{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.0,
  "eval_steps": 5,
  "global_step": 2527,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0,
      "eval_PRM Accuracy": 0.15079365079365079,
      "eval_PRM F1": 0.0,
      "eval_PRM F1 AUC": 0.5,
      "eval_PRM F1 AUC (fixed)": 0.5034431874077718,
      "eval_PRM F1 Neg": 0.2620689655172414,
      "eval_PRM NPV": 0.15079365079365079,
      "eval_PRM Precision": 0.0,
      "eval_PRM Recall": 0.0,
      "eval_PRM Specificty": 1.0,
      "eval_loss": 5.104166507720947,
      "eval_runtime": 5.645,
      "eval_samples_per_second": 5.314,
      "eval_steps_per_second": 0.177,
      "step": 0
    },
    {
      "epoch": 0.0003957261574990107,
      "grad_norm": 18.679440766969446,
      "learning_rate": 3.952569169960474e-07,
      "loss": 4.1606,
      "step": 1
    },
    {
      "epoch": 0.0007914523149980214,
      "grad_norm": 19.93434430744454,
      "learning_rate": 7.905138339920948e-07,
      "loss": 3.9287,
      "step": 2
    },
    {
      "epoch": 0.001187178472497032,
      "grad_norm": 18.764187337010576,
      "learning_rate": 1.1857707509881422e-06,
      "loss": 3.9258,
      "step": 3
    },
    {
      "epoch": 0.0015829046299960427,
      "grad_norm": 19.92512130907959,
      "learning_rate": 1.5810276679841897e-06,
      "loss": 4.1885,
      "step": 4
    },
    {
      "epoch": 0.0019786307874950534,
      "grad_norm": 17.09367148451006,
      "learning_rate": 1.9762845849802374e-06,
      "loss": 3.5586,
      "step": 5
    },
    {
      "epoch": 0.0019786307874950534,
      "eval_PRM Accuracy": 0.15079365079365079,
      "eval_PRM F1": 0.0,
      "eval_PRM F1 AUC": 0.5,
      "eval_PRM F1 AUC (fixed)": 0.5027053615346778,
      "eval_PRM F1 Neg": 0.2620689655172414,
      "eval_PRM NPV": 0.15079365079365079,
      "eval_PRM Precision": 0.0,
      "eval_PRM Recall": 0.0,
      "eval_PRM Specificty": 1.0,
      "eval_loss": 5.102083206176758,
      "eval_runtime": 5.7709,
      "eval_samples_per_second": 5.198,
      "eval_steps_per_second": 0.173,
      "step": 5
    },
    {
      "epoch": 0.002374356944994064,
      "grad_norm": 17.288014813010893,
      "learning_rate": 2.3715415019762844e-06,
      "loss": 3.4521,
      "step": 6
    },
    {
      "epoch": 0.002770083102493075,
      "grad_norm": 19.783148250791655,
      "learning_rate": 2.7667984189723323e-06,
      "loss": 3.832,
      "step": 7
    },
    {
      "epoch": 0.0031658092599920855,
      "grad_norm": 18.56069488836899,
      "learning_rate": 3.1620553359683794e-06,
      "loss": 4.0889,
      "step": 8
    },
    {
      "epoch": 0.003561535417491096,
      "grad_norm": 18.151830981159222,
      "learning_rate": 3.5573122529644273e-06,
      "loss": 3.5774,
      "step": 9
    },
    {
      "epoch": 0.003957261574990107,
      "grad_norm": 17.60401013651056,
      "learning_rate": 3.952569169960475e-06,
      "loss": 3.5596,
      "step": 10
    },
    {
      "epoch": 0.003957261574990107,
      "eval_PRM Accuracy": 0.15079365079365079,
      "eval_PRM F1": 0.0,
      "eval_PRM F1 AUC": 0.5,
      "eval_PRM F1 AUC (fixed)": 0.5068863748155436,
      "eval_PRM F1 Neg": 0.2620689655172414,
      "eval_PRM NPV": 0.15079365079365079,
      "eval_PRM Precision": 0.0,
      "eval_PRM Recall": 0.0,
      "eval_PRM Specificty": 1.0,
      "eval_loss": 5.022916793823242,
      "eval_runtime": 5.3601,
      "eval_samples_per_second": 5.597,
      "eval_steps_per_second": 0.187,
      "step": 10
    },
    {
      "epoch": 0.004352987732489117,
      "grad_norm": 18.843141225173827,
      "learning_rate": 4.347826086956522e-06,
      "loss": 3.5303,
      "step": 11
    },
    {
      "epoch": 0.004748713889988128,
      "grad_norm": 17.902273627028176,
      "learning_rate": 4.743083003952569e-06,
      "loss": 3.3682,
      "step": 12
    },
    {
      "epoch": 0.0051444400474871385,
      "grad_norm": 19.787789936375873,
      "learning_rate": 5.138339920948617e-06,
      "loss": 3.4863,
      "step": 13
    },
    {
      "epoch": 0.00554016620498615,
      "grad_norm": 22.32404701195085,
      "learning_rate": 5.533596837944665e-06,
      "loss": 3.6895,
      "step": 14
    },
    {
      "epoch": 0.00593589236248516,
      "grad_norm": 20.26194882666582,
      "learning_rate": 5.928853754940711e-06,
      "loss": 3.4512,
      "step": 15
    },
    {
      "epoch": 0.00593589236248516,
      "eval_PRM Accuracy": 0.15079365079365079,
      "eval_PRM F1": 0.0,
      "eval_PRM F1 AUC": 0.5,
      "eval_PRM F1 AUC (fixed)": 0.5049188391539596,
      "eval_PRM F1 Neg": 0.2620689655172414,
      "eval_PRM NPV": 0.15079365079365079,
      "eval_PRM Precision": 0.0,
      "eval_PRM Recall": 0.0,
      "eval_PRM Specificty": 1.0,
      "eval_loss": 4.793749809265137,
      "eval_runtime": 5.2891,
      "eval_samples_per_second": 5.672,
      "eval_steps_per_second": 0.189,
      "step": 15
    },
    {
      "epoch": 0.006331618519984171,
      "grad_norm": 19.77688244477446,
      "learning_rate": 6.324110671936759e-06,
      "loss": 3.2798,
      "step": 16
    },
    {
      "epoch": 0.006727344677483181,
      "grad_norm": 20.673340406580706,
      "learning_rate": 6.719367588932807e-06,
      "loss": 3.1404,
      "step": 17
    },
    {
      "epoch": 0.007123070834982192,
      "grad_norm": 21.981025672661666,
      "learning_rate": 7.1146245059288545e-06,
      "loss": 3.3759,
      "step": 18
    },
    {
      "epoch": 0.007518796992481203,
      "grad_norm": 20.457962505081607,
      "learning_rate": 7.509881422924901e-06,
      "loss": 2.917,
      "step": 19
    },
    {
      "epoch": 0.007914523149980214,
      "grad_norm": 23.374264614038623,
      "learning_rate": 7.90513833992095e-06,
      "loss": 2.894,
      "step": 20
    },
    {
      "epoch": 0.007914523149980214,
      "eval_PRM Accuracy": 0.15079365079365079,
      "eval_PRM F1": 0.0,
      "eval_PRM F1 AUC": 0.5,
      "eval_PRM F1 AUC (fixed)": 0.5002459419576979,
      "eval_PRM F1 Neg": 0.2620689655172414,
      "eval_PRM NPV": 0.15079365079365079,
      "eval_PRM Precision": 0.0,
      "eval_PRM Recall": 0.0,
      "eval_PRM Specificty": 1.0,
      "eval_loss": 4.183333396911621,
      "eval_runtime": 5.212,
      "eval_samples_per_second": 5.756,
      "eval_steps_per_second": 0.192,
      "step": 20
    },
    {
      "epoch": 0.008310249307479225,
      "grad_norm": 20.430795312660443,
      "learning_rate": 8.300395256916998e-06,
      "loss": 2.2808,
      "step": 21
    },
    {
      "epoch": 0.008705975464978234,
      "grad_norm": 18.6760840600575,
      "learning_rate": 8.695652173913044e-06,
      "loss": 1.8269,
      "step": 22
    },
    {
      "epoch": 0.009101701622477245,
      "grad_norm": 20.19597080657976,
      "learning_rate": 9.090909090909091e-06,
      "loss": 1.8501,
      "step": 23
    },
    {
      "epoch": 0.009497427779976256,
      "grad_norm": 19.41223634980227,
      "learning_rate": 9.486166007905138e-06,
      "loss": 1.4681,
      "step": 24
    },
    {
      "epoch": 0.009893153937475268,
      "grad_norm": 15.159492022053273,
      "learning_rate": 9.881422924901186e-06,
      "loss": 1.2933,
      "step": 25
    },
    {
      "epoch": 0.009893153937475268,
      "eval_PRM Accuracy": 0.1984126984126984,
      "eval_PRM F1": 0.13675213675213677,
      "eval_PRM F1 AUC": 0.48475159862272504,
      "eval_PRM F1 AUC (fixed)": 0.47884899163797345,
      "eval_PRM F1 Neg": 0.2518518518518518,
      "eval_PRM NPV": 0.14655172413793102,
      "eval_PRM Precision": 0.8,
      "eval_PRM Recall": 0.07476635514018691,
      "eval_PRM Specificty": 0.8947368421052632,
      "eval_loss": 2.5885417461395264,
      "eval_runtime": 5.0187,
      "eval_samples_per_second": 5.978,
      "eval_steps_per_second": 0.199,
      "step": 25
    },
    {
      "epoch": 0.010288880094974277,
      "grad_norm": 11.195465424334387,
      "learning_rate": 1.0276679841897234e-05,
      "loss": 0.8041,
      "step": 26
    },
    {
      "epoch": 0.010684606252473288,
      "grad_norm": 6.156751897014624,
      "learning_rate": 1.0671936758893281e-05,
      "loss": 0.6689,
      "step": 27
    },
    {
      "epoch": 0.0110803324099723,
      "grad_norm": 3.7995404914247657,
      "learning_rate": 1.106719367588933e-05,
      "loss": 0.5178,
      "step": 28
    },
    {
      "epoch": 0.01147605856747131,
      "grad_norm": 3.174129656529322,
      "learning_rate": 1.1462450592885376e-05,
      "loss": 0.5728,
      "step": 29
    },
    {
      "epoch": 0.01187178472497032,
      "grad_norm": 4.941466575923112,
      "learning_rate": 1.1857707509881423e-05,
      "loss": 0.5383,
      "step": 30
    },
    {
      "epoch": 0.01187178472497032,
      "eval_PRM Accuracy": 0.6428571428571429,
      "eval_PRM F1": 0.782608695652174,
      "eval_PRM F1 AUC": 0.37850467289719625,
      "eval_PRM F1 AUC (fixed)": 0.45917363502213476,
      "eval_PRM F1 Neg": 0.0,
      "eval_PRM NPV": 0.0,
      "eval_PRM Precision": 0.81,
      "eval_PRM Recall": 0.7570093457943925,
      "eval_PRM Specificty": 0.0,
      "eval_loss": 1.1531250476837158,
      "eval_runtime": 5.2389,
      "eval_samples_per_second": 5.726,
      "eval_steps_per_second": 0.191,
      "step": 30
    },
    {
      "epoch": 0.01226751088246933,
      "grad_norm": 1.7533835915646967,
      "learning_rate": 1.225296442687747e-05,
      "loss": 0.3236,
      "step": 31
    },
    {
      "epoch": 0.012663237039968342,
      "grad_norm": 4.8618928869998665,
      "learning_rate": 1.2648221343873517e-05,
      "loss": 0.6709,
      "step": 32
    },
    {
      "epoch": 0.013058963197467353,
      "grad_norm": 5.833727450640927,
      "learning_rate": 1.3043478260869566e-05,
      "loss": 0.7535,
      "step": 33
    },
    {
      "epoch": 0.013454689354966362,
      "grad_norm": 5.013599184272929,
      "learning_rate": 1.3438735177865614e-05,
      "loss": 0.5645,
      "step": 34
    },
    {
      "epoch": 0.013850415512465374,
      "grad_norm": 3.841617194736358,
      "learning_rate": 1.383399209486166e-05,
      "loss": 0.4477,
      "step": 35
    },
    {
      "epoch": 0.013850415512465374,
      "eval_PRM Accuracy": 0.7222222222222222,
      "eval_PRM F1": 0.8387096774193549,
      "eval_PRM F1 AUC": 0.4252336448598131,
      "eval_PRM F1 AUC (fixed)": 0.4899163797343827,
      "eval_PRM F1 Neg": 0.0,
      "eval_PRM NPV": 0.0,
      "eval_PRM Precision": 0.8272727272727273,
      "eval_PRM Recall": 0.8504672897196262,
      "eval_PRM Specificty": 0.0,
      "eval_loss": 0.8377603888511658,
      "eval_runtime": 5.1008,
      "eval_samples_per_second": 5.881,
      "eval_steps_per_second": 0.196,
      "step": 35
    },
    {
      "epoch": 0.014246141669964385,
      "grad_norm": 2.9464506470041365,
      "learning_rate": 1.4229249011857709e-05,
      "loss": 0.3883,
      "step": 36
    },
    {
      "epoch": 0.014641867827463396,
      "grad_norm": 2.3395513500376905,
      "learning_rate": 1.4624505928853754e-05,
      "loss": 0.3986,
      "step": 37
    },
    {
      "epoch": 0.015037593984962405,
      "grad_norm": 1.4167566579602888,
      "learning_rate": 1.5019762845849802e-05,
      "loss": 0.3546,
      "step": 38
    },
    {
      "epoch": 0.015433320142461416,
      "grad_norm": 2.0367146336429593,
      "learning_rate": 1.541501976284585e-05,
      "loss": 0.478,
      "step": 39
    },
    {
      "epoch": 0.015829046299960427,
      "grad_norm": 4.047508246077786,
      "learning_rate": 1.58102766798419e-05,
      "loss": 0.3996,
      "step": 40
    },
    {
      "epoch": 0.015829046299960427,
      "eval_PRM Accuracy": 0.6349206349206349,
      "eval_PRM F1": 0.7745098039215687,
      "eval_PRM F1 AUC": 0.3954746679783571,
      "eval_PRM F1 AUC (fixed)": 0.5243482538121004,
      "eval_PRM F1 Neg": 0.041666666666666664,
      "eval_PRM NPV": 0.034482758620689655,
      "eval_PRM Precision": 0.8144329896907216,
      "eval_PRM Recall": 0.7383177570093458,
      "eval_PRM Specificty": 0.05263157894736842,
      "eval_loss": 0.7927083373069763,
      "eval_runtime": 5.3924,
      "eval_samples_per_second": 5.563,
      "eval_steps_per_second": 0.185,
      "step": 40
    },
    {
      "epoch": 0.016224772457459437,
      "grad_norm": 3.3239278410198474,
      "learning_rate": 1.6205533596837947e-05,
      "loss": 0.5471,
      "step": 41
    },
    {
      "epoch": 0.01662049861495845,
      "grad_norm": 4.355448382606326,
      "learning_rate": 1.6600790513833996e-05,
      "loss": 0.4165,
      "step": 42
    },
    {
      "epoch": 0.01701622477245746,
      "grad_norm": 4.730438132075204,
      "learning_rate": 1.699604743083004e-05,
      "loss": 0.4449,
      "step": 43
    },
    {
      "epoch": 0.01741195092995647,
      "grad_norm": 2.466837950012606,
      "learning_rate": 1.739130434782609e-05,
      "loss": 0.4586,
      "step": 44
    },
    {
      "epoch": 0.01780767708745548,
      "grad_norm": 2.3655525165338935,
      "learning_rate": 1.7786561264822134e-05,
      "loss": 0.3229,
      "step": 45
    },
    {
      "epoch": 0.01780767708745548,
      "eval_PRM Accuracy": 0.7222222222222222,
      "eval_PRM F1": 0.8372093023255814,
      "eval_PRM F1 AUC": 0.44687653713723563,
      "eval_PRM F1 AUC (fixed)": 0.5553369404820462,
      "eval_PRM F1 Neg": 0.05405405405405406,
      "eval_PRM NPV": 0.05555555555555555,
      "eval_PRM Precision": 0.8333333333333334,
      "eval_PRM Recall": 0.8411214953271028,
      "eval_PRM Specificty": 0.05263157894736842,
      "eval_loss": 0.6333333253860474,
      "eval_runtime": 5.1383,
      "eval_samples_per_second": 5.838,
      "eval_steps_per_second": 0.195,
      "step": 45
    },
    {
      "epoch": 0.01820340324495449,
      "grad_norm": 1.9000119738790562,
      "learning_rate": 1.8181818181818182e-05,
      "loss": 0.2971,
      "step": 46
    },
    {
      "epoch": 0.018599129402453504,
      "grad_norm": 1.0739447125176689,
      "learning_rate": 1.857707509881423e-05,
      "loss": 0.3701,
      "step": 47
    },
    {
      "epoch": 0.018994855559952513,
      "grad_norm": 1.0930446768487763,
      "learning_rate": 1.8972332015810275e-05,
      "loss": 0.2685,
      "step": 48
    },
    {
      "epoch": 0.019390581717451522,
      "grad_norm": 2.286243908202953,
      "learning_rate": 1.9367588932806324e-05,
      "loss": 0.4901,
      "step": 49
    },
    {
      "epoch": 0.019786307874950535,
      "grad_norm": 1.820785279394848,
      "learning_rate": 1.9762845849802372e-05,
      "loss": 0.3174,
      "step": 50
    },
    {
      "epoch": 0.019786307874950535,
      "eval_PRM Accuracy": 0.8333333333333334,
      "eval_PRM F1": 0.9090909090909091,
      "eval_PRM F1 AUC": 0.49065420560747663,
      "eval_PRM F1 AUC (fixed)": 0.6089522872602066,
      "eval_PRM F1 Neg": 0.0,
      "eval_PRM NPV": 0.0,
      "eval_PRM Precision": 0.8467741935483871,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.0,
      "eval_loss": 0.5188801884651184,
      "eval_runtime": 5.1873,
      "eval_samples_per_second": 5.783,
      "eval_steps_per_second": 0.193,
      "step": 50
    },
    {
      "epoch": 0.020182034032449545,
      "grad_norm": 2.116954246335264,
      "learning_rate": 2.015810276679842e-05,
      "loss": 0.3754,
      "step": 51
    },
    {
      "epoch": 0.020577760189948554,
      "grad_norm": 1.5562673959829776,
      "learning_rate": 2.055335968379447e-05,
      "loss": 0.3152,
      "step": 52
    },
    {
      "epoch": 0.020973486347447567,
      "grad_norm": 2.0021756500849315,
      "learning_rate": 2.0948616600790517e-05,
      "loss": 0.3499,
      "step": 53
    },
    {
      "epoch": 0.021369212504946576,
      "grad_norm": 1.3364495175327136,
      "learning_rate": 2.1343873517786562e-05,
      "loss": 0.387,
      "step": 54
    },
    {
      "epoch": 0.02176493866244559,
      "grad_norm": 1.570371776127056,
      "learning_rate": 2.173913043478261e-05,
      "loss": 0.3652,
      "step": 55
    },
    {
      "epoch": 0.02176493866244559,
      "eval_PRM Accuracy": 0.8412698412698413,
      "eval_PRM F1": 0.9130434782608695,
      "eval_PRM F1 AUC": 0.5169699950811608,
      "eval_PRM F1 AUC (fixed)": 0.6328086571569109,
      "eval_PRM F1 Neg": 0.09090909090909091,
      "eval_PRM NPV": 0.3333333333333333,
      "eval_PRM Precision": 0.8536585365853658,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.05263157894736842,
      "eval_loss": 0.4619791805744171,
      "eval_runtime": 5.1608,
      "eval_samples_per_second": 5.813,
      "eval_steps_per_second": 0.194,
      "step": 55
    },
    {
      "epoch": 0.0221606648199446,
      "grad_norm": 0.7617945507743419,
      "learning_rate": 2.213438735177866e-05,
      "loss": 0.3217,
      "step": 56
    },
    {
      "epoch": 0.022556390977443608,
      "grad_norm": 1.524009595673118,
      "learning_rate": 2.2529644268774703e-05,
      "loss": 0.3338,
      "step": 57
    },
    {
      "epoch": 0.02295211713494262,
      "grad_norm": 1.3376491978579645,
      "learning_rate": 2.2924901185770752e-05,
      "loss": 0.352,
      "step": 58
    },
    {
      "epoch": 0.02334784329244163,
      "grad_norm": 1.2102881704805017,
      "learning_rate": 2.33201581027668e-05,
      "loss": 0.3458,
      "step": 59
    },
    {
      "epoch": 0.02374356944994064,
      "grad_norm": 1.3015321149253491,
      "learning_rate": 2.3715415019762845e-05,
      "loss": 0.4152,
      "step": 60
    },
    {
      "epoch": 0.02374356944994064,
      "eval_PRM Accuracy": 0.8412698412698413,
      "eval_PRM F1": 0.9130434782608695,
      "eval_PRM F1 AUC": 0.5169699950811608,
      "eval_PRM F1 AUC (fixed)": 0.6519921298573536,
      "eval_PRM F1 Neg": 0.09090909090909091,
      "eval_PRM NPV": 0.3333333333333333,
      "eval_PRM Precision": 0.8536585365853658,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.05263157894736842,
      "eval_loss": 0.4494791626930237,
      "eval_runtime": 5.1859,
      "eval_samples_per_second": 5.785,
      "eval_steps_per_second": 0.193,
      "step": 60
    },
    {
      "epoch": 0.024139295607439652,
      "grad_norm": 2.870493246763942,
      "learning_rate": 2.4110671936758893e-05,
      "loss": 0.3011,
      "step": 61
    },
    {
      "epoch": 0.02453502176493866,
      "grad_norm": 1.8638626392658066,
      "learning_rate": 2.450592885375494e-05,
      "loss": 0.3317,
      "step": 62
    },
    {
      "epoch": 0.024930747922437674,
      "grad_norm": 0.8756457676198278,
      "learning_rate": 2.490118577075099e-05,
      "loss": 0.3231,
      "step": 63
    },
    {
      "epoch": 0.025326474079936684,
      "grad_norm": 1.089305270497872,
      "learning_rate": 2.5296442687747035e-05,
      "loss": 0.3354,
      "step": 64
    },
    {
      "epoch": 0.025722200237435693,
      "grad_norm": 2.5979890390130764,
      "learning_rate": 2.5691699604743087e-05,
      "loss": 0.4165,
      "step": 65
    },
    {
      "epoch": 0.025722200237435693,
      "eval_PRM Accuracy": 0.8571428571428571,
      "eval_PRM F1": 0.9224137931034483,
      "eval_PRM F1 AUC": 0.5263157894736843,
      "eval_PRM F1 AUC (fixed)": 0.6874077717658633,
      "eval_PRM F1 Neg": 0.1,
      "eval_PRM NPV": 1.0,
      "eval_PRM Precision": 0.856,
      "eval_PRM Recall": 1.0,
      "eval_PRM Specificty": 0.05263157894736842,
      "eval_loss": 0.4258463680744171,
      "eval_runtime": 5.0916,
      "eval_samples_per_second": 5.892,
      "eval_steps_per_second": 0.196,
      "step": 65
    },
    {
      "epoch": 0.026117926394934706,
      "grad_norm": 2.2676484289457775,
      "learning_rate": 2.608695652173913e-05,
      "loss": 0.3829,
      "step": 66
    },
    {
      "epoch": 0.026513652552433715,
      "grad_norm": 1.480591503241869,
      "learning_rate": 2.6482213438735183e-05,
      "loss": 0.3344,
      "step": 67
    },
    {
      "epoch": 0.026909378709932725,
      "grad_norm": 1.9032369881350584,
      "learning_rate": 2.6877470355731228e-05,
      "loss": 0.4024,
      "step": 68
    },
    {
      "epoch": 0.027305104867431738,
      "grad_norm": 0.7853162517569824,
      "learning_rate": 2.7272727272727273e-05,
      "loss": 0.2465,
      "step": 69
    },
    {
      "epoch": 0.027700831024930747,
      "grad_norm": 0.9899304740153296,
      "learning_rate": 2.766798418972332e-05,
      "loss": 0.2842,
      "step": 70
    },
    {
      "epoch": 0.027700831024930747,
      "eval_PRM Accuracy": 0.8492063492063492,
      "eval_PRM F1": 0.9177489177489178,
      "eval_PRM F1 AUC": 0.5216428922774226,
      "eval_PRM F1 AUC (fixed)": 0.691342843089031,
      "eval_PRM F1 Neg": 0.09523809523809523,
      "eval_PRM NPV": 0.5,
      "eval_PRM Precision": 0.8548387096774194,
      "eval_PRM Recall": 0.9906542056074766,
      "eval_PRM Specificty": 0.05263157894736842,
      "eval_loss": 0.41523438692092896,
      "eval_runtime": 5.2108,
      "eval_samples_per_second": 5.757,
      "eval_steps_per_second": 0.192,
      "step": 70
    },
    {
      "epoch": 0.02809655718242976,
      "grad_norm": 1.0455841735483649,
      "learning_rate": 2.8063241106719366e-05,
      "loss": 0.3616,
      "step": 71
    },
    {
      "epoch": 0.02849228333992877,
      "grad_norm": 1.3692130591804181,
      "learning_rate": 2.8458498023715418e-05,
      "loss": 0.3801,
      "step": 72
    },
    {
      "epoch": 0.02888800949742778,
      "grad_norm": 2.6586144803573806,
      "learning_rate": 2.8853754940711463e-05,
      "loss": 0.3188,
      "step": 73
    },
    {
      "epoch": 0.02928373565492679,
      "grad_norm": 1.7817113798850117,
      "learning_rate": 2.9249011857707508e-05,
      "loss": 0.4109,
      "step": 74
    },
    {
      "epoch": 0.0296794618124258,
      "grad_norm": 3.2318845283795654,
      "learning_rate": 2.964426877470356e-05,
      "loss": 0.3106,
      "step": 75
    },
    {
      "epoch": 0.0296794618124258,
      "eval_PRM Accuracy": 0.8412698412698413,
      "eval_PRM F1": 0.9130434782608695,
      "eval_PRM F1 AUC": 0.5169699950811608,
      "eval_PRM F1 AUC (fixed)": 0.6974913920314806,
      "eval_PRM F1 Neg": 0.09090909090909091,
      "eval_PRM NPV": 0.3333333333333333,
      "eval_PRM Precision": 0.8536585365853658,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.05263157894736842,
      "eval_loss": 0.4033854305744171,
      "eval_runtime": 5.2736,
      "eval_samples_per_second": 5.689,
      "eval_steps_per_second": 0.19,
      "step": 75
    },
    {
      "epoch": 0.03007518796992481,
      "grad_norm": 1.7909426411134697,
      "learning_rate": 3.0039525691699605e-05,
      "loss": 0.3859,
      "step": 76
    },
    {
      "epoch": 0.030470914127423823,
      "grad_norm": 1.3774726761743339,
      "learning_rate": 3.0434782608695656e-05,
      "loss": 0.2687,
      "step": 77
    },
    {
      "epoch": 0.030866640284922833,
      "grad_norm": 0.9814212497780863,
      "learning_rate": 3.08300395256917e-05,
      "loss": 0.2458,
      "step": 78
    },
    {
      "epoch": 0.03126236644242184,
      "grad_norm": 2.711898808984566,
      "learning_rate": 3.1225296442687746e-05,
      "loss": 0.3655,
      "step": 79
    },
    {
      "epoch": 0.031658092599920855,
      "grad_norm": 2.180725341318851,
      "learning_rate": 3.16205533596838e-05,
      "loss": 0.294,
      "step": 80
    },
    {
      "epoch": 0.031658092599920855,
      "eval_PRM Accuracy": 0.8571428571428571,
      "eval_PRM F1": 0.9224137931034483,
      "eval_PRM F1 AUC": 0.5263157894736843,
      "eval_PRM F1 AUC (fixed)": 0.7120019675356615,
      "eval_PRM F1 Neg": 0.1,
      "eval_PRM NPV": 1.0,
      "eval_PRM Precision": 0.856,
      "eval_PRM Recall": 1.0,
      "eval_PRM Specificty": 0.05263157894736842,
      "eval_loss": 0.42337238788604736,
      "eval_runtime": 5.2813,
      "eval_samples_per_second": 5.68,
      "eval_steps_per_second": 0.189,
      "step": 80
    },
    {
      "epoch": 0.03205381875741987,
      "grad_norm": 1.730652370020414,
      "learning_rate": 3.201581027667984e-05,
      "loss": 0.2952,
      "step": 81
    },
    {
      "epoch": 0.032449544914918874,
      "grad_norm": 2.9913050569584634,
      "learning_rate": 3.2411067193675894e-05,
      "loss": 0.4053,
      "step": 82
    },
    {
      "epoch": 0.032845271072417886,
      "grad_norm": 1.0609369534237518,
      "learning_rate": 3.280632411067194e-05,
      "loss": 0.237,
      "step": 83
    },
    {
      "epoch": 0.0332409972299169,
      "grad_norm": 1.504838778747724,
      "learning_rate": 3.320158102766799e-05,
      "loss": 0.3227,
      "step": 84
    },
    {
      "epoch": 0.033636723387415905,
      "grad_norm": 1.5043466193721529,
      "learning_rate": 3.3596837944664036e-05,
      "loss": 0.3158,
      "step": 85
    },
    {
      "epoch": 0.033636723387415905,
      "eval_PRM Accuracy": 0.8492063492063492,
      "eval_PRM F1": 0.9177489177489178,
      "eval_PRM F1 AUC": 0.5216428922774226,
      "eval_PRM F1 AUC (fixed)": 0.7274963108706344,
      "eval_PRM F1 Neg": 0.09523809523809523,
      "eval_PRM NPV": 0.5,
      "eval_PRM Precision": 0.8548387096774194,
      "eval_PRM Recall": 0.9906542056074766,
      "eval_PRM Specificty": 0.05263157894736842,
      "eval_loss": 0.3841145932674408,
      "eval_runtime": 5.1431,
      "eval_samples_per_second": 5.833,
      "eval_steps_per_second": 0.194,
      "step": 85
    },
    {
      "epoch": 0.03403244954491492,
      "grad_norm": 3.929722829050404,
      "learning_rate": 3.399209486166008e-05,
      "loss": 0.2845,
      "step": 86
    },
    {
      "epoch": 0.03442817570241393,
      "grad_norm": 3.103114457537566,
      "learning_rate": 3.438735177865613e-05,
      "loss": 0.3327,
      "step": 87
    },
    {
      "epoch": 0.03482390185991294,
      "grad_norm": 2.6349915618562707,
      "learning_rate": 3.478260869565218e-05,
      "loss": 0.2698,
      "step": 88
    },
    {
      "epoch": 0.03521962801741195,
      "grad_norm": 1.7545206935775584,
      "learning_rate": 3.517786561264822e-05,
      "loss": 0.2223,
      "step": 89
    },
    {
      "epoch": 0.03561535417491096,
      "grad_norm": 3.266788647502398,
      "learning_rate": 3.557312252964427e-05,
      "loss": 0.4256,
      "step": 90
    },
    {
      "epoch": 0.03561535417491096,
      "eval_PRM Accuracy": 0.8571428571428571,
      "eval_PRM F1": 0.9224137931034483,
      "eval_PRM F1 AUC": 0.5263157894736843,
      "eval_PRM F1 AUC (fixed)": 0.7356123954746678,
      "eval_PRM F1 Neg": 0.1,
      "eval_PRM NPV": 1.0,
      "eval_PRM Precision": 0.856,
      "eval_PRM Recall": 1.0,
      "eval_PRM Specificty": 0.05263157894736842,
      "eval_loss": 0.42851561307907104,
      "eval_runtime": 4.9755,
      "eval_samples_per_second": 6.03,
      "eval_steps_per_second": 0.201,
      "step": 90
    },
    {
      "epoch": 0.036011080332409975,
      "grad_norm": 1.027944848557743,
      "learning_rate": 3.596837944664031e-05,
      "loss": 0.1986,
      "step": 91
    },
    {
      "epoch": 0.03640680648990898,
      "grad_norm": 4.277804218438809,
      "learning_rate": 3.6363636363636364e-05,
      "loss": 0.4712,
      "step": 92
    },
    {
      "epoch": 0.036802532647407994,
      "grad_norm": 2.635565838869095,
      "learning_rate": 3.675889328063241e-05,
      "loss": 0.3304,
      "step": 93
    },
    {
      "epoch": 0.03719825880490701,
      "grad_norm": 2.8178407973377277,
      "learning_rate": 3.715415019762846e-05,
      "loss": 0.3158,
      "step": 94
    },
    {
      "epoch": 0.03759398496240601,
      "grad_norm": 0.8948854518829017,
      "learning_rate": 3.7549407114624506e-05,
      "loss": 0.1175,
      "step": 95
    },
    {
      "epoch": 0.03759398496240601,
      "eval_PRM Accuracy": 0.8571428571428571,
      "eval_PRM F1": 0.9224137931034483,
      "eval_PRM F1 AUC": 0.5263157894736843,
      "eval_PRM F1 AUC (fixed)": 0.7385636989670438,
      "eval_PRM F1 Neg": 0.1,
      "eval_PRM NPV": 1.0,
      "eval_PRM Precision": 0.856,
      "eval_PRM Recall": 1.0,
      "eval_PRM Specificty": 0.05263157894736842,
      "eval_loss": 0.3970703184604645,
      "eval_runtime": 5.0611,
      "eval_samples_per_second": 5.928,
      "eval_steps_per_second": 0.198,
      "step": 95
    },
    {
      "epoch": 0.037989711119905026,
      "grad_norm": 1.5523517867148884,
      "learning_rate": 3.794466403162055e-05,
      "loss": 0.2827,
      "step": 96
    },
    {
      "epoch": 0.03838543727740404,
      "grad_norm": 1.2316878136884037,
      "learning_rate": 3.83399209486166e-05,
      "loss": 0.2935,
      "step": 97
    },
    {
      "epoch": 0.038781163434903045,
      "grad_norm": 1.6115305122370436,
      "learning_rate": 3.873517786561265e-05,
      "loss": 0.1908,
      "step": 98
    },
    {
      "epoch": 0.03917688959240206,
      "grad_norm": 1.318558125157373,
      "learning_rate": 3.91304347826087e-05,
      "loss": 0.291,
      "step": 99
    },
    {
      "epoch": 0.03957261574990107,
      "grad_norm": 2.6638735395301287,
      "learning_rate": 3.9525691699604744e-05,
      "loss": 0.2182,
      "step": 100
    },
    {
      "epoch": 0.03957261574990107,
      "eval_PRM Accuracy": 0.8571428571428571,
      "eval_PRM F1": 0.9217391304347826,
      "eval_PRM F1 AUC": 0.5479586817511067,
      "eval_PRM F1 AUC (fixed)": 0.7515986227250369,
      "eval_PRM F1 Neg": 0.18181818181818182,
      "eval_PRM NPV": 0.6666666666666666,
      "eval_PRM Precision": 0.8617886178861789,
      "eval_PRM Recall": 0.9906542056074766,
      "eval_PRM Specificty": 0.10526315789473684,
      "eval_loss": 0.38951823115348816,
      "eval_runtime": 5.1666,
      "eval_samples_per_second": 5.807,
      "eval_steps_per_second": 0.194,
      "step": 100
    },
    {
      "epoch": 0.039968341907400076,
      "grad_norm": 0.9643087620304042,
      "learning_rate": 3.9920948616600796e-05,
      "loss": 0.1804,
      "step": 101
    },
    {
      "epoch": 0.04036406806489909,
      "grad_norm": 1.613929197330765,
      "learning_rate": 4.031620553359684e-05,
      "loss": 0.2626,
      "step": 102
    },
    {
      "epoch": 0.0407597942223981,
      "grad_norm": 2.955399791178141,
      "learning_rate": 4.0711462450592886e-05,
      "loss": 0.3522,
      "step": 103
    },
    {
      "epoch": 0.04115552037989711,
      "grad_norm": 1.6852133078387364,
      "learning_rate": 4.110671936758894e-05,
      "loss": 0.2877,
      "step": 104
    },
    {
      "epoch": 0.04155124653739612,
      "grad_norm": 1.7499190920574095,
      "learning_rate": 4.150197628458498e-05,
      "loss": 0.2952,
      "step": 105
    },
    {
      "epoch": 0.04155124653739612,
      "eval_PRM Accuracy": 0.8571428571428571,
      "eval_PRM F1": 0.9217391304347826,
      "eval_PRM F1 AUC": 0.5479586817511067,
      "eval_PRM F1 AUC (fixed)": 0.7803738317757009,
      "eval_PRM F1 Neg": 0.18181818181818182,
      "eval_PRM NPV": 0.6666666666666666,
      "eval_PRM Precision": 0.8617886178861789,
      "eval_PRM Recall": 0.9906542056074766,
      "eval_PRM Specificty": 0.10526315789473684,
      "eval_loss": 0.3997395932674408,
      "eval_runtime": 5.0138,
      "eval_samples_per_second": 5.983,
      "eval_steps_per_second": 0.199,
      "step": 105
    },
    {
      "epoch": 0.041946972694895134,
      "grad_norm": 2.261535997190661,
      "learning_rate": 4.1897233201581034e-05,
      "loss": 0.3005,
      "step": 106
    },
    {
      "epoch": 0.042342698852394146,
      "grad_norm": 2.451377333545837,
      "learning_rate": 4.229249011857708e-05,
      "loss": 0.2937,
      "step": 107
    },
    {
      "epoch": 0.04273842500989315,
      "grad_norm": 1.6329569247243219,
      "learning_rate": 4.2687747035573124e-05,
      "loss": 0.2243,
      "step": 108
    },
    {
      "epoch": 0.043134151167392165,
      "grad_norm": 1.3271824854083534,
      "learning_rate": 4.3083003952569175e-05,
      "loss": 0.2728,
      "step": 109
    },
    {
      "epoch": 0.04352987732489118,
      "grad_norm": 1.7152258452977818,
      "learning_rate": 4.347826086956522e-05,
      "loss": 0.3356,
      "step": 110
    },
    {
      "epoch": 0.04352987732489118,
      "eval_PRM Accuracy": 0.8571428571428571,
      "eval_PRM F1": 0.9217391304347826,
      "eval_PRM F1 AUC": 0.5479586817511067,
      "eval_PRM F1 AUC (fixed)": 0.794638465322184,
      "eval_PRM F1 Neg": 0.18181818181818182,
      "eval_PRM NPV": 0.6666666666666666,
      "eval_PRM Precision": 0.8617886178861789,
      "eval_PRM Recall": 0.9906542056074766,
      "eval_PRM Specificty": 0.10526315789473684,
      "eval_loss": 0.4004557430744171,
      "eval_runtime": 5.2442,
      "eval_samples_per_second": 5.721,
      "eval_steps_per_second": 0.191,
      "step": 110
    },
    {
      "epoch": 0.043925603482390184,
      "grad_norm": 1.699776281060092,
      "learning_rate": 4.387351778656127e-05,
      "loss": 0.2784,
      "step": 111
    },
    {
      "epoch": 0.0443213296398892,
      "grad_norm": 1.4948043135883748,
      "learning_rate": 4.426877470355732e-05,
      "loss": 0.2808,
      "step": 112
    },
    {
      "epoch": 0.04471705579738821,
      "grad_norm": 1.9758117190865245,
      "learning_rate": 4.466403162055336e-05,
      "loss": 0.3023,
      "step": 113
    },
    {
      "epoch": 0.045112781954887216,
      "grad_norm": 3.454023285461921,
      "learning_rate": 4.505928853754941e-05,
      "loss": 0.2808,
      "step": 114
    },
    {
      "epoch": 0.04550850811238623,
      "grad_norm": 1.9989238541236272,
      "learning_rate": 4.545454545454546e-05,
      "loss": 0.1921,
      "step": 115
    },
    {
      "epoch": 0.04550850811238623,
      "eval_PRM Accuracy": 0.8571428571428571,
      "eval_PRM F1": 0.9217391304347826,
      "eval_PRM F1 AUC": 0.5479586817511067,
      "eval_PRM F1 AUC (fixed)": 0.7889818002951302,
      "eval_PRM F1 Neg": 0.18181818181818182,
      "eval_PRM NPV": 0.6666666666666666,
      "eval_PRM Precision": 0.8617886178861789,
      "eval_PRM Recall": 0.9906542056074766,
      "eval_PRM Specificty": 0.10526315789473684,
      "eval_loss": 0.3909505307674408,
      "eval_runtime": 5.1055,
      "eval_samples_per_second": 5.876,
      "eval_steps_per_second": 0.196,
      "step": 115
    },
    {
      "epoch": 0.04590423426988524,
      "grad_norm": 1.096742114098721,
      "learning_rate": 4.5849802371541504e-05,
      "loss": 0.2413,
      "step": 116
    },
    {
      "epoch": 0.04629996042738425,
      "grad_norm": 2.2713661239067036,
      "learning_rate": 4.624505928853755e-05,
      "loss": 0.3195,
      "step": 117
    },
    {
      "epoch": 0.04669568658488326,
      "grad_norm": 1.5536702621287604,
      "learning_rate": 4.66403162055336e-05,
      "loss": 0.2578,
      "step": 118
    },
    {
      "epoch": 0.04709141274238227,
      "grad_norm": 1.0218391497773849,
      "learning_rate": 4.7035573122529645e-05,
      "loss": 0.209,
      "step": 119
    },
    {
      "epoch": 0.04748713889988128,
      "grad_norm": 1.5213079067906112,
      "learning_rate": 4.743083003952569e-05,
      "loss": 0.2915,
      "step": 120
    },
    {
      "epoch": 0.04748713889988128,
      "eval_PRM Accuracy": 0.8571428571428571,
      "eval_PRM F1": 0.9203539823008849,
      "eval_PRM F1 AUC": 0.5912444663059518,
      "eval_PRM F1 AUC (fixed)": 0.7953762911952778,
      "eval_PRM F1 Neg": 0.3076923076923077,
      "eval_PRM NPV": 0.5714285714285714,
      "eval_PRM Precision": 0.8739495798319328,
      "eval_PRM Recall": 0.9719626168224299,
      "eval_PRM Specificty": 0.21052631578947367,
      "eval_loss": 0.35664063692092896,
      "eval_runtime": 5.182,
      "eval_samples_per_second": 5.789,
      "eval_steps_per_second": 0.193,
      "step": 120
    },
    {
      "epoch": 0.04788286505738029,
      "grad_norm": 2.125450802219854,
      "learning_rate": 4.782608695652174e-05,
      "loss": 0.3243,
      "step": 121
    },
    {
      "epoch": 0.048278591214879304,
      "grad_norm": 1.6091177857858354,
      "learning_rate": 4.822134387351779e-05,
      "loss": 0.2841,
      "step": 122
    },
    {
      "epoch": 0.04867431737237832,
      "grad_norm": 2.8232863051515986,
      "learning_rate": 4.861660079051384e-05,
      "loss": 0.3283,
      "step": 123
    },
    {
      "epoch": 0.04907004352987732,
      "grad_norm": 1.3279890401688152,
      "learning_rate": 4.901185770750988e-05,
      "loss": 0.2414,
      "step": 124
    },
    {
      "epoch": 0.049465769687376336,
      "grad_norm": 1.4678986924745383,
      "learning_rate": 4.940711462450593e-05,
      "loss": 0.2589,
      "step": 125
    },
    {
      "epoch": 0.049465769687376336,
      "eval_PRM Accuracy": 0.8571428571428571,
      "eval_PRM F1": 0.9224137931034483,
      "eval_PRM F1 AUC": 0.5263157894736843,
      "eval_PRM F1 AUC (fixed)": 0.8212001967535659,
      "eval_PRM F1 Neg": 0.1,
      "eval_PRM NPV": 1.0,
      "eval_PRM Precision": 0.856,
      "eval_PRM Recall": 1.0,
      "eval_PRM Specificty": 0.05263157894736842,
      "eval_loss": 0.4049479067325592,
      "eval_runtime": 5.4146,
      "eval_samples_per_second": 5.541,
      "eval_steps_per_second": 0.185,
      "step": 125
    },
    {
      "epoch": 0.04986149584487535,
      "grad_norm": 1.1559564295941434,
      "learning_rate": 4.980237154150198e-05,
      "loss": 0.2222,
      "step": 126
    },
    {
      "epoch": 0.050257222002374355,
      "grad_norm": 1.9597752126020416,
      "learning_rate": 5.0197628458498025e-05,
      "loss": 0.2693,
      "step": 127
    },
    {
      "epoch": 0.05065294815987337,
      "grad_norm": 1.3689526731616795,
      "learning_rate": 5.059288537549407e-05,
      "loss": 0.2213,
      "step": 128
    },
    {
      "epoch": 0.05104867431737238,
      "grad_norm": 1.1014604418337837,
      "learning_rate": 5.098814229249013e-05,
      "loss": 0.255,
      "step": 129
    },
    {
      "epoch": 0.051444400474871387,
      "grad_norm": 1.1479733100742033,
      "learning_rate": 5.138339920948617e-05,
      "loss": 0.204,
      "step": 130
    },
    {
      "epoch": 0.051444400474871387,
      "eval_PRM Accuracy": 0.8650793650793651,
      "eval_PRM F1": 0.9251101321585903,
      "eval_PRM F1 AUC": 0.5959173635022135,
      "eval_PRM F1 AUC (fixed)": 0.8175110673880963,
      "eval_PRM F1 Neg": 0.32,
      "eval_PRM NPV": 0.6666666666666666,
      "eval_PRM Precision": 0.875,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.21052631578947367,
      "eval_loss": 0.34192708134651184,
      "eval_runtime": 5.1073,
      "eval_samples_per_second": 5.874,
      "eval_steps_per_second": 0.196,
      "step": 130
    },
    {
      "epoch": 0.0518401266323704,
      "grad_norm": 1.4006703561222118,
      "learning_rate": 5.177865612648222e-05,
      "loss": 0.2168,
      "step": 131
    },
    {
      "epoch": 0.05223585278986941,
      "grad_norm": 1.4684150660483872,
      "learning_rate": 5.217391304347826e-05,
      "loss": 0.243,
      "step": 132
    },
    {
      "epoch": 0.05263157894736842,
      "grad_norm": 1.5573221708921128,
      "learning_rate": 5.256916996047431e-05,
      "loss": 0.1855,
      "step": 133
    },
    {
      "epoch": 0.05302730510486743,
      "grad_norm": 1.1398886453802985,
      "learning_rate": 5.2964426877470366e-05,
      "loss": 0.2977,
      "step": 134
    },
    {
      "epoch": 0.053423031262366444,
      "grad_norm": 2.9618738605058224,
      "learning_rate": 5.335968379446641e-05,
      "loss": 0.2886,
      "step": 135
    },
    {
      "epoch": 0.053423031262366444,
      "eval_PRM Accuracy": 0.8571428571428571,
      "eval_PRM F1": 0.9217391304347826,
      "eval_PRM F1 AUC": 0.5479586817511067,
      "eval_PRM F1 AUC (fixed)": 0.8123462862764388,
      "eval_PRM F1 Neg": 0.18181818181818182,
      "eval_PRM NPV": 0.6666666666666666,
      "eval_PRM Precision": 0.8617886178861789,
      "eval_PRM Recall": 0.9906542056074766,
      "eval_PRM Specificty": 0.10526315789473684,
      "eval_loss": 0.4007161557674408,
      "eval_runtime": 5.5442,
      "eval_samples_per_second": 5.411,
      "eval_steps_per_second": 0.18,
      "step": 135
    },
    {
      "epoch": 0.05381875741986545,
      "grad_norm": 1.565540818573565,
      "learning_rate": 5.3754940711462456e-05,
      "loss": 0.1988,
      "step": 136
    },
    {
      "epoch": 0.05421448357736446,
      "grad_norm": 1.495014957094978,
      "learning_rate": 5.41501976284585e-05,
      "loss": 0.2204,
      "step": 137
    },
    {
      "epoch": 0.054610209734863475,
      "grad_norm": 1.2248016105245465,
      "learning_rate": 5.4545454545454546e-05,
      "loss": 0.2326,
      "step": 138
    },
    {
      "epoch": 0.05500593589236249,
      "grad_norm": 1.7373926239107826,
      "learning_rate": 5.49407114624506e-05,
      "loss": 0.2783,
      "step": 139
    },
    {
      "epoch": 0.055401662049861494,
      "grad_norm": 1.4994899705517453,
      "learning_rate": 5.533596837944664e-05,
      "loss": 0.2604,
      "step": 140
    },
    {
      "epoch": 0.055401662049861494,
      "eval_PRM Accuracy": 0.8650793650793651,
      "eval_PRM F1": 0.9237668161434978,
      "eval_PRM F1 AUC": 0.6392031480570586,
      "eval_PRM F1 AUC (fixed)": 0.82316773241515,
      "eval_PRM F1 Neg": 0.41379310344827586,
      "eval_PRM NPV": 0.6,
      "eval_PRM Precision": 0.8879310344827587,
      "eval_PRM Recall": 0.9626168224299065,
      "eval_PRM Specificty": 0.3157894736842105,
      "eval_loss": 0.3427734375,
      "eval_runtime": 5.1773,
      "eval_samples_per_second": 5.795,
      "eval_steps_per_second": 0.193,
      "step": 140
    },
    {
      "epoch": 0.05579738820736051,
      "grad_norm": 2.303864895570084,
      "learning_rate": 5.573122529644269e-05,
      "loss": 0.2971,
      "step": 141
    },
    {
      "epoch": 0.05619311436485952,
      "grad_norm": 2.288316867147724,
      "learning_rate": 5.612648221343873e-05,
      "loss": 0.28,
      "step": 142
    },
    {
      "epoch": 0.056588840522358526,
      "grad_norm": 2.245095025736702,
      "learning_rate": 5.652173913043478e-05,
      "loss": 0.2554,
      "step": 143
    },
    {
      "epoch": 0.05698456667985754,
      "grad_norm": 1.2883463466689713,
      "learning_rate": 5.6916996047430836e-05,
      "loss": 0.2144,
      "step": 144
    },
    {
      "epoch": 0.05738029283735655,
      "grad_norm": 1.4597842467629922,
      "learning_rate": 5.731225296442688e-05,
      "loss": 0.2017,
      "step": 145
    },
    {
      "epoch": 0.05738029283735655,
      "eval_PRM Accuracy": 0.8571428571428571,
      "eval_PRM F1": 0.9217391304347826,
      "eval_PRM F1 AUC": 0.5479586817511067,
      "eval_PRM F1 AUC (fixed)": 0.8315297589768814,
      "eval_PRM F1 Neg": 0.18181818181818182,
      "eval_PRM NPV": 0.6666666666666666,
      "eval_PRM Precision": 0.8617886178861789,
      "eval_PRM Recall": 0.9906542056074766,
      "eval_PRM Specificty": 0.10526315789473684,
      "eval_loss": 0.4089192748069763,
      "eval_runtime": 5.1596,
      "eval_samples_per_second": 5.814,
      "eval_steps_per_second": 0.194,
      "step": 145
    },
    {
      "epoch": 0.05777601899485556,
      "grad_norm": 1.233341524177808,
      "learning_rate": 5.7707509881422926e-05,
      "loss": 0.1829,
      "step": 146
    },
    {
      "epoch": 0.05817174515235457,
      "grad_norm": 1.9138426132930946,
      "learning_rate": 5.810276679841897e-05,
      "loss": 0.2902,
      "step": 147
    },
    {
      "epoch": 0.05856747130985358,
      "grad_norm": 0.9978418885286504,
      "learning_rate": 5.8498023715415016e-05,
      "loss": 0.1895,
      "step": 148
    },
    {
      "epoch": 0.05896319746735259,
      "grad_norm": 2.1072103709598236,
      "learning_rate": 5.8893280632411074e-05,
      "loss": 0.2599,
      "step": 149
    },
    {
      "epoch": 0.0593589236248516,
      "grad_norm": 1.6725269059527457,
      "learning_rate": 5.928853754940712e-05,
      "loss": 0.2315,
      "step": 150
    },
    {
      "epoch": 0.0593589236248516,
      "eval_PRM Accuracy": 0.873015873015873,
      "eval_PRM F1": 0.9298245614035088,
      "eval_PRM F1 AUC": 0.6005902606984752,
      "eval_PRM F1 AUC (fixed)": 0.8307919331037875,
      "eval_PRM F1 Neg": 0.3333333333333333,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.8760330578512396,
      "eval_PRM Recall": 0.9906542056074766,
      "eval_PRM Specificty": 0.21052631578947367,
      "eval_loss": 0.3349609375,
      "eval_runtime": 5.4009,
      "eval_samples_per_second": 5.555,
      "eval_steps_per_second": 0.185,
      "step": 150
    },
    {
      "epoch": 0.059754649782350615,
      "grad_norm": 1.3586844877082114,
      "learning_rate": 5.9683794466403164e-05,
      "loss": 0.2984,
      "step": 151
    },
    {
      "epoch": 0.06015037593984962,
      "grad_norm": 2.786379743121634,
      "learning_rate": 6.007905138339921e-05,
      "loss": 0.2945,
      "step": 152
    },
    {
      "epoch": 0.060546102097348634,
      "grad_norm": 1.5039454417658424,
      "learning_rate": 6.0474308300395254e-05,
      "loss": 0.2597,
      "step": 153
    },
    {
      "epoch": 0.060941828254847646,
      "grad_norm": 1.1947267375427337,
      "learning_rate": 6.086956521739131e-05,
      "loss": 0.2005,
      "step": 154
    },
    {
      "epoch": 0.06133755441234666,
      "grad_norm": 1.1758811153435025,
      "learning_rate": 6.126482213438736e-05,
      "loss": 0.2868,
      "step": 155
    },
    {
      "epoch": 0.06133755441234666,
      "eval_PRM Accuracy": 0.8650793650793651,
      "eval_PRM F1": 0.925764192139738,
      "eval_PRM F1 AUC": 0.5742744712247909,
      "eval_PRM F1 AUC (fixed)": 0.8359567142154452,
      "eval_PRM F1 Neg": 0.2608695652173913,
      "eval_PRM NPV": 0.75,
      "eval_PRM Precision": 0.8688524590163934,
      "eval_PRM Recall": 0.9906542056074766,
      "eval_PRM Specificty": 0.15789473684210525,
      "eval_loss": 0.3711588680744171,
      "eval_runtime": 5.3751,
      "eval_samples_per_second": 5.581,
      "eval_steps_per_second": 0.186,
      "step": 155
    },
    {
      "epoch": 0.061733280569845665,
      "grad_norm": 1.8336176830650266,
      "learning_rate": 6.16600790513834e-05,
      "loss": 0.2875,
      "step": 156
    },
    {
      "epoch": 0.06212900672734468,
      "grad_norm": 2.072198112402528,
      "learning_rate": 6.205533596837945e-05,
      "loss": 0.3571,
      "step": 157
    },
    {
      "epoch": 0.06252473288484368,
      "grad_norm": 1.6474731027868925,
      "learning_rate": 6.245059288537549e-05,
      "loss": 0.3023,
      "step": 158
    },
    {
      "epoch": 0.0629204590423427,
      "grad_norm": 0.9986757071545244,
      "learning_rate": 6.284584980237155e-05,
      "loss": 0.2186,
      "step": 159
    },
    {
      "epoch": 0.06331618519984171,
      "grad_norm": 1.2990604666461316,
      "learning_rate": 6.32411067193676e-05,
      "loss": 0.2975,
      "step": 160
    },
    {
      "epoch": 0.06331618519984171,
      "eval_PRM Accuracy": 0.8571428571428571,
      "eval_PRM F1": 0.9210526315789473,
      "eval_PRM F1 AUC": 0.5696015740285293,
      "eval_PRM F1 AUC (fixed)": 0.8202164289227742,
      "eval_PRM F1 Neg": 0.25,
      "eval_PRM NPV": 0.6,
      "eval_PRM Precision": 0.8677685950413223,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.15789473684210525,
      "eval_loss": 0.3473958373069763,
      "eval_runtime": 5.001,
      "eval_samples_per_second": 5.999,
      "eval_steps_per_second": 0.2,
      "step": 160
    },
    {
      "epoch": 0.06371191135734072,
      "grad_norm": 1.0648872975811778,
      "learning_rate": 6.363636363636364e-05,
      "loss": 0.2697,
      "step": 161
    },
    {
      "epoch": 0.06410763751483974,
      "grad_norm": 2.361891944586372,
      "learning_rate": 6.403162055335969e-05,
      "loss": 0.2922,
      "step": 162
    },
    {
      "epoch": 0.06450336367233875,
      "grad_norm": 2.4306862174047588,
      "learning_rate": 6.442687747035574e-05,
      "loss": 0.2909,
      "step": 163
    },
    {
      "epoch": 0.06489908982983775,
      "grad_norm": 1.7139361560341446,
      "learning_rate": 6.482213438735179e-05,
      "loss": 0.3111,
      "step": 164
    },
    {
      "epoch": 0.06529481598733676,
      "grad_norm": 1.1156092091125824,
      "learning_rate": 6.521739130434783e-05,
      "loss": 0.239,
      "step": 165
    },
    {
      "epoch": 0.06529481598733676,
      "eval_PRM Accuracy": 0.8650793650793651,
      "eval_PRM F1": 0.925764192139738,
      "eval_PRM F1 AUC": 0.5742744712247909,
      "eval_PRM F1 AUC (fixed)": 0.8162813575996065,
      "eval_PRM F1 Neg": 0.2608695652173913,
      "eval_PRM NPV": 0.75,
      "eval_PRM Precision": 0.8688524590163934,
      "eval_PRM Recall": 0.9906542056074766,
      "eval_PRM Specificty": 0.15789473684210525,
      "eval_loss": 0.3817708194255829,
      "eval_runtime": 4.8896,
      "eval_samples_per_second": 6.135,
      "eval_steps_per_second": 0.205,
      "step": 165
    },
    {
      "epoch": 0.06569054214483577,
      "grad_norm": 2.290693649853952,
      "learning_rate": 6.561264822134388e-05,
      "loss": 0.1944,
      "step": 166
    },
    {
      "epoch": 0.06608626830233479,
      "grad_norm": 0.9996789652946064,
      "learning_rate": 6.600790513833992e-05,
      "loss": 0.2382,
      "step": 167
    },
    {
      "epoch": 0.0664819944598338,
      "grad_norm": 1.9728676319629503,
      "learning_rate": 6.640316205533598e-05,
      "loss": 0.3577,
      "step": 168
    },
    {
      "epoch": 0.06687772061733281,
      "grad_norm": 1.399037552654936,
      "learning_rate": 6.679841897233203e-05,
      "loss": 0.2326,
      "step": 169
    },
    {
      "epoch": 0.06727344677483181,
      "grad_norm": 1.711639348538408,
      "learning_rate": 6.719367588932807e-05,
      "loss": 0.2298,
      "step": 170
    },
    {
      "epoch": 0.06727344677483181,
      "eval_PRM Accuracy": 0.8650793650793651,
      "eval_PRM F1": 0.925764192139738,
      "eval_PRM F1 AUC": 0.5742744712247909,
      "eval_PRM F1 AUC (fixed)": 0.8278406296114118,
      "eval_PRM F1 Neg": 0.2608695652173913,
      "eval_PRM NPV": 0.75,
      "eval_PRM Precision": 0.8688524590163934,
      "eval_PRM Recall": 0.9906542056074766,
      "eval_PRM Specificty": 0.15789473684210525,
      "eval_loss": 0.375,
      "eval_runtime": 5.2933,
      "eval_samples_per_second": 5.668,
      "eval_steps_per_second": 0.189,
      "step": 170
    },
    {
      "epoch": 0.06766917293233082,
      "grad_norm": 1.090873751851087,
      "learning_rate": 6.758893280632412e-05,
      "loss": 0.2071,
      "step": 171
    },
    {
      "epoch": 0.06806489908982984,
      "grad_norm": 1.0528725839121942,
      "learning_rate": 6.798418972332016e-05,
      "loss": 0.2097,
      "step": 172
    },
    {
      "epoch": 0.06846062524732885,
      "grad_norm": 1.564611352824747,
      "learning_rate": 6.837944664031622e-05,
      "loss": 0.1678,
      "step": 173
    },
    {
      "epoch": 0.06885635140482786,
      "grad_norm": 1.7184526814417416,
      "learning_rate": 6.877470355731227e-05,
      "loss": 0.2539,
      "step": 174
    },
    {
      "epoch": 0.06925207756232687,
      "grad_norm": 1.5523488421815403,
      "learning_rate": 6.916996047430831e-05,
      "loss": 0.2153,
      "step": 175
    },
    {
      "epoch": 0.06925207756232687,
      "eval_PRM Accuracy": 0.8650793650793651,
      "eval_PRM F1": 0.925764192139738,
      "eval_PRM F1 AUC": 0.5742744712247909,
      "eval_PRM F1 AUC (fixed)": 0.8509591736350219,
      "eval_PRM F1 Neg": 0.2608695652173913,
      "eval_PRM NPV": 0.75,
      "eval_PRM Precision": 0.8688524590163934,
      "eval_PRM Recall": 0.9906542056074766,
      "eval_PRM Specificty": 0.15789473684210525,
      "eval_loss": 0.3604166805744171,
      "eval_runtime": 5.3446,
      "eval_samples_per_second": 5.613,
      "eval_steps_per_second": 0.187,
      "step": 175
    },
    {
      "epoch": 0.06964780371982587,
      "grad_norm": 0.9338854274836507,
      "learning_rate": 6.956521739130436e-05,
      "loss": 0.1609,
      "step": 176
    },
    {
      "epoch": 0.07004352987732489,
      "grad_norm": 1.918425900220694,
      "learning_rate": 6.99604743083004e-05,
      "loss": 0.3119,
      "step": 177
    },
    {
      "epoch": 0.0704392560348239,
      "grad_norm": 1.174433070795973,
      "learning_rate": 7.035573122529645e-05,
      "loss": 0.2123,
      "step": 178
    },
    {
      "epoch": 0.07083498219232291,
      "grad_norm": 1.2418185542233562,
      "learning_rate": 7.075098814229249e-05,
      "loss": 0.2086,
      "step": 179
    },
    {
      "epoch": 0.07123070834982193,
      "grad_norm": 1.5109792612782922,
      "learning_rate": 7.114624505928854e-05,
      "loss": 0.2646,
      "step": 180
    },
    {
      "epoch": 0.07123070834982193,
      "eval_PRM Accuracy": 0.873015873015873,
      "eval_PRM F1": 0.9279279279279279,
      "eval_PRM F1 AUC": 0.6655189375307428,
      "eval_PRM F1 AUC (fixed)": 0.8622725036891294,
      "eval_PRM F1 Neg": 0.4666666666666667,
      "eval_PRM NPV": 0.6363636363636364,
      "eval_PRM Precision": 0.8956521739130435,
      "eval_PRM Recall": 0.9626168224299065,
      "eval_PRM Specificty": 0.3684210526315789,
      "eval_loss": 0.3208984434604645,
      "eval_runtime": 5.1541,
      "eval_samples_per_second": 5.821,
      "eval_steps_per_second": 0.194,
      "step": 180
    },
    {
      "epoch": 0.07162643450732094,
      "grad_norm": 1.5574165262021364,
      "learning_rate": 7.154150197628458e-05,
      "loss": 0.2786,
      "step": 181
    },
    {
      "epoch": 0.07202216066481995,
      "grad_norm": 1.7800358189484389,
      "learning_rate": 7.193675889328062e-05,
      "loss": 0.1803,
      "step": 182
    },
    {
      "epoch": 0.07241788682231895,
      "grad_norm": 1.0747136227042327,
      "learning_rate": 7.233201581027668e-05,
      "loss": 0.2416,
      "step": 183
    },
    {
      "epoch": 0.07281361297981796,
      "grad_norm": 1.6040584368559547,
      "learning_rate": 7.272727272727273e-05,
      "loss": 0.271,
      "step": 184
    },
    {
      "epoch": 0.07320933913731698,
      "grad_norm": 1.3537700370672423,
      "learning_rate": 7.312252964426877e-05,
      "loss": 0.2527,
      "step": 185
    },
    {
      "epoch": 0.07320933913731698,
      "eval_PRM Accuracy": 0.8650793650793651,
      "eval_PRM F1": 0.9251101321585903,
      "eval_PRM F1 AUC": 0.5959173635022135,
      "eval_PRM F1 AUC (fixed)": 0.8610427939006394,
      "eval_PRM F1 Neg": 0.32,
      "eval_PRM NPV": 0.6666666666666666,
      "eval_PRM Precision": 0.875,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.21052631578947367,
      "eval_loss": 0.3151041567325592,
      "eval_runtime": 5.2916,
      "eval_samples_per_second": 5.669,
      "eval_steps_per_second": 0.189,
      "step": 185
    },
    {
      "epoch": 0.07360506529481599,
      "grad_norm": 0.6761748961609548,
      "learning_rate": 7.351778656126482e-05,
      "loss": 0.1759,
      "step": 186
    },
    {
      "epoch": 0.074000791452315,
      "grad_norm": 0.8729138605124771,
      "learning_rate": 7.391304347826086e-05,
      "loss": 0.2607,
      "step": 187
    },
    {
      "epoch": 0.07439651760981401,
      "grad_norm": 0.8200508709377767,
      "learning_rate": 7.430830039525692e-05,
      "loss": 0.2634,
      "step": 188
    },
    {
      "epoch": 0.07479224376731301,
      "grad_norm": 1.239198743416465,
      "learning_rate": 7.470355731225297e-05,
      "loss": 0.2095,
      "step": 189
    },
    {
      "epoch": 0.07518796992481203,
      "grad_norm": 0.8414227436143085,
      "learning_rate": 7.509881422924901e-05,
      "loss": 0.2401,
      "step": 190
    },
    {
      "epoch": 0.07518796992481203,
      "eval_PRM Accuracy": 0.8571428571428571,
      "eval_PRM F1": 0.9210526315789473,
      "eval_PRM F1 AUC": 0.5696015740285293,
      "eval_PRM F1 AUC (fixed)": 0.8430890309886867,
      "eval_PRM F1 Neg": 0.25,
      "eval_PRM NPV": 0.6,
      "eval_PRM Precision": 0.8677685950413223,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.15789473684210525,
      "eval_loss": 0.3287109434604645,
      "eval_runtime": 5.0329,
      "eval_samples_per_second": 5.961,
      "eval_steps_per_second": 0.199,
      "step": 190
    },
    {
      "epoch": 0.07558369608231104,
      "grad_norm": 1.0553310781861904,
      "learning_rate": 7.549407114624506e-05,
      "loss": 0.1618,
      "step": 191
    },
    {
      "epoch": 0.07597942223981005,
      "grad_norm": 0.5619230002995388,
      "learning_rate": 7.58893280632411e-05,
      "loss": 0.1691,
      "step": 192
    },
    {
      "epoch": 0.07637514839730906,
      "grad_norm": 0.8517539406417115,
      "learning_rate": 7.628458498023716e-05,
      "loss": 0.2236,
      "step": 193
    },
    {
      "epoch": 0.07677087455480808,
      "grad_norm": 1.4422012975184149,
      "learning_rate": 7.66798418972332e-05,
      "loss": 0.3376,
      "step": 194
    },
    {
      "epoch": 0.07716660071230709,
      "grad_norm": 0.9935483122606977,
      "learning_rate": 7.707509881422925e-05,
      "loss": 0.2773,
      "step": 195
    },
    {
      "epoch": 0.07716660071230709,
      "eval_PRM Accuracy": 0.8809523809523809,
      "eval_PRM F1": 0.9333333333333333,
      "eval_PRM F1 AUC": 0.6485489424495818,
      "eval_PRM F1 AUC (fixed)": 0.838170191834727,
      "eval_PRM F1 Neg": 0.4444444444444444,
      "eval_PRM NPV": 0.75,
      "eval_PRM Precision": 0.8898305084745762,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.3157894736842105,
      "eval_loss": 0.3184244930744171,
      "eval_runtime": 4.9712,
      "eval_samples_per_second": 6.035,
      "eval_steps_per_second": 0.201,
      "step": 195
    },
    {
      "epoch": 0.07756232686980609,
      "grad_norm": 1.1973683080119206,
      "learning_rate": 7.74703557312253e-05,
      "loss": 0.2544,
      "step": 196
    },
    {
      "epoch": 0.0779580530273051,
      "grad_norm": 1.8585460585260873,
      "learning_rate": 7.786561264822134e-05,
      "loss": 0.2083,
      "step": 197
    },
    {
      "epoch": 0.07835377918480411,
      "grad_norm": 1.287945904589627,
      "learning_rate": 7.82608695652174e-05,
      "loss": 0.252,
      "step": 198
    },
    {
      "epoch": 0.07874950534230313,
      "grad_norm": 0.7920011868098351,
      "learning_rate": 7.865612648221344e-05,
      "loss": 0.2184,
      "step": 199
    },
    {
      "epoch": 0.07914523149980214,
      "grad_norm": 1.5441373183702873,
      "learning_rate": 7.905138339920949e-05,
      "loss": 0.2678,
      "step": 200
    },
    {
      "epoch": 0.07914523149980214,
      "eval_PRM Accuracy": 0.8571428571428571,
      "eval_PRM F1": 0.9210526315789473,
      "eval_PRM F1 AUC": 0.5696015740285293,
      "eval_PRM F1 AUC (fixed)": 0.8455484505656665,
      "eval_PRM F1 Neg": 0.25,
      "eval_PRM NPV": 0.6,
      "eval_PRM Precision": 0.8677685950413223,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.15789473684210525,
      "eval_loss": 0.3359375,
      "eval_runtime": 5.1477,
      "eval_samples_per_second": 5.828,
      "eval_steps_per_second": 0.194,
      "step": 200
    },
    {
      "epoch": 0.07954095765730115,
      "grad_norm": 1.7498047378823298,
      "learning_rate": 7.944664031620553e-05,
      "loss": 0.2156,
      "step": 201
    },
    {
      "epoch": 0.07993668381480015,
      "grad_norm": 0.7914969036534353,
      "learning_rate": 7.984189723320159e-05,
      "loss": 0.1594,
      "step": 202
    },
    {
      "epoch": 0.08033240997229917,
      "grad_norm": 1.4693699270347629,
      "learning_rate": 8.023715415019764e-05,
      "loss": 0.26,
      "step": 203
    },
    {
      "epoch": 0.08072813612979818,
      "grad_norm": 2.90232378523646,
      "learning_rate": 8.063241106719368e-05,
      "loss": 0.2366,
      "step": 204
    },
    {
      "epoch": 0.08112386228729719,
      "grad_norm": 1.9206441874267852,
      "learning_rate": 8.102766798418973e-05,
      "loss": 0.3147,
      "step": 205
    },
    {
      "epoch": 0.08112386228729719,
      "eval_PRM Accuracy": 0.8809523809523809,
      "eval_PRM F1": 0.9333333333333333,
      "eval_PRM F1 AUC": 0.6485489424495818,
      "eval_PRM F1 AUC (fixed)": 0.8541564190850959,
      "eval_PRM F1 Neg": 0.4444444444444444,
      "eval_PRM NPV": 0.75,
      "eval_PRM Precision": 0.8898305084745762,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.3157894736842105,
      "eval_loss": 0.3179687559604645,
      "eval_runtime": 5.1803,
      "eval_samples_per_second": 5.791,
      "eval_steps_per_second": 0.193,
      "step": 205
    },
    {
      "epoch": 0.0815195884447962,
      "grad_norm": 1.2105328533998576,
      "learning_rate": 8.142292490118577e-05,
      "loss": 0.1757,
      "step": 206
    },
    {
      "epoch": 0.08191531460229522,
      "grad_norm": 1.4430350619244348,
      "learning_rate": 8.181818181818183e-05,
      "loss": 0.2095,
      "step": 207
    },
    {
      "epoch": 0.08231104075979422,
      "grad_norm": 1.146391461565794,
      "learning_rate": 8.221343873517787e-05,
      "loss": 0.1514,
      "step": 208
    },
    {
      "epoch": 0.08270676691729323,
      "grad_norm": 1.9607721975101102,
      "learning_rate": 8.260869565217392e-05,
      "loss": 0.2995,
      "step": 209
    },
    {
      "epoch": 0.08310249307479224,
      "grad_norm": 2.449872281022898,
      "learning_rate": 8.300395256916996e-05,
      "loss": 0.4339,
      "step": 210
    },
    {
      "epoch": 0.08310249307479224,
      "eval_PRM Accuracy": 0.8650793650793651,
      "eval_PRM F1": 0.925764192139738,
      "eval_PRM F1 AUC": 0.5742744712247909,
      "eval_PRM F1 AUC (fixed)": 0.8612887358583375,
      "eval_PRM F1 Neg": 0.2608695652173913,
      "eval_PRM NPV": 0.75,
      "eval_PRM Precision": 0.8688524590163934,
      "eval_PRM Recall": 0.9906542056074766,
      "eval_PRM Specificty": 0.15789473684210525,
      "eval_loss": 0.3447916805744171,
      "eval_runtime": 5.1114,
      "eval_samples_per_second": 5.869,
      "eval_steps_per_second": 0.196,
      "step": 210
    },
    {
      "epoch": 0.08349821923229125,
      "grad_norm": 1.177023762186669,
      "learning_rate": 8.339920948616601e-05,
      "loss": 0.168,
      "step": 211
    },
    {
      "epoch": 0.08389394538979027,
      "grad_norm": 1.5995463953350837,
      "learning_rate": 8.379446640316207e-05,
      "loss": 0.2363,
      "step": 212
    },
    {
      "epoch": 0.08428967154728928,
      "grad_norm": 1.003097082648222,
      "learning_rate": 8.418972332015811e-05,
      "loss": 0.2565,
      "step": 213
    },
    {
      "epoch": 0.08468539770478829,
      "grad_norm": 2.7142190062016125,
      "learning_rate": 8.458498023715416e-05,
      "loss": 0.2661,
      "step": 214
    },
    {
      "epoch": 0.08508112386228729,
      "grad_norm": 2.76848978150072,
      "learning_rate": 8.49802371541502e-05,
      "loss": 0.3083,
      "step": 215
    },
    {
      "epoch": 0.08508112386228729,
      "eval_PRM Accuracy": 0.873015873015873,
      "eval_PRM F1": 0.926605504587156,
      "eval_PRM F1 AUC": 0.7088047220855878,
      "eval_PRM F1 AUC (fixed)": 0.8455484505656665,
      "eval_PRM F1 Neg": 0.5294117647058824,
      "eval_PRM NPV": 0.6,
      "eval_PRM Precision": 0.9099099099099099,
      "eval_PRM Recall": 0.9439252336448598,
      "eval_PRM Specificty": 0.47368421052631576,
      "eval_loss": 0.3160156309604645,
      "eval_runtime": 5.4929,
      "eval_samples_per_second": 5.462,
      "eval_steps_per_second": 0.182,
      "step": 215
    },
    {
      "epoch": 0.0854768500197863,
      "grad_norm": 1.875375450618787,
      "learning_rate": 8.537549407114625e-05,
      "loss": 0.2555,
      "step": 216
    },
    {
      "epoch": 0.08587257617728532,
      "grad_norm": 1.1347133907648337,
      "learning_rate": 8.57707509881423e-05,
      "loss": 0.2831,
      "step": 217
    },
    {
      "epoch": 0.08626830233478433,
      "grad_norm": 1.3466133793289627,
      "learning_rate": 8.616600790513835e-05,
      "loss": 0.272,
      "step": 218
    },
    {
      "epoch": 0.08666402849228334,
      "grad_norm": 1.0665527978871852,
      "learning_rate": 8.65612648221344e-05,
      "loss": 0.2105,
      "step": 219
    },
    {
      "epoch": 0.08705975464978236,
      "grad_norm": 1.144635175140197,
      "learning_rate": 8.695652173913044e-05,
      "loss": 0.2174,
      "step": 220
    },
    {
      "epoch": 0.08705975464978236,
      "eval_PRM Accuracy": 0.873015873015873,
      "eval_PRM F1": 0.9298245614035088,
      "eval_PRM F1 AUC": 0.6005902606984752,
      "eval_PRM F1 AUC (fixed)": 0.8563698967043777,
      "eval_PRM F1 Neg": 0.3333333333333333,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.8760330578512396,
      "eval_PRM Recall": 0.9906542056074766,
      "eval_PRM Specificty": 0.21052631578947367,
      "eval_loss": 0.3269205689430237,
      "eval_runtime": 5.2243,
      "eval_samples_per_second": 5.742,
      "eval_steps_per_second": 0.191,
      "step": 220
    },
    {
      "epoch": 0.08745548080728135,
      "grad_norm": 0.9633190515538607,
      "learning_rate": 8.735177865612649e-05,
      "loss": 0.2542,
      "step": 221
    },
    {
      "epoch": 0.08785120696478037,
      "grad_norm": 0.8678739847092348,
      "learning_rate": 8.774703557312254e-05,
      "loss": 0.173,
      "step": 222
    },
    {
      "epoch": 0.08824693312227938,
      "grad_norm": 1.106862169280851,
      "learning_rate": 8.814229249011859e-05,
      "loss": 0.2533,
      "step": 223
    },
    {
      "epoch": 0.0886426592797784,
      "grad_norm": 1.6096321049896682,
      "learning_rate": 8.853754940711463e-05,
      "loss": 0.2735,
      "step": 224
    },
    {
      "epoch": 0.0890383854372774,
      "grad_norm": 1.3072048830902998,
      "learning_rate": 8.893280632411068e-05,
      "loss": 0.2564,
      "step": 225
    },
    {
      "epoch": 0.0890383854372774,
      "eval_PRM Accuracy": 0.8888888888888888,
      "eval_PRM F1": 0.9380530973451328,
      "eval_PRM F1 AUC": 0.6532218396458436,
      "eval_PRM F1 AUC (fixed)": 0.8470241023118543,
      "eval_PRM F1 Neg": 0.46153846153846156,
      "eval_PRM NPV": 0.8571428571428571,
      "eval_PRM Precision": 0.8907563025210085,
      "eval_PRM Recall": 0.9906542056074766,
      "eval_PRM Specificty": 0.3157894736842105,
      "eval_loss": 0.3187499940395355,
      "eval_runtime": 5.1405,
      "eval_samples_per_second": 5.836,
      "eval_steps_per_second": 0.195,
      "step": 225
    },
    {
      "epoch": 0.08943411159477642,
      "grad_norm": 1.161608459147029,
      "learning_rate": 8.932806324110672e-05,
      "loss": 0.225,
      "step": 226
    },
    {
      "epoch": 0.08982983775227543,
      "grad_norm": 1.1778486661442251,
      "learning_rate": 8.972332015810277e-05,
      "loss": 0.2729,
      "step": 227
    },
    {
      "epoch": 0.09022556390977443,
      "grad_norm": 1.0913941998044796,
      "learning_rate": 9.011857707509881e-05,
      "loss": 0.1974,
      "step": 228
    },
    {
      "epoch": 0.09062129006727344,
      "grad_norm": 1.056959365446791,
      "learning_rate": 9.051383399209486e-05,
      "loss": 0.2323,
      "step": 229
    },
    {
      "epoch": 0.09101701622477246,
      "grad_norm": 1.2558494214060387,
      "learning_rate": 9.090909090909092e-05,
      "loss": 0.2363,
      "step": 230
    },
    {
      "epoch": 0.09101701622477246,
      "eval_PRM Accuracy": 0.8650793650793651,
      "eval_PRM F1": 0.925764192139738,
      "eval_PRM F1 AUC": 0.5742744712247909,
      "eval_PRM F1 AUC (fixed)": 0.8583374323659617,
      "eval_PRM F1 Neg": 0.2608695652173913,
      "eval_PRM NPV": 0.75,
      "eval_PRM Precision": 0.8688524590163934,
      "eval_PRM Recall": 0.9906542056074766,
      "eval_PRM Specificty": 0.15789473684210525,
      "eval_loss": 0.3174479305744171,
      "eval_runtime": 5.2308,
      "eval_samples_per_second": 5.735,
      "eval_steps_per_second": 0.191,
      "step": 230
    },
    {
      "epoch": 0.09141274238227147,
      "grad_norm": 1.0360688099651376,
      "learning_rate": 9.130434782608696e-05,
      "loss": 0.2554,
      "step": 231
    },
    {
      "epoch": 0.09180846853977048,
      "grad_norm": 0.8582464112266317,
      "learning_rate": 9.169960474308301e-05,
      "loss": 0.2762,
      "step": 232
    },
    {
      "epoch": 0.0922041946972695,
      "grad_norm": 0.9274943349895965,
      "learning_rate": 9.209486166007905e-05,
      "loss": 0.2091,
      "step": 233
    },
    {
      "epoch": 0.0925999208547685,
      "grad_norm": 1.1594653771729335,
      "learning_rate": 9.24901185770751e-05,
      "loss": 0.2196,
      "step": 234
    },
    {
      "epoch": 0.09299564701226751,
      "grad_norm": 0.829752859205453,
      "learning_rate": 9.288537549407114e-05,
      "loss": 0.1932,
      "step": 235
    },
    {
      "epoch": 0.09299564701226751,
      "eval_PRM Accuracy": 0.8809523809523809,
      "eval_PRM F1": 0.933920704845815,
      "eval_PRM F1 AUC": 0.6269060501721594,
      "eval_PRM F1 AUC (fixed)": 0.8514510575504181,
      "eval_PRM F1 Neg": 0.4,
      "eval_PRM NPV": 0.8333333333333334,
      "eval_PRM Precision": 0.8833333333333333,
      "eval_PRM Recall": 0.9906542056074766,
      "eval_PRM Specificty": 0.2631578947368421,
      "eval_loss": 0.31399738788604736,
      "eval_runtime": 5.2972,
      "eval_samples_per_second": 5.663,
      "eval_steps_per_second": 0.189,
      "step": 235
    },
    {
      "epoch": 0.09339137316976652,
      "grad_norm": 0.8250115723885185,
      "learning_rate": 9.32806324110672e-05,
      "loss": 0.2012,
      "step": 236
    },
    {
      "epoch": 0.09378709932726553,
      "grad_norm": 1.2558087006660452,
      "learning_rate": 9.367588932806325e-05,
      "loss": 0.2531,
      "step": 237
    },
    {
      "epoch": 0.09418282548476455,
      "grad_norm": 0.9946412275943232,
      "learning_rate": 9.407114624505929e-05,
      "loss": 0.1825,
      "step": 238
    },
    {
      "epoch": 0.09457855164226356,
      "grad_norm": 1.4633465246756725,
      "learning_rate": 9.446640316205534e-05,
      "loss": 0.2834,
      "step": 239
    },
    {
      "epoch": 0.09497427779976256,
      "grad_norm": 1.7421557902320985,
      "learning_rate": 9.486166007905138e-05,
      "loss": 0.2705,
      "step": 240
    },
    {
      "epoch": 0.09497427779976256,
      "eval_PRM Accuracy": 0.873015873015873,
      "eval_PRM F1": 0.926605504587156,
      "eval_PRM F1 AUC": 0.7088047220855878,
      "eval_PRM F1 AUC (fixed)": 0.8571077225774718,
      "eval_PRM F1 Neg": 0.5294117647058824,
      "eval_PRM NPV": 0.6,
      "eval_PRM Precision": 0.9099099099099099,
      "eval_PRM Recall": 0.9439252336448598,
      "eval_PRM Specificty": 0.47368421052631576,
      "eval_loss": 0.3016926944255829,
      "eval_runtime": 5.0028,
      "eval_samples_per_second": 5.997,
      "eval_steps_per_second": 0.2,
      "step": 240
    },
    {
      "epoch": 0.09537000395726157,
      "grad_norm": 1.428635764785657,
      "learning_rate": 9.525691699604744e-05,
      "loss": 0.2511,
      "step": 241
    },
    {
      "epoch": 0.09576573011476058,
      "grad_norm": 1.1933830808299823,
      "learning_rate": 9.565217391304348e-05,
      "loss": 0.2014,
      "step": 242
    },
    {
      "epoch": 0.0961614562722596,
      "grad_norm": 1.5877875387397131,
      "learning_rate": 9.604743083003953e-05,
      "loss": 0.317,
      "step": 243
    },
    {
      "epoch": 0.09655718242975861,
      "grad_norm": 1.0930814880388284,
      "learning_rate": 9.644268774703557e-05,
      "loss": 0.1752,
      "step": 244
    },
    {
      "epoch": 0.09695290858725762,
      "grad_norm": 0.9366911184178074,
      "learning_rate": 9.683794466403162e-05,
      "loss": 0.2385,
      "step": 245
    },
    {
      "epoch": 0.09695290858725762,
      "eval_PRM Accuracy": 0.873015873015873,
      "eval_PRM F1": 0.9298245614035088,
      "eval_PRM F1 AUC": 0.6005902606984752,
      "eval_PRM F1 AUC (fixed)": 0.853172651254304,
      "eval_PRM F1 Neg": 0.3333333333333333,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.8760330578512396,
      "eval_PRM Recall": 0.9906542056074766,
      "eval_PRM Specificty": 0.21052631578947367,
      "eval_loss": 0.3461262881755829,
      "eval_runtime": 5.0216,
      "eval_samples_per_second": 5.974,
      "eval_steps_per_second": 0.199,
      "step": 245
    },
    {
      "epoch": 0.09734863474475663,
      "grad_norm": 1.5642817021626845,
      "learning_rate": 9.723320158102768e-05,
      "loss": 0.2859,
      "step": 246
    },
    {
      "epoch": 0.09774436090225563,
      "grad_norm": 0.9228916964648477,
      "learning_rate": 9.762845849802372e-05,
      "loss": 0.1865,
      "step": 247
    },
    {
      "epoch": 0.09814008705975465,
      "grad_norm": 1.834239459619384,
      "learning_rate": 9.802371541501977e-05,
      "loss": 0.3112,
      "step": 248
    },
    {
      "epoch": 0.09853581321725366,
      "grad_norm": 0.8583064177610271,
      "learning_rate": 9.841897233201581e-05,
      "loss": 0.2794,
      "step": 249
    },
    {
      "epoch": 0.09893153937475267,
      "grad_norm": 1.4101251707770837,
      "learning_rate": 9.881422924901186e-05,
      "loss": 0.2374,
      "step": 250
    },
    {
      "epoch": 0.09893153937475267,
      "eval_PRM Accuracy": 0.8809523809523809,
      "eval_PRM F1": 0.9315068493150684,
      "eval_PRM F1 AUC": 0.7134776192818495,
      "eval_PRM F1 AUC (fixed)": 0.838416133792425,
      "eval_PRM F1 Neg": 0.5454545454545454,
      "eval_PRM NPV": 0.6428571428571429,
      "eval_PRM Precision": 0.9107142857142857,
      "eval_PRM Recall": 0.9532710280373832,
      "eval_PRM Specificty": 0.47368421052631576,
      "eval_loss": 0.32760417461395264,
      "eval_runtime": 5.2183,
      "eval_samples_per_second": 5.749,
      "eval_steps_per_second": 0.192,
      "step": 250
    },
    {
      "epoch": 0.09932726553225169,
      "grad_norm": 1.5146570123653844,
      "learning_rate": 9.920948616600791e-05,
      "loss": 0.2844,
      "step": 251
    },
    {
      "epoch": 0.0997229916897507,
      "grad_norm": 2.1373876601209068,
      "learning_rate": 9.960474308300396e-05,
      "loss": 0.2694,
      "step": 252
    },
    {
      "epoch": 0.1001187178472497,
      "grad_norm": 1.0271558400601226,
      "learning_rate": 0.0001,
      "loss": 0.2281,
      "step": 253
    },
    {
      "epoch": 0.10051444400474871,
      "grad_norm": 1.2796333872637746,
      "learning_rate": 9.999995228457853e-05,
      "loss": 0.2788,
      "step": 254
    },
    {
      "epoch": 0.10091017016224772,
      "grad_norm": 0.9128345535627792,
      "learning_rate": 9.99998091384052e-05,
      "loss": 0.1895,
      "step": 255
    },
    {
      "epoch": 0.10091017016224772,
      "eval_PRM Accuracy": 0.8650793650793651,
      "eval_PRM F1": 0.9264069264069265,
      "eval_PRM F1 AUC": 0.5526315789473684,
      "eval_PRM F1 AUC (fixed)": 0.8610427939006395,
      "eval_PRM F1 Neg": 0.19047619047619047,
      "eval_PRM NPV": 1.0,
      "eval_PRM Precision": 0.8629032258064516,
      "eval_PRM Recall": 1.0,
      "eval_PRM Specificty": 0.10526315789473684,
      "eval_loss": 0.37649738788604736,
      "eval_runtime": 4.9898,
      "eval_samples_per_second": 6.012,
      "eval_steps_per_second": 0.2,
      "step": 255
    },
    {
      "epoch": 0.10130589631974674,
      "grad_norm": 1.183245380563395,
      "learning_rate": 9.999957056175321e-05,
      "loss": 0.1924,
      "step": 256
    },
    {
      "epoch": 0.10170162247724575,
      "grad_norm": 0.9298576014658254,
      "learning_rate": 9.999923655507792e-05,
      "loss": 0.2081,
      "step": 257
    },
    {
      "epoch": 0.10209734863474476,
      "grad_norm": 0.8789574502583992,
      "learning_rate": 9.999880711901682e-05,
      "loss": 0.1722,
      "step": 258
    },
    {
      "epoch": 0.10249307479224377,
      "grad_norm": 1.1415498146511225,
      "learning_rate": 9.999828225438954e-05,
      "loss": 0.1649,
      "step": 259
    },
    {
      "epoch": 0.10288880094974277,
      "grad_norm": 1.3009427976925718,
      "learning_rate": 9.999766196219784e-05,
      "loss": 0.2399,
      "step": 260
    },
    {
      "epoch": 0.10288880094974277,
      "eval_PRM Accuracy": 0.8888888888888888,
      "eval_PRM F1": 0.9380530973451328,
      "eval_PRM F1 AUC": 0.6532218396458436,
      "eval_PRM F1 AUC (fixed)": 0.8595671421544514,
      "eval_PRM F1 Neg": 0.46153846153846156,
      "eval_PRM NPV": 0.8571428571428571,
      "eval_PRM Precision": 0.8907563025210085,
      "eval_PRM Recall": 0.9906542056074766,
      "eval_PRM Specificty": 0.3157894736842105,
      "eval_loss": 0.3143880069255829,
      "eval_runtime": 5.1441,
      "eval_samples_per_second": 5.832,
      "eval_steps_per_second": 0.194,
      "step": 260
    },
    {
      "epoch": 0.10328452710724179,
      "grad_norm": 1.1352674593331524,
      "learning_rate": 9.99969462436256e-05,
      "loss": 0.2004,
      "step": 261
    },
    {
      "epoch": 0.1036802532647408,
      "grad_norm": 1.4953862767195778,
      "learning_rate": 9.999613510003891e-05,
      "loss": 0.2354,
      "step": 262
    },
    {
      "epoch": 0.10407597942223981,
      "grad_norm": 1.9296494594100453,
      "learning_rate": 9.999522853298589e-05,
      "loss": 0.2936,
      "step": 263
    },
    {
      "epoch": 0.10447170557973882,
      "grad_norm": 0.8113504181052489,
      "learning_rate": 9.999422654419682e-05,
      "loss": 0.2143,
      "step": 264
    },
    {
      "epoch": 0.10486743173723784,
      "grad_norm": 1.1104817906610782,
      "learning_rate": 9.999312913558413e-05,
      "loss": 0.2882,
      "step": 265
    },
    {
      "epoch": 0.10486743173723784,
      "eval_PRM Accuracy": 0.8809523809523809,
      "eval_PRM F1": 0.933920704845815,
      "eval_PRM F1 AUC": 0.6269060501721594,
      "eval_PRM F1 AUC (fixed)": 0.8571077225774717,
      "eval_PRM F1 Neg": 0.4,
      "eval_PRM NPV": 0.8333333333333334,
      "eval_PRM Precision": 0.8833333333333333,
      "eval_PRM Recall": 0.9906542056074766,
      "eval_PRM Specificty": 0.2631578947368421,
      "eval_loss": 0.30546873807907104,
      "eval_runtime": 5.164,
      "eval_samples_per_second": 5.809,
      "eval_steps_per_second": 0.194,
      "step": 265
    },
    {
      "epoch": 0.10526315789473684,
      "grad_norm": 1.1601254855815446,
      "learning_rate": 9.999193630924236e-05,
      "loss": 0.2889,
      "step": 266
    },
    {
      "epoch": 0.10565888405223585,
      "grad_norm": 0.7848753248273791,
      "learning_rate": 9.999064806744816e-05,
      "loss": 0.1975,
      "step": 267
    },
    {
      "epoch": 0.10605461020973486,
      "grad_norm": 1.1626303940674743,
      "learning_rate": 9.998926441266026e-05,
      "loss": 0.2703,
      "step": 268
    },
    {
      "epoch": 0.10645033636723387,
      "grad_norm": 1.1554886525777526,
      "learning_rate": 9.998778534751956e-05,
      "loss": 0.2719,
      "step": 269
    },
    {
      "epoch": 0.10684606252473289,
      "grad_norm": 0.8180652665693736,
      "learning_rate": 9.998621087484901e-05,
      "loss": 0.2001,
      "step": 270
    },
    {
      "epoch": 0.10684606252473289,
      "eval_PRM Accuracy": 0.873015873015873,
      "eval_PRM F1": 0.9298245614035088,
      "eval_PRM F1 AUC": 0.6005902606984752,
      "eval_PRM F1 AUC (fixed)": 0.8467781603541564,
      "eval_PRM F1 Neg": 0.3333333333333333,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.8760330578512396,
      "eval_PRM Recall": 0.9906542056074766,
      "eval_PRM Specificty": 0.21052631578947367,
      "eval_loss": 0.3245442807674408,
      "eval_runtime": 5.1644,
      "eval_samples_per_second": 5.809,
      "eval_steps_per_second": 0.194,
      "step": 270
    },
    {
      "epoch": 0.1072417886822319,
      "grad_norm": 0.9863699033545287,
      "learning_rate": 9.998454099765368e-05,
      "loss": 0.2373,
      "step": 271
    },
    {
      "epoch": 0.1076375148397309,
      "grad_norm": 0.8246783888944057,
      "learning_rate": 9.998277571912073e-05,
      "loss": 0.1456,
      "step": 272
    },
    {
      "epoch": 0.10803324099722991,
      "grad_norm": 0.8474312355323622,
      "learning_rate": 9.99809150426194e-05,
      "loss": 0.2183,
      "step": 273
    },
    {
      "epoch": 0.10842896715472893,
      "grad_norm": 0.8415999725913421,
      "learning_rate": 9.9978958971701e-05,
      "loss": 0.2074,
      "step": 274
    },
    {
      "epoch": 0.10882469331222794,
      "grad_norm": 1.4159013929201012,
      "learning_rate": 9.997690751009892e-05,
      "loss": 0.3035,
      "step": 275
    },
    {
      "epoch": 0.10882469331222794,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.8435809149040826,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.3040364682674408,
      "eval_runtime": 5.0664,
      "eval_samples_per_second": 5.921,
      "eval_steps_per_second": 0.197,
      "step": 275
    },
    {
      "epoch": 0.10922041946972695,
      "grad_norm": 0.8032189766847755,
      "learning_rate": 9.997476066172863e-05,
      "loss": 0.1381,
      "step": 276
    },
    {
      "epoch": 0.10961614562722596,
      "grad_norm": 1.485921094149362,
      "learning_rate": 9.997251843068762e-05,
      "loss": 0.2787,
      "step": 277
    },
    {
      "epoch": 0.11001187178472498,
      "grad_norm": 1.580917742842857,
      "learning_rate": 9.997018082125546e-05,
      "loss": 0.1975,
      "step": 278
    },
    {
      "epoch": 0.11040759794222398,
      "grad_norm": 1.2831791263250123,
      "learning_rate": 9.996774783789377e-05,
      "loss": 0.2232,
      "step": 279
    },
    {
      "epoch": 0.11080332409972299,
      "grad_norm": 0.927647590301699,
      "learning_rate": 9.996521948524615e-05,
      "loss": 0.1815,
      "step": 280
    },
    {
      "epoch": 0.11080332409972299,
      "eval_PRM Accuracy": 0.873015873015873,
      "eval_PRM F1": 0.9298245614035088,
      "eval_PRM F1 AUC": 0.6005902606984752,
      "eval_PRM F1 AUC (fixed)": 0.8637481554353172,
      "eval_PRM F1 Neg": 0.3333333333333333,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.8760330578512396,
      "eval_PRM Recall": 0.9906542056074766,
      "eval_PRM Specificty": 0.21052631578947367,
      "eval_loss": 0.33525389432907104,
      "eval_runtime": 5.271,
      "eval_samples_per_second": 5.692,
      "eval_steps_per_second": 0.19,
      "step": 280
    },
    {
      "epoch": 0.111199050257222,
      "grad_norm": 1.3954702518092017,
      "learning_rate": 9.996259576813828e-05,
      "loss": 0.2555,
      "step": 281
    },
    {
      "epoch": 0.11159477641472101,
      "grad_norm": 1.4230080409477388,
      "learning_rate": 9.995987669157781e-05,
      "loss": 0.257,
      "step": 282
    },
    {
      "epoch": 0.11199050257222003,
      "grad_norm": 2.206257207395009,
      "learning_rate": 9.995706226075445e-05,
      "loss": 0.3274,
      "step": 283
    },
    {
      "epoch": 0.11238622872971904,
      "grad_norm": 0.9692411308181991,
      "learning_rate": 9.995415248103982e-05,
      "loss": 0.2275,
      "step": 284
    },
    {
      "epoch": 0.11278195488721804,
      "grad_norm": 1.3141857890806372,
      "learning_rate": 9.995114735798761e-05,
      "loss": 0.2318,
      "step": 285
    },
    {
      "epoch": 0.11278195488721804,
      "eval_PRM Accuracy": 0.8809523809523809,
      "eval_PRM F1": 0.9327354260089686,
      "eval_PRM F1 AUC": 0.6701918347270044,
      "eval_PRM F1 AUC (fixed)": 0.8423512051155927,
      "eval_PRM F1 Neg": 0.4827586206896552,
      "eval_PRM NPV": 0.7,
      "eval_PRM Precision": 0.896551724137931,
      "eval_PRM Recall": 0.9719626168224299,
      "eval_PRM Specificty": 0.3684210526315789,
      "eval_loss": 0.2985677123069763,
      "eval_runtime": 5.0325,
      "eval_samples_per_second": 5.961,
      "eval_steps_per_second": 0.199,
      "step": 285
    },
    {
      "epoch": 0.11317768104471705,
      "grad_norm": 1.9304656673414036,
      "learning_rate": 9.994804689733344e-05,
      "loss": 0.2594,
      "step": 286
    },
    {
      "epoch": 0.11357340720221606,
      "grad_norm": 0.734774262183603,
      "learning_rate": 9.99448511049949e-05,
      "loss": 0.1599,
      "step": 287
    },
    {
      "epoch": 0.11396913335971508,
      "grad_norm": 1.131100839114183,
      "learning_rate": 9.994155998707155e-05,
      "loss": 0.174,
      "step": 288
    },
    {
      "epoch": 0.11436485951721409,
      "grad_norm": 0.7482016646974031,
      "learning_rate": 9.993817354984486e-05,
      "loss": 0.1518,
      "step": 289
    },
    {
      "epoch": 0.1147605856747131,
      "grad_norm": 0.8413429547614211,
      "learning_rate": 9.993469179977821e-05,
      "loss": 0.1622,
      "step": 290
    },
    {
      "epoch": 0.1147605856747131,
      "eval_PRM Accuracy": 0.8809523809523809,
      "eval_PRM F1": 0.933920704845815,
      "eval_PRM F1 AUC": 0.6269060501721594,
      "eval_PRM F1 AUC (fixed)": 0.837924249877029,
      "eval_PRM F1 Neg": 0.4,
      "eval_PRM NPV": 0.8333333333333334,
      "eval_PRM Precision": 0.8833333333333333,
      "eval_PRM Recall": 0.9906542056074766,
      "eval_PRM Specificty": 0.2631578947368421,
      "eval_loss": 0.33027344942092896,
      "eval_runtime": 5.2643,
      "eval_samples_per_second": 5.699,
      "eval_steps_per_second": 0.19,
      "step": 290
    },
    {
      "epoch": 0.11515631183221212,
      "grad_norm": 2.0453925987101083,
      "learning_rate": 9.993111474351698e-05,
      "loss": 0.3581,
      "step": 291
    },
    {
      "epoch": 0.11555203798971111,
      "grad_norm": 1.1289598031461439,
      "learning_rate": 9.992744238788836e-05,
      "loss": 0.1954,
      "step": 292
    },
    {
      "epoch": 0.11594776414721013,
      "grad_norm": 0.9469501646996655,
      "learning_rate": 9.99236747399015e-05,
      "loss": 0.2,
      "step": 293
    },
    {
      "epoch": 0.11634349030470914,
      "grad_norm": 0.7404051811597198,
      "learning_rate": 9.991981180674737e-05,
      "loss": 0.1776,
      "step": 294
    },
    {
      "epoch": 0.11673921646220815,
      "grad_norm": 1.495367589248185,
      "learning_rate": 9.991585359579884e-05,
      "loss": 0.2673,
      "step": 295
    },
    {
      "epoch": 0.11673921646220815,
      "eval_PRM Accuracy": 0.8650793650793651,
      "eval_PRM F1": 0.9223744292237442,
      "eval_PRM F1 AUC": 0.6824889326119036,
      "eval_PRM F1 AUC (fixed)": 0.8261190359075258,
      "eval_PRM F1 Neg": 0.48484848484848486,
      "eval_PRM NPV": 0.5714285714285714,
      "eval_PRM Precision": 0.9017857142857143,
      "eval_PRM Recall": 0.9439252336448598,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.3164713680744171,
      "eval_runtime": 5.0855,
      "eval_samples_per_second": 5.899,
      "eval_steps_per_second": 0.197,
      "step": 295
    },
    {
      "epoch": 0.11713494261970717,
      "grad_norm": 1.3916230330347663,
      "learning_rate": 9.991180011461063e-05,
      "loss": 0.2719,
      "step": 296
    },
    {
      "epoch": 0.11753066877720618,
      "grad_norm": 1.4981842952663937,
      "learning_rate": 9.990765137091927e-05,
      "loss": 0.3036,
      "step": 297
    },
    {
      "epoch": 0.11792639493470518,
      "grad_norm": 0.8921097366961067,
      "learning_rate": 9.990340737264311e-05,
      "loss": 0.2214,
      "step": 298
    },
    {
      "epoch": 0.11832212109220419,
      "grad_norm": 0.6011491751212544,
      "learning_rate": 9.989906812788235e-05,
      "loss": 0.1254,
      "step": 299
    },
    {
      "epoch": 0.1187178472497032,
      "grad_norm": 1.0259347512053194,
      "learning_rate": 9.989463364491893e-05,
      "loss": 0.2208,
      "step": 300
    },
    {
      "epoch": 0.1187178472497032,
      "eval_PRM Accuracy": 0.873015873015873,
      "eval_PRM F1": 0.9304347826086956,
      "eval_PRM F1 AUC": 0.5789473684210527,
      "eval_PRM F1 AUC (fixed)": 0.8393999016232169,
      "eval_PRM F1 Neg": 0.2727272727272727,
      "eval_PRM NPV": 1.0,
      "eval_PRM Precision": 0.8699186991869918,
      "eval_PRM Recall": 1.0,
      "eval_PRM Specificty": 0.15789473684210525,
      "eval_loss": 0.3674153685569763,
      "eval_runtime": 5.2469,
      "eval_samples_per_second": 5.718,
      "eval_steps_per_second": 0.191,
      "step": 300
    },
    {
      "epoch": 0.11911357340720222,
      "grad_norm": 1.250546832822412,
      "learning_rate": 9.989010393221656e-05,
      "loss": 0.2572,
      "step": 301
    },
    {
      "epoch": 0.11950929956470123,
      "grad_norm": 0.9395942123920941,
      "learning_rate": 9.988547899842076e-05,
      "loss": 0.1837,
      "step": 302
    },
    {
      "epoch": 0.11990502572220024,
      "grad_norm": 1.3530243734510679,
      "learning_rate": 9.988075885235873e-05,
      "loss": 0.3279,
      "step": 303
    },
    {
      "epoch": 0.12030075187969924,
      "grad_norm": 0.9765077752217082,
      "learning_rate": 9.987594350303941e-05,
      "loss": 0.2229,
      "step": 304
    },
    {
      "epoch": 0.12069647803719825,
      "grad_norm": 1.0545077927262563,
      "learning_rate": 9.98710329596535e-05,
      "loss": 0.2161,
      "step": 305
    },
    {
      "epoch": 0.12069647803719825,
      "eval_PRM Accuracy": 0.8650793650793651,
      "eval_PRM F1": 0.9223744292237442,
      "eval_PRM F1 AUC": 0.6824889326119036,
      "eval_PRM F1 AUC (fixed)": 0.8393999016232169,
      "eval_PRM F1 Neg": 0.48484848484848486,
      "eval_PRM NPV": 0.5714285714285714,
      "eval_PRM Precision": 0.9017857142857143,
      "eval_PRM Recall": 0.9439252336448598,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.30553385615348816,
      "eval_runtime": 5.3692,
      "eval_samples_per_second": 5.587,
      "eval_steps_per_second": 0.186,
      "step": 305
    },
    {
      "epoch": 0.12109220419469727,
      "grad_norm": 1.1795462130232517,
      "learning_rate": 9.986602723157332e-05,
      "loss": 0.2487,
      "step": 306
    },
    {
      "epoch": 0.12148793035219628,
      "grad_norm": 1.0058697005107944,
      "learning_rate": 9.986092632835286e-05,
      "loss": 0.3177,
      "step": 307
    },
    {
      "epoch": 0.12188365650969529,
      "grad_norm": 0.8430723023497687,
      "learning_rate": 9.985573025972785e-05,
      "loss": 0.2092,
      "step": 308
    },
    {
      "epoch": 0.1222793826671943,
      "grad_norm": 1.2163734543499976,
      "learning_rate": 9.985043903561555e-05,
      "loss": 0.2518,
      "step": 309
    },
    {
      "epoch": 0.12267510882469332,
      "grad_norm": 1.1247340666262227,
      "learning_rate": 9.984505266611491e-05,
      "loss": 0.2535,
      "step": 310
    },
    {
      "epoch": 0.12267510882469332,
      "eval_PRM Accuracy": 0.8888888888888888,
      "eval_PRM F1": 0.9380530973451328,
      "eval_PRM F1 AUC": 0.6532218396458436,
      "eval_PRM F1 AUC (fixed)": 0.838416133792425,
      "eval_PRM F1 Neg": 0.46153846153846156,
      "eval_PRM NPV": 0.8571428571428571,
      "eval_PRM Precision": 0.8907563025210085,
      "eval_PRM Recall": 0.9906542056074766,
      "eval_PRM Specificty": 0.3157894736842105,
      "eval_loss": 0.3548828065395355,
      "eval_runtime": 5.1907,
      "eval_samples_per_second": 5.78,
      "eval_steps_per_second": 0.193,
      "step": 310
    },
    {
      "epoch": 0.12307083498219232,
      "grad_norm": 0.8470639680714415,
      "learning_rate": 9.98395711615064e-05,
      "loss": 0.1742,
      "step": 311
    },
    {
      "epoch": 0.12346656113969133,
      "grad_norm": 0.9707165612657758,
      "learning_rate": 9.983399453225216e-05,
      "loss": 0.2156,
      "step": 312
    },
    {
      "epoch": 0.12386228729719034,
      "grad_norm": 0.792127345348363,
      "learning_rate": 9.982832278899582e-05,
      "loss": 0.1683,
      "step": 313
    },
    {
      "epoch": 0.12425801345468936,
      "grad_norm": 1.055078180826939,
      "learning_rate": 9.982255594256253e-05,
      "loss": 0.2223,
      "step": 314
    },
    {
      "epoch": 0.12465373961218837,
      "grad_norm": 0.8177205122347739,
      "learning_rate": 9.981669400395906e-05,
      "loss": 0.211,
      "step": 315
    },
    {
      "epoch": 0.12465373961218837,
      "eval_PRM Accuracy": 0.8888888888888888,
      "eval_PRM F1": 0.9369369369369369,
      "eval_PRM F1 AUC": 0.6965076242006887,
      "eval_PRM F1 AUC (fixed)": 0.8374323659616331,
      "eval_PRM F1 Neg": 0.5333333333333333,
      "eval_PRM NPV": 0.7272727272727273,
      "eval_PRM Precision": 0.9043478260869565,
      "eval_PRM Recall": 0.9719626168224299,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.31171876192092896,
      "eval_runtime": 5.1605,
      "eval_samples_per_second": 5.813,
      "eval_steps_per_second": 0.194,
      "step": 315
    },
    {
      "epoch": 0.12504946576968737,
      "grad_norm": 0.8238792553312159,
      "learning_rate": 9.981073698437355e-05,
      "loss": 0.2213,
      "step": 316
    },
    {
      "epoch": 0.1254451919271864,
      "grad_norm": 1.0718017518131142,
      "learning_rate": 9.98046848951757e-05,
      "loss": 0.2396,
      "step": 317
    },
    {
      "epoch": 0.1258409180846854,
      "grad_norm": 0.7683394451532709,
      "learning_rate": 9.97985377479166e-05,
      "loss": 0.1901,
      "step": 318
    },
    {
      "epoch": 0.12623664424218442,
      "grad_norm": 0.9797483340445025,
      "learning_rate": 9.979229555432882e-05,
      "loss": 0.2459,
      "step": 319
    },
    {
      "epoch": 0.12663237039968342,
      "grad_norm": 0.9647973102452009,
      "learning_rate": 9.978595832632632e-05,
      "loss": 0.1955,
      "step": 320
    },
    {
      "epoch": 0.12663237039968342,
      "eval_PRM Accuracy": 0.873015873015873,
      "eval_PRM F1": 0.9292035398230089,
      "eval_PRM F1 AUC": 0.6222331529758978,
      "eval_PRM F1 AUC (fixed)": 0.8499754058042301,
      "eval_PRM F1 Neg": 0.38461538461538464,
      "eval_PRM NPV": 0.7142857142857143,
      "eval_PRM Precision": 0.8823529411764706,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.2631578947368421,
      "eval_loss": 0.3257812559604645,
      "eval_runtime": 5.0409,
      "eval_samples_per_second": 5.951,
      "eval_steps_per_second": 0.198,
      "step": 320
    },
    {
      "epoch": 0.12702809655718242,
      "grad_norm": 0.7472444243359343,
      "learning_rate": 9.977952607600442e-05,
      "loss": 0.1753,
      "step": 321
    },
    {
      "epoch": 0.12742382271468145,
      "grad_norm": 0.7697177526460143,
      "learning_rate": 9.977299881563984e-05,
      "loss": 0.1979,
      "step": 322
    },
    {
      "epoch": 0.12781954887218044,
      "grad_norm": 1.321938759641418,
      "learning_rate": 9.976637655769061e-05,
      "loss": 0.2902,
      "step": 323
    },
    {
      "epoch": 0.12821527502967947,
      "grad_norm": 1.1740807914013507,
      "learning_rate": 9.975965931479607e-05,
      "loss": 0.2133,
      "step": 324
    },
    {
      "epoch": 0.12861100118717847,
      "grad_norm": 0.7366182647350779,
      "learning_rate": 9.97528470997769e-05,
      "loss": 0.1921,
      "step": 325
    },
    {
      "epoch": 0.12861100118717847,
      "eval_PRM Accuracy": 0.8809523809523809,
      "eval_PRM F1": 0.9327354260089686,
      "eval_PRM F1 AUC": 0.6701918347270044,
      "eval_PRM F1 AUC (fixed)": 0.8548942449581899,
      "eval_PRM F1 Neg": 0.4827586206896552,
      "eval_PRM NPV": 0.7,
      "eval_PRM Precision": 0.896551724137931,
      "eval_PRM Recall": 0.9719626168224299,
      "eval_PRM Specificty": 0.3684210526315789,
      "eval_loss": 0.2997395694255829,
      "eval_runtime": 5.0623,
      "eval_samples_per_second": 5.926,
      "eval_steps_per_second": 0.198,
      "step": 325
    },
    {
      "epoch": 0.1290067273446775,
      "grad_norm": 0.8896912481225402,
      "learning_rate": 9.974593992563498e-05,
      "loss": 0.2559,
      "step": 326
    },
    {
      "epoch": 0.1294024535021765,
      "grad_norm": 0.9695093010075738,
      "learning_rate": 9.973893780555346e-05,
      "loss": 0.2718,
      "step": 327
    },
    {
      "epoch": 0.1297981796596755,
      "grad_norm": 0.6357808888573262,
      "learning_rate": 9.973184075289672e-05,
      "loss": 0.1819,
      "step": 328
    },
    {
      "epoch": 0.13019390581717452,
      "grad_norm": 0.8888502692301904,
      "learning_rate": 9.972464878121028e-05,
      "loss": 0.2223,
      "step": 329
    },
    {
      "epoch": 0.13058963197467352,
      "grad_norm": 0.9594382776671175,
      "learning_rate": 9.97173619042209e-05,
      "loss": 0.2789,
      "step": 330
    },
    {
      "epoch": 0.13058963197467352,
      "eval_PRM Accuracy": 0.873015873015873,
      "eval_PRM F1": 0.9298245614035088,
      "eval_PRM F1 AUC": 0.6005902606984752,
      "eval_PRM F1 AUC (fixed)": 0.8590752582390555,
      "eval_PRM F1 Neg": 0.3333333333333333,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.8760330578512396,
      "eval_PRM Recall": 0.9906542056074766,
      "eval_PRM Specificty": 0.21052631578947367,
      "eval_loss": 0.3152669370174408,
      "eval_runtime": 5.1645,
      "eval_samples_per_second": 5.809,
      "eval_steps_per_second": 0.194,
      "step": 330
    },
    {
      "epoch": 0.13098535813217255,
      "grad_norm": 0.6130473102316657,
      "learning_rate": 9.970998013583643e-05,
      "loss": 0.2048,
      "step": 331
    },
    {
      "epoch": 0.13138108428967155,
      "grad_norm": 0.823459766763559,
      "learning_rate": 9.97025034901458e-05,
      "loss": 0.1988,
      "step": 332
    },
    {
      "epoch": 0.13177681044717054,
      "grad_norm": 0.9735050043211828,
      "learning_rate": 9.96949319814191e-05,
      "loss": 0.2353,
      "step": 333
    },
    {
      "epoch": 0.13217253660466957,
      "grad_norm": 1.0910141354775398,
      "learning_rate": 9.968726562410744e-05,
      "loss": 0.2309,
      "step": 334
    },
    {
      "epoch": 0.13256826276216857,
      "grad_norm": 0.8365728032037729,
      "learning_rate": 9.967950443284293e-05,
      "loss": 0.2284,
      "step": 335
    },
    {
      "epoch": 0.13256826276216857,
      "eval_PRM Accuracy": 0.8888888888888888,
      "eval_PRM F1": 0.9375,
      "eval_PRM F1 AUC": 0.6748647319232661,
      "eval_PRM F1 AUC (fixed)": 0.8580914904082637,
      "eval_PRM F1 Neg": 0.5,
      "eval_PRM NPV": 0.7777777777777778,
      "eval_PRM Precision": 0.8974358974358975,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.3684210526315789,
      "eval_loss": 0.29342448711395264,
      "eval_runtime": 5.0072,
      "eval_samples_per_second": 5.991,
      "eval_steps_per_second": 0.2,
      "step": 335
    },
    {
      "epoch": 0.1329639889196676,
      "grad_norm": 1.0082810285089199,
      "learning_rate": 9.967164842243872e-05,
      "loss": 0.2653,
      "step": 336
    },
    {
      "epoch": 0.1333597150771666,
      "grad_norm": 0.8990981173594887,
      "learning_rate": 9.966369760788895e-05,
      "loss": 0.2215,
      "step": 337
    },
    {
      "epoch": 0.13375544123466562,
      "grad_norm": 0.9830980248851123,
      "learning_rate": 9.965565200436865e-05,
      "loss": 0.2166,
      "step": 338
    },
    {
      "epoch": 0.13415116739216462,
      "grad_norm": 0.867151466408736,
      "learning_rate": 9.96475116272338e-05,
      "loss": 0.1822,
      "step": 339
    },
    {
      "epoch": 0.13454689354966362,
      "grad_norm": 1.31489859787621,
      "learning_rate": 9.963927649202127e-05,
      "loss": 0.2548,
      "step": 340
    },
    {
      "epoch": 0.13454689354966362,
      "eval_PRM Accuracy": 0.8809523809523809,
      "eval_PRM F1": 0.933920704845815,
      "eval_PRM F1 AUC": 0.6269060501721594,
      "eval_PRM F1 AUC (fixed)": 0.8706345302508608,
      "eval_PRM F1 Neg": 0.4,
      "eval_PRM NPV": 0.8333333333333334,
      "eval_PRM Precision": 0.8833333333333333,
      "eval_PRM Recall": 0.9906542056074766,
      "eval_PRM Specificty": 0.2631578947368421,
      "eval_loss": 0.3238932192325592,
      "eval_runtime": 5.1658,
      "eval_samples_per_second": 5.807,
      "eval_steps_per_second": 0.194,
      "step": 340
    },
    {
      "epoch": 0.13494261970716265,
      "grad_norm": 0.6966773582270539,
      "learning_rate": 9.963094661444878e-05,
      "loss": 0.1503,
      "step": 341
    },
    {
      "epoch": 0.13533834586466165,
      "grad_norm": 0.8887232899725811,
      "learning_rate": 9.962252201041486e-05,
      "loss": 0.1899,
      "step": 342
    },
    {
      "epoch": 0.13573407202216067,
      "grad_norm": 1.2646791592227051,
      "learning_rate": 9.961400269599886e-05,
      "loss": 0.2255,
      "step": 343
    },
    {
      "epoch": 0.13612979817965967,
      "grad_norm": 1.7623683677474806,
      "learning_rate": 9.960538868746087e-05,
      "loss": 0.2685,
      "step": 344
    },
    {
      "epoch": 0.1365255243371587,
      "grad_norm": 1.62914005315704,
      "learning_rate": 9.959668000124177e-05,
      "loss": 0.216,
      "step": 345
    },
    {
      "epoch": 0.1365255243371587,
      "eval_PRM Accuracy": 0.8888888888888888,
      "eval_PRM F1": 0.9369369369369369,
      "eval_PRM F1 AUC": 0.6965076242006887,
      "eval_PRM F1 AUC (fixed)": 0.8753074274471224,
      "eval_PRM F1 Neg": 0.5333333333333333,
      "eval_PRM NPV": 0.7272727272727273,
      "eval_PRM Precision": 0.9043478260869565,
      "eval_PRM Recall": 0.9719626168224299,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.2906900942325592,
      "eval_runtime": 5.192,
      "eval_samples_per_second": 5.778,
      "eval_steps_per_second": 0.193,
      "step": 345
    },
    {
      "epoch": 0.1369212504946577,
      "grad_norm": 1.3767602974689317,
      "learning_rate": 9.958787665396308e-05,
      "loss": 0.219,
      "step": 346
    },
    {
      "epoch": 0.1373169766521567,
      "grad_norm": 2.3282257840117855,
      "learning_rate": 9.957897866242702e-05,
      "loss": 0.2678,
      "step": 347
    },
    {
      "epoch": 0.13771270280965572,
      "grad_norm": 1.9681003835046509,
      "learning_rate": 9.956998604361644e-05,
      "loss": 0.2789,
      "step": 348
    },
    {
      "epoch": 0.13810842896715472,
      "grad_norm": 0.907908829178822,
      "learning_rate": 9.956089881469482e-05,
      "loss": 0.1907,
      "step": 349
    },
    {
      "epoch": 0.13850415512465375,
      "grad_norm": 1.3362738911096135,
      "learning_rate": 9.955171699300617e-05,
      "loss": 0.2371,
      "step": 350
    },
    {
      "epoch": 0.13850415512465375,
      "eval_PRM Accuracy": 0.8809523809523809,
      "eval_PRM F1": 0.933920704845815,
      "eval_PRM F1 AUC": 0.6269060501721594,
      "eval_PRM F1 AUC (fixed)": 0.8598130841121495,
      "eval_PRM F1 Neg": 0.4,
      "eval_PRM NPV": 0.8333333333333334,
      "eval_PRM Precision": 0.8833333333333333,
      "eval_PRM Recall": 0.9906542056074766,
      "eval_PRM Specificty": 0.2631578947368421,
      "eval_loss": 0.33688151836395264,
      "eval_runtime": 5.1772,
      "eval_samples_per_second": 5.795,
      "eval_steps_per_second": 0.193,
      "step": 350
    },
    {
      "epoch": 0.13889988128215275,
      "grad_norm": 1.3672787888299363,
      "learning_rate": 9.95424405960751e-05,
      "loss": 0.2614,
      "step": 351
    },
    {
      "epoch": 0.13929560743965175,
      "grad_norm": 0.8475475035358652,
      "learning_rate": 9.953306964160669e-05,
      "loss": 0.1748,
      "step": 352
    },
    {
      "epoch": 0.13969133359715077,
      "grad_norm": 0.841326974870686,
      "learning_rate": 9.952360414748649e-05,
      "loss": 0.1412,
      "step": 353
    },
    {
      "epoch": 0.14008705975464977,
      "grad_norm": 0.706401463403191,
      "learning_rate": 9.951404413178052e-05,
      "loss": 0.1674,
      "step": 354
    },
    {
      "epoch": 0.1404827859121488,
      "grad_norm": 0.7334831776284896,
      "learning_rate": 9.950438961273517e-05,
      "loss": 0.2502,
      "step": 355
    },
    {
      "epoch": 0.1404827859121488,
      "eval_PRM Accuracy": 0.8809523809523809,
      "eval_PRM F1": 0.9333333333333333,
      "eval_PRM F1 AUC": 0.6485489424495818,
      "eval_PRM F1 AUC (fixed)": 0.8364485981308412,
      "eval_PRM F1 Neg": 0.4444444444444444,
      "eval_PRM NPV": 0.75,
      "eval_PRM Precision": 0.8898305084745762,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.3157894736842105,
      "eval_loss": 0.3207356631755829,
      "eval_runtime": 5.6724,
      "eval_samples_per_second": 5.289,
      "eval_steps_per_second": 0.176,
      "step": 355
    },
    {
      "epoch": 0.1408785120696478,
      "grad_norm": 1.169454666033358,
      "learning_rate": 9.949464060877723e-05,
      "loss": 0.2633,
      "step": 356
    },
    {
      "epoch": 0.14127423822714683,
      "grad_norm": 0.6655323966174417,
      "learning_rate": 9.948479713851379e-05,
      "loss": 0.2104,
      "step": 357
    },
    {
      "epoch": 0.14166996438464582,
      "grad_norm": 1.105528441828879,
      "learning_rate": 9.947485922073231e-05,
      "loss": 0.2246,
      "step": 358
    },
    {
      "epoch": 0.14206569054214482,
      "grad_norm": 0.7732703486644537,
      "learning_rate": 9.946482687440042e-05,
      "loss": 0.1996,
      "step": 359
    },
    {
      "epoch": 0.14246141669964385,
      "grad_norm": 1.0412262937824934,
      "learning_rate": 9.945470011866604e-05,
      "loss": 0.2213,
      "step": 360
    },
    {
      "epoch": 0.14246141669964385,
      "eval_PRM Accuracy": 0.8809523809523809,
      "eval_PRM F1": 0.9333333333333333,
      "eval_PRM F1 AUC": 0.6485489424495818,
      "eval_PRM F1 AUC (fixed)": 0.8266109198229217,
      "eval_PRM F1 Neg": 0.4444444444444444,
      "eval_PRM NPV": 0.75,
      "eval_PRM Precision": 0.8898305084745762,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.3157894736842105,
      "eval_loss": 0.3063802123069763,
      "eval_runtime": 5.4261,
      "eval_samples_per_second": 5.529,
      "eval_steps_per_second": 0.184,
      "step": 360
    },
    {
      "epoch": 0.14285714285714285,
      "grad_norm": 0.7698025566577554,
      "learning_rate": 9.944447897285729e-05,
      "loss": 0.1353,
      "step": 361
    },
    {
      "epoch": 0.14325286901464188,
      "grad_norm": 1.0319691799316077,
      "learning_rate": 9.943416345648238e-05,
      "loss": 0.2365,
      "step": 362
    },
    {
      "epoch": 0.14364859517214087,
      "grad_norm": 0.9563986730994617,
      "learning_rate": 9.942375358922971e-05,
      "loss": 0.2088,
      "step": 363
    },
    {
      "epoch": 0.1440443213296399,
      "grad_norm": 1.0952805260909144,
      "learning_rate": 9.941324939096772e-05,
      "loss": 0.2111,
      "step": 364
    },
    {
      "epoch": 0.1444400474871389,
      "grad_norm": 1.3390066068610906,
      "learning_rate": 9.94026508817449e-05,
      "loss": 0.1756,
      "step": 365
    },
    {
      "epoch": 0.1444400474871389,
      "eval_PRM Accuracy": 0.8888888888888888,
      "eval_PRM F1": 0.9375,
      "eval_PRM F1 AUC": 0.6748647319232661,
      "eval_PRM F1 AUC (fixed)": 0.851942941465814,
      "eval_PRM F1 Neg": 0.5,
      "eval_PRM NPV": 0.7777777777777778,
      "eval_PRM Precision": 0.8974358974358975,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.3684210526315789,
      "eval_loss": 0.31253254413604736,
      "eval_runtime": 5.1502,
      "eval_samples_per_second": 5.825,
      "eval_steps_per_second": 0.194,
      "step": 365
    },
    {
      "epoch": 0.1448357736446379,
      "grad_norm": 1.3896729818869984,
      "learning_rate": 9.939195808178974e-05,
      "loss": 0.2269,
      "step": 366
    },
    {
      "epoch": 0.14523149980213693,
      "grad_norm": 1.096496697601604,
      "learning_rate": 9.93811710115107e-05,
      "loss": 0.2103,
      "step": 367
    },
    {
      "epoch": 0.14562722595963593,
      "grad_norm": 1.627563341278145,
      "learning_rate": 9.937028969149617e-05,
      "loss": 0.306,
      "step": 368
    },
    {
      "epoch": 0.14602295211713495,
      "grad_norm": 1.482460015518166,
      "learning_rate": 9.93593141425144e-05,
      "loss": 0.2673,
      "step": 369
    },
    {
      "epoch": 0.14641867827463395,
      "grad_norm": 1.4920757526323924,
      "learning_rate": 9.934824438551353e-05,
      "loss": 0.3004,
      "step": 370
    },
    {
      "epoch": 0.14641867827463395,
      "eval_PRM Accuracy": 0.8888888888888888,
      "eval_PRM F1": 0.9380530973451328,
      "eval_PRM F1 AUC": 0.6532218396458436,
      "eval_PRM F1 AUC (fixed)": 0.8716182980816528,
      "eval_PRM F1 Neg": 0.46153846153846156,
      "eval_PRM NPV": 0.8571428571428571,
      "eval_PRM Precision": 0.8907563025210085,
      "eval_PRM Recall": 0.9906542056074766,
      "eval_PRM Specificty": 0.3157894736842105,
      "eval_loss": 0.31985676288604736,
      "eval_runtime": 5.0419,
      "eval_samples_per_second": 5.95,
      "eval_steps_per_second": 0.198,
      "step": 370
    },
    {
      "epoch": 0.14681440443213298,
      "grad_norm": 0.7554380511114847,
      "learning_rate": 9.933708044162149e-05,
      "loss": 0.1703,
      "step": 371
    },
    {
      "epoch": 0.14721013058963198,
      "grad_norm": 0.9818963538490313,
      "learning_rate": 9.932582233214593e-05,
      "loss": 0.2075,
      "step": 372
    },
    {
      "epoch": 0.14760585674713098,
      "grad_norm": 1.0053359257785752,
      "learning_rate": 9.931447007857432e-05,
      "loss": 0.2281,
      "step": 373
    },
    {
      "epoch": 0.14800158290463,
      "grad_norm": 1.1019921540211615,
      "learning_rate": 9.930302370257374e-05,
      "loss": 0.2599,
      "step": 374
    },
    {
      "epoch": 0.148397309062129,
      "grad_norm": 0.8993492170365734,
      "learning_rate": 9.929148322599093e-05,
      "loss": 0.2268,
      "step": 375
    },
    {
      "epoch": 0.148397309062129,
      "eval_PRM Accuracy": 0.873015873015873,
      "eval_PRM F1": 0.9298245614035088,
      "eval_PRM F1 AUC": 0.6005902606984752,
      "eval_PRM F1 AUC (fixed)": 0.8701426463354649,
      "eval_PRM F1 Neg": 0.3333333333333333,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.8760330578512396,
      "eval_PRM Recall": 0.9906542056074766,
      "eval_PRM Specificty": 0.21052631578947367,
      "eval_loss": 0.3114257752895355,
      "eval_runtime": 5.2769,
      "eval_samples_per_second": 5.685,
      "eval_steps_per_second": 0.19,
      "step": 375
    },
    {
      "epoch": 0.14879303521962803,
      "grad_norm": 1.084242791231953,
      "learning_rate": 9.927984867085224e-05,
      "loss": 0.2426,
      "step": 376
    },
    {
      "epoch": 0.14918876137712703,
      "grad_norm": 0.6338408899181103,
      "learning_rate": 9.926812005936359e-05,
      "loss": 0.2107,
      "step": 377
    },
    {
      "epoch": 0.14958448753462603,
      "grad_norm": 0.8577307195658436,
      "learning_rate": 9.925629741391038e-05,
      "loss": 0.2256,
      "step": 378
    },
    {
      "epoch": 0.14998021369212505,
      "grad_norm": 0.986026436466095,
      "learning_rate": 9.924438075705756e-05,
      "loss": 0.1965,
      "step": 379
    },
    {
      "epoch": 0.15037593984962405,
      "grad_norm": 1.0569187352960723,
      "learning_rate": 9.92323701115494e-05,
      "loss": 0.1889,
      "step": 380
    },
    {
      "epoch": 0.15037593984962405,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9422222222222222,
      "eval_PRM F1 AUC": 0.6795376291195278,
      "eval_PRM F1 AUC (fixed)": 0.8703885882931628,
      "eval_PRM F1 Neg": 0.5185185185185185,
      "eval_PRM NPV": 0.875,
      "eval_PRM Precision": 0.8983050847457628,
      "eval_PRM Recall": 0.9906542056074766,
      "eval_PRM Specificty": 0.3684210526315789,
      "eval_loss": 0.29033201932907104,
      "eval_runtime": 5.3494,
      "eval_samples_per_second": 5.608,
      "eval_steps_per_second": 0.187,
      "step": 380
    },
    {
      "epoch": 0.15077166600712308,
      "grad_norm": 1.2605523383406032,
      "learning_rate": 9.922026550030965e-05,
      "loss": 0.2204,
      "step": 381
    },
    {
      "epoch": 0.15116739216462208,
      "grad_norm": 0.8260289962696835,
      "learning_rate": 9.92080669464414e-05,
      "loss": 0.1768,
      "step": 382
    },
    {
      "epoch": 0.1515631183221211,
      "grad_norm": 0.6518838181399441,
      "learning_rate": 9.919577447322697e-05,
      "loss": 0.1653,
      "step": 383
    },
    {
      "epoch": 0.1519588444796201,
      "grad_norm": 1.1826921723081798,
      "learning_rate": 9.918338810412801e-05,
      "loss": 0.1728,
      "step": 384
    },
    {
      "epoch": 0.1523545706371191,
      "grad_norm": 1.4179737906616703,
      "learning_rate": 9.917090786278533e-05,
      "loss": 0.228,
      "step": 385
    },
    {
      "epoch": 0.1523545706371191,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9422222222222222,
      "eval_PRM F1 AUC": 0.6795376291195278,
      "eval_PRM F1 AUC (fixed)": 0.8762911952779144,
      "eval_PRM F1 Neg": 0.5185185185185185,
      "eval_PRM NPV": 0.875,
      "eval_PRM Precision": 0.8983050847457628,
      "eval_PRM Recall": 0.9906542056074766,
      "eval_PRM Specificty": 0.3684210526315789,
      "eval_loss": 0.30677083134651184,
      "eval_runtime": 5.2909,
      "eval_samples_per_second": 5.67,
      "eval_steps_per_second": 0.189,
      "step": 385
    },
    {
      "epoch": 0.15275029679461813,
      "grad_norm": 1.1578945566674441,
      "learning_rate": 9.915833377301898e-05,
      "loss": 0.2205,
      "step": 386
    },
    {
      "epoch": 0.15314602295211713,
      "grad_norm": 0.8585571856540902,
      "learning_rate": 9.914566585882801e-05,
      "loss": 0.1929,
      "step": 387
    },
    {
      "epoch": 0.15354174910961615,
      "grad_norm": 1.792104707688705,
      "learning_rate": 9.913290414439068e-05,
      "loss": 0.1997,
      "step": 388
    },
    {
      "epoch": 0.15393747526711515,
      "grad_norm": 1.1743712044158157,
      "learning_rate": 9.912004865406415e-05,
      "loss": 0.1936,
      "step": 389
    },
    {
      "epoch": 0.15433320142461418,
      "grad_norm": 1.3008181859254757,
      "learning_rate": 9.910709941238467e-05,
      "loss": 0.1713,
      "step": 390
    },
    {
      "epoch": 0.15433320142461418,
      "eval_PRM Accuracy": 0.9047619047619048,
      "eval_PRM F1": 0.9464285714285714,
      "eval_PRM F1 AUC": 0.7058534185932119,
      "eval_PRM F1 AUC (fixed)": 0.8716182980816527,
      "eval_PRM F1 Neg": 0.5714285714285714,
      "eval_PRM NPV": 0.8888888888888888,
      "eval_PRM Precision": 0.905982905982906,
      "eval_PRM Recall": 0.9906542056074766,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.2810221314430237,
      "eval_runtime": 5.1475,
      "eval_samples_per_second": 5.828,
      "eval_steps_per_second": 0.194,
      "step": 390
    },
    {
      "epoch": 0.15472892758211318,
      "grad_norm": 1.240862311563138,
      "learning_rate": 9.909405644406738e-05,
      "loss": 0.1991,
      "step": 391
    },
    {
      "epoch": 0.15512465373961218,
      "grad_norm": 0.9361534992273919,
      "learning_rate": 9.908091977400629e-05,
      "loss": 0.1806,
      "step": 392
    },
    {
      "epoch": 0.1555203798971112,
      "grad_norm": 1.4114250852078498,
      "learning_rate": 9.906768942727427e-05,
      "loss": 0.2864,
      "step": 393
    },
    {
      "epoch": 0.1559161060546102,
      "grad_norm": 1.2260932555252384,
      "learning_rate": 9.905436542912301e-05,
      "loss": 0.2536,
      "step": 394
    },
    {
      "epoch": 0.15631183221210923,
      "grad_norm": 1.6525328253857028,
      "learning_rate": 9.904094780498288e-05,
      "loss": 0.345,
      "step": 395
    },
    {
      "epoch": 0.15631183221210923,
      "eval_PRM Accuracy": 0.8650793650793651,
      "eval_PRM F1": 0.925764192139738,
      "eval_PRM F1 AUC": 0.5742744712247909,
      "eval_PRM F1 AUC (fixed)": 0.8723561239547467,
      "eval_PRM F1 Neg": 0.2608695652173913,
      "eval_PRM NPV": 0.75,
      "eval_PRM Precision": 0.8688524590163934,
      "eval_PRM Recall": 0.9906542056074766,
      "eval_PRM Specificty": 0.15789473684210525,
      "eval_loss": 0.32011717557907104,
      "eval_runtime": 4.9245,
      "eval_samples_per_second": 6.092,
      "eval_steps_per_second": 0.203,
      "step": 395
    },
    {
      "epoch": 0.15670755836960823,
      "grad_norm": 0.8309626636232432,
      "learning_rate": 9.902743658046301e-05,
      "loss": 0.2328,
      "step": 396
    },
    {
      "epoch": 0.15710328452710723,
      "grad_norm": 0.6975758308838723,
      "learning_rate": 9.901383178135113e-05,
      "loss": 0.2085,
      "step": 397
    },
    {
      "epoch": 0.15749901068460626,
      "grad_norm": 0.8218021256812938,
      "learning_rate": 9.900013343361361e-05,
      "loss": 0.2183,
      "step": 398
    },
    {
      "epoch": 0.15789473684210525,
      "grad_norm": 0.8764692476896122,
      "learning_rate": 9.898634156339534e-05,
      "loss": 0.1802,
      "step": 399
    },
    {
      "epoch": 0.15829046299960428,
      "grad_norm": 0.9927961927952914,
      "learning_rate": 9.897245619701972e-05,
      "loss": 0.1982,
      "step": 400
    },
    {
      "epoch": 0.15829046299960428,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.8546483030004918,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.2874349057674408,
      "eval_runtime": 5.1824,
      "eval_samples_per_second": 5.789,
      "eval_steps_per_second": 0.193,
      "step": 400
    },
    {
      "epoch": 0.15868618915710328,
      "grad_norm": 0.782945547898975,
      "learning_rate": 9.89584773609886e-05,
      "loss": 0.227,
      "step": 401
    },
    {
      "epoch": 0.1590819153146023,
      "grad_norm": 0.6988366912221621,
      "learning_rate": 9.894440508198219e-05,
      "loss": 0.2346,
      "step": 402
    },
    {
      "epoch": 0.1594776414721013,
      "grad_norm": 1.1754822643408613,
      "learning_rate": 9.893023938685911e-05,
      "loss": 0.247,
      "step": 403
    },
    {
      "epoch": 0.1598733676296003,
      "grad_norm": 0.9649274593902464,
      "learning_rate": 9.891598030265623e-05,
      "loss": 0.2903,
      "step": 404
    },
    {
      "epoch": 0.16026909378709933,
      "grad_norm": 0.7841368053872065,
      "learning_rate": 9.89016278565887e-05,
      "loss": 0.2288,
      "step": 405
    },
    {
      "epoch": 0.16026909378709933,
      "eval_PRM Accuracy": 0.8809523809523809,
      "eval_PRM F1": 0.933920704845815,
      "eval_PRM F1 AUC": 0.6269060501721594,
      "eval_PRM F1 AUC (fixed)": 0.8652238071815052,
      "eval_PRM F1 Neg": 0.4,
      "eval_PRM NPV": 0.8333333333333334,
      "eval_PRM Precision": 0.8833333333333333,
      "eval_PRM Recall": 0.9906542056074766,
      "eval_PRM Specificty": 0.2631578947368421,
      "eval_loss": 0.29055988788604736,
      "eval_runtime": 5.1745,
      "eval_samples_per_second": 5.798,
      "eval_steps_per_second": 0.193,
      "step": 405
    },
    {
      "epoch": 0.16066481994459833,
      "grad_norm": 0.8325349839097159,
      "learning_rate": 9.88871820760498e-05,
      "loss": 0.1579,
      "step": 406
    },
    {
      "epoch": 0.16106054610209736,
      "grad_norm": 0.8240296874341528,
      "learning_rate": 9.887264298861103e-05,
      "loss": 0.2281,
      "step": 407
    },
    {
      "epoch": 0.16145627225959636,
      "grad_norm": 0.9157494959887805,
      "learning_rate": 9.88580106220219e-05,
      "loss": 0.2262,
      "step": 408
    },
    {
      "epoch": 0.16185199841709538,
      "grad_norm": 1.4685427377561266,
      "learning_rate": 9.884328500421005e-05,
      "loss": 0.2113,
      "step": 409
    },
    {
      "epoch": 0.16224772457459438,
      "grad_norm": 1.3600235444934314,
      "learning_rate": 9.882846616328099e-05,
      "loss": 0.2509,
      "step": 410
    },
    {
      "epoch": 0.16224772457459438,
      "eval_PRM Accuracy": 0.8809523809523809,
      "eval_PRM F1": 0.9321266968325792,
      "eval_PRM F1 AUC": 0.6918347270044269,
      "eval_PRM F1 AUC (fixed)": 0.868666994589277,
      "eval_PRM F1 Neg": 0.5161290322580645,
      "eval_PRM NPV": 0.6666666666666666,
      "eval_PRM Precision": 0.9035087719298246,
      "eval_PRM Recall": 0.9626168224299065,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.2912760376930237,
      "eval_runtime": 5.2052,
      "eval_samples_per_second": 5.763,
      "eval_steps_per_second": 0.192,
      "step": 410
    },
    {
      "epoch": 0.16264345073209338,
      "grad_norm": 1.376461926922242,
      "learning_rate": 9.881355412751822e-05,
      "loss": 0.3067,
      "step": 411
    },
    {
      "epoch": 0.1630391768895924,
      "grad_norm": 0.881208694997927,
      "learning_rate": 9.87985489253831e-05,
      "loss": 0.1424,
      "step": 412
    },
    {
      "epoch": 0.1634349030470914,
      "grad_norm": 0.8607549492243324,
      "learning_rate": 9.878345058551484e-05,
      "loss": 0.1672,
      "step": 413
    },
    {
      "epoch": 0.16383062920459043,
      "grad_norm": 1.1797007312476324,
      "learning_rate": 9.876825913673036e-05,
      "loss": 0.226,
      "step": 414
    },
    {
      "epoch": 0.16422635536208943,
      "grad_norm": 0.8436626857735583,
      "learning_rate": 9.875297460802431e-05,
      "loss": 0.1724,
      "step": 415
    },
    {
      "epoch": 0.16422635536208943,
      "eval_PRM Accuracy": 0.873015873015873,
      "eval_PRM F1": 0.9298245614035088,
      "eval_PRM F1 AUC": 0.6005902606984752,
      "eval_PRM F1 AUC (fixed)": 0.8546483030004919,
      "eval_PRM F1 Neg": 0.3333333333333333,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.8760330578512396,
      "eval_PRM Recall": 0.9906542056074766,
      "eval_PRM Specificty": 0.21052631578947367,
      "eval_loss": 0.32906901836395264,
      "eval_runtime": 5.2012,
      "eval_samples_per_second": 5.768,
      "eval_steps_per_second": 0.192,
      "step": 415
    },
    {
      "epoch": 0.16462208151958843,
      "grad_norm": 1.3318480631785306,
      "learning_rate": 9.873759702856901e-05,
      "loss": 0.2726,
      "step": 416
    },
    {
      "epoch": 0.16501780767708746,
      "grad_norm": 0.752632643815522,
      "learning_rate": 9.872212642771439e-05,
      "loss": 0.147,
      "step": 417
    },
    {
      "epoch": 0.16541353383458646,
      "grad_norm": 0.8256280123403222,
      "learning_rate": 9.870656283498786e-05,
      "loss": 0.1734,
      "step": 418
    },
    {
      "epoch": 0.16580925999208548,
      "grad_norm": 0.7778366327127897,
      "learning_rate": 9.869090628009438e-05,
      "loss": 0.1882,
      "step": 419
    },
    {
      "epoch": 0.16620498614958448,
      "grad_norm": 0.7268919844563321,
      "learning_rate": 9.86751567929163e-05,
      "loss": 0.1448,
      "step": 420
    },
    {
      "epoch": 0.16620498614958448,
      "eval_PRM Accuracy": 0.8888888888888888,
      "eval_PRM F1": 0.9375,
      "eval_PRM F1 AUC": 0.6748647319232661,
      "eval_PRM F1 AUC (fixed)": 0.8428430890309887,
      "eval_PRM F1 Neg": 0.5,
      "eval_PRM NPV": 0.7777777777777778,
      "eval_PRM Precision": 0.8974358974358975,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.3684210526315789,
      "eval_loss": 0.32180988788604736,
      "eval_runtime": 5.3336,
      "eval_samples_per_second": 5.625,
      "eval_steps_per_second": 0.187,
      "step": 420
    },
    {
      "epoch": 0.1666007123070835,
      "grad_norm": 0.7964057658543404,
      "learning_rate": 9.865931440351337e-05,
      "loss": 0.1282,
      "step": 421
    },
    {
      "epoch": 0.1669964384645825,
      "grad_norm": 0.8358088064920918,
      "learning_rate": 9.864337914212263e-05,
      "loss": 0.1679,
      "step": 422
    },
    {
      "epoch": 0.1673921646220815,
      "grad_norm": 0.9007579879808606,
      "learning_rate": 9.86273510391584e-05,
      "loss": 0.1463,
      "step": 423
    },
    {
      "epoch": 0.16778789077958053,
      "grad_norm": 0.8237939831731644,
      "learning_rate": 9.861123012521219e-05,
      "loss": 0.1546,
      "step": 424
    },
    {
      "epoch": 0.16818361693707953,
      "grad_norm": 0.7456620296348686,
      "learning_rate": 9.859501643105262e-05,
      "loss": 0.1781,
      "step": 425
    },
    {
      "epoch": 0.16818361693707953,
      "eval_PRM Accuracy": 0.8888888888888888,
      "eval_PRM F1": 0.9369369369369369,
      "eval_PRM F1 AUC": 0.6965076242006887,
      "eval_PRM F1 AUC (fixed)": 0.8539104771273979,
      "eval_PRM F1 Neg": 0.5333333333333333,
      "eval_PRM NPV": 0.7272727272727273,
      "eval_PRM Precision": 0.9043478260869565,
      "eval_PRM Recall": 0.9719626168224299,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.32255858182907104,
      "eval_runtime": 4.727,
      "eval_samples_per_second": 6.347,
      "eval_steps_per_second": 0.212,
      "step": 425
    },
    {
      "epoch": 0.16857934309457856,
      "grad_norm": 0.9709925984122385,
      "learning_rate": 9.857870998762544e-05,
      "loss": 0.1897,
      "step": 426
    },
    {
      "epoch": 0.16897506925207756,
      "grad_norm": 1.1905147231479112,
      "learning_rate": 9.856231082605342e-05,
      "loss": 0.2153,
      "step": 427
    },
    {
      "epoch": 0.16937079540957659,
      "grad_norm": 0.9991178556819051,
      "learning_rate": 9.854581897763626e-05,
      "loss": 0.1695,
      "step": 428
    },
    {
      "epoch": 0.16976652156707558,
      "grad_norm": 0.9334313761797003,
      "learning_rate": 9.852923447385056e-05,
      "loss": 0.157,
      "step": 429
    },
    {
      "epoch": 0.17016224772457458,
      "grad_norm": 0.9741867864822177,
      "learning_rate": 9.851255734634983e-05,
      "loss": 0.1436,
      "step": 430
    },
    {
      "epoch": 0.17016224772457458,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9411764705882353,
      "eval_PRM F1 AUC": 0.7228234136743729,
      "eval_PRM F1 AUC (fixed)": 0.865715691096901,
      "eval_PRM F1 Neg": 0.5806451612903226,
      "eval_PRM NPV": 0.75,
      "eval_PRM Precision": 0.9122807017543859,
      "eval_PRM Recall": 0.9719626168224299,
      "eval_PRM Specificty": 0.47368421052631576,
      "eval_loss": 0.3036458194255829,
      "eval_runtime": 5.1452,
      "eval_samples_per_second": 5.831,
      "eval_steps_per_second": 0.194,
      "step": 430
    },
    {
      "epoch": 0.1705579738820736,
      "grad_norm": 1.2848545499996646,
      "learning_rate": 9.849578762696426e-05,
      "loss": 0.1791,
      "step": 431
    },
    {
      "epoch": 0.1709537000395726,
      "grad_norm": 1.3007941726582894,
      "learning_rate": 9.847892534770086e-05,
      "loss": 0.2304,
      "step": 432
    },
    {
      "epoch": 0.17134942619707164,
      "grad_norm": 1.3002150956886682,
      "learning_rate": 9.846197054074325e-05,
      "loss": 0.1966,
      "step": 433
    },
    {
      "epoch": 0.17174515235457063,
      "grad_norm": 1.0644083825553325,
      "learning_rate": 9.844492323845167e-05,
      "loss": 0.2084,
      "step": 434
    },
    {
      "epoch": 0.17214087851206966,
      "grad_norm": 0.9651572025987847,
      "learning_rate": 9.842778347336286e-05,
      "loss": 0.2191,
      "step": 435
    },
    {
      "epoch": 0.17214087851206966,
      "eval_PRM Accuracy": 0.873015873015873,
      "eval_PRM F1": 0.9298245614035088,
      "eval_PRM F1 AUC": 0.6005902606984752,
      "eval_PRM F1 AUC (fixed)": 0.8703885882931628,
      "eval_PRM F1 Neg": 0.3333333333333333,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.8760330578512396,
      "eval_PRM Recall": 0.9906542056074766,
      "eval_PRM Specificty": 0.21052631578947367,
      "eval_loss": 0.3273111879825592,
      "eval_runtime": 5.1399,
      "eval_samples_per_second": 5.837,
      "eval_steps_per_second": 0.195,
      "step": 435
    },
    {
      "epoch": 0.17253660466956866,
      "grad_norm": 0.8248507421737991,
      "learning_rate": 9.841055127819009e-05,
      "loss": 0.1464,
      "step": 436
    },
    {
      "epoch": 0.17293233082706766,
      "grad_norm": 0.8808171345651463,
      "learning_rate": 9.8393226685823e-05,
      "loss": 0.191,
      "step": 437
    },
    {
      "epoch": 0.1733280569845667,
      "grad_norm": 0.7174727552036037,
      "learning_rate": 9.837580972932762e-05,
      "loss": 0.1699,
      "step": 438
    },
    {
      "epoch": 0.17372378314206569,
      "grad_norm": 1.0549911008959587,
      "learning_rate": 9.835830044194625e-05,
      "loss": 0.2365,
      "step": 439
    },
    {
      "epoch": 0.1741195092995647,
      "grad_norm": 1.0764702097362318,
      "learning_rate": 9.834069885709738e-05,
      "loss": 0.2608,
      "step": 440
    },
    {
      "epoch": 0.1741195092995647,
      "eval_PRM Accuracy": 0.8888888888888888,
      "eval_PRM F1": 0.9363636363636364,
      "eval_PRM F1 AUC": 0.7181505164781112,
      "eval_PRM F1 AUC (fixed)": 0.8598130841121494,
      "eval_PRM F1 Neg": 0.5625,
      "eval_PRM NPV": 0.6923076923076923,
      "eval_PRM Precision": 0.911504424778761,
      "eval_PRM Recall": 0.9626168224299065,
      "eval_PRM Specificty": 0.47368421052631576,
      "eval_loss": 0.29023438692092896,
      "eval_runtime": 5.1473,
      "eval_samples_per_second": 5.828,
      "eval_steps_per_second": 0.194,
      "step": 440
    },
    {
      "epoch": 0.1745152354570637,
      "grad_norm": 1.507816191114127,
      "learning_rate": 9.83230050083757e-05,
      "loss": 0.2383,
      "step": 441
    },
    {
      "epoch": 0.1749109616145627,
      "grad_norm": 1.2173394026481812,
      "learning_rate": 9.830521892955202e-05,
      "loss": 0.2277,
      "step": 442
    },
    {
      "epoch": 0.17530668777206174,
      "grad_norm": 1.156748942545076,
      "learning_rate": 9.828734065457313e-05,
      "loss": 0.2219,
      "step": 443
    },
    {
      "epoch": 0.17570241392956074,
      "grad_norm": 1.0696524188080094,
      "learning_rate": 9.826937021756177e-05,
      "loss": 0.2053,
      "step": 444
    },
    {
      "epoch": 0.17609814008705976,
      "grad_norm": 0.6772780331197916,
      "learning_rate": 9.825130765281668e-05,
      "loss": 0.1206,
      "step": 445
    },
    {
      "epoch": 0.17609814008705976,
      "eval_PRM Accuracy": 0.8650793650793651,
      "eval_PRM F1": 0.925764192139738,
      "eval_PRM F1 AUC": 0.5742744712247909,
      "eval_PRM F1 AUC (fixed)": 0.8620265617314313,
      "eval_PRM F1 Neg": 0.2608695652173913,
      "eval_PRM NPV": 0.75,
      "eval_PRM Precision": 0.8688524590163934,
      "eval_PRM Recall": 0.9906542056074766,
      "eval_PRM Specificty": 0.15789473684210525,
      "eval_loss": 0.34436848759651184,
      "eval_runtime": 5.1589,
      "eval_samples_per_second": 5.815,
      "eval_steps_per_second": 0.194,
      "step": 445
    },
    {
      "epoch": 0.17649386624455876,
      "grad_norm": 1.6058506163829567,
      "learning_rate": 9.823315299481235e-05,
      "loss": 0.2915,
      "step": 446
    },
    {
      "epoch": 0.1768895924020578,
      "grad_norm": 1.532309323119185,
      "learning_rate": 9.821490627819904e-05,
      "loss": 0.241,
      "step": 447
    },
    {
      "epoch": 0.1772853185595568,
      "grad_norm": 2.43027550013657,
      "learning_rate": 9.81965675378028e-05,
      "loss": 0.1774,
      "step": 448
    },
    {
      "epoch": 0.17768104471705579,
      "grad_norm": 1.2828904326228279,
      "learning_rate": 9.81781368086252e-05,
      "loss": 0.1487,
      "step": 449
    },
    {
      "epoch": 0.1780767708745548,
      "grad_norm": 0.7713967290502675,
      "learning_rate": 9.815961412584347e-05,
      "loss": 0.188,
      "step": 450
    },
    {
      "epoch": 0.1780767708745548,
      "eval_PRM Accuracy": 0.8888888888888888,
      "eval_PRM F1": 0.9375,
      "eval_PRM F1 AUC": 0.6748647319232661,
      "eval_PRM F1 AUC (fixed)": 0.8620265617314314,
      "eval_PRM F1 Neg": 0.5,
      "eval_PRM NPV": 0.7777777777777778,
      "eval_PRM Precision": 0.8974358974358975,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.3684210526315789,
      "eval_loss": 0.30725911259651184,
      "eval_runtime": 5.2096,
      "eval_samples_per_second": 5.759,
      "eval_steps_per_second": 0.192,
      "step": 450
    },
    {
      "epoch": 0.1784724970320538,
      "grad_norm": 1.1963818551662022,
      "learning_rate": 9.814099952481032e-05,
      "loss": 0.1916,
      "step": 451
    },
    {
      "epoch": 0.17886822318955284,
      "grad_norm": 1.0308637746858087,
      "learning_rate": 9.812229304105387e-05,
      "loss": 0.1818,
      "step": 452
    },
    {
      "epoch": 0.17926394934705184,
      "grad_norm": 1.1222282905450551,
      "learning_rate": 9.810349471027765e-05,
      "loss": 0.1794,
      "step": 453
    },
    {
      "epoch": 0.17965967550455086,
      "grad_norm": 1.1244269374574152,
      "learning_rate": 9.808460456836047e-05,
      "loss": 0.2561,
      "step": 454
    },
    {
      "epoch": 0.18005540166204986,
      "grad_norm": 1.7094228415535384,
      "learning_rate": 9.806562265135635e-05,
      "loss": 0.2284,
      "step": 455
    },
    {
      "epoch": 0.18005540166204986,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.853172651254304,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.31051433086395264,
      "eval_runtime": 4.9345,
      "eval_samples_per_second": 6.08,
      "eval_steps_per_second": 0.203,
      "step": 455
    },
    {
      "epoch": 0.18045112781954886,
      "grad_norm": 1.137280961780425,
      "learning_rate": 9.804654899549451e-05,
      "loss": 0.2141,
      "step": 456
    },
    {
      "epoch": 0.1808468539770479,
      "grad_norm": 1.496837047531952,
      "learning_rate": 9.802738363717928e-05,
      "loss": 0.2694,
      "step": 457
    },
    {
      "epoch": 0.1812425801345469,
      "grad_norm": 1.250545215062278,
      "learning_rate": 9.800812661298992e-05,
      "loss": 0.211,
      "step": 458
    },
    {
      "epoch": 0.18163830629204591,
      "grad_norm": 1.4604350769296355,
      "learning_rate": 9.798877795968078e-05,
      "loss": 0.2178,
      "step": 459
    },
    {
      "epoch": 0.1820340324495449,
      "grad_norm": 0.9088694011668346,
      "learning_rate": 9.796933771418098e-05,
      "loss": 0.2169,
      "step": 460
    },
    {
      "epoch": 0.1820340324495449,
      "eval_PRM Accuracy": 0.8888888888888888,
      "eval_PRM F1": 0.9369369369369369,
      "eval_PRM F1 AUC": 0.6965076242006887,
      "eval_PRM F1 AUC (fixed)": 0.8467781603541564,
      "eval_PRM F1 Neg": 0.5333333333333333,
      "eval_PRM NPV": 0.7272727272727273,
      "eval_PRM Precision": 0.9043478260869565,
      "eval_PRM Recall": 0.9719626168224299,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.30423176288604736,
      "eval_runtime": 4.865,
      "eval_samples_per_second": 6.167,
      "eval_steps_per_second": 0.206,
      "step": 460
    },
    {
      "epoch": 0.1824297586070439,
      "grad_norm": 1.2863171743495188,
      "learning_rate": 9.794980591359453e-05,
      "loss": 0.1991,
      "step": 461
    },
    {
      "epoch": 0.18282548476454294,
      "grad_norm": 1.1857003231197765,
      "learning_rate": 9.793018259520014e-05,
      "loss": 0.1593,
      "step": 462
    },
    {
      "epoch": 0.18322121092204194,
      "grad_norm": 1.0413845286286343,
      "learning_rate": 9.791046779645121e-05,
      "loss": 0.2232,
      "step": 463
    },
    {
      "epoch": 0.18361693707954096,
      "grad_norm": 1.0062633991526322,
      "learning_rate": 9.789066155497573e-05,
      "loss": 0.1731,
      "step": 464
    },
    {
      "epoch": 0.18401266323703996,
      "grad_norm": 1.408974424486651,
      "learning_rate": 9.787076390857623e-05,
      "loss": 0.2142,
      "step": 465
    },
    {
      "epoch": 0.18401266323703996,
      "eval_PRM Accuracy": 0.8809523809523809,
      "eval_PRM F1": 0.933920704845815,
      "eval_PRM F1 AUC": 0.6269060501721594,
      "eval_PRM F1 AUC (fixed)": 0.8649778652238072,
      "eval_PRM F1 Neg": 0.4,
      "eval_PRM NPV": 0.8333333333333334,
      "eval_PRM Precision": 0.8833333333333333,
      "eval_PRM Recall": 0.9906542056074766,
      "eval_PRM Specificty": 0.2631578947368421,
      "eval_loss": 0.35185545682907104,
      "eval_runtime": 5.1784,
      "eval_samples_per_second": 5.793,
      "eval_steps_per_second": 0.193,
      "step": 465
    },
    {
      "epoch": 0.184408389394539,
      "grad_norm": 1.2291668566653438,
      "learning_rate": 9.78507748952297e-05,
      "loss": 0.1815,
      "step": 466
    },
    {
      "epoch": 0.184804115552038,
      "grad_norm": 0.7961457074977336,
      "learning_rate": 9.783069455308749e-05,
      "loss": 0.1498,
      "step": 467
    },
    {
      "epoch": 0.185199841709537,
      "grad_norm": 1.8456966784990103,
      "learning_rate": 9.78105229204753e-05,
      "loss": 0.2613,
      "step": 468
    },
    {
      "epoch": 0.18559556786703602,
      "grad_norm": 1.2877100962816268,
      "learning_rate": 9.779026003589304e-05,
      "loss": 0.1955,
      "step": 469
    },
    {
      "epoch": 0.18599129402453501,
      "grad_norm": 1.267302401798338,
      "learning_rate": 9.77699059380148e-05,
      "loss": 0.2361,
      "step": 470
    },
    {
      "epoch": 0.18599129402453501,
      "eval_PRM Accuracy": 0.9047619047619048,
      "eval_PRM F1": 0.9464285714285714,
      "eval_PRM F1 AUC": 0.7058534185932119,
      "eval_PRM F1 AUC (fixed)": 0.8689129365469748,
      "eval_PRM F1 Neg": 0.5714285714285714,
      "eval_PRM NPV": 0.8888888888888888,
      "eval_PRM Precision": 0.905982905982906,
      "eval_PRM Recall": 0.9906542056074766,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.3060872256755829,
      "eval_runtime": 4.9107,
      "eval_samples_per_second": 6.109,
      "eval_steps_per_second": 0.204,
      "step": 470
    },
    {
      "epoch": 0.18638702018203404,
      "grad_norm": 1.1960094258040463,
      "learning_rate": 9.774946066568873e-05,
      "loss": 0.1815,
      "step": 471
    },
    {
      "epoch": 0.18678274633953304,
      "grad_norm": 1.5934914556010202,
      "learning_rate": 9.772892425793705e-05,
      "loss": 0.2188,
      "step": 472
    },
    {
      "epoch": 0.18717847249703207,
      "grad_norm": 1.2091755268659399,
      "learning_rate": 9.770829675395587e-05,
      "loss": 0.1823,
      "step": 473
    },
    {
      "epoch": 0.18757419865453107,
      "grad_norm": 1.235062546844469,
      "learning_rate": 9.768757819311523e-05,
      "loss": 0.1667,
      "step": 474
    },
    {
      "epoch": 0.18796992481203006,
      "grad_norm": 1.1992116630214977,
      "learning_rate": 9.766676861495888e-05,
      "loss": 0.2079,
      "step": 475
    },
    {
      "epoch": 0.18796992481203006,
      "eval_PRM Accuracy": 0.873015873015873,
      "eval_PRM F1": 0.9298245614035088,
      "eval_PRM F1 AUC": 0.6005902606984752,
      "eval_PRM F1 AUC (fixed)": 0.867191342843089,
      "eval_PRM F1 Neg": 0.3333333333333333,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.8760330578512396,
      "eval_PRM Recall": 0.9906542056074766,
      "eval_PRM Specificty": 0.21052631578947367,
      "eval_loss": 0.32223308086395264,
      "eval_runtime": 5.2368,
      "eval_samples_per_second": 5.729,
      "eval_steps_per_second": 0.191,
      "step": 475
    },
    {
      "epoch": 0.1883656509695291,
      "grad_norm": 0.9100189558273402,
      "learning_rate": 9.764586805920434e-05,
      "loss": 0.1956,
      "step": 476
    },
    {
      "epoch": 0.1887613771270281,
      "grad_norm": 1.0054252441051408,
      "learning_rate": 9.76248765657428e-05,
      "loss": 0.2282,
      "step": 477
    },
    {
      "epoch": 0.18915710328452712,
      "grad_norm": 1.0047134966770168,
      "learning_rate": 9.760379417463894e-05,
      "loss": 0.2355,
      "step": 478
    },
    {
      "epoch": 0.18955282944202612,
      "grad_norm": 0.8125513539439929,
      "learning_rate": 9.758262092613099e-05,
      "loss": 0.2278,
      "step": 479
    },
    {
      "epoch": 0.18994855559952512,
      "grad_norm": 0.7611823051482737,
      "learning_rate": 9.756135686063055e-05,
      "loss": 0.2033,
      "step": 480
    },
    {
      "epoch": 0.18994855559952512,
      "eval_PRM Accuracy": 0.8809523809523809,
      "eval_PRM F1": 0.933920704845815,
      "eval_PRM F1 AUC": 0.6269060501721594,
      "eval_PRM F1 AUC (fixed)": 0.8607968519429414,
      "eval_PRM F1 Neg": 0.4,
      "eval_PRM NPV": 0.8333333333333334,
      "eval_PRM Precision": 0.8833333333333333,
      "eval_PRM Recall": 0.9906542056074766,
      "eval_PRM Specificty": 0.2631578947368421,
      "eval_loss": 0.30712890625,
      "eval_runtime": 5.0564,
      "eval_samples_per_second": 5.933,
      "eval_steps_per_second": 0.198,
      "step": 480
    },
    {
      "epoch": 0.19034428175702414,
      "grad_norm": 0.6433764553200512,
      "learning_rate": 9.754000201872258e-05,
      "loss": 0.1526,
      "step": 481
    },
    {
      "epoch": 0.19074000791452314,
      "grad_norm": 0.7881003350239213,
      "learning_rate": 9.75185564411653e-05,
      "loss": 0.2109,
      "step": 482
    },
    {
      "epoch": 0.19113573407202217,
      "grad_norm": 0.826672657143874,
      "learning_rate": 9.749702016889008e-05,
      "loss": 0.1467,
      "step": 483
    },
    {
      "epoch": 0.19153146022952117,
      "grad_norm": 1.026482727630389,
      "learning_rate": 9.747539324300143e-05,
      "loss": 0.2369,
      "step": 484
    },
    {
      "epoch": 0.1919271863870202,
      "grad_norm": 1.3195179269839081,
      "learning_rate": 9.745367570477688e-05,
      "loss": 0.3038,
      "step": 485
    },
    {
      "epoch": 0.1919271863870202,
      "eval_PRM Accuracy": 0.873015873015873,
      "eval_PRM F1": 0.9298245614035088,
      "eval_PRM F1 AUC": 0.6005902606984752,
      "eval_PRM F1 AUC (fixed)": 0.8666994589276931,
      "eval_PRM F1 Neg": 0.3333333333333333,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.8760330578512396,
      "eval_PRM Recall": 0.9906542056074766,
      "eval_PRM Specificty": 0.21052631578947367,
      "eval_loss": 0.32001954317092896,
      "eval_runtime": 4.9903,
      "eval_samples_per_second": 6.012,
      "eval_steps_per_second": 0.2,
      "step": 485
    },
    {
      "epoch": 0.1923229125445192,
      "grad_norm": 0.8319515966711912,
      "learning_rate": 9.743186759566685e-05,
      "loss": 0.1669,
      "step": 486
    },
    {
      "epoch": 0.1927186387020182,
      "grad_norm": 0.972732634979424,
      "learning_rate": 9.74099689572947e-05,
      "loss": 0.2254,
      "step": 487
    },
    {
      "epoch": 0.19311436485951722,
      "grad_norm": 1.0125530271904788,
      "learning_rate": 9.738797983145654e-05,
      "loss": 0.1997,
      "step": 488
    },
    {
      "epoch": 0.19351009101701622,
      "grad_norm": 0.8485645922174874,
      "learning_rate": 9.736590026012114e-05,
      "loss": 0.1444,
      "step": 489
    },
    {
      "epoch": 0.19390581717451524,
      "grad_norm": 0.7367979409086853,
      "learning_rate": 9.734373028543001e-05,
      "loss": 0.132,
      "step": 490
    },
    {
      "epoch": 0.19390581717451524,
      "eval_PRM Accuracy": 0.8888888888888888,
      "eval_PRM F1": 0.9380530973451328,
      "eval_PRM F1 AUC": 0.6532218396458436,
      "eval_PRM F1 AUC (fixed)": 0.8652238071815052,
      "eval_PRM F1 Neg": 0.46153846153846156,
      "eval_PRM NPV": 0.8571428571428571,
      "eval_PRM Precision": 0.8907563025210085,
      "eval_PRM Recall": 0.9906542056074766,
      "eval_PRM Specificty": 0.3157894736842105,
      "eval_loss": 0.31181639432907104,
      "eval_runtime": 4.9899,
      "eval_samples_per_second": 6.012,
      "eval_steps_per_second": 0.2,
      "step": 490
    },
    {
      "epoch": 0.19430154333201424,
      "grad_norm": 1.0451585604362188,
      "learning_rate": 9.732146994969706e-05,
      "loss": 0.2093,
      "step": 491
    },
    {
      "epoch": 0.19469726948951327,
      "grad_norm": 0.8972052689670954,
      "learning_rate": 9.729911929540883e-05,
      "loss": 0.1578,
      "step": 492
    },
    {
      "epoch": 0.19509299564701227,
      "grad_norm": 1.2723243916022555,
      "learning_rate": 9.727667836522407e-05,
      "loss": 0.2714,
      "step": 493
    },
    {
      "epoch": 0.19548872180451127,
      "grad_norm": 0.8887536975525306,
      "learning_rate": 9.725414720197399e-05,
      "loss": 0.1458,
      "step": 494
    },
    {
      "epoch": 0.1958844479620103,
      "grad_norm": 1.052302895536357,
      "learning_rate": 9.72315258486619e-05,
      "loss": 0.2147,
      "step": 495
    },
    {
      "epoch": 0.1958844479620103,
      "eval_PRM Accuracy": 0.8809523809523809,
      "eval_PRM F1": 0.9333333333333333,
      "eval_PRM F1 AUC": 0.6485489424495818,
      "eval_PRM F1 AUC (fixed)": 0.8568617806197737,
      "eval_PRM F1 Neg": 0.4444444444444444,
      "eval_PRM NPV": 0.75,
      "eval_PRM Precision": 0.8898305084745762,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.3157894736842105,
      "eval_loss": 0.32522785663604736,
      "eval_runtime": 5.0028,
      "eval_samples_per_second": 5.997,
      "eval_steps_per_second": 0.2,
      "step": 495
    },
    {
      "epoch": 0.1962801741195093,
      "grad_norm": 1.107885519591802,
      "learning_rate": 9.720881434846332e-05,
      "loss": 0.2153,
      "step": 496
    },
    {
      "epoch": 0.19667590027700832,
      "grad_norm": 1.0550679159424838,
      "learning_rate": 9.718601274472578e-05,
      "loss": 0.2705,
      "step": 497
    },
    {
      "epoch": 0.19707162643450732,
      "grad_norm": 1.1302368605928312,
      "learning_rate": 9.716312108096884e-05,
      "loss": 0.1845,
      "step": 498
    },
    {
      "epoch": 0.19746735259200635,
      "grad_norm": 1.089049657778467,
      "learning_rate": 9.714013940088388e-05,
      "loss": 0.1842,
      "step": 499
    },
    {
      "epoch": 0.19786307874950534,
      "grad_norm": 1.401334561425866,
      "learning_rate": 9.711706774833414e-05,
      "loss": 0.2216,
      "step": 500
    },
    {
      "epoch": 0.19786307874950534,
      "eval_PRM Accuracy": 0.8888888888888888,
      "eval_PRM F1": 0.9375,
      "eval_PRM F1 AUC": 0.6748647319232661,
      "eval_PRM F1 AUC (fixed)": 0.8612887358583373,
      "eval_PRM F1 Neg": 0.5,
      "eval_PRM NPV": 0.7777777777777778,
      "eval_PRM Precision": 0.8974358974358975,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.3684210526315789,
      "eval_loss": 0.3042968809604645,
      "eval_runtime": 5.1886,
      "eval_samples_per_second": 5.782,
      "eval_steps_per_second": 0.193,
      "step": 500
    },
    {
      "epoch": 0.19825880490700434,
      "grad_norm": 1.1264535554417732,
      "learning_rate": 9.709390616735456e-05,
      "loss": 0.1603,
      "step": 501
    },
    {
      "epoch": 0.19865453106450337,
      "grad_norm": 1.2549995485933023,
      "learning_rate": 9.707065470215174e-05,
      "loss": 0.1631,
      "step": 502
    },
    {
      "epoch": 0.19905025722200237,
      "grad_norm": 1.4298566453240185,
      "learning_rate": 9.70473133971038e-05,
      "loss": 0.2256,
      "step": 503
    },
    {
      "epoch": 0.1994459833795014,
      "grad_norm": 1.243183708125135,
      "learning_rate": 9.702388229676033e-05,
      "loss": 0.1871,
      "step": 504
    },
    {
      "epoch": 0.1998417095370004,
      "grad_norm": 0.8346136506497759,
      "learning_rate": 9.700036144584237e-05,
      "loss": 0.157,
      "step": 505
    },
    {
      "epoch": 0.1998417095370004,
      "eval_PRM Accuracy": 0.8571428571428571,
      "eval_PRM F1": 0.9217391304347826,
      "eval_PRM F1 AUC": 0.5479586817511067,
      "eval_PRM F1 AUC (fixed)": 0.8603049680275455,
      "eval_PRM F1 Neg": 0.18181818181818182,
      "eval_PRM NPV": 0.6666666666666666,
      "eval_PRM Precision": 0.8617886178861789,
      "eval_PRM Recall": 0.9906542056074766,
      "eval_PRM Specificty": 0.10526315789473684,
      "eval_loss": 0.3991861939430237,
      "eval_runtime": 5.0295,
      "eval_samples_per_second": 5.965,
      "eval_steps_per_second": 0.199,
      "step": 505
    },
    {
      "epoch": 0.2002374356944994,
      "grad_norm": 1.3900068270293604,
      "learning_rate": 9.697675088924218e-05,
      "loss": 0.2165,
      "step": 506
    },
    {
      "epoch": 0.20063316185199842,
      "grad_norm": 0.5945249548749268,
      "learning_rate": 9.695305067202328e-05,
      "loss": 0.1352,
      "step": 507
    },
    {
      "epoch": 0.20102888800949742,
      "grad_norm": 1.043679639075919,
      "learning_rate": 9.692926083942029e-05,
      "loss": 0.1934,
      "step": 508
    },
    {
      "epoch": 0.20142461416699645,
      "grad_norm": 0.9441783291481128,
      "learning_rate": 9.690538143683891e-05,
      "loss": 0.2495,
      "step": 509
    },
    {
      "epoch": 0.20182034032449545,
      "grad_norm": 0.7709899617459975,
      "learning_rate": 9.688141250985574e-05,
      "loss": 0.2009,
      "step": 510
    },
    {
      "epoch": 0.20182034032449545,
      "eval_PRM Accuracy": 0.8650793650793651,
      "eval_PRM F1": 0.9230769230769231,
      "eval_PRM F1 AUC": 0.6608460403344811,
      "eval_PRM F1 AUC (fixed)": 0.8499754058042301,
      "eval_PRM F1 Neg": 0.45161290322580644,
      "eval_PRM NPV": 0.5833333333333334,
      "eval_PRM Precision": 0.8947368421052632,
      "eval_PRM Recall": 0.9532710280373832,
      "eval_PRM Specificty": 0.3684210526315789,
      "eval_loss": 0.32386067509651184,
      "eval_runtime": 5.0065,
      "eval_samples_per_second": 5.992,
      "eval_steps_per_second": 0.2,
      "step": 510
    },
    {
      "epoch": 0.20221606648199447,
      "grad_norm": 1.1223110605792201,
      "learning_rate": 9.68573541042183e-05,
      "loss": 0.298,
      "step": 511
    },
    {
      "epoch": 0.20261179263949347,
      "grad_norm": 1.4007344018582633,
      "learning_rate": 9.683320626584486e-05,
      "loss": 0.2643,
      "step": 512
    },
    {
      "epoch": 0.20300751879699247,
      "grad_norm": 1.7682189070436682,
      "learning_rate": 9.680896904082439e-05,
      "loss": 0.2133,
      "step": 513
    },
    {
      "epoch": 0.2034032449544915,
      "grad_norm": 1.036671216282991,
      "learning_rate": 9.678464247541648e-05,
      "loss": 0.1837,
      "step": 514
    },
    {
      "epoch": 0.2037989711119905,
      "grad_norm": 0.695210891373307,
      "learning_rate": 9.67602266160512e-05,
      "loss": 0.1752,
      "step": 515
    },
    {
      "epoch": 0.2037989711119905,
      "eval_PRM Accuracy": 0.8650793650793651,
      "eval_PRM F1": 0.9230769230769231,
      "eval_PRM F1 AUC": 0.6608460403344811,
      "eval_PRM F1 AUC (fixed)": 0.867437284800787,
      "eval_PRM F1 Neg": 0.45161290322580644,
      "eval_PRM NPV": 0.5833333333333334,
      "eval_PRM Precision": 0.8947368421052632,
      "eval_PRM Recall": 0.9532710280373832,
      "eval_PRM Specificty": 0.3684210526315789,
      "eval_loss": 0.33037108182907104,
      "eval_runtime": 5.3133,
      "eval_samples_per_second": 5.646,
      "eval_steps_per_second": 0.188,
      "step": 515
    },
    {
      "epoch": 0.20419469726948952,
      "grad_norm": 1.0365457684608814,
      "learning_rate": 9.673572150932909e-05,
      "loss": 0.2064,
      "step": 516
    },
    {
      "epoch": 0.20459042342698852,
      "grad_norm": 1.4331670555504556,
      "learning_rate": 9.6711127202021e-05,
      "loss": 0.3918,
      "step": 517
    },
    {
      "epoch": 0.20498614958448755,
      "grad_norm": 1.4725826262605737,
      "learning_rate": 9.668644374106805e-05,
      "loss": 0.2939,
      "step": 518
    },
    {
      "epoch": 0.20538187574198655,
      "grad_norm": 0.7531585558002998,
      "learning_rate": 9.666167117358149e-05,
      "loss": 0.1518,
      "step": 519
    },
    {
      "epoch": 0.20577760189948555,
      "grad_norm": 1.1173060271029716,
      "learning_rate": 9.663680954684268e-05,
      "loss": 0.1819,
      "step": 520
    },
    {
      "epoch": 0.20577760189948555,
      "eval_PRM Accuracy": 0.873015873015873,
      "eval_PRM F1": 0.9285714285714286,
      "eval_PRM F1 AUC": 0.6438760452533202,
      "eval_PRM F1 AUC (fixed)": 0.8718642400393507,
      "eval_PRM F1 Neg": 0.42857142857142855,
      "eval_PRM NPV": 0.6666666666666666,
      "eval_PRM Precision": 0.8888888888888888,
      "eval_PRM Recall": 0.9719626168224299,
      "eval_PRM Specificty": 0.3157894736842105,
      "eval_loss": 0.3360026180744171,
      "eval_runtime": 5.1335,
      "eval_samples_per_second": 5.844,
      "eval_steps_per_second": 0.195,
      "step": 520
    },
    {
      "epoch": 0.20617332805698457,
      "grad_norm": 1.2406857590155362,
      "learning_rate": 9.661185890830293e-05,
      "loss": 0.1799,
      "step": 521
    },
    {
      "epoch": 0.20656905421448357,
      "grad_norm": 0.8502138580876508,
      "learning_rate": 9.658681930558345e-05,
      "loss": 0.1954,
      "step": 522
    },
    {
      "epoch": 0.2069647803719826,
      "grad_norm": 0.8615146218402908,
      "learning_rate": 9.656169078647526e-05,
      "loss": 0.224,
      "step": 523
    },
    {
      "epoch": 0.2073605065294816,
      "grad_norm": 1.0325978121139723,
      "learning_rate": 9.653647339893905e-05,
      "loss": 0.2473,
      "step": 524
    },
    {
      "epoch": 0.2077562326869806,
      "grad_norm": 0.8077612731881857,
      "learning_rate": 9.651116719110517e-05,
      "loss": 0.1659,
      "step": 525
    },
    {
      "epoch": 0.2077562326869806,
      "eval_PRM Accuracy": 0.8809523809523809,
      "eval_PRM F1": 0.9321266968325792,
      "eval_PRM F1 AUC": 0.6918347270044269,
      "eval_PRM F1 AUC (fixed)": 0.8644859813084113,
      "eval_PRM F1 Neg": 0.5161290322580645,
      "eval_PRM NPV": 0.6666666666666666,
      "eval_PRM Precision": 0.9035087719298246,
      "eval_PRM Recall": 0.9626168224299065,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.3466796875,
      "eval_runtime": 5.1442,
      "eval_samples_per_second": 5.832,
      "eval_steps_per_second": 0.194,
      "step": 525
    },
    {
      "epoch": 0.20815195884447962,
      "grad_norm": 1.0624343543580033,
      "learning_rate": 9.648577221127346e-05,
      "loss": 0.1215,
      "step": 526
    },
    {
      "epoch": 0.20854768500197862,
      "grad_norm": 5.168173680614494,
      "learning_rate": 9.646028850791325e-05,
      "loss": 0.1992,
      "step": 527
    },
    {
      "epoch": 0.20894341115947765,
      "grad_norm": 1.8652195410740853,
      "learning_rate": 9.64347161296631e-05,
      "loss": 0.2237,
      "step": 528
    },
    {
      "epoch": 0.20933913731697665,
      "grad_norm": 0.7799574378963159,
      "learning_rate": 9.640905512533091e-05,
      "loss": 0.2083,
      "step": 529
    },
    {
      "epoch": 0.20973486347447567,
      "grad_norm": 0.9037284373512658,
      "learning_rate": 9.638330554389374e-05,
      "loss": 0.1608,
      "step": 530
    },
    {
      "epoch": 0.20973486347447567,
      "eval_PRM Accuracy": 0.8888888888888888,
      "eval_PRM F1": 0.9369369369369369,
      "eval_PRM F1 AUC": 0.6965076242006887,
      "eval_PRM F1 AUC (fixed)": 0.8644859813084113,
      "eval_PRM F1 Neg": 0.5333333333333333,
      "eval_PRM NPV": 0.7272727272727273,
      "eval_PRM Precision": 0.9043478260869565,
      "eval_PRM Recall": 0.9719626168224299,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.3483235538005829,
      "eval_runtime": 5.0089,
      "eval_samples_per_second": 5.989,
      "eval_steps_per_second": 0.2,
      "step": 530
    },
    {
      "epoch": 0.21013058963197467,
      "grad_norm": 0.9656568325957725,
      "learning_rate": 9.635746743449763e-05,
      "loss": 0.1943,
      "step": 531
    },
    {
      "epoch": 0.21052631578947367,
      "grad_norm": 1.1758744585978005,
      "learning_rate": 9.633154084645766e-05,
      "loss": 0.1958,
      "step": 532
    },
    {
      "epoch": 0.2109220419469727,
      "grad_norm": 1.1040540916105812,
      "learning_rate": 9.630552582925772e-05,
      "loss": 0.2008,
      "step": 533
    },
    {
      "epoch": 0.2113177681044717,
      "grad_norm": 0.9935287884699731,
      "learning_rate": 9.627942243255055e-05,
      "loss": 0.1666,
      "step": 534
    },
    {
      "epoch": 0.21171349426197072,
      "grad_norm": 0.972737886868156,
      "learning_rate": 9.625323070615751e-05,
      "loss": 0.1853,
      "step": 535
    },
    {
      "epoch": 0.21171349426197072,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.8703885882931629,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.30195313692092896,
      "eval_runtime": 4.9376,
      "eval_samples_per_second": 6.076,
      "eval_steps_per_second": 0.203,
      "step": 535
    },
    {
      "epoch": 0.21210922041946972,
      "grad_norm": 0.7467895741053363,
      "learning_rate": 9.622695070006855e-05,
      "loss": 0.134,
      "step": 536
    },
    {
      "epoch": 0.21250494657696875,
      "grad_norm": 0.9403120886314019,
      "learning_rate": 9.620058246444218e-05,
      "loss": 0.2194,
      "step": 537
    },
    {
      "epoch": 0.21290067273446775,
      "grad_norm": 1.2576960118400111,
      "learning_rate": 9.617412604960523e-05,
      "loss": 0.2212,
      "step": 538
    },
    {
      "epoch": 0.21329639889196675,
      "grad_norm": 0.8308266600414512,
      "learning_rate": 9.614758150605286e-05,
      "loss": 0.1599,
      "step": 539
    },
    {
      "epoch": 0.21369212504946578,
      "grad_norm": 0.960188799817658,
      "learning_rate": 9.612094888444845e-05,
      "loss": 0.1733,
      "step": 540
    },
    {
      "epoch": 0.21369212504946578,
      "eval_PRM Accuracy": 0.8888888888888888,
      "eval_PRM F1": 0.9375,
      "eval_PRM F1 AUC": 0.6748647319232661,
      "eval_PRM F1 AUC (fixed)": 0.8607968519429415,
      "eval_PRM F1 Neg": 0.5,
      "eval_PRM NPV": 0.7777777777777778,
      "eval_PRM Precision": 0.8974358974358975,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.3684210526315789,
      "eval_loss": 0.3046875,
      "eval_runtime": 5.0892,
      "eval_samples_per_second": 5.895,
      "eval_steps_per_second": 0.196,
      "step": 540
    },
    {
      "epoch": 0.21408785120696477,
      "grad_norm": 0.8821226244088822,
      "learning_rate": 9.609422823562345e-05,
      "loss": 0.1838,
      "step": 541
    },
    {
      "epoch": 0.2144835773644638,
      "grad_norm": 0.8779396786356636,
      "learning_rate": 9.606741961057736e-05,
      "loss": 0.1994,
      "step": 542
    },
    {
      "epoch": 0.2148793035219628,
      "grad_norm": 0.9278291992515524,
      "learning_rate": 9.604052306047755e-05,
      "loss": 0.1424,
      "step": 543
    },
    {
      "epoch": 0.2152750296794618,
      "grad_norm": 1.3601463696375726,
      "learning_rate": 9.601353863665925e-05,
      "loss": 0.2202,
      "step": 544
    },
    {
      "epoch": 0.21567075583696083,
      "grad_norm": 0.9571474831287845,
      "learning_rate": 9.598646639062538e-05,
      "loss": 0.2449,
      "step": 545
    },
    {
      "epoch": 0.21567075583696083,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.8630103295622233,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.3026692569255829,
      "eval_runtime": 4.9598,
      "eval_samples_per_second": 6.049,
      "eval_steps_per_second": 0.202,
      "step": 545
    },
    {
      "epoch": 0.21606648199445982,
      "grad_norm": 0.6974800822103099,
      "learning_rate": 9.595930637404649e-05,
      "loss": 0.1742,
      "step": 546
    },
    {
      "epoch": 0.21646220815195885,
      "grad_norm": 0.7413282402564647,
      "learning_rate": 9.593205863876062e-05,
      "loss": 0.17,
      "step": 547
    },
    {
      "epoch": 0.21685793430945785,
      "grad_norm": 1.0502428203770906,
      "learning_rate": 9.590472323677328e-05,
      "loss": 0.1984,
      "step": 548
    },
    {
      "epoch": 0.21725366046695688,
      "grad_norm": 1.0691413471866484,
      "learning_rate": 9.58773002202573e-05,
      "loss": 0.1281,
      "step": 549
    },
    {
      "epoch": 0.21764938662445588,
      "grad_norm": 0.6566165275714569,
      "learning_rate": 9.584978964155266e-05,
      "loss": 0.1253,
      "step": 550
    },
    {
      "epoch": 0.21764938662445588,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.8605509099852434,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.3195638060569763,
      "eval_runtime": 5.1351,
      "eval_samples_per_second": 5.842,
      "eval_steps_per_second": 0.195,
      "step": 550
    },
    {
      "epoch": 0.21804511278195488,
      "grad_norm": 0.9558714705740771,
      "learning_rate": 9.582219155316656e-05,
      "loss": 0.1737,
      "step": 551
    },
    {
      "epoch": 0.2184408389394539,
      "grad_norm": 0.9211184932049711,
      "learning_rate": 9.579450600777314e-05,
      "loss": 0.1874,
      "step": 552
    },
    {
      "epoch": 0.2188365650969529,
      "grad_norm": 0.9822808185532677,
      "learning_rate": 9.576673305821353e-05,
      "loss": 0.2077,
      "step": 553
    },
    {
      "epoch": 0.21923229125445193,
      "grad_norm": 1.0456582618812262,
      "learning_rate": 9.573887275749564e-05,
      "loss": 0.1653,
      "step": 554
    },
    {
      "epoch": 0.21962801741195093,
      "grad_norm": 2.7426170342114937,
      "learning_rate": 9.57109251587941e-05,
      "loss": 0.3692,
      "step": 555
    },
    {
      "epoch": 0.21962801741195093,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.8644859813084111,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.3179687559604645,
      "eval_runtime": 4.8311,
      "eval_samples_per_second": 6.21,
      "eval_steps_per_second": 0.207,
      "step": 555
    },
    {
      "epoch": 0.22002374356944995,
      "grad_norm": 0.9584841161621672,
      "learning_rate": 9.568289031545017e-05,
      "loss": 0.2137,
      "step": 556
    },
    {
      "epoch": 0.22041946972694895,
      "grad_norm": 0.902570061001305,
      "learning_rate": 9.565476828097163e-05,
      "loss": 0.1766,
      "step": 557
    },
    {
      "epoch": 0.22081519588444795,
      "grad_norm": 0.6781830371649948,
      "learning_rate": 9.562655910903267e-05,
      "loss": 0.1375,
      "step": 558
    },
    {
      "epoch": 0.22121092204194698,
      "grad_norm": 0.7032453896001626,
      "learning_rate": 9.55982628534738e-05,
      "loss": 0.1655,
      "step": 559
    },
    {
      "epoch": 0.22160664819944598,
      "grad_norm": 1.2747164883201627,
      "learning_rate": 9.55698795683017e-05,
      "loss": 0.1696,
      "step": 560
    },
    {
      "epoch": 0.22160664819944598,
      "eval_PRM Accuracy": 0.8888888888888888,
      "eval_PRM F1": 0.9369369369369369,
      "eval_PRM F1 AUC": 0.6965076242006887,
      "eval_PRM F1 AUC (fixed)": 0.8664535169699951,
      "eval_PRM F1 Neg": 0.5333333333333333,
      "eval_PRM NPV": 0.7272727272727273,
      "eval_PRM Precision": 0.9043478260869565,
      "eval_PRM Recall": 0.9719626168224299,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.3035481870174408,
      "eval_runtime": 4.7243,
      "eval_samples_per_second": 6.35,
      "eval_steps_per_second": 0.212,
      "step": 560
    },
    {
      "epoch": 0.222002374356945,
      "grad_norm": 0.6865135453027532,
      "learning_rate": 9.554140930768922e-05,
      "loss": 0.1609,
      "step": 561
    },
    {
      "epoch": 0.222398100514444,
      "grad_norm": 1.1115197727130854,
      "learning_rate": 9.551285212597516e-05,
      "loss": 0.1638,
      "step": 562
    },
    {
      "epoch": 0.222793826671943,
      "grad_norm": 0.6641373048947078,
      "learning_rate": 9.548420807766425e-05,
      "loss": 0.1424,
      "step": 563
    },
    {
      "epoch": 0.22318955282944203,
      "grad_norm": 1.4565119763313692,
      "learning_rate": 9.545547721742698e-05,
      "loss": 0.2122,
      "step": 564
    },
    {
      "epoch": 0.22358527898694103,
      "grad_norm": 1.174160384376405,
      "learning_rate": 9.542665960009959e-05,
      "loss": 0.1643,
      "step": 565
    },
    {
      "epoch": 0.22358527898694103,
      "eval_PRM Accuracy": 0.8650793650793651,
      "eval_PRM F1": 0.925764192139738,
      "eval_PRM F1 AUC": 0.5742744712247909,
      "eval_PRM F1 AUC (fixed)": 0.8792424987702901,
      "eval_PRM F1 Neg": 0.2608695652173913,
      "eval_PRM NPV": 0.75,
      "eval_PRM Precision": 0.8688524590163934,
      "eval_PRM Recall": 0.9906542056074766,
      "eval_PRM Specificty": 0.15789473684210525,
      "eval_loss": 0.36243489384651184,
      "eval_runtime": 5.0113,
      "eval_samples_per_second": 5.986,
      "eval_steps_per_second": 0.2,
      "step": 565
    },
    {
      "epoch": 0.22398100514444005,
      "grad_norm": 1.1951671765700973,
      "learning_rate": 9.539775528068384e-05,
      "loss": 0.2291,
      "step": 566
    },
    {
      "epoch": 0.22437673130193905,
      "grad_norm": 1.0704835132881214,
      "learning_rate": 9.536876431434703e-05,
      "loss": 0.2534,
      "step": 567
    },
    {
      "epoch": 0.22477245745943808,
      "grad_norm": 1.1412400220486025,
      "learning_rate": 9.533968675642178e-05,
      "loss": 0.2867,
      "step": 568
    },
    {
      "epoch": 0.22516818361693708,
      "grad_norm": 0.8719591915839799,
      "learning_rate": 9.531052266240601e-05,
      "loss": 0.1715,
      "step": 569
    },
    {
      "epoch": 0.22556390977443608,
      "grad_norm": 0.6894055076847694,
      "learning_rate": 9.528127208796282e-05,
      "loss": 0.1572,
      "step": 570
    },
    {
      "epoch": 0.22556390977443608,
      "eval_PRM Accuracy": 0.8888888888888888,
      "eval_PRM F1": 0.9375,
      "eval_PRM F1 AUC": 0.6748647319232661,
      "eval_PRM F1 AUC (fixed)": 0.8748155435317265,
      "eval_PRM F1 Neg": 0.5,
      "eval_PRM NPV": 0.7777777777777778,
      "eval_PRM Precision": 0.8974358974358975,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.3684210526315789,
      "eval_loss": 0.31067708134651184,
      "eval_runtime": 5.3395,
      "eval_samples_per_second": 5.619,
      "eval_steps_per_second": 0.187,
      "step": 570
    },
    {
      "epoch": 0.2259596359319351,
      "grad_norm": 0.9686660329013871,
      "learning_rate": 9.525193508892034e-05,
      "loss": 0.1939,
      "step": 571
    },
    {
      "epoch": 0.2263553620894341,
      "grad_norm": 0.8864659755188832,
      "learning_rate": 9.522251172127166e-05,
      "loss": 0.1876,
      "step": 572
    },
    {
      "epoch": 0.22675108824693313,
      "grad_norm": 0.7742166668044917,
      "learning_rate": 9.519300204117469e-05,
      "loss": 0.1888,
      "step": 573
    },
    {
      "epoch": 0.22714681440443213,
      "grad_norm": 0.8043464652581477,
      "learning_rate": 9.516340610495215e-05,
      "loss": 0.2561,
      "step": 574
    },
    {
      "epoch": 0.22754254056193116,
      "grad_norm": 0.6464540788353937,
      "learning_rate": 9.513372396909133e-05,
      "loss": 0.1456,
      "step": 575
    },
    {
      "epoch": 0.22754254056193116,
      "eval_PRM Accuracy": 0.873015873015873,
      "eval_PRM F1": 0.9298245614035088,
      "eval_PRM F1 AUC": 0.6005902606984752,
      "eval_PRM F1 AUC (fixed)": 0.8701426463354648,
      "eval_PRM F1 Neg": 0.3333333333333333,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.8760330578512396,
      "eval_PRM Recall": 0.9906542056074766,
      "eval_PRM Specificty": 0.21052631578947367,
      "eval_loss": 0.3508463501930237,
      "eval_runtime": 5.0857,
      "eval_samples_per_second": 5.899,
      "eval_steps_per_second": 0.197,
      "step": 575
    },
    {
      "epoch": 0.22793826671943015,
      "grad_norm": 0.5592506160676732,
      "learning_rate": 9.510395569024404e-05,
      "loss": 0.1366,
      "step": 576
    },
    {
      "epoch": 0.22833399287692915,
      "grad_norm": 0.9384888980101656,
      "learning_rate": 9.507410132522652e-05,
      "loss": 0.1857,
      "step": 577
    },
    {
      "epoch": 0.22872971903442818,
      "grad_norm": 0.9045955332067419,
      "learning_rate": 9.50441609310193e-05,
      "loss": 0.2088,
      "step": 578
    },
    {
      "epoch": 0.22912544519192718,
      "grad_norm": 1.2276700932407885,
      "learning_rate": 9.501413456476717e-05,
      "loss": 0.2593,
      "step": 579
    },
    {
      "epoch": 0.2295211713494262,
      "grad_norm": 0.7015735002553686,
      "learning_rate": 9.498402228377892e-05,
      "loss": 0.1231,
      "step": 580
    },
    {
      "epoch": 0.2295211713494262,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9422222222222222,
      "eval_PRM F1 AUC": 0.6795376291195278,
      "eval_PRM F1 AUC (fixed)": 0.868666994589277,
      "eval_PRM F1 Neg": 0.5185185185185185,
      "eval_PRM NPV": 0.875,
      "eval_PRM Precision": 0.8983050847457628,
      "eval_PRM Recall": 0.9906542056074766,
      "eval_PRM Specificty": 0.3684210526315789,
      "eval_loss": 0.3221028745174408,
      "eval_runtime": 5.0205,
      "eval_samples_per_second": 5.975,
      "eval_steps_per_second": 0.199,
      "step": 580
    },
    {
      "epoch": 0.2299168975069252,
      "grad_norm": 1.1283779128227227,
      "learning_rate": 9.495382414552737e-05,
      "loss": 0.1915,
      "step": 581
    },
    {
      "epoch": 0.23031262366442423,
      "grad_norm": 0.7567864418022118,
      "learning_rate": 9.492354020764919e-05,
      "loss": 0.137,
      "step": 582
    },
    {
      "epoch": 0.23070834982192323,
      "grad_norm": 1.0400915649584914,
      "learning_rate": 9.489317052794481e-05,
      "loss": 0.1729,
      "step": 583
    },
    {
      "epoch": 0.23110407597942223,
      "grad_norm": 1.0751961953781162,
      "learning_rate": 9.486271516437832e-05,
      "loss": 0.2185,
      "step": 584
    },
    {
      "epoch": 0.23149980213692126,
      "grad_norm": 1.16836913547842,
      "learning_rate": 9.483217417507734e-05,
      "loss": 0.1408,
      "step": 585
    },
    {
      "epoch": 0.23149980213692126,
      "eval_PRM Accuracy": 0.8888888888888888,
      "eval_PRM F1": 0.9375,
      "eval_PRM F1 AUC": 0.6748647319232661,
      "eval_PRM F1 AUC (fixed)": 0.8745696015740285,
      "eval_PRM F1 Neg": 0.5,
      "eval_PRM NPV": 0.7777777777777778,
      "eval_PRM Precision": 0.8974358974358975,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.3684210526315789,
      "eval_loss": 0.31787109375,
      "eval_runtime": 4.9632,
      "eval_samples_per_second": 6.045,
      "eval_steps_per_second": 0.201,
      "step": 585
    },
    {
      "epoch": 0.23189552829442026,
      "grad_norm": 1.128841484632027,
      "learning_rate": 9.480154761833293e-05,
      "loss": 0.2409,
      "step": 586
    },
    {
      "epoch": 0.23229125445191928,
      "grad_norm": 1.0484734719659377,
      "learning_rate": 9.477083555259943e-05,
      "loss": 0.2563,
      "step": 587
    },
    {
      "epoch": 0.23268698060941828,
      "grad_norm": 1.1994550998647855,
      "learning_rate": 9.474003803649441e-05,
      "loss": 0.2347,
      "step": 588
    },
    {
      "epoch": 0.23308270676691728,
      "grad_norm": 1.4231828524881203,
      "learning_rate": 9.470915512879852e-05,
      "loss": 0.2093,
      "step": 589
    },
    {
      "epoch": 0.2334784329244163,
      "grad_norm": 0.9663546927104574,
      "learning_rate": 9.467818688845544e-05,
      "loss": 0.2118,
      "step": 590
    },
    {
      "epoch": 0.2334784329244163,
      "eval_PRM Accuracy": 0.8888888888888888,
      "eval_PRM F1": 0.9369369369369369,
      "eval_PRM F1 AUC": 0.6965076242006887,
      "eval_PRM F1 AUC (fixed)": 0.8735858337432366,
      "eval_PRM F1 Neg": 0.5333333333333333,
      "eval_PRM NPV": 0.7272727272727273,
      "eval_PRM Precision": 0.9043478260869565,
      "eval_PRM Recall": 0.9719626168224299,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.29755860567092896,
      "eval_runtime": 5.2713,
      "eval_samples_per_second": 5.691,
      "eval_steps_per_second": 0.19,
      "step": 590
    },
    {
      "epoch": 0.2338741590819153,
      "grad_norm": 0.907891643306674,
      "learning_rate": 9.464713337457163e-05,
      "loss": 0.1468,
      "step": 591
    },
    {
      "epoch": 0.23426988523941433,
      "grad_norm": 0.9215718171161111,
      "learning_rate": 9.461599464641638e-05,
      "loss": 0.1335,
      "step": 592
    },
    {
      "epoch": 0.23466561139691333,
      "grad_norm": 1.5082953660735747,
      "learning_rate": 9.458477076342157e-05,
      "loss": 0.3044,
      "step": 593
    },
    {
      "epoch": 0.23506133755441236,
      "grad_norm": 0.6348812107302226,
      "learning_rate": 9.455346178518164e-05,
      "loss": 0.2146,
      "step": 594
    },
    {
      "epoch": 0.23545706371191136,
      "grad_norm": 1.122679016046897,
      "learning_rate": 9.452206777145342e-05,
      "loss": 0.2414,
      "step": 595
    },
    {
      "epoch": 0.23545706371191136,
      "eval_PRM Accuracy": 0.8650793650793651,
      "eval_PRM F1": 0.925764192139738,
      "eval_PRM F1 AUC": 0.5742744712247909,
      "eval_PRM F1 AUC (fixed)": 0.8622725036891293,
      "eval_PRM F1 Neg": 0.2608695652173913,
      "eval_PRM NPV": 0.75,
      "eval_PRM Precision": 0.8688524590163934,
      "eval_PRM Recall": 0.9906542056074766,
      "eval_PRM Specificty": 0.15789473684210525,
      "eval_loss": 0.33574217557907104,
      "eval_runtime": 5.1043,
      "eval_samples_per_second": 5.877,
      "eval_steps_per_second": 0.196,
      "step": 595
    },
    {
      "epoch": 0.23585278986941036,
      "grad_norm": 0.7336962804636922,
      "learning_rate": 9.44905887821561e-05,
      "loss": 0.1946,
      "step": 596
    },
    {
      "epoch": 0.23624851602690938,
      "grad_norm": 1.5076773027459303,
      "learning_rate": 9.445902487737095e-05,
      "loss": 0.2643,
      "step": 597
    },
    {
      "epoch": 0.23664424218440838,
      "grad_norm": 0.9990565626793124,
      "learning_rate": 9.442737611734141e-05,
      "loss": 0.2163,
      "step": 598
    },
    {
      "epoch": 0.2370399683419074,
      "grad_norm": 0.7677618013103612,
      "learning_rate": 9.439564256247281e-05,
      "loss": 0.237,
      "step": 599
    },
    {
      "epoch": 0.2374356944994064,
      "grad_norm": 0.8864752202385261,
      "learning_rate": 9.436382427333237e-05,
      "loss": 0.2222,
      "step": 600
    },
    {
      "epoch": 0.2374356944994064,
      "eval_PRM Accuracy": 0.8809523809523809,
      "eval_PRM F1": 0.9321266968325792,
      "eval_PRM F1 AUC": 0.6918347270044269,
      "eval_PRM F1 AUC (fixed)": 0.852680767338908,
      "eval_PRM F1 Neg": 0.5161290322580645,
      "eval_PRM NPV": 0.6666666666666666,
      "eval_PRM Precision": 0.9035087719298246,
      "eval_PRM Recall": 0.9626168224299065,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.29798176884651184,
      "eval_runtime": 4.819,
      "eval_samples_per_second": 6.225,
      "eval_steps_per_second": 0.208,
      "step": 600
    },
    {
      "epoch": 0.23783142065690543,
      "grad_norm": 0.843227711539601,
      "learning_rate": 9.4331921310649e-05,
      "loss": 0.2179,
      "step": 601
    },
    {
      "epoch": 0.23822714681440443,
      "grad_norm": 0.9023599319655619,
      "learning_rate": 9.429993373531326e-05,
      "loss": 0.2213,
      "step": 602
    },
    {
      "epoch": 0.23862287297190343,
      "grad_norm": 0.5349869383171925,
      "learning_rate": 9.426786160837713e-05,
      "loss": 0.1216,
      "step": 603
    },
    {
      "epoch": 0.23901859912940246,
      "grad_norm": 0.6550727960830575,
      "learning_rate": 9.423570499105403e-05,
      "loss": 0.1914,
      "step": 604
    },
    {
      "epoch": 0.23941432528690146,
      "grad_norm": 0.6256497404073587,
      "learning_rate": 9.420346394471864e-05,
      "loss": 0.1665,
      "step": 605
    },
    {
      "epoch": 0.23941432528690146,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9422222222222222,
      "eval_PRM F1 AUC": 0.6795376291195278,
      "eval_PRM F1 AUC (fixed)": 0.8460403344810624,
      "eval_PRM F1 Neg": 0.5185185185185185,
      "eval_PRM NPV": 0.875,
      "eval_PRM Precision": 0.8983050847457628,
      "eval_PRM Recall": 0.9906542056074766,
      "eval_PRM Specificty": 0.3684210526315789,
      "eval_loss": 0.36315104365348816,
      "eval_runtime": 4.9856,
      "eval_samples_per_second": 6.017,
      "eval_steps_per_second": 0.201,
      "step": 605
    },
    {
      "epoch": 0.23981005144440048,
      "grad_norm": 0.6183163375383558,
      "learning_rate": 9.417113853090675e-05,
      "loss": 0.137,
      "step": 606
    },
    {
      "epoch": 0.24020577760189948,
      "grad_norm": 0.8027547947526502,
      "learning_rate": 9.413872881131518e-05,
      "loss": 0.1754,
      "step": 607
    },
    {
      "epoch": 0.24060150375939848,
      "grad_norm": 0.671864441942157,
      "learning_rate": 9.410623484780168e-05,
      "loss": 0.1527,
      "step": 608
    },
    {
      "epoch": 0.2409972299168975,
      "grad_norm": 1.1548928413207609,
      "learning_rate": 9.407365670238479e-05,
      "loss": 0.2031,
      "step": 609
    },
    {
      "epoch": 0.2413929560743965,
      "grad_norm": 1.3152261646506562,
      "learning_rate": 9.404099443724368e-05,
      "loss": 0.2406,
      "step": 610
    },
    {
      "epoch": 0.2413929560743965,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9422222222222222,
      "eval_PRM F1 AUC": 0.6795376291195278,
      "eval_PRM F1 AUC (fixed)": 0.8430890309886866,
      "eval_PRM F1 Neg": 0.5185185185185185,
      "eval_PRM NPV": 0.875,
      "eval_PRM Precision": 0.8983050847457628,
      "eval_PRM Recall": 0.9906542056074766,
      "eval_PRM Specificty": 0.3684210526315789,
      "eval_loss": 0.38823240995407104,
      "eval_runtime": 5.115,
      "eval_samples_per_second": 5.865,
      "eval_steps_per_second": 0.196,
      "step": 610
    },
    {
      "epoch": 0.24178868223189554,
      "grad_norm": 1.8054059950134695,
      "learning_rate": 9.400824811471811e-05,
      "loss": 0.2868,
      "step": 611
    },
    {
      "epoch": 0.24218440838939453,
      "grad_norm": 1.483288223294385,
      "learning_rate": 9.397541779730827e-05,
      "loss": 0.2897,
      "step": 612
    },
    {
      "epoch": 0.24258013454689356,
      "grad_norm": 0.8151158082075218,
      "learning_rate": 9.394250354767467e-05,
      "loss": 0.2069,
      "step": 613
    },
    {
      "epoch": 0.24297586070439256,
      "grad_norm": 0.9609947966439928,
      "learning_rate": 9.390950542863797e-05,
      "loss": 0.2153,
      "step": 614
    },
    {
      "epoch": 0.24337158686189156,
      "grad_norm": 1.0979707104056993,
      "learning_rate": 9.387642350317894e-05,
      "loss": 0.2064,
      "step": 615
    },
    {
      "epoch": 0.24337158686189156,
      "eval_PRM Accuracy": 0.8809523809523809,
      "eval_PRM F1": 0.9321266968325792,
      "eval_PRM F1 AUC": 0.6918347270044269,
      "eval_PRM F1 AUC (fixed)": 0.8428430890309887,
      "eval_PRM F1 Neg": 0.5161290322580645,
      "eval_PRM NPV": 0.6666666666666666,
      "eval_PRM Precision": 0.9035087719298246,
      "eval_PRM Recall": 0.9626168224299065,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.3037109375,
      "eval_runtime": 4.9687,
      "eval_samples_per_second": 6.038,
      "eval_steps_per_second": 0.201,
      "step": 615
    },
    {
      "epoch": 0.24376731301939059,
      "grad_norm": 0.8619863219027731,
      "learning_rate": 9.384325783443832e-05,
      "loss": 0.1569,
      "step": 616
    },
    {
      "epoch": 0.24416303917688958,
      "grad_norm": 0.8612757328902113,
      "learning_rate": 9.381000848571666e-05,
      "loss": 0.2054,
      "step": 617
    },
    {
      "epoch": 0.2445587653343886,
      "grad_norm": 1.1283574341271685,
      "learning_rate": 9.377667552047423e-05,
      "loss": 0.1545,
      "step": 618
    },
    {
      "epoch": 0.2449544914918876,
      "grad_norm": 0.7739505180687086,
      "learning_rate": 9.374325900233088e-05,
      "loss": 0.1364,
      "step": 619
    },
    {
      "epoch": 0.24535021764938664,
      "grad_norm": 0.9957171774689368,
      "learning_rate": 9.370975899506593e-05,
      "loss": 0.1954,
      "step": 620
    },
    {
      "epoch": 0.24535021764938664,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9422222222222222,
      "eval_PRM F1 AUC": 0.6795376291195278,
      "eval_PRM F1 AUC (fixed)": 0.8438268568617807,
      "eval_PRM F1 Neg": 0.5185185185185185,
      "eval_PRM NPV": 0.875,
      "eval_PRM Precision": 0.8983050847457628,
      "eval_PRM Recall": 0.9906542056074766,
      "eval_PRM Specificty": 0.3684210526315789,
      "eval_loss": 0.32962238788604736,
      "eval_runtime": 5.0662,
      "eval_samples_per_second": 5.922,
      "eval_steps_per_second": 0.197,
      "step": 620
    },
    {
      "epoch": 0.24574594380688564,
      "grad_norm": 1.4168595803088824,
      "learning_rate": 9.367617556261808e-05,
      "loss": 0.21,
      "step": 621
    },
    {
      "epoch": 0.24614166996438464,
      "grad_norm": 0.9480516220046619,
      "learning_rate": 9.364250876908522e-05,
      "loss": 0.1425,
      "step": 622
    },
    {
      "epoch": 0.24653739612188366,
      "grad_norm": 0.8553451927402599,
      "learning_rate": 9.360875867872437e-05,
      "loss": 0.1364,
      "step": 623
    },
    {
      "epoch": 0.24693312227938266,
      "grad_norm": 2.176078009465126,
      "learning_rate": 9.357492535595151e-05,
      "loss": 0.2277,
      "step": 624
    },
    {
      "epoch": 0.2473288484368817,
      "grad_norm": 1.1193803787713226,
      "learning_rate": 9.354100886534152e-05,
      "loss": 0.1802,
      "step": 625
    },
    {
      "epoch": 0.2473288484368817,
      "eval_PRM Accuracy": 0.9047619047619048,
      "eval_PRM F1": 0.9464285714285714,
      "eval_PRM F1 AUC": 0.7058534185932119,
      "eval_PRM F1 AUC (fixed)": 0.8524348253812101,
      "eval_PRM F1 Neg": 0.5714285714285714,
      "eval_PRM NPV": 0.8888888888888888,
      "eval_PRM Precision": 0.905982905982906,
      "eval_PRM Recall": 0.9906542056074766,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.34028321504592896,
      "eval_runtime": 4.8865,
      "eval_samples_per_second": 6.139,
      "eval_steps_per_second": 0.205,
      "step": 625
    },
    {
      "epoch": 0.2477245745943807,
      "grad_norm": 1.2630058603031382,
      "learning_rate": 9.350700927162794e-05,
      "loss": 0.1891,
      "step": 626
    },
    {
      "epoch": 0.24812030075187969,
      "grad_norm": 1.1552702223291675,
      "learning_rate": 9.347292663970301e-05,
      "loss": 0.1836,
      "step": 627
    },
    {
      "epoch": 0.2485160269093787,
      "grad_norm": 1.033400900315006,
      "learning_rate": 9.34387610346174e-05,
      "loss": 0.1597,
      "step": 628
    },
    {
      "epoch": 0.2489117530668777,
      "grad_norm": 1.068934167801897,
      "learning_rate": 9.340451252158015e-05,
      "loss": 0.1603,
      "step": 629
    },
    {
      "epoch": 0.24930747922437674,
      "grad_norm": 0.6160709854200137,
      "learning_rate": 9.337018116595855e-05,
      "loss": 0.1179,
      "step": 630
    },
    {
      "epoch": 0.24930747922437674,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.8553861288735858,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.3403157591819763,
      "eval_runtime": 5.114,
      "eval_samples_per_second": 5.866,
      "eval_steps_per_second": 0.196,
      "step": 630
    },
    {
      "epoch": 0.24970320538187574,
      "grad_norm": 0.7854758690178264,
      "learning_rate": 9.333576703327803e-05,
      "loss": 0.1222,
      "step": 631
    },
    {
      "epoch": 0.25009893153937474,
      "grad_norm": 0.9830013903638098,
      "learning_rate": 9.330127018922194e-05,
      "loss": 0.1573,
      "step": 632
    },
    {
      "epoch": 0.2504946576968738,
      "grad_norm": 0.8726529687641472,
      "learning_rate": 9.326669069963156e-05,
      "loss": 0.1647,
      "step": 633
    },
    {
      "epoch": 0.2508903838543728,
      "grad_norm": 1.3151345227205198,
      "learning_rate": 9.32320286305059e-05,
      "loss": 0.2453,
      "step": 634
    },
    {
      "epoch": 0.2512861100118718,
      "grad_norm": 0.7023228968713068,
      "learning_rate": 9.319728404800157e-05,
      "loss": 0.1018,
      "step": 635
    },
    {
      "epoch": 0.2512861100118718,
      "eval_PRM Accuracy": 0.9047619047619048,
      "eval_PRM F1": 0.9464285714285714,
      "eval_PRM F1 AUC": 0.7058534185932119,
      "eval_PRM F1 AUC (fixed)": 0.8598130841121495,
      "eval_PRM F1 Neg": 0.5714285714285714,
      "eval_PRM NPV": 0.8888888888888888,
      "eval_PRM Precision": 0.905982905982906,
      "eval_PRM Recall": 0.9906542056074766,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.3390462100505829,
      "eval_runtime": 4.9693,
      "eval_samples_per_second": 6.037,
      "eval_steps_per_second": 0.201,
      "step": 635
    },
    {
      "epoch": 0.2516818361693708,
      "grad_norm": 0.7983733224023781,
      "learning_rate": 9.316245701843266e-05,
      "loss": 0.1775,
      "step": 636
    },
    {
      "epoch": 0.2520775623268698,
      "grad_norm": 1.4342366435295932,
      "learning_rate": 9.312754760827061e-05,
      "loss": 0.1809,
      "step": 637
    },
    {
      "epoch": 0.25247328848436884,
      "grad_norm": 0.9750865337714575,
      "learning_rate": 9.309255588414412e-05,
      "loss": 0.1549,
      "step": 638
    },
    {
      "epoch": 0.25286901464186784,
      "grad_norm": 0.9562341987446537,
      "learning_rate": 9.305748191283898e-05,
      "loss": 0.1988,
      "step": 639
    },
    {
      "epoch": 0.25326474079936684,
      "grad_norm": 1.0368633553672248,
      "learning_rate": 9.302232576129797e-05,
      "loss": 0.118,
      "step": 640
    },
    {
      "epoch": 0.25326474079936684,
      "eval_PRM Accuracy": 0.8809523809523809,
      "eval_PRM F1": 0.9321266968325792,
      "eval_PRM F1 AUC": 0.6918347270044269,
      "eval_PRM F1 AUC (fixed)": 0.8625184456468273,
      "eval_PRM F1 Neg": 0.5161290322580645,
      "eval_PRM NPV": 0.6666666666666666,
      "eval_PRM Precision": 0.9035087719298246,
      "eval_PRM Recall": 0.9626168224299065,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.30970051884651184,
      "eval_runtime": 5.1568,
      "eval_samples_per_second": 5.818,
      "eval_steps_per_second": 0.194,
      "step": 640
    },
    {
      "epoch": 0.25366046695686584,
      "grad_norm": 0.8142380529929587,
      "learning_rate": 9.29870874966207e-05,
      "loss": 0.1667,
      "step": 641
    },
    {
      "epoch": 0.25405619311436484,
      "grad_norm": 0.8805785694244057,
      "learning_rate": 9.295176718606355e-05,
      "loss": 0.1907,
      "step": 642
    },
    {
      "epoch": 0.2544519192718639,
      "grad_norm": 0.9212718889915351,
      "learning_rate": 9.291636489703943e-05,
      "loss": 0.1987,
      "step": 643
    },
    {
      "epoch": 0.2548476454293629,
      "grad_norm": 0.9962379537993588,
      "learning_rate": 9.288088069711774e-05,
      "loss": 0.2247,
      "step": 644
    },
    {
      "epoch": 0.2552433715868619,
      "grad_norm": 1.169838474412042,
      "learning_rate": 9.284531465402424e-05,
      "loss": 0.2164,
      "step": 645
    },
    {
      "epoch": 0.2552433715868619,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9422222222222222,
      "eval_PRM F1 AUC": 0.6795376291195278,
      "eval_PRM F1 AUC (fixed)": 0.8632562715199213,
      "eval_PRM F1 Neg": 0.5185185185185185,
      "eval_PRM NPV": 0.875,
      "eval_PRM Precision": 0.8983050847457628,
      "eval_PRM Recall": 0.9906542056074766,
      "eval_PRM Specificty": 0.3684210526315789,
      "eval_loss": 0.3142741024494171,
      "eval_runtime": 5.0218,
      "eval_samples_per_second": 5.974,
      "eval_steps_per_second": 0.199,
      "step": 645
    },
    {
      "epoch": 0.2556390977443609,
      "grad_norm": 0.9185654433989892,
      "learning_rate": 9.280966683564088e-05,
      "loss": 0.2482,
      "step": 646
    },
    {
      "epoch": 0.2560348239018599,
      "grad_norm": 1.3013352290446198,
      "learning_rate": 9.277393731000568e-05,
      "loss": 0.285,
      "step": 647
    },
    {
      "epoch": 0.25643055005935894,
      "grad_norm": 0.763502671129575,
      "learning_rate": 9.273812614531261e-05,
      "loss": 0.1462,
      "step": 648
    },
    {
      "epoch": 0.25682627621685794,
      "grad_norm": 0.8298355780986364,
      "learning_rate": 9.270223340991147e-05,
      "loss": 0.1856,
      "step": 649
    },
    {
      "epoch": 0.25722200237435694,
      "grad_norm": 0.8506194786434039,
      "learning_rate": 9.266625917230774e-05,
      "loss": 0.1553,
      "step": 650
    },
    {
      "epoch": 0.25722200237435694,
      "eval_PRM Accuracy": 0.8809523809523809,
      "eval_PRM F1": 0.9321266968325792,
      "eval_PRM F1 AUC": 0.6918347270044269,
      "eval_PRM F1 AUC (fixed)": 0.8637481554353172,
      "eval_PRM F1 Neg": 0.5161290322580645,
      "eval_PRM NPV": 0.6666666666666666,
      "eval_PRM Precision": 0.9035087719298246,
      "eval_PRM Recall": 0.9626168224299065,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.28649088740348816,
      "eval_runtime": 5.1222,
      "eval_samples_per_second": 5.857,
      "eval_steps_per_second": 0.195,
      "step": 650
    },
    {
      "epoch": 0.25761772853185594,
      "grad_norm": 1.22543292326297,
      "learning_rate": 9.263020350116247e-05,
      "loss": 0.2312,
      "step": 651
    },
    {
      "epoch": 0.258013454689355,
      "grad_norm": 0.7237414074956429,
      "learning_rate": 9.259406646529209e-05,
      "loss": 0.1754,
      "step": 652
    },
    {
      "epoch": 0.258409180846854,
      "grad_norm": 1.0021829845702939,
      "learning_rate": 9.255784813366837e-05,
      "loss": 0.1796,
      "step": 653
    },
    {
      "epoch": 0.258804907004353,
      "grad_norm": 0.8674056319285317,
      "learning_rate": 9.252154857541825e-05,
      "loss": 0.1464,
      "step": 654
    },
    {
      "epoch": 0.259200633161852,
      "grad_norm": 1.1397401398742075,
      "learning_rate": 9.248516785982364e-05,
      "loss": 0.2552,
      "step": 655
    },
    {
      "epoch": 0.259200633161852,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9422222222222222,
      "eval_PRM F1 AUC": 0.6795376291195278,
      "eval_PRM F1 AUC (fixed)": 0.8679291687161831,
      "eval_PRM F1 Neg": 0.5185185185185185,
      "eval_PRM NPV": 0.875,
      "eval_PRM Precision": 0.8983050847457628,
      "eval_PRM Recall": 0.9906542056074766,
      "eval_PRM Specificty": 0.3684210526315789,
      "eval_loss": 0.3284505307674408,
      "eval_runtime": 5.0817,
      "eval_samples_per_second": 5.904,
      "eval_steps_per_second": 0.197,
      "step": 655
    },
    {
      "epoch": 0.259596359319351,
      "grad_norm": 0.9211209637559996,
      "learning_rate": 9.244870605632142e-05,
      "loss": 0.2238,
      "step": 656
    },
    {
      "epoch": 0.25999208547685004,
      "grad_norm": 1.2824954013546763,
      "learning_rate": 9.24121632345032e-05,
      "loss": 0.2338,
      "step": 657
    },
    {
      "epoch": 0.26038781163434904,
      "grad_norm": 1.0663396352881847,
      "learning_rate": 9.237553946411519e-05,
      "loss": 0.2045,
      "step": 658
    },
    {
      "epoch": 0.26078353779184804,
      "grad_norm": 0.8001839653771903,
      "learning_rate": 9.233883481505817e-05,
      "loss": 0.2267,
      "step": 659
    },
    {
      "epoch": 0.26117926394934704,
      "grad_norm": 0.7498852961610969,
      "learning_rate": 9.230204935738725e-05,
      "loss": 0.172,
      "step": 660
    },
    {
      "epoch": 0.26117926394934704,
      "eval_PRM Accuracy": 0.8888888888888888,
      "eval_PRM F1": 0.9375,
      "eval_PRM F1 AUC": 0.6748647319232661,
      "eval_PRM F1 AUC (fixed)": 0.8647319232661091,
      "eval_PRM F1 Neg": 0.5,
      "eval_PRM NPV": 0.7777777777777778,
      "eval_PRM Precision": 0.8974358974358975,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.3684210526315789,
      "eval_loss": 0.2878580689430237,
      "eval_runtime": 5.1742,
      "eval_samples_per_second": 5.798,
      "eval_steps_per_second": 0.193,
      "step": 660
    },
    {
      "epoch": 0.26157499010684604,
      "grad_norm": 1.0755131951508787,
      "learning_rate": 9.226518316131176e-05,
      "loss": 0.1972,
      "step": 661
    },
    {
      "epoch": 0.2619707162643451,
      "grad_norm": 0.6873731378290169,
      "learning_rate": 9.222823629719516e-05,
      "loss": 0.1689,
      "step": 662
    },
    {
      "epoch": 0.2623664424218441,
      "grad_norm": 0.6996977850198727,
      "learning_rate": 9.219120883555486e-05,
      "loss": 0.1789,
      "step": 663
    },
    {
      "epoch": 0.2627621685793431,
      "grad_norm": 0.8330640110587905,
      "learning_rate": 9.215410084706206e-05,
      "loss": 0.1788,
      "step": 664
    },
    {
      "epoch": 0.2631578947368421,
      "grad_norm": 0.6960659280191442,
      "learning_rate": 9.211691240254173e-05,
      "loss": 0.1782,
      "step": 665
    },
    {
      "epoch": 0.2631578947368421,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9422222222222222,
      "eval_PRM F1 AUC": 0.6795376291195278,
      "eval_PRM F1 AUC (fixed)": 0.8735858337432366,
      "eval_PRM F1 Neg": 0.5185185185185185,
      "eval_PRM NPV": 0.875,
      "eval_PRM Precision": 0.8983050847457628,
      "eval_PRM Recall": 0.9906542056074766,
      "eval_PRM Specificty": 0.3684210526315789,
      "eval_loss": 0.30374348163604736,
      "eval_runtime": 4.9925,
      "eval_samples_per_second": 6.009,
      "eval_steps_per_second": 0.2,
      "step": 665
    },
    {
      "epoch": 0.2635536208943411,
      "grad_norm": 0.636245322646594,
      "learning_rate": 9.207964357297235e-05,
      "loss": 0.1565,
      "step": 666
    },
    {
      "epoch": 0.26394934705184014,
      "grad_norm": 0.7137372039946666,
      "learning_rate": 9.204229442948585e-05,
      "loss": 0.1796,
      "step": 667
    },
    {
      "epoch": 0.26434507320933914,
      "grad_norm": 1.0119871222857901,
      "learning_rate": 9.20048650433674e-05,
      "loss": 0.2521,
      "step": 668
    },
    {
      "epoch": 0.26474079936683814,
      "grad_norm": 0.810789129823208,
      "learning_rate": 9.196735548605541e-05,
      "loss": 0.1693,
      "step": 669
    },
    {
      "epoch": 0.26513652552433714,
      "grad_norm": 1.4243984555870466,
      "learning_rate": 9.19297658291412e-05,
      "loss": 0.1752,
      "step": 670
    },
    {
      "epoch": 0.26513652552433714,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9422222222222222,
      "eval_PRM F1 AUC": 0.6795376291195278,
      "eval_PRM F1 AUC (fixed)": 0.8812100344318741,
      "eval_PRM F1 Neg": 0.5185185185185185,
      "eval_PRM NPV": 0.875,
      "eval_PRM Precision": 0.8983050847457628,
      "eval_PRM Recall": 0.9906542056074766,
      "eval_PRM Specificty": 0.3684210526315789,
      "eval_loss": 0.30504557490348816,
      "eval_runtime": 5.1828,
      "eval_samples_per_second": 5.788,
      "eval_steps_per_second": 0.193,
      "step": 670
    },
    {
      "epoch": 0.2655322516818362,
      "grad_norm": 0.6983153438067933,
      "learning_rate": 9.189209614436906e-05,
      "loss": 0.1612,
      "step": 671
    },
    {
      "epoch": 0.2659279778393352,
      "grad_norm": 0.6844516702460982,
      "learning_rate": 9.185434650363596e-05,
      "loss": 0.1085,
      "step": 672
    },
    {
      "epoch": 0.2663237039968342,
      "grad_norm": 0.9623071331666166,
      "learning_rate": 9.181651697899152e-05,
      "loss": 0.1929,
      "step": 673
    },
    {
      "epoch": 0.2667194301543332,
      "grad_norm": 0.9791661878102461,
      "learning_rate": 9.177860764263779e-05,
      "loss": 0.1841,
      "step": 674
    },
    {
      "epoch": 0.2671151563118322,
      "grad_norm": 1.0575489340859652,
      "learning_rate": 9.174061856692919e-05,
      "loss": 0.1889,
      "step": 675
    },
    {
      "epoch": 0.2671151563118322,
      "eval_PRM Accuracy": 0.8809523809523809,
      "eval_PRM F1": 0.9327354260089686,
      "eval_PRM F1 AUC": 0.6701918347270044,
      "eval_PRM F1 AUC (fixed)": 0.8888342351205116,
      "eval_PRM F1 Neg": 0.4827586206896552,
      "eval_PRM NPV": 0.7,
      "eval_PRM Precision": 0.896551724137931,
      "eval_PRM Recall": 0.9719626168224299,
      "eval_PRM Specificty": 0.3684210526315789,
      "eval_loss": 0.30569660663604736,
      "eval_runtime": 5.2772,
      "eval_samples_per_second": 5.685,
      "eval_steps_per_second": 0.189,
      "step": 675
    },
    {
      "epoch": 0.26751088246933125,
      "grad_norm": 0.9711032560120395,
      "learning_rate": 9.17025498243723e-05,
      "loss": 0.1877,
      "step": 676
    },
    {
      "epoch": 0.26790660862683024,
      "grad_norm": 0.871454567740767,
      "learning_rate": 9.166440148762576e-05,
      "loss": 0.1415,
      "step": 677
    },
    {
      "epoch": 0.26830233478432924,
      "grad_norm": 1.0870337171879758,
      "learning_rate": 9.162617362950015e-05,
      "loss": 0.2121,
      "step": 678
    },
    {
      "epoch": 0.26869806094182824,
      "grad_norm": 1.0263145249551762,
      "learning_rate": 9.158786632295776e-05,
      "loss": 0.2044,
      "step": 679
    },
    {
      "epoch": 0.26909378709932724,
      "grad_norm": 1.5736627103678087,
      "learning_rate": 9.15494796411126e-05,
      "loss": 0.2348,
      "step": 680
    },
    {
      "epoch": 0.26909378709932724,
      "eval_PRM Accuracy": 0.8888888888888888,
      "eval_PRM F1": 0.9369369369369369,
      "eval_PRM F1 AUC": 0.6965076242006887,
      "eval_PRM F1 AUC (fixed)": 0.883177570093458,
      "eval_PRM F1 Neg": 0.5333333333333333,
      "eval_PRM NPV": 0.7272727272727273,
      "eval_PRM Precision": 0.9043478260869565,
      "eval_PRM Recall": 0.9719626168224299,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.294921875,
      "eval_runtime": 4.8969,
      "eval_samples_per_second": 6.126,
      "eval_steps_per_second": 0.204,
      "step": 680
    },
    {
      "epoch": 0.2694895132568263,
      "grad_norm": 0.9592710551250099,
      "learning_rate": 9.151101365723013e-05,
      "loss": 0.1873,
      "step": 681
    },
    {
      "epoch": 0.2698852394143253,
      "grad_norm": 1.3107103493867434,
      "learning_rate": 9.147246844472716e-05,
      "loss": 0.1817,
      "step": 682
    },
    {
      "epoch": 0.2702809655718243,
      "grad_norm": 0.7689651641593306,
      "learning_rate": 9.143384407717175e-05,
      "loss": 0.141,
      "step": 683
    },
    {
      "epoch": 0.2706766917293233,
      "grad_norm": 0.9300396795313893,
      "learning_rate": 9.1395140628283e-05,
      "loss": 0.1624,
      "step": 684
    },
    {
      "epoch": 0.2710724178868223,
      "grad_norm": 1.1186614704956632,
      "learning_rate": 9.1356358171931e-05,
      "loss": 0.1516,
      "step": 685
    },
    {
      "epoch": 0.2710724178868223,
      "eval_PRM Accuracy": 0.8809523809523809,
      "eval_PRM F1": 0.9327354260089686,
      "eval_PRM F1 AUC": 0.6701918347270044,
      "eval_PRM F1 AUC (fixed)": 0.8785046728971962,
      "eval_PRM F1 Neg": 0.4827586206896552,
      "eval_PRM NPV": 0.7,
      "eval_PRM Precision": 0.896551724137931,
      "eval_PRM Recall": 0.9719626168224299,
      "eval_PRM Specificty": 0.3684210526315789,
      "eval_loss": 0.3026367127895355,
      "eval_runtime": 5.1819,
      "eval_samples_per_second": 5.789,
      "eval_steps_per_second": 0.193,
      "step": 685
    },
    {
      "epoch": 0.27146814404432135,
      "grad_norm": 1.254189960241517,
      "learning_rate": 9.131749678213657e-05,
      "loss": 0.2783,
      "step": 686
    },
    {
      "epoch": 0.27186387020182035,
      "grad_norm": 1.0113966691212009,
      "learning_rate": 9.127855653307123e-05,
      "loss": 0.2462,
      "step": 687
    },
    {
      "epoch": 0.27225959635931934,
      "grad_norm": 0.8125794573843638,
      "learning_rate": 9.123953749905697e-05,
      "loss": 0.1396,
      "step": 688
    },
    {
      "epoch": 0.27265532251681834,
      "grad_norm": 0.9128450199165946,
      "learning_rate": 9.12004397545662e-05,
      "loss": 0.1608,
      "step": 689
    },
    {
      "epoch": 0.2730510486743174,
      "grad_norm": 0.5624585862098771,
      "learning_rate": 9.116126337422152e-05,
      "loss": 0.1436,
      "step": 690
    },
    {
      "epoch": 0.2730510486743174,
      "eval_PRM Accuracy": 0.8809523809523809,
      "eval_PRM F1": 0.9327354260089686,
      "eval_PRM F1 AUC": 0.6701918347270044,
      "eval_PRM F1 AUC (fixed)": 0.8669454008853911,
      "eval_PRM F1 Neg": 0.4827586206896552,
      "eval_PRM NPV": 0.7,
      "eval_PRM Precision": 0.896551724137931,
      "eval_PRM Recall": 0.9719626168224299,
      "eval_PRM Specificty": 0.3684210526315789,
      "eval_loss": 0.2922200560569763,
      "eval_runtime": 5.3417,
      "eval_samples_per_second": 5.616,
      "eval_steps_per_second": 0.187,
      "step": 690
    },
    {
      "epoch": 0.2734467748318164,
      "grad_norm": 0.7591065889823867,
      "learning_rate": 9.112200843279565e-05,
      "loss": 0.2158,
      "step": 691
    },
    {
      "epoch": 0.2738425009893154,
      "grad_norm": 0.9354338454813161,
      "learning_rate": 9.108267500521121e-05,
      "loss": 0.2026,
      "step": 692
    },
    {
      "epoch": 0.2742382271468144,
      "grad_norm": 0.7268205891367698,
      "learning_rate": 9.104326316654067e-05,
      "loss": 0.1468,
      "step": 693
    },
    {
      "epoch": 0.2746339533043134,
      "grad_norm": 1.4335552146734205,
      "learning_rate": 9.10037729920061e-05,
      "loss": 0.223,
      "step": 694
    },
    {
      "epoch": 0.27502967946181245,
      "grad_norm": 0.7555211334022334,
      "learning_rate": 9.096420455697914e-05,
      "loss": 0.1987,
      "step": 695
    },
    {
      "epoch": 0.27502967946181245,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9422222222222222,
      "eval_PRM F1 AUC": 0.6795376291195278,
      "eval_PRM F1 AUC (fixed)": 0.8627643876045253,
      "eval_PRM F1 Neg": 0.5185185185185185,
      "eval_PRM NPV": 0.875,
      "eval_PRM Precision": 0.8983050847457628,
      "eval_PRM Recall": 0.9906542056074766,
      "eval_PRM Specificty": 0.3684210526315789,
      "eval_loss": 0.2944987118244171,
      "eval_runtime": 5.047,
      "eval_samples_per_second": 5.944,
      "eval_steps_per_second": 0.198,
      "step": 695
    },
    {
      "epoch": 0.27542540561931145,
      "grad_norm": 0.8189189658622638,
      "learning_rate": 9.092455793698075e-05,
      "loss": 0.1846,
      "step": 696
    },
    {
      "epoch": 0.27582113177681045,
      "grad_norm": 0.6793424965135729,
      "learning_rate": 9.088483320768115e-05,
      "loss": 0.1841,
      "step": 697
    },
    {
      "epoch": 0.27621685793430945,
      "grad_norm": 1.1346566755573384,
      "learning_rate": 9.08450304448996e-05,
      "loss": 0.2525,
      "step": 698
    },
    {
      "epoch": 0.27661258409180844,
      "grad_norm": 0.6534103036603515,
      "learning_rate": 9.080514972460439e-05,
      "loss": 0.1462,
      "step": 699
    },
    {
      "epoch": 0.2770083102493075,
      "grad_norm": 0.7628441572243372,
      "learning_rate": 9.076519112291246e-05,
      "loss": 0.1739,
      "step": 700
    },
    {
      "epoch": 0.2770083102493075,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9422222222222222,
      "eval_PRM F1 AUC": 0.6795376291195278,
      "eval_PRM F1 AUC (fixed)": 0.8750614854894245,
      "eval_PRM F1 Neg": 0.5185185185185185,
      "eval_PRM NPV": 0.875,
      "eval_PRM Precision": 0.8983050847457628,
      "eval_PRM Recall": 0.9906542056074766,
      "eval_PRM Specificty": 0.3684210526315789,
      "eval_loss": 0.2897135317325592,
      "eval_runtime": 5.2861,
      "eval_samples_per_second": 5.675,
      "eval_steps_per_second": 0.189,
      "step": 700
    },
    {
      "epoch": 0.2774040364068065,
      "grad_norm": 0.6607457838568496,
      "learning_rate": 9.07251547160895e-05,
      "loss": 0.1313,
      "step": 701
    },
    {
      "epoch": 0.2777997625643055,
      "grad_norm": 0.8533872530467852,
      "learning_rate": 9.06850405805497e-05,
      "loss": 0.1595,
      "step": 702
    },
    {
      "epoch": 0.2781954887218045,
      "grad_norm": 0.6476126920331056,
      "learning_rate": 9.064484879285555e-05,
      "loss": 0.1456,
      "step": 703
    },
    {
      "epoch": 0.2785912148793035,
      "grad_norm": 0.982952385932065,
      "learning_rate": 9.060457942971776e-05,
      "loss": 0.2164,
      "step": 704
    },
    {
      "epoch": 0.27898694103680255,
      "grad_norm": 0.7667149787248076,
      "learning_rate": 9.056423256799513e-05,
      "loss": 0.1383,
      "step": 705
    },
    {
      "epoch": 0.27898694103680255,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9422222222222222,
      "eval_PRM F1 AUC": 0.6795376291195278,
      "eval_PRM F1 AUC (fixed)": 0.883669454008854,
      "eval_PRM F1 Neg": 0.5185185185185185,
      "eval_PRM NPV": 0.875,
      "eval_PRM Precision": 0.8983050847457628,
      "eval_PRM Recall": 0.9906542056074766,
      "eval_PRM Specificty": 0.3684210526315789,
      "eval_loss": 0.2916015684604645,
      "eval_runtime": 4.9834,
      "eval_samples_per_second": 6.02,
      "eval_steps_per_second": 0.201,
      "step": 705
    },
    {
      "epoch": 0.27938266719430155,
      "grad_norm": 0.9769531041036345,
      "learning_rate": 9.052380828469436e-05,
      "loss": 0.1637,
      "step": 706
    },
    {
      "epoch": 0.27977839335180055,
      "grad_norm": 0.9214654827890637,
      "learning_rate": 9.048330665696993e-05,
      "loss": 0.1894,
      "step": 707
    },
    {
      "epoch": 0.28017411950929955,
      "grad_norm": 1.02055901185936,
      "learning_rate": 9.04427277621239e-05,
      "loss": 0.2138,
      "step": 708
    },
    {
      "epoch": 0.2805698456667986,
      "grad_norm": 1.638505514806321,
      "learning_rate": 9.040207167760586e-05,
      "loss": 0.293,
      "step": 709
    },
    {
      "epoch": 0.2809655718242976,
      "grad_norm": 0.5103129313867601,
      "learning_rate": 9.036133848101269e-05,
      "loss": 0.1144,
      "step": 710
    },
    {
      "epoch": 0.2809655718242976,
      "eval_PRM Accuracy": 0.8888888888888888,
      "eval_PRM F1": 0.9369369369369369,
      "eval_PRM F1 AUC": 0.6965076242006887,
      "eval_PRM F1 AUC (fixed)": 0.883423512051156,
      "eval_PRM F1 Neg": 0.5333333333333333,
      "eval_PRM NPV": 0.7272727272727273,
      "eval_PRM Precision": 0.9043478260869565,
      "eval_PRM Recall": 0.9719626168224299,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.271484375,
      "eval_runtime": 5.0799,
      "eval_samples_per_second": 5.906,
      "eval_steps_per_second": 0.197,
      "step": 710
    },
    {
      "epoch": 0.2813612979817966,
      "grad_norm": 0.851892128891332,
      "learning_rate": 9.032052825008845e-05,
      "loss": 0.1526,
      "step": 711
    },
    {
      "epoch": 0.2817570241392956,
      "grad_norm": 0.8483666128644993,
      "learning_rate": 9.027964106272423e-05,
      "loss": 0.1492,
      "step": 712
    },
    {
      "epoch": 0.2821527502967946,
      "grad_norm": 0.7632322196341024,
      "learning_rate": 9.023867699695804e-05,
      "loss": 0.1646,
      "step": 713
    },
    {
      "epoch": 0.28254847645429365,
      "grad_norm": 0.7612194930893317,
      "learning_rate": 9.019763613097455e-05,
      "loss": 0.1688,
      "step": 714
    },
    {
      "epoch": 0.28294420261179265,
      "grad_norm": 1.1141932691697012,
      "learning_rate": 9.015651854310506e-05,
      "loss": 0.2438,
      "step": 715
    },
    {
      "epoch": 0.28294420261179265,
      "eval_PRM Accuracy": 0.9047619047619048,
      "eval_PRM F1": 0.9464285714285714,
      "eval_PRM F1 AUC": 0.7058534185932119,
      "eval_PRM F1 AUC (fixed)": 0.8876045253320216,
      "eval_PRM F1 Neg": 0.5714285714285714,
      "eval_PRM NPV": 0.8888888888888888,
      "eval_PRM Precision": 0.905982905982906,
      "eval_PRM Recall": 0.9906542056074766,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.29472655057907104,
      "eval_runtime": 4.9817,
      "eval_samples_per_second": 6.022,
      "eval_steps_per_second": 0.201,
      "step": 715
    },
    {
      "epoch": 0.28333992876929165,
      "grad_norm": 1.1587755741290395,
      "learning_rate": 9.011532431182729e-05,
      "loss": 0.1748,
      "step": 716
    },
    {
      "epoch": 0.28373565492679065,
      "grad_norm": 1.5392021948756802,
      "learning_rate": 9.007405351576524e-05,
      "loss": 0.3444,
      "step": 717
    },
    {
      "epoch": 0.28413138108428965,
      "grad_norm": 0.7938171697497031,
      "learning_rate": 9.003270623368905e-05,
      "loss": 0.1653,
      "step": 718
    },
    {
      "epoch": 0.2845271072417887,
      "grad_norm": 1.0671784206576023,
      "learning_rate": 8.999128254451486e-05,
      "loss": 0.2085,
      "step": 719
    },
    {
      "epoch": 0.2849228333992877,
      "grad_norm": 0.8571675690584678,
      "learning_rate": 8.99497825273046e-05,
      "loss": 0.2005,
      "step": 720
    },
    {
      "epoch": 0.2849228333992877,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.8772749631087065,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.27314454317092896,
      "eval_runtime": 5.1322,
      "eval_samples_per_second": 5.845,
      "eval_steps_per_second": 0.195,
      "step": 720
    },
    {
      "epoch": 0.2853185595567867,
      "grad_norm": 0.7020104268585275,
      "learning_rate": 8.990820626126589e-05,
      "loss": 0.1178,
      "step": 721
    },
    {
      "epoch": 0.2857142857142857,
      "grad_norm": 0.551517656772806,
      "learning_rate": 8.986655382575192e-05,
      "loss": 0.1306,
      "step": 722
    },
    {
      "epoch": 0.2861100118717847,
      "grad_norm": 0.835937866856043,
      "learning_rate": 8.982482530026122e-05,
      "loss": 0.1967,
      "step": 723
    },
    {
      "epoch": 0.28650573802928375,
      "grad_norm": 0.8368102251324648,
      "learning_rate": 8.978302076443754e-05,
      "loss": 0.1727,
      "step": 724
    },
    {
      "epoch": 0.28690146418678275,
      "grad_norm": 0.8660458425103343,
      "learning_rate": 8.974114029806976e-05,
      "loss": 0.2213,
      "step": 725
    },
    {
      "epoch": 0.28690146418678275,
      "eval_PRM Accuracy": 0.8809523809523809,
      "eval_PRM F1": 0.9321266968325792,
      "eval_PRM F1 AUC": 0.6918347270044269,
      "eval_PRM F1 AUC (fixed)": 0.8856369896704377,
      "eval_PRM F1 Neg": 0.5161290322580645,
      "eval_PRM NPV": 0.6666666666666666,
      "eval_PRM Precision": 0.9035087719298246,
      "eval_PRM Recall": 0.9626168224299065,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.2822916805744171,
      "eval_runtime": 5.1756,
      "eval_samples_per_second": 5.796,
      "eval_steps_per_second": 0.193,
      "step": 725
    },
    {
      "epoch": 0.28729719034428175,
      "grad_norm": 1.4395936296050706,
      "learning_rate": 8.969918398109162e-05,
      "loss": 0.2175,
      "step": 726
    },
    {
      "epoch": 0.28769291650178075,
      "grad_norm": 0.8392540153410275,
      "learning_rate": 8.965715189358164e-05,
      "loss": 0.2052,
      "step": 727
    },
    {
      "epoch": 0.2880886426592798,
      "grad_norm": 0.8498842269373648,
      "learning_rate": 8.9615044115763e-05,
      "loss": 0.23,
      "step": 728
    },
    {
      "epoch": 0.2884843688167788,
      "grad_norm": 1.1484022806367529,
      "learning_rate": 8.95728607280033e-05,
      "loss": 0.1833,
      "step": 729
    },
    {
      "epoch": 0.2888800949742778,
      "grad_norm": 0.6436287635235969,
      "learning_rate": 8.953060181081447e-05,
      "loss": 0.1945,
      "step": 730
    },
    {
      "epoch": 0.2888800949742778,
      "eval_PRM Accuracy": 0.8809523809523809,
      "eval_PRM F1": 0.9327354260089686,
      "eval_PRM F1 AUC": 0.6701918347270044,
      "eval_PRM F1 AUC (fixed)": 0.8917855386128873,
      "eval_PRM F1 Neg": 0.4827586206896552,
      "eval_PRM NPV": 0.7,
      "eval_PRM Precision": 0.896551724137931,
      "eval_PRM Recall": 0.9719626168224299,
      "eval_PRM Specificty": 0.3684210526315789,
      "eval_loss": 0.31298828125,
      "eval_runtime": 5.1871,
      "eval_samples_per_second": 5.784,
      "eval_steps_per_second": 0.193,
      "step": 730
    },
    {
      "epoch": 0.2892758211317768,
      "grad_norm": 0.7279521368670537,
      "learning_rate": 8.948826744485258e-05,
      "loss": 0.1433,
      "step": 731
    },
    {
      "epoch": 0.2896715472892758,
      "grad_norm": 0.9415593375937553,
      "learning_rate": 8.944585771091773e-05,
      "loss": 0.1826,
      "step": 732
    },
    {
      "epoch": 0.29006727344677485,
      "grad_norm": 0.8353676957085796,
      "learning_rate": 8.940337268995385e-05,
      "loss": 0.1902,
      "step": 733
    },
    {
      "epoch": 0.29046299960427385,
      "grad_norm": 1.1591437479608018,
      "learning_rate": 8.936081246304855e-05,
      "loss": 0.2265,
      "step": 734
    },
    {
      "epoch": 0.29085872576177285,
      "grad_norm": 1.0589753650535267,
      "learning_rate": 8.931817711143302e-05,
      "loss": 0.2601,
      "step": 735
    },
    {
      "epoch": 0.29085872576177285,
      "eval_PRM Accuracy": 0.8809523809523809,
      "eval_PRM F1": 0.9327354260089686,
      "eval_PRM F1 AUC": 0.6701918347270044,
      "eval_PRM F1 AUC (fixed)": 0.8922774225282833,
      "eval_PRM F1 Neg": 0.4827586206896552,
      "eval_PRM NPV": 0.7,
      "eval_PRM Precision": 0.896551724137931,
      "eval_PRM Recall": 0.9719626168224299,
      "eval_PRM Specificty": 0.3684210526315789,
      "eval_loss": 0.29293620586395264,
      "eval_runtime": 5.1837,
      "eval_samples_per_second": 5.787,
      "eval_steps_per_second": 0.193,
      "step": 735
    },
    {
      "epoch": 0.29125445191927185,
      "grad_norm": 0.9466532951110602,
      "learning_rate": 8.92754667164818e-05,
      "loss": 0.269,
      "step": 736
    },
    {
      "epoch": 0.29165017807677085,
      "grad_norm": 0.5546897117290948,
      "learning_rate": 8.923268135971267e-05,
      "loss": 0.1516,
      "step": 737
    },
    {
      "epoch": 0.2920459042342699,
      "grad_norm": 0.726795549085925,
      "learning_rate": 8.91898211227865e-05,
      "loss": 0.1738,
      "step": 738
    },
    {
      "epoch": 0.2924416303917689,
      "grad_norm": 0.8758479495103402,
      "learning_rate": 8.914688608750701e-05,
      "loss": 0.1995,
      "step": 739
    },
    {
      "epoch": 0.2928373565492679,
      "grad_norm": 0.9058601441832761,
      "learning_rate": 8.91038763358208e-05,
      "loss": 0.2906,
      "step": 740
    },
    {
      "epoch": 0.2928373565492679,
      "eval_PRM Accuracy": 0.8650793650793651,
      "eval_PRM F1": 0.9251101321585903,
      "eval_PRM F1 AUC": 0.5959173635022135,
      "eval_PRM F1 AUC (fixed)": 0.8883423512051157,
      "eval_PRM F1 Neg": 0.32,
      "eval_PRM NPV": 0.6666666666666666,
      "eval_PRM Precision": 0.875,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.21052631578947367,
      "eval_loss": 0.2775065004825592,
      "eval_runtime": 5.1279,
      "eval_samples_per_second": 5.85,
      "eval_steps_per_second": 0.195,
      "step": 740
    },
    {
      "epoch": 0.2932330827067669,
      "grad_norm": 0.9063094013514543,
      "learning_rate": 8.906079194981695e-05,
      "loss": 0.1671,
      "step": 741
    },
    {
      "epoch": 0.29362880886426596,
      "grad_norm": 0.6651368106596527,
      "learning_rate": 8.901763301172708e-05,
      "loss": 0.1513,
      "step": 742
    },
    {
      "epoch": 0.29402453502176495,
      "grad_norm": 0.5973154384515552,
      "learning_rate": 8.897439960392507e-05,
      "loss": 0.1293,
      "step": 743
    },
    {
      "epoch": 0.29442026117926395,
      "grad_norm": 0.8955840947808993,
      "learning_rate": 8.89310918089269e-05,
      "loss": 0.2392,
      "step": 744
    },
    {
      "epoch": 0.29481598733676295,
      "grad_norm": 0.6206958893144313,
      "learning_rate": 8.888770970939057e-05,
      "loss": 0.1787,
      "step": 745
    },
    {
      "epoch": 0.29481598733676295,
      "eval_PRM Accuracy": 0.873015873015873,
      "eval_PRM F1": 0.9298245614035088,
      "eval_PRM F1 AUC": 0.6005902606984752,
      "eval_PRM F1 AUC (fixed)": 0.8910477127397934,
      "eval_PRM F1 Neg": 0.3333333333333333,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.8760330578512396,
      "eval_PRM Recall": 0.9906542056074766,
      "eval_PRM Specificty": 0.21052631578947367,
      "eval_loss": 0.30009764432907104,
      "eval_runtime": 4.8482,
      "eval_samples_per_second": 6.188,
      "eval_steps_per_second": 0.206,
      "step": 745
    },
    {
      "epoch": 0.29521171349426195,
      "grad_norm": 0.6786888424779063,
      "learning_rate": 8.88442533881159e-05,
      "loss": 0.1603,
      "step": 746
    },
    {
      "epoch": 0.295607439651761,
      "grad_norm": 0.9414471796153763,
      "learning_rate": 8.880072292804435e-05,
      "loss": 0.1964,
      "step": 747
    },
    {
      "epoch": 0.29600316580926,
      "grad_norm": 0.7837874338689773,
      "learning_rate": 8.875711841225888e-05,
      "loss": 0.2305,
      "step": 748
    },
    {
      "epoch": 0.296398891966759,
      "grad_norm": 1.2912999325498091,
      "learning_rate": 8.871343992398383e-05,
      "loss": 0.2515,
      "step": 749
    },
    {
      "epoch": 0.296794618124258,
      "grad_norm": 0.9026489779277271,
      "learning_rate": 8.866968754658466e-05,
      "loss": 0.206,
      "step": 750
    },
    {
      "epoch": 0.296794618124258,
      "eval_PRM Accuracy": 0.8888888888888888,
      "eval_PRM F1": 0.9369369369369369,
      "eval_PRM F1 AUC": 0.6965076242006887,
      "eval_PRM F1 AUC (fixed)": 0.8893261190359075,
      "eval_PRM F1 Neg": 0.5333333333333333,
      "eval_PRM NPV": 0.7272727272727273,
      "eval_PRM Precision": 0.9043478260869565,
      "eval_PRM Recall": 0.9719626168224299,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.26956379413604736,
      "eval_runtime": 5.3776,
      "eval_samples_per_second": 5.579,
      "eval_steps_per_second": 0.186,
      "step": 750
    },
    {
      "epoch": 0.297190344281757,
      "grad_norm": 0.7696797543394589,
      "learning_rate": 8.862586136356794e-05,
      "loss": 0.1674,
      "step": 751
    },
    {
      "epoch": 0.29758607043925606,
      "grad_norm": 0.9602999971402979,
      "learning_rate": 8.858196145858104e-05,
      "loss": 0.1898,
      "step": 752
    },
    {
      "epoch": 0.29798179659675506,
      "grad_norm": 0.8828895719430124,
      "learning_rate": 8.853798791541204e-05,
      "loss": 0.2273,
      "step": 753
    },
    {
      "epoch": 0.29837752275425405,
      "grad_norm": 0.9976380413908069,
      "learning_rate": 8.849394081798962e-05,
      "loss": 0.1945,
      "step": 754
    },
    {
      "epoch": 0.29877324891175305,
      "grad_norm": 0.8342263917602064,
      "learning_rate": 8.844982025038279e-05,
      "loss": 0.2335,
      "step": 755
    },
    {
      "epoch": 0.29877324891175305,
      "eval_PRM Accuracy": 0.8888888888888888,
      "eval_PRM F1": 0.9375,
      "eval_PRM F1 AUC": 0.6748647319232661,
      "eval_PRM F1 AUC (fixed)": 0.8893261190359075,
      "eval_PRM F1 Neg": 0.5,
      "eval_PRM NPV": 0.7777777777777778,
      "eval_PRM Precision": 0.8974358974358975,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.3684210526315789,
      "eval_loss": 0.2806640565395355,
      "eval_runtime": 4.8594,
      "eval_samples_per_second": 6.174,
      "eval_steps_per_second": 0.206,
      "step": 755
    },
    {
      "epoch": 0.29916897506925205,
      "grad_norm": 0.9151882315488452,
      "learning_rate": 8.84056262968008e-05,
      "loss": 0.2161,
      "step": 756
    },
    {
      "epoch": 0.2995647012267511,
      "grad_norm": 0.8459616664970567,
      "learning_rate": 8.836135904159302e-05,
      "loss": 0.1541,
      "step": 757
    },
    {
      "epoch": 0.2999604273842501,
      "grad_norm": 0.6746795344350985,
      "learning_rate": 8.831701856924864e-05,
      "loss": 0.162,
      "step": 758
    },
    {
      "epoch": 0.3003561535417491,
      "grad_norm": 0.7654457054114777,
      "learning_rate": 8.827260496439662e-05,
      "loss": 0.1826,
      "step": 759
    },
    {
      "epoch": 0.3007518796992481,
      "grad_norm": 1.1923245027585447,
      "learning_rate": 8.822811831180555e-05,
      "loss": 0.2223,
      "step": 760
    },
    {
      "epoch": 0.3007518796992481,
      "eval_PRM Accuracy": 0.8650793650793651,
      "eval_PRM F1": 0.925764192139738,
      "eval_PRM F1 AUC": 0.5742744712247909,
      "eval_PRM F1 AUC (fixed)": 0.8925233644859814,
      "eval_PRM F1 Neg": 0.2608695652173913,
      "eval_PRM NPV": 0.75,
      "eval_PRM Precision": 0.8688524590163934,
      "eval_PRM Recall": 0.9906542056074766,
      "eval_PRM Specificty": 0.15789473684210525,
      "eval_loss": 0.3451497256755829,
      "eval_runtime": 5.0736,
      "eval_samples_per_second": 5.913,
      "eval_steps_per_second": 0.197,
      "step": 760
    },
    {
      "epoch": 0.30114760585674716,
      "grad_norm": 1.315315242050645,
      "learning_rate": 8.818355869638339e-05,
      "loss": 0.2256,
      "step": 761
    },
    {
      "epoch": 0.30154333201424616,
      "grad_norm": 0.7026193936703187,
      "learning_rate": 8.81389262031774e-05,
      "loss": 0.2259,
      "step": 762
    },
    {
      "epoch": 0.30193905817174516,
      "grad_norm": 0.7378162205814667,
      "learning_rate": 8.809422091737387e-05,
      "loss": 0.1867,
      "step": 763
    },
    {
      "epoch": 0.30233478432924416,
      "grad_norm": 0.83139474139224,
      "learning_rate": 8.804944292429807e-05,
      "loss": 0.2439,
      "step": 764
    },
    {
      "epoch": 0.30273051048674315,
      "grad_norm": 0.6124547932922225,
      "learning_rate": 8.800459230941405e-05,
      "loss": 0.1712,
      "step": 765
    },
    {
      "epoch": 0.30273051048674315,
      "eval_PRM Accuracy": 0.8888888888888888,
      "eval_PRM F1": 0.9380530973451328,
      "eval_PRM F1 AUC": 0.6532218396458436,
      "eval_PRM F1 AUC (fixed)": 0.8812100344318741,
      "eval_PRM F1 Neg": 0.46153846153846156,
      "eval_PRM NPV": 0.8571428571428571,
      "eval_PRM Precision": 0.8907563025210085,
      "eval_PRM Recall": 0.9906542056074766,
      "eval_PRM Specificty": 0.3157894736842105,
      "eval_loss": 0.27991536259651184,
      "eval_runtime": 5.2088,
      "eval_samples_per_second": 5.759,
      "eval_steps_per_second": 0.192,
      "step": 765
    },
    {
      "epoch": 0.3031262366442422,
      "grad_norm": 0.8159197300922938,
      "learning_rate": 8.795966915832443e-05,
      "loss": 0.182,
      "step": 766
    },
    {
      "epoch": 0.3035219628017412,
      "grad_norm": 0.7908198253381381,
      "learning_rate": 8.791467355677031e-05,
      "loss": 0.2122,
      "step": 767
    },
    {
      "epoch": 0.3039176889592402,
      "grad_norm": 0.7234911177006348,
      "learning_rate": 8.786960559063105e-05,
      "loss": 0.222,
      "step": 768
    },
    {
      "epoch": 0.3043134151167392,
      "grad_norm": 0.7952637819088901,
      "learning_rate": 8.782446534592413e-05,
      "loss": 0.2116,
      "step": 769
    },
    {
      "epoch": 0.3047091412742382,
      "grad_norm": 0.9612974289352592,
      "learning_rate": 8.777925290880496e-05,
      "loss": 0.1918,
      "step": 770
    },
    {
      "epoch": 0.3047091412742382,
      "eval_PRM Accuracy": 0.8888888888888888,
      "eval_PRM F1": 0.9380530973451328,
      "eval_PRM F1 AUC": 0.6532218396458436,
      "eval_PRM F1 AUC (fixed)": 0.8804722085587801,
      "eval_PRM F1 Neg": 0.46153846153846156,
      "eval_PRM NPV": 0.8571428571428571,
      "eval_PRM Precision": 0.8907563025210085,
      "eval_PRM Recall": 0.9906542056074766,
      "eval_PRM Specificty": 0.3157894736842105,
      "eval_loss": 0.28177082538604736,
      "eval_runtime": 4.9519,
      "eval_samples_per_second": 6.058,
      "eval_steps_per_second": 0.202,
      "step": 770
    },
    {
      "epoch": 0.30510486743173726,
      "grad_norm": 0.9087054032483611,
      "learning_rate": 8.773396836556679e-05,
      "loss": 0.2429,
      "step": 771
    },
    {
      "epoch": 0.30550059358923626,
      "grad_norm": 0.7447575951133999,
      "learning_rate": 8.768861180264045e-05,
      "loss": 0.2014,
      "step": 772
    },
    {
      "epoch": 0.30589631974673526,
      "grad_norm": 0.9069651617022895,
      "learning_rate": 8.764318330659424e-05,
      "loss": 0.2401,
      "step": 773
    },
    {
      "epoch": 0.30629204590423426,
      "grad_norm": 0.9521867301096485,
      "learning_rate": 8.759768296413376e-05,
      "loss": 0.1792,
      "step": 774
    },
    {
      "epoch": 0.30668777206173325,
      "grad_norm": 0.6072237551406509,
      "learning_rate": 8.755211086210172e-05,
      "loss": 0.1562,
      "step": 775
    },
    {
      "epoch": 0.30668777206173325,
      "eval_PRM Accuracy": 0.8888888888888888,
      "eval_PRM F1": 0.9380530973451328,
      "eval_PRM F1 AUC": 0.6532218396458436,
      "eval_PRM F1 AUC (fixed)": 0.8799803246433842,
      "eval_PRM F1 Neg": 0.46153846153846156,
      "eval_PRM NPV": 0.8571428571428571,
      "eval_PRM Precision": 0.8907563025210085,
      "eval_PRM Recall": 0.9906542056074766,
      "eval_PRM Specificty": 0.3157894736842105,
      "eval_loss": 0.306640625,
      "eval_runtime": 5.0011,
      "eval_samples_per_second": 5.999,
      "eval_steps_per_second": 0.2,
      "step": 775
    },
    {
      "epoch": 0.3070834982192323,
      "grad_norm": 0.8792892279462375,
      "learning_rate": 8.750646708747781e-05,
      "loss": 0.1507,
      "step": 776
    },
    {
      "epoch": 0.3074792243767313,
      "grad_norm": 0.6571602790522012,
      "learning_rate": 8.746075172737851e-05,
      "loss": 0.1421,
      "step": 777
    },
    {
      "epoch": 0.3078749505342303,
      "grad_norm": 1.0283840524578438,
      "learning_rate": 8.741496486905691e-05,
      "loss": 0.1422,
      "step": 778
    },
    {
      "epoch": 0.3082706766917293,
      "grad_norm": 0.6998294955241987,
      "learning_rate": 8.736910659990261e-05,
      "loss": 0.1464,
      "step": 779
    },
    {
      "epoch": 0.30866640284922836,
      "grad_norm": 1.2720310775742825,
      "learning_rate": 8.732317700744146e-05,
      "loss": 0.2327,
      "step": 780
    },
    {
      "epoch": 0.30866640284922836,
      "eval_PRM Accuracy": 0.8888888888888888,
      "eval_PRM F1": 0.9369369369369369,
      "eval_PRM F1 AUC": 0.6965076242006887,
      "eval_PRM F1 AUC (fixed)": 0.8900639449090014,
      "eval_PRM F1 Neg": 0.5333333333333333,
      "eval_PRM NPV": 0.7272727272727273,
      "eval_PRM Precision": 0.9043478260869565,
      "eval_PRM Recall": 0.9719626168224299,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.2822265625,
      "eval_runtime": 5.103,
      "eval_samples_per_second": 5.879,
      "eval_steps_per_second": 0.196,
      "step": 780
    },
    {
      "epoch": 0.30906212900672736,
      "grad_norm": 0.7342901789664035,
      "learning_rate": 8.727717617933544e-05,
      "loss": 0.1264,
      "step": 781
    },
    {
      "epoch": 0.30945785516422636,
      "grad_norm": 1.1028064366961383,
      "learning_rate": 8.723110420338251e-05,
      "loss": 0.2212,
      "step": 782
    },
    {
      "epoch": 0.30985358132172536,
      "grad_norm": 1.2893229878786714,
      "learning_rate": 8.718496116751644e-05,
      "loss": 0.1484,
      "step": 783
    },
    {
      "epoch": 0.31024930747922436,
      "grad_norm": 0.5643435084286054,
      "learning_rate": 8.71387471598066e-05,
      "loss": 0.0959,
      "step": 784
    },
    {
      "epoch": 0.3106450336367234,
      "grad_norm": 0.7583173597008048,
      "learning_rate": 8.709246226845782e-05,
      "loss": 0.1048,
      "step": 785
    },
    {
      "epoch": 0.3106450336367234,
      "eval_PRM Accuracy": 0.8888888888888888,
      "eval_PRM F1": 0.9369369369369369,
      "eval_PRM F1 AUC": 0.6965076242006887,
      "eval_PRM F1 AUC (fixed)": 0.8999016232169208,
      "eval_PRM F1 Neg": 0.5333333333333333,
      "eval_PRM NPV": 0.7272727272727273,
      "eval_PRM Precision": 0.9043478260869565,
      "eval_PRM Recall": 0.9719626168224299,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.28942057490348816,
      "eval_runtime": 5.0981,
      "eval_samples_per_second": 5.885,
      "eval_steps_per_second": 0.196,
      "step": 785
    },
    {
      "epoch": 0.3110407597942224,
      "grad_norm": 0.8073987997530575,
      "learning_rate": 8.704610658181021e-05,
      "loss": 0.1581,
      "step": 786
    },
    {
      "epoch": 0.3114364859517214,
      "grad_norm": 1.114560450506898,
      "learning_rate": 8.699968018833904e-05,
      "loss": 0.1419,
      "step": 787
    },
    {
      "epoch": 0.3118322121092204,
      "grad_norm": 0.7316494749476418,
      "learning_rate": 8.69531831766545e-05,
      "loss": 0.0942,
      "step": 788
    },
    {
      "epoch": 0.3122279382667194,
      "grad_norm": 2.010956156016771,
      "learning_rate": 8.690661563550156e-05,
      "loss": 0.2824,
      "step": 789
    },
    {
      "epoch": 0.31262366442421846,
      "grad_norm": 1.153707937935504,
      "learning_rate": 8.68599776537598e-05,
      "loss": 0.2088,
      "step": 790
    },
    {
      "epoch": 0.31262366442421846,
      "eval_PRM Accuracy": 0.873015873015873,
      "eval_PRM F1": 0.9285714285714286,
      "eval_PRM F1 AUC": 0.6438760452533202,
      "eval_PRM F1 AUC (fixed)": 0.8991637973438268,
      "eval_PRM F1 Neg": 0.42857142857142855,
      "eval_PRM NPV": 0.6666666666666666,
      "eval_PRM Precision": 0.8888888888888888,
      "eval_PRM Recall": 0.9719626168224299,
      "eval_PRM Specificty": 0.3157894736842105,
      "eval_loss": 0.30595701932907104,
      "eval_runtime": 4.7699,
      "eval_samples_per_second": 6.289,
      "eval_steps_per_second": 0.21,
      "step": 790
    },
    {
      "epoch": 0.31301939058171746,
      "grad_norm": 1.1484907644773685,
      "learning_rate": 8.68132693204433e-05,
      "loss": 0.2056,
      "step": 791
    },
    {
      "epoch": 0.31341511673921646,
      "grad_norm": 1.6777980586879817,
      "learning_rate": 8.676649072470034e-05,
      "loss": 0.1962,
      "step": 792
    },
    {
      "epoch": 0.31381084289671546,
      "grad_norm": 0.7395973147724139,
      "learning_rate": 8.671964195581336e-05,
      "loss": 0.1298,
      "step": 793
    },
    {
      "epoch": 0.31420656905421446,
      "grad_norm": 0.9643682491142119,
      "learning_rate": 8.66727231031987e-05,
      "loss": 0.171,
      "step": 794
    },
    {
      "epoch": 0.3146022952117135,
      "grad_norm": 0.8899614793760491,
      "learning_rate": 8.662573425640645e-05,
      "loss": 0.165,
      "step": 795
    },
    {
      "epoch": 0.3146022952117135,
      "eval_PRM Accuracy": 0.8888888888888888,
      "eval_PRM F1": 0.9369369369369369,
      "eval_PRM F1 AUC": 0.6965076242006887,
      "eval_PRM F1 AUC (fixed)": 0.8962124938514511,
      "eval_PRM F1 Neg": 0.5333333333333333,
      "eval_PRM NPV": 0.7272727272727273,
      "eval_PRM Precision": 0.9043478260869565,
      "eval_PRM Recall": 0.9719626168224299,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.2797200381755829,
      "eval_runtime": 4.8426,
      "eval_samples_per_second": 6.195,
      "eval_steps_per_second": 0.207,
      "step": 795
    },
    {
      "epoch": 0.3149980213692125,
      "grad_norm": 0.9323404131120511,
      "learning_rate": 8.657867550512033e-05,
      "loss": 0.1627,
      "step": 796
    },
    {
      "epoch": 0.3153937475267115,
      "grad_norm": 0.7969863305269954,
      "learning_rate": 8.65315469391575e-05,
      "loss": 0.1364,
      "step": 797
    },
    {
      "epoch": 0.3157894736842105,
      "grad_norm": 0.6073649742768036,
      "learning_rate": 8.64843486484683e-05,
      "loss": 0.1061,
      "step": 798
    },
    {
      "epoch": 0.31618519984170956,
      "grad_norm": 0.9190067904070243,
      "learning_rate": 8.643708072313618e-05,
      "loss": 0.2136,
      "step": 799
    },
    {
      "epoch": 0.31658092599920856,
      "grad_norm": 0.5106188430854406,
      "learning_rate": 8.63897432533775e-05,
      "loss": 0.1453,
      "step": 800
    },
    {
      "epoch": 0.31658092599920856,
      "eval_PRM Accuracy": 0.873015873015873,
      "eval_PRM F1": 0.9292035398230089,
      "eval_PRM F1 AUC": 0.6222331529758978,
      "eval_PRM F1 AUC (fixed)": 0.8942449581898672,
      "eval_PRM F1 Neg": 0.38461538461538464,
      "eval_PRM NPV": 0.7142857142857143,
      "eval_PRM Precision": 0.8823529411764706,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.2631578947368421,
      "eval_loss": 0.3200846314430237,
      "eval_runtime": 5.0855,
      "eval_samples_per_second": 5.899,
      "eval_steps_per_second": 0.197,
      "step": 800
    },
    {
      "epoch": 0.31697665215670756,
      "grad_norm": 0.7289283720156537,
      "learning_rate": 8.634233632954139e-05,
      "loss": 0.1527,
      "step": 801
    },
    {
      "epoch": 0.31737237831420656,
      "grad_norm": 0.8570109277701821,
      "learning_rate": 8.629486004210945e-05,
      "loss": 0.1973,
      "step": 802
    },
    {
      "epoch": 0.31776810447170556,
      "grad_norm": 0.7857443785523225,
      "learning_rate": 8.624731448169576e-05,
      "loss": 0.1507,
      "step": 803
    },
    {
      "epoch": 0.3181638306292046,
      "grad_norm": 0.9553421553523828,
      "learning_rate": 8.619969973904655e-05,
      "loss": 0.2024,
      "step": 804
    },
    {
      "epoch": 0.3185595567867036,
      "grad_norm": 0.6153155522590844,
      "learning_rate": 8.615201590504017e-05,
      "loss": 0.1317,
      "step": 805
    },
    {
      "epoch": 0.3185595567867036,
      "eval_PRM Accuracy": 0.8888888888888888,
      "eval_PRM F1": 0.9375,
      "eval_PRM F1 AUC": 0.6748647319232661,
      "eval_PRM F1 AUC (fixed)": 0.8866207575012297,
      "eval_PRM F1 Neg": 0.5,
      "eval_PRM NPV": 0.7777777777777778,
      "eval_PRM Precision": 0.8974358974358975,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.3684210526315789,
      "eval_loss": 0.3003906309604645,
      "eval_runtime": 5.1121,
      "eval_samples_per_second": 5.868,
      "eval_steps_per_second": 0.196,
      "step": 805
    },
    {
      "epoch": 0.3189552829442026,
      "grad_norm": 0.7682299096727117,
      "learning_rate": 8.610426307068674e-05,
      "loss": 0.1388,
      "step": 806
    },
    {
      "epoch": 0.3193510091017016,
      "grad_norm": 0.8666234152885707,
      "learning_rate": 8.605644132712814e-05,
      "loss": 0.2311,
      "step": 807
    },
    {
      "epoch": 0.3197467352592006,
      "grad_norm": 1.5250090898838904,
      "learning_rate": 8.600855076563776e-05,
      "loss": 0.2397,
      "step": 808
    },
    {
      "epoch": 0.32014246141669966,
      "grad_norm": 1.045702461189082,
      "learning_rate": 8.596059147762034e-05,
      "loss": 0.1861,
      "step": 809
    },
    {
      "epoch": 0.32053818757419866,
      "grad_norm": 0.8749524500579257,
      "learning_rate": 8.591256355461176e-05,
      "loss": 0.1504,
      "step": 810
    },
    {
      "epoch": 0.32053818757419866,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.8863748155435317,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.2663411498069763,
      "eval_runtime": 5.168,
      "eval_samples_per_second": 5.805,
      "eval_steps_per_second": 0.193,
      "step": 810
    },
    {
      "epoch": 0.32093391373169766,
      "grad_norm": 1.383279312317896,
      "learning_rate": 8.586446708827896e-05,
      "loss": 0.1709,
      "step": 811
    },
    {
      "epoch": 0.32132963988919666,
      "grad_norm": 0.8403605155927515,
      "learning_rate": 8.581630217041963e-05,
      "loss": 0.1964,
      "step": 812
    },
    {
      "epoch": 0.32172536604669566,
      "grad_norm": 1.0912036401913052,
      "learning_rate": 8.576806889296216e-05,
      "loss": 0.1711,
      "step": 813
    },
    {
      "epoch": 0.3221210922041947,
      "grad_norm": 1.7141470559355845,
      "learning_rate": 8.57197673479654e-05,
      "loss": 0.2641,
      "step": 814
    },
    {
      "epoch": 0.3225168183616937,
      "grad_norm": 1.0645503918252057,
      "learning_rate": 8.56713976276185e-05,
      "loss": 0.2157,
      "step": 815
    },
    {
      "epoch": 0.3225168183616937,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9035907525823906,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.2689778506755829,
      "eval_runtime": 5.0017,
      "eval_samples_per_second": 5.998,
      "eval_steps_per_second": 0.2,
      "step": 815
    },
    {
      "epoch": 0.3229125445191927,
      "grad_norm": 0.7009662996692828,
      "learning_rate": 8.562295982424069e-05,
      "loss": 0.14,
      "step": 816
    },
    {
      "epoch": 0.3233082706766917,
      "grad_norm": 1.241855373078175,
      "learning_rate": 8.557445403028122e-05,
      "loss": 0.1907,
      "step": 817
    },
    {
      "epoch": 0.32370399683419077,
      "grad_norm": 0.6956449753509186,
      "learning_rate": 8.552588033831905e-05,
      "loss": 0.1316,
      "step": 818
    },
    {
      "epoch": 0.32409972299168976,
      "grad_norm": 1.2506381497847148,
      "learning_rate": 8.547723884106274e-05,
      "loss": 0.2073,
      "step": 819
    },
    {
      "epoch": 0.32449544914918876,
      "grad_norm": 0.7410411089502383,
      "learning_rate": 8.542852963135029e-05,
      "loss": 0.1695,
      "step": 820
    },
    {
      "epoch": 0.32449544914918876,
      "eval_PRM Accuracy": 0.8888888888888888,
      "eval_PRM F1": 0.9375,
      "eval_PRM F1 AUC": 0.6748647319232661,
      "eval_PRM F1 AUC (fixed)": 0.9043285784554845,
      "eval_PRM F1 Neg": 0.5,
      "eval_PRM NPV": 0.7777777777777778,
      "eval_PRM Precision": 0.8974358974358975,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.3684210526315789,
      "eval_loss": 0.29212239384651184,
      "eval_runtime": 5.1898,
      "eval_samples_per_second": 5.781,
      "eval_steps_per_second": 0.193,
      "step": 820
    },
    {
      "epoch": 0.32489117530668776,
      "grad_norm": 1.0926492341151244,
      "learning_rate": 8.537975280214889e-05,
      "loss": 0.1957,
      "step": 821
    },
    {
      "epoch": 0.32528690146418676,
      "grad_norm": 1.135039526246786,
      "learning_rate": 8.533090844655482e-05,
      "loss": 0.1414,
      "step": 822
    },
    {
      "epoch": 0.3256826276216858,
      "grad_norm": 1.3071292433628887,
      "learning_rate": 8.528199665779328e-05,
      "loss": 0.1876,
      "step": 823
    },
    {
      "epoch": 0.3260783537791848,
      "grad_norm": 0.8116672832951006,
      "learning_rate": 8.523301752921811e-05,
      "loss": 0.1551,
      "step": 824
    },
    {
      "epoch": 0.3264740799366838,
      "grad_norm": 1.027783869631157,
      "learning_rate": 8.518397115431169e-05,
      "loss": 0.2153,
      "step": 825
    },
    {
      "epoch": 0.3264740799366838,
      "eval_PRM Accuracy": 0.8888888888888888,
      "eval_PRM F1": 0.9369369369369369,
      "eval_PRM F1 AUC": 0.6965076242006887,
      "eval_PRM F1 AUC (fixed)": 0.9001475651746188,
      "eval_PRM F1 Neg": 0.5333333333333333,
      "eval_PRM NPV": 0.7272727272727273,
      "eval_PRM Precision": 0.9043478260869565,
      "eval_PRM Recall": 0.9719626168224299,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.2613281309604645,
      "eval_runtime": 5.3164,
      "eval_samples_per_second": 5.643,
      "eval_steps_per_second": 0.188,
      "step": 825
    },
    {
      "epoch": 0.3268698060941828,
      "grad_norm": 0.8780206436295659,
      "learning_rate": 8.513485762668476e-05,
      "loss": 0.2199,
      "step": 826
    },
    {
      "epoch": 0.3272655322516818,
      "grad_norm": 0.9425617394206817,
      "learning_rate": 8.508567704007627e-05,
      "loss": 0.145,
      "step": 827
    },
    {
      "epoch": 0.32766125840918087,
      "grad_norm": 1.2828138498832955,
      "learning_rate": 8.503642948835305e-05,
      "loss": 0.2381,
      "step": 828
    },
    {
      "epoch": 0.32805698456667987,
      "grad_norm": 0.8888925013939754,
      "learning_rate": 8.498711506550983e-05,
      "loss": 0.1869,
      "step": 829
    },
    {
      "epoch": 0.32845271072417886,
      "grad_norm": 1.0940759254085846,
      "learning_rate": 8.493773386566899e-05,
      "loss": 0.2316,
      "step": 830
    },
    {
      "epoch": 0.32845271072417886,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.897196261682243,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.2684895694255829,
      "eval_runtime": 5.1585,
      "eval_samples_per_second": 5.816,
      "eval_steps_per_second": 0.194,
      "step": 830
    },
    {
      "epoch": 0.32884843688167786,
      "grad_norm": 0.7567948350209694,
      "learning_rate": 8.488828598308028e-05,
      "loss": 0.1438,
      "step": 831
    },
    {
      "epoch": 0.32924416303917686,
      "grad_norm": 0.7678359729015216,
      "learning_rate": 8.483877151212077e-05,
      "loss": 0.1517,
      "step": 832
    },
    {
      "epoch": 0.3296398891966759,
      "grad_norm": 0.9923361241209745,
      "learning_rate": 8.47891905472946e-05,
      "loss": 0.1844,
      "step": 833
    },
    {
      "epoch": 0.3300356153541749,
      "grad_norm": 0.7085662804763562,
      "learning_rate": 8.473954318323287e-05,
      "loss": 0.1897,
      "step": 834
    },
    {
      "epoch": 0.3304313415116739,
      "grad_norm": 0.7147096103898686,
      "learning_rate": 8.468982951469333e-05,
      "loss": 0.1607,
      "step": 835
    },
    {
      "epoch": 0.3304313415116739,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9422222222222222,
      "eval_PRM F1 AUC": 0.6795376291195278,
      "eval_PRM F1 AUC (fixed)": 0.8944909001475652,
      "eval_PRM F1 Neg": 0.5185185185185185,
      "eval_PRM NPV": 0.875,
      "eval_PRM Precision": 0.8983050847457628,
      "eval_PRM Recall": 0.9906542056074766,
      "eval_PRM Specificty": 0.3684210526315789,
      "eval_loss": 0.2911783754825592,
      "eval_runtime": 5.1004,
      "eval_samples_per_second": 5.882,
      "eval_steps_per_second": 0.196,
      "step": 835
    },
    {
      "epoch": 0.3308270676691729,
      "grad_norm": 0.6596445924702422,
      "learning_rate": 8.464004963656037e-05,
      "loss": 0.1603,
      "step": 836
    },
    {
      "epoch": 0.33122279382667197,
      "grad_norm": 0.5971853090797905,
      "learning_rate": 8.45902036438447e-05,
      "loss": 0.1523,
      "step": 837
    },
    {
      "epoch": 0.33161851998417097,
      "grad_norm": 0.6966287567958265,
      "learning_rate": 8.454029163168317e-05,
      "loss": 0.1741,
      "step": 838
    },
    {
      "epoch": 0.33201424614166997,
      "grad_norm": 0.6851643180686253,
      "learning_rate": 8.449031369533876e-05,
      "loss": 0.1682,
      "step": 839
    },
    {
      "epoch": 0.33240997229916897,
      "grad_norm": 0.7036682208218882,
      "learning_rate": 8.444026993020017e-05,
      "loss": 0.2282,
      "step": 840
    },
    {
      "epoch": 0.33240997229916897,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.8947368421052633,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.28925782442092896,
      "eval_runtime": 5.0929,
      "eval_samples_per_second": 5.891,
      "eval_steps_per_second": 0.196,
      "step": 840
    },
    {
      "epoch": 0.33280569845666796,
      "grad_norm": 0.6675456484859625,
      "learning_rate": 8.439016043178176e-05,
      "loss": 0.1559,
      "step": 841
    },
    {
      "epoch": 0.333201424614167,
      "grad_norm": 0.7725292485843828,
      "learning_rate": 8.433998529572338e-05,
      "loss": 0.1756,
      "step": 842
    },
    {
      "epoch": 0.333597150771666,
      "grad_norm": 0.5651744758485415,
      "learning_rate": 8.428974461779014e-05,
      "loss": 0.1308,
      "step": 843
    },
    {
      "epoch": 0.333992876929165,
      "grad_norm": 0.8938081414973864,
      "learning_rate": 8.423943849387223e-05,
      "loss": 0.1983,
      "step": 844
    },
    {
      "epoch": 0.334388603086664,
      "grad_norm": 1.083079267779291,
      "learning_rate": 8.418906701998477e-05,
      "loss": 0.2044,
      "step": 845
    },
    {
      "epoch": 0.334388603086664,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9038366945400886,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.27307942509651184,
      "eval_runtime": 5.0396,
      "eval_samples_per_second": 5.953,
      "eval_steps_per_second": 0.198,
      "step": 845
    },
    {
      "epoch": 0.334784329244163,
      "grad_norm": 1.237028516692739,
      "learning_rate": 8.413863029226762e-05,
      "loss": 0.2041,
      "step": 846
    },
    {
      "epoch": 0.33518005540166207,
      "grad_norm": 0.7362456243045792,
      "learning_rate": 8.408812840698517e-05,
      "loss": 0.1298,
      "step": 847
    },
    {
      "epoch": 0.33557578155916107,
      "grad_norm": 0.8836264510234058,
      "learning_rate": 8.403756146052617e-05,
      "loss": 0.2198,
      "step": 848
    },
    {
      "epoch": 0.33597150771666007,
      "grad_norm": 0.5770378079109552,
      "learning_rate": 8.398692954940352e-05,
      "loss": 0.1073,
      "step": 849
    },
    {
      "epoch": 0.33636723387415907,
      "grad_norm": 1.4507493057396252,
      "learning_rate": 8.393623277025415e-05,
      "loss": 0.2093,
      "step": 850
    },
    {
      "epoch": 0.33636723387415907,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9011313330054107,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.27347004413604736,
      "eval_runtime": 5.0973,
      "eval_samples_per_second": 5.885,
      "eval_steps_per_second": 0.196,
      "step": 850
    },
    {
      "epoch": 0.33676296003165807,
      "grad_norm": 0.8132004482782825,
      "learning_rate": 8.388547121983881e-05,
      "loss": 0.1809,
      "step": 851
    },
    {
      "epoch": 0.3371586861891571,
      "grad_norm": 0.813328882882879,
      "learning_rate": 8.383464499504183e-05,
      "loss": 0.1313,
      "step": 852
    },
    {
      "epoch": 0.3375544123466561,
      "grad_norm": 0.8381291582903194,
      "learning_rate": 8.378375419287099e-05,
      "loss": 0.1769,
      "step": 853
    },
    {
      "epoch": 0.3379501385041551,
      "grad_norm": 0.6462416856495375,
      "learning_rate": 8.373279891045735e-05,
      "loss": 0.1444,
      "step": 854
    },
    {
      "epoch": 0.3383458646616541,
      "grad_norm": 0.6221755845055602,
      "learning_rate": 8.368177924505504e-05,
      "loss": 0.1343,
      "step": 855
    },
    {
      "epoch": 0.3383458646616541,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9021151008362027,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.2757161557674408,
      "eval_runtime": 4.9916,
      "eval_samples_per_second": 6.01,
      "eval_steps_per_second": 0.2,
      "step": 855
    },
    {
      "epoch": 0.33874159081915317,
      "grad_norm": 0.7442770131734728,
      "learning_rate": 8.363069529404102e-05,
      "loss": 0.1787,
      "step": 856
    },
    {
      "epoch": 0.33913731697665217,
      "grad_norm": 0.8298459969984038,
      "learning_rate": 8.357954715491498e-05,
      "loss": 0.1956,
      "step": 857
    },
    {
      "epoch": 0.33953304313415117,
      "grad_norm": 0.8822403855159155,
      "learning_rate": 8.352833492529914e-05,
      "loss": 0.1922,
      "step": 858
    },
    {
      "epoch": 0.33992876929165017,
      "grad_norm": 0.9979725148915223,
      "learning_rate": 8.3477058702938e-05,
      "loss": 0.1866,
      "step": 859
    },
    {
      "epoch": 0.34032449544914917,
      "grad_norm": 0.6889519291640593,
      "learning_rate": 8.342571858569826e-05,
      "loss": 0.1687,
      "step": 860
    },
    {
      "epoch": 0.34032449544914917,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9008853910477127,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.2668619751930237,
      "eval_runtime": 4.8792,
      "eval_samples_per_second": 6.149,
      "eval_steps_per_second": 0.205,
      "step": 860
    },
    {
      "epoch": 0.3407202216066482,
      "grad_norm": 0.8316984538300345,
      "learning_rate": 8.337431467156851e-05,
      "loss": 0.1574,
      "step": 861
    },
    {
      "epoch": 0.3411159477641472,
      "grad_norm": 1.7234635973632797,
      "learning_rate": 8.332284705865914e-05,
      "loss": 0.2331,
      "step": 862
    },
    {
      "epoch": 0.3415116739216462,
      "grad_norm": 0.9132787253556621,
      "learning_rate": 8.327131584520207e-05,
      "loss": 0.1717,
      "step": 863
    },
    {
      "epoch": 0.3419074000791452,
      "grad_norm": 1.1689814264870746,
      "learning_rate": 8.321972112955068e-05,
      "loss": 0.2204,
      "step": 864
    },
    {
      "epoch": 0.3423031262366442,
      "grad_norm": 0.668992889648423,
      "learning_rate": 8.31680630101795e-05,
      "loss": 0.1289,
      "step": 865
    },
    {
      "epoch": 0.3423031262366442,
      "eval_PRM Accuracy": 0.8888888888888888,
      "eval_PRM F1": 0.9369369369369369,
      "eval_PRM F1 AUC": 0.6965076242006887,
      "eval_PRM F1 AUC (fixed)": 0.8991637973438269,
      "eval_PRM F1 Neg": 0.5333333333333333,
      "eval_PRM NPV": 0.7272727272727273,
      "eval_PRM Precision": 0.9043478260869565,
      "eval_PRM Recall": 0.9719626168224299,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.2648111879825592,
      "eval_runtime": 5.1813,
      "eval_samples_per_second": 5.79,
      "eval_steps_per_second": 0.193,
      "step": 865
    },
    {
      "epoch": 0.34269885239414327,
      "grad_norm": 0.9898982801677739,
      "learning_rate": 8.311634158568408e-05,
      "loss": 0.2062,
      "step": 866
    },
    {
      "epoch": 0.34309457855164227,
      "grad_norm": 0.6630768646564437,
      "learning_rate": 8.306455695478081e-05,
      "loss": 0.141,
      "step": 867
    },
    {
      "epoch": 0.34349030470914127,
      "grad_norm": 1.3165642093943424,
      "learning_rate": 8.301270921630673e-05,
      "loss": 0.2335,
      "step": 868
    },
    {
      "epoch": 0.34388603086664027,
      "grad_norm": 0.9142797355976449,
      "learning_rate": 8.296079846921927e-05,
      "loss": 0.153,
      "step": 869
    },
    {
      "epoch": 0.3442817570241393,
      "grad_norm": 0.9216845264807406,
      "learning_rate": 8.290882481259618e-05,
      "loss": 0.1797,
      "step": 870
    },
    {
      "epoch": 0.3442817570241393,
      "eval_PRM Accuracy": 0.8809523809523809,
      "eval_PRM F1": 0.9321266968325792,
      "eval_PRM F1 AUC": 0.6918347270044269,
      "eval_PRM F1 AUC (fixed)": 0.8984259714707329,
      "eval_PRM F1 Neg": 0.5161290322580645,
      "eval_PRM NPV": 0.6666666666666666,
      "eval_PRM Precision": 0.9035087719298246,
      "eval_PRM Recall": 0.9626168224299065,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.26598307490348816,
      "eval_runtime": 5.1912,
      "eval_samples_per_second": 5.779,
      "eval_steps_per_second": 0.193,
      "step": 870
    },
    {
      "epoch": 0.3446774831816383,
      "grad_norm": 0.8651769085395622,
      "learning_rate": 8.285678834563524e-05,
      "loss": 0.155,
      "step": 871
    },
    {
      "epoch": 0.3450732093391373,
      "grad_norm": 0.9518289213674653,
      "learning_rate": 8.280468916765415e-05,
      "loss": 0.1035,
      "step": 872
    },
    {
      "epoch": 0.3454689354966363,
      "grad_norm": 1.087527989751469,
      "learning_rate": 8.275252737809028e-05,
      "loss": 0.1987,
      "step": 873
    },
    {
      "epoch": 0.3458646616541353,
      "grad_norm": 0.8612811027947653,
      "learning_rate": 8.270030307650048e-05,
      "loss": 0.1976,
      "step": 874
    },
    {
      "epoch": 0.3462603878116344,
      "grad_norm": 0.8266407382637784,
      "learning_rate": 8.264801636256094e-05,
      "loss": 0.1474,
      "step": 875
    },
    {
      "epoch": 0.3462603878116344,
      "eval_PRM Accuracy": 0.8888888888888888,
      "eval_PRM F1": 0.9369369369369369,
      "eval_PRM F1 AUC": 0.6965076242006887,
      "eval_PRM F1 AUC (fixed)": 0.9038366945400885,
      "eval_PRM F1 Neg": 0.5333333333333333,
      "eval_PRM NPV": 0.7272727272727273,
      "eval_PRM Precision": 0.9043478260869565,
      "eval_PRM Recall": 0.9719626168224299,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.2710612118244171,
      "eval_runtime": 5.1806,
      "eval_samples_per_second": 5.791,
      "eval_steps_per_second": 0.193,
      "step": 875
    },
    {
      "epoch": 0.3466561139691334,
      "grad_norm": 1.2250109101178033,
      "learning_rate": 8.259566733606696e-05,
      "loss": 0.2165,
      "step": 876
    },
    {
      "epoch": 0.34705184012663237,
      "grad_norm": 0.8999170721823463,
      "learning_rate": 8.25432560969328e-05,
      "loss": 0.1934,
      "step": 877
    },
    {
      "epoch": 0.34744756628413137,
      "grad_norm": 1.2885561911684744,
      "learning_rate": 8.24907827451914e-05,
      "loss": 0.2555,
      "step": 878
    },
    {
      "epoch": 0.34784329244163037,
      "grad_norm": 1.2109685924497229,
      "learning_rate": 8.243824738099431e-05,
      "loss": 0.2004,
      "step": 879
    },
    {
      "epoch": 0.3482390185991294,
      "grad_norm": 1.298688275406524,
      "learning_rate": 8.238565010461138e-05,
      "loss": 0.2294,
      "step": 880
    },
    {
      "epoch": 0.3482390185991294,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9055582882439744,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.2764648497104645,
      "eval_runtime": 5.1409,
      "eval_samples_per_second": 5.836,
      "eval_steps_per_second": 0.195,
      "step": 880
    },
    {
      "epoch": 0.3486347447566284,
      "grad_norm": 1.1408633103873533,
      "learning_rate": 8.23329910164307e-05,
      "loss": 0.2087,
      "step": 881
    },
    {
      "epoch": 0.3490304709141274,
      "grad_norm": 1.1464720838539775,
      "learning_rate": 8.228027021695827e-05,
      "loss": 0.2062,
      "step": 882
    },
    {
      "epoch": 0.3494261970716264,
      "grad_norm": 0.7193282757881734,
      "learning_rate": 8.22274878068179e-05,
      "loss": 0.153,
      "step": 883
    },
    {
      "epoch": 0.3498219232291254,
      "grad_norm": 1.0106511876092554,
      "learning_rate": 8.2174643886751e-05,
      "loss": 0.2208,
      "step": 884
    },
    {
      "epoch": 0.3502176493866245,
      "grad_norm": 0.9517391729236356,
      "learning_rate": 8.212173855761636e-05,
      "loss": 0.2146,
      "step": 885
    },
    {
      "epoch": 0.3502176493866245,
      "eval_PRM Accuracy": 0.8888888888888888,
      "eval_PRM F1": 0.9369369369369369,
      "eval_PRM F1 AUC": 0.6965076242006887,
      "eval_PRM F1 AUC (fixed)": 0.9080177078209543,
      "eval_PRM F1 Neg": 0.5333333333333333,
      "eval_PRM NPV": 0.7272727272727273,
      "eval_PRM Precision": 0.9043478260869565,
      "eval_PRM Recall": 0.9719626168224299,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.25589191913604736,
      "eval_runtime": 4.7861,
      "eval_samples_per_second": 6.268,
      "eval_steps_per_second": 0.209,
      "step": 885
    },
    {
      "epoch": 0.3506133755441235,
      "grad_norm": 0.8364283541995906,
      "learning_rate": 8.206877192038995e-05,
      "loss": 0.1736,
      "step": 886
    },
    {
      "epoch": 0.3510091017016225,
      "grad_norm": 0.9185753485607663,
      "learning_rate": 8.201574407616483e-05,
      "loss": 0.2102,
      "step": 887
    },
    {
      "epoch": 0.35140482785912147,
      "grad_norm": 0.7018609574067978,
      "learning_rate": 8.196265512615081e-05,
      "loss": 0.1564,
      "step": 888
    },
    {
      "epoch": 0.3518005540166205,
      "grad_norm": 1.1693031030621928,
      "learning_rate": 8.190950517167437e-05,
      "loss": 0.221,
      "step": 889
    },
    {
      "epoch": 0.3521962801741195,
      "grad_norm": 0.6653103853980346,
      "learning_rate": 8.18562943141784e-05,
      "loss": 0.1346,
      "step": 890
    },
    {
      "epoch": 0.3521962801741195,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9422222222222222,
      "eval_PRM F1 AUC": 0.6795376291195278,
      "eval_PRM F1 AUC (fixed)": 0.9067879980324643,
      "eval_PRM F1 Neg": 0.5185185185185185,
      "eval_PRM NPV": 0.875,
      "eval_PRM Precision": 0.8983050847457628,
      "eval_PRM Recall": 0.9906542056074766,
      "eval_PRM Specificty": 0.3684210526315789,
      "eval_loss": 0.2787109315395355,
      "eval_runtime": 4.8288,
      "eval_samples_per_second": 6.213,
      "eval_steps_per_second": 0.207,
      "step": 890
    },
    {
      "epoch": 0.3525920063316185,
      "grad_norm": 0.7886924397386983,
      "learning_rate": 8.180302265522206e-05,
      "loss": 0.208,
      "step": 891
    },
    {
      "epoch": 0.3529877324891175,
      "grad_norm": 0.7837903777469855,
      "learning_rate": 8.174969029648052e-05,
      "loss": 0.1589,
      "step": 892
    },
    {
      "epoch": 0.3533834586466165,
      "grad_norm": 0.9005986478126117,
      "learning_rate": 8.169629733974482e-05,
      "loss": 0.1926,
      "step": 893
    },
    {
      "epoch": 0.3537791848041156,
      "grad_norm": 0.8139863060862231,
      "learning_rate": 8.164284388692166e-05,
      "loss": 0.1706,
      "step": 894
    },
    {
      "epoch": 0.3541749109616146,
      "grad_norm": 0.6170196410524107,
      "learning_rate": 8.158933004003319e-05,
      "loss": 0.1529,
      "step": 895
    },
    {
      "epoch": 0.3541749109616146,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9422222222222222,
      "eval_PRM F1 AUC": 0.6795376291195278,
      "eval_PRM F1 AUC (fixed)": 0.9070339399901624,
      "eval_PRM F1 Neg": 0.5185185185185185,
      "eval_PRM NPV": 0.875,
      "eval_PRM Precision": 0.8983050847457628,
      "eval_PRM Recall": 0.9906542056074766,
      "eval_PRM Specificty": 0.3684210526315789,
      "eval_loss": 0.28422850370407104,
      "eval_runtime": 5.3817,
      "eval_samples_per_second": 5.574,
      "eval_steps_per_second": 0.186,
      "step": 895
    },
    {
      "epoch": 0.3545706371191136,
      "grad_norm": 0.7476950704069479,
      "learning_rate": 8.153575590121686e-05,
      "loss": 0.1748,
      "step": 896
    },
    {
      "epoch": 0.3549663632766126,
      "grad_norm": 0.8524765376473925,
      "learning_rate": 8.148212157272517e-05,
      "loss": 0.1749,
      "step": 897
    },
    {
      "epoch": 0.35536208943411157,
      "grad_norm": 0.7538529262656093,
      "learning_rate": 8.142842715692548e-05,
      "loss": 0.1635,
      "step": 898
    },
    {
      "epoch": 0.3557578155916106,
      "grad_norm": 0.8730864342562548,
      "learning_rate": 8.137467275629988e-05,
      "loss": 0.1708,
      "step": 899
    },
    {
      "epoch": 0.3561535417491096,
      "grad_norm": 0.5626621536324666,
      "learning_rate": 8.132085847344493e-05,
      "loss": 0.1412,
      "step": 900
    },
    {
      "epoch": 0.3561535417491096,
      "eval_PRM Accuracy": 0.8888888888888888,
      "eval_PRM F1": 0.9380530973451328,
      "eval_PRM F1 AUC": 0.6532218396458436,
      "eval_PRM F1 AUC (fixed)": 0.9026069847515986,
      "eval_PRM F1 Neg": 0.46153846153846156,
      "eval_PRM NPV": 0.8571428571428571,
      "eval_PRM Precision": 0.8907563025210085,
      "eval_PRM Recall": 0.9906542056074766,
      "eval_PRM Specificty": 0.3157894736842105,
      "eval_loss": 0.2946940064430237,
      "eval_runtime": 4.9661,
      "eval_samples_per_second": 6.041,
      "eval_steps_per_second": 0.201,
      "step": 900
    },
    {
      "epoch": 0.3565492679066086,
      "grad_norm": 0.7534223585165879,
      "learning_rate": 8.126698441107146e-05,
      "loss": 0.178,
      "step": 901
    },
    {
      "epoch": 0.3569449940641076,
      "grad_norm": 0.8770347695832404,
      "learning_rate": 8.121305067200442e-05,
      "loss": 0.1336,
      "step": 902
    },
    {
      "epoch": 0.3573407202216066,
      "grad_norm": 0.895664182084698,
      "learning_rate": 8.115905735918268e-05,
      "loss": 0.1736,
      "step": 903
    },
    {
      "epoch": 0.3577364463791057,
      "grad_norm": 0.8759322368558397,
      "learning_rate": 8.110500457565873e-05,
      "loss": 0.1802,
      "step": 904
    },
    {
      "epoch": 0.3581321725366047,
      "grad_norm": 1.0048722558881418,
      "learning_rate": 8.105089242459866e-05,
      "loss": 0.147,
      "step": 905
    },
    {
      "epoch": 0.3581321725366047,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9422222222222222,
      "eval_PRM F1 AUC": 0.6795376291195278,
      "eval_PRM F1 AUC (fixed)": 0.8947368421052632,
      "eval_PRM F1 Neg": 0.5185185185185185,
      "eval_PRM NPV": 0.875,
      "eval_PRM Precision": 0.8983050847457628,
      "eval_PRM Recall": 0.9906542056074766,
      "eval_PRM Specificty": 0.3684210526315789,
      "eval_loss": 0.2684570252895355,
      "eval_runtime": 5.1537,
      "eval_samples_per_second": 5.821,
      "eval_steps_per_second": 0.194,
      "step": 905
    },
    {
      "epoch": 0.3585278986941037,
      "grad_norm": 0.8539026591098068,
      "learning_rate": 8.099672100928184e-05,
      "loss": 0.2033,
      "step": 906
    },
    {
      "epoch": 0.3589236248516027,
      "grad_norm": 1.3646544143578732,
      "learning_rate": 8.094249043310073e-05,
      "loss": 0.2432,
      "step": 907
    },
    {
      "epoch": 0.35931935100910173,
      "grad_norm": 0.8815465613362381,
      "learning_rate": 8.088820079956074e-05,
      "loss": 0.1462,
      "step": 908
    },
    {
      "epoch": 0.3597150771666007,
      "grad_norm": 0.8491536790059588,
      "learning_rate": 8.083385221227997e-05,
      "loss": 0.1465,
      "step": 909
    },
    {
      "epoch": 0.3601108033240997,
      "grad_norm": 0.8261044063135888,
      "learning_rate": 8.077944477498905e-05,
      "loss": 0.154,
      "step": 910
    },
    {
      "epoch": 0.3601108033240997,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.8937530742744713,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.26663410663604736,
      "eval_runtime": 5.2415,
      "eval_samples_per_second": 5.724,
      "eval_steps_per_second": 0.191,
      "step": 910
    },
    {
      "epoch": 0.3605065294815987,
      "grad_norm": 0.9567908722972123,
      "learning_rate": 8.072497859153091e-05,
      "loss": 0.1667,
      "step": 911
    },
    {
      "epoch": 0.3609022556390977,
      "grad_norm": 1.1472506022125033,
      "learning_rate": 8.067045376586066e-05,
      "loss": 0.2597,
      "step": 912
    },
    {
      "epoch": 0.3612979817965968,
      "grad_norm": 1.0171193652971011,
      "learning_rate": 8.061587040204528e-05,
      "loss": 0.1249,
      "step": 913
    },
    {
      "epoch": 0.3616937079540958,
      "grad_norm": 1.3550690920076445,
      "learning_rate": 8.056122860426352e-05,
      "loss": 0.2233,
      "step": 914
    },
    {
      "epoch": 0.3620894341115948,
      "grad_norm": 0.8560245996279463,
      "learning_rate": 8.050652847680562e-05,
      "loss": 0.181,
      "step": 915
    },
    {
      "epoch": 0.3620894341115948,
      "eval_PRM Accuracy": 0.8809523809523809,
      "eval_PRM F1": 0.933920704845815,
      "eval_PRM F1 AUC": 0.6269060501721594,
      "eval_PRM F1 AUC (fixed)": 0.8935071323167733,
      "eval_PRM F1 Neg": 0.4,
      "eval_PRM NPV": 0.8333333333333334,
      "eval_PRM Precision": 0.8833333333333333,
      "eval_PRM Recall": 0.9906542056074766,
      "eval_PRM Specificty": 0.2631578947368421,
      "eval_loss": 0.31640625,
      "eval_runtime": 5.1641,
      "eval_samples_per_second": 5.809,
      "eval_steps_per_second": 0.194,
      "step": 915
    },
    {
      "epoch": 0.3624851602690938,
      "grad_norm": 1.34936282084295,
      "learning_rate": 8.045177012407316e-05,
      "loss": 0.1816,
      "step": 916
    },
    {
      "epoch": 0.3628808864265928,
      "grad_norm": 1.0256281829092346,
      "learning_rate": 8.039695365057887e-05,
      "loss": 0.1906,
      "step": 917
    },
    {
      "epoch": 0.36327661258409183,
      "grad_norm": 1.1601798823326257,
      "learning_rate": 8.034207916094638e-05,
      "loss": 0.2272,
      "step": 918
    },
    {
      "epoch": 0.36367233874159083,
      "grad_norm": 0.8806699686211412,
      "learning_rate": 8.028714675991006e-05,
      "loss": 0.1842,
      "step": 919
    },
    {
      "epoch": 0.3640680648990898,
      "grad_norm": 1.5865857329484852,
      "learning_rate": 8.023215655231488e-05,
      "loss": 0.292,
      "step": 920
    },
    {
      "epoch": 0.3640680648990898,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.8915395966551894,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.2671875059604645,
      "eval_runtime": 4.946,
      "eval_samples_per_second": 6.065,
      "eval_steps_per_second": 0.202,
      "step": 920
    },
    {
      "epoch": 0.3644637910565888,
      "grad_norm": 0.6705389484948697,
      "learning_rate": 8.017710864311599e-05,
      "loss": 0.1569,
      "step": 921
    },
    {
      "epoch": 0.3648595172140878,
      "grad_norm": 1.0123403523097558,
      "learning_rate": 8.012200313737881e-05,
      "loss": 0.1477,
      "step": 922
    },
    {
      "epoch": 0.3652552433715869,
      "grad_norm": 1.0947664605424559,
      "learning_rate": 8.006684014027862e-05,
      "loss": 0.247,
      "step": 923
    },
    {
      "epoch": 0.3656509695290859,
      "grad_norm": 0.6980648657843986,
      "learning_rate": 8.001161975710045e-05,
      "loss": 0.128,
      "step": 924
    },
    {
      "epoch": 0.3660466956865849,
      "grad_norm": 0.8033232131722131,
      "learning_rate": 7.995634209323886e-05,
      "loss": 0.221,
      "step": 925
    },
    {
      "epoch": 0.3660466956865849,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9422222222222222,
      "eval_PRM F1 AUC": 0.6795376291195278,
      "eval_PRM F1 AUC (fixed)": 0.896458435809149,
      "eval_PRM F1 Neg": 0.5185185185185185,
      "eval_PRM NPV": 0.875,
      "eval_PRM Precision": 0.8983050847457628,
      "eval_PRM Recall": 0.9906542056074766,
      "eval_PRM Specificty": 0.3684210526315789,
      "eval_loss": 0.26611328125,
      "eval_runtime": 5.1601,
      "eval_samples_per_second": 5.814,
      "eval_steps_per_second": 0.194,
      "step": 925
    },
    {
      "epoch": 0.3664424218440839,
      "grad_norm": 0.7763032809624354,
      "learning_rate": 7.990100725419771e-05,
      "loss": 0.1454,
      "step": 926
    },
    {
      "epoch": 0.36683814800158293,
      "grad_norm": 1.1357204737601847,
      "learning_rate": 7.984561534559003e-05,
      "loss": 0.1841,
      "step": 927
    },
    {
      "epoch": 0.36723387415908193,
      "grad_norm": 0.8015236100480437,
      "learning_rate": 7.979016647313774e-05,
      "loss": 0.1554,
      "step": 928
    },
    {
      "epoch": 0.36762960031658093,
      "grad_norm": 0.8381641470220285,
      "learning_rate": 7.97346607426715e-05,
      "loss": 0.1334,
      "step": 929
    },
    {
      "epoch": 0.36802532647407993,
      "grad_norm": 0.7889919595847641,
      "learning_rate": 7.96790982601305e-05,
      "loss": 0.1616,
      "step": 930
    },
    {
      "epoch": 0.36802532647407993,
      "eval_PRM Accuracy": 0.873015873015873,
      "eval_PRM F1": 0.9298245614035088,
      "eval_PRM F1 AUC": 0.6005902606984752,
      "eval_PRM F1 AUC (fixed)": 0.9050664043285784,
      "eval_PRM F1 Neg": 0.3333333333333333,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.8760330578512396,
      "eval_PRM Recall": 0.9906542056074766,
      "eval_PRM Specificty": 0.21052631578947367,
      "eval_loss": 0.32275390625,
      "eval_runtime": 5.2216,
      "eval_samples_per_second": 5.745,
      "eval_steps_per_second": 0.192,
      "step": 930
    },
    {
      "epoch": 0.3684210526315789,
      "grad_norm": 1.2137875075708977,
      "learning_rate": 7.962347913156218e-05,
      "loss": 0.1983,
      "step": 931
    },
    {
      "epoch": 0.368816778789078,
      "grad_norm": 1.0212924006651691,
      "learning_rate": 7.956780346312218e-05,
      "loss": 0.2174,
      "step": 932
    },
    {
      "epoch": 0.369212504946577,
      "grad_norm": 1.2280396662918818,
      "learning_rate": 7.951207136107401e-05,
      "loss": 0.2034,
      "step": 933
    },
    {
      "epoch": 0.369608231104076,
      "grad_norm": 1.4206643849057659,
      "learning_rate": 7.945628293178891e-05,
      "loss": 0.2465,
      "step": 934
    },
    {
      "epoch": 0.370003957261575,
      "grad_norm": 0.8171085263301563,
      "learning_rate": 7.940043828174562e-05,
      "loss": 0.1313,
      "step": 935
    },
    {
      "epoch": 0.370003957261575,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9411764705882353,
      "eval_PRM F1 AUC": 0.7228234136743729,
      "eval_PRM F1 AUC (fixed)": 0.9124446630595179,
      "eval_PRM F1 Neg": 0.5806451612903226,
      "eval_PRM NPV": 0.75,
      "eval_PRM Precision": 0.9122807017543859,
      "eval_PRM Recall": 0.9719626168224299,
      "eval_PRM Specificty": 0.47368421052631576,
      "eval_loss": 0.24602864682674408,
      "eval_runtime": 5.1448,
      "eval_samples_per_second": 5.831,
      "eval_steps_per_second": 0.194,
      "step": 935
    },
    {
      "epoch": 0.370399683419074,
      "grad_norm": 1.4021018408112313,
      "learning_rate": 7.934453751753017e-05,
      "loss": 0.1737,
      "step": 936
    },
    {
      "epoch": 0.37079540957657303,
      "grad_norm": 1.8067271031971475,
      "learning_rate": 7.928858074583569e-05,
      "loss": 0.2164,
      "step": 937
    },
    {
      "epoch": 0.37119113573407203,
      "grad_norm": 1.264366749628032,
      "learning_rate": 7.923256807346224e-05,
      "loss": 0.1756,
      "step": 938
    },
    {
      "epoch": 0.37158686189157103,
      "grad_norm": 1.2378286555676006,
      "learning_rate": 7.917649960731655e-05,
      "loss": 0.1816,
      "step": 939
    },
    {
      "epoch": 0.37198258804907003,
      "grad_norm": 0.7591404751997808,
      "learning_rate": 7.912037545441182e-05,
      "loss": 0.1421,
      "step": 940
    },
    {
      "epoch": 0.37198258804907003,
      "eval_PRM Accuracy": 0.8888888888888888,
      "eval_PRM F1": 0.9375,
      "eval_PRM F1 AUC": 0.6748647319232661,
      "eval_PRM F1 AUC (fixed)": 0.9146581406787998,
      "eval_PRM F1 Neg": 0.5,
      "eval_PRM NPV": 0.7777777777777778,
      "eval_PRM Precision": 0.8974358974358975,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.3684210526315789,
      "eval_loss": 0.24915364384651184,
      "eval_runtime": 5.0582,
      "eval_samples_per_second": 5.931,
      "eval_steps_per_second": 0.198,
      "step": 940
    },
    {
      "epoch": 0.372378314206569,
      "grad_norm": 0.9935140380300687,
      "learning_rate": 7.906419572186756e-05,
      "loss": 0.1339,
      "step": 941
    },
    {
      "epoch": 0.3727740403640681,
      "grad_norm": 1.1058985799495225,
      "learning_rate": 7.900796051690937e-05,
      "loss": 0.2501,
      "step": 942
    },
    {
      "epoch": 0.3731697665215671,
      "grad_norm": 0.7937010098446688,
      "learning_rate": 7.895166994686869e-05,
      "loss": 0.1948,
      "step": 943
    },
    {
      "epoch": 0.3735654926790661,
      "grad_norm": 1.0455700003601043,
      "learning_rate": 7.889532411918267e-05,
      "loss": 0.1841,
      "step": 944
    },
    {
      "epoch": 0.3739612188365651,
      "grad_norm": 0.9725602260528471,
      "learning_rate": 7.88389231413939e-05,
      "loss": 0.1684,
      "step": 945
    },
    {
      "epoch": 0.3739612188365651,
      "eval_PRM Accuracy": 0.8809523809523809,
      "eval_PRM F1": 0.933920704845815,
      "eval_PRM F1 AUC": 0.6269060501721594,
      "eval_PRM F1 AUC (fixed)": 0.9158878504672897,
      "eval_PRM F1 Neg": 0.4,
      "eval_PRM NPV": 0.8333333333333334,
      "eval_PRM Precision": 0.8833333333333333,
      "eval_PRM Recall": 0.9906542056074766,
      "eval_PRM Specificty": 0.2631578947368421,
      "eval_loss": 0.2781575620174408,
      "eval_runtime": 5.0131,
      "eval_samples_per_second": 5.984,
      "eval_steps_per_second": 0.199,
      "step": 945
    },
    {
      "epoch": 0.37435694499406413,
      "grad_norm": 1.078264910314406,
      "learning_rate": 7.878246712115022e-05,
      "loss": 0.2273,
      "step": 946
    },
    {
      "epoch": 0.37475267115156313,
      "grad_norm": 1.0811931291171262,
      "learning_rate": 7.872595616620458e-05,
      "loss": 0.2198,
      "step": 947
    },
    {
      "epoch": 0.37514839730906213,
      "grad_norm": 1.1226757702102015,
      "learning_rate": 7.86693903844147e-05,
      "loss": 0.2247,
      "step": 948
    },
    {
      "epoch": 0.37554412346656113,
      "grad_norm": 1.0826415708689405,
      "learning_rate": 7.861276988374302e-05,
      "loss": 0.1979,
      "step": 949
    },
    {
      "epoch": 0.37593984962406013,
      "grad_norm": 1.1862370372748008,
      "learning_rate": 7.855609477225635e-05,
      "loss": 0.2086,
      "step": 950
    },
    {
      "epoch": 0.37593984962406013,
      "eval_PRM Accuracy": 0.8888888888888888,
      "eval_PRM F1": 0.9375,
      "eval_PRM F1 AUC": 0.6748647319232661,
      "eval_PRM F1 AUC (fixed)": 0.9173635022134777,
      "eval_PRM F1 Neg": 0.5,
      "eval_PRM NPV": 0.7777777777777778,
      "eval_PRM Precision": 0.8974358974358975,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.3684210526315789,
      "eval_loss": 0.2525065243244171,
      "eval_runtime": 4.9205,
      "eval_samples_per_second": 6.097,
      "eval_steps_per_second": 0.203,
      "step": 950
    },
    {
      "epoch": 0.3763355757815592,
      "grad_norm": 0.8486565289453545,
      "learning_rate": 7.849936515812578e-05,
      "loss": 0.2056,
      "step": 951
    },
    {
      "epoch": 0.3767313019390582,
      "grad_norm": 0.7210638719103278,
      "learning_rate": 7.844258114962642e-05,
      "loss": 0.1255,
      "step": 952
    },
    {
      "epoch": 0.3771270280965572,
      "grad_norm": 0.6569444304353967,
      "learning_rate": 7.838574285513716e-05,
      "loss": 0.1598,
      "step": 953
    },
    {
      "epoch": 0.3775227542540562,
      "grad_norm": 0.9795940508799222,
      "learning_rate": 7.832885038314053e-05,
      "loss": 0.117,
      "step": 954
    },
    {
      "epoch": 0.3779184804115552,
      "grad_norm": 0.9137163087822452,
      "learning_rate": 7.82719038422225e-05,
      "loss": 0.1754,
      "step": 955
    },
    {
      "epoch": 0.3779184804115552,
      "eval_PRM Accuracy": 0.8888888888888888,
      "eval_PRM F1": 0.9375,
      "eval_PRM F1 AUC": 0.6748647319232661,
      "eval_PRM F1 AUC (fixed)": 0.9168716182980816,
      "eval_PRM F1 Neg": 0.5,
      "eval_PRM NPV": 0.7777777777777778,
      "eval_PRM Precision": 0.8974358974358975,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.3684210526315789,
      "eval_loss": 0.2566080689430237,
      "eval_runtime": 4.9491,
      "eval_samples_per_second": 6.062,
      "eval_steps_per_second": 0.202,
      "step": 955
    },
    {
      "epoch": 0.37831420656905423,
      "grad_norm": 1.3946163286152813,
      "learning_rate": 7.821490334107216e-05,
      "loss": 0.1732,
      "step": 956
    },
    {
      "epoch": 0.37870993272655323,
      "grad_norm": 0.8957317042914218,
      "learning_rate": 7.815784898848163e-05,
      "loss": 0.2181,
      "step": 957
    },
    {
      "epoch": 0.37910565888405223,
      "grad_norm": 1.2690009363253354,
      "learning_rate": 7.810074089334581e-05,
      "loss": 0.2025,
      "step": 958
    },
    {
      "epoch": 0.37950138504155123,
      "grad_norm": 0.5657547196893011,
      "learning_rate": 7.804357916466216e-05,
      "loss": 0.0974,
      "step": 959
    },
    {
      "epoch": 0.37989711119905023,
      "grad_norm": 0.9762911393922666,
      "learning_rate": 7.798636391153056e-05,
      "loss": 0.2395,
      "step": 960
    },
    {
      "epoch": 0.37989711119905023,
      "eval_PRM Accuracy": 0.8888888888888888,
      "eval_PRM F1": 0.9380530973451328,
      "eval_PRM F1 AUC": 0.6532218396458436,
      "eval_PRM F1 AUC (fixed)": 0.9109690113133301,
      "eval_PRM F1 Neg": 0.46153846153846156,
      "eval_PRM NPV": 0.8571428571428571,
      "eval_PRM Precision": 0.8907563025210085,
      "eval_PRM Recall": 0.9906542056074766,
      "eval_PRM Specificty": 0.3157894736842105,
      "eval_loss": 0.2858072817325592,
      "eval_runtime": 5.1336,
      "eval_samples_per_second": 5.844,
      "eval_steps_per_second": 0.195,
      "step": 960
    },
    {
      "epoch": 0.3802928373565493,
      "grad_norm": 1.047448753651798,
      "learning_rate": 7.792909524315298e-05,
      "loss": 0.22,
      "step": 961
    },
    {
      "epoch": 0.3806885635140483,
      "grad_norm": 1.1084392876329912,
      "learning_rate": 7.787177326883336e-05,
      "loss": 0.1425,
      "step": 962
    },
    {
      "epoch": 0.3810842896715473,
      "grad_norm": 1.350540964010863,
      "learning_rate": 7.781439809797743e-05,
      "loss": 0.2144,
      "step": 963
    },
    {
      "epoch": 0.3814800158290463,
      "grad_norm": 1.0413842349607318,
      "learning_rate": 7.775696984009236e-05,
      "loss": 0.1588,
      "step": 964
    },
    {
      "epoch": 0.38187574198654534,
      "grad_norm": 0.956582675892353,
      "learning_rate": 7.769948860478669e-05,
      "loss": 0.1631,
      "step": 965
    },
    {
      "epoch": 0.38187574198654534,
      "eval_PRM Accuracy": 0.9047619047619048,
      "eval_PRM F1": 0.9464285714285714,
      "eval_PRM F1 AUC": 0.7058534185932119,
      "eval_PRM F1 AUC (fixed)": 0.911706837186424,
      "eval_PRM F1 Neg": 0.5714285714285714,
      "eval_PRM NPV": 0.8888888888888888,
      "eval_PRM Precision": 0.905982905982906,
      "eval_PRM Recall": 0.9906542056074766,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.2758137881755829,
      "eval_runtime": 4.7485,
      "eval_samples_per_second": 6.318,
      "eval_steps_per_second": 0.211,
      "step": 965
    },
    {
      "epoch": 0.38227146814404434,
      "grad_norm": 0.8330337149079056,
      "learning_rate": 7.764195450177011e-05,
      "loss": 0.1409,
      "step": 966
    },
    {
      "epoch": 0.38266719430154333,
      "grad_norm": 0.9879636926532511,
      "learning_rate": 7.758436764085315e-05,
      "loss": 0.187,
      "step": 967
    },
    {
      "epoch": 0.38306292045904233,
      "grad_norm": 0.8126722501498516,
      "learning_rate": 7.752672813194707e-05,
      "loss": 0.1648,
      "step": 968
    },
    {
      "epoch": 0.38345864661654133,
      "grad_norm": 0.9812482162780913,
      "learning_rate": 7.746903608506362e-05,
      "loss": 0.1817,
      "step": 969
    },
    {
      "epoch": 0.3838543727740404,
      "grad_norm": 0.9625135680327048,
      "learning_rate": 7.74112916103148e-05,
      "loss": 0.1587,
      "step": 970
    },
    {
      "epoch": 0.3838543727740404,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9411764705882353,
      "eval_PRM F1 AUC": 0.7228234136743729,
      "eval_PRM F1 AUC (fixed)": 0.9163797343826856,
      "eval_PRM F1 Neg": 0.5806451612903226,
      "eval_PRM NPV": 0.75,
      "eval_PRM Precision": 0.9122807017543859,
      "eval_PRM Recall": 0.9719626168224299,
      "eval_PRM Specificty": 0.47368421052631576,
      "eval_loss": 0.24947915971279144,
      "eval_runtime": 5.0624,
      "eval_samples_per_second": 5.926,
      "eval_steps_per_second": 0.198,
      "step": 970
    },
    {
      "epoch": 0.3842500989315394,
      "grad_norm": 0.7805689586454879,
      "learning_rate": 7.73534948179127e-05,
      "loss": 0.1242,
      "step": 971
    },
    {
      "epoch": 0.3846458250890384,
      "grad_norm": 1.1125621100367127,
      "learning_rate": 7.729564581816923e-05,
      "loss": 0.1905,
      "step": 972
    },
    {
      "epoch": 0.3850415512465374,
      "grad_norm": 1.0460512273276006,
      "learning_rate": 7.723774472149601e-05,
      "loss": 0.1768,
      "step": 973
    },
    {
      "epoch": 0.3854372774040364,
      "grad_norm": 0.9058358879564835,
      "learning_rate": 7.717979163840401e-05,
      "loss": 0.1469,
      "step": 974
    },
    {
      "epoch": 0.38583300356153544,
      "grad_norm": 1.077238475230232,
      "learning_rate": 7.712178667950346e-05,
      "loss": 0.1348,
      "step": 975
    },
    {
      "epoch": 0.38583300356153544,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9181013280865715,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.25013020634651184,
      "eval_runtime": 4.8996,
      "eval_samples_per_second": 6.123,
      "eval_steps_per_second": 0.204,
      "step": 975
    },
    {
      "epoch": 0.38622872971903444,
      "grad_norm": 1.0799868694732282,
      "learning_rate": 7.706372995550361e-05,
      "loss": 0.1533,
      "step": 976
    },
    {
      "epoch": 0.38662445587653343,
      "grad_norm": 1.2633713733016638,
      "learning_rate": 7.700562157721254e-05,
      "loss": 0.2576,
      "step": 977
    },
    {
      "epoch": 0.38702018203403243,
      "grad_norm": 1.069770180047322,
      "learning_rate": 7.694746165553682e-05,
      "loss": 0.1861,
      "step": 978
    },
    {
      "epoch": 0.38741590819153143,
      "grad_norm": 0.8497199090648525,
      "learning_rate": 7.688925030148149e-05,
      "loss": 0.1686,
      "step": 979
    },
    {
      "epoch": 0.3878116343490305,
      "grad_norm": 1.4547279595576492,
      "learning_rate": 7.683098762614971e-05,
      "loss": 0.2087,
      "step": 980
    },
    {
      "epoch": 0.3878116343490305,
      "eval_PRM Accuracy": 0.9047619047619048,
      "eval_PRM F1": 0.9464285714285714,
      "eval_PRM F1 AUC": 0.7058534185932119,
      "eval_PRM F1 AUC (fixed)": 0.9153959665518938,
      "eval_PRM F1 Neg": 0.5714285714285714,
      "eval_PRM NPV": 0.8888888888888888,
      "eval_PRM Precision": 0.905982905982906,
      "eval_PRM Recall": 0.9906542056074766,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.2621419131755829,
      "eval_runtime": 5.0453,
      "eval_samples_per_second": 5.946,
      "eval_steps_per_second": 0.198,
      "step": 980
    },
    {
      "epoch": 0.3882073605065295,
      "grad_norm": 1.059626672976495,
      "learning_rate": 7.677267374074262e-05,
      "loss": 0.1883,
      "step": 981
    },
    {
      "epoch": 0.3886030866640285,
      "grad_norm": 0.9351305620279964,
      "learning_rate": 7.671430875655907e-05,
      "loss": 0.1755,
      "step": 982
    },
    {
      "epoch": 0.3889988128215275,
      "grad_norm": 0.4610310550817402,
      "learning_rate": 7.665589278499547e-05,
      "loss": 0.0865,
      "step": 983
    },
    {
      "epoch": 0.38939453897902654,
      "grad_norm": 1.5068778942952485,
      "learning_rate": 7.659742593754551e-05,
      "loss": 0.3118,
      "step": 984
    },
    {
      "epoch": 0.38979026513652554,
      "grad_norm": 1.2931522160932827,
      "learning_rate": 7.653890832580002e-05,
      "loss": 0.1822,
      "step": 985
    },
    {
      "epoch": 0.38979026513652554,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9082636497786523,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.24667969346046448,
      "eval_runtime": 5.548,
      "eval_samples_per_second": 5.407,
      "eval_steps_per_second": 0.18,
      "step": 985
    },
    {
      "epoch": 0.39018599129402454,
      "grad_norm": 0.6926202905142572,
      "learning_rate": 7.648034006144667e-05,
      "loss": 0.1558,
      "step": 986
    },
    {
      "epoch": 0.39058171745152354,
      "grad_norm": 0.6636695070528967,
      "learning_rate": 7.642172125626986e-05,
      "loss": 0.1385,
      "step": 987
    },
    {
      "epoch": 0.39097744360902253,
      "grad_norm": 0.8028049784237185,
      "learning_rate": 7.636305202215041e-05,
      "loss": 0.1265,
      "step": 988
    },
    {
      "epoch": 0.3913731697665216,
      "grad_norm": 0.7024509893389485,
      "learning_rate": 7.630433247106543e-05,
      "loss": 0.1471,
      "step": 989
    },
    {
      "epoch": 0.3917688959240206,
      "grad_norm": 0.8983355345926901,
      "learning_rate": 7.624556271508805e-05,
      "loss": 0.2198,
      "step": 990
    },
    {
      "epoch": 0.3917688959240206,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9109690113133301,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.244140625,
      "eval_runtime": 5.052,
      "eval_samples_per_second": 5.938,
      "eval_steps_per_second": 0.198,
      "step": 990
    },
    {
      "epoch": 0.3921646220815196,
      "grad_norm": 1.0022856692403896,
      "learning_rate": 7.61867428663872e-05,
      "loss": 0.2446,
      "step": 991
    },
    {
      "epoch": 0.3925603482390186,
      "grad_norm": 0.6454129825230569,
      "learning_rate": 7.612787303722744e-05,
      "loss": 0.1265,
      "step": 992
    },
    {
      "epoch": 0.3929560743965176,
      "grad_norm": 0.8218286001550995,
      "learning_rate": 7.606895333996871e-05,
      "loss": 0.1633,
      "step": 993
    },
    {
      "epoch": 0.39335180055401664,
      "grad_norm": 1.7572035348522699,
      "learning_rate": 7.600998388706615e-05,
      "loss": 0.1339,
      "step": 994
    },
    {
      "epoch": 0.39374752671151564,
      "grad_norm": 0.7835985377610045,
      "learning_rate": 7.595096479106985e-05,
      "loss": 0.2012,
      "step": 995
    },
    {
      "epoch": 0.39374752671151564,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9139203148057058,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.24492187798023224,
      "eval_runtime": 5.2828,
      "eval_samples_per_second": 5.679,
      "eval_steps_per_second": 0.189,
      "step": 995
    },
    {
      "epoch": 0.39414325286901464,
      "grad_norm": 1.0611986223039003,
      "learning_rate": 7.589189616462465e-05,
      "loss": 0.1715,
      "step": 996
    },
    {
      "epoch": 0.39453897902651364,
      "grad_norm": 1.2109979184538513,
      "learning_rate": 7.583277812046993e-05,
      "loss": 0.2168,
      "step": 997
    },
    {
      "epoch": 0.3949347051840127,
      "grad_norm": 0.9158287931741367,
      "learning_rate": 7.577361077143939e-05,
      "loss": 0.1774,
      "step": 998
    },
    {
      "epoch": 0.3953304313415117,
      "grad_norm": 0.8044180426512219,
      "learning_rate": 7.571439423046079e-05,
      "loss": 0.1577,
      "step": 999
    },
    {
      "epoch": 0.3957261574990107,
      "grad_norm": 1.0638141821573956,
      "learning_rate": 7.565512861055589e-05,
      "loss": 0.1678,
      "step": 1000
    },
    {
      "epoch": 0.3957261574990107,
      "eval_PRM Accuracy": 0.8888888888888888,
      "eval_PRM F1": 0.9375,
      "eval_PRM F1 AUC": 0.6748647319232661,
      "eval_PRM F1 AUC (fixed)": 0.9131824889326119,
      "eval_PRM F1 Neg": 0.5,
      "eval_PRM NPV": 0.7777777777777778,
      "eval_PRM Precision": 0.8974358974358975,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.3684210526315789,
      "eval_loss": 0.2607421875,
      "eval_runtime": 5.2004,
      "eval_samples_per_second": 5.769,
      "eval_steps_per_second": 0.192,
      "step": 1000
    },
    {
      "epoch": 0.3961218836565097,
      "grad_norm": 0.6114920439497844,
      "learning_rate": 7.559581402483999e-05,
      "loss": 0.1404,
      "step": 1001
    },
    {
      "epoch": 0.3965176098140087,
      "grad_norm": 0.5065952306864845,
      "learning_rate": 7.553645058652193e-05,
      "loss": 0.1233,
      "step": 1002
    },
    {
      "epoch": 0.39691333597150774,
      "grad_norm": 1.359677724184506,
      "learning_rate": 7.547703840890377e-05,
      "loss": 0.1868,
      "step": 1003
    },
    {
      "epoch": 0.39730906212900674,
      "grad_norm": 0.9047734904901806,
      "learning_rate": 7.54175776053806e-05,
      "loss": 0.1764,
      "step": 1004
    },
    {
      "epoch": 0.39770478828650574,
      "grad_norm": 0.8728615920707045,
      "learning_rate": 7.535806828944028e-05,
      "loss": 0.1307,
      "step": 1005
    },
    {
      "epoch": 0.39770478828650574,
      "eval_PRM Accuracy": 0.8888888888888888,
      "eval_PRM F1": 0.9375,
      "eval_PRM F1 AUC": 0.6748647319232661,
      "eval_PRM F1 AUC (fixed)": 0.9151500245941957,
      "eval_PRM F1 Neg": 0.5,
      "eval_PRM NPV": 0.7777777777777778,
      "eval_PRM Precision": 0.8974358974358975,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.3684210526315789,
      "eval_loss": 0.2756510376930237,
      "eval_runtime": 5.172,
      "eval_samples_per_second": 5.801,
      "eval_steps_per_second": 0.193,
      "step": 1005
    },
    {
      "epoch": 0.39810051444400474,
      "grad_norm": 0.9139078570646113,
      "learning_rate": 7.529851057466336e-05,
      "loss": 0.1817,
      "step": 1006
    },
    {
      "epoch": 0.39849624060150374,
      "grad_norm": 1.1135123319023568,
      "learning_rate": 7.523890457472261e-05,
      "loss": 0.2115,
      "step": 1007
    },
    {
      "epoch": 0.3988919667590028,
      "grad_norm": 1.3559851732994588,
      "learning_rate": 7.517925040338312e-05,
      "loss": 0.2417,
      "step": 1008
    },
    {
      "epoch": 0.3992876929165018,
      "grad_norm": 0.7682760832279395,
      "learning_rate": 7.511954817450181e-05,
      "loss": 0.1244,
      "step": 1009
    },
    {
      "epoch": 0.3996834190740008,
      "grad_norm": 0.8284102581750562,
      "learning_rate": 7.505979800202739e-05,
      "loss": 0.1473,
      "step": 1010
    },
    {
      "epoch": 0.3996834190740008,
      "eval_PRM Accuracy": 0.8888888888888888,
      "eval_PRM F1": 0.9369369369369369,
      "eval_PRM F1 AUC": 0.6965076242006887,
      "eval_PRM F1 AUC (fixed)": 0.9149040826364978,
      "eval_PRM F1 Neg": 0.5333333333333333,
      "eval_PRM NPV": 0.7272727272727273,
      "eval_PRM Precision": 0.9043478260869565,
      "eval_PRM Recall": 0.9719626168224299,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.2512369751930237,
      "eval_runtime": 5.3038,
      "eval_samples_per_second": 5.656,
      "eval_steps_per_second": 0.189,
      "step": 1010
    },
    {
      "epoch": 0.4000791452314998,
      "grad_norm": 0.9975694048937381,
      "learning_rate": 7.500000000000001e-05,
      "loss": 0.2158,
      "step": 1011
    },
    {
      "epoch": 0.4004748713889988,
      "grad_norm": 0.8668827103037339,
      "learning_rate": 7.494015428255116e-05,
      "loss": 0.1822,
      "step": 1012
    },
    {
      "epoch": 0.40087059754649784,
      "grad_norm": 0.6263729414635655,
      "learning_rate": 7.488026096390339e-05,
      "loss": 0.1099,
      "step": 1013
    },
    {
      "epoch": 0.40126632370399684,
      "grad_norm": 0.7504103196205205,
      "learning_rate": 7.482032015837009e-05,
      "loss": 0.151,
      "step": 1014
    },
    {
      "epoch": 0.40166204986149584,
      "grad_norm": 0.6650329577029296,
      "learning_rate": 7.476033198035531e-05,
      "loss": 0.1496,
      "step": 1015
    },
    {
      "epoch": 0.40166204986149584,
      "eval_PRM Accuracy": 0.8888888888888888,
      "eval_PRM F1": 0.9369369369369369,
      "eval_PRM F1 AUC": 0.6965076242006887,
      "eval_PRM F1 AUC (fixed)": 0.911706837186424,
      "eval_PRM F1 Neg": 0.5333333333333333,
      "eval_PRM NPV": 0.7272727272727273,
      "eval_PRM Precision": 0.9043478260869565,
      "eval_PRM Recall": 0.9719626168224299,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.2530273497104645,
      "eval_runtime": 4.872,
      "eval_samples_per_second": 6.158,
      "eval_steps_per_second": 0.205,
      "step": 1015
    },
    {
      "epoch": 0.40205777601899484,
      "grad_norm": 0.7359735527288536,
      "learning_rate": 7.470029654435349e-05,
      "loss": 0.1315,
      "step": 1016
    },
    {
      "epoch": 0.4024535021764939,
      "grad_norm": 0.8276551257497571,
      "learning_rate": 7.464021396494925e-05,
      "loss": 0.1375,
      "step": 1017
    },
    {
      "epoch": 0.4028492283339929,
      "grad_norm": 1.214910283526394,
      "learning_rate": 7.458008435681725e-05,
      "loss": 0.2021,
      "step": 1018
    },
    {
      "epoch": 0.4032449544914919,
      "grad_norm": 0.6157842160231813,
      "learning_rate": 7.451990783472186e-05,
      "loss": 0.1607,
      "step": 1019
    },
    {
      "epoch": 0.4036406806489909,
      "grad_norm": 0.8111756497331858,
      "learning_rate": 7.4459684513517e-05,
      "loss": 0.1807,
      "step": 1020
    },
    {
      "epoch": 0.4036406806489909,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9109690113133301,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.2736165225505829,
      "eval_runtime": 5.1249,
      "eval_samples_per_second": 5.854,
      "eval_steps_per_second": 0.195,
      "step": 1020
    },
    {
      "epoch": 0.4040364068064899,
      "grad_norm": 0.8675431607811697,
      "learning_rate": 7.439941450814591e-05,
      "loss": 0.19,
      "step": 1021
    },
    {
      "epoch": 0.40443213296398894,
      "grad_norm": 1.2709001389564516,
      "learning_rate": 7.433909793364093e-05,
      "loss": 0.2242,
      "step": 1022
    },
    {
      "epoch": 0.40482785912148794,
      "grad_norm": 1.0414325771646424,
      "learning_rate": 7.427873490512334e-05,
      "loss": 0.2457,
      "step": 1023
    },
    {
      "epoch": 0.40522358527898694,
      "grad_norm": 1.02278098111723,
      "learning_rate": 7.421832553780299e-05,
      "loss": 0.1733,
      "step": 1024
    },
    {
      "epoch": 0.40561931143648594,
      "grad_norm": 0.8813327683696582,
      "learning_rate": 7.415786994697823e-05,
      "loss": 0.1713,
      "step": 1025
    },
    {
      "epoch": 0.40561931143648594,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9030988686669945,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.2758951783180237,
      "eval_runtime": 5.1148,
      "eval_samples_per_second": 5.865,
      "eval_steps_per_second": 0.196,
      "step": 1025
    },
    {
      "epoch": 0.40601503759398494,
      "grad_norm": 1.026290285158073,
      "learning_rate": 7.40973682480356e-05,
      "loss": 0.1695,
      "step": 1026
    },
    {
      "epoch": 0.406410763751484,
      "grad_norm": 1.6127639759375876,
      "learning_rate": 7.40368205564497e-05,
      "loss": 0.2266,
      "step": 1027
    },
    {
      "epoch": 0.406806489908983,
      "grad_norm": 0.7328498940865347,
      "learning_rate": 7.397622698778286e-05,
      "loss": 0.1516,
      "step": 1028
    },
    {
      "epoch": 0.407202216066482,
      "grad_norm": 0.8014013512206176,
      "learning_rate": 7.391558765768496e-05,
      "loss": 0.1921,
      "step": 1029
    },
    {
      "epoch": 0.407597942223981,
      "grad_norm": 0.6235870183727873,
      "learning_rate": 7.38549026818933e-05,
      "loss": 0.153,
      "step": 1030
    },
    {
      "epoch": 0.407597942223981,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.8930152484013772,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.2869466245174408,
      "eval_runtime": 4.9913,
      "eval_samples_per_second": 6.011,
      "eval_steps_per_second": 0.2,
      "step": 1030
    },
    {
      "epoch": 0.40799366838148,
      "grad_norm": 0.6213657394884893,
      "learning_rate": 7.379417217623221e-05,
      "loss": 0.1483,
      "step": 1031
    },
    {
      "epoch": 0.40838939453897904,
      "grad_norm": 0.9885060592813676,
      "learning_rate": 7.373339625661295e-05,
      "loss": 0.1945,
      "step": 1032
    },
    {
      "epoch": 0.40878512069647804,
      "grad_norm": 0.7479553490173853,
      "learning_rate": 7.367257503903349e-05,
      "loss": 0.1333,
      "step": 1033
    },
    {
      "epoch": 0.40918084685397704,
      "grad_norm": 0.6874306059404787,
      "learning_rate": 7.361170863957822e-05,
      "loss": 0.1059,
      "step": 1034
    },
    {
      "epoch": 0.40957657301147604,
      "grad_norm": 0.7609920894097508,
      "learning_rate": 7.355079717441777e-05,
      "loss": 0.1407,
      "step": 1035
    },
    {
      "epoch": 0.40957657301147604,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.897688145597639,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.2816406190395355,
      "eval_runtime": 5.0679,
      "eval_samples_per_second": 5.92,
      "eval_steps_per_second": 0.197,
      "step": 1035
    },
    {
      "epoch": 0.4099722991689751,
      "grad_norm": 0.7253695585907347,
      "learning_rate": 7.348984075980882e-05,
      "loss": 0.1325,
      "step": 1036
    },
    {
      "epoch": 0.4103680253264741,
      "grad_norm": 1.0767579293816631,
      "learning_rate": 7.342883951209377e-05,
      "loss": 0.2081,
      "step": 1037
    },
    {
      "epoch": 0.4107637514839731,
      "grad_norm": 1.0151996541820276,
      "learning_rate": 7.336779354770066e-05,
      "loss": 0.1648,
      "step": 1038
    },
    {
      "epoch": 0.4111594776414721,
      "grad_norm": 1.5622694614536354,
      "learning_rate": 7.330670298314281e-05,
      "loss": 0.2698,
      "step": 1039
    },
    {
      "epoch": 0.4115552037989711,
      "grad_norm": 0.7247536716568549,
      "learning_rate": 7.324556793501875e-05,
      "loss": 0.122,
      "step": 1040
    },
    {
      "epoch": 0.4115552037989711,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9062961141170683,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.2953124940395355,
      "eval_runtime": 5.0976,
      "eval_samples_per_second": 5.885,
      "eval_steps_per_second": 0.196,
      "step": 1040
    },
    {
      "epoch": 0.41195092995647015,
      "grad_norm": 1.5536323648564274,
      "learning_rate": 7.318438852001181e-05,
      "loss": 0.1949,
      "step": 1041
    },
    {
      "epoch": 0.41234665611396915,
      "grad_norm": 1.1662109969652108,
      "learning_rate": 7.31231648548901e-05,
      "loss": 0.1936,
      "step": 1042
    },
    {
      "epoch": 0.41274238227146814,
      "grad_norm": 0.7970019561537186,
      "learning_rate": 7.306189705650613e-05,
      "loss": 0.1297,
      "step": 1043
    },
    {
      "epoch": 0.41313810842896714,
      "grad_norm": 0.9094175309802561,
      "learning_rate": 7.300058524179662e-05,
      "loss": 0.1494,
      "step": 1044
    },
    {
      "epoch": 0.41353383458646614,
      "grad_norm": 1.0860686234491435,
      "learning_rate": 7.293922952778239e-05,
      "loss": 0.1859,
      "step": 1045
    },
    {
      "epoch": 0.41353383458646614,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9070339399901624,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.2765462100505829,
      "eval_runtime": 5.2889,
      "eval_samples_per_second": 5.672,
      "eval_steps_per_second": 0.189,
      "step": 1045
    },
    {
      "epoch": 0.4139295607439652,
      "grad_norm": 0.8547541676984697,
      "learning_rate": 7.287783003156793e-05,
      "loss": 0.1221,
      "step": 1046
    },
    {
      "epoch": 0.4143252869014642,
      "grad_norm": 1.1990645073044954,
      "learning_rate": 7.281638687034139e-05,
      "loss": 0.2334,
      "step": 1047
    },
    {
      "epoch": 0.4147210130589632,
      "grad_norm": 1.3926775085056746,
      "learning_rate": 7.275490016137421e-05,
      "loss": 0.1765,
      "step": 1048
    },
    {
      "epoch": 0.4151167392164622,
      "grad_norm": 0.5881583902547192,
      "learning_rate": 7.269337002202096e-05,
      "loss": 0.0933,
      "step": 1049
    },
    {
      "epoch": 0.4155124653739612,
      "grad_norm": 1.2824078894771334,
      "learning_rate": 7.263179656971911e-05,
      "loss": 0.2319,
      "step": 1050
    },
    {
      "epoch": 0.4155124653739612,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9163797343826857,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.2482096403837204,
      "eval_runtime": 4.8471,
      "eval_samples_per_second": 6.189,
      "eval_steps_per_second": 0.206,
      "step": 1050
    },
    {
      "epoch": 0.41590819153146025,
      "grad_norm": 1.2143427034356677,
      "learning_rate": 7.257017992198878e-05,
      "loss": 0.1842,
      "step": 1051
    },
    {
      "epoch": 0.41630391768895925,
      "grad_norm": 0.8681199136967512,
      "learning_rate": 7.250852019643256e-05,
      "loss": 0.1626,
      "step": 1052
    },
    {
      "epoch": 0.41669964384645825,
      "grad_norm": 0.776003479898443,
      "learning_rate": 7.244681751073522e-05,
      "loss": 0.1441,
      "step": 1053
    },
    {
      "epoch": 0.41709537000395724,
      "grad_norm": 0.8829476486837633,
      "learning_rate": 7.238507198266356e-05,
      "loss": 0.1614,
      "step": 1054
    },
    {
      "epoch": 0.4174910961614563,
      "grad_norm": 0.6634719308376149,
      "learning_rate": 7.232328373006613e-05,
      "loss": 0.1719,
      "step": 1055
    },
    {
      "epoch": 0.4174910961614563,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9161337924249878,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.24472656846046448,
      "eval_runtime": 5.3797,
      "eval_samples_per_second": 5.577,
      "eval_steps_per_second": 0.186,
      "step": 1055
    },
    {
      "epoch": 0.4178868223189553,
      "grad_norm": 0.7968457229963968,
      "learning_rate": 7.226145287087302e-05,
      "loss": 0.2202,
      "step": 1056
    },
    {
      "epoch": 0.4182825484764543,
      "grad_norm": 0.7454144969172052,
      "learning_rate": 7.219957952309567e-05,
      "loss": 0.1569,
      "step": 1057
    },
    {
      "epoch": 0.4186782746339533,
      "grad_norm": 0.692989164692621,
      "learning_rate": 7.213766380482658e-05,
      "loss": 0.1118,
      "step": 1058
    },
    {
      "epoch": 0.4190740007914523,
      "grad_norm": 0.9460530113629996,
      "learning_rate": 7.207570583423915e-05,
      "loss": 0.1672,
      "step": 1059
    },
    {
      "epoch": 0.41946972694895135,
      "grad_norm": 0.9249444937628459,
      "learning_rate": 7.201370572958738e-05,
      "loss": 0.1443,
      "step": 1060
    },
    {
      "epoch": 0.41946972694895135,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9104771273979341,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.2552734315395355,
      "eval_runtime": 5.1715,
      "eval_samples_per_second": 5.801,
      "eval_steps_per_second": 0.193,
      "step": 1060
    },
    {
      "epoch": 0.41986545310645035,
      "grad_norm": 0.833714549579215,
      "learning_rate": 7.195166360920574e-05,
      "loss": 0.1976,
      "step": 1061
    },
    {
      "epoch": 0.42026117926394935,
      "grad_norm": 1.0551741242274266,
      "learning_rate": 7.188957959150886e-05,
      "loss": 0.274,
      "step": 1062
    },
    {
      "epoch": 0.42065690542144835,
      "grad_norm": 0.8056137694266688,
      "learning_rate": 7.182745379499135e-05,
      "loss": 0.1746,
      "step": 1063
    },
    {
      "epoch": 0.42105263157894735,
      "grad_norm": 1.0458729944535654,
      "learning_rate": 7.176528633822752e-05,
      "loss": 0.2019,
      "step": 1064
    },
    {
      "epoch": 0.4214483577364464,
      "grad_norm": 1.043950105141731,
      "learning_rate": 7.170307733987127e-05,
      "loss": 0.1484,
      "step": 1065
    },
    {
      "epoch": 0.4214483577364464,
      "eval_PRM Accuracy": 0.8888888888888888,
      "eval_PRM F1": 0.9369369369369369,
      "eval_PRM F1 AUC": 0.6965076242006887,
      "eval_PRM F1 AUC (fixed)": 0.9053123462862765,
      "eval_PRM F1 Neg": 0.5333333333333333,
      "eval_PRM NPV": 0.7272727272727273,
      "eval_PRM Precision": 0.9043478260869565,
      "eval_PRM Recall": 0.9719626168224299,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.25533854961395264,
      "eval_runtime": 4.8562,
      "eval_samples_per_second": 6.178,
      "eval_steps_per_second": 0.206,
      "step": 1065
    },
    {
      "epoch": 0.4218440838939454,
      "grad_norm": 0.6169931967285929,
      "learning_rate": 7.164082691865573e-05,
      "loss": 0.1293,
      "step": 1066
    },
    {
      "epoch": 0.4222398100514444,
      "grad_norm": 0.8953133811208986,
      "learning_rate": 7.157853519339306e-05,
      "loss": 0.2189,
      "step": 1067
    },
    {
      "epoch": 0.4226355362089434,
      "grad_norm": 0.8655811306878989,
      "learning_rate": 7.151620228297436e-05,
      "loss": 0.159,
      "step": 1068
    },
    {
      "epoch": 0.4230312623664424,
      "grad_norm": 1.1171323535229134,
      "learning_rate": 7.145382830636924e-05,
      "loss": 0.2128,
      "step": 1069
    },
    {
      "epoch": 0.42342698852394145,
      "grad_norm": 1.3896583322187834,
      "learning_rate": 7.139141338262573e-05,
      "loss": 0.1971,
      "step": 1070
    },
    {
      "epoch": 0.42342698852394145,
      "eval_PRM Accuracy": 0.8888888888888888,
      "eval_PRM F1": 0.9369369369369369,
      "eval_PRM F1 AUC": 0.6965076242006887,
      "eval_PRM F1 AUC (fixed)": 0.8996556812592228,
      "eval_PRM F1 Neg": 0.5333333333333333,
      "eval_PRM NPV": 0.7272727272727273,
      "eval_PRM Precision": 0.9043478260869565,
      "eval_PRM Recall": 0.9719626168224299,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.2591796815395355,
      "eval_runtime": 5.0581,
      "eval_samples_per_second": 5.931,
      "eval_steps_per_second": 0.198,
      "step": 1070
    },
    {
      "epoch": 0.42382271468144045,
      "grad_norm": 0.9660879797074645,
      "learning_rate": 7.132895763087002e-05,
      "loss": 0.1822,
      "step": 1071
    },
    {
      "epoch": 0.42421844083893945,
      "grad_norm": 0.7917397034104919,
      "learning_rate": 7.126646117030619e-05,
      "loss": 0.1969,
      "step": 1072
    },
    {
      "epoch": 0.42461416699643845,
      "grad_norm": 0.6248062171228119,
      "learning_rate": 7.120392412021605e-05,
      "loss": 0.1226,
      "step": 1073
    },
    {
      "epoch": 0.4250098931539375,
      "grad_norm": 0.7682433865946915,
      "learning_rate": 7.114134659995886e-05,
      "loss": 0.134,
      "step": 1074
    },
    {
      "epoch": 0.4254056193114365,
      "grad_norm": 0.6869070512744374,
      "learning_rate": 7.107872872897113e-05,
      "loss": 0.201,
      "step": 1075
    },
    {
      "epoch": 0.4254056193114365,
      "eval_PRM Accuracy": 0.8888888888888888,
      "eval_PRM F1": 0.9369369369369369,
      "eval_PRM F1 AUC": 0.6965076242006887,
      "eval_PRM F1 AUC (fixed)": 0.896950319724545,
      "eval_PRM F1 Neg": 0.5333333333333333,
      "eval_PRM NPV": 0.7272727272727273,
      "eval_PRM Precision": 0.9043478260869565,
      "eval_PRM Recall": 0.9719626168224299,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.2595052123069763,
      "eval_runtime": 5.1649,
      "eval_samples_per_second": 5.808,
      "eval_steps_per_second": 0.194,
      "step": 1075
    },
    {
      "epoch": 0.4258013454689355,
      "grad_norm": 0.6164907856893128,
      "learning_rate": 7.101607062676638e-05,
      "loss": 0.1278,
      "step": 1076
    },
    {
      "epoch": 0.4261970716264345,
      "grad_norm": 0.8867766154804855,
      "learning_rate": 7.095337241293493e-05,
      "loss": 0.1929,
      "step": 1077
    },
    {
      "epoch": 0.4265927977839335,
      "grad_norm": 0.6227772743485064,
      "learning_rate": 7.089063420714366e-05,
      "loss": 0.1173,
      "step": 1078
    },
    {
      "epoch": 0.42698852394143255,
      "grad_norm": 1.0871358330833987,
      "learning_rate": 7.082785612913573e-05,
      "loss": 0.2433,
      "step": 1079
    },
    {
      "epoch": 0.42738425009893155,
      "grad_norm": 1.0032166621865515,
      "learning_rate": 7.076503829873048e-05,
      "loss": 0.2379,
      "step": 1080
    },
    {
      "epoch": 0.42738425009893155,
      "eval_PRM Accuracy": 0.873015873015873,
      "eval_PRM F1": 0.9272727272727272,
      "eval_PRM F1 AUC": 0.6871618298081652,
      "eval_PRM F1 AUC (fixed)": 0.8979340875553369,
      "eval_PRM F1 Neg": 0.5,
      "eval_PRM NPV": 0.6153846153846154,
      "eval_PRM Precision": 0.9026548672566371,
      "eval_PRM Recall": 0.9532710280373832,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.2586914002895355,
      "eval_runtime": 5.0751,
      "eval_samples_per_second": 5.911,
      "eval_steps_per_second": 0.197,
      "step": 1080
    },
    {
      "epoch": 0.42777997625643055,
      "grad_norm": 0.764980978700694,
      "learning_rate": 7.070218083582306e-05,
      "loss": 0.1599,
      "step": 1081
    },
    {
      "epoch": 0.42817570241392955,
      "grad_norm": 0.8101485404714679,
      "learning_rate": 7.063928386038428e-05,
      "loss": 0.1558,
      "step": 1082
    },
    {
      "epoch": 0.42857142857142855,
      "grad_norm": 0.78224904838527,
      "learning_rate": 7.057634749246037e-05,
      "loss": 0.1148,
      "step": 1083
    },
    {
      "epoch": 0.4289671547289276,
      "grad_norm": 0.6475554533796173,
      "learning_rate": 7.051337185217273e-05,
      "loss": 0.1108,
      "step": 1084
    },
    {
      "epoch": 0.4293628808864266,
      "grad_norm": 0.9462148739074969,
      "learning_rate": 7.045035705971778e-05,
      "loss": 0.2112,
      "step": 1085
    },
    {
      "epoch": 0.4293628808864266,
      "eval_PRM Accuracy": 0.8888888888888888,
      "eval_PRM F1": 0.9375,
      "eval_PRM F1 AUC": 0.6748647319232661,
      "eval_PRM F1 AUC (fixed)": 0.896704377766847,
      "eval_PRM F1 Neg": 0.5,
      "eval_PRM NPV": 0.7777777777777778,
      "eval_PRM Precision": 0.8974358974358975,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.3684210526315789,
      "eval_loss": 0.27356770634651184,
      "eval_runtime": 4.9386,
      "eval_samples_per_second": 6.075,
      "eval_steps_per_second": 0.202,
      "step": 1085
    },
    {
      "epoch": 0.4297586070439256,
      "grad_norm": 0.8754379936540506,
      "learning_rate": 7.038730323536655e-05,
      "loss": 0.2021,
      "step": 1086
    },
    {
      "epoch": 0.4301543332014246,
      "grad_norm": 1.0789109475816612,
      "learning_rate": 7.032421049946467e-05,
      "loss": 0.2107,
      "step": 1087
    },
    {
      "epoch": 0.4305500593589236,
      "grad_norm": 0.592238816929124,
      "learning_rate": 7.0261078972432e-05,
      "loss": 0.1212,
      "step": 1088
    },
    {
      "epoch": 0.43094578551642265,
      "grad_norm": 0.5522184724339104,
      "learning_rate": 7.019790877476241e-05,
      "loss": 0.1013,
      "step": 1089
    },
    {
      "epoch": 0.43134151167392165,
      "grad_norm": 1.0574270528303438,
      "learning_rate": 7.013470002702363e-05,
      "loss": 0.1532,
      "step": 1090
    },
    {
      "epoch": 0.43134151167392165,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9422222222222222,
      "eval_PRM F1 AUC": 0.6795376291195278,
      "eval_PRM F1 AUC (fixed)": 0.898180029513035,
      "eval_PRM F1 Neg": 0.5185185185185185,
      "eval_PRM NPV": 0.875,
      "eval_PRM Precision": 0.8983050847457628,
      "eval_PRM Recall": 0.9906542056074766,
      "eval_PRM Specificty": 0.3684210526315789,
      "eval_loss": 0.2925618588924408,
      "eval_runtime": 4.9335,
      "eval_samples_per_second": 6.081,
      "eval_steps_per_second": 0.203,
      "step": 1090
    },
    {
      "epoch": 0.43173723783142065,
      "grad_norm": 0.7675040313831717,
      "learning_rate": 7.007145284985694e-05,
      "loss": 0.1504,
      "step": 1091
    },
    {
      "epoch": 0.43213296398891965,
      "grad_norm": 0.7752955380783298,
      "learning_rate": 7.000816736397695e-05,
      "loss": 0.1395,
      "step": 1092
    },
    {
      "epoch": 0.4325286901464187,
      "grad_norm": 0.9757693810994306,
      "learning_rate": 6.994484369017143e-05,
      "loss": 0.1866,
      "step": 1093
    },
    {
      "epoch": 0.4329244163039177,
      "grad_norm": 1.1376595609453177,
      "learning_rate": 6.988148194930099e-05,
      "loss": 0.1804,
      "step": 1094
    },
    {
      "epoch": 0.4333201424614167,
      "grad_norm": 0.667210832374645,
      "learning_rate": 6.981808226229892e-05,
      "loss": 0.1453,
      "step": 1095
    },
    {
      "epoch": 0.4333201424614167,
      "eval_PRM Accuracy": 0.8809523809523809,
      "eval_PRM F1": 0.9321266968325792,
      "eval_PRM F1 AUC": 0.6918347270044269,
      "eval_PRM F1 AUC (fixed)": 0.9028529267092965,
      "eval_PRM F1 Neg": 0.5161290322580645,
      "eval_PRM NPV": 0.6666666666666666,
      "eval_PRM Precision": 0.9035087719298246,
      "eval_PRM Recall": 0.9626168224299065,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.2683919370174408,
      "eval_runtime": 5.032,
      "eval_samples_per_second": 5.962,
      "eval_steps_per_second": 0.199,
      "step": 1095
    },
    {
      "epoch": 0.4337158686189157,
      "grad_norm": 0.842522589665804,
      "learning_rate": 6.975464475017093e-05,
      "loss": 0.2022,
      "step": 1096
    },
    {
      "epoch": 0.4341115947764147,
      "grad_norm": 1.2057974584412778,
      "learning_rate": 6.969116953399496e-05,
      "loss": 0.2223,
      "step": 1097
    },
    {
      "epoch": 0.43450732093391375,
      "grad_norm": 1.0009068096875529,
      "learning_rate": 6.962765673492083e-05,
      "loss": 0.2061,
      "step": 1098
    },
    {
      "epoch": 0.43490304709141275,
      "grad_norm": 0.853091167842377,
      "learning_rate": 6.956410647417017e-05,
      "loss": 0.1622,
      "step": 1099
    },
    {
      "epoch": 0.43529877324891175,
      "grad_norm": 1.3196140461877635,
      "learning_rate": 6.950051887303606e-05,
      "loss": 0.2357,
      "step": 1100
    },
    {
      "epoch": 0.43529877324891175,
      "eval_PRM Accuracy": 0.8809523809523809,
      "eval_PRM F1": 0.9321266968325792,
      "eval_PRM F1 AUC": 0.6918347270044269,
      "eval_PRM F1 AUC (fixed)": 0.9050664043285784,
      "eval_PRM F1 Neg": 0.5161290322580645,
      "eval_PRM NPV": 0.6666666666666666,
      "eval_PRM Precision": 0.9035087719298246,
      "eval_PRM Recall": 0.9626168224299065,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.2555338442325592,
      "eval_runtime": 5.381,
      "eval_samples_per_second": 5.575,
      "eval_steps_per_second": 0.186,
      "step": 1100
    },
    {
      "epoch": 0.43569449940641075,
      "grad_norm": 1.0593154206903501,
      "learning_rate": 6.943689405288288e-05,
      "loss": 0.2401,
      "step": 1101
    },
    {
      "epoch": 0.43609022556390975,
      "grad_norm": 0.7128541908595966,
      "learning_rate": 6.937323213514601e-05,
      "loss": 0.1474,
      "step": 1102
    },
    {
      "epoch": 0.4364859517214088,
      "grad_norm": 1.0864970238845069,
      "learning_rate": 6.930953324133169e-05,
      "loss": 0.1621,
      "step": 1103
    },
    {
      "epoch": 0.4368816778789078,
      "grad_norm": 0.9761075426450818,
      "learning_rate": 6.924579749301671e-05,
      "loss": 0.1968,
      "step": 1104
    },
    {
      "epoch": 0.4372774040364068,
      "grad_norm": 0.6152469458364157,
      "learning_rate": 6.918202501184815e-05,
      "loss": 0.1251,
      "step": 1105
    },
    {
      "epoch": 0.4372774040364068,
      "eval_PRM Accuracy": 0.8888888888888888,
      "eval_PRM F1": 0.9380530973451328,
      "eval_PRM F1 AUC": 0.6532218396458436,
      "eval_PRM F1 AUC (fixed)": 0.909001475651746,
      "eval_PRM F1 Neg": 0.46153846153846156,
      "eval_PRM NPV": 0.8571428571428571,
      "eval_PRM Precision": 0.8907563025210085,
      "eval_PRM Recall": 0.9906542056074766,
      "eval_PRM Specificty": 0.3157894736842105,
      "eval_loss": 0.2842773497104645,
      "eval_runtime": 5.1837,
      "eval_samples_per_second": 5.787,
      "eval_steps_per_second": 0.193,
      "step": 1105
    },
    {
      "epoch": 0.4376731301939058,
      "grad_norm": 0.6476226992411395,
      "learning_rate": 6.911821591954328e-05,
      "loss": 0.156,
      "step": 1106
    },
    {
      "epoch": 0.4380688563514048,
      "grad_norm": 0.5750138955229323,
      "learning_rate": 6.90543703378892e-05,
      "loss": 0.0985,
      "step": 1107
    },
    {
      "epoch": 0.43846458250890386,
      "grad_norm": 0.5221460846576396,
      "learning_rate": 6.899048838874267e-05,
      "loss": 0.1173,
      "step": 1108
    },
    {
      "epoch": 0.43886030866640285,
      "grad_norm": 1.3222602938830674,
      "learning_rate": 6.892657019402983e-05,
      "loss": 0.2054,
      "step": 1109
    },
    {
      "epoch": 0.43925603482390185,
      "grad_norm": 0.9310232343319954,
      "learning_rate": 6.886261587574604e-05,
      "loss": 0.1819,
      "step": 1110
    },
    {
      "epoch": 0.43925603482390185,
      "eval_PRM Accuracy": 0.8888888888888888,
      "eval_PRM F1": 0.9380530973451328,
      "eval_PRM F1 AUC": 0.6532218396458436,
      "eval_PRM F1 AUC (fixed)": 0.9077717658632563,
      "eval_PRM F1 Neg": 0.46153846153846156,
      "eval_PRM NPV": 0.8571428571428571,
      "eval_PRM Precision": 0.8907563025210085,
      "eval_PRM Recall": 0.9906542056074766,
      "eval_PRM Specificty": 0.3157894736842105,
      "eval_loss": 0.3307291567325592,
      "eval_runtime": 5.0128,
      "eval_samples_per_second": 5.985,
      "eval_steps_per_second": 0.199,
      "step": 1110
    },
    {
      "epoch": 0.43965176098140085,
      "grad_norm": 0.7050040817051295,
      "learning_rate": 6.879862555595559e-05,
      "loss": 0.1287,
      "step": 1111
    },
    {
      "epoch": 0.4400474871388999,
      "grad_norm": 0.955967274178146,
      "learning_rate": 6.87345993567915e-05,
      "loss": 0.2285,
      "step": 1112
    },
    {
      "epoch": 0.4404432132963989,
      "grad_norm": 0.9711233459670032,
      "learning_rate": 6.867053740045521e-05,
      "loss": 0.2061,
      "step": 1113
    },
    {
      "epoch": 0.4408389394538979,
      "grad_norm": 1.0188610482149132,
      "learning_rate": 6.860643980921648e-05,
      "loss": 0.242,
      "step": 1114
    },
    {
      "epoch": 0.4412346656113969,
      "grad_norm": 0.862920130020995,
      "learning_rate": 6.854230670541306e-05,
      "loss": 0.1311,
      "step": 1115
    },
    {
      "epoch": 0.4412346656113969,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9080177078209541,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.2684895694255829,
      "eval_runtime": 5.0392,
      "eval_samples_per_second": 5.953,
      "eval_steps_per_second": 0.198,
      "step": 1115
    },
    {
      "epoch": 0.4416303917688959,
      "grad_norm": 0.6443102963178605,
      "learning_rate": 6.847813821145045e-05,
      "loss": 0.1215,
      "step": 1116
    },
    {
      "epoch": 0.44202611792639496,
      "grad_norm": 1.0591525804588973,
      "learning_rate": 6.841393444980177e-05,
      "loss": 0.1923,
      "step": 1117
    },
    {
      "epoch": 0.44242184408389396,
      "grad_norm": 1.1107267697369232,
      "learning_rate": 6.834969554300732e-05,
      "loss": 0.1976,
      "step": 1118
    },
    {
      "epoch": 0.44281757024139295,
      "grad_norm": 1.50588141120044,
      "learning_rate": 6.828542161367462e-05,
      "loss": 0.1642,
      "step": 1119
    },
    {
      "epoch": 0.44321329639889195,
      "grad_norm": 0.8918585278526379,
      "learning_rate": 6.822111278447796e-05,
      "loss": 0.1672,
      "step": 1120
    },
    {
      "epoch": 0.44321329639889195,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9013772749631087,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.2665039002895355,
      "eval_runtime": 5.1833,
      "eval_samples_per_second": 5.788,
      "eval_steps_per_second": 0.193,
      "step": 1120
    },
    {
      "epoch": 0.44360902255639095,
      "grad_norm": 0.8871335582525137,
      "learning_rate": 6.815676917815826e-05,
      "loss": 0.2036,
      "step": 1121
    },
    {
      "epoch": 0.44400474871389,
      "grad_norm": 0.665613125663357,
      "learning_rate": 6.80923909175228e-05,
      "loss": 0.1355,
      "step": 1122
    },
    {
      "epoch": 0.444400474871389,
      "grad_norm": 1.118527707241081,
      "learning_rate": 6.802797812544502e-05,
      "loss": 0.2284,
      "step": 1123
    },
    {
      "epoch": 0.444796201028888,
      "grad_norm": 1.0754073872799603,
      "learning_rate": 6.796353092486427e-05,
      "loss": 0.1771,
      "step": 1124
    },
    {
      "epoch": 0.445191927186387,
      "grad_norm": 0.752453229662109,
      "learning_rate": 6.789904943878554e-05,
      "loss": 0.1554,
      "step": 1125
    },
    {
      "epoch": 0.445191927186387,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9422222222222222,
      "eval_PRM F1 AUC": 0.6795376291195278,
      "eval_PRM F1 AUC (fixed)": 0.9075258239055584,
      "eval_PRM F1 Neg": 0.5185185185185185,
      "eval_PRM NPV": 0.875,
      "eval_PRM Precision": 0.8983050847457628,
      "eval_PRM Recall": 0.9906542056074766,
      "eval_PRM Specificty": 0.3684210526315789,
      "eval_loss": 0.2969563901424408,
      "eval_runtime": 5.3958,
      "eval_samples_per_second": 5.56,
      "eval_steps_per_second": 0.185,
      "step": 1125
    },
    {
      "epoch": 0.445587653343886,
      "grad_norm": 1.0292726490257276,
      "learning_rate": 6.783453379027931e-05,
      "loss": 0.1752,
      "step": 1126
    },
    {
      "epoch": 0.44598337950138506,
      "grad_norm": 0.6487250128911204,
      "learning_rate": 6.776998410248122e-05,
      "loss": 0.1306,
      "step": 1127
    },
    {
      "epoch": 0.44637910565888406,
      "grad_norm": 0.529340671671959,
      "learning_rate": 6.770540049859188e-05,
      "loss": 0.1179,
      "step": 1128
    },
    {
      "epoch": 0.44677483181638306,
      "grad_norm": 1.028411738180361,
      "learning_rate": 6.764078310187668e-05,
      "loss": 0.1545,
      "step": 1129
    },
    {
      "epoch": 0.44717055797388205,
      "grad_norm": 1.107562885732197,
      "learning_rate": 6.757613203566542e-05,
      "loss": 0.2573,
      "step": 1130
    },
    {
      "epoch": 0.44717055797388205,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9067879980324643,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.2767903506755829,
      "eval_runtime": 5.0259,
      "eval_samples_per_second": 5.969,
      "eval_steps_per_second": 0.199,
      "step": 1130
    },
    {
      "epoch": 0.4475662841313811,
      "grad_norm": 0.9861032570397578,
      "learning_rate": 6.751144742335227e-05,
      "loss": 0.1425,
      "step": 1131
    },
    {
      "epoch": 0.4479620102888801,
      "grad_norm": 0.5368364700298076,
      "learning_rate": 6.744672938839534e-05,
      "loss": 0.0897,
      "step": 1132
    },
    {
      "epoch": 0.4483577364463791,
      "grad_norm": 0.8859818818040079,
      "learning_rate": 6.738197805431657e-05,
      "loss": 0.1354,
      "step": 1133
    },
    {
      "epoch": 0.4487534626038781,
      "grad_norm": 0.8886138342391837,
      "learning_rate": 6.731719354470143e-05,
      "loss": 0.1126,
      "step": 1134
    },
    {
      "epoch": 0.4491491887613771,
      "grad_norm": 1.0256762689391408,
      "learning_rate": 6.725237598319877e-05,
      "loss": 0.2269,
      "step": 1135
    },
    {
      "epoch": 0.4491491887613771,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9094933595671423,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.27778321504592896,
      "eval_runtime": 5.122,
      "eval_samples_per_second": 5.857,
      "eval_steps_per_second": 0.195,
      "step": 1135
    },
    {
      "epoch": 0.44954491491887616,
      "grad_norm": 1.0006349927217255,
      "learning_rate": 6.718752549352045e-05,
      "loss": 0.1332,
      "step": 1136
    },
    {
      "epoch": 0.44994064107637516,
      "grad_norm": 0.7965441836026523,
      "learning_rate": 6.71226421994412e-05,
      "loss": 0.1701,
      "step": 1137
    },
    {
      "epoch": 0.45033636723387416,
      "grad_norm": 0.9907952338440558,
      "learning_rate": 6.70577262247984e-05,
      "loss": 0.1673,
      "step": 1138
    },
    {
      "epoch": 0.45073209339137316,
      "grad_norm": 0.8792094827217769,
      "learning_rate": 6.699277769349174e-05,
      "loss": 0.1658,
      "step": 1139
    },
    {
      "epoch": 0.45112781954887216,
      "grad_norm": 0.8851569659333709,
      "learning_rate": 6.69277967294831e-05,
      "loss": 0.1622,
      "step": 1140
    },
    {
      "epoch": 0.45112781954887216,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9131824889326119,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.29736328125,
      "eval_runtime": 5.1163,
      "eval_samples_per_second": 5.864,
      "eval_steps_per_second": 0.195,
      "step": 1140
    },
    {
      "epoch": 0.4515235457063712,
      "grad_norm": 0.8338521515796626,
      "learning_rate": 6.686278345679625e-05,
      "loss": 0.1387,
      "step": 1141
    },
    {
      "epoch": 0.4519192718638702,
      "grad_norm": 1.0046475507637396,
      "learning_rate": 6.679773799951662e-05,
      "loss": 0.1604,
      "step": 1142
    },
    {
      "epoch": 0.4523149980213692,
      "grad_norm": 0.8542707175038639,
      "learning_rate": 6.673266048179103e-05,
      "loss": 0.1597,
      "step": 1143
    },
    {
      "epoch": 0.4527107241788682,
      "grad_norm": 0.8358989784559276,
      "learning_rate": 6.666755102782758e-05,
      "loss": 0.1187,
      "step": 1144
    },
    {
      "epoch": 0.45310645033636726,
      "grad_norm": 1.2113148835473992,
      "learning_rate": 6.660240976189523e-05,
      "loss": 0.1893,
      "step": 1145
    },
    {
      "epoch": 0.45310645033636726,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9168716182980816,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.29277342557907104,
      "eval_runtime": 5.2377,
      "eval_samples_per_second": 5.728,
      "eval_steps_per_second": 0.191,
      "step": 1145
    },
    {
      "epoch": 0.45350217649386626,
      "grad_norm": 0.8994370250141286,
      "learning_rate": 6.653723680832371e-05,
      "loss": 0.1818,
      "step": 1146
    },
    {
      "epoch": 0.45389790265136526,
      "grad_norm": 0.8110339616095306,
      "learning_rate": 6.647203229150322e-05,
      "loss": 0.1822,
      "step": 1147
    },
    {
      "epoch": 0.45429362880886426,
      "grad_norm": 1.171659601524543,
      "learning_rate": 6.640679633588421e-05,
      "loss": 0.1953,
      "step": 1148
    },
    {
      "epoch": 0.45468935496636326,
      "grad_norm": 1.4346097612938402,
      "learning_rate": 6.63415290659771e-05,
      "loss": 0.2195,
      "step": 1149
    },
    {
      "epoch": 0.4550850811238623,
      "grad_norm": 0.5683693035673459,
      "learning_rate": 6.627623060635214e-05,
      "loss": 0.1676,
      "step": 1150
    },
    {
      "epoch": 0.4550850811238623,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9195769798327595,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.26215821504592896,
      "eval_runtime": 5.1083,
      "eval_samples_per_second": 5.873,
      "eval_steps_per_second": 0.196,
      "step": 1150
    },
    {
      "epoch": 0.4554808072813613,
      "grad_norm": 0.7216955630584146,
      "learning_rate": 6.621090108163904e-05,
      "loss": 0.1274,
      "step": 1151
    },
    {
      "epoch": 0.4558765334388603,
      "grad_norm": 0.8874018323598207,
      "learning_rate": 6.614554061652683e-05,
      "loss": 0.1755,
      "step": 1152
    },
    {
      "epoch": 0.4562722595963593,
      "grad_norm": 0.6308012982475653,
      "learning_rate": 6.608014933576362e-05,
      "loss": 0.14,
      "step": 1153
    },
    {
      "epoch": 0.4566679857538583,
      "grad_norm": 0.8043056556053905,
      "learning_rate": 6.601472736415629e-05,
      "loss": 0.206,
      "step": 1154
    },
    {
      "epoch": 0.45706371191135736,
      "grad_norm": 0.7532543042969924,
      "learning_rate": 6.594927482657033e-05,
      "loss": 0.1986,
      "step": 1155
    },
    {
      "epoch": 0.45706371191135736,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9185932120019675,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.26184895634651184,
      "eval_runtime": 5.153,
      "eval_samples_per_second": 5.822,
      "eval_steps_per_second": 0.194,
      "step": 1155
    },
    {
      "epoch": 0.45745943806885636,
      "grad_norm": 0.6434595278635993,
      "learning_rate": 6.588379184792954e-05,
      "loss": 0.1657,
      "step": 1156
    },
    {
      "epoch": 0.45785516422635536,
      "grad_norm": 0.6641560063692152,
      "learning_rate": 6.581827855321587e-05,
      "loss": 0.1387,
      "step": 1157
    },
    {
      "epoch": 0.45825089038385436,
      "grad_norm": 0.8308891982185348,
      "learning_rate": 6.575273506746905e-05,
      "loss": 0.1578,
      "step": 1158
    },
    {
      "epoch": 0.45864661654135336,
      "grad_norm": 1.2185339319181352,
      "learning_rate": 6.568716151578653e-05,
      "loss": 0.1864,
      "step": 1159
    },
    {
      "epoch": 0.4590423426988524,
      "grad_norm": 0.81190074935988,
      "learning_rate": 6.562155802332307e-05,
      "loss": 0.1612,
      "step": 1160
    },
    {
      "epoch": 0.4590423426988524,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9217904574520414,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.25517576932907104,
      "eval_runtime": 5.0096,
      "eval_samples_per_second": 5.989,
      "eval_steps_per_second": 0.2,
      "step": 1160
    },
    {
      "epoch": 0.4594380688563514,
      "grad_norm": 0.7731972379632047,
      "learning_rate": 6.555592471529059e-05,
      "loss": 0.1795,
      "step": 1161
    },
    {
      "epoch": 0.4598337950138504,
      "grad_norm": 1.4111703158541495,
      "learning_rate": 6.549026171695799e-05,
      "loss": 0.1628,
      "step": 1162
    },
    {
      "epoch": 0.4602295211713494,
      "grad_norm": 0.9259779793654891,
      "learning_rate": 6.54245691536507e-05,
      "loss": 0.1814,
      "step": 1163
    },
    {
      "epoch": 0.46062524732884846,
      "grad_norm": 0.8093043893412636,
      "learning_rate": 6.535884715075067e-05,
      "loss": 0.1556,
      "step": 1164
    },
    {
      "epoch": 0.46102097348634746,
      "grad_norm": 1.2880839068274712,
      "learning_rate": 6.529309583369605e-05,
      "loss": 0.2285,
      "step": 1165
    },
    {
      "epoch": 0.46102097348634746,
      "eval_PRM Accuracy": 0.9047619047619048,
      "eval_PRM F1": 0.9454545454545454,
      "eval_PRM F1 AUC": 0.7491392031480572,
      "eval_PRM F1 AUC (fixed)": 0.9235120511559273,
      "eval_PRM F1 Neg": 0.625,
      "eval_PRM NPV": 0.7692307692307693,
      "eval_PRM Precision": 0.9203539823008849,
      "eval_PRM Recall": 0.9719626168224299,
      "eval_PRM Specificty": 0.5263157894736842,
      "eval_loss": 0.24290364980697632,
      "eval_runtime": 4.9763,
      "eval_samples_per_second": 6.029,
      "eval_steps_per_second": 0.201,
      "step": 1165
    },
    {
      "epoch": 0.46141669964384646,
      "grad_norm": 1.2057938803953123,
      "learning_rate": 6.522731532798091e-05,
      "loss": 0.2585,
      "step": 1166
    },
    {
      "epoch": 0.46181242580134546,
      "grad_norm": 1.261684806128178,
      "learning_rate": 6.516150575915502e-05,
      "loss": 0.2081,
      "step": 1167
    },
    {
      "epoch": 0.46220815195884446,
      "grad_norm": 0.7961911508652638,
      "learning_rate": 6.509566725282362e-05,
      "loss": 0.1264,
      "step": 1168
    },
    {
      "epoch": 0.4626038781163435,
      "grad_norm": 1.0289455243270862,
      "learning_rate": 6.502979993464723e-05,
      "loss": 0.1689,
      "step": 1169
    },
    {
      "epoch": 0.4629996042738425,
      "grad_norm": 1.1324814023589889,
      "learning_rate": 6.496390393034129e-05,
      "loss": 0.1993,
      "step": 1170
    },
    {
      "epoch": 0.4629996042738425,
      "eval_PRM Accuracy": 0.8888888888888888,
      "eval_PRM F1": 0.9369369369369369,
      "eval_PRM F1 AUC": 0.6965076242006887,
      "eval_PRM F1 AUC (fixed)": 0.9220363994097394,
      "eval_PRM F1 Neg": 0.5333333333333333,
      "eval_PRM NPV": 0.7272727272727273,
      "eval_PRM Precision": 0.9043478260869565,
      "eval_PRM Recall": 0.9719626168224299,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.25183919072151184,
      "eval_runtime": 5.2571,
      "eval_samples_per_second": 5.707,
      "eval_steps_per_second": 0.19,
      "step": 1170
    },
    {
      "epoch": 0.4633953304313415,
      "grad_norm": 0.7775846183421931,
      "learning_rate": 6.489797936567603e-05,
      "loss": 0.1318,
      "step": 1171
    },
    {
      "epoch": 0.4637910565888405,
      "grad_norm": 0.8403784984596453,
      "learning_rate": 6.48320263664762e-05,
      "loss": 0.1429,
      "step": 1172
    },
    {
      "epoch": 0.4641867827463395,
      "grad_norm": 0.5494352155704674,
      "learning_rate": 6.47660450586208e-05,
      "loss": 0.0922,
      "step": 1173
    },
    {
      "epoch": 0.46458250890383856,
      "grad_norm": 1.1300655946136449,
      "learning_rate": 6.470003556804286e-05,
      "loss": 0.1839,
      "step": 1174
    },
    {
      "epoch": 0.46497823506133756,
      "grad_norm": 0.6693066955599658,
      "learning_rate": 6.46339980207292e-05,
      "loss": 0.1322,
      "step": 1175
    },
    {
      "epoch": 0.46497823506133756,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9422222222222222,
      "eval_PRM F1 AUC": 0.6795376291195278,
      "eval_PRM F1 AUC (fixed)": 0.9212985735366453,
      "eval_PRM F1 Neg": 0.5185185185185185,
      "eval_PRM NPV": 0.875,
      "eval_PRM Precision": 0.8983050847457628,
      "eval_PRM Recall": 0.9906542056074766,
      "eval_PRM Specificty": 0.3684210526315789,
      "eval_loss": 0.28372395038604736,
      "eval_runtime": 5.0356,
      "eval_samples_per_second": 5.958,
      "eval_steps_per_second": 0.199,
      "step": 1175
    },
    {
      "epoch": 0.46537396121883656,
      "grad_norm": 1.7296410650735405,
      "learning_rate": 6.456793254272023e-05,
      "loss": 0.2192,
      "step": 1176
    },
    {
      "epoch": 0.46576968737633556,
      "grad_norm": 2.457712266051786,
      "learning_rate": 6.45018392601096e-05,
      "loss": 0.1705,
      "step": 1177
    },
    {
      "epoch": 0.46616541353383456,
      "grad_norm": 0.9573637613921339,
      "learning_rate": 6.443571829904408e-05,
      "loss": 0.1298,
      "step": 1178
    },
    {
      "epoch": 0.4665611396913336,
      "grad_norm": 0.640223509839757,
      "learning_rate": 6.436956978572324e-05,
      "loss": 0.0881,
      "step": 1179
    },
    {
      "epoch": 0.4669568658488326,
      "grad_norm": 1.223430170695839,
      "learning_rate": 6.430339384639927e-05,
      "loss": 0.1462,
      "step": 1180
    },
    {
      "epoch": 0.4669568658488326,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9205607476635513,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.28129881620407104,
      "eval_runtime": 4.9031,
      "eval_samples_per_second": 6.119,
      "eval_steps_per_second": 0.204,
      "step": 1180
    },
    {
      "epoch": 0.4673525920063316,
      "grad_norm": 0.6648895831230512,
      "learning_rate": 6.423719060737665e-05,
      "loss": 0.1217,
      "step": 1181
    },
    {
      "epoch": 0.4677483181638306,
      "grad_norm": 1.18671155334821,
      "learning_rate": 6.417096019501203e-05,
      "loss": 0.2012,
      "step": 1182
    },
    {
      "epoch": 0.46814404432132967,
      "grad_norm": 1.644761369023588,
      "learning_rate": 6.410470273571387e-05,
      "loss": 0.3058,
      "step": 1183
    },
    {
      "epoch": 0.46853977047882867,
      "grad_norm": 1.0696174108542842,
      "learning_rate": 6.403841835594228e-05,
      "loss": 0.1758,
      "step": 1184
    },
    {
      "epoch": 0.46893549663632766,
      "grad_norm": 0.8647100282116553,
      "learning_rate": 6.397210718220874e-05,
      "loss": 0.1299,
      "step": 1185
    },
    {
      "epoch": 0.46893549663632766,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9411764705882353,
      "eval_PRM F1 AUC": 0.7228234136743729,
      "eval_PRM F1 AUC (fixed)": 0.9195769798327595,
      "eval_PRM F1 Neg": 0.5806451612903226,
      "eval_PRM NPV": 0.75,
      "eval_PRM Precision": 0.9122807017543859,
      "eval_PRM Recall": 0.9719626168224299,
      "eval_PRM Specificty": 0.47368421052631576,
      "eval_loss": 0.2601887881755829,
      "eval_runtime": 5.2108,
      "eval_samples_per_second": 5.757,
      "eval_steps_per_second": 0.192,
      "step": 1185
    },
    {
      "epoch": 0.46933122279382666,
      "grad_norm": 1.0679510415952915,
      "learning_rate": 6.390576934107589e-05,
      "loss": 0.1404,
      "step": 1186
    },
    {
      "epoch": 0.46972694895132566,
      "grad_norm": 1.0132937689229329,
      "learning_rate": 6.383940495915723e-05,
      "loss": 0.1398,
      "step": 1187
    },
    {
      "epoch": 0.4701226751088247,
      "grad_norm": 1.419010378085849,
      "learning_rate": 6.377301416311696e-05,
      "loss": 0.1728,
      "step": 1188
    },
    {
      "epoch": 0.4705184012663237,
      "grad_norm": 0.962856417397252,
      "learning_rate": 6.370659707966967e-05,
      "loss": 0.1484,
      "step": 1189
    },
    {
      "epoch": 0.4709141274238227,
      "grad_norm": 1.1061821300016605,
      "learning_rate": 6.36401538355801e-05,
      "loss": 0.1897,
      "step": 1190
    },
    {
      "epoch": 0.4709141274238227,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9171175602557796,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.26953125,
      "eval_runtime": 5.1337,
      "eval_samples_per_second": 5.844,
      "eval_steps_per_second": 0.195,
      "step": 1190
    },
    {
      "epoch": 0.4713098535813217,
      "grad_norm": 0.9736006904333611,
      "learning_rate": 6.357368455766299e-05,
      "loss": 0.188,
      "step": 1191
    },
    {
      "epoch": 0.4717055797388207,
      "grad_norm": 1.038184162387294,
      "learning_rate": 6.350718937278269e-05,
      "loss": 0.198,
      "step": 1192
    },
    {
      "epoch": 0.47210130589631977,
      "grad_norm": 1.0768791841608756,
      "learning_rate": 6.344066840785302e-05,
      "loss": 0.2056,
      "step": 1193
    },
    {
      "epoch": 0.47249703205381877,
      "grad_norm": 0.7131285423214887,
      "learning_rate": 6.337412178983704e-05,
      "loss": 0.1295,
      "step": 1194
    },
    {
      "epoch": 0.47289275821131777,
      "grad_norm": 1.0602051510555484,
      "learning_rate": 6.330754964574676e-05,
      "loss": 0.2253,
      "step": 1195
    },
    {
      "epoch": 0.47289275821131777,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9178553861288735,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.2586914002895355,
      "eval_runtime": 4.8797,
      "eval_samples_per_second": 6.148,
      "eval_steps_per_second": 0.205,
      "step": 1195
    },
    {
      "epoch": 0.47328848436881676,
      "grad_norm": 0.8869746342527639,
      "learning_rate": 6.324095210264286e-05,
      "loss": 0.2011,
      "step": 1196
    },
    {
      "epoch": 0.47368421052631576,
      "grad_norm": 0.8780073055965166,
      "learning_rate": 6.317432928763456e-05,
      "loss": 0.1736,
      "step": 1197
    },
    {
      "epoch": 0.4740799366838148,
      "grad_norm": 0.8397557045959201,
      "learning_rate": 6.310768132787928e-05,
      "loss": 0.1341,
      "step": 1198
    },
    {
      "epoch": 0.4744756628413138,
      "grad_norm": 0.8010634043987951,
      "learning_rate": 6.304100835058244e-05,
      "loss": 0.1645,
      "step": 1199
    },
    {
      "epoch": 0.4748713889988128,
      "grad_norm": 1.080326066602254,
      "learning_rate": 6.29743104829972e-05,
      "loss": 0.2335,
      "step": 1200
    },
    {
      "epoch": 0.4748713889988128,
      "eval_PRM Accuracy": 0.8888888888888888,
      "eval_PRM F1": 0.9369369369369369,
      "eval_PRM F1 AUC": 0.6965076242006887,
      "eval_PRM F1 AUC (fixed)": 0.9183472700442695,
      "eval_PRM F1 Neg": 0.5333333333333333,
      "eval_PRM NPV": 0.7272727272727273,
      "eval_PRM Precision": 0.9043478260869565,
      "eval_PRM Recall": 0.9719626168224299,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.2522786557674408,
      "eval_runtime": 4.9755,
      "eval_samples_per_second": 6.03,
      "eval_steps_per_second": 0.201,
      "step": 1200
    },
    {
      "epoch": 0.4752671151563118,
      "grad_norm": 0.6436342751392566,
      "learning_rate": 6.290758785242425e-05,
      "loss": 0.1145,
      "step": 1201
    },
    {
      "epoch": 0.47566284131381087,
      "grad_norm": 0.7496013707287245,
      "learning_rate": 6.28408405862115e-05,
      "loss": 0.2087,
      "step": 1202
    },
    {
      "epoch": 0.47605856747130987,
      "grad_norm": 0.9089122119180205,
      "learning_rate": 6.277406881175395e-05,
      "loss": 0.1789,
      "step": 1203
    },
    {
      "epoch": 0.47645429362880887,
      "grad_norm": 0.6394960085556239,
      "learning_rate": 6.27072726564933e-05,
      "loss": 0.1213,
      "step": 1204
    },
    {
      "epoch": 0.47685001978630787,
      "grad_norm": 0.5094712700151979,
      "learning_rate": 6.264045224791784e-05,
      "loss": 0.1227,
      "step": 1205
    },
    {
      "epoch": 0.47685001978630787,
      "eval_PRM Accuracy": 0.8888888888888888,
      "eval_PRM F1": 0.9375,
      "eval_PRM F1 AUC": 0.6748647319232661,
      "eval_PRM F1 AUC (fixed)": 0.9131824889326119,
      "eval_PRM F1 Neg": 0.5,
      "eval_PRM NPV": 0.7777777777777778,
      "eval_PRM Precision": 0.8974358974358975,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.3684210526315789,
      "eval_loss": 0.2726074159145355,
      "eval_runtime": 5.1091,
      "eval_samples_per_second": 5.872,
      "eval_steps_per_second": 0.196,
      "step": 1205
    },
    {
      "epoch": 0.47724574594380686,
      "grad_norm": 0.9731631552947037,
      "learning_rate": 6.25736077135621e-05,
      "loss": 0.2645,
      "step": 1206
    },
    {
      "epoch": 0.4776414721013059,
      "grad_norm": 0.7602244579134391,
      "learning_rate": 6.250673918100671e-05,
      "loss": 0.2106,
      "step": 1207
    },
    {
      "epoch": 0.4780371982588049,
      "grad_norm": 0.7384163745495891,
      "learning_rate": 6.243984677787808e-05,
      "loss": 0.1544,
      "step": 1208
    },
    {
      "epoch": 0.4784329244163039,
      "grad_norm": 0.7262152689101322,
      "learning_rate": 6.237293063184816e-05,
      "loss": 0.1369,
      "step": 1209
    },
    {
      "epoch": 0.4788286505738029,
      "grad_norm": 0.9026487803436578,
      "learning_rate": 6.230599087063426e-05,
      "loss": 0.2031,
      "step": 1210
    },
    {
      "epoch": 0.4788286505738029,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9422222222222222,
      "eval_PRM F1 AUC": 0.6795376291195278,
      "eval_PRM F1 AUC (fixed)": 0.9158878504672898,
      "eval_PRM F1 Neg": 0.5185185185185185,
      "eval_PRM NPV": 0.875,
      "eval_PRM Precision": 0.8983050847457628,
      "eval_PRM Recall": 0.9906542056074766,
      "eval_PRM Specificty": 0.3684210526315789,
      "eval_loss": 0.27071940898895264,
      "eval_runtime": 4.8221,
      "eval_samples_per_second": 6.221,
      "eval_steps_per_second": 0.207,
      "step": 1210
    },
    {
      "epoch": 0.4792243767313019,
      "grad_norm": 0.7188204065356958,
      "learning_rate": 6.223902762199871e-05,
      "loss": 0.1731,
      "step": 1211
    },
    {
      "epoch": 0.47962010288880097,
      "grad_norm": 0.604930109786888,
      "learning_rate": 6.21720410137487e-05,
      "loss": 0.123,
      "step": 1212
    },
    {
      "epoch": 0.48001582904629997,
      "grad_norm": 0.7829529833023485,
      "learning_rate": 6.2105031173736e-05,
      "loss": 0.1669,
      "step": 1213
    },
    {
      "epoch": 0.48041155520379897,
      "grad_norm": 0.5953277695430208,
      "learning_rate": 6.203799822985676e-05,
      "loss": 0.1625,
      "step": 1214
    },
    {
      "epoch": 0.48080728136129797,
      "grad_norm": 1.3415921184872912,
      "learning_rate": 6.197094231005112e-05,
      "loss": 0.2279,
      "step": 1215
    },
    {
      "epoch": 0.48080728136129797,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9198229217904574,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.2523437440395355,
      "eval_runtime": 5.0891,
      "eval_samples_per_second": 5.895,
      "eval_steps_per_second": 0.197,
      "step": 1215
    },
    {
      "epoch": 0.48120300751879697,
      "grad_norm": 0.5714214948155252,
      "learning_rate": 6.19038635423032e-05,
      "loss": 0.0881,
      "step": 1216
    },
    {
      "epoch": 0.481598733676296,
      "grad_norm": 0.8232087013423751,
      "learning_rate": 6.183676205464062e-05,
      "loss": 0.128,
      "step": 1217
    },
    {
      "epoch": 0.481994459833795,
      "grad_norm": 0.7390902590751279,
      "learning_rate": 6.176963797513443e-05,
      "loss": 0.1316,
      "step": 1218
    },
    {
      "epoch": 0.482390185991294,
      "grad_norm": 1.175398095573704,
      "learning_rate": 6.170249143189878e-05,
      "loss": 0.191,
      "step": 1219
    },
    {
      "epoch": 0.482785912148793,
      "grad_norm": 1.0304710540061501,
      "learning_rate": 6.16353225530907e-05,
      "loss": 0.1252,
      "step": 1220
    },
    {
      "epoch": 0.482785912148793,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9411764705882353,
      "eval_PRM F1 AUC": 0.7228234136743729,
      "eval_PRM F1 AUC (fixed)": 0.9210526315789473,
      "eval_PRM F1 Neg": 0.5806451612903226,
      "eval_PRM NPV": 0.75,
      "eval_PRM Precision": 0.9122807017543859,
      "eval_PRM Recall": 0.9719626168224299,
      "eval_PRM Specificty": 0.47368421052631576,
      "eval_loss": 0.2556803524494171,
      "eval_runtime": 4.9947,
      "eval_samples_per_second": 6.006,
      "eval_steps_per_second": 0.2,
      "step": 1220
    },
    {
      "epoch": 0.48318163830629207,
      "grad_norm": 0.937419113542309,
      "learning_rate": 6.156813146690983e-05,
      "loss": 0.1526,
      "step": 1221
    },
    {
      "epoch": 0.48357736446379107,
      "grad_norm": 0.9944453358390976,
      "learning_rate": 6.150091830159823e-05,
      "loss": 0.1846,
      "step": 1222
    },
    {
      "epoch": 0.48397309062129007,
      "grad_norm": 0.6237762702091242,
      "learning_rate": 6.143368318544006e-05,
      "loss": 0.1321,
      "step": 1223
    },
    {
      "epoch": 0.48436881677878907,
      "grad_norm": 0.7716116048495874,
      "learning_rate": 6.13664262467614e-05,
      "loss": 0.1881,
      "step": 1224
    },
    {
      "epoch": 0.48476454293628807,
      "grad_norm": 1.4999087967853473,
      "learning_rate": 6.129914761393001e-05,
      "loss": 0.2003,
      "step": 1225
    },
    {
      "epoch": 0.48476454293628807,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9422222222222222,
      "eval_PRM F1 AUC": 0.6795376291195278,
      "eval_PRM F1 AUC (fixed)": 0.9212985735366453,
      "eval_PRM F1 Neg": 0.5185185185185185,
      "eval_PRM NPV": 0.875,
      "eval_PRM Precision": 0.8983050847457628,
      "eval_PRM Recall": 0.9906542056074766,
      "eval_PRM Specificty": 0.3684210526315789,
      "eval_loss": 0.2835530638694763,
      "eval_runtime": 5.1981,
      "eval_samples_per_second": 5.771,
      "eval_steps_per_second": 0.192,
      "step": 1225
    },
    {
      "epoch": 0.4851602690937871,
      "grad_norm": 0.8185426730725777,
      "learning_rate": 6.123184741535495e-05,
      "loss": 0.1713,
      "step": 1226
    },
    {
      "epoch": 0.4855559952512861,
      "grad_norm": 1.606934979187631,
      "learning_rate": 6.11645257794866e-05,
      "loss": 0.2313,
      "step": 1227
    },
    {
      "epoch": 0.4859517214087851,
      "grad_norm": 1.4560090540493398,
      "learning_rate": 6.109718283481611e-05,
      "loss": 0.1796,
      "step": 1228
    },
    {
      "epoch": 0.4863474475662841,
      "grad_norm": 1.7939926844544278,
      "learning_rate": 6.1029818709875374e-05,
      "loss": 0.1502,
      "step": 1229
    },
    {
      "epoch": 0.4867431737237831,
      "grad_norm": 0.75349366658544,
      "learning_rate": 6.0962433533236705e-05,
      "loss": 0.1729,
      "step": 1230
    },
    {
      "epoch": 0.4867431737237831,
      "eval_PRM Accuracy": 0.9047619047619048,
      "eval_PRM F1": 0.9464285714285714,
      "eval_PRM F1 AUC": 0.7058534185932119,
      "eval_PRM F1 AUC (fixed)": 0.9225282833251353,
      "eval_PRM F1 Neg": 0.5714285714285714,
      "eval_PRM NPV": 0.8888888888888888,
      "eval_PRM Precision": 0.905982905982906,
      "eval_PRM Recall": 0.9906542056074766,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.2630371153354645,
      "eval_runtime": 5.0778,
      "eval_samples_per_second": 5.908,
      "eval_steps_per_second": 0.197,
      "step": 1230
    },
    {
      "epoch": 0.4871388998812822,
      "grad_norm": 1.3218422672275125,
      "learning_rate": 6.089502743351259e-05,
      "loss": 0.1866,
      "step": 1231
    },
    {
      "epoch": 0.48753462603878117,
      "grad_norm": 0.8852562021304003,
      "learning_rate": 6.082760053935541e-05,
      "loss": 0.1534,
      "step": 1232
    },
    {
      "epoch": 0.48793035219628017,
      "grad_norm": 0.6898448573533574,
      "learning_rate": 6.076015297945733e-05,
      "loss": 0.1216,
      "step": 1233
    },
    {
      "epoch": 0.48832607835377917,
      "grad_norm": 0.9071798587505693,
      "learning_rate": 6.0692684882549864e-05,
      "loss": 0.1783,
      "step": 1234
    },
    {
      "epoch": 0.48872180451127817,
      "grad_norm": 0.8145988876723135,
      "learning_rate": 6.062519637740378e-05,
      "loss": 0.186,
      "step": 1235
    },
    {
      "epoch": 0.48872180451127817,
      "eval_PRM Accuracy": 0.9047619047619048,
      "eval_PRM F1": 0.9464285714285714,
      "eval_PRM F1 AUC": 0.7058534185932119,
      "eval_PRM F1 AUC (fixed)": 0.9193310378750614,
      "eval_PRM F1 Neg": 0.5714285714285714,
      "eval_PRM NPV": 0.8888888888888888,
      "eval_PRM Precision": 0.905982905982906,
      "eval_PRM Recall": 0.9906542056074766,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.25060221552848816,
      "eval_runtime": 5.1805,
      "eval_samples_per_second": 5.791,
      "eval_steps_per_second": 0.193,
      "step": 1235
    },
    {
      "epoch": 0.4891175306687772,
      "grad_norm": 0.8525005363756072,
      "learning_rate": 6.055768759282874e-05,
      "loss": 0.2101,
      "step": 1236
    },
    {
      "epoch": 0.4895132568262762,
      "grad_norm": 0.8566513028659172,
      "learning_rate": 6.049015865767318e-05,
      "loss": 0.2438,
      "step": 1237
    },
    {
      "epoch": 0.4899089829837752,
      "grad_norm": 0.9041281750931823,
      "learning_rate": 6.042260970082395e-05,
      "loss": 0.1795,
      "step": 1238
    },
    {
      "epoch": 0.4903047091412742,
      "grad_norm": 1.5277362926305358,
      "learning_rate": 6.035504085120613e-05,
      "loss": 0.2089,
      "step": 1239
    },
    {
      "epoch": 0.4907004352987733,
      "grad_norm": 0.7355586067171286,
      "learning_rate": 6.028745223778278e-05,
      "loss": 0.1215,
      "step": 1240
    },
    {
      "epoch": 0.4907004352987733,
      "eval_PRM Accuracy": 0.9047619047619048,
      "eval_PRM F1": 0.9464285714285714,
      "eval_PRM F1 AUC": 0.7058534185932119,
      "eval_PRM F1 AUC (fixed)": 0.9215445154943434,
      "eval_PRM F1 Neg": 0.5714285714285714,
      "eval_PRM NPV": 0.8888888888888888,
      "eval_PRM Precision": 0.905982905982906,
      "eval_PRM Recall": 0.9906542056074766,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.25515949726104736,
      "eval_runtime": 4.9416,
      "eval_samples_per_second": 6.071,
      "eval_steps_per_second": 0.202,
      "step": 1240
    },
    {
      "epoch": 0.4910961614562723,
      "grad_norm": 0.6044613239958214,
      "learning_rate": 6.021984398955466e-05,
      "loss": 0.1288,
      "step": 1241
    },
    {
      "epoch": 0.49149188761377127,
      "grad_norm": 0.4668583351135625,
      "learning_rate": 6.015221623555999e-05,
      "loss": 0.1129,
      "step": 1242
    },
    {
      "epoch": 0.49188761377127027,
      "grad_norm": 0.7636514561520606,
      "learning_rate": 6.008456910487428e-05,
      "loss": 0.1675,
      "step": 1243
    },
    {
      "epoch": 0.49228333992876927,
      "grad_norm": 0.5837820138597654,
      "learning_rate": 6.0016902726609945e-05,
      "loss": 0.156,
      "step": 1244
    },
    {
      "epoch": 0.4926790660862683,
      "grad_norm": 0.6329801625698934,
      "learning_rate": 5.99492172299162e-05,
      "loss": 0.1696,
      "step": 1245
    },
    {
      "epoch": 0.4926790660862683,
      "eval_PRM Accuracy": 0.9047619047619048,
      "eval_PRM F1": 0.9464285714285714,
      "eval_PRM F1 AUC": 0.7058534185932119,
      "eval_PRM F1 AUC (fixed)": 0.9188391539596655,
      "eval_PRM F1 Neg": 0.5714285714285714,
      "eval_PRM NPV": 0.8888888888888888,
      "eval_PRM Precision": 0.905982905982906,
      "eval_PRM Recall": 0.9906542056074766,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.25678712129592896,
      "eval_runtime": 4.9834,
      "eval_samples_per_second": 6.02,
      "eval_steps_per_second": 0.201,
      "step": 1245
    },
    {
      "epoch": 0.4930747922437673,
      "grad_norm": 0.8750988690177731,
      "learning_rate": 5.988151274397873e-05,
      "loss": 0.1554,
      "step": 1246
    },
    {
      "epoch": 0.4934705184012663,
      "grad_norm": 0.6724957460922982,
      "learning_rate": 5.981378939801942e-05,
      "loss": 0.1422,
      "step": 1247
    },
    {
      "epoch": 0.4938662445587653,
      "grad_norm": 0.5411033303693988,
      "learning_rate": 5.974604732129625e-05,
      "loss": 0.0903,
      "step": 1248
    },
    {
      "epoch": 0.4942619707162643,
      "grad_norm": 0.957114851322142,
      "learning_rate": 5.967828664310283e-05,
      "loss": 0.1617,
      "step": 1249
    },
    {
      "epoch": 0.4946576968737634,
      "grad_norm": 0.9341282646903378,
      "learning_rate": 5.961050749276838e-05,
      "loss": 0.1471,
      "step": 1250
    },
    {
      "epoch": 0.4946576968737634,
      "eval_PRM Accuracy": 0.9047619047619048,
      "eval_PRM F1": 0.9464285714285714,
      "eval_PRM F1 AUC": 0.7058534185932119,
      "eval_PRM F1 AUC (fixed)": 0.9193310378750615,
      "eval_PRM F1 Neg": 0.5714285714285714,
      "eval_PRM NPV": 0.8888888888888888,
      "eval_PRM Precision": 0.905982905982906,
      "eval_PRM Recall": 0.9906542056074766,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.27774250507354736,
      "eval_runtime": 4.9448,
      "eval_samples_per_second": 6.067,
      "eval_steps_per_second": 0.202,
      "step": 1250
    },
    {
      "epoch": 0.4950534230312624,
      "grad_norm": 1.3223040581761416,
      "learning_rate": 5.9542709999657286e-05,
      "loss": 0.1842,
      "step": 1251
    },
    {
      "epoch": 0.4954491491887614,
      "grad_norm": 1.437478283294331,
      "learning_rate": 5.947489429316904e-05,
      "loss": 0.308,
      "step": 1252
    },
    {
      "epoch": 0.49584487534626037,
      "grad_norm": 0.5766331287742914,
      "learning_rate": 5.940706050273779e-05,
      "loss": 0.1116,
      "step": 1253
    },
    {
      "epoch": 0.49624060150375937,
      "grad_norm": 1.1308868121595035,
      "learning_rate": 5.933920875783228e-05,
      "loss": 0.1853,
      "step": 1254
    },
    {
      "epoch": 0.4966363276612584,
      "grad_norm": 0.9996995330571914,
      "learning_rate": 5.9271339187955475e-05,
      "loss": 0.1985,
      "step": 1255
    },
    {
      "epoch": 0.4966363276612584,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9217904574520412,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.25581055879592896,
      "eval_runtime": 5.008,
      "eval_samples_per_second": 5.99,
      "eval_steps_per_second": 0.2,
      "step": 1255
    },
    {
      "epoch": 0.4970320538187574,
      "grad_norm": 0.9358627128547623,
      "learning_rate": 5.92034519226444e-05,
      "loss": 0.1497,
      "step": 1256
    },
    {
      "epoch": 0.4974277799762564,
      "grad_norm": 0.9593487068416724,
      "learning_rate": 5.913554709146983e-05,
      "loss": 0.1313,
      "step": 1257
    },
    {
      "epoch": 0.4978235061337554,
      "grad_norm": 1.0474536600325186,
      "learning_rate": 5.906762482403607e-05,
      "loss": 0.2126,
      "step": 1258
    },
    {
      "epoch": 0.4982192322912545,
      "grad_norm": 1.102967046342181,
      "learning_rate": 5.8999685249980696e-05,
      "loss": 0.1472,
      "step": 1259
    },
    {
      "epoch": 0.4986149584487535,
      "grad_norm": 1.1540654096412584,
      "learning_rate": 5.8931728498974336e-05,
      "loss": 0.2085,
      "step": 1260
    },
    {
      "epoch": 0.4986149584487535,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9176094441711756,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.2561849057674408,
      "eval_runtime": 5.1348,
      "eval_samples_per_second": 5.843,
      "eval_steps_per_second": 0.195,
      "step": 1260
    },
    {
      "epoch": 0.4990106846062525,
      "grad_norm": 0.7451620863660723,
      "learning_rate": 5.88637547007204e-05,
      "loss": 0.187,
      "step": 1261
    },
    {
      "epoch": 0.4994064107637515,
      "grad_norm": 1.203363073854477,
      "learning_rate": 5.8795763984954776e-05,
      "loss": 0.1893,
      "step": 1262
    },
    {
      "epoch": 0.4998021369212505,
      "grad_norm": 0.9754846271341009,
      "learning_rate": 5.872775648144575e-05,
      "loss": 0.155,
      "step": 1263
    },
    {
      "epoch": 0.5001978630787495,
      "grad_norm": 0.7786825862926908,
      "learning_rate": 5.8659732319993555e-05,
      "loss": 0.1511,
      "step": 1264
    },
    {
      "epoch": 0.5005935892362485,
      "grad_norm": 0.661736921313685,
      "learning_rate": 5.859169163043027e-05,
      "loss": 0.1661,
      "step": 1265
    },
    {
      "epoch": 0.5005935892362485,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9185932120019676,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.2579589784145355,
      "eval_runtime": 5.1381,
      "eval_samples_per_second": 5.839,
      "eval_steps_per_second": 0.195,
      "step": 1265
    },
    {
      "epoch": 0.5009893153937476,
      "grad_norm": 0.9275708937695162,
      "learning_rate": 5.852363454261949e-05,
      "loss": 0.2063,
      "step": 1266
    },
    {
      "epoch": 0.5013850415512465,
      "grad_norm": 1.0315629317183914,
      "learning_rate": 5.845556118645612e-05,
      "loss": 0.1732,
      "step": 1267
    },
    {
      "epoch": 0.5017807677087456,
      "grad_norm": 0.596369131523145,
      "learning_rate": 5.838747169186611e-05,
      "loss": 0.1386,
      "step": 1268
    },
    {
      "epoch": 0.5021764938662445,
      "grad_norm": 0.7515547987746115,
      "learning_rate": 5.8319366188806256e-05,
      "loss": 0.198,
      "step": 1269
    },
    {
      "epoch": 0.5025722200237436,
      "grad_norm": 0.8318288446686654,
      "learning_rate": 5.8251244807263825e-05,
      "loss": 0.2171,
      "step": 1270
    },
    {
      "epoch": 0.5025722200237436,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9166256763403837,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.26656901836395264,
      "eval_runtime": 5.0953,
      "eval_samples_per_second": 5.888,
      "eval_steps_per_second": 0.196,
      "step": 1270
    },
    {
      "epoch": 0.5029679461812426,
      "grad_norm": 0.5201103992343463,
      "learning_rate": 5.8183107677256456e-05,
      "loss": 0.1517,
      "step": 1271
    },
    {
      "epoch": 0.5033636723387416,
      "grad_norm": 0.8481231158609983,
      "learning_rate": 5.81149549288318e-05,
      "loss": 0.1871,
      "step": 1272
    },
    {
      "epoch": 0.5037593984962406,
      "grad_norm": 1.5663429119778447,
      "learning_rate": 5.804678669206738e-05,
      "loss": 0.194,
      "step": 1273
    },
    {
      "epoch": 0.5041551246537396,
      "grad_norm": 0.8604830694166269,
      "learning_rate": 5.797860309707021e-05,
      "loss": 0.1742,
      "step": 1274
    },
    {
      "epoch": 0.5045508508112386,
      "grad_norm": 0.8594072493253806,
      "learning_rate": 5.791040427397666e-05,
      "loss": 0.2171,
      "step": 1275
    },
    {
      "epoch": 0.5045508508112386,
      "eval_PRM Accuracy": 0.8888888888888888,
      "eval_PRM F1": 0.9369369369369369,
      "eval_PRM F1 AUC": 0.6965076242006887,
      "eval_PRM F1 AUC (fixed)": 0.9166256763403837,
      "eval_PRM F1 Neg": 0.5333333333333333,
      "eval_PRM NPV": 0.7272727272727273,
      "eval_PRM Precision": 0.9043478260869565,
      "eval_PRM Recall": 0.9719626168224299,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.25244140625,
      "eval_runtime": 5.2093,
      "eval_samples_per_second": 5.759,
      "eval_steps_per_second": 0.192,
      "step": 1275
    },
    {
      "epoch": 0.5049465769687377,
      "grad_norm": 0.6676443115770968,
      "learning_rate": 5.7842190352952143e-05,
      "loss": 0.1343,
      "step": 1276
    },
    {
      "epoch": 0.5053423031262366,
      "grad_norm": 0.6537613019710552,
      "learning_rate": 5.777396146419093e-05,
      "loss": 0.1525,
      "step": 1277
    },
    {
      "epoch": 0.5057380292837357,
      "grad_norm": 0.6697577326441106,
      "learning_rate": 5.770571773791579e-05,
      "loss": 0.1574,
      "step": 1278
    },
    {
      "epoch": 0.5061337554412346,
      "grad_norm": 0.7175588717606146,
      "learning_rate": 5.7637459304377874e-05,
      "loss": 0.1903,
      "step": 1279
    },
    {
      "epoch": 0.5065294815987337,
      "grad_norm": 0.7088823285175888,
      "learning_rate": 5.756918629385638e-05,
      "loss": 0.1837,
      "step": 1280
    },
    {
      "epoch": 0.5065294815987337,
      "eval_PRM Accuracy": 0.8809523809523809,
      "eval_PRM F1": 0.9321266968325792,
      "eval_PRM F1 AUC": 0.6918347270044269,
      "eval_PRM F1 AUC (fixed)": 0.9242498770290212,
      "eval_PRM F1 Neg": 0.5161290322580645,
      "eval_PRM NPV": 0.6666666666666666,
      "eval_PRM Precision": 0.9035087719298246,
      "eval_PRM Recall": 0.9626168224299065,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.24498698115348816,
      "eval_runtime": 4.9719,
      "eval_samples_per_second": 6.034,
      "eval_steps_per_second": 0.201,
      "step": 1280
    },
    {
      "epoch": 0.5069252077562327,
      "grad_norm": 0.49974458734644467,
      "learning_rate": 5.75008988366583e-05,
      "loss": 0.1076,
      "step": 1281
    },
    {
      "epoch": 0.5073209339137317,
      "grad_norm": 0.8296951670736253,
      "learning_rate": 5.743259706311827e-05,
      "loss": 0.1806,
      "step": 1282
    },
    {
      "epoch": 0.5077166600712307,
      "grad_norm": 0.8532139685636388,
      "learning_rate": 5.736428110359815e-05,
      "loss": 0.2065,
      "step": 1283
    },
    {
      "epoch": 0.5081123862287297,
      "grad_norm": 0.7929357009673553,
      "learning_rate": 5.7295951088486985e-05,
      "loss": 0.12,
      "step": 1284
    },
    {
      "epoch": 0.5085081123862287,
      "grad_norm": 0.9495138944443309,
      "learning_rate": 5.722760714820057e-05,
      "loss": 0.1592,
      "step": 1285
    },
    {
      "epoch": 0.5085081123862287,
      "eval_PRM Accuracy": 0.8888888888888888,
      "eval_PRM F1": 0.9369369369369369,
      "eval_PRM F1 AUC": 0.6965076242006887,
      "eval_PRM F1 AUC (fixed)": 0.9259714707329071,
      "eval_PRM F1 Neg": 0.5333333333333333,
      "eval_PRM NPV": 0.7272727272727273,
      "eval_PRM Precision": 0.9043478260869565,
      "eval_PRM Recall": 0.9719626168224299,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.25255534052848816,
      "eval_runtime": 4.9432,
      "eval_samples_per_second": 6.069,
      "eval_steps_per_second": 0.202,
      "step": 1285
    },
    {
      "epoch": 0.5089038385437278,
      "grad_norm": 0.9729834465528704,
      "learning_rate": 5.7159249413181303e-05,
      "loss": 0.1603,
      "step": 1286
    },
    {
      "epoch": 0.5092995647012267,
      "grad_norm": 0.7720065464077377,
      "learning_rate": 5.70908780138979e-05,
      "loss": 0.1713,
      "step": 1287
    },
    {
      "epoch": 0.5096952908587258,
      "grad_norm": 1.0286893487242033,
      "learning_rate": 5.7022493080845194e-05,
      "loss": 0.1964,
      "step": 1288
    },
    {
      "epoch": 0.5100910170162247,
      "grad_norm": 1.3577615575193391,
      "learning_rate": 5.69540947445438e-05,
      "loss": 0.184,
      "step": 1289
    },
    {
      "epoch": 0.5104867431737238,
      "grad_norm": 0.717904360817334,
      "learning_rate": 5.688568313553994e-05,
      "loss": 0.1305,
      "step": 1290
    },
    {
      "epoch": 0.5104867431737238,
      "eval_PRM Accuracy": 0.8888888888888888,
      "eval_PRM F1": 0.9369369369369369,
      "eval_PRM F1 AUC": 0.6965076242006887,
      "eval_PRM F1 AUC (fixed)": 0.9291687161829808,
      "eval_PRM F1 Neg": 0.5333333333333333,
      "eval_PRM NPV": 0.7272727272727273,
      "eval_PRM Precision": 0.9043478260869565,
      "eval_PRM Recall": 0.9719626168224299,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.25586751103401184,
      "eval_runtime": 4.9805,
      "eval_samples_per_second": 6.023,
      "eval_steps_per_second": 0.201,
      "step": 1290
    },
    {
      "epoch": 0.5108824693312228,
      "grad_norm": 1.354644220707343,
      "learning_rate": 5.681725838440515e-05,
      "loss": 0.1903,
      "step": 1291
    },
    {
      "epoch": 0.5112781954887218,
      "grad_norm": 0.7862713112744212,
      "learning_rate": 5.6748820621736084e-05,
      "loss": 0.137,
      "step": 1292
    },
    {
      "epoch": 0.5116739216462208,
      "grad_norm": 1.1492624123949011,
      "learning_rate": 5.66803699781542e-05,
      "loss": 0.1922,
      "step": 1293
    },
    {
      "epoch": 0.5120696478037198,
      "grad_norm": 0.7378456887339425,
      "learning_rate": 5.661190658430556e-05,
      "loss": 0.1559,
      "step": 1294
    },
    {
      "epoch": 0.5124653739612188,
      "grad_norm": 0.7860158833261784,
      "learning_rate": 5.654343057086053e-05,
      "loss": 0.1682,
      "step": 1295
    },
    {
      "epoch": 0.5124653739612188,
      "eval_PRM Accuracy": 0.8888888888888888,
      "eval_PRM F1": 0.9369369369369369,
      "eval_PRM F1 AUC": 0.6965076242006887,
      "eval_PRM F1 AUC (fixed)": 0.9306443679291687,
      "eval_PRM F1 Neg": 0.5333333333333333,
      "eval_PRM NPV": 0.7272727272727273,
      "eval_PRM Precision": 0.9043478260869565,
      "eval_PRM Recall": 0.9719626168224299,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.2553955018520355,
      "eval_runtime": 5.2157,
      "eval_samples_per_second": 5.752,
      "eval_steps_per_second": 0.192,
      "step": 1295
    },
    {
      "epoch": 0.5128611001187179,
      "grad_norm": 0.86572065673799,
      "learning_rate": 5.647494206851363e-05,
      "loss": 0.1702,
      "step": 1296
    },
    {
      "epoch": 0.5132568262762168,
      "grad_norm": 1.156236947293318,
      "learning_rate": 5.640644120798312e-05,
      "loss": 0.1988,
      "step": 1297
    },
    {
      "epoch": 0.5136525524337159,
      "grad_norm": 0.5843493476289539,
      "learning_rate": 5.6337928120010906e-05,
      "loss": 0.1263,
      "step": 1298
    },
    {
      "epoch": 0.5140482785912148,
      "grad_norm": 0.887632617337494,
      "learning_rate": 5.626940293536225e-05,
      "loss": 0.1914,
      "step": 1299
    },
    {
      "epoch": 0.5144440047487139,
      "grad_norm": 0.7500639036464125,
      "learning_rate": 5.620086578482544e-05,
      "loss": 0.168,
      "step": 1300
    },
    {
      "epoch": 0.5144440047487139,
      "eval_PRM Accuracy": 0.8888888888888888,
      "eval_PRM F1": 0.9369369369369369,
      "eval_PRM F1 AUC": 0.6965076242006887,
      "eval_PRM F1 AUC (fixed)": 0.9281849483521889,
      "eval_PRM F1 Neg": 0.5333333333333333,
      "eval_PRM NPV": 0.7272727272727273,
      "eval_PRM Precision": 0.9043478260869565,
      "eval_PRM Recall": 0.9719626168224299,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.25045573711395264,
      "eval_runtime": 5.0426,
      "eval_samples_per_second": 5.949,
      "eval_steps_per_second": 0.198,
      "step": 1300
    },
    {
      "epoch": 0.5148397309062129,
      "grad_norm": 0.6875792767445299,
      "learning_rate": 5.613231679921167e-05,
      "loss": 0.1499,
      "step": 1301
    },
    {
      "epoch": 0.5152354570637119,
      "grad_norm": 0.998674602427952,
      "learning_rate": 5.606375610935466e-05,
      "loss": 0.2302,
      "step": 1302
    },
    {
      "epoch": 0.5156311832212109,
      "grad_norm": 1.3984470346937108,
      "learning_rate": 5.5995183846110524e-05,
      "loss": 0.1025,
      "step": 1303
    },
    {
      "epoch": 0.51602690937871,
      "grad_norm": 0.600460557271782,
      "learning_rate": 5.592660014035742e-05,
      "loss": 0.0965,
      "step": 1304
    },
    {
      "epoch": 0.5164226355362089,
      "grad_norm": 0.5812534636442083,
      "learning_rate": 5.585800512299539e-05,
      "loss": 0.1321,
      "step": 1305
    },
    {
      "epoch": 0.5164226355362089,
      "eval_PRM Accuracy": 0.8888888888888888,
      "eval_PRM F1": 0.9369369369369369,
      "eval_PRM F1 AUC": 0.6965076242006887,
      "eval_PRM F1 AUC (fixed)": 0.928184948352189,
      "eval_PRM F1 Neg": 0.5333333333333333,
      "eval_PRM NPV": 0.7272727272727273,
      "eval_PRM Precision": 0.9043478260869565,
      "eval_PRM Recall": 0.9719626168224299,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.25369465351104736,
      "eval_runtime": 5.1764,
      "eval_samples_per_second": 5.796,
      "eval_steps_per_second": 0.193,
      "step": 1305
    },
    {
      "epoch": 0.516818361693708,
      "grad_norm": 0.7161989569890755,
      "learning_rate": 5.578939892494601e-05,
      "loss": 0.1496,
      "step": 1306
    },
    {
      "epoch": 0.5172140878512069,
      "grad_norm": 1.3019151052338234,
      "learning_rate": 5.572078167715225e-05,
      "loss": 0.2622,
      "step": 1307
    },
    {
      "epoch": 0.517609814008706,
      "grad_norm": 1.1099355163335185,
      "learning_rate": 5.565215351057812e-05,
      "loss": 0.1921,
      "step": 1308
    },
    {
      "epoch": 0.518005540166205,
      "grad_norm": 0.7690316744376192,
      "learning_rate": 5.5583514556208514e-05,
      "loss": 0.1361,
      "step": 1309
    },
    {
      "epoch": 0.518401266323704,
      "grad_norm": 0.9266420961344689,
      "learning_rate": 5.5514864945048904e-05,
      "loss": 0.1411,
      "step": 1310
    },
    {
      "epoch": 0.518401266323704,
      "eval_PRM Accuracy": 0.8888888888888888,
      "eval_PRM F1": 0.9369369369369369,
      "eval_PRM F1 AUC": 0.6965076242006887,
      "eval_PRM F1 AUC (fixed)": 0.9259714707329071,
      "eval_PRM F1 Neg": 0.5333333333333333,
      "eval_PRM NPV": 0.7272727272727273,
      "eval_PRM Precision": 0.9043478260869565,
      "eval_PRM Recall": 0.9719626168224299,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.2521321475505829,
      "eval_runtime": 4.9673,
      "eval_samples_per_second": 6.039,
      "eval_steps_per_second": 0.201,
      "step": 1310
    },
    {
      "epoch": 0.518796992481203,
      "grad_norm": 0.9336610327589714,
      "learning_rate": 5.544620480812508e-05,
      "loss": 0.1627,
      "step": 1311
    },
    {
      "epoch": 0.519192718638702,
      "grad_norm": 4.357334786279429,
      "learning_rate": 5.537753427648295e-05,
      "loss": 0.1618,
      "step": 1312
    },
    {
      "epoch": 0.519588444796201,
      "grad_norm": 1.2448446229084362,
      "learning_rate": 5.5308853481188236e-05,
      "loss": 0.1983,
      "step": 1313
    },
    {
      "epoch": 0.5199841709537001,
      "grad_norm": 0.8612216114199529,
      "learning_rate": 5.524016255332627e-05,
      "loss": 0.2022,
      "step": 1314
    },
    {
      "epoch": 0.520379897111199,
      "grad_norm": 1.0072577932443008,
      "learning_rate": 5.517146162400171e-05,
      "loss": 0.1807,
      "step": 1315
    },
    {
      "epoch": 0.520379897111199,
      "eval_PRM Accuracy": 0.8888888888888888,
      "eval_PRM F1": 0.9369369369369369,
      "eval_PRM F1 AUC": 0.6965076242006887,
      "eval_PRM F1 AUC (fixed)": 0.9294146581406788,
      "eval_PRM F1 Neg": 0.5333333333333333,
      "eval_PRM NPV": 0.7272727272727273,
      "eval_PRM Precision": 0.9043478260869565,
      "eval_PRM Recall": 0.9719626168224299,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.24965819716453552,
      "eval_runtime": 5.1319,
      "eval_samples_per_second": 5.846,
      "eval_steps_per_second": 0.195,
      "step": 1315
    },
    {
      "epoch": 0.5207756232686981,
      "grad_norm": 0.5895166175989843,
      "learning_rate": 5.510275082433831e-05,
      "loss": 0.1776,
      "step": 1316
    },
    {
      "epoch": 0.521171349426197,
      "grad_norm": 1.0214312482867896,
      "learning_rate": 5.503403028547867e-05,
      "loss": 0.2498,
      "step": 1317
    },
    {
      "epoch": 0.5215670755836961,
      "grad_norm": 0.6681968355440214,
      "learning_rate": 5.4965300138583955e-05,
      "loss": 0.1408,
      "step": 1318
    },
    {
      "epoch": 0.5219628017411951,
      "grad_norm": 0.9072693257431519,
      "learning_rate": 5.4896560514833675e-05,
      "loss": 0.1919,
      "step": 1319
    },
    {
      "epoch": 0.5223585278986941,
      "grad_norm": 0.5944097016353368,
      "learning_rate": 5.4827811545425454e-05,
      "loss": 0.1256,
      "step": 1320
    },
    {
      "epoch": 0.5223585278986941,
      "eval_PRM Accuracy": 0.8888888888888888,
      "eval_PRM F1": 0.9369369369369369,
      "eval_PRM F1 AUC": 0.6965076242006887,
      "eval_PRM F1 AUC (fixed)": 0.927693064436793,
      "eval_PRM F1 Neg": 0.5333333333333333,
      "eval_PRM NPV": 0.7272727272727273,
      "eval_PRM Precision": 0.9043478260869565,
      "eval_PRM Recall": 0.9719626168224299,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.24230143427848816,
      "eval_runtime": 5.0758,
      "eval_samples_per_second": 5.91,
      "eval_steps_per_second": 0.197,
      "step": 1320
    },
    {
      "epoch": 0.5227542540561931,
      "grad_norm": 0.9535350043066437,
      "learning_rate": 5.475905336157473e-05,
      "loss": 0.1568,
      "step": 1321
    },
    {
      "epoch": 0.5231499802136921,
      "grad_norm": 0.5937254813773883,
      "learning_rate": 5.4690286094514534e-05,
      "loss": 0.1395,
      "step": 1322
    },
    {
      "epoch": 0.5235457063711911,
      "grad_norm": 0.8128019433392004,
      "learning_rate": 5.4621509875495216e-05,
      "loss": 0.2007,
      "step": 1323
    },
    {
      "epoch": 0.5239414325286902,
      "grad_norm": 0.8249606856756874,
      "learning_rate": 5.4552724835784244e-05,
      "loss": 0.1433,
      "step": 1324
    },
    {
      "epoch": 0.5243371586861891,
      "grad_norm": 1.013367480163338,
      "learning_rate": 5.448393110666588e-05,
      "loss": 0.2529,
      "step": 1325
    },
    {
      "epoch": 0.5243371586861891,
      "eval_PRM Accuracy": 0.8888888888888888,
      "eval_PRM F1": 0.9369369369369369,
      "eval_PRM F1 AUC": 0.6965076242006887,
      "eval_PRM F1 AUC (fixed)": 0.9328578455484505,
      "eval_PRM F1 Neg": 0.5333333333333333,
      "eval_PRM NPV": 0.7272727272727273,
      "eval_PRM Precision": 0.9043478260869565,
      "eval_PRM Recall": 0.9719626168224299,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.24285481870174408,
      "eval_runtime": 4.9824,
      "eval_samples_per_second": 6.021,
      "eval_steps_per_second": 0.201,
      "step": 1325
    },
    {
      "epoch": 0.5247328848436882,
      "grad_norm": 0.6141285047427931,
      "learning_rate": 5.441512881944104e-05,
      "loss": 0.0966,
      "step": 1326
    },
    {
      "epoch": 0.5251286110011871,
      "grad_norm": 0.7757912581343077,
      "learning_rate": 5.434631810542687e-05,
      "loss": 0.1561,
      "step": 1327
    },
    {
      "epoch": 0.5255243371586862,
      "grad_norm": 0.735806126651385,
      "learning_rate": 5.427749909595672e-05,
      "loss": 0.1654,
      "step": 1328
    },
    {
      "epoch": 0.5259200633161852,
      "grad_norm": 0.6099423731346483,
      "learning_rate": 5.420867192237966e-05,
      "loss": 0.1128,
      "step": 1329
    },
    {
      "epoch": 0.5263157894736842,
      "grad_norm": 0.6502506382097482,
      "learning_rate": 5.4139836716060435e-05,
      "loss": 0.1568,
      "step": 1330
    },
    {
      "epoch": 0.5263157894736842,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9422222222222222,
      "eval_PRM F1 AUC": 0.6795376291195278,
      "eval_PRM F1 AUC (fixed)": 0.9316281357599606,
      "eval_PRM F1 Neg": 0.5185185185185185,
      "eval_PRM NPV": 0.875,
      "eval_PRM Precision": 0.8983050847457628,
      "eval_PRM Recall": 0.9906542056074766,
      "eval_PRM Specificty": 0.3684210526315789,
      "eval_loss": 0.261474609375,
      "eval_runtime": 4.8722,
      "eval_samples_per_second": 6.157,
      "eval_steps_per_second": 0.205,
      "step": 1330
    },
    {
      "epoch": 0.5267115156311832,
      "grad_norm": 0.5790046140681657,
      "learning_rate": 5.407099360837905e-05,
      "loss": 0.093,
      "step": 1331
    },
    {
      "epoch": 0.5271072417886822,
      "grad_norm": 0.6318388717371626,
      "learning_rate": 5.400214273073065e-05,
      "loss": 0.1557,
      "step": 1332
    },
    {
      "epoch": 0.5275029679461812,
      "grad_norm": 0.8645136276819803,
      "learning_rate": 5.393328421452514e-05,
      "loss": 0.1922,
      "step": 1333
    },
    {
      "epoch": 0.5278986941036803,
      "grad_norm": 0.9952529685753582,
      "learning_rate": 5.3864418191187074e-05,
      "loss": 0.1561,
      "step": 1334
    },
    {
      "epoch": 0.5282944202611792,
      "grad_norm": 0.7482900562617728,
      "learning_rate": 5.3795544792155306e-05,
      "loss": 0.1933,
      "step": 1335
    },
    {
      "epoch": 0.5282944202611792,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9422222222222222,
      "eval_PRM F1 AUC": 0.6795376291195278,
      "eval_PRM F1 AUC (fixed)": 0.9323659616330547,
      "eval_PRM F1 Neg": 0.5185185185185185,
      "eval_PRM NPV": 0.875,
      "eval_PRM Precision": 0.8983050847457628,
      "eval_PRM Recall": 0.9906542056074766,
      "eval_PRM Specificty": 0.3684210526315789,
      "eval_loss": 0.28011882305145264,
      "eval_runtime": 4.8456,
      "eval_samples_per_second": 6.191,
      "eval_steps_per_second": 0.206,
      "step": 1335
    },
    {
      "epoch": 0.5286901464186783,
      "grad_norm": 0.7146309204853534,
      "learning_rate": 5.372666414888274e-05,
      "loss": 0.1487,
      "step": 1336
    },
    {
      "epoch": 0.5290858725761773,
      "grad_norm": 0.969053059135273,
      "learning_rate": 5.3657776392836175e-05,
      "loss": 0.138,
      "step": 1337
    },
    {
      "epoch": 0.5294815987336763,
      "grad_norm": 0.9919312007917611,
      "learning_rate": 5.3588881655495914e-05,
      "loss": 0.2185,
      "step": 1338
    },
    {
      "epoch": 0.5298773248911753,
      "grad_norm": 0.7927749701448865,
      "learning_rate": 5.351998006835562e-05,
      "loss": 0.1439,
      "step": 1339
    },
    {
      "epoch": 0.5302730510486743,
      "grad_norm": 0.72883730990061,
      "learning_rate": 5.345107176292202e-05,
      "loss": 0.1718,
      "step": 1340
    },
    {
      "epoch": 0.5302730510486743,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9422222222222222,
      "eval_PRM F1 AUC": 0.6795376291195278,
      "eval_PRM F1 AUC (fixed)": 0.9360550909985242,
      "eval_PRM F1 Neg": 0.5185185185185185,
      "eval_PRM NPV": 0.875,
      "eval_PRM Precision": 0.8983050847457628,
      "eval_PRM Recall": 0.9906542056074766,
      "eval_PRM Specificty": 0.3684210526315789,
      "eval_loss": 0.2523437440395355,
      "eval_runtime": 5.0493,
      "eval_samples_per_second": 5.941,
      "eval_steps_per_second": 0.198,
      "step": 1340
    },
    {
      "epoch": 0.5306687772061733,
      "grad_norm": 0.714933077279408,
      "learning_rate": 5.338215687071469e-05,
      "loss": 0.1405,
      "step": 1341
    },
    {
      "epoch": 0.5310645033636724,
      "grad_norm": 0.6629158241940782,
      "learning_rate": 5.331323552326573e-05,
      "loss": 0.1812,
      "step": 1342
    },
    {
      "epoch": 0.5314602295211713,
      "grad_norm": 1.7162642692607628,
      "learning_rate": 5.324430785211959e-05,
      "loss": 0.1331,
      "step": 1343
    },
    {
      "epoch": 0.5318559556786704,
      "grad_norm": 0.9605554159026668,
      "learning_rate": 5.317537398883279e-05,
      "loss": 0.1939,
      "step": 1344
    },
    {
      "epoch": 0.5322516818361693,
      "grad_norm": 0.7107028280100226,
      "learning_rate": 5.3106434064973665e-05,
      "loss": 0.1627,
      "step": 1345
    },
    {
      "epoch": 0.5322516818361693,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9360550909985244,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.23561197519302368,
      "eval_runtime": 5.3389,
      "eval_samples_per_second": 5.619,
      "eval_steps_per_second": 0.187,
      "step": 1345
    },
    {
      "epoch": 0.5326474079936684,
      "grad_norm": 0.4776712304531224,
      "learning_rate": 5.30374882121221e-05,
      "loss": 0.1007,
      "step": 1346
    },
    {
      "epoch": 0.5330431341511674,
      "grad_norm": 0.7613595761162486,
      "learning_rate": 5.296853656186934e-05,
      "loss": 0.1395,
      "step": 1347
    },
    {
      "epoch": 0.5334388603086664,
      "grad_norm": 0.5843256358849226,
      "learning_rate": 5.289957924581764e-05,
      "loss": 0.0926,
      "step": 1348
    },
    {
      "epoch": 0.5338345864661654,
      "grad_norm": 0.7437080927457614,
      "learning_rate": 5.283061639558011e-05,
      "loss": 0.1302,
      "step": 1349
    },
    {
      "epoch": 0.5342303126236644,
      "grad_norm": 0.9662153859348727,
      "learning_rate": 5.276164814278043e-05,
      "loss": 0.2347,
      "step": 1350
    },
    {
      "epoch": 0.5342303126236644,
      "eval_PRM Accuracy": 0.8888888888888888,
      "eval_PRM F1": 0.9369369369369369,
      "eval_PRM F1 AUC": 0.6965076242006887,
      "eval_PRM F1 AUC (fixed)": 0.9358091490408263,
      "eval_PRM F1 Neg": 0.5333333333333333,
      "eval_PRM NPV": 0.7272727272727273,
      "eval_PRM Precision": 0.9043478260869565,
      "eval_PRM Recall": 0.9719626168224299,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.23479817807674408,
      "eval_runtime": 5.0857,
      "eval_samples_per_second": 5.899,
      "eval_steps_per_second": 0.197,
      "step": 1350
    },
    {
      "epoch": 0.5346260387811634,
      "grad_norm": 0.9800309890579693,
      "learning_rate": 5.269267461905253e-05,
      "loss": 0.1586,
      "step": 1351
    },
    {
      "epoch": 0.5350217649386625,
      "grad_norm": 0.5288599921762195,
      "learning_rate": 5.2623695956040445e-05,
      "loss": 0.0806,
      "step": 1352
    },
    {
      "epoch": 0.5354174910961614,
      "grad_norm": 0.7202248114934156,
      "learning_rate": 5.2554712285398034e-05,
      "loss": 0.1118,
      "step": 1353
    },
    {
      "epoch": 0.5358132172536605,
      "grad_norm": 0.8436598977278604,
      "learning_rate": 5.24857237387887e-05,
      "loss": 0.1821,
      "step": 1354
    },
    {
      "epoch": 0.5362089434111594,
      "grad_norm": 1.0124374717920286,
      "learning_rate": 5.241673044788511e-05,
      "loss": 0.1635,
      "step": 1355
    },
    {
      "epoch": 0.5362089434111594,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9340875553369404,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.24814453721046448,
      "eval_runtime": 5.1505,
      "eval_samples_per_second": 5.825,
      "eval_steps_per_second": 0.194,
      "step": 1355
    },
    {
      "epoch": 0.5366046695686585,
      "grad_norm": 0.8502513369438582,
      "learning_rate": 5.2347732544369055e-05,
      "loss": 0.1412,
      "step": 1356
    },
    {
      "epoch": 0.5370003957261575,
      "grad_norm": 0.637521845298401,
      "learning_rate": 5.2278730159931076e-05,
      "loss": 0.1309,
      "step": 1357
    },
    {
      "epoch": 0.5373961218836565,
      "grad_norm": 0.620754515455489,
      "learning_rate": 5.220972342627032e-05,
      "loss": 0.0789,
      "step": 1358
    },
    {
      "epoch": 0.5377918480411555,
      "grad_norm": 0.8561992125971556,
      "learning_rate": 5.2140712475094166e-05,
      "loss": 0.155,
      "step": 1359
    },
    {
      "epoch": 0.5381875741986545,
      "grad_norm": 0.7793029962316026,
      "learning_rate": 5.207169743811809e-05,
      "loss": 0.1538,
      "step": 1360
    },
    {
      "epoch": 0.5381875741986545,
      "eval_PRM Accuracy": 0.9047619047619048,
      "eval_PRM F1": 0.9464285714285714,
      "eval_PRM F1 AUC": 0.7058534185932119,
      "eval_PRM F1 AUC (fixed)": 0.9318740777176586,
      "eval_PRM F1 Neg": 0.5714285714285714,
      "eval_PRM NPV": 0.8888888888888888,
      "eval_PRM Precision": 0.905982905982906,
      "eval_PRM Recall": 0.9906542056074766,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.2732340395450592,
      "eval_runtime": 5.1088,
      "eval_samples_per_second": 5.872,
      "eval_steps_per_second": 0.196,
      "step": 1360
    },
    {
      "epoch": 0.5385833003561535,
      "grad_norm": 1.273282783574179,
      "learning_rate": 5.200267844706537e-05,
      "loss": 0.1529,
      "step": 1361
    },
    {
      "epoch": 0.5389790265136526,
      "grad_norm": 1.0368121518258484,
      "learning_rate": 5.193365563366679e-05,
      "loss": 0.1367,
      "step": 1362
    },
    {
      "epoch": 0.5393747526711515,
      "grad_norm": 1.1829637162995341,
      "learning_rate": 5.186462912966047e-05,
      "loss": 0.1473,
      "step": 1363
    },
    {
      "epoch": 0.5397704788286506,
      "grad_norm": 1.6414468370034483,
      "learning_rate": 5.179559906679157e-05,
      "loss": 0.3343,
      "step": 1364
    },
    {
      "epoch": 0.5401662049861495,
      "grad_norm": 0.9841282699539626,
      "learning_rate": 5.172656557681199e-05,
      "loss": 0.1698,
      "step": 1365
    },
    {
      "epoch": 0.5401662049861495,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.927201180521397,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.2699137330055237,
      "eval_runtime": 5.2473,
      "eval_samples_per_second": 5.717,
      "eval_steps_per_second": 0.191,
      "step": 1365
    },
    {
      "epoch": 0.5405619311436486,
      "grad_norm": 0.9916364237249542,
      "learning_rate": 5.165752879148027e-05,
      "loss": 0.1748,
      "step": 1366
    },
    {
      "epoch": 0.5409576573011476,
      "grad_norm": 0.714329448951714,
      "learning_rate": 5.1588488842561145e-05,
      "loss": 0.1246,
      "step": 1367
    },
    {
      "epoch": 0.5413533834586466,
      "grad_norm": 0.6980850194027967,
      "learning_rate": 5.151944586182545e-05,
      "loss": 0.1316,
      "step": 1368
    },
    {
      "epoch": 0.5417491096161456,
      "grad_norm": 0.95637621536108,
      "learning_rate": 5.145039998104974e-05,
      "loss": 0.1662,
      "step": 1369
    },
    {
      "epoch": 0.5421448357736446,
      "grad_norm": 0.8222591280009837,
      "learning_rate": 5.13813513320162e-05,
      "loss": 0.1549,
      "step": 1370
    },
    {
      "epoch": 0.5421448357736446,
      "eval_PRM Accuracy": 0.8888888888888888,
      "eval_PRM F1": 0.9369369369369369,
      "eval_PRM F1 AUC": 0.6965076242006887,
      "eval_PRM F1 AUC (fixed)": 0.9240039350713232,
      "eval_PRM F1 Neg": 0.5333333333333333,
      "eval_PRM NPV": 0.7272727272727273,
      "eval_PRM Precision": 0.9043478260869565,
      "eval_PRM Recall": 0.9719626168224299,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.2532552182674408,
      "eval_runtime": 5.2213,
      "eval_samples_per_second": 5.746,
      "eval_steps_per_second": 0.192,
      "step": 1370
    },
    {
      "epoch": 0.5425405619311436,
      "grad_norm": 1.0064447608093403,
      "learning_rate": 5.1312300046512205e-05,
      "loss": 0.2727,
      "step": 1371
    },
    {
      "epoch": 0.5429362880886427,
      "grad_norm": 0.881911766376721,
      "learning_rate": 5.124324625633021e-05,
      "loss": 0.2399,
      "step": 1372
    },
    {
      "epoch": 0.5433320142461416,
      "grad_norm": 1.1208194862156973,
      "learning_rate": 5.117419009326747e-05,
      "loss": 0.1996,
      "step": 1373
    },
    {
      "epoch": 0.5437277404036407,
      "grad_norm": 0.808982329347689,
      "learning_rate": 5.110513168912571e-05,
      "loss": 0.1733,
      "step": 1374
    },
    {
      "epoch": 0.5441234665611397,
      "grad_norm": 1.0182816002803343,
      "learning_rate": 5.1036071175710986e-05,
      "loss": 0.159,
      "step": 1375
    },
    {
      "epoch": 0.5441234665611397,
      "eval_PRM Accuracy": 0.8888888888888888,
      "eval_PRM F1": 0.9369369369369369,
      "eval_PRM F1 AUC": 0.6965076242006887,
      "eval_PRM F1 AUC (fixed)": 0.9232661091982292,
      "eval_PRM F1 Neg": 0.5333333333333333,
      "eval_PRM NPV": 0.7272727272727273,
      "eval_PRM Precision": 0.9043478260869565,
      "eval_PRM Recall": 0.9719626168224299,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.25182291865348816,
      "eval_runtime": 5.2107,
      "eval_samples_per_second": 5.757,
      "eval_steps_per_second": 0.192,
      "step": 1375
    },
    {
      "epoch": 0.5445191927186387,
      "grad_norm": 0.7946278013386661,
      "learning_rate": 5.096700868483334e-05,
      "loss": 0.1722,
      "step": 1376
    },
    {
      "epoch": 0.5449149188761377,
      "grad_norm": 0.9282126616781327,
      "learning_rate": 5.0897944348306636e-05,
      "loss": 0.1652,
      "step": 1377
    },
    {
      "epoch": 0.5453106450336367,
      "grad_norm": 0.766439167370745,
      "learning_rate": 5.082887829794819e-05,
      "loss": 0.2082,
      "step": 1378
    },
    {
      "epoch": 0.5457063711911357,
      "grad_norm": 0.5645815405697221,
      "learning_rate": 5.075981066557866e-05,
      "loss": 0.1239,
      "step": 1379
    },
    {
      "epoch": 0.5461020973486348,
      "grad_norm": 0.5264690755334512,
      "learning_rate": 5.069074158302167e-05,
      "loss": 0.1362,
      "step": 1380
    },
    {
      "epoch": 0.5461020973486348,
      "eval_PRM Accuracy": 0.9047619047619048,
      "eval_PRM F1": 0.9464285714285714,
      "eval_PRM F1 AUC": 0.7058534185932119,
      "eval_PRM F1 AUC (fixed)": 0.9242498770290212,
      "eval_PRM F1 Neg": 0.5714285714285714,
      "eval_PRM NPV": 0.8888888888888888,
      "eval_PRM Precision": 0.905982905982906,
      "eval_PRM Recall": 0.9906542056074766,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.2627929747104645,
      "eval_runtime": 4.9822,
      "eval_samples_per_second": 6.021,
      "eval_steps_per_second": 0.201,
      "step": 1380
    },
    {
      "epoch": 0.5464978235061337,
      "grad_norm": 1.0207114387437277,
      "learning_rate": 5.062167118210367e-05,
      "loss": 0.1879,
      "step": 1381
    },
    {
      "epoch": 0.5468935496636328,
      "grad_norm": 0.6584602367568244,
      "learning_rate": 5.055259959465355e-05,
      "loss": 0.1595,
      "step": 1382
    },
    {
      "epoch": 0.5472892758211317,
      "grad_norm": 0.7447217151349483,
      "learning_rate": 5.0483526952502545e-05,
      "loss": 0.1597,
      "step": 1383
    },
    {
      "epoch": 0.5476850019786308,
      "grad_norm": 0.5469315493385634,
      "learning_rate": 5.041445338748383e-05,
      "loss": 0.1202,
      "step": 1384
    },
    {
      "epoch": 0.5480807281361298,
      "grad_norm": 0.4353024253384949,
      "learning_rate": 5.0345379031432414e-05,
      "loss": 0.0855,
      "step": 1385
    },
    {
      "epoch": 0.5480807281361298,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9422222222222222,
      "eval_PRM F1 AUC": 0.6795376291195278,
      "eval_PRM F1 AUC (fixed)": 0.9230201672405313,
      "eval_PRM F1 Neg": 0.5185185185185185,
      "eval_PRM NPV": 0.875,
      "eval_PRM Precision": 0.8983050847457628,
      "eval_PRM Recall": 0.9906542056074766,
      "eval_PRM Specificty": 0.3684210526315789,
      "eval_loss": 0.26632487773895264,
      "eval_runtime": 5.1525,
      "eval_samples_per_second": 5.822,
      "eval_steps_per_second": 0.194,
      "step": 1385
    },
    {
      "epoch": 0.5484764542936288,
      "grad_norm": 0.6205684652893204,
      "learning_rate": 5.027630401618475e-05,
      "loss": 0.1464,
      "step": 1386
    },
    {
      "epoch": 0.5488721804511278,
      "grad_norm": 0.6492098560521636,
      "learning_rate": 5.020722847357858e-05,
      "loss": 0.1266,
      "step": 1387
    },
    {
      "epoch": 0.5492679066086268,
      "grad_norm": 0.7012703485540707,
      "learning_rate": 5.0138152535452646e-05,
      "loss": 0.1743,
      "step": 1388
    },
    {
      "epoch": 0.5496636327661258,
      "grad_norm": 0.6900843534081865,
      "learning_rate": 5.006907633364646e-05,
      "loss": 0.1611,
      "step": 1389
    },
    {
      "epoch": 0.5500593589236249,
      "grad_norm": 0.5831943666977494,
      "learning_rate": 5e-05,
      "loss": 0.1307,
      "step": 1390
    },
    {
      "epoch": 0.5500593589236249,
      "eval_PRM Accuracy": 0.8888888888888888,
      "eval_PRM F1": 0.9375,
      "eval_PRM F1 AUC": 0.6748647319232661,
      "eval_PRM F1 AUC (fixed)": 0.9225282833251353,
      "eval_PRM F1 Neg": 0.5,
      "eval_PRM NPV": 0.7777777777777778,
      "eval_PRM Precision": 0.8974358974358975,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.3684210526315789,
      "eval_loss": 0.26818034052848816,
      "eval_runtime": 4.8994,
      "eval_samples_per_second": 6.123,
      "eval_steps_per_second": 0.204,
      "step": 1390
    },
    {
      "epoch": 0.5504550850811238,
      "grad_norm": 0.6964289018199051,
      "learning_rate": 4.9930923666353565e-05,
      "loss": 0.1317,
      "step": 1391
    },
    {
      "epoch": 0.5508508112386229,
      "grad_norm": 1.179270365647178,
      "learning_rate": 4.9861847464547366e-05,
      "loss": 0.2332,
      "step": 1392
    },
    {
      "epoch": 0.5512465373961218,
      "grad_norm": 0.9614488760271286,
      "learning_rate": 4.9792771526421445e-05,
      "loss": 0.2532,
      "step": 1393
    },
    {
      "epoch": 0.5516422635536209,
      "grad_norm": 0.8409619610377148,
      "learning_rate": 4.972369598381527e-05,
      "loss": 0.1307,
      "step": 1394
    },
    {
      "epoch": 0.55203798971112,
      "grad_norm": 1.0443315588015185,
      "learning_rate": 4.96546209685676e-05,
      "loss": 0.17,
      "step": 1395
    },
    {
      "epoch": 0.55203798971112,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9244958189867192,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.2539225220680237,
      "eval_runtime": 5.3734,
      "eval_samples_per_second": 5.583,
      "eval_steps_per_second": 0.186,
      "step": 1395
    },
    {
      "epoch": 0.5524337158686189,
      "grad_norm": 0.8706957188656315,
      "learning_rate": 4.958554661251616e-05,
      "loss": 0.156,
      "step": 1396
    },
    {
      "epoch": 0.552829442026118,
      "grad_norm": 0.7183250285424764,
      "learning_rate": 4.9516473047497466e-05,
      "loss": 0.1868,
      "step": 1397
    },
    {
      "epoch": 0.5532251681836169,
      "grad_norm": 0.7443674979006363,
      "learning_rate": 4.944740040534645e-05,
      "loss": 0.0992,
      "step": 1398
    },
    {
      "epoch": 0.5536208943411159,
      "grad_norm": 1.3328489845632765,
      "learning_rate": 4.9378328817896347e-05,
      "loss": 0.2437,
      "step": 1399
    },
    {
      "epoch": 0.554016620498615,
      "grad_norm": 1.3291037097114378,
      "learning_rate": 4.9309258416978346e-05,
      "loss": 0.1573,
      "step": 1400
    },
    {
      "epoch": 0.554016620498615,
      "eval_PRM Accuracy": 0.8888888888888888,
      "eval_PRM F1": 0.9363636363636364,
      "eval_PRM F1 AUC": 0.7181505164781112,
      "eval_PRM F1 AUC (fixed)": 0.9237579931136252,
      "eval_PRM F1 Neg": 0.5625,
      "eval_PRM NPV": 0.6923076923076923,
      "eval_PRM Precision": 0.911504424778761,
      "eval_PRM Recall": 0.9626168224299065,
      "eval_PRM Specificty": 0.47368421052631576,
      "eval_loss": 0.24594727158546448,
      "eval_runtime": 5.1318,
      "eval_samples_per_second": 5.846,
      "eval_steps_per_second": 0.195,
      "step": 1400
    },
    {
      "epoch": 0.5544123466561139,
      "grad_norm": 0.9587269897518773,
      "learning_rate": 4.924018933442135e-05,
      "loss": 0.1459,
      "step": 1401
    },
    {
      "epoch": 0.554808072813613,
      "grad_norm": 0.9755434927012591,
      "learning_rate": 4.917112170205183e-05,
      "loss": 0.1475,
      "step": 1402
    },
    {
      "epoch": 0.5552037989711119,
      "grad_norm": 1.0594524093028612,
      "learning_rate": 4.9102055651693376e-05,
      "loss": 0.1433,
      "step": 1403
    },
    {
      "epoch": 0.555599525128611,
      "grad_norm": 0.8735588297754491,
      "learning_rate": 4.9032991315166674e-05,
      "loss": 0.1328,
      "step": 1404
    },
    {
      "epoch": 0.55599525128611,
      "grad_norm": 1.0411603933175342,
      "learning_rate": 4.896392882428901e-05,
      "loss": 0.1903,
      "step": 1405
    },
    {
      "epoch": 0.55599525128611,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9411764705882353,
      "eval_PRM F1 AUC": 0.7228234136743729,
      "eval_PRM F1 AUC (fixed)": 0.9237579931136252,
      "eval_PRM F1 Neg": 0.5806451612903226,
      "eval_PRM NPV": 0.75,
      "eval_PRM Precision": 0.9122807017543859,
      "eval_PRM Recall": 0.9719626168224299,
      "eval_PRM Specificty": 0.47368421052631576,
      "eval_loss": 0.2561849057674408,
      "eval_runtime": 5.1098,
      "eval_samples_per_second": 5.871,
      "eval_steps_per_second": 0.196,
      "step": 1405
    },
    {
      "epoch": 0.556390977443609,
      "grad_norm": 0.7558160732117621,
      "learning_rate": 4.8894868310874296e-05,
      "loss": 0.1363,
      "step": 1406
    },
    {
      "epoch": 0.556786703601108,
      "grad_norm": 1.087986057357664,
      "learning_rate": 4.882580990673253e-05,
      "loss": 0.2294,
      "step": 1407
    },
    {
      "epoch": 0.557182429758607,
      "grad_norm": 0.7499505977906048,
      "learning_rate": 4.875675374366979e-05,
      "loss": 0.1465,
      "step": 1408
    },
    {
      "epoch": 0.557578155916106,
      "grad_norm": 0.5565413930327318,
      "learning_rate": 4.868769995348781e-05,
      "loss": 0.1133,
      "step": 1409
    },
    {
      "epoch": 0.5579738820736051,
      "grad_norm": 0.9112737424658822,
      "learning_rate": 4.861864866798381e-05,
      "loss": 0.2037,
      "step": 1410
    },
    {
      "epoch": 0.5579738820736051,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9235120511559273,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.272705078125,
      "eval_runtime": 4.8722,
      "eval_samples_per_second": 6.157,
      "eval_steps_per_second": 0.205,
      "step": 1410
    },
    {
      "epoch": 0.558369608231104,
      "grad_norm": 0.7005516439911487,
      "learning_rate": 4.854960001895027e-05,
      "loss": 0.1165,
      "step": 1411
    },
    {
      "epoch": 0.5587653343886031,
      "grad_norm": 1.0432235496789841,
      "learning_rate": 4.848055413817456e-05,
      "loss": 0.2053,
      "step": 1412
    },
    {
      "epoch": 0.5591610605461022,
      "grad_norm": 0.8223197092288267,
      "learning_rate": 4.841151115743887e-05,
      "loss": 0.1914,
      "step": 1413
    },
    {
      "epoch": 0.5595567867036011,
      "grad_norm": 0.8055827572405424,
      "learning_rate": 4.8342471208519726e-05,
      "loss": 0.198,
      "step": 1414
    },
    {
      "epoch": 0.5599525128611001,
      "grad_norm": 0.6110964295792504,
      "learning_rate": 4.827343442318801e-05,
      "loss": 0.1333,
      "step": 1415
    },
    {
      "epoch": 0.5599525128611001,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9183472700442695,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.2725585997104645,
      "eval_runtime": 4.9694,
      "eval_samples_per_second": 6.037,
      "eval_steps_per_second": 0.201,
      "step": 1415
    },
    {
      "epoch": 0.5603482390185991,
      "grad_norm": 0.6651429005914177,
      "learning_rate": 4.8204400933208435e-05,
      "loss": 0.141,
      "step": 1416
    },
    {
      "epoch": 0.5607439651760981,
      "grad_norm": 0.7848321270052059,
      "learning_rate": 4.8135370870339545e-05,
      "loss": 0.1782,
      "step": 1417
    },
    {
      "epoch": 0.5611396913335972,
      "grad_norm": 0.6317464845694473,
      "learning_rate": 4.806634436633323e-05,
      "loss": 0.0935,
      "step": 1418
    },
    {
      "epoch": 0.5615354174910961,
      "grad_norm": 0.7907526877083627,
      "learning_rate": 4.799732155293465e-05,
      "loss": 0.1836,
      "step": 1419
    },
    {
      "epoch": 0.5619311436485952,
      "grad_norm": 0.6640550306764561,
      "learning_rate": 4.792830256188192e-05,
      "loss": 0.1619,
      "step": 1420
    },
    {
      "epoch": 0.5619311436485952,
      "eval_PRM Accuracy": 0.8888888888888888,
      "eval_PRM F1": 0.9375,
      "eval_PRM F1 AUC": 0.6748647319232661,
      "eval_PRM F1 AUC (fixed)": 0.9163797343826857,
      "eval_PRM F1 Neg": 0.5,
      "eval_PRM NPV": 0.7777777777777778,
      "eval_PRM Precision": 0.8974358974358975,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.3684210526315789,
      "eval_loss": 0.2805826961994171,
      "eval_runtime": 4.9121,
      "eval_samples_per_second": 6.107,
      "eval_steps_per_second": 0.204,
      "step": 1420
    },
    {
      "epoch": 0.5623268698060941,
      "grad_norm": 0.7439904275633714,
      "learning_rate": 4.7859287524905845e-05,
      "loss": 0.1455,
      "step": 1421
    },
    {
      "epoch": 0.5627225959635932,
      "grad_norm": 0.6548890015560155,
      "learning_rate": 4.77902765737297e-05,
      "loss": 0.1152,
      "step": 1422
    },
    {
      "epoch": 0.5631183221210923,
      "grad_norm": 0.7932007123944221,
      "learning_rate": 4.772126984006892e-05,
      "loss": 0.159,
      "step": 1423
    },
    {
      "epoch": 0.5635140482785912,
      "grad_norm": 0.5210131523643514,
      "learning_rate": 4.7652267455630956e-05,
      "loss": 0.1193,
      "step": 1424
    },
    {
      "epoch": 0.5639097744360902,
      "grad_norm": 0.6840097801346285,
      "learning_rate": 4.75832695521149e-05,
      "loss": 0.177,
      "step": 1425
    },
    {
      "epoch": 0.5639097744360902,
      "eval_PRM Accuracy": 0.8888888888888888,
      "eval_PRM F1": 0.9380530973451328,
      "eval_PRM F1 AUC": 0.6532218396458436,
      "eval_PRM F1 AUC (fixed)": 0.913674372848008,
      "eval_PRM F1 Neg": 0.46153846153846156,
      "eval_PRM NPV": 0.8571428571428571,
      "eval_PRM Precision": 0.8907563025210085,
      "eval_PRM Recall": 0.9906542056074766,
      "eval_PRM Specificty": 0.3157894736842105,
      "eval_loss": 0.29594725370407104,
      "eval_runtime": 4.7302,
      "eval_samples_per_second": 6.342,
      "eval_steps_per_second": 0.211,
      "step": 1425
    },
    {
      "epoch": 0.5643055005935892,
      "grad_norm": 0.50796355910448,
      "learning_rate": 4.751427626121132e-05,
      "loss": 0.1065,
      "step": 1426
    },
    {
      "epoch": 0.5647012267510882,
      "grad_norm": 0.9355132053293667,
      "learning_rate": 4.744528771460198e-05,
      "loss": 0.2641,
      "step": 1427
    },
    {
      "epoch": 0.5650969529085873,
      "grad_norm": 0.8288009372478395,
      "learning_rate": 4.737630404395956e-05,
      "loss": 0.1592,
      "step": 1428
    },
    {
      "epoch": 0.5654926790660862,
      "grad_norm": 1.0511139835432355,
      "learning_rate": 4.730732538094749e-05,
      "loss": 0.2287,
      "step": 1429
    },
    {
      "epoch": 0.5658884052235853,
      "grad_norm": 0.9805332296050199,
      "learning_rate": 4.7238351857219587e-05,
      "loss": 0.2655,
      "step": 1430
    },
    {
      "epoch": 0.5658884052235853,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9158878504672897,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.2774088680744171,
      "eval_runtime": 4.9705,
      "eval_samples_per_second": 6.036,
      "eval_steps_per_second": 0.201,
      "step": 1430
    },
    {
      "epoch": 0.5662841313810842,
      "grad_norm": 0.8497233652275535,
      "learning_rate": 4.716938360441989e-05,
      "loss": 0.1792,
      "step": 1431
    },
    {
      "epoch": 0.5666798575385833,
      "grad_norm": 0.9016653471227217,
      "learning_rate": 4.710042075418236e-05,
      "loss": 0.2109,
      "step": 1432
    },
    {
      "epoch": 0.5670755836960824,
      "grad_norm": 0.9136907075796074,
      "learning_rate": 4.7031463438130676e-05,
      "loss": 0.1663,
      "step": 1433
    },
    {
      "epoch": 0.5674713098535813,
      "grad_norm": 0.6335276718866922,
      "learning_rate": 4.69625117878779e-05,
      "loss": 0.1477,
      "step": 1434
    },
    {
      "epoch": 0.5678670360110804,
      "grad_norm": 0.8634019863960144,
      "learning_rate": 4.6893565935026354e-05,
      "loss": 0.1851,
      "step": 1435
    },
    {
      "epoch": 0.5678670360110804,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9411764705882353,
      "eval_PRM F1 AUC": 0.7228234136743729,
      "eval_PRM F1 AUC (fixed)": 0.911460895228726,
      "eval_PRM F1 Neg": 0.5806451612903226,
      "eval_PRM NPV": 0.75,
      "eval_PRM Precision": 0.9122807017543859,
      "eval_PRM Recall": 0.9719626168224299,
      "eval_PRM Specificty": 0.47368421052631576,
      "eval_loss": 0.2605631649494171,
      "eval_runtime": 5.0125,
      "eval_samples_per_second": 5.985,
      "eval_steps_per_second": 0.2,
      "step": 1435
    },
    {
      "epoch": 0.5682627621685793,
      "grad_norm": 0.8476459302256096,
      "learning_rate": 4.682462601116724e-05,
      "loss": 0.1887,
      "step": 1436
    },
    {
      "epoch": 0.5686584883260783,
      "grad_norm": 0.6678185625512246,
      "learning_rate": 4.6755692147880414e-05,
      "loss": 0.1752,
      "step": 1437
    },
    {
      "epoch": 0.5690542144835774,
      "grad_norm": 0.9778488158413574,
      "learning_rate": 4.668676447673429e-05,
      "loss": 0.1681,
      "step": 1438
    },
    {
      "epoch": 0.5694499406410763,
      "grad_norm": 0.7171701498142868,
      "learning_rate": 4.661784312928531e-05,
      "loss": 0.1687,
      "step": 1439
    },
    {
      "epoch": 0.5698456667985754,
      "grad_norm": 0.9745514696984362,
      "learning_rate": 4.654892823707799e-05,
      "loss": 0.1449,
      "step": 1440
    },
    {
      "epoch": 0.5698456667985754,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9094933595671422,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.27249348163604736,
      "eval_runtime": 4.9637,
      "eval_samples_per_second": 6.044,
      "eval_steps_per_second": 0.201,
      "step": 1440
    },
    {
      "epoch": 0.5702413929560743,
      "grad_norm": 0.5247733913400936,
      "learning_rate": 4.648001993164438e-05,
      "loss": 0.1247,
      "step": 1441
    },
    {
      "epoch": 0.5706371191135734,
      "grad_norm": 0.6351217634295374,
      "learning_rate": 4.64111183445041e-05,
      "loss": 0.1404,
      "step": 1442
    },
    {
      "epoch": 0.5710328452710725,
      "grad_norm": 0.912387304851361,
      "learning_rate": 4.634222360716382e-05,
      "loss": 0.2193,
      "step": 1443
    },
    {
      "epoch": 0.5714285714285714,
      "grad_norm": 0.669024290034336,
      "learning_rate": 4.6273335851117264e-05,
      "loss": 0.119,
      "step": 1444
    },
    {
      "epoch": 0.5718242975860705,
      "grad_norm": 0.6756909251503679,
      "learning_rate": 4.620445520784472e-05,
      "loss": 0.1421,
      "step": 1445
    },
    {
      "epoch": 0.5718242975860705,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.912444663059518,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.28372395038604736,
      "eval_runtime": 5.161,
      "eval_samples_per_second": 5.813,
      "eval_steps_per_second": 0.194,
      "step": 1445
    },
    {
      "epoch": 0.5722200237435694,
      "grad_norm": 0.6495760805922457,
      "learning_rate": 4.613558180881294e-05,
      "loss": 0.1447,
      "step": 1446
    },
    {
      "epoch": 0.5726157499010684,
      "grad_norm": 1.3342175120483484,
      "learning_rate": 4.606671578547488e-05,
      "loss": 0.2212,
      "step": 1447
    },
    {
      "epoch": 0.5730114760585675,
      "grad_norm": 1.4066531024943802,
      "learning_rate": 4.599785726926936e-05,
      "loss": 0.155,
      "step": 1448
    },
    {
      "epoch": 0.5734072022160664,
      "grad_norm": 0.8988606186444456,
      "learning_rate": 4.5929006391620963e-05,
      "loss": 0.1747,
      "step": 1449
    },
    {
      "epoch": 0.5738029283735655,
      "grad_norm": 1.042090192375632,
      "learning_rate": 4.586016328393956e-05,
      "loss": 0.1892,
      "step": 1450
    },
    {
      "epoch": 0.5738029283735655,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9092474176094442,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.27527669072151184,
      "eval_runtime": 5.2233,
      "eval_samples_per_second": 5.743,
      "eval_steps_per_second": 0.191,
      "step": 1450
    },
    {
      "epoch": 0.5741986545310646,
      "grad_norm": 0.8340287374857285,
      "learning_rate": 4.5791328077620344e-05,
      "loss": 0.2,
      "step": 1451
    },
    {
      "epoch": 0.5745943806885635,
      "grad_norm": 0.5712425283337229,
      "learning_rate": 4.572250090404328e-05,
      "loss": 0.108,
      "step": 1452
    },
    {
      "epoch": 0.5749901068460626,
      "grad_norm": 0.7860448870357001,
      "learning_rate": 4.565368189457313e-05,
      "loss": 0.1804,
      "step": 1453
    },
    {
      "epoch": 0.5753858330035615,
      "grad_norm": 0.7547134700304947,
      "learning_rate": 4.558487118055898e-05,
      "loss": 0.1593,
      "step": 1454
    },
    {
      "epoch": 0.5757815591610606,
      "grad_norm": 0.7786650935872544,
      "learning_rate": 4.551606889333412e-05,
      "loss": 0.1643,
      "step": 1455
    },
    {
      "epoch": 0.5757815591610606,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9129365469749139,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.27094727754592896,
      "eval_runtime": 5.077,
      "eval_samples_per_second": 5.909,
      "eval_steps_per_second": 0.197,
      "step": 1455
    },
    {
      "epoch": 0.5761772853185596,
      "grad_norm": 1.0707203810407482,
      "learning_rate": 4.5447275164215774e-05,
      "loss": 0.2864,
      "step": 1456
    },
    {
      "epoch": 0.5765730114760586,
      "grad_norm": 1.011582014402992,
      "learning_rate": 4.5378490124504796e-05,
      "loss": 0.2171,
      "step": 1457
    },
    {
      "epoch": 0.5769687376335576,
      "grad_norm": 0.5118060660813321,
      "learning_rate": 4.5309713905485485e-05,
      "loss": 0.1271,
      "step": 1458
    },
    {
      "epoch": 0.5773644637910565,
      "grad_norm": 0.6603286494289827,
      "learning_rate": 4.5240946638425275e-05,
      "loss": 0.0937,
      "step": 1459
    },
    {
      "epoch": 0.5777601899485556,
      "grad_norm": 0.8449186302396923,
      "learning_rate": 4.517218845457456e-05,
      "loss": 0.1659,
      "step": 1460
    },
    {
      "epoch": 0.5777601899485556,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9411764705882353,
      "eval_PRM F1 AUC": 0.7228234136743729,
      "eval_PRM F1 AUC (fixed)": 0.9134284308903099,
      "eval_PRM F1 Neg": 0.5806451612903226,
      "eval_PRM NPV": 0.75,
      "eval_PRM Precision": 0.9122807017543859,
      "eval_PRM Recall": 0.9719626168224299,
      "eval_PRM Specificty": 0.47368421052631576,
      "eval_loss": 0.2576334774494171,
      "eval_runtime": 4.6594,
      "eval_samples_per_second": 6.439,
      "eval_steps_per_second": 0.215,
      "step": 1460
    },
    {
      "epoch": 0.5781559161060547,
      "grad_norm": 0.9444435138941004,
      "learning_rate": 4.510343948516633e-05,
      "loss": 0.1728,
      "step": 1461
    },
    {
      "epoch": 0.5785516422635536,
      "grad_norm": 0.9202527050846183,
      "learning_rate": 4.503469986141606e-05,
      "loss": 0.1559,
      "step": 1462
    },
    {
      "epoch": 0.5789473684210527,
      "grad_norm": 1.06639477686937,
      "learning_rate": 4.496596971452135e-05,
      "loss": 0.1827,
      "step": 1463
    },
    {
      "epoch": 0.5793430945785516,
      "grad_norm": 0.9217869641675038,
      "learning_rate": 4.489724917566169e-05,
      "loss": 0.2029,
      "step": 1464
    },
    {
      "epoch": 0.5797388207360507,
      "grad_norm": 0.7137614982060287,
      "learning_rate": 4.48285383759983e-05,
      "loss": 0.1434,
      "step": 1465
    },
    {
      "epoch": 0.5797388207360507,
      "eval_PRM Accuracy": 0.9047619047619048,
      "eval_PRM F1": 0.9464285714285714,
      "eval_PRM F1 AUC": 0.7058534185932119,
      "eval_PRM F1 AUC (fixed)": 0.9136743728480079,
      "eval_PRM F1 Neg": 0.5714285714285714,
      "eval_PRM NPV": 0.8888888888888888,
      "eval_PRM Precision": 0.905982905982906,
      "eval_PRM Recall": 0.9906542056074766,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.26907551288604736,
      "eval_runtime": 4.813,
      "eval_samples_per_second": 6.233,
      "eval_steps_per_second": 0.208,
      "step": 1465
    },
    {
      "epoch": 0.5801345468935497,
      "grad_norm": 0.6495924860334613,
      "learning_rate": 4.475983744667374e-05,
      "loss": 0.1239,
      "step": 1466
    },
    {
      "epoch": 0.5805302730510487,
      "grad_norm": 1.034444912832059,
      "learning_rate": 4.4691146518811775e-05,
      "loss": 0.2245,
      "step": 1467
    },
    {
      "epoch": 0.5809259992085477,
      "grad_norm": 0.6767046143862024,
      "learning_rate": 4.462246572351706e-05,
      "loss": 0.1222,
      "step": 1468
    },
    {
      "epoch": 0.5813217253660466,
      "grad_norm": 1.072372080188341,
      "learning_rate": 4.4553795191874924e-05,
      "loss": 0.1389,
      "step": 1469
    },
    {
      "epoch": 0.5817174515235457,
      "grad_norm": 1.1485374040001943,
      "learning_rate": 4.44851350549511e-05,
      "loss": 0.1903,
      "step": 1470
    },
    {
      "epoch": 0.5817174515235457,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9422222222222222,
      "eval_PRM F1 AUC": 0.6795376291195278,
      "eval_PRM F1 AUC (fixed)": 0.9171175602557796,
      "eval_PRM F1 Neg": 0.5185185185185185,
      "eval_PRM NPV": 0.875,
      "eval_PRM Precision": 0.8983050847457628,
      "eval_PRM Recall": 0.9906542056074766,
      "eval_PRM Specificty": 0.3684210526315789,
      "eval_loss": 0.27096354961395264,
      "eval_runtime": 5.1197,
      "eval_samples_per_second": 5.86,
      "eval_steps_per_second": 0.195,
      "step": 1470
    },
    {
      "epoch": 0.5821131776810448,
      "grad_norm": 0.7567450455374135,
      "learning_rate": 4.441648544379149e-05,
      "loss": 0.103,
      "step": 1471
    },
    {
      "epoch": 0.5825089038385437,
      "grad_norm": 0.8594857474655224,
      "learning_rate": 4.434784648942191e-05,
      "loss": 0.1841,
      "step": 1472
    },
    {
      "epoch": 0.5829046299960428,
      "grad_norm": 0.9155088266130493,
      "learning_rate": 4.4279218322847764e-05,
      "loss": 0.1701,
      "step": 1473
    },
    {
      "epoch": 0.5833003561535417,
      "grad_norm": 1.1203155731027175,
      "learning_rate": 4.421060107505401e-05,
      "loss": 0.1956,
      "step": 1474
    },
    {
      "epoch": 0.5836960823110408,
      "grad_norm": 0.8844476319633571,
      "learning_rate": 4.4141994877004614e-05,
      "loss": 0.1675,
      "step": 1475
    },
    {
      "epoch": 0.5836960823110408,
      "eval_PRM Accuracy": 0.9047619047619048,
      "eval_PRM F1": 0.9464285714285714,
      "eval_PRM F1 AUC": 0.7058534185932119,
      "eval_PRM F1 AUC (fixed)": 0.9146581406787997,
      "eval_PRM F1 Neg": 0.5714285714285714,
      "eval_PRM NPV": 0.8888888888888888,
      "eval_PRM Precision": 0.905982905982906,
      "eval_PRM Recall": 0.9906542056074766,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.2571614682674408,
      "eval_runtime": 5.1197,
      "eval_samples_per_second": 5.86,
      "eval_steps_per_second": 0.195,
      "step": 1475
    },
    {
      "epoch": 0.5840918084685398,
      "grad_norm": 0.5634285985467712,
      "learning_rate": 4.407339985964259e-05,
      "loss": 0.1058,
      "step": 1476
    },
    {
      "epoch": 0.5844875346260388,
      "grad_norm": 0.6202564881147808,
      "learning_rate": 4.400481615388948e-05,
      "loss": 0.1305,
      "step": 1477
    },
    {
      "epoch": 0.5848832607835378,
      "grad_norm": 0.6819460695044784,
      "learning_rate": 4.393624389064535e-05,
      "loss": 0.1317,
      "step": 1478
    },
    {
      "epoch": 0.5852789869410367,
      "grad_norm": 0.7076236758791421,
      "learning_rate": 4.3867683200788334e-05,
      "loss": 0.1334,
      "step": 1479
    },
    {
      "epoch": 0.5856747130985358,
      "grad_norm": 0.8389500782567075,
      "learning_rate": 4.379913421517458e-05,
      "loss": 0.1912,
      "step": 1480
    },
    {
      "epoch": 0.5856747130985358,
      "eval_PRM Accuracy": 0.9047619047619048,
      "eval_PRM F1": 0.9464285714285714,
      "eval_PRM F1 AUC": 0.7058534185932119,
      "eval_PRM F1 AUC (fixed)": 0.9171175602557796,
      "eval_PRM F1 Neg": 0.5714285714285714,
      "eval_PRM NPV": 0.8888888888888888,
      "eval_PRM Precision": 0.905982905982906,
      "eval_PRM Recall": 0.9906542056074766,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.24899087846279144,
      "eval_runtime": 5.1697,
      "eval_samples_per_second": 5.803,
      "eval_steps_per_second": 0.193,
      "step": 1480
    },
    {
      "epoch": 0.5860704392560349,
      "grad_norm": 0.6436330738749902,
      "learning_rate": 4.373059706463778e-05,
      "loss": 0.1321,
      "step": 1481
    },
    {
      "epoch": 0.5864661654135338,
      "grad_norm": 0.6924271461607078,
      "learning_rate": 4.3662071879989106e-05,
      "loss": 0.1403,
      "step": 1482
    },
    {
      "epoch": 0.5868618915710329,
      "grad_norm": 0.636498167300724,
      "learning_rate": 4.359355879201691e-05,
      "loss": 0.1228,
      "step": 1483
    },
    {
      "epoch": 0.5872576177285319,
      "grad_norm": 0.6873484766774083,
      "learning_rate": 4.352505793148639e-05,
      "loss": 0.1482,
      "step": 1484
    },
    {
      "epoch": 0.5876533438860309,
      "grad_norm": 0.7277647544954131,
      "learning_rate": 4.345656942913947e-05,
      "loss": 0.1338,
      "step": 1485
    },
    {
      "epoch": 0.5876533438860309,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9203148057058533,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.24363607168197632,
      "eval_runtime": 5.0491,
      "eval_samples_per_second": 5.942,
      "eval_steps_per_second": 0.198,
      "step": 1485
    },
    {
      "epoch": 0.5880490700435299,
      "grad_norm": 0.7693659732023566,
      "learning_rate": 4.338809341569444e-05,
      "loss": 0.1827,
      "step": 1486
    },
    {
      "epoch": 0.5884447962010289,
      "grad_norm": 0.5139220236767955,
      "learning_rate": 4.331963002184581e-05,
      "loss": 0.0884,
      "step": 1487
    },
    {
      "epoch": 0.5888405223585279,
      "grad_norm": 0.9802724030952206,
      "learning_rate": 4.325117937826392e-05,
      "loss": 0.173,
      "step": 1488
    },
    {
      "epoch": 0.589236248516027,
      "grad_norm": 0.9771942738582647,
      "learning_rate": 4.318274161559487e-05,
      "loss": 0.1293,
      "step": 1489
    },
    {
      "epoch": 0.5896319746735259,
      "grad_norm": 0.6788382452535272,
      "learning_rate": 4.311431686446009e-05,
      "loss": 0.175,
      "step": 1490
    },
    {
      "epoch": 0.5896319746735259,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.925971470732907,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.25011393427848816,
      "eval_runtime": 4.9535,
      "eval_samples_per_second": 6.056,
      "eval_steps_per_second": 0.202,
      "step": 1490
    },
    {
      "epoch": 0.590027700831025,
      "grad_norm": 0.836904014287633,
      "learning_rate": 4.304590525545622e-05,
      "loss": 0.1417,
      "step": 1491
    },
    {
      "epoch": 0.5904234269885239,
      "grad_norm": 0.5901206021280098,
      "learning_rate": 4.297750691915482e-05,
      "loss": 0.1158,
      "step": 1492
    },
    {
      "epoch": 0.590819153146023,
      "grad_norm": 0.7479267042211478,
      "learning_rate": 4.29091219861021e-05,
      "loss": 0.1242,
      "step": 1493
    },
    {
      "epoch": 0.591214879303522,
      "grad_norm": 0.830915379528522,
      "learning_rate": 4.2840750586818715e-05,
      "loss": 0.1339,
      "step": 1494
    },
    {
      "epoch": 0.591610605461021,
      "grad_norm": 0.977506335489616,
      "learning_rate": 4.2772392851799434e-05,
      "loss": 0.148,
      "step": 1495
    },
    {
      "epoch": 0.591610605461021,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.927693064436793,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.2668701112270355,
      "eval_runtime": 5.0937,
      "eval_samples_per_second": 5.89,
      "eval_steps_per_second": 0.196,
      "step": 1495
    },
    {
      "epoch": 0.59200633161852,
      "grad_norm": 0.9644103285471061,
      "learning_rate": 4.270404891151302e-05,
      "loss": 0.16,
      "step": 1496
    },
    {
      "epoch": 0.592402057776019,
      "grad_norm": 0.7609870060261933,
      "learning_rate": 4.263571889640184e-05,
      "loss": 0.1088,
      "step": 1497
    },
    {
      "epoch": 0.592797783933518,
      "grad_norm": 0.7930472471362493,
      "learning_rate": 4.256740293688175e-05,
      "loss": 0.1352,
      "step": 1498
    },
    {
      "epoch": 0.5931935100910171,
      "grad_norm": 0.7699152694294746,
      "learning_rate": 4.249910116334171e-05,
      "loss": 0.1485,
      "step": 1499
    },
    {
      "epoch": 0.593589236248516,
      "grad_norm": 0.8323189521119218,
      "learning_rate": 4.2430813706143636e-05,
      "loss": 0.1234,
      "step": 1500
    },
    {
      "epoch": 0.593589236248516,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9335956714215445,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.26510417461395264,
      "eval_runtime": 5.212,
      "eval_samples_per_second": 5.756,
      "eval_steps_per_second": 0.192,
      "step": 1500
    },
    {
      "epoch": 0.5939849624060151,
      "grad_norm": 0.9657162594443577,
      "learning_rate": 4.236254069562213e-05,
      "loss": 0.1377,
      "step": 1501
    },
    {
      "epoch": 0.594380688563514,
      "grad_norm": 0.8362494293065121,
      "learning_rate": 4.2294282262084215e-05,
      "loss": 0.1602,
      "step": 1502
    },
    {
      "epoch": 0.5947764147210131,
      "grad_norm": 1.09110093143604,
      "learning_rate": 4.2226038535809084e-05,
      "loss": 0.185,
      "step": 1503
    },
    {
      "epoch": 0.5951721408785121,
      "grad_norm": 0.9265133499115938,
      "learning_rate": 4.2157809647047855e-05,
      "loss": 0.1589,
      "step": 1504
    },
    {
      "epoch": 0.5955678670360111,
      "grad_norm": 0.9512366675900349,
      "learning_rate": 4.208959572602336e-05,
      "loss": 0.1688,
      "step": 1505
    },
    {
      "epoch": 0.5955678670360111,
      "eval_PRM Accuracy": 0.8888888888888888,
      "eval_PRM F1": 0.9363636363636364,
      "eval_PRM F1 AUC": 0.7181505164781112,
      "eval_PRM F1 AUC (fixed)": 0.9323659616330546,
      "eval_PRM F1 Neg": 0.5625,
      "eval_PRM NPV": 0.6923076923076923,
      "eval_PRM Precision": 0.911504424778761,
      "eval_PRM Recall": 0.9626168224299065,
      "eval_PRM Specificty": 0.47368421052631576,
      "eval_loss": 0.24205729365348816,
      "eval_runtime": 5.0521,
      "eval_samples_per_second": 5.938,
      "eval_steps_per_second": 0.198,
      "step": 1505
    },
    {
      "epoch": 0.5959635931935101,
      "grad_norm": 0.8347863770857155,
      "learning_rate": 4.2021396902929796e-05,
      "loss": 0.1074,
      "step": 1506
    },
    {
      "epoch": 0.596359319351009,
      "grad_norm": 0.9174571416952996,
      "learning_rate": 4.195321330793264e-05,
      "loss": 0.1365,
      "step": 1507
    },
    {
      "epoch": 0.5967550455085081,
      "grad_norm": 0.961727521958764,
      "learning_rate": 4.1885045071168216e-05,
      "loss": 0.1801,
      "step": 1508
    },
    {
      "epoch": 0.5971507716660072,
      "grad_norm": 1.5596445200441809,
      "learning_rate": 4.1816892322743555e-05,
      "loss": 0.2624,
      "step": 1509
    },
    {
      "epoch": 0.5975464978235061,
      "grad_norm": 1.2274557057587892,
      "learning_rate": 4.1748755192736194e-05,
      "loss": 0.2332,
      "step": 1510
    },
    {
      "epoch": 0.5975464978235061,
      "eval_PRM Accuracy": 0.8888888888888888,
      "eval_PRM F1": 0.9363636363636364,
      "eval_PRM F1 AUC": 0.7181505164781112,
      "eval_PRM F1 AUC (fixed)": 0.926709296606001,
      "eval_PRM F1 Neg": 0.5625,
      "eval_PRM NPV": 0.6923076923076923,
      "eval_PRM Precision": 0.911504424778761,
      "eval_PRM Recall": 0.9626168224299065,
      "eval_PRM Specificty": 0.47368421052631576,
      "eval_loss": 0.23916015028953552,
      "eval_runtime": 4.9892,
      "eval_samples_per_second": 6.013,
      "eval_steps_per_second": 0.2,
      "step": 1510
    },
    {
      "epoch": 0.5979422239810052,
      "grad_norm": 0.983222930061282,
      "learning_rate": 4.168063381119375e-05,
      "loss": 0.1546,
      "step": 1511
    },
    {
      "epoch": 0.5983379501385041,
      "grad_norm": 0.865980475019263,
      "learning_rate": 4.1612528308133895e-05,
      "loss": 0.1249,
      "step": 1512
    },
    {
      "epoch": 0.5987336762960032,
      "grad_norm": 1.128120691530934,
      "learning_rate": 4.154443881354388e-05,
      "loss": 0.1437,
      "step": 1513
    },
    {
      "epoch": 0.5991294024535022,
      "grad_norm": 0.8569085474043757,
      "learning_rate": 4.147636545738053e-05,
      "loss": 0.1311,
      "step": 1514
    },
    {
      "epoch": 0.5995251286110012,
      "grad_norm": 0.6451534056624597,
      "learning_rate": 4.1408308369569734e-05,
      "loss": 0.1258,
      "step": 1515
    },
    {
      "epoch": 0.5995251286110012,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9212985735366453,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.24664713442325592,
      "eval_runtime": 4.8704,
      "eval_samples_per_second": 6.16,
      "eval_steps_per_second": 0.205,
      "step": 1515
    },
    {
      "epoch": 0.5999208547685002,
      "grad_norm": 0.6635836615247669,
      "learning_rate": 4.134026768000646e-05,
      "loss": 0.1482,
      "step": 1516
    },
    {
      "epoch": 0.6003165809259992,
      "grad_norm": 0.6741555214082703,
      "learning_rate": 4.1272243518554274e-05,
      "loss": 0.1292,
      "step": 1517
    },
    {
      "epoch": 0.6007123070834982,
      "grad_norm": 0.7726731361403109,
      "learning_rate": 4.120423601504523e-05,
      "loss": 0.1346,
      "step": 1518
    },
    {
      "epoch": 0.6011080332409973,
      "grad_norm": 0.6613645124365088,
      "learning_rate": 4.113624529927963e-05,
      "loss": 0.1275,
      "step": 1519
    },
    {
      "epoch": 0.6015037593984962,
      "grad_norm": 0.42239658530944607,
      "learning_rate": 4.106827150102567e-05,
      "loss": 0.1068,
      "step": 1520
    },
    {
      "epoch": 0.6015037593984962,
      "eval_PRM Accuracy": 0.8888888888888888,
      "eval_PRM F1": 0.9375,
      "eval_PRM F1 AUC": 0.6748647319232661,
      "eval_PRM F1 AUC (fixed)": 0.9188391539596655,
      "eval_PRM F1 Neg": 0.5,
      "eval_PRM NPV": 0.7777777777777778,
      "eval_PRM Precision": 0.8974358974358975,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.3684210526315789,
      "eval_loss": 0.26232096552848816,
      "eval_runtime": 4.9996,
      "eval_samples_per_second": 6.001,
      "eval_steps_per_second": 0.2,
      "step": 1520
    },
    {
      "epoch": 0.6018994855559953,
      "grad_norm": 0.6899929167806808,
      "learning_rate": 4.1000314750019316e-05,
      "loss": 0.1451,
      "step": 1521
    },
    {
      "epoch": 0.6022952117134943,
      "grad_norm": 1.062108164021016,
      "learning_rate": 4.093237517596394e-05,
      "loss": 0.2167,
      "step": 1522
    },
    {
      "epoch": 0.6026909378709933,
      "grad_norm": 0.8301629620237868,
      "learning_rate": 4.086445290853018e-05,
      "loss": 0.1325,
      "step": 1523
    },
    {
      "epoch": 0.6030866640284923,
      "grad_norm": 0.6187463221157489,
      "learning_rate": 4.079654807735559e-05,
      "loss": 0.101,
      "step": 1524
    },
    {
      "epoch": 0.6034823901859913,
      "grad_norm": 0.9698425710140381,
      "learning_rate": 4.0728660812044536e-05,
      "loss": 0.1506,
      "step": 1525
    },
    {
      "epoch": 0.6034823901859913,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9166256763403837,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.2593750059604645,
      "eval_runtime": 5.1491,
      "eval_samples_per_second": 5.826,
      "eval_steps_per_second": 0.194,
      "step": 1525
    },
    {
      "epoch": 0.6038781163434903,
      "grad_norm": 0.8549946171718574,
      "learning_rate": 4.066079124216775e-05,
      "loss": 0.1602,
      "step": 1526
    },
    {
      "epoch": 0.6042738425009894,
      "grad_norm": 0.925905508424247,
      "learning_rate": 4.059293949726222e-05,
      "loss": 0.158,
      "step": 1527
    },
    {
      "epoch": 0.6046695686584883,
      "grad_norm": 0.7941244976149617,
      "learning_rate": 4.052510570683098e-05,
      "loss": 0.1513,
      "step": 1528
    },
    {
      "epoch": 0.6050652948159874,
      "grad_norm": 0.825734372484861,
      "learning_rate": 4.045729000034271e-05,
      "loss": 0.1223,
      "step": 1529
    },
    {
      "epoch": 0.6054610209734863,
      "grad_norm": 0.9572101878662733,
      "learning_rate": 4.0389492507231635e-05,
      "loss": 0.1524,
      "step": 1530
    },
    {
      "epoch": 0.6054610209734863,
      "eval_PRM Accuracy": 0.8888888888888888,
      "eval_PRM F1": 0.9369369369369369,
      "eval_PRM F1 AUC": 0.6965076242006887,
      "eval_PRM F1 AUC (fixed)": 0.9193310378750615,
      "eval_PRM F1 Neg": 0.5333333333333333,
      "eval_PRM NPV": 0.7272727272727273,
      "eval_PRM Precision": 0.9043478260869565,
      "eval_PRM Recall": 0.9719626168224299,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.24744465947151184,
      "eval_runtime": 5.2706,
      "eval_samples_per_second": 5.692,
      "eval_steps_per_second": 0.19,
      "step": 1530
    },
    {
      "epoch": 0.6058567471309854,
      "grad_norm": 0.9916901134432358,
      "learning_rate": 4.0321713356897176e-05,
      "loss": 0.145,
      "step": 1531
    },
    {
      "epoch": 0.6062524732884844,
      "grad_norm": 0.8939529906807856,
      "learning_rate": 4.025395267870376e-05,
      "loss": 0.1485,
      "step": 1532
    },
    {
      "epoch": 0.6066481994459834,
      "grad_norm": 1.0505352006228945,
      "learning_rate": 4.0186210601980575e-05,
      "loss": 0.151,
      "step": 1533
    },
    {
      "epoch": 0.6070439256034824,
      "grad_norm": 1.8656784491331981,
      "learning_rate": 4.011848725602129e-05,
      "loss": 0.1519,
      "step": 1534
    },
    {
      "epoch": 0.6074396517609814,
      "grad_norm": 0.8564592800051644,
      "learning_rate": 4.005078277008381e-05,
      "loss": 0.1328,
      "step": 1535
    },
    {
      "epoch": 0.6074396517609814,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9411764705882353,
      "eval_PRM F1 AUC": 0.7228234136743729,
      "eval_PRM F1 AUC (fixed)": 0.9190850959173635,
      "eval_PRM F1 Neg": 0.5806451612903226,
      "eval_PRM NPV": 0.75,
      "eval_PRM Precision": 0.9122807017543859,
      "eval_PRM Recall": 0.9719626168224299,
      "eval_PRM Specificty": 0.47368421052631576,
      "eval_loss": 0.24637044966220856,
      "eval_runtime": 5.0085,
      "eval_samples_per_second": 5.99,
      "eval_steps_per_second": 0.2,
      "step": 1535
    },
    {
      "epoch": 0.6078353779184804,
      "grad_norm": 0.9204410340492978,
      "learning_rate": 3.998309727339006e-05,
      "loss": 0.1744,
      "step": 1536
    },
    {
      "epoch": 0.6082311040759795,
      "grad_norm": 0.48482758018876304,
      "learning_rate": 3.991543089512574e-05,
      "loss": 0.0687,
      "step": 1537
    },
    {
      "epoch": 0.6086268302334784,
      "grad_norm": 0.852226247973999,
      "learning_rate": 3.984778376444001e-05,
      "loss": 0.1466,
      "step": 1538
    },
    {
      "epoch": 0.6090225563909775,
      "grad_norm": 0.9733601758719616,
      "learning_rate": 3.9780156010445355e-05,
      "loss": 0.1693,
      "step": 1539
    },
    {
      "epoch": 0.6094182825484764,
      "grad_norm": 0.8396406312102926,
      "learning_rate": 3.9712547762217226e-05,
      "loss": 0.1784,
      "step": 1540
    },
    {
      "epoch": 0.6094182825484764,
      "eval_PRM Accuracy": 0.9047619047619048,
      "eval_PRM F1": 0.9459459459459459,
      "eval_PRM F1 AUC": 0.7274963108706345,
      "eval_PRM F1 AUC (fixed)": 0.9208066896212493,
      "eval_PRM F1 Neg": 0.6,
      "eval_PRM NPV": 0.8181818181818182,
      "eval_PRM Precision": 0.9130434782608695,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.47368421052631576,
      "eval_loss": 0.25515949726104736,
      "eval_runtime": 5.1802,
      "eval_samples_per_second": 5.791,
      "eval_steps_per_second": 0.193,
      "step": 1540
    },
    {
      "epoch": 0.6098140087059755,
      "grad_norm": 0.9999142064520963,
      "learning_rate": 3.964495914879387e-05,
      "loss": 0.1948,
      "step": 1541
    },
    {
      "epoch": 0.6102097348634745,
      "grad_norm": 0.8574450866660717,
      "learning_rate": 3.957739029917605e-05,
      "loss": 0.1099,
      "step": 1542
    },
    {
      "epoch": 0.6106054610209735,
      "grad_norm": 0.7354426980780667,
      "learning_rate": 3.950984134232683e-05,
      "loss": 0.1091,
      "step": 1543
    },
    {
      "epoch": 0.6110011871784725,
      "grad_norm": 1.0571908132710006,
      "learning_rate": 3.9442312407171285e-05,
      "loss": 0.1914,
      "step": 1544
    },
    {
      "epoch": 0.6113969133359715,
      "grad_norm": 0.9269918154343854,
      "learning_rate": 3.9374803622596236e-05,
      "loss": 0.1852,
      "step": 1545
    },
    {
      "epoch": 0.6113969133359715,
      "eval_PRM Accuracy": 0.9047619047619048,
      "eval_PRM F1": 0.9459459459459459,
      "eval_PRM F1 AUC": 0.7274963108706345,
      "eval_PRM F1 AUC (fixed)": 0.9225282833251353,
      "eval_PRM F1 Neg": 0.6,
      "eval_PRM NPV": 0.8181818181818182,
      "eval_PRM Precision": 0.9130434782608695,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.47368421052631576,
      "eval_loss": 0.25353190302848816,
      "eval_runtime": 4.7575,
      "eval_samples_per_second": 6.306,
      "eval_steps_per_second": 0.21,
      "step": 1545
    },
    {
      "epoch": 0.6117926394934705,
      "grad_norm": 0.9111487612122596,
      "learning_rate": 3.930731511745015e-05,
      "loss": 0.1544,
      "step": 1546
    },
    {
      "epoch": 0.6121883656509696,
      "grad_norm": 0.7061738249095735,
      "learning_rate": 3.923984702054266e-05,
      "loss": 0.1204,
      "step": 1547
    },
    {
      "epoch": 0.6125840918084685,
      "grad_norm": 0.7124853702930833,
      "learning_rate": 3.917239946064459e-05,
      "loss": 0.1337,
      "step": 1548
    },
    {
      "epoch": 0.6129798179659676,
      "grad_norm": 0.5098215114351303,
      "learning_rate": 3.910497256648742e-05,
      "loss": 0.095,
      "step": 1549
    },
    {
      "epoch": 0.6133755441234665,
      "grad_norm": 0.6290031867916469,
      "learning_rate": 3.903756646676331e-05,
      "loss": 0.1415,
      "step": 1550
    },
    {
      "epoch": 0.6133755441234665,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9411764705882353,
      "eval_PRM F1 AUC": 0.7228234136743729,
      "eval_PRM F1 AUC (fixed)": 0.9220363994097392,
      "eval_PRM F1 Neg": 0.5806451612903226,
      "eval_PRM NPV": 0.75,
      "eval_PRM Precision": 0.9122807017543859,
      "eval_PRM Recall": 0.9719626168224299,
      "eval_PRM Specificty": 0.47368421052631576,
      "eval_loss": 0.24627278745174408,
      "eval_runtime": 4.9771,
      "eval_samples_per_second": 6.028,
      "eval_steps_per_second": 0.201,
      "step": 1550
    },
    {
      "epoch": 0.6137712702809656,
      "grad_norm": 1.9656948899813809,
      "learning_rate": 3.897018129012462e-05,
      "loss": 0.2101,
      "step": 1551
    },
    {
      "epoch": 0.6141669964384646,
      "grad_norm": 1.030881832929198,
      "learning_rate": 3.89028171651839e-05,
      "loss": 0.1652,
      "step": 1552
    },
    {
      "epoch": 0.6145627225959636,
      "grad_norm": 0.8262573746782119,
      "learning_rate": 3.883547422051343e-05,
      "loss": 0.1777,
      "step": 1553
    },
    {
      "epoch": 0.6149584487534626,
      "grad_norm": 0.7733978134798383,
      "learning_rate": 3.8768152584645045e-05,
      "loss": 0.1182,
      "step": 1554
    },
    {
      "epoch": 0.6153541749109616,
      "grad_norm": 0.768337290299473,
      "learning_rate": 3.870085238607002e-05,
      "loss": 0.1392,
      "step": 1555
    },
    {
      "epoch": 0.6153541749109616,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9411764705882353,
      "eval_PRM F1 AUC": 0.7228234136743729,
      "eval_PRM F1 AUC (fixed)": 0.9222823413674373,
      "eval_PRM F1 Neg": 0.5806451612903226,
      "eval_PRM NPV": 0.75,
      "eval_PRM Precision": 0.9122807017543859,
      "eval_PRM Recall": 0.9719626168224299,
      "eval_PRM Specificty": 0.47368421052631576,
      "eval_loss": 0.24786783754825592,
      "eval_runtime": 4.6683,
      "eval_samples_per_second": 6.426,
      "eval_steps_per_second": 0.214,
      "step": 1555
    },
    {
      "epoch": 0.6157499010684606,
      "grad_norm": 0.7888434873726559,
      "learning_rate": 3.86335737532386e-05,
      "loss": 0.1641,
      "step": 1556
    },
    {
      "epoch": 0.6161456272259597,
      "grad_norm": 1.7799382949674674,
      "learning_rate": 3.8566316814559955e-05,
      "loss": 0.2277,
      "step": 1557
    },
    {
      "epoch": 0.6165413533834586,
      "grad_norm": 1.6850091612061884,
      "learning_rate": 3.8499081698401786e-05,
      "loss": 0.2123,
      "step": 1558
    },
    {
      "epoch": 0.6169370795409577,
      "grad_norm": 1.1612977554282127,
      "learning_rate": 3.843186853309018e-05,
      "loss": 0.1833,
      "step": 1559
    },
    {
      "epoch": 0.6173328056984567,
      "grad_norm": 0.8620798718482171,
      "learning_rate": 3.836467744690931e-05,
      "loss": 0.1441,
      "step": 1560
    },
    {
      "epoch": 0.6173328056984567,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9227742252828334,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.2503255307674408,
      "eval_runtime": 4.9088,
      "eval_samples_per_second": 6.111,
      "eval_steps_per_second": 0.204,
      "step": 1560
    },
    {
      "epoch": 0.6177285318559557,
      "grad_norm": 0.5741179519626246,
      "learning_rate": 3.829750856810123e-05,
      "loss": 0.0854,
      "step": 1561
    },
    {
      "epoch": 0.6181242580134547,
      "grad_norm": 0.6628778408364507,
      "learning_rate": 3.823036202486559e-05,
      "loss": 0.1235,
      "step": 1562
    },
    {
      "epoch": 0.6185199841709537,
      "grad_norm": 0.6088495108954647,
      "learning_rate": 3.81632379453594e-05,
      "loss": 0.1169,
      "step": 1563
    },
    {
      "epoch": 0.6189157103284527,
      "grad_norm": 0.5917842799427135,
      "learning_rate": 3.809613645769682e-05,
      "loss": 0.0955,
      "step": 1564
    },
    {
      "epoch": 0.6193114364859518,
      "grad_norm": 0.5480350336060075,
      "learning_rate": 3.8029057689948885e-05,
      "loss": 0.1124,
      "step": 1565
    },
    {
      "epoch": 0.6193114364859518,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9235120511559272,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.25939127802848816,
      "eval_runtime": 5.1571,
      "eval_samples_per_second": 5.817,
      "eval_steps_per_second": 0.194,
      "step": 1565
    },
    {
      "epoch": 0.6197071626434507,
      "grad_norm": 0.8035889475920374,
      "learning_rate": 3.796200177014326e-05,
      "loss": 0.1421,
      "step": 1566
    },
    {
      "epoch": 0.6201028888009498,
      "grad_norm": 1.3907990943178097,
      "learning_rate": 3.789496882626399e-05,
      "loss": 0.2392,
      "step": 1567
    },
    {
      "epoch": 0.6204986149584487,
      "grad_norm": 0.7660525536909165,
      "learning_rate": 3.782795898625131e-05,
      "loss": 0.1264,
      "step": 1568
    },
    {
      "epoch": 0.6208943411159478,
      "grad_norm": 0.8500644692001158,
      "learning_rate": 3.7760972378001304e-05,
      "loss": 0.1652,
      "step": 1569
    },
    {
      "epoch": 0.6212900672734468,
      "grad_norm": 1.4035006307606788,
      "learning_rate": 3.7694009129365756e-05,
      "loss": 0.3161,
      "step": 1570
    },
    {
      "epoch": 0.6212900672734468,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9232661091982293,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.2596191465854645,
      "eval_runtime": 5.0954,
      "eval_samples_per_second": 5.888,
      "eval_steps_per_second": 0.196,
      "step": 1570
    },
    {
      "epoch": 0.6216857934309458,
      "grad_norm": 1.5819492572117542,
      "learning_rate": 3.7627069368151845e-05,
      "loss": 0.1741,
      "step": 1571
    },
    {
      "epoch": 0.6220815195884448,
      "grad_norm": 0.9237331419482251,
      "learning_rate": 3.756015322212193e-05,
      "loss": 0.2366,
      "step": 1572
    },
    {
      "epoch": 0.6224772457459438,
      "grad_norm": 0.818173417363629,
      "learning_rate": 3.749326081899329e-05,
      "loss": 0.1711,
      "step": 1573
    },
    {
      "epoch": 0.6228729719034428,
      "grad_norm": 0.9082183165469345,
      "learning_rate": 3.742639228643791e-05,
      "loss": 0.1663,
      "step": 1574
    },
    {
      "epoch": 0.6232686980609419,
      "grad_norm": 1.0668296407053135,
      "learning_rate": 3.735954775208218e-05,
      "loss": 0.1689,
      "step": 1575
    },
    {
      "epoch": 0.6232686980609419,
      "eval_PRM Accuracy": 0.8888888888888888,
      "eval_PRM F1": 0.9369369369369369,
      "eval_PRM F1 AUC": 0.6965076242006887,
      "eval_PRM F1 AUC (fixed)": 0.9215445154943434,
      "eval_PRM F1 Neg": 0.5333333333333333,
      "eval_PRM NPV": 0.7272727272727273,
      "eval_PRM Precision": 0.9043478260869565,
      "eval_PRM Recall": 0.9719626168224299,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.24464517831802368,
      "eval_runtime": 5.182,
      "eval_samples_per_second": 5.789,
      "eval_steps_per_second": 0.193,
      "step": 1575
    },
    {
      "epoch": 0.6236644242184408,
      "grad_norm": 0.6640194797498201,
      "learning_rate": 3.729272734350671e-05,
      "loss": 0.1371,
      "step": 1576
    },
    {
      "epoch": 0.6240601503759399,
      "grad_norm": 0.662503330902093,
      "learning_rate": 3.722593118824606e-05,
      "loss": 0.1749,
      "step": 1577
    },
    {
      "epoch": 0.6244558765334388,
      "grad_norm": 0.7087841794401485,
      "learning_rate": 3.715915941378849e-05,
      "loss": 0.1457,
      "step": 1578
    },
    {
      "epoch": 0.6248516026909379,
      "grad_norm": 0.8161266288908837,
      "learning_rate": 3.709241214757576e-05,
      "loss": 0.1333,
      "step": 1579
    },
    {
      "epoch": 0.6252473288484369,
      "grad_norm": 1.161517248433403,
      "learning_rate": 3.7025689517002826e-05,
      "loss": 0.1719,
      "step": 1580
    },
    {
      "epoch": 0.6252473288484369,
      "eval_PRM Accuracy": 0.8888888888888888,
      "eval_PRM F1": 0.9369369369369369,
      "eval_PRM F1 AUC": 0.6965076242006887,
      "eval_PRM F1 AUC (fixed)": 0.9247417609444171,
      "eval_PRM F1 Neg": 0.5333333333333333,
      "eval_PRM NPV": 0.7272727272727273,
      "eval_PRM Precision": 0.9043478260869565,
      "eval_PRM Recall": 0.9719626168224299,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.24296875298023224,
      "eval_runtime": 5.1606,
      "eval_samples_per_second": 5.813,
      "eval_steps_per_second": 0.194,
      "step": 1580
    },
    {
      "epoch": 0.6256430550059359,
      "grad_norm": 0.8915166708051876,
      "learning_rate": 3.695899164941757e-05,
      "loss": 0.1565,
      "step": 1581
    },
    {
      "epoch": 0.6260387811634349,
      "grad_norm": 0.7275993274960687,
      "learning_rate": 3.689231867212074e-05,
      "loss": 0.1502,
      "step": 1582
    },
    {
      "epoch": 0.6264345073209339,
      "grad_norm": 0.9882838418687355,
      "learning_rate": 3.682567071236544e-05,
      "loss": 0.2004,
      "step": 1583
    },
    {
      "epoch": 0.6268302334784329,
      "grad_norm": 0.8497961365729795,
      "learning_rate": 3.675904789735716e-05,
      "loss": 0.1872,
      "step": 1584
    },
    {
      "epoch": 0.627225959635932,
      "grad_norm": 0.6654183776206689,
      "learning_rate": 3.6692450354253246e-05,
      "loss": 0.0944,
      "step": 1585
    },
    {
      "epoch": 0.627225959635932,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.926955238563699,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.2514811158180237,
      "eval_runtime": 4.9184,
      "eval_samples_per_second": 6.1,
      "eval_steps_per_second": 0.203,
      "step": 1585
    },
    {
      "epoch": 0.6276216857934309,
      "grad_norm": 0.7505532563868831,
      "learning_rate": 3.6625878210162966e-05,
      "loss": 0.1643,
      "step": 1586
    },
    {
      "epoch": 0.62801741195093,
      "grad_norm": 0.8802668321523966,
      "learning_rate": 3.6559331592147e-05,
      "loss": 0.1757,
      "step": 1587
    },
    {
      "epoch": 0.6284131381084289,
      "grad_norm": 0.8812587374045024,
      "learning_rate": 3.649281062721733e-05,
      "loss": 0.1443,
      "step": 1588
    },
    {
      "epoch": 0.628808864265928,
      "grad_norm": 0.6010125850815679,
      "learning_rate": 3.642631544233703e-05,
      "loss": 0.1327,
      "step": 1589
    },
    {
      "epoch": 0.629204590423427,
      "grad_norm": 0.7077558998839296,
      "learning_rate": 3.63598461644199e-05,
      "loss": 0.1166,
      "step": 1590
    },
    {
      "epoch": 0.629204590423427,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9272011805213969,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.2510579526424408,
      "eval_runtime": 5.3252,
      "eval_samples_per_second": 5.634,
      "eval_steps_per_second": 0.188,
      "step": 1590
    },
    {
      "epoch": 0.629600316580926,
      "grad_norm": 0.949185022980317,
      "learning_rate": 3.6293402920330346e-05,
      "loss": 0.1688,
      "step": 1591
    },
    {
      "epoch": 0.629996042738425,
      "grad_norm": 0.6186071728252178,
      "learning_rate": 3.622698583688304e-05,
      "loss": 0.1791,
      "step": 1592
    },
    {
      "epoch": 0.630391768895924,
      "grad_norm": 0.6304083801077957,
      "learning_rate": 3.616059504084278e-05,
      "loss": 0.1244,
      "step": 1593
    },
    {
      "epoch": 0.630787495053423,
      "grad_norm": 0.7062535416501295,
      "learning_rate": 3.609423065892412e-05,
      "loss": 0.1491,
      "step": 1594
    },
    {
      "epoch": 0.6311832212109221,
      "grad_norm": 0.719747570030943,
      "learning_rate": 3.6027892817791275e-05,
      "loss": 0.1908,
      "step": 1595
    },
    {
      "epoch": 0.6311832212109221,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9242498770290212,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.2559244930744171,
      "eval_runtime": 4.871,
      "eval_samples_per_second": 6.159,
      "eval_steps_per_second": 0.205,
      "step": 1595
    },
    {
      "epoch": 0.631578947368421,
      "grad_norm": 0.5650329560203284,
      "learning_rate": 3.5961581644057744e-05,
      "loss": 0.1409,
      "step": 1596
    },
    {
      "epoch": 0.6319746735259201,
      "grad_norm": 1.0987871409912282,
      "learning_rate": 3.589529726428615e-05,
      "loss": 0.1846,
      "step": 1597
    },
    {
      "epoch": 0.6323703996834191,
      "grad_norm": 0.8728289476461197,
      "learning_rate": 3.5829039804988e-05,
      "loss": 0.1761,
      "step": 1598
    },
    {
      "epoch": 0.6327661258409181,
      "grad_norm": 0.8223132300459512,
      "learning_rate": 3.576280939262336e-05,
      "loss": 0.1184,
      "step": 1599
    },
    {
      "epoch": 0.6331618519984171,
      "grad_norm": 0.7745023100215828,
      "learning_rate": 3.5696606153600754e-05,
      "loss": 0.1541,
      "step": 1600
    },
    {
      "epoch": 0.6331618519984171,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9249877029021151,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.24694010615348816,
      "eval_runtime": 4.9804,
      "eval_samples_per_second": 6.024,
      "eval_steps_per_second": 0.201,
      "step": 1600
    },
    {
      "epoch": 0.6335575781559161,
      "grad_norm": 0.618237545390577,
      "learning_rate": 3.563043021427677e-05,
      "loss": 0.103,
      "step": 1601
    },
    {
      "epoch": 0.6339533043134151,
      "grad_norm": 0.760799307925824,
      "learning_rate": 3.556428170095593e-05,
      "loss": 0.1676,
      "step": 1602
    },
    {
      "epoch": 0.6343490304709142,
      "grad_norm": 0.9756936827721346,
      "learning_rate": 3.5498160739890404e-05,
      "loss": 0.176,
      "step": 1603
    },
    {
      "epoch": 0.6347447566284131,
      "grad_norm": 0.7292806996057538,
      "learning_rate": 3.5432067457279775e-05,
      "loss": 0.1387,
      "step": 1604
    },
    {
      "epoch": 0.6351404827859122,
      "grad_norm": 0.7944460992734116,
      "learning_rate": 3.5366001979270805e-05,
      "loss": 0.1501,
      "step": 1605
    },
    {
      "epoch": 0.6351404827859122,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9244958189867191,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.24435220658779144,
      "eval_runtime": 5.2837,
      "eval_samples_per_second": 5.678,
      "eval_steps_per_second": 0.189,
      "step": 1605
    },
    {
      "epoch": 0.6355362089434111,
      "grad_norm": 0.9692676907945897,
      "learning_rate": 3.529996443195714e-05,
      "loss": 0.1493,
      "step": 1606
    },
    {
      "epoch": 0.6359319351009102,
      "grad_norm": 0.6194671017107998,
      "learning_rate": 3.523395494137921e-05,
      "loss": 0.1512,
      "step": 1607
    },
    {
      "epoch": 0.6363276612584092,
      "grad_norm": 0.7070073655609458,
      "learning_rate": 3.5167973633523804e-05,
      "loss": 0.1004,
      "step": 1608
    },
    {
      "epoch": 0.6367233874159082,
      "grad_norm": 0.8890423393855097,
      "learning_rate": 3.510202063432397e-05,
      "loss": 0.1157,
      "step": 1609
    },
    {
      "epoch": 0.6371191135734072,
      "grad_norm": 0.8414660654124163,
      "learning_rate": 3.503609606965872e-05,
      "loss": 0.1984,
      "step": 1610
    },
    {
      "epoch": 0.6371191135734072,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9240039350713231,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.25208333134651184,
      "eval_runtime": 4.9037,
      "eval_samples_per_second": 6.118,
      "eval_steps_per_second": 0.204,
      "step": 1610
    },
    {
      "epoch": 0.6375148397309062,
      "grad_norm": 1.2827861941865317,
      "learning_rate": 3.497020006535278e-05,
      "loss": 0.2692,
      "step": 1611
    },
    {
      "epoch": 0.6379105658884052,
      "grad_norm": 0.9299602619133756,
      "learning_rate": 3.4904332747176373e-05,
      "loss": 0.1413,
      "step": 1612
    },
    {
      "epoch": 0.6383062920459043,
      "grad_norm": 0.9253425223596159,
      "learning_rate": 3.483849424084499e-05,
      "loss": 0.1619,
      "step": 1613
    },
    {
      "epoch": 0.6387020182034032,
      "grad_norm": 1.653405104104079,
      "learning_rate": 3.477268467201911e-05,
      "loss": 0.2736,
      "step": 1614
    },
    {
      "epoch": 0.6390977443609023,
      "grad_norm": 1.153863457706101,
      "learning_rate": 3.470690416630395e-05,
      "loss": 0.1732,
      "step": 1615
    },
    {
      "epoch": 0.6390977443609023,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9230201672405312,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.25019532442092896,
      "eval_runtime": 5.1233,
      "eval_samples_per_second": 5.856,
      "eval_steps_per_second": 0.195,
      "step": 1615
    },
    {
      "epoch": 0.6394934705184012,
      "grad_norm": 0.9692509270708172,
      "learning_rate": 3.4641152849249346e-05,
      "loss": 0.1829,
      "step": 1616
    },
    {
      "epoch": 0.6398891966759003,
      "grad_norm": 0.8552004445245153,
      "learning_rate": 3.457543084634932e-05,
      "loss": 0.1357,
      "step": 1617
    },
    {
      "epoch": 0.6402849228333993,
      "grad_norm": 0.8423305232122711,
      "learning_rate": 3.4509738283042046e-05,
      "loss": 0.1286,
      "step": 1618
    },
    {
      "epoch": 0.6406806489908983,
      "grad_norm": 1.0444060825385282,
      "learning_rate": 3.44440752847094e-05,
      "loss": 0.1623,
      "step": 1619
    },
    {
      "epoch": 0.6410763751483973,
      "grad_norm": 1.0332518155086392,
      "learning_rate": 3.437844197667696e-05,
      "loss": 0.1784,
      "step": 1620
    },
    {
      "epoch": 0.6410763751483973,
      "eval_PRM Accuracy": 0.9047619047619048,
      "eval_PRM F1": 0.9459459459459459,
      "eval_PRM F1 AUC": 0.7274963108706345,
      "eval_PRM F1 AUC (fixed)": 0.9200688637481553,
      "eval_PRM F1 Neg": 0.6,
      "eval_PRM NPV": 0.8181818181818182,
      "eval_PRM Precision": 0.9130434782608695,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.47368421052631576,
      "eval_loss": 0.24469400942325592,
      "eval_runtime": 5.1993,
      "eval_samples_per_second": 5.77,
      "eval_steps_per_second": 0.192,
      "step": 1620
    },
    {
      "epoch": 0.6414721013058963,
      "grad_norm": 0.5817713233669263,
      "learning_rate": 3.431283848421347e-05,
      "loss": 0.0917,
      "step": 1621
    },
    {
      "epoch": 0.6418678274633953,
      "grad_norm": 0.6813452022057115,
      "learning_rate": 3.4247264932530964e-05,
      "loss": 0.1173,
      "step": 1622
    },
    {
      "epoch": 0.6422635536208944,
      "grad_norm": 1.0405022904962467,
      "learning_rate": 3.418172144678416e-05,
      "loss": 0.1629,
      "step": 1623
    },
    {
      "epoch": 0.6426592797783933,
      "grad_norm": 1.014668822575636,
      "learning_rate": 3.4116208152070473e-05,
      "loss": 0.1649,
      "step": 1624
    },
    {
      "epoch": 0.6430550059358924,
      "grad_norm": 0.6041421991100032,
      "learning_rate": 3.4050725173429695e-05,
      "loss": 0.12,
      "step": 1625
    },
    {
      "epoch": 0.6430550059358924,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9188391539596655,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.2541666626930237,
      "eval_runtime": 4.9228,
      "eval_samples_per_second": 6.094,
      "eval_steps_per_second": 0.203,
      "step": 1625
    },
    {
      "epoch": 0.6434507320933913,
      "grad_norm": 0.6045546392702725,
      "learning_rate": 3.3985272635843725e-05,
      "loss": 0.1427,
      "step": 1626
    },
    {
      "epoch": 0.6438464582508904,
      "grad_norm": 0.8189187484681875,
      "learning_rate": 3.39198506642364e-05,
      "loss": 0.1963,
      "step": 1627
    },
    {
      "epoch": 0.6442421844083894,
      "grad_norm": 0.7773420731827327,
      "learning_rate": 3.3854459383473174e-05,
      "loss": 0.1648,
      "step": 1628
    },
    {
      "epoch": 0.6446379105658884,
      "grad_norm": 0.8912136359019834,
      "learning_rate": 3.378909891836098e-05,
      "loss": 0.1877,
      "step": 1629
    },
    {
      "epoch": 0.6450336367233874,
      "grad_norm": 0.7129557098247674,
      "learning_rate": 3.372376939364787e-05,
      "loss": 0.1346,
      "step": 1630
    },
    {
      "epoch": 0.6450336367233874,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9422222222222222,
      "eval_PRM F1 AUC": 0.6795376291195278,
      "eval_PRM F1 AUC (fixed)": 0.9149040826364978,
      "eval_PRM F1 Neg": 0.5185185185185185,
      "eval_PRM NPV": 0.875,
      "eval_PRM Precision": 0.8983050847457628,
      "eval_PRM Recall": 0.9906542056074766,
      "eval_PRM Specificty": 0.3684210526315789,
      "eval_loss": 0.267578125,
      "eval_runtime": 4.7079,
      "eval_samples_per_second": 6.372,
      "eval_steps_per_second": 0.212,
      "step": 1630
    },
    {
      "epoch": 0.6454293628808865,
      "grad_norm": 0.8812031044552366,
      "learning_rate": 3.36584709340229e-05,
      "loss": 0.2078,
      "step": 1631
    },
    {
      "epoch": 0.6458250890383854,
      "grad_norm": 0.6538499392032535,
      "learning_rate": 3.3593203664115807e-05,
      "loss": 0.1142,
      "step": 1632
    },
    {
      "epoch": 0.6462208151958845,
      "grad_norm": 0.5942485118920212,
      "learning_rate": 3.352796770849679e-05,
      "loss": 0.1433,
      "step": 1633
    },
    {
      "epoch": 0.6466165413533834,
      "grad_norm": 0.6144719088984341,
      "learning_rate": 3.3462763191676305e-05,
      "loss": 0.1222,
      "step": 1634
    },
    {
      "epoch": 0.6470122675108825,
      "grad_norm": 0.8327972854183461,
      "learning_rate": 3.339759023810478e-05,
      "loss": 0.1813,
      "step": 1635
    },
    {
      "epoch": 0.6470122675108825,
      "eval_PRM Accuracy": 0.8888888888888888,
      "eval_PRM F1": 0.9380530973451328,
      "eval_PRM F1 AUC": 0.6532218396458436,
      "eval_PRM F1 AUC (fixed)": 0.9141662567634039,
      "eval_PRM F1 Neg": 0.46153846153846156,
      "eval_PRM NPV": 0.8571428571428571,
      "eval_PRM Precision": 0.8907563025210085,
      "eval_PRM Recall": 0.9906542056074766,
      "eval_PRM Specificty": 0.3157894736842105,
      "eval_loss": 0.2788737118244171,
      "eval_runtime": 5.1964,
      "eval_samples_per_second": 5.773,
      "eval_steps_per_second": 0.192,
      "step": 1635
    },
    {
      "epoch": 0.6474079936683815,
      "grad_norm": 0.6650308669705507,
      "learning_rate": 3.3332448972172434e-05,
      "loss": 0.1413,
      "step": 1636
    },
    {
      "epoch": 0.6478037198258805,
      "grad_norm": 1.0840000308928164,
      "learning_rate": 3.3267339518208976e-05,
      "loss": 0.1912,
      "step": 1637
    },
    {
      "epoch": 0.6481994459833795,
      "grad_norm": 0.787224505067319,
      "learning_rate": 3.32022620004834e-05,
      "loss": 0.1697,
      "step": 1638
    },
    {
      "epoch": 0.6485951721408785,
      "grad_norm": 0.7664209452418511,
      "learning_rate": 3.313721654320375e-05,
      "loss": 0.1555,
      "step": 1639
    },
    {
      "epoch": 0.6489908982983775,
      "grad_norm": 0.7503675496270195,
      "learning_rate": 3.30722032705169e-05,
      "loss": 0.1867,
      "step": 1640
    },
    {
      "epoch": 0.6489908982983775,
      "eval_PRM Accuracy": 0.9047619047619048,
      "eval_PRM F1": 0.9464285714285714,
      "eval_PRM F1 AUC": 0.7058534185932119,
      "eval_PRM F1 AUC (fixed)": 0.9136743728480079,
      "eval_PRM F1 Neg": 0.5714285714285714,
      "eval_PRM NPV": 0.8888888888888888,
      "eval_PRM Precision": 0.905982905982906,
      "eval_PRM Recall": 0.9906542056074766,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.2711588442325592,
      "eval_runtime": 5.0687,
      "eval_samples_per_second": 5.919,
      "eval_steps_per_second": 0.197,
      "step": 1640
    },
    {
      "epoch": 0.6493866244558766,
      "grad_norm": 0.8394211100379778,
      "learning_rate": 3.300722230650827e-05,
      "loss": 0.1712,
      "step": 1641
    },
    {
      "epoch": 0.6497823506133755,
      "grad_norm": 0.7218193739223439,
      "learning_rate": 3.294227377520161e-05,
      "loss": 0.1561,
      "step": 1642
    },
    {
      "epoch": 0.6501780767708746,
      "grad_norm": 0.7586954516919955,
      "learning_rate": 3.2877357800558804e-05,
      "loss": 0.123,
      "step": 1643
    },
    {
      "epoch": 0.6505738029283735,
      "grad_norm": 1.186653808526199,
      "learning_rate": 3.281247450647956e-05,
      "loss": 0.2529,
      "step": 1644
    },
    {
      "epoch": 0.6509695290858726,
      "grad_norm": 0.857976019099429,
      "learning_rate": 3.274762401680124e-05,
      "loss": 0.1382,
      "step": 1645
    },
    {
      "epoch": 0.6509695290858726,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9151500245941958,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.25605469942092896,
      "eval_runtime": 5.1437,
      "eval_samples_per_second": 5.832,
      "eval_steps_per_second": 0.194,
      "step": 1645
    },
    {
      "epoch": 0.6513652552433716,
      "grad_norm": 0.7480119711488435,
      "learning_rate": 3.268280645529857e-05,
      "loss": 0.1258,
      "step": 1646
    },
    {
      "epoch": 0.6517609814008706,
      "grad_norm": 1.0612554222669093,
      "learning_rate": 3.2618021945683455e-05,
      "loss": 0.1888,
      "step": 1647
    },
    {
      "epoch": 0.6521567075583696,
      "grad_norm": 1.218558267882022,
      "learning_rate": 3.255327061160467e-05,
      "loss": 0.2162,
      "step": 1648
    },
    {
      "epoch": 0.6525524337158686,
      "grad_norm": 0.9550403767591656,
      "learning_rate": 3.2488552576647746e-05,
      "loss": 0.1804,
      "step": 1649
    },
    {
      "epoch": 0.6529481598733676,
      "grad_norm": 0.8075922213497077,
      "learning_rate": 3.24238679643346e-05,
      "loss": 0.1678,
      "step": 1650
    },
    {
      "epoch": 0.6529481598733676,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9411764705882353,
      "eval_PRM F1 AUC": 0.7228234136743729,
      "eval_PRM F1 AUC (fixed)": 0.9153959665518936,
      "eval_PRM F1 Neg": 0.5806451612903226,
      "eval_PRM NPV": 0.75,
      "eval_PRM Precision": 0.9122807017543859,
      "eval_PRM Recall": 0.9719626168224299,
      "eval_PRM Specificty": 0.47368421052631576,
      "eval_loss": 0.2516927123069763,
      "eval_runtime": 5.3396,
      "eval_samples_per_second": 5.618,
      "eval_steps_per_second": 0.187,
      "step": 1650
    },
    {
      "epoch": 0.6533438860308667,
      "grad_norm": 1.1759267766414092,
      "learning_rate": 3.235921689812334e-05,
      "loss": 0.2797,
      "step": 1651
    },
    {
      "epoch": 0.6537396121883656,
      "grad_norm": 0.9986688104463152,
      "learning_rate": 3.229459950140814e-05,
      "loss": 0.2049,
      "step": 1652
    },
    {
      "epoch": 0.6541353383458647,
      "grad_norm": 1.1000952924471674,
      "learning_rate": 3.22300158975188e-05,
      "loss": 0.1558,
      "step": 1653
    },
    {
      "epoch": 0.6545310645033636,
      "grad_norm": 0.9164561969127896,
      "learning_rate": 3.2165466209720705e-05,
      "loss": 0.1374,
      "step": 1654
    },
    {
      "epoch": 0.6549267906608627,
      "grad_norm": 0.7527894698896691,
      "learning_rate": 3.210095056121446e-05,
      "loss": 0.1584,
      "step": 1655
    },
    {
      "epoch": 0.6549267906608627,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9149040826364978,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.2583984434604645,
      "eval_runtime": 5.0874,
      "eval_samples_per_second": 5.897,
      "eval_steps_per_second": 0.197,
      "step": 1655
    },
    {
      "epoch": 0.6553225168183617,
      "grad_norm": 0.8170571291583314,
      "learning_rate": 3.203646907513575e-05,
      "loss": 0.1419,
      "step": 1656
    },
    {
      "epoch": 0.6557182429758607,
      "grad_norm": 0.5939985859431135,
      "learning_rate": 3.197202187455498e-05,
      "loss": 0.1275,
      "step": 1657
    },
    {
      "epoch": 0.6561139691333597,
      "grad_norm": 0.8939795146819883,
      "learning_rate": 3.1907609082477215e-05,
      "loss": 0.1489,
      "step": 1658
    },
    {
      "epoch": 0.6565096952908587,
      "grad_norm": 1.044116217452929,
      "learning_rate": 3.184323082184176e-05,
      "loss": 0.1904,
      "step": 1659
    },
    {
      "epoch": 0.6569054214483577,
      "grad_norm": 0.6819371138436929,
      "learning_rate": 3.1778887215522044e-05,
      "loss": 0.1223,
      "step": 1660
    },
    {
      "epoch": 0.6569054214483577,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9422222222222222,
      "eval_PRM F1 AUC": 0.6795376291195278,
      "eval_PRM F1 AUC (fixed)": 0.9107230693556321,
      "eval_PRM F1 Neg": 0.5185185185185185,
      "eval_PRM NPV": 0.875,
      "eval_PRM Precision": 0.8983050847457628,
      "eval_PRM Recall": 0.9906542056074766,
      "eval_PRM Specificty": 0.3684210526315789,
      "eval_loss": 0.27617186307907104,
      "eval_runtime": 4.8275,
      "eval_samples_per_second": 6.214,
      "eval_steps_per_second": 0.207,
      "step": 1660
    },
    {
      "epoch": 0.6573011476058568,
      "grad_norm": 0.911899996276412,
      "learning_rate": 3.171457838632539e-05,
      "loss": 0.2473,
      "step": 1661
    },
    {
      "epoch": 0.6576968737633557,
      "grad_norm": 0.9490085475168927,
      "learning_rate": 3.165030445699269e-05,
      "loss": 0.1727,
      "step": 1662
    },
    {
      "epoch": 0.6580925999208548,
      "grad_norm": 0.8297175488221508,
      "learning_rate": 3.158606555019826e-05,
      "loss": 0.1458,
      "step": 1663
    },
    {
      "epoch": 0.6584883260783537,
      "grad_norm": 0.6378900203189766,
      "learning_rate": 3.1521861788549544e-05,
      "loss": 0.1568,
      "step": 1664
    },
    {
      "epoch": 0.6588840522358528,
      "grad_norm": 0.930754440456445,
      "learning_rate": 3.145769329458695e-05,
      "loss": 0.1168,
      "step": 1665
    },
    {
      "epoch": 0.6588840522358528,
      "eval_PRM Accuracy": 0.9047619047619048,
      "eval_PRM F1": 0.9464285714285714,
      "eval_PRM F1 AUC": 0.7058534185932119,
      "eval_PRM F1 AUC (fixed)": 0.9117068371864241,
      "eval_PRM F1 Neg": 0.5714285714285714,
      "eval_PRM NPV": 0.8888888888888888,
      "eval_PRM Precision": 0.905982905982906,
      "eval_PRM Recall": 0.9906542056074766,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.26559245586395264,
      "eval_runtime": 5.3569,
      "eval_samples_per_second": 5.6,
      "eval_steps_per_second": 0.187,
      "step": 1665
    },
    {
      "epoch": 0.6592797783933518,
      "grad_norm": 0.637509194436935,
      "learning_rate": 3.139356019078352e-05,
      "loss": 0.1563,
      "step": 1666
    },
    {
      "epoch": 0.6596755045508508,
      "grad_norm": 0.8084565735950835,
      "learning_rate": 3.13294625995448e-05,
      "loss": 0.1052,
      "step": 1667
    },
    {
      "epoch": 0.6600712307083498,
      "grad_norm": 0.6143164711886805,
      "learning_rate": 3.126540064320853e-05,
      "loss": 0.121,
      "step": 1668
    },
    {
      "epoch": 0.6604669568658489,
      "grad_norm": 0.9445339911627708,
      "learning_rate": 3.120137444404442e-05,
      "loss": 0.2158,
      "step": 1669
    },
    {
      "epoch": 0.6608626830233478,
      "grad_norm": 0.5929224640844797,
      "learning_rate": 3.1137384124253974e-05,
      "loss": 0.1536,
      "step": 1670
    },
    {
      "epoch": 0.6608626830233478,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.91096901131333,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.2622233033180237,
      "eval_runtime": 5.0428,
      "eval_samples_per_second": 5.949,
      "eval_steps_per_second": 0.198,
      "step": 1670
    },
    {
      "epoch": 0.6612584091808469,
      "grad_norm": 0.7841873576164498,
      "learning_rate": 3.107342980597019e-05,
      "loss": 0.1054,
      "step": 1671
    },
    {
      "epoch": 0.6616541353383458,
      "grad_norm": 0.70820798306573,
      "learning_rate": 3.1009511611257354e-05,
      "loss": 0.1291,
      "step": 1672
    },
    {
      "epoch": 0.6620498614958449,
      "grad_norm": 0.846618243132325,
      "learning_rate": 3.0945629662110805e-05,
      "loss": 0.14,
      "step": 1673
    },
    {
      "epoch": 0.6624455876533439,
      "grad_norm": 0.6411665715812614,
      "learning_rate": 3.088178408045672e-05,
      "loss": 0.1403,
      "step": 1674
    },
    {
      "epoch": 0.6628413138108429,
      "grad_norm": 0.7870212728656223,
      "learning_rate": 3.0817974988151846e-05,
      "loss": 0.1485,
      "step": 1675
    },
    {
      "epoch": 0.6628413138108429,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.911460895228726,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.2684732973575592,
      "eval_runtime": 5.0451,
      "eval_samples_per_second": 5.946,
      "eval_steps_per_second": 0.198,
      "step": 1675
    },
    {
      "epoch": 0.6632370399683419,
      "grad_norm": 0.6406667783607488,
      "learning_rate": 3.075420250698331e-05,
      "loss": 0.1117,
      "step": 1676
    },
    {
      "epoch": 0.6636327661258409,
      "grad_norm": 1.0351196536488345,
      "learning_rate": 3.069046675866831e-05,
      "loss": 0.0966,
      "step": 1677
    },
    {
      "epoch": 0.6640284922833399,
      "grad_norm": 1.2950462103388103,
      "learning_rate": 3.062676786485399e-05,
      "loss": 0.2094,
      "step": 1678
    },
    {
      "epoch": 0.664424218440839,
      "grad_norm": 0.8393167820905282,
      "learning_rate": 3.0563105947117147e-05,
      "loss": 0.184,
      "step": 1679
    },
    {
      "epoch": 0.6648199445983379,
      "grad_norm": 1.2406977099661578,
      "learning_rate": 3.0499481126963947e-05,
      "loss": 0.2081,
      "step": 1680
    },
    {
      "epoch": 0.6648199445983379,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.909985243482538,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.2720377743244171,
      "eval_runtime": 5.0297,
      "eval_samples_per_second": 5.965,
      "eval_steps_per_second": 0.199,
      "step": 1680
    },
    {
      "epoch": 0.665215670755837,
      "grad_norm": 0.6777976461634672,
      "learning_rate": 3.0435893525829846e-05,
      "loss": 0.1085,
      "step": 1681
    },
    {
      "epoch": 0.6656113969133359,
      "grad_norm": 0.9067404798371308,
      "learning_rate": 3.0372343265079172e-05,
      "loss": 0.1577,
      "step": 1682
    },
    {
      "epoch": 0.666007123070835,
      "grad_norm": 0.7576189170624374,
      "learning_rate": 3.030883046600505e-05,
      "loss": 0.1117,
      "step": 1683
    },
    {
      "epoch": 0.666402849228334,
      "grad_norm": 0.8810048708597216,
      "learning_rate": 3.0245355249829055e-05,
      "loss": 0.1513,
      "step": 1684
    },
    {
      "epoch": 0.666798575385833,
      "grad_norm": 0.8002553330291301,
      "learning_rate": 3.018191773770108e-05,
      "loss": 0.1716,
      "step": 1685
    },
    {
      "epoch": 0.666798575385833,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9131824889326119,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.27470701932907104,
      "eval_runtime": 5.0441,
      "eval_samples_per_second": 5.947,
      "eval_steps_per_second": 0.198,
      "step": 1685
    },
    {
      "epoch": 0.667194301543332,
      "grad_norm": 0.9254054059954825,
      "learning_rate": 3.011851805069904e-05,
      "loss": 0.1289,
      "step": 1686
    },
    {
      "epoch": 0.667590027700831,
      "grad_norm": 1.0204182931034567,
      "learning_rate": 3.005515630982858e-05,
      "loss": 0.2235,
      "step": 1687
    },
    {
      "epoch": 0.66798575385833,
      "grad_norm": 0.5220672838150713,
      "learning_rate": 2.9991832636023065e-05,
      "loss": 0.1091,
      "step": 1688
    },
    {
      "epoch": 0.6683814800158291,
      "grad_norm": 0.661670531815788,
      "learning_rate": 2.9928547150143066e-05,
      "loss": 0.1099,
      "step": 1689
    },
    {
      "epoch": 0.668777206173328,
      "grad_norm": 0.6758057965250595,
      "learning_rate": 2.9865299972976386e-05,
      "loss": 0.1459,
      "step": 1690
    },
    {
      "epoch": 0.668777206173328,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9141662567634039,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.27224934101104736,
      "eval_runtime": 5.0298,
      "eval_samples_per_second": 5.964,
      "eval_steps_per_second": 0.199,
      "step": 1690
    },
    {
      "epoch": 0.6691729323308271,
      "grad_norm": 0.7757461204124356,
      "learning_rate": 2.980209122523759e-05,
      "loss": 0.1301,
      "step": 1691
    },
    {
      "epoch": 0.669568658488326,
      "grad_norm": 0.9072837541279577,
      "learning_rate": 2.9738921027568024e-05,
      "loss": 0.1435,
      "step": 1692
    },
    {
      "epoch": 0.6699643846458251,
      "grad_norm": 0.9186257200102006,
      "learning_rate": 2.9675789500535328e-05,
      "loss": 0.2005,
      "step": 1693
    },
    {
      "epoch": 0.6703601108033241,
      "grad_norm": 0.8616896608654981,
      "learning_rate": 2.9612696764633465e-05,
      "loss": 0.161,
      "step": 1694
    },
    {
      "epoch": 0.6707558369608231,
      "grad_norm": 1.4746013378335232,
      "learning_rate": 2.9549642940282246e-05,
      "loss": 0.2358,
      "step": 1695
    },
    {
      "epoch": 0.6707558369608231,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9156419085095917,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.2687825560569763,
      "eval_runtime": 5.075,
      "eval_samples_per_second": 5.911,
      "eval_steps_per_second": 0.197,
      "step": 1695
    },
    {
      "epoch": 0.6711515631183221,
      "grad_norm": 0.9346803410165002,
      "learning_rate": 2.9486628147827273e-05,
      "loss": 0.1895,
      "step": 1696
    },
    {
      "epoch": 0.6715472892758211,
      "grad_norm": 0.6807351729965638,
      "learning_rate": 2.9423652507539655e-05,
      "loss": 0.0793,
      "step": 1697
    },
    {
      "epoch": 0.6719430154333201,
      "grad_norm": 0.7741113757155825,
      "learning_rate": 2.936071613961574e-05,
      "loss": 0.1577,
      "step": 1698
    },
    {
      "epoch": 0.6723387415908192,
      "grad_norm": 0.8259174954237541,
      "learning_rate": 2.9297819164176965e-05,
      "loss": 0.1474,
      "step": 1699
    },
    {
      "epoch": 0.6727344677483181,
      "grad_norm": 0.8518234679048291,
      "learning_rate": 2.923496170126953e-05,
      "loss": 0.1675,
      "step": 1700
    },
    {
      "epoch": 0.6727344677483181,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9168716182980816,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.26111653447151184,
      "eval_runtime": 4.8927,
      "eval_samples_per_second": 6.132,
      "eval_steps_per_second": 0.204,
      "step": 1700
    },
    {
      "epoch": 0.6731301939058172,
      "grad_norm": 0.865580062633823,
      "learning_rate": 2.9172143870864276e-05,
      "loss": 0.1897,
      "step": 1701
    },
    {
      "epoch": 0.6735259200633161,
      "grad_norm": 0.8018772939654772,
      "learning_rate": 2.9109365792856358e-05,
      "loss": 0.1721,
      "step": 1702
    },
    {
      "epoch": 0.6739216462208152,
      "grad_norm": 0.9636502380662514,
      "learning_rate": 2.9046627587065066e-05,
      "loss": 0.2111,
      "step": 1703
    },
    {
      "epoch": 0.6743173723783142,
      "grad_norm": 0.628670255003344,
      "learning_rate": 2.898392937323364e-05,
      "loss": 0.1376,
      "step": 1704
    },
    {
      "epoch": 0.6747130985358132,
      "grad_norm": 0.5608547698521629,
      "learning_rate": 2.8921271271028894e-05,
      "loss": 0.1496,
      "step": 1705
    },
    {
      "epoch": 0.6747130985358132,
      "eval_PRM Accuracy": 0.9047619047619048,
      "eval_PRM F1": 0.9464285714285714,
      "eval_PRM F1 AUC": 0.7058534185932119,
      "eval_PRM F1 AUC (fixed)": 0.9163797343826857,
      "eval_PRM F1 Neg": 0.5714285714285714,
      "eval_PRM NPV": 0.8888888888888888,
      "eval_PRM Precision": 0.905982905982906,
      "eval_PRM Recall": 0.9906542056074766,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.2604817748069763,
      "eval_runtime": 5.0308,
      "eval_samples_per_second": 5.963,
      "eval_steps_per_second": 0.199,
      "step": 1705
    },
    {
      "epoch": 0.6751088246933122,
      "grad_norm": 0.7759052906685112,
      "learning_rate": 2.8858653400041176e-05,
      "loss": 0.1208,
      "step": 1706
    },
    {
      "epoch": 0.6755045508508113,
      "grad_norm": 0.8981517224223691,
      "learning_rate": 2.8796075879783956e-05,
      "loss": 0.1704,
      "step": 1707
    },
    {
      "epoch": 0.6759002770083102,
      "grad_norm": 0.7044619459285052,
      "learning_rate": 2.873353882969382e-05,
      "loss": 0.1491,
      "step": 1708
    },
    {
      "epoch": 0.6762960031658093,
      "grad_norm": 0.683527284408525,
      "learning_rate": 2.8671042369129984e-05,
      "loss": 0.1486,
      "step": 1709
    },
    {
      "epoch": 0.6766917293233082,
      "grad_norm": 0.636836018681674,
      "learning_rate": 2.860858661737428e-05,
      "loss": 0.1064,
      "step": 1710
    },
    {
      "epoch": 0.6766917293233082,
      "eval_PRM Accuracy": 0.9047619047619048,
      "eval_PRM F1": 0.9464285714285714,
      "eval_PRM F1 AUC": 0.7058534185932119,
      "eval_PRM F1 AUC (fixed)": 0.9168716182980816,
      "eval_PRM F1 Neg": 0.5714285714285714,
      "eval_PRM NPV": 0.8888888888888888,
      "eval_PRM Precision": 0.905982905982906,
      "eval_PRM Recall": 0.9906542056074766,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.26404622197151184,
      "eval_runtime": 5.008,
      "eval_samples_per_second": 5.99,
      "eval_steps_per_second": 0.2,
      "step": 1710
    },
    {
      "epoch": 0.6770874554808073,
      "grad_norm": 0.720224307516925,
      "learning_rate": 2.8546171693630746e-05,
      "loss": 0.1262,
      "step": 1711
    },
    {
      "epoch": 0.6774831816383063,
      "grad_norm": 1.0411798156204677,
      "learning_rate": 2.8483797717025646e-05,
      "loss": 0.2297,
      "step": 1712
    },
    {
      "epoch": 0.6778789077958053,
      "grad_norm": 0.6920023428702126,
      "learning_rate": 2.8421464806606955e-05,
      "loss": 0.1539,
      "step": 1713
    },
    {
      "epoch": 0.6782746339533043,
      "grad_norm": 0.5545419355580743,
      "learning_rate": 2.8359173081344305e-05,
      "loss": 0.116,
      "step": 1714
    },
    {
      "epoch": 0.6786703601108033,
      "grad_norm": 0.8240655113494316,
      "learning_rate": 2.8296922660128744e-05,
      "loss": 0.1782,
      "step": 1715
    },
    {
      "epoch": 0.6786703601108033,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9161337924249877,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.26171875,
      "eval_runtime": 5.0985,
      "eval_samples_per_second": 5.884,
      "eval_steps_per_second": 0.196,
      "step": 1715
    },
    {
      "epoch": 0.6790660862683023,
      "grad_norm": 0.4868084574715076,
      "learning_rate": 2.8234713661772484e-05,
      "loss": 0.1194,
      "step": 1716
    },
    {
      "epoch": 0.6794618124258014,
      "grad_norm": 0.8951421410496175,
      "learning_rate": 2.8172546205008683e-05,
      "loss": 0.1816,
      "step": 1717
    },
    {
      "epoch": 0.6798575385833003,
      "grad_norm": 0.6162112563533695,
      "learning_rate": 2.8110420408491155e-05,
      "loss": 0.1399,
      "step": 1718
    },
    {
      "epoch": 0.6802532647407994,
      "grad_norm": 0.5854854751130102,
      "learning_rate": 2.8048336390794272e-05,
      "loss": 0.1078,
      "step": 1719
    },
    {
      "epoch": 0.6806489908982983,
      "grad_norm": 1.1250226870307858,
      "learning_rate": 2.7986294270412623e-05,
      "loss": 0.1581,
      "step": 1720
    },
    {
      "epoch": 0.6806489908982983,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9205607476635514,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.2596842348575592,
      "eval_runtime": 4.974,
      "eval_samples_per_second": 6.031,
      "eval_steps_per_second": 0.201,
      "step": 1720
    },
    {
      "epoch": 0.6810447170557974,
      "grad_norm": 0.7806054956941448,
      "learning_rate": 2.7924294165760878e-05,
      "loss": 0.156,
      "step": 1721
    },
    {
      "epoch": 0.6814404432132964,
      "grad_norm": 0.674313727078266,
      "learning_rate": 2.7862336195173434e-05,
      "loss": 0.1257,
      "step": 1722
    },
    {
      "epoch": 0.6818361693707954,
      "grad_norm": 0.6485236002241265,
      "learning_rate": 2.7800420476904337e-05,
      "loss": 0.1732,
      "step": 1723
    },
    {
      "epoch": 0.6822318955282944,
      "grad_norm": 0.681551541340875,
      "learning_rate": 2.7738547129127002e-05,
      "loss": 0.141,
      "step": 1724
    },
    {
      "epoch": 0.6826276216857934,
      "grad_norm": 0.8048197093631744,
      "learning_rate": 2.767671626993389e-05,
      "loss": 0.1887,
      "step": 1725
    },
    {
      "epoch": 0.6826276216857934,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9235120511559273,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.25377604365348816,
      "eval_runtime": 5.0154,
      "eval_samples_per_second": 5.982,
      "eval_steps_per_second": 0.199,
      "step": 1725
    },
    {
      "epoch": 0.6830233478432924,
      "grad_norm": 0.511532489871784,
      "learning_rate": 2.761492801733645e-05,
      "loss": 0.1343,
      "step": 1726
    },
    {
      "epoch": 0.6834190740007915,
      "grad_norm": 0.7276985441022853,
      "learning_rate": 2.7553182489264777e-05,
      "loss": 0.1197,
      "step": 1727
    },
    {
      "epoch": 0.6838148001582904,
      "grad_norm": 0.7133094318026139,
      "learning_rate": 2.7491479803567453e-05,
      "loss": 0.1287,
      "step": 1728
    },
    {
      "epoch": 0.6842105263157895,
      "grad_norm": 0.6382113995112451,
      "learning_rate": 2.7429820078011214e-05,
      "loss": 0.1002,
      "step": 1729
    },
    {
      "epoch": 0.6846062524732884,
      "grad_norm": 0.8298623384619986,
      "learning_rate": 2.7368203430280887e-05,
      "loss": 0.1568,
      "step": 1730
    },
    {
      "epoch": 0.6846062524732884,
      "eval_PRM Accuracy": 0.8888888888888888,
      "eval_PRM F1": 0.9369369369369369,
      "eval_PRM F1 AUC": 0.6965076242006887,
      "eval_PRM F1 AUC (fixed)": 0.9249877029021151,
      "eval_PRM F1 Neg": 0.5333333333333333,
      "eval_PRM NPV": 0.7272727272727273,
      "eval_PRM Precision": 0.9043478260869565,
      "eval_PRM Recall": 0.9719626168224299,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.25410157442092896,
      "eval_runtime": 4.7666,
      "eval_samples_per_second": 6.294,
      "eval_steps_per_second": 0.21,
      "step": 1730
    },
    {
      "epoch": 0.6850019786307875,
      "grad_norm": 0.7297730751344316,
      "learning_rate": 2.7306629977979047e-05,
      "loss": 0.1657,
      "step": 1731
    },
    {
      "epoch": 0.6853977047882865,
      "grad_norm": 0.7895304169120664,
      "learning_rate": 2.7245099838625805e-05,
      "loss": 0.1448,
      "step": 1732
    },
    {
      "epoch": 0.6857934309457855,
      "grad_norm": 0.64222364456738,
      "learning_rate": 2.718361312965862e-05,
      "loss": 0.1285,
      "step": 1733
    },
    {
      "epoch": 0.6861891571032845,
      "grad_norm": 0.854080149749247,
      "learning_rate": 2.7122169968432075e-05,
      "loss": 0.148,
      "step": 1734
    },
    {
      "epoch": 0.6865848832607835,
      "grad_norm": 0.5438629153282041,
      "learning_rate": 2.7060770472217634e-05,
      "loss": 0.0757,
      "step": 1735
    },
    {
      "epoch": 0.6865848832607835,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9244958189867192,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.26378580927848816,
      "eval_runtime": 5.249,
      "eval_samples_per_second": 5.715,
      "eval_steps_per_second": 0.191,
      "step": 1735
    },
    {
      "epoch": 0.6869806094182825,
      "grad_norm": 1.546583182361692,
      "learning_rate": 2.6999414758203378e-05,
      "loss": 0.338,
      "step": 1736
    },
    {
      "epoch": 0.6873763355757816,
      "grad_norm": 0.8952895203955109,
      "learning_rate": 2.693810294349388e-05,
      "loss": 0.1339,
      "step": 1737
    },
    {
      "epoch": 0.6877720617332805,
      "grad_norm": 0.7588467325194502,
      "learning_rate": 2.6876835145109892e-05,
      "loss": 0.1069,
      "step": 1738
    },
    {
      "epoch": 0.6881677878907796,
      "grad_norm": 0.7889025780600908,
      "learning_rate": 2.681561147998819e-05,
      "loss": 0.1769,
      "step": 1739
    },
    {
      "epoch": 0.6885635140482786,
      "grad_norm": 0.7935678416988543,
      "learning_rate": 2.6754432064981285e-05,
      "loss": 0.143,
      "step": 1740
    },
    {
      "epoch": 0.6885635140482786,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9240039350713232,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.2676757872104645,
      "eval_runtime": 4.9345,
      "eval_samples_per_second": 6.08,
      "eval_steps_per_second": 0.203,
      "step": 1740
    },
    {
      "epoch": 0.6889592402057776,
      "grad_norm": 0.7456709906093306,
      "learning_rate": 2.6693297016857188e-05,
      "loss": 0.1362,
      "step": 1741
    },
    {
      "epoch": 0.6893549663632766,
      "grad_norm": 0.7079420886195056,
      "learning_rate": 2.6632206452299363e-05,
      "loss": 0.1209,
      "step": 1742
    },
    {
      "epoch": 0.6897506925207756,
      "grad_norm": 0.9708123609399834,
      "learning_rate": 2.657116048790624e-05,
      "loss": 0.1855,
      "step": 1743
    },
    {
      "epoch": 0.6901464186782746,
      "grad_norm": 0.5300513302107794,
      "learning_rate": 2.6510159240191202e-05,
      "loss": 0.0694,
      "step": 1744
    },
    {
      "epoch": 0.6905421448357737,
      "grad_norm": 0.8486731736527329,
      "learning_rate": 2.6449202825582214e-05,
      "loss": 0.1405,
      "step": 1745
    },
    {
      "epoch": 0.6905421448357737,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.924987702902115,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.25743815302848816,
      "eval_runtime": 4.9178,
      "eval_samples_per_second": 6.1,
      "eval_steps_per_second": 0.203,
      "step": 1745
    },
    {
      "epoch": 0.6909378709932726,
      "grad_norm": 0.7293208210838185,
      "learning_rate": 2.6388291360421784e-05,
      "loss": 0.1058,
      "step": 1746
    },
    {
      "epoch": 0.6913335971507717,
      "grad_norm": 0.9367885722104417,
      "learning_rate": 2.6327424960966506e-05,
      "loss": 0.1731,
      "step": 1747
    },
    {
      "epoch": 0.6917293233082706,
      "grad_norm": 0.8230177951764005,
      "learning_rate": 2.6266603743387063e-05,
      "loss": 0.1624,
      "step": 1748
    },
    {
      "epoch": 0.6921250494657697,
      "grad_norm": 0.8599648213964229,
      "learning_rate": 2.6205827823767808e-05,
      "loss": 0.172,
      "step": 1749
    },
    {
      "epoch": 0.6925207756232687,
      "grad_norm": 1.4015577323791675,
      "learning_rate": 2.6145097318106703e-05,
      "loss": 0.1821,
      "step": 1750
    },
    {
      "epoch": 0.6925207756232687,
      "eval_PRM Accuracy": 0.8888888888888888,
      "eval_PRM F1": 0.9369369369369369,
      "eval_PRM F1 AUC": 0.6965076242006887,
      "eval_PRM F1 AUC (fixed)": 0.926709296606001,
      "eval_PRM F1 Neg": 0.5333333333333333,
      "eval_PRM NPV": 0.7272727272727273,
      "eval_PRM Precision": 0.9043478260869565,
      "eval_PRM Recall": 0.9719626168224299,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.24737955629825592,
      "eval_runtime": 5.0107,
      "eval_samples_per_second": 5.987,
      "eval_steps_per_second": 0.2,
      "step": 1750
    },
    {
      "epoch": 0.6929165017807677,
      "grad_norm": 0.8213547178755115,
      "learning_rate": 2.6084412342315047e-05,
      "loss": 0.1296,
      "step": 1751
    },
    {
      "epoch": 0.6933122279382667,
      "grad_norm": 0.895677585669731,
      "learning_rate": 2.6023773012217155e-05,
      "loss": 0.1337,
      "step": 1752
    },
    {
      "epoch": 0.6937079540957657,
      "grad_norm": 0.7500314355880874,
      "learning_rate": 2.5963179443550302e-05,
      "loss": 0.1432,
      "step": 1753
    },
    {
      "epoch": 0.6941036802532647,
      "grad_norm": 0.7663835868461323,
      "learning_rate": 2.5902631751964395e-05,
      "loss": 0.1433,
      "step": 1754
    },
    {
      "epoch": 0.6944994064107638,
      "grad_norm": 0.6415386826257503,
      "learning_rate": 2.5842130053021796e-05,
      "loss": 0.1183,
      "step": 1755
    },
    {
      "epoch": 0.6944994064107638,
      "eval_PRM Accuracy": 0.8888888888888888,
      "eval_PRM F1": 0.9369369369369369,
      "eval_PRM F1 AUC": 0.6965076242006887,
      "eval_PRM F1 AUC (fixed)": 0.925725528775209,
      "eval_PRM F1 Neg": 0.5333333333333333,
      "eval_PRM NPV": 0.7272727272727273,
      "eval_PRM Precision": 0.9043478260869565,
      "eval_PRM Recall": 0.9719626168224299,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.24721679091453552,
      "eval_runtime": 5.0417,
      "eval_samples_per_second": 5.95,
      "eval_steps_per_second": 0.198,
      "step": 1755
    },
    {
      "epoch": 0.6948951325682627,
      "grad_norm": 0.8139877789577366,
      "learning_rate": 2.5781674462197026e-05,
      "loss": 0.121,
      "step": 1756
    },
    {
      "epoch": 0.6952908587257618,
      "grad_norm": 1.0306803509659623,
      "learning_rate": 2.5721265094876667e-05,
      "loss": 0.2132,
      "step": 1757
    },
    {
      "epoch": 0.6956865848832607,
      "grad_norm": 0.8762892272087696,
      "learning_rate": 2.5660902066359084e-05,
      "loss": 0.141,
      "step": 1758
    },
    {
      "epoch": 0.6960823110407598,
      "grad_norm": 0.7660744154803971,
      "learning_rate": 2.560058549185412e-05,
      "loss": 0.1398,
      "step": 1759
    },
    {
      "epoch": 0.6964780371982588,
      "grad_norm": 0.7147658527017271,
      "learning_rate": 2.5540315486483024e-05,
      "loss": 0.1315,
      "step": 1760
    },
    {
      "epoch": 0.6964780371982588,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9247417609444171,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.2645833194255829,
      "eval_runtime": 4.9154,
      "eval_samples_per_second": 6.103,
      "eval_steps_per_second": 0.203,
      "step": 1760
    },
    {
      "epoch": 0.6968737633557578,
      "grad_norm": 1.2225769121031047,
      "learning_rate": 2.5480092165278153e-05,
      "loss": 0.1484,
      "step": 1761
    },
    {
      "epoch": 0.6972694895132568,
      "grad_norm": 0.9689444469951302,
      "learning_rate": 2.5419915643182767e-05,
      "loss": 0.1865,
      "step": 1762
    },
    {
      "epoch": 0.6976652156707558,
      "grad_norm": 0.7512090628721629,
      "learning_rate": 2.5359786035050758e-05,
      "loss": 0.1002,
      "step": 1763
    },
    {
      "epoch": 0.6980609418282548,
      "grad_norm": 1.1961229065190835,
      "learning_rate": 2.5299703455646516e-05,
      "loss": 0.2278,
      "step": 1764
    },
    {
      "epoch": 0.6984566679857539,
      "grad_norm": 0.7141828849380475,
      "learning_rate": 2.523966801964468e-05,
      "loss": 0.1256,
      "step": 1765
    },
    {
      "epoch": 0.6984566679857539,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9235120511559273,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.2701985538005829,
      "eval_runtime": 5.1744,
      "eval_samples_per_second": 5.798,
      "eval_steps_per_second": 0.193,
      "step": 1765
    },
    {
      "epoch": 0.6988523941432528,
      "grad_norm": 0.6444027672175999,
      "learning_rate": 2.5179679841629912e-05,
      "loss": 0.1499,
      "step": 1766
    },
    {
      "epoch": 0.6992481203007519,
      "grad_norm": 0.7062316084910141,
      "learning_rate": 2.5119739036096613e-05,
      "loss": 0.1173,
      "step": 1767
    },
    {
      "epoch": 0.6996438464582508,
      "grad_norm": 0.6749115445084526,
      "learning_rate": 2.5059845717448843e-05,
      "loss": 0.1557,
      "step": 1768
    },
    {
      "epoch": 0.7000395726157499,
      "grad_norm": 0.6497584235201437,
      "learning_rate": 2.500000000000001e-05,
      "loss": 0.1299,
      "step": 1769
    },
    {
      "epoch": 0.700435298773249,
      "grad_norm": 0.8393825442091659,
      "learning_rate": 2.4940201997972628e-05,
      "loss": 0.1569,
      "step": 1770
    },
    {
      "epoch": 0.700435298773249,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9237579931136252,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.25867512822151184,
      "eval_runtime": 5.0511,
      "eval_samples_per_second": 5.939,
      "eval_steps_per_second": 0.198,
      "step": 1770
    },
    {
      "epoch": 0.7008310249307479,
      "grad_norm": 0.63761849969483,
      "learning_rate": 2.488045182549819e-05,
      "loss": 0.1468,
      "step": 1771
    },
    {
      "epoch": 0.701226751088247,
      "grad_norm": 0.8124455455201907,
      "learning_rate": 2.482074959661688e-05,
      "loss": 0.1688,
      "step": 1772
    },
    {
      "epoch": 0.7016224772457459,
      "grad_norm": 0.6621442684502544,
      "learning_rate": 2.4761095425277396e-05,
      "loss": 0.1238,
      "step": 1773
    },
    {
      "epoch": 0.702018203403245,
      "grad_norm": 0.67223349223692,
      "learning_rate": 2.4701489425336667e-05,
      "loss": 0.1468,
      "step": 1774
    },
    {
      "epoch": 0.702413929560744,
      "grad_norm": 0.9470100625396481,
      "learning_rate": 2.4641931710559717e-05,
      "loss": 0.1998,
      "step": 1775
    },
    {
      "epoch": 0.702413929560744,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9252336448598131,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.25634765625,
      "eval_runtime": 5.105,
      "eval_samples_per_second": 5.877,
      "eval_steps_per_second": 0.196,
      "step": 1775
    },
    {
      "epoch": 0.7028096557182429,
      "grad_norm": 0.5177412924220431,
      "learning_rate": 2.4582422394619427e-05,
      "loss": 0.0948,
      "step": 1776
    },
    {
      "epoch": 0.703205381875742,
      "grad_norm": 0.9312128661947334,
      "learning_rate": 2.4522961591096246e-05,
      "loss": 0.2638,
      "step": 1777
    },
    {
      "epoch": 0.703601108033241,
      "grad_norm": 0.7647043055460352,
      "learning_rate": 2.4463549413478098e-05,
      "loss": 0.1438,
      "step": 1778
    },
    {
      "epoch": 0.70399683419074,
      "grad_norm": 0.6273222039671842,
      "learning_rate": 2.4404185975160014e-05,
      "loss": 0.1199,
      "step": 1779
    },
    {
      "epoch": 0.704392560348239,
      "grad_norm": 0.7129719221346945,
      "learning_rate": 2.4344871389444128e-05,
      "loss": 0.1454,
      "step": 1780
    },
    {
      "epoch": 0.704392560348239,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9262174126906051,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.25882160663604736,
      "eval_runtime": 5.0527,
      "eval_samples_per_second": 5.937,
      "eval_steps_per_second": 0.198,
      "step": 1780
    },
    {
      "epoch": 0.704788286505738,
      "grad_norm": 0.6907453496234908,
      "learning_rate": 2.4285605769539204e-05,
      "loss": 0.1299,
      "step": 1781
    },
    {
      "epoch": 0.705184012663237,
      "grad_norm": 1.2220258745951802,
      "learning_rate": 2.4226389228560635e-05,
      "loss": 0.2075,
      "step": 1782
    },
    {
      "epoch": 0.7055797388207361,
      "grad_norm": 0.7463989674103331,
      "learning_rate": 2.416722187953006e-05,
      "loss": 0.1232,
      "step": 1783
    },
    {
      "epoch": 0.705975464978235,
      "grad_norm": 1.1653777092299322,
      "learning_rate": 2.410810383537535e-05,
      "loss": 0.1392,
      "step": 1784
    },
    {
      "epoch": 0.7063711911357341,
      "grad_norm": 0.720711674773519,
      "learning_rate": 2.4049035208930172e-05,
      "loss": 0.1251,
      "step": 1785
    },
    {
      "epoch": 0.7063711911357341,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9240039350713232,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.2592610716819763,
      "eval_runtime": 4.948,
      "eval_samples_per_second": 6.063,
      "eval_steps_per_second": 0.202,
      "step": 1785
    },
    {
      "epoch": 0.706766917293233,
      "grad_norm": 1.082201267832487,
      "learning_rate": 2.3990016112933866e-05,
      "loss": 0.1548,
      "step": 1786
    },
    {
      "epoch": 0.7071626434507321,
      "grad_norm": 0.7361189228404685,
      "learning_rate": 2.3931046660031302e-05,
      "loss": 0.1313,
      "step": 1787
    },
    {
      "epoch": 0.7075583696082312,
      "grad_norm": 0.6002673608478755,
      "learning_rate": 2.3872126962772572e-05,
      "loss": 0.1205,
      "step": 1788
    },
    {
      "epoch": 0.7079540957657301,
      "grad_norm": 0.8081772236953343,
      "learning_rate": 2.3813257133612827e-05,
      "loss": 0.1296,
      "step": 1789
    },
    {
      "epoch": 0.7083498219232292,
      "grad_norm": 0.6377485509465721,
      "learning_rate": 2.3754437284911968e-05,
      "loss": 0.1204,
      "step": 1790
    },
    {
      "epoch": 0.7083498219232292,
      "eval_PRM Accuracy": 0.8888888888888888,
      "eval_PRM F1": 0.9369369369369369,
      "eval_PRM F1 AUC": 0.6965076242006887,
      "eval_PRM F1 AUC (fixed)": 0.9222823413674373,
      "eval_PRM F1 Neg": 0.5333333333333333,
      "eval_PRM NPV": 0.7272727272727273,
      "eval_PRM Precision": 0.9043478260869565,
      "eval_PRM Recall": 0.9719626168224299,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.2582031190395355,
      "eval_runtime": 5.3866,
      "eval_samples_per_second": 5.569,
      "eval_steps_per_second": 0.186,
      "step": 1790
    },
    {
      "epoch": 0.7087455480807281,
      "grad_norm": 0.7771426302698183,
      "learning_rate": 2.3695667528934573e-05,
      "loss": 0.2042,
      "step": 1791
    },
    {
      "epoch": 0.7091412742382271,
      "grad_norm": 1.0587475144545846,
      "learning_rate": 2.3636947977849592e-05,
      "loss": 0.1252,
      "step": 1792
    },
    {
      "epoch": 0.7095370003957262,
      "grad_norm": 0.8953630043511814,
      "learning_rate": 2.357827874373017e-05,
      "loss": 0.1485,
      "step": 1793
    },
    {
      "epoch": 0.7099327265532251,
      "grad_norm": 0.8444992471566256,
      "learning_rate": 2.3519659938553352e-05,
      "loss": 0.1114,
      "step": 1794
    },
    {
      "epoch": 0.7103284527107242,
      "grad_norm": 0.7966988460144155,
      "learning_rate": 2.3461091674199998e-05,
      "loss": 0.1302,
      "step": 1795
    },
    {
      "epoch": 0.7103284527107242,
      "eval_PRM Accuracy": 0.8888888888888888,
      "eval_PRM F1": 0.9369369369369369,
      "eval_PRM F1 AUC": 0.6965076242006887,
      "eval_PRM F1 AUC (fixed)": 0.9220363994097392,
      "eval_PRM F1 Neg": 0.5333333333333333,
      "eval_PRM NPV": 0.7272727272727273,
      "eval_PRM Precision": 0.9043478260869565,
      "eval_PRM Recall": 0.9719626168224299,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.2626139223575592,
      "eval_runtime": 4.9067,
      "eval_samples_per_second": 6.114,
      "eval_steps_per_second": 0.204,
      "step": 1795
    },
    {
      "epoch": 0.7107241788682231,
      "grad_norm": 1.1093509103818675,
      "learning_rate": 2.3402574062454508e-05,
      "loss": 0.169,
      "step": 1796
    },
    {
      "epoch": 0.7111199050257222,
      "grad_norm": 0.881446740725918,
      "learning_rate": 2.334410721500454e-05,
      "loss": 0.1697,
      "step": 1797
    },
    {
      "epoch": 0.7115156311832213,
      "grad_norm": 0.6022257229410424,
      "learning_rate": 2.3285691243440927e-05,
      "loss": 0.1271,
      "step": 1798
    },
    {
      "epoch": 0.7119113573407202,
      "grad_norm": 1.344850322664759,
      "learning_rate": 2.3227326259257376e-05,
      "loss": 0.1287,
      "step": 1799
    },
    {
      "epoch": 0.7123070834982193,
      "grad_norm": 1.091061527984674,
      "learning_rate": 2.3169012373850298e-05,
      "loss": 0.2232,
      "step": 1800
    },
    {
      "epoch": 0.7123070834982193,
      "eval_PRM Accuracy": 0.8888888888888888,
      "eval_PRM F1": 0.9369369369369369,
      "eval_PRM F1 AUC": 0.6965076242006887,
      "eval_PRM F1 AUC (fixed)": 0.9208066896212493,
      "eval_PRM F1 Neg": 0.5333333333333333,
      "eval_PRM NPV": 0.7272727272727273,
      "eval_PRM Precision": 0.9043478260869565,
      "eval_PRM Recall": 0.9719626168224299,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.2690592408180237,
      "eval_runtime": 5.2356,
      "eval_samples_per_second": 5.73,
      "eval_steps_per_second": 0.191,
      "step": 1800
    },
    {
      "epoch": 0.7127028096557182,
      "grad_norm": 0.7667643167333761,
      "learning_rate": 2.311074969851852e-05,
      "loss": 0.1154,
      "step": 1801
    },
    {
      "epoch": 0.7130985358132172,
      "grad_norm": 0.7235188347019936,
      "learning_rate": 2.3052538344463187e-05,
      "loss": 0.1441,
      "step": 1802
    },
    {
      "epoch": 0.7134942619707163,
      "grad_norm": 1.061901458017209,
      "learning_rate": 2.2994378422787488e-05,
      "loss": 0.2128,
      "step": 1803
    },
    {
      "epoch": 0.7138899881282152,
      "grad_norm": 0.6174006163424876,
      "learning_rate": 2.293627004449639e-05,
      "loss": 0.1069,
      "step": 1804
    },
    {
      "epoch": 0.7142857142857143,
      "grad_norm": 0.78110244288932,
      "learning_rate": 2.2878213320496545e-05,
      "loss": 0.155,
      "step": 1805
    },
    {
      "epoch": 0.7142857142857143,
      "eval_PRM Accuracy": 0.8888888888888888,
      "eval_PRM F1": 0.9369369369369369,
      "eval_PRM F1 AUC": 0.6965076242006887,
      "eval_PRM F1 AUC (fixed)": 0.9225282833251353,
      "eval_PRM F1 Neg": 0.5333333333333333,
      "eval_PRM NPV": 0.7272727272727273,
      "eval_PRM Precision": 0.9043478260869565,
      "eval_PRM Recall": 0.9719626168224299,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.2673177123069763,
      "eval_runtime": 4.9645,
      "eval_samples_per_second": 6.043,
      "eval_steps_per_second": 0.201,
      "step": 1805
    },
    {
      "epoch": 0.7146814404432132,
      "grad_norm": 0.6647177984305762,
      "learning_rate": 2.2820208361596e-05,
      "loss": 0.0922,
      "step": 1806
    },
    {
      "epoch": 0.7150771666007123,
      "grad_norm": 0.872976643708582,
      "learning_rate": 2.2762255278504007e-05,
      "loss": 0.1663,
      "step": 1807
    },
    {
      "epoch": 0.7154728927582114,
      "grad_norm": 1.2288784922197935,
      "learning_rate": 2.2704354181830767e-05,
      "loss": 0.2664,
      "step": 1808
    },
    {
      "epoch": 0.7158686189157103,
      "grad_norm": 0.9915164163328413,
      "learning_rate": 2.26465051820873e-05,
      "loss": 0.1897,
      "step": 1809
    },
    {
      "epoch": 0.7162643450732094,
      "grad_norm": 0.8193256350929381,
      "learning_rate": 2.2588708389685193e-05,
      "loss": 0.1727,
      "step": 1810
    },
    {
      "epoch": 0.7162643450732094,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9198229217904573,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.26705729961395264,
      "eval_runtime": 5.045,
      "eval_samples_per_second": 5.946,
      "eval_steps_per_second": 0.198,
      "step": 1810
    },
    {
      "epoch": 0.7166600712307083,
      "grad_norm": 0.7925793957129699,
      "learning_rate": 2.2530963914936387e-05,
      "loss": 0.1755,
      "step": 1811
    },
    {
      "epoch": 0.7170557973882073,
      "grad_norm": 0.616622678794837,
      "learning_rate": 2.247327186805295e-05,
      "loss": 0.1209,
      "step": 1812
    },
    {
      "epoch": 0.7174515235457064,
      "grad_norm": 1.0347749318007307,
      "learning_rate": 2.2415632359146856e-05,
      "loss": 0.1541,
      "step": 1813
    },
    {
      "epoch": 0.7178472497032053,
      "grad_norm": 0.9997970613006104,
      "learning_rate": 2.2358045498229907e-05,
      "loss": 0.1933,
      "step": 1814
    },
    {
      "epoch": 0.7182429758607044,
      "grad_norm": 0.7938038481406889,
      "learning_rate": 2.2300511395213313e-05,
      "loss": 0.1557,
      "step": 1815
    },
    {
      "epoch": 0.7182429758607044,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9185932120019674,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.26321613788604736,
      "eval_runtime": 5.0179,
      "eval_samples_per_second": 5.979,
      "eval_steps_per_second": 0.199,
      "step": 1815
    },
    {
      "epoch": 0.7186387020182035,
      "grad_norm": 0.9984300276074307,
      "learning_rate": 2.224303015990767e-05,
      "loss": 0.1988,
      "step": 1816
    },
    {
      "epoch": 0.7190344281757024,
      "grad_norm": 0.6968040595401331,
      "learning_rate": 2.218560190202257e-05,
      "loss": 0.1731,
      "step": 1817
    },
    {
      "epoch": 0.7194301543332015,
      "grad_norm": 0.8232047773138893,
      "learning_rate": 2.2128226731166633e-05,
      "loss": 0.2086,
      "step": 1818
    },
    {
      "epoch": 0.7198258804907004,
      "grad_norm": 0.8184708896147453,
      "learning_rate": 2.2070904756847022e-05,
      "loss": 0.1338,
      "step": 1819
    },
    {
      "epoch": 0.7202216066481995,
      "grad_norm": 0.8360658863409396,
      "learning_rate": 2.2013636088469458e-05,
      "loss": 0.1531,
      "step": 1820
    },
    {
      "epoch": 0.7202216066481995,
      "eval_PRM Accuracy": 0.8888888888888888,
      "eval_PRM F1": 0.9369369369369369,
      "eval_PRM F1 AUC": 0.6965076242006887,
      "eval_PRM F1 AUC (fixed)": 0.9200688637481553,
      "eval_PRM F1 Neg": 0.5333333333333333,
      "eval_PRM NPV": 0.7272727272727273,
      "eval_PRM Precision": 0.9043478260869565,
      "eval_PRM Recall": 0.9719626168224299,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.2569010555744171,
      "eval_runtime": 5.0502,
      "eval_samples_per_second": 5.94,
      "eval_steps_per_second": 0.198,
      "step": 1820
    },
    {
      "epoch": 0.7206173328056985,
      "grad_norm": 0.935689418520536,
      "learning_rate": 2.1956420835337848e-05,
      "loss": 0.2041,
      "step": 1821
    },
    {
      "epoch": 0.7210130589631975,
      "grad_norm": 0.547843002459468,
      "learning_rate": 2.1899259106654215e-05,
      "loss": 0.1169,
      "step": 1822
    },
    {
      "epoch": 0.7214087851206965,
      "grad_norm": 0.7938716875293297,
      "learning_rate": 2.1842151011518413e-05,
      "loss": 0.1683,
      "step": 1823
    },
    {
      "epoch": 0.7218045112781954,
      "grad_norm": 0.7053770656801953,
      "learning_rate": 2.1785096658927873e-05,
      "loss": 0.1654,
      "step": 1824
    },
    {
      "epoch": 0.7222002374356945,
      "grad_norm": 0.7629462495716527,
      "learning_rate": 2.1728096157777517e-05,
      "loss": 0.1363,
      "step": 1825
    },
    {
      "epoch": 0.7222002374356945,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9210526315789473,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.2576334774494171,
      "eval_runtime": 5.0,
      "eval_samples_per_second": 6.0,
      "eval_steps_per_second": 0.2,
      "step": 1825
    },
    {
      "epoch": 0.7225959635931936,
      "grad_norm": 0.5864294821965705,
      "learning_rate": 2.1671149616859466e-05,
      "loss": 0.1087,
      "step": 1826
    },
    {
      "epoch": 0.7229916897506925,
      "grad_norm": 0.7113916442051033,
      "learning_rate": 2.161425714486286e-05,
      "loss": 0.1104,
      "step": 1827
    },
    {
      "epoch": 0.7233874159081916,
      "grad_norm": 0.5509631675986841,
      "learning_rate": 2.1557418850373602e-05,
      "loss": 0.1245,
      "step": 1828
    },
    {
      "epoch": 0.7237831420656905,
      "grad_norm": 0.7648877569661009,
      "learning_rate": 2.1500634841874224e-05,
      "loss": 0.1545,
      "step": 1829
    },
    {
      "epoch": 0.7241788682231896,
      "grad_norm": 0.8304892719391077,
      "learning_rate": 2.144390522774367e-05,
      "loss": 0.1897,
      "step": 1830
    },
    {
      "epoch": 0.7241788682231896,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9227742252828331,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.26695963740348816,
      "eval_runtime": 5.2719,
      "eval_samples_per_second": 5.691,
      "eval_steps_per_second": 0.19,
      "step": 1830
    },
    {
      "epoch": 0.7245745943806886,
      "grad_norm": 1.222408327159709,
      "learning_rate": 2.1387230116257e-05,
      "loss": 0.1842,
      "step": 1831
    },
    {
      "epoch": 0.7249703205381876,
      "grad_norm": 0.8785348974216282,
      "learning_rate": 2.1330609615585308e-05,
      "loss": 0.2035,
      "step": 1832
    },
    {
      "epoch": 0.7253660466956866,
      "grad_norm": 1.2074994467319153,
      "learning_rate": 2.1274043833795426e-05,
      "loss": 0.1578,
      "step": 1833
    },
    {
      "epoch": 0.7257617728531855,
      "grad_norm": 1.011105442602931,
      "learning_rate": 2.1217532878849787e-05,
      "loss": 0.2353,
      "step": 1834
    },
    {
      "epoch": 0.7261574990106846,
      "grad_norm": 0.6063287756839674,
      "learning_rate": 2.1161076858606115e-05,
      "loss": 0.1467,
      "step": 1835
    },
    {
      "epoch": 0.7261574990106846,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9222823413674373,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.26762694120407104,
      "eval_runtime": 4.963,
      "eval_samples_per_second": 6.045,
      "eval_steps_per_second": 0.201,
      "step": 1835
    },
    {
      "epoch": 0.7265532251681837,
      "grad_norm": 0.8788623002340055,
      "learning_rate": 2.1104675880817337e-05,
      "loss": 0.1898,
      "step": 1836
    },
    {
      "epoch": 0.7269489513256826,
      "grad_norm": 0.8949887512219598,
      "learning_rate": 2.104833005313131e-05,
      "loss": 0.2223,
      "step": 1837
    },
    {
      "epoch": 0.7273446774831817,
      "grad_norm": 0.4545095104266266,
      "learning_rate": 2.0992039483090647e-05,
      "loss": 0.088,
      "step": 1838
    },
    {
      "epoch": 0.7277404036406806,
      "grad_norm": 1.3154146779466707,
      "learning_rate": 2.0935804278132448e-05,
      "loss": 0.2578,
      "step": 1839
    },
    {
      "epoch": 0.7281361297981797,
      "grad_norm": 0.7802206534168505,
      "learning_rate": 2.087962454558819e-05,
      "loss": 0.1472,
      "step": 1840
    },
    {
      "epoch": 0.7281361297981797,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9227742252828333,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.26087239384651184,
      "eval_runtime": 5.1355,
      "eval_samples_per_second": 5.842,
      "eval_steps_per_second": 0.195,
      "step": 1840
    },
    {
      "epoch": 0.7285318559556787,
      "grad_norm": 0.6682288985252659,
      "learning_rate": 2.0823500392683476e-05,
      "loss": 0.1218,
      "step": 1841
    },
    {
      "epoch": 0.7289275821131777,
      "grad_norm": 1.0532540165424913,
      "learning_rate": 2.0767431926537766e-05,
      "loss": 0.2032,
      "step": 1842
    },
    {
      "epoch": 0.7293233082706767,
      "grad_norm": 0.706851700973422,
      "learning_rate": 2.071141925416431e-05,
      "loss": 0.1291,
      "step": 1843
    },
    {
      "epoch": 0.7297190344281757,
      "grad_norm": 1.611661542363454,
      "learning_rate": 2.0655462482469833e-05,
      "loss": 0.1793,
      "step": 1844
    },
    {
      "epoch": 0.7301147605856747,
      "grad_norm": 0.752767555773831,
      "learning_rate": 2.0599561718254397e-05,
      "loss": 0.1381,
      "step": 1845
    },
    {
      "epoch": 0.7301147605856747,
      "eval_PRM Accuracy": 0.8888888888888888,
      "eval_PRM F1": 0.9369369369369369,
      "eval_PRM F1 AUC": 0.6965076242006887,
      "eval_PRM F1 AUC (fixed)": 0.9235120511559272,
      "eval_PRM F1 Neg": 0.5333333333333333,
      "eval_PRM NPV": 0.7272727272727273,
      "eval_PRM Precision": 0.9043478260869565,
      "eval_PRM Recall": 0.9719626168224299,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.2558431029319763,
      "eval_runtime": 4.8512,
      "eval_samples_per_second": 6.184,
      "eval_steps_per_second": 0.206,
      "step": 1845
    },
    {
      "epoch": 0.7305104867431738,
      "grad_norm": 0.8808783337543461,
      "learning_rate": 2.0543717068211093e-05,
      "loss": 0.1647,
      "step": 1846
    },
    {
      "epoch": 0.7309062129006727,
      "grad_norm": 0.7959998264879183,
      "learning_rate": 2.048792863892599e-05,
      "loss": 0.2045,
      "step": 1847
    },
    {
      "epoch": 0.7313019390581718,
      "grad_norm": 1.598833931971828,
      "learning_rate": 2.043219653687784e-05,
      "loss": 0.1503,
      "step": 1848
    },
    {
      "epoch": 0.7316976652156707,
      "grad_norm": 0.8894720772872597,
      "learning_rate": 2.0376520868437838e-05,
      "loss": 0.1218,
      "step": 1849
    },
    {
      "epoch": 0.7320933913731698,
      "grad_norm": 1.0993024233554871,
      "learning_rate": 2.0320901739869537e-05,
      "loss": 0.203,
      "step": 1850
    },
    {
      "epoch": 0.7320933913731698,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9217904574520412,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.25882160663604736,
      "eval_runtime": 5.1301,
      "eval_samples_per_second": 5.848,
      "eval_steps_per_second": 0.195,
      "step": 1850
    },
    {
      "epoch": 0.7324891175306688,
      "grad_norm": 0.7854335067265968,
      "learning_rate": 2.026533925732849e-05,
      "loss": 0.2077,
      "step": 1851
    },
    {
      "epoch": 0.7328848436881678,
      "grad_norm": 0.7090756283038675,
      "learning_rate": 2.0209833526862267e-05,
      "loss": 0.1111,
      "step": 1852
    },
    {
      "epoch": 0.7332805698456668,
      "grad_norm": 1.0333691573481174,
      "learning_rate": 2.0154384654409975e-05,
      "loss": 0.2353,
      "step": 1853
    },
    {
      "epoch": 0.7336762960031659,
      "grad_norm": 0.7562127657862519,
      "learning_rate": 2.009899274580231e-05,
      "loss": 0.1349,
      "step": 1854
    },
    {
      "epoch": 0.7340720221606648,
      "grad_norm": 0.8448709028126433,
      "learning_rate": 2.0043657906761142e-05,
      "loss": 0.1642,
      "step": 1855
    },
    {
      "epoch": 0.7340720221606648,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9205607476635513,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.26012369990348816,
      "eval_runtime": 5.1147,
      "eval_samples_per_second": 5.865,
      "eval_steps_per_second": 0.196,
      "step": 1855
    },
    {
      "epoch": 0.7344677483181639,
      "grad_norm": 0.5946066961791077,
      "learning_rate": 1.998838024289956e-05,
      "loss": 0.1345,
      "step": 1856
    },
    {
      "epoch": 0.7348634744756628,
      "grad_norm": 0.5412120416126538,
      "learning_rate": 1.9933159859721408e-05,
      "loss": 0.1063,
      "step": 1857
    },
    {
      "epoch": 0.7352592006331619,
      "grad_norm": 0.562011803081114,
      "learning_rate": 1.9877996862621207e-05,
      "loss": 0.1434,
      "step": 1858
    },
    {
      "epoch": 0.7356549267906609,
      "grad_norm": 0.6991284129518043,
      "learning_rate": 1.9822891356884022e-05,
      "loss": 0.1797,
      "step": 1859
    },
    {
      "epoch": 0.7360506529481599,
      "grad_norm": 0.6700773353560388,
      "learning_rate": 1.9767843447685137e-05,
      "loss": 0.1359,
      "step": 1860
    },
    {
      "epoch": 0.7360506529481599,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9232661091982292,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.2613281309604645,
      "eval_runtime": 4.9601,
      "eval_samples_per_second": 6.048,
      "eval_steps_per_second": 0.202,
      "step": 1860
    },
    {
      "epoch": 0.7364463791056589,
      "grad_norm": 0.7397056952998681,
      "learning_rate": 1.971285324008994e-05,
      "loss": 0.1528,
      "step": 1861
    },
    {
      "epoch": 0.7368421052631579,
      "grad_norm": 0.685117972157196,
      "learning_rate": 1.9657920839053634e-05,
      "loss": 0.1357,
      "step": 1862
    },
    {
      "epoch": 0.7372378314206569,
      "grad_norm": 0.5866163026587115,
      "learning_rate": 1.9603046349421146e-05,
      "loss": 0.1265,
      "step": 1863
    },
    {
      "epoch": 0.737633557578156,
      "grad_norm": 0.5574623985714691,
      "learning_rate": 1.954822987592685e-05,
      "loss": 0.1357,
      "step": 1864
    },
    {
      "epoch": 0.7380292837356549,
      "grad_norm": 0.6801805629067953,
      "learning_rate": 1.9493471523194402e-05,
      "loss": 0.1537,
      "step": 1865
    },
    {
      "epoch": 0.7380292837356549,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9235120511559272,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.2666829526424408,
      "eval_runtime": 4.9638,
      "eval_samples_per_second": 6.044,
      "eval_steps_per_second": 0.201,
      "step": 1865
    },
    {
      "epoch": 0.738425009893154,
      "grad_norm": 0.6890667000540105,
      "learning_rate": 1.9438771395736495e-05,
      "loss": 0.1701,
      "step": 1866
    },
    {
      "epoch": 0.7388207360506529,
      "grad_norm": 0.7013380568225412,
      "learning_rate": 1.938412959795472e-05,
      "loss": 0.1766,
      "step": 1867
    },
    {
      "epoch": 0.739216462208152,
      "grad_norm": 0.8407166916867785,
      "learning_rate": 1.9329546234139356e-05,
      "loss": 0.2045,
      "step": 1868
    },
    {
      "epoch": 0.739612188365651,
      "grad_norm": 0.6777612034422675,
      "learning_rate": 1.92750214084691e-05,
      "loss": 0.158,
      "step": 1869
    },
    {
      "epoch": 0.74000791452315,
      "grad_norm": 0.6925321769242335,
      "learning_rate": 1.9220555225010966e-05,
      "loss": 0.1199,
      "step": 1870
    },
    {
      "epoch": 0.74000791452315,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9237579931136252,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.2687825560569763,
      "eval_runtime": 4.7512,
      "eval_samples_per_second": 6.314,
      "eval_steps_per_second": 0.21,
      "step": 1870
    },
    {
      "epoch": 0.740403640680649,
      "grad_norm": 0.6418918195567985,
      "learning_rate": 1.9166147787720036e-05,
      "loss": 0.1172,
      "step": 1871
    },
    {
      "epoch": 0.740799366838148,
      "grad_norm": 0.8511457143116246,
      "learning_rate": 1.9111799200439267e-05,
      "loss": 0.156,
      "step": 1872
    },
    {
      "epoch": 0.741195092995647,
      "grad_norm": 0.593750232907013,
      "learning_rate": 1.9057509566899266e-05,
      "loss": 0.1287,
      "step": 1873
    },
    {
      "epoch": 0.7415908191531461,
      "grad_norm": 0.9645618491293677,
      "learning_rate": 1.900327899071816e-05,
      "loss": 0.215,
      "step": 1874
    },
    {
      "epoch": 0.741986545310645,
      "grad_norm": 0.6291843933593182,
      "learning_rate": 1.8949107575401347e-05,
      "loss": 0.1238,
      "step": 1875
    },
    {
      "epoch": 0.741986545310645,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.925725528775209,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.2700358033180237,
      "eval_runtime": 5.0433,
      "eval_samples_per_second": 5.949,
      "eval_steps_per_second": 0.198,
      "step": 1875
    },
    {
      "epoch": 0.7423822714681441,
      "grad_norm": 0.7887121603574619,
      "learning_rate": 1.8894995424341285e-05,
      "loss": 0.1871,
      "step": 1876
    },
    {
      "epoch": 0.742777997625643,
      "grad_norm": 0.7646745464288452,
      "learning_rate": 1.8840942640817338e-05,
      "loss": 0.1718,
      "step": 1877
    },
    {
      "epoch": 0.7431737237831421,
      "grad_norm": 1.013361402933889,
      "learning_rate": 1.8786949327995574e-05,
      "loss": 0.1739,
      "step": 1878
    },
    {
      "epoch": 0.7435694499406411,
      "grad_norm": 0.6718042298138901,
      "learning_rate": 1.873301558892855e-05,
      "loss": 0.1057,
      "step": 1879
    },
    {
      "epoch": 0.7439651760981401,
      "grad_norm": 0.8412749202429344,
      "learning_rate": 1.8679141526555078e-05,
      "loss": 0.1897,
      "step": 1880
    },
    {
      "epoch": 0.7439651760981401,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9281849483521889,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.26184895634651184,
      "eval_runtime": 4.8657,
      "eval_samples_per_second": 6.166,
      "eval_steps_per_second": 0.206,
      "step": 1880
    },
    {
      "epoch": 0.7443609022556391,
      "grad_norm": 1.2701465098908922,
      "learning_rate": 1.862532724370012e-05,
      "loss": 0.3103,
      "step": 1881
    },
    {
      "epoch": 0.744756628413138,
      "grad_norm": 0.5270005801702822,
      "learning_rate": 1.857157284307452e-05,
      "loss": 0.1239,
      "step": 1882
    },
    {
      "epoch": 0.7451523545706371,
      "grad_norm": 0.6226596056956873,
      "learning_rate": 1.8517878427274848e-05,
      "loss": 0.1416,
      "step": 1883
    },
    {
      "epoch": 0.7455480807281362,
      "grad_norm": 0.6289485615149631,
      "learning_rate": 1.8464244098783163e-05,
      "loss": 0.1187,
      "step": 1884
    },
    {
      "epoch": 0.7459438068856351,
      "grad_norm": 0.7071818164504134,
      "learning_rate": 1.84106699599668e-05,
      "loss": 0.1958,
      "step": 1885
    },
    {
      "epoch": 0.7459438068856351,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.923757993113625,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.2503092586994171,
      "eval_runtime": 4.8571,
      "eval_samples_per_second": 6.176,
      "eval_steps_per_second": 0.206,
      "step": 1885
    },
    {
      "epoch": 0.7463395330431342,
      "grad_norm": 0.7457821761487807,
      "learning_rate": 1.8357156113078357e-05,
      "loss": 0.173,
      "step": 1886
    },
    {
      "epoch": 0.7467352592006332,
      "grad_norm": 0.6651083170515766,
      "learning_rate": 1.8303702660255184e-05,
      "loss": 0.1457,
      "step": 1887
    },
    {
      "epoch": 0.7471309853581322,
      "grad_norm": 0.5781612367551257,
      "learning_rate": 1.8250309703519496e-05,
      "loss": 0.1143,
      "step": 1888
    },
    {
      "epoch": 0.7475267115156312,
      "grad_norm": 0.6602681307700908,
      "learning_rate": 1.8196977344777933e-05,
      "loss": 0.1153,
      "step": 1889
    },
    {
      "epoch": 0.7479224376731302,
      "grad_norm": 0.6768341783130551,
      "learning_rate": 1.81437056858216e-05,
      "loss": 0.1221,
      "step": 1890
    },
    {
      "epoch": 0.7479224376731302,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9242498770290212,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.25029295682907104,
      "eval_runtime": 5.1109,
      "eval_samples_per_second": 5.87,
      "eval_steps_per_second": 0.196,
      "step": 1890
    },
    {
      "epoch": 0.7483181638306292,
      "grad_norm": 0.6061568247001031,
      "learning_rate": 1.809049482832563e-05,
      "loss": 0.1136,
      "step": 1891
    },
    {
      "epoch": 0.7487138899881283,
      "grad_norm": 0.8602003297227924,
      "learning_rate": 1.803734487384921e-05,
      "loss": 0.2234,
      "step": 1892
    },
    {
      "epoch": 0.7491096161456272,
      "grad_norm": 0.5288064051905645,
      "learning_rate": 1.798425592383519e-05,
      "loss": 0.106,
      "step": 1893
    },
    {
      "epoch": 0.7495053423031263,
      "grad_norm": 0.6390476855427978,
      "learning_rate": 1.7931228079610057e-05,
      "loss": 0.1516,
      "step": 1894
    },
    {
      "epoch": 0.7499010684606252,
      "grad_norm": 0.5448235403613532,
      "learning_rate": 1.787826144238367e-05,
      "loss": 0.1028,
      "step": 1895
    },
    {
      "epoch": 0.7499010684606252,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9244958189867192,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.24947915971279144,
      "eval_runtime": 5.1287,
      "eval_samples_per_second": 5.849,
      "eval_steps_per_second": 0.195,
      "step": 1895
    },
    {
      "epoch": 0.7502967946181243,
      "grad_norm": 0.6962915539093395,
      "learning_rate": 1.782535611324901e-05,
      "loss": 0.1303,
      "step": 1896
    },
    {
      "epoch": 0.7506925207756233,
      "grad_norm": 0.9712828059068266,
      "learning_rate": 1.7772512193182095e-05,
      "loss": 0.1736,
      "step": 1897
    },
    {
      "epoch": 0.7510882469331223,
      "grad_norm": 1.0102606709966984,
      "learning_rate": 1.7719729783041717e-05,
      "loss": 0.173,
      "step": 1898
    },
    {
      "epoch": 0.7514839730906213,
      "grad_norm": 0.9469177852957767,
      "learning_rate": 1.76670089835693e-05,
      "loss": 0.1374,
      "step": 1899
    },
    {
      "epoch": 0.7518796992481203,
      "grad_norm": 0.8060418419435674,
      "learning_rate": 1.7614349895388614e-05,
      "loss": 0.1734,
      "step": 1900
    },
    {
      "epoch": 0.7518796992481203,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9242498770290212,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.2535969913005829,
      "eval_runtime": 4.8397,
      "eval_samples_per_second": 6.199,
      "eval_steps_per_second": 0.207,
      "step": 1900
    },
    {
      "epoch": 0.7522754254056193,
      "grad_norm": 0.9774110880657569,
      "learning_rate": 1.7561752619005695e-05,
      "loss": 0.2109,
      "step": 1901
    },
    {
      "epoch": 0.7526711515631184,
      "grad_norm": 0.9606402915892724,
      "learning_rate": 1.7509217254808613e-05,
      "loss": 0.1851,
      "step": 1902
    },
    {
      "epoch": 0.7530668777206173,
      "grad_norm": 0.7696885166820605,
      "learning_rate": 1.745674390306722e-05,
      "loss": 0.124,
      "step": 1903
    },
    {
      "epoch": 0.7534626038781164,
      "grad_norm": 1.3881842536819593,
      "learning_rate": 1.7404332663933043e-05,
      "loss": 0.2536,
      "step": 1904
    },
    {
      "epoch": 0.7538583300356153,
      "grad_norm": 0.45290554027364655,
      "learning_rate": 1.735198363743907e-05,
      "loss": 0.0761,
      "step": 1905
    },
    {
      "epoch": 0.7538583300356153,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9237579931136252,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.26370441913604736,
      "eval_runtime": 4.7899,
      "eval_samples_per_second": 6.263,
      "eval_steps_per_second": 0.209,
      "step": 1905
    },
    {
      "epoch": 0.7542540561931144,
      "grad_norm": 0.6973996164170063,
      "learning_rate": 1.7299696923499543e-05,
      "loss": 0.138,
      "step": 1906
    },
    {
      "epoch": 0.7546497823506134,
      "grad_norm": 0.7156235199612323,
      "learning_rate": 1.7247472621909737e-05,
      "loss": 0.1421,
      "step": 1907
    },
    {
      "epoch": 0.7550455085081124,
      "grad_norm": 0.7399275790041239,
      "learning_rate": 1.7195310832345852e-05,
      "loss": 0.1179,
      "step": 1908
    },
    {
      "epoch": 0.7554412346656114,
      "grad_norm": 0.7628533558284784,
      "learning_rate": 1.7143211654364762e-05,
      "loss": 0.0991,
      "step": 1909
    },
    {
      "epoch": 0.7558369608231104,
      "grad_norm": 1.2374120942816305,
      "learning_rate": 1.7091175187403842e-05,
      "loss": 0.1673,
      "step": 1910
    },
    {
      "epoch": 0.7558369608231104,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9244958189867192,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.2711588442325592,
      "eval_runtime": 5.3104,
      "eval_samples_per_second": 5.649,
      "eval_steps_per_second": 0.188,
      "step": 1910
    },
    {
      "epoch": 0.7562326869806094,
      "grad_norm": 0.9587810384153634,
      "learning_rate": 1.7039201530780742e-05,
      "loss": 0.1688,
      "step": 1911
    },
    {
      "epoch": 0.7566284131381085,
      "grad_norm": 0.9346589052296339,
      "learning_rate": 1.6987290783693282e-05,
      "loss": 0.1563,
      "step": 1912
    },
    {
      "epoch": 0.7570241392956074,
      "grad_norm": 0.8323818304230431,
      "learning_rate": 1.6935443045219198e-05,
      "loss": 0.1523,
      "step": 1913
    },
    {
      "epoch": 0.7574198654531065,
      "grad_norm": 0.5736660369505782,
      "learning_rate": 1.6883658414315928e-05,
      "loss": 0.1112,
      "step": 1914
    },
    {
      "epoch": 0.7578155916106054,
      "grad_norm": 0.7325154994726154,
      "learning_rate": 1.6831936989820506e-05,
      "loss": 0.1329,
      "step": 1915
    },
    {
      "epoch": 0.7578155916106054,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9244958189867192,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.26451823115348816,
      "eval_runtime": 5.0963,
      "eval_samples_per_second": 5.887,
      "eval_steps_per_second": 0.196,
      "step": 1915
    },
    {
      "epoch": 0.7582113177681045,
      "grad_norm": 1.2498202159987677,
      "learning_rate": 1.6780278870449325e-05,
      "loss": 0.1501,
      "step": 1916
    },
    {
      "epoch": 0.7586070439256035,
      "grad_norm": 0.6324257092071747,
      "learning_rate": 1.6728684154797942e-05,
      "loss": 0.1307,
      "step": 1917
    },
    {
      "epoch": 0.7590027700831025,
      "grad_norm": 0.6905834612697092,
      "learning_rate": 1.6677152941340873e-05,
      "loss": 0.1969,
      "step": 1918
    },
    {
      "epoch": 0.7593984962406015,
      "grad_norm": 0.9190314941323324,
      "learning_rate": 1.662568532843149e-05,
      "loss": 0.1168,
      "step": 1919
    },
    {
      "epoch": 0.7597942223981005,
      "grad_norm": 0.541100149633461,
      "learning_rate": 1.6574281414301744e-05,
      "loss": 0.1163,
      "step": 1920
    },
    {
      "epoch": 0.7597942223981005,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9230201672405313,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.25712889432907104,
      "eval_runtime": 4.9833,
      "eval_samples_per_second": 6.02,
      "eval_steps_per_second": 0.201,
      "step": 1920
    },
    {
      "epoch": 0.7601899485555995,
      "grad_norm": 0.8897399836804901,
      "learning_rate": 1.6522941297061996e-05,
      "loss": 0.16,
      "step": 1921
    },
    {
      "epoch": 0.7605856747130986,
      "grad_norm": 0.8729767902260266,
      "learning_rate": 1.647166507470088e-05,
      "loss": 0.1208,
      "step": 1922
    },
    {
      "epoch": 0.7609814008705975,
      "grad_norm": 0.8441619096287705,
      "learning_rate": 1.642045284508502e-05,
      "loss": 0.1479,
      "step": 1923
    },
    {
      "epoch": 0.7613771270280966,
      "grad_norm": 0.6859854880173678,
      "learning_rate": 1.6369304705959e-05,
      "loss": 0.1089,
      "step": 1924
    },
    {
      "epoch": 0.7617728531855956,
      "grad_norm": 1.0202622051045571,
      "learning_rate": 1.631822075494497e-05,
      "loss": 0.1591,
      "step": 1925
    },
    {
      "epoch": 0.7617728531855956,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.923512051155927,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.24915364384651184,
      "eval_runtime": 4.9515,
      "eval_samples_per_second": 6.059,
      "eval_steps_per_second": 0.202,
      "step": 1925
    },
    {
      "epoch": 0.7621685793430946,
      "grad_norm": 0.6214881719485488,
      "learning_rate": 1.6267201089542657e-05,
      "loss": 0.1177,
      "step": 1926
    },
    {
      "epoch": 0.7625643055005936,
      "grad_norm": 0.7935109548830888,
      "learning_rate": 1.6216245807129004e-05,
      "loss": 0.151,
      "step": 1927
    },
    {
      "epoch": 0.7629600316580926,
      "grad_norm": 0.9625217471471244,
      "learning_rate": 1.616535500495818e-05,
      "loss": 0.1684,
      "step": 1928
    },
    {
      "epoch": 0.7633557578155916,
      "grad_norm": 0.9612372113985292,
      "learning_rate": 1.6114528780161213e-05,
      "loss": 0.1658,
      "step": 1929
    },
    {
      "epoch": 0.7637514839730907,
      "grad_norm": 0.7836931974187432,
      "learning_rate": 1.606376722974586e-05,
      "loss": 0.1692,
      "step": 1930
    },
    {
      "epoch": 0.7637514839730907,
      "eval_PRM Accuracy": 0.8888888888888888,
      "eval_PRM F1": 0.9369369369369369,
      "eval_PRM F1 AUC": 0.6965076242006887,
      "eval_PRM F1 AUC (fixed)": 0.9237579931136252,
      "eval_PRM F1 Neg": 0.5333333333333333,
      "eval_PRM NPV": 0.7272727272727273,
      "eval_PRM Precision": 0.9043478260869565,
      "eval_PRM Recall": 0.9719626168224299,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.2490234375,
      "eval_runtime": 5.1392,
      "eval_samples_per_second": 5.837,
      "eval_steps_per_second": 0.195,
      "step": 1930
    },
    {
      "epoch": 0.7641472101305896,
      "grad_norm": 0.6320690159725635,
      "learning_rate": 1.6013070450596492e-05,
      "loss": 0.1148,
      "step": 1931
    },
    {
      "epoch": 0.7645429362880887,
      "grad_norm": 0.8331262626825312,
      "learning_rate": 1.596243853947384e-05,
      "loss": 0.1706,
      "step": 1932
    },
    {
      "epoch": 0.7649386624455876,
      "grad_norm": 0.8289234826521927,
      "learning_rate": 1.5911871593014837e-05,
      "loss": 0.1645,
      "step": 1933
    },
    {
      "epoch": 0.7653343886030867,
      "grad_norm": 0.4996517959598237,
      "learning_rate": 1.586136970773238e-05,
      "loss": 0.0843,
      "step": 1934
    },
    {
      "epoch": 0.7657301147605857,
      "grad_norm": 0.9749394297349808,
      "learning_rate": 1.581093298001523e-05,
      "loss": 0.1514,
      "step": 1935
    },
    {
      "epoch": 0.7657301147605857,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9242498770290212,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.25175780057907104,
      "eval_runtime": 4.8736,
      "eval_samples_per_second": 6.156,
      "eval_steps_per_second": 0.205,
      "step": 1935
    },
    {
      "epoch": 0.7661258409180847,
      "grad_norm": 0.9698507140339012,
      "learning_rate": 1.576056150612778e-05,
      "loss": 0.1652,
      "step": 1936
    },
    {
      "epoch": 0.7665215670755837,
      "grad_norm": 0.5899250109867621,
      "learning_rate": 1.5710255382209887e-05,
      "loss": 0.1459,
      "step": 1937
    },
    {
      "epoch": 0.7669172932330827,
      "grad_norm": 0.7387431771529708,
      "learning_rate": 1.5660014704276638e-05,
      "loss": 0.1305,
      "step": 1938
    },
    {
      "epoch": 0.7673130193905817,
      "grad_norm": 0.7745268402020463,
      "learning_rate": 1.5609839568218247e-05,
      "loss": 0.1362,
      "step": 1939
    },
    {
      "epoch": 0.7677087455480808,
      "grad_norm": 0.7376208625680865,
      "learning_rate": 1.5559730069799857e-05,
      "loss": 0.114,
      "step": 1940
    },
    {
      "epoch": 0.7677087455480808,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9249877029021151,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.25719401240348816,
      "eval_runtime": 5.1689,
      "eval_samples_per_second": 5.804,
      "eval_steps_per_second": 0.193,
      "step": 1940
    },
    {
      "epoch": 0.7681044717055797,
      "grad_norm": 0.8254472521363346,
      "learning_rate": 1.5509686304661247e-05,
      "loss": 0.1895,
      "step": 1941
    },
    {
      "epoch": 0.7685001978630788,
      "grad_norm": 0.7458558178260173,
      "learning_rate": 1.5459708368316823e-05,
      "loss": 0.1351,
      "step": 1942
    },
    {
      "epoch": 0.7688959240205777,
      "grad_norm": 0.7248734844842177,
      "learning_rate": 1.5409796356155316e-05,
      "loss": 0.1257,
      "step": 1943
    },
    {
      "epoch": 0.7692916501780768,
      "grad_norm": 0.950536629653865,
      "learning_rate": 1.535995036343964e-05,
      "loss": 0.174,
      "step": 1944
    },
    {
      "epoch": 0.7696873763355758,
      "grad_norm": 0.6440801880601715,
      "learning_rate": 1.5310170485306672e-05,
      "loss": 0.1141,
      "step": 1945
    },
    {
      "epoch": 0.7696873763355758,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.925971470732907,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.2586100399494171,
      "eval_runtime": 4.991,
      "eval_samples_per_second": 6.011,
      "eval_steps_per_second": 0.2,
      "step": 1945
    },
    {
      "epoch": 0.7700831024930748,
      "grad_norm": 0.6884330770400127,
      "learning_rate": 1.526045681676715e-05,
      "loss": 0.1506,
      "step": 1946
    },
    {
      "epoch": 0.7704788286505738,
      "grad_norm": 1.0214379577455484,
      "learning_rate": 1.5210809452705415e-05,
      "loss": 0.1667,
      "step": 1947
    },
    {
      "epoch": 0.7708745548080728,
      "grad_norm": 0.5586595423535153,
      "learning_rate": 1.5161228487879253e-05,
      "loss": 0.1057,
      "step": 1948
    },
    {
      "epoch": 0.7712702809655718,
      "grad_norm": 0.7297946461912072,
      "learning_rate": 1.5111714016919732e-05,
      "loss": 0.1309,
      "step": 1949
    },
    {
      "epoch": 0.7716660071230709,
      "grad_norm": 0.5773660211167126,
      "learning_rate": 1.5062266134331016e-05,
      "loss": 0.0812,
      "step": 1950
    },
    {
      "epoch": 0.7716660071230709,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.925233644859813,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.25493162870407104,
      "eval_runtime": 5.0596,
      "eval_samples_per_second": 5.929,
      "eval_steps_per_second": 0.198,
      "step": 1950
    },
    {
      "epoch": 0.7720617332805698,
      "grad_norm": 1.0547670883999753,
      "learning_rate": 1.5012884934490167e-05,
      "loss": 0.2116,
      "step": 1951
    },
    {
      "epoch": 0.7724574594380689,
      "grad_norm": 1.000326118103544,
      "learning_rate": 1.4963570511646963e-05,
      "loss": 0.1148,
      "step": 1952
    },
    {
      "epoch": 0.7728531855955678,
      "grad_norm": 0.5817142385356567,
      "learning_rate": 1.491432295992375e-05,
      "loss": 0.0855,
      "step": 1953
    },
    {
      "epoch": 0.7732489117530669,
      "grad_norm": 0.5842996380214173,
      "learning_rate": 1.4865142373315227e-05,
      "loss": 0.0922,
      "step": 1954
    },
    {
      "epoch": 0.7736446379105659,
      "grad_norm": 1.0032121825622722,
      "learning_rate": 1.4816028845688323e-05,
      "loss": 0.1512,
      "step": 1955
    },
    {
      "epoch": 0.7736446379105659,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9227742252828333,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.2508789002895355,
      "eval_runtime": 5.1565,
      "eval_samples_per_second": 5.818,
      "eval_steps_per_second": 0.194,
      "step": 1955
    },
    {
      "epoch": 0.7740403640680649,
      "grad_norm": 0.7476451697325677,
      "learning_rate": 1.4766982470781915e-05,
      "loss": 0.1531,
      "step": 1956
    },
    {
      "epoch": 0.7744360902255639,
      "grad_norm": 1.299508253332237,
      "learning_rate": 1.4718003342206722e-05,
      "loss": 0.1317,
      "step": 1957
    },
    {
      "epoch": 0.7748318163830629,
      "grad_norm": 0.8700567899717432,
      "learning_rate": 1.4669091553445185e-05,
      "loss": 0.1963,
      "step": 1958
    },
    {
      "epoch": 0.7752275425405619,
      "grad_norm": 0.9867674541392951,
      "learning_rate": 1.4620247197851128e-05,
      "loss": 0.1448,
      "step": 1959
    },
    {
      "epoch": 0.775623268698061,
      "grad_norm": 0.9071568434015389,
      "learning_rate": 1.4571470368649742e-05,
      "loss": 0.1379,
      "step": 1960
    },
    {
      "epoch": 0.775623268698061,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9235120511559272,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.24951171875,
      "eval_runtime": 5.2599,
      "eval_samples_per_second": 5.704,
      "eval_steps_per_second": 0.19,
      "step": 1960
    },
    {
      "epoch": 0.7760189948555599,
      "grad_norm": 0.7298476313648533,
      "learning_rate": 1.452276115893726e-05,
      "loss": 0.172,
      "step": 1961
    },
    {
      "epoch": 0.776414721013059,
      "grad_norm": 1.210691735867347,
      "learning_rate": 1.4474119661680962e-05,
      "loss": 0.2156,
      "step": 1962
    },
    {
      "epoch": 0.776810447170558,
      "grad_norm": 0.7994271718937888,
      "learning_rate": 1.4425545969718801e-05,
      "loss": 0.187,
      "step": 1963
    },
    {
      "epoch": 0.777206173328057,
      "grad_norm": 1.0195110705501482,
      "learning_rate": 1.437704017575932e-05,
      "loss": 0.1561,
      "step": 1964
    },
    {
      "epoch": 0.777601899485556,
      "grad_norm": 0.7240433407015456,
      "learning_rate": 1.4328602372381522e-05,
      "loss": 0.1241,
      "step": 1965
    },
    {
      "epoch": 0.777601899485556,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9227742252828331,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.25110676884651184,
      "eval_runtime": 5.0522,
      "eval_samples_per_second": 5.938,
      "eval_steps_per_second": 0.198,
      "step": 1965
    },
    {
      "epoch": 0.777997625643055,
      "grad_norm": 1.0966803506866127,
      "learning_rate": 1.428023265203461e-05,
      "loss": 0.2095,
      "step": 1966
    },
    {
      "epoch": 0.778393351800554,
      "grad_norm": 1.0466756914844653,
      "learning_rate": 1.4231931107037861e-05,
      "loss": 0.2042,
      "step": 1967
    },
    {
      "epoch": 0.7787890779580531,
      "grad_norm": 0.7438021090210153,
      "learning_rate": 1.4183697829580389e-05,
      "loss": 0.1501,
      "step": 1968
    },
    {
      "epoch": 0.779184804115552,
      "grad_norm": 0.5095964872484632,
      "learning_rate": 1.4135532911721061e-05,
      "loss": 0.0931,
      "step": 1969
    },
    {
      "epoch": 0.7795805302730511,
      "grad_norm": 0.8596075988231007,
      "learning_rate": 1.4087436445388242e-05,
      "loss": 0.1669,
      "step": 1970
    },
    {
      "epoch": 0.7795805302730511,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9225282833251353,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.2524251341819763,
      "eval_runtime": 5.3927,
      "eval_samples_per_second": 5.563,
      "eval_steps_per_second": 0.185,
      "step": 1970
    },
    {
      "epoch": 0.77997625643055,
      "grad_norm": 0.8549203354868359,
      "learning_rate": 1.4039408522379683e-05,
      "loss": 0.1463,
      "step": 1971
    },
    {
      "epoch": 0.7803719825880491,
      "grad_norm": 0.6225642813207956,
      "learning_rate": 1.3991449234362246e-05,
      "loss": 0.1428,
      "step": 1972
    },
    {
      "epoch": 0.7807677087455481,
      "grad_norm": 0.5159406799915701,
      "learning_rate": 1.3943558672871858e-05,
      "loss": 0.0816,
      "step": 1973
    },
    {
      "epoch": 0.7811634349030471,
      "grad_norm": 0.5639015060064828,
      "learning_rate": 1.3895736929313274e-05,
      "loss": 0.1078,
      "step": 1974
    },
    {
      "epoch": 0.7815591610605461,
      "grad_norm": 0.793573796829072,
      "learning_rate": 1.3847984094959842e-05,
      "loss": 0.1688,
      "step": 1975
    },
    {
      "epoch": 0.7815591610605461,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.925971470732907,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.2577962279319763,
      "eval_runtime": 5.1356,
      "eval_samples_per_second": 5.842,
      "eval_steps_per_second": 0.195,
      "step": 1975
    },
    {
      "epoch": 0.7819548872180451,
      "grad_norm": 0.59123941597089,
      "learning_rate": 1.3800300260953442e-05,
      "loss": 0.1173,
      "step": 1976
    },
    {
      "epoch": 0.7823506133755441,
      "grad_norm": 1.0013061176765425,
      "learning_rate": 1.3752685518304243e-05,
      "loss": 0.1589,
      "step": 1977
    },
    {
      "epoch": 0.7827463395330432,
      "grad_norm": 1.1049300686769525,
      "learning_rate": 1.3705139957890561e-05,
      "loss": 0.2629,
      "step": 1978
    },
    {
      "epoch": 0.7831420656905421,
      "grad_norm": 0.7530204201072208,
      "learning_rate": 1.3657663670458631e-05,
      "loss": 0.1239,
      "step": 1979
    },
    {
      "epoch": 0.7835377918480412,
      "grad_norm": 0.7796923357322306,
      "learning_rate": 1.3610256746622501e-05,
      "loss": 0.1458,
      "step": 1980
    },
    {
      "epoch": 0.7835377918480412,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9227742252828333,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.26289063692092896,
      "eval_runtime": 5.1425,
      "eval_samples_per_second": 5.834,
      "eval_steps_per_second": 0.194,
      "step": 1980
    },
    {
      "epoch": 0.7839335180055401,
      "grad_norm": 1.4504959754945084,
      "learning_rate": 1.3562919276863844e-05,
      "loss": 0.1998,
      "step": 1981
    },
    {
      "epoch": 0.7843292441630392,
      "grad_norm": 0.5848446796465915,
      "learning_rate": 1.3515651351531722e-05,
      "loss": 0.0937,
      "step": 1982
    },
    {
      "epoch": 0.7847249703205382,
      "grad_norm": 0.980713451170035,
      "learning_rate": 1.3468453060842512e-05,
      "loss": 0.1161,
      "step": 1983
    },
    {
      "epoch": 0.7851206964780372,
      "grad_norm": 0.9340843265366097,
      "learning_rate": 1.342132449487966e-05,
      "loss": 0.1904,
      "step": 1984
    },
    {
      "epoch": 0.7855164226355362,
      "grad_norm": 0.507401778841016,
      "learning_rate": 1.3374265743593568e-05,
      "loss": 0.0784,
      "step": 1985
    },
    {
      "epoch": 0.7855164226355362,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9244958189867191,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.2591308653354645,
      "eval_runtime": 4.9238,
      "eval_samples_per_second": 6.093,
      "eval_steps_per_second": 0.203,
      "step": 1985
    },
    {
      "epoch": 0.7859121487930352,
      "grad_norm": 0.6527687145253552,
      "learning_rate": 1.3327276896801321e-05,
      "loss": 0.108,
      "step": 1986
    },
    {
      "epoch": 0.7863078749505342,
      "grad_norm": 0.6365441330238715,
      "learning_rate": 1.3280358044186647e-05,
      "loss": 0.1384,
      "step": 1987
    },
    {
      "epoch": 0.7867036011080333,
      "grad_norm": 0.6130895696765969,
      "learning_rate": 1.3233509275299654e-05,
      "loss": 0.1561,
      "step": 1988
    },
    {
      "epoch": 0.7870993272655322,
      "grad_norm": 0.8704377057489306,
      "learning_rate": 1.3186730679556708e-05,
      "loss": 0.1734,
      "step": 1989
    },
    {
      "epoch": 0.7874950534230313,
      "grad_norm": 1.709491894692586,
      "learning_rate": 1.3140022346240216e-05,
      "loss": 0.2139,
      "step": 1990
    },
    {
      "epoch": 0.7874950534230313,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9242498770290212,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.2558431029319763,
      "eval_runtime": 4.9251,
      "eval_samples_per_second": 6.091,
      "eval_steps_per_second": 0.203,
      "step": 1990
    },
    {
      "epoch": 0.7878907795805302,
      "grad_norm": 0.5493831162248407,
      "learning_rate": 1.309338436449845e-05,
      "loss": 0.1189,
      "step": 1991
    },
    {
      "epoch": 0.7882865057380293,
      "grad_norm": 0.864166802128415,
      "learning_rate": 1.304681682334552e-05,
      "loss": 0.1538,
      "step": 1992
    },
    {
      "epoch": 0.7886822318955283,
      "grad_norm": 0.8458759201606376,
      "learning_rate": 1.3000319811660967e-05,
      "loss": 0.1993,
      "step": 1993
    },
    {
      "epoch": 0.7890779580530273,
      "grad_norm": 0.7853835014018782,
      "learning_rate": 1.2953893418189806e-05,
      "loss": 0.1571,
      "step": 1994
    },
    {
      "epoch": 0.7894736842105263,
      "grad_norm": 0.5843454490833054,
      "learning_rate": 1.290753773154218e-05,
      "loss": 0.1119,
      "step": 1995
    },
    {
      "epoch": 0.7894736842105263,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9237579931136252,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.2513834536075592,
      "eval_runtime": 5.2961,
      "eval_samples_per_second": 5.665,
      "eval_steps_per_second": 0.189,
      "step": 1995
    },
    {
      "epoch": 0.7898694103680254,
      "grad_norm": 0.7822997089466512,
      "learning_rate": 1.2861252840193406e-05,
      "loss": 0.1689,
      "step": 1996
    },
    {
      "epoch": 0.7902651365255243,
      "grad_norm": 0.7821233088584514,
      "learning_rate": 1.2815038832483556e-05,
      "loss": 0.1706,
      "step": 1997
    },
    {
      "epoch": 0.7906608626830234,
      "grad_norm": 0.6043844075854122,
      "learning_rate": 1.2768895796617497e-05,
      "loss": 0.1343,
      "step": 1998
    },
    {
      "epoch": 0.7910565888405223,
      "grad_norm": 0.872936651029278,
      "learning_rate": 1.2722823820664575e-05,
      "loss": 0.1668,
      "step": 1999
    },
    {
      "epoch": 0.7914523149980214,
      "grad_norm": 1.4632565068427705,
      "learning_rate": 1.2676822992558556e-05,
      "loss": 0.1187,
      "step": 2000
    },
    {
      "epoch": 0.7914523149980214,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9232661091982292,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.2472330778837204,
      "eval_runtime": 4.9012,
      "eval_samples_per_second": 6.121,
      "eval_steps_per_second": 0.204,
      "step": 2000
    },
    {
      "epoch": 0.7918480411555204,
      "grad_norm": 1.2159335608037602,
      "learning_rate": 1.2630893400097404e-05,
      "loss": 0.1981,
      "step": 2001
    },
    {
      "epoch": 0.7922437673130194,
      "grad_norm": 1.7062302998592431,
      "learning_rate": 1.2585035130943096e-05,
      "loss": 0.1501,
      "step": 2002
    },
    {
      "epoch": 0.7926394934705184,
      "grad_norm": 0.7775123439925503,
      "learning_rate": 1.2539248272621501e-05,
      "loss": 0.1599,
      "step": 2003
    },
    {
      "epoch": 0.7930352196280174,
      "grad_norm": 0.8235827260709309,
      "learning_rate": 1.2493532912522193e-05,
      "loss": 0.1266,
      "step": 2004
    },
    {
      "epoch": 0.7934309457855164,
      "grad_norm": 0.7499589084292341,
      "learning_rate": 1.2447889137898293e-05,
      "loss": 0.1289,
      "step": 2005
    },
    {
      "epoch": 0.7934309457855164,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9242498770290212,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.24602864682674408,
      "eval_runtime": 4.5702,
      "eval_samples_per_second": 6.564,
      "eval_steps_per_second": 0.219,
      "step": 2005
    },
    {
      "epoch": 0.7938266719430155,
      "grad_norm": 0.7595382472156097,
      "learning_rate": 1.2402317035866251e-05,
      "loss": 0.1704,
      "step": 2006
    },
    {
      "epoch": 0.7942223981005144,
      "grad_norm": 0.9338154936869791,
      "learning_rate": 1.2356816693405766e-05,
      "loss": 0.2018,
      "step": 2007
    },
    {
      "epoch": 0.7946181242580135,
      "grad_norm": 0.6871552963197894,
      "learning_rate": 1.2311388197359569e-05,
      "loss": 0.1872,
      "step": 2008
    },
    {
      "epoch": 0.7950138504155124,
      "grad_norm": 0.6203918289978421,
      "learning_rate": 1.2266031634433223e-05,
      "loss": 0.109,
      "step": 2009
    },
    {
      "epoch": 0.7954095765730115,
      "grad_norm": 0.8827216785502419,
      "learning_rate": 1.222074709119505e-05,
      "loss": 0.2012,
      "step": 2010
    },
    {
      "epoch": 0.7954095765730115,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9244958189867191,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.24965819716453552,
      "eval_runtime": 5.2063,
      "eval_samples_per_second": 5.762,
      "eval_steps_per_second": 0.192,
      "step": 2010
    },
    {
      "epoch": 0.7958053027305105,
      "grad_norm": 0.97193578128228,
      "learning_rate": 1.2175534654075888e-05,
      "loss": 0.1743,
      "step": 2011
    },
    {
      "epoch": 0.7962010288880095,
      "grad_norm": 1.1195058843861392,
      "learning_rate": 1.2130394409368967e-05,
      "loss": 0.1636,
      "step": 2012
    },
    {
      "epoch": 0.7965967550455085,
      "grad_norm": 0.7883575969046394,
      "learning_rate": 1.20853264432297e-05,
      "loss": 0.1622,
      "step": 2013
    },
    {
      "epoch": 0.7969924812030075,
      "grad_norm": 0.9987726628839008,
      "learning_rate": 1.2040330841675573e-05,
      "loss": 0.1958,
      "step": 2014
    },
    {
      "epoch": 0.7973882073605065,
      "grad_norm": 0.5409254273681513,
      "learning_rate": 1.1995407690585953e-05,
      "loss": 0.0951,
      "step": 2015
    },
    {
      "epoch": 0.7973882073605065,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.924987702902115,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.25013020634651184,
      "eval_runtime": 4.9832,
      "eval_samples_per_second": 6.02,
      "eval_steps_per_second": 0.201,
      "step": 2015
    },
    {
      "epoch": 0.7977839335180056,
      "grad_norm": 0.7525615904204186,
      "learning_rate": 1.1950557075701941e-05,
      "loss": 0.1626,
      "step": 2016
    },
    {
      "epoch": 0.7981796596755045,
      "grad_norm": 1.0528606270817817,
      "learning_rate": 1.190577908262614e-05,
      "loss": 0.1498,
      "step": 2017
    },
    {
      "epoch": 0.7985753858330036,
      "grad_norm": 0.6237560941869832,
      "learning_rate": 1.1861073796822608e-05,
      "loss": 0.1182,
      "step": 2018
    },
    {
      "epoch": 0.7989711119905025,
      "grad_norm": 0.9112880172178663,
      "learning_rate": 1.181644130361661e-05,
      "loss": 0.1528,
      "step": 2019
    },
    {
      "epoch": 0.7993668381480016,
      "grad_norm": 0.7445839218103913,
      "learning_rate": 1.1771881688194452e-05,
      "loss": 0.1273,
      "step": 2020
    },
    {
      "epoch": 0.7993668381480016,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9249877029021151,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.25393879413604736,
      "eval_runtime": 4.9475,
      "eval_samples_per_second": 6.064,
      "eval_steps_per_second": 0.202,
      "step": 2020
    },
    {
      "epoch": 0.7997625643055006,
      "grad_norm": 0.7208918290506557,
      "learning_rate": 1.1727395035603384e-05,
      "loss": 0.1176,
      "step": 2021
    },
    {
      "epoch": 0.8001582904629996,
      "grad_norm": 0.5477988096303354,
      "learning_rate": 1.1682981430751378e-05,
      "loss": 0.0767,
      "step": 2022
    },
    {
      "epoch": 0.8005540166204986,
      "grad_norm": 0.6821008389713935,
      "learning_rate": 1.1638640958406999e-05,
      "loss": 0.096,
      "step": 2023
    },
    {
      "epoch": 0.8009497427779976,
      "grad_norm": 0.6100701984851055,
      "learning_rate": 1.1594373703199195e-05,
      "loss": 0.1049,
      "step": 2024
    },
    {
      "epoch": 0.8013454689354966,
      "grad_norm": 0.667826943129846,
      "learning_rate": 1.1550179749617219e-05,
      "loss": 0.1155,
      "step": 2025
    },
    {
      "epoch": 0.8013454689354966,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9264633546483029,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.2593587338924408,
      "eval_runtime": 4.986,
      "eval_samples_per_second": 6.017,
      "eval_steps_per_second": 0.201,
      "step": 2025
    },
    {
      "epoch": 0.8017411950929957,
      "grad_norm": 0.6030424098621106,
      "learning_rate": 1.1506059182010393e-05,
      "loss": 0.109,
      "step": 2026
    },
    {
      "epoch": 0.8021369212504946,
      "grad_norm": 1.3049532919132147,
      "learning_rate": 1.1462012084587964e-05,
      "loss": 0.2157,
      "step": 2027
    },
    {
      "epoch": 0.8025326474079937,
      "grad_norm": 0.7881433236462037,
      "learning_rate": 1.141803854141898e-05,
      "loss": 0.1465,
      "step": 2028
    },
    {
      "epoch": 0.8029283735654926,
      "grad_norm": 0.8626606972084313,
      "learning_rate": 1.1374138636432053e-05,
      "loss": 0.1964,
      "step": 2029
    },
    {
      "epoch": 0.8033240997229917,
      "grad_norm": 0.616377966727902,
      "learning_rate": 1.1330312453415332e-05,
      "loss": 0.082,
      "step": 2030
    },
    {
      "epoch": 0.8033240997229917,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.926463354648303,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.2610026001930237,
      "eval_runtime": 4.8328,
      "eval_samples_per_second": 6.208,
      "eval_steps_per_second": 0.207,
      "step": 2030
    },
    {
      "epoch": 0.8037198258804907,
      "grad_norm": 1.1166740805353914,
      "learning_rate": 1.1286560076016172e-05,
      "loss": 0.1568,
      "step": 2031
    },
    {
      "epoch": 0.8041155520379897,
      "grad_norm": 0.7428075027467106,
      "learning_rate": 1.1242881587741127e-05,
      "loss": 0.1442,
      "step": 2032
    },
    {
      "epoch": 0.8045112781954887,
      "grad_norm": 1.3561411167083395,
      "learning_rate": 1.1199277071955649e-05,
      "loss": 0.2127,
      "step": 2033
    },
    {
      "epoch": 0.8049070043529878,
      "grad_norm": 0.5498468389795856,
      "learning_rate": 1.1155746611884105e-05,
      "loss": 0.1059,
      "step": 2034
    },
    {
      "epoch": 0.8053027305104867,
      "grad_norm": 0.9845277519022093,
      "learning_rate": 1.1112290290609445e-05,
      "loss": 0.1615,
      "step": 2035
    },
    {
      "epoch": 0.8053027305104867,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9254795868175111,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.2564453184604645,
      "eval_runtime": 4.9516,
      "eval_samples_per_second": 6.059,
      "eval_steps_per_second": 0.202,
      "step": 2035
    },
    {
      "epoch": 0.8056984566679858,
      "grad_norm": 1.0152567903269543,
      "learning_rate": 1.1068908191073123e-05,
      "loss": 0.2132,
      "step": 2036
    },
    {
      "epoch": 0.8060941828254847,
      "grad_norm": 1.0387595605725022,
      "learning_rate": 1.1025600396074954e-05,
      "loss": 0.1962,
      "step": 2037
    },
    {
      "epoch": 0.8064899089829838,
      "grad_norm": 0.7421075483289555,
      "learning_rate": 1.0982366988272924e-05,
      "loss": 0.1584,
      "step": 2038
    },
    {
      "epoch": 0.8068856351404828,
      "grad_norm": 0.7166010187670492,
      "learning_rate": 1.0939208050183064e-05,
      "loss": 0.1193,
      "step": 2039
    },
    {
      "epoch": 0.8072813612979818,
      "grad_norm": 0.6048831257563484,
      "learning_rate": 1.089612366417922e-05,
      "loss": 0.1129,
      "step": 2040
    },
    {
      "epoch": 0.8072813612979818,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.925725528775209,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.252197265625,
      "eval_runtime": 5.0329,
      "eval_samples_per_second": 5.961,
      "eval_steps_per_second": 0.199,
      "step": 2040
    },
    {
      "epoch": 0.8076770874554808,
      "grad_norm": 0.7587510002970207,
      "learning_rate": 1.085311391249299e-05,
      "loss": 0.1454,
      "step": 2041
    },
    {
      "epoch": 0.8080728136129798,
      "grad_norm": 0.7468303907419029,
      "learning_rate": 1.0810178877213517e-05,
      "loss": 0.1025,
      "step": 2042
    },
    {
      "epoch": 0.8084685397704788,
      "grad_norm": 0.7064439414203635,
      "learning_rate": 1.0767318640287343e-05,
      "loss": 0.1501,
      "step": 2043
    },
    {
      "epoch": 0.8088642659279779,
      "grad_norm": 1.0400952906128809,
      "learning_rate": 1.0724533283518206e-05,
      "loss": 0.1403,
      "step": 2044
    },
    {
      "epoch": 0.8092599920854768,
      "grad_norm": 0.8435264427427267,
      "learning_rate": 1.0681822888566984e-05,
      "loss": 0.154,
      "step": 2045
    },
    {
      "epoch": 0.8092599920854768,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.925233644859813,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.25192058086395264,
      "eval_runtime": 5.1501,
      "eval_samples_per_second": 5.825,
      "eval_steps_per_second": 0.194,
      "step": 2045
    },
    {
      "epoch": 0.8096557182429759,
      "grad_norm": 0.6532493036336242,
      "learning_rate": 1.0639187536951462e-05,
      "loss": 0.0975,
      "step": 2046
    },
    {
      "epoch": 0.8100514444004748,
      "grad_norm": 1.1718841644349773,
      "learning_rate": 1.0596627310046165e-05,
      "loss": 0.1878,
      "step": 2047
    },
    {
      "epoch": 0.8104471705579739,
      "grad_norm": 0.7814491271826595,
      "learning_rate": 1.0554142289082275e-05,
      "loss": 0.0938,
      "step": 2048
    },
    {
      "epoch": 0.8108428967154729,
      "grad_norm": 0.4834294526026405,
      "learning_rate": 1.0511732555147419e-05,
      "loss": 0.0847,
      "step": 2049
    },
    {
      "epoch": 0.8112386228729719,
      "grad_norm": 0.6610897614277756,
      "learning_rate": 1.0469398189185542e-05,
      "loss": 0.1211,
      "step": 2050
    },
    {
      "epoch": 0.8112386228729719,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.926217412690605,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.25231119990348816,
      "eval_runtime": 4.9707,
      "eval_samples_per_second": 6.035,
      "eval_steps_per_second": 0.201,
      "step": 2050
    },
    {
      "epoch": 0.8116343490304709,
      "grad_norm": 0.806992161033756,
      "learning_rate": 1.0427139271996705e-05,
      "loss": 0.1514,
      "step": 2051
    },
    {
      "epoch": 0.8120300751879699,
      "grad_norm": 1.4900571455276737,
      "learning_rate": 1.0384955884237003e-05,
      "loss": 0.1688,
      "step": 2052
    },
    {
      "epoch": 0.8124258013454689,
      "grad_norm": 0.9674595811241086,
      "learning_rate": 1.0342848106418368e-05,
      "loss": 0.1633,
      "step": 2053
    },
    {
      "epoch": 0.812821527502968,
      "grad_norm": 1.1059680583321914,
      "learning_rate": 1.0300816018908393e-05,
      "loss": 0.1611,
      "step": 2054
    },
    {
      "epoch": 0.8132172536604669,
      "grad_norm": 0.8479407872481435,
      "learning_rate": 1.0258859701930246e-05,
      "loss": 0.1664,
      "step": 2055
    },
    {
      "epoch": 0.8132172536604669,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9247417609444171,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.2554687559604645,
      "eval_runtime": 4.9025,
      "eval_samples_per_second": 6.119,
      "eval_steps_per_second": 0.204,
      "step": 2055
    },
    {
      "epoch": 0.813612979817966,
      "grad_norm": 0.6455428097935911,
      "learning_rate": 1.0216979235562451e-05,
      "loss": 0.1379,
      "step": 2056
    },
    {
      "epoch": 0.8140087059754649,
      "grad_norm": 1.0519160034922128,
      "learning_rate": 1.0175174699738793e-05,
      "loss": 0.1651,
      "step": 2057
    },
    {
      "epoch": 0.814404432132964,
      "grad_norm": 1.2672967133153847,
      "learning_rate": 1.013344617424809e-05,
      "loss": 0.2005,
      "step": 2058
    },
    {
      "epoch": 0.814800158290463,
      "grad_norm": 0.7492988028006651,
      "learning_rate": 1.0091793738734113e-05,
      "loss": 0.1371,
      "step": 2059
    },
    {
      "epoch": 0.815195884447962,
      "grad_norm": 2.2548333046025455,
      "learning_rate": 1.005021747269541e-05,
      "loss": 0.1891,
      "step": 2060
    },
    {
      "epoch": 0.815195884447962,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9249877029021151,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.2577311098575592,
      "eval_runtime": 4.9914,
      "eval_samples_per_second": 6.01,
      "eval_steps_per_second": 0.2,
      "step": 2060
    },
    {
      "epoch": 0.815591610605461,
      "grad_norm": 0.7544599852415839,
      "learning_rate": 1.000871745548515e-05,
      "loss": 0.144,
      "step": 2061
    },
    {
      "epoch": 0.81598733676296,
      "grad_norm": 1.030494436437544,
      "learning_rate": 9.96729376631096e-06,
      "loss": 0.19,
      "step": 2062
    },
    {
      "epoch": 0.816383062920459,
      "grad_norm": 0.8587497085041976,
      "learning_rate": 9.92594648423476e-06,
      "loss": 0.1804,
      "step": 2063
    },
    {
      "epoch": 0.8167787890779581,
      "grad_norm": 1.663810349146248,
      "learning_rate": 9.884675688172723e-06,
      "loss": 0.1916,
      "step": 2064
    },
    {
      "epoch": 0.817174515235457,
      "grad_norm": 0.5623854137617148,
      "learning_rate": 9.843481456894948e-06,
      "loss": 0.0951,
      "step": 2065
    },
    {
      "epoch": 0.817174515235457,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.926955238563699,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.2550455629825592,
      "eval_runtime": 5.6103,
      "eval_samples_per_second": 5.347,
      "eval_steps_per_second": 0.178,
      "step": 2065
    },
    {
      "epoch": 0.8175702413929561,
      "grad_norm": 0.5142510638461822,
      "learning_rate": 9.802363869025467e-06,
      "loss": 0.0954,
      "step": 2066
    },
    {
      "epoch": 0.817965967550455,
      "grad_norm": 0.7585483478680642,
      "learning_rate": 9.761323003041955e-06,
      "loss": 0.1094,
      "step": 2067
    },
    {
      "epoch": 0.8183616937079541,
      "grad_norm": 1.225766554796266,
      "learning_rate": 9.720358937275764e-06,
      "loss": 0.1631,
      "step": 2068
    },
    {
      "epoch": 0.8187574198654531,
      "grad_norm": 0.7371777759738903,
      "learning_rate": 9.67947174991155e-06,
      "loss": 0.1586,
      "step": 2069
    },
    {
      "epoch": 0.8191531460229521,
      "grad_norm": 0.8623641815453774,
      "learning_rate": 9.638661518987324e-06,
      "loss": 0.1303,
      "step": 2070
    },
    {
      "epoch": 0.8191531460229521,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9262174126906051,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.2561197876930237,
      "eval_runtime": 5.0984,
      "eval_samples_per_second": 5.884,
      "eval_steps_per_second": 0.196,
      "step": 2070
    },
    {
      "epoch": 0.8195488721804511,
      "grad_norm": 0.8639706462083075,
      "learning_rate": 9.59792832239415e-06,
      "loss": 0.1716,
      "step": 2071
    },
    {
      "epoch": 0.8199445983379502,
      "grad_norm": 0.7357460994001227,
      "learning_rate": 9.557272237876102e-06,
      "loss": 0.1094,
      "step": 2072
    },
    {
      "epoch": 0.8203403244954491,
      "grad_norm": 0.6921685237752392,
      "learning_rate": 9.516693343030093e-06,
      "loss": 0.1311,
      "step": 2073
    },
    {
      "epoch": 0.8207360506529482,
      "grad_norm": 0.6464627634756616,
      "learning_rate": 9.476191715305649e-06,
      "loss": 0.1251,
      "step": 2074
    },
    {
      "epoch": 0.8211317768104471,
      "grad_norm": 0.7359707215084921,
      "learning_rate": 9.435767432004877e-06,
      "loss": 0.1099,
      "step": 2075
    },
    {
      "epoch": 0.8211317768104471,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9252336448598131,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.25467121601104736,
      "eval_runtime": 5.0111,
      "eval_samples_per_second": 5.987,
      "eval_steps_per_second": 0.2,
      "step": 2075
    },
    {
      "epoch": 0.8215275029679462,
      "grad_norm": 0.6314287788934156,
      "learning_rate": 9.395420570282248e-06,
      "loss": 0.1034,
      "step": 2076
    },
    {
      "epoch": 0.8219232291254452,
      "grad_norm": 0.6715091978240232,
      "learning_rate": 9.35515120714447e-06,
      "loss": 0.1415,
      "step": 2077
    },
    {
      "epoch": 0.8223189552829442,
      "grad_norm": 0.915094253402362,
      "learning_rate": 9.314959419450303e-06,
      "loss": 0.1779,
      "step": 2078
    },
    {
      "epoch": 0.8227146814404432,
      "grad_norm": 0.7526590576605944,
      "learning_rate": 9.274845283910493e-06,
      "loss": 0.1251,
      "step": 2079
    },
    {
      "epoch": 0.8231104075979422,
      "grad_norm": 1.1147181041655092,
      "learning_rate": 9.234808877087554e-06,
      "loss": 0.1822,
      "step": 2080
    },
    {
      "epoch": 0.8231104075979422,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9244958189867191,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.25379231572151184,
      "eval_runtime": 5.1928,
      "eval_samples_per_second": 5.777,
      "eval_steps_per_second": 0.193,
      "step": 2080
    },
    {
      "epoch": 0.8235061337554412,
      "grad_norm": 0.7690596881182328,
      "learning_rate": 9.194850275395633e-06,
      "loss": 0.1035,
      "step": 2081
    },
    {
      "epoch": 0.8239018599129403,
      "grad_norm": 0.7374342560416068,
      "learning_rate": 9.154969555100396e-06,
      "loss": 0.1576,
      "step": 2082
    },
    {
      "epoch": 0.8242975860704392,
      "grad_norm": 0.562150260497393,
      "learning_rate": 9.115166792318858e-06,
      "loss": 0.1009,
      "step": 2083
    },
    {
      "epoch": 0.8246933122279383,
      "grad_norm": 0.9219526645261468,
      "learning_rate": 9.075442063019263e-06,
      "loss": 0.1908,
      "step": 2084
    },
    {
      "epoch": 0.8250890383854372,
      "grad_norm": 0.8119692146377178,
      "learning_rate": 9.035795443020873e-06,
      "loss": 0.1216,
      "step": 2085
    },
    {
      "epoch": 0.8250890383854372,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9252336448598132,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.24982096254825592,
      "eval_runtime": 5.0949,
      "eval_samples_per_second": 5.888,
      "eval_steps_per_second": 0.196,
      "step": 2085
    },
    {
      "epoch": 0.8254847645429363,
      "grad_norm": 0.6147092131098841,
      "learning_rate": 8.996227007993896e-06,
      "loss": 0.1246,
      "step": 2086
    },
    {
      "epoch": 0.8258804907004353,
      "grad_norm": 1.1117802456809973,
      "learning_rate": 8.956736833459328e-06,
      "loss": 0.1697,
      "step": 2087
    },
    {
      "epoch": 0.8262762168579343,
      "grad_norm": 0.9179634359224617,
      "learning_rate": 8.91732499478879e-06,
      "loss": 0.1537,
      "step": 2088
    },
    {
      "epoch": 0.8266719430154333,
      "grad_norm": 1.1292164316772388,
      "learning_rate": 8.877991567204352e-06,
      "loss": 0.2151,
      "step": 2089
    },
    {
      "epoch": 0.8270676691729323,
      "grad_norm": 1.1027373500259001,
      "learning_rate": 8.838736625778476e-06,
      "loss": 0.2189,
      "step": 2090
    },
    {
      "epoch": 0.8270676691729323,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9262174126906051,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.2473958283662796,
      "eval_runtime": 4.9397,
      "eval_samples_per_second": 6.073,
      "eval_steps_per_second": 0.202,
      "step": 2090
    },
    {
      "epoch": 0.8274633953304313,
      "grad_norm": 0.9769606669583665,
      "learning_rate": 8.799560245433814e-06,
      "loss": 0.1741,
      "step": 2091
    },
    {
      "epoch": 0.8278591214879304,
      "grad_norm": 0.826800540837405,
      "learning_rate": 8.76046250094304e-06,
      "loss": 0.1778,
      "step": 2092
    },
    {
      "epoch": 0.8282548476454293,
      "grad_norm": 0.8446785142808437,
      "learning_rate": 8.721443466928786e-06,
      "loss": 0.15,
      "step": 2093
    },
    {
      "epoch": 0.8286505738029284,
      "grad_norm": 0.8633231052020104,
      "learning_rate": 8.68250321786343e-06,
      "loss": 0.1485,
      "step": 2094
    },
    {
      "epoch": 0.8290462999604273,
      "grad_norm": 1.4517127690016542,
      "learning_rate": 8.643641828069005e-06,
      "loss": 0.23,
      "step": 2095
    },
    {
      "epoch": 0.8290462999604273,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9254795868175111,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.24484048783779144,
      "eval_runtime": 4.8268,
      "eval_samples_per_second": 6.215,
      "eval_steps_per_second": 0.207,
      "step": 2095
    },
    {
      "epoch": 0.8294420261179264,
      "grad_norm": 0.7412021202111235,
      "learning_rate": 8.604859371716994e-06,
      "loss": 0.1363,
      "step": 2096
    },
    {
      "epoch": 0.8298377522754254,
      "grad_norm": 1.0989623479656467,
      "learning_rate": 8.56615592282825e-06,
      "loss": 0.1436,
      "step": 2097
    },
    {
      "epoch": 0.8302334784329244,
      "grad_norm": 0.8220607455591905,
      "learning_rate": 8.527531555272849e-06,
      "loss": 0.1411,
      "step": 2098
    },
    {
      "epoch": 0.8306292045904234,
      "grad_norm": 0.8067191689632638,
      "learning_rate": 8.488986342769883e-06,
      "loss": 0.1595,
      "step": 2099
    },
    {
      "epoch": 0.8310249307479224,
      "grad_norm": 0.7563790437155197,
      "learning_rate": 8.450520358887415e-06,
      "loss": 0.1472,
      "step": 2100
    },
    {
      "epoch": 0.8310249307479224,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9257255287752091,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.24537760019302368,
      "eval_runtime": 4.8962,
      "eval_samples_per_second": 6.127,
      "eval_steps_per_second": 0.204,
      "step": 2100
    },
    {
      "epoch": 0.8314206569054214,
      "grad_norm": 0.6775786821660054,
      "learning_rate": 8.41213367704224e-06,
      "loss": 0.1571,
      "step": 2101
    },
    {
      "epoch": 0.8318163830629205,
      "grad_norm": 0.7160643965541019,
      "learning_rate": 8.37382637049987e-06,
      "loss": 0.1393,
      "step": 2102
    },
    {
      "epoch": 0.8322121092204194,
      "grad_norm": 0.7854208408469532,
      "learning_rate": 8.335598512374243e-06,
      "loss": 0.1083,
      "step": 2103
    },
    {
      "epoch": 0.8326078353779185,
      "grad_norm": 0.8826276542781595,
      "learning_rate": 8.297450175627714e-06,
      "loss": 0.2523,
      "step": 2104
    },
    {
      "epoch": 0.8330035615354174,
      "grad_norm": 0.8989597316696435,
      "learning_rate": 8.259381433070801e-06,
      "loss": 0.1603,
      "step": 2105
    },
    {
      "epoch": 0.8330035615354174,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9247417609444172,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.24734701216220856,
      "eval_runtime": 5.2698,
      "eval_samples_per_second": 5.693,
      "eval_steps_per_second": 0.19,
      "step": 2105
    },
    {
      "epoch": 0.8333992876929165,
      "grad_norm": 1.0773810788659635,
      "learning_rate": 8.221392357362211e-06,
      "loss": 0.161,
      "step": 2106
    },
    {
      "epoch": 0.8337950138504155,
      "grad_norm": 0.6887314492292407,
      "learning_rate": 8.183483021008498e-06,
      "loss": 0.1197,
      "step": 2107
    },
    {
      "epoch": 0.8341907400079145,
      "grad_norm": 0.6720322414991993,
      "learning_rate": 8.145653496364054e-06,
      "loss": 0.1289,
      "step": 2108
    },
    {
      "epoch": 0.8345864661654135,
      "grad_norm": 0.7516227336586865,
      "learning_rate": 8.107903855630956e-06,
      "loss": 0.0912,
      "step": 2109
    },
    {
      "epoch": 0.8349821923229126,
      "grad_norm": 0.9843758540628557,
      "learning_rate": 8.070234170858803e-06,
      "loss": 0.2199,
      "step": 2110
    },
    {
      "epoch": 0.8349821923229126,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.926463354648303,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.25068360567092896,
      "eval_runtime": 5.0225,
      "eval_samples_per_second": 5.973,
      "eval_steps_per_second": 0.199,
      "step": 2110
    },
    {
      "epoch": 0.8353779184804115,
      "grad_norm": 0.8115574881340666,
      "learning_rate": 8.032644513944609e-06,
      "loss": 0.1126,
      "step": 2111
    },
    {
      "epoch": 0.8357736446379106,
      "grad_norm": 0.6937953982103526,
      "learning_rate": 7.995134956632599e-06,
      "loss": 0.1436,
      "step": 2112
    },
    {
      "epoch": 0.8361693707954095,
      "grad_norm": 0.9762508996485155,
      "learning_rate": 7.957705570514163e-06,
      "loss": 0.1313,
      "step": 2113
    },
    {
      "epoch": 0.8365650969529086,
      "grad_norm": 0.5649270647356713,
      "learning_rate": 7.920356427027648e-06,
      "loss": 0.0988,
      "step": 2114
    },
    {
      "epoch": 0.8369608231104076,
      "grad_norm": 0.6211842179744085,
      "learning_rate": 7.883087597458278e-06,
      "loss": 0.0888,
      "step": 2115
    },
    {
      "epoch": 0.8369608231104076,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.927201180521397,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.2557942569255829,
      "eval_runtime": 5.1158,
      "eval_samples_per_second": 5.864,
      "eval_steps_per_second": 0.195,
      "step": 2115
    },
    {
      "epoch": 0.8373565492679066,
      "grad_norm": 0.5638575184828931,
      "learning_rate": 7.845899152937946e-06,
      "loss": 0.1268,
      "step": 2116
    },
    {
      "epoch": 0.8377522754254056,
      "grad_norm": 0.9739319581178364,
      "learning_rate": 7.808791164445156e-06,
      "loss": 0.1488,
      "step": 2117
    },
    {
      "epoch": 0.8381480015829046,
      "grad_norm": 0.7013382631114277,
      "learning_rate": 7.771763702804852e-06,
      "loss": 0.076,
      "step": 2118
    },
    {
      "epoch": 0.8385437277404036,
      "grad_norm": 0.8021500785045441,
      "learning_rate": 7.734816838688248e-06,
      "loss": 0.208,
      "step": 2119
    },
    {
      "epoch": 0.8389394538979027,
      "grad_norm": 1.1908877134783324,
      "learning_rate": 7.697950642612756e-06,
      "loss": 0.1762,
      "step": 2120
    },
    {
      "epoch": 0.8389394538979027,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9279390063944909,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.2587727904319763,
      "eval_runtime": 5.1276,
      "eval_samples_per_second": 5.851,
      "eval_steps_per_second": 0.195,
      "step": 2120
    },
    {
      "epoch": 0.8393351800554016,
      "grad_norm": 0.5503842550393047,
      "learning_rate": 7.661165184941832e-06,
      "loss": 0.0839,
      "step": 2121
    },
    {
      "epoch": 0.8397309062129007,
      "grad_norm": 0.6384443358340002,
      "learning_rate": 7.62446053588482e-06,
      "loss": 0.1111,
      "step": 2122
    },
    {
      "epoch": 0.8401266323703996,
      "grad_norm": 0.4615175798116959,
      "learning_rate": 7.587836765496819e-06,
      "loss": 0.0735,
      "step": 2123
    },
    {
      "epoch": 0.8405223585278987,
      "grad_norm": 0.5694659307292461,
      "learning_rate": 7.551293943678583e-06,
      "loss": 0.131,
      "step": 2124
    },
    {
      "epoch": 0.8409180846853977,
      "grad_norm": 0.7682571662273591,
      "learning_rate": 7.51483214017637e-06,
      "loss": 0.1224,
      "step": 2125
    },
    {
      "epoch": 0.8409180846853977,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9274471224790948,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.26022136211395264,
      "eval_runtime": 5.1062,
      "eval_samples_per_second": 5.875,
      "eval_steps_per_second": 0.196,
      "step": 2125
    },
    {
      "epoch": 0.8413138108428967,
      "grad_norm": 0.8302699393949629,
      "learning_rate": 7.478451424581761e-06,
      "loss": 0.2036,
      "step": 2126
    },
    {
      "epoch": 0.8417095370003957,
      "grad_norm": 0.7362658788824431,
      "learning_rate": 7.442151866331631e-06,
      "loss": 0.1356,
      "step": 2127
    },
    {
      "epoch": 0.8421052631578947,
      "grad_norm": 0.8693907983629763,
      "learning_rate": 7.405933534707915e-06,
      "loss": 0.1336,
      "step": 2128
    },
    {
      "epoch": 0.8425009893153937,
      "grad_norm": 0.7394769841827123,
      "learning_rate": 7.3697964988375444e-06,
      "loss": 0.1556,
      "step": 2129
    },
    {
      "epoch": 0.8428967154728928,
      "grad_norm": 1.1060012864887494,
      "learning_rate": 7.3337408276922594e-06,
      "loss": 0.1754,
      "step": 2130
    },
    {
      "epoch": 0.8428967154728928,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9286768322675848,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.26280924677848816,
      "eval_runtime": 5.2209,
      "eval_samples_per_second": 5.746,
      "eval_steps_per_second": 0.192,
      "step": 2130
    },
    {
      "epoch": 0.8432924416303917,
      "grad_norm": 1.23259057272544,
      "learning_rate": 7.2977665900885285e-06,
      "loss": 0.148,
      "step": 2131
    },
    {
      "epoch": 0.8436881677878908,
      "grad_norm": 1.010807828529088,
      "learning_rate": 7.26187385468739e-06,
      "loss": 0.1671,
      "step": 2132
    },
    {
      "epoch": 0.8440838939453897,
      "grad_norm": 0.8504317817630274,
      "learning_rate": 7.226062689994328e-06,
      "loss": 0.1427,
      "step": 2133
    },
    {
      "epoch": 0.8444796201028888,
      "grad_norm": 0.892470546749732,
      "learning_rate": 7.190333164359137e-06,
      "loss": 0.1585,
      "step": 2134
    },
    {
      "epoch": 0.8448753462603878,
      "grad_norm": 0.8753497664220152,
      "learning_rate": 7.154685345975759e-06,
      "loss": 0.1713,
      "step": 2135
    },
    {
      "epoch": 0.8448753462603878,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.926709296606001,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.2620605528354645,
      "eval_runtime": 4.96,
      "eval_samples_per_second": 6.048,
      "eval_steps_per_second": 0.202,
      "step": 2135
    },
    {
      "epoch": 0.8452710724178868,
      "grad_norm": 0.6446270923407261,
      "learning_rate": 7.119119302882271e-06,
      "loss": 0.0904,
      "step": 2136
    },
    {
      "epoch": 0.8456667985753858,
      "grad_norm": 0.7249639672901916,
      "learning_rate": 7.083635102960584e-06,
      "loss": 0.1562,
      "step": 2137
    },
    {
      "epoch": 0.8460625247328848,
      "grad_norm": 0.9462197153145087,
      "learning_rate": 7.048232813936467e-06,
      "loss": 0.2016,
      "step": 2138
    },
    {
      "epoch": 0.8464582508903838,
      "grad_norm": 0.6935257323360903,
      "learning_rate": 7.012912503379287e-06,
      "loss": 0.1542,
      "step": 2139
    },
    {
      "epoch": 0.8468539770478829,
      "grad_norm": 0.8733388039091272,
      "learning_rate": 6.977674238702036e-06,
      "loss": 0.1818,
      "step": 2140
    },
    {
      "epoch": 0.8468539770478829,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9274471224790949,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.2576334774494171,
      "eval_runtime": 5.2403,
      "eval_samples_per_second": 5.725,
      "eval_steps_per_second": 0.191,
      "step": 2140
    },
    {
      "epoch": 0.8472497032053818,
      "grad_norm": 0.7213248512509336,
      "learning_rate": 6.942518087161026e-06,
      "loss": 0.1133,
      "step": 2141
    },
    {
      "epoch": 0.8476454293628809,
      "grad_norm": 1.0255099206805611,
      "learning_rate": 6.907444115855899e-06,
      "loss": 0.1418,
      "step": 2142
    },
    {
      "epoch": 0.84804115552038,
      "grad_norm": 1.0709456015733834,
      "learning_rate": 6.8724523917294e-06,
      "loss": 0.1767,
      "step": 2143
    },
    {
      "epoch": 0.8484368816778789,
      "grad_norm": 1.1103151090448384,
      "learning_rate": 6.837542981567346e-06,
      "loss": 0.2251,
      "step": 2144
    },
    {
      "epoch": 0.848832607835378,
      "grad_norm": 1.2521914879532863,
      "learning_rate": 6.802715951998434e-06,
      "loss": 0.3122,
      "step": 2145
    },
    {
      "epoch": 0.848832607835378,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9274471224790949,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.25133463740348816,
      "eval_runtime": 4.8676,
      "eval_samples_per_second": 6.163,
      "eval_steps_per_second": 0.205,
      "step": 2145
    },
    {
      "epoch": 0.8492283339928769,
      "grad_norm": 0.7554870372865617,
      "learning_rate": 6.767971369494097e-06,
      "loss": 0.175,
      "step": 2146
    },
    {
      "epoch": 0.849624060150376,
      "grad_norm": 0.7835641177117965,
      "learning_rate": 6.733309300368435e-06,
      "loss": 0.1622,
      "step": 2147
    },
    {
      "epoch": 0.850019786307875,
      "grad_norm": 0.7120032472218877,
      "learning_rate": 6.698729810778065e-06,
      "loss": 0.1778,
      "step": 2148
    },
    {
      "epoch": 0.850415512465374,
      "grad_norm": 0.9699253826567068,
      "learning_rate": 6.664232966721995e-06,
      "loss": 0.1405,
      "step": 2149
    },
    {
      "epoch": 0.850811238622873,
      "grad_norm": 0.8291135048711319,
      "learning_rate": 6.629818834041457e-06,
      "loss": 0.1482,
      "step": 2150
    },
    {
      "epoch": 0.850811238622873,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9281849483521888,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.247314453125,
      "eval_runtime": 4.94,
      "eval_samples_per_second": 6.073,
      "eval_steps_per_second": 0.202,
      "step": 2150
    },
    {
      "epoch": 0.8512069647803719,
      "grad_norm": 0.7667815441011655,
      "learning_rate": 6.595487478419859e-06,
      "loss": 0.167,
      "step": 2151
    },
    {
      "epoch": 0.851602690937871,
      "grad_norm": 0.8400647072981933,
      "learning_rate": 6.5612389653826215e-06,
      "loss": 0.1671,
      "step": 2152
    },
    {
      "epoch": 0.85199841709537,
      "grad_norm": 0.472328094688762,
      "learning_rate": 6.527073360296998e-06,
      "loss": 0.076,
      "step": 2153
    },
    {
      "epoch": 0.852394143252869,
      "grad_norm": 0.7510527086655423,
      "learning_rate": 6.492990728372056e-06,
      "loss": 0.1288,
      "step": 2154
    },
    {
      "epoch": 0.852789869410368,
      "grad_norm": 0.9593142317647102,
      "learning_rate": 6.458991134658487e-06,
      "loss": 0.1824,
      "step": 2155
    },
    {
      "epoch": 0.852789869410368,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.927201180521397,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.24428710341453552,
      "eval_runtime": 5.1056,
      "eval_samples_per_second": 5.876,
      "eval_steps_per_second": 0.196,
      "step": 2155
    },
    {
      "epoch": 0.853185595567867,
      "grad_norm": 0.7168452867613544,
      "learning_rate": 6.42507464404849e-06,
      "loss": 0.1361,
      "step": 2156
    },
    {
      "epoch": 0.853581321725366,
      "grad_norm": 0.6620712075071548,
      "learning_rate": 6.391241321275637e-06,
      "loss": 0.152,
      "step": 2157
    },
    {
      "epoch": 0.8539770478828651,
      "grad_norm": 1.1686736300061498,
      "learning_rate": 6.357491230914786e-06,
      "loss": 0.2314,
      "step": 2158
    },
    {
      "epoch": 0.854372774040364,
      "grad_norm": 0.6254742796693729,
      "learning_rate": 6.323824437381931e-06,
      "loss": 0.1135,
      "step": 2159
    },
    {
      "epoch": 0.8547685001978631,
      "grad_norm": 0.6339742142101213,
      "learning_rate": 6.290241004934083e-06,
      "loss": 0.1602,
      "step": 2160
    },
    {
      "epoch": 0.8547685001978631,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9276930644367928,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.24293619394302368,
      "eval_runtime": 5.0464,
      "eval_samples_per_second": 5.945,
      "eval_steps_per_second": 0.198,
      "step": 2160
    },
    {
      "epoch": 0.855164226355362,
      "grad_norm": 0.8965989554043279,
      "learning_rate": 6.256740997669142e-06,
      "loss": 0.157,
      "step": 2161
    },
    {
      "epoch": 0.8555599525128611,
      "grad_norm": 0.7420574497235716,
      "learning_rate": 6.223324479525778e-06,
      "loss": 0.1401,
      "step": 2162
    },
    {
      "epoch": 0.8559556786703602,
      "grad_norm": 0.5258313090846062,
      "learning_rate": 6.18999151428335e-06,
      "loss": 0.0846,
      "step": 2163
    },
    {
      "epoch": 0.8563514048278591,
      "grad_norm": 0.6352766821560462,
      "learning_rate": 6.1567421655616856e-06,
      "loss": 0.1119,
      "step": 2164
    },
    {
      "epoch": 0.8567471309853582,
      "grad_norm": 0.5970107477792718,
      "learning_rate": 6.12357649682106e-06,
      "loss": 0.1297,
      "step": 2165
    },
    {
      "epoch": 0.8567471309853582,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9286768322675848,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.243408203125,
      "eval_runtime": 4.9625,
      "eval_samples_per_second": 6.045,
      "eval_steps_per_second": 0.202,
      "step": 2165
    },
    {
      "epoch": 0.8571428571428571,
      "grad_norm": 0.7606985896810577,
      "learning_rate": 6.090494571362037e-06,
      "loss": 0.1496,
      "step": 2166
    },
    {
      "epoch": 0.8575385833003561,
      "grad_norm": 0.7750190096806079,
      "learning_rate": 6.057496452325346e-06,
      "loss": 0.0999,
      "step": 2167
    },
    {
      "epoch": 0.8579343094578552,
      "grad_norm": 0.8772335193769163,
      "learning_rate": 6.0245822026917256e-06,
      "loss": 0.1461,
      "step": 2168
    },
    {
      "epoch": 0.8583300356153541,
      "grad_norm": 0.7569183635580057,
      "learning_rate": 5.991751885281882e-06,
      "loss": 0.1245,
      "step": 2169
    },
    {
      "epoch": 0.8587257617728532,
      "grad_norm": 0.682711271361936,
      "learning_rate": 5.9590055627563256e-06,
      "loss": 0.102,
      "step": 2170
    },
    {
      "epoch": 0.8587257617728532,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9276930644367928,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.24516601860523224,
      "eval_runtime": 5.0492,
      "eval_samples_per_second": 5.942,
      "eval_steps_per_second": 0.198,
      "step": 2170
    },
    {
      "epoch": 0.8591214879303521,
      "grad_norm": 0.6154470362240125,
      "learning_rate": 5.926343297615216e-06,
      "loss": 0.127,
      "step": 2171
    },
    {
      "epoch": 0.8595172140878512,
      "grad_norm": 0.7120709457614607,
      "learning_rate": 5.893765152198327e-06,
      "loss": 0.1292,
      "step": 2172
    },
    {
      "epoch": 0.8599129402453503,
      "grad_norm": 1.0897337812478771,
      "learning_rate": 5.8612711886848196e-06,
      "loss": 0.2512,
      "step": 2173
    },
    {
      "epoch": 0.8603086664028492,
      "grad_norm": 0.5337095317476303,
      "learning_rate": 5.828861469093266e-06,
      "loss": 0.0957,
      "step": 2174
    },
    {
      "epoch": 0.8607043925603483,
      "grad_norm": 0.6870274668163368,
      "learning_rate": 5.796536055281371e-06,
      "loss": 0.1362,
      "step": 2175
    },
    {
      "epoch": 0.8607043925603483,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9284308903098869,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.2465006560087204,
      "eval_runtime": 5.0078,
      "eval_samples_per_second": 5.991,
      "eval_steps_per_second": 0.2,
      "step": 2175
    },
    {
      "epoch": 0.8611001187178472,
      "grad_norm": 0.9348128866554337,
      "learning_rate": 5.7642950089459805e-06,
      "loss": 0.1882,
      "step": 2176
    },
    {
      "epoch": 0.8614958448753463,
      "grad_norm": 0.755569960760724,
      "learning_rate": 5.7321383916228764e-06,
      "loss": 0.1033,
      "step": 2177
    },
    {
      "epoch": 0.8618915710328453,
      "grad_norm": 0.7282658149840272,
      "learning_rate": 5.700066264686759e-06,
      "loss": 0.1221,
      "step": 2178
    },
    {
      "epoch": 0.8622872971903442,
      "grad_norm": 1.0410815372378537,
      "learning_rate": 5.668078689351009e-06,
      "loss": 0.197,
      "step": 2179
    },
    {
      "epoch": 0.8626830233478433,
      "grad_norm": 0.6811890382487824,
      "learning_rate": 5.636175726667636e-06,
      "loss": 0.1262,
      "step": 2180
    },
    {
      "epoch": 0.8626830233478433,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9299065420560747,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.24768880009651184,
      "eval_runtime": 4.6629,
      "eval_samples_per_second": 6.434,
      "eval_steps_per_second": 0.214,
      "step": 2180
    },
    {
      "epoch": 0.8630787495053424,
      "grad_norm": 0.5448460825882789,
      "learning_rate": 5.604357437527191e-06,
      "loss": 0.1494,
      "step": 2181
    },
    {
      "epoch": 0.8634744756628413,
      "grad_norm": 0.4554738968431922,
      "learning_rate": 5.572623882658595e-06,
      "loss": 0.0755,
      "step": 2182
    },
    {
      "epoch": 0.8638702018203404,
      "grad_norm": 1.0374084042664145,
      "learning_rate": 5.540975122629061e-06,
      "loss": 0.1298,
      "step": 2183
    },
    {
      "epoch": 0.8642659279778393,
      "grad_norm": 0.8169865896784426,
      "learning_rate": 5.509411217843913e-06,
      "loss": 0.1279,
      "step": 2184
    },
    {
      "epoch": 0.8646616541353384,
      "grad_norm": 0.7696480357682703,
      "learning_rate": 5.477932228546573e-06,
      "loss": 0.1252,
      "step": 2185
    },
    {
      "epoch": 0.8646616541353384,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9291687161829808,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.24930013716220856,
      "eval_runtime": 5.0628,
      "eval_samples_per_second": 5.926,
      "eval_steps_per_second": 0.198,
      "step": 2185
    },
    {
      "epoch": 0.8650573802928374,
      "grad_norm": 1.2912697187132232,
      "learning_rate": 5.4465382148183645e-06,
      "loss": 0.1349,
      "step": 2186
    },
    {
      "epoch": 0.8654531064503364,
      "grad_norm": 0.8704044974664962,
      "learning_rate": 5.41522923657844e-06,
      "loss": 0.1674,
      "step": 2187
    },
    {
      "epoch": 0.8658488326078354,
      "grad_norm": 0.7510600802747459,
      "learning_rate": 5.384005353583632e-06,
      "loss": 0.1387,
      "step": 2188
    },
    {
      "epoch": 0.8662445587653343,
      "grad_norm": 0.8112705789791733,
      "learning_rate": 5.352866625428371e-06,
      "loss": 0.1284,
      "step": 2189
    },
    {
      "epoch": 0.8666402849228334,
      "grad_norm": 0.6591252890527719,
      "learning_rate": 5.321813111544577e-06,
      "loss": 0.1245,
      "step": 2190
    },
    {
      "epoch": 0.8666402849228334,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9303984259714706,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.24913737177848816,
      "eval_runtime": 5.1118,
      "eval_samples_per_second": 5.869,
      "eval_steps_per_second": 0.196,
      "step": 2190
    },
    {
      "epoch": 0.8670360110803325,
      "grad_norm": 0.7762106308986856,
      "learning_rate": 5.290844871201484e-06,
      "loss": 0.1433,
      "step": 2191
    },
    {
      "epoch": 0.8674317372378314,
      "grad_norm": 0.967717555974649,
      "learning_rate": 5.259961963505606e-06,
      "loss": 0.1964,
      "step": 2192
    },
    {
      "epoch": 0.8678274633953305,
      "grad_norm": 0.9768531263910336,
      "learning_rate": 5.229164447400587e-06,
      "loss": 0.1356,
      "step": 2193
    },
    {
      "epoch": 0.8682231895528294,
      "grad_norm": 1.107814255040027,
      "learning_rate": 5.19845238166709e-06,
      "loss": 0.1572,
      "step": 2194
    },
    {
      "epoch": 0.8686189157103285,
      "grad_norm": 0.5990956047298069,
      "learning_rate": 5.1678258249226615e-06,
      "loss": 0.112,
      "step": 2195
    },
    {
      "epoch": 0.8686189157103285,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9318740777176586,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.24734701216220856,
      "eval_runtime": 5.1986,
      "eval_samples_per_second": 5.771,
      "eval_steps_per_second": 0.192,
      "step": 2195
    },
    {
      "epoch": 0.8690146418678275,
      "grad_norm": 0.8784487000584996,
      "learning_rate": 5.137284835621681e-06,
      "loss": 0.1322,
      "step": 2196
    },
    {
      "epoch": 0.8694103680253265,
      "grad_norm": 0.975150632477361,
      "learning_rate": 5.106829472055202e-06,
      "loss": 0.1421,
      "step": 2197
    },
    {
      "epoch": 0.8698060941828255,
      "grad_norm": 0.6930704810344208,
      "learning_rate": 5.0764597923508235e-06,
      "loss": 0.1272,
      "step": 2198
    },
    {
      "epoch": 0.8702018203403245,
      "grad_norm": 0.715168917807267,
      "learning_rate": 5.046175854472634e-06,
      "loss": 0.1623,
      "step": 2199
    },
    {
      "epoch": 0.8705975464978235,
      "grad_norm": 0.8825385295170945,
      "learning_rate": 5.015977716221076e-06,
      "loss": 0.1282,
      "step": 2200
    },
    {
      "epoch": 0.8705975464978235,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9311362518445646,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.24720051884651184,
      "eval_runtime": 5.1099,
      "eval_samples_per_second": 5.871,
      "eval_steps_per_second": 0.196,
      "step": 2200
    },
    {
      "epoch": 0.8709932726553226,
      "grad_norm": 0.8197033902255443,
      "learning_rate": 4.985865435232834e-06,
      "loss": 0.0809,
      "step": 2201
    },
    {
      "epoch": 0.8713889988128215,
      "grad_norm": 0.5859303036122966,
      "learning_rate": 4.955839068980689e-06,
      "loss": 0.0954,
      "step": 2202
    },
    {
      "epoch": 0.8717847249703206,
      "grad_norm": 1.07651751662836,
      "learning_rate": 4.925898674773488e-06,
      "loss": 0.1891,
      "step": 2203
    },
    {
      "epoch": 0.8721804511278195,
      "grad_norm": 0.8394849525247677,
      "learning_rate": 4.896044309755965e-06,
      "loss": 0.13,
      "step": 2204
    },
    {
      "epoch": 0.8725761772853186,
      "grad_norm": 0.7364926883118647,
      "learning_rate": 4.866276030908678e-06,
      "loss": 0.1528,
      "step": 2205
    },
    {
      "epoch": 0.8725761772853186,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9303984259714707,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.24638672173023224,
      "eval_runtime": 5.2096,
      "eval_samples_per_second": 5.759,
      "eval_steps_per_second": 0.192,
      "step": 2205
    },
    {
      "epoch": 0.8729719034428176,
      "grad_norm": 0.8074544130302107,
      "learning_rate": 4.836593895047853e-06,
      "loss": 0.1171,
      "step": 2206
    },
    {
      "epoch": 0.8733676296003166,
      "grad_norm": 0.7031327717903912,
      "learning_rate": 4.806997958825299e-06,
      "loss": 0.1187,
      "step": 2207
    },
    {
      "epoch": 0.8737633557578156,
      "grad_norm": 0.6424257456682662,
      "learning_rate": 4.777488278728354e-06,
      "loss": 0.1478,
      "step": 2208
    },
    {
      "epoch": 0.8741590819153146,
      "grad_norm": 0.9728132581222644,
      "learning_rate": 4.748064911079669e-06,
      "loss": 0.14,
      "step": 2209
    },
    {
      "epoch": 0.8745548080728136,
      "grad_norm": 0.8382843485598854,
      "learning_rate": 4.7187279120371905e-06,
      "loss": 0.1212,
      "step": 2210
    },
    {
      "epoch": 0.8745548080728136,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9306443679291686,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.24653320014476776,
      "eval_runtime": 4.8976,
      "eval_samples_per_second": 6.125,
      "eval_steps_per_second": 0.204,
      "step": 2210
    },
    {
      "epoch": 0.8749505342303127,
      "grad_norm": 0.8594058848220121,
      "learning_rate": 4.6894773375939836e-06,
      "loss": 0.1991,
      "step": 2211
    },
    {
      "epoch": 0.8753462603878116,
      "grad_norm": 1.1323621978598417,
      "learning_rate": 4.6603132435782295e-06,
      "loss": 0.1518,
      "step": 2212
    },
    {
      "epoch": 0.8757419865453107,
      "grad_norm": 0.7641177811803257,
      "learning_rate": 4.631235685652979e-06,
      "loss": 0.1292,
      "step": 2213
    },
    {
      "epoch": 0.8761377127028096,
      "grad_norm": 0.9954093753531614,
      "learning_rate": 4.6022447193161625e-06,
      "loss": 0.1187,
      "step": 2214
    },
    {
      "epoch": 0.8765334388603087,
      "grad_norm": 1.0391254689994038,
      "learning_rate": 4.573340399900417e-06,
      "loss": 0.1702,
      "step": 2215
    },
    {
      "epoch": 0.8765334388603087,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9321200196753566,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.24558919668197632,
      "eval_runtime": 5.1707,
      "eval_samples_per_second": 5.802,
      "eval_steps_per_second": 0.193,
      "step": 2215
    },
    {
      "epoch": 0.8769291650178077,
      "grad_norm": 1.1304159906403821,
      "learning_rate": 4.544522782573019e-06,
      "loss": 0.2121,
      "step": 2216
    },
    {
      "epoch": 0.8773248911753067,
      "grad_norm": 0.4495622409311928,
      "learning_rate": 4.515791922335772e-06,
      "loss": 0.1118,
      "step": 2217
    },
    {
      "epoch": 0.8777206173328057,
      "grad_norm": 0.873624917741848,
      "learning_rate": 4.487147874024855e-06,
      "loss": 0.1714,
      "step": 2218
    },
    {
      "epoch": 0.8781163434903048,
      "grad_norm": 0.9670911078730067,
      "learning_rate": 4.458590692310793e-06,
      "loss": 0.1695,
      "step": 2219
    },
    {
      "epoch": 0.8785120696478037,
      "grad_norm": 0.7922088727246017,
      "learning_rate": 4.4301204316983035e-06,
      "loss": 0.1665,
      "step": 2220
    },
    {
      "epoch": 0.8785120696478037,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9316281357599606,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.24402669072151184,
      "eval_runtime": 4.9706,
      "eval_samples_per_second": 6.035,
      "eval_steps_per_second": 0.201,
      "step": 2220
    },
    {
      "epoch": 0.8789077958053028,
      "grad_norm": 0.716207214998658,
      "learning_rate": 4.401737146526219e-06,
      "loss": 0.1048,
      "step": 2221
    },
    {
      "epoch": 0.8793035219628017,
      "grad_norm": 0.7970104807431878,
      "learning_rate": 4.37344089096734e-06,
      "loss": 0.1536,
      "step": 2222
    },
    {
      "epoch": 0.8796992481203008,
      "grad_norm": 1.0797396744434662,
      "learning_rate": 4.3452317190283755e-06,
      "loss": 0.2005,
      "step": 2223
    },
    {
      "epoch": 0.8800949742777998,
      "grad_norm": 1.090500440780956,
      "learning_rate": 4.317109684549847e-06,
      "loss": 0.2638,
      "step": 2224
    },
    {
      "epoch": 0.8804907004352988,
      "grad_norm": 0.6940385191431523,
      "learning_rate": 4.289074841205914e-06,
      "loss": 0.1192,
      "step": 2225
    },
    {
      "epoch": 0.8804907004352988,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9331037875061485,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.24557290971279144,
      "eval_runtime": 5.007,
      "eval_samples_per_second": 5.992,
      "eval_steps_per_second": 0.2,
      "step": 2225
    },
    {
      "epoch": 0.8808864265927978,
      "grad_norm": 0.9745683605291304,
      "learning_rate": 4.261127242504376e-06,
      "loss": 0.1685,
      "step": 2226
    },
    {
      "epoch": 0.8812821527502968,
      "grad_norm": 0.6598043430303845,
      "learning_rate": 4.2332669417864735e-06,
      "loss": 0.079,
      "step": 2227
    },
    {
      "epoch": 0.8816778789077958,
      "grad_norm": 0.7452596347409093,
      "learning_rate": 4.205493992226867e-06,
      "loss": 0.1204,
      "step": 2228
    },
    {
      "epoch": 0.8820736050652949,
      "grad_norm": 0.8466749849627204,
      "learning_rate": 4.177808446833453e-06,
      "loss": 0.1381,
      "step": 2229
    },
    {
      "epoch": 0.8824693312227938,
      "grad_norm": 0.9447502764576206,
      "learning_rate": 4.150210358447343e-06,
      "loss": 0.1518,
      "step": 2230
    },
    {
      "epoch": 0.8824693312227938,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9328578455484505,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.24591471254825592,
      "eval_runtime": 5.0789,
      "eval_samples_per_second": 5.907,
      "eval_steps_per_second": 0.197,
      "step": 2230
    },
    {
      "epoch": 0.8828650573802929,
      "grad_norm": 1.0353845903875933,
      "learning_rate": 4.122699779742711e-06,
      "loss": 0.1794,
      "step": 2231
    },
    {
      "epoch": 0.8832607835377918,
      "grad_norm": 1.3078407377030685,
      "learning_rate": 4.095276763226719e-06,
      "loss": 0.219,
      "step": 2232
    },
    {
      "epoch": 0.8836565096952909,
      "grad_norm": 0.8948227337391861,
      "learning_rate": 4.067941361239386e-06,
      "loss": 0.1677,
      "step": 2233
    },
    {
      "epoch": 0.8840522358527899,
      "grad_norm": 0.6395053794334896,
      "learning_rate": 4.040693625953523e-06,
      "loss": 0.116,
      "step": 2234
    },
    {
      "epoch": 0.8844479620102889,
      "grad_norm": 0.9276942912260081,
      "learning_rate": 4.013533609374631e-06,
      "loss": 0.2145,
      "step": 2235
    },
    {
      "epoch": 0.8844479620102889,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9313821938022626,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.24562174081802368,
      "eval_runtime": 5.1382,
      "eval_samples_per_second": 5.839,
      "eval_steps_per_second": 0.195,
      "step": 2235
    },
    {
      "epoch": 0.8848436881677879,
      "grad_norm": 0.8214577675106776,
      "learning_rate": 3.986461363340754e-06,
      "loss": 0.1457,
      "step": 2236
    },
    {
      "epoch": 0.8852394143252869,
      "grad_norm": 0.8372969144863791,
      "learning_rate": 3.959476939522455e-06,
      "loss": 0.1562,
      "step": 2237
    },
    {
      "epoch": 0.8856351404827859,
      "grad_norm": 0.9928341153197203,
      "learning_rate": 3.932580389422647e-06,
      "loss": 0.2097,
      "step": 2238
    },
    {
      "epoch": 0.886030866640285,
      "grad_norm": 0.8217680624952052,
      "learning_rate": 3.905771764376553e-06,
      "loss": 0.1449,
      "step": 2239
    },
    {
      "epoch": 0.8864265927977839,
      "grad_norm": 0.6284779935028032,
      "learning_rate": 3.879051115551557e-06,
      "loss": 0.1033,
      "step": 2240
    },
    {
      "epoch": 0.8864265927977839,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9301524840137728,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.2445475310087204,
      "eval_runtime": 5.0914,
      "eval_samples_per_second": 5.892,
      "eval_steps_per_second": 0.196,
      "step": 2240
    },
    {
      "epoch": 0.886822318955283,
      "grad_norm": 0.9137044662592305,
      "learning_rate": 3.852418493947135e-06,
      "loss": 0.1118,
      "step": 2241
    },
    {
      "epoch": 0.8872180451127819,
      "grad_norm": 0.48454521853238985,
      "learning_rate": 3.825873950394776e-06,
      "loss": 0.114,
      "step": 2242
    },
    {
      "epoch": 0.887613771270281,
      "grad_norm": 0.7746245762697936,
      "learning_rate": 3.7994175355578256e-06,
      "loss": 0.1407,
      "step": 2243
    },
    {
      "epoch": 0.88800949742778,
      "grad_norm": 0.9334715291276695,
      "learning_rate": 3.7730492999314583e-06,
      "loss": 0.1457,
      "step": 2244
    },
    {
      "epoch": 0.888405223585279,
      "grad_norm": 1.0120324788192048,
      "learning_rate": 3.7467692938425057e-06,
      "loss": 0.1555,
      "step": 2245
    },
    {
      "epoch": 0.888405223585279,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9316281357599606,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.24443359673023224,
      "eval_runtime": 5.0491,
      "eval_samples_per_second": 5.942,
      "eval_steps_per_second": 0.198,
      "step": 2245
    },
    {
      "epoch": 0.888800949742778,
      "grad_norm": 1.1647461480873509,
      "learning_rate": 3.7205775674494624e-06,
      "loss": 0.1537,
      "step": 2246
    },
    {
      "epoch": 0.889196675900277,
      "grad_norm": 0.7798740101268412,
      "learning_rate": 3.694474170742279e-06,
      "loss": 0.1346,
      "step": 2247
    },
    {
      "epoch": 0.889592402057776,
      "grad_norm": 1.0137864206983274,
      "learning_rate": 3.6684591535423586e-06,
      "loss": 0.1216,
      "step": 2248
    },
    {
      "epoch": 0.8899881282152751,
      "grad_norm": 0.8631393961201419,
      "learning_rate": 3.6425325655023656e-06,
      "loss": 0.1046,
      "step": 2249
    },
    {
      "epoch": 0.890383854372774,
      "grad_norm": 0.6789978341492537,
      "learning_rate": 3.6166944561062622e-06,
      "loss": 0.1651,
      "step": 2250
    },
    {
      "epoch": 0.890383854372774,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9313821938022626,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.24526366591453552,
      "eval_runtime": 4.9914,
      "eval_samples_per_second": 6.01,
      "eval_steps_per_second": 0.2,
      "step": 2250
    },
    {
      "epoch": 0.8907795805302731,
      "grad_norm": 0.7217143306382184,
      "learning_rate": 3.590944874669089e-06,
      "loss": 0.1596,
      "step": 2251
    },
    {
      "epoch": 0.891175306687772,
      "grad_norm": 1.0474479263943033,
      "learning_rate": 3.565283870336911e-06,
      "loss": 0.1832,
      "step": 2252
    },
    {
      "epoch": 0.8915710328452711,
      "grad_norm": 1.298565597330361,
      "learning_rate": 3.5397114920867725e-06,
      "loss": 0.2266,
      "step": 2253
    },
    {
      "epoch": 0.8919667590027701,
      "grad_norm": 0.777351949111962,
      "learning_rate": 3.514227788726537e-06,
      "loss": 0.1586,
      "step": 2254
    },
    {
      "epoch": 0.8923624851602691,
      "grad_norm": 1.0469299232455658,
      "learning_rate": 3.488832808894843e-06,
      "loss": 0.2616,
      "step": 2255
    },
    {
      "epoch": 0.8923624851602691,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9321200196753565,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.2451985627412796,
      "eval_runtime": 5.0939,
      "eval_samples_per_second": 5.889,
      "eval_steps_per_second": 0.196,
      "step": 2255
    },
    {
      "epoch": 0.8927582113177681,
      "grad_norm": 0.7464953188084694,
      "learning_rate": 3.4635266010609624e-06,
      "loss": 0.1218,
      "step": 2256
    },
    {
      "epoch": 0.8931539374752672,
      "grad_norm": 0.5689038028332999,
      "learning_rate": 3.4383092135247543e-06,
      "loss": 0.0984,
      "step": 2257
    },
    {
      "epoch": 0.8935496636327661,
      "grad_norm": 1.1300439661450687,
      "learning_rate": 3.413180694416551e-06,
      "loss": 0.1718,
      "step": 2258
    },
    {
      "epoch": 0.8939453897902652,
      "grad_norm": 0.5247760631000785,
      "learning_rate": 3.388141091697078e-06,
      "loss": 0.0695,
      "step": 2259
    },
    {
      "epoch": 0.8943411159477641,
      "grad_norm": 1.0681085747091035,
      "learning_rate": 3.3631904531573277e-06,
      "loss": 0.1996,
      "step": 2260
    },
    {
      "epoch": 0.8943411159477641,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9321200196753566,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.24431966245174408,
      "eval_runtime": 5.1204,
      "eval_samples_per_second": 5.859,
      "eval_steps_per_second": 0.195,
      "step": 2260
    },
    {
      "epoch": 0.8947368421052632,
      "grad_norm": 1.0397697690275594,
      "learning_rate": 3.338328826418513e-06,
      "loss": 0.1432,
      "step": 2261
    },
    {
      "epoch": 0.8951325682627622,
      "grad_norm": 0.9111874286931895,
      "learning_rate": 3.3135562589319656e-06,
      "loss": 0.1283,
      "step": 2262
    },
    {
      "epoch": 0.8955282944202612,
      "grad_norm": 0.9525890326219971,
      "learning_rate": 3.288872797979009e-06,
      "loss": 0.1689,
      "step": 2263
    },
    {
      "epoch": 0.8959240205777602,
      "grad_norm": 1.051407377399643,
      "learning_rate": 3.264278490670919e-06,
      "loss": 0.2121,
      "step": 2264
    },
    {
      "epoch": 0.8963197467352592,
      "grad_norm": 0.7311757168275806,
      "learning_rate": 3.239773383948802e-06,
      "loss": 0.1372,
      "step": 2265
    },
    {
      "epoch": 0.8963197467352592,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9291687161829807,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.24514974653720856,
      "eval_runtime": 4.6992,
      "eval_samples_per_second": 6.384,
      "eval_steps_per_second": 0.213,
      "step": 2265
    },
    {
      "epoch": 0.8967154728927582,
      "grad_norm": 0.8261748268345058,
      "learning_rate": 3.2153575245835344e-06,
      "loss": 0.145,
      "step": 2266
    },
    {
      "epoch": 0.8971111990502573,
      "grad_norm": 0.8871994407693755,
      "learning_rate": 3.1910309591756172e-06,
      "loss": 0.1476,
      "step": 2267
    },
    {
      "epoch": 0.8975069252077562,
      "grad_norm": 0.8056710550738623,
      "learning_rate": 3.166793734155149e-06,
      "loss": 0.1469,
      "step": 2268
    },
    {
      "epoch": 0.8979026513652553,
      "grad_norm": 0.8080981282853406,
      "learning_rate": 3.142645895781715e-06,
      "loss": 0.1825,
      "step": 2269
    },
    {
      "epoch": 0.8982983775227542,
      "grad_norm": 0.9443455251284217,
      "learning_rate": 3.1185874901442703e-06,
      "loss": 0.1477,
      "step": 2270
    },
    {
      "epoch": 0.8982983775227542,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9306443679291687,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.2459309846162796,
      "eval_runtime": 4.9546,
      "eval_samples_per_second": 6.055,
      "eval_steps_per_second": 0.202,
      "step": 2270
    },
    {
      "epoch": 0.8986941036802533,
      "grad_norm": 0.5347161591715075,
      "learning_rate": 3.0946185631611002e-06,
      "loss": 0.0696,
      "step": 2271
    },
    {
      "epoch": 0.8990898298377523,
      "grad_norm": 0.7541299410745835,
      "learning_rate": 3.070739160579711e-06,
      "loss": 0.1421,
      "step": 2272
    },
    {
      "epoch": 0.8994855559952513,
      "grad_norm": 0.6672381154264779,
      "learning_rate": 3.0469493279767335e-06,
      "loss": 0.1268,
      "step": 2273
    },
    {
      "epoch": 0.8998812821527503,
      "grad_norm": 0.8089624178294565,
      "learning_rate": 3.0232491107578253e-06,
      "loss": 0.144,
      "step": 2274
    },
    {
      "epoch": 0.9002770083102493,
      "grad_norm": 0.9652041772523724,
      "learning_rate": 2.9996385541576353e-06,
      "loss": 0.1968,
      "step": 2275
    },
    {
      "epoch": 0.9002770083102493,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9306443679291687,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.24545899033546448,
      "eval_runtime": 5.1294,
      "eval_samples_per_second": 5.849,
      "eval_steps_per_second": 0.195,
      "step": 2275
    },
    {
      "epoch": 0.9006727344677483,
      "grad_norm": 0.5833603034146356,
      "learning_rate": 2.976117703239667e-06,
      "loss": 0.1013,
      "step": 2276
    },
    {
      "epoch": 0.9010684606252474,
      "grad_norm": 0.5542397091559673,
      "learning_rate": 2.9526866028962206e-06,
      "loss": 0.08,
      "step": 2277
    },
    {
      "epoch": 0.9014641867827463,
      "grad_norm": 1.425115156073236,
      "learning_rate": 2.9293452978482793e-06,
      "loss": 0.2648,
      "step": 2278
    },
    {
      "epoch": 0.9018599129402454,
      "grad_norm": 1.0670272488854673,
      "learning_rate": 2.90609383264544e-06,
      "loss": 0.2093,
      "step": 2279
    },
    {
      "epoch": 0.9022556390977443,
      "grad_norm": 0.8950779608594098,
      "learning_rate": 2.882932251665871e-06,
      "loss": 0.1769,
      "step": 2280
    },
    {
      "epoch": 0.9022556390977443,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9313821938022626,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.24596354365348816,
      "eval_runtime": 5.06,
      "eval_samples_per_second": 5.929,
      "eval_steps_per_second": 0.198,
      "step": 2280
    },
    {
      "epoch": 0.9026513652552434,
      "grad_norm": 0.7419704102540784,
      "learning_rate": 2.8598605991161264e-06,
      "loss": 0.1612,
      "step": 2281
    },
    {
      "epoch": 0.9030470914127424,
      "grad_norm": 0.744678244019709,
      "learning_rate": 2.8368789190311773e-06,
      "loss": 0.104,
      "step": 2282
    },
    {
      "epoch": 0.9034428175702414,
      "grad_norm": 0.722703467380455,
      "learning_rate": 2.81398725527422e-06,
      "loss": 0.1458,
      "step": 2283
    },
    {
      "epoch": 0.9038385437277404,
      "grad_norm": 0.8203214782726644,
      "learning_rate": 2.791185651536693e-06,
      "loss": 0.1453,
      "step": 2284
    },
    {
      "epoch": 0.9042342698852394,
      "grad_norm": 0.7469174327860593,
      "learning_rate": 2.7684741513381074e-06,
      "loss": 0.1718,
      "step": 2285
    },
    {
      "epoch": 0.9042342698852394,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9311362518445647,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.2461751252412796,
      "eval_runtime": 5.0478,
      "eval_samples_per_second": 5.943,
      "eval_steps_per_second": 0.198,
      "step": 2285
    },
    {
      "epoch": 0.9046299960427384,
      "grad_norm": 0.6798659751140131,
      "learning_rate": 2.7458527980260216e-06,
      "loss": 0.115,
      "step": 2286
    },
    {
      "epoch": 0.9050257222002375,
      "grad_norm": 0.6793258899224335,
      "learning_rate": 2.7233216347759272e-06,
      "loss": 0.1215,
      "step": 2287
    },
    {
      "epoch": 0.9054214483577364,
      "grad_norm": 0.9066840561244014,
      "learning_rate": 2.7008807045911855e-06,
      "loss": 0.1911,
      "step": 2288
    },
    {
      "epoch": 0.9058171745152355,
      "grad_norm": 0.8764945318302403,
      "learning_rate": 2.6785300503029407e-06,
      "loss": 0.1826,
      "step": 2289
    },
    {
      "epoch": 0.9062129006727345,
      "grad_norm": 0.9487261091884898,
      "learning_rate": 2.656269714570009e-06,
      "loss": 0.2272,
      "step": 2290
    },
    {
      "epoch": 0.9062129006727345,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9303984259714707,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.24573567509651184,
      "eval_runtime": 5.0696,
      "eval_samples_per_second": 5.918,
      "eval_steps_per_second": 0.197,
      "step": 2290
    },
    {
      "epoch": 0.9066086268302335,
      "grad_norm": 0.5968770657573935,
      "learning_rate": 2.6340997398788593e-06,
      "loss": 0.0807,
      "step": 2291
    },
    {
      "epoch": 0.9070043529877325,
      "grad_norm": 0.7942754326535966,
      "learning_rate": 2.6120201685434776e-06,
      "loss": 0.1474,
      "step": 2292
    },
    {
      "epoch": 0.9074000791452315,
      "grad_norm": 0.8897541402252085,
      "learning_rate": 2.5900310427053044e-06,
      "loss": 0.1376,
      "step": 2293
    },
    {
      "epoch": 0.9077958053027305,
      "grad_norm": 0.7346998873322627,
      "learning_rate": 2.5681324043331455e-06,
      "loss": 0.1711,
      "step": 2294
    },
    {
      "epoch": 0.9081915314602296,
      "grad_norm": 1.0165177368816904,
      "learning_rate": 2.5463242952231235e-06,
      "loss": 0.1488,
      "step": 2295
    },
    {
      "epoch": 0.9081915314602296,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9316281357599606,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.24720051884651184,
      "eval_runtime": 5.1578,
      "eval_samples_per_second": 5.816,
      "eval_steps_per_second": 0.194,
      "step": 2295
    },
    {
      "epoch": 0.9085872576177285,
      "grad_norm": 0.6852360196463074,
      "learning_rate": 2.524606756998571e-06,
      "loss": 0.1545,
      "step": 2296
    },
    {
      "epoch": 0.9089829837752276,
      "grad_norm": 0.9982700874479719,
      "learning_rate": 2.502979831109925e-06,
      "loss": 0.1662,
      "step": 2297
    },
    {
      "epoch": 0.9093787099327265,
      "grad_norm": 1.0567593785827034,
      "learning_rate": 2.481443558834712e-06,
      "loss": 0.2025,
      "step": 2298
    },
    {
      "epoch": 0.9097744360902256,
      "grad_norm": 0.671469616921964,
      "learning_rate": 2.459997981277423e-06,
      "loss": 0.1614,
      "step": 2299
    },
    {
      "epoch": 0.9101701622477246,
      "grad_norm": 0.7865270985714553,
      "learning_rate": 2.438643139369462e-06,
      "loss": 0.1605,
      "step": 2300
    },
    {
      "epoch": 0.9101701622477246,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9306443679291686,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.24676106870174408,
      "eval_runtime": 4.8875,
      "eval_samples_per_second": 6.138,
      "eval_steps_per_second": 0.205,
      "step": 2300
    },
    {
      "epoch": 0.9105658884052236,
      "grad_norm": 0.7243140865391168,
      "learning_rate": 2.4173790738690195e-06,
      "loss": 0.1324,
      "step": 2301
    },
    {
      "epoch": 0.9109616145627226,
      "grad_norm": 0.8105830991931131,
      "learning_rate": 2.3962058253610587e-06,
      "loss": 0.1244,
      "step": 2302
    },
    {
      "epoch": 0.9113573407202216,
      "grad_norm": 0.4713284289812932,
      "learning_rate": 2.375123434257198e-06,
      "loss": 0.1141,
      "step": 2303
    },
    {
      "epoch": 0.9117530668777206,
      "grad_norm": 0.7282643672818592,
      "learning_rate": 2.354131940795651e-06,
      "loss": 0.1242,
      "step": 2304
    },
    {
      "epoch": 0.9121487930352197,
      "grad_norm": 0.8754582479861359,
      "learning_rate": 2.3332313850411236e-06,
      "loss": 0.2384,
      "step": 2305
    },
    {
      "epoch": 0.9121487930352197,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9321200196753566,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.24728189408779144,
      "eval_runtime": 4.5804,
      "eval_samples_per_second": 6.55,
      "eval_steps_per_second": 0.218,
      "step": 2305
    },
    {
      "epoch": 0.9125445191927186,
      "grad_norm": 0.7832360577839531,
      "learning_rate": 2.312421806884779e-06,
      "loss": 0.1319,
      "step": 2306
    },
    {
      "epoch": 0.9129402453502177,
      "grad_norm": 0.7488781877821481,
      "learning_rate": 2.29170324604413e-06,
      "loss": 0.1413,
      "step": 2307
    },
    {
      "epoch": 0.9133359715077166,
      "grad_norm": 0.8542936372054603,
      "learning_rate": 2.2710757420629558e-06,
      "loss": 0.1594,
      "step": 2308
    },
    {
      "epoch": 0.9137316976652157,
      "grad_norm": 0.5205134536647634,
      "learning_rate": 2.2505393343112745e-06,
      "loss": 0.1148,
      "step": 2309
    },
    {
      "epoch": 0.9141274238227147,
      "grad_norm": 0.5176587595309369,
      "learning_rate": 2.2300940619852107e-06,
      "loss": 0.0929,
      "step": 2310
    },
    {
      "epoch": 0.9141274238227147,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9303984259714707,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.24643555283546448,
      "eval_runtime": 4.8481,
      "eval_samples_per_second": 6.188,
      "eval_steps_per_second": 0.206,
      "step": 2310
    },
    {
      "epoch": 0.9145231499802137,
      "grad_norm": 0.7226377961236182,
      "learning_rate": 2.209739964106966e-06,
      "loss": 0.1122,
      "step": 2311
    },
    {
      "epoch": 0.9149188761377127,
      "grad_norm": 0.5856186496303655,
      "learning_rate": 2.1894770795247042e-06,
      "loss": 0.0914,
      "step": 2312
    },
    {
      "epoch": 0.9153146022952117,
      "grad_norm": 0.7118883006910611,
      "learning_rate": 2.1693054469125118e-06,
      "loss": 0.1056,
      "step": 2313
    },
    {
      "epoch": 0.9157103284527107,
      "grad_norm": 0.8623956970355978,
      "learning_rate": 2.149225104770314e-06,
      "loss": 0.1486,
      "step": 2314
    },
    {
      "epoch": 0.9161060546102098,
      "grad_norm": 0.8501029702803289,
      "learning_rate": 2.1292360914237753e-06,
      "loss": 0.1644,
      "step": 2315
    },
    {
      "epoch": 0.9161060546102098,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9299065420560747,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.24697265028953552,
      "eval_runtime": 5.0387,
      "eval_samples_per_second": 5.954,
      "eval_steps_per_second": 0.198,
      "step": 2315
    },
    {
      "epoch": 0.9165017807677087,
      "grad_norm": 1.193630316398348,
      "learning_rate": 2.109338445024284e-06,
      "loss": 0.1663,
      "step": 2316
    },
    {
      "epoch": 0.9168975069252078,
      "grad_norm": 0.8512156515093804,
      "learning_rate": 2.089532203548794e-06,
      "loss": 0.1857,
      "step": 2317
    },
    {
      "epoch": 0.9172932330827067,
      "grad_norm": 0.6789388897844072,
      "learning_rate": 2.0698174047998618e-06,
      "loss": 0.139,
      "step": 2318
    },
    {
      "epoch": 0.9176889592402058,
      "grad_norm": 0.5113862337968058,
      "learning_rate": 2.0501940864054715e-06,
      "loss": 0.0882,
      "step": 2319
    },
    {
      "epoch": 0.9180846853977048,
      "grad_norm": 0.6148129275804831,
      "learning_rate": 2.030662285819024e-06,
      "loss": 0.1308,
      "step": 2320
    },
    {
      "epoch": 0.9180846853977048,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9308903098868666,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.2461751252412796,
      "eval_runtime": 5.1407,
      "eval_samples_per_second": 5.836,
      "eval_steps_per_second": 0.195,
      "step": 2320
    },
    {
      "epoch": 0.9184804115552038,
      "grad_norm": 1.2831911813629184,
      "learning_rate": 2.0112220403192215e-06,
      "loss": 0.2225,
      "step": 2321
    },
    {
      "epoch": 0.9188761377127028,
      "grad_norm": 0.7031060710742897,
      "learning_rate": 1.9918733870100793e-06,
      "loss": 0.1321,
      "step": 2322
    },
    {
      "epoch": 0.9192718638702018,
      "grad_norm": 0.6541442234210689,
      "learning_rate": 1.972616362820745e-06,
      "loss": 0.124,
      "step": 2323
    },
    {
      "epoch": 0.9196675900277008,
      "grad_norm": 0.9022683134285963,
      "learning_rate": 1.9534510045054967e-06,
      "loss": 0.1842,
      "step": 2324
    },
    {
      "epoch": 0.9200633161851999,
      "grad_norm": 1.0275945024247726,
      "learning_rate": 1.934377348643662e-06,
      "loss": 0.1822,
      "step": 2325
    },
    {
      "epoch": 0.9200633161851999,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9311362518445647,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.24767252802848816,
      "eval_runtime": 4.8423,
      "eval_samples_per_second": 6.195,
      "eval_steps_per_second": 0.207,
      "step": 2325
    },
    {
      "epoch": 0.9204590423426988,
      "grad_norm": 0.8363835960305179,
      "learning_rate": 1.915395431639544e-06,
      "loss": 0.1763,
      "step": 2326
    },
    {
      "epoch": 0.9208547685001979,
      "grad_norm": 0.9407344122066565,
      "learning_rate": 1.8965052897223611e-06,
      "loss": 0.1688,
      "step": 2327
    },
    {
      "epoch": 0.9212504946576969,
      "grad_norm": 1.0098848415843023,
      "learning_rate": 1.8777069589461348e-06,
      "loss": 0.2079,
      "step": 2328
    },
    {
      "epoch": 0.9216462208151959,
      "grad_norm": 0.6654966399281359,
      "learning_rate": 1.8590004751896871e-06,
      "loss": 0.1057,
      "step": 2329
    },
    {
      "epoch": 0.9220419469726949,
      "grad_norm": 1.0668723184095703,
      "learning_rate": 1.8403858741565306e-06,
      "loss": 0.1655,
      "step": 2330
    },
    {
      "epoch": 0.9220419469726949,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9308903098868667,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.24659830331802368,
      "eval_runtime": 5.0635,
      "eval_samples_per_second": 5.925,
      "eval_steps_per_second": 0.197,
      "step": 2330
    },
    {
      "epoch": 0.9224376731301939,
      "grad_norm": 0.7558786225506254,
      "learning_rate": 1.8218631913748062e-06,
      "loss": 0.1491,
      "step": 2331
    },
    {
      "epoch": 0.9228333992876929,
      "grad_norm": 0.8591819941297277,
      "learning_rate": 1.8034324621972132e-06,
      "loss": 0.2078,
      "step": 2332
    },
    {
      "epoch": 0.923229125445192,
      "grad_norm": 0.8625550739982581,
      "learning_rate": 1.7850937218009567e-06,
      "loss": 0.1761,
      "step": 2333
    },
    {
      "epoch": 0.9236248516026909,
      "grad_norm": 0.7642124031539225,
      "learning_rate": 1.7668470051876662e-06,
      "loss": 0.1713,
      "step": 2334
    },
    {
      "epoch": 0.92402057776019,
      "grad_norm": 0.6965065977139256,
      "learning_rate": 1.7486923471833284e-06,
      "loss": 0.1505,
      "step": 2335
    },
    {
      "epoch": 0.92402057776019,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9323659616330545,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.24654947221279144,
      "eval_runtime": 5.0894,
      "eval_samples_per_second": 5.895,
      "eval_steps_per_second": 0.196,
      "step": 2335
    },
    {
      "epoch": 0.9244163039176889,
      "grad_norm": 0.8906154877484328,
      "learning_rate": 1.7306297824382312e-06,
      "loss": 0.2023,
      "step": 2336
    },
    {
      "epoch": 0.924812030075188,
      "grad_norm": 0.8844419887997305,
      "learning_rate": 1.712659345426887e-06,
      "loss": 0.1655,
      "step": 2337
    },
    {
      "epoch": 0.925207756232687,
      "grad_norm": 0.6694311309114312,
      "learning_rate": 1.6947810704479873e-06,
      "loss": 0.1954,
      "step": 2338
    },
    {
      "epoch": 0.925603482390186,
      "grad_norm": 0.6935445429853461,
      "learning_rate": 1.6769949916242977e-06,
      "loss": 0.1416,
      "step": 2339
    },
    {
      "epoch": 0.925999208547685,
      "grad_norm": 0.7529683184437345,
      "learning_rate": 1.659301142902625e-06,
      "loss": 0.1426,
      "step": 2340
    },
    {
      "epoch": 0.925999208547685,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9318740777176586,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.24680989980697632,
      "eval_runtime": 5.0738,
      "eval_samples_per_second": 5.913,
      "eval_steps_per_second": 0.197,
      "step": 2340
    },
    {
      "epoch": 0.926394934705184,
      "grad_norm": 0.8844078906984691,
      "learning_rate": 1.6416995580537664e-06,
      "loss": 0.254,
      "step": 2341
    },
    {
      "epoch": 0.926790660862683,
      "grad_norm": 0.6234132125905922,
      "learning_rate": 1.6241902706723766e-06,
      "loss": 0.1299,
      "step": 2342
    },
    {
      "epoch": 0.9271863870201821,
      "grad_norm": 1.317549425865593,
      "learning_rate": 1.6067733141769958e-06,
      "loss": 0.2506,
      "step": 2343
    },
    {
      "epoch": 0.927582113177681,
      "grad_norm": 0.4966459549244899,
      "learning_rate": 1.5894487218099164e-06,
      "loss": 0.0837,
      "step": 2344
    },
    {
      "epoch": 0.9279778393351801,
      "grad_norm": 0.47213527981642744,
      "learning_rate": 1.5722165266371492e-06,
      "loss": 0.0893,
      "step": 2345
    },
    {
      "epoch": 0.9279778393351801,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9321200196753566,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.24685873091220856,
      "eval_runtime": 5.0565,
      "eval_samples_per_second": 5.933,
      "eval_steps_per_second": 0.198,
      "step": 2345
    },
    {
      "epoch": 0.928373565492679,
      "grad_norm": 0.9508720126999562,
      "learning_rate": 1.5550767615483408e-06,
      "loss": 0.1788,
      "step": 2346
    },
    {
      "epoch": 0.9287692916501781,
      "grad_norm": 1.3273145559071609,
      "learning_rate": 1.5380294592567513e-06,
      "loss": 0.1748,
      "step": 2347
    },
    {
      "epoch": 0.9291650178076771,
      "grad_norm": 0.9954780344138726,
      "learning_rate": 1.5210746522991425e-06,
      "loss": 0.2131,
      "step": 2348
    },
    {
      "epoch": 0.9295607439651761,
      "grad_norm": 0.5920072889550725,
      "learning_rate": 1.504212373035746e-06,
      "loss": 0.1405,
      "step": 2349
    },
    {
      "epoch": 0.9299564701226751,
      "grad_norm": 0.8909762545802551,
      "learning_rate": 1.4874426536501895e-06,
      "loss": 0.1803,
      "step": 2350
    },
    {
      "epoch": 0.9299564701226751,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9316281357599607,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.24594727158546448,
      "eval_runtime": 5.0283,
      "eval_samples_per_second": 5.966,
      "eval_steps_per_second": 0.199,
      "step": 2350
    },
    {
      "epoch": 0.9303521962801741,
      "grad_norm": 1.177686980602943,
      "learning_rate": 1.4707655261494368e-06,
      "loss": 0.1855,
      "step": 2351
    },
    {
      "epoch": 0.9307479224376731,
      "grad_norm": 0.9008828566191269,
      "learning_rate": 1.4541810223637487e-06,
      "loss": 0.1412,
      "step": 2352
    },
    {
      "epoch": 0.9311436485951722,
      "grad_norm": 0.866257345194914,
      "learning_rate": 1.4376891739465826e-06,
      "loss": 0.1813,
      "step": 2353
    },
    {
      "epoch": 0.9315393747526711,
      "grad_norm": 0.8347391258726958,
      "learning_rate": 1.42129001237456e-06,
      "loss": 0.2278,
      "step": 2354
    },
    {
      "epoch": 0.9319351009101702,
      "grad_norm": 0.7212418435263043,
      "learning_rate": 1.4049835689473822e-06,
      "loss": 0.1194,
      "step": 2355
    },
    {
      "epoch": 0.9319351009101702,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9308903098868667,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.24558919668197632,
      "eval_runtime": 4.634,
      "eval_samples_per_second": 6.474,
      "eval_steps_per_second": 0.216,
      "step": 2355
    },
    {
      "epoch": 0.9323308270676691,
      "grad_norm": 0.8131654418770823,
      "learning_rate": 1.3887698747878263e-06,
      "loss": 0.1409,
      "step": 2356
    },
    {
      "epoch": 0.9327265532251682,
      "grad_norm": 1.058738985934202,
      "learning_rate": 1.3726489608416104e-06,
      "loss": 0.1991,
      "step": 2357
    },
    {
      "epoch": 0.9331222793826672,
      "grad_norm": 0.45565516700928477,
      "learning_rate": 1.3566208578773775e-06,
      "loss": 0.0841,
      "step": 2358
    },
    {
      "epoch": 0.9335180055401662,
      "grad_norm": 0.9928214921505215,
      "learning_rate": 1.3406855964866405e-06,
      "loss": 0.1274,
      "step": 2359
    },
    {
      "epoch": 0.9339137316976652,
      "grad_norm": 0.7417888938816193,
      "learning_rate": 1.3248432070837035e-06,
      "loss": 0.1253,
      "step": 2360
    },
    {
      "epoch": 0.9339137316976652,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9313821938022626,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.24565429985523224,
      "eval_runtime": 4.7163,
      "eval_samples_per_second": 6.361,
      "eval_steps_per_second": 0.212,
      "step": 2360
    },
    {
      "epoch": 0.9343094578551642,
      "grad_norm": 0.521254687739885,
      "learning_rate": 1.3090937199056352e-06,
      "loss": 0.1126,
      "step": 2361
    },
    {
      "epoch": 0.9347051840126632,
      "grad_norm": 1.2327313415044423,
      "learning_rate": 1.2934371650121458e-06,
      "loss": 0.2252,
      "step": 2362
    },
    {
      "epoch": 0.9351009101701623,
      "grad_norm": 0.8045702998089796,
      "learning_rate": 1.2778735722856205e-06,
      "loss": 0.1315,
      "step": 2363
    },
    {
      "epoch": 0.9354966363276612,
      "grad_norm": 0.6193581247545799,
      "learning_rate": 1.2624029714309872e-06,
      "loss": 0.1494,
      "step": 2364
    },
    {
      "epoch": 0.9358923624851603,
      "grad_norm": 0.5383755998604777,
      "learning_rate": 1.247025391975698e-06,
      "loss": 0.0962,
      "step": 2365
    },
    {
      "epoch": 0.9358923624851603,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9308903098868667,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.24669596552848816,
      "eval_runtime": 4.6724,
      "eval_samples_per_second": 6.421,
      "eval_steps_per_second": 0.214,
      "step": 2365
    },
    {
      "epoch": 0.9362880886426593,
      "grad_norm": 1.1037105483446057,
      "learning_rate": 1.2317408632696537e-06,
      "loss": 0.1699,
      "step": 2366
    },
    {
      "epoch": 0.9366838148001583,
      "grad_norm": 0.8464858946058663,
      "learning_rate": 1.2165494144851686e-06,
      "loss": 0.1686,
      "step": 2367
    },
    {
      "epoch": 0.9370795409576573,
      "grad_norm": 0.6273187145927843,
      "learning_rate": 1.2014510746168994e-06,
      "loss": 0.112,
      "step": 2368
    },
    {
      "epoch": 0.9374752671151563,
      "grad_norm": 0.601571403704037,
      "learning_rate": 1.1864458724817895e-06,
      "loss": 0.1333,
      "step": 2369
    },
    {
      "epoch": 0.9378709932726553,
      "grad_norm": 1.0579516660959085,
      "learning_rate": 1.1715338367190188e-06,
      "loss": 0.194,
      "step": 2370
    },
    {
      "epoch": 0.9378709932726553,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9313821938022626,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.24679361283779144,
      "eval_runtime": 4.9878,
      "eval_samples_per_second": 6.015,
      "eval_steps_per_second": 0.2,
      "step": 2370
    },
    {
      "epoch": 0.9382667194301544,
      "grad_norm": 0.9394397382095822,
      "learning_rate": 1.1567149957899592e-06,
      "loss": 0.1701,
      "step": 2371
    },
    {
      "epoch": 0.9386624455876533,
      "grad_norm": 0.8723818137028356,
      "learning_rate": 1.1419893779780922e-06,
      "loss": 0.1597,
      "step": 2372
    },
    {
      "epoch": 0.9390581717451524,
      "grad_norm": 0.6029101073132475,
      "learning_rate": 1.1273570113889798e-06,
      "loss": 0.1012,
      "step": 2373
    },
    {
      "epoch": 0.9394538979026513,
      "grad_norm": 0.721636846965824,
      "learning_rate": 1.1128179239502046e-06,
      "loss": 0.1313,
      "step": 2374
    },
    {
      "epoch": 0.9398496240601504,
      "grad_norm": 0.8598433176964458,
      "learning_rate": 1.0983721434113192e-06,
      "loss": 0.1881,
      "step": 2375
    },
    {
      "epoch": 0.9398496240601504,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9313821938022626,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.24560546875,
      "eval_runtime": 5.0583,
      "eval_samples_per_second": 5.931,
      "eval_steps_per_second": 0.198,
      "step": 2375
    },
    {
      "epoch": 0.9402453502176494,
      "grad_norm": 0.6141854272130054,
      "learning_rate": 1.0840196973437744e-06,
      "loss": 0.1198,
      "step": 2376
    },
    {
      "epoch": 0.9406410763751484,
      "grad_norm": 0.6835049966151376,
      "learning_rate": 1.0697606131408966e-06,
      "loss": 0.168,
      "step": 2377
    },
    {
      "epoch": 0.9410368025326474,
      "grad_norm": 0.742926512307513,
      "learning_rate": 1.0555949180178104e-06,
      "loss": 0.1159,
      "step": 2378
    },
    {
      "epoch": 0.9414325286901464,
      "grad_norm": 0.4101079285315537,
      "learning_rate": 1.0415226390114108e-06,
      "loss": 0.0623,
      "step": 2379
    },
    {
      "epoch": 0.9418282548476454,
      "grad_norm": 0.5516064219376665,
      "learning_rate": 1.0275438029802797e-06,
      "loss": 0.0894,
      "step": 2380
    },
    {
      "epoch": 0.9418282548476454,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9308903098868667,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.24563802778720856,
      "eval_runtime": 4.9073,
      "eval_samples_per_second": 6.113,
      "eval_steps_per_second": 0.204,
      "step": 2380
    },
    {
      "epoch": 0.9422239810051445,
      "grad_norm": 0.889486461896786,
      "learning_rate": 1.0136584366046531e-06,
      "loss": 0.1319,
      "step": 2381
    },
    {
      "epoch": 0.9426197071626434,
      "grad_norm": 0.9240871143981982,
      "learning_rate": 9.99866566386387e-07,
      "loss": 0.1575,
      "step": 2382
    },
    {
      "epoch": 0.9430154333201425,
      "grad_norm": 0.5646337100218698,
      "learning_rate": 9.861682186488697e-07,
      "loss": 0.1357,
      "step": 2383
    },
    {
      "epoch": 0.9434111594776414,
      "grad_norm": 0.6471240948547796,
      "learning_rate": 9.725634195370036e-07,
      "loss": 0.145,
      "step": 2384
    },
    {
      "epoch": 0.9438068856351405,
      "grad_norm": 0.7238296004053348,
      "learning_rate": 9.590521950171293e-07,
      "loss": 0.1326,
      "step": 2385
    },
    {
      "epoch": 0.9438068856351405,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9313821938022626,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.24702148139476776,
      "eval_runtime": 5.0849,
      "eval_samples_per_second": 5.9,
      "eval_steps_per_second": 0.197,
      "step": 2385
    },
    {
      "epoch": 0.9442026117926395,
      "grad_norm": 0.8328971547148584,
      "learning_rate": 9.456345708770076e-07,
      "loss": 0.1433,
      "step": 2386
    },
    {
      "epoch": 0.9445983379501385,
      "grad_norm": 0.5252819391201454,
      "learning_rate": 9.323105727257308e-07,
      "loss": 0.1156,
      "step": 2387
    },
    {
      "epoch": 0.9449940641076375,
      "grad_norm": 0.7991088277249658,
      "learning_rate": 9.190802259937237e-07,
      "loss": 0.1582,
      "step": 2388
    },
    {
      "epoch": 0.9453897902651365,
      "grad_norm": 0.8325878517604974,
      "learning_rate": 9.059435559326257e-07,
      "loss": 0.146,
      "step": 2389
    },
    {
      "epoch": 0.9457855164226355,
      "grad_norm": 0.795571994105661,
      "learning_rate": 8.929005876153307e-07,
      "loss": 0.1285,
      "step": 2390
    },
    {
      "epoch": 0.9457855164226355,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9299065420560747,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.24708658456802368,
      "eval_runtime": 4.7738,
      "eval_samples_per_second": 6.284,
      "eval_steps_per_second": 0.209,
      "step": 2390
    },
    {
      "epoch": 0.9461812425801346,
      "grad_norm": 0.5737633043743479,
      "learning_rate": 8.799513459358533e-07,
      "loss": 0.1094,
      "step": 2391
    },
    {
      "epoch": 0.9465769687376335,
      "grad_norm": 1.0971987401426833,
      "learning_rate": 8.670958556093401e-07,
      "loss": 0.1368,
      "step": 2392
    },
    {
      "epoch": 0.9469726948951326,
      "grad_norm": 1.1235655807082656,
      "learning_rate": 8.543341411719918e-07,
      "loss": 0.2238,
      "step": 2393
    },
    {
      "epoch": 0.9473684210526315,
      "grad_norm": 0.7223380379769588,
      "learning_rate": 8.41666226981036e-07,
      "loss": 0.1606,
      "step": 2394
    },
    {
      "epoch": 0.9477641472101306,
      "grad_norm": 0.5733447790842316,
      "learning_rate": 8.290921372146654e-07,
      "loss": 0.104,
      "step": 2395
    },
    {
      "epoch": 0.9477641472101306,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9308903098868666,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.2479654997587204,
      "eval_runtime": 5.1268,
      "eval_samples_per_second": 5.852,
      "eval_steps_per_second": 0.195,
      "step": 2395
    },
    {
      "epoch": 0.9481598733676296,
      "grad_norm": 0.9563617683533557,
      "learning_rate": 8.166118958719992e-07,
      "loss": 0.1934,
      "step": 2396
    },
    {
      "epoch": 0.9485555995251286,
      "grad_norm": 0.7123942028728772,
      "learning_rate": 8.042255267730392e-07,
      "loss": 0.134,
      "step": 2397
    },
    {
      "epoch": 0.9489513256826276,
      "grad_norm": 0.8504689792705687,
      "learning_rate": 7.919330535586134e-07,
      "loss": 0.1694,
      "step": 2398
    },
    {
      "epoch": 0.9493470518401267,
      "grad_norm": 0.7711323899197261,
      "learning_rate": 7.797344996903544e-07,
      "loss": 0.1752,
      "step": 2399
    },
    {
      "epoch": 0.9497427779976256,
      "grad_norm": 0.6383787192128396,
      "learning_rate": 7.676298884506106e-07,
      "loss": 0.1435,
      "step": 2400
    },
    {
      "epoch": 0.9497427779976256,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9301524840137727,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.24742838740348816,
      "eval_runtime": 5.2459,
      "eval_samples_per_second": 5.719,
      "eval_steps_per_second": 0.191,
      "step": 2400
    },
    {
      "epoch": 0.9501385041551247,
      "grad_norm": 0.66428074472605,
      "learning_rate": 7.556192429424569e-07,
      "loss": 0.1047,
      "step": 2401
    },
    {
      "epoch": 0.9505342303126236,
      "grad_norm": 0.6182222004980799,
      "learning_rate": 7.43702586089623e-07,
      "loss": 0.1058,
      "step": 2402
    },
    {
      "epoch": 0.9509299564701227,
      "grad_norm": 0.8426409008525981,
      "learning_rate": 7.318799406364208e-07,
      "loss": 0.1781,
      "step": 2403
    },
    {
      "epoch": 0.9513256826276217,
      "grad_norm": 0.7077985198244826,
      "learning_rate": 7.201513291477669e-07,
      "loss": 0.1482,
      "step": 2404
    },
    {
      "epoch": 0.9517214087851207,
      "grad_norm": 0.8303908915635464,
      "learning_rate": 7.085167740090771e-07,
      "loss": 0.1911,
      "step": 2405
    },
    {
      "epoch": 0.9517214087851207,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9313821938022627,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.24710287153720856,
      "eval_runtime": 5.0588,
      "eval_samples_per_second": 5.93,
      "eval_steps_per_second": 0.198,
      "step": 2405
    },
    {
      "epoch": 0.9521171349426197,
      "grad_norm": 0.974936675939218,
      "learning_rate": 6.969762974262717e-07,
      "loss": 0.1838,
      "step": 2406
    },
    {
      "epoch": 0.9525128611001187,
      "grad_norm": 1.1996662894131347,
      "learning_rate": 6.855299214256817e-07,
      "loss": 0.1835,
      "step": 2407
    },
    {
      "epoch": 0.9529085872576177,
      "grad_norm": 0.8504649438722559,
      "learning_rate": 6.741776678540645e-07,
      "loss": 0.1741,
      "step": 2408
    },
    {
      "epoch": 0.9533043134151168,
      "grad_norm": 0.4730385190472341,
      "learning_rate": 6.629195583785219e-07,
      "loss": 0.0988,
      "step": 2409
    },
    {
      "epoch": 0.9537000395726157,
      "grad_norm": 0.5734419533059284,
      "learning_rate": 6.517556144864711e-07,
      "loss": 0.1017,
      "step": 2410
    },
    {
      "epoch": 0.9537000395726157,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9301524840137727,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.24816080927848816,
      "eval_runtime": 4.9958,
      "eval_samples_per_second": 6.005,
      "eval_steps_per_second": 0.2,
      "step": 2410
    },
    {
      "epoch": 0.9540957657301148,
      "grad_norm": 0.7203093906137981,
      "learning_rate": 6.406858574856067e-07,
      "loss": 0.1381,
      "step": 2411
    },
    {
      "epoch": 0.9544914918876137,
      "grad_norm": 0.8386685545481133,
      "learning_rate": 6.297103085038391e-07,
      "loss": 0.1432,
      "step": 2412
    },
    {
      "epoch": 0.9548872180451128,
      "grad_norm": 0.5638128929852357,
      "learning_rate": 6.188289884893062e-07,
      "loss": 0.1154,
      "step": 2413
    },
    {
      "epoch": 0.9552829442026118,
      "grad_norm": 0.7647360868727683,
      "learning_rate": 6.080419182102615e-07,
      "loss": 0.174,
      "step": 2414
    },
    {
      "epoch": 0.9556786703601108,
      "grad_norm": 0.8273465509937044,
      "learning_rate": 5.973491182551028e-07,
      "loss": 0.1332,
      "step": 2415
    },
    {
      "epoch": 0.9556786703601108,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9311362518445647,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.24671223759651184,
      "eval_runtime": 4.9379,
      "eval_samples_per_second": 6.075,
      "eval_steps_per_second": 0.203,
      "step": 2415
    },
    {
      "epoch": 0.9560743965176098,
      "grad_norm": 0.7167648805174767,
      "learning_rate": 5.867506090322772e-07,
      "loss": 0.1585,
      "step": 2416
    },
    {
      "epoch": 0.9564701226751088,
      "grad_norm": 0.865770167381506,
      "learning_rate": 5.76246410770287e-07,
      "loss": 0.1741,
      "step": 2417
    },
    {
      "epoch": 0.9568658488326078,
      "grad_norm": 0.9699957273764583,
      "learning_rate": 5.658365435176171e-07,
      "loss": 0.2198,
      "step": 2418
    },
    {
      "epoch": 0.9572615749901069,
      "grad_norm": 0.7187997762648886,
      "learning_rate": 5.555210271427192e-07,
      "loss": 0.1271,
      "step": 2419
    },
    {
      "epoch": 0.9576573011476058,
      "grad_norm": 1.2440588141048479,
      "learning_rate": 5.452998813339605e-07,
      "loss": 0.1623,
      "step": 2420
    },
    {
      "epoch": 0.9576573011476058,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9316281357599606,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.24744465947151184,
      "eval_runtime": 5.0719,
      "eval_samples_per_second": 5.915,
      "eval_steps_per_second": 0.197,
      "step": 2420
    },
    {
      "epoch": 0.9580530273051049,
      "grad_norm": 1.1523531091658583,
      "learning_rate": 5.351731255995862e-07,
      "loss": 0.161,
      "step": 2421
    },
    {
      "epoch": 0.9584487534626038,
      "grad_norm": 0.6604203497026335,
      "learning_rate": 5.251407792677021e-07,
      "loss": 0.1227,
      "step": 2422
    },
    {
      "epoch": 0.9588444796201029,
      "grad_norm": 0.8518925594872353,
      "learning_rate": 5.152028614862026e-07,
      "loss": 0.1678,
      "step": 2423
    },
    {
      "epoch": 0.9592402057776019,
      "grad_norm": 0.7517461253587917,
      "learning_rate": 5.05359391222776e-07,
      "loss": 0.1594,
      "step": 2424
    },
    {
      "epoch": 0.9596359319351009,
      "grad_norm": 0.5758214832731425,
      "learning_rate": 4.956103872648333e-07,
      "loss": 0.1048,
      "step": 2425
    },
    {
      "epoch": 0.9596359319351009,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9316281357599607,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.24711914360523224,
      "eval_runtime": 5.1123,
      "eval_samples_per_second": 5.868,
      "eval_steps_per_second": 0.196,
      "step": 2425
    },
    {
      "epoch": 0.9600316580925999,
      "grad_norm": 1.7051204644291214,
      "learning_rate": 4.859558682194898e-07,
      "loss": 0.3023,
      "step": 2426
    },
    {
      "epoch": 0.9604273842500989,
      "grad_norm": 0.8981537772349597,
      "learning_rate": 4.7639585251350593e-07,
      "loss": 0.1427,
      "step": 2427
    },
    {
      "epoch": 0.9608231104075979,
      "grad_norm": 0.9320401984724555,
      "learning_rate": 4.669303583933138e-07,
      "loss": 0.1688,
      "step": 2428
    },
    {
      "epoch": 0.961218836565097,
      "grad_norm": 0.6123138593928498,
      "learning_rate": 4.575594039249065e-07,
      "loss": 0.1166,
      "step": 2429
    },
    {
      "epoch": 0.9616145627225959,
      "grad_norm": 0.8435792478572272,
      "learning_rate": 4.4828300699383264e-07,
      "loss": 0.1586,
      "step": 2430
    },
    {
      "epoch": 0.9616145627225959,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9308903098868666,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.2469889372587204,
      "eval_runtime": 5.293,
      "eval_samples_per_second": 5.668,
      "eval_steps_per_second": 0.189,
      "step": 2430
    },
    {
      "epoch": 0.962010288880095,
      "grad_norm": 0.80848027038971,
      "learning_rate": 4.3910118530519626e-07,
      "loss": 0.1592,
      "step": 2431
    },
    {
      "epoch": 0.9624060150375939,
      "grad_norm": 0.8117385093728984,
      "learning_rate": 4.300139563835681e-07,
      "loss": 0.2036,
      "step": 2432
    },
    {
      "epoch": 0.962801741195093,
      "grad_norm": 0.9836827630930483,
      "learning_rate": 4.2102133757299103e-07,
      "loss": 0.2079,
      "step": 2433
    },
    {
      "epoch": 0.963197467352592,
      "grad_norm": 0.7890214445075321,
      "learning_rate": 4.1212334603693003e-07,
      "loss": 0.1488,
      "step": 2434
    },
    {
      "epoch": 0.963593193510091,
      "grad_norm": 1.0341219910022466,
      "learning_rate": 4.033199987582337e-07,
      "loss": 0.1931,
      "step": 2435
    },
    {
      "epoch": 0.963593193510091,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9301524840137727,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.24781900644302368,
      "eval_runtime": 4.8326,
      "eval_samples_per_second": 6.208,
      "eval_steps_per_second": 0.207,
      "step": 2435
    },
    {
      "epoch": 0.96398891966759,
      "grad_norm": 0.7603779734409521,
      "learning_rate": 3.9461131253912266e-07,
      "loss": 0.1835,
      "step": 2436
    },
    {
      "epoch": 0.9643846458250891,
      "grad_norm": 0.8282839425998522,
      "learning_rate": 3.8599730400115107e-07,
      "loss": 0.1744,
      "step": 2437
    },
    {
      "epoch": 0.964780371982588,
      "grad_norm": 0.9123367977543965,
      "learning_rate": 3.7747798958515103e-07,
      "loss": 0.1442,
      "step": 2438
    },
    {
      "epoch": 0.9651760981400871,
      "grad_norm": 0.7492812392252703,
      "learning_rate": 3.690533855512268e-07,
      "loss": 0.1366,
      "step": 2439
    },
    {
      "epoch": 0.965571824297586,
      "grad_norm": 0.7825820435138836,
      "learning_rate": 3.60723507978733e-07,
      "loss": 0.1527,
      "step": 2440
    },
    {
      "epoch": 0.965571824297586,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9316281357599606,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.24729818105697632,
      "eval_runtime": 4.938,
      "eval_samples_per_second": 6.075,
      "eval_steps_per_second": 0.203,
      "step": 2440
    },
    {
      "epoch": 0.9659675504550851,
      "grad_norm": 1.073219670625734,
      "learning_rate": 3.5248837276620205e-07,
      "loss": 0.1456,
      "step": 2441
    },
    {
      "epoch": 0.9663632766125841,
      "grad_norm": 0.48454430145899896,
      "learning_rate": 3.4434799563135556e-07,
      "loss": 0.0831,
      "step": 2442
    },
    {
      "epoch": 0.9667590027700831,
      "grad_norm": 0.6910564562581203,
      "learning_rate": 3.363023921110542e-07,
      "loss": 0.1536,
      "step": 2443
    },
    {
      "epoch": 0.9671547289275821,
      "grad_norm": 0.7859314742863688,
      "learning_rate": 3.283515775612811e-07,
      "loss": 0.1679,
      "step": 2444
    },
    {
      "epoch": 0.9675504550850811,
      "grad_norm": 0.8338376376774453,
      "learning_rate": 3.2049556715708083e-07,
      "loss": 0.1557,
      "step": 2445
    },
    {
      "epoch": 0.9675504550850811,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9291687161829808,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.24669596552848816,
      "eval_runtime": 5.1539,
      "eval_samples_per_second": 5.821,
      "eval_steps_per_second": 0.194,
      "step": 2445
    },
    {
      "epoch": 0.9679461812425801,
      "grad_norm": 0.7475600293511832,
      "learning_rate": 3.127343758925705e-07,
      "loss": 0.1493,
      "step": 2446
    },
    {
      "epoch": 0.9683419074000792,
      "grad_norm": 0.7617574594269578,
      "learning_rate": 3.0506801858090073e-07,
      "loss": 0.1436,
      "step": 2447
    },
    {
      "epoch": 0.9687376335575781,
      "grad_norm": 0.722356520948722,
      "learning_rate": 2.9749650985420043e-07,
      "loss": 0.1253,
      "step": 2448
    },
    {
      "epoch": 0.9691333597150772,
      "grad_norm": 0.931694945629189,
      "learning_rate": 2.900198641635876e-07,
      "loss": 0.177,
      "step": 2449
    },
    {
      "epoch": 0.9695290858725761,
      "grad_norm": 0.6294279822235682,
      "learning_rate": 2.826380957790975e-07,
      "loss": 0.1623,
      "step": 2450
    },
    {
      "epoch": 0.9695290858725761,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9284308903098868,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.24806314706802368,
      "eval_runtime": 4.9037,
      "eval_samples_per_second": 6.118,
      "eval_steps_per_second": 0.204,
      "step": 2450
    },
    {
      "epoch": 0.9699248120300752,
      "grad_norm": 0.888351841092406,
      "learning_rate": 2.753512187897211e-07,
      "loss": 0.1743,
      "step": 2451
    },
    {
      "epoch": 0.9703205381875742,
      "grad_norm": 0.7910792194025019,
      "learning_rate": 2.6815924710329456e-07,
      "loss": 0.2057,
      "step": 2452
    },
    {
      "epoch": 0.9707162643450732,
      "grad_norm": 1.0229093187972313,
      "learning_rate": 2.61062194446543e-07,
      "loss": 0.1728,
      "step": 2453
    },
    {
      "epoch": 0.9711119905025722,
      "grad_norm": 1.092258672429503,
      "learning_rate": 2.5406007436502566e-07,
      "loss": 0.1887,
      "step": 2454
    },
    {
      "epoch": 0.9715077166600712,
      "grad_norm": 0.42158108228758867,
      "learning_rate": 2.471529002231021e-07,
      "loss": 0.0696,
      "step": 2455
    },
    {
      "epoch": 0.9715077166600712,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9308903098868666,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.24654947221279144,
      "eval_runtime": 4.9273,
      "eval_samples_per_second": 6.089,
      "eval_steps_per_second": 0.203,
      "step": 2455
    },
    {
      "epoch": 0.9719034428175702,
      "grad_norm": 0.5811022496232213,
      "learning_rate": 2.4034068520392693e-07,
      "loss": 0.135,
      "step": 2456
    },
    {
      "epoch": 0.9722991689750693,
      "grad_norm": 0.92418328218446,
      "learning_rate": 2.336234423093997e-07,
      "loss": 0.1591,
      "step": 2457
    },
    {
      "epoch": 0.9726948951325682,
      "grad_norm": 0.8268547080093145,
      "learning_rate": 2.2700118436016494e-07,
      "loss": 0.1256,
      "step": 2458
    },
    {
      "epoch": 0.9730906212900673,
      "grad_norm": 1.012905405992622,
      "learning_rate": 2.2047392399558443e-07,
      "loss": 0.1992,
      "step": 2459
    },
    {
      "epoch": 0.9734863474475662,
      "grad_norm": 0.8280246339511267,
      "learning_rate": 2.1404167367368721e-07,
      "loss": 0.1697,
      "step": 2460
    },
    {
      "epoch": 0.9734863474475662,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9323659616330545,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.24680989980697632,
      "eval_runtime": 5.1571,
      "eval_samples_per_second": 5.817,
      "eval_steps_per_second": 0.194,
      "step": 2460
    },
    {
      "epoch": 0.9738820736050653,
      "grad_norm": 0.9244855564456367,
      "learning_rate": 2.0770444567118075e-07,
      "loss": 0.1854,
      "step": 2461
    },
    {
      "epoch": 0.9742777997625643,
      "grad_norm": 0.5267723384766674,
      "learning_rate": 2.014622520834064e-07,
      "loss": 0.0766,
      "step": 2462
    },
    {
      "epoch": 0.9746735259200633,
      "grad_norm": 0.6653119230120595,
      "learning_rate": 1.9531510482431182e-07,
      "loss": 0.1176,
      "step": 2463
    },
    {
      "epoch": 0.9750692520775623,
      "grad_norm": 0.7859638511318909,
      "learning_rate": 1.8926301562645632e-07,
      "loss": 0.1527,
      "step": 2464
    },
    {
      "epoch": 0.9754649782350613,
      "grad_norm": 0.5895733794946284,
      "learning_rate": 1.8330599604095e-07,
      "loss": 0.1061,
      "step": 2465
    },
    {
      "epoch": 0.9754649782350613,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9316281357599606,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.2467447966337204,
      "eval_runtime": 5.2981,
      "eval_samples_per_second": 5.662,
      "eval_steps_per_second": 0.189,
      "step": 2465
    },
    {
      "epoch": 0.9758607043925603,
      "grad_norm": 0.814098691315181,
      "learning_rate": 1.774440574374703e-07,
      "loss": 0.1083,
      "step": 2466
    },
    {
      "epoch": 0.9762564305500594,
      "grad_norm": 0.698868931555832,
      "learning_rate": 1.7167721100420087e-07,
      "loss": 0.1151,
      "step": 2467
    },
    {
      "epoch": 0.9766521567075583,
      "grad_norm": 0.7292134110425602,
      "learning_rate": 1.6600546774785398e-07,
      "loss": 0.1195,
      "step": 2468
    },
    {
      "epoch": 0.9770478828650574,
      "grad_norm": 0.5006360520161892,
      "learning_rate": 1.604288384936037e-07,
      "loss": 0.0974,
      "step": 2469
    },
    {
      "epoch": 0.9774436090225563,
      "grad_norm": 1.1231035683191963,
      "learning_rate": 1.5494733388510817e-07,
      "loss": 0.1576,
      "step": 2470
    },
    {
      "epoch": 0.9774436090225563,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9311362518445646,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.2472330778837204,
      "eval_runtime": 5.0119,
      "eval_samples_per_second": 5.986,
      "eval_steps_per_second": 0.2,
      "step": 2470
    },
    {
      "epoch": 0.9778393351800554,
      "grad_norm": 0.8651052306900577,
      "learning_rate": 1.4956096438445423e-07,
      "loss": 0.1987,
      "step": 2471
    },
    {
      "epoch": 0.9782350613375544,
      "grad_norm": 0.9514257100213481,
      "learning_rate": 1.4426974027215713e-07,
      "loss": 0.1701,
      "step": 2472
    },
    {
      "epoch": 0.9786307874950534,
      "grad_norm": 1.0827726169294678,
      "learning_rate": 1.3907367164713303e-07,
      "loss": 0.1283,
      "step": 2473
    },
    {
      "epoch": 0.9790265136525524,
      "grad_norm": 0.8497954897547443,
      "learning_rate": 1.3397276842669892e-07,
      "loss": 0.174,
      "step": 2474
    },
    {
      "epoch": 0.9794222398100515,
      "grad_norm": 0.7642443722361096,
      "learning_rate": 1.2896704034651152e-07,
      "loss": 0.105,
      "step": 2475
    },
    {
      "epoch": 0.9794222398100515,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9316281357599607,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.24694010615348816,
      "eval_runtime": 4.8657,
      "eval_samples_per_second": 6.166,
      "eval_steps_per_second": 0.206,
      "step": 2475
    },
    {
      "epoch": 0.9798179659675504,
      "grad_norm": 0.7502953110890435,
      "learning_rate": 1.2405649696058953e-07,
      "loss": 0.1152,
      "step": 2476
    },
    {
      "epoch": 0.9802136921250495,
      "grad_norm": 0.8571420692335163,
      "learning_rate": 1.192411476412858e-07,
      "loss": 0.1439,
      "step": 2477
    },
    {
      "epoch": 0.9806094182825484,
      "grad_norm": 0.7142100462732285,
      "learning_rate": 1.1452100157925416e-07,
      "loss": 0.1631,
      "step": 2478
    },
    {
      "epoch": 0.9810051444400475,
      "grad_norm": 0.7980132526602703,
      "learning_rate": 1.0989606778344375e-07,
      "loss": 0.1724,
      "step": 2479
    },
    {
      "epoch": 0.9814008705975465,
      "grad_norm": 0.4129464805854385,
      "learning_rate": 1.0536635508107684e-07,
      "loss": 0.0886,
      "step": 2480
    },
    {
      "epoch": 0.9814008705975465,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9291687161829808,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.24871419370174408,
      "eval_runtime": 5.1153,
      "eval_samples_per_second": 5.865,
      "eval_steps_per_second": 0.195,
      "step": 2480
    },
    {
      "epoch": 0.9817965967550455,
      "grad_norm": 0.770834764410502,
      "learning_rate": 1.0093187211764887e-07,
      "loss": 0.1245,
      "step": 2481
    },
    {
      "epoch": 0.9821923229125445,
      "grad_norm": 0.8241331103258956,
      "learning_rate": 9.659262735688401e-08,
      "loss": 0.1448,
      "step": 2482
    },
    {
      "epoch": 0.9825880490700435,
      "grad_norm": 1.058550017077056,
      "learning_rate": 9.234862908074071e-08,
      "loss": 0.1578,
      "step": 2483
    },
    {
      "epoch": 0.9829837752275425,
      "grad_norm": 0.7262527113206975,
      "learning_rate": 8.81998853893784e-08,
      "loss": 0.1388,
      "step": 2484
    },
    {
      "epoch": 0.9833795013850416,
      "grad_norm": 0.9031304694898294,
      "learning_rate": 8.414640420116305e-08,
      "loss": 0.1544,
      "step": 2485
    },
    {
      "epoch": 0.9833795013850416,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9313821938022627,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.2474772185087204,
      "eval_runtime": 4.7699,
      "eval_samples_per_second": 6.289,
      "eval_steps_per_second": 0.21,
      "step": 2485
    },
    {
      "epoch": 0.9837752275425405,
      "grad_norm": 0.8038321877023986,
      "learning_rate": 8.018819325263937e-08,
      "loss": 0.1685,
      "step": 2486
    },
    {
      "epoch": 0.9841709537000396,
      "grad_norm": 0.7147091285661772,
      "learning_rate": 7.632526009851981e-08,
      "loss": 0.1344,
      "step": 2487
    },
    {
      "epoch": 0.9845666798575385,
      "grad_norm": 0.489683938156519,
      "learning_rate": 7.255761211165113e-08,
      "loss": 0.0848,
      "step": 2488
    },
    {
      "epoch": 0.9849624060150376,
      "grad_norm": 0.6951532506760213,
      "learning_rate": 6.888525648303667e-08,
      "loss": 0.1443,
      "step": 2489
    },
    {
      "epoch": 0.9853581321725366,
      "grad_norm": 0.759593113988845,
      "learning_rate": 6.530820022179751e-08,
      "loss": 0.1716,
      "step": 2490
    },
    {
      "epoch": 0.9853581321725366,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9301524840137727,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.24734701216220856,
      "eval_runtime": 4.9953,
      "eval_samples_per_second": 6.006,
      "eval_steps_per_second": 0.2,
      "step": 2490
    },
    {
      "epoch": 0.9857538583300356,
      "grad_norm": 0.4483048981359248,
      "learning_rate": 6.182645015516131e-08,
      "loss": 0.1127,
      "step": 2491
    },
    {
      "epoch": 0.9861495844875346,
      "grad_norm": 0.7529115065376876,
      "learning_rate": 5.844001292846235e-08,
      "loss": 0.153,
      "step": 2492
    },
    {
      "epoch": 0.9865453106450336,
      "grad_norm": 1.4321210666187865,
      "learning_rate": 5.514889500509712e-08,
      "loss": 0.2271,
      "step": 2493
    },
    {
      "epoch": 0.9869410368025326,
      "grad_norm": 1.2406480393177448,
      "learning_rate": 5.195310266656317e-08,
      "loss": 0.1555,
      "step": 2494
    },
    {
      "epoch": 0.9873367629600317,
      "grad_norm": 1.7317648261380225,
      "learning_rate": 4.885264201239248e-08,
      "loss": 0.1768,
      "step": 2495
    },
    {
      "epoch": 0.9873367629600317,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9311362518445647,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.24825845658779144,
      "eval_runtime": 4.962,
      "eval_samples_per_second": 6.046,
      "eval_steps_per_second": 0.202,
      "step": 2495
    },
    {
      "epoch": 0.9877324891175306,
      "grad_norm": 0.8281240116243923,
      "learning_rate": 4.5847518960184796e-08,
      "loss": 0.1736,
      "step": 2496
    },
    {
      "epoch": 0.9881282152750297,
      "grad_norm": 1.556198582369784,
      "learning_rate": 4.293773924556321e-08,
      "loss": 0.1719,
      "step": 2497
    },
    {
      "epoch": 0.9885239414325286,
      "grad_norm": 0.9740178172772898,
      "learning_rate": 4.012330842219081e-08,
      "loss": 0.1551,
      "step": 2498
    },
    {
      "epoch": 0.9889196675900277,
      "grad_norm": 0.5575840740570712,
      "learning_rate": 3.7404231861726255e-08,
      "loss": 0.1243,
      "step": 2499
    },
    {
      "epoch": 0.9893153937475268,
      "grad_norm": 0.7788329000687713,
      "learning_rate": 3.478051475385158e-08,
      "loss": 0.1271,
      "step": 2500
    },
    {
      "epoch": 0.9893153937475268,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9316281357599606,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.24619141221046448,
      "eval_runtime": 4.8893,
      "eval_samples_per_second": 6.136,
      "eval_steps_per_second": 0.205,
      "step": 2500
    },
    {
      "epoch": 0.9897111199050257,
      "grad_norm": 0.8039947711452612,
      "learning_rate": 3.225216210623327e-08,
      "loss": 0.1647,
      "step": 2501
    },
    {
      "epoch": 0.9901068460625247,
      "grad_norm": 0.7476619529213824,
      "learning_rate": 2.981917874453344e-08,
      "loss": 0.1253,
      "step": 2502
    },
    {
      "epoch": 0.9905025722200237,
      "grad_norm": 0.5241491291045124,
      "learning_rate": 2.7481569312381995e-08,
      "loss": 0.0823,
      "step": 2503
    },
    {
      "epoch": 0.9908982983775227,
      "grad_norm": 0.9939223746529062,
      "learning_rate": 2.52393382713767e-08,
      "loss": 0.1324,
      "step": 2504
    },
    {
      "epoch": 0.9912940245350218,
      "grad_norm": 0.6458497118973661,
      "learning_rate": 2.3092489901083148e-08,
      "loss": 0.15,
      "step": 2505
    },
    {
      "epoch": 0.9912940245350218,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9301524840137727,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.24746093153953552,
      "eval_runtime": 4.9162,
      "eval_samples_per_second": 6.102,
      "eval_steps_per_second": 0.203,
      "step": 2505
    },
    {
      "epoch": 0.9916897506925207,
      "grad_norm": 0.7602970248130199,
      "learning_rate": 2.1041028299012555e-08,
      "loss": 0.1582,
      "step": 2506
    },
    {
      "epoch": 0.9920854768500198,
      "grad_norm": 0.5423786801507522,
      "learning_rate": 1.908495738061067e-08,
      "loss": 0.0966,
      "step": 2507
    },
    {
      "epoch": 0.9924812030075187,
      "grad_norm": 0.6253780666388552,
      "learning_rate": 1.7224280879279964e-08,
      "loss": 0.1219,
      "step": 2508
    },
    {
      "epoch": 0.9928769291650178,
      "grad_norm": 1.1878637961106282,
      "learning_rate": 1.5459002346324135e-08,
      "loss": 0.1512,
      "step": 2509
    },
    {
      "epoch": 0.9932726553225169,
      "grad_norm": 0.5906597247343434,
      "learning_rate": 1.3789125150998061e-08,
      "loss": 0.1143,
      "step": 2510
    },
    {
      "epoch": 0.9932726553225169,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9308903098868667,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.24822591245174408,
      "eval_runtime": 5.2395,
      "eval_samples_per_second": 5.726,
      "eval_steps_per_second": 0.191,
      "step": 2510
    },
    {
      "epoch": 0.9936683814800158,
      "grad_norm": 0.7766810763883317,
      "learning_rate": 1.2214652480452282e-08,
      "loss": 0.1674,
      "step": 2511
    },
    {
      "epoch": 0.9940641076375148,
      "grad_norm": 0.6071055624471351,
      "learning_rate": 1.0735587339749665e-08,
      "loss": 0.1007,
      "step": 2512
    },
    {
      "epoch": 0.9944598337950139,
      "grad_norm": 0.7587109482992928,
      "learning_rate": 9.351932551854292e-09,
      "loss": 0.174,
      "step": 2513
    },
    {
      "epoch": 0.9948555599525128,
      "grad_norm": 0.9723362059523731,
      "learning_rate": 8.063690757642572e-09,
      "loss": 0.2211,
      "step": 2514
    },
    {
      "epoch": 0.9952512861100119,
      "grad_norm": 0.8585323746115432,
      "learning_rate": 6.8708644158754775e-09,
      "loss": 0.1269,
      "step": 2515
    },
    {
      "epoch": 0.9952512861100119,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9311362518445647,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.24700520932674408,
      "eval_runtime": 5.2251,
      "eval_samples_per_second": 5.742,
      "eval_steps_per_second": 0.191,
      "step": 2515
    },
    {
      "epoch": 0.9956470122675108,
      "grad_norm": 0.5898253204229328,
      "learning_rate": 5.773455803187444e-09,
      "loss": 0.0996,
      "step": 2516
    },
    {
      "epoch": 0.9960427384250099,
      "grad_norm": 0.5804500055147189,
      "learning_rate": 4.771467014125231e-09,
      "loss": 0.1425,
      "step": 2517
    },
    {
      "epoch": 0.996438464582509,
      "grad_norm": 0.7889287195136169,
      "learning_rate": 3.864899961097956e-09,
      "loss": 0.1534,
      "step": 2518
    },
    {
      "epoch": 0.9968341907400079,
      "grad_norm": 0.9666558103457217,
      "learning_rate": 3.053756374393757e-09,
      "loss": 0.1731,
      "step": 2519
    },
    {
      "epoch": 0.997229916897507,
      "grad_norm": 0.8618885558551689,
      "learning_rate": 2.338037802174231e-09,
      "loss": 0.1279,
      "step": 2520
    },
    {
      "epoch": 0.997229916897507,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9316281357599606,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.24785156548023224,
      "eval_runtime": 5.0732,
      "eval_samples_per_second": 5.913,
      "eval_steps_per_second": 0.197,
      "step": 2520
    },
    {
      "epoch": 0.9976256430550059,
      "grad_norm": 0.7093407752751012,
      "learning_rate": 1.7177456104688905e-09,
      "loss": 0.1339,
      "step": 2521
    },
    {
      "epoch": 0.998021369212505,
      "grad_norm": 0.9592888390561018,
      "learning_rate": 1.1928809831807108e-09,
      "loss": 0.1844,
      "step": 2522
    },
    {
      "epoch": 0.998417095370004,
      "grad_norm": 0.6765076345571852,
      "learning_rate": 7.634449220805806e-10,
      "loss": 0.1263,
      "step": 2523
    },
    {
      "epoch": 0.998812821527503,
      "grad_norm": 0.597491796120058,
      "learning_rate": 4.294382467906477e-10,
      "loss": 0.1135,
      "step": 2524
    },
    {
      "epoch": 0.999208547685002,
      "grad_norm": 0.8136990563223105,
      "learning_rate": 1.9086159480097287e-10,
      "loss": 0.1752,
      "step": 2525
    },
    {
      "epoch": 0.999208547685002,
      "eval_PRM Accuracy": 0.8968253968253969,
      "eval_PRM F1": 0.9417040358744395,
      "eval_PRM F1 AUC": 0.7011805213969503,
      "eval_PRM F1 AUC (fixed)": 0.9318740777176585,
      "eval_PRM F1 Neg": 0.5517241379310345,
      "eval_PRM NPV": 0.8,
      "eval_PRM Precision": 0.9051724137931034,
      "eval_PRM Recall": 0.9813084112149533,
      "eval_PRM Specificty": 0.42105263157894735,
      "eval_loss": 0.24759115278720856,
      "eval_runtime": 4.7407,
      "eval_samples_per_second": 6.328,
      "eval_steps_per_second": 0.211,
      "step": 2525
    },
    {
      "epoch": 0.999604273842501,
      "grad_norm": 0.9911556527429359,
      "learning_rate": 4.771542146952967e-11,
      "loss": 0.1765,
      "step": 2526
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.9876160839837966,
      "learning_rate": 0.0,
      "loss": 0.1651,
      "step": 2527
    }
  ],
  "logging_steps": 1,
  "max_steps": 2527,
  "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": true
      },
      "attributes": {}
    }
  },
  "total_flos": 3818088336261120.0,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}