diff --git "a/last-checkpoint/trainer_state.json" "b/last-checkpoint/trainer_state.json"
--- "a/last-checkpoint/trainer_state.json"
+++ "b/last-checkpoint/trainer_state.json"
@@ -1,9 +1,9 @@
 {
   "best_metric": null,
   "best_model_checkpoint": null,
-  "epoch": 0.9064128710627691,
+  "epoch": 0.9999028843352433,
   "eval_steps": 5,
-  "global_step": 3500,
+  "global_step": 3861,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
@@ -35723,6 +35723,3685 @@
       "eval_samples_per_second": 2.357,
       "eval_steps_per_second": 0.161,
       "step": 3500
+    },
+    {
+      "epoch": 0.906671846168787,
+      "grad_norm": 0.13423647403153086,
+      "learning_rate": 2.6263100921208482e-06,
+      "loss": 0.2854,
+      "step": 3501
+    },
+    {
+      "epoch": 0.906930821274805,
+      "grad_norm": 0.09574244399002062,
+      "learning_rate": 2.61186794717439e-06,
+      "loss": 0.3449,
+      "step": 3502
+    },
+    {
+      "epoch": 0.9071897963808229,
+      "grad_norm": 0.10872403782219404,
+      "learning_rate": 2.597464555616025e-06,
+      "loss": 0.408,
+      "step": 3503
+    },
+    {
+      "epoch": 0.9074487714868408,
+      "grad_norm": 0.11294136587055507,
+      "learning_rate": 2.583099929224658e-06,
+      "loss": 0.3506,
+      "step": 3504
+    },
+    {
+      "epoch": 0.9077077465928588,
+      "grad_norm": 0.08739093882495587,
+      "learning_rate": 2.568774079747488e-06,
+      "loss": 0.3155,
+      "step": 3505
+    },
+    {
+      "epoch": 0.9077077465928588,
+      "eval_PRM Accuracy": 0.9056603773584906,
+      "eval_PRM F1": 0.9390243902439024,
+      "eval_PRM F1 AUC": 0.8768988999476165,
+      "eval_PRM F1 Neg": 0.7916666666666666,
+      "eval_PRM NPV": 0.76,
+      "eval_PRM Precision": 0.9506172839506173,
+      "eval_PRM Recall": 0.927710843373494,
+      "eval_PRM Specificty": 0.8260869565217391,
+      "eval_loss": 0.224853515625,
+      "eval_runtime": 18.1708,
+      "eval_samples_per_second": 2.421,
+      "eval_steps_per_second": 0.165,
+      "step": 3505
+    },
+    {
+      "epoch": 0.9079667216988767,
+      "grad_norm": 0.07720558334000181,
+      "learning_rate": 2.5544870189000093e-06,
+      "loss": 0.2867,
+      "step": 3506
+    },
+    {
+      "epoch": 0.9082256968048946,
+      "grad_norm": 0.1102451388700012,
+      "learning_rate": 2.540238758365987e-06,
+      "loss": 0.351,
+      "step": 3507
+    },
+    {
+      "epoch": 0.9084846719109125,
+      "grad_norm": 0.09337654965628764,
+      "learning_rate": 2.5260293097974675e-06,
+      "loss": 0.3237,
+      "step": 3508
+    },
+    {
+      "epoch": 0.9087436470169304,
+      "grad_norm": 0.11920855993183896,
+      "learning_rate": 2.5118586848147386e-06,
+      "loss": 0.3469,
+      "step": 3509
+    },
+    {
+      "epoch": 0.9090026221229485,
+      "grad_norm": 0.10230906075271297,
+      "learning_rate": 2.4977268950063524e-06,
+      "loss": 0.3193,
+      "step": 3510
+    },
+    {
+      "epoch": 0.9090026221229485,
+      "eval_PRM Accuracy": 0.9056603773584906,
+      "eval_PRM F1": 0.9390243902439024,
+      "eval_PRM F1 AUC": 0.8768988999476165,
+      "eval_PRM F1 Neg": 0.7916666666666666,
+      "eval_PRM NPV": 0.76,
+      "eval_PRM Precision": 0.9506172839506173,
+      "eval_PRM Recall": 0.927710843373494,
+      "eval_PRM Specificty": 0.8260869565217391,
+      "eval_loss": 0.2252197265625,
+      "eval_runtime": 19.0903,
+      "eval_samples_per_second": 2.305,
+      "eval_steps_per_second": 0.157,
+      "step": 3510
+    },
+    {
+      "epoch": 0.9092615972289664,
+      "grad_norm": 0.08768902891508182,
+      "learning_rate": 2.483633951929104e-06,
+      "loss": 0.2599,
+      "step": 3511
+    },
+    {
+      "epoch": 0.9095205723349843,
+      "grad_norm": 0.088879496311207,
+      "learning_rate": 2.4695798671080183e-06,
+      "loss": 0.3033,
+      "step": 3512
+    },
+    {
+      "epoch": 0.9097795474410022,
+      "grad_norm": 0.08470900769089199,
+      "learning_rate": 2.455564652036324e-06,
+      "loss": 0.286,
+      "step": 3513
+    },
+    {
+      "epoch": 0.9100385225470201,
+      "grad_norm": 0.08462203007675333,
+      "learning_rate": 2.441588318175503e-06,
+      "loss": 0.2866,
+      "step": 3514
+    },
+    {
+      "epoch": 0.9102974976530381,
+      "grad_norm": 0.1230050801950961,
+      "learning_rate": 2.4276508769552064e-06,
+      "loss": 0.2744,
+      "step": 3515
+    },
+    {
+      "epoch": 0.9102974976530381,
+      "eval_PRM Accuracy": 0.8962264150943396,
+      "eval_PRM F1": 0.9333333333333333,
+      "eval_PRM F1 AUC": 0.855159769512834,
+      "eval_PRM F1 Neg": 0.7659574468085106,
+      "eval_PRM NPV": 0.75,
+      "eval_PRM Precision": 0.9390243902439024,
+      "eval_PRM Recall": 0.927710843373494,
+      "eval_PRM Specificty": 0.782608695652174,
+      "eval_loss": 0.22446511685848236,
+      "eval_runtime": 18.4952,
+      "eval_samples_per_second": 2.379,
+      "eval_steps_per_second": 0.162,
+      "step": 3515
+    },
+    {
+      "epoch": 0.910556472759056,
+      "grad_norm": 0.10260936904886261,
+      "learning_rate": 2.413752339773284e-06,
+      "loss": 0.3648,
+      "step": 3516
+    },
+    {
+      "epoch": 0.910815447865074,
+      "grad_norm": 0.09164539796223421,
+      "learning_rate": 2.3998927179957887e-06,
+      "loss": 0.3458,
+      "step": 3517
+    },
+    {
+      "epoch": 0.9110744229710919,
+      "grad_norm": 0.10675429533005563,
+      "learning_rate": 2.38607202295692e-06,
+      "loss": 0.3311,
+      "step": 3518
+    },
+    {
+      "epoch": 0.9113333980771099,
+      "grad_norm": 0.10366682186402575,
+      "learning_rate": 2.372290265959065e-06,
+      "loss": 0.3268,
+      "step": 3519
+    },
+    {
+      "epoch": 0.9115923731831278,
+      "grad_norm": 0.17835478265266552,
+      "learning_rate": 2.3585474582727695e-06,
+      "loss": 0.4438,
+      "step": 3520
+    },
+    {
+      "epoch": 0.9115923731831278,
+      "eval_PRM Accuracy": 0.9150943396226415,
+      "eval_PRM F1": 0.9447852760736196,
+      "eval_PRM F1 AUC": 0.8986380303823991,
+      "eval_PRM F1 Neg": 0.8163265306122449,
+      "eval_PRM NPV": 0.7692307692307693,
+      "eval_PRM Precision": 0.9625,
+      "eval_PRM Recall": 0.927710843373494,
+      "eval_PRM Specificty": 0.8695652173913043,
+      "eval_loss": 0.2239990234375,
+      "eval_runtime": 18.6514,
+      "eval_samples_per_second": 2.359,
+      "eval_steps_per_second": 0.161,
+      "step": 3520
+    },
+    {
+      "epoch": 0.9118513482891457,
+      "grad_norm": 0.10498125404377973,
+      "learning_rate": 2.34484361113671e-06,
+      "loss": 0.3767,
+      "step": 3521
+    },
+    {
+      "epoch": 0.9121103233951636,
+      "grad_norm": 0.1158862525825441,
+      "learning_rate": 2.3311787357577166e-06,
+      "loss": 0.3874,
+      "step": 3522
+    },
+    {
+      "epoch": 0.9123692985011815,
+      "grad_norm": 0.08002069711272743,
+      "learning_rate": 2.317552843310744e-06,
+      "loss": 0.2715,
+      "step": 3523
+    },
+    {
+      "epoch": 0.9126282736071996,
+      "grad_norm": 0.08934030097544853,
+      "learning_rate": 2.3039659449388573e-06,
+      "loss": 0.3332,
+      "step": 3524
+    },
+    {
+      "epoch": 0.9128872487132175,
+      "grad_norm": 0.09934304380120498,
+      "learning_rate": 2.290418051753268e-06,
+      "loss": 0.3078,
+      "step": 3525
+    },
+    {
+      "epoch": 0.9128872487132175,
+      "eval_PRM Accuracy": 0.9056603773584906,
+      "eval_PRM F1": 0.9390243902439024,
+      "eval_PRM F1 AUC": 0.8768988999476165,
+      "eval_PRM F1 Neg": 0.7916666666666666,
+      "eval_PRM NPV": 0.76,
+      "eval_PRM Precision": 0.9506172839506173,
+      "eval_PRM Recall": 0.927710843373494,
+      "eval_PRM Specificty": 0.8260869565217391,
+      "eval_loss": 0.2252197265625,
+      "eval_runtime": 18.4621,
+      "eval_samples_per_second": 2.383,
+      "eval_steps_per_second": 0.162,
+      "step": 3525
+    },
+    {
+      "epoch": 0.9131462238192354,
+      "grad_norm": 0.09666354060313337,
+      "learning_rate": 2.2769091748332418e-06,
+      "loss": 0.3076,
+      "step": 3526
+    },
+    {
+      "epoch": 0.9134051989252533,
+      "grad_norm": 0.10914740695070031,
+      "learning_rate": 2.2634393252261633e-06,
+      "loss": 0.2997,
+      "step": 3527
+    },
+    {
+      "epoch": 0.9136641740312712,
+      "grad_norm": 0.0760558232699135,
+      "learning_rate": 2.250008513947499e-06,
+      "loss": 0.2575,
+      "step": 3528
+    },
+    {
+      "epoch": 0.9139231491372892,
+      "grad_norm": 0.08962789317075943,
+      "learning_rate": 2.236616751980791e-06,
+      "loss": 0.3248,
+      "step": 3529
+    },
+    {
+      "epoch": 0.9141821242433071,
+      "grad_norm": 0.11465011269429774,
+      "learning_rate": 2.2232640502776457e-06,
+      "loss": 0.3719,
+      "step": 3530
+    },
+    {
+      "epoch": 0.9141821242433071,
+      "eval_PRM Accuracy": 0.9056603773584906,
+      "eval_PRM F1": 0.9390243902439024,
+      "eval_PRM F1 AUC": 0.8768988999476165,
+      "eval_PRM F1 Neg": 0.7916666666666666,
+      "eval_PRM NPV": 0.76,
+      "eval_PRM Precision": 0.9506172839506173,
+      "eval_PRM Recall": 0.927710843373494,
+      "eval_PRM Specificty": 0.8260869565217391,
+      "eval_loss": 0.2246759533882141,
+      "eval_runtime": 18.4705,
+      "eval_samples_per_second": 2.382,
+      "eval_steps_per_second": 0.162,
+      "step": 3530
+    },
+    {
+      "epoch": 0.914441099349325,
+      "grad_norm": 0.11820778109952361,
+      "learning_rate": 2.209950419757717e-06,
+      "loss": 0.4073,
+      "step": 3531
+    },
+    {
+      "epoch": 0.914700074455343,
+      "grad_norm": 0.09126097717994588,
+      "learning_rate": 2.1966758713087286e-06,
+      "loss": 0.2985,
+      "step": 3532
+    },
+    {
+      "epoch": 0.9149590495613609,
+      "grad_norm": 0.08188717495426173,
+      "learning_rate": 2.183440415786425e-06,
+      "loss": 0.3432,
+      "step": 3533
+    },
+    {
+      "epoch": 0.9152180246673789,
+      "grad_norm": 0.08924239668913243,
+      "learning_rate": 2.170244064014587e-06,
+      "loss": 0.307,
+      "step": 3534
+    },
+    {
+      "epoch": 0.9154769997733968,
+      "grad_norm": 0.10421591156480886,
+      "learning_rate": 2.1570868267850207e-06,
+      "loss": 0.3346,
+      "step": 3535
+    },
+    {
+      "epoch": 0.9154769997733968,
+      "eval_PRM Accuracy": 0.8962264150943396,
+      "eval_PRM F1": 0.9333333333333333,
+      "eval_PRM F1 AUC": 0.855159769512834,
+      "eval_PRM F1 Neg": 0.7659574468085106,
+      "eval_PRM NPV": 0.75,
+      "eval_PRM Precision": 0.9390243902439024,
+      "eval_PRM Recall": 0.927710843373494,
+      "eval_PRM Specificty": 0.782608695652174,
+      "eval_loss": 0.2255193591117859,
+      "eval_runtime": 18.1202,
+      "eval_samples_per_second": 2.428,
+      "eval_steps_per_second": 0.166,
+      "step": 3535
+    },
+    {
+      "epoch": 0.9157359748794147,
+      "grad_norm": 0.08343092470673229,
+      "learning_rate": 2.1439687148575306e-06,
+      "loss": 0.3382,
+      "step": 3536
+    },
+    {
+      "epoch": 0.9159949499854326,
+      "grad_norm": 0.0926368018245957,
+      "learning_rate": 2.130889738959946e-06,
+      "loss": 0.3672,
+      "step": 3537
+    },
+    {
+      "epoch": 0.9162539250914505,
+      "grad_norm": 0.09108156719035451,
+      "learning_rate": 2.1178499097880722e-06,
+      "loss": 0.3052,
+      "step": 3538
+    },
+    {
+      "epoch": 0.9165129001974686,
+      "grad_norm": 0.09316624136676999,
+      "learning_rate": 2.1048492380057126e-06,
+      "loss": 0.3276,
+      "step": 3539
+    },
+    {
+      "epoch": 0.9167718753034865,
+      "grad_norm": 0.11376298261341627,
+      "learning_rate": 2.091887734244652e-06,
+      "loss": 0.3844,
+      "step": 3540
+    },
+    {
+      "epoch": 0.9167718753034865,
+      "eval_PRM Accuracy": 0.9056603773584906,
+      "eval_PRM F1": 0.9390243902439024,
+      "eval_PRM F1 AUC": 0.8768988999476165,
+      "eval_PRM F1 Neg": 0.7916666666666666,
+      "eval_PRM NPV": 0.76,
+      "eval_PRM Precision": 0.9506172839506173,
+      "eval_PRM Recall": 0.927710843373494,
+      "eval_PRM Specificty": 0.8260869565217391,
+      "eval_loss": 0.22362171113491058,
+      "eval_runtime": 18.5552,
+      "eval_samples_per_second": 2.371,
+      "eval_steps_per_second": 0.162,
+      "step": 3540
+    },
+    {
+      "epoch": 0.9170308504095044,
+      "grad_norm": 0.08280863479085442,
+      "learning_rate": 2.078965409104622e-06,
+      "loss": 0.3016,
+      "step": 3541
+    },
+    {
+      "epoch": 0.9172898255155223,
+      "grad_norm": 0.09667067188226519,
+      "learning_rate": 2.0660822731533413e-06,
+      "loss": 0.3085,
+      "step": 3542
+    },
+    {
+      "epoch": 0.9175488006215402,
+      "grad_norm": 0.107532983693228,
+      "learning_rate": 2.0532383369264662e-06,
+      "loss": 0.3443,
+      "step": 3543
+    },
+    {
+      "epoch": 0.9178077757275582,
+      "grad_norm": 0.07981002509105538,
+      "learning_rate": 2.0404336109275944e-06,
+      "loss": 0.3379,
+      "step": 3544
+    },
+    {
+      "epoch": 0.9180667508335761,
+      "grad_norm": 0.10828869927777779,
+      "learning_rate": 2.0276681056282608e-06,
+      "loss": 0.3084,
+      "step": 3545
+    },
+    {
+      "epoch": 0.9180667508335761,
+      "eval_PRM Accuracy": 0.9056603773584906,
+      "eval_PRM F1": 0.9390243902439024,
+      "eval_PRM F1 AUC": 0.8768988999476165,
+      "eval_PRM F1 Neg": 0.7916666666666666,
+      "eval_PRM NPV": 0.76,
+      "eval_PRM Precision": 0.9506172839506173,
+      "eval_PRM Recall": 0.927710843373494,
+      "eval_PRM Specificty": 0.8260869565217391,
+      "eval_loss": 0.22402121126651764,
+      "eval_runtime": 18.7305,
+      "eval_samples_per_second": 2.349,
+      "eval_steps_per_second": 0.16,
+      "step": 3545
+    },
+    {
+      "epoch": 0.918325725939594,
+      "grad_norm": 0.1132998082263914,
+      "learning_rate": 2.0149418314679313e-06,
+      "loss": 0.3809,
+      "step": 3546
+    },
+    {
+      "epoch": 0.918584701045612,
+      "grad_norm": 0.09146593239908724,
+      "learning_rate": 2.0022547988539863e-06,
+      "loss": 0.304,
+      "step": 3547
+    },
+    {
+      "epoch": 0.9188436761516299,
+      "grad_norm": 0.08165379354498084,
+      "learning_rate": 1.9896070181617043e-06,
+      "loss": 0.2769,
+      "step": 3548
+    },
+    {
+      "epoch": 0.9191026512576479,
+      "grad_norm": 0.11447548063841524,
+      "learning_rate": 1.976998499734284e-06,
+      "loss": 0.422,
+      "step": 3549
+    },
+    {
+      "epoch": 0.9193616263636658,
+      "grad_norm": 0.10072150725196698,
+      "learning_rate": 1.9644292538827992e-06,
+      "loss": 0.3265,
+      "step": 3550
+    },
+    {
+      "epoch": 0.9193616263636658,
+      "eval_PRM Accuracy": 0.9056603773584906,
+      "eval_PRM F1": 0.9390243902439024,
+      "eval_PRM F1 AUC": 0.8768988999476165,
+      "eval_PRM F1 Neg": 0.7916666666666666,
+      "eval_PRM NPV": 0.76,
+      "eval_PRM Precision": 0.9506172839506173,
+      "eval_PRM Recall": 0.927710843373494,
+      "eval_PRM Specificty": 0.8260869565217391,
+      "eval_loss": 0.22364391386508942,
+      "eval_runtime": 18.2461,
+      "eval_samples_per_second": 2.411,
+      "eval_steps_per_second": 0.164,
+      "step": 3550
+    },
+    {
+      "epoch": 0.9196206014696837,
+      "grad_norm": 0.13312616712317882,
+      "learning_rate": 1.9518992908862168e-06,
+      "loss": 0.3733,
+      "step": 3551
+    },
+    {
+      "epoch": 0.9198795765757016,
+      "grad_norm": 0.0892283289249243,
+      "learning_rate": 1.9394086209913563e-06,
+      "loss": 0.3281,
+      "step": 3552
+    },
+    {
+      "epoch": 0.9201385516817195,
+      "grad_norm": 0.11810980552161127,
+      "learning_rate": 1.9269572544129532e-06,
+      "loss": 0.411,
+      "step": 3553
+    },
+    {
+      "epoch": 0.9203975267877376,
+      "grad_norm": 0.08385957873259561,
+      "learning_rate": 1.9145452013335506e-06,
+      "loss": 0.3065,
+      "step": 3554
+    },
+    {
+      "epoch": 0.9206565018937555,
+      "grad_norm": 0.11171647325084338,
+      "learning_rate": 1.9021724719035628e-06,
+      "loss": 0.2654,
+      "step": 3555
+    },
+    {
+      "epoch": 0.9206565018937555,
+      "eval_PRM Accuracy": 0.9056603773584906,
+      "eval_PRM F1": 0.9390243902439024,
+      "eval_PRM F1 AUC": 0.8768988999476165,
+      "eval_PRM F1 Neg": 0.7916666666666666,
+      "eval_PRM NPV": 0.76,
+      "eval_PRM Precision": 0.9506172839506173,
+      "eval_PRM Recall": 0.927710843373494,
+      "eval_PRM Specificty": 0.8260869565217391,
+      "eval_loss": 0.22574129700660706,
+      "eval_runtime": 18.287,
+      "eval_samples_per_second": 2.406,
+      "eval_steps_per_second": 0.164,
+      "step": 3555
+    },
+    {
+      "epoch": 0.9209154769997734,
+      "grad_norm": 0.10539260009326258,
+      "learning_rate": 1.8898390762412566e-06,
+      "loss": 0.3813,
+      "step": 3556
+    },
+    {
+      "epoch": 0.9211744521057913,
+      "grad_norm": 0.10417784512360244,
+      "learning_rate": 1.877545024432703e-06,
+      "loss": 0.343,
+      "step": 3557
+    },
+    {
+      "epoch": 0.9214334272118093,
+      "grad_norm": 0.14123116595735288,
+      "learning_rate": 1.8652903265318323e-06,
+      "loss": 0.3837,
+      "step": 3558
+    },
+    {
+      "epoch": 0.9216924023178272,
+      "grad_norm": 0.08193793782314669,
+      "learning_rate": 1.8530749925603608e-06,
+      "loss": 0.295,
+      "step": 3559
+    },
+    {
+      "epoch": 0.9219513774238451,
+      "grad_norm": 0.09090184886879703,
+      "learning_rate": 1.8408990325078313e-06,
+      "loss": 0.3454,
+      "step": 3560
+    },
+    {
+      "epoch": 0.9219513774238451,
+      "eval_PRM Accuracy": 0.9056603773584906,
+      "eval_PRM F1": 0.9390243902439024,
+      "eval_PRM F1 AUC": 0.8768988999476165,
+      "eval_PRM F1 Neg": 0.7916666666666666,
+      "eval_PRM NPV": 0.76,
+      "eval_PRM Precision": 0.9506172839506173,
+      "eval_PRM Recall": 0.927710843373494,
+      "eval_PRM Specificty": 0.8260869565217391,
+      "eval_loss": 0.22428755462169647,
+      "eval_runtime": 18.7023,
+      "eval_samples_per_second": 2.353,
+      "eval_steps_per_second": 0.16,
+      "step": 3560
+    },
+    {
+      "epoch": 0.922210352529863,
+      "grad_norm": 0.13661510498457352,
+      "learning_rate": 1.8287624563315841e-06,
+      "loss": 0.3709,
+      "step": 3561
+    },
+    {
+      "epoch": 0.922469327635881,
+      "grad_norm": 0.12624571608918805,
+      "learning_rate": 1.8166652739567413e-06,
+      "loss": 0.3841,
+      "step": 3562
+    },
+    {
+      "epoch": 0.922728302741899,
+      "grad_norm": 0.10371161289191488,
+      "learning_rate": 1.8046074952762393e-06,
+      "loss": 0.3336,
+      "step": 3563
+    },
+    {
+      "epoch": 0.9229872778479169,
+      "grad_norm": 0.1126383670009335,
+      "learning_rate": 1.792589130150757e-06,
+      "loss": 0.332,
+      "step": 3564
+    },
+    {
+      "epoch": 0.9232462529539348,
+      "grad_norm": 0.0911321078303712,
+      "learning_rate": 1.7806101884087544e-06,
+      "loss": 0.2784,
+      "step": 3565
+    },
+    {
+      "epoch": 0.9232462529539348,
+      "eval_PRM Accuracy": 0.9150943396226415,
+      "eval_PRM F1": 0.9447852760736196,
+      "eval_PRM F1 AUC": 0.8986380303823991,
+      "eval_PRM F1 Neg": 0.8163265306122449,
+      "eval_PRM NPV": 0.7692307692307693,
+      "eval_PRM Precision": 0.9625,
+      "eval_PRM Recall": 0.927710843373494,
+      "eval_PRM Specificty": 0.8695652173913043,
+      "eval_loss": 0.2246759533882141,
+      "eval_runtime": 19.1409,
+      "eval_samples_per_second": 2.299,
+      "eval_steps_per_second": 0.157,
+      "step": 3565
+    },
+    {
+      "epoch": 0.9235052280599527,
+      "grad_norm": 0.1213933793931651,
+      "learning_rate": 1.768670679846457e-06,
+      "loss": 0.348,
+      "step": 3566
+    },
+    {
+      "epoch": 0.9237642031659706,
+      "grad_norm": 0.09603766075813287,
+      "learning_rate": 1.7567706142278317e-06,
+      "loss": 0.3627,
+      "step": 3567
+    },
+    {
+      "epoch": 0.9240231782719887,
+      "grad_norm": 0.08731222298461713,
+      "learning_rate": 1.7449100012845942e-06,
+      "loss": 0.322,
+      "step": 3568
+    },
+    {
+      "epoch": 0.9242821533780066,
+      "grad_norm": 0.24315897296309427,
+      "learning_rate": 1.7330888507162023e-06,
+      "loss": 0.3502,
+      "step": 3569
+    },
+    {
+      "epoch": 0.9245411284840245,
+      "grad_norm": 0.10629631222617211,
+      "learning_rate": 1.7213071721898344e-06,
+      "loss": 0.3702,
+      "step": 3570
+    },
+    {
+      "epoch": 0.9245411284840245,
+      "eval_PRM Accuracy": 0.9056603773584906,
+      "eval_PRM F1": 0.9390243902439024,
+      "eval_PRM F1 AUC": 0.8768988999476165,
+      "eval_PRM F1 Neg": 0.7916666666666666,
+      "eval_PRM NPV": 0.76,
+      "eval_PRM Precision": 0.9506172839506173,
+      "eval_PRM Recall": 0.927710843373494,
+      "eval_PRM Specificty": 0.8260869565217391,
+      "eval_loss": 0.22579678893089294,
+      "eval_runtime": 18.5262,
+      "eval_samples_per_second": 2.375,
+      "eval_steps_per_second": 0.162,
+      "step": 3570
+    },
+    {
+      "epoch": 0.9248001035900424,
+      "grad_norm": 0.09237349493810404,
+      "learning_rate": 1.7095649753403887e-06,
+      "loss": 0.3127,
+      "step": 3571
+    },
+    {
+      "epoch": 0.9250590786960603,
+      "grad_norm": 0.09266671910207464,
+      "learning_rate": 1.6978622697704727e-06,
+      "loss": 0.3301,
+      "step": 3572
+    },
+    {
+      "epoch": 0.9253180538020783,
+      "grad_norm": 0.10002993867164743,
+      "learning_rate": 1.6861990650504255e-06,
+      "loss": 0.3213,
+      "step": 3573
+    },
+    {
+      "epoch": 0.9255770289080962,
+      "grad_norm": 0.10323222583912185,
+      "learning_rate": 1.67457537071824e-06,
+      "loss": 0.3205,
+      "step": 3574
+    },
+    {
+      "epoch": 0.9258360040141141,
+      "grad_norm": 0.10596902791544177,
+      "learning_rate": 1.6629911962796397e-06,
+      "loss": 0.3512,
+      "step": 3575
+    },
+    {
+      "epoch": 0.9258360040141141,
+      "eval_PRM Accuracy": 0.9056603773584906,
+      "eval_PRM F1": 0.9390243902439024,
+      "eval_PRM F1 AUC": 0.8768988999476165,
+      "eval_PRM F1 Neg": 0.7916666666666666,
+      "eval_PRM NPV": 0.76,
+      "eval_PRM Precision": 0.9506172839506173,
+      "eval_PRM Recall": 0.927710843373494,
+      "eval_PRM Specificty": 0.8260869565217391,
+      "eval_loss": 0.22654031217098236,
+      "eval_runtime": 18.7259,
+      "eval_samples_per_second": 2.35,
+      "eval_steps_per_second": 0.16,
+      "step": 3575
+    },
+    {
+      "epoch": 0.9260949791201321,
+      "grad_norm": 0.10085112686321387,
+      "learning_rate": 1.651446551207997e-06,
+      "loss": 0.3993,
+      "step": 3576
+    },
+    {
+      "epoch": 0.92635395422615,
+      "grad_norm": 0.11494210943222248,
+      "learning_rate": 1.639941444944365e-06,
+      "loss": 0.4243,
+      "step": 3577
+    },
+    {
+      "epoch": 0.926612929332168,
+      "grad_norm": 0.06779234169606624,
+      "learning_rate": 1.6284758868974737e-06,
+      "loss": 0.2865,
+      "step": 3578
+    },
+    {
+      "epoch": 0.9268719044381859,
+      "grad_norm": 0.16661008565924487,
+      "learning_rate": 1.617049886443711e-06,
+      "loss": 0.3595,
+      "step": 3579
+    },
+    {
+      "epoch": 0.9271308795442038,
+      "grad_norm": 0.12645216691826167,
+      "learning_rate": 1.6056634529270976e-06,
+      "loss": 0.3419,
+      "step": 3580
+    },
+    {
+      "epoch": 0.9271308795442038,
+      "eval_PRM Accuracy": 0.9056603773584906,
+      "eval_PRM F1": 0.9390243902439024,
+      "eval_PRM F1 AUC": 0.8768988999476165,
+      "eval_PRM F1 Neg": 0.7916666666666666,
+      "eval_PRM NPV": 0.76,
+      "eval_PRM Precision": 0.9506172839506173,
+      "eval_PRM Recall": 0.927710843373494,
+      "eval_PRM Specificty": 0.8260869565217391,
+      "eval_loss": 0.22531960904598236,
+      "eval_runtime": 18.912,
+      "eval_samples_per_second": 2.327,
+      "eval_steps_per_second": 0.159,
+      "step": 3580
+    },
+    {
+      "epoch": 0.9273898546502217,
+      "grad_norm": 0.10583473752729956,
+      "learning_rate": 1.5943165956593176e-06,
+      "loss": 0.4103,
+      "step": 3581
+    },
+    {
+      "epoch": 0.9276488297562396,
+      "grad_norm": 0.1328157919975935,
+      "learning_rate": 1.5830093239196764e-06,
+      "loss": 0.4228,
+      "step": 3582
+    },
+    {
+      "epoch": 0.9279078048622577,
+      "grad_norm": 0.09297802010235966,
+      "learning_rate": 1.5717416469551105e-06,
+      "loss": 0.3961,
+      "step": 3583
+    },
+    {
+      "epoch": 0.9281667799682756,
+      "grad_norm": 0.10086525622979277,
+      "learning_rate": 1.5605135739801824e-06,
+      "loss": 0.4108,
+      "step": 3584
+    },
+    {
+      "epoch": 0.9284257550742935,
+      "grad_norm": 0.11762400780726545,
+      "learning_rate": 1.5493251141770525e-06,
+      "loss": 0.4119,
+      "step": 3585
+    },
+    {
+      "epoch": 0.9284257550742935,
+      "eval_PRM Accuracy": 0.9150943396226415,
+      "eval_PRM F1": 0.9447852760736196,
+      "eval_PRM F1 AUC": 0.8986380303823991,
+      "eval_PRM F1 Neg": 0.8163265306122449,
+      "eval_PRM NPV": 0.7692307692307693,
+      "eval_PRM Precision": 0.9625,
+      "eval_PRM Recall": 0.927710843373494,
+      "eval_PRM Specificty": 0.8695652173913043,
+      "eval_loss": 0.2257080078125,
+      "eval_runtime": 18.401,
+      "eval_samples_per_second": 2.391,
+      "eval_steps_per_second": 0.163,
+      "step": 3585
+    },
+    {
+      "epoch": 0.9286847301803114,
+      "grad_norm": 0.15070292537465907,
+      "learning_rate": 1.5381762766955077e-06,
+      "loss": 0.4072,
+      "step": 3586
+    },
+    {
+      "epoch": 0.9289437052863293,
+      "grad_norm": 0.09903602312232185,
+      "learning_rate": 1.52706707065291e-06,
+      "loss": 0.3195,
+      "step": 3587
+    },
+    {
+      "epoch": 0.9292026803923473,
+      "grad_norm": 0.10333240456943571,
+      "learning_rate": 1.5159975051342256e-06,
+      "loss": 0.3748,
+      "step": 3588
+    },
+    {
+      "epoch": 0.9294616554983652,
+      "grad_norm": 0.10743198095807449,
+      "learning_rate": 1.5049675891919968e-06,
+      "loss": 0.3262,
+      "step": 3589
+    },
+    {
+      "epoch": 0.9297206306043831,
+      "grad_norm": 0.15036510684459076,
+      "learning_rate": 1.493977331846347e-06,
+      "loss": 0.3834,
+      "step": 3590
+    },
+    {
+      "epoch": 0.9297206306043831,
+      "eval_PRM Accuracy": 0.9150943396226415,
+      "eval_PRM F1": 0.9447852760736196,
+      "eval_PRM F1 AUC": 0.8986380303823991,
+      "eval_PRM F1 Neg": 0.8163265306122449,
+      "eval_PRM NPV": 0.7692307692307693,
+      "eval_PRM Precision": 0.9625,
+      "eval_PRM Recall": 0.927710843373494,
+      "eval_PRM Specificty": 0.8695652173913043,
+      "eval_loss": 0.2261407971382141,
+      "eval_runtime": 18.8141,
+      "eval_samples_per_second": 2.339,
+      "eval_steps_per_second": 0.159,
+      "step": 3590
+    },
+    {
+      "epoch": 0.9299796057104011,
+      "grad_norm": 0.12494442231887389,
+      "learning_rate": 1.4830267420849585e-06,
+      "loss": 0.3193,
+      "step": 3591
+    },
+    {
+      "epoch": 0.930238580816419,
+      "grad_norm": 0.08747387641847369,
+      "learning_rate": 1.4721158288630854e-06,
+      "loss": 0.3041,
+      "step": 3592
+    },
+    {
+      "epoch": 0.930497555922437,
+      "grad_norm": 0.08396679546067763,
+      "learning_rate": 1.4612446011035174e-06,
+      "loss": 0.2845,
+      "step": 3593
+    },
+    {
+      "epoch": 0.9307565310284549,
+      "grad_norm": 0.10698222955962103,
+      "learning_rate": 1.4504130676966155e-06,
+      "loss": 0.3104,
+      "step": 3594
+    },
+    {
+      "epoch": 0.9310155061344728,
+      "grad_norm": 0.13397290796286093,
+      "learning_rate": 1.4396212375002494e-06,
+      "loss": 0.4074,
+      "step": 3595
+    },
+    {
+      "epoch": 0.9310155061344728,
+      "eval_PRM Accuracy": 0.8962264150943396,
+      "eval_PRM F1": 0.9333333333333333,
+      "eval_PRM F1 AUC": 0.855159769512834,
+      "eval_PRM F1 Neg": 0.7659574468085106,
+      "eval_PRM NPV": 0.75,
+      "eval_PRM Precision": 0.9390243902439024,
+      "eval_PRM Recall": 0.927710843373494,
+      "eval_PRM Specificty": 0.782608695652174,
+      "eval_loss": 0.22743918001651764,
+      "eval_runtime": 18.349,
+      "eval_samples_per_second": 2.398,
+      "eval_steps_per_second": 0.163,
+      "step": 3595
+    },
+    {
+      "epoch": 0.9312744812404907,
+      "grad_norm": 0.08710733594249344,
+      "learning_rate": 1.4288691193398486e-06,
+      "loss": 0.2953,
+      "step": 3596
+    },
+    {
+      "epoch": 0.9315334563465086,
+      "grad_norm": 0.11295805821958349,
+      "learning_rate": 1.4181567220083403e-06,
+      "loss": 0.3008,
+      "step": 3597
+    },
+    {
+      "epoch": 0.9317924314525267,
+      "grad_norm": 0.1241124134705868,
+      "learning_rate": 1.407484054266195e-06,
+      "loss": 0.3757,
+      "step": 3598
+    },
+    {
+      "epoch": 0.9320514065585446,
+      "grad_norm": 0.0936476211152951,
+      "learning_rate": 1.396851124841364e-06,
+      "loss": 0.3198,
+      "step": 3599
+    },
+    {
+      "epoch": 0.9323103816645625,
+      "grad_norm": 0.10964181400570848,
+      "learning_rate": 1.3862579424293365e-06,
+      "loss": 0.3485,
+      "step": 3600
+    },
+    {
+      "epoch": 0.9323103816645625,
+      "eval_PRM Accuracy": 0.9150943396226415,
+      "eval_PRM F1": 0.9447852760736196,
+      "eval_PRM F1 AUC": 0.8986380303823991,
+      "eval_PRM F1 Neg": 0.8163265306122449,
+      "eval_PRM NPV": 0.7692307692307693,
+      "eval_PRM Precision": 0.9625,
+      "eval_PRM Recall": 0.927710843373494,
+      "eval_PRM Specificty": 0.8695652173913043,
+      "eval_loss": 0.22537508606910706,
+      "eval_runtime": 18.9763,
+      "eval_samples_per_second": 2.319,
+      "eval_steps_per_second": 0.158,
+      "step": 3600
+    },
+    {
+      "epoch": 0.9325693567705804,
+      "grad_norm": 0.09739916386290368,
+      "learning_rate": 1.3757045156930548e-06,
+      "loss": 0.3259,
+      "step": 3601
+    },
+    {
+      "epoch": 0.9328283318765984,
+      "grad_norm": 0.08006200450639955,
+      "learning_rate": 1.3651908532629877e-06,
+      "loss": 0.3195,
+      "step": 3602
+    },
+    {
+      "epoch": 0.9330873069826163,
+      "grad_norm": 0.09242188579141536,
+      "learning_rate": 1.3547169637370626e-06,
+      "loss": 0.2899,
+      "step": 3603
+    },
+    {
+      "epoch": 0.9333462820886342,
+      "grad_norm": 0.08353567321669103,
+      "learning_rate": 1.3442828556806896e-06,
+      "loss": 0.3087,
+      "step": 3604
+    },
+    {
+      "epoch": 0.9336052571946521,
+      "grad_norm": 0.10089107835607052,
+      "learning_rate": 1.3338885376267486e-06,
+      "loss": 0.4317,
+      "step": 3605
+    },
+    {
+      "epoch": 0.9336052571946521,
+      "eval_PRM Accuracy": 0.9150943396226415,
+      "eval_PRM F1": 0.9447852760736196,
+      "eval_PRM F1 AUC": 0.8986380303823991,
+      "eval_PRM F1 Neg": 0.8163265306122449,
+      "eval_PRM NPV": 0.7692307692307693,
+      "eval_PRM Precision": 0.9625,
+      "eval_PRM Recall": 0.927710843373494,
+      "eval_PRM Specificty": 0.8695652173913043,
+      "eval_loss": 0.22606311738491058,
+      "eval_runtime": 18.5717,
+      "eval_samples_per_second": 2.369,
+      "eval_steps_per_second": 0.162,
+      "step": 3605
+    },
+    {
+      "epoch": 0.9338642323006701,
+      "grad_norm": 0.11157162695694962,
+      "learning_rate": 1.3235340180755673e-06,
+      "loss": 0.3032,
+      "step": 3606
+    },
+    {
+      "epoch": 0.9341232074066881,
+      "grad_norm": 0.08418067712195422,
+      "learning_rate": 1.3132193054949394e-06,
+      "loss": 0.2889,
+      "step": 3607
+    },
+    {
+      "epoch": 0.934382182512706,
+      "grad_norm": 0.11270593387288827,
+      "learning_rate": 1.3029444083201058e-06,
+      "loss": 0.3984,
+      "step": 3608
+    },
+    {
+      "epoch": 0.9346411576187239,
+      "grad_norm": 0.07796434380501502,
+      "learning_rate": 1.292709334953729e-06,
+      "loss": 0.3162,
+      "step": 3609
+    },
+    {
+      "epoch": 0.9349001327247418,
+      "grad_norm": 0.11693373305810187,
+      "learning_rate": 1.2825140937659242e-06,
+      "loss": 0.268,
+      "step": 3610
+    },
+    {
+      "epoch": 0.9349001327247418,
+      "eval_PRM Accuracy": 0.9150943396226415,
+      "eval_PRM F1": 0.9447852760736196,
+      "eval_PRM F1 AUC": 0.8986380303823991,
+      "eval_PRM F1 Neg": 0.8163265306122449,
+      "eval_PRM NPV": 0.7692307692307693,
+      "eval_PRM Precision": 0.9625,
+      "eval_PRM Recall": 0.927710843373494,
+      "eval_PRM Specificty": 0.8695652173913043,
+      "eval_loss": 0.22525301575660706,
+      "eval_runtime": 18.3518,
+      "eval_samples_per_second": 2.398,
+      "eval_steps_per_second": 0.163,
+      "step": 3610
+    },
+    {
+      "epoch": 0.9351591078307597,
+      "grad_norm": 0.1464448621158321,
+      "learning_rate": 1.2723586930942221e-06,
+      "loss": 0.3719,
+      "step": 3611
+    },
+    {
+      "epoch": 0.9354180829367778,
+      "grad_norm": 0.10758545732247717,
+      "learning_rate": 1.2622431412435799e-06,
+      "loss": 0.2895,
+      "step": 3612
+    },
+    {
+      "epoch": 0.9356770580427957,
+      "grad_norm": 0.12754220261163823,
+      "learning_rate": 1.2521674464863576e-06,
+      "loss": 0.3623,
+      "step": 3613
+    },
+    {
+      "epoch": 0.9359360331488136,
+      "grad_norm": 0.11228429107248397,
+      "learning_rate": 1.242131617062331e-06,
+      "loss": 0.4263,
+      "step": 3614
+    },
+    {
+      "epoch": 0.9361950082548315,
+      "grad_norm": 0.08046814469745639,
+      "learning_rate": 1.2321356611786627e-06,
+      "loss": 0.2662,
+      "step": 3615
+    },
+    {
+      "epoch": 0.9361950082548315,
+      "eval_PRM Accuracy": 0.9150943396226415,
+      "eval_PRM F1": 0.9447852760736196,
+      "eval_PRM F1 AUC": 0.8986380303823991,
+      "eval_PRM F1 Neg": 0.8163265306122449,
+      "eval_PRM NPV": 0.7692307692307693,
+      "eval_PRM Precision": 0.9625,
+      "eval_PRM Recall": 0.927710843373494,
+      "eval_PRM Specificty": 0.8695652173913043,
+      "eval_loss": 0.22554154694080353,
+      "eval_runtime": 18.8854,
+      "eval_samples_per_second": 2.33,
+      "eval_steps_per_second": 0.159,
+      "step": 3615
+    },
+    {
+      "epoch": 0.9364539833608494,
+      "grad_norm": 0.0872818568564886,
+      "learning_rate": 1.2221795870099195e-06,
+      "loss": 0.2685,
+      "step": 3616
+    },
+    {
+      "epoch": 0.9367129584668674,
+      "grad_norm": 0.11421026114058384,
+      "learning_rate": 1.212263402698044e-06,
+      "loss": 0.4479,
+      "step": 3617
+    },
+    {
+      "epoch": 0.9369719335728853,
+      "grad_norm": 0.09061094476483011,
+      "learning_rate": 1.2023871163523549e-06,
+      "loss": 0.2815,
+      "step": 3618
+    },
+    {
+      "epoch": 0.9372309086789032,
+      "grad_norm": 0.08204366239605856,
+      "learning_rate": 1.192550736049558e-06,
+      "loss": 0.313,
+      "step": 3619
+    },
+    {
+      "epoch": 0.9374898837849212,
+      "grad_norm": 0.06964065504874097,
+      "learning_rate": 1.182754269833708e-06,
+      "loss": 0.2756,
+      "step": 3620
+    },
+    {
+      "epoch": 0.9374898837849212,
+      "eval_PRM Accuracy": 0.9150943396226415,
+      "eval_PRM F1": 0.9447852760736196,
+      "eval_PRM F1 AUC": 0.8986380303823991,
+      "eval_PRM F1 Neg": 0.8163265306122449,
+      "eval_PRM NPV": 0.7692307692307693,
+      "eval_PRM Precision": 0.9625,
+      "eval_PRM Recall": 0.927710843373494,
+      "eval_PRM Specificty": 0.8695652173913043,
+      "eval_loss": 0.22608532011508942,
+      "eval_runtime": 18.4563,
+      "eval_samples_per_second": 2.384,
+      "eval_steps_per_second": 0.163,
+      "step": 3620
+    },
+    {
+      "epoch": 0.9377488588909391,
+      "grad_norm": 0.10169554692300896,
+      "learning_rate": 1.1729977257162239e-06,
+      "loss": 0.3062,
+      "step": 3621
+    },
+    {
+      "epoch": 0.9380078339969571,
+      "grad_norm": 0.09702243276312277,
+      "learning_rate": 1.1632811116758846e-06,
+      "loss": 0.3102,
+      "step": 3622
+    },
+    {
+      "epoch": 0.938266809102975,
+      "grad_norm": 0.08870170822062395,
+      "learning_rate": 1.1536044356588116e-06,
+      "loss": 0.351,
+      "step": 3623
+    },
+    {
+      "epoch": 0.9385257842089929,
+      "grad_norm": 0.12378921239424223,
+      "learning_rate": 1.143967705578458e-06,
+      "loss": 0.3665,
+      "step": 3624
+    },
+    {
+      "epoch": 0.9387847593150108,
+      "grad_norm": 0.12612374115144812,
+      "learning_rate": 1.1343709293156201e-06,
+      "loss": 0.3441,
+      "step": 3625
+    },
+    {
+      "epoch": 0.9387847593150108,
+      "eval_PRM Accuracy": 0.9150943396226415,
+      "eval_PRM F1": 0.9447852760736196,
+      "eval_PRM F1 AUC": 0.8986380303823991,
+      "eval_PRM F1 Neg": 0.8163265306122449,
+      "eval_PRM NPV": 0.7692307692307693,
+      "eval_PRM Precision": 0.9625,
+      "eval_PRM Recall": 0.927710843373494,
+      "eval_PRM Specificty": 0.8695652173913043,
+      "eval_loss": 0.22636274993419647,
+      "eval_runtime": 19.1501,
+      "eval_samples_per_second": 2.298,
+      "eval_steps_per_second": 0.157,
+      "step": 3625
+    },
+    {
+      "epoch": 0.9390437344210287,
+      "grad_norm": 0.08546975773794105,
+      "learning_rate": 1.1248141147184033e-06,
+      "loss": 0.2978,
+      "step": 3626
+    },
+    {
+      "epoch": 0.9393027095270468,
+      "grad_norm": 0.1451168289131301,
+      "learning_rate": 1.1152972696022445e-06,
+      "loss": 0.4364,
+      "step": 3627
+    },
+    {
+      "epoch": 0.9395616846330647,
+      "grad_norm": 0.08853201754452705,
+      "learning_rate": 1.1058204017499073e-06,
+      "loss": 0.3291,
+      "step": 3628
+    },
+    {
+      "epoch": 0.9398206597390826,
+      "grad_norm": 0.11478234636160796,
+      "learning_rate": 1.0963835189114368e-06,
+      "loss": 0.3681,
+      "step": 3629
+    },
+    {
+      "epoch": 0.9400796348451005,
+      "grad_norm": 0.08829912173241744,
+      "learning_rate": 1.0869866288041986e-06,
+      "loss": 0.3329,
+      "step": 3630
+    },
+    {
+      "epoch": 0.9400796348451005,
+      "eval_PRM Accuracy": 0.9150943396226415,
+      "eval_PRM F1": 0.9447852760736196,
+      "eval_PRM F1 AUC": 0.8986380303823991,
+      "eval_PRM F1 Neg": 0.8163265306122449,
+      "eval_PRM NPV": 0.7692307692307693,
+      "eval_PRM Precision": 0.9625,
+      "eval_PRM Recall": 0.927710843373494,
+      "eval_PRM Specificty": 0.8695652173913043,
+      "eval_loss": 0.22632946074008942,
+      "eval_runtime": 18.2062,
+      "eval_samples_per_second": 2.417,
+      "eval_steps_per_second": 0.165,
+      "step": 3630
+    },
+    {
+      "epoch": 0.9403386099511184,
+      "grad_norm": 0.12303949860424165,
+      "learning_rate": 1.0776297391128399e-06,
+      "loss": 0.4269,
+      "step": 3631
+    },
+    {
+      "epoch": 0.9405975850571364,
+      "grad_norm": 0.08515169938822552,
+      "learning_rate": 1.0683128574892953e-06,
+      "loss": 0.3149,
+      "step": 3632
+    },
+    {
+      "epoch": 0.9408565601631543,
+      "grad_norm": 0.11638916235685758,
+      "learning_rate": 1.0590359915528092e-06,
+      "loss": 0.412,
+      "step": 3633
+    },
+    {
+      "epoch": 0.9411155352691722,
+      "grad_norm": 0.10050987960890728,
+      "learning_rate": 1.049799148889863e-06,
+      "loss": 0.3879,
+      "step": 3634
+    },
+    {
+      "epoch": 0.9413745103751902,
+      "grad_norm": 0.08183166340406944,
+      "learning_rate": 1.040602337054225e-06,
+      "loss": 0.3295,
+      "step": 3635
+    },
+    {
+      "epoch": 0.9413745103751902,
+      "eval_PRM Accuracy": 0.9150943396226415,
+      "eval_PRM F1": 0.9447852760736196,
+      "eval_PRM F1 AUC": 0.8986380303823991,
+      "eval_PRM F1 Neg": 0.8163265306122449,
+      "eval_PRM NPV": 0.7692307692307693,
+      "eval_PRM Precision": 0.9625,
+      "eval_PRM Recall": 0.927710843373494,
+      "eval_PRM Specificty": 0.8695652173913043,
+      "eval_loss": 0.2269842028617859,
+      "eval_runtime": 18.3309,
+      "eval_samples_per_second": 2.4,
+      "eval_steps_per_second": 0.164,
+      "step": 3635
+    },
+    {
+      "epoch": 0.9416334854812081,
+      "grad_norm": 0.17009816629912455,
+      "learning_rate": 1.0314455635669295e-06,
+      "loss": 0.4053,
+      "step": 3636
+    },
+    {
+      "epoch": 0.9418924605872261,
+      "grad_norm": 0.09348388794498838,
+      "learning_rate": 1.0223288359162642e-06,
+      "loss": 0.3621,
+      "step": 3637
+    },
+    {
+      "epoch": 0.942151435693244,
+      "grad_norm": 0.11993760317063545,
+      "learning_rate": 1.0132521615577707e-06,
+      "loss": 0.3754,
+      "step": 3638
+    },
+    {
+      "epoch": 0.9424104107992619,
+      "grad_norm": 0.09604337994143786,
+      "learning_rate": 1.0042155479142334e-06,
+      "loss": 0.3151,
+      "step": 3639
+    },
+    {
+      "epoch": 0.9426693859052798,
+      "grad_norm": 0.08778420938755499,
+      "learning_rate": 9.952190023756692e-07,
+      "loss": 0.3151,
+      "step": 3640
+    },
+    {
+      "epoch": 0.9426693859052798,
+      "eval_PRM Accuracy": 0.9150943396226415,
+      "eval_PRM F1": 0.9447852760736196,
+      "eval_PRM F1 AUC": 0.8986380303823991,
+      "eval_PRM F1 Neg": 0.8163265306122449,
+      "eval_PRM NPV": 0.7692307692307693,
+      "eval_PRM Precision": 0.9625,
+      "eval_PRM Recall": 0.927710843373494,
+      "eval_PRM Specificty": 0.8695652173913043,
+      "eval_loss": 0.22701749205589294,
+      "eval_runtime": 18.3403,
+      "eval_samples_per_second": 2.399,
+      "eval_steps_per_second": 0.164,
+      "step": 3640
+    },
+    {
+      "epoch": 0.9429283610112978,
+      "grad_norm": 0.12089551137093389,
+      "learning_rate": 9.862625322993423e-07,
+      "loss": 0.2798,
+      "step": 3641
+    },
+    {
+      "epoch": 0.9431873361173158,
+      "grad_norm": 0.10801932244428895,
+      "learning_rate": 9.773461450097277e-07,
+      "loss": 0.3269,
+      "step": 3642
+    },
+    {
+      "epoch": 0.9434463112233337,
+      "grad_norm": 0.0957428724034601,
+      "learning_rate": 9.684698477985366e-07,
+      "loss": 0.2274,
+      "step": 3643
+    },
+    {
+      "epoch": 0.9437052863293516,
+      "grad_norm": 0.1183556976739253,
+      "learning_rate": 9.596336479246848e-07,
+      "loss": 0.3833,
+      "step": 3644
+    },
+    {
+      "epoch": 0.9439642614353695,
+      "grad_norm": 0.09185542663292395,
+      "learning_rate": 9.508375526142976e-07,
+      "loss": 0.3099,
+      "step": 3645
+    },
+    {
+      "epoch": 0.9439642614353695,
+      "eval_PRM Accuracy": 0.9150943396226415,
+      "eval_PRM F1": 0.9447852760736196,
+      "eval_PRM F1 AUC": 0.8986380303823991,
+      "eval_PRM F1 Neg": 0.8163265306122449,
+      "eval_PRM NPV": 0.7692307692307693,
+      "eval_PRM Precision": 0.9625,
+      "eval_PRM Recall": 0.927710843373494,
+      "eval_PRM Specificty": 0.8695652173913043,
+      "eval_loss": 0.22652921080589294,
+      "eval_runtime": 18.6209,
+      "eval_samples_per_second": 2.363,
+      "eval_steps_per_second": 0.161,
+      "step": 3645
+    },
+    {
+      "epoch": 0.9442232365413875,
+      "grad_norm": 0.10091627256444724,
+      "learning_rate": 9.420815690607099e-07,
+      "loss": 0.3714,
+      "step": 3646
+    },
+    {
+      "epoch": 0.9444822116474054,
+      "grad_norm": 0.08022715015293055,
+      "learning_rate": 9.333657044244437e-07,
+      "loss": 0.2584,
+      "step": 3647
+    },
+    {
+      "epoch": 0.9447411867534233,
+      "grad_norm": 0.11596029999356494,
+      "learning_rate": 9.246899658332198e-07,
+      "loss": 0.3078,
+      "step": 3648
+    },
+    {
+      "epoch": 0.9450001618594412,
+      "grad_norm": 0.08804138919610968,
+      "learning_rate": 9.16054360381946e-07,
+      "loss": 0.3391,
+      "step": 3649
+    },
+    {
+      "epoch": 0.9452591369654592,
+      "grad_norm": 0.11098957301052466,
+      "learning_rate": 9.074588951327012e-07,
+      "loss": 0.3548,
+      "step": 3650
+    },
+    {
+      "epoch": 0.9452591369654592,
+      "eval_PRM Accuracy": 0.9150943396226415,
+      "eval_PRM F1": 0.9447852760736196,
+      "eval_PRM F1 AUC": 0.8986380303823991,
+      "eval_PRM F1 Neg": 0.8163265306122449,
+      "eval_PRM NPV": 0.7692307692307693,
+      "eval_PRM Precision": 0.9625,
+      "eval_PRM Recall": 0.927710843373494,
+      "eval_PRM Specificty": 0.8695652173913043,
+      "eval_loss": 0.22654031217098236,
+      "eval_runtime": 18.8005,
+      "eval_samples_per_second": 2.34,
+      "eval_steps_per_second": 0.16,
+      "step": 3650
+    },
+    {
+      "epoch": 0.9455181120714772,
+      "grad_norm": 0.0929468874939957,
+      "learning_rate": 8.989035771147458e-07,
+      "loss": 0.3856,
+      "step": 3651
+    },
+    {
+      "epoch": 0.9457770871774951,
+      "grad_norm": 0.11538838508671108,
+      "learning_rate": 8.903884133245055e-07,
+      "loss": 0.3007,
+      "step": 3652
+    },
+    {
+      "epoch": 0.946036062283513,
+      "grad_norm": 0.11474897160896774,
+      "learning_rate": 8.819134107255711e-07,
+      "loss": 0.3334,
+      "step": 3653
+    },
+    {
+      "epoch": 0.9462950373895309,
+      "grad_norm": 0.14748566589341697,
+      "learning_rate": 8.734785762486875e-07,
+      "loss": 0.342,
+      "step": 3654
+    },
+    {
+      "epoch": 0.9465540124955488,
+      "grad_norm": 0.12762969547058145,
+      "learning_rate": 8.650839167917479e-07,
+      "loss": 0.4551,
+      "step": 3655
+    },
+    {
+      "epoch": 0.9465540124955488,
+      "eval_PRM Accuracy": 0.9150943396226415,
+      "eval_PRM F1": 0.9447852760736196,
+      "eval_PRM F1 AUC": 0.8986380303823991,
+      "eval_PRM F1 Neg": 0.8163265306122449,
+      "eval_PRM NPV": 0.7692307692307693,
+      "eval_PRM Precision": 0.9625,
+      "eval_PRM Recall": 0.927710843373494,
+      "eval_PRM Specificty": 0.8695652173913043,
+      "eval_loss": 0.22630725800991058,
+      "eval_runtime": 18.4157,
+      "eval_samples_per_second": 2.389,
+      "eval_steps_per_second": 0.163,
+      "step": 3655
+    },
+    {
+      "epoch": 0.9468129876015668,
+      "grad_norm": 0.08539587703781182,
+      "learning_rate": 8.567294392197944e-07,
+      "loss": 0.2818,
+      "step": 3656
+    },
+    {
+      "epoch": 0.9470719627075848,
+      "grad_norm": 0.12180944803502,
+      "learning_rate": 8.484151503650062e-07,
+      "loss": 0.3767,
+      "step": 3657
+    },
+    {
+      "epoch": 0.9473309378136027,
+      "grad_norm": 0.09045119456210972,
+      "learning_rate": 8.401410570267054e-07,
+      "loss": 0.2778,
+      "step": 3658
+    },
+    {
+      "epoch": 0.9475899129196206,
+      "grad_norm": 0.0997640173003017,
+      "learning_rate": 8.319071659713351e-07,
+      "loss": 0.3507,
+      "step": 3659
+    },
+    {
+      "epoch": 0.9478488880256385,
+      "grad_norm": 0.08259407161217419,
+      "learning_rate": 8.237134839324479e-07,
+      "loss": 0.2752,
+      "step": 3660
+    },
+    {
+      "epoch": 0.9478488880256385,
+      "eval_PRM Accuracy": 0.9056603773584906,
+      "eval_PRM F1": 0.9390243902439024,
+      "eval_PRM F1 AUC": 0.8768988999476165,
+      "eval_PRM F1 Neg": 0.7916666666666666,
+      "eval_PRM NPV": 0.76,
+      "eval_PRM Precision": 0.9506172839506173,
+      "eval_PRM Recall": 0.927710843373494,
+      "eval_PRM Specificty": 0.8260869565217391,
+      "eval_loss": 0.22629617154598236,
+      "eval_runtime": 17.9513,
+      "eval_samples_per_second": 2.451,
+      "eval_steps_per_second": 0.167,
+      "step": 3660
+    },
+    {
+      "epoch": 0.9481078631316565,
+      "grad_norm": 0.09779587628759123,
+      "learning_rate": 8.155600176107503e-07,
+      "loss": 0.3059,
+      "step": 3661
+    },
+    {
+      "epoch": 0.9483668382376744,
+      "grad_norm": 0.1014513616417956,
+      "learning_rate": 8.074467736740254e-07,
+      "loss": 0.3413,
+      "step": 3662
+    },
+    {
+      "epoch": 0.9486258133436923,
+      "grad_norm": 0.12033694883782667,
+      "learning_rate": 7.993737587571826e-07,
+      "loss": 0.3738,
+      "step": 3663
+    },
+    {
+      "epoch": 0.9488847884497102,
+      "grad_norm": 0.09689683163413962,
+      "learning_rate": 7.913409794622295e-07,
+      "loss": 0.3212,
+      "step": 3664
+    },
+    {
+      "epoch": 0.9491437635557282,
+      "grad_norm": 0.1010382071051823,
+      "learning_rate": 7.83348442358256e-07,
+      "loss": 0.3219,
+      "step": 3665
+    },
+    {
+      "epoch": 0.9491437635557282,
+      "eval_PRM Accuracy": 0.9150943396226415,
+      "eval_PRM F1": 0.9447852760736196,
+      "eval_PRM F1 AUC": 0.8986380303823991,
+      "eval_PRM F1 Neg": 0.8163265306122449,
+      "eval_PRM NPV": 0.7692307692307693,
+      "eval_PRM Precision": 0.9625,
+      "eval_PRM Recall": 0.927710843373494,
+      "eval_PRM Specificty": 0.8695652173913043,
+      "eval_loss": 0.22635164856910706,
+      "eval_runtime": 19.0851,
+      "eval_samples_per_second": 2.305,
+      "eval_steps_per_second": 0.157,
+      "step": 3665
+    },
+    {
+      "epoch": 0.9494027386617462,
+      "grad_norm": 0.08601352806272027,
+      "learning_rate": 7.75396153981467e-07,
+      "loss": 0.3243,
+      "step": 3666
+    },
+    {
+      "epoch": 0.9496617137677641,
+      "grad_norm": 0.1017097706873693,
+      "learning_rate": 7.674841208351269e-07,
+      "loss": 0.4028,
+      "step": 3667
+    },
+    {
+      "epoch": 0.949920688873782,
+      "grad_norm": 0.11942107205019582,
+      "learning_rate": 7.596123493895991e-07,
+      "loss": 0.3623,
+      "step": 3668
+    },
+    {
+      "epoch": 0.9501796639797999,
+      "grad_norm": 0.11636618672772103,
+      "learning_rate": 7.517808460823117e-07,
+      "loss": 0.3429,
+      "step": 3669
+    },
+    {
+      "epoch": 0.9504386390858178,
+      "grad_norm": 0.10240518227004253,
+      "learning_rate": 7.439896173177585e-07,
+      "loss": 0.3585,
+      "step": 3670
+    },
+    {
+      "epoch": 0.9504386390858178,
+      "eval_PRM Accuracy": 0.9150943396226415,
+      "eval_PRM F1": 0.9447852760736196,
+      "eval_PRM F1 AUC": 0.8986380303823991,
+      "eval_PRM F1 Neg": 0.8163265306122449,
+      "eval_PRM NPV": 0.7692307692307693,
+      "eval_PRM Precision": 0.9625,
+      "eval_PRM Recall": 0.927710843373494,
+      "eval_PRM Specificty": 0.8695652173913043,
+      "eval_loss": 0.22620739042758942,
+      "eval_runtime": 18.2206,
+      "eval_samples_per_second": 2.415,
+      "eval_steps_per_second": 0.165,
+      "step": 3670
+    },
+    {
+      "epoch": 0.9506976141918358,
+      "grad_norm": 0.08981990776588318,
+      "learning_rate": 7.36238669467515e-07,
+      "loss": 0.341,
+      "step": 3671
+    },
+    {
+      "epoch": 0.9509565892978538,
+      "grad_norm": 0.10922759592859453,
+      "learning_rate": 7.285280088701996e-07,
+      "loss": 0.269,
+      "step": 3672
+    },
+    {
+      "epoch": 0.9512155644038717,
+      "grad_norm": 0.10458018237654504,
+      "learning_rate": 7.208576418314849e-07,
+      "loss": 0.3297,
+      "step": 3673
+    },
+    {
+      "epoch": 0.9514745395098896,
+      "grad_norm": 0.12857621761089671,
+      "learning_rate": 7.13227574624098e-07,
+      "loss": 0.3506,
+      "step": 3674
+    },
+    {
+      "epoch": 0.9517335146159075,
+      "grad_norm": 0.09517849652628264,
+      "learning_rate": 7.056378134878139e-07,
+      "loss": 0.3139,
+      "step": 3675
+    },
+    {
+      "epoch": 0.9517335146159075,
+      "eval_PRM Accuracy": 0.9056603773584906,
+      "eval_PRM F1": 0.9390243902439024,
+      "eval_PRM F1 AUC": 0.8768988999476165,
+      "eval_PRM F1 Neg": 0.7916666666666666,
+      "eval_PRM NPV": 0.76,
+      "eval_PRM Precision": 0.9506172839506173,
+      "eval_PRM Recall": 0.927710843373494,
+      "eval_PRM Specificty": 0.8260869565217391,
+      "eval_loss": 0.226318359375,
+      "eval_runtime": 18.9356,
+      "eval_samples_per_second": 2.324,
+      "eval_steps_per_second": 0.158,
+      "step": 3675
+    },
+    {
+      "epoch": 0.9519924897219255,
+      "grad_norm": 0.10045297428124696,
+      "learning_rate": 6.98088364629429e-07,
+      "loss": 0.3213,
+      "step": 3676
+    },
+    {
+      "epoch": 0.9522514648279434,
+      "grad_norm": 0.09812689394398072,
+      "learning_rate": 6.905792342227823e-07,
+      "loss": 0.3127,
+      "step": 3677
+    },
+    {
+      "epoch": 0.9525104399339613,
+      "grad_norm": 0.11555229241551024,
+      "learning_rate": 6.831104284087509e-07,
+      "loss": 0.3037,
+      "step": 3678
+    },
+    {
+      "epoch": 0.9527694150399793,
+      "grad_norm": 0.08862686504367377,
+      "learning_rate": 6.756819532952152e-07,
+      "loss": 0.3454,
+      "step": 3679
+    },
+    {
+      "epoch": 0.9530283901459972,
+      "grad_norm": 0.10936124493042092,
+      "learning_rate": 6.682938149570883e-07,
+      "loss": 0.2526,
+      "step": 3680
+    },
+    {
+      "epoch": 0.9530283901459972,
+      "eval_PRM Accuracy": 0.9056603773584906,
+      "eval_PRM F1": 0.9390243902439024,
+      "eval_PRM F1 AUC": 0.8768988999476165,
+      "eval_PRM F1 Neg": 0.7916666666666666,
+      "eval_PRM NPV": 0.76,
+      "eval_PRM Precision": 0.9506172839506173,
+      "eval_PRM Recall": 0.927710843373494,
+      "eval_PRM Specificty": 0.8260869565217391,
+      "eval_loss": 0.22763894498348236,
+      "eval_runtime": 18.8028,
+      "eval_samples_per_second": 2.34,
+      "eval_steps_per_second": 0.16,
+      "step": 3680
+    },
+    {
+      "epoch": 0.9532873652520152,
+      "grad_norm": 0.087450856603701,
+      "learning_rate": 6.609460194362927e-07,
+      "loss": 0.3525,
+      "step": 3681
+    },
+    {
+      "epoch": 0.9535463403580331,
+      "grad_norm": 0.08981180434110808,
+      "learning_rate": 6.536385727417604e-07,
+      "loss": 0.2972,
+      "step": 3682
+    },
+    {
+      "epoch": 0.953805315464051,
+      "grad_norm": 0.08538266214693427,
+      "learning_rate": 6.463714808494281e-07,
+      "loss": 0.3116,
+      "step": 3683
+    },
+    {
+      "epoch": 0.9540642905700689,
+      "grad_norm": 0.09006775756584584,
+      "learning_rate": 6.391447497022196e-07,
+      "loss": 0.3714,
+      "step": 3684
+    },
+    {
+      "epoch": 0.9543232656760869,
+      "grad_norm": 0.09989435538629539,
+      "learning_rate": 6.319583852100575e-07,
+      "loss": 0.3012,
+      "step": 3685
+    },
+    {
+      "epoch": 0.9543232656760869,
+      "eval_PRM Accuracy": 0.9150943396226415,
+      "eval_PRM F1": 0.9447852760736196,
+      "eval_PRM F1 AUC": 0.8986380303823991,
+      "eval_PRM F1 Neg": 0.8163265306122449,
+      "eval_PRM NPV": 0.7692307692307693,
+      "eval_PRM Precision": 0.9625,
+      "eval_PRM Recall": 0.927710843373494,
+      "eval_PRM Specificty": 0.8695652173913043,
+      "eval_loss": 0.2266845703125,
+      "eval_runtime": 18.8789,
+      "eval_samples_per_second": 2.331,
+      "eval_steps_per_second": 0.159,
+      "step": 3685
+    },
+    {
+      "epoch": 0.9545822407821049,
+      "grad_norm": 0.07348370953657105,
+      "learning_rate": 6.248123932498629e-07,
+      "loss": 0.2649,
+      "step": 3686
+    },
+    {
+      "epoch": 0.9548412158881228,
+      "grad_norm": 0.12473302674445111,
+      "learning_rate": 6.177067796655333e-07,
+      "loss": 0.3815,
+      "step": 3687
+    },
+    {
+      "epoch": 0.9551001909941407,
+      "grad_norm": 0.08708333203791523,
+      "learning_rate": 6.106415502679375e-07,
+      "loss": 0.2913,
+      "step": 3688
+    },
+    {
+      "epoch": 0.9553591661001586,
+      "grad_norm": 0.10164637912222271,
+      "learning_rate": 6.036167108349422e-07,
+      "loss": 0.3421,
+      "step": 3689
+    },
+    {
+      "epoch": 0.9556181412061766,
+      "grad_norm": 0.08272516199681351,
+      "learning_rate": 5.966322671113522e-07,
+      "loss": 0.3211,
+      "step": 3690
+    },
+    {
+      "epoch": 0.9556181412061766,
+      "eval_PRM Accuracy": 0.9150943396226415,
+      "eval_PRM F1": 0.9447852760736196,
+      "eval_PRM F1 AUC": 0.8986380303823991,
+      "eval_PRM F1 Neg": 0.8163265306122449,
+      "eval_PRM NPV": 0.7692307692307693,
+      "eval_PRM Precision": 0.9625,
+      "eval_PRM Recall": 0.927710843373494,
+      "eval_PRM Specificty": 0.8695652173913043,
+      "eval_loss": 0.22533069550991058,
+      "eval_runtime": 19.1151,
+      "eval_samples_per_second": 2.302,
+      "eval_steps_per_second": 0.157,
+      "step": 3690
+    },
+    {
+      "epoch": 0.9558771163121945,
+      "grad_norm": 0.09103199410344145,
+      "learning_rate": 5.896882248089597e-07,
+      "loss": 0.3482,
+      "step": 3691
+    },
+    {
+      "epoch": 0.9561360914182124,
+      "grad_norm": 0.10959139015215107,
+      "learning_rate": 5.827845896065165e-07,
+      "loss": 0.317,
+      "step": 3692
+    },
+    {
+      "epoch": 0.9563950665242303,
+      "grad_norm": 0.08463192745637686,
+      "learning_rate": 5.759213671497121e-07,
+      "loss": 0.3068,
+      "step": 3693
+    },
+    {
+      "epoch": 0.9566540416302483,
+      "grad_norm": 0.10667780893329848,
+      "learning_rate": 5.690985630512124e-07,
+      "loss": 0.2455,
+      "step": 3694
+    },
+    {
+      "epoch": 0.9569130167362663,
+      "grad_norm": 0.09443537303609133,
+      "learning_rate": 5.623161828906043e-07,
+      "loss": 0.3111,
+      "step": 3695
+    },
+    {
+      "epoch": 0.9569130167362663,
+      "eval_PRM Accuracy": 0.8962264150943396,
+      "eval_PRM F1": 0.9333333333333333,
+      "eval_PRM F1 AUC": 0.855159769512834,
+      "eval_PRM F1 Neg": 0.7659574468085106,
+      "eval_PRM NPV": 0.75,
+      "eval_PRM Precision": 0.9390243902439024,
+      "eval_PRM Recall": 0.927710843373494,
+      "eval_PRM Specificty": 0.782608695652174,
+      "eval_loss": 0.22683992981910706,
+      "eval_runtime": 18.4485,
+      "eval_samples_per_second": 2.385,
+      "eval_steps_per_second": 0.163,
+      "step": 3695
+    },
+    {
+      "epoch": 0.9571719918422842,
+      "grad_norm": 0.08095077767534335,
+      "learning_rate": 5.555742322144342e-07,
+      "loss": 0.2742,
+      "step": 3696
+    },
+    {
+      "epoch": 0.9574309669483021,
+      "grad_norm": 0.08671746999508634,
+      "learning_rate": 5.488727165361862e-07,
+      "loss": 0.3285,
+      "step": 3697
+    },
+    {
+      "epoch": 0.95768994205432,
+      "grad_norm": 0.1271392258408676,
+      "learning_rate": 5.422116413362599e-07,
+      "loss": 0.3732,
+      "step": 3698
+    },
+    {
+      "epoch": 0.9579489171603379,
+      "grad_norm": 0.09654699804086661,
+      "learning_rate": 5.355910120620034e-07,
+      "loss": 0.3805,
+      "step": 3699
+    },
+    {
+      "epoch": 0.9582078922663559,
+      "grad_norm": 0.0904339701173781,
+      "learning_rate": 5.290108341276745e-07,
+      "loss": 0.358,
+      "step": 3700
+    },
+    {
+      "epoch": 0.9582078922663559,
+      "eval_PRM Accuracy": 0.9056603773584906,
+      "eval_PRM F1": 0.9390243902439024,
+      "eval_PRM F1 AUC": 0.8768988999476165,
+      "eval_PRM F1 Neg": 0.7916666666666666,
+      "eval_PRM NPV": 0.76,
+      "eval_PRM Precision": 0.9506172839506173,
+      "eval_PRM Recall": 0.927710843373494,
+      "eval_PRM Specificty": 0.8260869565217391,
+      "eval_loss": 0.22634054720401764,
+      "eval_runtime": 18.4038,
+      "eval_samples_per_second": 2.391,
+      "eval_steps_per_second": 0.163,
+      "step": 3700
+    },
+    {
+      "epoch": 0.9584668673723739,
+      "grad_norm": 0.1228351574941874,
+      "learning_rate": 5.224711129144577e-07,
+      "loss": 0.4247,
+      "step": 3701
+    },
+    {
+      "epoch": 0.9587258424783918,
+      "grad_norm": 0.09586568782950836,
+      "learning_rate": 5.159718537704528e-07,
+      "loss": 0.2818,
+      "step": 3702
+    },
+    {
+      "epoch": 0.9589848175844097,
+      "grad_norm": 0.10197365382037611,
+      "learning_rate": 5.095130620106581e-07,
+      "loss": 0.3,
+      "step": 3703
+    },
+    {
+      "epoch": 0.9592437926904276,
+      "grad_norm": 0.09892851920127262,
+      "learning_rate": 5.030947429170041e-07,
+      "loss": 0.332,
+      "step": 3704
+    },
+    {
+      "epoch": 0.9595027677964456,
+      "grad_norm": 0.07992137934893262,
+      "learning_rate": 4.967169017382978e-07,
+      "loss": 0.2649,
+      "step": 3705
+    },
+    {
+      "epoch": 0.9595027677964456,
+      "eval_PRM Accuracy": 0.9150943396226415,
+      "eval_PRM F1": 0.9447852760736196,
+      "eval_PRM F1 AUC": 0.8986380303823991,
+      "eval_PRM F1 Neg": 0.8163265306122449,
+      "eval_PRM NPV": 0.7692307692307693,
+      "eval_PRM Precision": 0.9625,
+      "eval_PRM Recall": 0.927710843373494,
+      "eval_PRM Specificty": 0.8695652173913043,
+      "eval_loss": 0.22611860930919647,
+      "eval_runtime": 18.8884,
+      "eval_samples_per_second": 2.329,
+      "eval_steps_per_second": 0.159,
+      "step": 3705
+    },
+    {
+      "epoch": 0.9597617429024635,
+      "grad_norm": 0.091572626652188,
+      "learning_rate": 4.903795436902503e-07,
+      "loss": 0.3076,
+      "step": 3706
+    },
+    {
+      "epoch": 0.9600207180084814,
+      "grad_norm": 0.19895015059693597,
+      "learning_rate": 4.840826739554771e-07,
+      "loss": 0.3993,
+      "step": 3707
+    },
+    {
+      "epoch": 0.9602796931144993,
+      "grad_norm": 0.1344861831039975,
+      "learning_rate": 4.778262976834758e-07,
+      "loss": 0.3785,
+      "step": 3708
+    },
+    {
+      "epoch": 0.9605386682205173,
+      "grad_norm": 0.08754570478911258,
+      "learning_rate": 4.716104199906146e-07,
+      "loss": 0.3109,
+      "step": 3709
+    },
+    {
+      "epoch": 0.9607976433265353,
+      "grad_norm": 0.17677527810532703,
+      "learning_rate": 4.654350459601664e-07,
+      "loss": 0.4461,
+      "step": 3710
+    },
+    {
+      "epoch": 0.9607976433265353,
+      "eval_PRM Accuracy": 0.9150943396226415,
+      "eval_PRM F1": 0.9447852760736196,
+      "eval_PRM F1 AUC": 0.8986380303823991,
+      "eval_PRM F1 Neg": 0.8163265306122449,
+      "eval_PRM NPV": 0.7692307692307693,
+      "eval_PRM Precision": 0.9625,
+      "eval_PRM Recall": 0.927710843373494,
+      "eval_PRM Specificty": 0.8695652173913043,
+      "eval_loss": 0.22541947662830353,
+      "eval_runtime": 19.1158,
+      "eval_samples_per_second": 2.302,
+      "eval_steps_per_second": 0.157,
+      "step": 3710
+    },
+    {
+      "epoch": 0.9610566184325532,
+      "grad_norm": 0.08777489275323948,
+      "learning_rate": 4.5930018064227476e-07,
+      "loss": 0.3244,
+      "step": 3711
+    },
+    {
+      "epoch": 0.9613155935385711,
+      "grad_norm": 0.09385468219953985,
+      "learning_rate": 4.5320582905394316e-07,
+      "loss": 0.3487,
+      "step": 3712
+    },
+    {
+      "epoch": 0.961574568644589,
+      "grad_norm": 0.10310174439143652,
+      "learning_rate": 4.471519961790571e-07,
+      "loss": 0.3882,
+      "step": 3713
+    },
+    {
+      "epoch": 0.9618335437506069,
+      "grad_norm": 0.08453034300402293,
+      "learning_rate": 4.411386869683565e-07,
+      "loss": 0.3255,
+      "step": 3714
+    },
+    {
+      "epoch": 0.962092518856625,
+      "grad_norm": 0.13889385831675685,
+      "learning_rate": 4.3516590633945197e-07,
+      "loss": 0.2833,
+      "step": 3715
+    },
+    {
+      "epoch": 0.962092518856625,
+      "eval_PRM Accuracy": 0.9150943396226415,
+      "eval_PRM F1": 0.9447852760736196,
+      "eval_PRM F1 AUC": 0.8986380303823991,
+      "eval_PRM F1 Neg": 0.8163265306122449,
+      "eval_PRM NPV": 0.7692307692307693,
+      "eval_PRM Precision": 0.9625,
+      "eval_PRM Recall": 0.927710843373494,
+      "eval_PRM Specificty": 0.8695652173913043,
+      "eval_loss": 0.22594104707241058,
+      "eval_runtime": 19.1517,
+      "eval_samples_per_second": 2.297,
+      "eval_steps_per_second": 0.157,
+      "step": 3715
+    },
+    {
+      "epoch": 0.9623514939626429,
+      "grad_norm": 0.08188408570432006,
+      "learning_rate": 4.292336591768031e-07,
+      "loss": 0.2365,
+      "step": 3716
+    },
+    {
+      "epoch": 0.9626104690686608,
+      "grad_norm": 0.08997165545640902,
+      "learning_rate": 4.233419503317182e-07,
+      "loss": 0.3502,
+      "step": 3717
+    },
+    {
+      "epoch": 0.9628694441746787,
+      "grad_norm": 0.13124187400704493,
+      "learning_rate": 4.174907846223597e-07,
+      "loss": 0.3596,
+      "step": 3718
+    },
+    {
+      "epoch": 0.9631284192806966,
+      "grad_norm": 0.07535739870993395,
+      "learning_rate": 4.116801668337333e-07,
+      "loss": 0.2887,
+      "step": 3719
+    },
+    {
+      "epoch": 0.9633873943867146,
+      "grad_norm": 0.08605352363716899,
+      "learning_rate": 4.059101017176936e-07,
+      "loss": 0.304,
+      "step": 3720
+    },
+    {
+      "epoch": 0.9633873943867146,
+      "eval_PRM Accuracy": 0.9056603773584906,
+      "eval_PRM F1": 0.9390243902439024,
+      "eval_PRM F1 AUC": 0.8768988999476165,
+      "eval_PRM F1 Neg": 0.7916666666666666,
+      "eval_PRM NPV": 0.76,
+      "eval_PRM Precision": 0.9506172839506173,
+      "eval_PRM Recall": 0.927710843373494,
+      "eval_PRM Specificty": 0.8260869565217391,
+      "eval_loss": 0.22537508606910706,
+      "eval_runtime": 18.9279,
+      "eval_samples_per_second": 2.325,
+      "eval_steps_per_second": 0.158,
+      "step": 3720
+    },
+    {
+      "epoch": 0.9636463694927325,
+      "grad_norm": 0.10134088846619156,
+      "learning_rate": 4.0018059399291596e-07,
+      "loss": 0.2975,
+      "step": 3721
+    },
+    {
+      "epoch": 0.9639053445987504,
+      "grad_norm": 0.09184612869876348,
+      "learning_rate": 3.944916483449135e-07,
+      "loss": 0.3218,
+      "step": 3722
+    },
+    {
+      "epoch": 0.9641643197047683,
+      "grad_norm": 0.09232173848597357,
+      "learning_rate": 3.888432694260424e-07,
+      "loss": 0.3066,
+      "step": 3723
+    },
+    {
+      "epoch": 0.9644232948107863,
+      "grad_norm": 0.13306580670014978,
+      "learning_rate": 3.8323546185546345e-07,
+      "loss": 0.2859,
+      "step": 3724
+    },
+    {
+      "epoch": 0.9646822699168043,
+      "grad_norm": 0.14891437037558003,
+      "learning_rate": 3.776682302191692e-07,
+      "loss": 0.3604,
+      "step": 3725
+    },
+    {
+      "epoch": 0.9646822699168043,
+      "eval_PRM Accuracy": 0.9056603773584906,
+      "eval_PRM F1": 0.9390243902439024,
+      "eval_PRM F1 AUC": 0.8768988999476165,
+      "eval_PRM F1 Neg": 0.7916666666666666,
+      "eval_PRM NPV": 0.76,
+      "eval_PRM Precision": 0.9506172839506173,
+      "eval_PRM Recall": 0.927710843373494,
+      "eval_PRM Specificty": 0.8260869565217391,
+      "eval_loss": 0.22457608580589294,
+      "eval_runtime": 18.2544,
+      "eval_samples_per_second": 2.41,
+      "eval_steps_per_second": 0.164,
+      "step": 3725
+    },
+    {
+      "epoch": 0.9649412450228222,
+      "grad_norm": 0.1088084302480197,
+      "learning_rate": 3.7214157906996784e-07,
+      "loss": 0.377,
+      "step": 3726
+    },
+    {
+      "epoch": 0.9652002201288401,
+      "grad_norm": 0.11397560739356946,
+      "learning_rate": 3.666555129274829e-07,
+      "loss": 0.3601,
+      "step": 3727
+    },
+    {
+      "epoch": 0.965459195234858,
+      "grad_norm": 0.12121259621416776,
+      "learning_rate": 3.6121003627814234e-07,
+      "loss": 0.2948,
+      "step": 3728
+    },
+    {
+      "epoch": 0.965718170340876,
+      "grad_norm": 0.07434971532229942,
+      "learning_rate": 3.558051535751894e-07,
+      "loss": 0.27,
+      "step": 3729
+    },
+    {
+      "epoch": 0.965977145446894,
+      "grad_norm": 0.14354318860113313,
+      "learning_rate": 3.5044086923866623e-07,
+      "loss": 0.4398,
+      "step": 3730
+    },
+    {
+      "epoch": 0.965977145446894,
+      "eval_PRM Accuracy": 0.9056603773584906,
+      "eval_PRM F1": 0.9390243902439024,
+      "eval_PRM F1 AUC": 0.8768988999476165,
+      "eval_PRM F1 Neg": 0.7916666666666666,
+      "eval_PRM NPV": 0.76,
+      "eval_PRM Precision": 0.9506172839506173,
+      "eval_PRM Recall": 0.927710843373494,
+      "eval_PRM Specificty": 0.8260869565217391,
+      "eval_loss": 0.2257080078125,
+      "eval_runtime": 17.9687,
+      "eval_samples_per_second": 2.449,
+      "eval_steps_per_second": 0.167,
+      "step": 3730
+    },
+    {
+      "epoch": 0.9662361205529119,
+      "grad_norm": 0.1111545326136044,
+      "learning_rate": 3.451171876554138e-07,
+      "loss": 0.3892,
+      "step": 3731
+    },
+    {
+      "epoch": 0.9664950956589298,
+      "grad_norm": 0.08024652330208376,
+      "learning_rate": 3.398341131790661e-07,
+      "loss": 0.2947,
+      "step": 3732
+    },
+    {
+      "epoch": 0.9667540707649477,
+      "grad_norm": 0.10198484619783761,
+      "learning_rate": 3.345916501300561e-07,
+      "loss": 0.3068,
+      "step": 3733
+    },
+    {
+      "epoch": 0.9670130458709657,
+      "grad_norm": 0.09641375349434936,
+      "learning_rate": 3.293898027955933e-07,
+      "loss": 0.3608,
+      "step": 3734
+    },
+    {
+      "epoch": 0.9672720209769836,
+      "grad_norm": 0.12637581148863575,
+      "learning_rate": 3.242285754296859e-07,
+      "loss": 0.3047,
+      "step": 3735
+    },
+    {
+      "epoch": 0.9672720209769836,
+      "eval_PRM Accuracy": 0.8962264150943396,
+      "eval_PRM F1": 0.9333333333333333,
+      "eval_PRM F1 AUC": 0.855159769512834,
+      "eval_PRM F1 Neg": 0.7659574468085106,
+      "eval_PRM NPV": 0.75,
+      "eval_PRM Precision": 0.9390243902439024,
+      "eval_PRM Recall": 0.927710843373494,
+      "eval_PRM Specificty": 0.782608695652174,
+      "eval_loss": 0.22634054720401764,
+      "eval_runtime": 19.1493,
+      "eval_samples_per_second": 2.298,
+      "eval_steps_per_second": 0.157,
+      "step": 3735
+    },
+    {
+      "epoch": 0.9675309960830015,
+      "grad_norm": 0.13988293514322395,
+      "learning_rate": 3.191079722531132e-07,
+      "loss": 0.3958,
+      "step": 3736
+    },
+    {
+      "epoch": 0.9677899711890194,
+      "grad_norm": 0.12120773706556932,
+      "learning_rate": 3.140279974534366e-07,
+      "loss": 0.3447,
+      "step": 3737
+    },
+    {
+      "epoch": 0.9680489462950373,
+      "grad_norm": 0.08999461583455147,
+      "learning_rate": 3.089886551849996e-07,
+      "loss": 0.3196,
+      "step": 3738
+    },
+    {
+      "epoch": 0.9683079214010554,
+      "grad_norm": 0.11581699510936184,
+      "learning_rate": 3.0398994956890024e-07,
+      "loss": 0.2924,
+      "step": 3739
+    },
+    {
+      "epoch": 0.9685668965070733,
+      "grad_norm": 0.1049589132957951,
+      "learning_rate": 2.990318846930185e-07,
+      "loss": 0.3719,
+      "step": 3740
+    },
+    {
+      "epoch": 0.9685668965070733,
+      "eval_PRM Accuracy": 0.9150943396226415,
+      "eval_PRM F1": 0.9447852760736196,
+      "eval_PRM F1 AUC": 0.8986380303823991,
+      "eval_PRM F1 Neg": 0.8163265306122449,
+      "eval_PRM NPV": 0.7692307692307693,
+      "eval_PRM Precision": 0.9625,
+      "eval_PRM Recall": 0.927710843373494,
+      "eval_PRM Specificty": 0.8695652173913043,
+      "eval_loss": 0.22520862519741058,
+      "eval_runtime": 18.9975,
+      "eval_samples_per_second": 2.316,
+      "eval_steps_per_second": 0.158,
+      "step": 3740
+    },
+    {
+      "epoch": 0.9688258716130912,
+      "grad_norm": 0.07947383223231992,
+      "learning_rate": 2.9411446461199975e-07,
+      "loss": 0.3405,
+      "step": 3741
+    },
+    {
+      "epoch": 0.9690848467191091,
+      "grad_norm": 0.11616810972548364,
+      "learning_rate": 2.8923769334723296e-07,
+      "loss": 0.3334,
+      "step": 3742
+    },
+    {
+      "epoch": 0.969343821825127,
+      "grad_norm": 0.08615752486047827,
+      "learning_rate": 2.844015748868889e-07,
+      "loss": 0.2878,
+      "step": 3743
+    },
+    {
+      "epoch": 0.969602796931145,
+      "grad_norm": 0.09785638284378743,
+      "learning_rate": 2.796061131858707e-07,
+      "loss": 0.2852,
+      "step": 3744
+    },
+    {
+      "epoch": 0.969861772037163,
+      "grad_norm": 0.09664024305536802,
+      "learning_rate": 2.748513121658525e-07,
+      "loss": 0.3365,
+      "step": 3745
+    },
+    {
+      "epoch": 0.969861772037163,
+      "eval_PRM Accuracy": 0.9056603773584906,
+      "eval_PRM F1": 0.9390243902439024,
+      "eval_PRM F1 AUC": 0.8768988999476165,
+      "eval_PRM F1 Neg": 0.7916666666666666,
+      "eval_PRM NPV": 0.76,
+      "eval_PRM Precision": 0.9506172839506173,
+      "eval_PRM Recall": 0.927710843373494,
+      "eval_PRM Specificty": 0.8260869565217391,
+      "eval_loss": 0.2257745862007141,
+      "eval_runtime": 18.4654,
+      "eval_samples_per_second": 2.383,
+      "eval_steps_per_second": 0.162,
+      "step": 3745
+    },
+    {
+      "epoch": 0.9701207471431809,
+      "grad_norm": 0.10030792424094177,
+      "learning_rate": 2.70137175715246e-07,
+      "loss": 0.3996,
+      "step": 3746
+    },
+    {
+      "epoch": 0.9703797222491988,
+      "grad_norm": 0.08300469401875417,
+      "learning_rate": 2.6546370768920635e-07,
+      "loss": 0.3537,
+      "step": 3747
+    },
+    {
+      "epoch": 0.9706386973552167,
+      "grad_norm": 0.11269079537457256,
+      "learning_rate": 2.608309119096319e-07,
+      "loss": 0.3561,
+      "step": 3748
+    },
+    {
+      "epoch": 0.9708976724612347,
+      "grad_norm": 0.08705343464518406,
+      "learning_rate": 2.5623879216516987e-07,
+      "loss": 0.3117,
+      "step": 3749
+    },
+    {
+      "epoch": 0.9711566475672526,
+      "grad_norm": 0.1752748972716979,
+      "learning_rate": 2.516873522111884e-07,
+      "loss": 0.3964,
+      "step": 3750
+    },
+    {
+      "epoch": 0.9711566475672526,
+      "eval_PRM Accuracy": 0.9056603773584906,
+      "eval_PRM F1": 0.9390243902439024,
+      "eval_PRM F1 AUC": 0.8768988999476165,
+      "eval_PRM F1 Neg": 0.7916666666666666,
+      "eval_PRM NPV": 0.76,
+      "eval_PRM Precision": 0.9506172839506173,
+      "eval_PRM Recall": 0.927710843373494,
+      "eval_PRM Specificty": 0.8260869565217391,
+      "eval_loss": 0.22465376555919647,
+      "eval_runtime": 18.5536,
+      "eval_samples_per_second": 2.372,
+      "eval_steps_per_second": 0.162,
+      "step": 3750
+    },
+    {
+      "epoch": 0.9714156226732705,
+      "grad_norm": 0.08202759773367019,
+      "learning_rate": 2.471765957697991e-07,
+      "loss": 0.2629,
+      "step": 3751
+    },
+    {
+      "epoch": 0.9716745977792884,
+      "grad_norm": 0.0935623351028988,
+      "learning_rate": 2.427065265298345e-07,
+      "loss": 0.3771,
+      "step": 3752
+    },
+    {
+      "epoch": 0.9719335728853064,
+      "grad_norm": 0.08410284324239865,
+      "learning_rate": 2.3827714814686486e-07,
+      "loss": 0.3435,
+      "step": 3753
+    },
+    {
+      "epoch": 0.9721925479913244,
+      "grad_norm": 0.08414137557471188,
+      "learning_rate": 2.3388846424317045e-07,
+      "loss": 0.3094,
+      "step": 3754
+    },
+    {
+      "epoch": 0.9724515230973423,
+      "grad_norm": 0.08947179227189277,
+      "learning_rate": 2.2954047840776372e-07,
+      "loss": 0.2816,
+      "step": 3755
+    },
+    {
+      "epoch": 0.9724515230973423,
+      "eval_PRM Accuracy": 0.9150943396226415,
+      "eval_PRM F1": 0.9447852760736196,
+      "eval_PRM F1 AUC": 0.8986380303823991,
+      "eval_PRM F1 Neg": 0.8163265306122449,
+      "eval_PRM NPV": 0.7692307692307693,
+      "eval_PRM Precision": 0.9625,
+      "eval_PRM Recall": 0.927710843373494,
+      "eval_PRM Specificty": 0.8695652173913043,
+      "eval_loss": 0.22536398470401764,
+      "eval_runtime": 18.9388,
+      "eval_samples_per_second": 2.323,
+      "eval_steps_per_second": 0.158,
+      "step": 3755
+    },
+    {
+      "epoch": 0.9727104982033602,
+      "grad_norm": 0.11663532757359332,
+      "learning_rate": 2.2523319419636702e-07,
+      "loss": 0.3617,
+      "step": 3756
+    },
+    {
+      "epoch": 0.9729694733093781,
+      "grad_norm": 0.15546807995667503,
+      "learning_rate": 2.2096661513141825e-07,
+      "loss": 0.4197,
+      "step": 3757
+    },
+    {
+      "epoch": 0.973228448415396,
+      "grad_norm": 0.08325459304144577,
+      "learning_rate": 2.1674074470207638e-07,
+      "loss": 0.3124,
+      "step": 3758
+    },
+    {
+      "epoch": 0.973487423521414,
+      "grad_norm": 0.06626144212671896,
+      "learning_rate": 2.1255558636419925e-07,
+      "loss": 0.2698,
+      "step": 3759
+    },
+    {
+      "epoch": 0.973746398627432,
+      "grad_norm": 0.11434985102261404,
+      "learning_rate": 2.0841114354034907e-07,
+      "loss": 0.3475,
+      "step": 3760
+    },
+    {
+      "epoch": 0.973746398627432,
+      "eval_PRM Accuracy": 0.9150943396226415,
+      "eval_PRM F1": 0.9447852760736196,
+      "eval_PRM F1 AUC": 0.8986380303823991,
+      "eval_PRM F1 Neg": 0.8163265306122449,
+      "eval_PRM NPV": 0.7692307692307693,
+      "eval_PRM Precision": 0.9625,
+      "eval_PRM Recall": 0.927710843373494,
+      "eval_PRM Specificty": 0.8695652173913043,
+      "eval_loss": 0.22423206269741058,
+      "eval_runtime": 18.2102,
+      "eval_samples_per_second": 2.416,
+      "eval_steps_per_second": 0.165,
+      "step": 3760
+    },
+    {
+      "epoch": 0.9740053737334499,
+      "grad_norm": 0.10817927422879262,
+      "learning_rate": 2.043074196198036e-07,
+      "loss": 0.3463,
+      "step": 3761
+    },
+    {
+      "epoch": 0.9742643488394678,
+      "grad_norm": 0.06954337412308464,
+      "learning_rate": 2.0024441795853388e-07,
+      "loss": 0.2277,
+      "step": 3762
+    },
+    {
+      "epoch": 0.9745233239454857,
+      "grad_norm": 0.10588347433145766,
+      "learning_rate": 1.9622214187919874e-07,
+      "loss": 0.3998,
+      "step": 3763
+    },
+    {
+      "epoch": 0.9747822990515037,
+      "grad_norm": 0.07759486145302771,
+      "learning_rate": 1.9224059467117805e-07,
+      "loss": 0.33,
+      "step": 3764
+    },
+    {
+      "epoch": 0.9750412741575216,
+      "grad_norm": 0.12492742276095857,
+      "learning_rate": 1.8829977959051726e-07,
+      "loss": 0.3374,
+      "step": 3765
+    },
+    {
+      "epoch": 0.9750412741575216,
+      "eval_PRM Accuracy": 0.9150943396226415,
+      "eval_PRM F1": 0.9447852760736196,
+      "eval_PRM F1 AUC": 0.8986380303823991,
+      "eval_PRM F1 Neg": 0.8163265306122449,
+      "eval_PRM NPV": 0.7692307692307693,
+      "eval_PRM Precision": 0.9625,
+      "eval_PRM Recall": 0.927710843373494,
+      "eval_PRM Specificty": 0.8695652173913043,
+      "eval_loss": 0.22616299986839294,
+      "eval_runtime": 18.3898,
+      "eval_samples_per_second": 2.393,
+      "eval_steps_per_second": 0.163,
+      "step": 3765
+    },
+    {
+      "epoch": 0.9753002492635395,
+      "grad_norm": 0.08100303180592666,
+      "learning_rate": 1.843996998599662e-07,
+      "loss": 0.3223,
+      "step": 3766
+    },
+    {
+      "epoch": 0.9755592243695574,
+      "grad_norm": 0.08643421261841984,
+      "learning_rate": 1.805403586689569e-07,
+      "loss": 0.2428,
+      "step": 3767
+    },
+    {
+      "epoch": 0.9758181994755755,
+      "grad_norm": 0.08649173350695663,
+      "learning_rate": 1.7672175917360923e-07,
+      "loss": 0.279,
+      "step": 3768
+    },
+    {
+      "epoch": 0.9760771745815934,
+      "grad_norm": 0.15239655024412016,
+      "learning_rate": 1.729439044967196e-07,
+      "loss": 0.4353,
+      "step": 3769
+    },
+    {
+      "epoch": 0.9763361496876113,
+      "grad_norm": 0.08484555537259249,
+      "learning_rate": 1.6920679772777782e-07,
+      "loss": 0.3113,
+      "step": 3770
+    },
+    {
+      "epoch": 0.9763361496876113,
+      "eval_PRM Accuracy": 0.9056603773584906,
+      "eval_PRM F1": 0.9390243902439024,
+      "eval_PRM F1 AUC": 0.8768988999476165,
+      "eval_PRM F1 Neg": 0.7916666666666666,
+      "eval_PRM NPV": 0.76,
+      "eval_PRM Precision": 0.9506172839506173,
+      "eval_PRM Recall": 0.927710843373494,
+      "eval_PRM Specificty": 0.8260869565217391,
+      "eval_loss": 0.22548605501651764,
+      "eval_runtime": 18.5642,
+      "eval_samples_per_second": 2.37,
+      "eval_steps_per_second": 0.162,
+      "step": 3770
+    },
+    {
+      "epoch": 0.9765951247936292,
+      "grad_norm": 0.11399615303462896,
+      "learning_rate": 1.655104419229281e-07,
+      "loss": 0.3357,
+      "step": 3771
+    },
+    {
+      "epoch": 0.9768540998996471,
+      "grad_norm": 0.10437236819080888,
+      "learning_rate": 1.618548401050135e-07,
+      "loss": 0.3358,
+      "step": 3772
+    },
+    {
+      "epoch": 0.9771130750056651,
+      "grad_norm": 0.11292965615773823,
+      "learning_rate": 1.5823999526352606e-07,
+      "loss": 0.3971,
+      "step": 3773
+    },
+    {
+      "epoch": 0.977372050111683,
+      "grad_norm": 0.10400501733451095,
+      "learning_rate": 1.5466591035463994e-07,
+      "loss": 0.326,
+      "step": 3774
+    },
+    {
+      "epoch": 0.977631025217701,
+      "grad_norm": 0.10580290811845071,
+      "learning_rate": 1.5113258830120046e-07,
+      "loss": 0.2461,
+      "step": 3775
+    },
+    {
+      "epoch": 0.977631025217701,
+      "eval_PRM Accuracy": 0.8962264150943396,
+      "eval_PRM F1": 0.9333333333333333,
+      "eval_PRM F1 AUC": 0.855159769512834,
+      "eval_PRM F1 Neg": 0.7659574468085106,
+      "eval_PRM NPV": 0.75,
+      "eval_PRM Precision": 0.9390243902439024,
+      "eval_PRM Recall": 0.927710843373494,
+      "eval_PRM Specificty": 0.782608695652174,
+      "eval_loss": 0.2265070080757141,
+      "eval_runtime": 18.6702,
+      "eval_samples_per_second": 2.357,
+      "eval_steps_per_second": 0.161,
+      "step": 3775
+    },
+    {
+      "epoch": 0.9778900003237189,
+      "grad_norm": 0.10461592611156097,
+      "learning_rate": 1.4764003199270736e-07,
+      "loss": 0.2626,
+      "step": 3776
+    },
+    {
+      "epoch": 0.9781489754297368,
+      "grad_norm": 0.15444422688395668,
+      "learning_rate": 1.441882442853315e-07,
+      "loss": 0.3407,
+      "step": 3777
+    },
+    {
+      "epoch": 0.9784079505357548,
+      "grad_norm": 0.08894471295622203,
+      "learning_rate": 1.4077722800188708e-07,
+      "loss": 0.3168,
+      "step": 3778
+    },
+    {
+      "epoch": 0.9786669256417727,
+      "grad_norm": 0.06933662532151363,
+      "learning_rate": 1.3740698593187606e-07,
+      "loss": 0.2709,
+      "step": 3779
+    },
+    {
+      "epoch": 0.9789259007477906,
+      "grad_norm": 0.0928132032413491,
+      "learning_rate": 1.3407752083142154e-07,
+      "loss": 0.2784,
+      "step": 3780
+    },
+    {
+      "epoch": 0.9789259007477906,
+      "eval_PRM Accuracy": 0.8962264150943396,
+      "eval_PRM F1": 0.9333333333333333,
+      "eval_PRM F1 AUC": 0.855159769512834,
+      "eval_PRM F1 Neg": 0.7659574468085106,
+      "eval_PRM NPV": 0.75,
+      "eval_PRM Precision": 0.9390243902439024,
+      "eval_PRM Recall": 0.927710843373494,
+      "eval_PRM Specificty": 0.782608695652174,
+      "eval_loss": 0.22660689055919647,
+      "eval_runtime": 19.0584,
+      "eval_samples_per_second": 2.309,
+      "eval_steps_per_second": 0.157,
+      "step": 3780
+    },
+    {
+      "epoch": 0.9791848758538085,
+      "grad_norm": 0.08413317716089637,
+      "learning_rate": 1.3078883542332886e-07,
+      "loss": 0.2782,
+      "step": 3781
+    },
+    {
+      "epoch": 0.9794438509598264,
+      "grad_norm": 0.07551667594913106,
+      "learning_rate": 1.2754093239703003e-07,
+      "loss": 0.2463,
+      "step": 3782
+    },
+    {
+      "epoch": 0.9797028260658445,
+      "grad_norm": 0.11619020126566788,
+      "learning_rate": 1.2433381440862812e-07,
+      "loss": 0.3415,
+      "step": 3783
+    },
+    {
+      "epoch": 0.9799618011718624,
+      "grad_norm": 0.10508037835859242,
+      "learning_rate": 1.21167484080853e-07,
+      "loss": 0.3471,
+      "step": 3784
+    },
+    {
+      "epoch": 0.9802207762778803,
+      "grad_norm": 0.09978517404005727,
+      "learning_rate": 1.1804194400309442e-07,
+      "loss": 0.2794,
+      "step": 3785
+    },
+    {
+      "epoch": 0.9802207762778803,
+      "eval_PRM Accuracy": 0.9056603773584906,
+      "eval_PRM F1": 0.9390243902439024,
+      "eval_PRM F1 AUC": 0.8768988999476165,
+      "eval_PRM F1 Neg": 0.7916666666666666,
+      "eval_PRM NPV": 0.76,
+      "eval_PRM Precision": 0.9506172839506173,
+      "eval_PRM Recall": 0.927710843373494,
+      "eval_PRM Specificty": 0.8260869565217391,
+      "eval_loss": 0.22605203092098236,
+      "eval_runtime": 18.2782,
+      "eval_samples_per_second": 2.407,
+      "eval_steps_per_second": 0.164,
+      "step": 3785
+    },
+    {
+      "epoch": 0.9804797513838982,
+      "grad_norm": 0.0975387199176699,
+      "learning_rate": 1.149571967313745e-07,
+      "loss": 0.403,
+      "step": 3786
+    },
+    {
+      "epoch": 0.9807387264899161,
+      "grad_norm": 0.11203208275073363,
+      "learning_rate": 1.1191324478835863e-07,
+      "loss": 0.3372,
+      "step": 3787
+    },
+    {
+      "epoch": 0.9809977015959341,
+      "grad_norm": 0.12217412790275088,
+      "learning_rate": 1.0891009066334445e-07,
+      "loss": 0.4423,
+      "step": 3788
+    },
+    {
+      "epoch": 0.981256676701952,
+      "grad_norm": 0.1049115645186349,
+      "learning_rate": 1.059477368122841e-07,
+      "loss": 0.3131,
+      "step": 3789
+    },
+    {
+      "epoch": 0.98151565180797,
+      "grad_norm": 0.10431075060939143,
+      "learning_rate": 1.0302618565774525e-07,
+      "loss": 0.415,
+      "step": 3790
+    },
+    {
+      "epoch": 0.98151565180797,
+      "eval_PRM Accuracy": 0.9056603773584906,
+      "eval_PRM F1": 0.9390243902439024,
+      "eval_PRM F1 AUC": 0.8768988999476165,
+      "eval_PRM F1 Neg": 0.7916666666666666,
+      "eval_PRM NPV": 0.76,
+      "eval_PRM Precision": 0.9506172839506173,
+      "eval_PRM Recall": 0.927710843373494,
+      "eval_PRM Specificty": 0.8260869565217391,
+      "eval_loss": 0.22531960904598236,
+      "eval_runtime": 18.7954,
+      "eval_samples_per_second": 2.341,
+      "eval_steps_per_second": 0.16,
+      "step": 3790
+    },
+    {
+      "epoch": 0.9817746269139879,
+      "grad_norm": 0.10064170554293002,
+      "learning_rate": 1.0014543958893341e-07,
+      "loss": 0.3616,
+      "step": 3791
+    },
+    {
+      "epoch": 0.9820336020200058,
+      "grad_norm": 0.11774097917326624,
+      "learning_rate": 9.730550096168633e-08,
+      "loss": 0.3477,
+      "step": 3792
+    },
+    {
+      "epoch": 0.9822925771260238,
+      "grad_norm": 0.10723165468451555,
+      "learning_rate": 9.450637209846292e-08,
+      "loss": 0.3754,
+      "step": 3793
+    },
+    {
+      "epoch": 0.9825515522320417,
+      "grad_norm": 0.09861110731375088,
+      "learning_rate": 9.174805528836539e-08,
+      "loss": 0.3731,
+      "step": 3794
+    },
+    {
+      "epoch": 0.9828105273380596,
+      "grad_norm": 0.09646182546344875,
+      "learning_rate": 8.903055278709494e-08,
+      "loss": 0.381,
+      "step": 3795
+    },
+    {
+      "epoch": 0.9828105273380596,
+      "eval_PRM Accuracy": 0.9056603773584906,
+      "eval_PRM F1": 0.9390243902439024,
+      "eval_PRM F1 AUC": 0.8768988999476165,
+      "eval_PRM F1 Neg": 0.7916666666666666,
+      "eval_PRM NPV": 0.76,
+      "eval_PRM Precision": 0.9506172839506173,
+      "eval_PRM Recall": 0.927710843373494,
+      "eval_PRM Specificty": 0.8260869565217391,
+      "eval_loss": 0.22642932832241058,
+      "eval_runtime": 18.7687,
+      "eval_samples_per_second": 2.344,
+      "eval_steps_per_second": 0.16,
+      "step": 3795
+    },
+    {
+      "epoch": 0.9830695024440775,
+      "grad_norm": 0.08683395305868136,
+      "learning_rate": 8.635386681699608e-08,
+      "loss": 0.3287,
+      "step": 3796
+    },
+    {
+      "epoch": 0.9833284775500954,
+      "grad_norm": 0.09488655997400364,
+      "learning_rate": 8.371799956702897e-08,
+      "loss": 0.2576,
+      "step": 3797
+    },
+    {
+      "epoch": 0.9835874526561135,
+      "grad_norm": 0.12775160820785042,
+      "learning_rate": 8.11229531927693e-08,
+      "loss": 0.3271,
+      "step": 3798
+    },
+    {
+      "epoch": 0.9838464277621314,
+      "grad_norm": 0.11124627635652057,
+      "learning_rate": 7.85687298164084e-08,
+      "loss": 0.3856,
+      "step": 3799
+    },
+    {
+      "epoch": 0.9841054028681493,
+      "grad_norm": 0.09316534587439892,
+      "learning_rate": 7.605533152676426e-08,
+      "loss": 0.2724,
+      "step": 3800
+    },
+    {
+      "epoch": 0.9841054028681493,
+      "eval_PRM Accuracy": 0.9056603773584906,
+      "eval_PRM F1": 0.9390243902439024,
+      "eval_PRM F1 AUC": 0.8768988999476165,
+      "eval_PRM F1 Neg": 0.7916666666666666,
+      "eval_PRM NPV": 0.76,
+      "eval_PRM Precision": 0.9506172839506173,
+      "eval_PRM Recall": 0.927710843373494,
+      "eval_PRM Specificty": 0.8260869565217391,
+      "eval_loss": 0.22547496855258942,
+      "eval_runtime": 19.0655,
+      "eval_samples_per_second": 2.308,
+      "eval_steps_per_second": 0.157,
+      "step": 3800
+    },
+    {
+      "epoch": 0.9843643779741672,
+      "grad_norm": 0.13523526267140148,
+      "learning_rate": 7.358276037924828e-08,
+      "loss": 0.2732,
+      "step": 3801
+    },
+    {
+      "epoch": 0.9846233530801851,
+      "grad_norm": 0.12177779875816264,
+      "learning_rate": 7.115101839590966e-08,
+      "loss": 0.3463,
+      "step": 3802
+    },
+    {
+      "epoch": 0.9848823281862031,
+      "grad_norm": 0.11522258036268547,
+      "learning_rate": 6.876010756538542e-08,
+      "loss": 0.2955,
+      "step": 3803
+    },
+    {
+      "epoch": 0.985141303292221,
+      "grad_norm": 0.09672699144480919,
+      "learning_rate": 6.641002984293931e-08,
+      "loss": 0.3444,
+      "step": 3804
+    },
+    {
+      "epoch": 0.985400278398239,
+      "grad_norm": 0.089247099420499,
+      "learning_rate": 6.41007871504229e-08,
+      "loss": 0.3075,
+      "step": 3805
+    },
+    {
+      "epoch": 0.985400278398239,
+      "eval_PRM Accuracy": 0.9056603773584906,
+      "eval_PRM F1": 0.9390243902439024,
+      "eval_PRM F1 AUC": 0.8768988999476165,
+      "eval_PRM F1 Neg": 0.7916666666666666,
+      "eval_PRM NPV": 0.76,
+      "eval_PRM Precision": 0.9506172839506173,
+      "eval_PRM Recall": 0.927710843373494,
+      "eval_PRM Specificty": 0.8260869565217391,
+      "eval_loss": 0.2249090075492859,
+      "eval_runtime": 19.0312,
+      "eval_samples_per_second": 2.312,
+      "eval_steps_per_second": 0.158,
+      "step": 3805
+    },
+    {
+      "epoch": 0.9856592535042569,
+      "grad_norm": 0.10398095894978453,
+      "learning_rate": 6.183238137631442e-08,
+      "loss": 0.3253,
+      "step": 3806
+    },
+    {
+      "epoch": 0.9859182286102748,
+      "grad_norm": 0.08427275077364413,
+      "learning_rate": 5.960481437568555e-08,
+      "loss": 0.3389,
+      "step": 3807
+    },
+    {
+      "epoch": 0.9861772037162928,
+      "grad_norm": 0.09249028344051778,
+      "learning_rate": 5.7418087970201315e-08,
+      "loss": 0.358,
+      "step": 3808
+    },
+    {
+      "epoch": 0.9864361788223107,
+      "grad_norm": 0.10022977649895663,
+      "learning_rate": 5.527220394814792e-08,
+      "loss": 0.3391,
+      "step": 3809
+    },
+    {
+      "epoch": 0.9866951539283286,
+      "grad_norm": 0.13118098432653125,
+      "learning_rate": 5.3167164064393815e-08,
+      "loss": 0.3283,
+      "step": 3810
+    },
+    {
+      "epoch": 0.9866951539283286,
+      "eval_PRM Accuracy": 0.9056603773584906,
+      "eval_PRM F1": 0.9390243902439024,
+      "eval_PRM F1 AUC": 0.8768988999476165,
+      "eval_PRM F1 Neg": 0.7916666666666666,
+      "eval_PRM NPV": 0.76,
+      "eval_PRM Precision": 0.9506172839506173,
+      "eval_PRM Recall": 0.927710843373494,
+      "eval_PRM Specificty": 0.8260869565217391,
+      "eval_loss": 0.2249090075492859,
+      "eval_runtime": 18.8389,
+      "eval_samples_per_second": 2.336,
+      "eval_steps_per_second": 0.159,
+      "step": 3810
+    },
+    {
+      "epoch": 0.9869541290343465,
+      "grad_norm": 0.08214687156840358,
+      "learning_rate": 5.110297004042308e-08,
+      "loss": 0.309,
+      "step": 3811
+    },
+    {
+      "epoch": 0.9872131041403646,
+      "grad_norm": 0.10798164925665907,
+      "learning_rate": 4.907962356429097e-08,
+      "loss": 0.3147,
+      "step": 3812
+    },
+    {
+      "epoch": 0.9874720792463825,
+      "grad_norm": 0.08803338817712282,
+      "learning_rate": 4.7097126290679415e-08,
+      "loss": 0.2496,
+      "step": 3813
+    },
+    {
+      "epoch": 0.9877310543524004,
+      "grad_norm": 0.08396227255614713,
+      "learning_rate": 4.5155479840836015e-08,
+      "loss": 0.2867,
+      "step": 3814
+    },
+    {
+      "epoch": 0.9879900294584183,
+      "grad_norm": 0.12229694536875016,
+      "learning_rate": 4.325468580261838e-08,
+      "loss": 0.2995,
+      "step": 3815
+    },
+    {
+      "epoch": 0.9879900294584183,
+      "eval_PRM Accuracy": 0.9056603773584906,
+      "eval_PRM F1": 0.9390243902439024,
+      "eval_PRM F1 AUC": 0.8768988999476165,
+      "eval_PRM F1 Neg": 0.7916666666666666,
+      "eval_PRM NPV": 0.76,
+      "eval_PRM Precision": 0.9506172839506173,
+      "eval_PRM Recall": 0.927710843373494,
+      "eval_PRM Specificty": 0.8260869565217391,
+      "eval_loss": 0.22526411712169647,
+      "eval_runtime": 18.5564,
+      "eval_samples_per_second": 2.371,
+      "eval_steps_per_second": 0.162,
+      "step": 3815
+    },
+    {
+      "epoch": 0.9882490045644362,
+      "grad_norm": 0.10642952502545472,
+      "learning_rate": 4.139474573047197e-08,
+      "loss": 0.387,
+      "step": 3816
+    },
+    {
+      "epoch": 0.9885079796704542,
+      "grad_norm": 0.129165600515501,
+      "learning_rate": 3.9575661145430095e-08,
+      "loss": 0.3786,
+      "step": 3817
+    },
+    {
+      "epoch": 0.9887669547764721,
+      "grad_norm": 0.10383180097041594,
+      "learning_rate": 3.7797433535119444e-08,
+      "loss": 0.2504,
+      "step": 3818
+    },
+    {
+      "epoch": 0.98902592988249,
+      "grad_norm": 0.09640530256748744,
+      "learning_rate": 3.6060064353743426e-08,
+      "loss": 0.3496,
+      "step": 3819
+    },
+    {
+      "epoch": 0.989284904988508,
+      "grad_norm": 0.12428439621511016,
+      "learning_rate": 3.436355502210442e-08,
+      "loss": 0.3997,
+      "step": 3820
+    },
+    {
+      "epoch": 0.989284904988508,
+      "eval_PRM Accuracy": 0.9056603773584906,
+      "eval_PRM F1": 0.9390243902439024,
+      "eval_PRM F1 AUC": 0.8768988999476165,
+      "eval_PRM F1 Neg": 0.7916666666666666,
+      "eval_PRM NPV": 0.76,
+      "eval_PRM Precision": 0.9506172839506173,
+      "eval_PRM Recall": 0.927710843373494,
+      "eval_PRM Specificty": 0.8260869565217391,
+      "eval_loss": 0.22484241425991058,
+      "eval_runtime": 18.5766,
+      "eval_samples_per_second": 2.369,
+      "eval_steps_per_second": 0.161,
+      "step": 3820
+    },
+    {
+      "epoch": 0.9895438800945259,
+      "grad_norm": 0.10961842325292329,
+      "learning_rate": 3.270790692758707e-08,
+      "loss": 0.3497,
+      "step": 3821
+    },
+    {
+      "epoch": 0.9898028552005439,
+      "grad_norm": 0.090272688680197,
+      "learning_rate": 3.109312142415832e-08,
+      "loss": 0.3206,
+      "step": 3822
+    },
+    {
+      "epoch": 0.9900618303065618,
+      "grad_norm": 0.11658775328873373,
+      "learning_rate": 2.9519199832361842e-08,
+      "loss": 0.3292,
+      "step": 3823
+    },
+    {
+      "epoch": 0.9903208054125797,
+      "grad_norm": 0.11367625580508493,
+      "learning_rate": 2.79861434393347e-08,
+      "loss": 0.3395,
+      "step": 3824
+    },
+    {
+      "epoch": 0.9905797805185976,
+      "grad_norm": 0.08006504529868755,
+      "learning_rate": 2.649395349879069e-08,
+      "loss": 0.2946,
+      "step": 3825
+    },
+    {
+      "epoch": 0.9905797805185976,
+      "eval_PRM Accuracy": 0.9056603773584906,
+      "eval_PRM F1": 0.9390243902439024,
+      "eval_PRM F1 AUC": 0.8768988999476165,
+      "eval_PRM F1 Neg": 0.7916666666666666,
+      "eval_PRM NPV": 0.76,
+      "eval_PRM Precision": 0.9506172839506173,
+      "eval_PRM Recall": 0.927710843373494,
+      "eval_PRM Specificty": 0.8260869565217391,
+      "eval_loss": 0.2247314453125,
+      "eval_runtime": 18.6677,
+      "eval_samples_per_second": 2.357,
+      "eval_steps_per_second": 0.161,
+      "step": 3825
+    },
+    {
+      "epoch": 0.9908387556246155,
+      "grad_norm": 0.09199773003966358,
+      "learning_rate": 2.5042631231025903e-08,
+      "loss": 0.3302,
+      "step": 3826
+    },
+    {
+      "epoch": 0.9910977307306336,
+      "grad_norm": 0.09633315960954299,
+      "learning_rate": 2.3632177822902057e-08,
+      "loss": 0.353,
+      "step": 3827
+    },
+    {
+      "epoch": 0.9913567058366515,
+      "grad_norm": 0.11828936156768534,
+      "learning_rate": 2.2262594427879813e-08,
+      "loss": 0.4348,
+      "step": 3828
+    },
+    {
+      "epoch": 0.9916156809426694,
+      "grad_norm": 0.16085689058963845,
+      "learning_rate": 2.0933882165979914e-08,
+      "loss": 0.3046,
+      "step": 3829
+    },
+    {
+      "epoch": 0.9918746560486873,
+      "grad_norm": 0.07276257063935491,
+      "learning_rate": 1.9646042123810938e-08,
+      "loss": 0.2672,
+      "step": 3830
+    },
+    {
+      "epoch": 0.9918746560486873,
+      "eval_PRM Accuracy": 0.9056603773584906,
+      "eval_PRM F1": 0.9390243902439024,
+      "eval_PRM F1 AUC": 0.8768988999476165,
+      "eval_PRM F1 Neg": 0.7916666666666666,
+      "eval_PRM NPV": 0.76,
+      "eval_PRM Precision": 0.9506172839506173,
+      "eval_PRM Recall": 0.927710843373494,
+      "eval_PRM Specificty": 0.8260869565217391,
+      "eval_loss": 0.22536398470401764,
+      "eval_runtime": 18.6479,
+      "eval_samples_per_second": 2.36,
+      "eval_steps_per_second": 0.161,
+      "step": 3830
+    },
+    {
+      "epoch": 0.9921336311547052,
+      "grad_norm": 0.09844963440524152,
+      "learning_rate": 1.839907535455265e-08,
+      "loss": 0.3096,
+      "step": 3831
+    },
+    {
+      "epoch": 0.9923926062607232,
+      "grad_norm": 0.11657489534739605,
+      "learning_rate": 1.719298287795046e-08,
+      "loss": 0.4156,
+      "step": 3832
+    },
+    {
+      "epoch": 0.9926515813667411,
+      "grad_norm": 0.12450046342582112,
+      "learning_rate": 1.602776568033204e-08,
+      "loss": 0.457,
+      "step": 3833
+    },
+    {
+      "epoch": 0.992910556472759,
+      "grad_norm": 0.07515160551099585,
+      "learning_rate": 1.490342471460182e-08,
+      "loss": 0.2756,
+      "step": 3834
+    },
+    {
+      "epoch": 0.993169531578777,
+      "grad_norm": 0.10284685981781568,
+      "learning_rate": 1.3819960900229855e-08,
+      "loss": 0.2594,
+      "step": 3835
+    },
+    {
+      "epoch": 0.993169531578777,
+      "eval_PRM Accuracy": 0.9056603773584906,
+      "eval_PRM F1": 0.9390243902439024,
+      "eval_PRM F1 AUC": 0.8768988999476165,
+      "eval_PRM F1 Neg": 0.7916666666666666,
+      "eval_PRM NPV": 0.76,
+      "eval_PRM Precision": 0.9506172839506173,
+      "eval_PRM Recall": 0.927710843373494,
+      "eval_PRM Specificty": 0.8260869565217391,
+      "eval_loss": 0.22530850768089294,
+      "eval_runtime": 18.6644,
+      "eval_samples_per_second": 2.357,
+      "eval_steps_per_second": 0.161,
+      "step": 3835
+    },
+    {
+      "epoch": 0.9934285066847949,
+      "grad_norm": 0.1446373530729935,
+      "learning_rate": 1.2777375123257385e-08,
+      "loss": 0.3416,
+      "step": 3836
+    },
+    {
+      "epoch": 0.9936874817908129,
+      "grad_norm": 0.08809779230160165,
+      "learning_rate": 1.1775668236291281e-08,
+      "loss": 0.3115,
+      "step": 3837
+    },
+    {
+      "epoch": 0.9939464568968308,
+      "grad_norm": 0.10100865439415825,
+      "learning_rate": 1.081484105852626e-08,
+      "loss": 0.3962,
+      "step": 3838
+    },
+    {
+      "epoch": 0.9942054320028487,
+      "grad_norm": 0.08511029134929173,
+      "learning_rate": 9.89489437570601e-09,
+      "loss": 0.2907,
+      "step": 3839
+    },
+    {
+      "epoch": 0.9944644071088666,
+      "grad_norm": 0.212268921837846,
+      "learning_rate": 9.015828940150961e-09,
+      "loss": 0.3498,
+      "step": 3840
+    },
+    {
+      "epoch": 0.9944644071088666,
+      "eval_PRM Accuracy": 0.9056603773584906,
+      "eval_PRM F1": 0.9390243902439024,
+      "eval_PRM F1 AUC": 0.8768988999476165,
+      "eval_PRM F1 Neg": 0.7916666666666666,
+      "eval_PRM NPV": 0.76,
+      "eval_PRM Precision": 0.9506172839506173,
+      "eval_PRM Recall": 0.927710843373494,
+      "eval_PRM Specificty": 0.8260869565217391,
+      "eval_loss": 0.2253972887992859,
+      "eval_runtime": 18.8825,
+      "eval_samples_per_second": 2.33,
+      "eval_steps_per_second": 0.159,
+      "step": 3840
+    },
+    {
+      "epoch": 0.9947233822148845,
+      "grad_norm": 0.09983903729116173,
+      "learning_rate": 8.177645470752727e-09,
+      "loss": 0.373,
+      "step": 3841
+    },
+    {
+      "epoch": 0.9949823573209026,
+      "grad_norm": 0.10268957888803255,
+      "learning_rate": 7.380344652963e-09,
+      "loss": 0.3078,
+      "step": 3842
+    },
+    {
+      "epoch": 0.9952413324269205,
+      "grad_norm": 0.11461091548324738,
+      "learning_rate": 6.623927138804664e-09,
+      "loss": 0.327,
+      "step": 3843
+    },
+    {
+      "epoch": 0.9955003075329384,
+      "grad_norm": 0.12067595976043448,
+      "learning_rate": 5.908393546866231e-09,
+      "loss": 0.3057,
+      "step": 3844
+    },
+    {
+      "epoch": 0.9957592826389563,
+      "grad_norm": 0.12328742569083354,
+      "learning_rate": 5.2337444623074e-09,
+      "loss": 0.3603,
+      "step": 3845
+    },
+    {
+      "epoch": 0.9957592826389563,
+      "eval_PRM Accuracy": 0.8962264150943396,
+      "eval_PRM F1": 0.9333333333333333,
+      "eval_PRM F1 AUC": 0.855159769512834,
+      "eval_PRM F1 Neg": 0.7659574468085106,
+      "eval_PRM NPV": 0.75,
+      "eval_PRM Precision": 0.9390243902439024,
+      "eval_PRM Recall": 0.927710843373494,
+      "eval_PRM Specificty": 0.782608695652174,
+      "eval_loss": 0.225830078125,
+      "eval_runtime": 18.678,
+      "eval_samples_per_second": 2.356,
+      "eval_steps_per_second": 0.161,
+      "step": 3845
+    },
+    {
+      "epoch": 0.9960182577449742,
+      "grad_norm": 0.1194765284283544,
+      "learning_rate": 4.5999804368368485e-09,
+      "loss": 0.4059,
+      "step": 3846
+    },
+    {
+      "epoch": 0.9962772328509922,
+      "grad_norm": 0.08473091220405443,
+      "learning_rate": 4.007101988745543e-09,
+      "loss": 0.2715,
+      "step": 3847
+    },
+    {
+      "epoch": 0.9965362079570101,
+      "grad_norm": 0.13218741741767376,
+      "learning_rate": 3.455109602878981e-09,
+      "loss": 0.3579,
+      "step": 3848
+    },
+    {
+      "epoch": 0.9967951830630281,
+      "grad_norm": 0.10304599344538268,
+      "learning_rate": 2.9440037306538437e-09,
+      "loss": 0.3417,
+      "step": 3849
+    },
+    {
+      "epoch": 0.997054158169046,
+      "grad_norm": 0.10540989435316424,
+      "learning_rate": 2.4737847900357935e-09,
+      "loss": 0.3443,
+      "step": 3850
+    },
+    {
+      "epoch": 0.997054158169046,
+      "eval_PRM Accuracy": 0.8962264150943396,
+      "eval_PRM F1": 0.9333333333333333,
+      "eval_PRM F1 AUC": 0.855159769512834,
+      "eval_PRM F1 Neg": 0.7659574468085106,
+      "eval_PRM NPV": 0.75,
+      "eval_PRM Precision": 0.9390243902439024,
+      "eval_PRM Recall": 0.927710843373494,
+      "eval_PRM Specificty": 0.782608695652174,
+      "eval_loss": 0.22578568756580353,
+      "eval_runtime": 18.349,
+      "eval_samples_per_second": 2.398,
+      "eval_steps_per_second": 0.163,
+      "step": 3850
+    },
+    {
+      "epoch": 0.997313133275064,
+      "grad_norm": 0.1273980080792991,
+      "learning_rate": 2.0444531655727794e-09,
+      "loss": 0.3442,
+      "step": 3851
+    },
+    {
+      "epoch": 0.9975721083810819,
+      "grad_norm": 0.13297212173034,
+      "learning_rate": 1.6560092083672817e-09,
+      "loss": 0.3719,
+      "step": 3852
+    },
+    {
+      "epoch": 0.9978310834870998,
+      "grad_norm": 0.141906827839403,
+      "learning_rate": 1.3084532360763125e-09,
+      "loss": 0.3913,
+      "step": 3853
+    },
+    {
+      "epoch": 0.9980900585931177,
+      "grad_norm": 0.1282560885505363,
+      "learning_rate": 1.0017855329280678e-09,
+      "loss": 0.3299,
+      "step": 3854
+    },
+    {
+      "epoch": 0.9983490336991356,
+      "grad_norm": 0.17131302652610803,
+      "learning_rate": 7.360063497163783e-10,
+      "loss": 0.416,
+      "step": 3855
+    },
+    {
+      "epoch": 0.9983490336991356,
+      "eval_PRM Accuracy": 0.8962264150943396,
+      "eval_PRM F1": 0.9333333333333333,
+      "eval_PRM F1 AUC": 0.855159769512834,
+      "eval_PRM F1 Neg": 0.7659574468085106,
+      "eval_PRM NPV": 0.75,
+      "eval_PRM Precision": 0.9390243902439024,
+      "eval_PRM Recall": 0.927710843373494,
+      "eval_PRM Specificty": 0.782608695652174,
+      "eval_loss": 0.22671785950660706,
+      "eval_runtime": 18.7928,
+      "eval_samples_per_second": 2.341,
+      "eval_steps_per_second": 0.16,
+      "step": 3855
+    },
+    {
+      "epoch": 0.9986080088051537,
+      "grad_norm": 0.11279819848057338,
+      "learning_rate": 5.111159037896052e-10,
+      "loss": 0.3398,
+      "step": 3856
+    },
+    {
+      "epoch": 0.9988669839111716,
+      "grad_norm": 0.13027046619834068,
+      "learning_rate": 3.271143790561926e-10,
+      "loss": 0.4107,
+      "step": 3857
+    },
+    {
+      "epoch": 0.9991259590171895,
+      "grad_norm": 0.08692802535830696,
+      "learning_rate": 1.8400192599576925e-10,
+      "loss": 0.2445,
+      "step": 3858
+    },
+    {
+      "epoch": 0.9993849341232074,
+      "grad_norm": 0.09703795362165403,
+      "learning_rate": 8.177866164249537e-11,
+      "loss": 0.3709,
+      "step": 3859
+    },
+    {
+      "epoch": 0.9996439092292253,
+      "grad_norm": 0.0854284964675507,
+      "learning_rate": 2.044466959061353e-11,
+      "loss": 0.3123,
+      "step": 3860
+    },
+    {
+      "epoch": 0.9996439092292253,
+      "eval_PRM Accuracy": 0.9056603773584906,
+      "eval_PRM F1": 0.9390243902439024,
+      "eval_PRM F1 AUC": 0.8768988999476165,
+      "eval_PRM F1 Neg": 0.7916666666666666,
+      "eval_PRM NPV": 0.76,
+      "eval_PRM Precision": 0.9506172839506173,
+      "eval_PRM Recall": 0.927710843373494,
+      "eval_PRM Specificty": 0.8260869565217391,
+      "eval_loss": 0.2269287109375,
+      "eval_runtime": 19.5839,
+      "eval_samples_per_second": 2.247,
+      "eval_steps_per_second": 0.153,
+      "step": 3860
+    },
+    {
+      "epoch": 0.9999028843352433,
+      "grad_norm": 0.08518419956920283,
+      "learning_rate": 0.0,
+      "loss": 0.2596,
+      "step": 3861
     }
   ],
   "logging_steps": 1,
@@ -35737,12 +39416,12 @@
         "should_evaluate": false,
         "should_log": false,
         "should_save": true,
-        "should_training_stop": false
+        "should_training_stop": true
       },
       "attributes": {}
     }
   },
-  "total_flos": 5891744914341888.0,
+  "total_flos": 6503291547418624.0,
   "train_batch_size": 4,
   "trial_name": null,
   "trial_params": null