      "eval_steps_per_second": 0.139,
      "step": 645
    },
    {
      "epoch": 0.16729791848758538,
      "grad_norm": 8.26070229351453,
      "learning_rate": 1.2329351143342127e-06,
      "loss": 0.4367,
      "step": 646
    },
    {
      "epoch": 0.16755689359360332,
      "grad_norm": 3.5692756270657853,
      "learning_rate": 1.232803693553837e-06,
      "loss": 0.3088,
      "step": 647
    },
    {
      "epoch": 0.16781586869962126,
      "grad_norm": 6.322229970930503,
      "learning_rate": 1.2326717757196335e-06,
      "loss": 0.3815,
      "step": 648
    },
    {
      "epoch": 0.1680748438056392,
      "grad_norm": 5.4301781501602955,
      "learning_rate": 1.232539360939483e-06,
      "loss": 0.4233,
      "step": 649
    },
    {
      "epoch": 0.1683338189116571,
      "grad_norm": 4.037554188342987,
      "learning_rate": 1.2324064493216729e-06,
      "loss": 0.4186,
      "step": 650
    },
    {
      "epoch": 0.1683338189116571,
      "eval_PRM Accuracy": 0.8301886792452831,
      "eval_PRM F1": 0.8846153846153846,
      "eval_PRM F1 AUC": 0.8287061288632793,
      "eval_PRM F1 AUC (fixed)": 0.7037716081718176,
      "eval_PRM F1 Neg": 0.6785714285714286,
      "eval_PRM NPV": 0.5757575757575758,
      "eval_PRM Precision": 0.9452054794520548,
      "eval_PRM Recall": 0.8313253012048193,
      "eval_PRM Specificty": 0.8260869565217391,
      "eval_loss": 0.3051375150680542,
      "eval_runtime": 14.6736,
      "eval_samples_per_second": 2.999,
      "eval_steps_per_second": 0.136,
      "step": 650
    },
    {
      "epoch": 0.16859279401767505,
      "grad_norm": 5.903651242662739,
      "learning_rate": 1.2322730409748958e-06,
      "loss": 0.4349,
      "step": 651
    },
    {
      "epoch": 0.16885176912369299,
      "grad_norm": 4.79677556638012,
      "learning_rate": 1.2321391360082521e-06,
      "loss": 0.3621,
      "step": 652
    },
    {
      "epoch": 0.16911074422971092,
      "grad_norm": 5.0682489678833536,
      "learning_rate": 1.2320047345312468e-06,
      "loss": 0.4286,
      "step": 653
    },
    {
      "epoch": 0.16936971933572886,
      "grad_norm": 10.204257727871076,
      "learning_rate": 1.2318698366537924e-06,
      "loss": 0.418,
      "step": 654
    },
    {
      "epoch": 0.16962869444174677,
      "grad_norm": 5.9807619771124845,
      "learning_rate": 1.2317344424862062e-06,
      "loss": 0.4272,
      "step": 655
    },
    {
      "epoch": 0.16962869444174677,
      "eval_PRM Accuracy": 0.8301886792452831,
      "eval_PRM F1": 0.8846153846153846,
      "eval_PRM F1 AUC": 0.8287061288632793,
      "eval_PRM F1 AUC (fixed)": 0.7003666841278156,
      "eval_PRM F1 Neg": 0.6785714285714286,
      "eval_PRM NPV": 0.5757575757575758,
      "eval_PRM Precision": 0.9452054794520548,
      "eval_PRM Recall": 0.8313253012048193,
      "eval_PRM Specificty": 0.8260869565217391,
      "eval_loss": 0.3076668977737427,
      "eval_runtime": 14.5657,
      "eval_samples_per_second": 3.021,
      "eval_steps_per_second": 0.137,
      "step": 655
    },
    {
      "epoch": 0.1698876695477647,
      "grad_norm": 8.183477991339371,
      "learning_rate": 1.2315985521392116e-06,
      "loss": 0.3848,
      "step": 656
    },
    {
      "epoch": 0.17014664465378265,
      "grad_norm": 4.350350964575722,
      "learning_rate": 1.2314621657239382e-06,
      "loss": 0.4005,
      "step": 657
    },
    {
      "epoch": 0.1704056197598006,
      "grad_norm": 3.8484281205534305,
      "learning_rate": 1.2313252833519207e-06,
      "loss": 0.3491,
      "step": 658
    },
    {
      "epoch": 0.17066459486581853,
      "grad_norm": 3.8215249184870355,
      "learning_rate": 1.2311879051351e-06,
      "loss": 0.365,
      "step": 659
    },
    {
      "epoch": 0.17092356997183647,
      "grad_norm": 3.8839060530983347,
      "learning_rate": 1.2310500311858224e-06,
      "loss": 0.3976,
      "step": 660
    },
    {
      "epoch": 0.17092356997183647,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.89171974522293,
      "eval_PRM F1 AUC": 0.8347302252488213,
      "eval_PRM F1 AUC (fixed)": 0.6909376636982713,
      "eval_PRM F1 Neg": 0.6909090909090909,
      "eval_PRM NPV": 0.59375,
      "eval_PRM Precision": 0.9459459459459459,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.8260869565217391,
      "eval_loss": 0.2925107777118683,
      "eval_runtime": 14.7018,
      "eval_samples_per_second": 2.993,
      "eval_steps_per_second": 0.136,
      "step": 660
    },
    {
      "epoch": 0.17118254507785438,
      "grad_norm": 3.2291152137659855,
      "learning_rate": 1.2309116616168388e-06,
      "loss": 0.3187,
      "step": 661
    },
    {
      "epoch": 0.17144152018387232,
      "grad_norm": 3.6725860645126867,
      "learning_rate": 1.2307727965413063e-06,
      "loss": 0.3757,
      "step": 662
    },
    {
      "epoch": 0.17170049528989026,
      "grad_norm": 3.1002808823226986,
      "learning_rate": 1.230633436072787e-06,
      "loss": 0.3074,
      "step": 663
    },
    {
      "epoch": 0.1719594703959082,
      "grad_norm": 8.746795861527746,
      "learning_rate": 1.2304935803252479e-06,
      "loss": 0.4095,
      "step": 664
    },
    {
      "epoch": 0.17221844550192614,
      "grad_norm": 10.82989204467529,
      "learning_rate": 1.2303532294130613e-06,
      "loss": 0.4814,
      "step": 665
    },
    {
      "epoch": 0.17221844550192614,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.8930817610062893,
      "eval_PRM F1 AUC": 0.819015191199581,
      "eval_PRM F1 AUC (fixed)": 0.6875327396542692,
      "eval_PRM F1 Neg": 0.6792452830188679,
      "eval_PRM NPV": 0.6,
      "eval_PRM Precision": 0.9342105263157895,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.30050474405288696,
      "eval_runtime": 14.9203,
      "eval_samples_per_second": 2.949,
      "eval_steps_per_second": 0.134,
      "step": 665
    },
    {
      "epoch": 0.17247742060794405,
      "grad_norm": 8.264683855258928,
      "learning_rate": 1.2302123834510042e-06,
      "loss": 0.4312,
      "step": 666
    },
    {
      "epoch": 0.172736395713962,
      "grad_norm": 4.929731323927467,
      "learning_rate": 1.2300710425542586e-06,
      "loss": 0.3498,
      "step": 667
    },
    {
      "epoch": 0.17299537081997993,
      "grad_norm": 4.665278640952237,
      "learning_rate": 1.2299292068384114e-06,
      "loss": 0.3288,
      "step": 668
    },
    {
      "epoch": 0.17325434592599787,
      "grad_norm": 4.455799796574909,
      "learning_rate": 1.2297868764194538e-06,
      "loss": 0.3569,
      "step": 669
    },
    {
      "epoch": 0.1735133210320158,
      "grad_norm": 3.97792111313062,
      "learning_rate": 1.2296440514137816e-06,
      "loss": 0.3906,
      "step": 670
    },
    {
      "epoch": 0.1735133210320158,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.8903225806451613,
      "eval_PRM F1 AUC": 0.8504452592980618,
      "eval_PRM F1 AUC (fixed)": 0.6836039811419592,
      "eval_PRM F1 Neg": 0.7017543859649122,
      "eval_PRM NPV": 0.5882352941176471,
      "eval_PRM Precision": 0.9583333333333334,
      "eval_PRM Recall": 0.8313253012048193,
      "eval_PRM Specificty": 0.8695652173913043,
      "eval_loss": 0.32599514722824097,
      "eval_runtime": 14.7092,
      "eval_samples_per_second": 2.991,
      "eval_steps_per_second": 0.136,
      "step": 670
    },
    {
      "epoch": 0.17377229613803374,
      "grad_norm": 8.57169595641781,
      "learning_rate": 1.2295007319381956e-06,
      "loss": 0.3743,
      "step": 671
    },
    {
      "epoch": 0.17403127124405166,
      "grad_norm": 9.193160162768763,
      "learning_rate": 1.2293569181099002e-06,
      "loss": 0.5468,
      "step": 672
    },
    {
      "epoch": 0.1742902463500696,
      "grad_norm": 7.97975844919332,
      "learning_rate": 1.2292126100465047e-06,
      "loss": 0.4063,
      "step": 673
    },
    {
      "epoch": 0.17454922145608753,
      "grad_norm": 3.8452977252610747,
      "learning_rate": 1.229067807866022e-06,
      "loss": 0.3421,
      "step": 674
    },
    {
      "epoch": 0.17480819656210547,
      "grad_norm": 10.931204111910318,
      "learning_rate": 1.2289225116868696e-06,
      "loss": 0.3536,
      "step": 675
    },
    {
      "epoch": 0.17480819656210547,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.8734177215189873,
      "eval_PRM F1 AUC": 0.785227867993714,
      "eval_PRM F1 AUC (fixed)": 0.6864850707176532,
      "eval_PRM F1 Neg": 0.6296296296296297,
      "eval_PRM NPV": 0.5483870967741935,
      "eval_PRM Precision": 0.92,
      "eval_PRM Recall": 0.8313253012048193,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.32675302028656006,
      "eval_runtime": 14.7963,
      "eval_samples_per_second": 2.974,
      "eval_steps_per_second": 0.135,
      "step": 675
    },
    {
      "epoch": 0.1750671716681234,
      "grad_norm": 5.2549991101724425,
      "learning_rate": 1.228776721627869e-06,
      "loss": 0.4517,
      "step": 676
    },
    {
      "epoch": 0.17532614677414132,
      "grad_norm": 5.623571928637202,
      "learning_rate": 1.2286304378082454e-06,
      "loss": 0.403,
      "step": 677
    },
    {
      "epoch": 0.17558512188015926,
      "grad_norm": 7.871866515710695,
      "learning_rate": 1.2284836603476272e-06,
      "loss": 0.3853,
      "step": 678
    },
    {
      "epoch": 0.1758440969861772,
      "grad_norm": 3.150820890154312,
      "learning_rate": 1.2283363893660478e-06,
      "loss": 0.3173,
      "step": 679
    },
    {
      "epoch": 0.17610307209219514,
      "grad_norm": 4.865805406292621,
      "learning_rate": 1.2281886249839429e-06,
      "loss": 0.3076,
      "step": 680
    },
    {
      "epoch": 0.17610307209219514,
      "eval_PRM Accuracy": 0.8301886792452831,
      "eval_PRM F1": 0.8860759493670886,
      "eval_PRM F1 AUC": 0.8129910948140389,
      "eval_PRM F1 AUC (fixed)": 0.6946045049764276,
      "eval_PRM F1 Neg": 0.6666666666666666,
      "eval_PRM NPV": 0.5806451612903226,
      "eval_PRM Precision": 0.9333333333333333,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.30507519841194153,
      "eval_runtime": 15.0032,
      "eval_samples_per_second": 2.933,
      "eval_steps_per_second": 0.133,
      "step": 680
    },
    {
      "epoch": 0.17636204719821308,
      "grad_norm": 3.6089302563745327,
      "learning_rate": 1.2280403673221522e-06,
      "loss": 0.3024,
      "step": 681
    },
    {
      "epoch": 0.17662102230423102,
      "grad_norm": 6.5551214656445485,
      "learning_rate": 1.2278916165019195e-06,
      "loss": 0.395,
      "step": 682
    },
    {
      "epoch": 0.17687999741024893,
      "grad_norm": 3.729361087732829,
      "learning_rate": 1.2277423726448905e-06,
      "loss": 0.3875,
      "step": 683
    },
    {
      "epoch": 0.17713897251626687,
      "grad_norm": 5.2636108819488365,
      "learning_rate": 1.2275926358731156e-06,
      "loss": 0.3809,
      "step": 684
    },
    {
      "epoch": 0.1773979476222848,
      "grad_norm": 3.124690927806883,
      "learning_rate": 1.227442406309047e-06,
      "loss": 0.3772,
      "step": 685
    },
    {
      "epoch": 0.1773979476222848,
      "eval_PRM Accuracy": 0.8301886792452831,
      "eval_PRM F1": 0.8860759493670886,
      "eval_PRM F1 AUC": 0.8129910948140389,
      "eval_PRM F1 AUC (fixed)": 0.7019381875327396,
      "eval_PRM F1 Neg": 0.6666666666666666,
      "eval_PRM NPV": 0.5806451612903226,
      "eval_PRM Precision": 0.9333333333333333,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.29255253076553345,
      "eval_runtime": 14.5029,
      "eval_samples_per_second": 3.034,
      "eval_steps_per_second": 0.138,
      "step": 685
    },
    {
      "epoch": 0.17765692272830275,
      "grad_norm": 6.064065924164856,
      "learning_rate": 1.227291684075541e-06,
      "loss": 0.4751,
      "step": 686
    },
    {
      "epoch": 0.17791589783432069,
      "grad_norm": 3.8170721320489993,
      "learning_rate": 1.2271404692958553e-06,
      "loss": 0.327,
      "step": 687
    },
    {
      "epoch": 0.1781748729403386,
      "grad_norm": 6.515137825706018,
      "learning_rate": 1.2269887620936524e-06,
      "loss": 0.3708,
      "step": 688
    },
    {
      "epoch": 0.17843384804635654,
      "grad_norm": 2.9103376080730987,
      "learning_rate": 1.2268365625929956e-06,
      "loss": 0.3284,
      "step": 689
    },
    {
      "epoch": 0.17869282315237447,
      "grad_norm": 4.771365939250701,
      "learning_rate": 1.2266838709183523e-06,
      "loss": 0.4002,
      "step": 690
    },
    {
      "epoch": 0.17869282315237447,
      "eval_PRM Accuracy": 0.8301886792452831,
      "eval_PRM F1": 0.8860759493670886,
      "eval_PRM F1 AUC": 0.8129910948140389,
      "eval_PRM F1 AUC (fixed)": 0.6938187532739654,
      "eval_PRM F1 Neg": 0.6666666666666666,
      "eval_PRM NPV": 0.5806451612903226,
      "eval_PRM Precision": 0.9333333333333333,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.29877424240112305,
      "eval_runtime": 14.538,
      "eval_samples_per_second": 3.027,
      "eval_steps_per_second": 0.138,
      "step": 690
    },
    {
      "epoch": 0.1789517982583924,
      "grad_norm": 6.18545393540289,
      "learning_rate": 1.2265306871945913e-06,
      "loss": 0.2963,
      "step": 691
    },
    {
      "epoch": 0.17921077336441035,
      "grad_norm": 3.6651890147345014,
      "learning_rate": 1.2263770115469844e-06,
      "loss": 0.3844,
      "step": 692
    },
    {
      "epoch": 0.1794697484704283,
      "grad_norm": 3.0288719572638274,
      "learning_rate": 1.2262228441012055e-06,
      "loss": 0.3152,
      "step": 693
    },
    {
      "epoch": 0.1797287235764462,
      "grad_norm": 4.805350206780974,
      "learning_rate": 1.2260681849833307e-06,
      "loss": 0.4042,
      "step": 694
    },
    {
      "epoch": 0.17998769868246414,
      "grad_norm": 4.11172461983298,
      "learning_rate": 1.2259130343198382e-06,
      "loss": 0.3348,
      "step": 695
    },
    {
      "epoch": 0.17998769868246414,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.8930817610062893,
      "eval_PRM F1 AUC": 0.819015191199581,
      "eval_PRM F1 AUC (fixed)": 0.6953902566788894,
      "eval_PRM F1 Neg": 0.6792452830188679,
      "eval_PRM NPV": 0.6,
      "eval_PRM Precision": 0.9342105263157895,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.298240065574646,
      "eval_runtime": 14.723,
      "eval_samples_per_second": 2.989,
      "eval_steps_per_second": 0.136,
      "step": 695
    },
    {
      "epoch": 0.18024667378848208,
      "grad_norm": 4.577272985951415,
      "learning_rate": 1.2257573922376082e-06,
      "loss": 0.3048,
      "step": 696
    },
    {
      "epoch": 0.18050564889450002,
      "grad_norm": 6.9027920443285655,
      "learning_rate": 1.2256012588639224e-06,
      "loss": 0.4526,
      "step": 697
    },
    {
      "epoch": 0.18076462400051796,
      "grad_norm": 6.007320601254431,
      "learning_rate": 1.2254446343264652e-06,
      "loss": 0.4382,
      "step": 698
    },
    {
      "epoch": 0.18102359910653587,
      "grad_norm": 6.4059971503781705,
      "learning_rate": 1.2252875187533216e-06,
      "loss": 0.2843,
      "step": 699
    },
    {
      "epoch": 0.1812825742125538,
      "grad_norm": 4.816960316434854,
      "learning_rate": 1.2251299122729788e-06,
      "loss": 0.3413,
      "step": 700
    },
    {
      "epoch": 0.1812825742125538,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.89171974522293,
      "eval_PRM F1 AUC": 0.8347302252488213,
      "eval_PRM F1 AUC (fixed)": 0.6927710843373494,
      "eval_PRM F1 Neg": 0.6909090909090909,
      "eval_PRM NPV": 0.59375,
      "eval_PRM Precision": 0.9459459459459459,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.8260869565217391,
      "eval_loss": 0.309047669172287,
      "eval_runtime": 14.3751,
      "eval_samples_per_second": 3.061,
      "eval_steps_per_second": 0.139,
      "step": 700
    },
    {
      "epoch": 0.18154154931857175,
      "grad_norm": 5.101486987466127,
      "learning_rate": 1.2249718150143253e-06,
      "loss": 0.3951,
      "step": 701
    },
    {
      "epoch": 0.1818005244245897,
      "grad_norm": 8.502311209246374,
      "learning_rate": 1.224813227106651e-06,
      "loss": 0.4991,
      "step": 702
    },
    {
      "epoch": 0.18205949953060763,
      "grad_norm": 5.728862100407766,
      "learning_rate": 1.2246541486796468e-06,
      "loss": 0.3105,
      "step": 703
    },
    {
      "epoch": 0.18231847463662557,
      "grad_norm": 10.721025982358185,
      "learning_rate": 1.2244945798634053e-06,
      "loss": 0.5076,
      "step": 704
    },
    {
      "epoch": 0.18257744974264348,
      "grad_norm": 7.180199889514728,
      "learning_rate": 1.2243345207884193e-06,
      "loss": 0.3735,
      "step": 705
    },
    {
      "epoch": 0.18257744974264348,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.8944099378881988,
      "eval_PRM F1 AUC": 0.8033001571503404,
      "eval_PRM F1 AUC (fixed)": 0.6888423258250393,
      "eval_PRM F1 Neg": 0.6666666666666666,
      "eval_PRM NPV": 0.6071428571428571,
      "eval_PRM Precision": 0.9230769230769231,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.30524346232414246,
      "eval_runtime": 14.8186,
      "eval_samples_per_second": 2.969,
      "eval_steps_per_second": 0.135,
      "step": 705
    },
    {
      "epoch": 0.18283642484866142,
      "grad_norm": 4.033301201721461,
      "learning_rate": 1.2241739715855833e-06,
      "loss": 0.3323,
      "step": 706
    },
    {
      "epoch": 0.18309539995467936,
      "grad_norm": 7.540320350893967,
      "learning_rate": 1.2240129323861923e-06,
      "loss": 0.4018,
      "step": 707
    },
    {
      "epoch": 0.1833543750606973,
      "grad_norm": 4.023342552193731,
      "learning_rate": 1.223851403321942e-06,
      "loss": 0.4608,
      "step": 708
    },
    {
      "epoch": 0.18361335016671523,
      "grad_norm": 3.4906951118001435,
      "learning_rate": 1.2236893845249286e-06,
      "loss": 0.3207,
      "step": 709
    },
    {
      "epoch": 0.18387232527273314,
      "grad_norm": 3.516295428942307,
      "learning_rate": 1.2235268761276492e-06,
      "loss": 0.4116,
      "step": 710
    },
    {
      "epoch": 0.18387232527273314,
      "eval_PRM Accuracy": 0.8490566037735849,
      "eval_PRM F1": 0.9,
      "eval_PRM F1 AUC": 0.8250392875851232,
      "eval_PRM F1 AUC (fixed)": 0.6888423258250392,
      "eval_PRM F1 Neg": 0.6923076923076923,
      "eval_PRM NPV": 0.6206896551724138,
      "eval_PRM Precision": 0.935064935064935,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.2983972430229187,
      "eval_runtime": 14.7691,
      "eval_samples_per_second": 2.979,
      "eval_steps_per_second": 0.135,
      "step": 710
    },
    {
      "epoch": 0.18413130037875108,
      "grad_norm": 6.3107471370478345,
      "learning_rate": 1.2233638782630008e-06,
      "loss": 0.3446,
      "step": 711
    },
    {
      "epoch": 0.18439027548476902,
      "grad_norm": 4.766463349421268,
      "learning_rate": 1.223200391064281e-06,
      "loss": 0.4559,
      "step": 712
    },
    {
      "epoch": 0.18464925059078696,
      "grad_norm": 4.079425145872921,
      "learning_rate": 1.2230364146651874e-06,
      "loss": 0.3296,
      "step": 713
    },
    {
      "epoch": 0.1849082256968049,
      "grad_norm": 3.558588858327177,
      "learning_rate": 1.2228719491998178e-06,
      "loss": 0.3722,
      "step": 714
    },
    {
      "epoch": 0.18516720080282284,
      "grad_norm": 3.6740128427409187,
      "learning_rate": 1.22270699480267e-06,
      "loss": 0.3337,
      "step": 715
    },
    {
      "epoch": 0.18516720080282284,
      "eval_PRM Accuracy": 0.8301886792452831,
      "eval_PRM F1": 0.8831168831168831,
      "eval_PRM F1 AUC": 0.8444211629125196,
      "eval_PRM F1 AUC (fixed)": 0.6893661602933473,
      "eval_PRM F1 Neg": 0.6896551724137931,
      "eval_PRM NPV": 0.5714285714285714,
      "eval_PRM Precision": 0.9577464788732394,
      "eval_PRM Recall": 0.8192771084337349,
      "eval_PRM Specificty": 0.8695652173913043,
      "eval_loss": 0.3114849328994751,
      "eval_runtime": 14.6726,
      "eval_samples_per_second": 2.999,
      "eval_steps_per_second": 0.136,
      "step": 715
    },
    {
      "epoch": 0.18542617590884075,
      "grad_norm": 6.842617200338519,
      "learning_rate": 1.222541551608641e-06,
      "loss": 0.4139,
      "step": 716
    },
    {
      "epoch": 0.1856851510148587,
      "grad_norm": 8.077453143122918,
      "learning_rate": 1.2223756197530287e-06,
      "loss": 0.4138,
      "step": 717
    },
    {
      "epoch": 0.18594412612087663,
      "grad_norm": 10.71942617315954,
      "learning_rate": 1.2222091993715296e-06,
      "loss": 0.3837,
      "step": 718
    },
    {
      "epoch": 0.18620310122689457,
      "grad_norm": 5.7851074969807526,
      "learning_rate": 1.2220422906002402e-06,
      "loss": 0.4341,
      "step": 719
    },
    {
      "epoch": 0.1864620763329125,
      "grad_norm": 4.993275457162052,
      "learning_rate": 1.2218748935756563e-06,
      "loss": 0.4591,
      "step": 720
    },
    {
      "epoch": 0.1864620763329125,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.89171974522293,
      "eval_PRM F1 AUC": 0.8347302252488213,
      "eval_PRM F1 AUC (fixed)": 0.6901519119958094,
      "eval_PRM F1 Neg": 0.6909090909090909,
      "eval_PRM NPV": 0.59375,
      "eval_PRM Precision": 0.9459459459459459,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.8260869565217391,
      "eval_loss": 0.30387333035469055,
      "eval_runtime": 15.0727,
      "eval_samples_per_second": 2.919,
      "eval_steps_per_second": 0.133,
      "step": 720
    },
    {
      "epoch": 0.18672105143893045,
      "grad_norm": 9.287087083955917,
      "learning_rate": 1.221707008434673e-06,
      "loss": 0.3178,
      "step": 721
    },
    {
      "epoch": 0.18698002654494836,
      "grad_norm": 8.33504855523225,
      "learning_rate": 1.2215386353145848e-06,
      "loss": 0.3372,
      "step": 722
    },
    {
      "epoch": 0.1872390016509663,
      "grad_norm": 3.0068926378723506,
      "learning_rate": 1.2213697743530841e-06,
      "loss": 0.3144,
      "step": 723
    },
    {
      "epoch": 0.18749797675698424,
      "grad_norm": 4.173934848386845,
      "learning_rate": 1.2212004256882643e-06,
      "loss": 0.4498,
      "step": 724
    },
    {
      "epoch": 0.18775695186300217,
      "grad_norm": 3.919799887702488,
      "learning_rate": 1.2210305894586158e-06,
      "loss": 0.3239,
      "step": 725
    },
    {
      "epoch": 0.18775695186300217,
      "eval_PRM Accuracy": 0.8301886792452831,
      "eval_PRM F1": 0.8860759493670886,
      "eval_PRM F1 AUC": 0.8129910948140389,
      "eval_PRM F1 AUC (fixed)": 0.6930330015715034,
      "eval_PRM F1 Neg": 0.6666666666666666,
      "eval_PRM NPV": 0.5806451612903226,
      "eval_PRM Precision": 0.9333333333333333,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.3035014569759369,
      "eval_runtime": 14.3599,
      "eval_samples_per_second": 3.064,
      "eval_steps_per_second": 0.139,
      "step": 725
    },
    {
      "epoch": 0.18801592696902011,
      "grad_norm": 3.465868954032309,
      "learning_rate": 1.2208602658030287e-06,
      "loss": 0.3427,
      "step": 726
    },
    {
      "epoch": 0.18827490207503803,
      "grad_norm": 3.9040806750182657,
      "learning_rate": 1.2206894548607914e-06,
      "loss": 0.3492,
      "step": 727
    },
    {
      "epoch": 0.18853387718105596,
      "grad_norm": 4.379059220952106,
      "learning_rate": 1.2205181567715905e-06,
      "loss": 0.4384,
      "step": 728
    },
    {
      "epoch": 0.1887928522870739,
      "grad_norm": 4.134944823065672,
      "learning_rate": 1.2203463716755117e-06,
      "loss": 0.2962,
      "step": 729
    },
    {
      "epoch": 0.18905182739309184,
      "grad_norm": 4.76895358133757,
      "learning_rate": 1.2201740997130388e-06,
      "loss": 0.4152,
      "step": 730
    },
    {
      "epoch": 0.18905182739309184,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.8930817610062893,
      "eval_PRM F1 AUC": 0.819015191199581,
      "eval_PRM F1 AUC (fixed)": 0.6951283394447355,
      "eval_PRM F1 Neg": 0.6792452830188679,
      "eval_PRM NPV": 0.6,
      "eval_PRM Precision": 0.9342105263157895,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.29401424527168274,
      "eval_runtime": 14.755,
      "eval_samples_per_second": 2.982,
      "eval_steps_per_second": 0.136,
      "step": 730
    },
    {
      "epoch": 0.18931080249910978,
      "grad_norm": 3.5289293203995435,
      "learning_rate": 1.2200013410250528e-06,
      "loss": 0.3676,
      "step": 731
    },
    {
      "epoch": 0.18956977760512772,
      "grad_norm": 3.544637575803375,
      "learning_rate": 1.219828095752834e-06,
      "loss": 0.3552,
      "step": 732
    },
    {
      "epoch": 0.18982875271114563,
      "grad_norm": 8.342332576696858,
      "learning_rate": 1.2196543640380601e-06,
      "loss": 0.4627,
      "step": 733
    },
    {
      "epoch": 0.19008772781716357,
      "grad_norm": 6.969226338650388,
      "learning_rate": 1.2194801460228063e-06,
      "loss": 0.4385,
      "step": 734
    },
    {
      "epoch": 0.1903467029231815,
      "grad_norm": 4.161712948947366,
      "learning_rate": 1.219305441849546e-06,
      "loss": 0.3898,
      "step": 735
    },
    {
      "epoch": 0.1903467029231815,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.8944099378881988,
      "eval_PRM F1 AUC": 0.8033001571503404,
      "eval_PRM F1 AUC (fixed)": 0.6938187532739655,
      "eval_PRM F1 Neg": 0.6666666666666666,
      "eval_PRM NPV": 0.6071428571428571,
      "eval_PRM Precision": 0.9230769230769231,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.2999347448348999,
      "eval_runtime": 15.0306,
      "eval_samples_per_second": 2.927,
      "eval_steps_per_second": 0.133,
      "step": 735
    },
    {
      "epoch": 0.19060567802919945,
      "grad_norm": 6.84476769969144,
      "learning_rate": 1.21913025166115e-06,
      "loss": 0.4164,
      "step": 736
    },
    {
      "epoch": 0.1908646531352174,
      "grad_norm": 5.136634266840099,
      "learning_rate": 1.2189545756008864e-06,
      "loss": 0.3763,
      "step": 737
    },
    {
      "epoch": 0.1911236282412353,
      "grad_norm": 7.2580426628791255,
      "learning_rate": 1.2187784138124207e-06,
      "loss": 0.3112,
      "step": 738
    },
    {
      "epoch": 0.19138260334725324,
      "grad_norm": 5.718167380444347,
      "learning_rate": 1.2186017664398158e-06,
      "loss": 0.3891,
      "step": 739
    },
    {
      "epoch": 0.19164157845327118,
      "grad_norm": 6.763967879398377,
      "learning_rate": 1.218424633627532e-06,
      "loss": 0.342,
      "step": 740
    },
    {
      "epoch": 0.19164157845327118,
      "eval_PRM Accuracy": 0.8679245283018868,
      "eval_PRM F1": 0.9135802469135802,
      "eval_PRM F1 AUC": 0.8370874803562075,
      "eval_PRM F1 AUC (fixed)": 0.6940806705081194,
      "eval_PRM F1 Neg": 0.72,
      "eval_PRM NPV": 0.6666666666666666,
      "eval_PRM Precision": 0.9367088607594937,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.2982929050922394,
      "eval_runtime": 14.3814,
      "eval_samples_per_second": 3.06,
      "eval_steps_per_second": 0.139,
      "step": 740
    },
    {
      "epoch": 0.19190055355928912,
      "grad_norm": 4.5433183235959005,
      "learning_rate": 1.2182470155204253e-06,
      "loss": 0.3741,
      "step": 741
    },
    {
      "epoch": 0.19215952866530706,
      "grad_norm": 5.452456164324737,
      "learning_rate": 1.2180689122637499e-06,
      "loss": 0.397,
      "step": 742
    },
    {
      "epoch": 0.192418503771325,
      "grad_norm": 3.4223013167649468,
      "learning_rate": 1.2178903240031563e-06,
      "loss": 0.2803,
      "step": 743
    },
    {
      "epoch": 0.1926774788773429,
      "grad_norm": 5.385205369812769,
      "learning_rate": 1.217711250884692e-06,
      "loss": 0.3934,
      "step": 744
    },
    {
      "epoch": 0.19293645398336084,
      "grad_norm": 6.183148792048128,
      "learning_rate": 1.2175316930547997e-06,
      "loss": 0.4244,
      "step": 745
    },
    {
      "epoch": 0.19293645398336084,
      "eval_PRM Accuracy": 0.8584905660377359,
      "eval_PRM F1": 0.906832298136646,
      "eval_PRM F1 AUC": 0.8310633839706653,
      "eval_PRM F1 AUC (fixed)": 0.6911995809324253,
      "eval_PRM F1 Neg": 0.7058823529411765,
      "eval_PRM NPV": 0.6428571428571429,
      "eval_PRM Precision": 0.9358974358974359,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.29950782656669617,
      "eval_runtime": 14.3784,
      "eval_samples_per_second": 3.06,
      "eval_steps_per_second": 0.139,
      "step": 745
    },
    {
      "epoch": 0.19319542908937878,
      "grad_norm": 3.935237830232094,
      "learning_rate": 1.2173516506603203e-06,
      "loss": 0.3224,
      "step": 746
    },
    {
      "epoch": 0.19345440419539672,
      "grad_norm": 4.276810366583379,
      "learning_rate": 1.2171711238484896e-06,
      "loss": 0.3685,
      "step": 747
    },
    {
      "epoch": 0.19371337930141466,
      "grad_norm": 8.221097875723666,
      "learning_rate": 1.2169901127669401e-06,
      "loss": 0.4421,
      "step": 748
    },
    {
      "epoch": 0.19397235440743257,
      "grad_norm": 3.1276241732515877,
      "learning_rate": 1.2168086175637006e-06,
      "loss": 0.3708,
      "step": 749
    },
    {
      "epoch": 0.1942313295134505,
      "grad_norm": 5.711025740988542,
      "learning_rate": 1.216626638387195e-06,
      "loss": 0.3505,
      "step": 750
    },
    {
      "epoch": 0.1942313295134505,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.89171974522293,
      "eval_PRM F1 AUC": 0.8347302252488213,
      "eval_PRM F1 AUC (fixed)": 0.6893661602933472,
      "eval_PRM F1 Neg": 0.6909090909090909,
      "eval_PRM NPV": 0.59375,
      "eval_PRM Precision": 0.9459459459459459,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.8260869565217391,
      "eval_loss": 0.3052200376987457,
      "eval_runtime": 14.9201,
      "eval_samples_per_second": 2.949,
      "eval_steps_per_second": 0.134,
      "step": 750
    },
    {
      "epoch": 0.19449030461946845,
      "grad_norm": 4.218636733039021,
      "learning_rate": 1.2164441753862438e-06,
      "loss": 0.3969,
      "step": 751
    },
    {
      "epoch": 0.1947492797254864,
      "grad_norm": 6.769566136370222,
      "learning_rate": 1.2162612287100627e-06,
      "loss": 0.3348,
      "step": 752
    },
    {
      "epoch": 0.19500825483150433,
      "grad_norm": 3.4857801672629707,
      "learning_rate": 1.216077798508263e-06,
      "loss": 0.3256,
      "step": 753
    },
    {
      "epoch": 0.19526722993752227,
      "grad_norm": 9.174244423244135,
      "learning_rate": 1.2158938849308519e-06,
      "loss": 0.368,
      "step": 754
    },
    {
      "epoch": 0.19552620504354018,
      "grad_norm": 9.747075045547675,
      "learning_rate": 1.2157094881282308e-06,
      "loss": 0.336,
      "step": 755
    },
    {
      "epoch": 0.19552620504354018,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.89171974522293,
      "eval_PRM F1 AUC": 0.8347302252488213,
      "eval_PRM F1 AUC (fixed)": 0.6990570979570455,
      "eval_PRM F1 Neg": 0.6909090909090909,
      "eval_PRM NPV": 0.59375,
      "eval_PRM Precision": 0.9459459459459459,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.8260869565217391,
      "eval_loss": 0.2928756773471832,
      "eval_runtime": 14.2527,
      "eval_samples_per_second": 3.087,
      "eval_steps_per_second": 0.14,
      "step": 755
    },
    {
      "epoch": 0.19578518014955812,
      "grad_norm": 3.251806890809459,
      "learning_rate": 1.2155246082511975e-06,
      "loss": 0.2617,
      "step": 756
    },
    {
      "epoch": 0.19604415525557606,
      "grad_norm": 4.98149048932274,
      "learning_rate": 1.2153392454509442e-06,
      "loss": 0.3699,
      "step": 757
    },
    {
      "epoch": 0.196303130361594,
      "grad_norm": 3.9074566654750433,
      "learning_rate": 1.2151533998790583e-06,
      "loss": 0.4173,
      "step": 758
    },
    {
      "epoch": 0.19656210546761194,
      "grad_norm": 3.752364680965662,
      "learning_rate": 1.2149670716875215e-06,
      "loss": 0.3821,
      "step": 759
    },
    {
      "epoch": 0.19682108057362985,
      "grad_norm": 5.348304560974054,
      "learning_rate": 1.2147802610287107e-06,
      "loss": 0.3873,
      "step": 760
    },
    {
      "epoch": 0.19682108057362985,
      "eval_PRM Accuracy": 0.8301886792452831,
      "eval_PRM F1": 0.8860759493670886,
      "eval_PRM F1 AUC": 0.8129910948140389,
      "eval_PRM F1 AUC (fixed)": 0.711629125196438,
      "eval_PRM F1 Neg": 0.6666666666666666,
      "eval_PRM NPV": 0.5806451612903226,
      "eval_PRM Precision": 0.9333333333333333,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.28274041414260864,
      "eval_runtime": 14.5446,
      "eval_samples_per_second": 3.025,
      "eval_steps_per_second": 0.138,
      "step": 760
    },
    {
      "epoch": 0.1970800556796478,
      "grad_norm": 7.55785178982506,
      "learning_rate": 1.2145929680553973e-06,
      "loss": 0.324,
      "step": 761
    },
    {
      "epoch": 0.19733903078566573,
      "grad_norm": 3.0868830616823852,
      "learning_rate": 1.2144051929207466e-06,
      "loss": 0.2994,
      "step": 762
    },
    {
      "epoch": 0.19759800589168366,
      "grad_norm": 3.5417146803588118,
      "learning_rate": 1.2142169357783194e-06,
      "loss": 0.3483,
      "step": 763
    },
    {
      "epoch": 0.1978569809977016,
      "grad_norm": 4.0208469768721224,
      "learning_rate": 1.214028196782069e-06,
      "loss": 0.4194,
      "step": 764
    },
    {
      "epoch": 0.19811595610371954,
      "grad_norm": 3.8431699488974886,
      "learning_rate": 1.2138389760863443e-06,
      "loss": 0.3649,
      "step": 765
    },
    {
      "epoch": 0.19811595610371954,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.8930817610062893,
      "eval_PRM F1 AUC": 0.819015191199581,
      "eval_PRM F1 AUC (fixed)": 0.7257726558407543,
      "eval_PRM F1 Neg": 0.6792452830188679,
      "eval_PRM NPV": 0.6,
      "eval_PRM Precision": 0.9342105263157895,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.274412602186203,
      "eval_runtime": 14.9209,
      "eval_samples_per_second": 2.949,
      "eval_steps_per_second": 0.134,
      "step": 765
    },
    {
      "epoch": 0.19837493120973745,
      "grad_norm": 4.5792155524540386,
      "learning_rate": 1.2136492738458874e-06,
      "loss": 0.3576,
      "step": 766
    },
    {
      "epoch": 0.1986339063157554,
      "grad_norm": 8.636004621584396,
      "learning_rate": 1.2134590902158339e-06,
      "loss": 0.4316,
      "step": 767
    },
    {
      "epoch": 0.19889288142177333,
      "grad_norm": 4.140044364503692,
      "learning_rate": 1.213268425351714e-06,
      "loss": 0.3677,
      "step": 768
    },
    {
      "epoch": 0.19915185652779127,
      "grad_norm": 5.324756102689371,
      "learning_rate": 1.2130772794094504e-06,
      "loss": 0.3403,
      "step": 769
    },
    {
      "epoch": 0.1994108316338092,
      "grad_norm": 8.4411250018076,
      "learning_rate": 1.21288565254536e-06,
      "loss": 0.391,
      "step": 770
    },
    {
      "epoch": 0.1994108316338092,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.8930817610062893,
      "eval_PRM F1 AUC": 0.819015191199581,
      "eval_PRM F1 AUC (fixed)": 0.7247249869041383,
      "eval_PRM F1 Neg": 0.6792452830188679,
      "eval_PRM NPV": 0.6,
      "eval_PRM Precision": 0.9342105263157895,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.2825393080711365,
      "eval_runtime": 15.1448,
      "eval_samples_per_second": 2.905,
      "eval_steps_per_second": 0.132,
      "step": 770
    },
    {
      "epoch": 0.19966980673982712,
      "grad_norm": 7.501946899653349,
      "learning_rate": 1.2126935449161529e-06,
      "loss": 0.3256,
      "step": 771
    },
    {
      "epoch": 0.19992878184584506,
      "grad_norm": 5.065470925957987,
      "learning_rate": 1.2125009566789318e-06,
      "loss": 0.3201,
      "step": 772
    },
    {
      "epoch": 0.200187756951863,
      "grad_norm": 3.8048849885969136,
      "learning_rate": 1.2123078879911928e-06,
      "loss": 0.3268,
      "step": 773
    },
    {
      "epoch": 0.20044673205788094,
      "grad_norm": 7.635096076804379,
      "learning_rate": 1.2121143390108253e-06,
      "loss": 0.4506,
      "step": 774
    },
    {
      "epoch": 0.20070570716389888,
      "grad_norm": 3.4186284893273022,
      "learning_rate": 1.2119203098961107e-06,
      "loss": 0.3222,
      "step": 775
    },
    {
      "epoch": 0.20070570716389888,
      "eval_PRM Accuracy": 0.8490566037735849,
      "eval_PRM F1": 0.8974358974358975,
      "eval_PRM F1 AUC": 0.8564693556836039,
      "eval_PRM F1 AUC (fixed)": 0.714772132006286,
      "eval_PRM F1 Neg": 0.7142857142857143,
      "eval_PRM NPV": 0.6060606060606061,
      "eval_PRM Precision": 0.958904109589041,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.8695652173913043,
      "eval_loss": 0.2990555763244629,
      "eval_runtime": 14.5907,
      "eval_samples_per_second": 3.016,
      "eval_steps_per_second": 0.137,
      "step": 775
    },
    {
      "epoch": 0.20096468226991682,
      "grad_norm": 5.857241588906598,
      "learning_rate": 1.211725800805724e-06,
      "loss": 0.412,
      "step": 776
    },
    {
      "epoch": 0.20122365737593473,
      "grad_norm": 8.636257256129669,
      "learning_rate": 1.2115308118987314e-06,
      "loss": 0.3583,
      "step": 777
    },
    {
      "epoch": 0.20148263248195267,
      "grad_norm": 4.001799400777121,
      "learning_rate": 1.2113353433345928e-06,
      "loss": 0.4281,
      "step": 778
    },
    {
      "epoch": 0.2017416075879706,
      "grad_norm": 5.457582326684946,
      "learning_rate": 1.2111393952731599e-06,
      "loss": 0.426,
      "step": 779
    },
    {
      "epoch": 0.20200058269398855,
      "grad_norm": 5.88931960974729,
      "learning_rate": 1.2109429678746757e-06,
      "loss": 0.3624,
      "step": 780
    },
    {
      "epoch": 0.20200058269398855,
      "eval_PRM Accuracy": 0.8490566037735849,
      "eval_PRM F1": 0.8987341772151899,
      "eval_PRM F1 AUC": 0.8407543216343635,
      "eval_PRM F1 AUC (fixed)": 0.7197485594552121,
      "eval_PRM F1 Neg": 0.7037037037037037,
      "eval_PRM NPV": 0.6129032258064516,
      "eval_PRM Precision": 0.9466666666666667,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.8260869565217391,
      "eval_loss": 0.2832641005516052,
      "eval_runtime": 15.2286,
      "eval_samples_per_second": 2.889,
      "eval_steps_per_second": 0.131,
      "step": 780
    },
    {
      "epoch": 0.20225955780000648,
      "grad_norm": 4.5606319970193905,
      "learning_rate": 1.2107460612997768e-06,
      "loss": 0.4612,
      "step": 781
    },
    {
      "epoch": 0.2025185329060244,
      "grad_norm": 10.640629778364074,
      "learning_rate": 1.2105486757094902e-06,
      "loss": 0.3714,
      "step": 782
    },
    {
      "epoch": 0.20277750801204233,
      "grad_norm": 10.81610998670503,
      "learning_rate": 1.2103508112652357e-06,
      "loss": 0.4135,
      "step": 783
    },
    {
      "epoch": 0.20303648311806027,
      "grad_norm": 5.278795898437136,
      "learning_rate": 1.2101524681288238e-06,
      "loss": 0.4048,
      "step": 784
    },
    {
      "epoch": 0.2032954582240782,
      "grad_norm": 9.07962614012763,
      "learning_rate": 1.209953646462457e-06,
      "loss": 0.3948,
      "step": 785
    },
    {
      "epoch": 0.2032954582240782,
      "eval_PRM Accuracy": 0.8584905660377359,
      "eval_PRM F1": 0.9056603773584906,
      "eval_PRM F1 AUC": 0.8467784180199058,
      "eval_PRM F1 AUC (fixed)": 0.7100576217915139,
      "eval_PRM F1 Neg": 0.7169811320754716,
      "eval_PRM NPV": 0.6333333333333333,
      "eval_PRM Precision": 0.9473684210526315,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.8260869565217391,
      "eval_loss": 0.28009822964668274,
      "eval_runtime": 14.868,
      "eval_samples_per_second": 2.959,
      "eval_steps_per_second": 0.135,
      "step": 785
    },
    {
      "epoch": 0.20355443333009615,
      "grad_norm": 4.945794786482562,
      "learning_rate": 1.2097543464287291e-06,
      "loss": 0.3626,
      "step": 786
    },
    {
      "epoch": 0.2038134084361141,
      "grad_norm": 4.210535900275439,
      "learning_rate": 1.2095545681906252e-06,
      "loss": 0.4574,
      "step": 787
    },
    {
      "epoch": 0.204072383542132,
      "grad_norm": 7.8934875649755885,
      "learning_rate": 1.2093543119115211e-06,
      "loss": 0.4204,
      "step": 788
    },
    {
      "epoch": 0.20433135864814994,
      "grad_norm": 5.9298063027561705,
      "learning_rate": 1.2091535777551837e-06,
      "loss": 0.3544,
      "step": 789
    },
    {
      "epoch": 0.20459033375416788,
      "grad_norm": 4.61640368794494,
      "learning_rate": 1.2089523658857708e-06,
      "loss": 0.3672,
      "step": 790
    },
    {
      "epoch": 0.20459033375416788,
      "eval_PRM Accuracy": 0.8301886792452831,
      "eval_PRM F1": 0.8831168831168831,
      "eval_PRM F1 AUC": 0.8444211629125196,
      "eval_PRM F1 AUC (fixed)": 0.6946045049764275,
      "eval_PRM F1 Neg": 0.6896551724137931,
      "eval_PRM NPV": 0.5714285714285714,
      "eval_PRM Precision": 0.9577464788732394,
      "eval_PRM Recall": 0.8192771084337349,
      "eval_PRM Specificty": 0.8695652173913043,
      "eval_loss": 0.30553391575813293,
      "eval_runtime": 14.5411,
      "eval_samples_per_second": 3.026,
      "eval_steps_per_second": 0.138,
      "step": 790
    },
    {
      "epoch": 0.20484930886018582,
      "grad_norm": 3.8906893870399473,
      "learning_rate": 1.2087506764678309e-06,
      "loss": 0.3811,
      "step": 791
    },
    {
      "epoch": 0.20510828396620376,
      "grad_norm": 5.048510996715175,
      "learning_rate": 1.2085485096663026e-06,
      "loss": 0.5011,
      "step": 792
    },
    {
      "epoch": 0.20536725907222167,
      "grad_norm": 11.837946428436359,
      "learning_rate": 1.2083458656465159e-06,
      "loss": 0.3957,
      "step": 793
    },
    {
      "epoch": 0.2056262341782396,
      "grad_norm": 11.305888044956315,
      "learning_rate": 1.2081427445741896e-06,
      "loss": 0.3496,
      "step": 794
    },
    {
      "epoch": 0.20588520928425755,
      "grad_norm": 9.260364149994357,
      "learning_rate": 1.207939146615434e-06,
      "loss": 0.3768,
      "step": 795
    },
    {
      "epoch": 0.20588520928425755,
      "eval_PRM Accuracy": 0.8301886792452831,
      "eval_PRM F1": 0.8846153846153846,
      "eval_PRM F1 AUC": 0.8287061288632793,
      "eval_PRM F1 AUC (fixed)": 0.6946045049764275,
      "eval_PRM F1 Neg": 0.6785714285714286,
      "eval_PRM NPV": 0.5757575757575758,
      "eval_PRM Precision": 0.9452054794520548,
      "eval_PRM Recall": 0.8313253012048193,
      "eval_PRM Specificty": 0.8260869565217391,
      "eval_loss": 0.3177688419818878,
      "eval_runtime": 14.5323,
      "eval_samples_per_second": 3.028,
      "eval_steps_per_second": 0.138,
      "step": 795
    },
    {
      "epoch": 0.2061441843902755,
      "grad_norm": 6.166600089935319,
      "learning_rate": 1.207735071936749e-06,
      "loss": 0.3425,
      "step": 796
    },
    {
      "epoch": 0.20640315949629343,
      "grad_norm": 5.291436952965745,
      "learning_rate": 1.2075305207050233e-06,
      "loss": 0.3083,
      "step": 797
    },
    {
      "epoch": 0.20666213460231136,
      "grad_norm": 4.946759609608186,
      "learning_rate": 1.2073254930875372e-06,
      "loss": 0.346,
      "step": 798
    },
    {
      "epoch": 0.20692110970832928,
      "grad_norm": 3.2952179929570984,
      "learning_rate": 1.2071199892519587e-06,
      "loss": 0.3274,
      "step": 799
    },
    {
      "epoch": 0.20718008481434722,
      "grad_norm": 3.9838080756980765,
      "learning_rate": 1.2069140093663468e-06,
      "loss": 0.4127,
      "step": 800
    },
    {
      "epoch": 0.20718008481434722,
      "eval_PRM Accuracy": 0.8584905660377359,
      "eval_PRM F1": 0.9079754601226994,
      "eval_PRM F1 AUC": 0.8153483499214248,
      "eval_PRM F1 AUC (fixed)": 0.710843373493976,
      "eval_PRM F1 Neg": 0.6938775510204082,
      "eval_PRM NPV": 0.6538461538461539,
      "eval_PRM Precision": 0.925,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.3012973964214325,
      "eval_runtime": 14.7148,
      "eval_samples_per_second": 2.99,
      "eval_steps_per_second": 0.136,
      "step": 800
    },
    {
      "epoch": 0.20743905992036515,
      "grad_norm": 5.84440802349982,
      "learning_rate": 1.2067075535991485e-06,
      "loss": 0.3121,
      "step": 801
    },
    {
      "epoch": 0.2076980350263831,
      "grad_norm": 18.634255107630327,
      "learning_rate": 1.206500622119201e-06,
      "loss": 0.6074,
      "step": 802
    },
    {
      "epoch": 0.20795701013240103,
      "grad_norm": 3.4684820156592164,
      "learning_rate": 1.2062932150957302e-06,
      "loss": 0.2838,
      "step": 803
    },
    {
      "epoch": 0.20821598523841894,
      "grad_norm": 6.985530687578916,
      "learning_rate": 1.2060853326983506e-06,
      "loss": 0.3661,
      "step": 804
    },
    {
      "epoch": 0.20847496034443688,
      "grad_norm": 15.523633503922179,
      "learning_rate": 1.205876975097066e-06,
      "loss": 0.5244,
      "step": 805
    },
    {
      "epoch": 0.20847496034443688,
      "eval_PRM Accuracy": 0.8584905660377359,
      "eval_PRM F1": 0.9079754601226994,
      "eval_PRM F1 AUC": 0.8153483499214248,
      "eval_PRM F1 AUC (fixed)": 0.717391304347826,
      "eval_PRM F1 Neg": 0.6938775510204082,
      "eval_PRM NPV": 0.6538461538461539,
      "eval_PRM Precision": 0.925,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.3000315725803375,
      "eval_runtime": 14.3543,
      "eval_samples_per_second": 3.065,
      "eval_steps_per_second": 0.139,
      "step": 805
    },
    {
      "epoch": 0.20873393545045482,
      "grad_norm": 3.215330481636461,
      "learning_rate": 1.2056681424622682e-06,
      "loss": 0.3055,
      "step": 806
    },
    {
      "epoch": 0.20899291055647276,
      "grad_norm": 3.3904213572865647,
      "learning_rate": 1.2054588349647376e-06,
      "loss": 0.3612,
      "step": 807
    },
    {
      "epoch": 0.2092518856624907,
      "grad_norm": 4.089578401872159,
      "learning_rate": 1.2052490527756435e-06,
      "loss": 0.3688,
      "step": 808
    },
    {
      "epoch": 0.20951086076850864,
      "grad_norm": 5.149244379565398,
      "learning_rate": 1.205038796066543e-06,
      "loss": 0.3794,
      "step": 809
    },
    {
      "epoch": 0.20976983587452655,
      "grad_norm": 7.982346861326755,
      "learning_rate": 1.2048280650093808e-06,
      "loss": 0.4222,
      "step": 810
    },
    {
      "epoch": 0.20976983587452655,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.89171974522293,
      "eval_PRM F1 AUC": 0.8347302252488213,
      "eval_PRM F1 AUC (fixed)": 0.7192247249869042,
      "eval_PRM F1 Neg": 0.6909090909090909,
      "eval_PRM NPV": 0.59375,
      "eval_PRM Precision": 0.9459459459459459,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.8260869565217391,
      "eval_loss": 0.3014485538005829,
      "eval_runtime": 14.5923,
      "eval_samples_per_second": 3.015,
      "eval_steps_per_second": 0.137,
      "step": 810
    },
    {
      "epoch": 0.2100288109805445,
      "grad_norm": 3.3870290321601293,
      "learning_rate": 1.2046168597764905e-06,
      "loss": 0.3392,
      "step": 811
    },
    {
      "epoch": 0.21028778608656243,
      "grad_norm": 5.422644541842962,
      "learning_rate": 1.2044051805405923e-06,
      "loss": 0.3788,
      "step": 812
    },
    {
      "epoch": 0.21054676119258037,
      "grad_norm": 12.103695395274976,
      "learning_rate": 1.2041930274747955e-06,
      "loss": 0.394,
      "step": 813
    },
    {
      "epoch": 0.2108057362985983,
      "grad_norm": 5.780235540532406,
      "learning_rate": 1.2039804007525956e-06,
      "loss": 0.4104,
      "step": 814
    },
    {
      "epoch": 0.21106471140461622,
      "grad_norm": 8.248186220584325,
      "learning_rate": 1.203767300547876e-06,
      "loss": 0.3479,
      "step": 815
    },
    {
      "epoch": 0.21106471140461622,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.8903225806451613,
      "eval_PRM F1 AUC": 0.8504452592980618,
      "eval_PRM F1 AUC (fixed)": 0.7168674698795181,
      "eval_PRM F1 Neg": 0.7017543859649122,
      "eval_PRM NPV": 0.5882352941176471,
      "eval_PRM Precision": 0.9583333333333334,
      "eval_PRM Recall": 0.8313253012048193,
      "eval_PRM Specificty": 0.8695652173913043,
      "eval_loss": 0.3021652400493622,
      "eval_runtime": 14.4687,
      "eval_samples_per_second": 3.041,
      "eval_steps_per_second": 0.138,
      "step": 815
    },
    {
      "epoch": 0.21132368651063416,
      "grad_norm": 10.434436452029244,
      "learning_rate": 1.203553727034907e-06,
      "loss": 0.3728,
      "step": 816
    },
    {
      "epoch": 0.2115826616166521,
      "grad_norm": 4.209869566782344,
      "learning_rate": 1.2033396803883467e-06,
      "loss": 0.3499,
      "step": 817
    },
    {
      "epoch": 0.21184163672267003,
      "grad_norm": 5.604499786309078,
      "learning_rate": 1.2031251607832393e-06,
      "loss": 0.4211,
      "step": 818
    },
    {
      "epoch": 0.21210061182868797,
      "grad_norm": 3.4492243819967734,
      "learning_rate": 1.2029101683950161e-06,
      "loss": 0.4104,
      "step": 819
    },
    {
      "epoch": 0.2123595869347059,
      "grad_norm": 5.627721634673,
      "learning_rate": 1.202694703399495e-06,
      "loss": 0.4138,
      "step": 820
    },
    {
      "epoch": 0.2123595869347059,
      "eval_PRM Accuracy": 0.8679245283018868,
      "eval_PRM F1": 0.9125,
      "eval_PRM F1 AUC": 0.8528025144054479,
      "eval_PRM F1 AUC (fixed)": 0.7187008905185961,
      "eval_PRM F1 Neg": 0.7307692307692307,
      "eval_PRM NPV": 0.6551724137931034,
      "eval_PRM Precision": 0.948051948051948,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.8260869565217391,
      "eval_loss": 0.272720068693161,
      "eval_runtime": 14.4143,
      "eval_samples_per_second": 3.053,
      "eval_steps_per_second": 0.139,
      "step": 820
    },
    {
      "epoch": 0.21261856204072382,
      "grad_norm": 4.1223242524960275,
      "learning_rate": 1.2024787659728805e-06,
      "loss": 0.3767,
      "step": 821
    },
    {
      "epoch": 0.21287753714674176,
      "grad_norm": 4.041731071615488,
      "learning_rate": 1.2022623562917634e-06,
      "loss": 0.3681,
      "step": 822
    },
    {
      "epoch": 0.2131365122527597,
      "grad_norm": 12.213894926599707,
      "learning_rate": 1.2020454745331207e-06,
      "loss": 0.5263,
      "step": 823
    },
    {
      "epoch": 0.21339548735877764,
      "grad_norm": 4.048567978079006,
      "learning_rate": 1.2018281208743155e-06,
      "loss": 0.2961,
      "step": 824
    },
    {
      "epoch": 0.21365446246479558,
      "grad_norm": 3.4978630432290467,
      "learning_rate": 1.2016102954930965e-06,
      "loss": 0.2978,
      "step": 825
    },
    {
      "epoch": 0.21365446246479558,
      "eval_PRM Accuracy": 0.8679245283018868,
      "eval_PRM F1": 0.9125,
      "eval_PRM F1 AUC": 0.8528025144054479,
      "eval_PRM F1 AUC (fixed)": 0.7160817181770561,
      "eval_PRM F1 Neg": 0.7307692307692307,
      "eval_PRM NPV": 0.6551724137931034,
      "eval_PRM Precision": 0.948051948051948,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.8260869565217391,
      "eval_loss": 0.27165505290031433,
      "eval_runtime": 14.2242,
      "eval_samples_per_second": 3.093,
      "eval_steps_per_second": 0.141,
      "step": 825
    },
    {
      "epoch": 0.21391343757081352,
      "grad_norm": 6.136696971864226,
      "learning_rate": 1.2013919985675985e-06,
      "loss": 0.4115,
      "step": 826
    },
    {
      "epoch": 0.21417241267683143,
      "grad_norm": 7.112543853206797,
      "learning_rate": 1.2011732302763419e-06,
      "loss": 0.3849,
      "step": 827
    },
    {
      "epoch": 0.21443138778284937,
      "grad_norm": 3.7784105346607646,
      "learning_rate": 1.2009539907982324e-06,
      "loss": 0.3712,
      "step": 828
    },
    {
      "epoch": 0.2146903628888673,
      "grad_norm": 4.433101135916309,
      "learning_rate": 1.2007342803125616e-06,
      "loss": 0.3525,
      "step": 829
    },
    {
      "epoch": 0.21494933799488525,
      "grad_norm": 3.219684081039623,
      "learning_rate": 1.2005140989990051e-06,
      "loss": 0.325,
      "step": 830
    },
    {
      "epoch": 0.21494933799488525,
      "eval_PRM Accuracy": 0.8584905660377359,
      "eval_PRM F1": 0.9056603773584906,
      "eval_PRM F1 AUC": 0.8467784180199058,
      "eval_PRM F1 AUC (fixed)": 0.7145102147721321,
      "eval_PRM F1 Neg": 0.7169811320754716,
      "eval_PRM NPV": 0.6333333333333333,
      "eval_PRM Precision": 0.9473684210526315,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.8260869565217391,
      "eval_loss": 0.2757134437561035,
      "eval_runtime": 14.4793,
      "eval_samples_per_second": 3.039,
      "eval_steps_per_second": 0.138,
      "step": 830
    },
    {
      "epoch": 0.2152083131009032,
      "grad_norm": 10.491754275926994,
      "learning_rate": 1.2002934470376248e-06,
      "loss": 0.4024,
      "step": 831
    },
    {
      "epoch": 0.2154672882069211,
      "grad_norm": 4.998222363731994,
      "learning_rate": 1.200072324608867e-06,
      "loss": 0.3625,
      "step": 832
    },
    {
      "epoch": 0.21572626331293904,
      "grad_norm": 4.470005863558178,
      "learning_rate": 1.1998507318935624e-06,
      "loss": 0.3332,
      "step": 833
    },
    {
      "epoch": 0.21598523841895698,
      "grad_norm": 7.993509681116096,
      "learning_rate": 1.1996286690729266e-06,
      "loss": 0.369,
      "step": 834
    },
    {
      "epoch": 0.21624421352497492,
      "grad_norm": 4.708532369425207,
      "learning_rate": 1.1994061363285598e-06,
      "loss": 0.3882,
      "step": 835
    },
    {
      "epoch": 0.21624421352497492,
      "eval_PRM Accuracy": 0.8679245283018868,
      "eval_PRM F1": 0.9113924050632911,
      "eval_PRM F1 AUC": 0.8685175484546883,
      "eval_PRM F1 AUC (fixed)": 0.7145102147721321,
      "eval_PRM F1 Neg": 0.7407407407407407,
      "eval_PRM NPV": 0.6451612903225806,
      "eval_PRM Precision": 0.96,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.8695652173913043,
      "eval_loss": 0.27347585558891296,
      "eval_runtime": 15.0518,
      "eval_samples_per_second": 2.923,
      "eval_steps_per_second": 0.133,
      "step": 835
    },
    {
      "epoch": 0.21650318863099285,
      "grad_norm": 3.187417433180603,
      "learning_rate": 1.1991831338424465e-06,
      "loss": 0.3189,
      "step": 836
    },
    {
      "epoch": 0.2167621637370108,
      "grad_norm": 5.108923083012557,
      "learning_rate": 1.1989596617969546e-06,
      "loss": 0.468,
      "step": 837
    },
    {
      "epoch": 0.2170211388430287,
      "grad_norm": 3.5092540984088227,
      "learning_rate": 1.198735720374837e-06,
      "loss": 0.2836,
      "step": 838
    },
    {
      "epoch": 0.21728011394904664,
      "grad_norm": 6.365958781576424,
      "learning_rate": 1.1985113097592302e-06,
      "loss": 0.3103,
      "step": 839
    },
    {
      "epoch": 0.21753908905506458,
      "grad_norm": 7.018918766044205,
      "learning_rate": 1.1982864301336541e-06,
      "loss": 0.3434,
      "step": 840
    },
    {
      "epoch": 0.21753908905506458,
      "eval_PRM Accuracy": 0.8679245283018868,
      "eval_PRM F1": 0.9125,
      "eval_PRM F1 AUC": 0.8528025144054479,
      "eval_PRM F1 AUC (fixed)": 0.7176532215819801,
      "eval_PRM F1 Neg": 0.7307692307692307,
      "eval_PRM NPV": 0.6551724137931034,
      "eval_PRM Precision": 0.948051948051948,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.8260869565217391,
      "eval_loss": 0.26224979758262634,
      "eval_runtime": 14.8565,
      "eval_samples_per_second": 2.962,
      "eval_steps_per_second": 0.135,
      "step": 840
    },
    {
      "epoch": 0.21779806416108252,
      "grad_norm": 3.179988198620702,
      "learning_rate": 1.198061081682012e-06,
      "loss": 0.292,
      "step": 841
    },
    {
      "epoch": 0.21805703926710046,
      "grad_norm": 5.252994006042356,
      "learning_rate": 1.1978352645885913e-06,
      "loss": 0.469,
      "step": 842
    },
    {
      "epoch": 0.21831601437311837,
      "grad_norm": 5.41230067605829,
      "learning_rate": 1.1976089790380619e-06,
      "loss": 0.3918,
      "step": 843
    },
    {
      "epoch": 0.2185749894791363,
      "grad_norm": 6.237162379573823,
      "learning_rate": 1.197382225215477e-06,
      "loss": 0.4238,
      "step": 844
    },
    {
      "epoch": 0.21883396458515425,
      "grad_norm": 5.260596050836425,
      "learning_rate": 1.1971550033062737e-06,
      "loss": 0.437,
      "step": 845
    },
    {
      "epoch": 0.21883396458515425,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.8903225806451613,
      "eval_PRM F1 AUC": 0.8504452592980618,
      "eval_PRM F1 AUC (fixed)": 0.7082242011524359,
      "eval_PRM F1 Neg": 0.7017543859649122,
      "eval_PRM NPV": 0.5882352941176471,
      "eval_PRM Precision": 0.9583333333333334,
      "eval_PRM Recall": 0.8313253012048193,
      "eval_PRM Specificty": 0.8695652173913043,
      "eval_loss": 0.2731127440929413,
      "eval_runtime": 15.0799,
      "eval_samples_per_second": 2.918,
      "eval_steps_per_second": 0.133,
      "step": 845
    },
    {
      "epoch": 0.2190929396911722,
      "grad_norm": 3.8809428532933423,
      "learning_rate": 1.19692731349627e-06,
      "loss": 0.4406,
      "step": 846
    },
    {
      "epoch": 0.21935191479719013,
      "grad_norm": 10.17440956775982,
      "learning_rate": 1.1966991559716685e-06,
      "loss": 0.3781,
      "step": 847
    },
    {
      "epoch": 0.21961088990320807,
      "grad_norm": 8.224360422397567,
      "learning_rate": 1.1964705309190531e-06,
      "loss": 0.3955,
      "step": 848
    },
    {
      "epoch": 0.21986986500922598,
      "grad_norm": 6.917395590081431,
      "learning_rate": 1.1962414385253898e-06,
      "loss": 0.3605,
      "step": 849
    },
    {
      "epoch": 0.22012884011524392,
      "grad_norm": 5.291250727527,
      "learning_rate": 1.1960118789780282e-06,
      "loss": 0.3373,
      "step": 850
    },
    {
      "epoch": 0.22012884011524392,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.8903225806451613,
      "eval_PRM F1 AUC": 0.8504452592980618,
      "eval_PRM F1 AUC (fixed)": 0.7048192771084338,
      "eval_PRM F1 Neg": 0.7017543859649122,
      "eval_PRM NPV": 0.5882352941176471,
      "eval_PRM Precision": 0.9583333333333334,
      "eval_PRM Recall": 0.8313253012048193,
      "eval_PRM Specificty": 0.8695652173913043,
      "eval_loss": 0.2853865921497345,
      "eval_runtime": 14.6096,
      "eval_samples_per_second": 3.012,
      "eval_steps_per_second": 0.137,
      "step": 850
    },
    {
      "epoch": 0.22038781522126186,
      "grad_norm": 4.169023666456844,
      "learning_rate": 1.1957818524646987e-06,
      "loss": 0.4194,
      "step": 851
    },
    {
      "epoch": 0.2206467903272798,
      "grad_norm": 3.8927309085937742,
      "learning_rate": 1.1955513591735137e-06,
      "loss": 0.3856,
      "step": 852
    },
    {
      "epoch": 0.22090576543329773,
      "grad_norm": 4.616525144202818,
      "learning_rate": 1.195320399292968e-06,
      "loss": 0.3722,
      "step": 853
    },
    {
      "epoch": 0.22116474053931565,
      "grad_norm": 4.846970920598354,
      "learning_rate": 1.1950889730119374e-06,
      "loss": 0.4251,
      "step": 854
    },
    {
      "epoch": 0.22142371564533359,
      "grad_norm": 3.389416043470305,
      "learning_rate": 1.194857080519679e-06,
      "loss": 0.3592,
      "step": 855
    },
    {
      "epoch": 0.22142371564533359,
      "eval_PRM Accuracy": 0.8867924528301887,
      "eval_PRM F1": 0.925,
      "eval_PRM F1 AUC": 0.8805657412257727,
      "eval_PRM F1 AUC (fixed)": 0.7032477737035098,
      "eval_PRM F1 Neg": 0.7692307692307693,
      "eval_PRM NPV": 0.6896551724137931,
      "eval_PRM Precision": 0.961038961038961,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.8695652173913043,
      "eval_loss": 0.26908186078071594,
      "eval_runtime": 14.562,
      "eval_samples_per_second": 3.022,
      "eval_steps_per_second": 0.137,
      "step": 855
    },
    {
      "epoch": 0.22168269075135152,
      "grad_norm": 5.107553289534113,
      "learning_rate": 1.1946247220058315e-06,
      "loss": 0.3464,
      "step": 856
    },
    {
      "epoch": 0.22194166585736946,
      "grad_norm": 4.546089365240697,
      "learning_rate": 1.194391897660415e-06,
      "loss": 0.4066,
      "step": 857
    },
    {
      "epoch": 0.2222006409633874,
      "grad_norm": 3.0429013571459382,
      "learning_rate": 1.1941586076738298e-06,
      "loss": 0.3494,
      "step": 858
    },
    {
      "epoch": 0.22245961606940534,
      "grad_norm": 4.515604834735779,
      "learning_rate": 1.1939248522368574e-06,
      "loss": 0.3247,
      "step": 859
    },
    {
      "epoch": 0.22271859117542325,
      "grad_norm": 6.328796625019218,
      "learning_rate": 1.1936906315406602e-06,
      "loss": 0.4285,
      "step": 860
    },
    {
      "epoch": 0.22271859117542325,
      "eval_PRM Accuracy": 0.8773584905660378,
      "eval_PRM F1": 0.9202453987730062,
      "eval_PRM F1 AUC": 0.8431115767417497,
      "eval_PRM F1 AUC (fixed)": 0.7066526977475118,
      "eval_PRM F1 Neg": 0.7346938775510204,
      "eval_PRM NPV": 0.6923076923076923,
      "eval_PRM Precision": 0.9375,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.2668800354003906,
      "eval_runtime": 14.4032,
      "eval_samples_per_second": 3.055,
      "eval_steps_per_second": 0.139,
      "step": 860
    },
    {
      "epoch": 0.2229775662814412,
      "grad_norm": 4.003627406455415,
      "learning_rate": 1.1934559457767803e-06,
      "loss": 0.3303,
      "step": 861
    },
    {
      "epoch": 0.22323654138745913,
      "grad_norm": 2.9568343764262015,
      "learning_rate": 1.193220795137141e-06,
      "loss": 0.3514,
      "step": 862
    },
    {
      "epoch": 0.22349551649347707,
      "grad_norm": 8.797213234245561,
      "learning_rate": 1.1929851798140452e-06,
      "loss": 0.3842,
      "step": 863
    },
    {
      "epoch": 0.223754491599495,
      "grad_norm": 2.5474802896901565,
      "learning_rate": 1.1927491000001762e-06,
      "loss": 0.2886,
      "step": 864
    },
    {
      "epoch": 0.22401346670551292,
      "grad_norm": 3.3874524772324275,
      "learning_rate": 1.1925125558885966e-06,
      "loss": 0.3237,
      "step": 865
    },
    {
      "epoch": 0.22401346670551292,
      "eval_PRM Accuracy": 0.8867924528301887,
      "eval_PRM F1": 0.9259259259259259,
      "eval_PRM F1 AUC": 0.8648507071765322,
      "eval_PRM F1 AUC (fixed)": 0.7056050288108958,
      "eval_PRM F1 Neg": 0.76,
      "eval_PRM NPV": 0.7037037037037037,
      "eval_PRM Precision": 0.9493670886075949,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.8260869565217391,
      "eval_loss": 0.26945021748542786,
      "eval_runtime": 14.3503,
      "eval_samples_per_second": 3.066,
      "eval_steps_per_second": 0.139,
      "step": 865
    },
    {
      "epoch": 0.22427244181153086,
      "grad_norm": 6.980144378433547,
      "learning_rate": 1.1922755476727493e-06,
      "loss": 0.4359,
      "step": 866
    },
    {
      "epoch": 0.2245314169175488,
      "grad_norm": 3.675259629117122,
      "learning_rate": 1.1920380755464567e-06,
      "loss": 0.4484,
      "step": 867
    },
    {
      "epoch": 0.22479039202356674,
      "grad_norm": 3.970800178023882,
      "learning_rate": 1.19180013970392e-06,
      "loss": 0.4649,
      "step": 868
    },
    {
      "epoch": 0.22504936712958468,
      "grad_norm": 5.593595197945199,
      "learning_rate": 1.1915617403397198e-06,
      "loss": 0.406,
      "step": 869
    },
    {
      "epoch": 0.22530834223560262,
      "grad_norm": 3.4055613682905475,
      "learning_rate": 1.1913228776488164e-06,
      "loss": 0.3813,
      "step": 870
    },
    {
      "epoch": 0.22530834223560262,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.89171974522293,
      "eval_PRM F1 AUC": 0.8347302252488213,
      "eval_PRM F1 AUC (fixed)": 0.6998428496595076,
      "eval_PRM F1 Neg": 0.6909090909090909,
      "eval_PRM NPV": 0.59375,
      "eval_PRM Precision": 0.9459459459459459,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.8260869565217391,
      "eval_loss": 0.2825016975402832,
      "eval_runtime": 14.7056,
      "eval_samples_per_second": 2.992,
      "eval_steps_per_second": 0.136,
      "step": 870
    },
    {
      "epoch": 0.22556731734162053,
      "grad_norm": 4.766583139189498,
      "learning_rate": 1.1910835518265488e-06,
      "loss": 0.406,
      "step": 871
    },
    {
      "epoch": 0.22582629244763847,
      "grad_norm": 5.360701882443408,
      "learning_rate": 1.1908437630686337e-06,
      "loss": 0.2926,
      "step": 872
    },
    {
      "epoch": 0.2260852675536564,
      "grad_norm": 8.516057938457399,
      "learning_rate": 1.1906035115711675e-06,
      "loss": 0.3621,
      "step": 873
    },
    {
      "epoch": 0.22634424265967434,
      "grad_norm": 6.787723362329659,
      "learning_rate": 1.190362797530625e-06,
      "loss": 0.327,
      "step": 874
    },
    {
      "epoch": 0.22660321776569228,
      "grad_norm": 3.657201137384946,
      "learning_rate": 1.1901216211438586e-06,
      "loss": 0.3264,
      "step": 875
    },
    {
      "epoch": 0.22660321776569228,
      "eval_PRM Accuracy": 0.8490566037735849,
      "eval_PRM F1": 0.8974358974358975,
      "eval_PRM F1 AUC": 0.8564693556836039,
      "eval_PRM F1 AUC (fixed)": 0.6948664222105815,
      "eval_PRM F1 Neg": 0.7142857142857143,
      "eval_PRM NPV": 0.6060606060606061,
      "eval_PRM Precision": 0.958904109589041,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.8695652173913043,
      "eval_loss": 0.29068121314048767,
      "eval_runtime": 14.7585,
      "eval_samples_per_second": 2.981,
      "eval_steps_per_second": 0.136,
      "step": 875
    },
    {
      "epoch": 0.2268621928717102,
      "grad_norm": 4.078588375063669,
      "learning_rate": 1.1898799826080994e-06,
      "loss": 0.4119,
      "step": 876
    },
    {
      "epoch": 0.22712116797772813,
      "grad_norm": 3.2953125555811984,
      "learning_rate": 1.189637882120956e-06,
      "loss": 0.3044,
      "step": 877
    },
    {
      "epoch": 0.22738014308374607,
      "grad_norm": 7.182771434003031,
      "learning_rate": 1.1893953198804152e-06,
      "loss": 0.3758,
      "step": 878
    },
    {
      "epoch": 0.227639118189764,
      "grad_norm": 4.247871537462031,
      "learning_rate": 1.1891522960848409e-06,
      "loss": 0.3703,
      "step": 879
    },
    {
      "epoch": 0.22789809329578195,
      "grad_norm": 3.4826649057243224,
      "learning_rate": 1.1889088109329752e-06,
      "loss": 0.3483,
      "step": 880
    },
    {
      "epoch": 0.22789809329578195,
      "eval_PRM Accuracy": 0.8584905660377359,
      "eval_PRM F1": 0.9056603773584906,
      "eval_PRM F1 AUC": 0.8467784180199058,
      "eval_PRM F1 AUC (fixed)": 0.7103195390256679,
      "eval_PRM F1 Neg": 0.7169811320754716,
      "eval_PRM NPV": 0.6333333333333333,
      "eval_PRM Precision": 0.9473684210526315,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.8260869565217391,
      "eval_loss": 0.2728223502635956,
      "eval_runtime": 15.0548,
      "eval_samples_per_second": 2.923,
      "eval_steps_per_second": 0.133,
      "step": 880
    },
    {
      "epoch": 0.2281570684017999,
      "grad_norm": 5.741320223413004,
      "learning_rate": 1.1886648646239365e-06,
      "loss": 0.3574,
      "step": 881
    },
    {
      "epoch": 0.2284160435078178,
      "grad_norm": 4.062167022584372,
      "learning_rate": 1.188420457357221e-06,
      "loss": 0.4174,
      "step": 882
    },
    {
      "epoch": 0.22867501861383574,
      "grad_norm": 2.526453290531781,
      "learning_rate": 1.1881755893327022e-06,
      "loss": 0.317,
      "step": 883
    },
    {
      "epoch": 0.22893399371985368,
      "grad_norm": 3.0807541085440304,
      "learning_rate": 1.1879302607506295e-06,
      "loss": 0.3288,
      "step": 884
    },
    {
      "epoch": 0.22919296882587162,
      "grad_norm": 3.8308465141169763,
      "learning_rate": 1.1876844718116293e-06,
      "loss": 0.4456,
      "step": 885
    },
    {
      "epoch": 0.22919296882587162,
      "eval_PRM Accuracy": 0.8584905660377359,
      "eval_PRM F1": 0.9056603773584906,
      "eval_PRM F1 AUC": 0.8467784180199058,
      "eval_PRM F1 AUC (fixed)": 0.71634363541121,
      "eval_PRM F1 Neg": 0.7169811320754716,
      "eval_PRM NPV": 0.6333333333333333,
      "eval_PRM Precision": 0.9473684210526315,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.8260869565217391,
      "eval_loss": 0.26763397455215454,
      "eval_runtime": 14.9954,
      "eval_samples_per_second": 2.934,
      "eval_steps_per_second": 0.133,
      "step": 885
    },
    {
      "epoch": 0.22945194393188956,
      "grad_norm": 3.024120524976241,
      "learning_rate": 1.1874382227167049e-06,
      "loss": 0.3371,
      "step": 886
    },
    {
      "epoch": 0.22971091903790747,
      "grad_norm": 3.82411760033903,
      "learning_rate": 1.1871915136672352e-06,
      "loss": 0.4065,
      "step": 887
    },
    {
      "epoch": 0.2299698941439254,
      "grad_norm": 3.8886412588863,
      "learning_rate": 1.1869443448649758e-06,
      "loss": 0.269,
      "step": 888
    },
    {
      "epoch": 0.23022886924994335,
      "grad_norm": 4.071708730287361,
      "learning_rate": 1.1866967165120583e-06,
      "loss": 0.2922,
      "step": 889
    },
    {
      "epoch": 0.23048784435596129,
      "grad_norm": 5.072664995850144,
      "learning_rate": 1.1864486288109894e-06,
      "loss": 0.4196,
      "step": 890
    },
    {
      "epoch": 0.23048784435596129,
      "eval_PRM Accuracy": 0.8679245283018868,
      "eval_PRM F1": 0.9113924050632911,
      "eval_PRM F1 AUC": 0.8685175484546883,
      "eval_PRM F1 AUC (fixed)": 0.7239392352016762,
      "eval_PRM F1 Neg": 0.7407407407407407,
      "eval_PRM NPV": 0.6451612903225806,
      "eval_PRM Precision": 0.96,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.8695652173913043,
      "eval_loss": 0.2615525424480438,
      "eval_runtime": 14.7332,
      "eval_samples_per_second": 2.986,
      "eval_steps_per_second": 0.136,
      "step": 890
    },
    {
      "epoch": 0.23074681946197922,
      "grad_norm": 3.0759707900811657,
      "learning_rate": 1.186200081964652e-06,
      "loss": 0.3235,
      "step": 891
    },
    {
      "epoch": 0.23100579456799716,
      "grad_norm": 4.164881209108914,
      "learning_rate": 1.185951076176305e-06,
      "loss": 0.3374,
      "step": 892
    },
    {
      "epoch": 0.23126476967401507,
      "grad_norm": 4.746338390873242,
      "learning_rate": 1.1857016116495815e-06,
      "loss": 0.3589,
      "step": 893
    },
    {
      "epoch": 0.23152374478003301,
      "grad_norm": 4.619158986837625,
      "learning_rate": 1.1854516885884903e-06,
      "loss": 0.2584,
      "step": 894
    },
    {
      "epoch": 0.23178271988605095,
      "grad_norm": 3.02965756184151,
      "learning_rate": 1.1852013071974155e-06,
      "loss": 0.3908,
      "step": 895
    },
    {
      "epoch": 0.23178271988605095,
      "eval_PRM Accuracy": 0.8584905660377359,
      "eval_PRM F1": 0.9044585987261147,
      "eval_PRM F1 AUC": 0.8624934520691462,
      "eval_PRM F1 AUC (fixed)": 0.7202723939235202,
      "eval_PRM F1 Neg": 0.7272727272727273,
      "eval_PRM NPV": 0.625,
      "eval_PRM Precision": 0.9594594594594594,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.8695652173913043,
      "eval_loss": 0.2613525986671448,
      "eval_runtime": 14.5307,
      "eval_samples_per_second": 3.028,
      "eval_steps_per_second": 0.138,
      "step": 895
    },
    {
      "epoch": 0.2320416949920689,
      "grad_norm": 6.098313346871557,
      "learning_rate": 1.1849504676811156e-06,
      "loss": 0.3305,
      "step": 896
    },
    {
      "epoch": 0.23230067009808683,
      "grad_norm": 6.268537137431423,
      "learning_rate": 1.1846991702447233e-06,
      "loss": 0.3817,
      "step": 897
    },
    {
      "epoch": 0.23255964520410474,
      "grad_norm": 4.935579305034822,
      "learning_rate": 1.184447415093747e-06,
      "loss": 0.3596,
      "step": 898
    },
    {
      "epoch": 0.23281862031012268,
      "grad_norm": 4.467145665647315,
      "learning_rate": 1.1841952024340688e-06,
      "loss": 0.416,
      "step": 899
    },
    {
      "epoch": 0.23307759541614062,
      "grad_norm": 4.170804728572236,
      "learning_rate": 1.1839425324719442e-06,
      "loss": 0.3559,
      "step": 900
    },
    {
      "epoch": 0.23307759541614062,
      "eval_PRM Accuracy": 0.8490566037735849,
      "eval_PRM F1": 0.8987341772151899,
      "eval_PRM F1 AUC": 0.8407543216343635,
      "eval_PRM F1 AUC (fixed)": 0.71503404924044,
      "eval_PRM F1 Neg": 0.7037037037037037,
      "eval_PRM NPV": 0.6129032258064516,
      "eval_PRM Precision": 0.9466666666666667,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.8260869565217391,
      "eval_loss": 0.2619946002960205,
      "eval_runtime": 14.6743,
      "eval_samples_per_second": 2.998,
      "eval_steps_per_second": 0.136,
      "step": 900
    },
    {
      "epoch": 0.23333657052215856,
      "grad_norm": 3.561851713362729,
      "learning_rate": 1.183689405414004e-06,
      "loss": 0.37,
      "step": 901
    },
    {
      "epoch": 0.2335955456281765,
      "grad_norm": 5.391340666602356,
      "learning_rate": 1.183435821467252e-06,
      "loss": 0.2998,
      "step": 902
    },
    {
      "epoch": 0.23385452073419444,
      "grad_norm": 4.479966992834465,
      "learning_rate": 1.1831817808390655e-06,
      "loss": 0.4816,
      "step": 903
    },
    {
      "epoch": 0.23411349584021235,
      "grad_norm": 3.788280129003505,
      "learning_rate": 1.1829272837371958e-06,
      "loss": 0.4445,
      "step": 904
    },
    {
      "epoch": 0.2343724709462303,
      "grad_norm": 6.620416351536968,
      "learning_rate": 1.1826723303697673e-06,
      "loss": 0.4148,
      "step": 905
    },
    {
      "epoch": 0.2343724709462303,
      "eval_PRM Accuracy": 0.8584905660377359,
      "eval_PRM F1": 0.9056603773584906,
      "eval_PRM F1 AUC": 0.8467784180199058,
      "eval_PRM F1 AUC (fixed)": 0.7116291251964378,
      "eval_PRM F1 Neg": 0.7169811320754716,
      "eval_PRM NPV": 0.6333333333333333,
      "eval_PRM Precision": 0.9473684210526315,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.8260869565217391,
      "eval_loss": 0.2720625102519989,
      "eval_runtime": 15.08,
      "eval_samples_per_second": 2.918,
      "eval_steps_per_second": 0.133,
      "step": 905
    },
    {
      "epoch": 0.23463144605224823,
      "grad_norm": 3.4257473666984324,
      "learning_rate": 1.1824169209452774e-06,
      "loss": 0.3041,
      "step": 906
    },
    {
      "epoch": 0.23489042115826617,
      "grad_norm": 4.939739069415202,
      "learning_rate": 1.1821610556725966e-06,
      "loss": 0.435,
      "step": 907
    },
    {
      "epoch": 0.2351493962642841,
      "grad_norm": 5.255878093381102,
      "learning_rate": 1.181904734760968e-06,
      "loss": 0.4064,
      "step": 908
    },
    {
      "epoch": 0.23540837137030202,
      "grad_norm": 8.85032004547169,
      "learning_rate": 1.1816479584200078e-06,
      "loss": 0.3155,
      "step": 909
    },
    {
      "epoch": 0.23566734647631996,
      "grad_norm": 3.6772122279451294,
      "learning_rate": 1.1813907268597038e-06,
      "loss": 0.3787,
      "step": 910
    },
    {
      "epoch": 0.23566734647631996,
      "eval_PRM Accuracy": 0.8584905660377359,
      "eval_PRM F1": 0.9056603773584906,
      "eval_PRM F1 AUC": 0.8467784180199058,
      "eval_PRM F1 AUC (fixed)": 0.712676794133054,
      "eval_PRM F1 Neg": 0.7169811320754716,
      "eval_PRM NPV": 0.6333333333333333,
      "eval_PRM Precision": 0.9473684210526315,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.8260869565217391,
      "eval_loss": 0.280823677778244,
      "eval_runtime": 13.5068,
      "eval_samples_per_second": 3.258,
      "eval_steps_per_second": 0.148,
      "step": 910
    },
    {
      "epoch": 0.2359263215823379,
      "grad_norm": 4.360653413098666,
      "learning_rate": 1.1811330402904171e-06,
      "loss": 0.3677,
      "step": 911
    },
    {
      "epoch": 0.23618529668835583,
      "grad_norm": 3.384814109696639,
      "learning_rate": 1.18087489892288e-06,
      "loss": 0.313,
      "step": 912
    },
    {
      "epoch": 0.23644427179437377,
      "grad_norm": 3.0761177527594104,
      "learning_rate": 1.1806163029681972e-06,
      "loss": 0.2976,
      "step": 913
    },
    {
      "epoch": 0.2367032469003917,
      "grad_norm": 3.0945972922107345,
      "learning_rate": 1.1803572526378453e-06,
      "loss": 0.3056,
      "step": 914
    },
    {
      "epoch": 0.23696222200640962,
      "grad_norm": 7.625548103055213,
      "learning_rate": 1.1800977481436718e-06,
      "loss": 0.3014,
      "step": 915
    },
    {
      "epoch": 0.23696222200640962,
      "eval_PRM Accuracy": 0.8584905660377359,
      "eval_PRM F1": 0.9056603773584906,
      "eval_PRM F1 AUC": 0.8467784180199058,
      "eval_PRM F1 AUC (fixed)": 0.7210581456259822,
      "eval_PRM F1 Neg": 0.7169811320754716,
      "eval_PRM NPV": 0.6333333333333333,
      "eval_PRM Precision": 0.9473684210526315,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.8260869565217391,
      "eval_loss": 0.2804136276245117,
      "eval_runtime": 14.4877,
      "eval_samples_per_second": 3.037,
      "eval_steps_per_second": 0.138,
      "step": 915
    },
    {
      "epoch": 0.23722119711242756,
      "grad_norm": 4.106895071256707,
      "learning_rate": 1.1798377896978964e-06,
      "loss": 0.3447,
      "step": 916
    },
    {
      "epoch": 0.2374801722184455,
      "grad_norm": 3.4879165088232758,
      "learning_rate": 1.1795773775131094e-06,
      "loss": 0.3609,
      "step": 917
    },
    {
      "epoch": 0.23773914732446344,
      "grad_norm": 4.94351615997525,
      "learning_rate": 1.1793165118022725e-06,
      "loss": 0.464,
      "step": 918
    },
    {
      "epoch": 0.23799812243048138,
      "grad_norm": 9.395892112887031,
      "learning_rate": 1.1790551927787186e-06,
      "loss": 0.3857,
      "step": 919
    },
    {
      "epoch": 0.2382570975364993,
      "grad_norm": 4.4953539881083575,
      "learning_rate": 1.1787934206561506e-06,
      "loss": 0.3387,
      "step": 920
    },
    {
      "epoch": 0.2382570975364993,
      "eval_PRM Accuracy": 0.8490566037735849,
      "eval_PRM F1": 0.8987341772151899,
      "eval_PRM F1 AUC": 0.8407543216343635,
      "eval_PRM F1 AUC (fixed)": 0.7482975379779989,
      "eval_PRM F1 Neg": 0.7037037037037037,
      "eval_PRM NPV": 0.6129032258064516,
      "eval_PRM Precision": 0.9466666666666667,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.8260869565217391,
      "eval_loss": 0.2711288332939148,
      "eval_runtime": 14.5297,
      "eval_samples_per_second": 3.028,
      "eval_steps_per_second": 0.138,
      "step": 920
    },
    {
      "epoch": 0.23851607264251723,
      "grad_norm": 4.288113954542933,
      "learning_rate": 1.178531195648642e-06,
      "loss": 0.3936,
      "step": 921
    },
    {
      "epoch": 0.23877504774853517,
      "grad_norm": 4.977864839923052,
      "learning_rate": 1.1782685179706375e-06,
      "loss": 0.4172,
      "step": 922
    },
    {
      "epoch": 0.2390340228545531,
      "grad_norm": 6.67201876896054,
      "learning_rate": 1.1780053878369513e-06,
      "loss": 0.388,
      "step": 923
    },
    {
      "epoch": 0.23929299796057105,
      "grad_norm": 6.880680332320413,
      "learning_rate": 1.1777418054627677e-06,
      "loss": 0.3382,
      "step": 924
    },
    {
      "epoch": 0.23955197306658899,
      "grad_norm": 3.9508208402930105,
      "learning_rate": 1.1774777710636406e-06,
      "loss": 0.3966,
      "step": 925
    },
    {
      "epoch": 0.23955197306658899,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.8888888888888888,
      "eval_PRM F1 AUC": 0.8661602933473023,
      "eval_PRM F1 AUC (fixed)": 0.7482975379779989,
      "eval_PRM F1 Neg": 0.711864406779661,
      "eval_PRM NPV": 0.5833333333333334,
      "eval_PRM Precision": 0.9714285714285714,
      "eval_PRM Recall": 0.8192771084337349,
      "eval_PRM Specificty": 0.9130434782608695,
      "eval_loss": 0.2894640564918518,
      "eval_runtime": 14.5853,
      "eval_samples_per_second": 3.017,
      "eval_steps_per_second": 0.137,
      "step": 925
    },
    {
      "epoch": 0.2398109481726069,
      "grad_norm": 10.535536833190756,
      "learning_rate": 1.1772132848554942e-06,
      "loss": 0.3728,
      "step": 926
    },
    {
      "epoch": 0.24006992327862484,
      "grad_norm": 6.648969924884001,
      "learning_rate": 1.1769483470546216e-06,
      "loss": 0.3397,
      "step": 927
    },
    {
      "epoch": 0.24032889838464278,
      "grad_norm": 3.638658941745989,
      "learning_rate": 1.1766829578776858e-06,
      "loss": 0.3116,
      "step": 928
    },
    {
      "epoch": 0.24058787349066071,
      "grad_norm": 6.1916710161098525,
      "learning_rate": 1.176417117541718e-06,
      "loss": 0.3743,
      "step": 929
    },
    {
      "epoch": 0.24084684859667865,
      "grad_norm": 6.466301555599048,
      "learning_rate": 1.1761508262641195e-06,
      "loss": 0.432,
      "step": 930
    },
    {
      "epoch": 0.24084684859667865,
      "eval_PRM Accuracy": 0.8679245283018868,
      "eval_PRM F1": 0.9102564102564102,
      "eval_PRM F1 AUC": 0.8842325825039288,
      "eval_PRM F1 AUC (fixed)": 0.7551073860660031,
      "eval_PRM F1 Neg": 0.75,
      "eval_PRM NPV": 0.6363636363636364,
      "eval_PRM Precision": 0.9726027397260274,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.9130434782608695,
      "eval_loss": 0.26407501101493835,
      "eval_runtime": 14.429,
      "eval_samples_per_second": 3.049,
      "eval_steps_per_second": 0.139,
      "step": 930
    },
    {
      "epoch": 0.24110582370269656,
      "grad_norm": 3.5072923022044473,
      "learning_rate": 1.1758840842626593e-06,
      "loss": 0.3742,
      "step": 931
    },
    {
      "epoch": 0.2413647988087145,
      "grad_norm": 4.137152539221647,
      "learning_rate": 1.1756168917554759e-06,
      "loss": 0.3656,
      "step": 932
    },
    {
      "epoch": 0.24162377391473244,
      "grad_norm": 3.7639799069758637,
      "learning_rate": 1.1753492489610755e-06,
      "loss": 0.3303,
      "step": 933
    },
    {
      "epoch": 0.24188274902075038,
      "grad_norm": 2.896848508314599,
      "learning_rate": 1.175081156098333e-06,
      "loss": 0.3401,
      "step": 934
    },
    {
      "epoch": 0.24214172412676832,
      "grad_norm": 9.300379803883216,
      "learning_rate": 1.1748126133864908e-06,
      "loss": 0.4256,
      "step": 935
    },
    {
      "epoch": 0.24214172412676832,
      "eval_PRM Accuracy": 0.8490566037735849,
      "eval_PRM F1": 0.8987341772151899,
      "eval_PRM F1 AUC": 0.8407543216343635,
      "eval_PRM F1 AUC (fixed)": 0.7503928758512309,
      "eval_PRM F1 Neg": 0.7037037037037037,
      "eval_PRM NPV": 0.6129032258064516,
      "eval_PRM Precision": 0.9466666666666667,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.8260869565217391,
      "eval_loss": 0.25802791118621826,
      "eval_runtime": 14.7761,
      "eval_samples_per_second": 2.978,
      "eval_steps_per_second": 0.135,
      "step": 935
    },
    {
      "epoch": 0.24240069923278626,
      "grad_norm": 6.945735757870045,
      "learning_rate": 1.1745436210451603e-06,
      "loss": 0.403,
      "step": 936
    },
    {
      "epoch": 0.24265967433880417,
      "grad_norm": 5.168673525936932,
      "learning_rate": 1.1742741792943193e-06,
      "loss": 0.3643,
      "step": 937
    },
    {
      "epoch": 0.2429186494448221,
      "grad_norm": 3.9301017982836597,
      "learning_rate": 1.1740042883543139e-06,
      "loss": 0.2894,
      "step": 938
    },
    {
      "epoch": 0.24317762455084005,
      "grad_norm": 3.8086762466056783,
      "learning_rate": 1.1737339484458574e-06,
      "loss": 0.3478,
      "step": 939
    },
    {
      "epoch": 0.243436599656858,
      "grad_norm": 10.598569677321471,
      "learning_rate": 1.1734631597900301e-06,
      "loss": 0.4391,
      "step": 940
    },
    {
      "epoch": 0.243436599656858,
      "eval_PRM Accuracy": 0.8301886792452831,
      "eval_PRM F1": 0.881578947368421,
      "eval_PRM F1 AUC": 0.8601361969617601,
      "eval_PRM F1 AUC (fixed)": 0.7454164484023048,
      "eval_PRM F1 Neg": 0.7,
      "eval_PRM NPV": 0.5675675675675675,
      "eval_PRM Precision": 0.9710144927536232,
      "eval_PRM Recall": 0.8072289156626506,
      "eval_PRM Specificty": 0.9130434782608695,
      "eval_loss": 0.26957109570503235,
      "eval_runtime": 14.7665,
      "eval_samples_per_second": 2.98,
      "eval_steps_per_second": 0.135,
      "step": 940
    },
    {
      "epoch": 0.24369557476287593,
      "grad_norm": 5.764724413638612,
      "learning_rate": 1.1731919226082793e-06,
      "loss": 0.365,
      "step": 941
    },
    {
      "epoch": 0.24395454986889387,
      "grad_norm": 5.9502788053858895,
      "learning_rate": 1.1729202371224194e-06,
      "loss": 0.3258,
      "step": 942
    },
    {
      "epoch": 0.24421352497491178,
      "grad_norm": 6.031412246354583,
      "learning_rate": 1.1726481035546311e-06,
      "loss": 0.3288,
      "step": 943
    },
    {
      "epoch": 0.24447250008092972,
      "grad_norm": 3.945656622656256,
      "learning_rate": 1.1723755221274619e-06,
      "loss": 0.28,
      "step": 944
    },
    {
      "epoch": 0.24473147518694766,
      "grad_norm": 3.3620655142691787,
      "learning_rate": 1.1721024930638248e-06,
      "loss": 0.34,
      "step": 945
    },
    {
      "epoch": 0.24473147518694766,
      "eval_PRM Accuracy": 0.8490566037735849,
      "eval_PRM F1": 0.8974358974358975,
      "eval_PRM F1 AUC": 0.8564693556836039,
      "eval_PRM F1 AUC (fixed)": 0.7407019381875327,
      "eval_PRM F1 Neg": 0.7142857142857143,
      "eval_PRM NPV": 0.6060606060606061,
      "eval_PRM Precision": 0.958904109589041,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.8695652173913043,
      "eval_loss": 0.2563490867614746,
      "eval_runtime": 14.652,
      "eval_samples_per_second": 3.003,
      "eval_steps_per_second": 0.136,
      "step": 945
    },
    {
      "epoch": 0.2449904502929656,
      "grad_norm": 3.5293565846531973,
      "learning_rate": 1.1718290165869995e-06,
      "loss": 0.3143,
      "step": 946
    },
    {
      "epoch": 0.24524942539898353,
      "grad_norm": 3.4890087445954756,
      "learning_rate": 1.1715550929206317e-06,
      "loss": 0.3351,
      "step": 947
    },
    {
      "epoch": 0.24550840050500145,
      "grad_norm": 2.986678147022999,
      "learning_rate": 1.1712807222887324e-06,
      "loss": 0.3392,
      "step": 948
    },
    {
      "epoch": 0.24576737561101938,
      "grad_norm": 11.359804178521635,
      "learning_rate": 1.1710059049156783e-06,
      "loss": 0.4761,
      "step": 949
    },
    {
      "epoch": 0.24602635071703732,
      "grad_norm": 10.1120871833462,
      "learning_rate": 1.1707306410262112e-06,
      "loss": 0.4559,
      "step": 950
    },
    {
      "epoch": 0.24602635071703732,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.8930817610062893,
      "eval_PRM F1 AUC": 0.819015191199581,
      "eval_PRM F1 AUC (fixed)": 0.7393923520167627,
      "eval_PRM F1 Neg": 0.6792452830188679,
      "eval_PRM NPV": 0.6,
      "eval_PRM Precision": 0.9342105263157895,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.2529611885547638,
      "eval_runtime": 14.6632,
      "eval_samples_per_second": 3.001,
      "eval_steps_per_second": 0.136,
      "step": 950
    },
    {
      "epoch": 0.24628532582305526,
      "grad_norm": 6.21559932436371,
      "learning_rate": 1.1704549308454385e-06,
      "loss": 0.3531,
      "step": 951
    },
    {
      "epoch": 0.2465443009290732,
      "grad_norm": 4.139390120568164,
      "learning_rate": 1.1701787745988325e-06,
      "loss": 0.4025,
      "step": 952
    },
    {
      "epoch": 0.24680327603509114,
      "grad_norm": 4.066293812063238,
      "learning_rate": 1.1699021725122297e-06,
      "loss": 0.3229,
      "step": 953
    },
    {
      "epoch": 0.24706225114110905,
      "grad_norm": 2.899098952795315,
      "learning_rate": 1.169625124811832e-06,
      "loss": 0.3156,
      "step": 954
    },
    {
      "epoch": 0.247321226247127,
      "grad_norm": 2.8517635816301694,
      "learning_rate": 1.169347631724205e-06,
      "loss": 0.3273,
      "step": 955
    },
    {
      "epoch": 0.247321226247127,
      "eval_PRM Accuracy": 0.8490566037735849,
      "eval_PRM F1": 0.8974358974358975,
      "eval_PRM F1 AUC": 0.8564693556836039,
      "eval_PRM F1 AUC (fixed)": 0.7333682556312205,
      "eval_PRM F1 Neg": 0.7142857142857143,
      "eval_PRM NPV": 0.6060606060606061,
      "eval_PRM Precision": 0.958904109589041,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.8695652173913043,
      "eval_loss": 0.2633054554462433,
      "eval_runtime": 14.3335,
      "eval_samples_per_second": 3.07,
      "eval_steps_per_second": 0.14,
      "step": 955
    },
    {
      "epoch": 0.24758020135314493,
      "grad_norm": 3.3564577121473618,
      "learning_rate": 1.1690696934762794e-06,
      "loss": 0.3227,
      "step": 956
    },
    {
      "epoch": 0.24783917645916287,
      "grad_norm": 3.8662849481785635,
      "learning_rate": 1.168791310295349e-06,
      "loss": 0.357,
      "step": 957
    },
    {
      "epoch": 0.2480981515651808,
      "grad_norm": 3.5997682609517634,
      "learning_rate": 1.1685124824090719e-06,
      "loss": 0.417,
      "step": 958
    },
    {
      "epoch": 0.24835712667119872,
      "grad_norm": 2.7034258306479293,
      "learning_rate": 1.1682332100454699e-06,
      "loss": 0.3167,
      "step": 959
    },
    {
      "epoch": 0.24861610177721666,
      "grad_norm": 4.988676707709703,
      "learning_rate": 1.1679534934329283e-06,
      "loss": 0.2986,
      "step": 960
    },
    {
      "epoch": 0.24861610177721666,
      "eval_PRM Accuracy": 0.8490566037735849,
      "eval_PRM F1": 0.9,
      "eval_PRM F1 AUC": 0.8250392875851232,
      "eval_PRM F1 AUC (fixed)": 0.7354635935044526,
      "eval_PRM F1 Neg": 0.6923076923076923,
      "eval_PRM NPV": 0.6206896551724138,
      "eval_PRM Precision": 0.935064935064935,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.26185938715934753,
      "eval_runtime": 14.5293,
      "eval_samples_per_second": 3.028,
      "eval_steps_per_second": 0.138,
      "step": 960
    },
    {
      "epoch": 0.2488750768832346,
      "grad_norm": 5.29107206320586,
      "learning_rate": 1.167673332800196e-06,
      "loss": 0.4215,
      "step": 961
    },
    {
      "epoch": 0.24913405198925254,
      "grad_norm": 5.437935849441125,
      "learning_rate": 1.167392728376384e-06,
      "loss": 0.4476,
      "step": 962
    },
    {
      "epoch": 0.24939302709527048,
      "grad_norm": 3.8839556426836372,
      "learning_rate": 1.1671116803909674e-06,
      "loss": 0.3246,
      "step": 963
    },
    {
      "epoch": 0.24965200220128841,
      "grad_norm": 10.37270496386567,
      "learning_rate": 1.1668301890737833e-06,
      "loss": 0.4899,
      "step": 964
    },
    {
      "epoch": 0.24991097730730633,
      "grad_norm": 5.908063708001343,
      "learning_rate": 1.1665482546550314e-06,
      "loss": 0.3955,
      "step": 965
    },
    {
      "epoch": 0.24991097730730633,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.89171974522293,
      "eval_PRM F1 AUC": 0.8347302252488213,
      "eval_PRM F1 AUC (fixed)": 0.7260345730749084,
      "eval_PRM F1 Neg": 0.6909090909090909,
      "eval_PRM NPV": 0.59375,
      "eval_PRM Precision": 0.9459459459459459,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.8260869565217391,
      "eval_loss": 0.2727225124835968,
      "eval_runtime": 14.3383,
      "eval_samples_per_second": 3.069,
      "eval_steps_per_second": 0.139,
      "step": 965
    },
    {
      "epoch": 0.2501699524133243,
      "grad_norm": 5.6447939854256735,
      "learning_rate": 1.1662658773652744e-06,
      "loss": 0.4093,
      "step": 966
    },
    {
      "epoch": 0.2504289275193422,
      "grad_norm": 2.850052957391009,
      "learning_rate": 1.1659830574354363e-06,
      "loss": 0.3873,
      "step": 967
    },
    {
      "epoch": 0.2506879026253601,
      "grad_norm": 4.787640638975102,
      "learning_rate": 1.1656997950968037e-06,
      "loss": 0.4238,
      "step": 968
    },
    {
      "epoch": 0.2509468777313781,
      "grad_norm": 5.680659064340013,
      "learning_rate": 1.1654160905810246e-06,
      "loss": 0.4614,
      "step": 969
    },
    {
      "epoch": 0.251205852837396,
      "grad_norm": 3.3362762459579023,
      "learning_rate": 1.1651319441201094e-06,
      "loss": 0.344,
      "step": 970
    },
    {
      "epoch": 0.251205852837396,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.89171974522293,
      "eval_PRM F1 AUC": 0.8347302252488213,
      "eval_PRM F1 AUC (fixed)": 0.716081718177056,
      "eval_PRM F1 Neg": 0.6909090909090909,
      "eval_PRM NPV": 0.59375,
      "eval_PRM Precision": 0.9459459459459459,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.8260869565217391,
      "eval_loss": 0.28485599160194397,
      "eval_runtime": 14.3151,
      "eval_samples_per_second": 3.074,
      "eval_steps_per_second": 0.14,
      "step": 970
    },
    {
      "epoch": 0.25146482794341396,
      "grad_norm": 5.758747112832516,
      "learning_rate": 1.1648473559464287e-06,
      "loss": 0.3374,
      "step": 971
    },
    {
      "epoch": 0.25172380304943187,
      "grad_norm": 8.58153415473367,
      "learning_rate": 1.164562326292715e-06,
      "loss": 0.538,
      "step": 972
    },
    {
      "epoch": 0.2519827781554498,
      "grad_norm": 5.950271313129463,
      "learning_rate": 1.1642768553920621e-06,
      "loss": 0.3176,
      "step": 973
    },
    {
      "epoch": 0.25224175326146775,
      "grad_norm": 3.1460089924178947,
      "learning_rate": 1.163990943477924e-06,
      "loss": 0.3241,
      "step": 974
    },
    {
      "epoch": 0.25250072836748566,
      "grad_norm": 6.985297468419992,
      "learning_rate": 1.163704590784116e-06,
      "loss": 0.3705,
      "step": 975
    },
    {
      "epoch": 0.25250072836748566,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.89171974522293,
      "eval_PRM F1 AUC": 0.8347302252488213,
      "eval_PRM F1 AUC (fixed)": 0.7087480356207437,
      "eval_PRM F1 Neg": 0.6909090909090909,
      "eval_PRM NPV": 0.59375,
      "eval_PRM Precision": 0.9459459459459459,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.8260869565217391,
      "eval_loss": 0.27933061122894287,
      "eval_runtime": 14.5344,
      "eval_samples_per_second": 3.027,
      "eval_steps_per_second": 0.138,
      "step": 975
    },
    {
      "epoch": 0.2527597034735036,
      "grad_norm": 8.012879551894242,
      "learning_rate": 1.1634177975448133e-06,
      "loss": 0.3826,
      "step": 976
    },
    {
      "epoch": 0.25301867857952154,
      "grad_norm": 7.96887475220341,
      "learning_rate": 1.1631305639945517e-06,
      "loss": 0.3364,
      "step": 977
    },
    {
      "epoch": 0.25327765368553945,
      "grad_norm": 3.612692739178533,
      "learning_rate": 1.162842890368227e-06,
      "loss": 0.4073,
      "step": 978
    },
    {
      "epoch": 0.2535366287915574,
      "grad_norm": 7.720990027200896,
      "learning_rate": 1.162554776901095e-06,
      "loss": 0.3994,
      "step": 979
    },
    {
      "epoch": 0.25379560389757533,
      "grad_norm": 2.783639066391506,
      "learning_rate": 1.1622662238287708e-06,
      "loss": 0.3031,
      "step": 980
    },
    {
      "epoch": 0.25379560389757533,
      "eval_PRM Accuracy": 0.8584905660377359,
      "eval_PRM F1": 0.9056603773584906,
      "eval_PRM F1 AUC": 0.8467784180199058,
      "eval_PRM F1 AUC (fixed)": 0.7145102147721321,
      "eval_PRM F1 Neg": 0.7169811320754716,
      "eval_PRM NPV": 0.6333333333333333,
      "eval_PRM Precision": 0.9473684210526315,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.8260869565217391,
      "eval_loss": 0.27657219767570496,
      "eval_runtime": 14.9306,
      "eval_samples_per_second": 2.947,
      "eval_steps_per_second": 0.134,
      "step": 980
    },
    {
      "epoch": 0.2540545790035933,
      "grad_norm": 3.4357231189414783,
      "learning_rate": 1.1619772313872295e-06,
      "loss": 0.3553,
      "step": 981
    },
    {
      "epoch": 0.2543135541096112,
      "grad_norm": 4.935806590676655,
      "learning_rate": 1.1616877998128053e-06,
      "loss": 0.3536,
      "step": 982
    },
    {
      "epoch": 0.2545725292156292,
      "grad_norm": 3.193095180565718,
      "learning_rate": 1.1613979293421918e-06,
      "loss": 0.2857,
      "step": 983
    },
    {
      "epoch": 0.2548315043216471,
      "grad_norm": 5.53188970528004,
      "learning_rate": 1.1611076202124404e-06,
      "loss": 0.3127,
      "step": 984
    },
    {
      "epoch": 0.255090479427665,
      "grad_norm": 3.229345173101998,
      "learning_rate": 1.160816872660963e-06,
      "loss": 0.4002,
      "step": 985
    },
    {
      "epoch": 0.255090479427665,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.89171974522293,
      "eval_PRM F1 AUC": 0.8347302252488213,
      "eval_PRM F1 AUC (fixed)": 0.7160817181770561,
      "eval_PRM F1 Neg": 0.6909090909090909,
      "eval_PRM NPV": 0.59375,
      "eval_PRM Precision": 0.9459459459459459,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.8260869565217391,
      "eval_loss": 0.27515214681625366,
      "eval_runtime": 14.4985,
      "eval_samples_per_second": 3.035,
      "eval_steps_per_second": 0.138,
      "step": 985
    },
    {
      "epoch": 0.25534945453368296,
      "grad_norm": 4.293361296415176,
      "learning_rate": 1.1605256869255285e-06,
      "loss": 0.3489,
      "step": 986
    },
    {
      "epoch": 0.2556084296397009,
      "grad_norm": 4.129626468223985,
      "learning_rate": 1.1602340632442652e-06,
      "loss": 0.3799,
      "step": 987
    },
    {
      "epoch": 0.25586740474571884,
      "grad_norm": 9.096540369766284,
      "learning_rate": 1.1599420018556586e-06,
      "loss": 0.4538,
      "step": 988
    },
    {
      "epoch": 0.25612637985173675,
      "grad_norm": 5.247960343985199,
      "learning_rate": 1.159649502998553e-06,
      "loss": 0.3674,
      "step": 989
    },
    {
      "epoch": 0.25638535495775466,
      "grad_norm": 7.994078985517792,
      "learning_rate": 1.1593565669121497e-06,
      "loss": 0.3333,
      "step": 990
    },
    {
      "epoch": 0.25638535495775466,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.8903225806451613,
      "eval_PRM F1 AUC": 0.8504452592980618,
      "eval_PRM F1 AUC (fixed)": 0.7239392352016761,
      "eval_PRM F1 Neg": 0.7017543859649122,
      "eval_PRM NPV": 0.5882352941176471,
      "eval_PRM Precision": 0.9583333333333334,
      "eval_PRM Recall": 0.8313253012048193,
      "eval_PRM Specificty": 0.8695652173913043,
      "eval_loss": 0.287078857421875,
      "eval_runtime": 14.6202,
      "eval_samples_per_second": 3.01,
      "eval_steps_per_second": 0.137,
      "step": 990
    },
    {
      "epoch": 0.25664433006377263,
      "grad_norm": 8.594303185516962,
      "learning_rate": 1.1590631938360086e-06,
      "loss": 0.3525,
      "step": 991
    },
    {
      "epoch": 0.25690330516979054,
      "grad_norm": 7.048469269863221,
      "learning_rate": 1.158769384010046e-06,
      "loss": 0.3893,
      "step": 992
    },
    {
      "epoch": 0.2571622802758085,
      "grad_norm": 13.317768952534582,
      "learning_rate": 1.1584751376745355e-06,
      "loss": 0.3186,
      "step": 993
    },
    {
      "epoch": 0.2574212553818264,
      "grad_norm": 5.726905993306342,
      "learning_rate": 1.1581804550701078e-06,
      "loss": 0.3263,
      "step": 994
    },
    {
      "epoch": 0.25768023048784433,
      "grad_norm": 4.893574502139262,
      "learning_rate": 1.157885336437751e-06,
      "loss": 0.4972,
      "step": 995
    },
    {
      "epoch": 0.25768023048784433,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.89171974522293,
      "eval_PRM F1 AUC": 0.8347302252488213,
      "eval_PRM F1 AUC (fixed)": 0.7409638554216867,
      "eval_PRM F1 Neg": 0.6909090909090909,
      "eval_PRM NPV": 0.59375,
      "eval_PRM Precision": 0.9459459459459459,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.8260869565217391,
      "eval_loss": 0.2642107307910919,
      "eval_runtime": 14.5194,
      "eval_samples_per_second": 3.03,
      "eval_steps_per_second": 0.138,
      "step": 995
    },
    {
      "epoch": 0.2579392055938623,
      "grad_norm": 3.5760726549084616,
      "learning_rate": 1.1575897820188087e-06,
      "loss": 0.3395,
      "step": 996
    },
    {
      "epoch": 0.2581981806998802,
      "grad_norm": 4.246352890964505,
      "learning_rate": 1.1572937920549816e-06,
      "loss": 0.4059,
      "step": 997
    },
    {
      "epoch": 0.2584571558058982,
      "grad_norm": 5.18629173452879,
      "learning_rate": 1.1569973667883264e-06,
      "loss": 0.3453,
      "step": 998
    },
    {
      "epoch": 0.2587161309119161,
      "grad_norm": 3.3078206651369015,
      "learning_rate": 1.1567005064612558e-06,
      "loss": 0.3284,
      "step": 999
    },
    {
      "epoch": 0.25897510601793405,
      "grad_norm": 3.0615848926933,
      "learning_rate": 1.156403211316538e-06,
      "loss": 0.2652,
      "step": 1000
    },
    {
      "epoch": 0.25897510601793405,
      "eval_PRM Accuracy": 0.8584905660377359,
      "eval_PRM F1": 0.9056603773584906,
      "eval_PRM F1 AUC": 0.8467784180199058,
      "eval_PRM F1 AUC (fixed)": 0.7443687794656888,
      "eval_PRM F1 Neg": 0.7169811320754716,
      "eval_PRM NPV": 0.6333333333333333,
      "eval_PRM Precision": 0.9473684210526315,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.8260869565217391,
      "eval_loss": 0.25816041231155396,
      "eval_runtime": 14.3987,
      "eval_samples_per_second": 3.056,
      "eval_steps_per_second": 0.139,
      "step": 1000
    },
    {
      "epoch": 0.25923408112395196,
      "grad_norm": 8.62964960291262,
      "learning_rate": 1.1561054815972973e-06,
      "loss": 0.4371,
      "step": 1001
    },
    {
      "epoch": 0.2594930562299699,
      "grad_norm": 3.752764520211669,
      "learning_rate": 1.155807317547013e-06,
      "loss": 0.3212,
      "step": 1002
    },
    {
      "epoch": 0.25975203133598784,
      "grad_norm": 11.589612656246816,
      "learning_rate": 1.1555087194095199e-06,
      "loss": 0.3769,
      "step": 1003
    },
    {
      "epoch": 0.26001100644200575,
      "grad_norm": 3.8830914204755986,
      "learning_rate": 1.1552096874290074e-06,
      "loss": 0.3566,
      "step": 1004
    },
    {
      "epoch": 0.2602699815480237,
      "grad_norm": 4.374775346676012,
      "learning_rate": 1.1549102218500199e-06,
      "loss": 0.3932,
      "step": 1005
    },
    {
      "epoch": 0.2602699815480237,
      "eval_PRM Accuracy": 0.8584905660377359,
      "eval_PRM F1": 0.9056603773584906,
      "eval_PRM F1 AUC": 0.8467784180199058,
      "eval_PRM F1 AUC (fixed)": 0.7412257726558408,
      "eval_PRM F1 Neg": 0.7169811320754716,
      "eval_PRM NPV": 0.6333333333333333,
      "eval_PRM Precision": 0.9473684210526315,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.8260869565217391,
      "eval_loss": 0.2624988853931427,
      "eval_runtime": 14.6088,
      "eval_samples_per_second": 3.012,
      "eval_steps_per_second": 0.137,
      "step": 1005
    },
    {
      "epoch": 0.26052895665404163,
      "grad_norm": 6.288991137216826,
      "learning_rate": 1.1546103229174566e-06,
      "loss": 0.3921,
      "step": 1006
    },
    {
      "epoch": 0.26078793176005954,
      "grad_norm": 3.4011240081976823,
      "learning_rate": 1.1543099908765706e-06,
      "loss": 0.3222,
      "step": 1007
    },
    {
      "epoch": 0.2610469068660775,
      "grad_norm": 8.545602163381849,
      "learning_rate": 1.1540092259729698e-06,
      "loss": 0.3048,
      "step": 1008
    },
    {
      "epoch": 0.2613058819720954,
      "grad_norm": 3.5424382509880132,
      "learning_rate": 1.1537080284526156e-06,
      "loss": 0.3191,
      "step": 1009
    },
    {
      "epoch": 0.2615648570781134,
      "grad_norm": 7.447874599354588,
      "learning_rate": 1.1534063985618231e-06,
      "loss": 0.3002,
      "step": 1010
    },
    {
      "epoch": 0.2615648570781134,
      "eval_PRM Accuracy": 0.8490566037735849,
      "eval_PRM F1": 0.8987341772151899,
      "eval_PRM F1 AUC": 0.8407543216343635,
      "eval_PRM F1 AUC (fixed)": 0.7391304347826088,
      "eval_PRM F1 Neg": 0.7037037037037037,
      "eval_PRM NPV": 0.6129032258064516,
      "eval_PRM Precision": 0.9466666666666667,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.8260869565217391,
      "eval_loss": 0.2732199430465698,
      "eval_runtime": 14.6566,
      "eval_samples_per_second": 3.002,
      "eval_steps_per_second": 0.136,
      "step": 1010
    },
    {
      "epoch": 0.2618238321841313,
      "grad_norm": 4.934117068568929,
      "learning_rate": 1.153104336547262e-06,
      "loss": 0.4552,
      "step": 1011
    },
    {
      "epoch": 0.2620828072901492,
      "grad_norm": 5.94232166921095,
      "learning_rate": 1.1528018426559534e-06,
      "loss": 0.3295,
      "step": 1012
    },
    {
      "epoch": 0.2623417823961672,
      "grad_norm": 4.68230926071834,
      "learning_rate": 1.152498917135274e-06,
      "loss": 0.3659,
      "step": 1013
    },
    {
      "epoch": 0.2626007575021851,
      "grad_norm": 3.9028364151194155,
      "learning_rate": 1.1521955602329516e-06,
      "loss": 0.375,
      "step": 1014
    },
    {
      "epoch": 0.26285973260820306,
      "grad_norm": 3.820941550762621,
      "learning_rate": 1.1518917721970679e-06,
      "loss": 0.387,
      "step": 1015
    },
    {
      "epoch": 0.26285973260820306,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.89171974522293,
      "eval_PRM F1 AUC": 0.8347302252488213,
      "eval_PRM F1 AUC (fixed)": 0.7252488213724463,
      "eval_PRM F1 Neg": 0.6909090909090909,
      "eval_PRM NPV": 0.59375,
      "eval_PRM Precision": 0.9459459459459459,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.8260869565217391,
      "eval_loss": 0.28247568011283875,
      "eval_runtime": 15.3011,
      "eval_samples_per_second": 2.876,
      "eval_steps_per_second": 0.131,
      "step": 1015
    },
    {
      "epoch": 0.26311870771422097,
      "grad_norm": 6.272094131495291,
      "learning_rate": 1.1515875532760563e-06,
      "loss": 0.4004,
      "step": 1016
    },
    {
      "epoch": 0.2633776828202389,
      "grad_norm": 9.61148233057241,
      "learning_rate": 1.1512829037187032e-06,
      "loss": 0.3623,
      "step": 1017
    },
    {
      "epoch": 0.26363665792625685,
      "grad_norm": 3.5927848620945237,
      "learning_rate": 1.150977823774147e-06,
      "loss": 0.3681,
      "step": 1018
    },
    {
      "epoch": 0.26389563303227476,
      "grad_norm": 4.669022408853148,
      "learning_rate": 1.150672313691878e-06,
      "loss": 0.329,
      "step": 1019
    },
    {
      "epoch": 0.2641546081382927,
      "grad_norm": 4.151941292781019,
      "learning_rate": 1.1503663737217385e-06,
      "loss": 0.3425,
      "step": 1020
    },
    {
      "epoch": 0.2641546081382927,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.89171974522293,
      "eval_PRM F1 AUC": 0.8347302252488213,
      "eval_PRM F1 AUC (fixed)": 0.7155578837087481,
      "eval_PRM F1 Neg": 0.6909090909090909,
      "eval_PRM NPV": 0.59375,
      "eval_PRM Precision": 0.9459459459459459,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.8260869565217391,
      "eval_loss": 0.29420801997184753,
      "eval_runtime": 14.6768,
      "eval_samples_per_second": 2.998,
      "eval_steps_per_second": 0.136,
      "step": 1020
    },
    {
      "epoch": 0.26441358324431063,
      "grad_norm": 3.1002475083020897,
      "learning_rate": 1.150060004113922e-06,
      "loss": 0.3625,
      "step": 1021
    },
    {
      "epoch": 0.2646725583503286,
      "grad_norm": 7.00755462604261,
      "learning_rate": 1.1497532051189734e-06,
      "loss": 0.3134,
      "step": 1022
    },
    {
      "epoch": 0.2649315334563465,
      "grad_norm": 3.205399859551509,
      "learning_rate": 1.1494459769877892e-06,
      "loss": 0.4001,
      "step": 1023
    },
    {
      "epoch": 0.2651905085623644,
      "grad_norm": 4.163906104739272,
      "learning_rate": 1.1491383199716162e-06,
      "loss": 0.3586,
      "step": 1024
    },
    {
      "epoch": 0.2654494836683824,
      "grad_norm": 3.3459054714717604,
      "learning_rate": 1.1488302343220522e-06,
      "loss": 0.3757,
      "step": 1025
    },
    {
      "epoch": 0.2654494836683824,
      "eval_PRM Accuracy": 0.8584905660377359,
      "eval_PRM F1": 0.9056603773584906,
      "eval_PRM F1 AUC": 0.8467784180199058,
      "eval_PRM F1 AUC (fixed)": 0.7097957045573599,
      "eval_PRM F1 Neg": 0.7169811320754716,
      "eval_PRM NPV": 0.6333333333333333,
      "eval_PRM Precision": 0.9473684210526315,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.8260869565217391,
      "eval_loss": 0.2874738276004791,
      "eval_runtime": 14.7665,
      "eval_samples_per_second": 2.98,
      "eval_steps_per_second": 0.135,
      "step": 1025
    },
    {
      "epoch": 0.2657084587744003,
      "grad_norm": 3.062475684970944,
      "learning_rate": 1.148521720291046e-06,
      "loss": 0.3169,
      "step": 1026
    },
    {
      "epoch": 0.26596743388041827,
      "grad_norm": 5.559405808823225,
      "learning_rate": 1.1482127781308959e-06,
      "loss": 0.3497,
      "step": 1027
    },
    {
      "epoch": 0.2662264089864362,
      "grad_norm": 2.8845109010736802,
      "learning_rate": 1.1479034080942506e-06,
      "loss": 0.2926,
      "step": 1028
    },
    {
      "epoch": 0.2664853840924541,
      "grad_norm": 2.7290208271682483,
      "learning_rate": 1.1475936104341092e-06,
      "loss": 0.2948,
      "step": 1029
    },
    {
      "epoch": 0.26674435919847206,
      "grad_norm": 4.275627121965084,
      "learning_rate": 1.1472833854038202e-06,
      "loss": 0.3048,
      "step": 1030
    },
    {
      "epoch": 0.26674435919847206,
      "eval_PRM Accuracy": 0.8490566037735849,
      "eval_PRM F1": 0.9,
      "eval_PRM F1 AUC": 0.8250392875851232,
      "eval_PRM F1 AUC (fixed)": 0.7176532215819801,
      "eval_PRM F1 Neg": 0.6923076923076923,
      "eval_PRM NPV": 0.6206896551724138,
      "eval_PRM Precision": 0.935064935064935,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.2746083736419678,
      "eval_runtime": 14.8637,
      "eval_samples_per_second": 2.96,
      "eval_steps_per_second": 0.135,
      "step": 1030
    },
    {
      "epoch": 0.26700333430448997,
      "grad_norm": 5.096607444133815,
      "learning_rate": 1.146972733257081e-06,
      "loss": 0.3847,
      "step": 1031
    },
    {
      "epoch": 0.26726230941050794,
      "grad_norm": 5.906775619461052,
      "learning_rate": 1.1466616542479392e-06,
      "loss": 0.4176,
      "step": 1032
    },
    {
      "epoch": 0.26752128451652585,
      "grad_norm": 3.3864533892662276,
      "learning_rate": 1.1463501486307913e-06,
      "loss": 0.4073,
      "step": 1033
    },
    {
      "epoch": 0.26778025962254376,
      "grad_norm": 7.674881770284153,
      "learning_rate": 1.146038216660382e-06,
      "loss": 0.3887,
      "step": 1034
    },
    {
      "epoch": 0.2680392347285617,
      "grad_norm": 3.9737886137856604,
      "learning_rate": 1.1457258585918055e-06,
      "loss": 0.3936,
      "step": 1035
    },
    {
      "epoch": 0.2680392347285617,
      "eval_PRM Accuracy": 0.8490566037735849,
      "eval_PRM F1": 0.8974358974358975,
      "eval_PRM F1 AUC": 0.8564693556836039,
      "eval_PRM F1 AUC (fixed)": 0.7317967522262965,
      "eval_PRM F1 Neg": 0.7142857142857143,
      "eval_PRM NPV": 0.6060606060606061,
      "eval_PRM Precision": 0.958904109589041,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.8695652173913043,
      "eval_loss": 0.2778846025466919,
      "eval_runtime": 14.5607,
      "eval_samples_per_second": 3.022,
      "eval_steps_per_second": 0.137,
      "step": 1035
    },
    {
      "epoch": 0.26829820983457964,
      "grad_norm": 3.299595057990445,
      "learning_rate": 1.1454130746805036e-06,
      "loss": 0.3571,
      "step": 1036
    },
    {
      "epoch": 0.2685571849405976,
      "grad_norm": 3.1859027589559115,
      "learning_rate": 1.1450998651822676e-06,
      "loss": 0.2808,
      "step": 1037
    },
    {
      "epoch": 0.2688161600466155,
      "grad_norm": 5.266318626904086,
      "learning_rate": 1.1447862303532357e-06,
      "loss": 0.3535,
      "step": 1038
    },
    {
      "epoch": 0.2690751351526334,
      "grad_norm": 6.0832686097080835,
      "learning_rate": 1.144472170449894e-06,
      "loss": 0.2934,
      "step": 1039
    },
    {
      "epoch": 0.2693341102586514,
      "grad_norm": 5.449410893831182,
      "learning_rate": 1.144157685729077e-06,
      "loss": 0.3511,
      "step": 1040
    },
    {
      "epoch": 0.2693341102586514,
      "eval_PRM Accuracy": 0.8301886792452831,
      "eval_PRM F1": 0.8831168831168831,
      "eval_PRM F1 AUC": 0.8444211629125196,
      "eval_PRM F1 AUC (fixed)": 0.7393923520167627,
      "eval_PRM F1 Neg": 0.6896551724137931,
      "eval_PRM NPV": 0.5714285714285714,
      "eval_PRM Precision": 0.9577464788732394,
      "eval_PRM Recall": 0.8192771084337349,
      "eval_PRM Specificty": 0.8695652173913043,
      "eval_loss": 0.2817172706127167,
      "eval_runtime": 14.4198,
      "eval_samples_per_second": 3.051,
      "eval_steps_per_second": 0.139,
      "step": 1040
    },
    {
      "epoch": 0.2695930853646693,
      "grad_norm": 10.491775913753377,
      "learning_rate": 1.143842776447966e-06,
      "loss": 0.4059,
      "step": 1041
    },
    {
      "epoch": 0.26985206047068727,
      "grad_norm": 4.163129039105927,
      "learning_rate": 1.1435274428640896e-06,
      "loss": 0.3379,
      "step": 1042
    },
    {
      "epoch": 0.2701110355767052,
      "grad_norm": 4.035370510840349,
      "learning_rate": 1.1432116852353236e-06,
      "loss": 0.3394,
      "step": 1043
    },
    {
      "epoch": 0.27037001068272315,
      "grad_norm": 5.014839844833506,
      "learning_rate": 1.1428955038198902e-06,
      "loss": 0.2666,
      "step": 1044
    },
    {
      "epoch": 0.27062898578874106,
      "grad_norm": 3.3777543126356533,
      "learning_rate": 1.1425788988763587e-06,
      "loss": 0.3968,
      "step": 1045
    },
    {
      "epoch": 0.27062898578874106,
      "eval_PRM Accuracy": 0.8584905660377359,
      "eval_PRM F1": 0.9044585987261147,
      "eval_PRM F1 AUC": 0.8624934520691462,
      "eval_PRM F1 AUC (fixed)": 0.7545835515976952,
      "eval_PRM F1 Neg": 0.7272727272727273,
      "eval_PRM NPV": 0.625,
      "eval_PRM Precision": 0.9594594594594594,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.8695652173913043,
      "eval_loss": 0.25391244888305664,
      "eval_runtime": 14.6404,
      "eval_samples_per_second": 3.005,
      "eval_steps_per_second": 0.137,
      "step": 1045
    },
    {
      "epoch": 0.270887960894759,
      "grad_norm": 4.769686269210307,
      "learning_rate": 1.142261870663644e-06,
      "loss": 0.3934,
      "step": 1046
    },
    {
      "epoch": 0.27114693600077694,
      "grad_norm": 4.340672008315693,
      "learning_rate": 1.1419444194410078e-06,
      "loss": 0.3723,
      "step": 1047
    },
    {
      "epoch": 0.27140591110679485,
      "grad_norm": 7.286389338199718,
      "learning_rate": 1.1416265454680578e-06,
      "loss": 0.3354,
      "step": 1048
    },
    {
      "epoch": 0.2716648862128128,
      "grad_norm": 6.061977229152866,
      "learning_rate": 1.1413082490047465e-06,
      "loss": 0.2896,
      "step": 1049
    },
    {
      "epoch": 0.27192386131883073,
      "grad_norm": 6.780070674559419,
      "learning_rate": 1.140989530311373e-06,
      "loss": 0.3464,
      "step": 1050
    },
    {
      "epoch": 0.27192386131883073,
      "eval_PRM Accuracy": 0.8584905660377359,
      "eval_PRM F1": 0.9056603773584906,
      "eval_PRM F1 AUC": 0.8467784180199058,
      "eval_PRM F1 AUC (fixed)": 0.7582503928758513,
      "eval_PRM F1 Neg": 0.7169811320754716,
      "eval_PRM NPV": 0.6333333333333333,
      "eval_PRM Precision": 0.9473684210526315,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.8260869565217391,
      "eval_loss": 0.250600129365921,
      "eval_runtime": 14.7351,
      "eval_samples_per_second": 2.986,
      "eval_steps_per_second": 0.136,
      "step": 1050
    },
    {
      "epoch": 0.27218283642484864,
      "grad_norm": 3.0057781248611124,
      "learning_rate": 1.1406703896485814e-06,
      "loss": 0.3507,
      "step": 1051
    },
    {
      "epoch": 0.2724418115308666,
      "grad_norm": 2.3656334187644323,
      "learning_rate": 1.14035082727736e-06,
      "loss": 0.2873,
      "step": 1052
    },
    {
      "epoch": 0.2727007866368845,
      "grad_norm": 4.327035964713289,
      "learning_rate": 1.1400308434590434e-06,
      "loss": 0.3582,
      "step": 1053
    },
    {
      "epoch": 0.2729597617429025,
      "grad_norm": 4.409821289635134,
      "learning_rate": 1.1397104384553099e-06,
      "loss": 0.3955,
      "step": 1054
    },
    {
      "epoch": 0.2732187368489204,
      "grad_norm": 6.070528106749317,
      "learning_rate": 1.1393896125281823e-06,
      "loss": 0.3542,
      "step": 1055
    },
    {
      "epoch": 0.2732187368489204,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.8903225806451613,
      "eval_PRM F1 AUC": 0.8504452592980618,
      "eval_PRM F1 AUC (fixed)": 0.7572027239392352,
      "eval_PRM F1 Neg": 0.7017543859649122,
      "eval_PRM NPV": 0.5882352941176471,
      "eval_PRM Precision": 0.9583333333333334,
      "eval_PRM Recall": 0.8313253012048193,
      "eval_PRM Specificty": 0.8695652173913043,
      "eval_loss": 0.256532222032547,
      "eval_runtime": 15.0852,
      "eval_samples_per_second": 2.917,
      "eval_steps_per_second": 0.133,
      "step": 1055
    },
    {
      "epoch": 0.2734777119549383,
      "grad_norm": 3.5562342168376575,
      "learning_rate": 1.139068365940028e-06,
      "loss": 0.3447,
      "step": 1056
    },
    {
      "epoch": 0.2737366870609563,
      "grad_norm": 5.912361465122927,
      "learning_rate": 1.1387466989535581e-06,
      "loss": 0.3846,
      "step": 1057
    },
    {
      "epoch": 0.2739956621669742,
      "grad_norm": 5.0568559263393595,
      "learning_rate": 1.1384246118318279e-06,
      "loss": 0.3529,
      "step": 1058
    },
    {
      "epoch": 0.27425463727299215,
      "grad_norm": 3.890988872505904,
      "learning_rate": 1.1381021048382354e-06,
      "loss": 0.3636,
      "step": 1059
    },
    {
      "epoch": 0.27451361237901006,
      "grad_norm": 5.104942614408514,
      "learning_rate": 1.1377791782365234e-06,
      "loss": 0.4364,
      "step": 1060
    },
    {
      "epoch": 0.27451361237901006,
      "eval_PRM Accuracy": 0.8490566037735849,
      "eval_PRM F1": 0.8961038961038961,
      "eval_PRM F1 AUC": 0.8721843897328445,
      "eval_PRM F1 AUC (fixed)": 0.7527501309586171,
      "eval_PRM F1 Neg": 0.7241379310344828,
      "eval_PRM NPV": 0.6,
      "eval_PRM Precision": 0.971830985915493,
      "eval_PRM Recall": 0.8313253012048193,
      "eval_PRM Specificty": 0.9130434782608695,
      "eval_loss": 0.26171520352363586,
      "eval_runtime": 14.5773,
      "eval_samples_per_second": 3.018,
      "eval_steps_per_second": 0.137,
      "step": 1060
    },
    {
      "epoch": 0.274772587485028,
      "grad_norm": 7.7662853415479525,
      "learning_rate": 1.137455832290776e-06,
      "loss": 0.3633,
      "step": 1061
    },
    {
      "epoch": 0.27503156259104594,
      "grad_norm": 5.938423981059565,
      "learning_rate": 1.137132067265422e-06,
      "loss": 0.3852,
      "step": 1062
    },
    {
      "epoch": 0.27529053769706385,
      "grad_norm": 5.107382832753897,
      "learning_rate": 1.136807883425232e-06,
      "loss": 0.2787,
      "step": 1063
    },
    {
      "epoch": 0.2755495128030818,
      "grad_norm": 3.6572572981324556,
      "learning_rate": 1.136483281035319e-06,
      "loss": 0.3835,
      "step": 1064
    },
    {
      "epoch": 0.27580848790909973,
      "grad_norm": 4.0782743725531505,
      "learning_rate": 1.136158260361139e-06,
      "loss": 0.3717,
      "step": 1065
    },
    {
      "epoch": 0.27580848790909973,
      "eval_PRM Accuracy": 0.8490566037735849,
      "eval_PRM F1": 0.8974358974358975,
      "eval_PRM F1 AUC": 0.8564693556836039,
      "eval_PRM F1 AUC (fixed)": 0.7456783656364588,
      "eval_PRM F1 Neg": 0.7142857142857143,
      "eval_PRM NPV": 0.6060606060606061,
      "eval_PRM Precision": 0.958904109589041,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.8695652173913043,
      "eval_loss": 0.26301947236061096,
      "eval_runtime": 14.8292,
      "eval_samples_per_second": 2.967,
      "eval_steps_per_second": 0.135,
      "step": 1065
    },
    {
      "epoch": 0.2760674630151177,
      "grad_norm": 5.013671476758816,
      "learning_rate": 1.1358328216684891e-06,
      "loss": 0.2519,
      "step": 1066
    },
    {
      "epoch": 0.2763264381211356,
      "grad_norm": 3.4108222416181158,
      "learning_rate": 1.135506965223509e-06,
      "loss": 0.2606,
      "step": 1067
    },
    {
      "epoch": 0.2765854132271535,
      "grad_norm": 6.796678122355816,
      "learning_rate": 1.13518069129268e-06,
      "loss": 0.4095,
      "step": 1068
    },
    {
      "epoch": 0.2768443883331715,
      "grad_norm": 4.326362848524545,
      "learning_rate": 1.1348540001428243e-06,
      "loss": 0.3535,
      "step": 1069
    },
    {
      "epoch": 0.2771033634391894,
      "grad_norm": 4.655760102750851,
      "learning_rate": 1.134526892041106e-06,
      "loss": 0.3831,
      "step": 1070
    },
    {
      "epoch": 0.2771033634391894,
      "eval_PRM Accuracy": 0.8490566037735849,
      "eval_PRM F1": 0.9,
      "eval_PRM F1 AUC": 0.8250392875851232,
      "eval_PRM F1 AUC (fixed)": 0.7569408067050812,
      "eval_PRM F1 Neg": 0.6923076923076923,
      "eval_PRM NPV": 0.6206896551724138,
      "eval_PRM Precision": 0.935064935064935,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.26515859365463257,
      "eval_runtime": 14.8422,
      "eval_samples_per_second": 2.965,
      "eval_steps_per_second": 0.135,
      "step": 1070
    },
    {
      "epoch": 0.27736233854520737,
      "grad_norm": 8.05051403901847,
      "learning_rate": 1.134199367255029e-06,
      "loss": 0.3539,
      "step": 1071
    },
    {
      "epoch": 0.2776213136512253,
      "grad_norm": 3.378675019359202,
      "learning_rate": 1.1338714260524393e-06,
      "loss": 0.3111,
      "step": 1072
    },
    {
      "epoch": 0.2778802887572432,
      "grad_norm": 3.7446683598641615,
      "learning_rate": 1.133543068701523e-06,
      "loss": 0.4096,
      "step": 1073
    },
    {
      "epoch": 0.27813926386326115,
      "grad_norm": 6.913150764414947,
      "learning_rate": 1.1332142954708064e-06,
      "loss": 0.3498,
      "step": 1074
    },
    {
      "epoch": 0.27839823896927907,
      "grad_norm": 3.781401402320106,
      "learning_rate": 1.1328851066291555e-06,
      "loss": 0.3295,
      "step": 1075
    },
    {
      "epoch": 0.27839823896927907,
      "eval_PRM Accuracy": 0.8490566037735849,
      "eval_PRM F1": 0.8987341772151899,
      "eval_PRM F1 AUC": 0.8407543216343635,
      "eval_PRM F1 AUC (fixed)": 0.7537977998952331,
      "eval_PRM F1 Neg": 0.7037037037037037,
      "eval_PRM NPV": 0.6129032258064516,
      "eval_PRM Precision": 0.9466666666666667,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.8260869565217391,
      "eval_loss": 0.2676940858364105,
      "eval_runtime": 14.9396,
      "eval_samples_per_second": 2.945,
      "eval_steps_per_second": 0.134,
      "step": 1075
    },
    {
      "epoch": 0.27865721407529703,
      "grad_norm": 5.819523266576147,
      "learning_rate": 1.1325555024457768e-06,
      "loss": 0.4179,
      "step": 1076
    },
    {
      "epoch": 0.27891618918131494,
      "grad_norm": 4.099575759852333,
      "learning_rate": 1.1322254831902162e-06,
      "loss": 0.4283,
      "step": 1077
    },
    {
      "epoch": 0.27917516428733286,
      "grad_norm": 2.388694788644085,
      "learning_rate": 1.1318950491323593e-06,
      "loss": 0.3189,
      "step": 1078
    },
    {
      "epoch": 0.2794341393933508,
      "grad_norm": 3.8430182333509477,
      "learning_rate": 1.1315642005424304e-06,
      "loss": 0.3609,
      "step": 1079
    },
    {
      "epoch": 0.27969311449936873,
      "grad_norm": 6.0484009009962865,
      "learning_rate": 1.1312329376909934e-06,
      "loss": 0.4167,
      "step": 1080
    },
    {
      "epoch": 0.27969311449936873,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8758169934640523,
      "eval_PRM F1 AUC": 0.8383970665269774,
      "eval_PRM F1 AUC (fixed)": 0.749083289680461,
      "eval_PRM F1 Neg": 0.6779661016949152,
      "eval_PRM NPV": 0.5555555555555556,
      "eval_PRM Precision": 0.9571428571428572,
      "eval_PRM Recall": 0.8072289156626506,
      "eval_PRM Specificty": 0.8695652173913043,
      "eval_loss": 0.28986474871635437,
      "eval_runtime": 14.6064,
      "eval_samples_per_second": 3.012,
      "eval_steps_per_second": 0.137,
      "step": 1080
    },
    {
      "epoch": 0.2799520896053867,
      "grad_norm": 7.64904237886481,
      "learning_rate": 1.1309012608489504e-06,
      "loss": 0.3803,
      "step": 1081
    },
    {
      "epoch": 0.2802110647114046,
      "grad_norm": 9.823767001803189,
      "learning_rate": 1.1305691702875423e-06,
      "loss": 0.2893,
      "step": 1082
    },
    {
      "epoch": 0.2804700398174225,
      "grad_norm": 5.004452607333411,
      "learning_rate": 1.1302366662783488e-06,
      "loss": 0.4203,
      "step": 1083
    },
    {
      "epoch": 0.2807290149234405,
      "grad_norm": 8.144551217692342,
      "learning_rate": 1.129903749093287e-06,
      "loss": 0.3629,
      "step": 1084
    },
    {
      "epoch": 0.2809879900294584,
      "grad_norm": 7.145958462099795,
      "learning_rate": 1.129570419004612e-06,
      "loss": 0.289,
      "step": 1085
    },
    {
      "epoch": 0.2809879900294584,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.868421052631579,
      "eval_PRM F1 AUC": 0.8323729701414354,
      "eval_PRM F1 AUC (fixed)": 0.748821372446307,
      "eval_PRM F1 Neg": 0.6666666666666666,
      "eval_PRM NPV": 0.5405405405405406,
      "eval_PRM Precision": 0.9565217391304348,
      "eval_PRM Recall": 0.7951807228915663,
      "eval_PRM Specificty": 0.8695652173913043,
      "eval_loss": 0.294299453496933,
      "eval_runtime": 14.5707,
      "eval_samples_per_second": 3.02,
      "eval_steps_per_second": 0.137,
      "step": 1085
    },
    {
      "epoch": 0.28124696513547637,
      "grad_norm": 4.960838845669801,
      "learning_rate": 1.129236676284917e-06,
      "loss": 0.3983,
      "step": 1086
    },
    {
      "epoch": 0.2815059402414943,
      "grad_norm": 4.9063190727136,
      "learning_rate": 1.1289025212071324e-06,
      "loss": 0.3552,
      "step": 1087
    },
    {
      "epoch": 0.28176491534751225,
      "grad_norm": 6.111329302463371,
      "learning_rate": 1.1285679540445259e-06,
      "loss": 0.4539,
      "step": 1088
    },
    {
      "epoch": 0.28202389045353016,
      "grad_norm": 3.1683233714486176,
      "learning_rate": 1.1282329750707018e-06,
      "loss": 0.3701,
      "step": 1089
    },
    {
      "epoch": 0.28228286555954807,
      "grad_norm": 5.4822402949350115,
      "learning_rate": 1.1278975845596016e-06,
      "loss": 0.3055,
      "step": 1090
    },
    {
      "epoch": 0.28228286555954807,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8789808917197452,
      "eval_PRM F1 AUC": 0.8069669984284967,
      "eval_PRM F1 AUC (fixed)": 0.753273965426925,
      "eval_PRM F1 Neg": 0.6545454545454545,
      "eval_PRM NPV": 0.5625,
      "eval_PRM Precision": 0.9324324324324325,
      "eval_PRM Recall": 0.8313253012048193,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.2796691656112671,
      "eval_runtime": 14.4984,
      "eval_samples_per_second": 3.035,
      "eval_steps_per_second": 0.138,
      "step": 1090
    },
    {
      "epoch": 0.28254184066556604,
      "grad_norm": 3.480490868188625,
      "learning_rate": 1.127561782785503e-06,
      "loss": 0.3824,
      "step": 1091
    },
    {
      "epoch": 0.28280081577158395,
      "grad_norm": 3.805686299465861,
      "learning_rate": 1.1272255700230209e-06,
      "loss": 0.3474,
      "step": 1092
    },
    {
      "epoch": 0.2830597908776019,
      "grad_norm": 4.48044589020009,
      "learning_rate": 1.126888946547105e-06,
      "loss": 0.3661,
      "step": 1093
    },
    {
      "epoch": 0.2833187659836198,
      "grad_norm": 2.877380173271708,
      "learning_rate": 1.1265519126330417e-06,
      "loss": 0.3087,
      "step": 1094
    },
    {
      "epoch": 0.28357774108963774,
      "grad_norm": 2.542242455848216,
      "learning_rate": 1.126214468556453e-06,
      "loss": 0.315,
      "step": 1095
    },
    {
      "epoch": 0.28357774108963774,
      "eval_PRM Accuracy": 0.8301886792452831,
      "eval_PRM F1": 0.8875,
      "eval_PRM F1 AUC": 0.7972760607647982,
      "eval_PRM F1 AUC (fixed)": 0.7637506547930855,
      "eval_PRM F1 Neg": 0.6538461538461539,
      "eval_PRM NPV": 0.5862068965517241,
      "eval_PRM Precision": 0.922077922077922,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.2715385854244232,
      "eval_runtime": 14.3759,
      "eval_samples_per_second": 3.061,
      "eval_steps_per_second": 0.139,
      "step": 1095
    },
    {
      "epoch": 0.2838367161956557,
      "grad_norm": 5.416422720177891,
      "learning_rate": 1.1258766145932964e-06,
      "loss": 0.5243,
      "step": 1096
    },
    {
      "epoch": 0.2840956913016736,
      "grad_norm": 7.765499713875308,
      "learning_rate": 1.125538351019864e-06,
      "loss": 0.4824,
      "step": 1097
    },
    {
      "epoch": 0.2843546664076916,
      "grad_norm": 4.784714891415138,
      "learning_rate": 1.1251996781127832e-06,
      "loss": 0.3516,
      "step": 1098
    },
    {
      "epoch": 0.2846136415137095,
      "grad_norm": 4.1611214148825315,
      "learning_rate": 1.1248605961490168e-06,
      "loss": 0.4139,
      "step": 1099
    },
    {
      "epoch": 0.2848726166197274,
      "grad_norm": 9.588537077132573,
      "learning_rate": 1.1245211054058611e-06,
      "loss": 0.3736,
      "step": 1100
    },
    {
      "epoch": 0.2848726166197274,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.89171974522293,
      "eval_PRM F1 AUC": 0.8347302252488213,
      "eval_PRM F1 AUC (fixed)": 0.7532739654269252,
      "eval_PRM F1 Neg": 0.6909090909090909,
      "eval_PRM NPV": 0.59375,
      "eval_PRM Precision": 0.9459459459459459,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.8260869565217391,
      "eval_loss": 0.27907219529151917,
      "eval_runtime": 14.5308,
      "eval_samples_per_second": 3.028,
      "eval_steps_per_second": 0.138,
      "step": 1100
    },
    {
      "epoch": 0.28513159172574537,
      "grad_norm": 8.28027340948901,
      "learning_rate": 1.1241812061609475e-06,
      "loss": 0.3066,
      "step": 1101
    },
    {
      "epoch": 0.2853905668317633,
      "grad_norm": 5.2275923345878565,
      "learning_rate": 1.1238408986922405e-06,
      "loss": 0.2485,
      "step": 1102
    },
    {
      "epoch": 0.28564954193778125,
      "grad_norm": 3.9191319661849118,
      "learning_rate": 1.1235001832780397e-06,
      "loss": 0.2648,
      "step": 1103
    },
    {
      "epoch": 0.28590851704379916,
      "grad_norm": 3.803575863891714,
      "learning_rate": 1.1231590601969773e-06,
      "loss": 0.3509,
      "step": 1104
    },
    {
      "epoch": 0.2861674921498171,
      "grad_norm": 5.062747812604048,
      "learning_rate": 1.1228175297280193e-06,
      "loss": 0.3306,
      "step": 1105
    },
    {
      "epoch": 0.2861674921498171,
      "eval_PRM Accuracy": 0.8490566037735849,
      "eval_PRM F1": 0.8987341772151899,
      "eval_PRM F1 AUC": 0.8407543216343635,
      "eval_PRM F1 AUC (fixed)": 0.7564169722367732,
      "eval_PRM F1 Neg": 0.7037037037037037,
      "eval_PRM NPV": 0.6129032258064516,
      "eval_PRM Precision": 0.9466666666666667,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.8260869565217391,
      "eval_loss": 0.2660982012748718,
      "eval_runtime": 14.627,
      "eval_samples_per_second": 3.008,
      "eval_steps_per_second": 0.137,
      "step": 1105
    },
    {
      "epoch": 0.28642646725583504,
      "grad_norm": 4.898710258003656,
      "learning_rate": 1.1224755921504648e-06,
      "loss": 0.4031,
      "step": 1106
    },
    {
      "epoch": 0.28668544236185295,
      "grad_norm": 3.561715323993499,
      "learning_rate": 1.1221332477439459e-06,
      "loss": 0.3674,
      "step": 1107
    },
    {
      "epoch": 0.2869444174678709,
      "grad_norm": 3.7429726684658853,
      "learning_rate": 1.1217904967884274e-06,
      "loss": 0.3967,
      "step": 1108
    },
    {
      "epoch": 0.2872033925738888,
      "grad_norm": 3.178179402931704,
      "learning_rate": 1.1214473395642064e-06,
      "loss": 0.3632,
      "step": 1109
    },
    {
      "epoch": 0.2874623676799068,
      "grad_norm": 6.2815507059756,
      "learning_rate": 1.1211037763519122e-06,
      "loss": 0.3744,
      "step": 1110
    },
    {
      "epoch": 0.2874623676799068,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.89171974522293,
      "eval_PRM F1 AUC": 0.8347302252488213,
      "eval_PRM F1 AUC (fixed)": 0.7608695652173912,
      "eval_PRM F1 Neg": 0.6909090909090909,
      "eval_PRM NPV": 0.59375,
      "eval_PRM Precision": 0.9459459459459459,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.8260869565217391,
      "eval_loss": 0.26640403270721436,
      "eval_runtime": 14.4791,
      "eval_samples_per_second": 3.039,
      "eval_steps_per_second": 0.138,
      "step": 1110
    },
    {
      "epoch": 0.2877213427859247,
      "grad_norm": 4.2386671161357405,
      "learning_rate": 1.1207598074325066e-06,
      "loss": 0.3827,
      "step": 1111
    },
    {
      "epoch": 0.2879803178919426,
      "grad_norm": 3.3655970950015295,
      "learning_rate": 1.1204154330872824e-06,
      "loss": 0.3525,
      "step": 1112
    },
    {
      "epoch": 0.2882392929979606,
      "grad_norm": 7.982714386079896,
      "learning_rate": 1.1200706535978647e-06,
      "loss": 0.3293,
      "step": 1113
    },
    {
      "epoch": 0.2884982681039785,
      "grad_norm": 9.602171181971372,
      "learning_rate": 1.1197254692462096e-06,
      "loss": 0.3855,
      "step": 1114
    },
    {
      "epoch": 0.28875724320999646,
      "grad_norm": 3.4844020908031554,
      "learning_rate": 1.1193798803146042e-06,
      "loss": 0.3128,
      "step": 1115
    },
    {
      "epoch": 0.28875724320999646,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8774193548387097,
      "eval_PRM F1 AUC": 0.8226820324777371,
      "eval_PRM F1 AUC (fixed)": 0.7627029858564695,
      "eval_PRM F1 Neg": 0.6666666666666666,
      "eval_PRM NPV": 0.5588235294117647,
      "eval_PRM Precision": 0.9444444444444444,
      "eval_PRM Recall": 0.8192771084337349,
      "eval_PRM Specificty": 0.8260869565217391,
      "eval_loss": 0.26669546961784363,
      "eval_runtime": 14.4819,
      "eval_samples_per_second": 3.038,
      "eval_steps_per_second": 0.138,
      "step": 1115
    },
    {
      "epoch": 0.2890162183160144,
      "grad_norm": 6.597406785592498,
      "learning_rate": 1.1190338870856666e-06,
      "loss": 0.2791,
      "step": 1116
    },
    {
      "epoch": 0.2892751934220323,
      "grad_norm": 5.9280718737016045,
      "learning_rate": 1.1186874898423455e-06,
      "loss": 0.3807,
      "step": 1117
    },
    {
      "epoch": 0.28953416852805025,
      "grad_norm": 2.8334208382831605,
      "learning_rate": 1.1183406888679199e-06,
      "loss": 0.3188,
      "step": 1118
    },
    {
      "epoch": 0.28979314363406816,
      "grad_norm": 7.612503860515019,
      "learning_rate": 1.117993484445999e-06,
      "loss": 0.3501,
      "step": 1119
    },
    {
      "epoch": 0.29005211874008613,
      "grad_norm": 8.11990763445414,
      "learning_rate": 1.1176458768605225e-06,
      "loss": 0.4403,
      "step": 1120
    },
    {
      "epoch": 0.29005211874008613,
      "eval_PRM Accuracy": 0.8490566037735849,
      "eval_PRM F1": 0.9,
      "eval_PRM F1 AUC": 0.8250392875851232,
      "eval_PRM F1 AUC (fixed)": 0.7747511786275538,
      "eval_PRM F1 Neg": 0.6923076923076923,
      "eval_PRM NPV": 0.6206896551724138,
      "eval_PRM Precision": 0.935064935064935,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.2532932758331299,
      "eval_runtime": 14.6892,
      "eval_samples_per_second": 2.995,
      "eval_steps_per_second": 0.136,
      "step": 1120
    },
    {
      "epoch": 0.29031109384610404,
      "grad_norm": 4.263431566665755,
      "learning_rate": 1.1172978663957587e-06,
      "loss": 0.3335,
      "step": 1121
    },
    {
      "epoch": 0.29057006895212195,
      "grad_norm": 5.260238174654083,
      "learning_rate": 1.1169494533363062e-06,
      "loss": 0.34,
      "step": 1122
    },
    {
      "epoch": 0.2908290440581399,
      "grad_norm": 4.355002531212862,
      "learning_rate": 1.1166006379670924e-06,
      "loss": 0.3562,
      "step": 1123
    },
    {
      "epoch": 0.29108801916415783,
      "grad_norm": 4.538866881853854,
      "learning_rate": 1.1162514205733744e-06,
      "loss": 0.343,
      "step": 1124
    },
    {
      "epoch": 0.2913469942701758,
      "grad_norm": 4.793637250566336,
      "learning_rate": 1.115901801440737e-06,
      "loss": 0.3588,
      "step": 1125
    },
    {
      "epoch": 0.2913469942701758,
      "eval_PRM Accuracy": 0.8584905660377359,
      "eval_PRM F1": 0.906832298136646,
      "eval_PRM F1 AUC": 0.8310633839706653,
      "eval_PRM F1 AUC (fixed)": 0.7755369303300158,
      "eval_PRM F1 Neg": 0.7058823529411765,
      "eval_PRM NPV": 0.6428571428571429,
      "eval_PRM Precision": 0.9358974358974359,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.2523413598537445,
      "eval_runtime": 14.579,
      "eval_samples_per_second": 3.018,
      "eval_steps_per_second": 0.137,
      "step": 1125
    },
    {
      "epoch": 0.2916059693761937,
      "grad_norm": 3.7268330544818213,
      "learning_rate": 1.1155517808550941e-06,
      "loss": 0.4181,
      "step": 1126
    },
    {
      "epoch": 0.2918649444822117,
      "grad_norm": 5.165577278078134,
      "learning_rate": 1.1152013591026884e-06,
      "loss": 0.3887,
      "step": 1127
    },
    {
      "epoch": 0.2921239195882296,
      "grad_norm": 5.040742010231609,
      "learning_rate": 1.11485053647009e-06,
      "loss": 0.3919,
      "step": 1128
    },
    {
      "epoch": 0.2923828946942475,
      "grad_norm": 4.725959357988391,
      "learning_rate": 1.1144993132441967e-06,
      "loss": 0.3668,
      "step": 1129
    },
    {
      "epoch": 0.29264186980026546,
      "grad_norm": 4.032660789825553,
      "learning_rate": 1.1141476897122348e-06,
      "loss": 0.3183,
      "step": 1130
    },
    {
      "epoch": 0.29264186980026546,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.89171974522293,
      "eval_PRM F1 AUC": 0.8347302252488213,
      "eval_PRM F1 AUC (fixed)": 0.7661079099004714,
      "eval_PRM F1 Neg": 0.6909090909090909,
      "eval_PRM NPV": 0.59375,
      "eval_PRM Precision": 0.9459459459459459,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.8260869565217391,
      "eval_loss": 0.26497408747673035,
      "eval_runtime": 14.3768,
      "eval_samples_per_second": 3.06,
      "eval_steps_per_second": 0.139,
      "step": 1130
    },
    {
      "epoch": 0.2929008449062834,
      "grad_norm": 7.673167100360908,
      "learning_rate": 1.1137956661617566e-06,
      "loss": 0.3832,
      "step": 1131
    },
    {
      "epoch": 0.29315982001230134,
      "grad_norm": 7.881068518051551,
      "learning_rate": 1.113443242880643e-06,
      "loss": 0.3027,
      "step": 1132
    },
    {
      "epoch": 0.29341879511831925,
      "grad_norm": 5.782281218579323,
      "learning_rate": 1.1130904201571006e-06,
      "loss": 0.3171,
      "step": 1133
    },
    {
      "epoch": 0.29367777022433716,
      "grad_norm": 3.762358561390526,
      "learning_rate": 1.1127371982796636e-06,
      "loss": 0.3372,
      "step": 1134
    },
    {
      "epoch": 0.29393674533035513,
      "grad_norm": 3.7955893572075787,
      "learning_rate": 1.1123835775371919e-06,
      "loss": 0.3413,
      "step": 1135
    },
    {
      "epoch": 0.29393674533035513,
      "eval_PRM Accuracy": 0.8584905660377359,
      "eval_PRM F1": 0.906832298136646,
      "eval_PRM F1 AUC": 0.8310633839706653,
      "eval_PRM F1 AUC (fixed)": 0.7619172341540074,
      "eval_PRM F1 Neg": 0.7058823529411765,
      "eval_PRM NPV": 0.6428571428571429,
      "eval_PRM Precision": 0.9358974358974359,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.25865188241004944,
      "eval_runtime": 14.9603,
      "eval_samples_per_second": 2.941,
      "eval_steps_per_second": 0.134,
      "step": 1135
    },
    {
      "epoch": 0.29419572043637304,
      "grad_norm": 2.959177369623632,
      "learning_rate": 1.1120295582188718e-06,
      "loss": 0.3441,
      "step": 1136
    },
    {
      "epoch": 0.294454695542391,
      "grad_norm": 3.3679792765512704,
      "learning_rate": 1.111675140614216e-06,
      "loss": 0.3377,
      "step": 1137
    },
    {
      "epoch": 0.2947136706484089,
      "grad_norm": 3.597086189327973,
      "learning_rate": 1.1113203250130623e-06,
      "loss": 0.3644,
      "step": 1138
    },
    {
      "epoch": 0.29497264575442683,
      "grad_norm": 2.9662259450420168,
      "learning_rate": 1.110965111705574e-06,
      "loss": 0.3793,
      "step": 1139
    },
    {
      "epoch": 0.2952316208604448,
      "grad_norm": 10.258782090900965,
      "learning_rate": 1.1106095009822401e-06,
      "loss": 0.4473,
      "step": 1140
    },
    {
      "epoch": 0.2952316208604448,
      "eval_PRM Accuracy": 0.8773584905660378,
      "eval_PRM F1": 0.9202453987730062,
      "eval_PRM F1 AUC": 0.8431115767417497,
      "eval_PRM F1 AUC (fixed)": 0.7621791513881614,
      "eval_PRM F1 Neg": 0.7346938775510204,
      "eval_PRM NPV": 0.6923076923076923,
      "eval_PRM Precision": 0.9375,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.2613238990306854,
      "eval_runtime": 14.3424,
      "eval_samples_per_second": 3.068,
      "eval_steps_per_second": 0.139,
      "step": 1140
    },
    {
      "epoch": 0.2954905959664627,
      "grad_norm": 9.086209755968815,
      "learning_rate": 1.1102534931338743e-06,
      "loss": 0.357,
      "step": 1141
    },
    {
      "epoch": 0.2957495710724807,
      "grad_norm": 4.4360234766203845,
      "learning_rate": 1.1098970884516153e-06,
      "loss": 0.3027,
      "step": 1142
    },
    {
      "epoch": 0.2960085461784986,
      "grad_norm": 3.6234523247900623,
      "learning_rate": 1.1095402872269257e-06,
      "loss": 0.3617,
      "step": 1143
    },
    {
      "epoch": 0.2962675212845165,
      "grad_norm": 3.214590685530097,
      "learning_rate": 1.1091830897515932e-06,
      "loss": 0.361,
      "step": 1144
    },
    {
      "epoch": 0.29652649639053447,
      "grad_norm": 4.836324193876224,
      "learning_rate": 1.108825496317729e-06,
      "loss": 0.3596,
      "step": 1145
    },
    {
      "epoch": 0.29652649639053447,
      "eval_PRM Accuracy": 0.8301886792452831,
      "eval_PRM F1": 0.8860759493670886,
      "eval_PRM F1 AUC": 0.8129910948140389,
      "eval_PRM F1 AUC (fixed)": 0.7619172341540074,
      "eval_PRM F1 Neg": 0.6666666666666666,
      "eval_PRM NPV": 0.5806451612903226,
      "eval_PRM Precision": 0.9333333333333333,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.2637127935886383,
      "eval_runtime": 14.6897,
      "eval_samples_per_second": 2.995,
      "eval_steps_per_second": 0.136,
      "step": 1145
    },
    {
      "epoch": 0.2967854714965524,
      "grad_norm": 4.40229983630881,
      "learning_rate": 1.1084675072177684e-06,
      "loss": 0.2748,
      "step": 1146
    },
    {
      "epoch": 0.29704444660257034,
      "grad_norm": 6.446443974590935,
      "learning_rate": 1.10810912274447e-06,
      "loss": 0.3959,
      "step": 1147
    },
    {
      "epoch": 0.29730342170858826,
      "grad_norm": 4.162136530137705,
      "learning_rate": 1.1077503431909161e-06,
      "loss": 0.3092,
      "step": 1148
    },
    {
      "epoch": 0.2975623968146062,
      "grad_norm": 3.379942752653197,
      "learning_rate": 1.1073911688505118e-06,
      "loss": 0.3569,
      "step": 1149
    },
    {
      "epoch": 0.29782137192062413,
      "grad_norm": 4.790526956890661,
      "learning_rate": 1.1070316000169847e-06,
      "loss": 0.3305,
      "step": 1150
    },
    {
      "epoch": 0.29782137192062413,
      "eval_PRM Accuracy": 0.8584905660377359,
      "eval_PRM F1": 0.906832298136646,
      "eval_PRM F1 AUC": 0.8310633839706653,
      "eval_PRM F1 AUC (fixed)": 0.7668936616029335,
      "eval_PRM F1 Neg": 0.7058823529411765,
      "eval_PRM NPV": 0.6428571428571429,
      "eval_PRM Precision": 0.9358974358974359,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.256858229637146,
      "eval_runtime": 14.6627,
      "eval_samples_per_second": 3.001,
      "eval_steps_per_second": 0.136,
      "step": 1150
    },
    {
      "epoch": 0.29808034702664205,
      "grad_norm": 4.646326337168363,
      "learning_rate": 1.1066716369843863e-06,
      "loss": 0.4038,
      "step": 1151
    },
    {
      "epoch": 0.29833932213266,
      "grad_norm": 6.96599879604997,
      "learning_rate": 1.1063112800470888e-06,
      "loss": 0.3765,
      "step": 1152
    },
    {
      "epoch": 0.2985982972386779,
      "grad_norm": 3.1915204817949783,
      "learning_rate": 1.105950529499788e-06,
      "loss": 0.2686,
      "step": 1153
    },
    {
      "epoch": 0.2988572723446959,
      "grad_norm": 3.798299543621722,
      "learning_rate": 1.1055893856375003e-06,
      "loss": 0.3423,
      "step": 1154
    },
    {
      "epoch": 0.2991162474507138,
      "grad_norm": 5.672302650751939,
      "learning_rate": 1.1052278487555649e-06,
      "loss": 0.383,
      "step": 1155
    },
    {
      "epoch": 0.2991162474507138,
      "eval_PRM Accuracy": 0.8584905660377359,
      "eval_PRM F1": 0.906832298136646,
      "eval_PRM F1 AUC": 0.8310633839706653,
      "eval_PRM F1 AUC (fixed)": 0.7750130958617076,
      "eval_PRM F1 Neg": 0.7058823529411765,
      "eval_PRM NPV": 0.6428571428571429,
      "eval_PRM Precision": 0.9358974358974359,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.25225791335105896,
      "eval_runtime": 14.7283,
      "eval_samples_per_second": 2.987,
      "eval_steps_per_second": 0.136,
      "step": 1155
    },
    {
      "epoch": 0.2993752225567317,
      "grad_norm": 5.845418089666656,
      "learning_rate": 1.1048659191496416e-06,
      "loss": 0.417,
      "step": 1156
    },
    {
      "epoch": 0.2996341976627497,
      "grad_norm": 5.937672783763917,
      "learning_rate": 1.1045035971157117e-06,
      "loss": 0.4291,
      "step": 1157
    },
    {
      "epoch": 0.2998931727687676,
      "grad_norm": 3.4584606230076447,
      "learning_rate": 1.1041408829500773e-06,
      "loss": 0.3194,
      "step": 1158
    },
    {
      "epoch": 0.30015214787478556,
      "grad_norm": 4.451876308852337,
      "learning_rate": 1.1037777769493613e-06,
      "loss": 0.3549,
      "step": 1159
    },
    {
      "epoch": 0.30041112298080347,
      "grad_norm": 3.602995497692067,
      "learning_rate": 1.103414279410507e-06,
      "loss": 0.3692,
      "step": 1160
    },
    {
      "epoch": 0.30041112298080347,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8789808917197452,
      "eval_PRM F1 AUC": 0.8069669984284967,
      "eval_PRM F1 AUC (fixed)": 0.7773703509690938,
      "eval_PRM F1 Neg": 0.6545454545454545,
      "eval_PRM NPV": 0.5625,
      "eval_PRM Precision": 0.9324324324324325,
      "eval_PRM Recall": 0.8313253012048193,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.26475226879119873,
      "eval_runtime": 14.7743,
      "eval_samples_per_second": 2.978,
      "eval_steps_per_second": 0.135,
      "step": 1160
    },
    {
      "epoch": 0.3006700980868214,
      "grad_norm": 5.62440934358399,
      "learning_rate": 1.1030503906307781e-06,
      "loss": 0.2798,
      "step": 1161
    },
    {
      "epoch": 0.30092907319283935,
      "grad_norm": 4.339810549098005,
      "learning_rate": 1.1026861109077579e-06,
      "loss": 0.3868,
      "step": 1162
    },
    {
      "epoch": 0.30118804829885726,
      "grad_norm": 12.165816417405841,
      "learning_rate": 1.1023214405393492e-06,
      "loss": 0.3984,
      "step": 1163
    },
    {
      "epoch": 0.3014470234048752,
      "grad_norm": 5.7935098656181365,
      "learning_rate": 1.1019563798237752e-06,
      "loss": 0.5339,
      "step": 1164
    },
    {
      "epoch": 0.30170599851089314,
      "grad_norm": 3.5268608216207467,
      "learning_rate": 1.1015909290595771e-06,
      "loss": 0.4078,
      "step": 1165
    },
    {
      "epoch": 0.30170599851089314,
      "eval_PRM Accuracy": 0.8301886792452831,
      "eval_PRM F1": 0.8860759493670886,
      "eval_PRM F1 AUC": 0.8129910948140389,
      "eval_PRM F1 AUC (fixed)": 0.7729177579884756,
      "eval_PRM F1 Neg": 0.6666666666666666,
      "eval_PRM NPV": 0.5806451612903226,
      "eval_PRM Precision": 0.9333333333333333,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.26175162196159363,
      "eval_runtime": 14.2847,
      "eval_samples_per_second": 3.08,
      "eval_steps_per_second": 0.14,
      "step": 1165
    },
    {
      "epoch": 0.30196497361691105,
      "grad_norm": 4.127649198536379,
      "learning_rate": 1.1012250885456164e-06,
      "loss": 0.4231,
      "step": 1166
    },
    {
      "epoch": 0.302223948722929,
      "grad_norm": 3.4331514655877773,
      "learning_rate": 1.100858858581072e-06,
      "loss": 0.4242,
      "step": 1167
    },
    {
      "epoch": 0.3024829238289469,
      "grad_norm": 6.40016116126962,
      "learning_rate": 1.1004922394654423e-06,
      "loss": 0.4285,
      "step": 1168
    },
    {
      "epoch": 0.3027418989349649,
      "grad_norm": 3.656133214972611,
      "learning_rate": 1.1001252314985433e-06,
      "loss": 0.3593,
      "step": 1169
    },
    {
      "epoch": 0.3030008740409828,
      "grad_norm": 7.158557853449496,
      "learning_rate": 1.0997578349805096e-06,
      "loss": 0.4875,
      "step": 1170
    },
    {
      "epoch": 0.3030008740409828,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.8930817610062893,
      "eval_PRM F1 AUC": 0.819015191199581,
      "eval_PRM F1 AUC (fixed)": 0.7590361445783133,
      "eval_PRM F1 Neg": 0.6792452830188679,
      "eval_PRM NPV": 0.6,
      "eval_PRM Precision": 0.9342105263157895,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.2685607969760895,
      "eval_runtime": 14.5565,
      "eval_samples_per_second": 3.023,
      "eval_steps_per_second": 0.137,
      "step": 1170
    },
    {
      "epoch": 0.30325984914700077,
      "grad_norm": 3.0556367903726644,
      "learning_rate": 1.0993900502117929e-06,
      "loss": 0.3849,
      "step": 1171
    },
    {
      "epoch": 0.3035188242530187,
      "grad_norm": 2.802202697649935,
      "learning_rate": 1.099021877493163e-06,
      "loss": 0.3063,
      "step": 1172
    },
    {
      "epoch": 0.3037777993590366,
      "grad_norm": 5.612109421532672,
      "learning_rate": 1.0986533171257064e-06,
      "loss": 0.3155,
      "step": 1173
    },
    {
      "epoch": 0.30403677446505456,
      "grad_norm": 7.349516478413747,
      "learning_rate": 1.0982843694108272e-06,
      "loss": 0.3551,
      "step": 1174
    },
    {
      "epoch": 0.30429574957107247,
      "grad_norm": 5.372861942029928,
      "learning_rate": 1.0979150346502456e-06,
      "loss": 0.3109,
      "step": 1175
    },
    {
      "epoch": 0.30429574957107247,
      "eval_PRM Accuracy": 0.8490566037735849,
      "eval_PRM F1": 0.9,
      "eval_PRM F1 AUC": 0.8250392875851232,
      "eval_PRM F1 AUC (fixed)": 0.7627029858564693,
      "eval_PRM F1 Neg": 0.6923076923076923,
      "eval_PRM NPV": 0.6206896551724138,
      "eval_PRM Precision": 0.935064935064935,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.2636399567127228,
      "eval_runtime": 14.6856,
      "eval_samples_per_second": 2.996,
      "eval_steps_per_second": 0.136,
      "step": 1175
    },
    {
      "epoch": 0.30455472467709044,
      "grad_norm": 2.9856816961371804,
      "learning_rate": 1.097545313145999e-06,
      "loss": 0.3356,
      "step": 1176
    },
    {
      "epoch": 0.30481369978310835,
      "grad_norm": 4.151685285025255,
      "learning_rate": 1.0971752052004406e-06,
      "loss": 0.3121,
      "step": 1177
    },
    {
      "epoch": 0.30507267488912626,
      "grad_norm": 3.2390232780479025,
      "learning_rate": 1.0968047111162399e-06,
      "loss": 0.3094,
      "step": 1178
    },
    {
      "epoch": 0.3053316499951442,
      "grad_norm": 4.673769546912608,
      "learning_rate": 1.096433831196382e-06,
      "loss": 0.3841,
      "step": 1179
    },
    {
      "epoch": 0.30559062510116214,
      "grad_norm": 3.818883587684225,
      "learning_rate": 1.0960625657441678e-06,
      "loss": 0.3289,
      "step": 1180
    },
    {
      "epoch": 0.30559062510116214,
      "eval_PRM Accuracy": 0.8490566037735849,
      "eval_PRM F1": 0.9,
      "eval_PRM F1 AUC": 0.8250392875851232,
      "eval_PRM F1 AUC (fixed)": 0.7797276060764797,
      "eval_PRM F1 Neg": 0.6923076923076923,
      "eval_PRM NPV": 0.6206896551724138,
      "eval_PRM Precision": 0.935064935064935,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.25430724024772644,
      "eval_runtime": 14.6663,
      "eval_samples_per_second": 3.0,
      "eval_steps_per_second": 0.136,
      "step": 1180
    },
    {
      "epoch": 0.3058496002071801,
      "grad_norm": 3.871524702893708,
      "learning_rate": 1.0956909150632127e-06,
      "loss": 0.3914,
      "step": 1181
    },
    {
      "epoch": 0.306108575313198,
      "grad_norm": 4.465570175748092,
      "learning_rate": 1.0953188794574485e-06,
      "loss": 0.4142,
      "step": 1182
    },
    {
      "epoch": 0.30636755041921593,
      "grad_norm": 6.871748586463232,
      "learning_rate": 1.0949464592311203e-06,
      "loss": 0.4566,
      "step": 1183
    },
    {
      "epoch": 0.3066265255252339,
      "grad_norm": 5.5128943255174985,
      "learning_rate": 1.0945736546887888e-06,
      "loss": 0.3573,
      "step": 1184
    },
    {
      "epoch": 0.3068855006312518,
      "grad_norm": 6.356115342327138,
      "learning_rate": 1.0942004661353286e-06,
      "loss": 0.4997,
      "step": 1185
    },
    {
      "epoch": 0.3068855006312518,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.8930817610062893,
      "eval_PRM F1 AUC": 0.819015191199581,
      "eval_PRM F1 AUC (fixed)": 0.7844421162912519,
      "eval_PRM F1 Neg": 0.6792452830188679,
      "eval_PRM NPV": 0.6,
      "eval_PRM Precision": 0.9342105263157895,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.25315046310424805,
      "eval_runtime": 14.4533,
      "eval_samples_per_second": 3.044,
      "eval_steps_per_second": 0.138,
      "step": 1185
    },
    {
      "epoch": 0.3071444757372698,
      "grad_norm": 3.5379039136569537,
      "learning_rate": 1.0938268938759283e-06,
      "loss": 0.379,
      "step": 1186
    },
    {
      "epoch": 0.3074034508432877,
      "grad_norm": 4.202834808405025,
      "learning_rate": 1.0934529382160903e-06,
      "loss": 0.3091,
      "step": 1187
    },
    {
      "epoch": 0.3076624259493056,
      "grad_norm": 3.8281927280708548,
      "learning_rate": 1.0930785994616308e-06,
      "loss": 0.4309,
      "step": 1188
    },
    {
      "epoch": 0.30792140105532356,
      "grad_norm": 6.333845590798601,
      "learning_rate": 1.092703877918679e-06,
      "loss": 0.4023,
      "step": 1189
    },
    {
      "epoch": 0.3081803761613415,
      "grad_norm": 4.1471122361908845,
      "learning_rate": 1.0923287738936771e-06,
      "loss": 0.4087,
      "step": 1190
    },
    {
      "epoch": 0.3081803761613415,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.8903225806451613,
      "eval_PRM F1 AUC": 0.8504452592980618,
      "eval_PRM F1 AUC (fixed)": 0.7799895233106339,
      "eval_PRM F1 Neg": 0.7017543859649122,
      "eval_PRM NPV": 0.5882352941176471,
      "eval_PRM Precision": 0.9583333333333334,
      "eval_PRM Recall": 0.8313253012048193,
      "eval_PRM Specificty": 0.8695652173913043,
      "eval_loss": 0.26387837529182434,
      "eval_runtime": 15.0772,
      "eval_samples_per_second": 2.918,
      "eval_steps_per_second": 0.133,
      "step": 1190
    },
    {
      "epoch": 0.30843935126735944,
      "grad_norm": 3.490983419168074,
      "learning_rate": 1.0919532876933803e-06,
      "loss": 0.4005,
      "step": 1191
    },
    {
      "epoch": 0.30869832637337735,
      "grad_norm": 4.300796741862126,
      "learning_rate": 1.0915774196248564e-06,
      "loss": 0.4054,
      "step": 1192
    },
    {
      "epoch": 0.3089573014793953,
      "grad_norm": 5.991455393032304,
      "learning_rate": 1.091201169995485e-06,
      "loss": 0.3517,
      "step": 1193
    },
    {
      "epoch": 0.30921627658541323,
      "grad_norm": 4.34016764245388,
      "learning_rate": 1.0908245391129584e-06,
      "loss": 0.3691,
      "step": 1194
    },
    {
      "epoch": 0.30947525169143114,
      "grad_norm": 3.006782408187908,
      "learning_rate": 1.0904475272852802e-06,
      "loss": 0.3554,
      "step": 1195
    },
    {
      "epoch": 0.30947525169143114,
      "eval_PRM Accuracy": 0.8301886792452831,
      "eval_PRM F1": 0.8860759493670886,
      "eval_PRM F1 AUC": 0.8129910948140389,
      "eval_PRM F1 AUC (fixed)": 0.788108957569408,
      "eval_PRM F1 Neg": 0.6666666666666666,
      "eval_PRM NPV": 0.5806451612903226,
      "eval_PRM Precision": 0.9333333333333333,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.25778552889823914,
      "eval_runtime": 14.6532,
      "eval_samples_per_second": 3.003,
      "eval_steps_per_second": 0.136,
      "step": 1195
    },
    {
      "epoch": 0.3097342267974491,
      "grad_norm": 7.834331309453101,
      "learning_rate": 1.0900701348207656e-06,
      "loss": 0.3341,
      "step": 1196
    },
    {
      "epoch": 0.309993201903467,
      "grad_norm": 4.837884152961991,
      "learning_rate": 1.0896923620280416e-06,
      "loss": 0.3391,
      "step": 1197
    },
    {
      "epoch": 0.310252177009485,
      "grad_norm": 3.458578607508161,
      "learning_rate": 1.0893142092160451e-06,
      "loss": 0.3735,
      "step": 1198
    },
    {
      "epoch": 0.3105111521155029,
      "grad_norm": 3.243852160400903,
      "learning_rate": 1.0889356766940252e-06,
      "loss": 0.3834,
      "step": 1199
    },
    {
      "epoch": 0.3107701272215208,
      "grad_norm": 4.370429860764285,
      "learning_rate": 1.0885567647715404e-06,
      "loss": 0.2463,
      "step": 1200
    },
    {
      "epoch": 0.3107701272215208,
      "eval_PRM Accuracy": 0.8679245283018868,
      "eval_PRM F1": 0.9135802469135802,
      "eval_PRM F1 AUC": 0.8370874803562075,
      "eval_PRM F1 AUC (fixed)": 0.7998952331063384,
      "eval_PRM F1 Neg": 0.72,
      "eval_PRM NPV": 0.6666666666666666,
      "eval_PRM Precision": 0.9367088607594937,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.25024715065956116,
      "eval_runtime": 14.8849,
      "eval_samples_per_second": 2.956,
      "eval_steps_per_second": 0.134,
      "step": 1200
    },
    {
      "epoch": 0.3110291023275388,
      "grad_norm": 3.9693767544882834,
      "learning_rate": 1.0881774737584597e-06,
      "loss": 0.2953,
      "step": 1201
    },
    {
      "epoch": 0.3112880774335567,
      "grad_norm": 3.45629528989537,
      "learning_rate": 1.0877978039649628e-06,
      "loss": 0.3799,
      "step": 1202
    },
    {
      "epoch": 0.31154705253957465,
      "grad_norm": 3.465792543374586,
      "learning_rate": 1.087417755701538e-06,
      "loss": 0.2799,
      "step": 1203
    },
    {
      "epoch": 0.31180602764559257,
      "grad_norm": 4.392253680638649,
      "learning_rate": 1.0870373292789844e-06,
      "loss": 0.4396,
      "step": 1204
    },
    {
      "epoch": 0.3120650027516105,
      "grad_norm": 7.80351799300313,
      "learning_rate": 1.0866565250084092e-06,
      "loss": 0.3509,
      "step": 1205
    },
    {
      "epoch": 0.3120650027516105,
      "eval_PRM Accuracy": 0.8773584905660378,
      "eval_PRM F1": 0.9212121212121213,
      "eval_PRM F1 AUC": 0.8273965426925091,
      "eval_PRM F1 AUC (fixed)": 0.8140387637506548,
      "eval_PRM F1 Neg": 0.723404255319149,
      "eval_PRM NPV": 0.7083333333333334,
      "eval_PRM Precision": 0.926829268292683,
      "eval_PRM Recall": 0.9156626506024096,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.2483535259962082,
      "eval_runtime": 14.8603,
      "eval_samples_per_second": 2.961,
      "eval_steps_per_second": 0.135,
      "step": 1205
    },
    {
      "epoch": 0.31232397785762844,
      "grad_norm": 5.156928269505667,
      "learning_rate": 1.0862753432012293e-06,
      "loss": 0.2728,
      "step": 1206
    },
    {
      "epoch": 0.31258295296364635,
      "grad_norm": 12.132211890537517,
      "learning_rate": 1.08589378416917e-06,
      "loss": 0.4319,
      "step": 1207
    },
    {
      "epoch": 0.3128419280696643,
      "grad_norm": 6.915697049487834,
      "learning_rate": 1.0855118482242655e-06,
      "loss": 0.4666,
      "step": 1208
    },
    {
      "epoch": 0.31310090317568223,
      "grad_norm": 6.968230933752234,
      "learning_rate": 1.085129535678858e-06,
      "loss": 0.2983,
      "step": 1209
    },
    {
      "epoch": 0.3133598782817002,
      "grad_norm": 3.594238368899922,
      "learning_rate": 1.084746846845597e-06,
      "loss": 0.4571,
      "step": 1210
    },
    {
      "epoch": 0.3133598782817002,
      "eval_PRM Accuracy": 0.8490566037735849,
      "eval_PRM F1": 0.8961038961038961,
      "eval_PRM F1 AUC": 0.8721843897328445,
      "eval_PRM F1 AUC (fixed)": 0.8085385018334206,
      "eval_PRM F1 Neg": 0.7241379310344828,
      "eval_PRM NPV": 0.6,
      "eval_PRM Precision": 0.971830985915493,
      "eval_PRM Recall": 0.8313253012048193,
      "eval_PRM Specificty": 0.9130434782608695,
      "eval_loss": 0.26208773255348206,
      "eval_runtime": 14.5572,
      "eval_samples_per_second": 3.023,
      "eval_steps_per_second": 0.137,
      "step": 1210
    },
    {
      "epoch": 0.3136188533877181,
      "grad_norm": 10.986999998465846,
      "learning_rate": 1.0843637820374415e-06,
      "loss": 0.348,
      "step": 1211
    },
    {
      "epoch": 0.313877828493736,
      "grad_norm": 6.007952874878346,
      "learning_rate": 1.0839803415676555e-06,
      "loss": 0.3719,
      "step": 1212
    },
    {
      "epoch": 0.314136803599754,
      "grad_norm": 6.548551815265104,
      "learning_rate": 1.0835965257498127e-06,
      "loss": 0.3069,
      "step": 1213
    },
    {
      "epoch": 0.3143957787057719,
      "grad_norm": 6.907837361489586,
      "learning_rate": 1.083212334897792e-06,
      "loss": 0.4288,
      "step": 1214
    },
    {
      "epoch": 0.31465475381178987,
      "grad_norm": 4.619638739193467,
      "learning_rate": 1.0828277693257797e-06,
      "loss": 0.3227,
      "step": 1215
    },
    {
      "epoch": 0.31465475381178987,
      "eval_PRM Accuracy": 0.8490566037735849,
      "eval_PRM F1": 0.8961038961038961,
      "eval_PRM F1 AUC": 0.8721843897328445,
      "eval_PRM F1 AUC (fixed)": 0.8129910948140388,
      "eval_PRM F1 Neg": 0.7241379310344828,
      "eval_PRM NPV": 0.6,
      "eval_PRM Precision": 0.971830985915493,
      "eval_PRM Recall": 0.8313253012048193,
      "eval_PRM Specificty": 0.9130434782608695,
      "eval_loss": 0.2673983871936798,
      "eval_runtime": 14.6708,
      "eval_samples_per_second": 2.999,
      "eval_steps_per_second": 0.136,
      "step": 1215
    },
    {
      "epoch": 0.3149137289178078,
      "grad_norm": 6.725501666695132,
      "learning_rate": 1.0824428293482682e-06,
      "loss": 0.3644,
      "step": 1216
    },
    {
      "epoch": 0.3151727040238257,
      "grad_norm": 4.1162153720484715,
      "learning_rate": 1.082057515280057e-06,
      "loss": 0.371,
      "step": 1217
    },
    {
      "epoch": 0.31543167912984366,
      "grad_norm": 6.834652646331434,
      "learning_rate": 1.0816718274362505e-06,
      "loss": 0.2873,
      "step": 1218
    },
    {
      "epoch": 0.31569065423586157,
      "grad_norm": 3.0933559887087174,
      "learning_rate": 1.0812857661322589e-06,
      "loss": 0.3626,
      "step": 1219
    },
    {
      "epoch": 0.31594962934187953,
      "grad_norm": 3.994407339934042,
      "learning_rate": 1.0808993316837982e-06,
      "loss": 0.418,
      "step": 1220
    },
    {
      "epoch": 0.31594962934187953,
      "eval_PRM Accuracy": 0.8867924528301887,
      "eval_PRM F1": 0.926829268292683,
      "eval_PRM F1 AUC": 0.8491356731272918,
      "eval_PRM F1 AUC (fixed)": 0.8122053431115768,
      "eval_PRM F1 Neg": 0.75,
      "eval_PRM NPV": 0.72,
      "eval_PRM Precision": 0.9382716049382716,
      "eval_PRM Recall": 0.9156626506024096,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.24850182235240936,
      "eval_runtime": 15.0241,
      "eval_samples_per_second": 2.929,
      "eval_steps_per_second": 0.133,
      "step": 1220
    },
    {
      "epoch": 0.31620860444789745,
      "grad_norm": 3.5620307606032946,
      "learning_rate": 1.0805125244068892e-06,
      "loss": 0.3168,
      "step": 1221
    },
    {
      "epoch": 0.31646757955391536,
      "grad_norm": 7.342654037434144,
      "learning_rate": 1.0801253446178583e-06,
      "loss": 0.4416,
      "step": 1222
    },
    {
      "epoch": 0.3167265546599333,
      "grad_norm": 9.728744933510296,
      "learning_rate": 1.0797377926333355e-06,
      "loss": 0.4929,
      "step": 1223
    },
    {
      "epoch": 0.31698552976595123,
      "grad_norm": 6.491475501077239,
      "learning_rate": 1.079349868770256e-06,
      "loss": 0.4135,
      "step": 1224
    },
    {
      "epoch": 0.3172445048719692,
      "grad_norm": 6.106952798989202,
      "learning_rate": 1.0789615733458586e-06,
      "loss": 0.4119,
      "step": 1225
    },
    {
      "epoch": 0.3172445048719692,
      "eval_PRM Accuracy": 0.8584905660377359,
      "eval_PRM F1": 0.906832298136646,
      "eval_PRM F1 AUC": 0.8310633839706653,
      "eval_PRM F1 AUC (fixed)": 0.7930853850183341,
      "eval_PRM F1 Neg": 0.7058823529411765,
      "eval_PRM NPV": 0.6428571428571429,
      "eval_PRM Precision": 0.9358974358974359,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.2541673481464386,
      "eval_runtime": 14.5856,
      "eval_samples_per_second": 3.017,
      "eval_steps_per_second": 0.137,
      "step": 1225
    },
    {
      "epoch": 0.3175034799779871,
      "grad_norm": 3.0338791452656144,
      "learning_rate": 1.0785729066776864e-06,
      "loss": 0.3631,
      "step": 1226
    },
    {
      "epoch": 0.317762455084005,
      "grad_norm": 3.6359197190099266,
      "learning_rate": 1.0781838690835856e-06,
      "loss": 0.3624,
      "step": 1227
    },
    {
      "epoch": 0.318021430190023,
      "grad_norm": 4.79808651922732,
      "learning_rate": 1.0777944608817061e-06,
      "loss": 0.353,
      "step": 1228
    },
    {
      "epoch": 0.3182804052960409,
      "grad_norm": 3.2920462609639523,
      "learning_rate": 1.0774046823905007e-06,
      "loss": 0.3106,
      "step": 1229
    },
    {
      "epoch": 0.31853938040205887,
      "grad_norm": 5.105740527499564,
      "learning_rate": 1.0770145339287254e-06,
      "loss": 0.3778,
      "step": 1230
    },
    {
      "epoch": 0.31853938040205887,
      "eval_PRM Accuracy": 0.8301886792452831,
      "eval_PRM F1": 0.8860759493670886,
      "eval_PRM F1 AUC": 0.8129910948140389,
      "eval_PRM F1 AUC (fixed)": 0.7794656888423258,
      "eval_PRM F1 Neg": 0.6666666666666666,
      "eval_PRM NPV": 0.5806451612903226,
      "eval_PRM Precision": 0.9333333333333333,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.264245867729187,
      "eval_runtime": 14.9669,
      "eval_samples_per_second": 2.94,
      "eval_steps_per_second": 0.134,
      "step": 1230
    },
    {
      "epoch": 0.3187983555080768,
      "grad_norm": 2.5851477388861754,
      "learning_rate": 1.0766240158154382e-06,
      "loss": 0.3024,
      "step": 1231
    },
    {
      "epoch": 0.31905733061409475,
      "grad_norm": 5.233677920244123,
      "learning_rate": 1.0762331283699997e-06,
      "loss": 0.3715,
      "step": 1232
    },
    {
      "epoch": 0.31931630572011266,
      "grad_norm": 2.695238376578191,
      "learning_rate": 1.0758418719120723e-06,
      "loss": 0.3182,
      "step": 1233
    },
    {
      "epoch": 0.31957528082613057,
      "grad_norm": 5.177819112800959,
      "learning_rate": 1.0754502467616207e-06,
      "loss": 0.3243,
      "step": 1234
    },
    {
      "epoch": 0.31983425593214854,
      "grad_norm": 4.881959830665304,
      "learning_rate": 1.0750582532389107e-06,
      "loss": 0.3412,
      "step": 1235
    },
    {
      "epoch": 0.31983425593214854,
      "eval_PRM Accuracy": 0.8301886792452831,
      "eval_PRM F1": 0.8860759493670886,
      "eval_PRM F1 AUC": 0.8129910948140389,
      "eval_PRM F1 AUC (fixed)": 0.7802514405447879,
      "eval_PRM F1 Neg": 0.6666666666666666,
      "eval_PRM NPV": 0.5806451612903226,
      "eval_PRM Precision": 0.9333333333333333,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.26019617915153503,
      "eval_runtime": 14.7283,
      "eval_samples_per_second": 2.987,
      "eval_steps_per_second": 0.136,
      "step": 1235
    },
    {
      "epoch": 0.32009323103816645,
      "grad_norm": 3.170497689713328,
      "learning_rate": 1.074665891664509e-06,
      "loss": 0.3291,
      "step": 1236
    },
    {
      "epoch": 0.3203522061441844,
      "grad_norm": 4.371669431913452,
      "learning_rate": 1.0742731623592846e-06,
      "loss": 0.3604,
      "step": 1237
    },
    {
      "epoch": 0.3206111812502023,
      "grad_norm": 6.077137989216843,
      "learning_rate": 1.0738800656444053e-06,
      "loss": 0.3834,
      "step": 1238
    },
    {
      "epoch": 0.32087015635622024,
      "grad_norm": 4.292347676028572,
      "learning_rate": 1.0734866018413413e-06,
      "loss": 0.3677,
      "step": 1239
    },
    {
      "epoch": 0.3211291314622382,
      "grad_norm": 6.327860085742873,
      "learning_rate": 1.0730927712718614e-06,
      "loss": 0.3754,
      "step": 1240
    },
    {
      "epoch": 0.3211291314622382,
      "eval_PRM Accuracy": 0.8773584905660378,
      "eval_PRM F1": 0.9202453987730062,
      "eval_PRM F1 AUC": 0.8431115767417497,
      "eval_PRM F1 AUC (fixed)": 0.7854897852278679,
      "eval_PRM F1 Neg": 0.7346938775510204,
      "eval_PRM NPV": 0.6923076923076923,
      "eval_PRM Precision": 0.9375,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.2534755766391754,
      "eval_runtime": 14.3876,
      "eval_samples_per_second": 3.058,
      "eval_steps_per_second": 0.139,
      "step": 1240
    },
    {
      "epoch": 0.3213881065682561,
      "grad_norm": 5.573037635059935,
      "learning_rate": 1.0726985742580355e-06,
      "loss": 0.399,
      "step": 1241
    },
    {
      "epoch": 0.3216470816742741,
      "grad_norm": 5.900603463933119,
      "learning_rate": 1.0723040111222325e-06,
      "loss": 0.3801,
      "step": 1242
    },
    {
      "epoch": 0.321906056780292,
      "grad_norm": 7.0207752592900485,
      "learning_rate": 1.0719090821871208e-06,
      "loss": 0.3531,
      "step": 1243
    },
    {
      "epoch": 0.3221650318863099,
      "grad_norm": 4.396846712017389,
      "learning_rate": 1.0715137877756683e-06,
      "loss": 0.3045,
      "step": 1244
    },
    {
      "epoch": 0.32242400699232787,
      "grad_norm": 3.10027769254378,
      "learning_rate": 1.0711181282111415e-06,
      "loss": 0.3093,
      "step": 1245
    },
    {
      "epoch": 0.32242400699232787,
      "eval_PRM Accuracy": 0.8301886792452831,
      "eval_PRM F1": 0.8860759493670886,
      "eval_PRM F1 AUC": 0.8129910948140389,
      "eval_PRM F1 AUC (fixed)": 0.7805133577789417,
      "eval_PRM F1 Neg": 0.6666666666666666,
      "eval_PRM NPV": 0.5806451612903226,
      "eval_PRM Precision": 0.9333333333333333,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.25553369522094727,
      "eval_runtime": 14.6501,
      "eval_samples_per_second": 3.003,
      "eval_steps_per_second": 0.137,
      "step": 1245
    },
    {
      "epoch": 0.3226829820983458,
      "grad_norm": 3.5643964755719777,
      "learning_rate": 1.0707221038171056e-06,
      "loss": 0.362,
      "step": 1246
    },
    {
      "epoch": 0.32294195720436375,
      "grad_norm": 4.684952259372974,
      "learning_rate": 1.0703257149174238e-06,
      "loss": 0.2909,
      "step": 1247
    },
    {
      "epoch": 0.32320093231038166,
      "grad_norm": 5.515834741111073,
      "learning_rate": 1.0699289618362579e-06,
      "loss": 0.3027,
      "step": 1248
    },
    {
      "epoch": 0.3234599074163996,
      "grad_norm": 3.841355440414757,
      "learning_rate": 1.0695318448980677e-06,
      "loss": 0.3314,
      "step": 1249
    },
    {
      "epoch": 0.32371888252241754,
      "grad_norm": 3.259683425761671,
      "learning_rate": 1.0691343644276095e-06,
      "loss": 0.3673,
      "step": 1250
    },
    {
      "epoch": 0.32371888252241754,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8789808917197452,
      "eval_PRM F1 AUC": 0.8069669984284967,
      "eval_PRM F1 AUC (fixed)": 0.7815610267155578,
      "eval_PRM F1 Neg": 0.6545454545454545,
      "eval_PRM NPV": 0.5625,
      "eval_PRM Precision": 0.9324324324324325,
      "eval_PRM Recall": 0.8313253012048193,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.2531985342502594,
      "eval_runtime": 14.525,
      "eval_samples_per_second": 3.029,
      "eval_steps_per_second": 0.138,
      "step": 1250
    },
    {
      "epoch": 0.32397785762843545,
      "grad_norm": 3.9427534379874714,
      "learning_rate": 1.0687365207499381e-06,
      "loss": 0.3362,
      "step": 1251
    },
    {
      "epoch": 0.3242368327344534,
      "grad_norm": 4.561687981910411,
      "learning_rate": 1.0683383141904046e-06,
      "loss": 0.2931,
      "step": 1252
    },
    {
      "epoch": 0.32449580784047133,
      "grad_norm": 4.234865118537673,
      "learning_rate": 1.0679397450746573e-06,
      "loss": 0.3682,
      "step": 1253
    },
    {
      "epoch": 0.3247547829464893,
      "grad_norm": 5.463852961847639,
      "learning_rate": 1.0675408137286403e-06,
      "loss": 0.3418,
      "step": 1254
    },
    {
      "epoch": 0.3250137580525072,
      "grad_norm": 3.8455546613574847,
      "learning_rate": 1.0671415204785948e-06,
      "loss": 0.3141,
      "step": 1255
    },
    {
      "epoch": 0.3250137580525072,
      "eval_PRM Accuracy": 0.8490566037735849,
      "eval_PRM F1": 0.9,
      "eval_PRM F1 AUC": 0.8250392875851232,
      "eval_PRM F1 AUC (fixed)": 0.7820848611838659,
      "eval_PRM F1 Neg": 0.6923076923076923,
      "eval_PRM NPV": 0.6206896551724138,
      "eval_PRM Precision": 0.935064935064935,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.2530316114425659,
      "eval_runtime": 14.6154,
      "eval_samples_per_second": 3.011,
      "eval_steps_per_second": 0.137,
      "step": 1255
    },
    {
      "epoch": 0.3252727331585251,
      "grad_norm": 3.095249401099387,
      "learning_rate": 1.066741865651057e-06,
      "loss": 0.3307,
      "step": 1256
    },
    {
      "epoch": 0.3255317082645431,
      "grad_norm": 4.028023928603981,
      "learning_rate": 1.06634184957286e-06,
      "loss": 0.4218,
      "step": 1257
    },
    {
      "epoch": 0.325790683370561,
      "grad_norm": 3.2652804916367,
      "learning_rate": 1.0659414725711314e-06,
      "loss": 0.3492,
      "step": 1258
    },
    {
      "epoch": 0.32604965847657896,
      "grad_norm": 6.934006280547388,
      "learning_rate": 1.0655407349732942e-06,
      "loss": 0.3953,
      "step": 1259
    },
    {
      "epoch": 0.3263086335825969,
      "grad_norm": 5.3012226263917,
      "learning_rate": 1.065139637107066e-06,
      "loss": 0.3463,
      "step": 1260
    },
    {
      "epoch": 0.3263086335825969,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.8930817610062893,
      "eval_PRM F1 AUC": 0.819015191199581,
      "eval_PRM F1 AUC (fixed)": 0.782870612886328,
      "eval_PRM F1 Neg": 0.6792452830188679,
      "eval_PRM NPV": 0.6,
      "eval_PRM Precision": 0.9342105263157895,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.25366419553756714,
      "eval_runtime": 14.8022,
      "eval_samples_per_second": 2.973,
      "eval_steps_per_second": 0.135,
      "step": 1260
    },
    {
      "epoch": 0.3265676086886148,
      "grad_norm": 5.814350884881637,
      "learning_rate": 1.06473817930046e-06,
      "loss": 0.2951,
      "step": 1261
    },
    {
      "epoch": 0.32682658379463275,
      "grad_norm": 5.444085139661884,
      "learning_rate": 1.0643363618817825e-06,
      "loss": 0.4134,
      "step": 1262
    },
    {
      "epoch": 0.32708555890065066,
      "grad_norm": 9.9526531408028,
      "learning_rate": 1.0639341851796348e-06,
      "loss": 0.5132,
      "step": 1263
    },
    {
      "epoch": 0.32734453400666863,
      "grad_norm": 4.3954758322259835,
      "learning_rate": 1.0635316495229113e-06,
      "loss": 0.2878,
      "step": 1264
    },
    {
      "epoch": 0.32760350911268654,
      "grad_norm": 4.885386307335036,
      "learning_rate": 1.063128755240801e-06,
      "loss": 0.3412,
      "step": 1265
    },
    {
      "epoch": 0.32760350911268654,
      "eval_PRM Accuracy": 0.8301886792452831,
      "eval_PRM F1": 0.8831168831168831,
      "eval_PRM F1 AUC": 0.8444211629125196,
      "eval_PRM F1 AUC (fixed)": 0.7799895233106339,
      "eval_PRM F1 Neg": 0.6896551724137931,
      "eval_PRM NPV": 0.5714285714285714,
      "eval_PRM Precision": 0.9577464788732394,
      "eval_PRM Recall": 0.8192771084337349,
      "eval_PRM Specificty": 0.8695652173913043,
      "eval_loss": 0.2644989490509033,
      "eval_runtime": 14.5446,
      "eval_samples_per_second": 3.025,
      "eval_steps_per_second": 0.138,
      "step": 1265
    },
    {
      "epoch": 0.32786248421870445,
      "grad_norm": 2.883792563856442,
      "learning_rate": 1.062725502662785e-06,
      "loss": 0.3397,
      "step": 1266
    },
    {
      "epoch": 0.3281214593247224,
      "grad_norm": 2.666039333847741,
      "learning_rate": 1.062321892118638e-06,
      "loss": 0.2813,
      "step": 1267
    },
    {
      "epoch": 0.32838043443074033,
      "grad_norm": 3.2249721487819643,
      "learning_rate": 1.0619179239384274e-06,
      "loss": 0.2993,
      "step": 1268
    },
    {
      "epoch": 0.3286394095367583,
      "grad_norm": 4.124141804714215,
      "learning_rate": 1.0615135984525133e-06,
      "loss": 0.3636,
      "step": 1269
    },
    {
      "epoch": 0.3288983846427762,
      "grad_norm": 4.252191767294118,
      "learning_rate": 1.0611089159915473e-06,
      "loss": 0.2837,
      "step": 1270
    },
    {
      "epoch": 0.3288983846427762,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8789808917197452,
      "eval_PRM F1 AUC": 0.8069669984284967,
      "eval_PRM F1 AUC (fixed)": 0.7852278679937139,
      "eval_PRM F1 Neg": 0.6545454545454545,
      "eval_PRM NPV": 0.5625,
      "eval_PRM Precision": 0.9324324324324325,
      "eval_PRM Recall": 0.8313253012048193,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.2600552439689636,
      "eval_runtime": 14.5574,
      "eval_samples_per_second": 3.023,
      "eval_steps_per_second": 0.137,
      "step": 1270
    },
    {
      "epoch": 0.3291573597487941,
      "grad_norm": 3.4884179116143366,
      "learning_rate": 1.0607038768864737e-06,
      "loss": 0.2719,
      "step": 1271
    },
    {
      "epoch": 0.3294163348548121,
      "grad_norm": 7.575658883276662,
      "learning_rate": 1.060298481468528e-06,
      "loss": 0.4286,
      "step": 1272
    },
    {
      "epoch": 0.32967530996083,
      "grad_norm": 2.7108707510626338,
      "learning_rate": 1.0598927300692372e-06,
      "loss": 0.2576,
      "step": 1273
    },
    {
      "epoch": 0.32993428506684797,
      "grad_norm": 4.385897164351068,
      "learning_rate": 1.0594866230204195e-06,
      "loss": 0.2862,
      "step": 1274
    },
    {
      "epoch": 0.3301932601728659,
      "grad_norm": 3.3979497479914396,
      "learning_rate": 1.0590801606541839e-06,
      "loss": 0.4011,
      "step": 1275
    },
    {
      "epoch": 0.3301932601728659,
      "eval_PRM Accuracy": 0.8679245283018868,
      "eval_PRM F1": 0.9135802469135802,
      "eval_PRM F1 AUC": 0.8370874803562075,
      "eval_PRM F1 AUC (fixed)": 0.7878470403352541,
      "eval_PRM F1 Neg": 0.72,
      "eval_PRM NPV": 0.6666666666666666,
      "eval_PRM Precision": 0.9367088607594937,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.25437629222869873,
      "eval_runtime": 14.7355,
      "eval_samples_per_second": 2.986,
      "eval_steps_per_second": 0.136,
      "step": 1275
    },
    {
      "epoch": 0.33045223527888384,
      "grad_norm": 3.0241464378789225,
      "learning_rate": 1.0586733433029298e-06,
      "loss": 0.3214,
      "step": 1276
    },
    {
      "epoch": 0.33071121038490175,
      "grad_norm": 4.362546620704584,
      "learning_rate": 1.058266171299347e-06,
      "loss": 0.3757,
      "step": 1277
    },
    {
      "epoch": 0.33097018549091967,
      "grad_norm": 3.935891246180922,
      "learning_rate": 1.0578586449764157e-06,
      "loss": 0.3495,
      "step": 1278
    },
    {
      "epoch": 0.33122916059693763,
      "grad_norm": 3.5114047359210065,
      "learning_rate": 1.0574507646674054e-06,
      "loss": 0.3644,
      "step": 1279
    },
    {
      "epoch": 0.33148813570295554,
      "grad_norm": 6.0952351579317625,
      "learning_rate": 1.0570425307058752e-06,
      "loss": 0.4252,
      "step": 1280
    },
    {
      "epoch": 0.33148813570295554,
      "eval_PRM Accuracy": 0.8867924528301887,
      "eval_PRM F1": 0.926829268292683,
      "eval_PRM F1 AUC": 0.8491356731272918,
      "eval_PRM F1 AUC (fixed)": 0.7930853850183343,
      "eval_PRM F1 Neg": 0.75,
      "eval_PRM NPV": 0.72,
      "eval_PRM Precision": 0.9382716049382716,
      "eval_PRM Recall": 0.9156626506024096,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.2512930631637573,
      "eval_runtime": 14.9129,
      "eval_samples_per_second": 2.95,
      "eval_steps_per_second": 0.134,
      "step": 1280
    },
    {
      "epoch": 0.3317471108089735,
      "grad_norm": 10.255531897227266,
      "learning_rate": 1.0566339434256734e-06,
      "loss": 0.3373,
      "step": 1281
    },
    {
      "epoch": 0.3320060859149914,
      "grad_norm": 2.6011141556068256,
      "learning_rate": 1.0562250031609372e-06,
      "loss": 0.201,
      "step": 1282
    },
    {
      "epoch": 0.33226506102100933,
      "grad_norm": 5.756435201271477,
      "learning_rate": 1.0558157102460929e-06,
      "loss": 0.3607,
      "step": 1283
    },
    {
      "epoch": 0.3325240361270273,
      "grad_norm": 3.256313178124926,
      "learning_rate": 1.0554060650158543e-06,
      "loss": 0.293,
      "step": 1284
    },
    {
      "epoch": 0.3327830112330452,
      "grad_norm": 4.038686227297466,
      "learning_rate": 1.0549960678052243e-06,
      "loss": 0.3206,
      "step": 1285
    },
    {
      "epoch": 0.3327830112330452,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.8930817610062893,
      "eval_PRM F1 AUC": 0.819015191199581,
      "eval_PRM F1 AUC (fixed)": 0.7886327920377161,
      "eval_PRM F1 Neg": 0.6792452830188679,
      "eval_PRM NPV": 0.6,
      "eval_PRM Precision": 0.9342105263157895,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.251687616109848,
      "eval_runtime": 14.924,
      "eval_samples_per_second": 2.948,
      "eval_steps_per_second": 0.134,
      "step": 1285
    },
    {
      "epoch": 0.3330419863390632,
      "grad_norm": 5.903103517069564,
      "learning_rate": 1.054585718949493e-06,
      "loss": 0.4008,
      "step": 1286
    },
    {
      "epoch": 0.3333009614450811,
      "grad_norm": 3.6280182528048797,
      "learning_rate": 1.0541750187842384e-06,
      "loss": 0.4035,
      "step": 1287
    },
    {
      "epoch": 0.333559936551099,
      "grad_norm": 7.746890673662093,
      "learning_rate": 1.0537639676453254e-06,
      "loss": 0.2507,
      "step": 1288
    },
    {
      "epoch": 0.33381891165711697,
      "grad_norm": 4.728549191922307,
      "learning_rate": 1.0533525658689064e-06,
      "loss": 0.3587,
      "step": 1289
    },
    {
      "epoch": 0.3340778867631349,
      "grad_norm": 8.67075750836026,
      "learning_rate": 1.0529408137914204e-06,
      "loss": 0.2709,
      "step": 1290
    },
    {
      "epoch": 0.3340778867631349,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.8717948717948718,
      "eval_PRM F1 AUC": 0.8009429020429545,
      "eval_PRM F1 AUC (fixed)": 0.7936092194866423,
      "eval_PRM F1 Neg": 0.6428571428571429,
      "eval_PRM NPV": 0.5454545454545454,
      "eval_PRM Precision": 0.9315068493150684,
      "eval_PRM Recall": 0.8192771084337349,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.2578117549419403,
      "eval_runtime": 14.7797,
      "eval_samples_per_second": 2.977,
      "eval_steps_per_second": 0.135,
      "step": 1290
    },
    {
      "epoch": 0.33433686186915285,
      "grad_norm": 4.534289687153858,
      "learning_rate": 1.0525287117495928e-06,
      "loss": 0.3352,
      "step": 1291
    },
    {
      "epoch": 0.33459583697517076,
      "grad_norm": 2.4350765866908084,
      "learning_rate": 1.0521162600804348e-06,
      "loss": 0.2627,
      "step": 1292
    },
    {
      "epoch": 0.33485481208118867,
      "grad_norm": 2.993341013394517,
      "learning_rate": 1.0517034591212447e-06,
      "loss": 0.3248,
      "step": 1293
    },
    {
      "epoch": 0.33511378718720664,
      "grad_norm": 6.045274041684217,
      "learning_rate": 1.0512903092096048e-06,
      "loss": 0.4829,
      "step": 1294
    },
    {
      "epoch": 0.33537276229322455,
      "grad_norm": 4.920969311687047,
      "learning_rate": 1.0508768106833844e-06,
      "loss": 0.4081,
      "step": 1295
    },
    {
      "epoch": 0.33537276229322455,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8789808917197452,
      "eval_PRM F1 AUC": 0.8069669984284967,
      "eval_PRM F1 AUC (fixed)": 0.8067050811943426,
      "eval_PRM F1 Neg": 0.6545454545454545,
      "eval_PRM NPV": 0.5625,
      "eval_PRM Precision": 0.9324324324324325,
      "eval_PRM Recall": 0.8313253012048193,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.24843551218509674,
      "eval_runtime": 14.645,
      "eval_samples_per_second": 3.004,
      "eval_steps_per_second": 0.137,
      "step": 1295
    },
    {
      "epoch": 0.3356317373992425,
      "grad_norm": 3.8198314302235783,
      "learning_rate": 1.0504629638807365e-06,
      "loss": 0.3589,
      "step": 1296
    },
    {
      "epoch": 0.3358907125052604,
      "grad_norm": 3.2373637606236456,
      "learning_rate": 1.0500487691400998e-06,
      "loss": 0.3175,
      "step": 1297
    },
    {
      "epoch": 0.3361496876112784,
      "grad_norm": 4.29295357505977,
      "learning_rate": 1.0496342268001975e-06,
      "loss": 0.3697,
      "step": 1298
    },
    {
      "epoch": 0.3364086627172963,
      "grad_norm": 5.516686019813932,
      "learning_rate": 1.0492193372000364e-06,
      "loss": 0.324,
      "step": 1299
    },
    {
      "epoch": 0.3366676378233142,
      "grad_norm": 3.812344311395989,
      "learning_rate": 1.0488041006789084e-06,
      "loss": 0.4356,
      "step": 1300
    },
    {
      "epoch": 0.3366676378233142,
      "eval_PRM Accuracy": 0.8301886792452831,
      "eval_PRM F1": 0.8860759493670886,
      "eval_PRM F1 AUC": 0.8129910948140389,
      "eval_PRM F1 AUC (fixed)": 0.8040859088528025,
      "eval_PRM F1 Neg": 0.6666666666666666,
      "eval_PRM NPV": 0.5806451612903226,
      "eval_PRM Precision": 0.9333333333333333,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.24917888641357422,
      "eval_runtime": 14.8531,
      "eval_samples_per_second": 2.962,
      "eval_steps_per_second": 0.135,
      "step": 1300
    },
    {
      "epoch": 0.3369266129293322,
      "grad_norm": 3.196056765448634,
      "learning_rate": 1.0483885175763879e-06,
      "loss": 0.3073,
      "step": 1301
    },
    {
      "epoch": 0.3371855880353501,
      "grad_norm": 4.95895879113035,
      "learning_rate": 1.047972588232333e-06,
      "loss": 0.4303,
      "step": 1302
    },
    {
      "epoch": 0.33744456314136806,
      "grad_norm": 2.4644183637611103,
      "learning_rate": 1.047556312986886e-06,
      "loss": 0.2904,
      "step": 1303
    },
    {
      "epoch": 0.33770353824738597,
      "grad_norm": 5.294206024896391,
      "learning_rate": 1.0471396921804708e-06,
      "loss": 0.3325,
      "step": 1304
    },
    {
      "epoch": 0.3379625133534039,
      "grad_norm": 3.3704941228453094,
      "learning_rate": 1.0467227261537944e-06,
      "loss": 0.2584,
      "step": 1305
    },
    {
      "epoch": 0.3379625133534039,
      "eval_PRM Accuracy": 0.8584905660377359,
      "eval_PRM F1": 0.906832298136646,
      "eval_PRM F1 AUC": 0.8310633839706653,
      "eval_PRM F1 AUC (fixed)": 0.7954426401257203,
      "eval_PRM F1 Neg": 0.7058823529411765,
      "eval_PRM NPV": 0.6428571428571429,
      "eval_PRM Precision": 0.9358974358974359,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.25340020656585693,
      "eval_runtime": 14.6219,
      "eval_samples_per_second": 3.009,
      "eval_steps_per_second": 0.137,
      "step": 1305
    },
    {
      "epoch": 0.33822148845942185,
      "grad_norm": 4.249922053779406,
      "learning_rate": 1.0463054152478464e-06,
      "loss": 0.3716,
      "step": 1306
    },
    {
      "epoch": 0.33848046356543976,
      "grad_norm": 4.501080318218483,
      "learning_rate": 1.0458877598038978e-06,
      "loss": 0.2519,
      "step": 1307
    },
    {
      "epoch": 0.3387394386714577,
      "grad_norm": 2.8975562640932733,
      "learning_rate": 1.045469760163502e-06,
      "loss": 0.3074,
      "step": 1308
    },
    {
      "epoch": 0.33899841377747564,
      "grad_norm": 8.021087356933908,
      "learning_rate": 1.0450514166684933e-06,
      "loss": 0.3931,
      "step": 1309
    },
    {
      "epoch": 0.33925738888349355,
      "grad_norm": 7.004507103777712,
      "learning_rate": 1.0446327296609876e-06,
      "loss": 0.4529,
      "step": 1310
    },
    {
      "epoch": 0.33925738888349355,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.8717948717948718,
      "eval_PRM F1 AUC": 0.8009429020429545,
      "eval_PRM F1 AUC (fixed)": 0.7833944473546359,
      "eval_PRM F1 Neg": 0.6428571428571429,
      "eval_PRM NPV": 0.5454545454545454,
      "eval_PRM Precision": 0.9315068493150684,
      "eval_PRM Recall": 0.8192771084337349,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.2645565867424011,
      "eval_runtime": 15.0492,
      "eval_samples_per_second": 2.924,
      "eval_steps_per_second": 0.133,
      "step": 1310
    },
    {
      "epoch": 0.3395163639895115,
      "grad_norm": 5.034753487824166,
      "learning_rate": 1.0442136994833815e-06,
      "loss": 0.3002,
      "step": 1311
    },
    {
      "epoch": 0.3397753390955294,
      "grad_norm": 4.159325525124912,
      "learning_rate": 1.0437943264783528e-06,
      "loss": 0.4791,
      "step": 1312
    },
    {
      "epoch": 0.3400343142015474,
      "grad_norm": 4.685979244764695,
      "learning_rate": 1.0433746109888585e-06,
      "loss": 0.3015,
      "step": 1313
    },
    {
      "epoch": 0.3402932893075653,
      "grad_norm": 3.849565505765079,
      "learning_rate": 1.042954553358137e-06,
      "loss": 0.333,
      "step": 1314
    },
    {
      "epoch": 0.3405522644135832,
      "grad_norm": 3.645168751223873,
      "learning_rate": 1.0425341539297054e-06,
      "loss": 0.3582,
      "step": 1315
    },
    {
      "epoch": 0.3405522644135832,
      "eval_PRM Accuracy": 0.8018867924528302,
      "eval_PRM F1": 0.864516129032258,
      "eval_PRM F1 AUC": 0.7949188056574124,
      "eval_PRM F1 AUC (fixed)": 0.7778941854374017,
      "eval_PRM F1 Neg": 0.631578947368421,
      "eval_PRM NPV": 0.5294117647058824,
      "eval_PRM Precision": 0.9305555555555556,
      "eval_PRM Recall": 0.8072289156626506,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.27547842264175415,
      "eval_runtime": 14.6792,
      "eval_samples_per_second": 2.997,
      "eval_steps_per_second": 0.136,
      "step": 1315
    },
    {
      "epoch": 0.3408112395196012,
      "grad_norm": 2.9138626397617093,
      "learning_rate": 1.0421134130473612e-06,
      "loss": 0.3854,
      "step": 1316
    },
    {
      "epoch": 0.3410702146256191,
      "grad_norm": 2.857443915356315,
      "learning_rate": 1.0416923310551805e-06,
      "loss": 0.3204,
      "step": 1317
    },
    {
      "epoch": 0.34132918973163706,
      "grad_norm": 6.003891257495076,
      "learning_rate": 1.0412709082975184e-06,
      "loss": 0.2993,
      "step": 1318
    },
    {
      "epoch": 0.341588164837655,
      "grad_norm": 4.153935299144768,
      "learning_rate": 1.0408491451190093e-06,
      "loss": 0.389,
      "step": 1319
    },
    {
      "epoch": 0.34184713994367294,
      "grad_norm": 4.6586000639572305,
      "learning_rate": 1.0404270418645654e-06,
      "loss": 0.3316,
      "step": 1320
    },
    {
      "epoch": 0.34184713994367294,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8789808917197452,
      "eval_PRM F1 AUC": 0.8069669984284967,
      "eval_PRM F1 AUC (fixed)": 0.7768465165007857,
      "eval_PRM F1 Neg": 0.6545454545454545,
      "eval_PRM NPV": 0.5625,
      "eval_PRM Precision": 0.9324324324324325,
      "eval_PRM Recall": 0.8313253012048193,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.27451756596565247,
      "eval_runtime": 15.085,
      "eval_samples_per_second": 2.917,
      "eval_steps_per_second": 0.133,
      "step": 1320
    },
    {
      "epoch": 0.34210611504969085,
      "grad_norm": 4.479214897680437,
      "learning_rate": 1.0400045988793768e-06,
      "loss": 0.3974,
      "step": 1321
    },
    {
      "epoch": 0.34236509015570876,
      "grad_norm": 3.175295099148635,
      "learning_rate": 1.0395818165089125e-06,
      "loss": 0.3704,
      "step": 1322
    },
    {
      "epoch": 0.34262406526172673,
      "grad_norm": 4.047456516055003,
      "learning_rate": 1.039158695098918e-06,
      "loss": 0.3562,
      "step": 1323
    },
    {
      "epoch": 0.34288304036774464,
      "grad_norm": 3.496307850241482,
      "learning_rate": 1.0387352349954158e-06,
      "loss": 0.3671,
      "step": 1324
    },
    {
      "epoch": 0.3431420154737626,
      "grad_norm": 6.735416072169709,
      "learning_rate": 1.038311436544707e-06,
      "loss": 0.4127,
      "step": 1325
    },
    {
      "epoch": 0.3431420154737626,
      "eval_PRM Accuracy": 0.8301886792452831,
      "eval_PRM F1": 0.8860759493670886,
      "eval_PRM F1 AUC": 0.8129910948140389,
      "eval_PRM F1 AUC (fixed)": 0.783918281822944,
      "eval_PRM F1 Neg": 0.6666666666666666,
      "eval_PRM NPV": 0.5806451612903226,
      "eval_PRM Precision": 0.9333333333333333,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.2647566795349121,
      "eval_runtime": 14.6853,
      "eval_samples_per_second": 2.996,
      "eval_steps_per_second": 0.136,
      "step": 1325
    },
    {
      "epoch": 0.3434009905797805,
      "grad_norm": 3.7720284176982295,
      "learning_rate": 1.0378873000933675e-06,
      "loss": 0.2992,
      "step": 1326
    },
    {
      "epoch": 0.34365996568579843,
      "grad_norm": 10.021450252138205,
      "learning_rate": 1.037462825988251e-06,
      "loss": 0.4193,
      "step": 1327
    },
    {
      "epoch": 0.3439189407918164,
      "grad_norm": 3.72757558238376,
      "learning_rate": 1.0370380145764868e-06,
      "loss": 0.2907,
      "step": 1328
    },
    {
      "epoch": 0.3441779158978343,
      "grad_norm": 5.12031159635696,
      "learning_rate": 1.03661286620548e-06,
      "loss": 0.4486,
      "step": 1329
    },
    {
      "epoch": 0.3444368910038523,
      "grad_norm": 2.935652764654381,
      "learning_rate": 1.0361873812229111e-06,
      "loss": 0.226,
      "step": 1330
    },
    {
      "epoch": 0.3444368910038523,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.8717948717948718,
      "eval_PRM F1 AUC": 0.8009429020429545,
      "eval_PRM F1 AUC (fixed)": 0.7771084337349398,
      "eval_PRM F1 Neg": 0.6428571428571429,
      "eval_PRM NPV": 0.5454545454545454,
      "eval_PRM Precision": 0.9315068493150684,
      "eval_PRM Recall": 0.8192771084337349,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.2631690204143524,
      "eval_runtime": 14.4887,
      "eval_samples_per_second": 3.037,
      "eval_steps_per_second": 0.138,
      "step": 1330
    },
    {
      "epoch": 0.3446958661098702,
      "grad_norm": 3.485190804963136,
      "learning_rate": 1.0357615599767362e-06,
      "loss": 0.3748,
      "step": 1331
    },
    {
      "epoch": 0.3449548412158881,
      "grad_norm": 4.250749573394851,
      "learning_rate": 1.0353354028151864e-06,
      "loss": 0.3145,
      "step": 1332
    },
    {
      "epoch": 0.34521381632190606,
      "grad_norm": 7.923349942246661,
      "learning_rate": 1.0349089100867673e-06,
      "loss": 0.2599,
      "step": 1333
    },
    {
      "epoch": 0.345472791427924,
      "grad_norm": 6.3912608889410905,
      "learning_rate": 1.034482082140259e-06,
      "loss": 0.3285,
      "step": 1334
    },
    {
      "epoch": 0.34573176653394194,
      "grad_norm": 4.472668850805793,
      "learning_rate": 1.034054919324716e-06,
      "loss": 0.3863,
      "step": 1335
    },
    {
      "epoch": 0.34573176653394194,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.8701298701298701,
      "eval_PRM F1 AUC": 0.8166579360921948,
      "eval_PRM F1 AUC (fixed)": 0.784442116291252,
      "eval_PRM F1 Neg": 0.6551724137931034,
      "eval_PRM NPV": 0.5428571428571428,
      "eval_PRM Precision": 0.9436619718309859,
      "eval_PRM Recall": 0.8072289156626506,
      "eval_PRM Specificty": 0.8260869565217391,
      "eval_loss": 0.25815823674201965,
      "eval_runtime": 14.7537,
      "eval_samples_per_second": 2.982,
      "eval_steps_per_second": 0.136,
      "step": 1335
    },
    {
      "epoch": 0.34599074163995985,
      "grad_norm": 4.239365211227992,
      "learning_rate": 1.033627421989466e-06,
      "loss": 0.3723,
      "step": 1336
    },
    {
      "epoch": 0.3462497167459778,
      "grad_norm": 3.6891728911365607,
      "learning_rate": 1.033199590484111e-06,
      "loss": 0.3842,
      "step": 1337
    },
    {
      "epoch": 0.34650869185199573,
      "grad_norm": 4.1821840352227255,
      "learning_rate": 1.0327714251585257e-06,
      "loss": 0.4589,
      "step": 1338
    },
    {
      "epoch": 0.34676766695801364,
      "grad_norm": 3.2232499807001544,
      "learning_rate": 1.0323429263628582e-06,
      "loss": 0.299,
      "step": 1339
    },
    {
      "epoch": 0.3470266420640316,
      "grad_norm": 5.441098902376691,
      "learning_rate": 1.0319140944475291e-06,
      "loss": 0.4802,
      "step": 1340
    },
    {
      "epoch": 0.3470266420640316,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8774193548387097,
      "eval_PRM F1 AUC": 0.8226820324777371,
      "eval_PRM F1 AUC (fixed)": 0.786013619696176,
      "eval_PRM F1 Neg": 0.6666666666666666,
      "eval_PRM NPV": 0.5588235294117647,
      "eval_PRM Precision": 0.9444444444444444,
      "eval_PRM Recall": 0.8192771084337349,
      "eval_PRM Specificty": 0.8260869565217391,
      "eval_loss": 0.2553994953632355,
      "eval_runtime": 14.633,
      "eval_samples_per_second": 3.007,
      "eval_steps_per_second": 0.137,
      "step": 1340
    },
    {
      "epoch": 0.3472856171700495,
      "grad_norm": 4.884035995601269,
      "learning_rate": 1.0314849297632318e-06,
      "loss": 0.3386,
      "step": 1341
    },
    {
      "epoch": 0.3475445922760675,
      "grad_norm": 3.2695838689315853,
      "learning_rate": 1.031055432660931e-06,
      "loss": 0.3565,
      "step": 1342
    },
    {
      "epoch": 0.3478035673820854,
      "grad_norm": 4.304615322479972,
      "learning_rate": 1.0306256034918637e-06,
      "loss": 0.3045,
      "step": 1343
    },
    {
      "epoch": 0.3480625424881033,
      "grad_norm": 3.3317439490288208,
      "learning_rate": 1.0301954426075391e-06,
      "loss": 0.343,
      "step": 1344
    },
    {
      "epoch": 0.3483215175941213,
      "grad_norm": 3.656639575005203,
      "learning_rate": 1.0297649503597366e-06,
      "loss": 0.3504,
      "step": 1345
    },
    {
      "epoch": 0.3483215175941213,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8774193548387097,
      "eval_PRM F1 AUC": 0.8226820324777371,
      "eval_PRM F1 AUC (fixed)": 0.784704033525406,
      "eval_PRM F1 Neg": 0.6666666666666666,
      "eval_PRM NPV": 0.5588235294117647,
      "eval_PRM Precision": 0.9444444444444444,
      "eval_PRM Recall": 0.8192771084337349,
      "eval_PRM Specificty": 0.8260869565217391,
      "eval_loss": 0.25615188479423523,
      "eval_runtime": 14.6776,
      "eval_samples_per_second": 2.998,
      "eval_steps_per_second": 0.136,
      "step": 1345
    },
    {
      "epoch": 0.3485804927001392,
      "grad_norm": 3.766623424533264,
      "learning_rate": 1.0293341271005073e-06,
      "loss": 0.284,
      "step": 1346
    },
    {
      "epoch": 0.34883946780615716,
      "grad_norm": 4.488867287571511,
      "learning_rate": 1.0289029731821725e-06,
      "loss": 0.3686,
      "step": 1347
    },
    {
      "epoch": 0.34909844291217507,
      "grad_norm": 3.1747978172581894,
      "learning_rate": 1.0284714889573248e-06,
      "loss": 0.3598,
      "step": 1348
    },
    {
      "epoch": 0.349357418018193,
      "grad_norm": 3.3137262712445894,
      "learning_rate": 1.0280396747788255e-06,
      "loss": 0.3896,
      "step": 1349
    },
    {
      "epoch": 0.34961639312421094,
      "grad_norm": 4.6326692615664316,
      "learning_rate": 1.027607530999807e-06,
      "loss": 0.3655,
      "step": 1350
    },
    {
      "epoch": 0.34961639312421094,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8789808917197452,
      "eval_PRM F1 AUC": 0.8069669984284967,
      "eval_PRM F1 AUC (fixed)": 0.7915138816134102,
      "eval_PRM F1 Neg": 0.6545454545454545,
      "eval_PRM NPV": 0.5625,
      "eval_PRM Precision": 0.9324324324324325,
      "eval_PRM Recall": 0.8313253012048193,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.2515624165534973,
      "eval_runtime": 14.6399,
      "eval_samples_per_second": 3.005,
      "eval_steps_per_second": 0.137,
      "step": 1350
    },
    {
      "epoch": 0.34987536823022886,
      "grad_norm": 5.029034145916732,
      "learning_rate": 1.0271750579736707e-06,
      "loss": 0.2834,
      "step": 1351
    },
    {
      "epoch": 0.3501343433362468,
      "grad_norm": 3.2262110030593965,
      "learning_rate": 1.026742256054087e-06,
      "loss": 0.3099,
      "step": 1352
    },
    {
      "epoch": 0.35039331844226473,
      "grad_norm": 4.277241669093875,
      "learning_rate": 1.0263091255949962e-06,
      "loss": 0.3033,
      "step": 1353
    },
    {
      "epoch": 0.35065229354828265,
      "grad_norm": 4.219679308248584,
      "learning_rate": 1.0258756669506065e-06,
      "loss": 0.3592,
      "step": 1354
    },
    {
      "epoch": 0.3509112686543006,
      "grad_norm": 3.873414393356266,
      "learning_rate": 1.0254418804753941e-06,
      "loss": 0.3591,
      "step": 1355
    },
    {
      "epoch": 0.3509112686543006,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.8717948717948718,
      "eval_PRM F1 AUC": 0.8009429020429545,
      "eval_PRM F1 AUC (fixed)": 0.7941330539549503,
      "eval_PRM F1 Neg": 0.6428571428571429,
      "eval_PRM NPV": 0.5454545454545454,
      "eval_PRM Precision": 0.9315068493150684,
      "eval_PRM Recall": 0.8192771084337349,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.25376003980636597,
      "eval_runtime": 14.6093,
      "eval_samples_per_second": 3.012,
      "eval_steps_per_second": 0.137,
      "step": 1355
    },
    {
      "epoch": 0.3511702437603185,
      "grad_norm": 3.596581202851472,
      "learning_rate": 1.0250077665241045e-06,
      "loss": 0.3553,
      "step": 1356
    },
    {
      "epoch": 0.3514292188663365,
      "grad_norm": 3.0834953882859586,
      "learning_rate": 1.02457332545175e-06,
      "loss": 0.2994,
      "step": 1357
    },
    {
      "epoch": 0.3516881939723544,
      "grad_norm": 3.5752729654852904,
      "learning_rate": 1.0241385576136109e-06,
      "loss": 0.3762,
      "step": 1358
    },
    {
      "epoch": 0.35194716907837237,
      "grad_norm": 6.601488366950558,
      "learning_rate": 1.0237034633652345e-06,
      "loss": 0.3852,
      "step": 1359
    },
    {
      "epoch": 0.3522061441843903,
      "grad_norm": 3.8142506524141284,
      "learning_rate": 1.0232680430624352e-06,
      "loss": 0.3078,
      "step": 1360
    },
    {
      "epoch": 0.3522061441843903,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8789808917197452,
      "eval_PRM F1 AUC": 0.8069669984284967,
      "eval_PRM F1 AUC (fixed)": 0.7977998952331063,
      "eval_PRM F1 Neg": 0.6545454545454545,
      "eval_PRM NPV": 0.5625,
      "eval_PRM Precision": 0.9324324324324325,
      "eval_PRM Recall": 0.8313253012048193,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.24866238236427307,
      "eval_runtime": 14.679,
      "eval_samples_per_second": 2.997,
      "eval_steps_per_second": 0.136,
      "step": 1360
    },
    {
      "epoch": 0.3524651192904082,
      "grad_norm": 7.2589367790025685,
      "learning_rate": 1.022832297061294e-06,
      "loss": 0.3643,
      "step": 1361
    },
    {
      "epoch": 0.35272409439642616,
      "grad_norm": 6.2858784402296495,
      "learning_rate": 1.022396225718158e-06,
      "loss": 0.4068,
      "step": 1362
    },
    {
      "epoch": 0.35298306950244407,
      "grad_norm": 7.788782355650248,
      "learning_rate": 1.021959829389641e-06,
      "loss": 0.3245,
      "step": 1363
    },
    {
      "epoch": 0.35324204460846204,
      "grad_norm": 3.599017538260975,
      "learning_rate": 1.021523108432622e-06,
      "loss": 0.3178,
      "step": 1364
    },
    {
      "epoch": 0.35350101971447995,
      "grad_norm": 6.426781605853537,
      "learning_rate": 1.0210860632042453e-06,
      "loss": 0.4676,
      "step": 1365
    },
    {
      "epoch": 0.35350101971447995,
      "eval_PRM Accuracy": 0.8301886792452831,
      "eval_PRM F1": 0.8860759493670886,
      "eval_PRM F1 AUC": 0.8129910948140389,
      "eval_PRM F1 AUC (fixed)": 0.8038239916186485,
      "eval_PRM F1 Neg": 0.6666666666666666,
      "eval_PRM NPV": 0.5806451612903226,
      "eval_PRM Precision": 0.9333333333333333,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.24092192947864532,
      "eval_runtime": 14.7974,
      "eval_samples_per_second": 2.974,
      "eval_steps_per_second": 0.135,
      "step": 1365
    },
    {
      "epoch": 0.35375999482049786,
      "grad_norm": 4.851902543247538,
      "learning_rate": 1.0206486940619207e-06,
      "loss": 0.3503,
      "step": 1366
    },
    {
      "epoch": 0.3540189699265158,
      "grad_norm": 3.260733464307621,
      "learning_rate": 1.0202110013633235e-06,
      "loss": 0.2659,
      "step": 1367
    },
    {
      "epoch": 0.35427794503253374,
      "grad_norm": 4.77562481552069,
      "learning_rate": 1.0197729854663925e-06,
      "loss": 0.3689,
      "step": 1368
    },
    {
      "epoch": 0.3545369201385517,
      "grad_norm": 6.9020702863177705,
      "learning_rate": 1.0193346467293314e-06,
      "loss": 0.3497,
      "step": 1369
    },
    {
      "epoch": 0.3547958952445696,
      "grad_norm": 2.8641402943880014,
      "learning_rate": 1.0188959855106082e-06,
      "loss": 0.2729,
      "step": 1370
    },
    {
      "epoch": 0.3547958952445696,
      "eval_PRM Accuracy": 0.8301886792452831,
      "eval_PRM F1": 0.8846153846153846,
      "eval_PRM F1 AUC": 0.8287061288632793,
      "eval_PRM F1 AUC (fixed)": 0.7909900471451021,
      "eval_PRM F1 Neg": 0.6785714285714286,
      "eval_PRM NPV": 0.5757575757575758,
      "eval_PRM Precision": 0.9452054794520548,
      "eval_PRM Recall": 0.8313253012048193,
      "eval_PRM Specificty": 0.8260869565217391,
      "eval_loss": 0.2476367950439453,
      "eval_runtime": 14.9235,
      "eval_samples_per_second": 2.948,
      "eval_steps_per_second": 0.134,
      "step": 1370
    },
    {
      "epoch": 0.3550548703505875,
      "grad_norm": 4.912052749316654,
      "learning_rate": 1.0184570021689536e-06,
      "loss": 0.3502,
      "step": 1371
    },
    {
      "epoch": 0.3553138454566055,
      "grad_norm": 5.308943310093734,
      "learning_rate": 1.018017697063363e-06,
      "loss": 0.3365,
      "step": 1372
    },
    {
      "epoch": 0.3555728205626234,
      "grad_norm": 3.630549995233943,
      "learning_rate": 1.0175780705530935e-06,
      "loss": 0.4088,
      "step": 1373
    },
    {
      "epoch": 0.35583179566864137,
      "grad_norm": 7.2185639503008545,
      "learning_rate": 1.0171381229976667e-06,
      "loss": 0.3197,
      "step": 1374
    },
    {
      "epoch": 0.3560907707746593,
      "grad_norm": 8.163397684150947,
      "learning_rate": 1.0166978547568656e-06,
      "loss": 0.2647,
      "step": 1375
    },
    {
      "epoch": 0.3560907707746593,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8789808917197452,
      "eval_PRM F1 AUC": 0.8069669984284967,
      "eval_PRM F1 AUC (fixed)": 0.788108957569408,
      "eval_PRM F1 Neg": 0.6545454545454545,
      "eval_PRM NPV": 0.5625,
      "eval_PRM Precision": 0.9324324324324325,
      "eval_PRM Recall": 0.8313253012048193,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.24819716811180115,
      "eval_runtime": 14.7698,
      "eval_samples_per_second": 2.979,
      "eval_steps_per_second": 0.135,
      "step": 1375
    },
    {
      "epoch": 0.3563497458806772,
      "grad_norm": 6.313737507680141,
      "learning_rate": 1.0162572661907358e-06,
      "loss": 0.3255,
      "step": 1376
    },
    {
      "epoch": 0.35660872098669516,
      "grad_norm": 4.288938839171069,
      "learning_rate": 1.0158163576595846e-06,
      "loss": 0.3833,
      "step": 1377
    },
    {
      "epoch": 0.35686769609271307,
      "grad_norm": 4.158728708310663,
      "learning_rate": 1.0153751295239813e-06,
      "loss": 0.464,
      "step": 1378
    },
    {
      "epoch": 0.35712667119873104,
      "grad_norm": 6.984820199164077,
      "learning_rate": 1.0149335821447565e-06,
      "loss": 0.3083,
      "step": 1379
    },
    {
      "epoch": 0.35738564630474895,
      "grad_norm": 7.100930397132101,
      "learning_rate": 1.0144917158830015e-06,
      "loss": 0.3468,
      "step": 1380
    },
    {
      "epoch": 0.35738564630474895,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8789808917197452,
      "eval_PRM F1 AUC": 0.8069669984284967,
      "eval_PRM F1 AUC (fixed)": 0.7870612886327921,
      "eval_PRM F1 Neg": 0.6545454545454545,
      "eval_PRM NPV": 0.5625,
      "eval_PRM Precision": 0.9324324324324325,
      "eval_PRM Recall": 0.8313253012048193,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.2420676201581955,
      "eval_runtime": 14.5031,
      "eval_samples_per_second": 3.034,
      "eval_steps_per_second": 0.138,
      "step": 1380
    },
    {
      "epoch": 0.3576446214107669,
      "grad_norm": 3.402060831951842,
      "learning_rate": 1.014049531100069e-06,
      "loss": 0.3431,
      "step": 1381
    },
    {
      "epoch": 0.3579035965167848,
      "grad_norm": 4.1631232404476215,
      "learning_rate": 1.0136070281575718e-06,
      "loss": 0.4072,
      "step": 1382
    },
    {
      "epoch": 0.35816257162280274,
      "grad_norm": 2.96097909341508,
      "learning_rate": 1.013164207417383e-06,
      "loss": 0.2762,
      "step": 1383
    },
    {
      "epoch": 0.3584215467288207,
      "grad_norm": 3.589280367075051,
      "learning_rate": 1.0127210692416356e-06,
      "loss": 0.3656,
      "step": 1384
    },
    {
      "epoch": 0.3586805218348386,
      "grad_norm": 10.198150056204957,
      "learning_rate": 1.012277613992722e-06,
      "loss": 0.3689,
      "step": 1385
    },
    {
      "epoch": 0.3586805218348386,
      "eval_PRM Accuracy": 0.8867924528301887,
      "eval_PRM F1": 0.927710843373494,
      "eval_PRM F1 AUC": 0.8334206390780513,
      "eval_PRM F1 AUC (fixed)": 0.7888947092718701,
      "eval_PRM F1 Neg": 0.7391304347826086,
      "eval_PRM NPV": 0.7391304347826086,
      "eval_PRM Precision": 0.927710843373494,
      "eval_PRM Recall": 0.927710843373494,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.2418714016675949,
      "eval_runtime": 14.7389,
      "eval_samples_per_second": 2.985,
      "eval_steps_per_second": 0.136,
      "step": 1385
    },
    {
      "epoch": 0.3589394969408566,
      "grad_norm": 9.174095092421487,
      "learning_rate": 1.011833842033294e-06,
      "loss": 0.385,
      "step": 1386
    },
    {
      "epoch": 0.3591984720468745,
      "grad_norm": 5.068293049956948,
      "learning_rate": 1.0113897537262627e-06,
      "loss": 0.314,
      "step": 1387
    },
    {
      "epoch": 0.3594574471528924,
      "grad_norm": 6.1186335582351745,
      "learning_rate": 1.0109453494347973e-06,
      "loss": 0.3665,
      "step": 1388
    },
    {
      "epoch": 0.3597164222589104,
      "grad_norm": 3.0993215013454383,
      "learning_rate": 1.010500629522326e-06,
      "loss": 0.3367,
      "step": 1389
    },
    {
      "epoch": 0.3599753973649283,
      "grad_norm": 7.933153861235534,
      "learning_rate": 1.0100555943525348e-06,
      "loss": 0.3925,
      "step": 1390
    },
    {
      "epoch": 0.3599753973649283,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8789808917197452,
      "eval_PRM F1 AUC": 0.8069669984284967,
      "eval_PRM F1 AUC (fixed)": 0.7891566265060241,
      "eval_PRM F1 Neg": 0.6545454545454545,
      "eval_PRM NPV": 0.5625,
      "eval_PRM Precision": 0.9324324324324325,
      "eval_PRM Recall": 0.8313253012048193,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.24478855729103088,
      "eval_runtime": 14.5608,
      "eval_samples_per_second": 3.022,
      "eval_steps_per_second": 0.137,
      "step": 1390
    },
    {
      "epoch": 0.36023437247094625,
      "grad_norm": 3.69707566141284,
      "learning_rate": 1.0096102442893676e-06,
      "loss": 0.4158,
      "step": 1391
    },
    {
      "epoch": 0.36049334757696416,
      "grad_norm": 5.0661368363238175,
      "learning_rate": 1.0091645796970257e-06,
      "loss": 0.3552,
      "step": 1392
    },
    {
      "epoch": 0.3607523226829821,
      "grad_norm": 6.248551686579097,
      "learning_rate": 1.008718600939968e-06,
      "loss": 0.342,
      "step": 1393
    },
    {
      "epoch": 0.36101129778900004,
      "grad_norm": 6.792166057295491,
      "learning_rate": 1.0082723083829096e-06,
      "loss": 0.323,
      "step": 1394
    },
    {
      "epoch": 0.36127027289501795,
      "grad_norm": 3.779026986393951,
      "learning_rate": 1.007825702390823e-06,
      "loss": 0.3358,
      "step": 1395
    },
    {
      "epoch": 0.36127027289501795,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8789808917197452,
      "eval_PRM F1 AUC": 0.8069669984284967,
      "eval_PRM F1 AUC (fixed)": 0.784180199057098,
      "eval_PRM F1 Neg": 0.6545454545454545,
      "eval_PRM NPV": 0.5625,
      "eval_PRM Precision": 0.9324324324324325,
      "eval_PRM Recall": 0.8313253012048193,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.25908514857292175,
      "eval_runtime": 14.8271,
      "eval_samples_per_second": 2.968,
      "eval_steps_per_second": 0.135,
      "step": 1395
    },
    {
      "epoch": 0.3615292480010359,
      "grad_norm": 7.177709456902315,
      "learning_rate": 1.0073787833289364e-06,
      "loss": 0.339,
      "step": 1396
    },
    {
      "epoch": 0.36178822310705383,
      "grad_norm": 4.534515548735088,
      "learning_rate": 1.0069315515627346e-06,
      "loss": 0.3624,
      "step": 1397
    },
    {
      "epoch": 0.36204719821307174,
      "grad_norm": 3.516634251586425,
      "learning_rate": 1.0064840074579578e-06,
      "loss": 0.3755,
      "step": 1398
    },
    {
      "epoch": 0.3623061733190897,
      "grad_norm": 2.9834443537357753,
      "learning_rate": 1.0060361513806013e-06,
      "loss": 0.2756,
      "step": 1399
    },
    {
      "epoch": 0.3625651484251076,
      "grad_norm": 6.302837815954074,
      "learning_rate": 1.0055879836969164e-06,
      "loss": 0.3814,
      "step": 1400
    },
    {
      "epoch": 0.3625651484251076,
      "eval_PRM Accuracy": 0.8301886792452831,
      "eval_PRM F1": 0.8860759493670886,
      "eval_PRM F1 AUC": 0.8129910948140389,
      "eval_PRM F1 AUC (fixed)": 0.7894185437401781,
      "eval_PRM F1 Neg": 0.6666666666666666,
      "eval_PRM NPV": 0.5806451612903226,
      "eval_PRM Precision": 0.9333333333333333,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.2521251142024994,
      "eval_runtime": 14.7735,
      "eval_samples_per_second": 2.978,
      "eval_steps_per_second": 0.135,
      "step": 1400
    },
    {
      "epoch": 0.3628241235311256,
      "grad_norm": 5.898802333208595,
      "learning_rate": 1.0051395047734083e-06,
      "loss": 0.3526,
      "step": 1401
    },
    {
      "epoch": 0.3630830986371435,
      "grad_norm": 3.827032648322295,
      "learning_rate": 1.004690714976837e-06,
      "loss": 0.3202,
      "step": 1402
    },
    {
      "epoch": 0.36334207374316146,
      "grad_norm": 3.767391984694187,
      "learning_rate": 1.0042416146742175e-06,
      "loss": 0.3372,
      "step": 1403
    },
    {
      "epoch": 0.3636010488491794,
      "grad_norm": 4.634860029918238,
      "learning_rate": 1.0037922042328177e-06,
      "loss": 0.4127,
      "step": 1404
    },
    {
      "epoch": 0.3638600239551973,
      "grad_norm": 3.1248109063152127,
      "learning_rate": 1.0033424840201595e-06,
      "loss": 0.3281,
      "step": 1405
    },
    {
      "epoch": 0.3638600239551973,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8789808917197452,
      "eval_PRM F1 AUC": 0.8069669984284967,
      "eval_PRM F1 AUC (fixed)": 0.788108957569408,
      "eval_PRM F1 Neg": 0.6545454545454545,
      "eval_PRM NPV": 0.5625,
      "eval_PRM Precision": 0.9324324324324325,
      "eval_PRM Recall": 0.8313253012048193,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.24834133684635162,
      "eval_runtime": 14.6414,
      "eval_samples_per_second": 3.005,
      "eval_steps_per_second": 0.137,
      "step": 1405
    },
    {
      "epoch": 0.36411899906121525,
      "grad_norm": 5.545453446821229,
      "learning_rate": 1.0028924544040179e-06,
      "loss": 0.384,
      "step": 1406
    },
    {
      "epoch": 0.36437797416723317,
      "grad_norm": 4.211034137415862,
      "learning_rate": 1.0024421157524216e-06,
      "loss": 0.3684,
      "step": 1407
    },
    {
      "epoch": 0.36463694927325113,
      "grad_norm": 4.057108424391074,
      "learning_rate": 1.0019914684336517e-06,
      "loss": 0.4328,
      "step": 1408
    },
    {
      "epoch": 0.36489592437926904,
      "grad_norm": 3.3320674863925537,
      "learning_rate": 1.0015405128162412e-06,
      "loss": 0.3834,
      "step": 1409
    },
    {
      "epoch": 0.36515489948528695,
      "grad_norm": 8.681780340750507,
      "learning_rate": 1.0010892492689758e-06,
      "loss": 0.351,
      "step": 1410
    },
    {
      "epoch": 0.36515489948528695,
      "eval_PRM Accuracy": 0.8301886792452831,
      "eval_PRM F1": 0.8860759493670886,
      "eval_PRM F1 AUC": 0.8129910948140389,
      "eval_PRM F1 AUC (fixed)": 0.788108957569408,
      "eval_PRM F1 Neg": 0.6666666666666666,
      "eval_PRM NPV": 0.5806451612903226,
      "eval_PRM Precision": 0.9333333333333333,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.24854445457458496,
      "eval_runtime": 14.7704,
      "eval_samples_per_second": 2.979,
      "eval_steps_per_second": 0.135,
      "step": 1410
    },
    {
      "epoch": 0.3654138745913049,
      "grad_norm": 3.47683496398654,
      "learning_rate": 1.000637678160893e-06,
      "loss": 0.3208,
      "step": 1411
    },
    {
      "epoch": 0.36567284969732283,
      "grad_norm": 3.9465011870396873,
      "learning_rate": 1.0001857998612815e-06,
      "loss": 0.4328,
      "step": 1412
    },
    {
      "epoch": 0.3659318248033408,
      "grad_norm": 2.7448030317458754,
      "learning_rate": 9.997336147396816e-07,
      "loss": 0.3005,
      "step": 1413
    },
    {
      "epoch": 0.3661907999093587,
      "grad_norm": 4.164297634945185,
      "learning_rate": 9.99281123165884e-07,
      "loss": 0.2506,
      "step": 1414
    },
    {
      "epoch": 0.3664497750153766,
      "grad_norm": 3.4588831820316877,
      "learning_rate": 9.988283255099306e-07,
      "loss": 0.3232,
      "step": 1415
    },
    {
      "epoch": 0.3664497750153766,
      "eval_PRM Accuracy": 0.8773584905660378,
      "eval_PRM F1": 0.9202453987730062,
      "eval_PRM F1 AUC": 0.8431115767417497,
      "eval_PRM F1 AUC (fixed)": 0.7909900471451021,
      "eval_PRM F1 Neg": 0.7346938775510204,
      "eval_PRM NPV": 0.6923076923076923,
      "eval_PRM Precision": 0.9375,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.2446984201669693,
      "eval_runtime": 14.2678,
      "eval_samples_per_second": 3.084,
      "eval_steps_per_second": 0.14,
      "step": 1415
    },
    {
      "epoch": 0.3667087501213946,
      "grad_norm": 5.711828100772439,
      "learning_rate": 9.983752221421134e-07,
      "loss": 0.3872,
      "step": 1416
    },
    {
      "epoch": 0.3669677252274125,
      "grad_norm": 3.83260643758672,
      "learning_rate": 9.979218134329739e-07,
      "loss": 0.2705,
      "step": 1417
    },
    {
      "epoch": 0.36722670033343047,
      "grad_norm": 7.755924384672744,
      "learning_rate": 9.974680997533043e-07,
      "loss": 0.4048,
      "step": 1418
    },
    {
      "epoch": 0.3674856754394484,
      "grad_norm": 3.4136788910922937,
      "learning_rate": 9.970140814741452e-07,
      "loss": 0.2815,
      "step": 1419
    },
    {
      "epoch": 0.3677446505454663,
      "grad_norm": 3.3743340350957682,
      "learning_rate": 9.965597589667871e-07,
      "loss": 0.3282,
      "step": 1420
    },
    {
      "epoch": 0.3677446505454663,
      "eval_PRM Accuracy": 0.8867924528301887,
      "eval_PRM F1": 0.926829268292683,
      "eval_PRM F1 AUC": 0.8491356731272918,
      "eval_PRM F1 AUC (fixed)": 0.8014667365112624,
      "eval_PRM F1 Neg": 0.75,
      "eval_PRM NPV": 0.72,
      "eval_PRM Precision": 0.9382716049382716,
      "eval_PRM Recall": 0.9156626506024096,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.23952960968017578,
      "eval_runtime": 14.9505,
      "eval_samples_per_second": 2.943,
      "eval_steps_per_second": 0.134,
      "step": 1420
    },
    {
      "epoch": 0.36800362565148426,
      "grad_norm": 3.4721221499465535,
      "learning_rate": 9.961051326027688e-07,
      "loss": 0.2982,
      "step": 1421
    },
    {
      "epoch": 0.36826260075750217,
      "grad_norm": 6.65201998896593,
      "learning_rate": 9.956502027538776e-07,
      "loss": 0.3241,
      "step": 1422
    },
    {
      "epoch": 0.36852157586352013,
      "grad_norm": 3.462624885557569,
      "learning_rate": 9.951949697921493e-07,
      "loss": 0.2788,
      "step": 1423
    },
    {
      "epoch": 0.36878055096953805,
      "grad_norm": 3.3189609048980264,
      "learning_rate": 9.947394340898675e-07,
      "loss": 0.3762,
      "step": 1424
    },
    {
      "epoch": 0.369039526075556,
      "grad_norm": 2.9790395939751733,
      "learning_rate": 9.94283596019563e-07,
      "loss": 0.2558,
      "step": 1425
    },
    {
      "epoch": 0.369039526075556,
      "eval_PRM Accuracy": 0.8490566037735849,
      "eval_PRM F1": 0.9,
      "eval_PRM F1 AUC": 0.8250392875851232,
      "eval_PRM F1 AUC (fixed)": 0.8043478260869564,
      "eval_PRM F1 Neg": 0.6923076923076923,
      "eval_PRM NPV": 0.6206896551724138,
      "eval_PRM Precision": 0.935064935064935,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.2378481775522232,
      "eval_runtime": 14.6257,
      "eval_samples_per_second": 3.008,
      "eval_steps_per_second": 0.137,
      "step": 1425
    },
    {
      "epoch": 0.3692985011815739,
      "grad_norm": 3.4004104085051905,
      "learning_rate": 9.938274559540144e-07,
      "loss": 0.3402,
      "step": 1426
    },
    {
      "epoch": 0.36955747628759184,
      "grad_norm": 4.1750959101791745,
      "learning_rate": 9.933710142662468e-07,
      "loss": 0.2943,
      "step": 1427
    },
    {
      "epoch": 0.3698164513936098,
      "grad_norm": 3.2247794918939894,
      "learning_rate": 9.92914271329532e-07,
      "loss": 0.35,
      "step": 1428
    },
    {
      "epoch": 0.3700754264996277,
      "grad_norm": 5.112086961413494,
      "learning_rate": 9.924572275173889e-07,
      "loss": 0.2642,
      "step": 1429
    },
    {
      "epoch": 0.3703344016056457,
      "grad_norm": 6.439724484325886,
      "learning_rate": 9.919998832035815e-07,
      "loss": 0.3536,
      "step": 1430
    },
    {
      "epoch": 0.3703344016056457,
      "eval_PRM Accuracy": 0.8584905660377359,
      "eval_PRM F1": 0.9056603773584906,
      "eval_PRM F1 AUC": 0.8467784180199058,
      "eval_PRM F1 AUC (fixed)": 0.8035620743844945,
      "eval_PRM F1 Neg": 0.7169811320754716,
      "eval_PRM NPV": 0.6333333333333333,
      "eval_PRM Precision": 0.9473684210526315,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.8260869565217391,
      "eval_loss": 0.23704619705677032,
      "eval_runtime": 14.4407,
      "eval_samples_per_second": 3.047,
      "eval_steps_per_second": 0.138,
      "step": 1430
    },
    {
      "epoch": 0.3705933767116636,
      "grad_norm": 3.443131454125624,
      "learning_rate": 9.9154223876212e-07,
      "loss": 0.3985,
      "step": 1431
    },
    {
      "epoch": 0.3708523518176815,
      "grad_norm": 5.567638713314668,
      "learning_rate": 9.910842945672602e-07,
      "loss": 0.3216,
      "step": 1432
    },
    {
      "epoch": 0.37111132692369947,
      "grad_norm": 2.9108306082811293,
      "learning_rate": 9.906260509935027e-07,
      "loss": 0.3545,
      "step": 1433
    },
    {
      "epoch": 0.3713703020297174,
      "grad_norm": 3.992555721910898,
      "learning_rate": 9.901675084155924e-07,
      "loss": 0.3514,
      "step": 1434
    },
    {
      "epoch": 0.37162927713573535,
      "grad_norm": 3.288250771978138,
      "learning_rate": 9.897086672085205e-07,
      "loss": 0.3328,
      "step": 1435
    },
    {
      "epoch": 0.37162927713573535,
      "eval_PRM Accuracy": 0.8490566037735849,
      "eval_PRM F1": 0.8987341772151899,
      "eval_PRM F1 AUC": 0.8407543216343635,
      "eval_PRM F1 AUC (fixed)": 0.8004190675746463,
      "eval_PRM F1 Neg": 0.7037037037037037,
      "eval_PRM NPV": 0.6129032258064516,
      "eval_PRM Precision": 0.9466666666666667,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.8260869565217391,
      "eval_loss": 0.23846612870693207,
      "eval_runtime": 14.7197,
      "eval_samples_per_second": 2.989,
      "eval_steps_per_second": 0.136,
      "step": 1435
    },
    {
      "epoch": 0.37188825224175326,
      "grad_norm": 3.3613173523286695,
      "learning_rate": 9.892495277475203e-07,
      "loss": 0.2902,
      "step": 1436
    },
    {
      "epoch": 0.37214722734777117,
      "grad_norm": 4.1674637022810845,
      "learning_rate": 9.887900904080704e-07,
      "loss": 0.2901,
      "step": 1437
    },
    {
      "epoch": 0.37240620245378914,
      "grad_norm": 4.785496501344485,
      "learning_rate": 9.883303555658928e-07,
      "loss": 0.3439,
      "step": 1438
    },
    {
      "epoch": 0.37266517755980705,
      "grad_norm": 4.6478897795229335,
      "learning_rate": 9.878703235969523e-07,
      "loss": 0.3821,
      "step": 1439
    },
    {
      "epoch": 0.372924152665825,
      "grad_norm": 4.519922141366181,
      "learning_rate": 9.874099948774568e-07,
      "loss": 0.4066,
      "step": 1440
    },
    {
      "epoch": 0.372924152665825,
      "eval_PRM Accuracy": 0.8584905660377359,
      "eval_PRM F1": 0.906832298136646,
      "eval_PRM F1 AUC": 0.8310633839706653,
      "eval_PRM F1 AUC (fixed)": 0.8064431639601886,
      "eval_PRM F1 Neg": 0.7058823529411765,
      "eval_PRM NPV": 0.6428571428571429,
      "eval_PRM Precision": 0.9358974358974359,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.23579718172550201,
      "eval_runtime": 14.5869,
      "eval_samples_per_second": 3.016,
      "eval_steps_per_second": 0.137,
      "step": 1440
    },
    {
      "epoch": 0.3731831277718429,
      "grad_norm": 8.082390323683258,
      "learning_rate": 9.869493697838573e-07,
      "loss": 0.3807,
      "step": 1441
    },
    {
      "epoch": 0.3734421028778609,
      "grad_norm": 3.9871931447775344,
      "learning_rate": 9.86488448692847e-07,
      "loss": 0.4278,
      "step": 1442
    },
    {
      "epoch": 0.3737010779838788,
      "grad_norm": 5.366527806558145,
      "learning_rate": 9.860272319813607e-07,
      "loss": 0.3911,
      "step": 1443
    },
    {
      "epoch": 0.3739600530898967,
      "grad_norm": 4.962218269420269,
      "learning_rate": 9.855657200265755e-07,
      "loss": 0.3959,
      "step": 1444
    },
    {
      "epoch": 0.3742190281959147,
      "grad_norm": 3.682545253283956,
      "learning_rate": 9.851039132059103e-07,
      "loss": 0.4014,
      "step": 1445
    },
    {
      "epoch": 0.3742190281959147,
      "eval_PRM Accuracy": 0.8490566037735849,
      "eval_PRM F1": 0.9,
      "eval_PRM F1 AUC": 0.8250392875851232,
      "eval_PRM F1 AUC (fixed)": 0.7991094814038764,
      "eval_PRM F1 Neg": 0.6923076923076923,
      "eval_PRM NPV": 0.6206896551724138,
      "eval_PRM Precision": 0.935064935064935,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.24163256585597992,
      "eval_runtime": 14.9003,
      "eval_samples_per_second": 2.953,
      "eval_steps_per_second": 0.134,
      "step": 1445
    },
    {
      "epoch": 0.3744780033019326,
      "grad_norm": 4.012447848004918,
      "learning_rate": 9.846418118970238e-07,
      "loss": 0.3358,
      "step": 1446
    },
    {
      "epoch": 0.37473697840795056,
      "grad_norm": 3.6285946249500456,
      "learning_rate": 9.841794164778167e-07,
      "loss": 0.3036,
      "step": 1447
    },
    {
      "epoch": 0.37499595351396847,
      "grad_norm": 7.966353046705746,
      "learning_rate": 9.8371672732643e-07,
      "loss": 0.2241,
      "step": 1448
    },
    {
      "epoch": 0.3752549286199864,
      "grad_norm": 3.212966323254794,
      "learning_rate": 9.832537448212445e-07,
      "loss": 0.3485,
      "step": 1449
    },
    {
      "epoch": 0.37551390372600435,
      "grad_norm": 5.815003987423003,
      "learning_rate": 9.827904693408814e-07,
      "loss": 0.34,
      "step": 1450
    },
    {
      "epoch": 0.37551390372600435,
      "eval_PRM Accuracy": 0.8773584905660378,
      "eval_PRM F1": 0.9202453987730062,
      "eval_PRM F1 AUC": 0.8431115767417497,
      "eval_PRM F1 AUC (fixed)": 0.8004190675746464,
      "eval_PRM F1 Neg": 0.7346938775510204,
      "eval_PRM NPV": 0.6923076923076923,
      "eval_PRM Precision": 0.9375,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.2403801530599594,
      "eval_runtime": 14.4933,
      "eval_samples_per_second": 3.036,
      "eval_steps_per_second": 0.138,
      "step": 1450
    },
    {
      "epoch": 0.37577287883202226,
      "grad_norm": 4.760451223227752,
      "learning_rate": 9.823269012642012e-07,
      "loss": 0.3352,
      "step": 1451
    },
    {
      "epoch": 0.37603185393804023,
      "grad_norm": 4.0186328881753335,
      "learning_rate": 9.818630409703036e-07,
      "loss": 0.2568,
      "step": 1452
    },
    {
      "epoch": 0.37629082904405814,
      "grad_norm": 4.164063275627754,
      "learning_rate": 9.813988888385275e-07,
      "loss": 0.3776,
      "step": 1453
    },
    {
      "epoch": 0.37654980415007605,
      "grad_norm": 8.55257640064162,
      "learning_rate": 9.809344452484508e-07,
      "loss": 0.3895,
      "step": 1454
    },
    {
      "epoch": 0.376808779256094,
      "grad_norm": 4.579909515061625,
      "learning_rate": 9.804697105798885e-07,
      "loss": 0.2611,
      "step": 1455
    },
    {
      "epoch": 0.376808779256094,
      "eval_PRM Accuracy": 0.8679245283018868,
      "eval_PRM F1": 0.9135802469135802,
      "eval_PRM F1 AUC": 0.8370874803562075,
      "eval_PRM F1 AUC (fixed)": 0.8067050811943426,
      "eval_PRM F1 Neg": 0.72,
      "eval_PRM NPV": 0.6666666666666666,
      "eval_PRM Precision": 0.9367088607594937,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.23761196434497833,
      "eval_runtime": 14.4681,
      "eval_samples_per_second": 3.041,
      "eval_steps_per_second": 0.138,
      "step": 1455
    },
    {
      "epoch": 0.37706775436211193,
      "grad_norm": 3.5982328749856105,
      "learning_rate": 9.80004685212895e-07,
      "loss": 0.3832,
      "step": 1456
    },
    {
      "epoch": 0.3773267294681299,
      "grad_norm": 5.440234201873794,
      "learning_rate": 9.79539369527762e-07,
      "loss": 0.3383,
      "step": 1457
    },
    {
      "epoch": 0.3775857045741478,
      "grad_norm": 5.14995747106743,
      "learning_rate": 9.790737639050182e-07,
      "loss": 0.4605,
      "step": 1458
    },
    {
      "epoch": 0.3778446796801657,
      "grad_norm": 3.3322124466293124,
      "learning_rate": 9.786078687254297e-07,
      "loss": 0.3834,
      "step": 1459
    },
    {
      "epoch": 0.3781036547861837,
      "grad_norm": 6.726995927479046,
      "learning_rate": 9.781416843699999e-07,
      "loss": 0.305,
      "step": 1460
    },
    {
      "epoch": 0.3781036547861837,
      "eval_PRM Accuracy": 0.8679245283018868,
      "eval_PRM F1": 0.9135802469135802,
      "eval_PRM F1 AUC": 0.8370874803562075,
      "eval_PRM F1 AUC (fixed)": 0.8114195914091147,
      "eval_PRM F1 Neg": 0.72,
      "eval_PRM NPV": 0.6666666666666666,
      "eval_PRM Precision": 0.9367088607594937,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.23749984800815582,
      "eval_runtime": 14.8026,
      "eval_samples_per_second": 2.972,
      "eval_steps_per_second": 0.135,
      "step": 1460
    },
    {
      "epoch": 0.3783626298922016,
      "grad_norm": 3.9906455982204188,
      "learning_rate": 9.776752112199674e-07,
      "loss": 0.3707,
      "step": 1461
    },
    {
      "epoch": 0.37862160499821956,
      "grad_norm": 5.0238473308596925,
      "learning_rate": 9.772084496568088e-07,
      "loss": 0.3443,
      "step": 1462
    },
    {
      "epoch": 0.3788805801042375,
      "grad_norm": 6.058458226117506,
      "learning_rate": 9.767414000622346e-07,
      "loss": 0.3752,
      "step": 1463
    },
    {
      "epoch": 0.37913955521025544,
      "grad_norm": 6.055971624389849,
      "learning_rate": 9.762740628181923e-07,
      "loss": 0.3526,
      "step": 1464
    },
    {
      "epoch": 0.37939853031627335,
      "grad_norm": 4.379766953792154,
      "learning_rate": 9.758064383068643e-07,
      "loss": 0.3024,
      "step": 1465
    },
    {
      "epoch": 0.37939853031627335,
      "eval_PRM Accuracy": 0.8867924528301887,
      "eval_PRM F1": 0.926829268292683,
      "eval_PRM F1 AUC": 0.8491356731272918,
      "eval_PRM F1 AUC (fixed)": 0.8171817705605029,
      "eval_PRM F1 Neg": 0.75,
      "eval_PRM NPV": 0.72,
      "eval_PRM Precision": 0.9382716049382716,
      "eval_PRM Recall": 0.9156626506024096,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.2357960194349289,
      "eval_runtime": 14.6708,
      "eval_samples_per_second": 2.999,
      "eval_steps_per_second": 0.136,
      "step": 1465
    },
    {
      "epoch": 0.37965750542229126,
      "grad_norm": 3.721510725794544,
      "learning_rate": 9.753385269106671e-07,
      "loss": 0.3405,
      "step": 1466
    },
    {
      "epoch": 0.37991648052830923,
      "grad_norm": 4.6197481619434795,
      "learning_rate": 9.74870329012253e-07,
      "loss": 0.3346,
      "step": 1467
    },
    {
      "epoch": 0.38017545563432714,
      "grad_norm": 3.318853840301337,
      "learning_rate": 9.744018449945078e-07,
      "loss": 0.342,
      "step": 1468
    },
    {
      "epoch": 0.3804344307403451,
      "grad_norm": 6.607534071714149,
      "learning_rate": 9.739330752405516e-07,
      "loss": 0.3281,
      "step": 1469
    },
    {
      "epoch": 0.380693405846363,
      "grad_norm": 3.3108665759335176,
      "learning_rate": 9.73464020133738e-07,
      "loss": 0.3388,
      "step": 1470
    },
    {
      "epoch": 0.380693405846363,
      "eval_PRM Accuracy": 0.8773584905660378,
      "eval_PRM F1": 0.9212121212121213,
      "eval_PRM F1 AUC": 0.8273965426925091,
      "eval_PRM F1 AUC (fixed)": 0.8192771084337349,
      "eval_PRM F1 Neg": 0.723404255319149,
      "eval_PRM NPV": 0.7083333333333334,
      "eval_PRM Precision": 0.926829268292683,
      "eval_PRM Recall": 0.9156626506024096,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.23642498254776,
      "eval_runtime": 14.734,
      "eval_samples_per_second": 2.986,
      "eval_steps_per_second": 0.136,
      "step": 1470
    },
    {
      "epoch": 0.38095238095238093,
      "grad_norm": 4.88496391866845,
      "learning_rate": 9.729946800576542e-07,
      "loss": 0.4142,
      "step": 1471
    },
    {
      "epoch": 0.3812113560583989,
      "grad_norm": 5.27916049939447,
      "learning_rate": 9.725250553961202e-07,
      "loss": 0.3391,
      "step": 1472
    },
    {
      "epoch": 0.3814703311644168,
      "grad_norm": 8.900314491547281,
      "learning_rate": 9.72055146533189e-07,
      "loss": 0.377,
      "step": 1473
    },
    {
      "epoch": 0.3817293062704348,
      "grad_norm": 3.0011588762933292,
      "learning_rate": 9.71584953853146e-07,
      "loss": 0.2884,
      "step": 1474
    },
    {
      "epoch": 0.3819882813764527,
      "grad_norm": 4.099551127706329,
      "learning_rate": 9.71114477740508e-07,
      "loss": 0.3726,
      "step": 1475
    },
    {
      "epoch": 0.3819882813764527,
      "eval_PRM Accuracy": 0.8773584905660378,
      "eval_PRM F1": 0.9212121212121213,
      "eval_PRM F1 AUC": 0.8273965426925091,
      "eval_PRM F1 AUC (fixed)": 0.8258250392875852,
      "eval_PRM F1 Neg": 0.723404255319149,
      "eval_PRM NPV": 0.7083333333333334,
      "eval_PRM Precision": 0.926829268292683,
      "eval_PRM Recall": 0.9156626506024096,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.23649762570858002,
      "eval_runtime": 14.8796,
      "eval_samples_per_second": 2.957,
      "eval_steps_per_second": 0.134,
      "step": 1475
    },
    {
      "epoch": 0.3822472564824706,
      "grad_norm": 3.6773421349404165,
      "learning_rate": 9.706437185800247e-07,
      "loss": 0.3779,
      "step": 1476
    },
    {
      "epoch": 0.38250623158848857,
      "grad_norm": 4.032342903491971,
      "learning_rate": 9.701726767566764e-07,
      "loss": 0.4382,
      "step": 1477
    },
    {
      "epoch": 0.3827652066945065,
      "grad_norm": 3.5491298285114325,
      "learning_rate": 9.697013526556751e-07,
      "loss": 0.3316,
      "step": 1478
    },
    {
      "epoch": 0.38302418180052444,
      "grad_norm": 4.341491391745633,
      "learning_rate": 9.692297466624633e-07,
      "loss": 0.3436,
      "step": 1479
    },
    {
      "epoch": 0.38328315690654235,
      "grad_norm": 2.8643056611618674,
      "learning_rate": 9.687578591627138e-07,
      "loss": 0.2758,
      "step": 1480
    },
    {
      "epoch": 0.38328315690654235,
      "eval_PRM Accuracy": 0.8490566037735849,
      "eval_PRM F1": 0.9,
      "eval_PRM F1 AUC": 0.8250392875851232,
      "eval_PRM F1 AUC (fixed)": 0.8187532739654269,
      "eval_PRM F1 Neg": 0.6923076923076923,
      "eval_PRM NPV": 0.6206896551724138,
      "eval_PRM Precision": 0.935064935064935,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.24352265894412994,
      "eval_runtime": 14.8616,
      "eval_samples_per_second": 2.961,
      "eval_steps_per_second": 0.135,
      "step": 1480
    },
    {
      "epoch": 0.38354213201256027,
      "grad_norm": 3.7035412141544444,
      "learning_rate": 9.682856905423307e-07,
      "loss": 0.2516,
      "step": 1481
    },
    {
      "epoch": 0.38380110711857823,
      "grad_norm": 6.508781566083587,
      "learning_rate": 9.678132411874468e-07,
      "loss": 0.4246,
      "step": 1482
    },
    {
      "epoch": 0.38406008222459614,
      "grad_norm": 3.258716176187973,
      "learning_rate": 9.67340511484425e-07,
      "loss": 0.3599,
      "step": 1483
    },
    {
      "epoch": 0.3843190573306141,
      "grad_norm": 3.4622217784743956,
      "learning_rate": 9.668675018198576e-07,
      "loss": 0.3031,
      "step": 1484
    },
    {
      "epoch": 0.384578032436632,
      "grad_norm": 5.0275928854286045,
      "learning_rate": 9.66394212580565e-07,
      "loss": 0.3924,
      "step": 1485
    },
    {
      "epoch": 0.384578032436632,
      "eval_PRM Accuracy": 0.8490566037735849,
      "eval_PRM F1": 0.9,
      "eval_PRM F1 AUC": 0.8250392875851232,
      "eval_PRM F1 AUC (fixed)": 0.8171817705605029,
      "eval_PRM F1 Neg": 0.6923076923076923,
      "eval_PRM NPV": 0.6206896551724138,
      "eval_PRM Precision": 0.935064935064935,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.24826860427856445,
      "eval_runtime": 14.271,
      "eval_samples_per_second": 3.083,
      "eval_steps_per_second": 0.14,
      "step": 1485
    },
    {
      "epoch": 0.38483700754265,
      "grad_norm": 6.109340461968093,
      "learning_rate": 9.659206441535976e-07,
      "loss": 0.3058,
      "step": 1486
    },
    {
      "epoch": 0.3850959826486679,
      "grad_norm": 5.412763780216841,
      "learning_rate": 9.654467969262333e-07,
      "loss": 0.3662,
      "step": 1487
    },
    {
      "epoch": 0.3853549577546858,
      "grad_norm": 4.1727011697393195,
      "learning_rate": 9.649726712859777e-07,
      "loss": 0.3745,
      "step": 1488
    },
    {
      "epoch": 0.3856139328607038,
      "grad_norm": 7.060653823640019,
      "learning_rate": 9.64498267620565e-07,
      "loss": 0.326,
      "step": 1489
    },
    {
      "epoch": 0.3858729079667217,
      "grad_norm": 3.036608182108932,
      "learning_rate": 9.640235863179559e-07,
      "loss": 0.2627,
      "step": 1490
    },
    {
      "epoch": 0.3858729079667217,
      "eval_PRM Accuracy": 0.8773584905660378,
      "eval_PRM F1": 0.9202453987730062,
      "eval_PRM F1 AUC": 0.8431115767417497,
      "eval_PRM F1 AUC (fixed)": 0.8268727082242011,
      "eval_PRM F1 Neg": 0.7346938775510204,
      "eval_PRM NPV": 0.6923076923076923,
      "eval_PRM Precision": 0.9375,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.23949016630649567,
      "eval_runtime": 14.7217,
      "eval_samples_per_second": 2.989,
      "eval_steps_per_second": 0.136,
      "step": 1490
    },
    {
      "epoch": 0.38613188307273966,
      "grad_norm": 3.9043233046211663,
      "learning_rate": 9.635486277663383e-07,
      "loss": 0.3002,
      "step": 1491
    },
    {
      "epoch": 0.38639085817875757,
      "grad_norm": 3.180806958564625,
      "learning_rate": 9.630733923541277e-07,
      "loss": 0.2761,
      "step": 1492
    },
    {
      "epoch": 0.3866498332847755,
      "grad_norm": 3.4766085563957825,
      "learning_rate": 9.625978804699646e-07,
      "loss": 0.2817,
      "step": 1493
    },
    {
      "epoch": 0.38690880839079345,
      "grad_norm": 12.109662314263637,
      "learning_rate": 9.621220925027168e-07,
      "loss": 0.4113,
      "step": 1494
    },
    {
      "epoch": 0.38716778349681136,
      "grad_norm": 4.082401544792815,
      "learning_rate": 9.616460288414776e-07,
      "loss": 0.3458,
      "step": 1495
    },
    {
      "epoch": 0.38716778349681136,
      "eval_PRM Accuracy": 0.8867924528301887,
      "eval_PRM F1": 0.9285714285714286,
      "eval_PRM F1 AUC": 0.817705605028811,
      "eval_PRM F1 AUC (fixed)": 0.8300157150340493,
      "eval_PRM F1 Neg": 0.7272727272727273,
      "eval_PRM NPV": 0.7619047619047619,
      "eval_PRM Precision": 0.9176470588235294,
      "eval_PRM Recall": 0.9397590361445783,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.23893995583057404,
      "eval_runtime": 14.7386,
      "eval_samples_per_second": 2.985,
      "eval_steps_per_second": 0.136,
      "step": 1495
    },
    {
      "epoch": 0.3874267586028293,
      "grad_norm": 3.5917626862055987,
      "learning_rate": 9.611696898755649e-07,
      "loss": 0.2712,
      "step": 1496
    },
    {
      "epoch": 0.38768573370884724,
      "grad_norm": 9.289904299091432,
      "learning_rate": 9.606930759945227e-07,
      "loss": 0.4151,
      "step": 1497
    },
    {
      "epoch": 0.38794470881486515,
      "grad_norm": 6.707271661263045,
      "learning_rate": 9.602161875881196e-07,
      "loss": 0.3717,
      "step": 1498
    },
    {
      "epoch": 0.3882036839208831,
      "grad_norm": 2.4515236018126743,
      "learning_rate": 9.59739025046349e-07,
      "loss": 0.2463,
      "step": 1499
    },
    {
      "epoch": 0.388462659026901,
      "grad_norm": 3.005546252622098,
      "learning_rate": 9.592615887594276e-07,
      "loss": 0.3076,
      "step": 1500
    },
    {
      "epoch": 0.388462659026901,
      "eval_PRM Accuracy": 0.9056603773584906,
      "eval_PRM F1": 0.9397590361445783,
      "eval_PRM F1 AUC": 0.8611838658983763,
      "eval_PRM F1 AUC (fixed)": 0.8328968046097434,
      "eval_PRM F1 Neg": 0.782608695652174,
      "eval_PRM NPV": 0.782608695652174,
      "eval_PRM Precision": 0.9397590361445783,
      "eval_PRM Recall": 0.9397590361445783,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.229969784617424,
      "eval_runtime": 14.7198,
      "eval_samples_per_second": 2.989,
      "eval_steps_per_second": 0.136,
      "step": 1500
    },
    {
      "epoch": 0.388721634132919,
      "grad_norm": 7.967579068140648,
      "learning_rate": 9.587838791177963e-07,
      "loss": 0.4003,
      "step": 1501
    },
    {
      "epoch": 0.3889806092389369,
      "grad_norm": 4.618845884100969,
      "learning_rate": 9.583058965121206e-07,
      "loss": 0.3345,
      "step": 1502
    },
    {
      "epoch": 0.3892395843449548,
      "grad_norm": 5.346673334225054,
      "learning_rate": 9.578276413332875e-07,
      "loss": 0.3575,
      "step": 1503
    },
    {
      "epoch": 0.3894985594509728,
      "grad_norm": 5.697466165554965,
      "learning_rate": 9.573491139724083e-07,
      "loss": 0.4453,
      "step": 1504
    },
    {
      "epoch": 0.3897575345569907,
      "grad_norm": 5.2630972374192755,
      "learning_rate": 9.56870314820816e-07,
      "loss": 0.3561,
      "step": 1505
    },
    {
      "epoch": 0.3897575345569907,
      "eval_PRM Accuracy": 0.8679245283018868,
      "eval_PRM F1": 0.9125,
      "eval_PRM F1 AUC": 0.8528025144054479,
      "eval_PRM F1 AUC (fixed)": 0.8247773703509691,
      "eval_PRM F1 Neg": 0.7307692307692307,
      "eval_PRM NPV": 0.6551724137931034,
      "eval_PRM Precision": 0.948051948051948,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.8260869565217391,
      "eval_loss": 0.2363291233778,
      "eval_runtime": 15.0616,
      "eval_samples_per_second": 2.921,
      "eval_steps_per_second": 0.133,
      "step": 1505
    },
    {
      "epoch": 0.39001650966300866,
      "grad_norm": 3.777040254982168,
      "learning_rate": 9.563912442700661e-07,
      "loss": 0.3645,
      "step": 1506
    },
    {
      "epoch": 0.39027548476902657,
      "grad_norm": 3.1829088852941547,
      "learning_rate": 9.559119027119371e-07,
      "loss": 0.3168,
      "step": 1507
    },
    {
      "epoch": 0.39053445987504454,
      "grad_norm": 3.6376117510857617,
      "learning_rate": 9.554322905384269e-07,
      "loss": 0.378,
      "step": 1508
    },
    {
      "epoch": 0.39079343498106245,
      "grad_norm": 8.22323548130612,
      "learning_rate": 9.54952408141757e-07,
      "loss": 0.3481,
      "step": 1509
    },
    {
      "epoch": 0.39105241008708036,
      "grad_norm": 7.604451735063203,
      "learning_rate": 9.544722559143684e-07,
      "loss": 0.3359,
      "step": 1510
    },
    {
      "epoch": 0.39105241008708036,
      "eval_PRM Accuracy": 0.8679245283018868,
      "eval_PRM F1": 0.9125,
      "eval_PRM F1 AUC": 0.8528025144054479,
      "eval_PRM F1 AUC (fixed)": 0.818753273965427,
      "eval_PRM F1 Neg": 0.7307692307692307,
      "eval_PRM NPV": 0.6551724137931034,
      "eval_PRM Precision": 0.948051948051948,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.8260869565217391,
      "eval_loss": 0.24453581869602203,
      "eval_runtime": 14.2697,
      "eval_samples_per_second": 3.083,
      "eval_steps_per_second": 0.14,
      "step": 1510
    },
    {
      "epoch": 0.3913113851930983,
      "grad_norm": 5.896979933649967,
      "learning_rate": 9.539918342489232e-07,
      "loss": 0.3824,
      "step": 1511
    },
    {
      "epoch": 0.39157036029911624,
      "grad_norm": 3.211554120427429,
      "learning_rate": 9.535111435383043e-07,
      "loss": 0.3195,
      "step": 1512
    },
    {
      "epoch": 0.3918293354051342,
      "grad_norm": 2.6798542655354125,
      "learning_rate": 9.530301841756139e-07,
      "loss": 0.3249,
      "step": 1513
    },
    {
      "epoch": 0.3920883105111521,
      "grad_norm": 6.785587452731779,
      "learning_rate": 9.52548956554174e-07,
      "loss": 0.3319,
      "step": 1514
    },
    {
      "epoch": 0.39234728561717,
      "grad_norm": 6.07488058517214,
      "learning_rate": 9.520674610675265e-07,
      "loss": 0.3176,
      "step": 1515
    },
    {
      "epoch": 0.39234728561717,
      "eval_PRM Accuracy": 0.9056603773584906,
      "eval_PRM F1": 0.9390243902439024,
      "eval_PRM F1 AUC": 0.8768988999476165,
      "eval_PRM F1 AUC (fixed)": 0.8213724463069669,
      "eval_PRM F1 Neg": 0.7916666666666666,
      "eval_PRM NPV": 0.76,
      "eval_PRM Precision": 0.9506172839506173,
      "eval_PRM Recall": 0.927710843373494,
      "eval_PRM Specificty": 0.8260869565217391,
      "eval_loss": 0.23415261507034302,
      "eval_runtime": 14.6219,
      "eval_samples_per_second": 3.009,
      "eval_steps_per_second": 0.137,
      "step": 1515
    },
    {
      "epoch": 0.392606260723188,
      "grad_norm": 3.0395299401082605,
      "learning_rate": 9.515856981094323e-07,
      "loss": 0.2384,
      "step": 1516
    },
    {
      "epoch": 0.3928652358292059,
      "grad_norm": 3.845802837421083,
      "learning_rate": 9.511036680738703e-07,
      "loss": 0.3435,
      "step": 1517
    },
    {
      "epoch": 0.3931242109352239,
      "grad_norm": 5.389403587129739,
      "learning_rate": 9.506213713550384e-07,
      "loss": 0.4791,
      "step": 1518
    },
    {
      "epoch": 0.3933831860412418,
      "grad_norm": 9.000139605946577,
      "learning_rate": 9.501388083473524e-07,
      "loss": 0.4559,
      "step": 1519
    },
    {
      "epoch": 0.3936421611472597,
      "grad_norm": 7.39259048113181,
      "learning_rate": 9.496559794454464e-07,
      "loss": 0.4693,
      "step": 1520
    },
    {
      "epoch": 0.3936421611472597,
      "eval_PRM Accuracy": 0.9056603773584906,
      "eval_PRM F1": 0.9397590361445783,
      "eval_PRM F1 AUC": 0.8611838658983763,
      "eval_PRM F1 AUC (fixed)": 0.8279203771608172,
      "eval_PRM F1 Neg": 0.782608695652174,
      "eval_PRM NPV": 0.782608695652174,
      "eval_PRM Precision": 0.9397590361445783,
      "eval_PRM Recall": 0.9397590361445783,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.22763299942016602,
      "eval_runtime": 15.2442,
      "eval_samples_per_second": 2.886,
      "eval_steps_per_second": 0.131,
      "step": 1520
    },
    {
      "epoch": 0.39390113625327766,
      "grad_norm": 6.186042956749198,
      "learning_rate": 9.491728850441711e-07,
      "loss": 0.4015,
      "step": 1521
    },
    {
      "epoch": 0.3941601113592956,
      "grad_norm": 7.0652245611169695,
      "learning_rate": 9.486895255385947e-07,
      "loss": 0.3989,
      "step": 1522
    },
    {
      "epoch": 0.39441908646531354,
      "grad_norm": 4.537471410549836,
      "learning_rate": 9.482059013240025e-07,
      "loss": 0.341,
      "step": 1523
    },
    {
      "epoch": 0.39467806157133145,
      "grad_norm": 2.8105350778496474,
      "learning_rate": 9.477220127958955e-07,
      "loss": 0.2762,
      "step": 1524
    },
    {
      "epoch": 0.39493703667734936,
      "grad_norm": 3.827521733683929,
      "learning_rate": 9.472378603499919e-07,
      "loss": 0.3103,
      "step": 1525
    },
    {
      "epoch": 0.39493703667734936,
      "eval_PRM Accuracy": 0.8773584905660378,
      "eval_PRM F1": 0.9202453987730062,
      "eval_PRM F1 AUC": 0.8431115767417497,
      "eval_PRM F1 AUC (fixed)": 0.8239916186485071,
      "eval_PRM F1 Neg": 0.7346938775510204,
      "eval_PRM NPV": 0.6923076923076923,
      "eval_PRM Precision": 0.9375,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.22646942734718323,
      "eval_runtime": 15.0861,
      "eval_samples_per_second": 2.917,
      "eval_steps_per_second": 0.133,
      "step": 1525
    },
    {
      "epoch": 0.39519601178336733,
      "grad_norm": 3.5684855750803823,
      "learning_rate": 9.467534443822249e-07,
      "loss": 0.3505,
      "step": 1526
    },
    {
      "epoch": 0.39545498688938524,
      "grad_norm": 4.211757250096138,
      "learning_rate": 9.462687652887434e-07,
      "loss": 0.3475,
      "step": 1527
    },
    {
      "epoch": 0.3957139619954032,
      "grad_norm": 3.9955346599067147,
      "learning_rate": 9.457838234659118e-07,
      "loss": 0.4099,
      "step": 1528
    },
    {
      "epoch": 0.3959729371014211,
      "grad_norm": 4.619680433983152,
      "learning_rate": 9.452986193103089e-07,
      "loss": 0.3414,
      "step": 1529
    },
    {
      "epoch": 0.3962319122074391,
      "grad_norm": 8.085827130049163,
      "learning_rate": 9.448131532187284e-07,
      "loss": 0.387,
      "step": 1530
    },
    {
      "epoch": 0.3962319122074391,
      "eval_PRM Accuracy": 0.8679245283018868,
      "eval_PRM F1": 0.9113924050632911,
      "eval_PRM F1 AUC": 0.8685175484546883,
      "eval_PRM F1 AUC (fixed)": 0.8163960188580408,
      "eval_PRM F1 Neg": 0.7407407407407407,
      "eval_PRM NPV": 0.6451612903225806,
      "eval_PRM Precision": 0.96,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.8695652173913043,
      "eval_loss": 0.2410266399383545,
      "eval_runtime": 14.8446,
      "eval_samples_per_second": 2.964,
      "eval_steps_per_second": 0.135,
      "step": 1530
    },
    {
      "epoch": 0.396490887313457,
      "grad_norm": 6.9949185617400245,
      "learning_rate": 9.44327425588178e-07,
      "loss": 0.3121,
      "step": 1531
    },
    {
      "epoch": 0.3967498624194749,
      "grad_norm": 3.396161701582718,
      "learning_rate": 9.438414368158792e-07,
      "loss": 0.3915,
      "step": 1532
    },
    {
      "epoch": 0.3970088375254929,
      "grad_norm": 3.253301134840018,
      "learning_rate": 9.433551872992671e-07,
      "loss": 0.2768,
      "step": 1533
    },
    {
      "epoch": 0.3972678126315108,
      "grad_norm": 11.366561643113847,
      "learning_rate": 9.428686774359906e-07,
      "loss": 0.3476,
      "step": 1534
    },
    {
      "epoch": 0.39752678773752875,
      "grad_norm": 5.6249170681103875,
      "learning_rate": 9.423819076239108e-07,
      "loss": 0.3954,
      "step": 1535
    },
    {
      "epoch": 0.39752678773752875,
      "eval_PRM Accuracy": 0.8773584905660378,
      "eval_PRM F1": 0.9192546583850931,
      "eval_PRM F1 AUC": 0.8588266107909901,
      "eval_PRM F1 AUC (fixed)": 0.8187532739654269,
      "eval_PRM F1 Neg": 0.7450980392156863,
      "eval_PRM NPV": 0.6785714285714286,
      "eval_PRM Precision": 0.9487179487179487,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.8260869565217391,
      "eval_loss": 0.2305150032043457,
      "eval_runtime": 14.877,
      "eval_samples_per_second": 2.958,
      "eval_steps_per_second": 0.134,
      "step": 1535
    },
    {
      "epoch": 0.39778576284354666,
      "grad_norm": 3.476121108479144,
      "learning_rate": 9.418948782611015e-07,
      "loss": 0.3245,
      "step": 1536
    },
    {
      "epoch": 0.3980447379495646,
      "grad_norm": 4.141270160746721,
      "learning_rate": 9.414075897458491e-07,
      "loss": 0.2943,
      "step": 1537
    },
    {
      "epoch": 0.39830371305558254,
      "grad_norm": 3.6135377415071797,
      "learning_rate": 9.409200424766513e-07,
      "loss": 0.2702,
      "step": 1538
    },
    {
      "epoch": 0.39856268816160045,
      "grad_norm": 2.717922771113031,
      "learning_rate": 9.404322368522181e-07,
      "loss": 0.3073,
      "step": 1539
    },
    {
      "epoch": 0.3988216632676184,
      "grad_norm": 3.5871641161734877,
      "learning_rate": 9.399441732714706e-07,
      "loss": 0.4095,
      "step": 1540
    },
    {
      "epoch": 0.3988216632676184,
      "eval_PRM Accuracy": 0.8773584905660378,
      "eval_PRM F1": 0.9202453987730062,
      "eval_PRM F1 AUC": 0.8431115767417497,
      "eval_PRM F1 AUC (fixed)": 0.8184913567312728,
      "eval_PRM F1 Neg": 0.7346938775510204,
      "eval_PRM NPV": 0.6923076923076923,
      "eval_PRM Precision": 0.9375,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.21936039626598358,
      "eval_runtime": 15.4346,
      "eval_samples_per_second": 2.851,
      "eval_steps_per_second": 0.13,
      "step": 1540
    },
    {
      "epoch": 0.39908063837363633,
      "grad_norm": 3.188644216295607,
      "learning_rate": 9.394558521335409e-07,
      "loss": 0.2844,
      "step": 1541
    },
    {
      "epoch": 0.39933961347965424,
      "grad_norm": 6.54859753722624,
      "learning_rate": 9.389672738377709e-07,
      "loss": 0.3931,
      "step": 1542
    },
    {
      "epoch": 0.3995985885856722,
      "grad_norm": 3.8231446280281642,
      "learning_rate": 9.38478438783714e-07,
      "loss": 0.2906,
      "step": 1543
    },
    {
      "epoch": 0.3998575636916901,
      "grad_norm": 3.7127784502781993,
      "learning_rate": 9.379893473711328e-07,
      "loss": 0.3128,
      "step": 1544
    },
    {
      "epoch": 0.4001165387977081,
      "grad_norm": 3.38321851419847,
      "learning_rate": 9.375000000000001e-07,
      "loss": 0.3173,
      "step": 1545
    },
    {
      "epoch": 0.4001165387977081,
      "eval_PRM Accuracy": 0.8773584905660378,
      "eval_PRM F1": 0.9202453987730062,
      "eval_PRM F1 AUC": 0.8431115767417497,
      "eval_PRM F1 AUC (fixed)": 0.8218962807752751,
      "eval_PRM F1 Neg": 0.7346938775510204,
      "eval_PRM NPV": 0.6923076923076923,
      "eval_PRM Precision": 0.9375,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.21774643659591675,
      "eval_runtime": 14.9003,
      "eval_samples_per_second": 2.953,
      "eval_steps_per_second": 0.134,
      "step": 1545
    },
    {
      "epoch": 0.400375513903726,
      "grad_norm": 3.557217840230169,
      "learning_rate": 9.370103970704975e-07,
      "loss": 0.3828,
      "step": 1546
    },
    {
      "epoch": 0.40063448900974397,
      "grad_norm": 8.849395387956283,
      "learning_rate": 9.365205389830156e-07,
      "loss": 0.4061,
      "step": 1547
    },
    {
      "epoch": 0.4008934641157619,
      "grad_norm": 2.921087421087627,
      "learning_rate": 9.360304261381542e-07,
      "loss": 0.3177,
      "step": 1548
    },
    {
      "epoch": 0.4011524392217798,
      "grad_norm": 4.234672156968746,
      "learning_rate": 9.355400589367213e-07,
      "loss": 0.3245,
      "step": 1549
    },
    {
      "epoch": 0.40141141432779776,
      "grad_norm": 5.725932469896439,
      "learning_rate": 9.350494377797322e-07,
      "loss": 0.3294,
      "step": 1550
    },
    {
      "epoch": 0.40141141432779776,
      "eval_PRM Accuracy": 0.8773584905660378,
      "eval_PRM F1": 0.9202453987730062,
      "eval_PRM F1 AUC": 0.8431115767417497,
      "eval_PRM F1 AUC (fixed)": 0.8268727082242011,
      "eval_PRM F1 Neg": 0.7346938775510204,
      "eval_PRM NPV": 0.6923076923076923,
      "eval_PRM Precision": 0.9375,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.22031234204769135,
      "eval_runtime": 14.3288,
      "eval_samples_per_second": 3.071,
      "eval_steps_per_second": 0.14,
      "step": 1550
    },
    {
      "epoch": 0.40167038943381567,
      "grad_norm": 8.642160997896136,
      "learning_rate": 9.345585630684105e-07,
      "loss": 0.3691,
      "step": 1551
    },
    {
      "epoch": 0.40192936453983363,
      "grad_norm": 10.358474583053521,
      "learning_rate": 9.340674352041873e-07,
      "loss": 0.3275,
      "step": 1552
    },
    {
      "epoch": 0.40218833964585154,
      "grad_norm": 4.968101854428275,
      "learning_rate": 9.335760545887003e-07,
      "loss": 0.4266,
      "step": 1553
    },
    {
      "epoch": 0.40244731475186946,
      "grad_norm": 3.271958598226771,
      "learning_rate": 9.330844216237942e-07,
      "loss": 0.3257,
      "step": 1554
    },
    {
      "epoch": 0.4027062898578874,
      "grad_norm": 4.12676860829523,
      "learning_rate": 9.325925367115197e-07,
      "loss": 0.4283,
      "step": 1555
    },
    {
      "epoch": 0.4027062898578874,
      "eval_PRM Accuracy": 0.8773584905660378,
      "eval_PRM F1": 0.9202453987730062,
      "eval_PRM F1 AUC": 0.8431115767417497,
      "eval_PRM F1 AUC (fixed)": 0.8232058669460451,
      "eval_PRM F1 Neg": 0.7346938775510204,
      "eval_PRM NPV": 0.6923076923076923,
      "eval_PRM Precision": 0.9375,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.2232244610786438,
      "eval_runtime": 14.8539,
      "eval_samples_per_second": 2.962,
      "eval_steps_per_second": 0.135,
      "step": 1555
    },
    {
      "epoch": 0.40296526496390533,
      "grad_norm": 2.8116528583993916,
      "learning_rate": 9.321004002541341e-07,
      "loss": 0.3272,
      "step": 1556
    },
    {
      "epoch": 0.4032242400699233,
      "grad_norm": 4.66248390250764,
      "learning_rate": 9.316080126540998e-07,
      "loss": 0.4285,
      "step": 1557
    },
    {
      "epoch": 0.4034832151759412,
      "grad_norm": 4.513837861618739,
      "learning_rate": 9.31115374314085e-07,
      "loss": 0.4012,
      "step": 1558
    },
    {
      "epoch": 0.4037421902819591,
      "grad_norm": 4.2647128198918685,
      "learning_rate": 9.306224856369629e-07,
      "loss": 0.24,
      "step": 1559
    },
    {
      "epoch": 0.4040011653879771,
      "grad_norm": 4.98465110903521,
      "learning_rate": 9.301293470258114e-07,
      "loss": 0.3474,
      "step": 1560
    },
    {
      "epoch": 0.4040011653879771,
      "eval_PRM Accuracy": 0.8773584905660378,
      "eval_PRM F1": 0.9192546583850931,
      "eval_PRM F1 AUC": 0.8588266107909901,
      "eval_PRM F1 AUC (fixed)": 0.819800942902043,
      "eval_PRM F1 Neg": 0.7450980392156863,
      "eval_PRM NPV": 0.6785714285714286,
      "eval_PRM Precision": 0.9487179487179487,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.8260869565217391,
      "eval_loss": 0.22973184287548065,
      "eval_runtime": 15.1897,
      "eval_samples_per_second": 2.897,
      "eval_steps_per_second": 0.132,
      "step": 1560
    },
    {
      "epoch": 0.404260140493995,
      "grad_norm": 6.340696302675487,
      "learning_rate": 9.296359588839125e-07,
      "loss": 0.3774,
      "step": 1561
    },
    {
      "epoch": 0.40451911560001297,
      "grad_norm": 4.009786635171831,
      "learning_rate": 9.291423216147524e-07,
      "loss": 0.3783,
      "step": 1562
    },
    {
      "epoch": 0.4047780907060309,
      "grad_norm": 4.095710727042009,
      "learning_rate": 9.286484356220215e-07,
      "loss": 0.3646,
      "step": 1563
    },
    {
      "epoch": 0.4050370658120488,
      "grad_norm": 3.3795032120863513,
      "learning_rate": 9.281543013096132e-07,
      "loss": 0.3555,
      "step": 1564
    },
    {
      "epoch": 0.40529604091806676,
      "grad_norm": 3.5500135879051746,
      "learning_rate": 9.276599190816238e-07,
      "loss": 0.2768,
      "step": 1565
    },
    {
      "epoch": 0.40529604091806676,
      "eval_PRM Accuracy": 0.8773584905660378,
      "eval_PRM F1": 0.9202453987730062,
      "eval_PRM F1 AUC": 0.8431115767417497,
      "eval_PRM F1 AUC (fixed)": 0.8122053431115768,
      "eval_PRM F1 Neg": 0.7346938775510204,
      "eval_PRM NPV": 0.6923076923076923,
      "eval_PRM Precision": 0.9375,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.22858896851539612,
      "eval_runtime": 14.798,
      "eval_samples_per_second": 2.973,
      "eval_steps_per_second": 0.135,
      "step": 1565
    },
    {
      "epoch": 0.40555501602408467,
      "grad_norm": 3.709950401983799,
      "learning_rate": 9.271652893423527e-07,
      "loss": 0.4145,
      "step": 1566
    },
    {
      "epoch": 0.40581399113010264,
      "grad_norm": 4.33652144371034,
      "learning_rate": 9.266704124963017e-07,
      "loss": 0.3367,
      "step": 1567
    },
    {
      "epoch": 0.40607296623612055,
      "grad_norm": 3.5503572210053713,
      "learning_rate": 9.261752889481738e-07,
      "loss": 0.3398,
      "step": 1568
    },
    {
      "epoch": 0.4063319413421385,
      "grad_norm": 3.5308579058818217,
      "learning_rate": 9.256799191028753e-07,
      "loss": 0.2959,
      "step": 1569
    },
    {
      "epoch": 0.4065909164481564,
      "grad_norm": 2.942235398456754,
      "learning_rate": 9.251843033655131e-07,
      "loss": 0.3303,
      "step": 1570
    },
    {
      "epoch": 0.4065909164481564,
      "eval_PRM Accuracy": 0.8867924528301887,
      "eval_PRM F1": 0.9259259259259259,
      "eval_PRM F1 AUC": 0.8648507071765322,
      "eval_PRM F1 AUC (fixed)": 0.8067050811943426,
      "eval_PRM F1 Neg": 0.76,
      "eval_PRM NPV": 0.7037037037037037,
      "eval_PRM Precision": 0.9493670886075949,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.8260869565217391,
      "eval_loss": 0.2316356599330902,
      "eval_runtime": 14.6623,
      "eval_samples_per_second": 3.001,
      "eval_steps_per_second": 0.136,
      "step": 1570
    },
    {
      "epoch": 0.40684989155417434,
      "grad_norm": 3.110905983808249,
      "learning_rate": 9.246884421413949e-07,
      "loss": 0.3391,
      "step": 1571
    },
    {
      "epoch": 0.4071088666601923,
      "grad_norm": 6.939051887466604,
      "learning_rate": 9.241923358360294e-07,
      "loss": 0.3787,
      "step": 1572
    },
    {
      "epoch": 0.4073678417662102,
      "grad_norm": 3.7045178313961844,
      "learning_rate": 9.236959848551258e-07,
      "loss": 0.3536,
      "step": 1573
    },
    {
      "epoch": 0.4076268168722282,
      "grad_norm": 4.9284738463473525,
      "learning_rate": 9.231993896045936e-07,
      "loss": 0.3511,
      "step": 1574
    },
    {
      "epoch": 0.4078857919782461,
      "grad_norm": 4.0170268942643315,
      "learning_rate": 9.227025504905417e-07,
      "loss": 0.3561,
      "step": 1575
    },
    {
      "epoch": 0.4078857919782461,
      "eval_PRM Accuracy": 0.8867924528301887,
      "eval_PRM F1": 0.9259259259259259,
      "eval_PRM F1 AUC": 0.8648507071765322,
      "eval_PRM F1 AUC (fixed)": 0.8098480880041907,
      "eval_PRM F1 Neg": 0.76,
      "eval_PRM NPV": 0.7037037037037037,
      "eval_PRM Precision": 0.9493670886075949,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.8260869565217391,
      "eval_loss": 0.2285064309835434,
      "eval_runtime": 14.3724,
      "eval_samples_per_second": 3.061,
      "eval_steps_per_second": 0.139,
      "step": 1575
    },
    {
      "epoch": 0.408144767084264,
      "grad_norm": 3.6292010693955086,
      "learning_rate": 9.222054679192785e-07,
      "loss": 0.2958,
      "step": 1576
    },
    {
      "epoch": 0.40840374219028197,
      "grad_norm": 3.074604666947679,
      "learning_rate": 9.217081422973116e-07,
      "loss": 0.3354,
      "step": 1577
    },
    {
      "epoch": 0.4086627172962999,
      "grad_norm": 4.533096410294307,
      "learning_rate": 9.212105740313473e-07,
      "loss": 0.2872,
      "step": 1578
    },
    {
      "epoch": 0.40892169240231785,
      "grad_norm": 6.79973062891781,
      "learning_rate": 9.207127635282906e-07,
      "loss": 0.301,
      "step": 1579
    },
    {
      "epoch": 0.40918066750833576,
      "grad_norm": 3.765681353148795,
      "learning_rate": 9.20214711195244e-07,
      "loss": 0.3086,
      "step": 1580
    },
    {
      "epoch": 0.40918066750833576,
      "eval_PRM Accuracy": 0.8867924528301887,
      "eval_PRM F1": 0.9259259259259259,
      "eval_PRM F1 AUC": 0.8648507071765322,
      "eval_PRM F1 AUC (fixed)": 0.8114195914091148,
      "eval_PRM F1 Neg": 0.76,
      "eval_PRM NPV": 0.7037037037037037,
      "eval_PRM Precision": 0.9493670886075949,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.8260869565217391,
      "eval_loss": 0.22827695310115814,
      "eval_runtime": 14.5918,
      "eval_samples_per_second": 3.015,
      "eval_steps_per_second": 0.137,
      "step": 1580
    },
    {
      "epoch": 0.40943964261435367,
      "grad_norm": 4.613099586737731,
      "learning_rate": 9.197164174395084e-07,
      "loss": 0.2904,
      "step": 1581
    },
    {
      "epoch": 0.40969861772037164,
      "grad_norm": 3.3282662950413666,
      "learning_rate": 9.192178826685814e-07,
      "loss": 0.3379,
      "step": 1582
    },
    {
      "epoch": 0.40995759282638955,
      "grad_norm": 3.7861278832972296,
      "learning_rate": 9.187191072901587e-07,
      "loss": 0.311,
      "step": 1583
    },
    {
      "epoch": 0.4102165679324075,
      "grad_norm": 4.181741160395208,
      "learning_rate": 9.182200917121318e-07,
      "loss": 0.3841,
      "step": 1584
    },
    {
      "epoch": 0.41047554303842543,
      "grad_norm": 6.821150870528454,
      "learning_rate": 9.177208363425894e-07,
      "loss": 0.3986,
      "step": 1585
    },
    {
      "epoch": 0.41047554303842543,
      "eval_PRM Accuracy": 0.8679245283018868,
      "eval_PRM F1": 0.9135802469135802,
      "eval_PRM F1 AUC": 0.8370874803562075,
      "eval_PRM F1 AUC (fixed)": 0.8127291775798847,
      "eval_PRM F1 Neg": 0.72,
      "eval_PRM NPV": 0.6666666666666666,
      "eval_PRM Precision": 0.9367088607594937,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.23270650207996368,
      "eval_runtime": 14.6306,
      "eval_samples_per_second": 3.007,
      "eval_steps_per_second": 0.137,
      "step": 1585
    },
    {
      "epoch": 0.41073451814444334,
      "grad_norm": 3.181380014402195,
      "learning_rate": 9.172213415898157e-07,
      "loss": 0.3496,
      "step": 1586
    },
    {
      "epoch": 0.4109934932504613,
      "grad_norm": 4.202969566505007,
      "learning_rate": 9.16721607862291e-07,
      "loss": 0.3644,
      "step": 1587
    },
    {
      "epoch": 0.4112524683564792,
      "grad_norm": 2.968705616731075,
      "learning_rate": 9.162216355686907e-07,
      "loss": 0.2971,
      "step": 1588
    },
    {
      "epoch": 0.4115114434624972,
      "grad_norm": 2.8197104250663507,
      "learning_rate": 9.157214251178859e-07,
      "loss": 0.318,
      "step": 1589
    },
    {
      "epoch": 0.4117704185685151,
      "grad_norm": 4.617171577840848,
      "learning_rate": 9.152209769189417e-07,
      "loss": 0.3833,
      "step": 1590
    },
    {
      "epoch": 0.4117704185685151,
      "eval_PRM Accuracy": 0.8867924528301887,
      "eval_PRM F1": 0.9259259259259259,
      "eval_PRM F1 AUC": 0.8648507071765322,
      "eval_PRM F1 AUC (fixed)": 0.8119434258774227,
      "eval_PRM F1 Neg": 0.76,
      "eval_PRM NPV": 0.7037037037037037,
      "eval_PRM Precision": 0.9493670886075949,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.8260869565217391,
      "eval_loss": 0.23377208411693573,
      "eval_runtime": 14.7727,
      "eval_samples_per_second": 2.978,
      "eval_steps_per_second": 0.135,
      "step": 1590
    },
    {
      "epoch": 0.41202939367453306,
      "grad_norm": 4.065809620813271,
      "learning_rate": 9.147202913811186e-07,
      "loss": 0.3289,
      "step": 1591
    },
    {
      "epoch": 0.412288368780551,
      "grad_norm": 4.268544709017804,
      "learning_rate": 9.142193689138699e-07,
      "loss": 0.2772,
      "step": 1592
    },
    {
      "epoch": 0.4125473438865689,
      "grad_norm": 6.581226734835593,
      "learning_rate": 9.137182099268439e-07,
      "loss": 0.3708,
      "step": 1593
    },
    {
      "epoch": 0.41280631899258685,
      "grad_norm": 3.5546798516442504,
      "learning_rate": 9.132168148298817e-07,
      "loss": 0.3476,
      "step": 1594
    },
    {
      "epoch": 0.41306529409860476,
      "grad_norm": 4.481149042660927,
      "learning_rate": 9.127151840330175e-07,
      "loss": 0.3521,
      "step": 1595
    },
    {
      "epoch": 0.41306529409860476,
      "eval_PRM Accuracy": 0.8867924528301887,
      "eval_PRM F1": 0.9259259259259259,
      "eval_PRM F1 AUC": 0.8648507071765322,
      "eval_PRM F1 AUC (fixed)": 0.8132530120481928,
      "eval_PRM F1 Neg": 0.76,
      "eval_PRM NPV": 0.7037037037037037,
      "eval_PRM Precision": 0.9493670886075949,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.8260869565217391,
      "eval_loss": 0.2304309904575348,
      "eval_runtime": 14.8136,
      "eval_samples_per_second": 2.97,
      "eval_steps_per_second": 0.135,
      "step": 1595
    },
    {
      "epoch": 0.41332426920462273,
      "grad_norm": 3.570569170599757,
      "learning_rate": 9.122133179464785e-07,
      "loss": 0.3589,
      "step": 1596
    },
    {
      "epoch": 0.41358324431064064,
      "grad_norm": 7.553813247270276,
      "learning_rate": 9.117112169806835e-07,
      "loss": 0.4214,
      "step": 1597
    },
    {
      "epoch": 0.41384221941665855,
      "grad_norm": 5.820835980150305,
      "learning_rate": 9.11208881546245e-07,
      "loss": 0.3483,
      "step": 1598
    },
    {
      "epoch": 0.4141011945226765,
      "grad_norm": 7.6282942209541496,
      "learning_rate": 9.107063120539658e-07,
      "loss": 0.404,
      "step": 1599
    },
    {
      "epoch": 0.41436016962869443,
      "grad_norm": 3.568743519308355,
      "learning_rate": 9.102035089148405e-07,
      "loss": 0.3004,
      "step": 1600
    },
    {
      "epoch": 0.41436016962869443,
      "eval_PRM Accuracy": 0.8773584905660378,
      "eval_PRM F1": 0.9192546583850931,
      "eval_PRM F1 AUC": 0.8588266107909901,
      "eval_PRM F1 AUC (fixed)": 0.8116815086432687,
      "eval_PRM F1 Neg": 0.7450980392156863,
      "eval_PRM NPV": 0.6785714285714286,
      "eval_PRM Precision": 0.9487179487179487,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.8260869565217391,
      "eval_loss": 0.22910362482070923,
      "eval_runtime": 14.6683,
      "eval_samples_per_second": 3.0,
      "eval_steps_per_second": 0.136,
      "step": 1600
    },
    {
      "epoch": 0.4146191447347124,
      "grad_norm": 4.109072376223985,
      "learning_rate": 9.097004725400549e-07,
      "loss": 0.4477,
      "step": 1601
    },
    {
      "epoch": 0.4148781198407303,
      "grad_norm": 4.930284637785742,
      "learning_rate": 9.091972033409855e-07,
      "loss": 0.3576,
      "step": 1602
    },
    {
      "epoch": 0.4151370949467482,
      "grad_norm": 3.3043274415947788,
      "learning_rate": 9.086937017291995e-07,
      "loss": 0.3254,
      "step": 1603
    },
    {
      "epoch": 0.4153960700527662,
      "grad_norm": 4.309456499070708,
      "learning_rate": 9.081899681164536e-07,
      "loss": 0.3716,
      "step": 1604
    },
    {
      "epoch": 0.4156550451587841,
      "grad_norm": 5.998403938471746,
      "learning_rate": 9.076860029146944e-07,
      "loss": 0.4528,
      "step": 1605
    },
    {
      "epoch": 0.4156550451587841,
      "eval_PRM Accuracy": 0.8773584905660378,
      "eval_PRM F1": 0.9192546583850931,
      "eval_PRM F1 AUC": 0.8588266107909901,
      "eval_PRM F1 AUC (fixed)": 0.8132530120481928,
      "eval_PRM F1 Neg": 0.7450980392156863,
      "eval_PRM NPV": 0.6785714285714286,
      "eval_PRM Precision": 0.9487179487179487,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.8260869565217391,
      "eval_loss": 0.2280660718679428,
      "eval_runtime": 14.1784,
      "eval_samples_per_second": 3.103,
      "eval_steps_per_second": 0.141,
      "step": 1605
    },
    {
      "epoch": 0.41591402026480206,
      "grad_norm": 8.11627497388889,
      "learning_rate": 9.071818065360583e-07,
      "loss": 0.3177,
      "step": 1606
    },
    {
      "epoch": 0.41617299537082,
      "grad_norm": 3.846957144646813,
      "learning_rate": 9.0667737939287e-07,
      "loss": 0.2999,
      "step": 1607
    },
    {
      "epoch": 0.4164319704768379,
      "grad_norm": 3.6591881455003787,
      "learning_rate": 9.061727218976437e-07,
      "loss": 0.3537,
      "step": 1608
    },
    {
      "epoch": 0.41669094558285585,
      "grad_norm": 4.10036283772463,
      "learning_rate": 9.056678344630814e-07,
      "loss": 0.3731,
      "step": 1609
    },
    {
      "epoch": 0.41694992068887377,
      "grad_norm": 3.5516059755511002,
      "learning_rate": 9.051627175020737e-07,
      "loss": 0.368,
      "step": 1610
    },
    {
      "epoch": 0.41694992068887377,
      "eval_PRM Accuracy": 0.8773584905660378,
      "eval_PRM F1": 0.9192546583850931,
      "eval_PRM F1 AUC": 0.8588266107909901,
      "eval_PRM F1 AUC (fixed)": 0.8095861707700367,
      "eval_PRM F1 Neg": 0.7450980392156863,
      "eval_PRM NPV": 0.6785714285714286,
      "eval_PRM Precision": 0.9487179487179487,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.8260869565217391,
      "eval_loss": 0.2249169796705246,
      "eval_runtime": 14.6382,
      "eval_samples_per_second": 3.006,
      "eval_steps_per_second": 0.137,
      "step": 1610
    },
    {
      "epoch": 0.41720889579489173,
      "grad_norm": 2.9603314781588126,
      "learning_rate": 9.046573714276983e-07,
      "loss": 0.2897,
      "step": 1611
    },
    {
      "epoch": 0.41746787090090964,
      "grad_norm": 3.0467493895826383,
      "learning_rate": 9.041517966532205e-07,
      "loss": 0.2727,
      "step": 1612
    },
    {
      "epoch": 0.4177268460069276,
      "grad_norm": 4.234041070121353,
      "learning_rate": 9.03645993592093e-07,
      "loss": 0.3128,
      "step": 1613
    },
    {
      "epoch": 0.4179858211129455,
      "grad_norm": 3.1692736034135915,
      "learning_rate": 9.031399626579544e-07,
      "loss": 0.3455,
      "step": 1614
    },
    {
      "epoch": 0.41824479621896343,
      "grad_norm": 3.723254642276008,
      "learning_rate": 9.026337042646305e-07,
      "loss": 0.4032,
      "step": 1615
    },
    {
      "epoch": 0.41824479621896343,
      "eval_PRM Accuracy": 0.8867924528301887,
      "eval_PRM F1": 0.926829268292683,
      "eval_PRM F1 AUC": 0.8491356731272918,
      "eval_PRM F1 AUC (fixed)": 0.8135149292823468,
      "eval_PRM F1 Neg": 0.75,
      "eval_PRM NPV": 0.72,
      "eval_PRM Precision": 0.9382716049382716,
      "eval_PRM Recall": 0.9156626506024096,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.22190698981285095,
      "eval_runtime": 14.3855,
      "eval_samples_per_second": 3.059,
      "eval_steps_per_second": 0.139,
      "step": 1615
    },
    {
      "epoch": 0.4185037713249814,
      "grad_norm": 5.1337874291904155,
      "learning_rate": 9.021272188261324e-07,
      "loss": 0.3979,
      "step": 1616
    },
    {
      "epoch": 0.4187627464309993,
      "grad_norm": 5.164141850343155,
      "learning_rate": 9.016205067566574e-07,
      "loss": 0.3202,
      "step": 1617
    },
    {
      "epoch": 0.4190217215370173,
      "grad_norm": 4.559613492730163,
      "learning_rate": 9.011135684705881e-07,
      "loss": 0.307,
      "step": 1618
    },
    {
      "epoch": 0.4192806966430352,
      "grad_norm": 4.084012653608465,
      "learning_rate": 9.006064043824915e-07,
      "loss": 0.3692,
      "step": 1619
    },
    {
      "epoch": 0.4195396717490531,
      "grad_norm": 6.417782731569771,
      "learning_rate": 9.000990149071201e-07,
      "loss": 0.3027,
      "step": 1620
    },
    {
      "epoch": 0.4195396717490531,
      "eval_PRM Accuracy": 0.8679245283018868,
      "eval_PRM F1": 0.9135802469135802,
      "eval_PRM F1 AUC": 0.8370874803562075,
      "eval_PRM F1 AUC (fixed)": 0.8082765845992667,
      "eval_PRM F1 Neg": 0.72,
      "eval_PRM NPV": 0.6666666666666666,
      "eval_PRM Precision": 0.9367088607594937,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.22353814542293549,
      "eval_runtime": 14.7809,
      "eval_samples_per_second": 2.977,
      "eval_steps_per_second": 0.135,
      "step": 1620
    },
    {
      "epoch": 0.41979864685507107,
      "grad_norm": 5.188024486858989,
      "learning_rate": 8.995914004594102e-07,
      "loss": 0.2677,
      "step": 1621
    },
    {
      "epoch": 0.420057621961089,
      "grad_norm": 2.997558416900511,
      "learning_rate": 8.990835614544816e-07,
      "loss": 0.3337,
      "step": 1622
    },
    {
      "epoch": 0.42031659706710695,
      "grad_norm": 3.376145696450939,
      "learning_rate": 8.985754983076393e-07,
      "loss": 0.3215,
      "step": 1623
    },
    {
      "epoch": 0.42057557217312486,
      "grad_norm": 4.053299327110271,
      "learning_rate": 8.980672114343702e-07,
      "loss": 0.3135,
      "step": 1624
    },
    {
      "epoch": 0.42083454727914277,
      "grad_norm": 3.3409444896076095,
      "learning_rate": 8.975587012503444e-07,
      "loss": 0.3845,
      "step": 1625
    },
    {
      "epoch": 0.42083454727914277,
      "eval_PRM Accuracy": 0.8679245283018868,
      "eval_PRM F1": 0.9135802469135802,
      "eval_PRM F1 AUC": 0.8370874803562075,
      "eval_PRM F1 AUC (fixed)": 0.8098480880041907,
      "eval_PRM F1 Neg": 0.72,
      "eval_PRM NPV": 0.6666666666666666,
      "eval_PRM Precision": 0.9367088607594937,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.22230681777000427,
      "eval_runtime": 15.0731,
      "eval_samples_per_second": 2.919,
      "eval_steps_per_second": 0.133,
      "step": 1625
    },
    {
      "epoch": 0.42109352238516073,
      "grad_norm": 3.0542917993365886,
      "learning_rate": 8.97049968171415e-07,
      "loss": 0.3275,
      "step": 1626
    },
    {
      "epoch": 0.42135249749117865,
      "grad_norm": 6.9321757463148765,
      "learning_rate": 8.965410126136168e-07,
      "loss": 0.3334,
      "step": 1627
    },
    {
      "epoch": 0.4216114725971966,
      "grad_norm": 3.503968279570744,
      "learning_rate": 8.960318349931678e-07,
      "loss": 0.3474,
      "step": 1628
    },
    {
      "epoch": 0.4218704477032145,
      "grad_norm": 3.703424325494324,
      "learning_rate": 8.95522435726466e-07,
      "loss": 0.2793,
      "step": 1629
    },
    {
      "epoch": 0.42212942280923244,
      "grad_norm": 4.18913896940311,
      "learning_rate": 8.950128152300916e-07,
      "loss": 0.339,
      "step": 1630
    },
    {
      "epoch": 0.42212942280923244,
      "eval_PRM Accuracy": 0.8679245283018868,
      "eval_PRM F1": 0.9135802469135802,
      "eval_PRM F1 AUC": 0.8370874803562075,
      "eval_PRM F1 AUC (fixed)": 0.8051335777894185,
      "eval_PRM F1 Neg": 0.72,
      "eval_PRM NPV": 0.6666666666666666,
      "eval_PRM Precision": 0.9367088607594937,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.2259032130241394,
      "eval_runtime": 14.8898,
      "eval_samples_per_second": 2.955,
      "eval_steps_per_second": 0.134,
      "step": 1630
    },
    {
      "epoch": 0.4223883979152504,
      "grad_norm": 8.73788308445668,
      "learning_rate": 8.945029739208054e-07,
      "loss": 0.3689,
      "step": 1631
    },
    {
      "epoch": 0.4226473730212683,
      "grad_norm": 9.167204914219283,
      "learning_rate": 8.939929122155491e-07,
      "loss": 0.3369,
      "step": 1632
    },
    {
      "epoch": 0.4229063481272863,
      "grad_norm": 6.917728506312492,
      "learning_rate": 8.934826305314442e-07,
      "loss": 0.3584,
      "step": 1633
    },
    {
      "epoch": 0.4231653232333042,
      "grad_norm": 4.450739565929513,
      "learning_rate": 8.929721292857925e-07,
      "loss": 0.336,
      "step": 1634
    },
    {
      "epoch": 0.42342429833932216,
      "grad_norm": 3.285587040095739,
      "learning_rate": 8.924614088960751e-07,
      "loss": 0.3152,
      "step": 1635
    },
    {
      "epoch": 0.42342429833932216,
      "eval_PRM Accuracy": 0.8773584905660378,
      "eval_PRM F1": 0.9192546583850931,
      "eval_PRM F1 AUC": 0.8588266107909901,
      "eval_PRM F1 AUC (fixed)": 0.8038239916186485,
      "eval_PRM F1 Neg": 0.7450980392156863,
      "eval_PRM NPV": 0.6785714285714286,
      "eval_PRM Precision": 0.9487179487179487,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.8260869565217391,
      "eval_loss": 0.23187941312789917,
      "eval_runtime": 14.6607,
      "eval_samples_per_second": 3.001,
      "eval_steps_per_second": 0.136,
      "step": 1635
    },
    {
      "epoch": 0.42368327344534007,
      "grad_norm": 7.060694586928436,
      "learning_rate": 8.919504697799524e-07,
      "loss": 0.3987,
      "step": 1636
    },
    {
      "epoch": 0.423942248551358,
      "grad_norm": 11.109205058450689,
      "learning_rate": 8.914393123552636e-07,
      "loss": 0.3281,
      "step": 1637
    },
    {
      "epoch": 0.42420122365737595,
      "grad_norm": 7.022811322885632,
      "learning_rate": 8.909279370400265e-07,
      "loss": 0.303,
      "step": 1638
    },
    {
      "epoch": 0.42446019876339386,
      "grad_norm": 3.6044168949453224,
      "learning_rate": 8.904163442524372e-07,
      "loss": 0.2626,
      "step": 1639
    },
    {
      "epoch": 0.4247191738694118,
      "grad_norm": 2.1812320431467125,
      "learning_rate": 8.899045344108694e-07,
      "loss": 0.2518,
      "step": 1640
    },
    {
      "epoch": 0.4247191738694118,
      "eval_PRM Accuracy": 0.8679245283018868,
      "eval_PRM F1": 0.9135802469135802,
      "eval_PRM F1 AUC": 0.8370874803562075,
      "eval_PRM F1 AUC (fixed)": 0.8085385018334206,
      "eval_PRM F1 Neg": 0.72,
      "eval_PRM NPV": 0.6666666666666666,
      "eval_PRM Precision": 0.9367088607594937,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.2213054746389389,
      "eval_runtime": 14.4822,
      "eval_samples_per_second": 3.038,
      "eval_steps_per_second": 0.138,
      "step": 1640
    },
    {
      "epoch": 0.42497814897542974,
      "grad_norm": 8.026212733033912,
      "learning_rate": 8.893925079338742e-07,
      "loss": 0.4836,
      "step": 1641
    },
    {
      "epoch": 0.42523712408144765,
      "grad_norm": 3.0719284366698014,
      "learning_rate": 8.888802652401804e-07,
      "loss": 0.3812,
      "step": 1642
    },
    {
      "epoch": 0.4254960991874656,
      "grad_norm": 4.162047005295492,
      "learning_rate": 8.883678067486934e-07,
      "loss": 0.3519,
      "step": 1643
    },
    {
      "epoch": 0.4257550742934835,
      "grad_norm": 4.138604905450256,
      "learning_rate": 8.878551328784947e-07,
      "loss": 0.3056,
      "step": 1644
    },
    {
      "epoch": 0.4260140493995015,
      "grad_norm": 3.1895684773842303,
      "learning_rate": 8.873422440488425e-07,
      "loss": 0.3433,
      "step": 1645
    },
    {
      "epoch": 0.4260140493995015,
      "eval_PRM Accuracy": 0.8773584905660378,
      "eval_PRM F1": 0.9202453987730062,
      "eval_PRM F1 AUC": 0.8431115767417497,
      "eval_PRM F1 AUC (fixed)": 0.8098480880041907,
      "eval_PRM F1 Neg": 0.7346938775510204,
      "eval_PRM NPV": 0.6923076923076923,
      "eval_PRM Precision": 0.9375,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.2170848846435547,
      "eval_runtime": 14.3406,
      "eval_samples_per_second": 3.068,
      "eval_steps_per_second": 0.139,
      "step": 1645
    },
    {
      "epoch": 0.4262730245055194,
      "grad_norm": 10.104451186337883,
      "learning_rate": 8.868291406791701e-07,
      "loss": 0.4841,
      "step": 1646
    },
    {
      "epoch": 0.4265319996115373,
      "grad_norm": 3.0506595381944717,
      "learning_rate": 8.86315823189087e-07,
      "loss": 0.2578,
      "step": 1647
    },
    {
      "epoch": 0.4267909747175553,
      "grad_norm": 4.742630810522798,
      "learning_rate": 8.858022919983773e-07,
      "loss": 0.3805,
      "step": 1648
    },
    {
      "epoch": 0.4270499498235732,
      "grad_norm": 4.7268284590438565,
      "learning_rate": 8.852885475270003e-07,
      "loss": 0.2897,
      "step": 1649
    },
    {
      "epoch": 0.42730892492959116,
      "grad_norm": 3.2082605322636737,
      "learning_rate": 8.84774590195089e-07,
      "loss": 0.3011,
      "step": 1650
    },
    {
      "epoch": 0.42730892492959116,
      "eval_PRM Accuracy": 0.8867924528301887,
      "eval_PRM F1": 0.926829268292683,
      "eval_PRM F1 AUC": 0.8491356731272918,
      "eval_PRM F1 AUC (fixed)": 0.8088004190675746,
      "eval_PRM F1 Neg": 0.75,
      "eval_PRM NPV": 0.72,
      "eval_PRM Precision": 0.9382716049382716,
      "eval_PRM Recall": 0.9156626506024096,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.21894888579845428,
      "eval_runtime": 14.9643,
      "eval_samples_per_second": 2.94,
      "eval_steps_per_second": 0.134,
      "step": 1650
    },
    {
      "epoch": 0.42756790003560907,
      "grad_norm": 3.0707583362104627,
      "learning_rate": 8.84260420422951e-07,
      "loss": 0.3105,
      "step": 1651
    },
    {
      "epoch": 0.42782687514162704,
      "grad_norm": 4.820078394832174,
      "learning_rate": 8.837460386310677e-07,
      "loss": 0.3234,
      "step": 1652
    },
    {
      "epoch": 0.42808585024764495,
      "grad_norm": 5.7500950199688745,
      "learning_rate": 8.832314452400939e-07,
      "loss": 0.3539,
      "step": 1653
    },
    {
      "epoch": 0.42834482535366286,
      "grad_norm": 5.8906413510144136,
      "learning_rate": 8.82716640670857e-07,
      "loss": 0.378,
      "step": 1654
    },
    {
      "epoch": 0.42860380045968083,
      "grad_norm": 5.1040179958597305,
      "learning_rate": 8.822016253443574e-07,
      "loss": 0.3758,
      "step": 1655
    },
    {
      "epoch": 0.42860380045968083,
      "eval_PRM Accuracy": 0.8773584905660378,
      "eval_PRM F1": 0.9202453987730062,
      "eval_PRM F1 AUC": 0.8431115767417497,
      "eval_PRM F1 AUC (fixed)": 0.8074908328968047,
      "eval_PRM F1 Neg": 0.7346938775510204,
      "eval_PRM NPV": 0.6923076923076923,
      "eval_PRM Precision": 0.9375,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.2200443148612976,
      "eval_runtime": 14.6274,
      "eval_samples_per_second": 3.008,
      "eval_steps_per_second": 0.137,
      "step": 1655
    },
    {
      "epoch": 0.42886277556569874,
      "grad_norm": 5.968232234324419,
      "learning_rate": 8.816863996817678e-07,
      "loss": 0.3191,
      "step": 1656
    },
    {
      "epoch": 0.4291217506717167,
      "grad_norm": 3.494625178948273,
      "learning_rate": 8.811709641044332e-07,
      "loss": 0.4133,
      "step": 1657
    },
    {
      "epoch": 0.4293807257777346,
      "grad_norm": 3.8755478263528986,
      "learning_rate": 8.806553190338696e-07,
      "loss": 0.2889,
      "step": 1658
    },
    {
      "epoch": 0.42963970088375253,
      "grad_norm": 3.378532477924372,
      "learning_rate": 8.80139464891765e-07,
      "loss": 0.3859,
      "step": 1659
    },
    {
      "epoch": 0.4298986759897705,
      "grad_norm": 3.6307844771657085,
      "learning_rate": 8.79623402099978e-07,
      "loss": 0.4118,
      "step": 1660
    },
    {
      "epoch": 0.4298986759897705,
      "eval_PRM Accuracy": 0.8679245283018868,
      "eval_PRM F1": 0.9135802469135802,
      "eval_PRM F1 AUC": 0.8370874803562075,
      "eval_PRM F1 AUC (fixed)": 0.8067050811943426,
      "eval_PRM F1 Neg": 0.72,
      "eval_PRM NPV": 0.6666666666666666,
      "eval_PRM Precision": 0.9367088607594937,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.22642911970615387,
      "eval_runtime": 14.3915,
      "eval_samples_per_second": 3.057,
      "eval_steps_per_second": 0.139,
      "step": 1660
    },
    {
      "epoch": 0.4301576510957884,
      "grad_norm": 3.77928032982321,
      "learning_rate": 8.791071310805381e-07,
      "loss": 0.3508,
      "step": 1661
    },
    {
      "epoch": 0.4304166262018064,
      "grad_norm": 7.492823741749537,
      "learning_rate": 8.785906522556447e-07,
      "loss": 0.3253,
      "step": 1662
    },
    {
      "epoch": 0.4306756013078243,
      "grad_norm": 5.060060882559026,
      "learning_rate": 8.780739660476675e-07,
      "loss": 0.3875,
      "step": 1663
    },
    {
      "epoch": 0.4309345764138422,
      "grad_norm": 4.951964181321862,
      "learning_rate": 8.775570728791454e-07,
      "loss": 0.4134,
      "step": 1664
    },
    {
      "epoch": 0.43119355151986016,
      "grad_norm": 3.7967973105566615,
      "learning_rate": 8.770399731727872e-07,
      "loss": 0.4639,
      "step": 1665
    },
    {
      "epoch": 0.43119355151986016,
      "eval_PRM Accuracy": 0.8679245283018868,
      "eval_PRM F1": 0.9135802469135802,
      "eval_PRM F1 AUC": 0.8370874803562075,
      "eval_PRM F1 AUC (fixed)": 0.8022524882137245,
      "eval_PRM F1 Neg": 0.72,
      "eval_PRM NPV": 0.6666666666666666,
      "eval_PRM Precision": 0.9367088607594937,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.23072762787342072,
      "eval_runtime": 14.7853,
      "eval_samples_per_second": 2.976,
      "eval_steps_per_second": 0.135,
      "step": 1665
    },
    {
      "epoch": 0.4314525266258781,
      "grad_norm": 3.3413203190585037,
      "learning_rate": 8.7652266735147e-07,
      "loss": 0.3265,
      "step": 1666
    },
    {
      "epoch": 0.43171150173189604,
      "grad_norm": 3.2590127051152864,
      "learning_rate": 8.760051558382396e-07,
      "loss": 0.3307,
      "step": 1667
    },
    {
      "epoch": 0.43197047683791395,
      "grad_norm": 3.015655163823686,
      "learning_rate": 8.754874390563101e-07,
      "loss": 0.2906,
      "step": 1668
    },
    {
      "epoch": 0.43222945194393186,
      "grad_norm": 3.75276624789197,
      "learning_rate": 8.749695174290635e-07,
      "loss": 0.3691,
      "step": 1669
    },
    {
      "epoch": 0.43248842704994983,
      "grad_norm": 2.8441005048542323,
      "learning_rate": 8.744513913800493e-07,
      "loss": 0.299,
      "step": 1670
    },
    {
      "epoch": 0.43248842704994983,
      "eval_PRM Accuracy": 0.8679245283018868,
      "eval_PRM F1": 0.9135802469135802,
      "eval_PRM F1 AUC": 0.8370874803562075,
      "eval_PRM F1 AUC (fixed)": 0.8006809848088005,
      "eval_PRM F1 Neg": 0.72,
      "eval_PRM NPV": 0.6666666666666666,
      "eval_PRM Precision": 0.9367088607594937,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.2328794002532959,
      "eval_runtime": 14.5475,
      "eval_samples_per_second": 3.025,
      "eval_steps_per_second": 0.137,
      "step": 1670
    },
    {
      "epoch": 0.43274740215596774,
      "grad_norm": 5.691176253313527,
      "learning_rate": 8.739330613329839e-07,
      "loss": 0.3112,
      "step": 1671
    },
    {
      "epoch": 0.4330063772619857,
      "grad_norm": 4.073152889158902,
      "learning_rate": 8.73414527711751e-07,
      "loss": 0.4038,
      "step": 1672
    },
    {
      "epoch": 0.4332653523680036,
      "grad_norm": 3.6181373476890712,
      "learning_rate": 8.728957909404006e-07,
      "loss": 0.2938,
      "step": 1673
    },
    {
      "epoch": 0.4335243274740216,
      "grad_norm": 5.103103893659234,
      "learning_rate": 8.723768514431487e-07,
      "loss": 0.4224,
      "step": 1674
    },
    {
      "epoch": 0.4337833025800395,
      "grad_norm": 12.451008210969361,
      "learning_rate": 8.718577096443769e-07,
      "loss": 0.4068,
      "step": 1675
    },
    {
      "epoch": 0.4337833025800395,
      "eval_PRM Accuracy": 0.8679245283018868,
      "eval_PRM F1": 0.9135802469135802,
      "eval_PRM F1 AUC": 0.8370874803562075,
      "eval_PRM F1 AUC (fixed)": 0.7959664745940283,
      "eval_PRM F1 Neg": 0.72,
      "eval_PRM NPV": 0.6666666666666666,
      "eval_PRM Precision": 0.9367088607594937,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.23811227083206177,
      "eval_runtime": 14.6527,
      "eval_samples_per_second": 3.003,
      "eval_steps_per_second": 0.136,
      "step": 1675
    },
    {
      "epoch": 0.4340422776860574,
      "grad_norm": 3.6393883954127024,
      "learning_rate": 8.713383659686329e-07,
      "loss": 0.366,
      "step": 1676
    },
    {
      "epoch": 0.4343012527920754,
      "grad_norm": 3.7561613045552287,
      "learning_rate": 8.708188208406287e-07,
      "loss": 0.3487,
      "step": 1677
    },
    {
      "epoch": 0.4345602278980933,
      "grad_norm": 4.112559032256317,
      "learning_rate": 8.702990746852419e-07,
      "loss": 0.3483,
      "step": 1678
    },
    {
      "epoch": 0.43481920300411125,
      "grad_norm": 6.360240374228111,
      "learning_rate": 8.697791279275138e-07,
      "loss": 0.367,
      "step": 1679
    },
    {
      "epoch": 0.43507817811012917,
      "grad_norm": 3.935694768901393,
      "learning_rate": 8.692589809926498e-07,
      "loss": 0.3179,
      "step": 1680
    },
    {
      "epoch": 0.43507817811012917,
      "eval_PRM Accuracy": 0.8679245283018868,
      "eval_PRM F1": 0.9135802469135802,
      "eval_PRM F1 AUC": 0.8370874803562075,
      "eval_PRM F1 AUC (fixed)": 0.7972760607647984,
      "eval_PRM F1 Neg": 0.72,
      "eval_PRM NPV": 0.6666666666666666,
      "eval_PRM Precision": 0.9367088607594937,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.23968106508255005,
      "eval_runtime": 14.3665,
      "eval_samples_per_second": 3.063,
      "eval_steps_per_second": 0.139,
      "step": 1680
    },
    {
      "epoch": 0.4353371532161471,
      "grad_norm": 3.5153240118439637,
      "learning_rate": 8.687386343060193e-07,
      "loss": 0.3567,
      "step": 1681
    },
    {
      "epoch": 0.43559612832216504,
      "grad_norm": 3.4525502369359056,
      "learning_rate": 8.682180882931552e-07,
      "loss": 0.2657,
      "step": 1682
    },
    {
      "epoch": 0.43585510342818296,
      "grad_norm": 2.8841332560401325,
      "learning_rate": 8.67697343379753e-07,
      "loss": 0.3499,
      "step": 1683
    },
    {
      "epoch": 0.4361140785342009,
      "grad_norm": 3.9017587222557473,
      "learning_rate": 8.671763999916708e-07,
      "loss": 0.3808,
      "step": 1684
    },
    {
      "epoch": 0.43637305364021883,
      "grad_norm": 3.3984910226128857,
      "learning_rate": 8.666552585549295e-07,
      "loss": 0.3615,
      "step": 1685
    },
    {
      "epoch": 0.43637305364021883,
      "eval_PRM Accuracy": 0.8584905660377359,
      "eval_PRM F1": 0.9079754601226994,
      "eval_PRM F1 AUC": 0.8153483499214248,
      "eval_PRM F1 AUC (fixed)": 0.7970141435306444,
      "eval_PRM F1 Neg": 0.6938775510204082,
      "eval_PRM NPV": 0.6538461538461539,
      "eval_PRM Precision": 0.925,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.24045367538928986,
      "eval_runtime": 14.7529,
      "eval_samples_per_second": 2.982,
      "eval_steps_per_second": 0.136,
      "step": 1685
    },
    {
      "epoch": 0.43663202874623674,
      "grad_norm": 3.3014876553458503,
      "learning_rate": 8.661339194957115e-07,
      "loss": 0.4039,
      "step": 1686
    },
    {
      "epoch": 0.4368910038522547,
      "grad_norm": 4.494385054391037,
      "learning_rate": 8.65612383240361e-07,
      "loss": 0.3515,
      "step": 1687
    },
    {
      "epoch": 0.4371499789582726,
      "grad_norm": 2.6955125987179436,
      "learning_rate": 8.650906502153836e-07,
      "loss": 0.2649,
      "step": 1688
    },
    {
      "epoch": 0.4374089540642906,
      "grad_norm": 6.297182129219428,
      "learning_rate": 8.645687208474455e-07,
      "loss": 0.2861,
      "step": 1689
    },
    {
      "epoch": 0.4376679291703085,
      "grad_norm": 5.525091990734904,
      "learning_rate": 8.640465955633739e-07,
      "loss": 0.3078,
      "step": 1690
    },
    {
      "epoch": 0.4376679291703085,
      "eval_PRM Accuracy": 0.8679245283018868,
      "eval_PRM F1": 0.9146341463414634,
      "eval_PRM F1 AUC": 0.8213724463069669,
      "eval_PRM F1 AUC (fixed)": 0.7980618124672604,
      "eval_PRM F1 Neg": 0.7083333333333334,
      "eval_PRM NPV": 0.68,
      "eval_PRM Precision": 0.9259259259259259,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.23908013105392456,
      "eval_runtime": 14.7326,
      "eval_samples_per_second": 2.987,
      "eval_steps_per_second": 0.136,
      "step": 1690
    },
    {
      "epoch": 0.4379269042763264,
      "grad_norm": 6.634706936547658,
      "learning_rate": 8.635242747901557e-07,
      "loss": 0.347,
      "step": 1691
    },
    {
      "epoch": 0.4381858793823444,
      "grad_norm": 4.002366725497352,
      "learning_rate": 8.630017589549379e-07,
      "loss": 0.3412,
      "step": 1692
    },
    {
      "epoch": 0.4384448544883623,
      "grad_norm": 4.134430196470716,
      "learning_rate": 8.624790484850272e-07,
      "loss": 0.3839,
      "step": 1693
    },
    {
      "epoch": 0.43870382959438026,
      "grad_norm": 3.0378928162284686,
      "learning_rate": 8.619561438078893e-07,
      "loss": 0.3878,
      "step": 1694
    },
    {
      "epoch": 0.43896280470039817,
      "grad_norm": 3.288723716176208,
      "learning_rate": 8.614330453511488e-07,
      "loss": 0.3348,
      "step": 1695
    },
    {
      "epoch": 0.43896280470039817,
      "eval_PRM Accuracy": 0.8584905660377359,
      "eval_PRM F1": 0.9079754601226994,
      "eval_PRM F1 AUC": 0.8153483499214248,
      "eval_PRM F1 AUC (fixed)": 0.7975379779989523,
      "eval_PRM F1 Neg": 0.6938775510204082,
      "eval_PRM NPV": 0.6538461538461539,
      "eval_PRM Precision": 0.925,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.23444853723049164,
      "eval_runtime": 14.8246,
      "eval_samples_per_second": 2.968,
      "eval_steps_per_second": 0.135,
      "step": 1695
    },
    {
      "epoch": 0.43922177980641613,
      "grad_norm": 8.049448219216604,
      "learning_rate": 8.609097535425883e-07,
      "loss": 0.3887,
      "step": 1696
    },
    {
      "epoch": 0.43948075491243405,
      "grad_norm": 3.3624078209970314,
      "learning_rate": 8.603862688101495e-07,
      "loss": 0.3259,
      "step": 1697
    },
    {
      "epoch": 0.43973973001845196,
      "grad_norm": 3.3402730111687,
      "learning_rate": 8.59862591581931e-07,
      "loss": 0.3547,
      "step": 1698
    },
    {
      "epoch": 0.4399987051244699,
      "grad_norm": 5.347674530617662,
      "learning_rate": 8.593387222861892e-07,
      "loss": 0.3592,
      "step": 1699
    },
    {
      "epoch": 0.44025768023048784,
      "grad_norm": 2.4360697552111823,
      "learning_rate": 8.588146613513372e-07,
      "loss": 0.2552,
      "step": 1700
    },
    {
      "epoch": 0.44025768023048784,
      "eval_PRM Accuracy": 0.8679245283018868,
      "eval_PRM F1": 0.9135802469135802,
      "eval_PRM F1 AUC": 0.8370874803562075,
      "eval_PRM F1 AUC (fixed)": 0.7985856469355683,
      "eval_PRM F1 Neg": 0.72,
      "eval_PRM NPV": 0.6666666666666666,
      "eval_PRM Precision": 0.9367088607594937,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.23092550039291382,
      "eval_runtime": 14.6344,
      "eval_samples_per_second": 3.007,
      "eval_steps_per_second": 0.137,
      "step": 1700
    },
    {
      "epoch": 0.4405166553365058,
      "grad_norm": 4.306812871104216,
      "learning_rate": 8.582904092059455e-07,
      "loss": 0.3437,
      "step": 1701
    },
    {
      "epoch": 0.4407756304425237,
      "grad_norm": 3.1278230337251665,
      "learning_rate": 8.577659662787402e-07,
      "loss": 0.3315,
      "step": 1702
    },
    {
      "epoch": 0.4410346055485416,
      "grad_norm": 3.9733312249032386,
      "learning_rate": 8.572413329986043e-07,
      "loss": 0.3765,
      "step": 1703
    },
    {
      "epoch": 0.4412935806545596,
      "grad_norm": 5.1439565481682745,
      "learning_rate": 8.567165097945755e-07,
      "loss": 0.3978,
      "step": 1704
    },
    {
      "epoch": 0.4415525557605775,
      "grad_norm": 3.1744327350930823,
      "learning_rate": 8.561914970958474e-07,
      "loss": 0.297,
      "step": 1705
    },
    {
      "epoch": 0.4415525557605775,
      "eval_PRM Accuracy": 0.8867924528301887,
      "eval_PRM F1": 0.926829268292683,
      "eval_PRM F1 AUC": 0.8491356731272918,
      "eval_PRM F1 AUC (fixed)": 0.7988475641697224,
      "eval_PRM F1 Neg": 0.75,
      "eval_PRM NPV": 0.72,
      "eval_PRM Precision": 0.9382716049382716,
      "eval_PRM Recall": 0.9156626506024096,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.2263193577528,
      "eval_runtime": 14.5952,
      "eval_samples_per_second": 3.015,
      "eval_steps_per_second": 0.137,
      "step": 1705
    },
    {
      "epoch": 0.44181153086659547,
      "grad_norm": 3.980867274635927,
      "learning_rate": 8.556662953317685e-07,
      "loss": 0.3984,
      "step": 1706
    },
    {
      "epoch": 0.4420705059726134,
      "grad_norm": 5.433404583374516,
      "learning_rate": 8.551409049318421e-07,
      "loss": 0.4307,
      "step": 1707
    },
    {
      "epoch": 0.4423294810786313,
      "grad_norm": 4.083500146112218,
      "learning_rate": 8.546153263257251e-07,
      "loss": 0.2691,
      "step": 1708
    },
    {
      "epoch": 0.44258845618464926,
      "grad_norm": 4.177337174850718,
      "learning_rate": 8.540895599432289e-07,
      "loss": 0.2824,
      "step": 1709
    },
    {
      "epoch": 0.44284743129066717,
      "grad_norm": 3.819657658493383,
      "learning_rate": 8.535636062143183e-07,
      "loss": 0.3729,
      "step": 1710
    },
    {
      "epoch": 0.44284743129066717,
      "eval_PRM Accuracy": 0.8679245283018868,
      "eval_PRM F1": 0.9135802469135802,
      "eval_PRM F1 AUC": 0.8370874803562075,
      "eval_PRM F1 AUC (fixed)": 0.8022524882137244,
      "eval_PRM F1 Neg": 0.72,
      "eval_PRM NPV": 0.6666666666666666,
      "eval_PRM Precision": 0.9367088607594937,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.22645670175552368,
      "eval_runtime": 14.5655,
      "eval_samples_per_second": 3.021,
      "eval_steps_per_second": 0.137,
      "step": 1710
    },
    {
      "epoch": 0.44310640639668514,
      "grad_norm": 6.149136739751437,
      "learning_rate": 8.530374655691114e-07,
      "loss": 0.2731,
      "step": 1711
    },
    {
      "epoch": 0.44336538150270305,
      "grad_norm": 2.9699105747130083,
      "learning_rate": 8.525111384378788e-07,
      "loss": 0.2906,
      "step": 1712
    },
    {
      "epoch": 0.44362435660872096,
      "grad_norm": 6.085554456562368,
      "learning_rate": 8.519846252510442e-07,
      "loss": 0.2938,
      "step": 1713
    },
    {
      "epoch": 0.4438833317147389,
      "grad_norm": 3.9239327771729,
      "learning_rate": 8.51457926439183e-07,
      "loss": 0.3472,
      "step": 1714
    },
    {
      "epoch": 0.44414230682075684,
      "grad_norm": 5.16591458430438,
      "learning_rate": 8.509310424330225e-07,
      "loss": 0.3619,
      "step": 1715
    },
    {
      "epoch": 0.44414230682075684,
      "eval_PRM Accuracy": 0.8679245283018868,
      "eval_PRM F1": 0.9135802469135802,
      "eval_PRM F1 AUC": 0.8370874803562075,
      "eval_PRM F1 AUC (fixed)": 0.8025144054478784,
      "eval_PRM F1 Neg": 0.72,
      "eval_PRM NPV": 0.6666666666666666,
      "eval_PRM Precision": 0.9367088607594937,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.22760477662086487,
      "eval_runtime": 14.6096,
      "eval_samples_per_second": 3.012,
      "eval_steps_per_second": 0.137,
      "step": 1715
    },
    {
      "epoch": 0.4444012819267748,
      "grad_norm": 3.2973839717958078,
      "learning_rate": 8.504039736634415e-07,
      "loss": 0.3162,
      "step": 1716
    },
    {
      "epoch": 0.4446602570327927,
      "grad_norm": 6.1341991140194825,
      "learning_rate": 8.498767205614698e-07,
      "loss": 0.2534,
      "step": 1717
    },
    {
      "epoch": 0.4449192321388107,
      "grad_norm": 4.229791341567047,
      "learning_rate": 8.493492835582881e-07,
      "loss": 0.3914,
      "step": 1718
    },
    {
      "epoch": 0.4451782072448286,
      "grad_norm": 3.127952731147352,
      "learning_rate": 8.488216630852276e-07,
      "loss": 0.3128,
      "step": 1719
    },
    {
      "epoch": 0.4454371823508465,
      "grad_norm": 5.744906103307722,
      "learning_rate": 8.482938595737688e-07,
      "loss": 0.3569,
      "step": 1720
    },
    {
      "epoch": 0.4454371823508465,
      "eval_PRM Accuracy": 0.8679245283018868,
      "eval_PRM F1": 0.9135802469135802,
      "eval_PRM F1 AUC": 0.8370874803562075,
      "eval_PRM F1 AUC (fixed)": 0.8030382399161864,
      "eval_PRM F1 Neg": 0.72,
      "eval_PRM NPV": 0.6666666666666666,
      "eval_PRM Precision": 0.9367088607594937,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.22972126305103302,
      "eval_runtime": 14.4328,
      "eval_samples_per_second": 3.049,
      "eval_steps_per_second": 0.139,
      "step": 1720
    },
    {
      "epoch": 0.4456961574568645,
      "grad_norm": 3.221749661822224,
      "learning_rate": 8.47765873455543e-07,
      "loss": 0.279,
      "step": 1721
    },
    {
      "epoch": 0.4459551325628824,
      "grad_norm": 4.723289273805276,
      "learning_rate": 8.472377051623299e-07,
      "loss": 0.3384,
      "step": 1722
    },
    {
      "epoch": 0.44621410766890035,
      "grad_norm": 5.190704872161203,
      "learning_rate": 8.46709355126059e-07,
      "loss": 0.3831,
      "step": 1723
    },
    {
      "epoch": 0.44647308277491826,
      "grad_norm": 3.700894177677377,
      "learning_rate": 8.461808237788073e-07,
      "loss": 0.3755,
      "step": 1724
    },
    {
      "epoch": 0.4467320578809362,
      "grad_norm": 5.361390881022496,
      "learning_rate": 8.456521115528013e-07,
      "loss": 0.2964,
      "step": 1725
    },
    {
      "epoch": 0.4467320578809362,
      "eval_PRM Accuracy": 0.8679245283018868,
      "eval_PRM F1": 0.9135802469135802,
      "eval_PRM F1 AUC": 0.8370874803562075,
      "eval_PRM F1 AUC (fixed)": 0.8061812467260345,
      "eval_PRM F1 Neg": 0.72,
      "eval_PRM NPV": 0.6666666666666666,
      "eval_PRM Precision": 0.9367088607594937,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.23026610910892487,
      "eval_runtime": 14.3767,
      "eval_samples_per_second": 3.061,
      "eval_steps_per_second": 0.139,
      "step": 1725
    },
    {
      "epoch": 0.44699103298695414,
      "grad_norm": 3.7260651984985,
      "learning_rate": 8.451232188804146e-07,
      "loss": 0.2309,
      "step": 1726
    },
    {
      "epoch": 0.44725000809297205,
      "grad_norm": 9.18710585126297,
      "learning_rate": 8.445941461941689e-07,
      "loss": 0.4108,
      "step": 1727
    },
    {
      "epoch": 0.44750898319899,
      "grad_norm": 4.3562980575068515,
      "learning_rate": 8.440648939267327e-07,
      "loss": 0.4047,
      "step": 1728
    },
    {
      "epoch": 0.44776795830500793,
      "grad_norm": 4.631338491954459,
      "learning_rate": 8.435354625109213e-07,
      "loss": 0.3096,
      "step": 1729
    },
    {
      "epoch": 0.44802693341102584,
      "grad_norm": 5.199946198691332,
      "learning_rate": 8.43005852379697e-07,
      "loss": 0.3592,
      "step": 1730
    },
    {
      "epoch": 0.44802693341102584,
      "eval_PRM Accuracy": 0.8584905660377359,
      "eval_PRM F1": 0.9079754601226994,
      "eval_PRM F1 AUC": 0.8153483499214248,
      "eval_PRM F1 AUC (fixed)": 0.8027763226820324,
      "eval_PRM F1 Neg": 0.6938775510204082,
      "eval_PRM NPV": 0.6538461538461539,
      "eval_PRM Precision": 0.925,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.23249457776546478,
      "eval_runtime": 14.4784,
      "eval_samples_per_second": 3.039,
      "eval_steps_per_second": 0.138,
      "step": 1730
    },
    {
      "epoch": 0.4482859085170438,
      "grad_norm": 4.252128206129468,
      "learning_rate": 8.424760639661679e-07,
      "loss": 0.2651,
      "step": 1731
    },
    {
      "epoch": 0.4485448836230617,
      "grad_norm": 2.9962798643102855,
      "learning_rate": 8.419460977035881e-07,
      "loss": 0.2554,
      "step": 1732
    },
    {
      "epoch": 0.4488038587290797,
      "grad_norm": 4.844131624007952,
      "learning_rate": 8.414159540253566e-07,
      "loss": 0.4041,
      "step": 1733
    },
    {
      "epoch": 0.4490628338350976,
      "grad_norm": 3.6798997934760007,
      "learning_rate": 8.408856333650182e-07,
      "loss": 0.3329,
      "step": 1734
    },
    {
      "epoch": 0.4493218089411155,
      "grad_norm": 6.515688754440896,
      "learning_rate": 8.40355136156262e-07,
      "loss": 0.3116,
      "step": 1735
    },
    {
      "epoch": 0.4493218089411155,
      "eval_PRM Accuracy": 0.8490566037735849,
      "eval_PRM F1": 0.9,
      "eval_PRM F1 AUC": 0.8250392875851232,
      "eval_PRM F1 AUC (fixed)": 0.7996333158721844,
      "eval_PRM F1 Neg": 0.6923076923076923,
      "eval_PRM NPV": 0.6206896551724138,
      "eval_PRM Precision": 0.935064935064935,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.23652400076389313,
      "eval_runtime": 14.8974,
      "eval_samples_per_second": 2.954,
      "eval_steps_per_second": 0.134,
      "step": 1735
    },
    {
      "epoch": 0.4495807840471335,
      "grad_norm": 3.9581638756947073,
      "learning_rate": 8.398244628329219e-07,
      "loss": 0.366,
      "step": 1736
    },
    {
      "epoch": 0.4498397591531514,
      "grad_norm": 3.620290892586666,
      "learning_rate": 8.39293613828975e-07,
      "loss": 0.3239,
      "step": 1737
    },
    {
      "epoch": 0.45009873425916935,
      "grad_norm": 3.9645238037365997,
      "learning_rate": 8.387625895785431e-07,
      "loss": 0.2711,
      "step": 1738
    },
    {
      "epoch": 0.45035770936518726,
      "grad_norm": 4.130706791680596,
      "learning_rate": 8.382313905158905e-07,
      "loss": 0.2967,
      "step": 1739
    },
    {
      "epoch": 0.45061668447120523,
      "grad_norm": 3.0491448534210375,
      "learning_rate": 8.377000170754249e-07,
      "loss": 0.3032,
      "step": 1740
    },
    {
      "epoch": 0.45061668447120523,
      "eval_PRM Accuracy": 0.8490566037735849,
      "eval_PRM F1": 0.9,
      "eval_PRM F1 AUC": 0.8250392875851232,
      "eval_PRM F1 AUC (fixed)": 0.8022524882137245,
      "eval_PRM F1 Neg": 0.6923076923076923,
      "eval_PRM NPV": 0.6206896551724138,
      "eval_PRM Precision": 0.935064935064935,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.23463712632656097,
      "eval_runtime": 14.3925,
      "eval_samples_per_second": 3.057,
      "eval_steps_per_second": 0.139,
      "step": 1740
    },
    {
      "epoch": 0.45087565957722314,
      "grad_norm": 5.554883859083742,
      "learning_rate": 8.371684696916965e-07,
      "loss": 0.4688,
      "step": 1741
    },
    {
      "epoch": 0.45113463468324105,
      "grad_norm": 5.151171640193862,
      "learning_rate": 8.366367487993978e-07,
      "loss": 0.3429,
      "step": 1742
    },
    {
      "epoch": 0.451393609789259,
      "grad_norm": 12.139519484164033,
      "learning_rate": 8.361048548333627e-07,
      "loss": 0.3423,
      "step": 1743
    },
    {
      "epoch": 0.45165258489527693,
      "grad_norm": 5.7588441409715845,
      "learning_rate": 8.355727882285676e-07,
      "loss": 0.3353,
      "step": 1744
    },
    {
      "epoch": 0.4519115600012949,
      "grad_norm": 4.245876555385781,
      "learning_rate": 8.350405494201293e-07,
      "loss": 0.3723,
      "step": 1745
    },
    {
      "epoch": 0.4519115600012949,
      "eval_PRM Accuracy": 0.8490566037735849,
      "eval_PRM F1": 0.9,
      "eval_PRM F1 AUC": 0.8250392875851232,
      "eval_PRM F1 AUC (fixed)": 0.8001571503404925,
      "eval_PRM F1 Neg": 0.6923076923076923,
      "eval_PRM NPV": 0.6206896551724138,
      "eval_PRM Precision": 0.935064935064935,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.23753543198108673,
      "eval_runtime": 14.6106,
      "eval_samples_per_second": 3.012,
      "eval_steps_per_second": 0.137,
      "step": 1745
    },
    {
      "epoch": 0.4521705351073128,
      "grad_norm": 5.105790492903071,
      "learning_rate": 8.345081388433056e-07,
      "loss": 0.3382,
      "step": 1746
    },
    {
      "epoch": 0.4524295102133307,
      "grad_norm": 4.329127432255067,
      "learning_rate": 8.339755569334948e-07,
      "loss": 0.3678,
      "step": 1747
    },
    {
      "epoch": 0.4526884853193487,
      "grad_norm": 2.4926356319218477,
      "learning_rate": 8.334428041262357e-07,
      "loss": 0.2392,
      "step": 1748
    },
    {
      "epoch": 0.4529474604253666,
      "grad_norm": 6.587054051496237,
      "learning_rate": 8.329098808572059e-07,
      "loss": 0.3374,
      "step": 1749
    },
    {
      "epoch": 0.45320643553138457,
      "grad_norm": 4.297210622041225,
      "learning_rate": 8.323767875622236e-07,
      "loss": 0.3333,
      "step": 1750
    },
    {
      "epoch": 0.45320643553138457,
      "eval_PRM Accuracy": 0.8584905660377359,
      "eval_PRM F1": 0.906832298136646,
      "eval_PRM F1 AUC": 0.8310633839706653,
      "eval_PRM F1 AUC (fixed)": 0.7964903090623363,
      "eval_PRM F1 Neg": 0.7058823529411765,
      "eval_PRM NPV": 0.6428571428571429,
      "eval_PRM Precision": 0.9358974358974359,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.23380161821842194,
      "eval_runtime": 15.005,
      "eval_samples_per_second": 2.932,
      "eval_steps_per_second": 0.133,
      "step": 1750
    },
    {
      "epoch": 0.4534654106374025,
      "grad_norm": 4.692593441858384,
      "learning_rate": 8.318435246772451e-07,
      "loss": 0.3315,
      "step": 1751
    },
    {
      "epoch": 0.4537243857434204,
      "grad_norm": 6.329874341355685,
      "learning_rate": 8.313100926383659e-07,
      "loss": 0.3813,
      "step": 1752
    },
    {
      "epoch": 0.45398336084943836,
      "grad_norm": 4.526285043124399,
      "learning_rate": 8.307764918818198e-07,
      "loss": 0.2917,
      "step": 1753
    },
    {
      "epoch": 0.45424233595545627,
      "grad_norm": 3.2859389770667335,
      "learning_rate": 8.302427228439779e-07,
      "loss": 0.3351,
      "step": 1754
    },
    {
      "epoch": 0.45450131106147423,
      "grad_norm": 2.7047710293212797,
      "learning_rate": 8.2970878596135e-07,
      "loss": 0.2893,
      "step": 1755
    },
    {
      "epoch": 0.45450131106147423,
      "eval_PRM Accuracy": 0.8773584905660378,
      "eval_PRM F1": 0.9202453987730062,
      "eval_PRM F1 AUC": 0.8431115767417497,
      "eval_PRM F1 AUC (fixed)": 0.7977998952331063,
      "eval_PRM F1 Neg": 0.7346938775510204,
      "eval_PRM NPV": 0.6923076923076923,
      "eval_PRM Precision": 0.9375,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.22954775393009186,
      "eval_runtime": 14.4544,
      "eval_samples_per_second": 3.044,
      "eval_steps_per_second": 0.138,
      "step": 1755
    },
    {
      "epoch": 0.45476028616749214,
      "grad_norm": 6.079841409527296,
      "learning_rate": 8.291746816705823e-07,
      "loss": 0.3106,
      "step": 1756
    },
    {
      "epoch": 0.45501926127351006,
      "grad_norm": 3.5916725818463653,
      "learning_rate": 8.286404104084587e-07,
      "loss": 0.3382,
      "step": 1757
    },
    {
      "epoch": 0.455278236379528,
      "grad_norm": 4.325582783032666,
      "learning_rate": 8.281059726118986e-07,
      "loss": 0.3001,
      "step": 1758
    },
    {
      "epoch": 0.45553721148554593,
      "grad_norm": 3.131271808003115,
      "learning_rate": 8.275713687179585e-07,
      "loss": 0.3106,
      "step": 1759
    },
    {
      "epoch": 0.4557961865915639,
      "grad_norm": 2.965484663724944,
      "learning_rate": 8.270365991638303e-07,
      "loss": 0.3372,
      "step": 1760
    },
    {
      "epoch": 0.4557961865915639,
      "eval_PRM Accuracy": 0.8584905660377359,
      "eval_PRM F1": 0.906832298136646,
      "eval_PRM F1 AUC": 0.8310633839706653,
      "eval_PRM F1 AUC (fixed)": 0.7951807228915664,
      "eval_PRM F1 Neg": 0.7058823529411765,
      "eval_PRM NPV": 0.6428571428571429,
      "eval_PRM Precision": 0.9358974358974359,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.23178471624851227,
      "eval_runtime": 14.8561,
      "eval_samples_per_second": 2.962,
      "eval_steps_per_second": 0.135,
      "step": 1760
    },
    {
      "epoch": 0.4560551616975818,
      "grad_norm": 3.7398543631485195,
      "learning_rate": 8.265016643868416e-07,
      "loss": 0.3386,
      "step": 1761
    },
    {
      "epoch": 0.4563141368035998,
      "grad_norm": 2.87885645422035,
      "learning_rate": 8.259665648244548e-07,
      "loss": 0.3598,
      "step": 1762
    },
    {
      "epoch": 0.4565731119096177,
      "grad_norm": 3.71470035190551,
      "learning_rate": 8.254313009142674e-07,
      "loss": 0.3184,
      "step": 1763
    },
    {
      "epoch": 0.4568320870156356,
      "grad_norm": 5.690174138505368,
      "learning_rate": 8.24895873094011e-07,
      "loss": 0.3148,
      "step": 1764
    },
    {
      "epoch": 0.45709106212165357,
      "grad_norm": 3.5570654965175654,
      "learning_rate": 8.243602818015515e-07,
      "loss": 0.2779,
      "step": 1765
    },
    {
      "epoch": 0.45709106212165357,
      "eval_PRM Accuracy": 0.8679245283018868,
      "eval_PRM F1": 0.9135802469135802,
      "eval_PRM F1 AUC": 0.8370874803562075,
      "eval_PRM F1 AUC (fixed)": 0.7920377160817181,
      "eval_PRM F1 Neg": 0.72,
      "eval_PRM NPV": 0.6666666666666666,
      "eval_PRM Precision": 0.9367088607594937,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.23534613847732544,
      "eval_runtime": 14.4916,
      "eval_samples_per_second": 3.036,
      "eval_steps_per_second": 0.138,
      "step": 1765
    },
    {
      "epoch": 0.4573500372276715,
      "grad_norm": 4.20905549559305,
      "learning_rate": 8.238245274748885e-07,
      "loss": 0.3097,
      "step": 1766
    },
    {
      "epoch": 0.45760901233368945,
      "grad_norm": 3.571089393427515,
      "learning_rate": 8.232886105521545e-07,
      "loss": 0.3772,
      "step": 1767
    },
    {
      "epoch": 0.45786798743970736,
      "grad_norm": 3.319074236687981,
      "learning_rate": 8.227525314716156e-07,
      "loss": 0.3209,
      "step": 1768
    },
    {
      "epoch": 0.45812696254572527,
      "grad_norm": 8.058991891459208,
      "learning_rate": 8.222162906716699e-07,
      "loss": 0.32,
      "step": 1769
    },
    {
      "epoch": 0.45838593765174324,
      "grad_norm": 9.635358259079585,
      "learning_rate": 8.216798885908483e-07,
      "loss": 0.4383,
      "step": 1770
    },
    {
      "epoch": 0.45838593765174324,
      "eval_PRM Accuracy": 0.8773584905660378,
      "eval_PRM F1": 0.9202453987730062,
      "eval_PRM F1 AUC": 0.8431115767417497,
      "eval_PRM F1 AUC (fixed)": 0.8022524882137245,
      "eval_PRM F1 Neg": 0.7346938775510204,
      "eval_PRM NPV": 0.6923076923076923,
      "eval_PRM Precision": 0.9375,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.2349451631307602,
      "eval_runtime": 14.3021,
      "eval_samples_per_second": 3.076,
      "eval_steps_per_second": 0.14,
      "step": 1770
    },
    {
      "epoch": 0.45864491275776115,
      "grad_norm": 5.322744420195617,
      "learning_rate": 8.211433256678131e-07,
      "loss": 0.3096,
      "step": 1771
    },
    {
      "epoch": 0.4589038878637791,
      "grad_norm": 5.765628097437391,
      "learning_rate": 8.206066023413585e-07,
      "loss": 0.2796,
      "step": 1772
    },
    {
      "epoch": 0.459162862969797,
      "grad_norm": 2.5397152562840755,
      "learning_rate": 8.200697190504096e-07,
      "loss": 0.2516,
      "step": 1773
    },
    {
      "epoch": 0.45942183807581494,
      "grad_norm": 4.517139406857374,
      "learning_rate": 8.195326762340228e-07,
      "loss": 0.2757,
      "step": 1774
    },
    {
      "epoch": 0.4596808131818329,
      "grad_norm": 2.8001331753915717,
      "learning_rate": 8.189954743313843e-07,
      "loss": 0.2975,
      "step": 1775
    },
    {
      "epoch": 0.4596808131818329,
      "eval_PRM Accuracy": 0.8773584905660378,
      "eval_PRM F1": 0.9202453987730062,
      "eval_PRM F1 AUC": 0.8431115767417497,
      "eval_PRM F1 AUC (fixed)": 0.8095861707700367,
      "eval_PRM F1 Neg": 0.7346938775510204,
      "eval_PRM NPV": 0.6923076923076923,
      "eval_PRM Precision": 0.9375,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.23248286545276642,
      "eval_runtime": 14.9726,
      "eval_samples_per_second": 2.939,
      "eval_steps_per_second": 0.134,
      "step": 1775
    },
    {
      "epoch": 0.4599397882878508,
      "grad_norm": 3.7805513339775425,
      "learning_rate": 8.184581137818108e-07,
      "loss": 0.3274,
      "step": 1776
    },
    {
      "epoch": 0.4601987633938688,
      "grad_norm": 3.8850221432331615,
      "learning_rate": 8.179205950247487e-07,
      "loss": 0.3689,
      "step": 1777
    },
    {
      "epoch": 0.4604577384998867,
      "grad_norm": 3.0121384408551557,
      "learning_rate": 8.173829184997737e-07,
      "loss": 0.2926,
      "step": 1778
    },
    {
      "epoch": 0.46071671360590466,
      "grad_norm": 3.53170730529767,
      "learning_rate": 8.168450846465906e-07,
      "loss": 0.352,
      "step": 1779
    },
    {
      "epoch": 0.46097568871192257,
      "grad_norm": 8.160401469509367,
      "learning_rate": 8.163070939050326e-07,
      "loss": 0.2963,
      "step": 1780
    },
    {
      "epoch": 0.46097568871192257,
      "eval_PRM Accuracy": 0.8679245283018868,
      "eval_PRM F1": 0.9135802469135802,
      "eval_PRM F1 AUC": 0.8370874803562075,
      "eval_PRM F1 AUC (fixed)": 0.8101100052383446,
      "eval_PRM F1 Neg": 0.72,
      "eval_PRM NPV": 0.6666666666666666,
      "eval_PRM Precision": 0.9367088607594937,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.23153018951416016,
      "eval_runtime": 14.7349,
      "eval_samples_per_second": 2.986,
      "eval_steps_per_second": 0.136,
      "step": 1780
    },
    {
      "epoch": 0.4612346638179405,
      "grad_norm": 4.367160519908248,
      "learning_rate": 8.157689467150617e-07,
      "loss": 0.3595,
      "step": 1781
    },
    {
      "epoch": 0.46149363892395845,
      "grad_norm": 6.551298425994705,
      "learning_rate": 8.152306435167675e-07,
      "loss": 0.432,
      "step": 1782
    },
    {
      "epoch": 0.46175261402997636,
      "grad_norm": 5.659351614486442,
      "learning_rate": 8.146921847503673e-07,
      "loss": 0.3914,
      "step": 1783
    },
    {
      "epoch": 0.4620115891359943,
      "grad_norm": 3.0275933641047263,
      "learning_rate": 8.141535708562052e-07,
      "loss": 0.3372,
      "step": 1784
    },
    {
      "epoch": 0.46227056424201224,
      "grad_norm": 4.730560492765888,
      "learning_rate": 8.136148022747529e-07,
      "loss": 0.32,
      "step": 1785
    },
    {
      "epoch": 0.46227056424201224,
      "eval_PRM Accuracy": 0.8679245283018868,
      "eval_PRM F1": 0.9113924050632911,
      "eval_PRM F1 AUC": 0.8685175484546883,
      "eval_PRM F1 AUC (fixed)": 0.8145625982189628,
      "eval_PRM F1 Neg": 0.7407407407407407,
      "eval_PRM NPV": 0.6451612903225806,
      "eval_PRM Precision": 0.96,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.8695652173913043,
      "eval_loss": 0.23782426118850708,
      "eval_runtime": 14.7677,
      "eval_samples_per_second": 2.979,
      "eval_steps_per_second": 0.135,
      "step": 1785
    },
    {
      "epoch": 0.46252953934803015,
      "grad_norm": 5.245711685236173,
      "learning_rate": 8.130758794466081e-07,
      "loss": 0.4484,
      "step": 1786
    },
    {
      "epoch": 0.4627885144540481,
      "grad_norm": 6.407074676859616,
      "learning_rate": 8.125368028124948e-07,
      "loss": 0.3551,
      "step": 1787
    },
    {
      "epoch": 0.46304748956006603,
      "grad_norm": 4.594617328789136,
      "learning_rate": 8.119975728132627e-07,
      "loss": 0.4103,
      "step": 1788
    },
    {
      "epoch": 0.463306464666084,
      "grad_norm": 3.812322209886043,
      "learning_rate": 8.114581898898868e-07,
      "loss": 0.3434,
      "step": 1789
    },
    {
      "epoch": 0.4635654397721019,
      "grad_norm": 4.42699867851653,
      "learning_rate": 8.109186544834679e-07,
      "loss": 0.3678,
      "step": 1790
    },
    {
      "epoch": 0.4635654397721019,
      "eval_PRM Accuracy": 0.8679245283018868,
      "eval_PRM F1": 0.9113924050632911,
      "eval_PRM F1 AUC": 0.8685175484546883,
      "eval_PRM F1 AUC (fixed)": 0.8088004190675747,
      "eval_PRM F1 Neg": 0.7407407407407407,
      "eval_PRM NPV": 0.6451612903225806,
      "eval_PRM Precision": 0.96,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.8695652173913043,
      "eval_loss": 0.24409064650535583,
      "eval_runtime": 14.7272,
      "eval_samples_per_second": 2.988,
      "eval_steps_per_second": 0.136,
      "step": 1790
    },
    {
      "epoch": 0.4638244148781198,
      "grad_norm": 10.198150664547708,
      "learning_rate": 8.103789670352304e-07,
      "loss": 0.3643,
      "step": 1791
    },
    {
      "epoch": 0.4640833899841378,
      "grad_norm": 9.547535616973445,
      "learning_rate": 8.098391279865236e-07,
      "loss": 0.3901,
      "step": 1792
    },
    {
      "epoch": 0.4643423650901557,
      "grad_norm": 3.682422647079557,
      "learning_rate": 8.092991377788209e-07,
      "loss": 0.3185,
      "step": 1793
    },
    {
      "epoch": 0.46460134019617366,
      "grad_norm": 4.563303944133527,
      "learning_rate": 8.08758996853719e-07,
      "loss": 0.2776,
      "step": 1794
    },
    {
      "epoch": 0.4648603153021916,
      "grad_norm": 2.8936492624830685,
      "learning_rate": 8.082187056529381e-07,
      "loss": 0.2755,
      "step": 1795
    },
    {
      "epoch": 0.4648603153021916,
      "eval_PRM Accuracy": 0.8490566037735849,
      "eval_PRM F1": 0.9,
      "eval_PRM F1 AUC": 0.8250392875851232,
      "eval_PRM F1 AUC (fixed)": 0.7996333158721844,
      "eval_PRM F1 Neg": 0.6923076923076923,
      "eval_PRM NPV": 0.6206896551724138,
      "eval_PRM Precision": 0.935064935064935,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.23865698277950287,
      "eval_runtime": 14.3072,
      "eval_samples_per_second": 3.075,
      "eval_steps_per_second": 0.14,
      "step": 1795
    },
    {
      "epoch": 0.4651192904082095,
      "grad_norm": 3.2301474276592947,
      "learning_rate": 8.076782646183214e-07,
      "loss": 0.3311,
      "step": 1796
    },
    {
      "epoch": 0.46537826551422745,
      "grad_norm": 3.3992629897493445,
      "learning_rate": 8.071376741918341e-07,
      "loss": 0.3538,
      "step": 1797
    },
    {
      "epoch": 0.46563724062024536,
      "grad_norm": 4.891436235900483,
      "learning_rate": 8.06596934815564e-07,
      "loss": 0.3409,
      "step": 1798
    },
    {
      "epoch": 0.46589621572626333,
      "grad_norm": 2.7678048082568214,
      "learning_rate": 8.060560469317208e-07,
      "loss": 0.3071,
      "step": 1799
    },
    {
      "epoch": 0.46615519083228124,
      "grad_norm": 3.901486260296828,
      "learning_rate": 8.055150109826352e-07,
      "loss": 0.2969,
      "step": 1800
    },
    {
      "epoch": 0.46615519083228124,
      "eval_PRM Accuracy": 0.8679245283018868,
      "eval_PRM F1": 0.9135802469135802,
      "eval_PRM F1 AUC": 0.8370874803562075,
      "eval_PRM F1 AUC (fixed)": 0.8006809848088005,
      "eval_PRM F1 Neg": 0.72,
      "eval_PRM NPV": 0.6666666666666666,
      "eval_PRM Precision": 0.9367088607594937,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.23264627158641815,
      "eval_runtime": 14.3344,
      "eval_samples_per_second": 3.07,
      "eval_steps_per_second": 0.14,
      "step": 1800
    },
    {
      "epoch": 0.4664141659382992,
      "grad_norm": 4.190246689208436,
      "learning_rate": 8.049738274107594e-07,
      "loss": 0.3229,
      "step": 1801
    },
    {
      "epoch": 0.4666731410443171,
      "grad_norm": 6.341832105693065,
      "learning_rate": 8.044324966586663e-07,
      "loss": 0.2824,
      "step": 1802
    },
    {
      "epoch": 0.46693211615033503,
      "grad_norm": 3.668661222275811,
      "learning_rate": 8.038910191690486e-07,
      "loss": 0.2618,
      "step": 1803
    },
    {
      "epoch": 0.467191091256353,
      "grad_norm": 3.1454731514946133,
      "learning_rate": 8.033493953847199e-07,
      "loss": 0.3348,
      "step": 1804
    },
    {
      "epoch": 0.4674500663623709,
      "grad_norm": 5.806055447923304,
      "learning_rate": 8.028076257486127e-07,
      "loss": 0.3321,
      "step": 1805
    },
    {
      "epoch": 0.4674500663623709,
      "eval_PRM Accuracy": 0.8679245283018868,
      "eval_PRM F1": 0.9146341463414634,
      "eval_PRM F1 AUC": 0.8213724463069669,
      "eval_PRM F1 AUC (fixed)": 0.8022524882137245,
      "eval_PRM F1 Neg": 0.7083333333333334,
      "eval_PRM NPV": 0.68,
      "eval_PRM Precision": 0.9259259259259259,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.22866840660572052,
      "eval_runtime": 14.4366,
      "eval_samples_per_second": 3.048,
      "eval_steps_per_second": 0.139,
      "step": 1805
    },
    {
      "epoch": 0.4677090414683889,
      "grad_norm": 3.411597990418709,
      "learning_rate": 8.022657107037793e-07,
      "loss": 0.3599,
      "step": 1806
    },
    {
      "epoch": 0.4679680165744068,
      "grad_norm": 3.4708399054715016,
      "learning_rate": 8.017236506933904e-07,
      "loss": 0.2624,
      "step": 1807
    },
    {
      "epoch": 0.4682269916804247,
      "grad_norm": 3.6053780892117873,
      "learning_rate": 8.011814461607355e-07,
      "loss": 0.2857,
      "step": 1808
    },
    {
      "epoch": 0.46848596678644266,
      "grad_norm": 4.7200909437066105,
      "learning_rate": 8.006390975492226e-07,
      "loss": 0.3552,
      "step": 1809
    },
    {
      "epoch": 0.4687449418924606,
      "grad_norm": 3.9219358614373485,
      "learning_rate": 8.000966053023769e-07,
      "loss": 0.3464,
      "step": 1810
    },
    {
      "epoch": 0.4687449418924606,
      "eval_PRM Accuracy": 0.8584905660377359,
      "eval_PRM F1": 0.9079754601226994,
      "eval_PRM F1 AUC": 0.8153483499214248,
      "eval_PRM F1 AUC (fixed)": 0.8017286537454165,
      "eval_PRM F1 Neg": 0.6938775510204082,
      "eval_PRM NPV": 0.6538461538461539,
      "eval_PRM Precision": 0.925,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.22844268381595612,
      "eval_runtime": 14.2589,
      "eval_samples_per_second": 3.086,
      "eval_steps_per_second": 0.14,
      "step": 1810
    },
    {
      "epoch": 0.46900391699847854,
      "grad_norm": 9.33731807939765,
      "learning_rate": 7.995539698638417e-07,
      "loss": 0.4929,
      "step": 1811
    },
    {
      "epoch": 0.46926289210449645,
      "grad_norm": 3.3611852528836503,
      "learning_rate": 7.990111916773768e-07,
      "loss": 0.3093,
      "step": 1812
    },
    {
      "epoch": 0.46952186721051437,
      "grad_norm": 7.494097866245163,
      "learning_rate": 7.984682711868591e-07,
      "loss": 0.456,
      "step": 1813
    },
    {
      "epoch": 0.46978084231653233,
      "grad_norm": 4.179278594454299,
      "learning_rate": 7.979252088362819e-07,
      "loss": 0.4174,
      "step": 1814
    },
    {
      "epoch": 0.47003981742255024,
      "grad_norm": 3.562185749898052,
      "learning_rate": 7.973820050697545e-07,
      "loss": 0.2682,
      "step": 1815
    },
    {
      "epoch": 0.47003981742255024,
      "eval_PRM Accuracy": 0.8490566037735849,
      "eval_PRM F1": 0.9,
      "eval_PRM F1 AUC": 0.8250392875851232,
      "eval_PRM F1 AUC (fixed)": 0.7991094814038764,
      "eval_PRM F1 Neg": 0.6923076923076923,
      "eval_PRM NPV": 0.6206896551724138,
      "eval_PRM Precision": 0.935064935064935,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.23334312438964844,
      "eval_runtime": 14.4683,
      "eval_samples_per_second": 3.041,
      "eval_steps_per_second": 0.138,
      "step": 1815
    },
    {
      "epoch": 0.4702987925285682,
      "grad_norm": 3.528772440221004,
      "learning_rate": 7.968386603315015e-07,
      "loss": 0.2492,
      "step": 1816
    },
    {
      "epoch": 0.4705577676345861,
      "grad_norm": 4.466703649793415,
      "learning_rate": 7.96295175065863e-07,
      "loss": 0.2534,
      "step": 1817
    },
    {
      "epoch": 0.47081674274060403,
      "grad_norm": 6.747629103938333,
      "learning_rate": 7.957515497172944e-07,
      "loss": 0.5479,
      "step": 1818
    },
    {
      "epoch": 0.471075717846622,
      "grad_norm": 3.383659302958618,
      "learning_rate": 7.952077847303652e-07,
      "loss": 0.2997,
      "step": 1819
    },
    {
      "epoch": 0.4713346929526399,
      "grad_norm": 5.80849933968738,
      "learning_rate": 7.946638805497592e-07,
      "loss": 0.265,
      "step": 1820
    },
    {
      "epoch": 0.4713346929526399,
      "eval_PRM Accuracy": 0.8490566037735849,
      "eval_PRM F1": 0.8987341772151899,
      "eval_PRM F1 AUC": 0.8407543216343635,
      "eval_PRM F1 AUC (fixed)": 0.7988475641697224,
      "eval_PRM F1 Neg": 0.7037037037037037,
      "eval_PRM NPV": 0.6129032258064516,
      "eval_PRM Precision": 0.9466666666666667,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.8260869565217391,
      "eval_loss": 0.2402866780757904,
      "eval_runtime": 14.4736,
      "eval_samples_per_second": 3.04,
      "eval_steps_per_second": 0.138,
      "step": 1820
    },
    {
      "epoch": 0.4715936680586579,
      "grad_norm": 8.48922152390816,
      "learning_rate": 7.941198376202739e-07,
      "loss": 0.3338,
      "step": 1821
    },
    {
      "epoch": 0.4718526431646758,
      "grad_norm": 5.267488243992496,
      "learning_rate": 7.935756563868205e-07,
      "loss": 0.2963,
      "step": 1822
    },
    {
      "epoch": 0.47211161827069376,
      "grad_norm": 6.023567792691506,
      "learning_rate": 7.930313372944234e-07,
      "loss": 0.3222,
      "step": 1823
    },
    {
      "epoch": 0.47237059337671167,
      "grad_norm": 3.1526189950294277,
      "learning_rate": 7.924868807882194e-07,
      "loss": 0.2923,
      "step": 1824
    },
    {
      "epoch": 0.4726295684827296,
      "grad_norm": 4.3801800497460555,
      "learning_rate": 7.919422873134579e-07,
      "loss": 0.3398,
      "step": 1825
    },
    {
      "epoch": 0.4726295684827296,
      "eval_PRM Accuracy": 0.8490566037735849,
      "eval_PRM F1": 0.9,
      "eval_PRM F1 AUC": 0.8250392875851232,
      "eval_PRM F1 AUC (fixed)": 0.8009429020429544,
      "eval_PRM F1 Neg": 0.6923076923076923,
      "eval_PRM NPV": 0.6206896551724138,
      "eval_PRM Precision": 0.935064935064935,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.23072166740894318,
      "eval_runtime": 14.5874,
      "eval_samples_per_second": 3.016,
      "eval_steps_per_second": 0.137,
      "step": 1825
    },
    {
      "epoch": 0.47288854358874755,
      "grad_norm": 4.115486615500542,
      "learning_rate": 7.913975573155002e-07,
      "loss": 0.3429,
      "step": 1826
    },
    {
      "epoch": 0.47314751869476546,
      "grad_norm": 6.191927084917738,
      "learning_rate": 7.908526912398193e-07,
      "loss": 0.4079,
      "step": 1827
    },
    {
      "epoch": 0.4734064938007834,
      "grad_norm": 6.07246804661019,
      "learning_rate": 7.903076895319994e-07,
      "loss": 0.3679,
      "step": 1828
    },
    {
      "epoch": 0.47366546890680133,
      "grad_norm": 3.280543854753348,
      "learning_rate": 7.897625526377357e-07,
      "loss": 0.3458,
      "step": 1829
    },
    {
      "epoch": 0.47392444401281925,
      "grad_norm": 3.5957909400024413,
      "learning_rate": 7.892172810028343e-07,
      "loss": 0.3064,
      "step": 1830
    },
    {
      "epoch": 0.47392444401281925,
      "eval_PRM Accuracy": 0.8584905660377359,
      "eval_PRM F1": 0.906832298136646,
      "eval_PRM F1 AUC": 0.8310633839706653,
      "eval_PRM F1 AUC (fixed)": 0.8048716605552646,
      "eval_PRM F1 Neg": 0.7058823529411765,
      "eval_PRM NPV": 0.6428571428571429,
      "eval_PRM Precision": 0.9358974358974359,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.22785210609436035,
      "eval_runtime": 14.6248,
      "eval_samples_per_second": 3.009,
      "eval_steps_per_second": 0.137,
      "step": 1830
    },
    {
      "epoch": 0.4741834191188372,
      "grad_norm": 3.677294337664697,
      "learning_rate": 7.886718750732105e-07,
      "loss": 0.3454,
      "step": 1831
    },
    {
      "epoch": 0.4744423942248551,
      "grad_norm": 3.569993762111734,
      "learning_rate": 7.881263352948908e-07,
      "loss": 0.2533,
      "step": 1832
    },
    {
      "epoch": 0.4747013693308731,
      "grad_norm": 6.430137577963266,
      "learning_rate": 7.875806621140097e-07,
      "loss": 0.3809,
      "step": 1833
    },
    {
      "epoch": 0.474960344436891,
      "grad_norm": 3.800419683459649,
      "learning_rate": 7.87034855976812e-07,
      "loss": 0.2992,
      "step": 1834
    },
    {
      "epoch": 0.4752193195429089,
      "grad_norm": 4.394140499507367,
      "learning_rate": 7.864889173296507e-07,
      "loss": 0.3417,
      "step": 1835
    },
    {
      "epoch": 0.4752193195429089,
      "eval_PRM Accuracy": 0.8584905660377359,
      "eval_PRM F1": 0.906832298136646,
      "eval_PRM F1 AUC": 0.8310633839706653,
      "eval_PRM F1 AUC (fixed)": 0.8051335777894185,
      "eval_PRM F1 Neg": 0.7058823529411765,
      "eval_PRM NPV": 0.6428571428571429,
      "eval_PRM Precision": 0.9358974358974359,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.2271353155374527,
      "eval_runtime": 14.45,
      "eval_samples_per_second": 3.045,
      "eval_steps_per_second": 0.138,
      "step": 1835
    },
    {
      "epoch": 0.4754782946489269,
      "grad_norm": 6.477449736941739,
      "learning_rate": 7.859428466189871e-07,
      "loss": 0.3568,
      "step": 1836
    },
    {
      "epoch": 0.4757372697549448,
      "grad_norm": 3.9059162852311253,
      "learning_rate": 7.853966442913905e-07,
      "loss": 0.3428,
      "step": 1837
    },
    {
      "epoch": 0.47599624486096276,
      "grad_norm": 3.782563962288157,
      "learning_rate": 7.848503107935379e-07,
      "loss": 0.3399,
      "step": 1838
    },
    {
      "epoch": 0.47625521996698067,
      "grad_norm": 5.323181969546517,
      "learning_rate": 7.84303846572214e-07,
      "loss": 0.2577,
      "step": 1839
    },
    {
      "epoch": 0.4765141950729986,
      "grad_norm": 4.6033390280073405,
      "learning_rate": 7.837572520743098e-07,
      "loss": 0.3835,
      "step": 1840
    },
    {
      "epoch": 0.4765141950729986,
      "eval_PRM Accuracy": 0.8584905660377359,
      "eval_PRM F1": 0.906832298136646,
      "eval_PRM F1 AUC": 0.8310633839706653,
      "eval_PRM F1 AUC (fixed)": 0.8077527501309587,
      "eval_PRM F1 Neg": 0.7058823529411765,
      "eval_PRM NPV": 0.6428571428571429,
      "eval_PRM Precision": 0.9358974358974359,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.22768282890319824,
      "eval_runtime": 14.3944,
      "eval_samples_per_second": 3.057,
      "eval_steps_per_second": 0.139,
      "step": 1840
    },
    {
      "epoch": 0.47677317017901655,
      "grad_norm": 4.496844293239733,
      "learning_rate": 7.832105277468232e-07,
      "loss": 0.3543,
      "step": 1841
    },
    {
      "epoch": 0.47703214528503446,
      "grad_norm": 5.435623613598174,
      "learning_rate": 7.826636740368578e-07,
      "loss": 0.2871,
      "step": 1842
    },
    {
      "epoch": 0.4772911203910524,
      "grad_norm": 3.331768864278108,
      "learning_rate": 7.821166913916233e-07,
      "loss": 0.3231,
      "step": 1843
    },
    {
      "epoch": 0.47755009549707034,
      "grad_norm": 4.819066933385308,
      "learning_rate": 7.815695802584354e-07,
      "loss": 0.3368,
      "step": 1844
    },
    {
      "epoch": 0.4778090706030883,
      "grad_norm": 8.174504593540913,
      "learning_rate": 7.810223410847139e-07,
      "loss": 0.2641,
      "step": 1845
    },
    {
      "epoch": 0.4778090706030883,
      "eval_PRM Accuracy": 0.8490566037735849,
      "eval_PRM F1": 0.9,
      "eval_PRM F1 AUC": 0.8250392875851232,
      "eval_PRM F1 AUC (fixed)": 0.8090623363017286,
      "eval_PRM F1 Neg": 0.6923076923076923,
      "eval_PRM NPV": 0.6206896551724138,
      "eval_PRM Precision": 0.935064935064935,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.2282373309135437,
      "eval_runtime": 14.7225,
      "eval_samples_per_second": 2.989,
      "eval_steps_per_second": 0.136,
      "step": 1845
    },
    {
      "epoch": 0.4780680457091062,
      "grad_norm": 7.713612608077943,
      "learning_rate": 7.80474974317984e-07,
      "loss": 0.4385,
      "step": 1846
    },
    {
      "epoch": 0.4783270208151241,
      "grad_norm": 5.501139892781156,
      "learning_rate": 7.79927480405875e-07,
      "loss": 0.4606,
      "step": 1847
    },
    {
      "epoch": 0.4785859959211421,
      "grad_norm": 4.028469917800099,
      "learning_rate": 7.793798597961199e-07,
      "loss": 0.3925,
      "step": 1848
    },
    {
      "epoch": 0.47884497102716,
      "grad_norm": 3.5977063939754035,
      "learning_rate": 7.788321129365558e-07,
      "loss": 0.3001,
      "step": 1849
    },
    {
      "epoch": 0.47910394613317797,
      "grad_norm": 4.1749602130213574,
      "learning_rate": 7.782842402751231e-07,
      "loss": 0.3117,
      "step": 1850
    },
    {
      "epoch": 0.47910394613317797,
      "eval_PRM Accuracy": 0.8584905660377359,
      "eval_PRM F1": 0.9056603773584906,
      "eval_PRM F1 AUC": 0.8467784180199058,
      "eval_PRM F1 AUC (fixed)": 0.8116815086432687,
      "eval_PRM F1 Neg": 0.7169811320754716,
      "eval_PRM NPV": 0.6333333333333333,
      "eval_PRM Precision": 0.9473684210526315,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.8260869565217391,
      "eval_loss": 0.23110289871692657,
      "eval_runtime": 14.5178,
      "eval_samples_per_second": 3.031,
      "eval_steps_per_second": 0.138,
      "step": 1850
    },
    {
      "epoch": 0.4793629212391959,
      "grad_norm": 4.392542629381872,
      "learning_rate": 7.777362422598645e-07,
      "loss": 0.319,
      "step": 1851
    },
    {
      "epoch": 0.4796218963452138,
      "grad_norm": 3.2429010046274267,
      "learning_rate": 7.771881193389254e-07,
      "loss": 0.319,
      "step": 1852
    },
    {
      "epoch": 0.47988087145123176,
      "grad_norm": 3.3117414902850566,
      "learning_rate": 7.766398719605538e-07,
      "loss": 0.2571,
      "step": 1853
    },
    {
      "epoch": 0.48013984655724967,
      "grad_norm": 5.246262430891637,
      "learning_rate": 7.76091500573099e-07,
      "loss": 0.3272,
      "step": 1854
    },
    {
      "epoch": 0.48039882166326764,
      "grad_norm": 3.034849808493933,
      "learning_rate": 7.75543005625012e-07,
      "loss": 0.2794,
      "step": 1855
    },
    {
      "epoch": 0.48039882166326764,
      "eval_PRM Accuracy": 0.8584905660377359,
      "eval_PRM F1": 0.9056603773584906,
      "eval_PRM F1 AUC": 0.8467784180199058,
      "eval_PRM F1 AUC (fixed)": 0.8169198533263489,
      "eval_PRM F1 Neg": 0.7169811320754716,
      "eval_PRM NPV": 0.6333333333333333,
      "eval_PRM Precision": 0.9473684210526315,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.8260869565217391,
      "eval_loss": 0.2294907569885254,
      "eval_runtime": 15.0458,
      "eval_samples_per_second": 2.924,
      "eval_steps_per_second": 0.133,
      "step": 1855
    },
    {
      "epoch": 0.48065779676928555,
      "grad_norm": 8.853086892103162,
      "learning_rate": 7.749943875648447e-07,
      "loss": 0.357,
      "step": 1856
    },
    {
      "epoch": 0.48091677187530346,
      "grad_norm": 9.246095988877377,
      "learning_rate": 7.744456468412493e-07,
      "loss": 0.3757,
      "step": 1857
    },
    {
      "epoch": 0.48117574698132143,
      "grad_norm": 5.900565173417949,
      "learning_rate": 7.738967839029793e-07,
      "loss": 0.3312,
      "step": 1858
    },
    {
      "epoch": 0.48143472208733934,
      "grad_norm": 3.862804686023086,
      "learning_rate": 7.733477991988872e-07,
      "loss": 0.3497,
      "step": 1859
    },
    {
      "epoch": 0.4816936971933573,
      "grad_norm": 4.952938875065223,
      "learning_rate": 7.727986931779257e-07,
      "loss": 0.3126,
      "step": 1860
    },
    {
      "epoch": 0.4816936971933573,
      "eval_PRM Accuracy": 0.8584905660377359,
      "eval_PRM F1": 0.906832298136646,
      "eval_PRM F1 AUC": 0.8310633839706653,
      "eval_PRM F1 AUC (fixed)": 0.8234677841801991,
      "eval_PRM F1 Neg": 0.7058823529411765,
      "eval_PRM NPV": 0.6428571428571429,
      "eval_PRM Precision": 0.9358974358974359,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.21761594712734222,
      "eval_runtime": 14.6095,
      "eval_samples_per_second": 3.012,
      "eval_steps_per_second": 0.137,
      "step": 1860
    },
    {
      "epoch": 0.4819526722993752,
      "grad_norm": 4.573100864306212,
      "learning_rate": 7.722494662891462e-07,
      "loss": 0.2673,
      "step": 1861
    },
    {
      "epoch": 0.48221164740539313,
      "grad_norm": 4.845352720816476,
      "learning_rate": 7.717001189816992e-07,
      "loss": 0.3863,
      "step": 1862
    },
    {
      "epoch": 0.4824706225114111,
      "grad_norm": 7.9411721265663395,
      "learning_rate": 7.711506517048338e-07,
      "loss": 0.4327,
      "step": 1863
    },
    {
      "epoch": 0.482729597617429,
      "grad_norm": 6.019250258631447,
      "learning_rate": 7.70601064907897e-07,
      "loss": 0.4286,
      "step": 1864
    },
    {
      "epoch": 0.482988572723447,
      "grad_norm": 10.046837357024193,
      "learning_rate": 7.700513590403338e-07,
      "loss": 0.4118,
      "step": 1865
    },
    {
      "epoch": 0.482988572723447,
      "eval_PRM Accuracy": 0.8962264150943396,
      "eval_PRM F1": 0.9333333333333333,
      "eval_PRM F1 AUC": 0.855159769512834,
      "eval_PRM F1 AUC (fixed)": 0.8284442116291252,
      "eval_PRM F1 Neg": 0.7659574468085106,
      "eval_PRM NPV": 0.75,
      "eval_PRM Precision": 0.9390243902439024,
      "eval_PRM Recall": 0.927710843373494,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.21364346146583557,
      "eval_runtime": 14.8981,
      "eval_samples_per_second": 2.953,
      "eval_steps_per_second": 0.134,
      "step": 1865
    },
    {
      "epoch": 0.4832475478294649,
      "grad_norm": 3.3702477649911944,
      "learning_rate": 7.695015345516861e-07,
      "loss": 0.3698,
      "step": 1866
    },
    {
      "epoch": 0.48350652293548285,
      "grad_norm": 6.242674576224769,
      "learning_rate": 7.689515918915931e-07,
      "loss": 0.338,
      "step": 1867
    },
    {
      "epoch": 0.48376549804150076,
      "grad_norm": 6.497090573684327,
      "learning_rate": 7.684015315097909e-07,
      "loss": 0.3487,
      "step": 1868
    },
    {
      "epoch": 0.4840244731475187,
      "grad_norm": 3.7831142105036686,
      "learning_rate": 7.678513538561117e-07,
      "loss": 0.2994,
      "step": 1869
    },
    {
      "epoch": 0.48428344825353664,
      "grad_norm": 12.046248358308517,
      "learning_rate": 7.673010593804829e-07,
      "loss": 0.3802,
      "step": 1870
    },
    {
      "epoch": 0.48428344825353664,
      "eval_PRM Accuracy": 0.8773584905660378,
      "eval_PRM F1": 0.9202453987730062,
      "eval_PRM F1 AUC": 0.8431115767417497,
      "eval_PRM F1 AUC (fixed)": 0.822420115243583,
      "eval_PRM F1 Neg": 0.7346938775510204,
      "eval_PRM NPV": 0.6923076923076923,
      "eval_PRM Precision": 0.9375,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.21648350358009338,
      "eval_runtime": 14.7542,
      "eval_samples_per_second": 2.982,
      "eval_steps_per_second": 0.136,
      "step": 1870
    },
    {
      "epoch": 0.48454242335955455,
      "grad_norm": 4.310071878641915,
      "learning_rate": 7.667506485329287e-07,
      "loss": 0.3622,
      "step": 1871
    },
    {
      "epoch": 0.4848013984655725,
      "grad_norm": 2.7821621342806693,
      "learning_rate": 7.662001217635673e-07,
      "loss": 0.2683,
      "step": 1872
    },
    {
      "epoch": 0.48506037357159043,
      "grad_norm": 3.0532767740012687,
      "learning_rate": 7.656494795226128e-07,
      "loss": 0.3373,
      "step": 1873
    },
    {
      "epoch": 0.48531934867760834,
      "grad_norm": 5.2910761333068805,
      "learning_rate": 7.650987222603728e-07,
      "loss": 0.3208,
      "step": 1874
    },
    {
      "epoch": 0.4855783237836263,
      "grad_norm": 3.2230981266809238,
      "learning_rate": 7.645478504272491e-07,
      "loss": 0.3622,
      "step": 1875
    },
    {
      "epoch": 0.4855783237836263,
      "eval_PRM Accuracy": 0.8584905660377359,
      "eval_PRM F1": 0.906832298136646,
      "eval_PRM F1 AUC": 0.8310633839706653,
      "eval_PRM F1 AUC (fixed)": 0.818753273965427,
      "eval_PRM F1 Neg": 0.7058823529411765,
      "eval_PRM NPV": 0.6428571428571429,
      "eval_PRM Precision": 0.9358974358974359,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.22220145165920258,
      "eval_runtime": 14.3859,
      "eval_samples_per_second": 3.059,
      "eval_steps_per_second": 0.139,
      "step": 1875
    },
    {
      "epoch": 0.4858372988896442,
      "grad_norm": 3.578950466609083,
      "learning_rate": 7.639968644737378e-07,
      "loss": 0.3411,
      "step": 1876
    },
    {
      "epoch": 0.4860962739956622,
      "grad_norm": 4.7903902264133995,
      "learning_rate": 7.634457648504277e-07,
      "loss": 0.2874,
      "step": 1877
    },
    {
      "epoch": 0.4863552491016801,
      "grad_norm": 4.419419406945147,
      "learning_rate": 7.628945520080008e-07,
      "loss": 0.249,
      "step": 1878
    },
    {
      "epoch": 0.486614224207698,
      "grad_norm": 3.544817731066733,
      "learning_rate": 7.623432263972319e-07,
      "loss": 0.3381,
      "step": 1879
    },
    {
      "epoch": 0.486873199313716,
      "grad_norm": 5.67829322818424,
      "learning_rate": 7.617917884689876e-07,
      "loss": 0.4002,
      "step": 1880
    },
    {
      "epoch": 0.486873199313716,
      "eval_PRM Accuracy": 0.8584905660377359,
      "eval_PRM F1": 0.906832298136646,
      "eval_PRM F1 AUC": 0.8310633839706653,
      "eval_PRM F1 AUC (fixed)": 0.8137768465165007,
      "eval_PRM F1 Neg": 0.7058823529411765,
      "eval_PRM NPV": 0.6428571428571429,
      "eval_PRM Precision": 0.9358974358974359,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.22564397752285004,
      "eval_runtime": 14.7604,
      "eval_samples_per_second": 2.981,
      "eval_steps_per_second": 0.135,
      "step": 1880
    },
    {
      "epoch": 0.4871321744197339,
      "grad_norm": 5.313345534177357,
      "learning_rate": 7.612402386742265e-07,
      "loss": 0.2828,
      "step": 1881
    },
    {
      "epoch": 0.48739114952575185,
      "grad_norm": 4.080298859481059,
      "learning_rate": 7.606885774639989e-07,
      "loss": 0.2898,
      "step": 1882
    },
    {
      "epoch": 0.48765012463176977,
      "grad_norm": 3.987064199446086,
      "learning_rate": 7.601368052894459e-07,
      "loss": 0.3066,
      "step": 1883
    },
    {
      "epoch": 0.48790909973778773,
      "grad_norm": 4.539410577961955,
      "learning_rate": 7.595849226017996e-07,
      "loss": 0.2902,
      "step": 1884
    },
    {
      "epoch": 0.48816807484380564,
      "grad_norm": 5.844603798205524,
      "learning_rate": 7.590329298523821e-07,
      "loss": 0.3285,
      "step": 1885
    },
    {
      "epoch": 0.48816807484380564,
      "eval_PRM Accuracy": 0.8773584905660378,
      "eval_PRM F1": 0.9202453987730062,
      "eval_PRM F1 AUC": 0.8431115767417497,
      "eval_PRM F1 AUC (fixed)": 0.8135149292823468,
      "eval_PRM F1 Neg": 0.7346938775510204,
      "eval_PRM NPV": 0.6923076923076923,
      "eval_PRM Precision": 0.9375,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.22327542304992676,
      "eval_runtime": 14.4271,
      "eval_samples_per_second": 3.05,
      "eval_steps_per_second": 0.139,
      "step": 1885
    },
    {
      "epoch": 0.48842704994982356,
      "grad_norm": 4.355824867251983,
      "learning_rate": 7.584808274926063e-07,
      "loss": 0.4263,
      "step": 1886
    },
    {
      "epoch": 0.4886860250558415,
      "grad_norm": 5.481873870990339,
      "learning_rate": 7.579286159739738e-07,
      "loss": 0.4366,
      "step": 1887
    },
    {
      "epoch": 0.48894500016185943,
      "grad_norm": 6.584261121249097,
      "learning_rate": 7.57376295748076e-07,
      "loss": 0.3342,
      "step": 1888
    },
    {
      "epoch": 0.4892039752678774,
      "grad_norm": 3.509521318462327,
      "learning_rate": 7.56823867266593e-07,
      "loss": 0.279,
      "step": 1889
    },
    {
      "epoch": 0.4894629503738953,
      "grad_norm": 5.087962103671689,
      "learning_rate": 7.562713309812939e-07,
      "loss": 0.3076,
      "step": 1890
    },
    {
      "epoch": 0.4894629503738953,
      "eval_PRM Accuracy": 0.8867924528301887,
      "eval_PRM F1": 0.926829268292683,
      "eval_PRM F1 AUC": 0.8491356731272918,
      "eval_PRM F1 AUC (fixed)": 0.8098480880041907,
      "eval_PRM F1 Neg": 0.75,
      "eval_PRM NPV": 0.72,
      "eval_PRM Precision": 0.9382716049382716,
      "eval_PRM Recall": 0.9156626506024096,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.2252662181854248,
      "eval_runtime": 14.9327,
      "eval_samples_per_second": 2.947,
      "eval_steps_per_second": 0.134,
      "step": 1890
    },
    {
      "epoch": 0.4897219254799132,
      "grad_norm": 11.314867139630405,
      "learning_rate": 7.557186873440348e-07,
      "loss": 0.4242,
      "step": 1891
    },
    {
      "epoch": 0.4899809005859312,
      "grad_norm": 2.251498860197715,
      "learning_rate": 7.55165936806761e-07,
      "loss": 0.2449,
      "step": 1892
    },
    {
      "epoch": 0.4902398756919491,
      "grad_norm": 3.8510977355179774,
      "learning_rate": 7.546130798215044e-07,
      "loss": 0.3536,
      "step": 1893
    },
    {
      "epoch": 0.49049885079796707,
      "grad_norm": 4.759798498619873,
      "learning_rate": 7.540601168403843e-07,
      "loss": 0.4067,
      "step": 1894
    },
    {
      "epoch": 0.490757825903985,
      "grad_norm": 4.494273430256073,
      "learning_rate": 7.535070483156062e-07,
      "loss": 0.3284,
      "step": 1895
    },
    {
      "epoch": 0.490757825903985,
      "eval_PRM Accuracy": 0.8773584905660378,
      "eval_PRM F1": 0.9202453987730062,
      "eval_PRM F1 AUC": 0.8431115767417497,
      "eval_PRM F1 AUC (fixed)": 0.8025144054478784,
      "eval_PRM F1 Neg": 0.7346938775510204,
      "eval_PRM NPV": 0.6923076923076923,
      "eval_PRM Precision": 0.9375,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.22756299376487732,
      "eval_runtime": 14.8443,
      "eval_samples_per_second": 2.964,
      "eval_steps_per_second": 0.135,
      "step": 1895
    },
    {
      "epoch": 0.4910168010100029,
      "grad_norm": 3.2714702635572137,
      "learning_rate": 7.529538746994623e-07,
      "loss": 0.3308,
      "step": 1896
    },
    {
      "epoch": 0.49127577611602086,
      "grad_norm": 3.4425896757391334,
      "learning_rate": 7.52400596444331e-07,
      "loss": 0.2968,
      "step": 1897
    },
    {
      "epoch": 0.49153475122203877,
      "grad_norm": 4.714291588919368,
      "learning_rate": 7.518472140026757e-07,
      "loss": 0.3404,
      "step": 1898
    },
    {
      "epoch": 0.49179372632805674,
      "grad_norm": 2.586142384819253,
      "learning_rate": 7.51293727827045e-07,
      "loss": 0.3105,
      "step": 1899
    },
    {
      "epoch": 0.49205270143407465,
      "grad_norm": 3.8451605864956835,
      "learning_rate": 7.50740138370073e-07,
      "loss": 0.3819,
      "step": 1900
    },
    {
      "epoch": 0.49205270143407465,
      "eval_PRM Accuracy": 0.8584905660377359,
      "eval_PRM F1": 0.906832298136646,
      "eval_PRM F1 AUC": 0.8310633839706653,
      "eval_PRM F1 AUC (fixed)": 0.8022524882137244,
      "eval_PRM F1 Neg": 0.7058823529411765,
      "eval_PRM NPV": 0.6428571428571429,
      "eval_PRM Precision": 0.9358974358974359,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.2354053109884262,
      "eval_runtime": 14.3986,
      "eval_samples_per_second": 3.056,
      "eval_steps_per_second": 0.139,
      "step": 1900
    },
    {
      "epoch": 0.49231167654009256,
      "grad_norm": 9.90430039984692,
      "learning_rate": 7.501864460844777e-07,
      "loss": 0.2935,
      "step": 1901
    },
    {
      "epoch": 0.4925706516461105,
      "grad_norm": 8.554044629123736,
      "learning_rate": 7.496326514230611e-07,
      "loss": 0.2935,
      "step": 1902
    },
    {
      "epoch": 0.49282962675212844,
      "grad_norm": 6.4037328948874395,
      "learning_rate": 7.490787548387096e-07,
      "loss": 0.2871,
      "step": 1903
    },
    {
      "epoch": 0.4930886018581464,
      "grad_norm": 6.576489357295404,
      "learning_rate": 7.485247567843921e-07,
      "loss": 0.3085,
      "step": 1904
    },
    {
      "epoch": 0.4933475769641643,
      "grad_norm": 4.315632213431601,
      "learning_rate": 7.479706577131611e-07,
      "loss": 0.3997,
      "step": 1905
    },
    {
      "epoch": 0.4933475769641643,
      "eval_PRM Accuracy": 0.8584905660377359,
      "eval_PRM F1": 0.906832298136646,
      "eval_PRM F1 AUC": 0.8310633839706653,
      "eval_PRM F1 AUC (fixed)": 0.8035620743844945,
      "eval_PRM F1 Neg": 0.7058823529411765,
      "eval_PRM NPV": 0.6428571428571429,
      "eval_PRM Precision": 0.9358974358974359,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.2289583534002304,
      "eval_runtime": 14.74,
      "eval_samples_per_second": 2.985,
      "eval_steps_per_second": 0.136,
      "step": 1905
    },
    {
      "epoch": 0.4936065520701823,
      "grad_norm": 4.1174298431364695,
      "learning_rate": 7.474164580781515e-07,
      "loss": 0.3716,
      "step": 1906
    },
    {
      "epoch": 0.4938655271762002,
      "grad_norm": 6.074577649510619,
      "learning_rate": 7.468621583325803e-07,
      "loss": 0.4052,
      "step": 1907
    },
    {
      "epoch": 0.4941245022822181,
      "grad_norm": 3.1570397691218117,
      "learning_rate": 7.463077589297467e-07,
      "loss": 0.3806,
      "step": 1908
    },
    {
      "epoch": 0.49438347738823607,
      "grad_norm": 3.238297089972081,
      "learning_rate": 7.45753260323031e-07,
      "loss": 0.3387,
      "step": 1909
    },
    {
      "epoch": 0.494642452494254,
      "grad_norm": 3.998534514193585,
      "learning_rate": 7.451986629658952e-07,
      "loss": 0.3421,
      "step": 1910
    },
    {
      "epoch": 0.494642452494254,
      "eval_PRM Accuracy": 0.8584905660377359,
      "eval_PRM F1": 0.906832298136646,
      "eval_PRM F1 AUC": 0.8310633839706653,
      "eval_PRM F1 AUC (fixed)": 0.8019905709795705,
      "eval_PRM F1 Neg": 0.7058823529411765,
      "eval_PRM NPV": 0.6428571428571429,
      "eval_PRM Precision": 0.9358974358974359,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.22805818915367126,
      "eval_runtime": 14.659,
      "eval_samples_per_second": 3.002,
      "eval_steps_per_second": 0.136,
      "step": 1910
    },
    {
      "epoch": 0.49490142760027195,
      "grad_norm": 5.485606573598385,
      "learning_rate": 7.44643967311881e-07,
      "loss": 0.429,
      "step": 1911
    },
    {
      "epoch": 0.49516040270628986,
      "grad_norm": 4.164661369960232,
      "learning_rate": 7.440891738146118e-07,
      "loss": 0.4009,
      "step": 1912
    },
    {
      "epoch": 0.49541937781230777,
      "grad_norm": 7.100390868750479,
      "learning_rate": 7.435342829277901e-07,
      "loss": 0.3876,
      "step": 1913
    },
    {
      "epoch": 0.49567835291832574,
      "grad_norm": 3.4589151338321975,
      "learning_rate": 7.429792951051985e-07,
      "loss": 0.3323,
      "step": 1914
    },
    {
      "epoch": 0.49593732802434365,
      "grad_norm": 4.653233157816339,
      "learning_rate": 7.424242108006988e-07,
      "loss": 0.2857,
      "step": 1915
    },
    {
      "epoch": 0.49593732802434365,
      "eval_PRM Accuracy": 0.8773584905660378,
      "eval_PRM F1": 0.9202453987730062,
      "eval_PRM F1 AUC": 0.8431115767417497,
      "eval_PRM F1 AUC (fixed)": 0.7967522262964903,
      "eval_PRM F1 Neg": 0.7346938775510204,
      "eval_PRM NPV": 0.6923076923076923,
      "eval_PRM Precision": 0.9375,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.23427662253379822,
      "eval_runtime": 14.4763,
      "eval_samples_per_second": 3.039,
      "eval_steps_per_second": 0.138,
      "step": 1915
    },
    {
      "epoch": 0.4961963031303616,
      "grad_norm": 3.722054051432413,
      "learning_rate": 7.418690304682311e-07,
      "loss": 0.2919,
      "step": 1916
    },
    {
      "epoch": 0.4964552782363795,
      "grad_norm": 3.7463561897294384,
      "learning_rate": 7.41313754561815e-07,
      "loss": 0.3363,
      "step": 1917
    },
    {
      "epoch": 0.49671425334239744,
      "grad_norm": 3.22192067410551,
      "learning_rate": 7.407583835355477e-07,
      "loss": 0.3378,
      "step": 1918
    },
    {
      "epoch": 0.4969732284484154,
      "grad_norm": 5.239676478746968,
      "learning_rate": 7.402029178436043e-07,
      "loss": 0.3523,
      "step": 1919
    },
    {
      "epoch": 0.4972322035544333,
      "grad_norm": 3.5421552493081947,
      "learning_rate": 7.396473579402372e-07,
      "loss": 0.3289,
      "step": 1920
    },
    {
      "epoch": 0.4972322035544333,
      "eval_PRM Accuracy": 0.8679245283018868,
      "eval_PRM F1": 0.9135802469135802,
      "eval_PRM F1 AUC": 0.8370874803562075,
      "eval_PRM F1 AUC (fixed)": 0.7943949711891043,
      "eval_PRM F1 Neg": 0.72,
      "eval_PRM NPV": 0.6666666666666666,
      "eval_PRM Precision": 0.9367088607594937,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.236124187707901,
      "eval_runtime": 14.6985,
      "eval_samples_per_second": 2.994,
      "eval_steps_per_second": 0.136,
      "step": 1920
    },
    {
      "epoch": 0.4974911786604513,
      "grad_norm": 3.4890272203065327,
      "learning_rate": 7.390917042797759e-07,
      "loss": 0.3269,
      "step": 1921
    },
    {
      "epoch": 0.4977501537664692,
      "grad_norm": 3.409924539361981,
      "learning_rate": 7.385359573166267e-07,
      "loss": 0.389,
      "step": 1922
    },
    {
      "epoch": 0.4980091288724871,
      "grad_norm": 3.7447981583847123,
      "learning_rate": 7.379801175052724e-07,
      "loss": 0.3386,
      "step": 1923
    },
    {
      "epoch": 0.4982681039785051,
      "grad_norm": 3.125092164233843,
      "learning_rate": 7.37424185300271e-07,
      "loss": 0.3098,
      "step": 1924
    },
    {
      "epoch": 0.498527079084523,
      "grad_norm": 6.446671391321472,
      "learning_rate": 7.368681611562568e-07,
      "loss": 0.4493,
      "step": 1925
    },
    {
      "epoch": 0.498527079084523,
      "eval_PRM Accuracy": 0.8584905660377359,
      "eval_PRM F1": 0.906832298136646,
      "eval_PRM F1 AUC": 0.8310633839706653,
      "eval_PRM F1 AUC (fixed)": 0.7983237297014144,
      "eval_PRM F1 Neg": 0.7058823529411765,
      "eval_PRM NPV": 0.6428571428571429,
      "eval_PRM Precision": 0.9358974358974359,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.2375415861606598,
      "eval_runtime": 14.8226,
      "eval_samples_per_second": 2.968,
      "eval_steps_per_second": 0.135,
      "step": 1925
    },
    {
      "epoch": 0.49878605419054095,
      "grad_norm": 3.208163207188384,
      "learning_rate": 7.363120455279386e-07,
      "loss": 0.3203,
      "step": 1926
    },
    {
      "epoch": 0.49904502929655886,
      "grad_norm": 4.218184608627159,
      "learning_rate": 7.35755838870101e-07,
      "loss": 0.3656,
      "step": 1927
    },
    {
      "epoch": 0.49930400440257683,
      "grad_norm": 6.76261794035762,
      "learning_rate": 7.351995416376024e-07,
      "loss": 0.3122,
      "step": 1928
    },
    {
      "epoch": 0.49956297950859474,
      "grad_norm": 4.402099460066203,
      "learning_rate": 7.346431542853747e-07,
      "loss": 0.3581,
      "step": 1929
    },
    {
      "epoch": 0.49982195461461265,
      "grad_norm": 10.558946238457926,
      "learning_rate": 7.340866772684247e-07,
      "loss": 0.339,
      "step": 1930
    },
    {
      "epoch": 0.49982195461461265,
      "eval_PRM Accuracy": 0.8584905660377359,
      "eval_PRM F1": 0.906832298136646,
      "eval_PRM F1 AUC": 0.8310633839706653,
      "eval_PRM F1 AUC (fixed)": 0.7964903090623363,
      "eval_PRM F1 Neg": 0.7058823529411765,
      "eval_PRM NPV": 0.6428571428571429,
      "eval_PRM Precision": 0.9358974358974359,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.23493939638137817,
      "eval_runtime": 14.5105,
      "eval_samples_per_second": 3.032,
      "eval_steps_per_second": 0.138,
      "step": 1930
    },
    {
      "epoch": 0.5000809297206306,
      "grad_norm": 5.684889369167639,
      "learning_rate": 7.335301110418317e-07,
      "loss": 0.3313,
      "step": 1931
    },
    {
      "epoch": 0.5003399048266486,
      "grad_norm": 4.128419726300969,
      "learning_rate": 7.329734560607481e-07,
      "loss": 0.3023,
      "step": 1932
    },
    {
      "epoch": 0.5005988799326665,
      "grad_norm": 3.191808202717404,
      "learning_rate": 7.324167127803994e-07,
      "loss": 0.3291,
      "step": 1933
    },
    {
      "epoch": 0.5008578550386844,
      "grad_norm": 3.3457819664225075,
      "learning_rate": 7.318598816560825e-07,
      "loss": 0.3111,
      "step": 1934
    },
    {
      "epoch": 0.5011168301447023,
      "grad_norm": 9.774334004828226,
      "learning_rate": 7.313029631431667e-07,
      "loss": 0.3708,
      "step": 1935
    },
    {
      "epoch": 0.5011168301447023,
      "eval_PRM Accuracy": 0.8962264150943396,
      "eval_PRM F1": 0.9333333333333333,
      "eval_PRM F1 AUC": 0.855159769512834,
      "eval_PRM F1 AUC (fixed)": 0.7964903090623363,
      "eval_PRM F1 Neg": 0.7659574468085106,
      "eval_PRM NPV": 0.75,
      "eval_PRM Precision": 0.9390243902439024,
      "eval_PRM Recall": 0.927710843373494,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.23073317110538483,
      "eval_runtime": 14.6507,
      "eval_samples_per_second": 3.003,
      "eval_steps_per_second": 0.137,
      "step": 1935
    },
    {
      "epoch": 0.5013758052507202,
      "grad_norm": 5.157407289004608,
      "learning_rate": 7.307459576970927e-07,
      "loss": 0.4173,
      "step": 1936
    },
    {
      "epoch": 0.5016347803567383,
      "grad_norm": 4.62674994744029,
      "learning_rate": 7.301888657733719e-07,
      "loss": 0.3263,
      "step": 1937
    },
    {
      "epoch": 0.5018937554627562,
      "grad_norm": 4.292305637281678,
      "learning_rate": 7.296316878275869e-07,
      "loss": 0.4149,
      "step": 1938
    },
    {
      "epoch": 0.5021527305687741,
      "grad_norm": 3.2197504841448543,
      "learning_rate": 7.290744243153905e-07,
      "loss": 0.3475,
      "step": 1939
    },
    {
      "epoch": 0.502411705674792,
      "grad_norm": 7.156108918431266,
      "learning_rate": 7.285170756925054e-07,
      "loss": 0.4301,
      "step": 1940
    },
    {
      "epoch": 0.502411705674792,
      "eval_PRM Accuracy": 0.9150943396226415,
      "eval_PRM F1": 0.9461077844311377,
      "eval_PRM F1 AUC": 0.8672079622839183,
      "eval_PRM F1 AUC (fixed)": 0.7988475641697224,
      "eval_PRM F1 Neg": 0.8,
      "eval_PRM NPV": 0.8181818181818182,
      "eval_PRM Precision": 0.9404761904761905,
      "eval_PRM Recall": 0.9518072289156626,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.23065905272960663,
      "eval_runtime": 14.6061,
      "eval_samples_per_second": 3.012,
      "eval_steps_per_second": 0.137,
      "step": 1940
    },
    {
      "epoch": 0.5026706807808099,
      "grad_norm": 8.050386581785924,
      "learning_rate": 7.279596424147239e-07,
      "loss": 0.3004,
      "step": 1941
    },
    {
      "epoch": 0.5029296558868279,
      "grad_norm": 2.79687624285831,
      "learning_rate": 7.274021249379076e-07,
      "loss": 0.2873,
      "step": 1942
    },
    {
      "epoch": 0.5031886309928458,
      "grad_norm": 2.4461020147386296,
      "learning_rate": 7.268445237179871e-07,
      "loss": 0.2805,
      "step": 1943
    },
    {
      "epoch": 0.5034476060988637,
      "grad_norm": 3.430225900119802,
      "learning_rate": 7.262868392109609e-07,
      "loss": 0.3657,
      "step": 1944
    },
    {
      "epoch": 0.5037065812048817,
      "grad_norm": 3.233469087202206,
      "learning_rate": 7.257290718728962e-07,
      "loss": 0.2989,
      "step": 1945
    },
    {
      "epoch": 0.5037065812048817,
      "eval_PRM Accuracy": 0.8679245283018868,
      "eval_PRM F1": 0.9135802469135802,
      "eval_PRM F1 AUC": 0.8370874803562075,
      "eval_PRM F1 AUC (fixed)": 0.7954426401257202,
      "eval_PRM F1 Neg": 0.72,
      "eval_PRM NPV": 0.6666666666666666,
      "eval_PRM Precision": 0.9367088607594937,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.22959648072719574,
      "eval_runtime": 15.3625,
      "eval_samples_per_second": 2.864,
      "eval_steps_per_second": 0.13,
      "step": 1945
    },
    {
      "epoch": 0.5039655563108996,
      "grad_norm": 7.165637789195796,
      "learning_rate": 7.251712221599279e-07,
      "loss": 0.4445,
      "step": 1946
    },
    {
      "epoch": 0.5042245314169176,
      "grad_norm": 4.457549586523684,
      "learning_rate": 7.24613290528258e-07,
      "loss": 0.2943,
      "step": 1947
    },
    {
      "epoch": 0.5044835065229355,
      "grad_norm": 7.04254739388138,
      "learning_rate": 7.240552774341558e-07,
      "loss": 0.4268,
      "step": 1948
    },
    {
      "epoch": 0.5047424816289534,
      "grad_norm": 4.989350197314329,
      "learning_rate": 7.234971833339567e-07,
      "loss": 0.3828,
      "step": 1949
    },
    {
      "epoch": 0.5050014567349713,
      "grad_norm": 4.356584758287479,
      "learning_rate": 7.229390086840627e-07,
      "loss": 0.3095,
      "step": 1950
    },
    {
      "epoch": 0.5050014567349713,
      "eval_PRM Accuracy": 0.8584905660377359,
      "eval_PRM F1": 0.906832298136646,
      "eval_PRM F1 AUC": 0.8310633839706653,
      "eval_PRM F1 AUC (fixed)": 0.7922996333158723,
      "eval_PRM F1 Neg": 0.7058823529411765,
      "eval_PRM NPV": 0.6428571428571429,
      "eval_PRM Precision": 0.9358974358974359,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.2342822551727295,
      "eval_runtime": 14.6327,
      "eval_samples_per_second": 3.007,
      "eval_steps_per_second": 0.137,
      "step": 1950
    },
    {
      "epoch": 0.5052604318409892,
      "grad_norm": 3.7530742401868706,
      "learning_rate": 7.223807539409419e-07,
      "loss": 0.3902,
      "step": 1951
    },
    {
      "epoch": 0.5055194069470073,
      "grad_norm": 3.170196963869656,
      "learning_rate": 7.218224195611279e-07,
      "loss": 0.3099,
      "step": 1952
    },
    {
      "epoch": 0.5057783820530252,
      "grad_norm": 3.978985677470167,
      "learning_rate": 7.212640060012184e-07,
      "loss": 0.3309,
      "step": 1953
    },
    {
      "epoch": 0.5060373571590431,
      "grad_norm": 5.49366863563685,
      "learning_rate": 7.207055137178772e-07,
      "loss": 0.3089,
      "step": 1954
    },
    {
      "epoch": 0.506296332265061,
      "grad_norm": 6.291045863720034,
      "learning_rate": 7.201469431678318e-07,
      "loss": 0.3901,
      "step": 1955
    },
    {
      "epoch": 0.506296332265061,
      "eval_PRM Accuracy": 0.8584905660377359,
      "eval_PRM F1": 0.906832298136646,
      "eval_PRM F1 AUC": 0.8310633839706653,
      "eval_PRM F1 AUC (fixed)": 0.7977998952331063,
      "eval_PRM F1 Neg": 0.7058823529411765,
      "eval_PRM NPV": 0.6428571428571429,
      "eval_PRM Precision": 0.9358974358974359,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.23335395753383636,
      "eval_runtime": 14.5705,
      "eval_samples_per_second": 3.02,
      "eval_steps_per_second": 0.137,
      "step": 1955
    },
    {
      "epoch": 0.5065553073710789,
      "grad_norm": 3.5881492367969616,
      "learning_rate": 7.195882948078736e-07,
      "loss": 0.3473,
      "step": 1956
    },
    {
      "epoch": 0.5068142824770969,
      "grad_norm": 4.016420150186458,
      "learning_rate": 7.190295690948582e-07,
      "loss": 0.3866,
      "step": 1957
    },
    {
      "epoch": 0.5070732575831148,
      "grad_norm": 2.9734538156884556,
      "learning_rate": 7.184707664857038e-07,
      "loss": 0.2515,
      "step": 1958
    },
    {
      "epoch": 0.5073322326891327,
      "grad_norm": 7.233224574554406,
      "learning_rate": 7.179118874373919e-07,
      "loss": 0.3226,
      "step": 1959
    },
    {
      "epoch": 0.5075912077951507,
      "grad_norm": 3.3409051496422006,
      "learning_rate": 7.173529324069666e-07,
      "loss": 0.3386,
      "step": 1960
    },
    {
      "epoch": 0.5075912077951507,
      "eval_PRM Accuracy": 0.8584905660377359,
      "eval_PRM F1": 0.906832298136646,
      "eval_PRM F1 AUC": 0.8310633839706653,
      "eval_PRM F1 AUC (fixed)": 0.8025144054478786,
      "eval_PRM F1 Neg": 0.7058823529411765,
      "eval_PRM NPV": 0.6428571428571429,
      "eval_PRM Precision": 0.9358974358974359,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.22894348204135895,
      "eval_runtime": 14.4555,
      "eval_samples_per_second": 3.044,
      "eval_steps_per_second": 0.138,
      "step": 1960
    },
    {
      "epoch": 0.5078501829011687,
      "grad_norm": 6.255980513266759,
      "learning_rate": 7.167939018515336e-07,
      "loss": 0.3296,
      "step": 1961
    },
    {
      "epoch": 0.5081091580071866,
      "grad_norm": 3.5421327405084195,
      "learning_rate": 7.162347962282609e-07,
      "loss": 0.3211,
      "step": 1962
    },
    {
      "epoch": 0.5083681331132045,
      "grad_norm": 3.7575166956029586,
      "learning_rate": 7.156756159943775e-07,
      "loss": 0.2999,
      "step": 1963
    },
    {
      "epoch": 0.5086271082192224,
      "grad_norm": 3.0050235106680683,
      "learning_rate": 7.151163616071741e-07,
      "loss": 0.3344,
      "step": 1964
    },
    {
      "epoch": 0.5088860833252403,
      "grad_norm": 4.871459919535215,
      "learning_rate": 7.145570335240008e-07,
      "loss": 0.3793,
      "step": 1965
    },
    {
      "epoch": 0.5088860833252403,
      "eval_PRM Accuracy": 0.8867924528301887,
      "eval_PRM F1": 0.926829268292683,
      "eval_PRM F1 AUC": 0.8491356731272918,
      "eval_PRM F1 AUC (fixed)": 0.8069669984284966,
      "eval_PRM F1 Neg": 0.75,
      "eval_PRM NPV": 0.72,
      "eval_PRM Precision": 0.9382716049382716,
      "eval_PRM Recall": 0.9156626506024096,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.2240259200334549,
      "eval_runtime": 14.7356,
      "eval_samples_per_second": 2.986,
      "eval_steps_per_second": 0.136,
      "step": 1965
    },
    {
      "epoch": 0.5091450584312583,
      "grad_norm": 4.697493483150097,
      "learning_rate": 7.139976322022693e-07,
      "loss": 0.3339,
      "step": 1966
    },
    {
      "epoch": 0.5094040335372763,
      "grad_norm": 7.148007632365984,
      "learning_rate": 7.134381580994504e-07,
      "loss": 0.3445,
      "step": 1967
    },
    {
      "epoch": 0.5096630086432942,
      "grad_norm": 5.098546551517129,
      "learning_rate": 7.128786116730746e-07,
      "loss": 0.3972,
      "step": 1968
    },
    {
      "epoch": 0.5099219837493121,
      "grad_norm": 2.9875694707737037,
      "learning_rate": 7.123189933807317e-07,
      "loss": 0.3332,
      "step": 1969
    },
    {
      "epoch": 0.51018095885533,
      "grad_norm": 3.328674448445375,
      "learning_rate": 7.117593036800699e-07,
      "loss": 0.2962,
      "step": 1970
    },
    {
      "epoch": 0.51018095885533,
      "eval_PRM Accuracy": 0.8584905660377359,
      "eval_PRM F1": 0.906832298136646,
      "eval_PRM F1 AUC": 0.8310633839706653,
      "eval_PRM F1 AUC (fixed)": 0.8061812467260345,
      "eval_PRM F1 Neg": 0.7058823529411765,
      "eval_PRM NPV": 0.6428571428571429,
      "eval_PRM Precision": 0.9358974358974359,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.2241746485233307,
      "eval_runtime": 14.3367,
      "eval_samples_per_second": 3.069,
      "eval_steps_per_second": 0.14,
      "step": 1970
    },
    {
      "epoch": 0.510439933961348,
      "grad_norm": 4.472354027200549,
      "learning_rate": 7.111995430287963e-07,
      "loss": 0.3792,
      "step": 1971
    },
    {
      "epoch": 0.5106989090673659,
      "grad_norm": 3.6439221844632623,
      "learning_rate": 7.106397118846757e-07,
      "loss": 0.2246,
      "step": 1972
    },
    {
      "epoch": 0.5109578841733838,
      "grad_norm": 6.424023285451845,
      "learning_rate": 7.100798107055307e-07,
      "loss": 0.3299,
      "step": 1973
    },
    {
      "epoch": 0.5112168592794017,
      "grad_norm": 3.063433753979935,
      "learning_rate": 7.095198399492406e-07,
      "loss": 0.2737,
      "step": 1974
    },
    {
      "epoch": 0.5114758343854197,
      "grad_norm": 4.793225172413915,
      "learning_rate": 7.089598000737426e-07,
      "loss": 0.3446,
      "step": 1975
    },
    {
      "epoch": 0.5114758343854197,
      "eval_PRM Accuracy": 0.8584905660377359,
      "eval_PRM F1": 0.906832298136646,
      "eval_PRM F1 AUC": 0.8310633839706653,
      "eval_PRM F1 AUC (fixed)": 0.8061812467260345,
      "eval_PRM F1 Neg": 0.7058823529411765,
      "eval_PRM NPV": 0.6428571428571429,
      "eval_PRM Precision": 0.9358974358974359,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.2251116782426834,
      "eval_runtime": 14.7645,
      "eval_samples_per_second": 2.98,
      "eval_steps_per_second": 0.135,
      "step": 1975
    },
    {
      "epoch": 0.5117348094914377,
      "grad_norm": 3.6469626418256222,
      "learning_rate": 7.083996915370296e-07,
      "loss": 0.3599,
      "step": 1976
    },
    {
      "epoch": 0.5119937845974556,
      "grad_norm": 4.0856015117149775,
      "learning_rate": 7.078395147971515e-07,
      "loss": 0.3135,
      "step": 1977
    },
    {
      "epoch": 0.5122527597034735,
      "grad_norm": 3.9583533919120897,
      "learning_rate": 7.072792703122128e-07,
      "loss": 0.4314,
      "step": 1978
    },
    {
      "epoch": 0.5125117348094914,
      "grad_norm": 4.016333899424493,
      "learning_rate": 7.067189585403742e-07,
      "loss": 0.3357,
      "step": 1979
    },
    {
      "epoch": 0.5127707099155093,
      "grad_norm": 5.310952516454567,
      "learning_rate": 7.061585799398512e-07,
      "loss": 0.4045,
      "step": 1980
    },
    {
      "epoch": 0.5127707099155093,
      "eval_PRM Accuracy": 0.8584905660377359,
      "eval_PRM F1": 0.906832298136646,
      "eval_PRM F1 AUC": 0.8310633839706653,
      "eval_PRM F1 AUC (fixed)": 0.8061812467260345,
      "eval_PRM F1 Neg": 0.7058823529411765,
      "eval_PRM NPV": 0.6428571428571429,
      "eval_PRM Precision": 0.9358974358974359,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.22383803129196167,
      "eval_runtime": 14.7046,
      "eval_samples_per_second": 2.992,
      "eval_steps_per_second": 0.136,
      "step": 1980
    },
    {
      "epoch": 0.5130296850215273,
      "grad_norm": 3.885097982351151,
      "learning_rate": 7.055981349689145e-07,
      "loss": 0.2756,
      "step": 1981
    },
    {
      "epoch": 0.5132886601275453,
      "grad_norm": 10.422904528375707,
      "learning_rate": 7.05037624085888e-07,
      "loss": 0.5268,
      "step": 1982
    },
    {
      "epoch": 0.5135476352335632,
      "grad_norm": 4.0199503005407875,
      "learning_rate": 7.044770477491505e-07,
      "loss": 0.407,
      "step": 1983
    },
    {
      "epoch": 0.5138066103395811,
      "grad_norm": 2.6508970248049764,
      "learning_rate": 7.039164064171336e-07,
      "loss": 0.2694,
      "step": 1984
    },
    {
      "epoch": 0.514065585445599,
      "grad_norm": 3.539180460581722,
      "learning_rate": 7.033557005483227e-07,
      "loss": 0.316,
      "step": 1985
    },
    {
      "epoch": 0.514065585445599,
      "eval_PRM Accuracy": 0.8584905660377359,
      "eval_PRM F1": 0.906832298136646,
      "eval_PRM F1 AUC": 0.8310633839706653,
      "eval_PRM F1 AUC (fixed)": 0.8069669984284966,
      "eval_PRM F1 Neg": 0.7058823529411765,
      "eval_PRM NPV": 0.6428571428571429,
      "eval_PRM Precision": 0.9358974358974359,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.22801996767520905,
      "eval_runtime": 14.604,
      "eval_samples_per_second": 3.013,
      "eval_steps_per_second": 0.137,
      "step": 1985
    },
    {
      "epoch": 0.514324560551617,
      "grad_norm": 5.983079771058838,
      "learning_rate": 7.027949306012553e-07,
      "loss": 0.3327,
      "step": 1986
    },
    {
      "epoch": 0.5145835356576349,
      "grad_norm": 4.197699657138563,
      "learning_rate": 7.022340970345219e-07,
      "loss": 0.3466,
      "step": 1987
    },
    {
      "epoch": 0.5148425107636528,
      "grad_norm": 3.6556760890180597,
      "learning_rate": 7.016732003067646e-07,
      "loss": 0.2702,
      "step": 1988
    },
    {
      "epoch": 0.5151014858696707,
      "grad_norm": 4.71178114730699,
      "learning_rate": 7.011122408766775e-07,
      "loss": 0.4095,
      "step": 1989
    },
    {
      "epoch": 0.5153604609756887,
      "grad_norm": 3.1234160042117463,
      "learning_rate": 7.005512192030057e-07,
      "loss": 0.3228,
      "step": 1990
    },
    {
      "epoch": 0.5153604609756887,
      "eval_PRM Accuracy": 0.8584905660377359,
      "eval_PRM F1": 0.906832298136646,
      "eval_PRM F1 AUC": 0.8310633839706653,
      "eval_PRM F1 AUC (fixed)": 0.8072289156626506,
      "eval_PRM F1 Neg": 0.7058823529411765,
      "eval_PRM NPV": 0.6428571428571429,
      "eval_PRM Precision": 0.9358974358974359,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.22756782174110413,
      "eval_runtime": 14.8077,
      "eval_samples_per_second": 2.971,
      "eval_steps_per_second": 0.135,
      "step": 1990
    },
    {
      "epoch": 0.5156194360817067,
      "grad_norm": 3.0558269274237686,
      "learning_rate": 6.999901357445454e-07,
      "loss": 0.2792,
      "step": 1991
    },
    {
      "epoch": 0.5158784111877246,
      "grad_norm": 11.217742729592652,
      "learning_rate": 6.994289909601433e-07,
      "loss": 0.4302,
      "step": 1992
    },
    {
      "epoch": 0.5161373862937425,
      "grad_norm": 4.617391772944836,
      "learning_rate": 6.988677853086959e-07,
      "loss": 0.3711,
      "step": 1993
    },
    {
      "epoch": 0.5163963613997604,
      "grad_norm": 5.061766432588815,
      "learning_rate": 6.983065192491501e-07,
      "loss": 0.3866,
      "step": 1994
    },
    {
      "epoch": 0.5166553365057783,
      "grad_norm": 5.619460823162033,
      "learning_rate": 6.977451932405016e-07,
      "loss": 0.3156,
      "step": 1995
    },
    {
      "epoch": 0.5166553365057783,
      "eval_PRM Accuracy": 0.8679245283018868,
      "eval_PRM F1": 0.9135802469135802,
      "eval_PRM F1 AUC": 0.8370874803562075,
      "eval_PRM F1 AUC (fixed)": 0.8064431639601886,
      "eval_PRM F1 Neg": 0.72,
      "eval_PRM NPV": 0.6666666666666666,
      "eval_PRM Precision": 0.9367088607594937,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.2291167825460434,
      "eval_runtime": 14.3967,
      "eval_samples_per_second": 3.056,
      "eval_steps_per_second": 0.139,
      "step": 1995
    },
    {
      "epoch": 0.5169143116117964,
      "grad_norm": 4.305794641981917,
      "learning_rate": 6.971838077417954e-07,
      "loss": 0.2998,
      "step": 1996
    },
    {
      "epoch": 0.5171732867178143,
      "grad_norm": 6.518512068960147,
      "learning_rate": 6.966223632121254e-07,
      "loss": 0.4694,
      "step": 1997
    },
    {
      "epoch": 0.5174322618238322,
      "grad_norm": 2.962729468014553,
      "learning_rate": 6.960608601106335e-07,
      "loss": 0.2479,
      "step": 1998
    },
    {
      "epoch": 0.5176912369298501,
      "grad_norm": 3.7166570901916303,
      "learning_rate": 6.954992988965089e-07,
      "loss": 0.3461,
      "step": 1999
    },
    {
      "epoch": 0.5179502120358681,
      "grad_norm": 3.160457655325982,
      "learning_rate": 6.949376800289898e-07,
      "loss": 0.2608,
      "step": 2000
    },
    {
      "epoch": 0.5179502120358681,
      "eval_PRM Accuracy": 0.8584905660377359,
      "eval_PRM F1": 0.9079754601226994,
      "eval_PRM F1 AUC": 0.8153483499214248,
      "eval_PRM F1 AUC (fixed)": 0.7988475641697224,
      "eval_PRM F1 Neg": 0.6938775510204082,
      "eval_PRM NPV": 0.6538461538461539,
      "eval_PRM Precision": 0.925,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.2310631424188614,
      "eval_runtime": 14.4545,
      "eval_samples_per_second": 3.044,
      "eval_steps_per_second": 0.138,
      "step": 2000
    },
    {
      "epoch": 0.518209187141886,
      "grad_norm": 3.2061715583921457,
      "learning_rate": 6.943760039673602e-07,
      "loss": 0.2724,
      "step": 2001
    },
    {
      "epoch": 0.5184681622479039,
      "grad_norm": 3.778193398909609,
      "learning_rate": 6.938142711709514e-07,
      "loss": 0.3118,
      "step": 2002
    },
    {
      "epoch": 0.5187271373539218,
      "grad_norm": 3.506670540864349,
      "learning_rate": 6.932524820991411e-07,
      "loss": 0.2801,
      "step": 2003
    },
    {
      "epoch": 0.5189861124599398,
      "grad_norm": 3.5026708274213334,
      "learning_rate": 6.92690637211353e-07,
      "loss": 0.3571,
      "step": 2004
    },
    {
      "epoch": 0.5192450875659578,
      "grad_norm": 10.584791254284506,
      "learning_rate": 6.921287369670564e-07,
      "loss": 0.4435,
      "step": 2005
    },
    {
      "epoch": 0.5192450875659578,
      "eval_PRM Accuracy": 0.8773584905660378,
      "eval_PRM F1": 0.9212121212121213,
      "eval_PRM F1 AUC": 0.8273965426925091,
      "eval_PRM F1 AUC (fixed)": 0.8004190675746464,
      "eval_PRM F1 Neg": 0.723404255319149,
      "eval_PRM NPV": 0.7083333333333334,
      "eval_PRM Precision": 0.926829268292683,
      "eval_PRM Recall": 0.9156626506024096,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.23249410092830658,
      "eval_runtime": 14.9452,
      "eval_samples_per_second": 2.944,
      "eval_steps_per_second": 0.134,
      "step": 2005
    },
    {
      "epoch": 0.5195040626719757,
      "grad_norm": 3.0760917762678393,
      "learning_rate": 6.915667818257661e-07,
      "loss": 0.2914,
      "step": 2006
    },
    {
      "epoch": 0.5197630377779936,
      "grad_norm": 6.133884437864586,
      "learning_rate": 6.910047722470413e-07,
      "loss": 0.3558,
      "step": 2007
    },
    {
      "epoch": 0.5200220128840115,
      "grad_norm": 8.729669508602132,
      "learning_rate": 6.904427086904859e-07,
      "loss": 0.4452,
      "step": 2008
    },
    {
      "epoch": 0.5202809879900294,
      "grad_norm": 9.636133173288515,
      "learning_rate": 6.898805916157484e-07,
      "loss": 0.3104,
      "step": 2009
    },
    {
      "epoch": 0.5205399630960474,
      "grad_norm": 4.797878966268347,
      "learning_rate": 6.893184214825206e-07,
      "loss": 0.3391,
      "step": 2010
    },
    {
      "epoch": 0.5205399630960474,
      "eval_PRM Accuracy": 0.8679245283018868,
      "eval_PRM F1": 0.9146341463414634,
      "eval_PRM F1 AUC": 0.8213724463069669,
      "eval_PRM F1 AUC (fixed)": 0.8004190675746464,
      "eval_PRM F1 Neg": 0.7083333333333334,
      "eval_PRM NPV": 0.68,
      "eval_PRM Precision": 0.9259259259259259,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.2306734025478363,
      "eval_runtime": 14.7172,
      "eval_samples_per_second": 2.99,
      "eval_steps_per_second": 0.136,
      "step": 2010
    },
    {
      "epoch": 0.5207989382020654,
      "grad_norm": 4.32422536050864,
      "learning_rate": 6.887561987505377e-07,
      "loss": 0.3406,
      "step": 2011
    },
    {
      "epoch": 0.5210579133080833,
      "grad_norm": 3.5272674414179814,
      "learning_rate": 6.881939238795782e-07,
      "loss": 0.2721,
      "step": 2012
    },
    {
      "epoch": 0.5213168884141012,
      "grad_norm": 5.120390312280001,
      "learning_rate": 6.876315973294628e-07,
      "loss": 0.3217,
      "step": 2013
    },
    {
      "epoch": 0.5215758635201191,
      "grad_norm": 3.399780207297747,
      "learning_rate": 6.870692195600549e-07,
      "loss": 0.2919,
      "step": 2014
    },
    {
      "epoch": 0.5218348386261371,
      "grad_norm": 3.578991259717659,
      "learning_rate": 6.865067910312596e-07,
      "loss": 0.3383,
      "step": 2015
    },
    {
      "epoch": 0.5218348386261371,
      "eval_PRM Accuracy": 0.8584905660377359,
      "eval_PRM F1": 0.906832298136646,
      "eval_PRM F1 AUC": 0.8310633839706653,
      "eval_PRM F1 AUC (fixed)": 0.8004190675746464,
      "eval_PRM F1 Neg": 0.7058823529411765,
      "eval_PRM NPV": 0.6428571428571429,
      "eval_PRM Precision": 0.9358974358974359,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.2322913557291031,
      "eval_runtime": 14.9883,
      "eval_samples_per_second": 2.936,
      "eval_steps_per_second": 0.133,
      "step": 2015
    },
    {
      "epoch": 0.522093813732155,
      "grad_norm": 3.2428141265055768,
      "learning_rate": 6.859443122030236e-07,
      "loss": 0.2898,
      "step": 2016
    },
    {
      "epoch": 0.5223527888381729,
      "grad_norm": 3.7615110198537063,
      "learning_rate": 6.853817835353344e-07,
      "loss": 0.3854,
      "step": 2017
    },
    {
      "epoch": 0.5226117639441908,
      "grad_norm": 7.48330466691377,
      "learning_rate": 6.848192054882207e-07,
      "loss": 0.3158,
      "step": 2018
    },
    {
      "epoch": 0.5228707390502088,
      "grad_norm": 4.108214890720237,
      "learning_rate": 6.842565785217513e-07,
      "loss": 0.4096,
      "step": 2019
    },
    {
      "epoch": 0.5231297141562268,
      "grad_norm": 3.0318620718932108,
      "learning_rate": 6.836939030960351e-07,
      "loss": 0.2582,
      "step": 2020
    },
    {
      "epoch": 0.5231297141562268,
      "eval_PRM Accuracy": 0.8584905660377359,
      "eval_PRM F1": 0.906832298136646,
      "eval_PRM F1 AUC": 0.8310633839706653,
      "eval_PRM F1 AUC (fixed)": 0.8017286537454165,
      "eval_PRM F1 Neg": 0.7058823529411765,
      "eval_PRM NPV": 0.6428571428571429,
      "eval_PRM Precision": 0.9358974358974359,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.23562628030776978,
      "eval_runtime": 14.959,
      "eval_samples_per_second": 2.941,
      "eval_steps_per_second": 0.134,
      "step": 2020
    },
    {
      "epoch": 0.5233886892622447,
      "grad_norm": 5.604343382295727,
      "learning_rate": 6.831311796712209e-07,
      "loss": 0.3478,
      "step": 2021
    },
    {
      "epoch": 0.5236476643682626,
      "grad_norm": 4.307305798786338,
      "learning_rate": 6.82568408707496e-07,
      "loss": 0.3431,
      "step": 2022
    },
    {
      "epoch": 0.5239066394742805,
      "grad_norm": 4.9450957065202115,
      "learning_rate": 6.820055906650877e-07,
      "loss": 0.3028,
      "step": 2023
    },
    {
      "epoch": 0.5241656145802984,
      "grad_norm": 3.3113193411423176,
      "learning_rate": 6.814427260042603e-07,
      "loss": 0.3598,
      "step": 2024
    },
    {
      "epoch": 0.5244245896863164,
      "grad_norm": 4.713896577692382,
      "learning_rate": 6.808798151853177e-07,
      "loss": 0.3413,
      "step": 2025
    },
    {
      "epoch": 0.5244245896863164,
      "eval_PRM Accuracy": 0.8584905660377359,
      "eval_PRM F1": 0.906832298136646,
      "eval_PRM F1 AUC": 0.8310633839706653,
      "eval_PRM F1 AUC (fixed)": 0.8040859088528025,
      "eval_PRM F1 Neg": 0.7058823529411765,
      "eval_PRM NPV": 0.6428571428571429,
      "eval_PRM Precision": 0.9358974358974359,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.2365666925907135,
      "eval_runtime": 15.4542,
      "eval_samples_per_second": 2.847,
      "eval_steps_per_second": 0.129,
      "step": 2025
    },
    {
      "epoch": 0.5246835647923344,
      "grad_norm": 4.274396697787456,
      "learning_rate": 6.803168586686007e-07,
      "loss": 0.2918,
      "step": 2026
    },
    {
      "epoch": 0.5249425398983523,
      "grad_norm": 2.788426991033734,
      "learning_rate": 6.79753856914488e-07,
      "loss": 0.284,
      "step": 2027
    },
    {
      "epoch": 0.5252015150043702,
      "grad_norm": 2.739642908055344,
      "learning_rate": 6.791908103833946e-07,
      "loss": 0.2675,
      "step": 2028
    },
    {
      "epoch": 0.5254604901103881,
      "grad_norm": 3.8106267406958705,
      "learning_rate": 6.786277195357727e-07,
      "loss": 0.2744,
      "step": 2029
    },
    {
      "epoch": 0.5257194652164061,
      "grad_norm": 5.625164632340796,
      "learning_rate": 6.780645848321104e-07,
      "loss": 0.4715,
      "step": 2030
    },
    {
      "epoch": 0.5257194652164061,
      "eval_PRM Accuracy": 0.8490566037735849,
      "eval_PRM F1": 0.9012345679012346,
      "eval_PRM F1 AUC": 0.8093242535358826,
      "eval_PRM F1 AUC (fixed)": 0.8053954950235725,
      "eval_PRM F1 Neg": 0.68,
      "eval_PRM NPV": 0.6296296296296297,
      "eval_PRM Precision": 0.9240506329113924,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.23772607743740082,
      "eval_runtime": 14.5648,
      "eval_samples_per_second": 3.021,
      "eval_steps_per_second": 0.137,
      "step": 2030
    },
    {
      "epoch": 0.525978440322424,
      "grad_norm": 3.074981358586057,
      "learning_rate": 6.775014067329322e-07,
      "loss": 0.2841,
      "step": 2031
    },
    {
      "epoch": 0.5262374154284419,
      "grad_norm": 2.9613004448677094,
      "learning_rate": 6.769381856987974e-07,
      "loss": 0.2888,
      "step": 2032
    },
    {
      "epoch": 0.5264963905344598,
      "grad_norm": 3.668184885026715,
      "learning_rate": 6.763749221903007e-07,
      "loss": 0.2671,
      "step": 2033
    },
    {
      "epoch": 0.5267553656404778,
      "grad_norm": 4.699499618822293,
      "learning_rate": 6.758116166680715e-07,
      "loss": 0.3835,
      "step": 2034
    },
    {
      "epoch": 0.5270143407464958,
      "grad_norm": 4.81171930848692,
      "learning_rate": 6.75248269592774e-07,
      "loss": 0.4363,
      "step": 2035
    },
    {
      "epoch": 0.5270143407464958,
      "eval_PRM Accuracy": 0.8490566037735849,
      "eval_PRM F1": 0.9012345679012346,
      "eval_PRM F1 AUC": 0.8093242535358826,
      "eval_PRM F1 AUC (fixed)": 0.8040859088528025,
      "eval_PRM F1 Neg": 0.68,
      "eval_PRM NPV": 0.6296296296296297,
      "eval_PRM Precision": 0.9240506329113924,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.23885823786258698,
      "eval_runtime": 14.8987,
      "eval_samples_per_second": 2.953,
      "eval_steps_per_second": 0.134,
      "step": 2035
    },
    {
      "epoch": 0.5272733158525137,
      "grad_norm": 3.9014169465534265,
      "learning_rate": 6.746848814251055e-07,
      "loss": 0.361,
      "step": 2036
    },
    {
      "epoch": 0.5275322909585316,
      "grad_norm": 4.3606932506239175,
      "learning_rate": 6.741214526257977e-07,
      "loss": 0.2537,
      "step": 2037
    },
    {
      "epoch": 0.5277912660645495,
      "grad_norm": 2.7459469503011085,
      "learning_rate": 6.735579836556151e-07,
      "loss": 0.2469,
      "step": 2038
    },
    {
      "epoch": 0.5280502411705674,
      "grad_norm": 7.848629951438361,
      "learning_rate": 6.729944749753553e-07,
      "loss": 0.425,
      "step": 2039
    },
    {
      "epoch": 0.5283092162765854,
      "grad_norm": 7.727017414472788,
      "learning_rate": 6.724309270458482e-07,
      "loss": 0.3805,
      "step": 2040
    },
    {
      "epoch": 0.5283092162765854,
      "eval_PRM Accuracy": 0.8490566037735849,
      "eval_PRM F1": 0.9012345679012346,
      "eval_PRM F1 AUC": 0.8093242535358826,
      "eval_PRM F1 AUC (fixed)": 0.8090623363017286,
      "eval_PRM F1 Neg": 0.68,
      "eval_PRM NPV": 0.6296296296296297,
      "eval_PRM Precision": 0.9240506329113924,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.24057665467262268,
      "eval_runtime": 15.1599,
      "eval_samples_per_second": 2.902,
      "eval_steps_per_second": 0.132,
      "step": 2040
    },
    {
      "epoch": 0.5285681913826034,
      "grad_norm": 4.0402233736534905,
      "learning_rate": 6.718673403279558e-07,
      "loss": 0.3427,
      "step": 2041
    },
    {
      "epoch": 0.5288271664886213,
      "grad_norm": 5.314057627737245,
      "learning_rate": 6.713037152825718e-07,
      "loss": 0.3512,
      "step": 2042
    },
    {
      "epoch": 0.5290861415946392,
      "grad_norm": 3.6578861001910927,
      "learning_rate": 6.707400523706216e-07,
      "loss": 0.2235,
      "step": 2043
    },
    {
      "epoch": 0.5293451167006572,
      "grad_norm": 4.086990485774963,
      "learning_rate": 6.701763520530609e-07,
      "loss": 0.3817,
      "step": 2044
    },
    {
      "epoch": 0.5296040918066751,
      "grad_norm": 5.2097591245407004,
      "learning_rate": 6.696126147908767e-07,
      "loss": 0.3781,
      "step": 2045
    },
    {
      "epoch": 0.5296040918066751,
      "eval_PRM Accuracy": 0.8490566037735849,
      "eval_PRM F1": 0.9012345679012346,
      "eval_PRM F1 AUC": 0.8093242535358826,
      "eval_PRM F1 AUC (fixed)": 0.8030382399161864,
      "eval_PRM F1 Neg": 0.68,
      "eval_PRM NPV": 0.6296296296296297,
      "eval_PRM Precision": 0.9240506329113924,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.23904496431350708,
      "eval_runtime": 14.7312,
      "eval_samples_per_second": 2.987,
      "eval_steps_per_second": 0.136,
      "step": 2045
    },
    {
      "epoch": 0.529863066912693,
      "grad_norm": 3.6676471864099187,
      "learning_rate": 6.690488410450858e-07,
      "loss": 0.3037,
      "step": 2046
    },
    {
      "epoch": 0.5301220420187109,
      "grad_norm": 3.4077188564554763,
      "learning_rate": 6.684850312767346e-07,
      "loss": 0.3151,
      "step": 2047
    },
    {
      "epoch": 0.5303810171247288,
      "grad_norm": 4.0960690466874405,
      "learning_rate": 6.679211859468997e-07,
      "loss": 0.3264,
      "step": 2048
    },
    {
      "epoch": 0.5306399922307469,
      "grad_norm": 6.5931766149599165,
      "learning_rate": 6.673573055166862e-07,
      "loss": 0.3726,
      "step": 2049
    },
    {
      "epoch": 0.5308989673367648,
      "grad_norm": 5.359954016263425,
      "learning_rate": 6.667933904472281e-07,
      "loss": 0.3984,
      "step": 2050
    },
    {
      "epoch": 0.5308989673367648,
      "eval_PRM Accuracy": 0.8490566037735849,
      "eval_PRM F1": 0.9,
      "eval_PRM F1 AUC": 0.8250392875851232,
      "eval_PRM F1 AUC (fixed)": 0.8048716605552646,
      "eval_PRM F1 Neg": 0.6923076923076923,
      "eval_PRM NPV": 0.6206896551724138,
      "eval_PRM Precision": 0.935064935064935,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.23912553489208221,
      "eval_runtime": 14.6669,
      "eval_samples_per_second": 3.0,
      "eval_steps_per_second": 0.136,
      "step": 2050
    },
    {
      "epoch": 0.5311579424427827,
      "grad_norm": 5.113226620551563,
      "learning_rate": 6.662294411996875e-07,
      "loss": 0.3139,
      "step": 2051
    },
    {
      "epoch": 0.5314169175488006,
      "grad_norm": 6.878043278298198,
      "learning_rate": 6.656654582352549e-07,
      "loss": 0.4242,
      "step": 2052
    },
    {
      "epoch": 0.5316758926548185,
      "grad_norm": 4.46387473685755,
      "learning_rate": 6.651014420151478e-07,
      "loss": 0.3279,
      "step": 2053
    },
    {
      "epoch": 0.5319348677608365,
      "grad_norm": 3.851585184773988,
      "learning_rate": 6.645373930006117e-07,
      "loss": 0.2896,
      "step": 2054
    },
    {
      "epoch": 0.5321938428668544,
      "grad_norm": 4.2827990079976175,
      "learning_rate": 6.639733116529181e-07,
      "loss": 0.3265,
      "step": 2055
    },
    {
      "epoch": 0.5321938428668544,
      "eval_PRM Accuracy": 0.8490566037735849,
      "eval_PRM F1": 0.9,
      "eval_PRM F1 AUC": 0.8250392875851232,
      "eval_PRM F1 AUC (fixed)": 0.8027763226820326,
      "eval_PRM F1 Neg": 0.6923076923076923,
      "eval_PRM NPV": 0.6206896551724138,
      "eval_PRM Precision": 0.935064935064935,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.23745904862880707,
      "eval_runtime": 14.9739,
      "eval_samples_per_second": 2.938,
      "eval_steps_per_second": 0.134,
      "step": 2055
    },
    {
      "epoch": 0.5324528179728724,
      "grad_norm": 4.5197844152093785,
      "learning_rate": 6.634091984333654e-07,
      "loss": 0.262,
      "step": 2056
    },
    {
      "epoch": 0.5327117930788903,
      "grad_norm": 3.5035410775267506,
      "learning_rate": 6.628450538032776e-07,
      "loss": 0.2829,
      "step": 2057
    },
    {
      "epoch": 0.5329707681849082,
      "grad_norm": 3.380795789228792,
      "learning_rate": 6.62280878224005e-07,
      "loss": 0.2794,
      "step": 2058
    },
    {
      "epoch": 0.5332297432909262,
      "grad_norm": 5.3953214832767635,
      "learning_rate": 6.617166721569231e-07,
      "loss": 0.3279,
      "step": 2059
    },
    {
      "epoch": 0.5334887183969441,
      "grad_norm": 5.944588390066497,
      "learning_rate": 6.611524360634319e-07,
      "loss": 0.3302,
      "step": 2060
    },
    {
      "epoch": 0.5334887183969441,
      "eval_PRM Accuracy": 0.8490566037735849,
      "eval_PRM F1": 0.9012345679012346,
      "eval_PRM F1 AUC": 0.8093242535358826,
      "eval_PRM F1 AUC (fixed)": 0.8048716605552645,
      "eval_PRM F1 Neg": 0.68,
      "eval_PRM NPV": 0.6296296296296297,
      "eval_PRM Precision": 0.9240506329113924,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.2332223504781723,
      "eval_runtime": 14.7553,
      "eval_samples_per_second": 2.982,
      "eval_steps_per_second": 0.136,
      "step": 2060
    },
    {
      "epoch": 0.533747693502962,
      "grad_norm": 5.9358873488695645,
      "learning_rate": 6.605881704049563e-07,
      "loss": 0.3577,
      "step": 2061
    },
    {
      "epoch": 0.5340066686089799,
      "grad_norm": 3.475541086117722,
      "learning_rate": 6.600238756429453e-07,
      "loss": 0.2575,
      "step": 2062
    },
    {
      "epoch": 0.5342656437149979,
      "grad_norm": 3.77209182431562,
      "learning_rate": 6.594595522388717e-07,
      "loss": 0.3014,
      "step": 2063
    },
    {
      "epoch": 0.5345246188210159,
      "grad_norm": 5.198027733972152,
      "learning_rate": 6.588952006542315e-07,
      "loss": 0.3867,
      "step": 2064
    },
    {
      "epoch": 0.5347835939270338,
      "grad_norm": 4.920148154606656,
      "learning_rate": 6.583308213505443e-07,
      "loss": 0.3741,
      "step": 2065
    },
    {
      "epoch": 0.5347835939270338,
      "eval_PRM Accuracy": 0.8490566037735849,
      "eval_PRM F1": 0.9012345679012346,
      "eval_PRM F1 AUC": 0.8093242535358826,
      "eval_PRM F1 AUC (fixed)": 0.8059193294918806,
      "eval_PRM F1 Neg": 0.68,
      "eval_PRM NPV": 0.6296296296296297,
      "eval_PRM Precision": 0.9240506329113924,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.2303762584924698,
      "eval_runtime": 14.2865,
      "eval_samples_per_second": 3.08,
      "eval_steps_per_second": 0.14,
      "step": 2065
    },
    {
      "epoch": 0.5350425690330517,
      "grad_norm": 3.1161871805244012,
      "learning_rate": 6.57766414789352e-07,
      "loss": 0.3141,
      "step": 2066
    },
    {
      "epoch": 0.5353015441390696,
      "grad_norm": 4.258492916030331,
      "learning_rate": 6.572019814322186e-07,
      "loss": 0.2905,
      "step": 2067
    },
    {
      "epoch": 0.5355605192450875,
      "grad_norm": 4.206402161450609,
      "learning_rate": 6.566375217407306e-07,
      "loss": 0.2897,
      "step": 2068
    },
    {
      "epoch": 0.5358194943511055,
      "grad_norm": 3.280595461725414,
      "learning_rate": 6.560730361764953e-07,
      "loss": 0.3508,
      "step": 2069
    },
    {
      "epoch": 0.5360784694571235,
      "grad_norm": 4.022051625501909,
      "learning_rate": 6.555085252011415e-07,
      "loss": 0.4384,
      "step": 2070
    },
    {
      "epoch": 0.5360784694571235,
      "eval_PRM Accuracy": 0.8584905660377359,
      "eval_PRM F1": 0.906832298136646,
      "eval_PRM F1 AUC": 0.8310633839706653,
      "eval_PRM F1 AUC (fixed)": 0.8038239916186485,
      "eval_PRM F1 Neg": 0.7058823529411765,
      "eval_PRM NPV": 0.6428571428571429,
      "eval_PRM Precision": 0.9358974358974359,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.23042291402816772,
      "eval_runtime": 14.624,
      "eval_samples_per_second": 3.009,
      "eval_steps_per_second": 0.137,
      "step": 2070
    },
    {
      "epoch": 0.5363374445631414,
      "grad_norm": 4.291897306295291,
      "learning_rate": 6.549439892763193e-07,
      "loss": 0.2319,
      "step": 2071
    },
    {
      "epoch": 0.5365964196691593,
      "grad_norm": 3.8251486317216434,
      "learning_rate": 6.543794288636983e-07,
      "loss": 0.3476,
      "step": 2072
    },
    {
      "epoch": 0.5368553947751772,
      "grad_norm": 4.709348869922407,
      "learning_rate": 6.538148444249687e-07,
      "loss": 0.2893,
      "step": 2073
    },
    {
      "epoch": 0.5371143698811952,
      "grad_norm": 3.617750986516083,
      "learning_rate": 6.532502364218402e-07,
      "loss": 0.3345,
      "step": 2074
    },
    {
      "epoch": 0.5373733449872131,
      "grad_norm": 5.489481909014706,
      "learning_rate": 6.526856053160415e-07,
      "loss": 0.3046,
      "step": 2075
    },
    {
      "epoch": 0.5373733449872131,
      "eval_PRM Accuracy": 0.8490566037735849,
      "eval_PRM F1": 0.9012345679012346,
      "eval_PRM F1 AUC": 0.8093242535358826,
      "eval_PRM F1 AUC (fixed)": 0.8025144054478786,
      "eval_PRM F1 Neg": 0.68,
      "eval_PRM NPV": 0.6296296296296297,
      "eval_PRM Precision": 0.9240506329113924,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.22988790273666382,
      "eval_runtime": 14.6497,
      "eval_samples_per_second": 3.003,
      "eval_steps_per_second": 0.137,
      "step": 2075
    },
    {
      "epoch": 0.537632320093231,
      "grad_norm": 4.444267827033878,
      "learning_rate": 6.52120951569321e-07,
      "loss": 0.3115,
      "step": 2076
    },
    {
      "epoch": 0.5378912951992489,
      "grad_norm": 3.014636283976347,
      "learning_rate": 6.515562756434447e-07,
      "loss": 0.335,
      "step": 2077
    },
    {
      "epoch": 0.5381502703052669,
      "grad_norm": 3.4280506268974564,
      "learning_rate": 6.509915780001969e-07,
      "loss": 0.3318,
      "step": 2078
    },
    {
      "epoch": 0.5384092454112849,
      "grad_norm": 8.846339510445935,
      "learning_rate": 6.504268591013803e-07,
      "loss": 0.4776,
      "step": 2079
    },
    {
      "epoch": 0.5386682205173028,
      "grad_norm": 3.5750039215362404,
      "learning_rate": 6.498621194088143e-07,
      "loss": 0.3052,
      "step": 2080
    },
    {
      "epoch": 0.5386682205173028,
      "eval_PRM Accuracy": 0.8490566037735849,
      "eval_PRM F1": 0.9012345679012346,
      "eval_PRM F1 AUC": 0.8093242535358826,
      "eval_PRM F1 AUC (fixed)": 0.8033001571503405,
      "eval_PRM F1 Neg": 0.68,
      "eval_PRM NPV": 0.6296296296296297,
      "eval_PRM Precision": 0.9240506329113924,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.22800038754940033,
      "eval_runtime": 14.4667,
      "eval_samples_per_second": 3.041,
      "eval_steps_per_second": 0.138,
      "step": 2080
    },
    {
      "epoch": 0.5389271956233207,
      "grad_norm": 3.3847015171094434,
      "learning_rate": 6.492973593843359e-07,
      "loss": 0.2751,
      "step": 2081
    },
    {
      "epoch": 0.5391861707293386,
      "grad_norm": 2.4474866586435424,
      "learning_rate": 6.487325794897978e-07,
      "loss": 0.2773,
      "step": 2082
    },
    {
      "epoch": 0.5394451458353565,
      "grad_norm": 3.506818566327238,
      "learning_rate": 6.481677801870699e-07,
      "loss": 0.3219,
      "step": 2083
    },
    {
      "epoch": 0.5397041209413745,
      "grad_norm": 3.844019830929198,
      "learning_rate": 6.476029619380376e-07,
      "loss": 0.2721,
      "step": 2084
    },
    {
      "epoch": 0.5399630960473925,
      "grad_norm": 4.735449386114506,
      "learning_rate": 6.470381252046019e-07,
      "loss": 0.3378,
      "step": 2085
    },
    {
      "epoch": 0.5399630960473925,
      "eval_PRM Accuracy": 0.8584905660377359,
      "eval_PRM F1": 0.906832298136646,
      "eval_PRM F1 AUC": 0.8310633839706653,
      "eval_PRM F1 AUC (fixed)": 0.8046097433211106,
      "eval_PRM F1 Neg": 0.7058823529411765,
      "eval_PRM NPV": 0.6428571428571429,
      "eval_PRM Precision": 0.9358974358974359,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.22756992280483246,
      "eval_runtime": 14.5471,
      "eval_samples_per_second": 3.025,
      "eval_steps_per_second": 0.137,
      "step": 2085
    },
    {
      "epoch": 0.5402220711534104,
      "grad_norm": 5.599981671270099,
      "learning_rate": 6.464732704486785e-07,
      "loss": 0.3872,
      "step": 2086
    },
    {
      "epoch": 0.5404810462594283,
      "grad_norm": 3.8197096162208353,
      "learning_rate": 6.459083981321984e-07,
      "loss": 0.2716,
      "step": 2087
    },
    {
      "epoch": 0.5407400213654463,
      "grad_norm": 5.135451291709805,
      "learning_rate": 6.453435087171064e-07,
      "loss": 0.3113,
      "step": 2088
    },
    {
      "epoch": 0.5409989964714642,
      "grad_norm": 4.67259568023336,
      "learning_rate": 6.447786026653623e-07,
      "loss": 0.2986,
      "step": 2089
    },
    {
      "epoch": 0.5412579715774821,
      "grad_norm": 5.29178364553776,
      "learning_rate": 6.442136804389381e-07,
      "loss": 0.2983,
      "step": 2090
    },
    {
      "epoch": 0.5412579715774821,
      "eval_PRM Accuracy": 0.8490566037735849,
      "eval_PRM F1": 0.9012345679012346,
      "eval_PRM F1 AUC": 0.8093242535358826,
      "eval_PRM F1 AUC (fixed)": 0.8114195914091147,
      "eval_PRM F1 Neg": 0.68,
      "eval_PRM NPV": 0.6296296296296297,
      "eval_PRM Precision": 0.9240506329113924,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.22176438570022583,
      "eval_runtime": 15.1537,
      "eval_samples_per_second": 2.904,
      "eval_steps_per_second": 0.132,
      "step": 2090
    },
    {
      "epoch": 0.5415169466835,
      "grad_norm": 7.350947918045169,
      "learning_rate": 6.436487424998199e-07,
      "loss": 0.2728,
      "step": 2091
    },
    {
      "epoch": 0.541775921789518,
      "grad_norm": 4.782952929924918,
      "learning_rate": 6.430837893100065e-07,
      "loss": 0.3695,
      "step": 2092
    },
    {
      "epoch": 0.542034896895536,
      "grad_norm": 3.192774878599221,
      "learning_rate": 6.425188213315095e-07,
      "loss": 0.274,
      "step": 2093
    },
    {
      "epoch": 0.5422938720015539,
      "grad_norm": 5.348528141913388,
      "learning_rate": 6.419538390263517e-07,
      "loss": 0.3629,
      "step": 2094
    },
    {
      "epoch": 0.5425528471075718,
      "grad_norm": 3.282991174687198,
      "learning_rate": 6.413888428565684e-07,
      "loss": 0.2911,
      "step": 2095
    },
    {
      "epoch": 0.5425528471075718,
      "eval_PRM Accuracy": 0.8679245283018868,
      "eval_PRM F1": 0.9146341463414634,
      "eval_PRM F1 AUC": 0.8213724463069669,
      "eval_PRM F1 AUC (fixed)": 0.823991618648507,
      "eval_PRM F1 Neg": 0.7083333333333334,
      "eval_PRM NPV": 0.68,
      "eval_PRM Precision": 0.9259259259259259,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.2215135246515274,
      "eval_runtime": 14.4124,
      "eval_samples_per_second": 3.053,
      "eval_steps_per_second": 0.139,
      "step": 2095
    },
    {
      "epoch": 0.5428118222135897,
      "grad_norm": 5.090130955007299,
      "learning_rate": 6.408238332842063e-07,
      "loss": 0.2991,
      "step": 2096
    },
    {
      "epoch": 0.5430707973196076,
      "grad_norm": 4.312208280057902,
      "learning_rate": 6.402588107713222e-07,
      "loss": 0.3835,
      "step": 2097
    },
    {
      "epoch": 0.5433297724256256,
      "grad_norm": 4.417074211468511,
      "learning_rate": 6.396937757799844e-07,
      "loss": 0.3476,
      "step": 2098
    },
    {
      "epoch": 0.5435887475316435,
      "grad_norm": 3.136210037797205,
      "learning_rate": 6.391287287722708e-07,
      "loss": 0.2704,
      "step": 2099
    },
    {
      "epoch": 0.5438477226376615,
      "grad_norm": 3.985730790683825,
      "learning_rate": 6.385636702102697e-07,
      "loss": 0.3407,
      "step": 2100
    },
    {
      "epoch": 0.5438477226376615,
      "eval_PRM Accuracy": 0.8773584905660378,
      "eval_PRM F1": 0.9212121212121213,
      "eval_PRM F1 AUC": 0.8273965426925091,
      "eval_PRM F1 AUC (fixed)": 0.8237297014143531,
      "eval_PRM F1 Neg": 0.723404255319149,
      "eval_PRM NPV": 0.7083333333333334,
      "eval_PRM Precision": 0.926829268292683,
      "eval_PRM Recall": 0.9156626506024096,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.22097660601139069,
      "eval_runtime": 14.8888,
      "eval_samples_per_second": 2.955,
      "eval_steps_per_second": 0.134,
      "step": 2100
    },
    {
      "epoch": 0.5441066977436794,
      "grad_norm": 6.365696198016137,
      "learning_rate": 6.379986005560785e-07,
      "loss": 0.2917,
      "step": 2101
    },
    {
      "epoch": 0.5443656728496973,
      "grad_norm": 6.409344303820183,
      "learning_rate": 6.374335202718033e-07,
      "loss": 0.3836,
      "step": 2102
    },
    {
      "epoch": 0.5446246479557153,
      "grad_norm": 7.194404909564871,
      "learning_rate": 6.368684298195598e-07,
      "loss": 0.2731,
      "step": 2103
    },
    {
      "epoch": 0.5448836230617332,
      "grad_norm": 4.03287235899355,
      "learning_rate": 6.363033296614711e-07,
      "loss": 0.3302,
      "step": 2104
    },
    {
      "epoch": 0.5451425981677511,
      "grad_norm": 5.17210115897139,
      "learning_rate": 6.357382202596687e-07,
      "loss": 0.2646,
      "step": 2105
    },
    {
      "epoch": 0.5451425981677511,
      "eval_PRM Accuracy": 0.8679245283018868,
      "eval_PRM F1": 0.9146341463414634,
      "eval_PRM F1 AUC": 0.8213724463069669,
      "eval_PRM F1 AUC (fixed)": 0.8184913567312729,
      "eval_PRM F1 Neg": 0.7083333333333334,
      "eval_PRM NPV": 0.68,
      "eval_PRM Precision": 0.9259259259259259,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.2218240201473236,
      "eval_runtime": 14.6176,
      "eval_samples_per_second": 3.01,
      "eval_steps_per_second": 0.137,
      "step": 2105
    },
    {
      "epoch": 0.545401573273769,
      "grad_norm": 3.5165925057228735,
      "learning_rate": 6.351731020762921e-07,
      "loss": 0.3177,
      "step": 2106
    },
    {
      "epoch": 0.545660548379787,
      "grad_norm": 3.604672100685866,
      "learning_rate": 6.346079755734866e-07,
      "loss": 0.3678,
      "step": 2107
    },
    {
      "epoch": 0.545919523485805,
      "grad_norm": 5.891185649487258,
      "learning_rate": 6.340428412134057e-07,
      "loss": 0.4143,
      "step": 2108
    },
    {
      "epoch": 0.5461784985918229,
      "grad_norm": 4.185357041957105,
      "learning_rate": 6.33477699458209e-07,
      "loss": 0.3495,
      "step": 2109
    },
    {
      "epoch": 0.5464374736978408,
      "grad_norm": 3.5320750068708713,
      "learning_rate": 6.329125507700618e-07,
      "loss": 0.3324,
      "step": 2110
    },
    {
      "epoch": 0.5464374736978408,
      "eval_PRM Accuracy": 0.8679245283018868,
      "eval_PRM F1": 0.9146341463414634,
      "eval_PRM F1 AUC": 0.8213724463069669,
      "eval_PRM F1 AUC (fixed)": 0.8166579360921948,
      "eval_PRM F1 Neg": 0.7083333333333334,
      "eval_PRM NPV": 0.68,
      "eval_PRM Precision": 0.9259259259259259,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.2260075956583023,
      "eval_runtime": 14.848,
      "eval_samples_per_second": 2.963,
      "eval_steps_per_second": 0.135,
      "step": 2110
    },
    {
      "epoch": 0.5466964488038587,
      "grad_norm": 6.217311998900788,
      "learning_rate": 6.32347395611135e-07,
      "loss": 0.3247,
      "step": 2111
    },
    {
      "epoch": 0.5469554239098766,
      "grad_norm": 4.666248326951082,
      "learning_rate": 6.317822344436049e-07,
      "loss": 0.2588,
      "step": 2112
    },
    {
      "epoch": 0.5472143990158946,
      "grad_norm": 4.120618869933418,
      "learning_rate": 6.312170677296533e-07,
      "loss": 0.3302,
      "step": 2113
    },
    {
      "epoch": 0.5474733741219125,
      "grad_norm": 3.604890814250355,
      "learning_rate": 6.306518959314662e-07,
      "loss": 0.2721,
      "step": 2114
    },
    {
      "epoch": 0.5477323492279305,
      "grad_norm": 2.5907657234593917,
      "learning_rate": 6.300867195112331e-07,
      "loss": 0.2649,
      "step": 2115
    },
    {
      "epoch": 0.5477323492279305,
      "eval_PRM Accuracy": 0.8584905660377359,
      "eval_PRM F1": 0.9079754601226994,
      "eval_PRM F1 AUC": 0.8153483499214248,
      "eval_PRM F1 AUC (fixed)": 0.8143006809848088,
      "eval_PRM F1 Neg": 0.6938775510204082,
      "eval_PRM NPV": 0.6538461538461539,
      "eval_PRM Precision": 0.925,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.2314336895942688,
      "eval_runtime": 14.6069,
      "eval_samples_per_second": 3.012,
      "eval_steps_per_second": 0.137,
      "step": 2115
    },
    {
      "epoch": 0.5479913243339484,
      "grad_norm": 3.0134554046606175,
      "learning_rate": 6.295215389311479e-07,
      "loss": 0.2883,
      "step": 2116
    },
    {
      "epoch": 0.5482502994399663,
      "grad_norm": 3.6809965143965764,
      "learning_rate": 6.289563546534079e-07,
      "loss": 0.3115,
      "step": 2117
    },
    {
      "epoch": 0.5485092745459843,
      "grad_norm": 4.682479721212389,
      "learning_rate": 6.283911671402136e-07,
      "loss": 0.32,
      "step": 2118
    },
    {
      "epoch": 0.5487682496520022,
      "grad_norm": 2.902815885637313,
      "learning_rate": 6.278259768537674e-07,
      "loss": 0.3453,
      "step": 2119
    },
    {
      "epoch": 0.5490272247580201,
      "grad_norm": 4.858166924362263,
      "learning_rate": 6.272607842562747e-07,
      "loss": 0.3338,
      "step": 2120
    },
    {
      "epoch": 0.5490272247580201,
      "eval_PRM Accuracy": 0.8584905660377359,
      "eval_PRM F1": 0.9079754601226994,
      "eval_PRM F1 AUC": 0.8153483499214248,
      "eval_PRM F1 AUC (fixed)": 0.8124672603457307,
      "eval_PRM F1 Neg": 0.6938775510204082,
      "eval_PRM NPV": 0.6538461538461539,
      "eval_PRM Precision": 0.925,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.23032961785793304,
      "eval_runtime": 14.4998,
      "eval_samples_per_second": 3.035,
      "eval_steps_per_second": 0.138,
      "step": 2120
    },
    {
      "epoch": 0.549286199864038,
      "grad_norm": 4.34167263740955,
      "learning_rate": 6.266955898099426e-07,
      "loss": 0.4101,
      "step": 2121
    },
    {
      "epoch": 0.549545174970056,
      "grad_norm": 2.9378142387940316,
      "learning_rate": 6.261303939769796e-07,
      "loss": 0.3473,
      "step": 2122
    },
    {
      "epoch": 0.549804150076074,
      "grad_norm": 3.4018943479956385,
      "learning_rate": 6.255651972195953e-07,
      "loss": 0.3717,
      "step": 2123
    },
    {
      "epoch": 0.5500631251820919,
      "grad_norm": 8.038346486365919,
      "learning_rate": 6.25e-07,
      "loss": 0.2704,
      "step": 2124
    },
    {
      "epoch": 0.5503221002881098,
      "grad_norm": 4.6478257733961295,
      "learning_rate": 6.24434802780405e-07,
      "loss": 0.3547,
      "step": 2125
    },
    {
      "epoch": 0.5503221002881098,
      "eval_PRM Accuracy": 0.8584905660377359,
      "eval_PRM F1": 0.9079754601226994,
      "eval_PRM F1 AUC": 0.8153483499214248,
      "eval_PRM F1 AUC (fixed)": 0.8080146673651126,
      "eval_PRM F1 Neg": 0.6938775510204082,
      "eval_PRM NPV": 0.6538461538461539,
      "eval_PRM Precision": 0.925,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.22999288141727448,
      "eval_runtime": 14.8411,
      "eval_samples_per_second": 2.965,
      "eval_steps_per_second": 0.135,
      "step": 2125
    },
    {
      "epoch": 0.5505810753941277,
      "grad_norm": 7.041848255803033,
      "learning_rate": 6.238696060230206e-07,
      "loss": 0.4228,
      "step": 2126
    },
    {
      "epoch": 0.5508400505001457,
      "grad_norm": 3.5728182187043243,
      "learning_rate": 6.233044101900575e-07,
      "loss": 0.347,
      "step": 2127
    },
    {
      "epoch": 0.5510990256061636,
      "grad_norm": 3.0651487067998704,
      "learning_rate": 6.227392157437252e-07,
      "loss": 0.3246,
      "step": 2128
    },
    {
      "epoch": 0.5513580007121816,
      "grad_norm": 4.141137313235903,
      "learning_rate": 6.221740231462329e-07,
      "loss": 0.3511,
      "step": 2129
    },
    {
      "epoch": 0.5516169758181995,
      "grad_norm": 2.7567031148589662,
      "learning_rate": 6.216088328597868e-07,
      "loss": 0.2791,
      "step": 2130
    },
    {
      "epoch": 0.5516169758181995,
      "eval_PRM Accuracy": 0.8584905660377359,
      "eval_PRM F1": 0.9079754601226994,
      "eval_PRM F1 AUC": 0.8153483499214248,
      "eval_PRM F1 AUC (fixed)": 0.8059193294918806,
      "eval_PRM F1 Neg": 0.6938775510204082,
      "eval_PRM NPV": 0.6538461538461539,
      "eval_PRM Precision": 0.925,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.23175816237926483,
      "eval_runtime": 14.6317,
      "eval_samples_per_second": 3.007,
      "eval_steps_per_second": 0.137,
      "step": 2130
    },
    {
      "epoch": 0.5518759509242174,
      "grad_norm": 4.796414638243899,
      "learning_rate": 6.210436453465922e-07,
      "loss": 0.2865,
      "step": 2131
    },
    {
      "epoch": 0.5521349260302354,
      "grad_norm": 3.060677378015724,
      "learning_rate": 6.204784610688523e-07,
      "loss": 0.2911,
      "step": 2132
    },
    {
      "epoch": 0.5523939011362533,
      "grad_norm": 5.77189323567779,
      "learning_rate": 6.199132804887671e-07,
      "loss": 0.3508,
      "step": 2133
    },
    {
      "epoch": 0.5526528762422712,
      "grad_norm": 2.8614243377835504,
      "learning_rate": 6.193481040685339e-07,
      "loss": 0.2875,
      "step": 2134
    },
    {
      "epoch": 0.5529118513482891,
      "grad_norm": 4.836800111222164,
      "learning_rate": 6.187829322703466e-07,
      "loss": 0.3365,
      "step": 2135
    },
    {
      "epoch": 0.5529118513482891,
      "eval_PRM Accuracy": 0.8584905660377359,
      "eval_PRM F1": 0.906832298136646,
      "eval_PRM F1 AUC": 0.8310633839706653,
      "eval_PRM F1 AUC (fixed)": 0.8059193294918805,
      "eval_PRM F1 Neg": 0.7058823529411765,
      "eval_PRM NPV": 0.6428571428571429,
      "eval_PRM Precision": 0.9358974358974359,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.2330990731716156,
      "eval_runtime": 14.5074,
      "eval_samples_per_second": 3.033,
      "eval_steps_per_second": 0.138,
      "step": 2135
    },
    {
      "epoch": 0.553170826454307,
      "grad_norm": 4.838169666356384,
      "learning_rate": 6.182177655563951e-07,
      "loss": 0.3986,
      "step": 2136
    },
    {
      "epoch": 0.5534298015603251,
      "grad_norm": 2.5687599795623415,
      "learning_rate": 6.176526043888654e-07,
      "loss": 0.3109,
      "step": 2137
    },
    {
      "epoch": 0.553688776666343,
      "grad_norm": 4.628514481444664,
      "learning_rate": 6.170874492299385e-07,
      "loss": 0.3793,
      "step": 2138
    },
    {
      "epoch": 0.5539477517723609,
      "grad_norm": 8.73534252137877,
      "learning_rate": 6.165223005417912e-07,
      "loss": 0.4094,
      "step": 2139
    },
    {
      "epoch": 0.5542067268783788,
      "grad_norm": 8.599399169963599,
      "learning_rate": 6.159571587865944e-07,
      "loss": 0.4208,
      "step": 2140
    },
    {
      "epoch": 0.5542067268783788,
      "eval_PRM Accuracy": 0.8584905660377359,
      "eval_PRM F1": 0.906832298136646,
      "eval_PRM F1 AUC": 0.8310633839706653,
      "eval_PRM F1 AUC (fixed)": 0.8043478260869565,
      "eval_PRM F1 Neg": 0.7058823529411765,
      "eval_PRM NPV": 0.6428571428571429,
      "eval_PRM Precision": 0.9358974358974359,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.23667941987514496,
      "eval_runtime": 14.8116,
      "eval_samples_per_second": 2.971,
      "eval_steps_per_second": 0.135,
      "step": 2140
    },
    {
      "epoch": 0.5544657019843967,
      "grad_norm": 6.299276361380223,
      "learning_rate": 6.153920244265136e-07,
      "loss": 0.3958,
      "step": 2141
    },
    {
      "epoch": 0.5547246770904147,
      "grad_norm": 4.553469617462408,
      "learning_rate": 6.148268979237082e-07,
      "loss": 0.3439,
      "step": 2142
    },
    {
      "epoch": 0.5549836521964326,
      "grad_norm": 3.658400210268422,
      "learning_rate": 6.142617797403313e-07,
      "loss": 0.3455,
      "step": 2143
    },
    {
      "epoch": 0.5552426273024506,
      "grad_norm": 4.212585141078089,
      "learning_rate": 6.13696670338529e-07,
      "loss": 0.3336,
      "step": 2144
    },
    {
      "epoch": 0.5555016024084685,
      "grad_norm": 4.302686781907926,
      "learning_rate": 6.131315701804405e-07,
      "loss": 0.2926,
      "step": 2145
    },
    {
      "epoch": 0.5555016024084685,
      "eval_PRM Accuracy": 0.8490566037735849,
      "eval_PRM F1": 0.9,
      "eval_PRM F1 AUC": 0.8250392875851232,
      "eval_PRM F1 AUC (fixed)": 0.7998952331063384,
      "eval_PRM F1 Neg": 0.6923076923076923,
      "eval_PRM NPV": 0.6206896551724138,
      "eval_PRM Precision": 0.935064935064935,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.24193677306175232,
      "eval_runtime": 14.5349,
      "eval_samples_per_second": 3.027,
      "eval_steps_per_second": 0.138,
      "step": 2145
    },
    {
      "epoch": 0.5557605775144864,
      "grad_norm": 4.6442253908420845,
      "learning_rate": 6.125664797281968e-07,
      "loss": 0.355,
      "step": 2146
    },
    {
      "epoch": 0.5560195526205044,
      "grad_norm": 7.060072762024737,
      "learning_rate": 6.120013994439217e-07,
      "loss": 0.3263,
      "step": 2147
    },
    {
      "epoch": 0.5562785277265223,
      "grad_norm": 5.1527709629499245,
      "learning_rate": 6.114363297897304e-07,
      "loss": 0.3009,
      "step": 2148
    },
    {
      "epoch": 0.5565375028325402,
      "grad_norm": 9.121733615711893,
      "learning_rate": 6.108712712277293e-07,
      "loss": 0.2985,
      "step": 2149
    },
    {
      "epoch": 0.5567964779385581,
      "grad_norm": 6.977071114570016,
      "learning_rate": 6.103062242200159e-07,
      "loss": 0.2989,
      "step": 2150
    },
    {
      "epoch": 0.5567964779385581,
      "eval_PRM Accuracy": 0.8584905660377359,
      "eval_PRM F1": 0.906832298136646,
      "eval_PRM F1 AUC": 0.8310633839706653,
      "eval_PRM F1 AUC (fixed)": 0.8059193294918805,
      "eval_PRM F1 Neg": 0.7058823529411765,
      "eval_PRM NPV": 0.6428571428571429,
      "eval_PRM Precision": 0.9358974358974359,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.2338414192199707,
      "eval_runtime": 15.039,
      "eval_samples_per_second": 2.926,
      "eval_steps_per_second": 0.133,
      "step": 2150
    },
    {
      "epoch": 0.557055453044576,
      "grad_norm": 4.725303332398599,
      "learning_rate": 6.09741189228678e-07,
      "loss": 0.3239,
      "step": 2151
    },
    {
      "epoch": 0.5573144281505941,
      "grad_norm": 3.9127766459865057,
      "learning_rate": 6.091761667157938e-07,
      "loss": 0.2754,
      "step": 2152
    },
    {
      "epoch": 0.557573403256612,
      "grad_norm": 6.138081110935778,
      "learning_rate": 6.086111571434315e-07,
      "loss": 0.3657,
      "step": 2153
    },
    {
      "epoch": 0.5578323783626299,
      "grad_norm": 2.9322038075625256,
      "learning_rate": 6.080461609736486e-07,
      "loss": 0.3628,
      "step": 2154
    },
    {
      "epoch": 0.5580913534686478,
      "grad_norm": 10.697951815069676,
      "learning_rate": 6.074811786684909e-07,
      "loss": 0.3969,
      "step": 2155
    },
    {
      "epoch": 0.5580913534686478,
      "eval_PRM Accuracy": 0.8679245283018868,
      "eval_PRM F1": 0.9146341463414634,
      "eval_PRM F1 AUC": 0.8213724463069669,
      "eval_PRM F1 AUC (fixed)": 0.8051335777894186,
      "eval_PRM F1 Neg": 0.7083333333333334,
      "eval_PRM NPV": 0.68,
      "eval_PRM Precision": 0.9259259259259259,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.2333838790655136,
      "eval_runtime": 14.4025,
      "eval_samples_per_second": 3.055,
      "eval_steps_per_second": 0.139,
      "step": 2155
    },
    {
      "epoch": 0.5583503285746657,
      "grad_norm": 2.804892360151983,
      "learning_rate": 6.069162106899936e-07,
      "loss": 0.2993,
      "step": 2156
    },
    {
      "epoch": 0.5586093036806837,
      "grad_norm": 3.220867529028216,
      "learning_rate": 6.063512575001803e-07,
      "loss": 0.3108,
      "step": 2157
    },
    {
      "epoch": 0.5588682787867016,
      "grad_norm": 3.347548550567105,
      "learning_rate": 6.057863195610621e-07,
      "loss": 0.2583,
      "step": 2158
    },
    {
      "epoch": 0.5591272538927196,
      "grad_norm": 5.228017173702376,
      "learning_rate": 6.052213973346379e-07,
      "loss": 0.3662,
      "step": 2159
    },
    {
      "epoch": 0.5593862289987375,
      "grad_norm": 8.48754766607804,
      "learning_rate": 6.046564912828936e-07,
      "loss": 0.3484,
      "step": 2160
    },
    {
      "epoch": 0.5593862289987375,
      "eval_PRM Accuracy": 0.8773584905660378,
      "eval_PRM F1": 0.9212121212121213,
      "eval_PRM F1 AUC": 0.8273965426925091,
      "eval_PRM F1 AUC (fixed)": 0.8119434258774227,
      "eval_PRM F1 Neg": 0.723404255319149,
      "eval_PRM NPV": 0.7083333333333334,
      "eval_PRM Precision": 0.926829268292683,
      "eval_PRM Recall": 0.9156626506024096,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.23568886518478394,
      "eval_runtime": 14.7148,
      "eval_samples_per_second": 2.99,
      "eval_steps_per_second": 0.136,
      "step": 2160
    },
    {
      "epoch": 0.5596452041047554,
      "grad_norm": 6.663250178270902,
      "learning_rate": 6.040916018678018e-07,
      "loss": 0.3508,
      "step": 2161
    },
    {
      "epoch": 0.5599041792107734,
      "grad_norm": 3.4601057879003547,
      "learning_rate": 6.035267295513218e-07,
      "loss": 0.3745,
      "step": 2162
    },
    {
      "epoch": 0.5601631543167913,
      "grad_norm": 4.795726866903443,
      "learning_rate": 6.029618747953984e-07,
      "loss": 0.3612,
      "step": 2163
    },
    {
      "epoch": 0.5604221294228092,
      "grad_norm": 10.930396850440806,
      "learning_rate": 6.023970380619626e-07,
      "loss": 0.4115,
      "step": 2164
    },
    {
      "epoch": 0.5606811045288271,
      "grad_norm": 2.3526199190829606,
      "learning_rate": 6.018322198129303e-07,
      "loss": 0.2706,
      "step": 2165
    },
    {
      "epoch": 0.5606811045288271,
      "eval_PRM Accuracy": 0.8584905660377359,
      "eval_PRM F1": 0.9079754601226994,
      "eval_PRM F1 AUC": 0.8153483499214248,
      "eval_PRM F1 AUC (fixed)": 0.8046097433211106,
      "eval_PRM F1 Neg": 0.6938775510204082,
      "eval_PRM NPV": 0.6538461538461539,
      "eval_PRM Precision": 0.925,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.2373889982700348,
      "eval_runtime": 14.4392,
      "eval_samples_per_second": 3.047,
      "eval_steps_per_second": 0.139,
      "step": 2165
    },
    {
      "epoch": 0.560940079634845,
      "grad_norm": 3.645285199273107,
      "learning_rate": 6.012674205102023e-07,
      "loss": 0.2444,
      "step": 2166
    },
    {
      "epoch": 0.5611990547408631,
      "grad_norm": 5.465470722155276,
      "learning_rate": 6.007026406156644e-07,
      "loss": 0.3594,
      "step": 2167
    },
    {
      "epoch": 0.561458029846881,
      "grad_norm": 4.802178660028119,
      "learning_rate": 6.001378805911857e-07,
      "loss": 0.3626,
      "step": 2168
    },
    {
      "epoch": 0.5617170049528989,
      "grad_norm": 3.652843241216237,
      "learning_rate": 5.995731408986197e-07,
      "loss": 0.2761,
      "step": 2169
    },
    {
      "epoch": 0.5619759800589168,
      "grad_norm": 3.1050035393196067,
      "learning_rate": 5.990084219998032e-07,
      "loss": 0.3172,
      "step": 2170
    },
    {
      "epoch": 0.5619759800589168,
      "eval_PRM Accuracy": 0.8584905660377359,
      "eval_PRM F1": 0.906832298136646,
      "eval_PRM F1 AUC": 0.8310633839706653,
      "eval_PRM F1 AUC (fixed)": 0.8048716605552646,
      "eval_PRM F1 Neg": 0.7058823529411765,
      "eval_PRM NPV": 0.6428571428571429,
      "eval_PRM Precision": 0.9358974358974359,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.23964986205101013,
      "eval_runtime": 14.4548,
      "eval_samples_per_second": 3.044,
      "eval_steps_per_second": 0.138,
      "step": 2170
    },
    {
      "epoch": 0.5622349551649348,
      "grad_norm": 2.7438622069234384,
      "learning_rate": 5.984437243565556e-07,
      "loss": 0.3192,
      "step": 2171
    },
    {
      "epoch": 0.5624939302709527,
      "grad_norm": 2.9370390839910634,
      "learning_rate": 5.978790484306792e-07,
      "loss": 0.3192,
      "step": 2172
    },
    {
      "epoch": 0.5627529053769706,
      "grad_norm": 3.998508745732631,
      "learning_rate": 5.973143946839586e-07,
      "loss": 0.3296,
      "step": 2173
    },
    {
      "epoch": 0.5630118804829886,
      "grad_norm": 3.395381152576033,
      "learning_rate": 5.967497635781601e-07,
      "loss": 0.3265,
      "step": 2174
    },
    {
      "epoch": 0.5632708555890065,
      "grad_norm": 3.2179260636542764,
      "learning_rate": 5.961851555750314e-07,
      "loss": 0.3445,
      "step": 2175
    },
    {
      "epoch": 0.5632708555890065,
      "eval_PRM Accuracy": 0.8679245283018868,
      "eval_PRM F1": 0.9135802469135802,
      "eval_PRM F1 AUC": 0.8370874803562075,
      "eval_PRM F1 AUC (fixed)": 0.8035620743844945,
      "eval_PRM F1 Neg": 0.72,
      "eval_PRM NPV": 0.6666666666666666,
      "eval_PRM Precision": 0.9367088607594937,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.23954223096370697,
      "eval_runtime": 14.7506,
      "eval_samples_per_second": 2.983,
      "eval_steps_per_second": 0.136,
      "step": 2175
    },
    {
      "epoch": 0.5635298306950245,
      "grad_norm": 4.333991750676386,
      "learning_rate": 5.956205711363018e-07,
      "loss": 0.3791,
      "step": 2176
    },
    {
      "epoch": 0.5637888058010424,
      "grad_norm": 4.73193791967382,
      "learning_rate": 5.950560107236807e-07,
      "loss": 0.2663,
      "step": 2177
    },
    {
      "epoch": 0.5640477809070603,
      "grad_norm": 5.10143049504143,
      "learning_rate": 5.944914747988583e-07,
      "loss": 0.31,
      "step": 2178
    },
    {
      "epoch": 0.5643067560130782,
      "grad_norm": 3.5094027717049046,
      "learning_rate": 5.939269638235051e-07,
      "loss": 0.3053,
      "step": 2179
    },
    {
      "epoch": 0.5645657311190961,
      "grad_norm": 3.150577034227375,
      "learning_rate": 5.933624782592696e-07,
      "loss": 0.2992,
      "step": 2180
    },
    {
      "epoch": 0.5645657311190961,
      "eval_PRM Accuracy": 0.8679245283018868,
      "eval_PRM F1": 0.9135802469135802,
      "eval_PRM F1 AUC": 0.8370874803562075,
      "eval_PRM F1 AUC (fixed)": 0.8046097433211106,
      "eval_PRM F1 Neg": 0.72,
      "eval_PRM NPV": 0.6666666666666666,
      "eval_PRM Precision": 0.9367088607594937,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.23694948852062225,
      "eval_runtime": 15.2488,
      "eval_samples_per_second": 2.885,
      "eval_steps_per_second": 0.131,
      "step": 2180
    },
    {
      "epoch": 0.5648247062251142,
      "grad_norm": 3.691441681200393,
      "learning_rate": 5.927980185677815e-07,
      "loss": 0.2441,
      "step": 2181
    },
    {
      "epoch": 0.5650836813311321,
      "grad_norm": 6.006657620529061,
      "learning_rate": 5.922335852106482e-07,
      "loss": 0.4381,
      "step": 2182
    },
    {
      "epoch": 0.56534265643715,
      "grad_norm": 2.960627634851511,
      "learning_rate": 5.916691786494558e-07,
      "loss": 0.3136,
      "step": 2183
    },
    {
      "epoch": 0.5656016315431679,
      "grad_norm": 4.100802391911433,
      "learning_rate": 5.911047993457687e-07,
      "loss": 0.3032,
      "step": 2184
    },
    {
      "epoch": 0.5658606066491858,
      "grad_norm": 4.305060391907341,
      "learning_rate": 5.905404477611286e-07,
      "loss": 0.3641,
      "step": 2185
    },
    {
      "epoch": 0.5658606066491858,
      "eval_PRM Accuracy": 0.8584905660377359,
      "eval_PRM F1": 0.9079754601226994,
      "eval_PRM F1 AUC": 0.8153483499214248,
      "eval_PRM F1 AUC (fixed)": 0.8085385018334207,
      "eval_PRM F1 Neg": 0.6938775510204082,
      "eval_PRM NPV": 0.6538461538461539,
      "eval_PRM Precision": 0.925,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.2373279184103012,
      "eval_runtime": 15.1415,
      "eval_samples_per_second": 2.906,
      "eval_steps_per_second": 0.132,
      "step": 2185
    },
    {
      "epoch": 0.5661195817552038,
      "grad_norm": 4.31925135823975,
      "learning_rate": 5.89976124357055e-07,
      "loss": 0.3467,
      "step": 2186
    },
    {
      "epoch": 0.5663785568612217,
      "grad_norm": 5.205890898411738,
      "learning_rate": 5.89411829595044e-07,
      "loss": 0.4189,
      "step": 2187
    },
    {
      "epoch": 0.5666375319672396,
      "grad_norm": 8.40049803719755,
      "learning_rate": 5.888475639365684e-07,
      "loss": 0.3568,
      "step": 2188
    },
    {
      "epoch": 0.5668965070732576,
      "grad_norm": 4.19722515922034,
      "learning_rate": 5.882833278430771e-07,
      "loss": 0.3649,
      "step": 2189
    },
    {
      "epoch": 0.5671554821792755,
      "grad_norm": 9.817230770195735,
      "learning_rate": 5.877191217759952e-07,
      "loss": 0.4202,
      "step": 2190
    },
    {
      "epoch": 0.5671554821792755,
      "eval_PRM Accuracy": 0.8679245283018868,
      "eval_PRM F1": 0.9135802469135802,
      "eval_PRM F1 AUC": 0.8370874803562075,
      "eval_PRM F1 AUC (fixed)": 0.8074908328968047,
      "eval_PRM F1 Neg": 0.72,
      "eval_PRM NPV": 0.6666666666666666,
      "eval_PRM Precision": 0.9367088607594937,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.23871129751205444,
      "eval_runtime": 14.7492,
      "eval_samples_per_second": 2.983,
      "eval_steps_per_second": 0.136,
      "step": 2190
    },
    {
      "epoch": 0.5674144572852935,
      "grad_norm": 4.665414141254776,
      "learning_rate": 5.871549461967225e-07,
      "loss": 0.2917,
      "step": 2191
    },
    {
      "epoch": 0.5676734323913114,
      "grad_norm": 3.4552742913869374,
      "learning_rate": 5.865908015666348e-07,
      "loss": 0.3337,
      "step": 2192
    },
    {
      "epoch": 0.5679324074973293,
      "grad_norm": 4.176413118542138,
      "learning_rate": 5.860266883470819e-07,
      "loss": 0.3603,
      "step": 2193
    },
    {
      "epoch": 0.5681913826033472,
      "grad_norm": 2.830483612853203,
      "learning_rate": 5.854626069993883e-07,
      "loss": 0.267,
      "step": 2194
    },
    {
      "epoch": 0.5684503577093651,
      "grad_norm": 3.4431523805157105,
      "learning_rate": 5.848985579848523e-07,
      "loss": 0.3493,
      "step": 2195
    },
    {
      "epoch": 0.5684503577093651,
      "eval_PRM Accuracy": 0.8490566037735849,
      "eval_PRM F1": 0.9,
      "eval_PRM F1 AUC": 0.8250392875851232,
      "eval_PRM F1 AUC (fixed)": 0.8027763226820326,
      "eval_PRM F1 Neg": 0.6923076923076923,
      "eval_PRM NPV": 0.6206896551724138,
      "eval_PRM Precision": 0.935064935064935,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.24241438508033752,
      "eval_runtime": 14.4215,
      "eval_samples_per_second": 3.051,
      "eval_steps_per_second": 0.139,
      "step": 2195
    },
    {
      "epoch": 0.5687093328153832,
      "grad_norm": 7.272621536523948,
      "learning_rate": 5.843345417647453e-07,
      "loss": 0.2999,
      "step": 2196
    },
    {
      "epoch": 0.5689683079214011,
      "grad_norm": 3.0882262034259975,
      "learning_rate": 5.837705588003127e-07,
      "loss": 0.2434,
      "step": 2197
    },
    {
      "epoch": 0.569227283027419,
      "grad_norm": 3.4638343344836016,
      "learning_rate": 5.832066095527722e-07,
      "loss": 0.3158,
      "step": 2198
    },
    {
      "epoch": 0.5694862581334369,
      "grad_norm": 3.370119849080314,
      "learning_rate": 5.826426944833139e-07,
      "loss": 0.3444,
      "step": 2199
    },
    {
      "epoch": 0.5697452332394548,
      "grad_norm": 4.715007655321245,
      "learning_rate": 5.820788140531005e-07,
      "loss": 0.3481,
      "step": 2200
    },
    {
      "epoch": 0.5697452332394548,
      "eval_PRM Accuracy": 0.8584905660377359,
      "eval_PRM F1": 0.906832298136646,
      "eval_PRM F1 AUC": 0.8310633839706653,
      "eval_PRM F1 AUC (fixed)": 0.8059193294918805,
      "eval_PRM F1 Neg": 0.7058823529411765,
      "eval_PRM NPV": 0.6428571428571429,
      "eval_PRM Precision": 0.9358974358974359,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.23918257653713226,
      "eval_runtime": 14.9911,
      "eval_samples_per_second": 2.935,
      "eval_steps_per_second": 0.133,
      "step": 2200
    },
    {
      "epoch": 0.5700042083454728,
      "grad_norm": 5.905858282740968,
      "learning_rate": 5.815149687232655e-07,
      "loss": 0.3373,
      "step": 2201
    },
    {
      "epoch": 0.5702631834514907,
      "grad_norm": 3.769987278447685,
      "learning_rate": 5.809511589549143e-07,
      "loss": 0.3141,
      "step": 2202
    },
    {
      "epoch": 0.5705221585575087,
      "grad_norm": 3.8585678173729168,
      "learning_rate": 5.803873852091236e-07,
      "loss": 0.3686,
      "step": 2203
    },
    {
      "epoch": 0.5707811336635266,
      "grad_norm": 2.886014729680767,
      "learning_rate": 5.798236479469394e-07,
      "loss": 0.2771,
      "step": 2204
    },
    {
      "epoch": 0.5710401087695445,
      "grad_norm": 4.0094727737267855,
      "learning_rate": 5.792599476293786e-07,
      "loss": 0.2987,
      "step": 2205
    },
    {
      "epoch": 0.5710401087695445,
      "eval_PRM Accuracy": 0.8679245283018868,
      "eval_PRM F1": 0.9146341463414634,
      "eval_PRM F1 AUC": 0.8213724463069669,
      "eval_PRM F1 AUC (fixed)": 0.8080146673651127,
      "eval_PRM F1 Neg": 0.7083333333333334,
      "eval_PRM NPV": 0.68,
      "eval_PRM Precision": 0.9259259259259259,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.2368496060371399,
      "eval_runtime": 14.87,
      "eval_samples_per_second": 2.959,
      "eval_steps_per_second": 0.134,
      "step": 2205
    },
    {
      "epoch": 0.5712990838755625,
      "grad_norm": 5.289767576604941,
      "learning_rate": 5.786962847174283e-07,
      "loss": 0.33,
      "step": 2206
    },
    {
      "epoch": 0.5715580589815804,
      "grad_norm": 4.5394204009609656,
      "learning_rate": 5.781326596720444e-07,
      "loss": 0.3767,
      "step": 2207
    },
    {
      "epoch": 0.5718170340875983,
      "grad_norm": 3.689013592610438,
      "learning_rate": 5.77569072954152e-07,
      "loss": 0.3102,
      "step": 2208
    },
    {
      "epoch": 0.5720760091936162,
      "grad_norm": 3.776834612394486,
      "learning_rate": 5.770055250246448e-07,
      "loss": 0.3533,
      "step": 2209
    },
    {
      "epoch": 0.5723349842996343,
      "grad_norm": 5.924542346459384,
      "learning_rate": 5.76442016344385e-07,
      "loss": 0.3442,
      "step": 2210
    },
    {
      "epoch": 0.5723349842996343,
      "eval_PRM Accuracy": 0.8679245283018868,
      "eval_PRM F1": 0.9156626506024096,
      "eval_PRM F1 AUC": 0.8056574122577265,
      "eval_PRM F1 AUC (fixed)": 0.8103719224724987,
      "eval_PRM F1 Neg": 0.6956521739130435,
      "eval_PRM NPV": 0.6956521739130435,
      "eval_PRM Precision": 0.9156626506024096,
      "eval_PRM Recall": 0.9156626506024096,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.23526550829410553,
      "eval_runtime": 14.5842,
      "eval_samples_per_second": 3.017,
      "eval_steps_per_second": 0.137,
      "step": 2210
    },
    {
      "epoch": 0.5725939594056522,
      "grad_norm": 4.823321847126282,
      "learning_rate": 5.758785473742024e-07,
      "loss": 0.3354,
      "step": 2211
    },
    {
      "epoch": 0.5728529345116701,
      "grad_norm": 6.529416793230183,
      "learning_rate": 5.753151185748948e-07,
      "loss": 0.3697,
      "step": 2212
    },
    {
      "epoch": 0.573111909617688,
      "grad_norm": 4.546228215649944,
      "learning_rate": 5.747517304072265e-07,
      "loss": 0.4414,
      "step": 2213
    },
    {
      "epoch": 0.5733708847237059,
      "grad_norm": 3.2145918920345506,
      "learning_rate": 5.741883833319288e-07,
      "loss": 0.2551,
      "step": 2214
    },
    {
      "epoch": 0.5736298598297239,
      "grad_norm": 6.215675257992527,
      "learning_rate": 5.736250778096996e-07,
      "loss": 0.3805,
      "step": 2215
    },
    {
      "epoch": 0.5736298598297239,
      "eval_PRM Accuracy": 0.8773584905660378,
      "eval_PRM F1": 0.9202453987730062,
      "eval_PRM F1 AUC": 0.8431115767417497,
      "eval_PRM F1 AUC (fixed)": 0.8074908328968047,
      "eval_PRM F1 Neg": 0.7346938775510204,
      "eval_PRM NPV": 0.6923076923076923,
      "eval_PRM Precision": 0.9375,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.2351292222738266,
      "eval_runtime": 15.3116,
      "eval_samples_per_second": 2.874,
      "eval_steps_per_second": 0.131,
      "step": 2215
    },
    {
      "epoch": 0.5738888349357418,
      "grad_norm": 5.074078758742773,
      "learning_rate": 5.730618143012028e-07,
      "loss": 0.3589,
      "step": 2216
    },
    {
      "epoch": 0.5741478100417597,
      "grad_norm": 4.597364264453503,
      "learning_rate": 5.724985932670679e-07,
      "loss": 0.3523,
      "step": 2217
    },
    {
      "epoch": 0.5744067851477777,
      "grad_norm": 3.366275484181149,
      "learning_rate": 5.719354151678896e-07,
      "loss": 0.265,
      "step": 2218
    },
    {
      "epoch": 0.5746657602537956,
      "grad_norm": 4.463756027833097,
      "learning_rate": 5.713722804642274e-07,
      "loss": 0.3731,
      "step": 2219
    },
    {
      "epoch": 0.5749247353598136,
      "grad_norm": 3.5438524256526165,
      "learning_rate": 5.708091896166057e-07,
      "loss": 0.3557,
      "step": 2220
    },
    {
      "epoch": 0.5749247353598136,
      "eval_PRM Accuracy": 0.8679245283018868,
      "eval_PRM F1": 0.9135802469135802,
      "eval_PRM F1 AUC": 0.8370874803562075,
      "eval_PRM F1 AUC (fixed)": 0.8048716605552646,
      "eval_PRM F1 Neg": 0.72,
      "eval_PRM NPV": 0.6666666666666666,
      "eval_PRM Precision": 0.9367088607594937,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.23860736191272736,
      "eval_runtime": 14.9573,
      "eval_samples_per_second": 2.942,
      "eval_steps_per_second": 0.134,
      "step": 2220
    },
    {
      "epoch": 0.5751837104658315,
      "grad_norm": 4.706802889861752,
      "learning_rate": 5.702461430855123e-07,
      "loss": 0.3033,
      "step": 2221
    },
    {
      "epoch": 0.5754426855718494,
      "grad_norm": 3.5408110592109594,
      "learning_rate": 5.696831413313994e-07,
      "loss": 0.3863,
      "step": 2222
    },
    {
      "epoch": 0.5757016606778673,
      "grad_norm": 3.2076864899003152,
      "learning_rate": 5.691201848146826e-07,
      "loss": 0.3129,
      "step": 2223
    },
    {
      "epoch": 0.5759606357838852,
      "grad_norm": 4.959563057577472,
      "learning_rate": 5.685572739957399e-07,
      "loss": 0.3461,
      "step": 2224
    },
    {
      "epoch": 0.5762196108899033,
      "grad_norm": 3.2221689968098395,
      "learning_rate": 5.679944093349125e-07,
      "loss": 0.2911,
      "step": 2225
    },
    {
      "epoch": 0.5762196108899033,
      "eval_PRM Accuracy": 0.8679245283018868,
      "eval_PRM F1": 0.9125,
      "eval_PRM F1 AUC": 0.8528025144054479,
      "eval_PRM F1 AUC (fixed)": 0.8009429020429545,
      "eval_PRM F1 Neg": 0.7307692307692307,
      "eval_PRM NPV": 0.6551724137931034,
      "eval_PRM Precision": 0.948051948051948,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.8260869565217391,
      "eval_loss": 0.23951144516468048,
      "eval_runtime": 14.5648,
      "eval_samples_per_second": 3.021,
      "eval_steps_per_second": 0.137,
      "step": 2225
    },
    {
      "epoch": 0.5764785859959212,
      "grad_norm": 3.779593604081406,
      "learning_rate": 5.67431591292504e-07,
      "loss": 0.29,
      "step": 2226
    },
    {
      "epoch": 0.5767375611019391,
      "grad_norm": 3.631030738159037,
      "learning_rate": 5.668688203287791e-07,
      "loss": 0.3503,
      "step": 2227
    },
    {
      "epoch": 0.576996536207957,
      "grad_norm": 3.339647513318244,
      "learning_rate": 5.663060969039651e-07,
      "loss": 0.3443,
      "step": 2228
    },
    {
      "epoch": 0.5772555113139749,
      "grad_norm": 2.7829476555373382,
      "learning_rate": 5.657434214782488e-07,
      "loss": 0.3167,
      "step": 2229
    },
    {
      "epoch": 0.5775144864199929,
      "grad_norm": 3.6085314219871587,
      "learning_rate": 5.651807945117794e-07,
      "loss": 0.2965,
      "step": 2230
    },
    {
      "epoch": 0.5775144864199929,
      "eval_PRM Accuracy": 0.8584905660377359,
      "eval_PRM F1": 0.906832298136646,
      "eval_PRM F1 AUC": 0.8310633839706653,
      "eval_PRM F1 AUC (fixed)": 0.8061812467260345,
      "eval_PRM F1 Neg": 0.7058823529411765,
      "eval_PRM NPV": 0.6428571428571429,
      "eval_PRM Precision": 0.9358974358974359,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.2356843501329422,
      "eval_runtime": 14.4739,
      "eval_samples_per_second": 3.04,
      "eval_steps_per_second": 0.138,
      "step": 2230
    },
    {
      "epoch": 0.5777734615260108,
      "grad_norm": 3.22894545550293,
      "learning_rate": 5.646182164646657e-07,
      "loss": 0.2348,
      "step": 2231
    },
    {
      "epoch": 0.5780324366320287,
      "grad_norm": 6.228289074191956,
      "learning_rate": 5.640556877969766e-07,
      "loss": 0.2419,
      "step": 2232
    },
    {
      "epoch": 0.5782914117380467,
      "grad_norm": 3.595117887907518,
      "learning_rate": 5.634932089687405e-07,
      "loss": 0.3374,
      "step": 2233
    },
    {
      "epoch": 0.5785503868440646,
      "grad_norm": 3.0669702778582257,
      "learning_rate": 5.629307804399452e-07,
      "loss": 0.3144,
      "step": 2234
    },
    {
      "epoch": 0.5788093619500826,
      "grad_norm": 4.0678924070890226,
      "learning_rate": 5.623684026705374e-07,
      "loss": 0.3021,
      "step": 2235
    },
    {
      "epoch": 0.5788093619500826,
      "eval_PRM Accuracy": 0.8584905660377359,
      "eval_PRM F1": 0.9079754601226994,
      "eval_PRM F1 AUC": 0.8153483499214248,
      "eval_PRM F1 AUC (fixed)": 0.8090623363017286,
      "eval_PRM F1 Neg": 0.6938775510204082,
      "eval_PRM NPV": 0.6538461538461539,
      "eval_PRM Precision": 0.925,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.22923964262008667,
      "eval_runtime": 14.9935,
      "eval_samples_per_second": 2.935,
      "eval_steps_per_second": 0.133,
      "step": 2235
    },
    {
      "epoch": 0.5790683370561005,
      "grad_norm": 5.035070357652491,
      "learning_rate": 5.618060761204218e-07,
      "loss": 0.338,
      "step": 2236
    },
    {
      "epoch": 0.5793273121621184,
      "grad_norm": 5.536224529666224,
      "learning_rate": 5.612438012494625e-07,
      "loss": 0.3413,
      "step": 2237
    },
    {
      "epoch": 0.5795862872681363,
      "grad_norm": 5.8582070592066815,
      "learning_rate": 5.606815785174797e-07,
      "loss": 0.3362,
      "step": 2238
    },
    {
      "epoch": 0.5798452623741542,
      "grad_norm": 3.5720002712287156,
      "learning_rate": 5.601194083842517e-07,
      "loss": 0.2763,
      "step": 2239
    },
    {
      "epoch": 0.5801042374801723,
      "grad_norm": 3.31702752622457,
      "learning_rate": 5.595572913095142e-07,
      "loss": 0.3153,
      "step": 2240
    },
    {
      "epoch": 0.5801042374801723,
      "eval_PRM Accuracy": 0.8679245283018868,
      "eval_PRM F1": 0.9146341463414634,
      "eval_PRM F1 AUC": 0.8213724463069669,
      "eval_PRM F1 AUC (fixed)": 0.8127291775798847,
      "eval_PRM F1 Neg": 0.7083333333333334,
      "eval_PRM NPV": 0.68,
      "eval_PRM Precision": 0.9259259259259259,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.22618189454078674,
      "eval_runtime": 14.7184,
      "eval_samples_per_second": 2.989,
      "eval_steps_per_second": 0.136,
      "step": 2240
    },
    {
      "epoch": 0.5803632125861902,
      "grad_norm": 4.135079055735316,
      "learning_rate": 5.589952277529589e-07,
      "loss": 0.2894,
      "step": 2241
    },
    {
      "epoch": 0.5806221876922081,
      "grad_norm": 3.331029874360077,
      "learning_rate": 5.584332181742341e-07,
      "loss": 0.3153,
      "step": 2242
    },
    {
      "epoch": 0.580881162798226,
      "grad_norm": 9.128158787277526,
      "learning_rate": 5.578712630329437e-07,
      "loss": 0.3798,
      "step": 2243
    },
    {
      "epoch": 0.5811401379042439,
      "grad_norm": 3.4476419488435277,
      "learning_rate": 5.573093627886471e-07,
      "loss": 0.2865,
      "step": 2244
    },
    {
      "epoch": 0.5813991130102619,
      "grad_norm": 3.34090527187964,
      "learning_rate": 5.567475179008591e-07,
      "loss": 0.2655,
      "step": 2245
    },
    {
      "epoch": 0.5813991130102619,
      "eval_PRM Accuracy": 0.8584905660377359,
      "eval_PRM F1": 0.9079754601226994,
      "eval_PRM F1 AUC": 0.8153483499214248,
      "eval_PRM F1 AUC (fixed)": 0.8135149292823467,
      "eval_PRM F1 Neg": 0.6938775510204082,
      "eval_PRM NPV": 0.6538461538461539,
      "eval_PRM Precision": 0.925,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.22423087060451508,
      "eval_runtime": 14.6161,
      "eval_samples_per_second": 3.01,
      "eval_steps_per_second": 0.137,
      "step": 2245
    },
    {
      "epoch": 0.5816580881162798,
      "grad_norm": 4.659373249265686,
      "learning_rate": 5.561857288290488e-07,
      "loss": 0.2826,
      "step": 2246
    },
    {
      "epoch": 0.5819170632222977,
      "grad_norm": 4.663837659525852,
      "learning_rate": 5.556239960326401e-07,
      "loss": 0.2924,
      "step": 2247
    },
    {
      "epoch": 0.5821760383283157,
      "grad_norm": 7.3393110411426985,
      "learning_rate": 5.550623199710105e-07,
      "loss": 0.4165,
      "step": 2248
    },
    {
      "epoch": 0.5824350134343336,
      "grad_norm": 4.060046683396946,
      "learning_rate": 5.545007011034913e-07,
      "loss": 0.3423,
      "step": 2249
    },
    {
      "epoch": 0.5826939885403516,
      "grad_norm": 3.1666829250293183,
      "learning_rate": 5.539391398893668e-07,
      "loss": 0.2963,
      "step": 2250
    },
    {
      "epoch": 0.5826939885403516,
      "eval_PRM Accuracy": 0.8773584905660378,
      "eval_PRM F1": 0.9202453987730062,
      "eval_PRM F1 AUC": 0.8431115767417497,
      "eval_PRM F1 AUC (fixed)": 0.8135149292823468,
      "eval_PRM F1 Neg": 0.7346938775510204,
      "eval_PRM NPV": 0.6923076923076923,
      "eval_PRM Precision": 0.9375,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.22282634675502777,
      "eval_runtime": 14.4828,
      "eval_samples_per_second": 3.038,
      "eval_steps_per_second": 0.138,
      "step": 2250
    },
    {
      "epoch": 0.5829529636463695,
      "grad_norm": 5.490205553533866,
      "learning_rate": 5.533776367878746e-07,
      "loss": 0.3081,
      "step": 2251
    },
    {
      "epoch": 0.5832119387523874,
      "grad_norm": 3.934700289875118,
      "learning_rate": 5.528161922582047e-07,
      "loss": 0.284,
      "step": 2252
    },
    {
      "epoch": 0.5834709138584053,
      "grad_norm": 3.5382586515194525,
      "learning_rate": 5.522548067594988e-07,
      "loss": 0.2645,
      "step": 2253
    },
    {
      "epoch": 0.5837298889644233,
      "grad_norm": 7.589217938028257,
      "learning_rate": 5.516934807508501e-07,
      "loss": 0.3666,
      "step": 2254
    },
    {
      "epoch": 0.5839888640704413,
      "grad_norm": 2.90897523325107,
      "learning_rate": 5.511322146913042e-07,
      "loss": 0.2214,
      "step": 2255
    },
    {
      "epoch": 0.5839888640704413,
      "eval_PRM Accuracy": 0.8679245283018868,
      "eval_PRM F1": 0.9135802469135802,
      "eval_PRM F1 AUC": 0.8370874803562075,
      "eval_PRM F1 AUC (fixed)": 0.8150864326872709,
      "eval_PRM F1 Neg": 0.72,
      "eval_PRM NPV": 0.6666666666666666,
      "eval_PRM Precision": 0.9367088607594937,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.22103890776634216,
      "eval_runtime": 14.4344,
      "eval_samples_per_second": 3.048,
      "eval_steps_per_second": 0.139,
      "step": 2255
    },
    {
      "epoch": 0.5842478391764592,
      "grad_norm": 3.7327610489730816,
      "learning_rate": 5.505710090398569e-07,
      "loss": 0.3349,
      "step": 2256
    },
    {
      "epoch": 0.5845068142824771,
      "grad_norm": 6.329829598767786,
      "learning_rate": 5.500098642554547e-07,
      "loss": 0.3548,
      "step": 2257
    },
    {
      "epoch": 0.584765789388495,
      "grad_norm": 3.6163394205203336,
      "learning_rate": 5.494487807969944e-07,
      "loss": 0.3523,
      "step": 2258
    },
    {
      "epoch": 0.585024764494513,
      "grad_norm": 3.9800518258515902,
      "learning_rate": 5.488877591233226e-07,
      "loss": 0.3641,
      "step": 2259
    },
    {
      "epoch": 0.5852837396005309,
      "grad_norm": 4.630266533736088,
      "learning_rate": 5.483267996932354e-07,
      "loss": 0.2183,
      "step": 2260
    },
    {
      "epoch": 0.5852837396005309,
      "eval_PRM Accuracy": 0.8584905660377359,
      "eval_PRM F1": 0.906832298136646,
      "eval_PRM F1 AUC": 0.8310633839706653,
      "eval_PRM F1 AUC (fixed)": 0.8101100052383448,
      "eval_PRM F1 Neg": 0.7058823529411765,
      "eval_PRM NPV": 0.6428571428571429,
      "eval_PRM Precision": 0.9358974358974359,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.22330118715763092,
      "eval_runtime": 14.6209,
      "eval_samples_per_second": 3.009,
      "eval_steps_per_second": 0.137,
      "step": 2260
    },
    {
      "epoch": 0.5855427147065488,
      "grad_norm": 3.291643139359937,
      "learning_rate": 5.477659029654782e-07,
      "loss": 0.2556,
      "step": 2261
    },
    {
      "epoch": 0.5858016898125668,
      "grad_norm": 6.8524311658322095,
      "learning_rate": 5.472050693987451e-07,
      "loss": 0.3613,
      "step": 2262
    },
    {
      "epoch": 0.5860606649185847,
      "grad_norm": 2.9061293621530804,
      "learning_rate": 5.466442994516777e-07,
      "loss": 0.2897,
      "step": 2263
    },
    {
      "epoch": 0.5863196400246027,
      "grad_norm": 4.798933814378702,
      "learning_rate": 5.460835935828665e-07,
      "loss": 0.2796,
      "step": 2264
    },
    {
      "epoch": 0.5865786151306206,
      "grad_norm": 3.2675723404747035,
      "learning_rate": 5.455229522508497e-07,
      "loss": 0.3383,
      "step": 2265
    },
    {
      "epoch": 0.5865786151306206,
      "eval_PRM Accuracy": 0.8584905660377359,
      "eval_PRM F1": 0.906832298136646,
      "eval_PRM F1 AUC": 0.8310633839706653,
      "eval_PRM F1 AUC (fixed)": 0.8114195914091147,
      "eval_PRM F1 Neg": 0.7058823529411765,
      "eval_PRM NPV": 0.6428571428571429,
      "eval_PRM Precision": 0.9358974358974359,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.22472737729549408,
      "eval_runtime": 14.4365,
      "eval_samples_per_second": 3.048,
      "eval_steps_per_second": 0.139,
      "step": 2265
    },
    {
      "epoch": 0.5868375902366385,
      "grad_norm": 4.278500753453795,
      "learning_rate": 5.449623759141121e-07,
      "loss": 0.3814,
      "step": 2266
    },
    {
      "epoch": 0.5870965653426564,
      "grad_norm": 5.43312604495216,
      "learning_rate": 5.444018650310857e-07,
      "loss": 0.4116,
      "step": 2267
    },
    {
      "epoch": 0.5873555404486743,
      "grad_norm": 6.850766088023858,
      "learning_rate": 5.438414200601489e-07,
      "loss": 0.4047,
      "step": 2268
    },
    {
      "epoch": 0.5876145155546924,
      "grad_norm": 3.8479515180117962,
      "learning_rate": 5.432810414596259e-07,
      "loss": 0.354,
      "step": 2269
    },
    {
      "epoch": 0.5878734906607103,
      "grad_norm": 2.9892305815172695,
      "learning_rate": 5.427207296877876e-07,
      "loss": 0.2646,
      "step": 2270
    },
    {
      "epoch": 0.5878734906607103,
      "eval_PRM Accuracy": 0.8679245283018868,
      "eval_PRM F1": 0.9135802469135802,
      "eval_PRM F1 AUC": 0.8370874803562075,
      "eval_PRM F1 AUC (fixed)": 0.8129910948140389,
      "eval_PRM F1 Neg": 0.72,
      "eval_PRM NPV": 0.6666666666666666,
      "eval_PRM Precision": 0.9367088607594937,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.22465883195400238,
      "eval_runtime": 14.8336,
      "eval_samples_per_second": 2.966,
      "eval_steps_per_second": 0.135,
      "step": 2270
    },
    {
      "epoch": 0.5881324657667282,
      "grad_norm": 6.27608121054562,
      "learning_rate": 5.421604852028488e-07,
      "loss": 0.4384,
      "step": 2271
    },
    {
      "epoch": 0.5883914408727461,
      "grad_norm": 5.9810879361448634,
      "learning_rate": 5.416003084629705e-07,
      "loss": 0.3532,
      "step": 2272
    },
    {
      "epoch": 0.588650415978764,
      "grad_norm": 3.1487868209796996,
      "learning_rate": 5.410401999262576e-07,
      "loss": 0.2876,
      "step": 2273
    },
    {
      "epoch": 0.588909391084782,
      "grad_norm": 4.930500480790229,
      "learning_rate": 5.404801600507595e-07,
      "loss": 0.3096,
      "step": 2274
    },
    {
      "epoch": 0.5891683661907999,
      "grad_norm": 3.197855315880685,
      "learning_rate": 5.399201892944696e-07,
      "loss": 0.3234,
      "step": 2275
    },
    {
      "epoch": 0.5891683661907999,
      "eval_PRM Accuracy": 0.8773584905660378,
      "eval_PRM F1": 0.9202453987730062,
      "eval_PRM F1 AUC": 0.8431115767417497,
      "eval_PRM F1 AUC (fixed)": 0.8064431639601886,
      "eval_PRM F1 Neg": 0.7346938775510204,
      "eval_PRM NPV": 0.6923076923076923,
      "eval_PRM Precision": 0.9375,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.22794176638126373,
      "eval_runtime": 14.6484,
      "eval_samples_per_second": 3.004,
      "eval_steps_per_second": 0.137,
      "step": 2275
    },
    {
      "epoch": 0.5894273412968178,
      "grad_norm": 4.422992831758531,
      "learning_rate": 5.393602881153244e-07,
      "loss": 0.4035,
      "step": 2276
    },
    {
      "epoch": 0.5896863164028358,
      "grad_norm": 3.6534325829111607,
      "learning_rate": 5.388004569712036e-07,
      "loss": 0.3473,
      "step": 2277
    },
    {
      "epoch": 0.5899452915088537,
      "grad_norm": 3.526368257106306,
      "learning_rate": 5.382406963199302e-07,
      "loss": 0.244,
      "step": 2278
    },
    {
      "epoch": 0.5902042666148717,
      "grad_norm": 5.577024576266806,
      "learning_rate": 5.376810066192685e-07,
      "loss": 0.36,
      "step": 2279
    },
    {
      "epoch": 0.5904632417208896,
      "grad_norm": 7.83817787329655,
      "learning_rate": 5.371213883269255e-07,
      "loss": 0.3629,
      "step": 2280
    },
    {
      "epoch": 0.5904632417208896,
      "eval_PRM Accuracy": 0.8773584905660378,
      "eval_PRM F1": 0.9202453987730062,
      "eval_PRM F1 AUC": 0.8431115767417497,
      "eval_PRM F1 AUC (fixed)": 0.8098480880041907,
      "eval_PRM F1 Neg": 0.7346938775510204,
      "eval_PRM NPV": 0.6923076923076923,
      "eval_PRM Precision": 0.9375,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.22685761749744415,
      "eval_runtime": 14.4598,
      "eval_samples_per_second": 3.043,
      "eval_steps_per_second": 0.138,
      "step": 2280
    },
    {
      "epoch": 0.5907222168269075,
      "grad_norm": 7.143694054161684,
      "learning_rate": 5.365618419005497e-07,
      "loss": 0.3257,
      "step": 2281
    },
    {
      "epoch": 0.5909811919329254,
      "grad_norm": 4.882218377371467,
      "learning_rate": 5.360023677977308e-07,
      "loss": 0.3546,
      "step": 2282
    },
    {
      "epoch": 0.5912401670389433,
      "grad_norm": 4.12450274748871,
      "learning_rate": 5.354429664759993e-07,
      "loss": 0.3561,
      "step": 2283
    },
    {
      "epoch": 0.5914991421449614,
      "grad_norm": 5.237754254451131,
      "learning_rate": 5.348836383928261e-07,
      "loss": 0.4105,
      "step": 2284
    },
    {
      "epoch": 0.5917581172509793,
      "grad_norm": 4.812706619755934,
      "learning_rate": 5.343243840056223e-07,
      "loss": 0.302,
      "step": 2285
    },
    {
      "epoch": 0.5917581172509793,
      "eval_PRM Accuracy": 0.8679245283018868,
      "eval_PRM F1": 0.9135802469135802,
      "eval_PRM F1 AUC": 0.8370874803562075,
      "eval_PRM F1 AUC (fixed)": 0.8080146673651126,
      "eval_PRM F1 Neg": 0.72,
      "eval_PRM NPV": 0.6666666666666666,
      "eval_PRM Precision": 0.9367088607594937,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.23035289347171783,
      "eval_runtime": 14.5377,
      "eval_samples_per_second": 3.027,
      "eval_steps_per_second": 0.138,
      "step": 2285
    },
    {
      "epoch": 0.5920170923569972,
      "grad_norm": 9.162043443385194,
      "learning_rate": 5.337652037717391e-07,
      "loss": 0.3305,
      "step": 2286
    },
    {
      "epoch": 0.5922760674630151,
      "grad_norm": 4.222127778394273,
      "learning_rate": 5.332060981484667e-07,
      "loss": 0.3094,
      "step": 2287
    },
    {
      "epoch": 0.592535042569033,
      "grad_norm": 3.489968734779064,
      "learning_rate": 5.326470675930335e-07,
      "loss": 0.324,
      "step": 2288
    },
    {
      "epoch": 0.592794017675051,
      "grad_norm": 2.9263327818547586,
      "learning_rate": 5.320881125626082e-07,
      "loss": 0.2825,
      "step": 2289
    },
    {
      "epoch": 0.5930529927810689,
      "grad_norm": 3.8872773634138635,
      "learning_rate": 5.315292335142963e-07,
      "loss": 0.3766,
      "step": 2290
    },
    {
      "epoch": 0.5930529927810689,
      "eval_PRM Accuracy": 0.8679245283018868,
      "eval_PRM F1": 0.9146341463414634,
      "eval_PRM F1 AUC": 0.8213724463069669,
      "eval_PRM F1 AUC (fixed)": 0.8077527501309586,
      "eval_PRM F1 Neg": 0.7083333333333334,
      "eval_PRM NPV": 0.68,
      "eval_PRM Precision": 0.9259259259259259,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.2290603518486023,
      "eval_runtime": 14.4607,
      "eval_samples_per_second": 3.043,
      "eval_steps_per_second": 0.138,
      "step": 2290
    },
    {
      "epoch": 0.5933119678870868,
      "grad_norm": 4.444858797481897,
      "learning_rate": 5.30970430905142e-07,
      "loss": 0.2666,
      "step": 2291
    },
    {
      "epoch": 0.5935709429931048,
      "grad_norm": 4.984446671939108,
      "learning_rate": 5.304117051921266e-07,
      "loss": 0.3139,
      "step": 2292
    },
    {
      "epoch": 0.5938299180991227,
      "grad_norm": 6.110633677057603,
      "learning_rate": 5.298530568321684e-07,
      "loss": 0.338,
      "step": 2293
    },
    {
      "epoch": 0.5940888932051407,
      "grad_norm": 4.011705385961452,
      "learning_rate": 5.292944862821229e-07,
      "loss": 0.2922,
      "step": 2294
    },
    {
      "epoch": 0.5943478683111586,
      "grad_norm": 2.5971981727513698,
      "learning_rate": 5.287359939987819e-07,
      "loss": 0.2626,
      "step": 2295
    },
    {
      "epoch": 0.5943478683111586,
      "eval_PRM Accuracy": 0.8679245283018868,
      "eval_PRM F1": 0.9146341463414634,
      "eval_PRM F1 AUC": 0.8213724463069669,
      "eval_PRM F1 AUC (fixed)": 0.8140387637506548,
      "eval_PRM F1 Neg": 0.7083333333333334,
      "eval_PRM NPV": 0.68,
      "eval_PRM Precision": 0.9259259259259259,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.22724123299121857,
      "eval_runtime": 14.9819,
      "eval_samples_per_second": 2.937,
      "eval_steps_per_second": 0.133,
      "step": 2295
    },
    {
      "epoch": 0.5946068434171765,
      "grad_norm": 4.506026919855117,
      "learning_rate": 5.281775804388725e-07,
      "loss": 0.3156,
      "step": 2296
    },
    {
      "epoch": 0.5948658185231944,
      "grad_norm": 3.5863945003201807,
      "learning_rate": 5.276192460590583e-07,
      "loss": 0.2951,
      "step": 2297
    },
    {
      "epoch": 0.5951247936292124,
      "grad_norm": 3.55230802619799,
      "learning_rate": 5.270609913159374e-07,
      "loss": 0.2985,
      "step": 2298
    },
    {
      "epoch": 0.5953837687352304,
      "grad_norm": 3.9496933069592184,
      "learning_rate": 5.265028166660435e-07,
      "loss": 0.3627,
      "step": 2299
    },
    {
      "epoch": 0.5956427438412483,
      "grad_norm": 3.884140136414765,
      "learning_rate": 5.259447225658444e-07,
      "loss": 0.3562,
      "step": 2300
    },
    {
      "epoch": 0.5956427438412483,
      "eval_PRM Accuracy": 0.8679245283018868,
      "eval_PRM F1": 0.9146341463414634,
      "eval_PRM F1 AUC": 0.8213724463069669,
      "eval_PRM F1 AUC (fixed)": 0.819800942902043,
      "eval_PRM F1 Neg": 0.7083333333333334,
      "eval_PRM NPV": 0.68,
      "eval_PRM Precision": 0.9259259259259259,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.22628051042556763,
      "eval_runtime": 14.6184,
      "eval_samples_per_second": 3.01,
      "eval_steps_per_second": 0.137,
      "step": 2300
    },
    {
      "epoch": 0.5959017189472662,
      "grad_norm": 2.803080960035547,
      "learning_rate": 5.25386709471742e-07,
      "loss": 0.2385,
      "step": 2301
    },
    {
      "epoch": 0.5961606940532841,
      "grad_norm": 3.590886795732654,
      "learning_rate": 5.248287778400721e-07,
      "loss": 0.3056,
      "step": 2302
    },
    {
      "epoch": 0.5964196691593021,
      "grad_norm": 6.223688301425503,
      "learning_rate": 5.242709281271039e-07,
      "loss": 0.3638,
      "step": 2303
    },
    {
      "epoch": 0.59667864426532,
      "grad_norm": 3.1440767779776904,
      "learning_rate": 5.237131607890393e-07,
      "loss": 0.2493,
      "step": 2304
    },
    {
      "epoch": 0.5969376193713379,
      "grad_norm": 6.546539151522905,
      "learning_rate": 5.231554762820131e-07,
      "loss": 0.4399,
      "step": 2305
    },
    {
      "epoch": 0.5969376193713379,
      "eval_PRM Accuracy": 0.8679245283018868,
      "eval_PRM F1": 0.9146341463414634,
      "eval_PRM F1 AUC": 0.8213724463069669,
      "eval_PRM F1 AUC (fixed)": 0.8234677841801991,
      "eval_PRM F1 Neg": 0.7083333333333334,
      "eval_PRM NPV": 0.68,
      "eval_PRM Precision": 0.9259259259259259,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.2260231077671051,
      "eval_runtime": 14.4648,
      "eval_samples_per_second": 3.042,
      "eval_steps_per_second": 0.138,
      "step": 2305
    },
    {
      "epoch": 0.5971965944773558,
      "grad_norm": 5.826692783459863,
      "learning_rate": 5.225978750620925e-07,
      "loss": 0.448,
      "step": 2306
    },
    {
      "epoch": 0.5974555695833738,
      "grad_norm": 4.585479626981312,
      "learning_rate": 5.220403575852762e-07,
      "loss": 0.2979,
      "step": 2307
    },
    {
      "epoch": 0.5977145446893918,
      "grad_norm": 5.369977285613527,
      "learning_rate": 5.214829243074947e-07,
      "loss": 0.3567,
      "step": 2308
    },
    {
      "epoch": 0.5979735197954097,
      "grad_norm": 7.260631025503597,
      "learning_rate": 5.209255756846095e-07,
      "loss": 0.3173,
      "step": 2309
    },
    {
      "epoch": 0.5982324949014276,
      "grad_norm": 4.824436102176374,
      "learning_rate": 5.203683121724131e-07,
      "loss": 0.3262,
      "step": 2310
    },
    {
      "epoch": 0.5982324949014276,
      "eval_PRM Accuracy": 0.8773584905660378,
      "eval_PRM F1": 0.9202453987730062,
      "eval_PRM F1 AUC": 0.8431115767417497,
      "eval_PRM F1 AUC (fixed)": 0.8205866946045051,
      "eval_PRM F1 Neg": 0.7346938775510204,
      "eval_PRM NPV": 0.6923076923076923,
      "eval_PRM Precision": 0.9375,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.22923235595226288,
      "eval_runtime": 14.6964,
      "eval_samples_per_second": 2.994,
      "eval_steps_per_second": 0.136,
      "step": 2310
    },
    {
      "epoch": 0.5984914700074455,
      "grad_norm": 4.351065677152797,
      "learning_rate": 5.198111342266285e-07,
      "loss": 0.2893,
      "step": 2311
    },
    {
      "epoch": 0.5987504451134634,
      "grad_norm": 9.003428247954938,
      "learning_rate": 5.192540423029078e-07,
      "loss": 0.3982,
      "step": 2312
    },
    {
      "epoch": 0.5990094202194814,
      "grad_norm": 3.927436107744623,
      "learning_rate": 5.186970368568335e-07,
      "loss": 0.3224,
      "step": 2313
    },
    {
      "epoch": 0.5992683953254994,
      "grad_norm": 3.573622542029136,
      "learning_rate": 5.181401183439177e-07,
      "loss": 0.2907,
      "step": 2314
    },
    {
      "epoch": 0.5995273704315173,
      "grad_norm": 4.489549657485354,
      "learning_rate": 5.175832872196009e-07,
      "loss": 0.3904,
      "step": 2315
    },
    {
      "epoch": 0.5995273704315173,
      "eval_PRM Accuracy": 0.8679245283018868,
      "eval_PRM F1": 0.9135802469135802,
      "eval_PRM F1 AUC": 0.8370874803562075,
      "eval_PRM F1 AUC (fixed)": 0.8190151911995809,
      "eval_PRM F1 Neg": 0.72,
      "eval_PRM NPV": 0.6666666666666666,
      "eval_PRM Precision": 0.9367088607594937,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.23295408487319946,
      "eval_runtime": 14.5816,
      "eval_samples_per_second": 3.017,
      "eval_steps_per_second": 0.137,
      "step": 2315
    },
    {
      "epoch": 0.5997863455375352,
      "grad_norm": 7.429627477344885,
      "learning_rate": 5.17026543939252e-07,
      "loss": 0.4014,
      "step": 2316
    },
    {
      "epoch": 0.6000453206435531,
      "grad_norm": 3.6016570031324155,
      "learning_rate": 5.164698889581686e-07,
      "loss": 0.3197,
      "step": 2317
    },
    {
      "epoch": 0.6003042957495711,
      "grad_norm": 5.730398851098755,
      "learning_rate": 5.159133227315756e-07,
      "loss": 0.3499,
      "step": 2318
    },
    {
      "epoch": 0.600563270855589,
      "grad_norm": 3.777252827295177,
      "learning_rate": 5.153568457146253e-07,
      "loss": 0.3371,
      "step": 2319
    },
    {
      "epoch": 0.6008222459616069,
      "grad_norm": 3.500151956044879,
      "learning_rate": 5.148004583623981e-07,
      "loss": 0.4022,
      "step": 2320
    },
    {
      "epoch": 0.6008222459616069,
      "eval_PRM Accuracy": 0.8679245283018868,
      "eval_PRM F1": 0.9135802469135802,
      "eval_PRM F1 AUC": 0.8370874803562075,
      "eval_PRM F1 AUC (fixed)": 0.8127291775798847,
      "eval_PRM F1 Neg": 0.72,
      "eval_PRM NPV": 0.6666666666666666,
      "eval_PRM Precision": 0.9367088607594937,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.23461176455020905,
      "eval_runtime": 14.5346,
      "eval_samples_per_second": 3.027,
      "eval_steps_per_second": 0.138,
      "step": 2320
    },
    {
      "epoch": 0.6010812210676248,
      "grad_norm": 5.52560126144034,
      "learning_rate": 5.142441611298992e-07,
      "loss": 0.2864,
      "step": 2321
    },
    {
      "epoch": 0.6013401961736428,
      "grad_norm": 5.127932387654273,
      "learning_rate": 5.136879544720615e-07,
      "loss": 0.4347,
      "step": 2322
    },
    {
      "epoch": 0.6015991712796608,
      "grad_norm": 4.736951514092038,
      "learning_rate": 5.131318388437434e-07,
      "loss": 0.3297,
      "step": 2323
    },
    {
      "epoch": 0.6018581463856787,
      "grad_norm": 5.099841867378184,
      "learning_rate": 5.125758146997291e-07,
      "loss": 0.2558,
      "step": 2324
    },
    {
      "epoch": 0.6021171214916966,
      "grad_norm": 7.287427114768257,
      "learning_rate": 5.120198824947277e-07,
      "loss": 0.2713,
      "step": 2325
    },
    {
      "epoch": 0.6021171214916966,
      "eval_PRM Accuracy": 0.8773584905660378,
      "eval_PRM F1": 0.9202453987730062,
      "eval_PRM F1 AUC": 0.8431115767417497,
      "eval_PRM F1 AUC (fixed)": 0.8124672603457308,
      "eval_PRM F1 Neg": 0.7346938775510204,
      "eval_PRM NPV": 0.6923076923076923,
      "eval_PRM Precision": 0.9375,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.23834198713302612,
      "eval_runtime": 14.5425,
      "eval_samples_per_second": 3.026,
      "eval_steps_per_second": 0.138,
      "step": 2325
    },
    {
      "epoch": 0.6023760965977145,
      "grad_norm": 6.162202130719232,
      "learning_rate": 5.114640426833733e-07,
      "loss": 0.2877,
      "step": 2326
    },
    {
      "epoch": 0.6026350717037324,
      "grad_norm": 5.7173371015592815,
      "learning_rate": 5.109082957202242e-07,
      "loss": 0.3901,
      "step": 2327
    },
    {
      "epoch": 0.6028940468097505,
      "grad_norm": 6.090873045974941,
      "learning_rate": 5.103526420597631e-07,
      "loss": 0.274,
      "step": 2328
    },
    {
      "epoch": 0.6031530219157684,
      "grad_norm": 2.748688128279512,
      "learning_rate": 5.09797082156396e-07,
      "loss": 0.3142,
      "step": 2329
    },
    {
      "epoch": 0.6034119970217863,
      "grad_norm": 3.7416385204574256,
      "learning_rate": 5.092416164644525e-07,
      "loss": 0.2645,
      "step": 2330
    },
    {
      "epoch": 0.6034119970217863,
      "eval_PRM Accuracy": 0.8679245283018868,
      "eval_PRM F1": 0.9146341463414634,
      "eval_PRM F1 AUC": 0.8213724463069669,
      "eval_PRM F1 AUC (fixed)": 0.8129910948140388,
      "eval_PRM F1 Neg": 0.7083333333333334,
      "eval_PRM NPV": 0.68,
      "eval_PRM Precision": 0.9259259259259259,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.23703013360500336,
      "eval_runtime": 14.3169,
      "eval_samples_per_second": 3.073,
      "eval_steps_per_second": 0.14,
      "step": 2330
    },
    {
      "epoch": 0.6036709721278042,
      "grad_norm": 3.049982448067719,
      "learning_rate": 5.086862454381853e-07,
      "loss": 0.235,
      "step": 2331
    },
    {
      "epoch": 0.6039299472338221,
      "grad_norm": 4.406067101790021,
      "learning_rate": 5.081309695317691e-07,
      "loss": 0.3462,
      "step": 2332
    },
    {
      "epoch": 0.6041889223398401,
      "grad_norm": 2.837764492129325,
      "learning_rate": 5.075757891993014e-07,
      "loss": 0.291,
      "step": 2333
    },
    {
      "epoch": 0.604447897445858,
      "grad_norm": 2.9969030812055695,
      "learning_rate": 5.070207048948015e-07,
      "loss": 0.3514,
      "step": 2334
    },
    {
      "epoch": 0.6047068725518759,
      "grad_norm": 4.4106084889161465,
      "learning_rate": 5.064657170722099e-07,
      "loss": 0.3263,
      "step": 2335
    },
    {
      "epoch": 0.6047068725518759,
      "eval_PRM Accuracy": 0.8679245283018868,
      "eval_PRM F1": 0.9146341463414634,
      "eval_PRM F1 AUC": 0.8213724463069669,
      "eval_PRM F1 AUC (fixed)": 0.8182294394971189,
      "eval_PRM F1 Neg": 0.7083333333333334,
      "eval_PRM NPV": 0.68,
      "eval_PRM Precision": 0.9259259259259259,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.2369740605354309,
      "eval_runtime": 14.8331,
      "eval_samples_per_second": 2.966,
      "eval_steps_per_second": 0.135,
      "step": 2335
    },
    {
      "epoch": 0.6049658476578939,
      "grad_norm": 3.473315940260668,
      "learning_rate": 5.059108261853885e-07,
      "loss": 0.323,
      "step": 2336
    },
    {
      "epoch": 0.6052248227639119,
      "grad_norm": 3.7234142446410616,
      "learning_rate": 5.053560326881192e-07,
      "loss": 0.301,
      "step": 2337
    },
    {
      "epoch": 0.6054837978699298,
      "grad_norm": 4.2045041030842585,
      "learning_rate": 5.048013370341051e-07,
      "loss": 0.3269,
      "step": 2338
    },
    {
      "epoch": 0.6057427729759477,
      "grad_norm": 2.933931553198188,
      "learning_rate": 5.042467396769692e-07,
      "loss": 0.3188,
      "step": 2339
    },
    {
      "epoch": 0.6060017480819656,
      "grad_norm": 6.01761944179701,
      "learning_rate": 5.036922410702535e-07,
      "loss": 0.3976,
      "step": 2340
    },
    {
      "epoch": 0.6060017480819656,
      "eval_PRM Accuracy": 0.8584905660377359,
      "eval_PRM F1": 0.9079754601226994,
      "eval_PRM F1 AUC": 0.8153483499214248,
      "eval_PRM F1 AUC (fixed)": 0.819015191199581,
      "eval_PRM F1 Neg": 0.6938775510204082,
      "eval_PRM NPV": 0.6538461538461539,
      "eval_PRM Precision": 0.925,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.23695515096187592,
      "eval_runtime": 14.2419,
      "eval_samples_per_second": 3.089,
      "eval_steps_per_second": 0.14,
      "step": 2340
    },
    {
      "epoch": 0.6062607231879835,
      "grad_norm": 3.136076734575955,
      "learning_rate": 5.031378416674198e-07,
      "loss": 0.351,
      "step": 2341
    },
    {
      "epoch": 0.6065196982940015,
      "grad_norm": 4.2714072870569115,
      "learning_rate": 5.025835419218487e-07,
      "loss": 0.3199,
      "step": 2342
    },
    {
      "epoch": 0.6067786734000195,
      "grad_norm": 3.4600802296061266,
      "learning_rate": 5.02029342286839e-07,
      "loss": 0.3245,
      "step": 2343
    },
    {
      "epoch": 0.6070376485060374,
      "grad_norm": 5.526098754094361,
      "learning_rate": 5.014752432156079e-07,
      "loss": 0.3737,
      "step": 2344
    },
    {
      "epoch": 0.6072966236120553,
      "grad_norm": 4.330743639683585,
      "learning_rate": 5.009212451612908e-07,
      "loss": 0.3357,
      "step": 2345
    },
    {
      "epoch": 0.6072966236120553,
      "eval_PRM Accuracy": 0.8679245283018868,
      "eval_PRM F1": 0.9135802469135802,
      "eval_PRM F1 AUC": 0.8370874803562075,
      "eval_PRM F1 AUC (fixed)": 0.817967522262965,
      "eval_PRM F1 Neg": 0.72,
      "eval_PRM NPV": 0.6666666666666666,
      "eval_PRM Precision": 0.9367088607594937,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.2365497201681137,
      "eval_runtime": 14.6843,
      "eval_samples_per_second": 2.996,
      "eval_steps_per_second": 0.136,
      "step": 2345
    },
    {
      "epoch": 0.6075555987180732,
      "grad_norm": 5.737102131056715,
      "learning_rate": 5.003673485769392e-07,
      "loss": 0.3686,
      "step": 2346
    },
    {
      "epoch": 0.6078145738240912,
      "grad_norm": 4.19893911591526,
      "learning_rate": 4.998135539155225e-07,
      "loss": 0.2999,
      "step": 2347
    },
    {
      "epoch": 0.6080735489301091,
      "grad_norm": 4.088030771593054,
      "learning_rate": 4.992598616299271e-07,
      "loss": 0.3421,
      "step": 2348
    },
    {
      "epoch": 0.608332524036127,
      "grad_norm": 6.204583231626178,
      "learning_rate": 4.987062721729551e-07,
      "loss": 0.4174,
      "step": 2349
    },
    {
      "epoch": 0.6085914991421449,
      "grad_norm": 4.002537414198088,
      "learning_rate": 4.981527859973245e-07,
      "loss": 0.4131,
      "step": 2350
    },
    {
      "epoch": 0.6085914991421449,
      "eval_PRM Accuracy": 0.8679245283018868,
      "eval_PRM F1": 0.9135802469135802,
      "eval_PRM F1 AUC": 0.8370874803562075,
      "eval_PRM F1 AUC (fixed)": 0.817705605028811,
      "eval_PRM F1 Neg": 0.72,
      "eval_PRM NPV": 0.6666666666666666,
      "eval_PRM Precision": 0.9367088607594937,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.23777227103710175,
      "eval_runtime": 14.6117,
      "eval_samples_per_second": 3.011,
      "eval_steps_per_second": 0.137,
      "step": 2350
    },
    {
      "epoch": 0.6088504742481629,
      "grad_norm": 4.2720702303454035,
      "learning_rate": 4.975994035556691e-07,
      "loss": 0.2849,
      "step": 2351
    },
    {
      "epoch": 0.6091094493541809,
      "grad_norm": 3.95336453653184,
      "learning_rate": 4.970461253005377e-07,
      "loss": 0.3715,
      "step": 2352
    },
    {
      "epoch": 0.6093684244601988,
      "grad_norm": 3.192350411167091,
      "learning_rate": 4.96492951684394e-07,
      "loss": 0.3248,
      "step": 2353
    },
    {
      "epoch": 0.6096273995662167,
      "grad_norm": 5.155666539142483,
      "learning_rate": 4.95939883159616e-07,
      "loss": 0.44,
      "step": 2354
    },
    {
      "epoch": 0.6098863746722346,
      "grad_norm": 3.924089888645864,
      "learning_rate": 4.953869201784957e-07,
      "loss": 0.3178,
      "step": 2355
    },
    {
      "epoch": 0.6098863746722346,
      "eval_PRM Accuracy": 0.8584905660377359,
      "eval_PRM F1": 0.906832298136646,
      "eval_PRM F1 AUC": 0.8310633839706653,
      "eval_PRM F1 AUC (fixed)": 0.8182294394971189,
      "eval_PRM F1 Neg": 0.7058823529411765,
      "eval_PRM NPV": 0.6428571428571429,
      "eval_PRM Precision": 0.9358974358974359,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.2385021150112152,
      "eval_runtime": 14.4461,
      "eval_samples_per_second": 3.046,
      "eval_steps_per_second": 0.138,
      "step": 2355
    },
    {
      "epoch": 0.6101453497782525,
      "grad_norm": 6.222350049276512,
      "learning_rate": 4.948340631932391e-07,
      "loss": 0.3054,
      "step": 2356
    },
    {
      "epoch": 0.6104043248842705,
      "grad_norm": 5.328580064917684,
      "learning_rate": 4.942813126559653e-07,
      "loss": 0.4344,
      "step": 2357
    },
    {
      "epoch": 0.6106632999902885,
      "grad_norm": 3.620327135954641,
      "learning_rate": 4.937286690187064e-07,
      "loss": 0.3425,
      "step": 2358
    },
    {
      "epoch": 0.6109222750963064,
      "grad_norm": 3.714367006600273,
      "learning_rate": 4.93176132733407e-07,
      "loss": 0.212,
      "step": 2359
    },
    {
      "epoch": 0.6111812502023243,
      "grad_norm": 5.907442102095968,
      "learning_rate": 4.926237042519241e-07,
      "loss": 0.2924,
      "step": 2360
    },
    {
      "epoch": 0.6111812502023243,
      "eval_PRM Accuracy": 0.8679245283018868,
      "eval_PRM F1": 0.9135802469135802,
      "eval_PRM F1 AUC": 0.8370874803562075,
      "eval_PRM F1 AUC (fixed)": 0.8161341016238868,
      "eval_PRM F1 Neg": 0.72,
      "eval_PRM NPV": 0.6666666666666666,
      "eval_PRM Precision": 0.9367088607594937,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.23777775466442108,
      "eval_runtime": 14.5792,
      "eval_samples_per_second": 3.018,
      "eval_steps_per_second": 0.137,
      "step": 2360
    },
    {
      "epoch": 0.6114402253083422,
      "grad_norm": 3.644225417297505,
      "learning_rate": 4.920713840260264e-07,
      "loss": 0.312,
      "step": 2361
    },
    {
      "epoch": 0.6116992004143602,
      "grad_norm": 3.3101561514933775,
      "learning_rate": 4.915191725073939e-07,
      "loss": 0.2833,
      "step": 2362
    },
    {
      "epoch": 0.6119581755203781,
      "grad_norm": 3.3293812090467014,
      "learning_rate": 4.90967070147618e-07,
      "loss": 0.2997,
      "step": 2363
    },
    {
      "epoch": 0.612217150626396,
      "grad_norm": 6.104526017326705,
      "learning_rate": 4.904150773982007e-07,
      "loss": 0.3987,
      "step": 2364
    },
    {
      "epoch": 0.612476125732414,
      "grad_norm": 4.676784701578161,
      "learning_rate": 4.898631947105543e-07,
      "loss": 0.3368,
      "step": 2365
    },
    {
      "epoch": 0.612476125732414,
      "eval_PRM Accuracy": 0.8584905660377359,
      "eval_PRM F1": 0.9079754601226994,
      "eval_PRM F1 AUC": 0.8153483499214248,
      "eval_PRM F1 AUC (fixed)": 0.8203247773703509,
      "eval_PRM F1 Neg": 0.6938775510204082,
      "eval_PRM NPV": 0.6538461538461539,
      "eval_PRM Precision": 0.925,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.23475049436092377,
      "eval_runtime": 14.8539,
      "eval_samples_per_second": 2.962,
      "eval_steps_per_second": 0.135,
      "step": 2365
    },
    {
      "epoch": 0.6127351008384319,
      "grad_norm": 3.2625583926007073,
      "learning_rate": 4.893114225360013e-07,
      "loss": 0.2969,
      "step": 2366
    },
    {
      "epoch": 0.6129940759444499,
      "grad_norm": 3.884855687053449,
      "learning_rate": 4.887597613257738e-07,
      "loss": 0.2576,
      "step": 2367
    },
    {
      "epoch": 0.6132530510504678,
      "grad_norm": 5.006277623096616,
      "learning_rate": 4.882082115310125e-07,
      "loss": 0.428,
      "step": 2368
    },
    {
      "epoch": 0.6135120261564857,
      "grad_norm": 3.764357692744137,
      "learning_rate": 4.876567736027682e-07,
      "loss": 0.333,
      "step": 2369
    },
    {
      "epoch": 0.6137710012625036,
      "grad_norm": 3.242239744435423,
      "learning_rate": 4.871054479919994e-07,
      "loss": 0.3275,
      "step": 2370
    },
    {
      "epoch": 0.6137710012625036,
      "eval_PRM Accuracy": 0.8679245283018868,
      "eval_PRM F1": 0.9135802469135802,
      "eval_PRM F1 AUC": 0.8370874803562075,
      "eval_PRM F1 AUC (fixed)": 0.8232058669460451,
      "eval_PRM F1 Neg": 0.72,
      "eval_PRM NPV": 0.6666666666666666,
      "eval_PRM Precision": 0.9367088607594937,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.23275423049926758,
      "eval_runtime": 14.5487,
      "eval_samples_per_second": 3.024,
      "eval_steps_per_second": 0.137,
      "step": 2370
    },
    {
      "epoch": 0.6140299763685215,
      "grad_norm": 3.5700555019641116,
      "learning_rate": 4.865542351495726e-07,
      "loss": 0.3052,
      "step": 2371
    },
    {
      "epoch": 0.6142889514745395,
      "grad_norm": 8.724531980269637,
      "learning_rate": 4.860031355262624e-07,
      "loss": 0.2398,
      "step": 2372
    },
    {
      "epoch": 0.6145479265805575,
      "grad_norm": 5.446197309078166,
      "learning_rate": 4.854521495727511e-07,
      "loss": 0.4156,
      "step": 2373
    },
    {
      "epoch": 0.6148069016865754,
      "grad_norm": 2.6675458187702557,
      "learning_rate": 4.849012777396275e-07,
      "loss": 0.2812,
      "step": 2374
    },
    {
      "epoch": 0.6150658767925933,
      "grad_norm": 4.1932196260069725,
      "learning_rate": 4.843505204773874e-07,
      "loss": 0.3317,
      "step": 2375
    },
    {
      "epoch": 0.6150658767925933,
      "eval_PRM Accuracy": 0.8679245283018868,
      "eval_PRM F1": 0.9135802469135802,
      "eval_PRM F1 AUC": 0.8370874803562075,
      "eval_PRM F1 AUC (fixed)": 0.8255631220534312,
      "eval_PRM F1 Neg": 0.72,
      "eval_PRM NPV": 0.6666666666666666,
      "eval_PRM Precision": 0.9367088607594937,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.23194923996925354,
      "eval_runtime": 14.7791,
      "eval_samples_per_second": 2.977,
      "eval_steps_per_second": 0.135,
      "step": 2375
    },
    {
      "epoch": 0.6153248518986112,
      "grad_norm": 6.684504764013728,
      "learning_rate": 4.837998782364326e-07,
      "loss": 0.3322,
      "step": 2376
    },
    {
      "epoch": 0.6155838270046292,
      "grad_norm": 4.9721202222180105,
      "learning_rate": 4.832493514670715e-07,
      "loss": 0.2714,
      "step": 2377
    },
    {
      "epoch": 0.6158428021106471,
      "grad_norm": 3.4444549153452204,
      "learning_rate": 4.826989406195174e-07,
      "loss": 0.3296,
      "step": 2378
    },
    {
      "epoch": 0.616101777216665,
      "grad_norm": 7.1400272483588,
      "learning_rate": 4.821486461438887e-07,
      "loss": 0.2508,
      "step": 2379
    },
    {
      "epoch": 0.616360752322683,
      "grad_norm": 4.592701792555213,
      "learning_rate": 4.815984684902093e-07,
      "loss": 0.3134,
      "step": 2380
    },
    {
      "epoch": 0.616360752322683,
      "eval_PRM Accuracy": 0.8679245283018868,
      "eval_PRM F1": 0.9135802469135802,
      "eval_PRM F1 AUC": 0.8370874803562075,
      "eval_PRM F1 AUC (fixed)": 0.8250392875851231,
      "eval_PRM F1 Neg": 0.72,
      "eval_PRM NPV": 0.6666666666666666,
      "eval_PRM Precision": 0.9367088607594937,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.2293485701084137,
      "eval_runtime": 14.7839,
      "eval_samples_per_second": 2.976,
      "eval_steps_per_second": 0.135,
      "step": 2380
    },
    {
      "epoch": 0.616619727428701,
      "grad_norm": 4.1378033567103,
      "learning_rate": 4.810484081084071e-07,
      "loss": 0.3202,
      "step": 2381
    },
    {
      "epoch": 0.6168787025347189,
      "grad_norm": 6.358922918852487,
      "learning_rate": 4.80498465448314e-07,
      "loss": 0.3207,
      "step": 2382
    },
    {
      "epoch": 0.6171376776407368,
      "grad_norm": 4.084546524317195,
      "learning_rate": 4.799486409596664e-07,
      "loss": 0.3072,
      "step": 2383
    },
    {
      "epoch": 0.6173966527467547,
      "grad_norm": 3.6986109532827554,
      "learning_rate": 4.79398935092103e-07,
      "loss": 0.2674,
      "step": 2384
    },
    {
      "epoch": 0.6176556278527726,
      "grad_norm": 6.587007465434324,
      "learning_rate": 4.788493482951663e-07,
      "loss": 0.4138,
      "step": 2385
    },
    {
      "epoch": 0.6176556278527726,
      "eval_PRM Accuracy": 0.8679245283018868,
      "eval_PRM F1": 0.9135802469135802,
      "eval_PRM F1 AUC": 0.8370874803562075,
      "eval_PRM F1 AUC (fixed)": 0.8308014667365113,
      "eval_PRM F1 Neg": 0.72,
      "eval_PRM NPV": 0.6666666666666666,
      "eval_PRM Precision": 0.9367088607594937,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.22619368135929108,
      "eval_runtime": 14.672,
      "eval_samples_per_second": 2.999,
      "eval_steps_per_second": 0.136,
      "step": 2385
    },
    {
      "epoch": 0.6179146029587906,
      "grad_norm": 5.698757399162513,
      "learning_rate": 4.782998810183009e-07,
      "loss": 0.3895,
      "step": 2386
    },
    {
      "epoch": 0.6181735780648085,
      "grad_norm": 5.597625908069988,
      "learning_rate": 4.77750533710854e-07,
      "loss": 0.3666,
      "step": 2387
    },
    {
      "epoch": 0.6184325531708265,
      "grad_norm": 5.837415297274771,
      "learning_rate": 4.772013068220745e-07,
      "loss": 0.3374,
      "step": 2388
    },
    {
      "epoch": 0.6186915282768444,
      "grad_norm": 7.604970810164243,
      "learning_rate": 4.766522008011129e-07,
      "loss": 0.3257,
      "step": 2389
    },
    {
      "epoch": 0.6189505033828623,
      "grad_norm": 3.619550445908711,
      "learning_rate": 4.761032160970209e-07,
      "loss": 0.3317,
      "step": 2390
    },
    {
      "epoch": 0.6189505033828623,
      "eval_PRM Accuracy": 0.8773584905660378,
      "eval_PRM F1": 0.9202453987730062,
      "eval_PRM F1 AUC": 0.8431115767417497,
      "eval_PRM F1 AUC (fixed)": 0.8331587218438974,
      "eval_PRM F1 Neg": 0.7346938775510204,
      "eval_PRM NPV": 0.6923076923076923,
      "eval_PRM Precision": 0.9375,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.22543154656887054,
      "eval_runtime": 14.6702,
      "eval_samples_per_second": 2.999,
      "eval_steps_per_second": 0.136,
      "step": 2390
    },
    {
      "epoch": 0.6192094784888803,
      "grad_norm": 4.380198615467105,
      "learning_rate": 4.7555435315875087e-07,
      "loss": 0.2112,
      "step": 2391
    },
    {
      "epoch": 0.6194684535948982,
      "grad_norm": 5.284679446627228,
      "learning_rate": 4.7500561243515553e-07,
      "loss": 0.267,
      "step": 2392
    },
    {
      "epoch": 0.6197274287009161,
      "grad_norm": 7.795372438423196,
      "learning_rate": 4.7445699437498813e-07,
      "loss": 0.3489,
      "step": 2393
    },
    {
      "epoch": 0.619986403806934,
      "grad_norm": 4.195507579896319,
      "learning_rate": 4.73908499426901e-07,
      "loss": 0.3786,
      "step": 2394
    },
    {
      "epoch": 0.620245378912952,
      "grad_norm": 3.545972131929893,
      "learning_rate": 4.733601280394465e-07,
      "loss": 0.316,
      "step": 2395
    },
    {
      "epoch": 0.620245378912952,
      "eval_PRM Accuracy": 0.8773584905660378,
      "eval_PRM F1": 0.9202453987730062,
      "eval_PRM F1 AUC": 0.8431115767417497,
      "eval_PRM F1 AUC (fixed)": 0.8347302252488213,
      "eval_PRM F1 Neg": 0.7346938775510204,
      "eval_PRM NPV": 0.6923076923076923,
      "eval_PRM Precision": 0.9375,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.22437819838523865,
      "eval_runtime": 14.5573,
      "eval_samples_per_second": 3.023,
      "eval_steps_per_second": 0.137,
      "step": 2395
    },
    {
      "epoch": 0.62050435401897,
      "grad_norm": 5.3927538256538,
      "learning_rate": 4.728118806610748e-07,
      "loss": 0.4223,
      "step": 2396
    },
    {
      "epoch": 0.6207633291249879,
      "grad_norm": 3.951021859249228,
      "learning_rate": 4.722637577401358e-07,
      "loss": 0.3573,
      "step": 2397
    },
    {
      "epoch": 0.6210223042310058,
      "grad_norm": 3.9469472547128457,
      "learning_rate": 4.7171575972487715e-07,
      "loss": 0.3193,
      "step": 2398
    },
    {
      "epoch": 0.6212812793370237,
      "grad_norm": 6.4923830122598885,
      "learning_rate": 4.7116788706344426e-07,
      "loss": 0.3595,
      "step": 2399
    },
    {
      "epoch": 0.6215402544430416,
      "grad_norm": 4.4851385015931315,
      "learning_rate": 4.706201402038803e-07,
      "loss": 0.3986,
      "step": 2400
    },
    {
      "epoch": 0.6215402544430416,
      "eval_PRM Accuracy": 0.8679245283018868,
      "eval_PRM F1": 0.9135802469135802,
      "eval_PRM F1 AUC": 0.8370874803562075,
      "eval_PRM F1 AUC (fixed)": 0.8373493975903614,
      "eval_PRM F1 Neg": 0.72,
      "eval_PRM NPV": 0.6666666666666666,
      "eval_PRM Precision": 0.9367088607594937,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.22535786032676697,
      "eval_runtime": 14.6734,
      "eval_samples_per_second": 2.999,
      "eval_steps_per_second": 0.136,
      "step": 2400
    },
    {
      "epoch": 0.6217992295490596,
      "grad_norm": 3.9365481487623986,
      "learning_rate": 4.7007251959412527e-07,
      "loss": 0.3561,
      "step": 2401
    },
    {
      "epoch": 0.6220582046550776,
      "grad_norm": 4.902452759226286,
      "learning_rate": 4.6952502568201606e-07,
      "loss": 0.37,
      "step": 2402
    },
    {
      "epoch": 0.6223171797610955,
      "grad_norm": 4.409358496752497,
      "learning_rate": 4.689776589152863e-07,
      "loss": 0.3756,
      "step": 2403
    },
    {
      "epoch": 0.6225761548671134,
      "grad_norm": 3.4620920301211275,
      "learning_rate": 4.6843041974156486e-07,
      "loss": 0.3295,
      "step": 2404
    },
    {
      "epoch": 0.6228351299731313,
      "grad_norm": 3.9973450116573055,
      "learning_rate": 4.678833086083769e-07,
      "loss": 0.3845,
      "step": 2405
    },
    {
      "epoch": 0.6228351299731313,
      "eval_PRM Accuracy": 0.8679245283018868,
      "eval_PRM F1": 0.9135802469135802,
      "eval_PRM F1 AUC": 0.8370874803562075,
      "eval_PRM F1 AUC (fixed)": 0.8310633839706654,
      "eval_PRM F1 Neg": 0.72,
      "eval_PRM NPV": 0.6666666666666666,
      "eval_PRM Precision": 0.9367088607594937,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.23237180709838867,
      "eval_runtime": 14.5746,
      "eval_samples_per_second": 3.019,
      "eval_steps_per_second": 0.137,
      "step": 2405
    },
    {
      "epoch": 0.6230941050791493,
      "grad_norm": 5.496089361153111,
      "learning_rate": 4.673363259631425e-07,
      "loss": 0.2672,
      "step": 2406
    },
    {
      "epoch": 0.6233530801851672,
      "grad_norm": 2.9297055614988183,
      "learning_rate": 4.6678947225317704e-07,
      "loss": 0.2731,
      "step": 2407
    },
    {
      "epoch": 0.6236120552911851,
      "grad_norm": 4.590123245959021,
      "learning_rate": 4.662427479256902e-07,
      "loss": 0.2708,
      "step": 2408
    },
    {
      "epoch": 0.623871030397203,
      "grad_norm": 3.748029917180353,
      "learning_rate": 4.6569615342778596e-07,
      "loss": 0.2897,
      "step": 2409
    },
    {
      "epoch": 0.624130005503221,
      "grad_norm": 3.9622205372767154,
      "learning_rate": 4.6514968920646205e-07,
      "loss": 0.3431,
      "step": 2410
    },
    {
      "epoch": 0.624130005503221,
      "eval_PRM Accuracy": 0.8773584905660378,
      "eval_PRM F1": 0.9202453987730062,
      "eval_PRM F1 AUC": 0.8431115767417497,
      "eval_PRM F1 AUC (fixed)": 0.8271346254583551,
      "eval_PRM F1 Neg": 0.7346938775510204,
      "eval_PRM NPV": 0.6923076923076923,
      "eval_PRM Precision": 0.9375,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.2327074110507965,
      "eval_runtime": 14.5209,
      "eval_samples_per_second": 3.03,
      "eval_steps_per_second": 0.138,
      "step": 2410
    },
    {
      "epoch": 0.624388980609239,
      "grad_norm": 5.563336448930735,
      "learning_rate": 4.646033557086098e-07,
      "loss": 0.3785,
      "step": 2411
    },
    {
      "epoch": 0.6246479557152569,
      "grad_norm": 2.6699065810035303,
      "learning_rate": 4.640571533810132e-07,
      "loss": 0.3148,
      "step": 2412
    },
    {
      "epoch": 0.6249069308212748,
      "grad_norm": 5.287050143539514,
      "learning_rate": 4.6351108267034956e-07,
      "loss": 0.3017,
      "step": 2413
    },
    {
      "epoch": 0.6251659059272927,
      "grad_norm": 2.6881857931726105,
      "learning_rate": 4.629651440231881e-07,
      "loss": 0.2911,
      "step": 2414
    },
    {
      "epoch": 0.6254248810333106,
      "grad_norm": 4.060825155854912,
      "learning_rate": 4.624193378859904e-07,
      "loss": 0.3856,
      "step": 2415
    },
    {
      "epoch": 0.6254248810333106,
      "eval_PRM Accuracy": 0.8679245283018868,
      "eval_PRM F1": 0.9146341463414634,
      "eval_PRM F1 AUC": 0.8213724463069669,
      "eval_PRM F1 AUC (fixed)": 0.8294918805657412,
      "eval_PRM F1 Neg": 0.7083333333333334,
      "eval_PRM NPV": 0.68,
      "eval_PRM Precision": 0.9259259259259259,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.23443764448165894,
      "eval_runtime": 14.1548,
      "eval_samples_per_second": 3.108,
      "eval_steps_per_second": 0.141,
      "step": 2415
    },
    {
      "epoch": 0.6256838561393286,
      "grad_norm": 2.634660859334268,
      "learning_rate": 4.6187366470510945e-07,
      "loss": 0.3395,
      "step": 2416
    },
    {
      "epoch": 0.6259428312453466,
      "grad_norm": 6.788400099682876,
      "learning_rate": 4.613281249267894e-07,
      "loss": 0.391,
      "step": 2417
    },
    {
      "epoch": 0.6262018063513645,
      "grad_norm": 8.467508934740962,
      "learning_rate": 4.6078271899716587e-07,
      "loss": 0.3447,
      "step": 2418
    },
    {
      "epoch": 0.6264607814573824,
      "grad_norm": 7.637652802279274,
      "learning_rate": 4.6023744736226424e-07,
      "loss": 0.3155,
      "step": 2419
    },
    {
      "epoch": 0.6267197565634004,
      "grad_norm": 3.6951904686780797,
      "learning_rate": 4.596923104680009e-07,
      "loss": 0.2853,
      "step": 2420
    },
    {
      "epoch": 0.6267197565634004,
      "eval_PRM Accuracy": 0.8773584905660378,
      "eval_PRM F1": 0.9202453987730062,
      "eval_PRM F1 AUC": 0.8431115767417497,
      "eval_PRM F1 AUC (fixed)": 0.8300157150340493,
      "eval_PRM F1 Neg": 0.7346938775510204,
      "eval_PRM NPV": 0.6923076923076923,
      "eval_PRM Precision": 0.9375,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.23425859212875366,
      "eval_runtime": 14.4571,
      "eval_samples_per_second": 3.043,
      "eval_steps_per_second": 0.138,
      "step": 2420
    },
    {
      "epoch": 0.6269787316694183,
      "grad_norm": 2.775299357914315,
      "learning_rate": 4.5914730876018095e-07,
      "loss": 0.3204,
      "step": 2421
    },
    {
      "epoch": 0.6272377067754362,
      "grad_norm": 4.843004094718668,
      "learning_rate": 4.5860244268450004e-07,
      "loss": 0.3935,
      "step": 2422
    },
    {
      "epoch": 0.6274966818814541,
      "grad_norm": 5.2338119096537365,
      "learning_rate": 4.580577126865423e-07,
      "loss": 0.3528,
      "step": 2423
    },
    {
      "epoch": 0.627755656987472,
      "grad_norm": 4.878112257331263,
      "learning_rate": 4.575131192117807e-07,
      "loss": 0.2757,
      "step": 2424
    },
    {
      "epoch": 0.6280146320934901,
      "grad_norm": 4.478184245729592,
      "learning_rate": 4.569686627055768e-07,
      "loss": 0.4101,
      "step": 2425
    },
    {
      "epoch": 0.6280146320934901,
      "eval_PRM Accuracy": 0.8679245283018868,
      "eval_PRM F1": 0.9135802469135802,
      "eval_PRM F1 AUC": 0.8370874803562075,
      "eval_PRM F1 AUC (fixed)": 0.8305395495023573,
      "eval_PRM F1 Neg": 0.72,
      "eval_PRM NPV": 0.6666666666666666,
      "eval_PRM Precision": 0.9367088607594937,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.23475131392478943,
      "eval_runtime": 14.5571,
      "eval_samples_per_second": 3.023,
      "eval_steps_per_second": 0.137,
      "step": 2425
    },
    {
      "epoch": 0.628273607199508,
      "grad_norm": 4.583318204331169,
      "learning_rate": 4.5642434361317963e-07,
      "loss": 0.3089,
      "step": 2426
    },
    {
      "epoch": 0.6285325823055259,
      "grad_norm": 3.4528835210487028,
      "learning_rate": 4.558801623797262e-07,
      "loss": 0.3092,
      "step": 2427
    },
    {
      "epoch": 0.6287915574115438,
      "grad_norm": 7.015543960608334,
      "learning_rate": 4.553361194502412e-07,
      "loss": 0.2807,
      "step": 2428
    },
    {
      "epoch": 0.6290505325175617,
      "grad_norm": 4.288363987578536,
      "learning_rate": 4.547922152696351e-07,
      "loss": 0.3854,
      "step": 2429
    },
    {
      "epoch": 0.6293095076235797,
      "grad_norm": 3.2831872361182692,
      "learning_rate": 4.542484502827059e-07,
      "loss": 0.3047,
      "step": 2430
    },
    {
      "epoch": 0.6293095076235797,
      "eval_PRM Accuracy": 0.8679245283018868,
      "eval_PRM F1": 0.9135802469135802,
      "eval_PRM F1 AUC": 0.8370874803562075,
      "eval_PRM F1 AUC (fixed)": 0.8268727082242011,
      "eval_PRM F1 Neg": 0.72,
      "eval_PRM NPV": 0.6666666666666666,
      "eval_PRM Precision": 0.9367088607594937,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.2335863560438156,
      "eval_runtime": 14.8457,
      "eval_samples_per_second": 2.964,
      "eval_steps_per_second": 0.135,
      "step": 2430
    },
    {
      "epoch": 0.6295684827295976,
      "grad_norm": 3.880366163018059,
      "learning_rate": 4.537048249341371e-07,
      "loss": 0.3522,
      "step": 2431
    },
    {
      "epoch": 0.6298274578356156,
      "grad_norm": 4.6346859806782215,
      "learning_rate": 4.531613396684988e-07,
      "loss": 0.3012,
      "step": 2432
    },
    {
      "epoch": 0.6300864329416335,
      "grad_norm": 3.1843215894549233,
      "learning_rate": 4.5261799493024574e-07,
      "loss": 0.2927,
      "step": 2433
    },
    {
      "epoch": 0.6303454080476514,
      "grad_norm": 3.5166843273639454,
      "learning_rate": 4.520747911637182e-07,
      "loss": 0.2737,
      "step": 2434
    },
    {
      "epoch": 0.6306043831536694,
      "grad_norm": 3.8549037852757313,
      "learning_rate": 4.51531728813141e-07,
      "loss": 0.3728,
      "step": 2435
    },
    {
      "epoch": 0.6306043831536694,
      "eval_PRM Accuracy": 0.8679245283018868,
      "eval_PRM F1": 0.9135802469135802,
      "eval_PRM F1 AUC": 0.8370874803562075,
      "eval_PRM F1 AUC (fixed)": 0.8281822943949712,
      "eval_PRM F1 Neg": 0.72,
      "eval_PRM NPV": 0.6666666666666666,
      "eval_PRM Precision": 0.9367088607594937,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.23153255879878998,
      "eval_runtime": 14.3152,
      "eval_samples_per_second": 3.074,
      "eval_steps_per_second": 0.14,
      "step": 2435
    },
    {
      "epoch": 0.6308633582596873,
      "grad_norm": 5.128048706368578,
      "learning_rate": 4.509888083226235e-07,
      "loss": 0.3285,
      "step": 2436
    },
    {
      "epoch": 0.6311223333657052,
      "grad_norm": 4.015458572127405,
      "learning_rate": 4.504460301361586e-07,
      "loss": 0.2406,
      "step": 2437
    },
    {
      "epoch": 0.6313813084717231,
      "grad_norm": 5.8038585756376255,
      "learning_rate": 4.4990339469762327e-07,
      "loss": 0.2525,
      "step": 2438
    },
    {
      "epoch": 0.631640283577741,
      "grad_norm": 5.073306570527701,
      "learning_rate": 4.4936090245077763e-07,
      "loss": 0.3723,
      "step": 2439
    },
    {
      "epoch": 0.6318992586837591,
      "grad_norm": 6.242601977880184,
      "learning_rate": 4.488185538392646e-07,
      "loss": 0.3549,
      "step": 2440
    },
    {
      "epoch": 0.6318992586837591,
      "eval_PRM Accuracy": 0.8773584905660378,
      "eval_PRM F1": 0.9202453987730062,
      "eval_PRM F1 AUC": 0.8431115767417497,
      "eval_PRM F1 AUC (fixed)": 0.8323729701414353,
      "eval_PRM F1 Neg": 0.7346938775510204,
      "eval_PRM NPV": 0.6923076923076923,
      "eval_PRM Precision": 0.9375,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.2316233068704605,
      "eval_runtime": 14.6473,
      "eval_samples_per_second": 3.004,
      "eval_steps_per_second": 0.137,
      "step": 2440
    },
    {
      "epoch": 0.632158233789777,
      "grad_norm": 3.232490018215951,
      "learning_rate": 4.4827634930660976e-07,
      "loss": 0.2646,
      "step": 2441
    },
    {
      "epoch": 0.6324172088957949,
      "grad_norm": 4.400947841804586,
      "learning_rate": 4.477342892962208e-07,
      "loss": 0.3567,
      "step": 2442
    },
    {
      "epoch": 0.6326761840018128,
      "grad_norm": 6.4022477658070045,
      "learning_rate": 4.4719237425138726e-07,
      "loss": 0.3297,
      "step": 2443
    },
    {
      "epoch": 0.6329351591078307,
      "grad_norm": 8.223653345996658,
      "learning_rate": 4.4665060461528046e-07,
      "loss": 0.4033,
      "step": 2444
    },
    {
      "epoch": 0.6331941342138487,
      "grad_norm": 3.357824631248715,
      "learning_rate": 4.4610898083095157e-07,
      "loss": 0.3338,
      "step": 2445
    },
    {
      "epoch": 0.6331941342138487,
      "eval_PRM Accuracy": 0.8773584905660378,
      "eval_PRM F1": 0.9202453987730062,
      "eval_PRM F1 AUC": 0.8431115767417497,
      "eval_PRM F1 AUC (fixed)": 0.8279203771608172,
      "eval_PRM F1 Neg": 0.7346938775510204,
      "eval_PRM NPV": 0.6923076923076923,
      "eval_PRM Precision": 0.9375,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.2321992814540863,
      "eval_runtime": 14.6204,
      "eval_samples_per_second": 3.009,
      "eval_steps_per_second": 0.137,
      "step": 2445
    },
    {
      "epoch": 0.6334531093198666,
      "grad_norm": 5.105391962201131,
      "learning_rate": 4.4556750334133404e-07,
      "loss": 0.3088,
      "step": 2446
    },
    {
      "epoch": 0.6337120844258846,
      "grad_norm": 4.576131988134334,
      "learning_rate": 4.450261725892408e-07,
      "loss": 0.2991,
      "step": 2447
    },
    {
      "epoch": 0.6339710595319025,
      "grad_norm": 3.4755489981775236,
      "learning_rate": 4.44484989017365e-07,
      "loss": 0.2737,
      "step": 2448
    },
    {
      "epoch": 0.6342300346379204,
      "grad_norm": 5.3637417486779455,
      "learning_rate": 4.4394395306827947e-07,
      "loss": 0.3067,
      "step": 2449
    },
    {
      "epoch": 0.6344890097439384,
      "grad_norm": 3.62687460694226,
      "learning_rate": 4.434030651844362e-07,
      "loss": 0.2826,
      "step": 2450
    },
    {
      "epoch": 0.6344890097439384,
      "eval_PRM Accuracy": 0.8679245283018868,
      "eval_PRM F1": 0.9135802469135802,
      "eval_PRM F1 AUC": 0.8370874803562075,
      "eval_PRM F1 AUC (fixed)": 0.8292299633315873,
      "eval_PRM F1 Neg": 0.72,
      "eval_PRM NPV": 0.6666666666666666,
      "eval_PRM Precision": 0.9367088607594937,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.23289035260677338,
      "eval_runtime": 14.7872,
      "eval_samples_per_second": 2.976,
      "eval_steps_per_second": 0.135,
      "step": 2450
    },
    {
      "epoch": 0.6347479848499563,
      "grad_norm": 3.604999126059274,
      "learning_rate": 4.4286232580816615e-07,
      "loss": 0.267,
      "step": 2451
    },
    {
      "epoch": 0.6350069599559742,
      "grad_norm": 3.9043915416950723,
      "learning_rate": 4.4232173538167876e-07,
      "loss": 0.3772,
      "step": 2452
    },
    {
      "epoch": 0.6352659350619921,
      "grad_norm": 4.673879586139769,
      "learning_rate": 4.417812943470622e-07,
      "loss": 0.3712,
      "step": 2453
    },
    {
      "epoch": 0.63552491016801,
      "grad_norm": 5.499986485950306,
      "learning_rate": 4.412410031462813e-07,
      "loss": 0.3261,
      "step": 2454
    },
    {
      "epoch": 0.6357838852740281,
      "grad_norm": 6.520420436189523,
      "learning_rate": 4.4070086222117933e-07,
      "loss": 0.3451,
      "step": 2455
    },
    {
      "epoch": 0.6357838852740281,
      "eval_PRM Accuracy": 0.8679245283018868,
      "eval_PRM F1": 0.9135802469135802,
      "eval_PRM F1 AUC": 0.8370874803562075,
      "eval_PRM F1 AUC (fixed)": 0.8279203771608172,
      "eval_PRM F1 Neg": 0.72,
      "eval_PRM NPV": 0.6666666666666666,
      "eval_PRM Precision": 0.9367088607594937,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.23285795748233795,
      "eval_runtime": 14.2523,
      "eval_samples_per_second": 3.087,
      "eval_steps_per_second": 0.14,
      "step": 2455
    },
    {
      "epoch": 0.636042860380046,
      "grad_norm": 3.1906064479453558,
      "learning_rate": 4.401608720134766e-07,
      "loss": 0.3565,
      "step": 2456
    },
    {
      "epoch": 0.6363018354860639,
      "grad_norm": 3.581629816876452,
      "learning_rate": 4.396210329647699e-07,
      "loss": 0.3832,
      "step": 2457
    },
    {
      "epoch": 0.6365608105920818,
      "grad_norm": 4.85407007370176,
      "learning_rate": 4.390813455165323e-07,
      "loss": 0.376,
      "step": 2458
    },
    {
      "epoch": 0.6368197856980997,
      "grad_norm": 3.6958757637898922,
      "learning_rate": 4.3854181011011313e-07,
      "loss": 0.3638,
      "step": 2459
    },
    {
      "epoch": 0.6370787608041177,
      "grad_norm": 4.159796671928392,
      "learning_rate": 4.380024271867374e-07,
      "loss": 0.3222,
      "step": 2460
    },
    {
      "epoch": 0.6370787608041177,
      "eval_PRM Accuracy": 0.8679245283018868,
      "eval_PRM F1": 0.9135802469135802,
      "eval_PRM F1 AUC": 0.8370874803562075,
      "eval_PRM F1 AUC (fixed)": 0.8281822943949712,
      "eval_PRM F1 Neg": 0.72,
      "eval_PRM NPV": 0.6666666666666666,
      "eval_PRM Precision": 0.9367088607594937,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.23250402510166168,
      "eval_runtime": 15.1199,
      "eval_samples_per_second": 2.91,
      "eval_steps_per_second": 0.132,
      "step": 2460
    },
    {
      "epoch": 0.6373377359101356,
      "grad_norm": 3.5362529188026586,
      "learning_rate": 4.374631971875054e-07,
      "loss": 0.3568,
      "step": 2461
    },
    {
      "epoch": 0.6375967110161536,
      "grad_norm": 4.466976275658977,
      "learning_rate": 4.369241205533921e-07,
      "loss": 0.2975,
      "step": 2462
    },
    {
      "epoch": 0.6378556861221715,
      "grad_norm": 3.3307374289086056,
      "learning_rate": 4.363851977252473e-07,
      "loss": 0.3285,
      "step": 2463
    },
    {
      "epoch": 0.6381146612281895,
      "grad_norm": 4.431362438064065,
      "learning_rate": 4.35846429143795e-07,
      "loss": 0.3114,
      "step": 2464
    },
    {
      "epoch": 0.6383736363342074,
      "grad_norm": 3.317543733298947,
      "learning_rate": 4.35307815249633e-07,
      "loss": 0.2941,
      "step": 2465
    },
    {
      "epoch": 0.6383736363342074,
      "eval_PRM Accuracy": 0.8679245283018868,
      "eval_PRM F1": 0.9146341463414634,
      "eval_PRM F1 AUC": 0.8213724463069669,
      "eval_PRM F1 AUC (fixed)": 0.8255631220534312,
      "eval_PRM F1 Neg": 0.7083333333333334,
      "eval_PRM NPV": 0.68,
      "eval_PRM Precision": 0.9259259259259259,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.23352384567260742,
      "eval_runtime": 14.4541,
      "eval_samples_per_second": 3.044,
      "eval_steps_per_second": 0.138,
      "step": 2465
    },
    {
      "epoch": 0.6386326114402253,
      "grad_norm": 6.117095894402694,
      "learning_rate": 4.347693564832325e-07,
      "loss": 0.3762,
      "step": 2466
    },
    {
      "epoch": 0.6388915865462432,
      "grad_norm": 6.454780954927651,
      "learning_rate": 4.342310532849384e-07,
      "loss": 0.2867,
      "step": 2467
    },
    {
      "epoch": 0.6391505616522611,
      "grad_norm": 3.975306437439717,
      "learning_rate": 4.336929060949674e-07,
      "loss": 0.2817,
      "step": 2468
    },
    {
      "epoch": 0.6394095367582792,
      "grad_norm": 7.367407886903491,
      "learning_rate": 4.3315491535340965e-07,
      "loss": 0.3969,
      "step": 2469
    },
    {
      "epoch": 0.6396685118642971,
      "grad_norm": 5.391577101908448,
      "learning_rate": 4.3261708150022644e-07,
      "loss": 0.3246,
      "step": 2470
    },
    {
      "epoch": 0.6396685118642971,
      "eval_PRM Accuracy": 0.8773584905660378,
      "eval_PRM F1": 0.9212121212121213,
      "eval_PRM F1 AUC": 0.8273965426925091,
      "eval_PRM F1 AUC (fixed)": 0.8229439497118911,
      "eval_PRM F1 Neg": 0.723404255319149,
      "eval_PRM NPV": 0.7083333333333334,
      "eval_PRM Precision": 0.926829268292683,
      "eval_PRM Recall": 0.9156626506024096,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.23500221967697144,
      "eval_runtime": 15.0653,
      "eval_samples_per_second": 2.921,
      "eval_steps_per_second": 0.133,
      "step": 2470
    },
    {
      "epoch": 0.639927486970315,
      "grad_norm": 3.459548396785892,
      "learning_rate": 4.320794049752515e-07,
      "loss": 0.3527,
      "step": 2471
    },
    {
      "epoch": 0.6401864620763329,
      "grad_norm": 3.8846105549364305,
      "learning_rate": 4.3154188621818936e-07,
      "loss": 0.2756,
      "step": 2472
    },
    {
      "epoch": 0.6404454371823508,
      "grad_norm": 3.640194635054852,
      "learning_rate": 4.3100452566861587e-07,
      "loss": 0.3461,
      "step": 2473
    },
    {
      "epoch": 0.6407044122883688,
      "grad_norm": 4.010649725908795,
      "learning_rate": 4.304673237659774e-07,
      "loss": 0.4071,
      "step": 2474
    },
    {
      "epoch": 0.6409633873943867,
      "grad_norm": 3.7521968423056857,
      "learning_rate": 4.299302809495905e-07,
      "loss": 0.2754,
      "step": 2475
    },
    {
      "epoch": 0.6409633873943867,
      "eval_PRM Accuracy": 0.8773584905660378,
      "eval_PRM F1": 0.9212121212121213,
      "eval_PRM F1 AUC": 0.8273965426925091,
      "eval_PRM F1 AUC (fixed)": 0.824253535882661,
      "eval_PRM F1 Neg": 0.723404255319149,
      "eval_PRM NPV": 0.7083333333333334,
      "eval_PRM Precision": 0.926829268292683,
      "eval_PRM Recall": 0.9156626506024096,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.23455935716629028,
      "eval_runtime": 14.7024,
      "eval_samples_per_second": 2.993,
      "eval_steps_per_second": 0.136,
      "step": 2475
    },
    {
      "epoch": 0.6412223625004047,
      "grad_norm": 7.527218016314557,
      "learning_rate": 4.293933976586416e-07,
      "loss": 0.3539,
      "step": 2476
    },
    {
      "epoch": 0.6414813376064226,
      "grad_norm": 7.3845277102770766,
      "learning_rate": 4.2885667433218705e-07,
      "loss": 0.3596,
      "step": 2477
    },
    {
      "epoch": 0.6417403127124405,
      "grad_norm": 5.471927637321291,
      "learning_rate": 4.283201114091521e-07,
      "loss": 0.2539,
      "step": 2478
    },
    {
      "epoch": 0.6419992878184585,
      "grad_norm": 2.965899751644036,
      "learning_rate": 4.2778370932833045e-07,
      "loss": 0.2784,
      "step": 2479
    },
    {
      "epoch": 0.6422582629244764,
      "grad_norm": 3.4546803855128165,
      "learning_rate": 4.2724746852838465e-07,
      "loss": 0.2586,
      "step": 2480
    },
    {
      "epoch": 0.6422582629244764,
      "eval_PRM Accuracy": 0.8773584905660378,
      "eval_PRM F1": 0.9202453987730062,
      "eval_PRM F1 AUC": 0.8431115767417497,
      "eval_PRM F1 AUC (fixed)": 0.822158198009429,
      "eval_PRM F1 Neg": 0.7346938775510204,
      "eval_PRM NPV": 0.6923076923076923,
      "eval_PRM Precision": 0.9375,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.23433376848697662,
      "eval_runtime": 14.9117,
      "eval_samples_per_second": 2.951,
      "eval_steps_per_second": 0.134,
      "step": 2480
    },
    {
      "epoch": 0.6425172380304943,
      "grad_norm": 4.247605746485939,
      "learning_rate": 4.267113894478457e-07,
      "loss": 0.2942,
      "step": 2481
    },
    {
      "epoch": 0.6427762131365122,
      "grad_norm": 3.8933199356862573,
      "learning_rate": 4.2617547252511173e-07,
      "loss": 0.3599,
      "step": 2482
    },
    {
      "epoch": 0.6430351882425301,
      "grad_norm": 4.955818398593934,
      "learning_rate": 4.256397181984486e-07,
      "loss": 0.3218,
      "step": 2483
    },
    {
      "epoch": 0.6432941633485482,
      "grad_norm": 4.689006967824344,
      "learning_rate": 4.2510412690598914e-07,
      "loss": 0.3498,
      "step": 2484
    },
    {
      "epoch": 0.6435531384545661,
      "grad_norm": 2.9804404468590375,
      "learning_rate": 4.2456869908573275e-07,
      "loss": 0.3283,
      "step": 2485
    },
    {
      "epoch": 0.6435531384545661,
      "eval_PRM Accuracy": 0.8773584905660378,
      "eval_PRM F1": 0.9202453987730062,
      "eval_PRM F1 AUC": 0.8431115767417497,
      "eval_PRM F1 AUC (fixed)": 0.8208486118386591,
      "eval_PRM F1 Neg": 0.7346938775510204,
      "eval_PRM NPV": 0.6923076923076923,
      "eval_PRM Precision": 0.9375,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.23384861648082733,
      "eval_runtime": 14.4432,
      "eval_samples_per_second": 3.046,
      "eval_steps_per_second": 0.138,
      "step": 2485
    },
    {
      "epoch": 0.643812113560584,
      "grad_norm": 3.7684121059834825,
      "learning_rate": 4.240334351755455e-07,
      "loss": 0.349,
      "step": 2486
    },
    {
      "epoch": 0.6440710886666019,
      "grad_norm": 3.839886053453229,
      "learning_rate": 4.234983356131587e-07,
      "loss": 0.3126,
      "step": 2487
    },
    {
      "epoch": 0.6443300637726198,
      "grad_norm": 3.537603305079429,
      "learning_rate": 4.2296340083616993e-07,
      "loss": 0.3246,
      "step": 2488
    },
    {
      "epoch": 0.6445890388786378,
      "grad_norm": 5.7928118960272785,
      "learning_rate": 4.224286312820417e-07,
      "loss": 0.3777,
      "step": 2489
    },
    {
      "epoch": 0.6448480139846557,
      "grad_norm": 3.466715424360159,
      "learning_rate": 4.2189402738810147e-07,
      "loss": 0.2673,
      "step": 2490
    },
    {
      "epoch": 0.6448480139846557,
      "eval_PRM Accuracy": 0.8679245283018868,
      "eval_PRM F1": 0.9135802469135802,
      "eval_PRM F1 AUC": 0.8370874803562075,
      "eval_PRM F1 AUC (fixed)": 0.820848611838659,
      "eval_PRM F1 Neg": 0.72,
      "eval_PRM NPV": 0.6666666666666666,
      "eval_PRM Precision": 0.9367088607594937,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.2335864156484604,
      "eval_runtime": 14.6996,
      "eval_samples_per_second": 2.993,
      "eval_steps_per_second": 0.136,
      "step": 2490
    },
    {
      "epoch": 0.6451069890906737,
      "grad_norm": 4.918040091188315,
      "learning_rate": 4.213595895915415e-07,
      "loss": 0.3056,
      "step": 2491
    },
    {
      "epoch": 0.6453659641966916,
      "grad_norm": 3.431189772943314,
      "learning_rate": 4.208253183294177e-07,
      "loss": 0.3127,
      "step": 2492
    },
    {
      "epoch": 0.6456249393027095,
      "grad_norm": 4.246523455247824,
      "learning_rate": 4.202912140386501e-07,
      "loss": 0.2833,
      "step": 2493
    },
    {
      "epoch": 0.6458839144087275,
      "grad_norm": 6.161699275142311,
      "learning_rate": 4.197572771560224e-07,
      "loss": 0.3366,
      "step": 2494
    },
    {
      "epoch": 0.6461428895147454,
      "grad_norm": 3.487125675764393,
      "learning_rate": 4.1922350811818056e-07,
      "loss": 0.3163,
      "step": 2495
    },
    {
      "epoch": 0.6461428895147454,
      "eval_PRM Accuracy": 0.8679245283018868,
      "eval_PRM F1": 0.9135802469135802,
      "eval_PRM F1 AUC": 0.8370874803562075,
      "eval_PRM F1 AUC (fixed)": 0.820062860136197,
      "eval_PRM F1 Neg": 0.72,
      "eval_PRM NPV": 0.6666666666666666,
      "eval_PRM Precision": 0.9367088607594937,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.23248301446437836,
      "eval_runtime": 14.2682,
      "eval_samples_per_second": 3.084,
      "eval_steps_per_second": 0.14,
      "step": 2495
    },
    {
      "epoch": 0.6464018646207633,
      "grad_norm": 2.5657849656151317,
      "learning_rate": 4.186899073616343e-07,
      "loss": 0.2549,
      "step": 2496
    },
    {
      "epoch": 0.6466608397267812,
      "grad_norm": 4.611991733340532,
      "learning_rate": 4.1815647532275503e-07,
      "loss": 0.4676,
      "step": 2497
    },
    {
      "epoch": 0.6469198148327991,
      "grad_norm": 4.89689593241627,
      "learning_rate": 4.176232124377766e-07,
      "loss": 0.3301,
      "step": 2498
    },
    {
      "epoch": 0.6471787899388172,
      "grad_norm": 4.477149834466212,
      "learning_rate": 4.1709011914279424e-07,
      "loss": 0.3941,
      "step": 2499
    },
    {
      "epoch": 0.6474377650448351,
      "grad_norm": 3.8271223544137425,
      "learning_rate": 4.165571958737645e-07,
      "loss": 0.302,
      "step": 2500
    },
    {
      "epoch": 0.6474377650448351,
      "eval_PRM Accuracy": 0.8679245283018868,
      "eval_PRM F1": 0.9135802469135802,
      "eval_PRM F1 AUC": 0.8370874803562075,
      "eval_PRM F1 AUC (fixed)": 0.821634363541121,
      "eval_PRM F1 Neg": 0.72,
      "eval_PRM NPV": 0.6666666666666666,
      "eval_PRM Precision": 0.9367088607594937,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.23355937004089355,
      "eval_runtime": 14.1529,
      "eval_samples_per_second": 3.109,
      "eval_steps_per_second": 0.141,
      "step": 2500
    },
    {
      "epoch": 0.647696740150853,
      "grad_norm": 6.726115964964346,
      "learning_rate": 4.160244430665052e-07,
      "loss": 0.3363,
      "step": 2501
    },
    {
      "epoch": 0.6479557152568709,
      "grad_norm": 4.521542775082122,
      "learning_rate": 4.1549186115669453e-07,
      "loss": 0.2607,
      "step": 2502
    },
    {
      "epoch": 0.6482146903628888,
      "grad_norm": 3.4921695888314916,
      "learning_rate": 4.149594505798711e-07,
      "loss": 0.2856,
      "step": 2503
    },
    {
      "epoch": 0.6484736654689068,
      "grad_norm": 6.90535942614918,
      "learning_rate": 4.144272117714326e-07,
      "loss": 0.303,
      "step": 2504
    },
    {
      "epoch": 0.6487326405749247,
      "grad_norm": 4.004982290311785,
      "learning_rate": 4.1389514516663746e-07,
      "loss": 0.3411,
      "step": 2505
    },
    {
      "epoch": 0.6487326405749247,
      "eval_PRM Accuracy": 0.8773584905660378,
      "eval_PRM F1": 0.9202453987730062,
      "eval_PRM F1 AUC": 0.8431115767417497,
      "eval_PRM F1 AUC (fixed)": 0.819800942902043,
      "eval_PRM F1 Neg": 0.7346938775510204,
      "eval_PRM NPV": 0.6923076923076923,
      "eval_PRM Precision": 0.9375,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.23301172256469727,
      "eval_runtime": 15.1569,
      "eval_samples_per_second": 2.903,
      "eval_steps_per_second": 0.132,
      "step": 2505
    },
    {
      "epoch": 0.6489916156809427,
      "grad_norm": 5.79550654655193,
      "learning_rate": 4.133632512006025e-07,
      "loss": 0.329,
      "step": 2506
    },
    {
      "epoch": 0.6492505907869606,
      "grad_norm": 5.2400749740031864,
      "learning_rate": 4.1283153030830365e-07,
      "loss": 0.4561,
      "step": 2507
    },
    {
      "epoch": 0.6495095658929786,
      "grad_norm": 3.8206946087326874,
      "learning_rate": 4.1229998292457526e-07,
      "loss": 0.3039,
      "step": 2508
    },
    {
      "epoch": 0.6497685409989965,
      "grad_norm": 2.8454074277297856,
      "learning_rate": 4.117686094841097e-07,
      "loss": 0.273,
      "step": 2509
    },
    {
      "epoch": 0.6500275161050144,
      "grad_norm": 3.302345353363104,
      "learning_rate": 4.112374104214571e-07,
      "loss": 0.2785,
      "step": 2510
    },
    {
      "epoch": 0.6500275161050144,
      "eval_PRM Accuracy": 0.8773584905660378,
      "eval_PRM F1": 0.9202453987730062,
      "eval_PRM F1 AUC": 0.8431115767417497,
      "eval_PRM F1 AUC (fixed)": 0.8171817705605029,
      "eval_PRM F1 Neg": 0.7346938775510204,
      "eval_PRM NPV": 0.6923076923076923,
      "eval_PRM Precision": 0.9375,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.23186974227428436,
      "eval_runtime": 14.2986,
      "eval_samples_per_second": 3.077,
      "eval_steps_per_second": 0.14,
      "step": 2510
    },
    {
      "epoch": 0.6502864912110323,
      "grad_norm": 7.842811210979961,
      "learning_rate": 4.1070638617102524e-07,
      "loss": 0.3926,
      "step": 2511
    },
    {
      "epoch": 0.6505454663170502,
      "grad_norm": 5.617047731830771,
      "learning_rate": 4.1017553716707846e-07,
      "loss": 0.3413,
      "step": 2512
    },
    {
      "epoch": 0.6508044414230683,
      "grad_norm": 5.852719964898465,
      "learning_rate": 4.096448638437382e-07,
      "loss": 0.3172,
      "step": 2513
    },
    {
      "epoch": 0.6510634165290862,
      "grad_norm": 2.492904061983185,
      "learning_rate": 4.0911436663498195e-07,
      "loss": 0.248,
      "step": 2514
    },
    {
      "epoch": 0.6513223916351041,
      "grad_norm": 3.270122469569195,
      "learning_rate": 4.085840459746435e-07,
      "loss": 0.3423,
      "step": 2515
    },
    {
      "epoch": 0.6513223916351041,
      "eval_PRM Accuracy": 0.8773584905660378,
      "eval_PRM F1": 0.9202453987730062,
      "eval_PRM F1 AUC": 0.8431115767417497,
      "eval_PRM F1 AUC (fixed)": 0.8174436877946569,
      "eval_PRM F1 Neg": 0.7346938775510204,
      "eval_PRM NPV": 0.6923076923076923,
      "eval_PRM Precision": 0.9375,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.22989903390407562,
      "eval_runtime": 14.6968,
      "eval_samples_per_second": 2.994,
      "eval_steps_per_second": 0.136,
      "step": 2515
    },
    {
      "epoch": 0.651581366741122,
      "grad_norm": 3.756666464236782,
      "learning_rate": 4.080539022964121e-07,
      "loss": 0.2967,
      "step": 2516
    },
    {
      "epoch": 0.6518403418471399,
      "grad_norm": 4.24922259317945,
      "learning_rate": 4.0752393603383205e-07,
      "loss": 0.3555,
      "step": 2517
    },
    {
      "epoch": 0.6520993169531579,
      "grad_norm": 2.88536564687218,
      "learning_rate": 4.06994147620303e-07,
      "loss": 0.2554,
      "step": 2518
    },
    {
      "epoch": 0.6523582920591758,
      "grad_norm": 3.062227840710602,
      "learning_rate": 4.0646453748907883e-07,
      "loss": 0.2961,
      "step": 2519
    },
    {
      "epoch": 0.6526172671651937,
      "grad_norm": 2.9687703820304683,
      "learning_rate": 4.059351060732675e-07,
      "loss": 0.3008,
      "step": 2520
    },
    {
      "epoch": 0.6526172671651937,
      "eval_PRM Accuracy": 0.8773584905660378,
      "eval_PRM F1": 0.9202453987730062,
      "eval_PRM F1 AUC": 0.8431115767417497,
      "eval_PRM F1 AUC (fixed)": 0.8161341016238869,
      "eval_PRM F1 Neg": 0.7346938775510204,
      "eval_PRM NPV": 0.6923076923076923,
      "eval_PRM Precision": 0.9375,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.22820109128952026,
      "eval_runtime": 14.7684,
      "eval_samples_per_second": 2.979,
      "eval_steps_per_second": 0.135,
      "step": 2520
    },
    {
      "epoch": 0.6528762422712117,
      "grad_norm": 2.9801936797406485,
      "learning_rate": 4.054058538058312e-07,
      "loss": 0.3215,
      "step": 2521
    },
    {
      "epoch": 0.6531352173772296,
      "grad_norm": 2.7429220346848204,
      "learning_rate": 4.048767811195854e-07,
      "loss": 0.2906,
      "step": 2522
    },
    {
      "epoch": 0.6533941924832476,
      "grad_norm": 3.4441876020892668,
      "learning_rate": 4.043478884471988e-07,
      "loss": 0.3237,
      "step": 2523
    },
    {
      "epoch": 0.6536531675892655,
      "grad_norm": 3.2841928323136016,
      "learning_rate": 4.038191762211928e-07,
      "loss": 0.3075,
      "step": 2524
    },
    {
      "epoch": 0.6539121426952834,
      "grad_norm": 3.5907061626519448,
      "learning_rate": 4.0329064487394113e-07,
      "loss": 0.2654,
      "step": 2525
    },
    {
      "epoch": 0.6539121426952834,
      "eval_PRM Accuracy": 0.8867924528301887,
      "eval_PRM F1": 0.926829268292683,
      "eval_PRM F1 AUC": 0.8491356731272918,
      "eval_PRM F1 AUC (fixed)": 0.8177056050288108,
      "eval_PRM F1 Neg": 0.75,
      "eval_PRM NPV": 0.72,
      "eval_PRM Precision": 0.9382716049382716,
      "eval_PRM Recall": 0.9156626506024096,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.22511066496372223,
      "eval_runtime": 14.8252,
      "eval_samples_per_second": 2.968,
      "eval_steps_per_second": 0.135,
      "step": 2525
    },
    {
      "epoch": 0.6541711178013013,
      "grad_norm": 2.6408963162463226,
      "learning_rate": 4.0276229483767005e-07,
      "loss": 0.2689,
      "step": 2526
    },
    {
      "epoch": 0.6544300929073192,
      "grad_norm": 5.23357071325616,
      "learning_rate": 4.022341265444571e-07,
      "loss": 0.3159,
      "step": 2527
    },
    {
      "epoch": 0.6546890680133373,
      "grad_norm": 8.368016356291607,
      "learning_rate": 4.017061404262313e-07,
      "loss": 0.2946,
      "step": 2528
    },
    {
      "epoch": 0.6549480431193552,
      "grad_norm": 4.709827167125874,
      "learning_rate": 4.011783369147727e-07,
      "loss": 0.3337,
      "step": 2529
    },
    {
      "epoch": 0.6552070182253731,
      "grad_norm": 10.42587611685267,
      "learning_rate": 4.00650716441712e-07,
      "loss": 0.3715,
      "step": 2530
    },
    {
      "epoch": 0.6552070182253731,
      "eval_PRM Accuracy": 0.8773584905660378,
      "eval_PRM F1": 0.9202453987730062,
      "eval_PRM F1 AUC": 0.8431115767417497,
      "eval_PRM F1 AUC (fixed)": 0.8203247773703509,
      "eval_PRM F1 Neg": 0.7346938775510204,
      "eval_PRM NPV": 0.6923076923076923,
      "eval_PRM Precision": 0.9375,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.22228409349918365,
      "eval_runtime": 15.0253,
      "eval_samples_per_second": 2.928,
      "eval_steps_per_second": 0.133,
      "step": 2530
    },
    {
      "epoch": 0.655465993331391,
      "grad_norm": 7.254118322871237,
      "learning_rate": 4.0012327943853034e-07,
      "loss": 0.2654,
      "step": 2531
    },
    {
      "epoch": 0.6557249684374089,
      "grad_norm": 4.69134301192646,
      "learning_rate": 3.9959602633655863e-07,
      "loss": 0.3685,
      "step": 2532
    },
    {
      "epoch": 0.6559839435434269,
      "grad_norm": 3.3222884724659862,
      "learning_rate": 3.9906895756697766e-07,
      "loss": 0.2892,
      "step": 2533
    },
    {
      "epoch": 0.6562429186494448,
      "grad_norm": 5.700597892594636,
      "learning_rate": 3.985420735608171e-07,
      "loss": 0.4284,
      "step": 2534
    },
    {
      "epoch": 0.6565018937554628,
      "grad_norm": 7.418149150302375,
      "learning_rate": 3.980153747489558e-07,
      "loss": 0.2704,
      "step": 2535
    },
    {
      "epoch": 0.6565018937554628,
      "eval_PRM Accuracy": 0.8773584905660378,
      "eval_PRM F1": 0.9202453987730062,
      "eval_PRM F1 AUC": 0.8431115767417497,
      "eval_PRM F1 AUC (fixed)": 0.8179675222629649,
      "eval_PRM F1 Neg": 0.7346938775510204,
      "eval_PRM NPV": 0.6923076923076923,
      "eval_PRM Precision": 0.9375,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.2223687618970871,
      "eval_runtime": 14.3823,
      "eval_samples_per_second": 3.059,
      "eval_steps_per_second": 0.139,
      "step": 2535
    },
    {
      "epoch": 0.6567608688614807,
      "grad_norm": 3.3413285100358547,
      "learning_rate": 3.974888615621214e-07,
      "loss": 0.3196,
      "step": 2536
    },
    {
      "epoch": 0.6570198439674986,
      "grad_norm": 5.338418947833522,
      "learning_rate": 3.96962534430889e-07,
      "loss": 0.4151,
      "step": 2537
    },
    {
      "epoch": 0.6572788190735166,
      "grad_norm": 4.966628717217341,
      "learning_rate": 3.9643639378568205e-07,
      "loss": 0.353,
      "step": 2538
    },
    {
      "epoch": 0.6575377941795345,
      "grad_norm": 5.079799275674761,
      "learning_rate": 3.9591044005677134e-07,
      "loss": 0.2764,
      "step": 2539
    },
    {
      "epoch": 0.6577967692855524,
      "grad_norm": 3.5838605012170626,
      "learning_rate": 3.9538467367427517e-07,
      "loss": 0.3099,
      "step": 2540
    },
    {
      "epoch": 0.6577967692855524,
      "eval_PRM Accuracy": 0.8679245283018868,
      "eval_PRM F1": 0.9135802469135802,
      "eval_PRM F1 AUC": 0.8370874803562075,
      "eval_PRM F1 AUC (fixed)": 0.821110529072813,
      "eval_PRM F1 Neg": 0.72,
      "eval_PRM NPV": 0.6666666666666666,
      "eval_PRM Precision": 0.9367088607594937,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.22301450371742249,
      "eval_runtime": 14.7633,
      "eval_samples_per_second": 2.98,
      "eval_steps_per_second": 0.135,
      "step": 2540
    },
    {
      "epoch": 0.6580557443915703,
      "grad_norm": 5.225021900075919,
      "learning_rate": 3.948590950681581e-07,
      "loss": 0.34,
      "step": 2541
    },
    {
      "epoch": 0.6583147194975882,
      "grad_norm": 7.777081410289656,
      "learning_rate": 3.943337046682315e-07,
      "loss": 0.3864,
      "step": 2542
    },
    {
      "epoch": 0.6585736946036063,
      "grad_norm": 3.136563318983399,
      "learning_rate": 3.938085029041527e-07,
      "loss": 0.2982,
      "step": 2543
    },
    {
      "epoch": 0.6588326697096242,
      "grad_norm": 4.275989384293558,
      "learning_rate": 3.932834902054247e-07,
      "loss": 0.4322,
      "step": 2544
    },
    {
      "epoch": 0.6590916448156421,
      "grad_norm": 3.1079261356744516,
      "learning_rate": 3.927586670013959e-07,
      "loss": 0.3208,
      "step": 2545
    },
    {
      "epoch": 0.6590916448156421,
      "eval_PRM Accuracy": 0.8773584905660378,
      "eval_PRM F1": 0.9202453987730062,
      "eval_PRM F1 AUC": 0.8431115767417497,
      "eval_PRM F1 AUC (fixed)": 0.8239916186485071,
      "eval_PRM F1 Neg": 0.7346938775510204,
      "eval_PRM NPV": 0.6923076923076923,
      "eval_PRM Precision": 0.9375,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.22407837212085724,
      "eval_runtime": 15.014,
      "eval_samples_per_second": 2.931,
      "eval_steps_per_second": 0.133,
      "step": 2545
    },
    {
      "epoch": 0.65935061992166,
      "grad_norm": 3.7935200926929102,
      "learning_rate": 3.922340337212599e-07,
      "loss": 0.3656,
      "step": 2546
    },
    {
      "epoch": 0.659609595027678,
      "grad_norm": 4.588194615195379,
      "learning_rate": 3.9170959079405474e-07,
      "loss": 0.3442,
      "step": 2547
    },
    {
      "epoch": 0.6598685701336959,
      "grad_norm": 6.145537094544368,
      "learning_rate": 3.9118533864866305e-07,
      "loss": 0.3815,
      "step": 2548
    },
    {
      "epoch": 0.6601275452397138,
      "grad_norm": 4.541310379208525,
      "learning_rate": 3.90661277713811e-07,
      "loss": 0.3435,
      "step": 2549
    },
    {
      "epoch": 0.6603865203457318,
      "grad_norm": 5.443733881407377,
      "learning_rate": 3.901374084180691e-07,
      "loss": 0.2928,
      "step": 2550
    },
    {
      "epoch": 0.6603865203457318,
      "eval_PRM Accuracy": 0.8679245283018868,
      "eval_PRM F1": 0.9135802469135802,
      "eval_PRM F1 AUC": 0.8370874803562075,
      "eval_PRM F1 AUC (fixed)": 0.818753273965427,
      "eval_PRM F1 Neg": 0.72,
      "eval_PRM NPV": 0.6666666666666666,
      "eval_PRM Precision": 0.9367088607594937,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.22702187299728394,
      "eval_runtime": 14.9156,
      "eval_samples_per_second": 2.95,
      "eval_steps_per_second": 0.134,
      "step": 2550
    },
    {
      "epoch": 0.6606454954517497,
      "grad_norm": 4.085150655578858,
      "learning_rate": 3.8961373118985056e-07,
      "loss": 0.3102,
      "step": 2551
    },
    {
      "epoch": 0.6609044705577677,
      "grad_norm": 4.968131664384284,
      "learning_rate": 3.890902464574116e-07,
      "loss": 0.2881,
      "step": 2552
    },
    {
      "epoch": 0.6611634456637856,
      "grad_norm": 3.1929748903178674,
      "learning_rate": 3.8856695464885144e-07,
      "loss": 0.3339,
      "step": 2553
    },
    {
      "epoch": 0.6614224207698035,
      "grad_norm": 3.190876021310957,
      "learning_rate": 3.8804385619211087e-07,
      "loss": 0.2589,
      "step": 2554
    },
    {
      "epoch": 0.6616813958758214,
      "grad_norm": 6.046362788082813,
      "learning_rate": 3.875209515149729e-07,
      "loss": 0.2515,
      "step": 2555
    },
    {
      "epoch": 0.6616813958758214,
      "eval_PRM Accuracy": 0.8773584905660378,
      "eval_PRM F1": 0.9202453987730062,
      "eval_PRM F1 AUC": 0.8431115767417497,
      "eval_PRM F1 AUC (fixed)": 0.8268727082242011,
      "eval_PRM F1 Neg": 0.7346938775510204,
      "eval_PRM NPV": 0.6923076923076923,
      "eval_PRM Precision": 0.9375,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.22394075989723206,
      "eval_runtime": 14.4114,
      "eval_samples_per_second": 3.053,
      "eval_steps_per_second": 0.139,
      "step": 2555
    },
    {
      "epoch": 0.6619403709818393,
      "grad_norm": 7.516474737044697,
      "learning_rate": 3.869982410450623e-07,
      "loss": 0.3187,
      "step": 2556
    },
    {
      "epoch": 0.6621993460878574,
      "grad_norm": 3.86794474780135,
      "learning_rate": 3.864757252098446e-07,
      "loss": 0.3685,
      "step": 2557
    },
    {
      "epoch": 0.6624583211938753,
      "grad_norm": 5.232139900583562,
      "learning_rate": 3.859534044366263e-07,
      "loss": 0.2953,
      "step": 2558
    },
    {
      "epoch": 0.6627172962998932,
      "grad_norm": 9.928342232031875,
      "learning_rate": 3.8543127915255446e-07,
      "loss": 0.4985,
      "step": 2559
    },
    {
      "epoch": 0.6629762714059111,
      "grad_norm": 3.4564980812429074,
      "learning_rate": 3.849093497846164e-07,
      "loss": 0.2918,
      "step": 2560
    },
    {
      "epoch": 0.6629762714059111,
      "eval_PRM Accuracy": 0.8867924528301887,
      "eval_PRM F1": 0.926829268292683,
      "eval_PRM F1 AUC": 0.8491356731272918,
      "eval_PRM F1 AUC (fixed)": 0.8255631220534311,
      "eval_PRM F1 Neg": 0.75,
      "eval_PRM NPV": 0.72,
      "eval_PRM Precision": 0.9382716049382716,
      "eval_PRM Recall": 0.9156626506024096,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.2248615175485611,
      "eval_runtime": 14.4295,
      "eval_samples_per_second": 3.049,
      "eval_steps_per_second": 0.139,
      "step": 2560
    },
    {
      "epoch": 0.663235246511929,
      "grad_norm": 7.21051916121377,
      "learning_rate": 3.8438761675963926e-07,
      "loss": 0.3453,
      "step": 2561
    },
    {
      "epoch": 0.663494221617947,
      "grad_norm": 4.546882934941051,
      "learning_rate": 3.838660805042888e-07,
      "loss": 0.293,
      "step": 2562
    },
    {
      "epoch": 0.6637531967239649,
      "grad_norm": 4.038845121283006,
      "learning_rate": 3.833447414450707e-07,
      "loss": 0.2634,
      "step": 2563
    },
    {
      "epoch": 0.6640121718299828,
      "grad_norm": 2.4625989762645615,
      "learning_rate": 3.8282360000832937e-07,
      "loss": 0.2696,
      "step": 2564
    },
    {
      "epoch": 0.6642711469360008,
      "grad_norm": 4.69252560266666,
      "learning_rate": 3.823026566202472e-07,
      "loss": 0.2965,
      "step": 2565
    },
    {
      "epoch": 0.6642711469360008,
      "eval_PRM Accuracy": 0.8867924528301887,
      "eval_PRM F1": 0.926829268292683,
      "eval_PRM F1 AUC": 0.8491356731272918,
      "eval_PRM F1 AUC (fixed)": 0.8271346254583551,
      "eval_PRM F1 Neg": 0.75,
      "eval_PRM NPV": 0.72,
      "eval_PRM Precision": 0.9382716049382716,
      "eval_PRM Recall": 0.9156626506024096,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.2231443077325821,
      "eval_runtime": 14.7604,
      "eval_samples_per_second": 2.981,
      "eval_steps_per_second": 0.135,
      "step": 2565
    },
    {
      "epoch": 0.6645301220420187,
      "grad_norm": 5.470840801795296,
      "learning_rate": 3.817819117068449e-07,
      "loss": 0.3668,
      "step": 2566
    },
    {
      "epoch": 0.6647890971480367,
      "grad_norm": 9.075406033173945,
      "learning_rate": 3.812613656939808e-07,
      "loss": 0.5016,
      "step": 2567
    },
    {
      "epoch": 0.6650480722540546,
      "grad_norm": 4.041190021259981,
      "learning_rate": 3.807410190073504e-07,
      "loss": 0.2861,
      "step": 2568
    },
    {
      "epoch": 0.6653070473600725,
      "grad_norm": 2.7495078899506273,
      "learning_rate": 3.802208720724865e-07,
      "loss": 0.2771,
      "step": 2569
    },
    {
      "epoch": 0.6655660224660904,
      "grad_norm": 8.147210530965568,
      "learning_rate": 3.7970092531475836e-07,
      "loss": 0.4153,
      "step": 2570
    },
    {
      "epoch": 0.6655660224660904,
      "eval_PRM Accuracy": 0.8773584905660378,
      "eval_PRM F1": 0.9202453987730062,
      "eval_PRM F1 AUC": 0.8431115767417497,
      "eval_PRM F1 AUC (fixed)": 0.8255631220534312,
      "eval_PRM F1 Neg": 0.7346938775510204,
      "eval_PRM NPV": 0.6923076923076923,
      "eval_PRM Precision": 0.9375,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.22377972304821014,
      "eval_runtime": 15.0567,
      "eval_samples_per_second": 2.922,
      "eval_steps_per_second": 0.133,
      "step": 2570
    },
    {
      "epoch": 0.6658249975721083,
      "grad_norm": 3.244044033249712,
      "learning_rate": 3.7918117915937145e-07,
      "loss": 0.3389,
      "step": 2571
    },
    {
      "epoch": 0.6660839726781264,
      "grad_norm": 3.1648246968439184,
      "learning_rate": 3.7866163403136746e-07,
      "loss": 0.2769,
      "step": 2572
    },
    {
      "epoch": 0.6663429477841443,
      "grad_norm": 2.8134022632434785,
      "learning_rate": 3.781422903556233e-07,
      "loss": 0.2703,
      "step": 2573
    },
    {
      "epoch": 0.6666019228901622,
      "grad_norm": 5.458096802053249,
      "learning_rate": 3.776231485568515e-07,
      "loss": 0.3156,
      "step": 2574
    },
    {
      "epoch": 0.6668608979961801,
      "grad_norm": 3.4681692804917597,
      "learning_rate": 3.7710420905959944e-07,
      "loss": 0.2956,
      "step": 2575
    },
    {
      "epoch": 0.6668608979961801,
      "eval_PRM Accuracy": 0.8773584905660378,
      "eval_PRM F1": 0.9202453987730062,
      "eval_PRM F1 AUC": 0.8431115767417497,
      "eval_PRM F1 AUC (fixed)": 0.8268727082242012,
      "eval_PRM F1 Neg": 0.7346938775510204,
      "eval_PRM NPV": 0.6923076923076923,
      "eval_PRM Precision": 0.9375,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.22493048012256622,
      "eval_runtime": 14.6735,
      "eval_samples_per_second": 2.999,
      "eval_steps_per_second": 0.136,
      "step": 2575
    },
    {
      "epoch": 0.667119873102198,
      "grad_norm": 3.1772027269940604,
      "learning_rate": 3.76585472288249e-07,
      "loss": 0.2839,
      "step": 2576
    },
    {
      "epoch": 0.667378848208216,
      "grad_norm": 4.66072986278749,
      "learning_rate": 3.760669386670163e-07,
      "loss": 0.2861,
      "step": 2577
    },
    {
      "epoch": 0.6676378233142339,
      "grad_norm": 3.3324250876526498,
      "learning_rate": 3.7554860861995096e-07,
      "loss": 0.3132,
      "step": 2578
    },
    {
      "epoch": 0.6678967984202518,
      "grad_norm": 5.190657652860604,
      "learning_rate": 3.7503048257093673e-07,
      "loss": 0.4206,
      "step": 2579
    },
    {
      "epoch": 0.6681557735262698,
      "grad_norm": 5.269749575297229,
      "learning_rate": 3.7451256094369016e-07,
      "loss": 0.3324,
      "step": 2580
    },
    {
      "epoch": 0.6681557735262698,
      "eval_PRM Accuracy": 0.8773584905660378,
      "eval_PRM F1": 0.9202453987730062,
      "eval_PRM F1 AUC": 0.8431115767417497,
      "eval_PRM F1 AUC (fixed)": 0.8268727082242011,
      "eval_PRM F1 Neg": 0.7346938775510204,
      "eval_PRM NPV": 0.6923076923076923,
      "eval_PRM Precision": 0.9375,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.22278118133544922,
      "eval_runtime": 14.8662,
      "eval_samples_per_second": 2.96,
      "eval_steps_per_second": 0.135,
      "step": 2580
    },
    {
      "epoch": 0.6684147486322877,
      "grad_norm": 4.143640213047591,
      "learning_rate": 3.7399484416176066e-07,
      "loss": 0.4053,
      "step": 2581
    },
    {
      "epoch": 0.6686737237383057,
      "grad_norm": 5.657923012647837,
      "learning_rate": 3.7347733264853024e-07,
      "loss": 0.2592,
      "step": 2582
    },
    {
      "epoch": 0.6689326988443236,
      "grad_norm": 3.9316794820107432,
      "learning_rate": 3.729600268272129e-07,
      "loss": 0.339,
      "step": 2583
    },
    {
      "epoch": 0.6691916739503415,
      "grad_norm": 5.834761950862783,
      "learning_rate": 3.7244292712085455e-07,
      "loss": 0.4571,
      "step": 2584
    },
    {
      "epoch": 0.6694506490563594,
      "grad_norm": 3.7049002186060487,
      "learning_rate": 3.7192603395233257e-07,
      "loss": 0.2707,
      "step": 2585
    },
    {
      "epoch": 0.6694506490563594,
      "eval_PRM Accuracy": 0.8679245283018868,
      "eval_PRM F1": 0.9135802469135802,
      "eval_PRM F1 AUC": 0.8370874803562075,
      "eval_PRM F1 AUC (fixed)": 0.8245154531168151,
      "eval_PRM F1 Neg": 0.72,
      "eval_PRM NPV": 0.6666666666666666,
      "eval_PRM Precision": 0.9367088607594937,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.22773408889770508,
      "eval_runtime": 14.8053,
      "eval_samples_per_second": 2.972,
      "eval_steps_per_second": 0.135,
      "step": 2585
    },
    {
      "epoch": 0.6697096241623773,
      "grad_norm": 4.674902378648941,
      "learning_rate": 3.714093477443556e-07,
      "loss": 0.3389,
      "step": 2586
    },
    {
      "epoch": 0.6699685992683954,
      "grad_norm": 7.752934462288362,
      "learning_rate": 3.7089286891946213e-07,
      "loss": 0.3357,
      "step": 2587
    },
    {
      "epoch": 0.6702275743744133,
      "grad_norm": 2.835128929950811,
      "learning_rate": 3.7037659790002206e-07,
      "loss": 0.3001,
      "step": 2588
    },
    {
      "epoch": 0.6704865494804312,
      "grad_norm": 6.820022009977163,
      "learning_rate": 3.698605351082352e-07,
      "loss": 0.2465,
      "step": 2589
    },
    {
      "epoch": 0.6707455245864491,
      "grad_norm": 4.9832945130647275,
      "learning_rate": 3.693446809661305e-07,
      "loss": 0.3374,
      "step": 2590
    },
    {
      "epoch": 0.6707455245864491,
      "eval_PRM Accuracy": 0.8679245283018868,
      "eval_PRM F1": 0.9135802469135802,
      "eval_PRM F1 AUC": 0.8370874803562075,
      "eval_PRM F1 AUC (fixed)": 0.8260869565217391,
      "eval_PRM F1 Neg": 0.72,
      "eval_PRM NPV": 0.6666666666666666,
      "eval_PRM Precision": 0.9367088607594937,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.22627678513526917,
      "eval_runtime": 14.7807,
      "eval_samples_per_second": 2.977,
      "eval_steps_per_second": 0.135,
      "step": 2590
    },
    {
      "epoch": 0.6710044996924671,
      "grad_norm": 6.4114936152046695,
      "learning_rate": 3.688290358955671e-07,
      "loss": 0.3776,
      "step": 2591
    },
    {
      "epoch": 0.671263474798485,
      "grad_norm": 3.7521743829113112,
      "learning_rate": 3.6831360031823237e-07,
      "loss": 0.3344,
      "step": 2592
    },
    {
      "epoch": 0.6715224499045029,
      "grad_norm": 6.665026035619128,
      "learning_rate": 3.677983746556428e-07,
      "loss": 0.4136,
      "step": 2593
    },
    {
      "epoch": 0.6717814250105208,
      "grad_norm": 5.21209505259137,
      "learning_rate": 3.6728335932914334e-07,
      "loss": 0.4212,
      "step": 2594
    },
    {
      "epoch": 0.6720404001165388,
      "grad_norm": 3.757679319422931,
      "learning_rate": 3.6676855475990633e-07,
      "loss": 0.3529,
      "step": 2595
    },
    {
      "epoch": 0.6720404001165388,
      "eval_PRM Accuracy": 0.8679245283018868,
      "eval_PRM F1": 0.9135802469135802,
      "eval_PRM F1 AUC": 0.8370874803562075,
      "eval_PRM F1 AUC (fixed)": 0.822420115243583,
      "eval_PRM F1 Neg": 0.72,
      "eval_PRM NPV": 0.6666666666666666,
      "eval_PRM Precision": 0.9367088607594937,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.2316768616437912,
      "eval_runtime": 14.5485,
      "eval_samples_per_second": 3.024,
      "eval_steps_per_second": 0.137,
      "step": 2595
    },
    {
      "epoch": 0.6722993752225568,
      "grad_norm": 6.455411295944331,
      "learning_rate": 3.662539613689324e-07,
      "loss": 0.239,
      "step": 2596
    },
    {
      "epoch": 0.6725583503285747,
      "grad_norm": 4.379827344950127,
      "learning_rate": 3.6573957957704924e-07,
      "loss": 0.3971,
      "step": 2597
    },
    {
      "epoch": 0.6728173254345926,
      "grad_norm": 4.984766520336958,
      "learning_rate": 3.652254098049113e-07,
      "loss": 0.2538,
      "step": 2598
    },
    {
      "epoch": 0.6730763005406105,
      "grad_norm": 4.374629476354842,
      "learning_rate": 3.6471145247300005e-07,
      "loss": 0.3199,
      "step": 2599
    },
    {
      "epoch": 0.6733352756466284,
      "grad_norm": 10.270684126371698,
      "learning_rate": 3.641977080016229e-07,
      "loss": 0.3131,
      "step": 2600
    },
    {
      "epoch": 0.6733352756466284,
      "eval_PRM Accuracy": 0.8679245283018868,
      "eval_PRM F1": 0.9135802469135802,
      "eval_PRM F1 AUC": 0.8370874803562075,
      "eval_PRM F1 AUC (fixed)": 0.822943949711891,
      "eval_PRM F1 Neg": 0.72,
      "eval_PRM NPV": 0.6666666666666666,
      "eval_PRM Precision": 0.9367088607594937,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.23243652284145355,
      "eval_runtime": 14.7154,
      "eval_samples_per_second": 2.99,
      "eval_steps_per_second": 0.136,
      "step": 2600
    },
    {
      "epoch": 0.6735942507526465,
      "grad_norm": 5.261439670477885,
      "learning_rate": 3.6368417681091297e-07,
      "loss": 0.3064,
      "step": 2601
    },
    {
      "epoch": 0.6738532258586644,
      "grad_norm": 3.214077839112643,
      "learning_rate": 3.631708593208301e-07,
      "loss": 0.3127,
      "step": 2602
    },
    {
      "epoch": 0.6741122009646823,
      "grad_norm": 4.996793816178899,
      "learning_rate": 3.626577559511578e-07,
      "loss": 0.2998,
      "step": 2603
    },
    {
      "epoch": 0.6743711760707002,
      "grad_norm": 5.231503215535616,
      "learning_rate": 3.621448671215054e-07,
      "loss": 0.3784,
      "step": 2604
    },
    {
      "epoch": 0.6746301511767181,
      "grad_norm": 3.664548677287629,
      "learning_rate": 3.616321932513068e-07,
      "loss": 0.2851,
      "step": 2605
    },
    {
      "epoch": 0.6746301511767181,
      "eval_PRM Accuracy": 0.8679245283018868,
      "eval_PRM F1": 0.9135802469135802,
      "eval_PRM F1 AUC": 0.8370874803562075,
      "eval_PRM F1 AUC (fixed)": 0.8237297014143531,
      "eval_PRM F1 Neg": 0.72,
      "eval_PRM NPV": 0.6666666666666666,
      "eval_PRM Precision": 0.9367088607594937,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.22809042036533356,
      "eval_runtime": 14.3707,
      "eval_samples_per_second": 3.062,
      "eval_steps_per_second": 0.139,
      "step": 2605
    },
    {
      "epoch": 0.6748891262827361,
      "grad_norm": 3.6839986151630066,
      "learning_rate": 3.611197347598197e-07,
      "loss": 0.3105,
      "step": 2606
    },
    {
      "epoch": 0.675148101388754,
      "grad_norm": 5.318613488123722,
      "learning_rate": 3.6060749206612596e-07,
      "loss": 0.3959,
      "step": 2607
    },
    {
      "epoch": 0.6754070764947719,
      "grad_norm": 5.24333043180103,
      "learning_rate": 3.6009546558913096e-07,
      "loss": 0.241,
      "step": 2608
    },
    {
      "epoch": 0.6756660516007899,
      "grad_norm": 7.590630458149315,
      "learning_rate": 3.595836557475631e-07,
      "loss": 0.4514,
      "step": 2609
    },
    {
      "epoch": 0.6759250267068078,
      "grad_norm": 3.3507620354213086,
      "learning_rate": 3.5907206295997363e-07,
      "loss": 0.2897,
      "step": 2610
    },
    {
      "epoch": 0.6759250267068078,
      "eval_PRM Accuracy": 0.8773584905660378,
      "eval_PRM F1": 0.9202453987730062,
      "eval_PRM F1 AUC": 0.8431115767417497,
      "eval_PRM F1 AUC (fixed)": 0.8245154531168151,
      "eval_PRM F1 Neg": 0.7346938775510204,
      "eval_PRM NPV": 0.6923076923076923,
      "eval_PRM Precision": 0.9375,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.22583723068237305,
      "eval_runtime": 14.5099,
      "eval_samples_per_second": 3.032,
      "eval_steps_per_second": 0.138,
      "step": 2610
    },
    {
      "epoch": 0.6761840018128258,
      "grad_norm": 3.3269322344856906,
      "learning_rate": 3.585606876447366e-07,
      "loss": 0.349,
      "step": 2611
    },
    {
      "epoch": 0.6764429769188437,
      "grad_norm": 4.458384141074137,
      "learning_rate": 3.5804953022004777e-07,
      "loss": 0.3401,
      "step": 2612
    },
    {
      "epoch": 0.6767019520248616,
      "grad_norm": 5.450303863076529,
      "learning_rate": 3.5753859110392505e-07,
      "loss": 0.3395,
      "step": 2613
    },
    {
      "epoch": 0.6769609271308795,
      "grad_norm": 5.412451729927383,
      "learning_rate": 3.570278707142076e-07,
      "loss": 0.3675,
      "step": 2614
    },
    {
      "epoch": 0.6772199022368974,
      "grad_norm": 5.2966153385789765,
      "learning_rate": 3.565173694685559e-07,
      "loss": 0.389,
      "step": 2615
    },
    {
      "epoch": 0.6772199022368974,
      "eval_PRM Accuracy": 0.8773584905660378,
      "eval_PRM F1": 0.9202453987730062,
      "eval_PRM F1 AUC": 0.8431115767417497,
      "eval_PRM F1 AUC (fixed)": 0.8253012048192772,
      "eval_PRM F1 Neg": 0.7346938775510204,
      "eval_PRM NPV": 0.6923076923076923,
      "eval_PRM Precision": 0.9375,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.22734102606773376,
      "eval_runtime": 14.654,
      "eval_samples_per_second": 3.003,
      "eval_steps_per_second": 0.136,
      "step": 2615
    },
    {
      "epoch": 0.6774788773429155,
      "grad_norm": 3.7981302736964593,
      "learning_rate": 3.560070877844511e-07,
      "loss": 0.3744,
      "step": 2616
    },
    {
      "epoch": 0.6777378524489334,
      "grad_norm": 5.5392428569242345,
      "learning_rate": 3.554970260791947e-07,
      "loss": 0.2934,
      "step": 2617
    },
    {
      "epoch": 0.6779968275549513,
      "grad_norm": 6.8459812282132955,
      "learning_rate": 3.5498718476990867e-07,
      "loss": 0.3507,
      "step": 2618
    },
    {
      "epoch": 0.6782558026609692,
      "grad_norm": 3.9631155433525493,
      "learning_rate": 3.544775642735342e-07,
      "loss": 0.2541,
      "step": 2619
    },
    {
      "epoch": 0.6785147777669871,
      "grad_norm": 4.013909454193455,
      "learning_rate": 3.539681650068324e-07,
      "loss": 0.3077,
      "step": 2620
    },
    {
      "epoch": 0.6785147777669871,
      "eval_PRM Accuracy": 0.8679245283018868,
      "eval_PRM F1": 0.9135802469135802,
      "eval_PRM F1 AUC": 0.8370874803562075,
      "eval_PRM F1 AUC (fixed)": 0.8239916186485071,
      "eval_PRM F1 Neg": 0.72,
      "eval_PRM NPV": 0.6666666666666666,
      "eval_PRM Precision": 0.9367088607594937,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.2303287386894226,
      "eval_runtime": 14.4627,
      "eval_samples_per_second": 3.042,
      "eval_steps_per_second": 0.138,
      "step": 2620
    },
    {
      "epoch": 0.6787737528730051,
      "grad_norm": 4.0434408468716505,
      "learning_rate": 3.5345898738638315e-07,
      "loss": 0.3357,
      "step": 2621
    },
    {
      "epoch": 0.679032727979023,
      "grad_norm": 5.665843252660804,
      "learning_rate": 3.5295003182858526e-07,
      "loss": 0.3907,
      "step": 2622
    },
    {
      "epoch": 0.6792917030850409,
      "grad_norm": 3.5434282927182243,
      "learning_rate": 3.524412987496558e-07,
      "loss": 0.3185,
      "step": 2623
    },
    {
      "epoch": 0.6795506781910589,
      "grad_norm": 5.394699982219372,
      "learning_rate": 3.5193278856562997e-07,
      "loss": 0.3446,
      "step": 2624
    },
    {
      "epoch": 0.6798096532970768,
      "grad_norm": 3.9860668072430645,
      "learning_rate": 3.514245016923608e-07,
      "loss": 0.3394,
      "step": 2625
    },
    {
      "epoch": 0.6798096532970768,
      "eval_PRM Accuracy": 0.8679245283018868,
      "eval_PRM F1": 0.9135802469135802,
      "eval_PRM F1 AUC": 0.8370874803562075,
      "eval_PRM F1 AUC (fixed)": 0.8245154531168151,
      "eval_PRM F1 Neg": 0.72,
      "eval_PRM NPV": 0.6666666666666666,
      "eval_PRM Precision": 0.9367088607594937,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.23323310911655426,
      "eval_runtime": 14.8355,
      "eval_samples_per_second": 2.966,
      "eval_steps_per_second": 0.135,
      "step": 2625
    },
    {
      "epoch": 0.6800686284030948,
      "grad_norm": 4.374407999350048,
      "learning_rate": 3.5091643854551825e-07,
      "loss": 0.4071,
      "step": 2626
    },
    {
      "epoch": 0.6803276035091127,
      "grad_norm": 3.957342990243323,
      "learning_rate": 3.504085995405903e-07,
      "loss": 0.3948,
      "step": 2627
    },
    {
      "epoch": 0.6805865786151306,
      "grad_norm": 5.1094346955408865,
      "learning_rate": 3.499009850928803e-07,
      "loss": 0.3291,
      "step": 2628
    },
    {
      "epoch": 0.6808455537211485,
      "grad_norm": 9.763514407395833,
      "learning_rate": 3.4939359561750846e-07,
      "loss": 0.3221,
      "step": 2629
    },
    {
      "epoch": 0.6811045288271664,
      "grad_norm": 3.5814585879565457,
      "learning_rate": 3.48886431529412e-07,
      "loss": 0.2792,
      "step": 2630
    },
    {
      "epoch": 0.6811045288271664,
      "eval_PRM Accuracy": 0.8679245283018868,
      "eval_PRM F1": 0.9135802469135802,
      "eval_PRM F1 AUC": 0.8370874803562075,
      "eval_PRM F1 AUC (fixed)": 0.8247773703509691,
      "eval_PRM F1 Neg": 0.72,
      "eval_PRM NPV": 0.6666666666666666,
      "eval_PRM Precision": 0.9367088607594937,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.23205780982971191,
      "eval_runtime": 14.6935,
      "eval_samples_per_second": 2.995,
      "eval_steps_per_second": 0.136,
      "step": 2630
    },
    {
      "epoch": 0.6813635039331845,
      "grad_norm": 4.696201403165561,
      "learning_rate": 3.483794932433425e-07,
      "loss": 0.302,
      "step": 2631
    },
    {
      "epoch": 0.6816224790392024,
      "grad_norm": 5.618815820131977,
      "learning_rate": 3.4787278117386763e-07,
      "loss": 0.3207,
      "step": 2632
    },
    {
      "epoch": 0.6818814541452203,
      "grad_norm": 8.293395148841052,
      "learning_rate": 3.473662957353696e-07,
      "loss": 0.2796,
      "step": 2633
    },
    {
      "epoch": 0.6821404292512382,
      "grad_norm": 3.5297940259603333,
      "learning_rate": 3.4686003734204567e-07,
      "loss": 0.2893,
      "step": 2634
    },
    {
      "epoch": 0.6823994043572562,
      "grad_norm": 3.2863981439116094,
      "learning_rate": 3.4635400640790714e-07,
      "loss": 0.3237,
      "step": 2635
    },
    {
      "epoch": 0.6823994043572562,
      "eval_PRM Accuracy": 0.8679245283018868,
      "eval_PRM F1": 0.9135802469135802,
      "eval_PRM F1 AUC": 0.8370874803562075,
      "eval_PRM F1 AUC (fixed)": 0.8271346254583553,
      "eval_PRM F1 Neg": 0.72,
      "eval_PRM NPV": 0.6666666666666666,
      "eval_PRM Precision": 0.9367088607594937,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.225877046585083,
      "eval_runtime": 14.8173,
      "eval_samples_per_second": 2.969,
      "eval_steps_per_second": 0.135,
      "step": 2635
    },
    {
      "epoch": 0.6826583794632741,
      "grad_norm": 5.882680181821401,
      "learning_rate": 3.458482033467797e-07,
      "loss": 0.3228,
      "step": 2636
    },
    {
      "epoch": 0.682917354569292,
      "grad_norm": 6.050439323232917,
      "learning_rate": 3.4534262857230204e-07,
      "loss": 0.351,
      "step": 2637
    },
    {
      "epoch": 0.68317632967531,
      "grad_norm": 2.9916254963010265,
      "learning_rate": 3.448372824979265e-07,
      "loss": 0.2683,
      "step": 2638
    },
    {
      "epoch": 0.6834353047813279,
      "grad_norm": 2.8548040280647116,
      "learning_rate": 3.4433216553691876e-07,
      "loss": 0.274,
      "step": 2639
    },
    {
      "epoch": 0.6836942798873459,
      "grad_norm": 8.009880999015257,
      "learning_rate": 3.438272781023563e-07,
      "loss": 0.3727,
      "step": 2640
    },
    {
      "epoch": 0.6836942798873459,
      "eval_PRM Accuracy": 0.8773584905660378,
      "eval_PRM F1": 0.9202453987730062,
      "eval_PRM F1 AUC": 0.8431115767417497,
      "eval_PRM F1 AUC (fixed)": 0.8271346254583551,
      "eval_PRM F1 Neg": 0.7346938775510204,
      "eval_PRM NPV": 0.6923076923076923,
      "eval_PRM Precision": 0.9375,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.22475957870483398,
      "eval_runtime": 14.6271,
      "eval_samples_per_second": 3.008,
      "eval_steps_per_second": 0.137,
      "step": 2640
    },
    {
      "epoch": 0.6839532549933638,
      "grad_norm": 3.306346434957066,
      "learning_rate": 3.4332262060713004e-07,
      "loss": 0.3597,
      "step": 2641
    },
    {
      "epoch": 0.6842122300993817,
      "grad_norm": 6.011927474573627,
      "learning_rate": 3.4281819346394187e-07,
      "loss": 0.3344,
      "step": 2642
    },
    {
      "epoch": 0.6844712052053996,
      "grad_norm": 3.5097708913224466,
      "learning_rate": 3.4231399708530553e-07,
      "loss": 0.351,
      "step": 2643
    },
    {
      "epoch": 0.6847301803114175,
      "grad_norm": 2.501548219172394,
      "learning_rate": 3.418100318835467e-07,
      "loss": 0.2436,
      "step": 2644
    },
    {
      "epoch": 0.6849891554174355,
      "grad_norm": 3.861429657120474,
      "learning_rate": 3.413062982708007e-07,
      "loss": 0.2624,
      "step": 2645
    },
    {
      "epoch": 0.6849891554174355,
      "eval_PRM Accuracy": 0.8773584905660378,
      "eval_PRM F1": 0.9202453987730062,
      "eval_PRM F1 AUC": 0.8431115767417497,
      "eval_PRM F1 AUC (fixed)": 0.8287061288632792,
      "eval_PRM F1 Neg": 0.7346938775510204,
      "eval_PRM NPV": 0.6923076923076923,
      "eval_PRM Precision": 0.9375,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.22434702515602112,
      "eval_runtime": 14.9648,
      "eval_samples_per_second": 2.94,
      "eval_steps_per_second": 0.134,
      "step": 2645
    },
    {
      "epoch": 0.6852481305234535,
      "grad_norm": 4.495833999830678,
      "learning_rate": 3.408027966590146e-07,
      "loss": 0.2413,
      "step": 2646
    },
    {
      "epoch": 0.6855071056294714,
      "grad_norm": 4.686422395330833,
      "learning_rate": 3.402995274599453e-07,
      "loss": 0.3332,
      "step": 2647
    },
    {
      "epoch": 0.6857660807354893,
      "grad_norm": 2.973031471402928,
      "learning_rate": 3.397964910851598e-07,
      "loss": 0.2725,
      "step": 2648
    },
    {
      "epoch": 0.6860250558415072,
      "grad_norm": 3.9507078014848953,
      "learning_rate": 3.3929368794603447e-07,
      "loss": 0.2799,
      "step": 2649
    },
    {
      "epoch": 0.6862840309475252,
      "grad_norm": 5.065435584750594,
      "learning_rate": 3.38791118453755e-07,
      "loss": 0.3145,
      "step": 2650
    },
    {
      "epoch": 0.6862840309475252,
      "eval_PRM Accuracy": 0.8773584905660378,
      "eval_PRM F1": 0.9202453987730062,
      "eval_PRM F1 AUC": 0.8431115767417497,
      "eval_PRM F1 AUC (fixed)": 0.8321110529072814,
      "eval_PRM F1 Neg": 0.7346938775510204,
      "eval_PRM NPV": 0.6923076923076923,
      "eval_PRM Precision": 0.9375,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.22204075753688812,
      "eval_runtime": 15.0916,
      "eval_samples_per_second": 2.916,
      "eval_steps_per_second": 0.133,
      "step": 2650
    },
    {
      "epoch": 0.6865430060535431,
      "grad_norm": 3.739397963160035,
      "learning_rate": 3.382887830193163e-07,
      "loss": 0.2906,
      "step": 2651
    },
    {
      "epoch": 0.686801981159561,
      "grad_norm": 3.3339655816083167,
      "learning_rate": 3.377866820535219e-07,
      "loss": 0.2278,
      "step": 2652
    },
    {
      "epoch": 0.687060956265579,
      "grad_norm": 3.6807899742299273,
      "learning_rate": 3.3728481596698267e-07,
      "loss": 0.2957,
      "step": 2653
    },
    {
      "epoch": 0.6873199313715969,
      "grad_norm": 4.719270478433307,
      "learning_rate": 3.367831851701185e-07,
      "loss": 0.3442,
      "step": 2654
    },
    {
      "epoch": 0.6875789064776149,
      "grad_norm": 4.215279695136565,
      "learning_rate": 3.362817900731562e-07,
      "loss": 0.3612,
      "step": 2655
    },
    {
      "epoch": 0.6875789064776149,
      "eval_PRM Accuracy": 0.8679245283018868,
      "eval_PRM F1": 0.9135802469135802,
      "eval_PRM F1 AUC": 0.8370874803562075,
      "eval_PRM F1 AUC (fixed)": 0.8331587218438973,
      "eval_PRM F1 Neg": 0.72,
      "eval_PRM NPV": 0.6666666666666666,
      "eval_PRM Precision": 0.9367088607594937,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.22059224545955658,
      "eval_runtime": 14.7783,
      "eval_samples_per_second": 2.977,
      "eval_steps_per_second": 0.135,
      "step": 2655
    },
    {
      "epoch": 0.6878378815836328,
      "grad_norm": 3.8811057762618764,
      "learning_rate": 3.3578063108613025e-07,
      "loss": 0.3378,
      "step": 2656
    },
    {
      "epoch": 0.6880968566896507,
      "grad_norm": 2.8331022896457996,
      "learning_rate": 3.3527970861888173e-07,
      "loss": 0.2923,
      "step": 2657
    },
    {
      "epoch": 0.6883558317956686,
      "grad_norm": 4.025221961435094,
      "learning_rate": 3.347790230810584e-07,
      "loss": 0.2703,
      "step": 2658
    },
    {
      "epoch": 0.6886148069016865,
      "grad_norm": 3.798660220804069,
      "learning_rate": 3.3427857488211437e-07,
      "loss": 0.4213,
      "step": 2659
    },
    {
      "epoch": 0.6888737820077045,
      "grad_norm": 4.368138757499022,
      "learning_rate": 3.337783644313093e-07,
      "loss": 0.353,
      "step": 2660
    },
    {
      "epoch": 0.6888737820077045,
      "eval_PRM Accuracy": 0.8679245283018868,
      "eval_PRM F1": 0.9135802469135802,
      "eval_PRM F1 AUC": 0.8370874803562075,
      "eval_PRM F1 AUC (fixed)": 0.8315872184389734,
      "eval_PRM F1 Neg": 0.72,
      "eval_PRM NPV": 0.6666666666666666,
      "eval_PRM Precision": 0.9367088607594937,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.22033046185970306,
      "eval_runtime": 14.8185,
      "eval_samples_per_second": 2.969,
      "eval_steps_per_second": 0.135,
      "step": 2660
    },
    {
      "epoch": 0.6891327571137225,
      "grad_norm": 3.7997981886027867,
      "learning_rate": 3.332783921377093e-07,
      "loss": 0.2605,
      "step": 2661
    },
    {
      "epoch": 0.6893917322197404,
      "grad_norm": 5.575815043208904,
      "learning_rate": 3.3277865841018454e-07,
      "loss": 0.2963,
      "step": 2662
    },
    {
      "epoch": 0.6896507073257583,
      "grad_norm": 4.531323797040318,
      "learning_rate": 3.3227916365741074e-07,
      "loss": 0.3384,
      "step": 2663
    },
    {
      "epoch": 0.6899096824317762,
      "grad_norm": 5.2451193623419305,
      "learning_rate": 3.317799082878683e-07,
      "loss": 0.31,
      "step": 2664
    },
    {
      "epoch": 0.6901686575377942,
      "grad_norm": 3.710015633472843,
      "learning_rate": 3.3128089270984155e-07,
      "loss": 0.2892,
      "step": 2665
    },
    {
      "epoch": 0.6901686575377942,
      "eval_PRM Accuracy": 0.8773584905660378,
      "eval_PRM F1": 0.9202453987730062,
      "eval_PRM F1 AUC": 0.8431115767417497,
      "eval_PRM F1 AUC (fixed)": 0.8323729701414353,
      "eval_PRM F1 Neg": 0.7346938775510204,
      "eval_PRM NPV": 0.6923076923076923,
      "eval_PRM Precision": 0.9375,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.21999172866344452,
      "eval_runtime": 14.7599,
      "eval_samples_per_second": 2.981,
      "eval_steps_per_second": 0.136,
      "step": 2665
    },
    {
      "epoch": 0.6904276326438121,
      "grad_norm": 6.062999497194352,
      "learning_rate": 3.3078211733141874e-07,
      "loss": 0.3209,
      "step": 2666
    },
    {
      "epoch": 0.69068660774983,
      "grad_norm": 3.2169484052503603,
      "learning_rate": 3.302835825604919e-07,
      "loss": 0.2139,
      "step": 2667
    },
    {
      "epoch": 0.690945582855848,
      "grad_norm": 3.0425702101012853,
      "learning_rate": 3.297852888047562e-07,
      "loss": 0.2895,
      "step": 2668
    },
    {
      "epoch": 0.6912045579618659,
      "grad_norm": 5.907388995702954,
      "learning_rate": 3.2928723647170956e-07,
      "loss": 0.3431,
      "step": 2669
    },
    {
      "epoch": 0.6914635330678839,
      "grad_norm": 3.4708325389805887,
      "learning_rate": 3.2878942596865274e-07,
      "loss": 0.2944,
      "step": 2670
    },
    {
      "epoch": 0.6914635330678839,
      "eval_PRM Accuracy": 0.8773584905660378,
      "eval_PRM F1": 0.9202453987730062,
      "eval_PRM F1 AUC": 0.8431115767417497,
      "eval_PRM F1 AUC (fixed)": 0.8326348873755893,
      "eval_PRM F1 Neg": 0.7346938775510204,
      "eval_PRM NPV": 0.6923076923076923,
      "eval_PRM Precision": 0.9375,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.2213703989982605,
      "eval_runtime": 14.548,
      "eval_samples_per_second": 3.024,
      "eval_steps_per_second": 0.137,
      "step": 2670
    },
    {
      "epoch": 0.6917225081739018,
      "grad_norm": 4.273014782730057,
      "learning_rate": 3.282918577026886e-07,
      "loss": 0.2777,
      "step": 2671
    },
    {
      "epoch": 0.6919814832799197,
      "grad_norm": 3.1429999888332496,
      "learning_rate": 3.2779453208072174e-07,
      "loss": 0.3004,
      "step": 2672
    },
    {
      "epoch": 0.6922404583859376,
      "grad_norm": 4.735525747528506,
      "learning_rate": 3.272974495094585e-07,
      "loss": 0.3431,
      "step": 2673
    },
    {
      "epoch": 0.6924994334919556,
      "grad_norm": 4.222359248869235,
      "learning_rate": 3.268006103954066e-07,
      "loss": 0.2699,
      "step": 2674
    },
    {
      "epoch": 0.6927584085979736,
      "grad_norm": 5.264383006228346,
      "learning_rate": 3.263040151448743e-07,
      "loss": 0.3508,
      "step": 2675
    },
    {
      "epoch": 0.6927584085979736,
      "eval_PRM Accuracy": 0.8773584905660378,
      "eval_PRM F1": 0.9202453987730062,
      "eval_PRM F1 AUC": 0.8431115767417497,
      "eval_PRM F1 AUC (fixed)": 0.8328968046097434,
      "eval_PRM F1 Neg": 0.7346938775510204,
      "eval_PRM NPV": 0.6923076923076923,
      "eval_PRM Precision": 0.9375,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.2207181304693222,
      "eval_runtime": 14.149,
      "eval_samples_per_second": 3.11,
      "eval_steps_per_second": 0.141,
      "step": 2675
    },
    {
      "epoch": 0.6930173837039915,
      "grad_norm": 5.5757396141372055,
      "learning_rate": 3.2580766416397084e-07,
      "loss": 0.2817,
      "step": 2676
    },
    {
      "epoch": 0.6932763588100094,
      "grad_norm": 4.219224271703278,
      "learning_rate": 3.2531155785860556e-07,
      "loss": 0.2379,
      "step": 2677
    },
    {
      "epoch": 0.6935353339160273,
      "grad_norm": 2.9437244436508965,
      "learning_rate": 3.2481569663448705e-07,
      "loss": 0.3242,
      "step": 2678
    },
    {
      "epoch": 0.6937943090220453,
      "grad_norm": 4.097682366601122,
      "learning_rate": 3.2432008089712463e-07,
      "loss": 0.2374,
      "step": 2679
    },
    {
      "epoch": 0.6940532841280632,
      "grad_norm": 4.510887895158138,
      "learning_rate": 3.2382471105182623e-07,
      "loss": 0.3146,
      "step": 2680
    },
    {
      "epoch": 0.6940532841280632,
      "eval_PRM Accuracy": 0.8773584905660378,
      "eval_PRM F1": 0.9202453987730062,
      "eval_PRM F1 AUC": 0.8431115767417497,
      "eval_PRM F1 AUC (fixed)": 0.8360398114195914,
      "eval_PRM F1 Neg": 0.7346938775510204,
      "eval_PRM NPV": 0.6923076923076923,
      "eval_PRM Precision": 0.9375,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.22125062346458435,
      "eval_runtime": 14.9641,
      "eval_samples_per_second": 2.94,
      "eval_steps_per_second": 0.134,
      "step": 2680
    },
    {
      "epoch": 0.6943122592340811,
      "grad_norm": 4.674920163613529,
      "learning_rate": 3.2332958750369863e-07,
      "loss": 0.4283,
      "step": 2681
    },
    {
      "epoch": 0.694571234340099,
      "grad_norm": 3.5744556908996183,
      "learning_rate": 3.2283471065764736e-07,
      "loss": 0.3613,
      "step": 2682
    },
    {
      "epoch": 0.694830209446117,
      "grad_norm": 5.326620229651874,
      "learning_rate": 3.2234008091837624e-07,
      "loss": 0.2009,
      "step": 2683
    },
    {
      "epoch": 0.695089184552135,
      "grad_norm": 3.2059252832989267,
      "learning_rate": 3.2184569869038685e-07,
      "loss": 0.2656,
      "step": 2684
    },
    {
      "epoch": 0.6953481596581529,
      "grad_norm": 3.2261200715517857,
      "learning_rate": 3.2135156437797867e-07,
      "loss": 0.3075,
      "step": 2685
    },
    {
      "epoch": 0.6953481596581529,
      "eval_PRM Accuracy": 0.8773584905660378,
      "eval_PRM F1": 0.9202453987730062,
      "eval_PRM F1 AUC": 0.8431115767417497,
      "eval_PRM F1 AUC (fixed)": 0.8347302252488215,
      "eval_PRM F1 Neg": 0.7346938775510204,
      "eval_PRM NPV": 0.6923076923076923,
      "eval_PRM Precision": 0.9375,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.2207738310098648,
      "eval_runtime": 14.579,
      "eval_samples_per_second": 3.018,
      "eval_steps_per_second": 0.137,
      "step": 2685
    },
    {
      "epoch": 0.6956071347641708,
      "grad_norm": 4.086141908290938,
      "learning_rate": 3.208576783852478e-07,
      "loss": 0.303,
      "step": 2686
    },
    {
      "epoch": 0.6958661098701887,
      "grad_norm": 4.755136659209133,
      "learning_rate": 3.203640411160879e-07,
      "loss": 0.3419,
      "step": 2687
    },
    {
      "epoch": 0.6961250849762066,
      "grad_norm": 7.1402960921731164,
      "learning_rate": 3.1987065297418874e-07,
      "loss": 0.3432,
      "step": 2688
    },
    {
      "epoch": 0.6963840600822246,
      "grad_norm": 4.086532213923248,
      "learning_rate": 3.19377514363037e-07,
      "loss": 0.3541,
      "step": 2689
    },
    {
      "epoch": 0.6966430351882426,
      "grad_norm": 6.542220581866447,
      "learning_rate": 3.18884625685915e-07,
      "loss": 0.4552,
      "step": 2690
    },
    {
      "epoch": 0.6966430351882426,
      "eval_PRM Accuracy": 0.8773584905660378,
      "eval_PRM F1": 0.9202453987730062,
      "eval_PRM F1 AUC": 0.8431115767417497,
      "eval_PRM F1 AUC (fixed)": 0.8318491356731273,
      "eval_PRM F1 Neg": 0.7346938775510204,
      "eval_PRM NPV": 0.6923076923076923,
      "eval_PRM Precision": 0.9375,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.22063879668712616,
      "eval_runtime": 14.2753,
      "eval_samples_per_second": 3.082,
      "eval_steps_per_second": 0.14,
      "step": 2690
    },
    {
      "epoch": 0.6969020102942605,
      "grad_norm": 3.0357902288432905,
      "learning_rate": 3.1839198734590024e-07,
      "loss": 0.2804,
      "step": 2691
    },
    {
      "epoch": 0.6971609854002784,
      "grad_norm": 3.89909619647914,
      "learning_rate": 3.17899599745866e-07,
      "loss": 0.3697,
      "step": 2692
    },
    {
      "epoch": 0.6974199605062963,
      "grad_norm": 4.367792488262871,
      "learning_rate": 3.1740746328848035e-07,
      "loss": 0.4168,
      "step": 2693
    },
    {
      "epoch": 0.6976789356123143,
      "grad_norm": 3.1261854011459134,
      "learning_rate": 3.1691557837620614e-07,
      "loss": 0.3144,
      "step": 2694
    },
    {
      "epoch": 0.6979379107183322,
      "grad_norm": 4.769934073600949,
      "learning_rate": 3.1642394541129997e-07,
      "loss": 0.3558,
      "step": 2695
    },
    {
      "epoch": 0.6979379107183322,
      "eval_PRM Accuracy": 0.8773584905660378,
      "eval_PRM F1": 0.9202453987730062,
      "eval_PRM F1 AUC": 0.8431115767417497,
      "eval_PRM F1 AUC (fixed)": 0.8302776322682033,
      "eval_PRM F1 Neg": 0.7346938775510204,
      "eval_PRM NPV": 0.6923076923076923,
      "eval_PRM Precision": 0.9375,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.22287343442440033,
      "eval_runtime": 14.7072,
      "eval_samples_per_second": 2.992,
      "eval_steps_per_second": 0.136,
      "step": 2695
    },
    {
      "epoch": 0.6981968858243501,
      "grad_norm": 3.992140824979468,
      "learning_rate": 3.15932564795813e-07,
      "loss": 0.3504,
      "step": 2696
    },
    {
      "epoch": 0.698455860930368,
      "grad_norm": 5.276370938336217,
      "learning_rate": 3.1544143693158975e-07,
      "loss": 0.411,
      "step": 2697
    },
    {
      "epoch": 0.698714836036386,
      "grad_norm": 6.573105713451107,
      "learning_rate": 3.149505622202682e-07,
      "loss": 0.2972,
      "step": 2698
    },
    {
      "epoch": 0.698973811142404,
      "grad_norm": 3.209707467753227,
      "learning_rate": 3.1445994106327884e-07,
      "loss": 0.3164,
      "step": 2699
    },
    {
      "epoch": 0.6992327862484219,
      "grad_norm": 3.08125380748184,
      "learning_rate": 3.1396957386184563e-07,
      "loss": 0.3146,
      "step": 2700
    },
    {
      "epoch": 0.6992327862484219,
      "eval_PRM Accuracy": 0.8773584905660378,
      "eval_PRM F1": 0.9202453987730062,
      "eval_PRM F1 AUC": 0.8431115767417497,
      "eval_PRM F1 AUC (fixed)": 0.8273965426925092,
      "eval_PRM F1 Neg": 0.7346938775510204,
      "eval_PRM NPV": 0.6923076923076923,
      "eval_PRM Precision": 0.9375,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.2260752171278,
      "eval_runtime": 14.6387,
      "eval_samples_per_second": 3.006,
      "eval_steps_per_second": 0.137,
      "step": 2700
    },
    {
      "epoch": 0.6994917613544398,
      "grad_norm": 3.6329018602024754,
      "learning_rate": 3.134794610169843e-07,
      "loss": 0.2941,
      "step": 2701
    },
    {
      "epoch": 0.6997507364604577,
      "grad_norm": 4.535228462741428,
      "learning_rate": 3.1298960292950276e-07,
      "loss": 0.4249,
      "step": 2702
    },
    {
      "epoch": 0.7000097115664756,
      "grad_norm": 5.569781291128044,
      "learning_rate": 3.125000000000002e-07,
      "loss": 0.3179,
      "step": 2703
    },
    {
      "epoch": 0.7002686866724936,
      "grad_norm": 3.679483610020434,
      "learning_rate": 3.1201065262886736e-07,
      "loss": 0.3134,
      "step": 2704
    },
    {
      "epoch": 0.7005276617785116,
      "grad_norm": 4.198320110955473,
      "learning_rate": 3.115215612162863e-07,
      "loss": 0.3673,
      "step": 2705
    },
    {
      "epoch": 0.7005276617785116,
      "eval_PRM Accuracy": 0.8679245283018868,
      "eval_PRM F1": 0.9135802469135802,
      "eval_PRM F1 AUC": 0.8370874803562075,
      "eval_PRM F1 AUC (fixed)": 0.8255631220534312,
      "eval_PRM F1 Neg": 0.72,
      "eval_PRM NPV": 0.6666666666666666,
      "eval_PRM Precision": 0.9367088607594937,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.2282734364271164,
      "eval_runtime": 14.7979,
      "eval_samples_per_second": 2.973,
      "eval_steps_per_second": 0.135,
      "step": 2705
    },
    {
      "epoch": 0.7007866368845295,
      "grad_norm": 2.6301430966803254,
      "learning_rate": 3.1103272616222935e-07,
      "loss": 0.2454,
      "step": 2706
    },
    {
      "epoch": 0.7010456119905474,
      "grad_norm": 3.5742194638576392,
      "learning_rate": 3.1054414786645947e-07,
      "loss": 0.3005,
      "step": 2707
    },
    {
      "epoch": 0.7013045870965653,
      "grad_norm": 4.116200991283721,
      "learning_rate": 3.1005582672852945e-07,
      "loss": 0.3659,
      "step": 2708
    },
    {
      "epoch": 0.7015635622025833,
      "grad_norm": 3.956223886898802,
      "learning_rate": 3.0956776314778176e-07,
      "loss": 0.315,
      "step": 2709
    },
    {
      "epoch": 0.7018225373086012,
      "grad_norm": 2.9122398422814277,
      "learning_rate": 3.090799575233489e-07,
      "loss": 0.3252,
      "step": 2710
    },
    {
      "epoch": 0.7018225373086012,
      "eval_PRM Accuracy": 0.8679245283018868,
      "eval_PRM F1": 0.9135802469135802,
      "eval_PRM F1 AUC": 0.8370874803562075,
      "eval_PRM F1 AUC (fixed)": 0.821896280775275,
      "eval_PRM F1 Neg": 0.72,
      "eval_PRM NPV": 0.6666666666666666,
      "eval_PRM Precision": 0.9367088607594937,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.2298944890499115,
      "eval_runtime": 15.0732,
      "eval_samples_per_second": 2.919,
      "eval_steps_per_second": 0.133,
      "step": 2710
    },
    {
      "epoch": 0.7020815124146191,
      "grad_norm": 4.67735415555063,
      "learning_rate": 3.085924102541513e-07,
      "loss": 0.3032,
      "step": 2711
    },
    {
      "epoch": 0.702340487520637,
      "grad_norm": 3.1109462260235365,
      "learning_rate": 3.0810512173889877e-07,
      "loss": 0.2875,
      "step": 2712
    },
    {
      "epoch": 0.702599462626655,
      "grad_norm": 7.5053708468861675,
      "learning_rate": 3.076180923760894e-07,
      "loss": 0.3005,
      "step": 2713
    },
    {
      "epoch": 0.702858437732673,
      "grad_norm": 3.3319128285406765,
      "learning_rate": 3.0713132256400954e-07,
      "loss": 0.3188,
      "step": 2714
    },
    {
      "epoch": 0.7031174128386909,
      "grad_norm": 3.298446767224876,
      "learning_rate": 3.06644812700733e-07,
      "loss": 0.3211,
      "step": 2715
    },
    {
      "epoch": 0.7031174128386909,
      "eval_PRM Accuracy": 0.8773584905660378,
      "eval_PRM F1": 0.9202453987730062,
      "eval_PRM F1 AUC": 0.8431115767417497,
      "eval_PRM F1 AUC (fixed)": 0.8273965426925092,
      "eval_PRM F1 Neg": 0.7346938775510204,
      "eval_PRM NPV": 0.6923076923076923,
      "eval_PRM Precision": 0.9375,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.2297452986240387,
      "eval_runtime": 14.4237,
      "eval_samples_per_second": 3.051,
      "eval_steps_per_second": 0.139,
      "step": 2715
    },
    {
      "epoch": 0.7033763879447088,
      "grad_norm": 3.9236319816200043,
      "learning_rate": 3.061585631841211e-07,
      "loss": 0.328,
      "step": 2716
    },
    {
      "epoch": 0.7036353630507267,
      "grad_norm": 4.256136830498222,
      "learning_rate": 3.056725744118223e-07,
      "loss": 0.2768,
      "step": 2717
    },
    {
      "epoch": 0.7038943381567447,
      "grad_norm": 7.743713943482351,
      "learning_rate": 3.051868467812718e-07,
      "loss": 0.3973,
      "step": 2718
    },
    {
      "epoch": 0.7041533132627626,
      "grad_norm": 3.2750561637422444,
      "learning_rate": 3.0470138068969123e-07,
      "loss": 0.3606,
      "step": 2719
    },
    {
      "epoch": 0.7044122883687806,
      "grad_norm": 4.369739235160086,
      "learning_rate": 3.042161765340883e-07,
      "loss": 0.3428,
      "step": 2720
    },
    {
      "epoch": 0.7044122883687806,
      "eval_PRM Accuracy": 0.8679245283018868,
      "eval_PRM F1": 0.9135802469135802,
      "eval_PRM F1 AUC": 0.8370874803562075,
      "eval_PRM F1 AUC (fixed)": 0.8258250392875851,
      "eval_PRM F1 Neg": 0.72,
      "eval_PRM NPV": 0.6666666666666666,
      "eval_PRM Precision": 0.9367088607594937,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.22882229089736938,
      "eval_runtime": 14.8597,
      "eval_samples_per_second": 2.961,
      "eval_steps_per_second": 0.135,
      "step": 2720
    },
    {
      "epoch": 0.7046712634747985,
      "grad_norm": 4.407453295582052,
      "learning_rate": 3.0373123471125665e-07,
      "loss": 0.3789,
      "step": 2721
    },
    {
      "epoch": 0.7049302385808164,
      "grad_norm": 3.731797470893279,
      "learning_rate": 3.0324655561777524e-07,
      "loss": 0.3131,
      "step": 2722
    },
    {
      "epoch": 0.7051892136868344,
      "grad_norm": 7.346784968386983,
      "learning_rate": 3.0276213965000824e-07,
      "loss": 0.3515,
      "step": 2723
    },
    {
      "epoch": 0.7054481887928523,
      "grad_norm": 7.748996846066618,
      "learning_rate": 3.022779872041046e-07,
      "loss": 0.3073,
      "step": 2724
    },
    {
      "epoch": 0.7057071638988702,
      "grad_norm": 2.9713856216056778,
      "learning_rate": 3.0179409867599777e-07,
      "loss": 0.2832,
      "step": 2725
    },
    {
      "epoch": 0.7057071638988702,
      "eval_PRM Accuracy": 0.8679245283018868,
      "eval_PRM F1": 0.9135802469135802,
      "eval_PRM F1 AUC": 0.8370874803562075,
      "eval_PRM F1 AUC (fixed)": 0.8271346254583551,
      "eval_PRM F1 Neg": 0.72,
      "eval_PRM NPV": 0.6666666666666666,
      "eval_PRM Precision": 0.9367088607594937,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.22889962792396545,
      "eval_runtime": 14.6227,
      "eval_samples_per_second": 3.009,
      "eval_steps_per_second": 0.137,
      "step": 2725
    },
    {
      "epoch": 0.7059661390048881,
      "grad_norm": 5.772033490053363,
      "learning_rate": 3.0131047446140544e-07,
      "loss": 0.2717,
      "step": 2726
    },
    {
      "epoch": 0.706225114110906,
      "grad_norm": 5.2826521222345635,
      "learning_rate": 3.008271149558291e-07,
      "loss": 0.3661,
      "step": 2727
    },
    {
      "epoch": 0.7064840892169241,
      "grad_norm": 2.9766358473803596,
      "learning_rate": 3.003440205545537e-07,
      "loss": 0.3547,
      "step": 2728
    },
    {
      "epoch": 0.706743064322942,
      "grad_norm": 3.8161783035500845,
      "learning_rate": 2.998611916526476e-07,
      "loss": 0.3694,
      "step": 2729
    },
    {
      "epoch": 0.7070020394289599,
      "grad_norm": 2.5160745780524927,
      "learning_rate": 2.993786286449617e-07,
      "loss": 0.302,
      "step": 2730
    },
    {
      "epoch": 0.7070020394289599,
      "eval_PRM Accuracy": 0.8773584905660378,
      "eval_PRM F1": 0.9202453987730062,
      "eval_PRM F1 AUC": 0.8431115767417497,
      "eval_PRM F1 AUC (fixed)": 0.8226820324777371,
      "eval_PRM F1 Neg": 0.7346938775510204,
      "eval_PRM NPV": 0.6923076923076923,
      "eval_PRM Precision": 0.9375,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.22934657335281372,
      "eval_runtime": 15.1529,
      "eval_samples_per_second": 2.904,
      "eval_steps_per_second": 0.132,
      "step": 2730
    },
    {
      "epoch": 0.7072610145349778,
      "grad_norm": 3.158118765653248,
      "learning_rate": 2.988963319261299e-07,
      "loss": 0.3542,
      "step": 2731
    },
    {
      "epoch": 0.7075199896409957,
      "grad_norm": 3.3105632832525025,
      "learning_rate": 2.984143018905678e-07,
      "loss": 0.3566,
      "step": 2732
    },
    {
      "epoch": 0.7077789647470137,
      "grad_norm": 4.482093609388006,
      "learning_rate": 2.9793253893247344e-07,
      "loss": 0.3153,
      "step": 2733
    },
    {
      "epoch": 0.7080379398530317,
      "grad_norm": 4.027985131951971,
      "learning_rate": 2.9745104344582603e-07,
      "loss": 0.318,
      "step": 2734
    },
    {
      "epoch": 0.7082969149590496,
      "grad_norm": 3.045412988014404,
      "learning_rate": 2.969698158243865e-07,
      "loss": 0.2648,
      "step": 2735
    },
    {
      "epoch": 0.7082969149590496,
      "eval_PRM Accuracy": 0.8679245283018868,
      "eval_PRM F1": 0.9135802469135802,
      "eval_PRM F1 AUC": 0.8370874803562075,
      "eval_PRM F1 AUC (fixed)": 0.823729701414353,
      "eval_PRM F1 Neg": 0.72,
      "eval_PRM NPV": 0.6666666666666666,
      "eval_PRM Precision": 0.9367088607594937,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.23092445731163025,
      "eval_runtime": 14.4166,
      "eval_samples_per_second": 3.052,
      "eval_steps_per_second": 0.139,
      "step": 2735
    },
    {
      "epoch": 0.7085558900650675,
      "grad_norm": 4.729840215380708,
      "learning_rate": 2.9648885646169607e-07,
      "loss": 0.3491,
      "step": 2736
    },
    {
      "epoch": 0.7088148651710854,
      "grad_norm": 6.936160505583626,
      "learning_rate": 2.9600816575107674e-07,
      "loss": 0.331,
      "step": 2737
    },
    {
      "epoch": 0.7090738402771034,
      "grad_norm": 3.2098180090970687,
      "learning_rate": 2.9552774408563166e-07,
      "loss": 0.3156,
      "step": 2738
    },
    {
      "epoch": 0.7093328153831213,
      "grad_norm": 4.118951090253375,
      "learning_rate": 2.950475918582431e-07,
      "loss": 0.2407,
      "step": 2739
    },
    {
      "epoch": 0.7095917904891392,
      "grad_norm": 3.5530108796069992,
      "learning_rate": 2.94567709461573e-07,
      "loss": 0.3421,
      "step": 2740
    },
    {
      "epoch": 0.7095917904891392,
      "eval_PRM Accuracy": 0.8679245283018868,
      "eval_PRM F1": 0.9135802469135802,
      "eval_PRM F1 AUC": 0.8370874803562075,
      "eval_PRM F1 AUC (fixed)": 0.8226820324777371,
      "eval_PRM F1 Neg": 0.72,
      "eval_PRM NPV": 0.6666666666666666,
      "eval_PRM Precision": 0.9367088607594937,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.23179985582828522,
      "eval_runtime": 14.3399,
      "eval_samples_per_second": 3.068,
      "eval_steps_per_second": 0.139,
      "step": 2740
    },
    {
      "epoch": 0.7098507655951571,
      "grad_norm": 3.0784118776422598,
      "learning_rate": 2.940880972880631e-07,
      "loss": 0.3326,
      "step": 2741
    },
    {
      "epoch": 0.710109740701175,
      "grad_norm": 3.7881194905128157,
      "learning_rate": 2.9360875572993374e-07,
      "loss": 0.3259,
      "step": 2742
    },
    {
      "epoch": 0.7103687158071931,
      "grad_norm": 3.4924284636461835,
      "learning_rate": 2.9312968517918416e-07,
      "loss": 0.3317,
      "step": 2743
    },
    {
      "epoch": 0.710627690913211,
      "grad_norm": 4.051277057057838,
      "learning_rate": 2.926508860275921e-07,
      "loss": 0.3595,
      "step": 2744
    },
    {
      "epoch": 0.7108866660192289,
      "grad_norm": 4.780006911646844,
      "learning_rate": 2.921723586667128e-07,
      "loss": 0.3708,
      "step": 2745
    },
    {
      "epoch": 0.7108866660192289,
      "eval_PRM Accuracy": 0.8773584905660378,
      "eval_PRM F1": 0.9202453987730062,
      "eval_PRM F1 AUC": 0.8431115767417497,
      "eval_PRM F1 AUC (fixed)": 0.8255631220534312,
      "eval_PRM F1 Neg": 0.7346938775510204,
      "eval_PRM NPV": 0.6923076923076923,
      "eval_PRM Precision": 0.9375,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.2295445054769516,
      "eval_runtime": 14.4745,
      "eval_samples_per_second": 3.04,
      "eval_steps_per_second": 0.138,
      "step": 2745
    },
    {
      "epoch": 0.7111456411252468,
      "grad_norm": 2.584971996423779,
      "learning_rate": 2.9169410348787975e-07,
      "loss": 0.2996,
      "step": 2746
    },
    {
      "epoch": 0.7114046162312647,
      "grad_norm": 5.85951685371132,
      "learning_rate": 2.912161208822038e-07,
      "loss": 0.3522,
      "step": 2747
    },
    {
      "epoch": 0.7116635913372827,
      "grad_norm": 5.395198569573732,
      "learning_rate": 2.907384112405726e-07,
      "loss": 0.2624,
      "step": 2748
    },
    {
      "epoch": 0.7119225664433007,
      "grad_norm": 3.8402400510307824,
      "learning_rate": 2.9026097495365104e-07,
      "loss": 0.3155,
      "step": 2749
    },
    {
      "epoch": 0.7121815415493186,
      "grad_norm": 2.821957319350055,
      "learning_rate": 2.8978381241188025e-07,
      "loss": 0.2703,
      "step": 2750
    },
    {
      "epoch": 0.7121815415493186,
      "eval_PRM Accuracy": 0.8679245283018868,
      "eval_PRM F1": 0.9135802469135802,
      "eval_PRM F1 AUC": 0.8370874803562075,
      "eval_PRM F1 AUC (fixed)": 0.8263488737558932,
      "eval_PRM F1 Neg": 0.72,
      "eval_PRM NPV": 0.6666666666666666,
      "eval_PRM Precision": 0.9367088607594937,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.2308908849954605,
      "eval_runtime": 14.5289,
      "eval_samples_per_second": 3.028,
      "eval_steps_per_second": 0.138,
      "step": 2750
    },
    {
      "epoch": 0.7124405166553365,
      "grad_norm": 5.624149809250447,
      "learning_rate": 2.893069240054773e-07,
      "loss": 0.2951,
      "step": 2751
    },
    {
      "epoch": 0.7126994917613544,
      "grad_norm": 3.4450269041553216,
      "learning_rate": 2.8883031012443553e-07,
      "loss": 0.2888,
      "step": 2752
    },
    {
      "epoch": 0.7129584668673724,
      "grad_norm": 7.635417524089154,
      "learning_rate": 2.8835397115852287e-07,
      "loss": 0.2882,
      "step": 2753
    },
    {
      "epoch": 0.7132174419733903,
      "grad_norm": 3.6687557108001454,
      "learning_rate": 2.878779074972834e-07,
      "loss": 0.2186,
      "step": 2754
    },
    {
      "epoch": 0.7134764170794082,
      "grad_norm": 4.666247797930841,
      "learning_rate": 2.8740211953003563e-07,
      "loss": 0.3213,
      "step": 2755
    },
    {
      "epoch": 0.7134764170794082,
      "eval_PRM Accuracy": 0.8679245283018868,
      "eval_PRM F1": 0.9135802469135802,
      "eval_PRM F1 AUC": 0.8370874803562075,
      "eval_PRM F1 AUC (fixed)": 0.8266107909900471,
      "eval_PRM F1 Neg": 0.72,
      "eval_PRM NPV": 0.6666666666666666,
      "eval_PRM Precision": 0.9367088607594937,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.22844244539737701,
      "eval_runtime": 14.9042,
      "eval_samples_per_second": 2.952,
      "eval_steps_per_second": 0.134,
      "step": 2755
    },
    {
      "epoch": 0.7137353921854261,
      "grad_norm": 3.132343943450176,
      "learning_rate": 2.8692660764587264e-07,
      "loss": 0.2921,
      "step": 2756
    },
    {
      "epoch": 0.7139943672914442,
      "grad_norm": 4.647140953339821,
      "learning_rate": 2.8645137223366185e-07,
      "loss": 0.3867,
      "step": 2757
    },
    {
      "epoch": 0.7142533423974621,
      "grad_norm": 4.601418043371969,
      "learning_rate": 2.859764136820442e-07,
      "loss": 0.2722,
      "step": 2758
    },
    {
      "epoch": 0.71451231750348,
      "grad_norm": 3.338615609866598,
      "learning_rate": 2.8550173237943507e-07,
      "loss": 0.3133,
      "step": 2759
    },
    {
      "epoch": 0.7147712926094979,
      "grad_norm": 3.3028123809758534,
      "learning_rate": 2.850273287140224e-07,
      "loss": 0.3316,
      "step": 2760
    },
    {
      "epoch": 0.7147712926094979,
      "eval_PRM Accuracy": 0.8773584905660378,
      "eval_PRM F1": 0.9202453987730062,
      "eval_PRM F1 AUC": 0.8431115767417497,
      "eval_PRM F1 AUC (fixed)": 0.8284442116291252,
      "eval_PRM F1 Neg": 0.7346938775510204,
      "eval_PRM NPV": 0.6923076923076923,
      "eval_PRM Precision": 0.9375,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.22737134993076324,
      "eval_runtime": 14.5273,
      "eval_samples_per_second": 3.029,
      "eval_steps_per_second": 0.138,
      "step": 2760
    },
    {
      "epoch": 0.7150302677155158,
      "grad_norm": 9.908498377530197,
      "learning_rate": 2.8455320307376695e-07,
      "loss": 0.3465,
      "step": 2761
    },
    {
      "epoch": 0.7152892428215338,
      "grad_norm": 5.522385598323615,
      "learning_rate": 2.840793558464025e-07,
      "loss": 0.3236,
      "step": 2762
    },
    {
      "epoch": 0.7155482179275517,
      "grad_norm": 3.439456747677689,
      "learning_rate": 2.8360578741943515e-07,
      "loss": 0.283,
      "step": 2763
    },
    {
      "epoch": 0.7158071930335697,
      "grad_norm": 4.936296251923971,
      "learning_rate": 2.8313249818014276e-07,
      "loss": 0.3567,
      "step": 2764
    },
    {
      "epoch": 0.7160661681395876,
      "grad_norm": 4.953758145236708,
      "learning_rate": 2.8265948851557515e-07,
      "loss": 0.4587,
      "step": 2765
    },
    {
      "epoch": 0.7160661681395876,
      "eval_PRM Accuracy": 0.8773584905660378,
      "eval_PRM F1": 0.9202453987730062,
      "eval_PRM F1 AUC": 0.8431115767417497,
      "eval_PRM F1 AUC (fixed)": 0.8328968046097434,
      "eval_PRM F1 Neg": 0.7346938775510204,
      "eval_PRM NPV": 0.6923076923076923,
      "eval_PRM Precision": 0.9375,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.22658222913742065,
      "eval_runtime": 14.7778,
      "eval_samples_per_second": 2.977,
      "eval_steps_per_second": 0.135,
      "step": 2765
    },
    {
      "epoch": 0.7163251432456055,
      "grad_norm": 2.9529735123891143,
      "learning_rate": 2.821867588125533e-07,
      "loss": 0.2637,
      "step": 2766
    },
    {
      "epoch": 0.7165841183516235,
      "grad_norm": 4.057901860418548,
      "learning_rate": 2.8171430945766944e-07,
      "loss": 0.4151,
      "step": 2767
    },
    {
      "epoch": 0.7168430934576414,
      "grad_norm": 3.785851483786009,
      "learning_rate": 2.8124214083728596e-07,
      "loss": 0.3077,
      "step": 2768
    },
    {
      "epoch": 0.7171020685636593,
      "grad_norm": 5.445833489708636,
      "learning_rate": 2.80770253337537e-07,
      "loss": 0.3616,
      "step": 2769
    },
    {
      "epoch": 0.7173610436696772,
      "grad_norm": 4.097754761836587,
      "learning_rate": 2.802986473443251e-07,
      "loss": 0.3259,
      "step": 2770
    },
    {
      "epoch": 0.7173610436696772,
      "eval_PRM Accuracy": 0.8773584905660378,
      "eval_PRM F1": 0.9202453987730062,
      "eval_PRM F1 AUC": 0.8431115767417497,
      "eval_PRM F1 AUC (fixed)": 0.8308014667365113,
      "eval_PRM F1 Neg": 0.7346938775510204,
      "eval_PRM NPV": 0.6923076923076923,
      "eval_PRM Precision": 0.9375,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.22802509367465973,
      "eval_runtime": 14.5813,
      "eval_samples_per_second": 3.018,
      "eval_steps_per_second": 0.137,
      "step": 2770
    },
    {
      "epoch": 0.7176200187756951,
      "grad_norm": 5.275176974728195,
      "learning_rate": 2.7982732324332373e-07,
      "loss": 0.3577,
      "step": 2771
    },
    {
      "epoch": 0.7178789938817132,
      "grad_norm": 11.229465162254435,
      "learning_rate": 2.7935628141997544e-07,
      "loss": 0.3865,
      "step": 2772
    },
    {
      "epoch": 0.7181379689877311,
      "grad_norm": 6.494821954803709,
      "learning_rate": 2.788855222594921e-07,
      "loss": 0.3819,
      "step": 2773
    },
    {
      "epoch": 0.718396944093749,
      "grad_norm": 3.912896879603631,
      "learning_rate": 2.7841504614685423e-07,
      "loss": 0.3245,
      "step": 2774
    },
    {
      "epoch": 0.7186559191997669,
      "grad_norm": 3.951233738530968,
      "learning_rate": 2.779448534668111e-07,
      "loss": 0.2909,
      "step": 2775
    },
    {
      "epoch": 0.7186559191997669,
      "eval_PRM Accuracy": 0.8773584905660378,
      "eval_PRM F1": 0.9202453987730062,
      "eval_PRM F1 AUC": 0.8431115767417497,
      "eval_PRM F1 AUC (fixed)": 0.8321110529072814,
      "eval_PRM F1 Neg": 0.7346938775510204,
      "eval_PRM NPV": 0.6923076923076923,
      "eval_PRM Precision": 0.9375,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.22936099767684937,
      "eval_runtime": 14.7318,
      "eval_samples_per_second": 2.987,
      "eval_steps_per_second": 0.136,
      "step": 2775
    },
    {
      "epoch": 0.7189148943057848,
      "grad_norm": 6.623712131546646,
      "learning_rate": 2.774749446038799e-07,
      "loss": 0.3456,
      "step": 2776
    },
    {
      "epoch": 0.7191738694118028,
      "grad_norm": 3.889194255164197,
      "learning_rate": 2.77005319942346e-07,
      "loss": 0.3106,
      "step": 2777
    },
    {
      "epoch": 0.7194328445178207,
      "grad_norm": 2.9219410898222633,
      "learning_rate": 2.765359798662622e-07,
      "loss": 0.3283,
      "step": 2778
    },
    {
      "epoch": 0.7196918196238387,
      "grad_norm": 3.030467581909445,
      "learning_rate": 2.7606692475944865e-07,
      "loss": 0.2734,
      "step": 2779
    },
    {
      "epoch": 0.7199507947298566,
      "grad_norm": 4.1590703022570645,
      "learning_rate": 2.7559815500549246e-07,
      "loss": 0.3172,
      "step": 2780
    },
    {
      "epoch": 0.7199507947298566,
      "eval_PRM Accuracy": 0.8773584905660378,
      "eval_PRM F1": 0.9202453987730062,
      "eval_PRM F1 AUC": 0.8431115767417497,
      "eval_PRM F1 AUC (fixed)": 0.8310633839706653,
      "eval_PRM F1 Neg": 0.7346938775510204,
      "eval_PRM NPV": 0.6923076923076923,
      "eval_PRM Precision": 0.9375,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.23109498620033264,
      "eval_runtime": 14.6441,
      "eval_samples_per_second": 3.005,
      "eval_steps_per_second": 0.137,
      "step": 2780
    },
    {
      "epoch": 0.7202097698358745,
      "grad_norm": 3.573057481785047,
      "learning_rate": 2.7512967098774717e-07,
      "loss": 0.3411,
      "step": 2781
    },
    {
      "epoch": 0.7204687449418925,
      "grad_norm": 3.456032487465389,
      "learning_rate": 2.7466147308933296e-07,
      "loss": 0.3582,
      "step": 2782
    },
    {
      "epoch": 0.7207277200479104,
      "grad_norm": 3.9893286771009078,
      "learning_rate": 2.7419356169313584e-07,
      "loss": 0.3438,
      "step": 2783
    },
    {
      "epoch": 0.7209866951539283,
      "grad_norm": 5.0072247059047665,
      "learning_rate": 2.737259371818076e-07,
      "loss": 0.2828,
      "step": 2784
    },
    {
      "epoch": 0.7212456702599462,
      "grad_norm": 2.6719707747525825,
      "learning_rate": 2.732585999377656e-07,
      "loss": 0.2914,
      "step": 2785
    },
    {
      "epoch": 0.7212456702599462,
      "eval_PRM Accuracy": 0.8773584905660378,
      "eval_PRM F1": 0.9202453987730062,
      "eval_PRM F1 AUC": 0.8431115767417497,
      "eval_PRM F1 AUC (fixed)": 0.8315872184389732,
      "eval_PRM F1 Neg": 0.7346938775510204,
      "eval_PRM NPV": 0.6923076923076923,
      "eval_PRM Precision": 0.9375,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.2312084138393402,
      "eval_runtime": 14.6489,
      "eval_samples_per_second": 3.004,
      "eval_steps_per_second": 0.137,
      "step": 2785
    },
    {
      "epoch": 0.7215046453659641,
      "grad_norm": 5.7040390664297105,
      "learning_rate": 2.7279155034319134e-07,
      "loss": 0.3172,
      "step": 2786
    },
    {
      "epoch": 0.7217636204719822,
      "grad_norm": 4.6038920245183155,
      "learning_rate": 2.723247887800325e-07,
      "loss": 0.327,
      "step": 2787
    },
    {
      "epoch": 0.7220225955780001,
      "grad_norm": 3.5157463961755617,
      "learning_rate": 2.718583156300003e-07,
      "loss": 0.2859,
      "step": 2788
    },
    {
      "epoch": 0.722281570684018,
      "grad_norm": 3.553675368447141,
      "learning_rate": 2.7139213127457033e-07,
      "loss": 0.2771,
      "step": 2789
    },
    {
      "epoch": 0.7225405457900359,
      "grad_norm": 3.9916755414477696,
      "learning_rate": 2.70926236094982e-07,
      "loss": 0.3123,
      "step": 2790
    },
    {
      "epoch": 0.7225405457900359,
      "eval_PRM Accuracy": 0.8679245283018868,
      "eval_PRM F1": 0.9135802469135802,
      "eval_PRM F1 AUC": 0.8370874803562075,
      "eval_PRM F1 AUC (fixed)": 0.8334206390780514,
      "eval_PRM F1 Neg": 0.72,
      "eval_PRM NPV": 0.6666666666666666,
      "eval_PRM Precision": 0.9367088607594937,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.23033572733402252,
      "eval_runtime": 14.7701,
      "eval_samples_per_second": 2.979,
      "eval_steps_per_second": 0.135,
      "step": 2790
    },
    {
      "epoch": 0.7227995208960538,
      "grad_norm": 3.91102909018941,
      "learning_rate": 2.704606304722382e-07,
      "loss": 0.3586,
      "step": 2791
    },
    {
      "epoch": 0.7230584960020718,
      "grad_norm": 3.3226282395848066,
      "learning_rate": 2.6999531478710505e-07,
      "loss": 0.3559,
      "step": 2792
    },
    {
      "epoch": 0.7233174711080897,
      "grad_norm": 3.364348136411542,
      "learning_rate": 2.695302894201116e-07,
      "loss": 0.3229,
      "step": 2793
    },
    {
      "epoch": 0.7235764462141077,
      "grad_norm": 4.283864103351954,
      "learning_rate": 2.690655547515497e-07,
      "loss": 0.399,
      "step": 2794
    },
    {
      "epoch": 0.7238354213201256,
      "grad_norm": 4.258464070427692,
      "learning_rate": 2.686011111614727e-07,
      "loss": 0.3306,
      "step": 2795
    },
    {
      "epoch": 0.7238354213201256,
      "eval_PRM Accuracy": 0.8679245283018868,
      "eval_PRM F1": 0.9135802469135802,
      "eval_PRM F1 AUC": 0.8370874803562075,
      "eval_PRM F1 AUC (fixed)": 0.8313253012048193,
      "eval_PRM F1 Neg": 0.72,
      "eval_PRM NPV": 0.6666666666666666,
      "eval_PRM Precision": 0.9367088607594937,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.23094840347766876,
      "eval_runtime": 14.5131,
      "eval_samples_per_second": 3.032,
      "eval_steps_per_second": 0.138,
      "step": 2795
    },
    {
      "epoch": 0.7240943964261435,
      "grad_norm": 3.5225838249851256,
      "learning_rate": 2.6813695902969644e-07,
      "loss": 0.2844,
      "step": 2796
    },
    {
      "epoch": 0.7243533715321615,
      "grad_norm": 2.6752120766850407,
      "learning_rate": 2.6767309873579897e-07,
      "loss": 0.2415,
      "step": 2797
    },
    {
      "epoch": 0.7246123466381794,
      "grad_norm": 4.03324436102591,
      "learning_rate": 2.6720953065911865e-07,
      "loss": 0.3395,
      "step": 2798
    },
    {
      "epoch": 0.7248713217441973,
      "grad_norm": 6.22211326508483,
      "learning_rate": 2.667462551787555e-07,
      "loss": 0.3968,
      "step": 2799
    },
    {
      "epoch": 0.7251302968502152,
      "grad_norm": 2.2852162270288168,
      "learning_rate": 2.6628327267357006e-07,
      "loss": 0.2358,
      "step": 2800
    },
    {
      "epoch": 0.7251302968502152,
      "eval_PRM Accuracy": 0.8773584905660378,
      "eval_PRM F1": 0.9202453987730062,
      "eval_PRM F1 AUC": 0.8431115767417497,
      "eval_PRM F1 AUC (fixed)": 0.8355159769512834,
      "eval_PRM F1 Neg": 0.7346938775510204,
      "eval_PRM NPV": 0.6923076923076923,
      "eval_PRM Precision": 0.9375,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.22936736047267914,
      "eval_runtime": 15.0082,
      "eval_samples_per_second": 2.932,
      "eval_steps_per_second": 0.133,
      "step": 2800
    },
    {
      "epoch": 0.7253892719562333,
      "grad_norm": 4.168228880207895,
      "learning_rate": 2.658205835221833e-07,
      "loss": 0.4486,
      "step": 2801
    },
    {
      "epoch": 0.7256482470622512,
      "grad_norm": 3.890461175616314,
      "learning_rate": 2.653581881029765e-07,
      "loss": 0.3627,
      "step": 2802
    },
    {
      "epoch": 0.7259072221682691,
      "grad_norm": 3.217976593816656,
      "learning_rate": 2.648960867940901e-07,
      "loss": 0.3476,
      "step": 2803
    },
    {
      "epoch": 0.726166197274287,
      "grad_norm": 2.5852822223785514,
      "learning_rate": 2.644342799734246e-07,
      "loss": 0.2718,
      "step": 2804
    },
    {
      "epoch": 0.7264251723803049,
      "grad_norm": 5.459196372535445,
      "learning_rate": 2.639727680186396e-07,
      "loss": 0.3099,
      "step": 2805
    },
    {
      "epoch": 0.7264251723803049,
      "eval_PRM Accuracy": 0.8773584905660378,
      "eval_PRM F1": 0.9202453987730062,
      "eval_PRM F1 AUC": 0.8431115767417497,
      "eval_PRM F1 AUC (fixed)": 0.8334206390780514,
      "eval_PRM F1 Neg": 0.7346938775510204,
      "eval_PRM NPV": 0.6923076923076923,
      "eval_PRM Precision": 0.9375,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.22806701064109802,
      "eval_runtime": 14.7998,
      "eval_samples_per_second": 2.973,
      "eval_steps_per_second": 0.135,
      "step": 2805
    },
    {
      "epoch": 0.7266841474863229,
      "grad_norm": 2.94972182205354,
      "learning_rate": 2.635115513071534e-07,
      "loss": 0.3107,
      "step": 2806
    },
    {
      "epoch": 0.7269431225923408,
      "grad_norm": 4.603103796868197,
      "learning_rate": 2.630506302161427e-07,
      "loss": 0.3042,
      "step": 2807
    },
    {
      "epoch": 0.7272020976983588,
      "grad_norm": 4.766639159438619,
      "learning_rate": 2.625900051225433e-07,
      "loss": 0.3144,
      "step": 2808
    },
    {
      "epoch": 0.7274610728043767,
      "grad_norm": 4.538074429319457,
      "learning_rate": 2.621296764030479e-07,
      "loss": 0.2841,
      "step": 2809
    },
    {
      "epoch": 0.7277200479103946,
      "grad_norm": 4.8053223645379335,
      "learning_rate": 2.616696444341075e-07,
      "loss": 0.2632,
      "step": 2810
    },
    {
      "epoch": 0.7277200479103946,
      "eval_PRM Accuracy": 0.8773584905660378,
      "eval_PRM F1": 0.9202453987730062,
      "eval_PRM F1 AUC": 0.8431115767417497,
      "eval_PRM F1 AUC (fixed)": 0.8328968046097434,
      "eval_PRM F1 Neg": 0.7346938775510204,
      "eval_PRM NPV": 0.6923076923076923,
      "eval_PRM Precision": 0.9375,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.22770456969738007,
      "eval_runtime": 14.453,
      "eval_samples_per_second": 3.044,
      "eval_steps_per_second": 0.138,
      "step": 2810
    },
    {
      "epoch": 0.7279790230164126,
      "grad_norm": 5.490776272600583,
      "learning_rate": 2.6120990959192974e-07,
      "loss": 0.4148,
      "step": 2811
    },
    {
      "epoch": 0.7282379981224305,
      "grad_norm": 3.0407375864649953,
      "learning_rate": 2.6075047225247995e-07,
      "loss": 0.3022,
      "step": 2812
    },
    {
      "epoch": 0.7284969732284484,
      "grad_norm": 4.110975338562775,
      "learning_rate": 2.602913327914799e-07,
      "loss": 0.3631,
      "step": 2813
    },
    {
      "epoch": 0.7287559483344663,
      "grad_norm": 3.5152651544613622,
      "learning_rate": 2.598324915844077e-07,
      "loss": 0.3514,
      "step": 2814
    },
    {
      "epoch": 0.7290149234404842,
      "grad_norm": 3.1665352503455946,
      "learning_rate": 2.5937394900649775e-07,
      "loss": 0.2798,
      "step": 2815
    },
    {
      "epoch": 0.7290149234404842,
      "eval_PRM Accuracy": 0.8773584905660378,
      "eval_PRM F1": 0.9202453987730062,
      "eval_PRM F1 AUC": 0.8431115767417497,
      "eval_PRM F1 AUC (fixed)": 0.8347302252488213,
      "eval_PRM F1 Neg": 0.7346938775510204,
      "eval_PRM NPV": 0.6923076923076923,
      "eval_PRM Precision": 0.9375,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.22800016403198242,
      "eval_runtime": 14.7622,
      "eval_samples_per_second": 2.981,
      "eval_steps_per_second": 0.135,
      "step": 2815
    },
    {
      "epoch": 0.7292738985465023,
      "grad_norm": 9.843543594820751,
      "learning_rate": 2.5891570543274e-07,
      "loss": 0.376,
      "step": 2816
    },
    {
      "epoch": 0.7295328736525202,
      "grad_norm": 10.362653930430206,
      "learning_rate": 2.584577612378799e-07,
      "loss": 0.5034,
      "step": 2817
    },
    {
      "epoch": 0.7297918487585381,
      "grad_norm": 3.3730893404789652,
      "learning_rate": 2.5800011679641866e-07,
      "loss": 0.2669,
      "step": 2818
    },
    {
      "epoch": 0.730050823864556,
      "grad_norm": 2.9556105849030176,
      "learning_rate": 2.575427724826113e-07,
      "loss": 0.2606,
      "step": 2819
    },
    {
      "epoch": 0.7303097989705739,
      "grad_norm": 2.9628751736549557,
      "learning_rate": 2.570857286704681e-07,
      "loss": 0.273,
      "step": 2820
    },
    {
      "epoch": 0.7303097989705739,
      "eval_PRM Accuracy": 0.8773584905660378,
      "eval_PRM F1": 0.9202453987730062,
      "eval_PRM F1 AUC": 0.8431115767417497,
      "eval_PRM F1 AUC (fixed)": 0.8363017286537454,
      "eval_PRM F1 Neg": 0.7346938775510204,
      "eval_PRM NPV": 0.6923076923076923,
      "eval_PRM Precision": 0.9375,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.2261267602443695,
      "eval_runtime": 15.0276,
      "eval_samples_per_second": 2.928,
      "eval_steps_per_second": 0.133,
      "step": 2820
    },
    {
      "epoch": 0.7305687740765919,
      "grad_norm": 3.7223882266259376,
      "learning_rate": 2.5662898573375353e-07,
      "loss": 0.341,
      "step": 2821
    },
    {
      "epoch": 0.7308277491826098,
      "grad_norm": 3.070170007338791,
      "learning_rate": 2.561725440459859e-07,
      "loss": 0.3269,
      "step": 2822
    },
    {
      "epoch": 0.7310867242886278,
      "grad_norm": 3.235670720245644,
      "learning_rate": 2.557164039804372e-07,
      "loss": 0.2742,
      "step": 2823
    },
    {
      "epoch": 0.7313456993946457,
      "grad_norm": 4.680980977414096,
      "learning_rate": 2.552605659101327e-07,
      "loss": 0.3334,
      "step": 2824
    },
    {
      "epoch": 0.7316046745006636,
      "grad_norm": 8.255172708569308,
      "learning_rate": 2.548050302078508e-07,
      "loss": 0.3104,
      "step": 2825
    },
    {
      "epoch": 0.7316046745006636,
      "eval_PRM Accuracy": 0.8773584905660378,
      "eval_PRM F1": 0.9202453987730062,
      "eval_PRM F1 AUC": 0.8431115767417497,
      "eval_PRM F1 AUC (fixed)": 0.8357778941854375,
      "eval_PRM F1 Neg": 0.7346938775510204,
      "eval_PRM NPV": 0.6923076923076923,
      "eval_PRM Precision": 0.9375,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.22597144544124603,
      "eval_runtime": 14.6794,
      "eval_samples_per_second": 2.997,
      "eval_steps_per_second": 0.136,
      "step": 2825
    },
    {
      "epoch": 0.7318636496066816,
      "grad_norm": 6.001310537609272,
      "learning_rate": 2.543497972461225e-07,
      "loss": 0.268,
      "step": 2826
    },
    {
      "epoch": 0.7321226247126995,
      "grad_norm": 3.966471198680594,
      "learning_rate": 2.5389486739723143e-07,
      "loss": 0.3831,
      "step": 2827
    },
    {
      "epoch": 0.7323815998187174,
      "grad_norm": 2.8498036800417066,
      "learning_rate": 2.5344024103321306e-07,
      "loss": 0.3553,
      "step": 2828
    },
    {
      "epoch": 0.7326405749247353,
      "grad_norm": 3.610811079029175,
      "learning_rate": 2.5298591852585495e-07,
      "loss": 0.3234,
      "step": 2829
    },
    {
      "epoch": 0.7328995500307532,
      "grad_norm": 2.7634222189939197,
      "learning_rate": 2.5253190024669597e-07,
      "loss": 0.2524,
      "step": 2830
    },
    {
      "epoch": 0.7328995500307532,
      "eval_PRM Accuracy": 0.8679245283018868,
      "eval_PRM F1": 0.9146341463414634,
      "eval_PRM F1 AUC": 0.8213724463069669,
      "eval_PRM F1 AUC (fixed)": 0.8347302252488213,
      "eval_PRM F1 Neg": 0.7083333333333334,
      "eval_PRM NPV": 0.68,
      "eval_PRM Precision": 0.9259259259259259,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.22537946701049805,
      "eval_runtime": 14.8945,
      "eval_samples_per_second": 2.954,
      "eval_steps_per_second": 0.134,
      "step": 2830
    },
    {
      "epoch": 0.7331585251367713,
      "grad_norm": 2.7948727244636267,
      "learning_rate": 2.5207818656702625e-07,
      "loss": 0.2377,
      "step": 2831
    },
    {
      "epoch": 0.7334175002427892,
      "grad_norm": 5.57452702810174,
      "learning_rate": 2.5162477785788687e-07,
      "loss": 0.303,
      "step": 2832
    },
    {
      "epoch": 0.7336764753488071,
      "grad_norm": 4.135884981978197,
      "learning_rate": 2.511716744900695e-07,
      "loss": 0.3628,
      "step": 2833
    },
    {
      "epoch": 0.733935450454825,
      "grad_norm": 4.1584245313484685,
      "learning_rate": 2.5071887683411613e-07,
      "loss": 0.3269,
      "step": 2834
    },
    {
      "epoch": 0.7341944255608429,
      "grad_norm": 3.0628872651493,
      "learning_rate": 2.5026638526031863e-07,
      "loss": 0.3164,
      "step": 2835
    },
    {
      "epoch": 0.7341944255608429,
      "eval_PRM Accuracy": 0.8679245283018868,
      "eval_PRM F1": 0.9146341463414634,
      "eval_PRM F1 AUC": 0.8213724463069669,
      "eval_PRM F1 AUC (fixed)": 0.8365636458878996,
      "eval_PRM F1 Neg": 0.7083333333333334,
      "eval_PRM NPV": 0.68,
      "eval_PRM Precision": 0.9259259259259259,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.22656245529651642,
      "eval_runtime": 14.9209,
      "eval_samples_per_second": 2.949,
      "eval_steps_per_second": 0.134,
      "step": 2835
    },
    {
      "epoch": 0.7344534006668609,
      "grad_norm": 6.9219015788227765,
      "learning_rate": 2.498142001387186e-07,
      "loss": 0.3157,
      "step": 2836
    },
    {
      "epoch": 0.7347123757728788,
      "grad_norm": 8.78552678816298,
      "learning_rate": 2.493623218391071e-07,
      "loss": 0.408,
      "step": 2837
    },
    {
      "epoch": 0.7349713508788968,
      "grad_norm": 7.812991477692669,
      "learning_rate": 2.4891075073102426e-07,
      "loss": 0.3717,
      "step": 2838
    },
    {
      "epoch": 0.7352303259849147,
      "grad_norm": 7.745557387083495,
      "learning_rate": 2.4845948718375886e-07,
      "loss": 0.3318,
      "step": 2839
    },
    {
      "epoch": 0.7354893010909326,
      "grad_norm": 5.344390075772926,
      "learning_rate": 2.4800853156634844e-07,
      "loss": 0.3917,
      "step": 2840
    },
    {
      "epoch": 0.7354893010909326,
      "eval_PRM Accuracy": 0.8773584905660378,
      "eval_PRM F1": 0.9202453987730062,
      "eval_PRM F1 AUC": 0.8431115767417497,
      "eval_PRM F1 AUC (fixed)": 0.8389209009952855,
      "eval_PRM F1 Neg": 0.7346938775510204,
      "eval_PRM NPV": 0.6923076923076923,
      "eval_PRM Precision": 0.9375,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.22631524503231049,
      "eval_runtime": 14.8274,
      "eval_samples_per_second": 2.967,
      "eval_steps_per_second": 0.135,
      "step": 2840
    },
    {
      "epoch": 0.7357482761969506,
      "grad_norm": 3.0819239127116678,
      "learning_rate": 2.4755788424757834e-07,
      "loss": 0.2343,
      "step": 2841
    },
    {
      "epoch": 0.7360072513029685,
      "grad_norm": 3.219107464320179,
      "learning_rate": 2.471075455959822e-07,
      "loss": 0.273,
      "step": 2842
    },
    {
      "epoch": 0.7362662264089864,
      "grad_norm": 3.5056140065058354,
      "learning_rate": 2.46657515979841e-07,
      "loss": 0.3951,
      "step": 2843
    },
    {
      "epoch": 0.7365252015150043,
      "grad_norm": 4.0445278596614465,
      "learning_rate": 2.4620779576718276e-07,
      "loss": 0.3452,
      "step": 2844
    },
    {
      "epoch": 0.7367841766210224,
      "grad_norm": 3.08839454822352,
      "learning_rate": 2.4575838532578255e-07,
      "loss": 0.2894,
      "step": 2845
    },
    {
      "epoch": 0.7367841766210224,
      "eval_PRM Accuracy": 0.8679245283018868,
      "eval_PRM F1": 0.9135802469135802,
      "eval_PRM F1 AUC": 0.8370874803562075,
      "eval_PRM F1 AUC (fixed)": 0.8355159769512833,
      "eval_PRM F1 Neg": 0.72,
      "eval_PRM NPV": 0.6666666666666666,
      "eval_PRM Precision": 0.9367088607594937,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.22941647469997406,
      "eval_runtime": 14.734,
      "eval_samples_per_second": 2.986,
      "eval_steps_per_second": 0.136,
      "step": 2845
    },
    {
      "epoch": 0.7370431517270403,
      "grad_norm": 4.547940040829685,
      "learning_rate": 2.4530928502316293e-07,
      "loss": 0.2846,
      "step": 2846
    },
    {
      "epoch": 0.7373021268330582,
      "grad_norm": 5.518643615637911,
      "learning_rate": 2.4486049522659193e-07,
      "loss": 0.3189,
      "step": 2847
    },
    {
      "epoch": 0.7375611019390761,
      "grad_norm": 3.9317294607853572,
      "learning_rate": 2.4441201630308375e-07,
      "loss": 0.3838,
      "step": 2848
    },
    {
      "epoch": 0.737820077045094,
      "grad_norm": 4.665553492156433,
      "learning_rate": 2.4396384861939867e-07,
      "loss": 0.3552,
      "step": 2849
    },
    {
      "epoch": 0.738079052151112,
      "grad_norm": 6.176293846974459,
      "learning_rate": 2.435159925420423e-07,
      "loss": 0.3442,
      "step": 2850
    },
    {
      "epoch": 0.738079052151112,
      "eval_PRM Accuracy": 0.8679245283018868,
      "eval_PRM F1": 0.9135802469135802,
      "eval_PRM F1 AUC": 0.8370874803562075,
      "eval_PRM F1 AUC (fixed)": 0.8331587218438973,
      "eval_PRM F1 Neg": 0.72,
      "eval_PRM NPV": 0.6666666666666666,
      "eval_PRM Precision": 0.9367088607594937,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.22980856895446777,
      "eval_runtime": 14.4603,
      "eval_samples_per_second": 3.043,
      "eval_steps_per_second": 0.138,
      "step": 2850
    },
    {
      "epoch": 0.7383380272571299,
      "grad_norm": 3.780760926165394,
      "learning_rate": 2.4306844843726535e-07,
      "loss": 0.3461,
      "step": 2851
    },
    {
      "epoch": 0.7385970023631478,
      "grad_norm": 3.7347945086353436,
      "learning_rate": 2.4262121667106376e-07,
      "loss": 0.3125,
      "step": 2852
    },
    {
      "epoch": 0.7388559774691658,
      "grad_norm": 4.892939683262253,
      "learning_rate": 2.4217429760917723e-07,
      "loss": 0.2818,
      "step": 2853
    },
    {
      "epoch": 0.7391149525751837,
      "grad_norm": 3.361601102366713,
      "learning_rate": 2.4172769161709063e-07,
      "loss": 0.3219,
      "step": 2854
    },
    {
      "epoch": 0.7393739276812017,
      "grad_norm": 5.196781137830133,
      "learning_rate": 2.4128139906003234e-07,
      "loss": 0.3155,
      "step": 2855
    },
    {
      "epoch": 0.7393739276812017,
      "eval_PRM Accuracy": 0.8584905660377359,
      "eval_PRM F1": 0.906832298136646,
      "eval_PRM F1 AUC": 0.8310633839706653,
      "eval_PRM F1 AUC (fixed)": 0.8318491356731272,
      "eval_PRM F1 Neg": 0.7058823529411765,
      "eval_PRM NPV": 0.6428571428571429,
      "eval_PRM Precision": 0.9358974358974359,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.23152877390384674,
      "eval_runtime": 14.5735,
      "eval_samples_per_second": 3.019,
      "eval_steps_per_second": 0.137,
      "step": 2855
    },
    {
      "epoch": 0.7396329027872196,
      "grad_norm": 3.7648538724264835,
      "learning_rate": 2.4083542030297425e-07,
      "loss": 0.3559,
      "step": 2856
    },
    {
      "epoch": 0.7398918778932375,
      "grad_norm": 3.4171251619846297,
      "learning_rate": 2.4038975571063243e-07,
      "loss": 0.327,
      "step": 2857
    },
    {
      "epoch": 0.7401508529992554,
      "grad_norm": 5.810335856418894,
      "learning_rate": 2.399444056474652e-07,
      "loss": 0.3546,
      "step": 2858
    },
    {
      "epoch": 0.7404098281052733,
      "grad_norm": 6.1661292801864604,
      "learning_rate": 2.39499370477674e-07,
      "loss": 0.354,
      "step": 2859
    },
    {
      "epoch": 0.7406688032112914,
      "grad_norm": 3.5806461401513494,
      "learning_rate": 2.390546505652029e-07,
      "loss": 0.3458,
      "step": 2860
    },
    {
      "epoch": 0.7406688032112914,
      "eval_PRM Accuracy": 0.8584905660377359,
      "eval_PRM F1": 0.906832298136646,
      "eval_PRM F1 AUC": 0.8310633839706653,
      "eval_PRM F1 AUC (fixed)": 0.8305395495023573,
      "eval_PRM F1 Neg": 0.7058823529411765,
      "eval_PRM NPV": 0.6428571428571429,
      "eval_PRM Precision": 0.9358974358974359,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.23339098691940308,
      "eval_runtime": 14.5476,
      "eval_samples_per_second": 3.025,
      "eval_steps_per_second": 0.137,
      "step": 2860
    },
    {
      "epoch": 0.7409277783173093,
      "grad_norm": 3.9760606187044254,
      "learning_rate": 2.386102462737376e-07,
      "loss": 0.4107,
      "step": 2861
    },
    {
      "epoch": 0.7411867534233272,
      "grad_norm": 3.6274336098691884,
      "learning_rate": 2.3816615796670613e-07,
      "loss": 0.3082,
      "step": 2862
    },
    {
      "epoch": 0.7414457285293451,
      "grad_norm": 2.8369560996151617,
      "learning_rate": 2.3772238600727817e-07,
      "loss": 0.2965,
      "step": 2863
    },
    {
      "epoch": 0.741704703635363,
      "grad_norm": 3.6440608267224355,
      "learning_rate": 2.3727893075836456e-07,
      "loss": 0.3783,
      "step": 2864
    },
    {
      "epoch": 0.741963678741381,
      "grad_norm": 4.19328031104658,
      "learning_rate": 2.3683579258261708e-07,
      "loss": 0.2657,
      "step": 2865
    },
    {
      "epoch": 0.741963678741381,
      "eval_PRM Accuracy": 0.8584905660377359,
      "eval_PRM F1": 0.906832298136646,
      "eval_PRM F1 AUC": 0.8310633839706653,
      "eval_PRM F1 AUC (fixed)": 0.8326348873755893,
      "eval_PRM F1 Neg": 0.7058823529411765,
      "eval_PRM NPV": 0.6428571428571429,
      "eval_PRM Precision": 0.9358974358974359,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.23373456299304962,
      "eval_runtime": 14.7855,
      "eval_samples_per_second": 2.976,
      "eval_steps_per_second": 0.135,
      "step": 2865
    },
    {
      "epoch": 0.7422226538473989,
      "grad_norm": 3.963623292660699,
      "learning_rate": 2.3639297184242807e-07,
      "loss": 0.2709,
      "step": 2866
    },
    {
      "epoch": 0.7424816289534169,
      "grad_norm": 3.5975678949529764,
      "learning_rate": 2.3595046889993097e-07,
      "loss": 0.3458,
      "step": 2867
    },
    {
      "epoch": 0.7427406040594348,
      "grad_norm": 3.50660624992548,
      "learning_rate": 2.3550828411699875e-07,
      "loss": 0.2768,
      "step": 2868
    },
    {
      "epoch": 0.7429995791654527,
      "grad_norm": 2.9617561744751373,
      "learning_rate": 2.3506641785524385e-07,
      "loss": 0.3322,
      "step": 2869
    },
    {
      "epoch": 0.7432585542714707,
      "grad_norm": 4.380888961878263,
      "learning_rate": 2.3462487047601894e-07,
      "loss": 0.2811,
      "step": 2870
    },
    {
      "epoch": 0.7432585542714707,
      "eval_PRM Accuracy": 0.8679245283018868,
      "eval_PRM F1": 0.9135802469135802,
      "eval_PRM F1 AUC": 0.8370874803562075,
      "eval_PRM F1 AUC (fixed)": 0.8310633839706653,
      "eval_PRM F1 Neg": 0.72,
      "eval_PRM NPV": 0.6666666666666666,
      "eval_PRM Precision": 0.9367088607594937,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.23268574476242065,
      "eval_runtime": 14.4043,
      "eval_samples_per_second": 3.055,
      "eval_steps_per_second": 0.139,
      "step": 2870
    },
    {
      "epoch": 0.7435175293774886,
      "grad_norm": 4.978503548264406,
      "learning_rate": 2.3418364234041563e-07,
      "loss": 0.3683,
      "step": 2871
    },
    {
      "epoch": 0.7437765044835065,
      "grad_norm": 4.153421078281767,
      "learning_rate": 2.337427338092644e-07,
      "loss": 0.2729,
      "step": 2872
    },
    {
      "epoch": 0.7440354795895244,
      "grad_norm": 4.323960523855149,
      "learning_rate": 2.333021452431345e-07,
      "loss": 0.3039,
      "step": 2873
    },
    {
      "epoch": 0.7442944546955423,
      "grad_norm": 3.1605004921290476,
      "learning_rate": 2.3286187700233338e-07,
      "loss": 0.2845,
      "step": 2874
    },
    {
      "epoch": 0.7445534298015604,
      "grad_norm": 4.371614997023461,
      "learning_rate": 2.3242192944690657e-07,
      "loss": 0.318,
      "step": 2875
    },
    {
      "epoch": 0.7445534298015604,
      "eval_PRM Accuracy": 0.8679245283018868,
      "eval_PRM F1": 0.9135802469135802,
      "eval_PRM F1 AUC": 0.8370874803562075,
      "eval_PRM F1 AUC (fixed)": 0.8326348873755893,
      "eval_PRM F1 Neg": 0.72,
      "eval_PRM NPV": 0.6666666666666666,
      "eval_PRM Precision": 0.9367088607594937,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.23041768372058868,
      "eval_runtime": 14.3415,
      "eval_samples_per_second": 3.068,
      "eval_steps_per_second": 0.139,
      "step": 2875
    },
    {
      "epoch": 0.7448124049075783,
      "grad_norm": 3.826345059193562,
      "learning_rate": 2.319823029366372e-07,
      "loss": 0.2571,
      "step": 2876
    },
    {
      "epoch": 0.7450713800135962,
      "grad_norm": 3.9414898156437532,
      "learning_rate": 2.3154299783104662e-07,
      "loss": 0.3898,
      "step": 2877
    },
    {
      "epoch": 0.7453303551196141,
      "grad_norm": 6.050644821852145,
      "learning_rate": 2.311040144893921e-07,
      "loss": 0.3876,
      "step": 2878
    },
    {
      "epoch": 0.745589330225632,
      "grad_norm": 4.682386363347069,
      "learning_rate": 2.3066535327066868e-07,
      "loss": 0.3359,
      "step": 2879
    },
    {
      "epoch": 0.74584830533165,
      "grad_norm": 6.050774818896306,
      "learning_rate": 2.3022701453360766e-07,
      "loss": 0.4133,
      "step": 2880
    },
    {
      "epoch": 0.74584830533165,
      "eval_PRM Accuracy": 0.8679245283018868,
      "eval_PRM F1": 0.9135802469135802,
      "eval_PRM F1 AUC": 0.8370874803562075,
      "eval_PRM F1 AUC (fixed)": 0.8334206390780513,
      "eval_PRM F1 Neg": 0.72,
      "eval_PRM NPV": 0.6666666666666666,
      "eval_PRM Precision": 0.9367088607594937,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.23084813356399536,
      "eval_runtime": 14.8745,
      "eval_samples_per_second": 2.958,
      "eval_steps_per_second": 0.134,
      "step": 2880
    },
    {
      "epoch": 0.7461072804376679,
      "grad_norm": 3.813196007657059,
      "learning_rate": 2.2978899863667672e-07,
      "loss": 0.2873,
      "step": 2881
    },
    {
      "epoch": 0.7463662555436859,
      "grad_norm": 2.9673692996695387,
      "learning_rate": 2.293513059380794e-07,
      "loss": 0.2259,
      "step": 2882
    },
    {
      "epoch": 0.7466252306497038,
      "grad_norm": 5.236060150671778,
      "learning_rate": 2.2891393679575504e-07,
      "loss": 0.3411,
      "step": 2883
    },
    {
      "epoch": 0.7468842057557218,
      "grad_norm": 4.583578116905563,
      "learning_rate": 2.2847689156737837e-07,
      "loss": 0.3127,
      "step": 2884
    },
    {
      "epoch": 0.7471431808617397,
      "grad_norm": 3.453540361399094,
      "learning_rate": 2.2804017061035916e-07,
      "loss": 0.3136,
      "step": 2885
    },
    {
      "epoch": 0.7471431808617397,
      "eval_PRM Accuracy": 0.8679245283018868,
      "eval_PRM F1": 0.9135802469135802,
      "eval_PRM F1 AUC": 0.8370874803562075,
      "eval_PRM F1 AUC (fixed)": 0.8334206390780514,
      "eval_PRM F1 Neg": 0.72,
      "eval_PRM NPV": 0.6666666666666666,
      "eval_PRM Precision": 0.9367088607594937,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.23110078275203705,
      "eval_runtime": 15.1525,
      "eval_samples_per_second": 2.904,
      "eval_steps_per_second": 0.132,
      "step": 2885
    },
    {
      "epoch": 0.7474021559677576,
      "grad_norm": 4.711406507156151,
      "learning_rate": 2.2760377428184204e-07,
      "loss": 0.346,
      "step": 2886
    },
    {
      "epoch": 0.7476611310737755,
      "grad_norm": 3.6619480446611123,
      "learning_rate": 2.2716770293870615e-07,
      "loss": 0.2331,
      "step": 2887
    },
    {
      "epoch": 0.7479201061797934,
      "grad_norm": 5.206664836200739,
      "learning_rate": 2.267319569375649e-07,
      "loss": 0.2893,
      "step": 2888
    },
    {
      "epoch": 0.7481790812858115,
      "grad_norm": 3.6466819730716624,
      "learning_rate": 2.262965366347656e-07,
      "loss": 0.2946,
      "step": 2889
    },
    {
      "epoch": 0.7484380563918294,
      "grad_norm": 3.30594274919817,
      "learning_rate": 2.2586144238638925e-07,
      "loss": 0.2459,
      "step": 2890
    },
    {
      "epoch": 0.7484380563918294,
      "eval_PRM Accuracy": 0.8679245283018868,
      "eval_PRM F1": 0.9135802469135802,
      "eval_PRM F1 AUC": 0.8370874803562075,
      "eval_PRM F1 AUC (fixed)": 0.8347302252488213,
      "eval_PRM F1 Neg": 0.72,
      "eval_PRM NPV": 0.6666666666666666,
      "eval_PRM Precision": 0.9367088607594937,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.22940531373023987,
      "eval_runtime": 14.6413,
      "eval_samples_per_second": 3.005,
      "eval_steps_per_second": 0.137,
      "step": 2890
    },
    {
      "epoch": 0.7486970314978473,
      "grad_norm": 3.267216096670298,
      "learning_rate": 2.2542667454825008e-07,
      "loss": 0.251,
      "step": 2891
    },
    {
      "epoch": 0.7489560066038652,
      "grad_norm": 3.1584722213236756,
      "learning_rate": 2.2499223347589562e-07,
      "loss": 0.3024,
      "step": 2892
    },
    {
      "epoch": 0.7492149817098831,
      "grad_norm": 2.981528709712225,
      "learning_rate": 2.2455811952460618e-07,
      "loss": 0.242,
      "step": 2893
    },
    {
      "epoch": 0.7494739568159011,
      "grad_norm": 3.5105587803324356,
      "learning_rate": 2.2412433304939369e-07,
      "loss": 0.3742,
      "step": 2894
    },
    {
      "epoch": 0.749732931921919,
      "grad_norm": 2.8974209868083616,
      "learning_rate": 2.2369087440500378e-07,
      "loss": 0.2706,
      "step": 2895
    },
    {
      "epoch": 0.749732931921919,
      "eval_PRM Accuracy": 0.8773584905660378,
      "eval_PRM F1": 0.9202453987730062,
      "eval_PRM F1 AUC": 0.8431115767417497,
      "eval_PRM F1 AUC (fixed)": 0.8355159769512833,
      "eval_PRM F1 Neg": 0.7346938775510204,
      "eval_PRM NPV": 0.6923076923076923,
      "eval_PRM Precision": 0.9375,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.2264910638332367,
      "eval_runtime": 14.6151,
      "eval_samples_per_second": 3.011,
      "eval_steps_per_second": 0.137,
      "step": 2895
    },
    {
      "epoch": 0.7499919070279369,
      "grad_norm": 3.6899537558256537,
      "learning_rate": 2.2325774394591292e-07,
      "loss": 0.3772,
      "step": 2896
    },
    {
      "epoch": 0.7502508821339549,
      "grad_norm": 4.264309926705439,
      "learning_rate": 2.2282494202632948e-07,
      "loss": 0.3937,
      "step": 2897
    },
    {
      "epoch": 0.7505098572399728,
      "grad_norm": 3.0071266872107976,
      "learning_rate": 2.2239246900019315e-07,
      "loss": 0.2923,
      "step": 2898
    },
    {
      "epoch": 0.7507688323459908,
      "grad_norm": 3.8316015054711725,
      "learning_rate": 2.2196032522117467e-07,
      "loss": 0.3405,
      "step": 2899
    },
    {
      "epoch": 0.7510278074520087,
      "grad_norm": 3.796302198335329,
      "learning_rate": 2.2152851104267542e-07,
      "loss": 0.3098,
      "step": 2900
    },
    {
      "epoch": 0.7510278074520087,
      "eval_PRM Accuracy": 0.8773584905660378,
      "eval_PRM F1": 0.9212121212121213,
      "eval_PRM F1 AUC": 0.8273965426925091,
      "eval_PRM F1 AUC (fixed)": 0.8391828182294394,
      "eval_PRM F1 Neg": 0.723404255319149,
      "eval_PRM NPV": 0.7083333333333334,
      "eval_PRM Precision": 0.926829268292683,
      "eval_PRM Recall": 0.9156626506024096,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.2272801250219345,
      "eval_runtime": 14.8644,
      "eval_samples_per_second": 2.96,
      "eval_steps_per_second": 0.135,
      "step": 2900
    },
    {
      "epoch": 0.7512867825580266,
      "grad_norm": 3.430510543628399,
      "learning_rate": 2.2109702681782746e-07,
      "loss": 0.3626,
      "step": 2901
    },
    {
      "epoch": 0.7515457576640445,
      "grad_norm": 3.3080963743653973,
      "learning_rate": 2.2066587289949307e-07,
      "loss": 0.375,
      "step": 2902
    },
    {
      "epoch": 0.7518047327700624,
      "grad_norm": 5.083432649408308,
      "learning_rate": 2.2023504964026374e-07,
      "loss": 0.2744,
      "step": 2903
    },
    {
      "epoch": 0.7520637078760805,
      "grad_norm": 3.3586996203144874,
      "learning_rate": 2.1980455739246105e-07,
      "loss": 0.3611,
      "step": 2904
    },
    {
      "epoch": 0.7523226829820984,
      "grad_norm": 3.208025635742934,
      "learning_rate": 2.1937439650813636e-07,
      "loss": 0.3244,
      "step": 2905
    },
    {
      "epoch": 0.7523226829820984,
      "eval_PRM Accuracy": 0.8773584905660378,
      "eval_PRM F1": 0.9212121212121213,
      "eval_PRM F1 AUC": 0.8273965426925091,
      "eval_PRM F1 AUC (fixed)": 0.8376113148245155,
      "eval_PRM F1 Neg": 0.723404255319149,
      "eval_PRM NPV": 0.7083333333333334,
      "eval_PRM Precision": 0.926829268292683,
      "eval_PRM Recall": 0.9156626506024096,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.22871938347816467,
      "eval_runtime": 14.3874,
      "eval_samples_per_second": 3.058,
      "eval_steps_per_second": 0.139,
      "step": 2905
    },
    {
      "epoch": 0.7525816580881163,
      "grad_norm": 3.1703004918514957,
      "learning_rate": 2.1894456733906922e-07,
      "loss": 0.2444,
      "step": 2906
    },
    {
      "epoch": 0.7528406331941342,
      "grad_norm": 3.6930726402642247,
      "learning_rate": 2.185150702367684e-07,
      "loss": 0.2366,
      "step": 2907
    },
    {
      "epoch": 0.7530996083001521,
      "grad_norm": 3.9988190512251225,
      "learning_rate": 2.180859055524709e-07,
      "loss": 0.3499,
      "step": 2908
    },
    {
      "epoch": 0.7533585834061701,
      "grad_norm": 2.473143859386706,
      "learning_rate": 2.176570736371418e-07,
      "loss": 0.2843,
      "step": 2909
    },
    {
      "epoch": 0.753617558512188,
      "grad_norm": 5.220788784607673,
      "learning_rate": 2.172285748414746e-07,
      "loss": 0.3176,
      "step": 2910
    },
    {
      "epoch": 0.753617558512188,
      "eval_PRM Accuracy": 0.8773584905660378,
      "eval_PRM F1": 0.9212121212121213,
      "eval_PRM F1 AUC": 0.8273965426925091,
      "eval_PRM F1 AUC (fixed)": 0.8383970665269775,
      "eval_PRM F1 Neg": 0.723404255319149,
      "eval_PRM NPV": 0.7083333333333334,
      "eval_PRM Precision": 0.926829268292683,
      "eval_PRM Recall": 0.9156626506024096,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.2287004590034485,
      "eval_runtime": 14.8936,
      "eval_samples_per_second": 2.954,
      "eval_steps_per_second": 0.134,
      "step": 2910
    },
    {
      "epoch": 0.753876533618206,
      "grad_norm": 4.834822034283974,
      "learning_rate": 2.1680040951588932e-07,
      "loss": 0.2934,
      "step": 2911
    },
    {
      "epoch": 0.7541355087242239,
      "grad_norm": 3.535190802724417,
      "learning_rate": 2.1637257801053426e-07,
      "loss": 0.3001,
      "step": 2912
    },
    {
      "epoch": 0.7543944838302418,
      "grad_norm": 2.487752453979307,
      "learning_rate": 2.1594508067528436e-07,
      "loss": 0.2628,
      "step": 2913
    },
    {
      "epoch": 0.7546534589362598,
      "grad_norm": 4.0949070224596635,
      "learning_rate": 2.1551791785974122e-07,
      "loss": 0.3194,
      "step": 2914
    },
    {
      "epoch": 0.7549124340422777,
      "grad_norm": 7.2854017402948985,
      "learning_rate": 2.1509108991323274e-07,
      "loss": 0.3456,
      "step": 2915
    },
    {
      "epoch": 0.7549124340422777,
      "eval_PRM Accuracy": 0.8867924528301887,
      "eval_PRM F1": 0.926829268292683,
      "eval_PRM F1 AUC": 0.8491356731272918,
      "eval_PRM F1 AUC (fixed)": 0.8368255631220535,
      "eval_PRM F1 Neg": 0.75,
      "eval_PRM NPV": 0.72,
      "eval_PRM Precision": 0.9382716049382716,
      "eval_PRM Recall": 0.9156626506024096,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.22821766138076782,
      "eval_runtime": 14.6003,
      "eval_samples_per_second": 3.014,
      "eval_steps_per_second": 0.137,
      "step": 2915
    },
    {
      "epoch": 0.7551714091482956,
      "grad_norm": 5.0713922295774685,
      "learning_rate": 2.1466459718481366e-07,
      "loss": 0.3532,
      "step": 2916
    },
    {
      "epoch": 0.7554303842543135,
      "grad_norm": 3.9088880858346706,
      "learning_rate": 2.142384400232639e-07,
      "loss": 0.3424,
      "step": 2917
    },
    {
      "epoch": 0.7556893593603314,
      "grad_norm": 4.415286103053356,
      "learning_rate": 2.138126187770892e-07,
      "loss": 0.3758,
      "step": 2918
    },
    {
      "epoch": 0.7559483344663495,
      "grad_norm": 3.5402475459653386,
      "learning_rate": 2.133871337945203e-07,
      "loss": 0.3029,
      "step": 2919
    },
    {
      "epoch": 0.7562073095723674,
      "grad_norm": 6.8584038720887275,
      "learning_rate": 2.1296198542351335e-07,
      "loss": 0.4299,
      "step": 2920
    },
    {
      "epoch": 0.7562073095723674,
      "eval_PRM Accuracy": 0.8773584905660378,
      "eval_PRM F1": 0.9202453987730062,
      "eval_PRM F1 AUC": 0.8431115767417497,
      "eval_PRM F1 AUC (fixed)": 0.8357778941854375,
      "eval_PRM F1 Neg": 0.7346938775510204,
      "eval_PRM NPV": 0.6923076923076923,
      "eval_PRM Precision": 0.9375,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.22816146910190582,
      "eval_runtime": 14.9008,
      "eval_samples_per_second": 2.953,
      "eval_steps_per_second": 0.134,
      "step": 2920
    },
    {
      "epoch": 0.7564662846783853,
      "grad_norm": 3.990607525904205,
      "learning_rate": 2.1253717401174905e-07,
      "loss": 0.2646,
      "step": 2921
    },
    {
      "epoch": 0.7567252597844032,
      "grad_norm": 3.340069736421438,
      "learning_rate": 2.121126999066326e-07,
      "loss": 0.3159,
      "step": 2922
    },
    {
      "epoch": 0.7569842348904211,
      "grad_norm": 3.1420851935386693,
      "learning_rate": 2.116885634552933e-07,
      "loss": 0.2746,
      "step": 2923
    },
    {
      "epoch": 0.7572432099964391,
      "grad_norm": 3.851717748884594,
      "learning_rate": 2.1126476500458433e-07,
      "loss": 0.3071,
      "step": 2924
    },
    {
      "epoch": 0.757502185102457,
      "grad_norm": 3.80244181751963,
      "learning_rate": 2.108413049010822e-07,
      "loss": 0.3727,
      "step": 2925
    },
    {
      "epoch": 0.757502185102457,
      "eval_PRM Accuracy": 0.8773584905660378,
      "eval_PRM F1": 0.9202453987730062,
      "eval_PRM F1 AUC": 0.8431115767417497,
      "eval_PRM F1 AUC (fixed)": 0.8349921424829754,
      "eval_PRM F1 Neg": 0.7346938775510204,
      "eval_PRM NPV": 0.6923076923076923,
      "eval_PRM Precision": 0.9375,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.2295248657464981,
      "eval_runtime": 14.5368,
      "eval_samples_per_second": 3.027,
      "eval_steps_per_second": 0.138,
      "step": 2925
    },
    {
      "epoch": 0.757761160208475,
      "grad_norm": 3.957427228586354,
      "learning_rate": 2.1041818349108747e-07,
      "loss": 0.3023,
      "step": 2926
    },
    {
      "epoch": 0.7580201353144929,
      "grad_norm": 6.320085307286381,
      "learning_rate": 2.0999540112062318e-07,
      "loss": 0.3292,
      "step": 2927
    },
    {
      "epoch": 0.7582791104205109,
      "grad_norm": 5.445152113130647,
      "learning_rate": 2.0957295813543477e-07,
      "loss": 0.3458,
      "step": 2928
    },
    {
      "epoch": 0.7585380855265288,
      "grad_norm": 3.81293640137749,
      "learning_rate": 2.0915085488099083e-07,
      "loss": 0.3558,
      "step": 2929
    },
    {
      "epoch": 0.7587970606325467,
      "grad_norm": 3.287615015635259,
      "learning_rate": 2.0872909170248167e-07,
      "loss": 0.3304,
      "step": 2930
    },
    {
      "epoch": 0.7587970606325467,
      "eval_PRM Accuracy": 0.8679245283018868,
      "eval_PRM F1": 0.9135802469135802,
      "eval_PRM F1 AUC": 0.8370874803562075,
      "eval_PRM F1 AUC (fixed)": 0.8347302252488213,
      "eval_PRM F1 Neg": 0.72,
      "eval_PRM NPV": 0.6666666666666666,
      "eval_PRM Precision": 0.9367088607594937,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.22973623871803284,
      "eval_runtime": 14.375,
      "eval_samples_per_second": 3.061,
      "eval_steps_per_second": 0.139,
      "step": 2930
    },
    {
      "epoch": 0.7590560357385646,
      "grad_norm": 4.535380904752414,
      "learning_rate": 2.0830766894481978e-07,
      "loss": 0.3579,
      "step": 2931
    },
    {
      "epoch": 0.7593150108445825,
      "grad_norm": 3.371851800323252,
      "learning_rate": 2.0788658695263905e-07,
      "loss": 0.2922,
      "step": 2932
    },
    {
      "epoch": 0.7595739859506005,
      "grad_norm": 3.8945593148001865,
      "learning_rate": 2.0746584607029475e-07,
      "loss": 0.2705,
      "step": 2933
    },
    {
      "epoch": 0.7598329610566185,
      "grad_norm": 3.541298761627775,
      "learning_rate": 2.0704544664186324e-07,
      "loss": 0.3304,
      "step": 2934
    },
    {
      "epoch": 0.7600919361626364,
      "grad_norm": 6.026919393517336,
      "learning_rate": 2.0662538901114165e-07,
      "loss": 0.3248,
      "step": 2935
    },
    {
      "epoch": 0.7600919361626364,
      "eval_PRM Accuracy": 0.8679245283018868,
      "eval_PRM F1": 0.9135802469135802,
      "eval_PRM F1 AUC": 0.8370874803562075,
      "eval_PRM F1 AUC (fixed)": 0.8339444735463594,
      "eval_PRM F1 Neg": 0.72,
      "eval_PRM NPV": 0.6666666666666666,
      "eval_PRM Precision": 0.9367088607594937,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.2292565405368805,
      "eval_runtime": 14.6423,
      "eval_samples_per_second": 3.005,
      "eval_steps_per_second": 0.137,
      "step": 2935
    },
    {
      "epoch": 0.7603509112686543,
      "grad_norm": 2.5845054456256142,
      "learning_rate": 2.062056735216475e-07,
      "loss": 0.3033,
      "step": 2936
    },
    {
      "epoch": 0.7606098863746722,
      "grad_norm": 5.701883405615083,
      "learning_rate": 2.0578630051661857e-07,
      "loss": 0.4074,
      "step": 2937
    },
    {
      "epoch": 0.7608688614806902,
      "grad_norm": 3.62521433203709,
      "learning_rate": 2.0536727033901264e-07,
      "loss": 0.342,
      "step": 2938
    },
    {
      "epoch": 0.7611278365867081,
      "grad_norm": 4.08478044721597,
      "learning_rate": 2.0494858333150695e-07,
      "loss": 0.2433,
      "step": 2939
    },
    {
      "epoch": 0.761386811692726,
      "grad_norm": 3.227801546256013,
      "learning_rate": 2.0453023983649823e-07,
      "loss": 0.3117,
      "step": 2940
    },
    {
      "epoch": 0.761386811692726,
      "eval_PRM Accuracy": 0.8773584905660378,
      "eval_PRM F1": 0.9202453987730062,
      "eval_PRM F1 AUC": 0.8431115767417497,
      "eval_PRM F1 AUC (fixed)": 0.8334206390780513,
      "eval_PRM F1 Neg": 0.7346938775510204,
      "eval_PRM NPV": 0.6923076923076923,
      "eval_PRM Precision": 0.9375,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.22916609048843384,
      "eval_runtime": 14.9741,
      "eval_samples_per_second": 2.938,
      "eval_steps_per_second": 0.134,
      "step": 2940
    },
    {
      "epoch": 0.761645786798744,
      "grad_norm": 3.3924559043095996,
      "learning_rate": 2.041122401961023e-07,
      "loss": 0.2279,
      "step": 2941
    },
    {
      "epoch": 0.7619047619047619,
      "grad_norm": 3.503255591179924,
      "learning_rate": 2.0369458475215377e-07,
      "loss": 0.4037,
      "step": 2942
    },
    {
      "epoch": 0.7621637370107799,
      "grad_norm": 4.068064126691298,
      "learning_rate": 2.032772738462057e-07,
      "loss": 0.2641,
      "step": 2943
    },
    {
      "epoch": 0.7624227121167978,
      "grad_norm": 3.604396911415084,
      "learning_rate": 2.0286030781952937e-07,
      "loss": 0.2988,
      "step": 2944
    },
    {
      "epoch": 0.7626816872228157,
      "grad_norm": 3.4105106103528313,
      "learning_rate": 2.0244368701311412e-07,
      "loss": 0.2599,
      "step": 2945
    },
    {
      "epoch": 0.7626816872228157,
      "eval_PRM Accuracy": 0.8773584905660378,
      "eval_PRM F1": 0.9202453987730062,
      "eval_PRM F1 AUC": 0.8431115767417497,
      "eval_PRM F1 AUC (fixed)": 0.8355159769512834,
      "eval_PRM F1 Neg": 0.7346938775510204,
      "eval_PRM NPV": 0.6923076923076923,
      "eval_PRM Precision": 0.9375,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.2277769148349762,
      "eval_runtime": 14.5516,
      "eval_samples_per_second": 3.024,
      "eval_steps_per_second": 0.137,
      "step": 2945
    },
    {
      "epoch": 0.7629406623288336,
      "grad_norm": 3.787524360952269,
      "learning_rate": 2.0202741176766704e-07,
      "loss": 0.3728,
      "step": 2946
    },
    {
      "epoch": 0.7631996374348515,
      "grad_norm": 2.972254243613975,
      "learning_rate": 2.0161148242361242e-07,
      "loss": 0.2578,
      "step": 2947
    },
    {
      "epoch": 0.7634586125408696,
      "grad_norm": 4.588614420592178,
      "learning_rate": 2.0119589932109177e-07,
      "loss": 0.3827,
      "step": 2948
    },
    {
      "epoch": 0.7637175876468875,
      "grad_norm": 3.2828696968398843,
      "learning_rate": 2.0078066279996353e-07,
      "loss": 0.2979,
      "step": 2949
    },
    {
      "epoch": 0.7639765627529054,
      "grad_norm": 6.290858049289823,
      "learning_rate": 2.0036577319980258e-07,
      "loss": 0.303,
      "step": 2950
    },
    {
      "epoch": 0.7639765627529054,
      "eval_PRM Accuracy": 0.8867924528301887,
      "eval_PRM F1": 0.926829268292683,
      "eval_PRM F1 AUC": 0.8491356731272918,
      "eval_PRM F1 AUC (fixed)": 0.8342063907805134,
      "eval_PRM F1 Neg": 0.75,
      "eval_PRM NPV": 0.72,
      "eval_PRM Precision": 0.9382716049382716,
      "eval_PRM Recall": 0.9156626506024096,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.22783786058425903,
      "eval_runtime": 14.8207,
      "eval_samples_per_second": 2.969,
      "eval_steps_per_second": 0.135,
      "step": 2950
    },
    {
      "epoch": 0.7642355378589233,
      "grad_norm": 5.135943410084388,
      "learning_rate": 1.9995123085990039e-07,
      "loss": 0.3588,
      "step": 2951
    },
    {
      "epoch": 0.7644945129649412,
      "grad_norm": 4.356743428889666,
      "learning_rate": 1.9953703611926382e-07,
      "loss": 0.3416,
      "step": 2952
    },
    {
      "epoch": 0.7647534880709592,
      "grad_norm": 4.074456310767771,
      "learning_rate": 1.991231893166158e-07,
      "loss": 0.2951,
      "step": 2953
    },
    {
      "epoch": 0.7650124631769771,
      "grad_norm": 3.3434995621714734,
      "learning_rate": 1.9870969079039521e-07,
      "loss": 0.3166,
      "step": 2954
    },
    {
      "epoch": 0.765271438282995,
      "grad_norm": 3.6192547909082213,
      "learning_rate": 1.9829654087875551e-07,
      "loss": 0.3071,
      "step": 2955
    },
    {
      "epoch": 0.765271438282995,
      "eval_PRM Accuracy": 0.8773584905660378,
      "eval_PRM F1": 0.9202453987730062,
      "eval_PRM F1 AUC": 0.8431115767417497,
      "eval_PRM F1 AUC (fixed)": 0.8383970665269775,
      "eval_PRM F1 Neg": 0.7346938775510204,
      "eval_PRM NPV": 0.6923076923076923,
      "eval_PRM Precision": 0.9375,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.22701311111450195,
      "eval_runtime": 14.3776,
      "eval_samples_per_second": 3.06,
      "eval_steps_per_second": 0.139,
      "step": 2955
    },
    {
      "epoch": 0.765530413389013,
      "grad_norm": 5.0995592777382575,
      "learning_rate": 1.978837399195651e-07,
      "loss": 0.3758,
      "step": 2956
    },
    {
      "epoch": 0.7657893884950309,
      "grad_norm": 2.9716818750306873,
      "learning_rate": 1.9747128825040734e-07,
      "loss": 0.2773,
      "step": 2957
    },
    {
      "epoch": 0.7660483636010489,
      "grad_norm": 2.801288348925496,
      "learning_rate": 1.970591862085796e-07,
      "loss": 0.2995,
      "step": 2958
    },
    {
      "epoch": 0.7663073387070668,
      "grad_norm": 4.394335199398056,
      "learning_rate": 1.9664743413109358e-07,
      "loss": 0.3208,
      "step": 2959
    },
    {
      "epoch": 0.7665663138130847,
      "grad_norm": 4.552067924542375,
      "learning_rate": 1.962360323546749e-07,
      "loss": 0.2732,
      "step": 2960
    },
    {
      "epoch": 0.7665663138130847,
      "eval_PRM Accuracy": 0.8679245283018868,
      "eval_PRM F1": 0.9135802469135802,
      "eval_PRM F1 AUC": 0.8370874803562075,
      "eval_PRM F1 AUC (fixed)": 0.8381351492928235,
      "eval_PRM F1 Neg": 0.72,
      "eval_PRM NPV": 0.6666666666666666,
      "eval_PRM Precision": 0.9367088607594937,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.22769373655319214,
      "eval_runtime": 14.9743,
      "eval_samples_per_second": 2.938,
      "eval_steps_per_second": 0.134,
      "step": 2960
    },
    {
      "epoch": 0.7668252889191026,
      "grad_norm": 7.130114403393841,
      "learning_rate": 1.9582498121576198e-07,
      "loss": 0.3779,
      "step": 2961
    },
    {
      "epoch": 0.7670842640251205,
      "grad_norm": 4.012868454252892,
      "learning_rate": 1.9541428105050732e-07,
      "loss": 0.3309,
      "step": 2962
    },
    {
      "epoch": 0.7673432391311386,
      "grad_norm": 3.1817172841724624,
      "learning_rate": 1.9500393219477582e-07,
      "loss": 0.3287,
      "step": 2963
    },
    {
      "epoch": 0.7676022142371565,
      "grad_norm": 3.1001554202069963,
      "learning_rate": 1.945939349841458e-07,
      "loss": 0.322,
      "step": 2964
    },
    {
      "epoch": 0.7678611893431744,
      "grad_norm": 2.8864519368875983,
      "learning_rate": 1.941842897539073e-07,
      "loss": 0.3645,
      "step": 2965
    },
    {
      "epoch": 0.7678611893431744,
      "eval_PRM Accuracy": 0.8679245283018868,
      "eval_PRM F1": 0.9135802469135802,
      "eval_PRM F1 AUC": 0.8370874803562075,
      "eval_PRM F1 AUC (fixed)": 0.8339444735463593,
      "eval_PRM F1 Neg": 0.72,
      "eval_PRM NPV": 0.6666666666666666,
      "eval_PRM Precision": 0.9367088607594937,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.23120974004268646,
      "eval_runtime": 14.5469,
      "eval_samples_per_second": 3.025,
      "eval_steps_per_second": 0.137,
      "step": 2965
    },
    {
      "epoch": 0.7681201644491923,
      "grad_norm": 3.9210303874365646,
      "learning_rate": 1.937749968390628e-07,
      "loss": 0.3284,
      "step": 2966
    },
    {
      "epoch": 0.7683791395552102,
      "grad_norm": 5.134356935350124,
      "learning_rate": 1.9336605657432672e-07,
      "loss": 0.3211,
      "step": 2967
    },
    {
      "epoch": 0.7686381146612282,
      "grad_norm": 4.394796681051276,
      "learning_rate": 1.929574692941251e-07,
      "loss": 0.2479,
      "step": 2968
    },
    {
      "epoch": 0.7688970897672461,
      "grad_norm": 4.1176727092029,
      "learning_rate": 1.925492353325948e-07,
      "loss": 0.4089,
      "step": 2969
    },
    {
      "epoch": 0.769156064873264,
      "grad_norm": 6.94792307259944,
      "learning_rate": 1.9214135502358448e-07,
      "loss": 0.3773,
      "step": 2970
    },
    {
      "epoch": 0.769156064873264,
      "eval_PRM Accuracy": 0.8679245283018868,
      "eval_PRM F1": 0.9135802469135802,
      "eval_PRM F1 AUC": 0.8370874803562075,
      "eval_PRM F1 AUC (fixed)": 0.8378732320586695,
      "eval_PRM F1 Neg": 0.72,
      "eval_PRM NPV": 0.6666666666666666,
      "eval_PRM Precision": 0.9367088607594937,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.23140259087085724,
      "eval_runtime": 14.2986,
      "eval_samples_per_second": 3.077,
      "eval_steps_per_second": 0.14,
      "step": 2970
    },
    {
      "epoch": 0.769415039979282,
      "grad_norm": 7.752866355737287,
      "learning_rate": 1.9173382870065316e-07,
      "loss": 0.3714,
      "step": 2971
    },
    {
      "epoch": 0.7696740150853,
      "grad_norm": 4.984707861256453,
      "learning_rate": 1.9132665669707048e-07,
      "loss": 0.3043,
      "step": 2972
    },
    {
      "epoch": 0.7699329901913179,
      "grad_norm": 4.8159645962756015,
      "learning_rate": 1.9091983934581644e-07,
      "loss": 0.3033,
      "step": 2973
    },
    {
      "epoch": 0.7701919652973358,
      "grad_norm": 3.9415651917856773,
      "learning_rate": 1.9051337697958055e-07,
      "loss": 0.294,
      "step": 2974
    },
    {
      "epoch": 0.7704509404033537,
      "grad_norm": 6.7247907471773924,
      "learning_rate": 1.901072699307628e-07,
      "loss": 0.402,
      "step": 2975
    },
    {
      "epoch": 0.7704509404033537,
      "eval_PRM Accuracy": 0.8584905660377359,
      "eval_PRM F1": 0.906832298136646,
      "eval_PRM F1 AUC": 0.8310633839706653,
      "eval_PRM F1 AUC (fixed)": 0.8349921424829754,
      "eval_PRM F1 Neg": 0.7058823529411765,
      "eval_PRM NPV": 0.6428571428571429,
      "eval_PRM Precision": 0.9358974358974359,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.22979949414730072,
      "eval_runtime": 14.7783,
      "eval_samples_per_second": 2.977,
      "eval_steps_per_second": 0.135,
      "step": 2975
    },
    {
      "epoch": 0.7707099155093716,
      "grad_norm": 4.105038568588277,
      "learning_rate": 1.8970151853147226e-07,
      "loss": 0.2915,
      "step": 2976
    },
    {
      "epoch": 0.7709688906153896,
      "grad_norm": 4.541493565871747,
      "learning_rate": 1.8929612311352653e-07,
      "loss": 0.3855,
      "step": 2977
    },
    {
      "epoch": 0.7712278657214076,
      "grad_norm": 7.650059600888274,
      "learning_rate": 1.8889108400845292e-07,
      "loss": 0.3329,
      "step": 2978
    },
    {
      "epoch": 0.7714868408274255,
      "grad_norm": 7.587293225289113,
      "learning_rate": 1.88486401547487e-07,
      "loss": 0.3406,
      "step": 2979
    },
    {
      "epoch": 0.7717458159334434,
      "grad_norm": 3.123925711260492,
      "learning_rate": 1.880820760615728e-07,
      "loss": 0.3187,
      "step": 2980
    },
    {
      "epoch": 0.7717458159334434,
      "eval_PRM Accuracy": 0.8679245283018868,
      "eval_PRM F1": 0.9135802469135802,
      "eval_PRM F1 AUC": 0.8370874803562075,
      "eval_PRM F1 AUC (fixed)": 0.8378732320586695,
      "eval_PRM F1 Neg": 0.72,
      "eval_PRM NPV": 0.6666666666666666,
      "eval_PRM Precision": 0.9367088607594937,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.2284248024225235,
      "eval_runtime": 14.8938,
      "eval_samples_per_second": 2.954,
      "eval_steps_per_second": 0.134,
      "step": 2980
    },
    {
      "epoch": 0.7720047910394613,
      "grad_norm": 4.546033709389239,
      "learning_rate": 1.8767810788136228e-07,
      "loss": 0.3236,
      "step": 2981
    },
    {
      "epoch": 0.7722637661454793,
      "grad_norm": 2.952387281088917,
      "learning_rate": 1.872744973372153e-07,
      "loss": 0.272,
      "step": 2982
    },
    {
      "epoch": 0.7725227412514972,
      "grad_norm": 5.983474289652768,
      "learning_rate": 1.868712447591993e-07,
      "loss": 0.4171,
      "step": 2983
    },
    {
      "epoch": 0.7727817163575151,
      "grad_norm": 5.359616578272896,
      "learning_rate": 1.8646835047708866e-07,
      "loss": 0.3666,
      "step": 2984
    },
    {
      "epoch": 0.773040691463533,
      "grad_norm": 3.833222108486842,
      "learning_rate": 1.8606581482036555e-07,
      "loss": 0.3145,
      "step": 2985
    },
    {
      "epoch": 0.773040691463533,
      "eval_PRM Accuracy": 0.8773584905660378,
      "eval_PRM F1": 0.9202453987730062,
      "eval_PRM F1 AUC": 0.8431115767417497,
      "eval_PRM F1 AUC (fixed)": 0.8355159769512834,
      "eval_PRM F1 Neg": 0.7346938775510204,
      "eval_PRM NPV": 0.6923076923076923,
      "eval_PRM Precision": 0.9375,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.2284346967935562,
      "eval_runtime": 14.67,
      "eval_samples_per_second": 2.999,
      "eval_steps_per_second": 0.136,
      "step": 2985
    },
    {
      "epoch": 0.773299666569551,
      "grad_norm": 3.1363161701544207,
      "learning_rate": 1.856636381182178e-07,
      "loss": 0.3007,
      "step": 2986
    },
    {
      "epoch": 0.773558641675569,
      "grad_norm": 4.0187458096732085,
      "learning_rate": 1.8526182069954033e-07,
      "loss": 0.2992,
      "step": 2987
    },
    {
      "epoch": 0.7738176167815869,
      "grad_norm": 4.184876062840923,
      "learning_rate": 1.8486036289293416e-07,
      "loss": 0.3693,
      "step": 2988
    },
    {
      "epoch": 0.7740765918876048,
      "grad_norm": 6.083137890623333,
      "learning_rate": 1.8445926502670608e-07,
      "loss": 0.3832,
      "step": 2989
    },
    {
      "epoch": 0.7743355669936227,
      "grad_norm": 4.806926998857229,
      "learning_rate": 1.840585274288688e-07,
      "loss": 0.2569,
      "step": 2990
    },
    {
      "epoch": 0.7743355669936227,
      "eval_PRM Accuracy": 0.8679245283018868,
      "eval_PRM F1": 0.9135802469135802,
      "eval_PRM F1 AUC": 0.8370874803562075,
      "eval_PRM F1 AUC (fixed)": 0.8365636458878993,
      "eval_PRM F1 Neg": 0.72,
      "eval_PRM NPV": 0.6666666666666666,
      "eval_PRM Precision": 0.9367088607594937,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.22727304697036743,
      "eval_runtime": 14.6615,
      "eval_samples_per_second": 3.001,
      "eval_steps_per_second": 0.136,
      "step": 2990
    },
    {
      "epoch": 0.7745945420996406,
      "grad_norm": 3.073691402486179,
      "learning_rate": 1.8365815042714008e-07,
      "loss": 0.274,
      "step": 2991
    },
    {
      "epoch": 0.7748535172056586,
      "grad_norm": 4.90741152098729,
      "learning_rate": 1.8325813434894305e-07,
      "loss": 0.282,
      "step": 2992
    },
    {
      "epoch": 0.7751124923116766,
      "grad_norm": 5.368983235329144,
      "learning_rate": 1.828584795214055e-07,
      "loss": 0.2899,
      "step": 2993
    },
    {
      "epoch": 0.7753714674176945,
      "grad_norm": 4.0449281813108895,
      "learning_rate": 1.8245918627135984e-07,
      "loss": 0.252,
      "step": 2994
    },
    {
      "epoch": 0.7756304425237124,
      "grad_norm": 2.8021023278083836,
      "learning_rate": 1.8206025492534286e-07,
      "loss": 0.2209,
      "step": 2995
    },
    {
      "epoch": 0.7756304425237124,
      "eval_PRM Accuracy": 0.8773584905660378,
      "eval_PRM F1": 0.9202453987730062,
      "eval_PRM F1 AUC": 0.8431115767417497,
      "eval_PRM F1 AUC (fixed)": 0.8357778941854375,
      "eval_PRM F1 Neg": 0.7346938775510204,
      "eval_PRM NPV": 0.6923076923076923,
      "eval_PRM Precision": 0.9375,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.22657732665538788,
      "eval_runtime": 14.341,
      "eval_samples_per_second": 3.068,
      "eval_steps_per_second": 0.139,
      "step": 2995
    },
    {
      "epoch": 0.7758894176297303,
      "grad_norm": 4.973221125397011,
      "learning_rate": 1.8166168580959536e-07,
      "loss": 0.3623,
      "step": 2996
    },
    {
      "epoch": 0.7761483927357483,
      "grad_norm": 6.772078715063489,
      "learning_rate": 1.8126347925006193e-07,
      "loss": 0.3588,
      "step": 2997
    },
    {
      "epoch": 0.7764073678417662,
      "grad_norm": 3.37575810495228,
      "learning_rate": 1.8086563557239048e-07,
      "loss": 0.3158,
      "step": 2998
    },
    {
      "epoch": 0.7766663429477841,
      "grad_norm": 3.0744650160450218,
      "learning_rate": 1.8046815510193245e-07,
      "loss": 0.2523,
      "step": 2999
    },
    {
      "epoch": 0.776925318053802,
      "grad_norm": 7.054157879790193,
      "learning_rate": 1.800710381637421e-07,
      "loss": 0.2923,
      "step": 3000
    },
    {
      "epoch": 0.776925318053802,
      "eval_PRM Accuracy": 0.8773584905660378,
      "eval_PRM F1": 0.9202453987730062,
      "eval_PRM F1 AUC": 0.8431115767417497,
      "eval_PRM F1 AUC (fixed)": 0.8376113148245155,
      "eval_PRM F1 Neg": 0.7346938775510204,
      "eval_PRM NPV": 0.6923076923076923,
      "eval_PRM Precision": 0.9375,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.22605527937412262,
      "eval_runtime": 14.7354,
      "eval_samples_per_second": 2.986,
      "eval_steps_per_second": 0.136,
      "step": 3000
    }
  ],
  "logging_steps": 1,
  "max_steps": 3861,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 253587145543680.0,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}