diff --git "a/last-checkpoint/trainer_state.json" "b/last-checkpoint/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/last-checkpoint/trainer_state.json"
@@ -0,0 +1,4647 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.9980287243030133,
+  "eval_steps": 5,
+  "global_step": 443,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0,
+      "eval_PRM Accuracy": 0.803921568627451,
+      "eval_PRM F1": 0.8507462686567164,
+      "eval_PRM F1 AUC": 0.8214454416627303,
+      "eval_PRM F1 AUC (fixed)": 0.8419933868682098,
+      "eval_PRM F1 Neg": 0.7142857142857143,
+      "eval_PRM NPV": 0.6097560975609756,
+      "eval_PRM Precision": 0.9344262295081968,
+      "eval_PRM Recall": 0.7808219178082192,
+      "eval_PRM Specificty": 0.8620689655172413,
+      "eval_loss": 0.4027876555919647,
+      "eval_runtime": 14.6583,
+      "eval_samples_per_second": 3.002,
+      "eval_steps_per_second": 0.136,
+      "step": 0
+    },
+    {
+      "epoch": 0.002252886510842016,
+      "grad_norm": 0.2515101123296272,
+      "learning_rate": 2.2222222222222225e-06,
+      "loss": 0.534,
+      "step": 1
+    },
+    {
+      "epoch": 0.004505773021684032,
+      "grad_norm": 0.2472829588089571,
+      "learning_rate": 4.444444444444445e-06,
+      "loss": 0.483,
+      "step": 2
+    },
+    {
+      "epoch": 0.006758659532526049,
+      "grad_norm": 0.2498680712960353,
+      "learning_rate": 6.666666666666667e-06,
+      "loss": 0.459,
+      "step": 3
+    },
+    {
+      "epoch": 0.009011546043368065,
+      "grad_norm": 0.25154785437869287,
+      "learning_rate": 8.88888888888889e-06,
+      "loss": 0.4815,
+      "step": 4
+    },
+    {
+      "epoch": 0.011264432554210082,
+      "grad_norm": 0.24738671902841008,
+      "learning_rate": 1.1111111111111112e-05,
+      "loss": 0.4957,
+      "step": 5
+    },
+    {
+      "epoch": 0.011264432554210082,
+      "eval_PRM Accuracy": 0.803921568627451,
+      "eval_PRM F1": 0.8507462686567164,
+      "eval_PRM F1 AUC": 0.8214454416627303,
+      "eval_PRM F1 AUC (fixed)": 0.8427019367028814,
+      "eval_PRM F1 Neg": 0.7142857142857143,
+      "eval_PRM NPV": 0.6097560975609756,
+      "eval_PRM Precision": 0.9344262295081968,
+      "eval_PRM Recall": 0.7808219178082192,
+      "eval_PRM Specificty": 0.8620689655172413,
+      "eval_loss": 0.4019886255264282,
+      "eval_runtime": 14.9449,
+      "eval_samples_per_second": 2.944,
+      "eval_steps_per_second": 0.134,
+      "step": 5
+    },
+    {
+      "epoch": 0.013517319065052097,
+      "grad_norm": 0.2521320576421226,
+      "learning_rate": 1.3333333333333333e-05,
+      "loss": 0.4734,
+      "step": 6
+    },
+    {
+      "epoch": 0.015770205575894116,
+      "grad_norm": 0.25614037491141056,
+      "learning_rate": 1.5555555555555555e-05,
+      "loss": 0.4749,
+      "step": 7
+    },
+    {
+      "epoch": 0.01802309208673613,
+      "grad_norm": 0.23988678817334272,
+      "learning_rate": 1.777777777777778e-05,
+      "loss": 0.5055,
+      "step": 8
+    },
+    {
+      "epoch": 0.020275978597578147,
+      "grad_norm": 0.23868331367530465,
+      "learning_rate": 2e-05,
+      "loss": 0.4546,
+      "step": 9
+    },
+    {
+      "epoch": 0.022528865108420164,
+      "grad_norm": 0.3494058102642062,
+      "learning_rate": 2.2222222222222223e-05,
+      "loss": 0.4552,
+      "step": 10
+    },
+    {
+      "epoch": 0.022528865108420164,
+      "eval_PRM Accuracy": 0.803921568627451,
+      "eval_PRM F1": 0.8507462686567164,
+      "eval_PRM F1 AUC": 0.8214454416627303,
+      "eval_PRM F1 AUC (fixed)": 0.8450637694851204,
+      "eval_PRM F1 Neg": 0.7142857142857143,
+      "eval_PRM NPV": 0.6097560975609756,
+      "eval_PRM Precision": 0.9344262295081968,
+      "eval_PRM Recall": 0.7808219178082192,
+      "eval_PRM Specificty": 0.8620689655172413,
+      "eval_loss": 0.40012428164482117,
+      "eval_runtime": 14.6759,
+      "eval_samples_per_second": 2.998,
+      "eval_steps_per_second": 0.136,
+      "step": 10
+    },
+    {
+      "epoch": 0.02478175161926218,
+      "grad_norm": 0.2713632370077711,
+      "learning_rate": 2.4444444444444445e-05,
+      "loss": 0.492,
+      "step": 11
+    },
+    {
+      "epoch": 0.027034638130104195,
+      "grad_norm": 0.24646690528373474,
+      "learning_rate": 2.6666666666666667e-05,
+      "loss": 0.4748,
+      "step": 12
+    },
+    {
+      "epoch": 0.02928752464094621,
+      "grad_norm": 0.3175281734676055,
+      "learning_rate": 2.8888888888888888e-05,
+      "loss": 0.4522,
+      "step": 13
+    },
+    {
+      "epoch": 0.03154041115178823,
+      "grad_norm": 0.2573856095550034,
+      "learning_rate": 3.111111111111111e-05,
+      "loss": 0.4762,
+      "step": 14
+    },
+    {
+      "epoch": 0.03379329766263024,
+      "grad_norm": 0.1935217613661525,
+      "learning_rate": 3.3333333333333335e-05,
+      "loss": 0.4974,
+      "step": 15
+    },
+    {
+      "epoch": 0.03379329766263024,
+      "eval_PRM Accuracy": 0.8431372549019608,
+      "eval_PRM F1": 0.8840579710144928,
+      "eval_PRM F1 AUC": 0.8488427019367027,
+      "eval_PRM F1 AUC (fixed)": 0.8467170524326878,
+      "eval_PRM F1 Neg": 0.7575757575757576,
+      "eval_PRM NPV": 0.6756756756756757,
+      "eval_PRM Precision": 0.9384615384615385,
+      "eval_PRM Recall": 0.8356164383561644,
+      "eval_PRM Specificty": 0.8620689655172413,
+      "eval_loss": 0.3894708752632141,
+      "eval_runtime": 13.781,
+      "eval_samples_per_second": 3.193,
+      "eval_steps_per_second": 0.145,
+      "step": 15
+    },
+    {
+      "epoch": 0.03604618417347226,
+      "grad_norm": 0.2796287302155928,
+      "learning_rate": 3.555555555555556e-05,
+      "loss": 0.4933,
+      "step": 16
+    },
+    {
+      "epoch": 0.038299070684314276,
+      "grad_norm": 0.28994120968737397,
+      "learning_rate": 3.777777777777778e-05,
+      "loss": 0.4491,
+      "step": 17
+    },
+    {
+      "epoch": 0.040551957195156293,
+      "grad_norm": 0.21514253746923748,
+      "learning_rate": 4e-05,
+      "loss": 0.4718,
+      "step": 18
+    },
+    {
+      "epoch": 0.04280484370599831,
+      "grad_norm": 0.15337531730100096,
+      "learning_rate": 4.222222222222222e-05,
+      "loss": 0.4446,
+      "step": 19
+    },
+    {
+      "epoch": 0.04505773021684033,
+      "grad_norm": 0.29809122120923176,
+      "learning_rate": 4.4444444444444447e-05,
+      "loss": 0.4715,
+      "step": 20
+    },
+    {
+      "epoch": 0.04505773021684033,
+      "eval_PRM Accuracy": 0.8823529411764706,
+      "eval_PRM F1": 0.9154929577464789,
+      "eval_PRM F1 AUC": 0.8762399622106755,
+      "eval_PRM F1 AUC (fixed)": 0.8516769012753898,
+      "eval_PRM F1 Neg": 0.8064516129032258,
+      "eval_PRM NPV": 0.7575757575757576,
+      "eval_PRM Precision": 0.9420289855072463,
+      "eval_PRM Recall": 0.8904109589041096,
+      "eval_PRM Specificty": 0.8620689655172413,
+      "eval_loss": 0.36545631289482117,
+      "eval_runtime": 15.1326,
+      "eval_samples_per_second": 2.908,
+      "eval_steps_per_second": 0.132,
+      "step": 20
+    },
+    {
+      "epoch": 0.047310616727682345,
+      "grad_norm": 0.20613724481972295,
+      "learning_rate": 4.666666666666667e-05,
+      "loss": 0.4114,
+      "step": 21
+    },
+    {
+      "epoch": 0.04956350323852436,
+      "grad_norm": 0.18613537246860848,
+      "learning_rate": 4.888888888888889e-05,
+      "loss": 0.468,
+      "step": 22
+    },
+    {
+      "epoch": 0.05181638974936638,
+      "grad_norm": 0.16471842730138056,
+      "learning_rate": 5.111111111111111e-05,
+      "loss": 0.4341,
+      "step": 23
+    },
+    {
+      "epoch": 0.05406927626020839,
+      "grad_norm": 0.17821237760742406,
+      "learning_rate": 5.333333333333333e-05,
+      "loss": 0.4125,
+      "step": 24
+    },
+    {
+      "epoch": 0.056322162771050406,
+      "grad_norm": 0.1690581147679887,
+      "learning_rate": 5.555555555555556e-05,
+      "loss": 0.4215,
+      "step": 25
+    },
+    {
+      "epoch": 0.056322162771050406,
+      "eval_PRM Accuracy": 0.9019607843137255,
+      "eval_PRM F1": 0.9305555555555556,
+      "eval_PRM F1 AUC": 0.8899385923476618,
+      "eval_PRM F1 AUC (fixed)": 0.8573452999527633,
+      "eval_PRM F1 Neg": 0.8333333333333334,
+      "eval_PRM NPV": 0.8064516129032258,
+      "eval_PRM Precision": 0.9436619718309859,
+      "eval_PRM Recall": 0.9178082191780822,
+      "eval_PRM Specificty": 0.8620689655172413,
+      "eval_loss": 0.33864524960517883,
+      "eval_runtime": 13.9395,
+      "eval_samples_per_second": 3.157,
+      "eval_steps_per_second": 0.143,
+      "step": 25
+    },
+    {
+      "epoch": 0.05857504928189242,
+      "grad_norm": 0.19347559569742234,
+      "learning_rate": 5.7777777777777776e-05,
+      "loss": 0.3543,
+      "step": 26
+    },
+    {
+      "epoch": 0.06082793579273444,
+      "grad_norm": 0.23736099384751042,
+      "learning_rate": 6e-05,
+      "loss": 0.4349,
+      "step": 27
+    },
+    {
+      "epoch": 0.06308082230357646,
+      "grad_norm": 0.21048218805465682,
+      "learning_rate": 6.222222222222222e-05,
+      "loss": 0.4202,
+      "step": 28
+    },
+    {
+      "epoch": 0.06533370881441847,
+      "grad_norm": 0.254397253074603,
+      "learning_rate": 6.444444444444446e-05,
+      "loss": 0.4039,
+      "step": 29
+    },
+    {
+      "epoch": 0.06758659532526048,
+      "grad_norm": 0.1870901461724942,
+      "learning_rate": 6.666666666666667e-05,
+      "loss": 0.4069,
+      "step": 30
+    },
+    {
+      "epoch": 0.06758659532526048,
+      "eval_PRM Accuracy": 0.9215686274509803,
+      "eval_PRM F1": 0.9459459459459459,
+      "eval_PRM F1 AUC": 0.8932451582427963,
+      "eval_PRM F1 AUC (fixed)": 0.863013698630137,
+      "eval_PRM F1 Neg": 0.8571428571428571,
+      "eval_PRM NPV": 0.8888888888888888,
+      "eval_PRM Precision": 0.9333333333333333,
+      "eval_PRM Recall": 0.958904109589041,
+      "eval_PRM Specificty": 0.8275862068965517,
+      "eval_loss": 0.32040128111839294,
+      "eval_runtime": 13.707,
+      "eval_samples_per_second": 3.21,
+      "eval_steps_per_second": 0.146,
+      "step": 30
+    },
+    {
+      "epoch": 0.0698394818361025,
+      "grad_norm": 0.2220791143130112,
+      "learning_rate": 6.88888888888889e-05,
+      "loss": 0.4386,
+      "step": 31
+    },
+    {
+      "epoch": 0.07209236834694452,
+      "grad_norm": 0.1883707469546962,
+      "learning_rate": 7.111111111111112e-05,
+      "loss": 0.3533,
+      "step": 32
+    },
+    {
+      "epoch": 0.07434525485778654,
+      "grad_norm": 0.17796013980682868,
+      "learning_rate": 7.333333333333333e-05,
+      "loss": 0.3716,
+      "step": 33
+    },
+    {
+      "epoch": 0.07659814136862855,
+      "grad_norm": 0.2838799777965568,
+      "learning_rate": 7.555555555555556e-05,
+      "loss": 0.4279,
+      "step": 34
+    },
+    {
+      "epoch": 0.07885102787947057,
+      "grad_norm": 0.18082247190300038,
+      "learning_rate": 7.777777777777778e-05,
+      "loss": 0.349,
+      "step": 35
+    },
+    {
+      "epoch": 0.07885102787947057,
+      "eval_PRM Accuracy": 0.9117647058823529,
+      "eval_PRM F1": 0.9395973154362416,
+      "eval_PRM F1 AUC": 0.8760037789324515,
+      "eval_PRM F1 AUC (fixed)": 0.8802550779404817,
+      "eval_PRM F1 Neg": 0.8363636363636363,
+      "eval_PRM NPV": 0.8846153846153846,
+      "eval_PRM Precision": 0.9210526315789473,
+      "eval_PRM Recall": 0.958904109589041,
+      "eval_PRM Specificty": 0.7931034482758621,
+      "eval_loss": 0.30135831236839294,
+      "eval_runtime": 13.5403,
+      "eval_samples_per_second": 3.25,
+      "eval_steps_per_second": 0.148,
+      "step": 35
+    },
+    {
+      "epoch": 0.08110391439031259,
+      "grad_norm": 0.25785848232803393,
+      "learning_rate": 8e-05,
+      "loss": 0.3689,
+      "step": 36
+    },
+    {
+      "epoch": 0.0833568009011546,
+      "grad_norm": 0.2784555050508512,
+      "learning_rate": 8.222222222222222e-05,
+      "loss": 0.3748,
+      "step": 37
+    },
+    {
+      "epoch": 0.08560968741199662,
+      "grad_norm": 0.21180955362645204,
+      "learning_rate": 8.444444444444444e-05,
+      "loss": 0.3654,
+      "step": 38
+    },
+    {
+      "epoch": 0.08786257392283864,
+      "grad_norm": 0.24850318332013638,
+      "learning_rate": 8.666666666666667e-05,
+      "loss": 0.3814,
+      "step": 39
+    },
+    {
+      "epoch": 0.09011546043368066,
+      "grad_norm": 0.3027070734265923,
+      "learning_rate": 8.888888888888889e-05,
+      "loss": 0.3483,
+      "step": 40
+    },
+    {
+      "epoch": 0.09011546043368066,
+      "eval_PRM Accuracy": 0.9117647058823529,
+      "eval_PRM F1": 0.9370629370629371,
+      "eval_PRM F1 AUC": 0.9071799716580067,
+      "eval_PRM F1 AUC (fixed)": 0.8863958431743033,
+      "eval_PRM F1 Neg": 0.8524590163934426,
+      "eval_PRM NPV": 0.8125,
+      "eval_PRM Precision": 0.9571428571428572,
+      "eval_PRM Recall": 0.9178082191780822,
+      "eval_PRM Specificty": 0.896551724137931,
+      "eval_loss": 0.28879615664482117,
+      "eval_runtime": 13.4639,
+      "eval_samples_per_second": 3.268,
+      "eval_steps_per_second": 0.149,
+      "step": 40
+    },
+    {
+      "epoch": 0.09236834694452267,
+      "grad_norm": 0.24016789243143516,
+      "learning_rate": 9.111111111111112e-05,
+      "loss": 0.3363,
+      "step": 41
+    },
+    {
+      "epoch": 0.09462123345536469,
+      "grad_norm": 0.26333269113271995,
+      "learning_rate": 9.333333333333334e-05,
+      "loss": 0.3484,
+      "step": 42
+    },
+    {
+      "epoch": 0.0968741199662067,
+      "grad_norm": 0.26089994289870105,
+      "learning_rate": 9.555555555555557e-05,
+      "loss": 0.3211,
+      "step": 43
+    },
+    {
+      "epoch": 0.09912700647704872,
+      "grad_norm": 0.43987117694649375,
+      "learning_rate": 9.777777777777778e-05,
+      "loss": 0.321,
+      "step": 44
+    },
+    {
+      "epoch": 0.10137989298789074,
+      "grad_norm": 0.48949236880217784,
+      "learning_rate": 0.0001,
+      "loss": 0.3461,
+      "step": 45
+    },
+    {
+      "epoch": 0.10137989298789074,
+      "eval_PRM Accuracy": 0.8627450980392157,
+      "eval_PRM F1": 0.8985507246376812,
+      "eval_PRM F1 AUC": 0.8729333963155408,
+      "eval_PRM F1 AUC (fixed)": 0.8823807274444969,
+      "eval_PRM F1 Neg": 0.7878787878787878,
+      "eval_PRM NPV": 0.7027027027027027,
+      "eval_PRM Precision": 0.9538461538461539,
+      "eval_PRM Recall": 0.8493150684931506,
+      "eval_PRM Specificty": 0.896551724137931,
+      "eval_loss": 0.2819158434867859,
+      "eval_runtime": 14.7204,
+      "eval_samples_per_second": 2.989,
+      "eval_steps_per_second": 0.136,
+      "step": 45
+    },
+    {
+      "epoch": 0.10363277949873276,
+      "grad_norm": 0.2845537005130612,
+      "learning_rate": 9.999844234470782e-05,
+      "loss": 0.3263,
+      "step": 46
+    },
+    {
+      "epoch": 0.10588566600957476,
+      "grad_norm": 0.28870132628126455,
+      "learning_rate": 9.999376947588288e-05,
+      "loss": 0.2917,
+      "step": 47
+    },
+    {
+      "epoch": 0.10813855252041678,
+      "grad_norm": 0.2964705719584474,
+      "learning_rate": 9.99859816846739e-05,
+      "loss": 0.252,
+      "step": 48
+    },
+    {
+      "epoch": 0.1103914390312588,
+      "grad_norm": 0.3050617694847216,
+      "learning_rate": 9.99750794563087e-05,
+      "loss": 0.265,
+      "step": 49
+    },
+    {
+      "epoch": 0.11264432554210081,
+      "grad_norm": 0.5836677203708822,
+      "learning_rate": 9.996106347006379e-05,
+      "loss": 0.3105,
+      "step": 50
+    },
+    {
+      "epoch": 0.11264432554210081,
+      "eval_PRM Accuracy": 0.8627450980392157,
+      "eval_PRM F1": 0.8985507246376812,
+      "eval_PRM F1 AUC": 0.8729333963155408,
+      "eval_PRM F1 AUC (fixed)": 0.884506376948512,
+      "eval_PRM F1 Neg": 0.7878787878787878,
+      "eval_PRM NPV": 0.7027027027027027,
+      "eval_PRM Precision": 0.9538461538461539,
+      "eval_PRM Recall": 0.8493150684931506,
+      "eval_PRM Specificty": 0.896551724137931,
+      "eval_loss": 0.2556818127632141,
+      "eval_runtime": 13.4262,
+      "eval_samples_per_second": 3.277,
+      "eval_steps_per_second": 0.149,
+      "step": 50
+    },
+    {
+      "epoch": 0.11489721205294283,
+      "grad_norm": 0.39701575729709887,
+      "learning_rate": 9.994393459922218e-05,
+      "loss": 0.3,
+      "step": 51
+    },
+    {
+      "epoch": 0.11715009856378485,
+      "grad_norm": 0.3753161334703991,
+      "learning_rate": 9.992369391101895e-05,
+      "loss": 0.288,
+      "step": 52
+    },
+    {
+      "epoch": 0.11940298507462686,
+      "grad_norm": 0.41833689505885724,
+      "learning_rate": 9.990034266657467e-05,
+      "loss": 0.3162,
+      "step": 53
+    },
+    {
+      "epoch": 0.12165587158546888,
+      "grad_norm": 0.5561971125744725,
+      "learning_rate": 9.987388232081694e-05,
+      "loss": 0.2691,
+      "step": 54
+    },
+    {
+      "epoch": 0.1239087580963109,
+      "grad_norm": 0.4275941689301913,
+      "learning_rate": 9.984431452238967e-05,
+      "loss": 0.2924,
+      "step": 55
+    },
+    {
+      "epoch": 0.1239087580963109,
+      "eval_PRM Accuracy": 0.8823529411764706,
+      "eval_PRM F1": 0.9117647058823529,
+      "eval_PRM F1 AUC": 0.9074161549362305,
+      "eval_PRM F1 AUC (fixed)": 0.8977326405290504,
+      "eval_PRM F1 Neg": 0.8235294117647058,
+      "eval_PRM NPV": 0.717948717948718,
+      "eval_PRM Precision": 0.9841269841269841,
+      "eval_PRM Recall": 0.8493150684931506,
+      "eval_PRM Specificty": 0.9655172413793104,
+      "eval_loss": 0.23599520325660706,
+      "eval_runtime": 13.3025,
+      "eval_samples_per_second": 3.308,
+      "eval_steps_per_second": 0.15,
+      "step": 55
+    },
+    {
+      "epoch": 0.12616164460715293,
+      "grad_norm": 0.349991312818601,
+      "learning_rate": 9.981164111355035e-05,
+      "loss": 0.2431,
+      "step": 56
+    },
+    {
+      "epoch": 0.12841453111799492,
+      "grad_norm": 0.6215772500225185,
+      "learning_rate": 9.977586413005531e-05,
+      "loss": 0.3137,
+      "step": 57
+    },
+    {
+      "epoch": 0.13066741762883693,
+      "grad_norm": 0.5214590014261731,
+      "learning_rate": 9.973698580103285e-05,
+      "loss": 0.3508,
+      "step": 58
+    },
+    {
+      "epoch": 0.13292030413967895,
+      "grad_norm": 0.47678677496939564,
+      "learning_rate": 9.96950085488444e-05,
+      "loss": 0.2854,
+      "step": 59
+    },
+    {
+      "epoch": 0.13517319065052097,
+      "grad_norm": 0.5162976401300434,
+      "learning_rate": 9.964993498893349e-05,
+      "loss": 0.3195,
+      "step": 60
+    },
+    {
+      "epoch": 0.13517319065052097,
+      "eval_PRM Accuracy": 0.8823529411764706,
+      "eval_PRM F1": 0.9117647058823529,
+      "eval_PRM F1 AUC": 0.9074161549362305,
+      "eval_PRM F1 AUC (fixed)": 0.9145016532829476,
+      "eval_PRM F1 Neg": 0.8235294117647058,
+      "eval_PRM NPV": 0.717948717948718,
+      "eval_PRM Precision": 0.9841269841269841,
+      "eval_PRM Recall": 0.8493150684931506,
+      "eval_PRM Specificty": 0.9655172413793104,
+      "eval_loss": 0.24027876555919647,
+      "eval_runtime": 13.2962,
+      "eval_samples_per_second": 3.309,
+      "eval_steps_per_second": 0.15,
+      "step": 60
+    },
+    {
+      "epoch": 0.137426077161363,
+      "grad_norm": 0.6781095199961917,
+      "learning_rate": 9.960176792966289e-05,
+      "loss": 0.349,
+      "step": 61
+    },
+    {
+      "epoch": 0.139678963672205,
+      "grad_norm": 0.5833900089314961,
+      "learning_rate": 9.95505103721396e-05,
+      "loss": 0.2625,
+      "step": 62
+    },
+    {
+      "epoch": 0.14193185018304702,
+      "grad_norm": 0.5084722245050096,
+      "learning_rate": 9.949616551002787e-05,
+      "loss": 0.3088,
+      "step": 63
+    },
+    {
+      "epoch": 0.14418473669388904,
+      "grad_norm": 0.40925068479342624,
+      "learning_rate": 9.943873672935014e-05,
+      "loss": 0.2621,
+      "step": 64
+    },
+    {
+      "epoch": 0.14643762320473105,
+      "grad_norm": 0.3784991366032602,
+      "learning_rate": 9.93782276082762e-05,
+      "loss": 0.3174,
+      "step": 65
+    },
+    {
+      "epoch": 0.14643762320473105,
+      "eval_PRM Accuracy": 0.9215686274509803,
+      "eval_PRM F1": 0.9428571428571428,
+      "eval_PRM F1 AUC": 0.9348134152102031,
+      "eval_PRM F1 AUC (fixed)": 0.8970240906943788,
+      "eval_PRM F1 Neg": 0.875,
+      "eval_PRM NPV": 0.8,
+      "eval_PRM Precision": 0.9850746268656716,
+      "eval_PRM Recall": 0.9041095890410958,
+      "eval_PRM Specificty": 0.9655172413793104,
+      "eval_loss": 0.21548739075660706,
+      "eval_runtime": 9.4155,
+      "eval_samples_per_second": 4.673,
+      "eval_steps_per_second": 0.212,
+      "step": 65
+    },
+    {
+      "epoch": 0.14869050971557307,
+      "grad_norm": 0.4534993567788596,
+      "learning_rate": 9.931464191690015e-05,
+      "loss": 0.295,
+      "step": 66
+    },
+    {
+      "epoch": 0.1509433962264151,
+      "grad_norm": 0.5016796203723086,
+      "learning_rate": 9.924798361700553e-05,
+      "loss": 0.2583,
+      "step": 67
+    },
+    {
+      "epoch": 0.1531962827372571,
+      "grad_norm": 0.37530785542303907,
+      "learning_rate": 9.91782568618185e-05,
+      "loss": 0.3137,
+      "step": 68
+    },
+    {
+      "epoch": 0.15544916924809912,
+      "grad_norm": 0.5589825951912164,
+      "learning_rate": 9.910546599574902e-05,
+      "loss": 0.3374,
+      "step": 69
+    },
+    {
+      "epoch": 0.15770205575894114,
+      "grad_norm": 0.4682803646220257,
+      "learning_rate": 9.90296155541202e-05,
+      "loss": 0.3069,
+      "step": 70
+    },
+    {
+      "epoch": 0.15770205575894114,
+      "eval_PRM Accuracy": 0.9019607843137255,
+      "eval_PRM F1": 0.927536231884058,
+      "eval_PRM F1 AUC": 0.9211147850732168,
+      "eval_PRM F1 AUC (fixed)": 0.8901747756258858,
+      "eval_PRM F1 Neg": 0.8484848484848485,
+      "eval_PRM NPV": 0.7567567567567568,
+      "eval_PRM Precision": 0.9846153846153847,
+      "eval_PRM Recall": 0.8767123287671232,
+      "eval_PRM Specificty": 0.9655172413793104,
+      "eval_loss": 0.2295587658882141,
+      "eval_runtime": 9.9609,
+      "eval_samples_per_second": 4.417,
+      "eval_steps_per_second": 0.201,
+      "step": 70
+    },
+    {
+      "epoch": 0.15995494226978316,
+      "grad_norm": 0.40899474687824006,
+      "learning_rate": 9.895071026288574e-05,
+      "loss": 0.3042,
+      "step": 71
+    },
+    {
+      "epoch": 0.16220782878062517,
+      "grad_norm": 0.41050009025994666,
+      "learning_rate": 9.886875503833536e-05,
+      "loss": 0.3415,
+      "step": 72
+    },
+    {
+      "epoch": 0.1644607152914672,
+      "grad_norm": 0.40937295359462034,
+      "learning_rate": 9.87837549867887e-05,
+      "loss": 0.2433,
+      "step": 73
+    },
+    {
+      "epoch": 0.1667136018023092,
+      "grad_norm": 0.3588489658348455,
+      "learning_rate": 9.869571540427689e-05,
+      "loss": 0.2788,
+      "step": 74
+    },
+    {
+      "epoch": 0.16896648831315123,
+      "grad_norm": 0.3624803899750991,
+      "learning_rate": 9.860464177621284e-05,
+      "loss": 0.2821,
+      "step": 75
+    },
+    {
+      "epoch": 0.16896648831315123,
+      "eval_PRM Accuracy": 0.8725490196078431,
+      "eval_PRM F1": 0.9037037037037037,
+      "eval_PRM F1 AUC": 0.9005668398677373,
+      "eval_PRM F1 AUC (fixed)": 0.8889938592347661,
+      "eval_PRM F1 Neg": 0.8115942028985508,
+      "eval_PRM NPV": 0.7,
+      "eval_PRM Precision": 0.9838709677419355,
+      "eval_PRM Recall": 0.8356164383561644,
+      "eval_PRM Specificty": 0.9655172413793104,
+      "eval_loss": 0.2620960474014282,
+      "eval_runtime": 9.805,
+      "eval_samples_per_second": 4.488,
+      "eval_steps_per_second": 0.204,
+      "step": 75
+    },
+    {
+      "epoch": 0.17121937482399324,
+      "grad_norm": 0.3827868819608748,
+      "learning_rate": 9.851053977704931e-05,
+      "loss": 0.2919,
+      "step": 76
+    },
+    {
+      "epoch": 0.17347226133483526,
+      "grad_norm": 0.4674449466768407,
+      "learning_rate": 9.841341526992536e-05,
+      "loss": 0.2889,
+      "step": 77
+    },
+    {
+      "epoch": 0.17572514784567728,
+      "grad_norm": 0.3543143366425332,
+      "learning_rate": 9.83132743063011e-05,
+      "loss": 0.2506,
+      "step": 78
+    },
+    {
+      "epoch": 0.1779780343565193,
+      "grad_norm": 0.3889354215083509,
+      "learning_rate": 9.821012312558058e-05,
+      "loss": 0.2471,
+      "step": 79
+    },
+    {
+      "epoch": 0.1802309208673613,
+      "grad_norm": 0.38782618082041825,
+      "learning_rate": 9.810396815472314e-05,
+      "loss": 0.2904,
+      "step": 80
+    },
+    {
+      "epoch": 0.1802309208673613,
+      "eval_PRM Accuracy": 0.8921568627450981,
+      "eval_PRM F1": 0.9197080291970803,
+      "eval_PRM F1 AUC": 0.9142654700047237,
+      "eval_PRM F1 AUC (fixed)": 0.8939537080774682,
+      "eval_PRM F1 Neg": 0.835820895522388,
+      "eval_PRM NPV": 0.7368421052631579,
+      "eval_PRM Precision": 0.984375,
+      "eval_PRM Recall": 0.863013698630137,
+      "eval_PRM Specificty": 0.9655172413793104,
+      "eval_loss": 0.23648348450660706,
+      "eval_runtime": 9.9993,
+      "eval_samples_per_second": 4.4,
+      "eval_steps_per_second": 0.2,
+      "step": 80
+    },
+    {
+      "epoch": 0.18248380737820333,
+      "grad_norm": 0.6065442457988935,
+      "learning_rate": 9.799481600784286e-05,
+      "loss": 0.2406,
+      "step": 81
+    },
+    {
+      "epoch": 0.18473669388904534,
+      "grad_norm": 0.3593408459220416,
+      "learning_rate": 9.788267348579648e-05,
+      "loss": 0.2223,
+      "step": 82
+    },
+    {
+      "epoch": 0.18698958039988736,
+      "grad_norm": 0.4097655659822076,
+      "learning_rate": 9.776754757575975e-05,
+      "loss": 0.2679,
+      "step": 83
+    },
+    {
+      "epoch": 0.18924246691072938,
+      "grad_norm": 0.379243222090275,
+      "learning_rate": 9.764944545079196e-05,
+      "loss": 0.2411,
+      "step": 84
+    },
+    {
+      "epoch": 0.1914953534215714,
+      "grad_norm": 0.48330678894058415,
+      "learning_rate": 9.752837446938915e-05,
+      "loss": 0.226,
+      "step": 85
+    },
+    {
+      "epoch": 0.1914953534215714,
+      "eval_PRM Accuracy": 0.9215686274509803,
+      "eval_PRM F1": 0.9428571428571428,
+      "eval_PRM F1 AUC": 0.9348134152102031,
+      "eval_PRM F1 AUC (fixed)": 0.9026924893717525,
+      "eval_PRM F1 Neg": 0.875,
+      "eval_PRM NPV": 0.8,
+      "eval_PRM Precision": 0.9850746268656716,
+      "eval_PRM Recall": 0.9041095890410958,
+      "eval_PRM Specificty": 0.9655172413793104,
+      "eval_loss": 0.20969460904598236,
+      "eval_runtime": 9.6543,
+      "eval_samples_per_second": 4.558,
+      "eval_steps_per_second": 0.207,
+      "step": 85
+    },
+    {
+      "epoch": 0.1937482399324134,
+      "grad_norm": 0.522205265428476,
+      "learning_rate": 9.740434217502547e-05,
+      "loss": 0.2939,
+      "step": 86
+    },
+    {
+      "epoch": 0.19600112644325543,
+      "grad_norm": 0.7052468385738452,
+      "learning_rate": 9.727735629568336e-05,
+      "loss": 0.2423,
+      "step": 87
+    },
+    {
+      "epoch": 0.19825401295409745,
+      "grad_norm": 0.43140383673467647,
+      "learning_rate": 9.714742474337186e-05,
+      "loss": 0.2205,
+      "step": 88
+    },
+    {
+      "epoch": 0.20050689946493946,
+      "grad_norm": 0.40292840995906376,
+      "learning_rate": 9.701455561363379e-05,
+      "loss": 0.2381,
+      "step": 89
+    },
+    {
+      "epoch": 0.20275978597578148,
+      "grad_norm": 0.5735785772636156,
+      "learning_rate": 9.687875718504126e-05,
+      "loss": 0.2534,
+      "step": 90
+    },
+    {
+      "epoch": 0.20275978597578148,
+      "eval_PRM Accuracy": 0.8823529411764706,
+      "eval_PRM F1": 0.9117647058823529,
+      "eval_PRM F1 AUC": 0.9074161549362305,
+      "eval_PRM F1 AUC (fixed)": 0.9192253188474256,
+      "eval_PRM F1 Neg": 0.8235294117647058,
+      "eval_PRM NPV": 0.717948717948718,
+      "eval_PRM Precision": 0.9841269841269841,
+      "eval_PRM Recall": 0.8493150684931506,
+      "eval_PRM Specificty": 0.9655172413793104,
+      "eval_loss": 0.22407670319080353,
+      "eval_runtime": 14.1488,
+      "eval_samples_per_second": 3.11,
+      "eval_steps_per_second": 0.141,
+      "step": 90
+    },
+    {
+      "epoch": 0.2050126724866235,
+      "grad_norm": 0.4145567522929326,
+      "learning_rate": 9.674003791867991e-05,
+      "loss": 0.2639,
+      "step": 91
+    },
+    {
+      "epoch": 0.20726555899746552,
+      "grad_norm": 0.6232738228861782,
+      "learning_rate": 9.659840645762175e-05,
+      "loss": 0.267,
+      "step": 92
+    },
+    {
+      "epoch": 0.20951844550830753,
+      "grad_norm": 0.6330512110807636,
+      "learning_rate": 9.645387162638652e-05,
+      "loss": 0.285,
+      "step": 93
+    },
+    {
+      "epoch": 0.21177133201914952,
+      "grad_norm": 0.4569457366597659,
+      "learning_rate": 9.630644243039207e-05,
+      "loss": 0.2375,
+      "step": 94
+    },
+    {
+      "epoch": 0.21402421852999154,
+      "grad_norm": 0.41916021516522795,
+      "learning_rate": 9.615612805539305e-05,
+      "loss": 0.2278,
+      "step": 95
+    },
+    {
+      "epoch": 0.21402421852999154,
+      "eval_PRM Accuracy": 0.9019607843137255,
+      "eval_PRM F1": 0.927536231884058,
+      "eval_PRM F1 AUC": 0.9211147850732168,
+      "eval_PRM F1 AUC (fixed)": 0.8977326405290506,
+      "eval_PRM F1 Neg": 0.8484848484848485,
+      "eval_PRM NPV": 0.7567567567567568,
+      "eval_PRM Precision": 0.9846153846153847,
+      "eval_PRM Recall": 0.8767123287671232,
+      "eval_PRM Specificty": 0.9655172413793104,
+      "eval_loss": 0.2196599841117859,
+      "eval_runtime": 14.5631,
+      "eval_samples_per_second": 3.021,
+      "eval_steps_per_second": 0.137,
+      "step": 95
+    },
+    {
+      "epoch": 0.21627710504083356,
+      "grad_norm": 0.3907042230794063,
+      "learning_rate": 9.600293786690872e-05,
+      "loss": 0.2604,
+      "step": 96
+    },
+    {
+      "epoch": 0.21852999155167557,
+      "grad_norm": 0.41318284701964014,
+      "learning_rate": 9.584688140963944e-05,
+      "loss": 0.2404,
+      "step": 97
+    },
+    {
+      "epoch": 0.2207828780625176,
+      "grad_norm": 0.559095628822006,
+      "learning_rate": 9.568796840687184e-05,
+      "loss": 0.2053,
+      "step": 98
+    },
+    {
+      "epoch": 0.2230357645733596,
+      "grad_norm": 0.45376733184445756,
+      "learning_rate": 9.552620875987311e-05,
+      "loss": 0.2235,
+      "step": 99
+    },
+    {
+      "epoch": 0.22528865108420162,
+      "grad_norm": 0.4299937730197361,
+      "learning_rate": 9.536161254727408e-05,
+      "loss": 0.198,
+      "step": 100
+    },
+    {
+      "epoch": 0.22528865108420162,
+      "eval_PRM Accuracy": 0.8823529411764706,
+      "eval_PRM F1": 0.9154929577464789,
+      "eval_PRM F1 AUC": 0.8762399622106755,
+      "eval_PRM F1 AUC (fixed)": 0.8904109589041096,
+      "eval_PRM F1 Neg": 0.8064516129032258,
+      "eval_PRM NPV": 0.7575757575757576,
+      "eval_PRM Precision": 0.9420289855072463,
+      "eval_PRM Recall": 0.8904109589041096,
+      "eval_PRM Specificty": 0.8620689655172413,
+      "eval_loss": 0.22010387480258942,
+      "eval_runtime": 14.383,
+      "eval_samples_per_second": 3.059,
+      "eval_steps_per_second": 0.139,
+      "step": 100
+    },
+    {
+      "epoch": 0.22754153759504364,
+      "grad_norm": 0.5585196713347745,
+      "learning_rate": 9.51941900244412e-05,
+      "loss": 0.2022,
+      "step": 101
+    },
+    {
+      "epoch": 0.22979442410588566,
+      "grad_norm": 0.495814475708339,
+      "learning_rate": 9.502395162283759e-05,
+      "loss": 0.2576,
+      "step": 102
+    },
+    {
+      "epoch": 0.23204731061672768,
+      "grad_norm": 0.4892471251632609,
+      "learning_rate": 9.485090794937319e-05,
+      "loss": 0.2539,
+      "step": 103
+    },
+    {
+      "epoch": 0.2343001971275697,
+      "grad_norm": 0.7682821324812346,
+      "learning_rate": 9.467506978574371e-05,
+      "loss": 0.2967,
+      "step": 104
+    },
+    {
+      "epoch": 0.2365530836384117,
+      "grad_norm": 0.47310691652196873,
+      "learning_rate": 9.449644808775902e-05,
+      "loss": 0.2287,
+      "step": 105
+    },
+    {
+      "epoch": 0.2365530836384117,
+      "eval_PRM Accuracy": 0.8921568627450981,
+      "eval_PRM F1": 0.9197080291970803,
+      "eval_PRM F1 AUC": 0.9142654700047237,
+      "eval_PRM F1 AUC (fixed)": 0.9140292867264999,
+      "eval_PRM F1 Neg": 0.835820895522388,
+      "eval_PRM NPV": 0.7368421052631579,
+      "eval_PRM Precision": 0.984375,
+      "eval_PRM Recall": 0.863013698630137,
+      "eval_PRM Specificty": 0.9655172413793104,
+      "eval_loss": 0.23409757018089294,
+      "eval_runtime": 13.9217,
+      "eval_samples_per_second": 3.161,
+      "eval_steps_per_second": 0.144,
+      "step": 105
+    },
+    {
+      "epoch": 0.23880597014925373,
+      "grad_norm": 0.572365200338062,
+      "learning_rate": 9.431505398466045e-05,
+      "loss": 0.2419,
+      "step": 106
+    },
+    {
+      "epoch": 0.24105885666009574,
+      "grad_norm": 0.5585049297825276,
+      "learning_rate": 9.413089877842736e-05,
+      "loss": 0.3065,
+      "step": 107
+    },
+    {
+      "epoch": 0.24331174317093776,
+      "grad_norm": 0.5723502866189658,
+      "learning_rate": 9.394399394307303e-05,
+      "loss": 0.3036,
+      "step": 108
+    },
+    {
+      "epoch": 0.24556462968177978,
+      "grad_norm": 0.498203689902634,
+      "learning_rate": 9.375435112392969e-05,
+      "loss": 0.2549,
+      "step": 109
+    },
+    {
+      "epoch": 0.2478175161926218,
+      "grad_norm": 0.5070759346070048,
+      "learning_rate": 9.356198213692297e-05,
+      "loss": 0.2597,
+      "step": 110
+    },
+    {
+      "epoch": 0.2478175161926218,
+      "eval_PRM Accuracy": 0.8823529411764706,
+      "eval_PRM F1": 0.9117647058823529,
+      "eval_PRM F1 AUC": 0.9074161549362305,
+      "eval_PRM F1 AUC (fixed)": 0.9279641001417099,
+      "eval_PRM F1 Neg": 0.8235294117647058,
+      "eval_PRM NPV": 0.717948717948718,
+      "eval_PRM Precision": 0.9841269841269841,
+      "eval_PRM Recall": 0.8493150684931506,
+      "eval_PRM Specificty": 0.9655172413793104,
+      "eval_loss": 0.23660555481910706,
+      "eval_runtime": 13.8508,
+      "eval_samples_per_second": 3.177,
+      "eval_steps_per_second": 0.144,
+      "step": 110
+    },
+    {
+      "epoch": 0.2500704027034638,
+      "grad_norm": 0.5105319862843105,
+      "learning_rate": 9.336689896783573e-05,
+      "loss": 0.2609,
+      "step": 111
+    },
+    {
+      "epoch": 0.25232328921430586,
+      "grad_norm": 0.5355761743373533,
+      "learning_rate": 9.316911377156117e-05,
+      "loss": 0.2521,
+      "step": 112
+    },
+    {
+      "epoch": 0.25457617572514785,
+      "grad_norm": 0.592620230599529,
+      "learning_rate": 9.29686388713456e-05,
+      "loss": 0.2489,
+      "step": 113
+    },
+    {
+      "epoch": 0.25682906223598984,
+      "grad_norm": 0.4830329219939314,
+      "learning_rate": 9.276548675802059e-05,
+      "loss": 0.2682,
+      "step": 114
+    },
+    {
+      "epoch": 0.2590819487468319,
+      "grad_norm": 0.5319912621201975,
+      "learning_rate": 9.255967008922474e-05,
+      "loss": 0.2479,
+      "step": 115
+    },
+    {
+      "epoch": 0.2590819487468319,
+      "eval_PRM Accuracy": 0.9117647058823529,
+      "eval_PRM F1": 0.9361702127659575,
+      "eval_PRM F1 AUC": 0.9175720358998584,
+      "eval_PRM F1 AUC (fixed)": 0.9315068493150684,
+      "eval_PRM F1 Neg": 0.8571428571428571,
+      "eval_PRM NPV": 0.7941176470588235,
+      "eval_PRM Precision": 0.9705882352941176,
+      "eval_PRM Recall": 0.9041095890410958,
+      "eval_PRM Specificty": 0.9310344827586207,
+      "eval_loss": 0.21530984342098236,
+      "eval_runtime": 13.4658,
+      "eval_samples_per_second": 3.268,
+      "eval_steps_per_second": 0.149,
+      "step": 115
+    },
+    {
+      "epoch": 0.26133483525767387,
+      "grad_norm": 0.5216692451136674,
+      "learning_rate": 9.235120168861496e-05,
+      "loss": 0.2315,
+      "step": 116
+    },
+    {
+      "epoch": 0.2635877217685159,
+      "grad_norm": 0.5221878314065512,
+      "learning_rate": 9.214009454506753e-05,
+      "loss": 0.2511,
+      "step": 117
+    },
+    {
+      "epoch": 0.2658406082793579,
+      "grad_norm": 0.5280571825274145,
+      "learning_rate": 9.192636181186888e-05,
+      "loss": 0.2703,
+      "step": 118
+    },
+    {
+      "epoch": 0.26809349479019995,
+      "grad_norm": 0.5134286170582513,
+      "learning_rate": 9.171001680589588e-05,
+      "loss": 0.2607,
+      "step": 119
+    },
+    {
+      "epoch": 0.27034638130104194,
+      "grad_norm": 0.47359596282347816,
+      "learning_rate": 9.14910730067863e-05,
+      "loss": 0.232,
+      "step": 120
+    },
+    {
+      "epoch": 0.27034638130104194,
+      "eval_PRM Accuracy": 0.9019607843137255,
+      "eval_PRM F1": 0.9295774647887324,
+      "eval_PRM F1 AUC": 0.9003306565895135,
+      "eval_PRM F1 AUC (fixed)": 0.9338686820973076,
+      "eval_PRM F1 Neg": 0.8387096774193549,
+      "eval_PRM NPV": 0.7878787878787878,
+      "eval_PRM Precision": 0.9565217391304348,
+      "eval_PRM Recall": 0.9041095890410958,
+      "eval_PRM Specificty": 0.896551724137931,
+      "eval_loss": 0.205078125,
+      "eval_runtime": 12.9316,
+      "eval_samples_per_second": 3.403,
+      "eval_steps_per_second": 0.155,
+      "step": 120
+    },
+    {
+      "epoch": 0.272599267811884,
+      "grad_norm": 0.4711168713508809,
+      "learning_rate": 9.126954405609882e-05,
+      "loss": 0.2689,
+      "step": 121
+    },
+    {
+      "epoch": 0.274852154322726,
+      "grad_norm": 0.6010548386584329,
+      "learning_rate": 9.104544375646313e-05,
+      "loss": 0.3001,
+      "step": 122
+    },
+    {
+      "epoch": 0.277105040833568,
+      "grad_norm": 0.5127902494596478,
+      "learning_rate": 9.081878607071996e-05,
+      "loss": 0.3029,
+      "step": 123
+    },
+    {
+      "epoch": 0.27935792734441,
+      "grad_norm": 0.44542644798327063,
+      "learning_rate": 9.058958512105104e-05,
+      "loss": 0.2072,
+      "step": 124
+    },
+    {
+      "epoch": 0.28161081385525205,
+      "grad_norm": 0.47928830424250285,
+      "learning_rate": 9.035785518809927e-05,
+      "loss": 0.2441,
+      "step": 125
+    },
+    {
+      "epoch": 0.28161081385525205,
+      "eval_PRM Accuracy": 0.9117647058823529,
+      "eval_PRM F1": 0.9361702127659575,
+      "eval_PRM F1 AUC": 0.9175720358998584,
+      "eval_PRM F1 AUC (fixed)": 0.9407179971658006,
+      "eval_PRM F1 Neg": 0.8571428571428571,
+      "eval_PRM NPV": 0.7941176470588235,
+      "eval_PRM Precision": 0.9705882352941176,
+      "eval_PRM Recall": 0.9041095890410958,
+      "eval_PRM Specificty": 0.9310344827586207,
+      "eval_loss": 0.20699796080589294,
+      "eval_runtime": 13.5767,
+      "eval_samples_per_second": 3.241,
+      "eval_steps_per_second": 0.147,
+      "step": 125
+    },
+    {
+      "epoch": 0.28386370036609404,
+      "grad_norm": 0.4392423453387639,
+      "learning_rate": 9.012361071007891e-05,
+      "loss": 0.246,
+      "step": 126
+    },
+    {
+      "epoch": 0.2861165868769361,
+      "grad_norm": 0.6192432101052482,
+      "learning_rate": 8.988686628187597e-05,
+      "loss": 0.2034,
+      "step": 127
+    },
+    {
+      "epoch": 0.2883694733877781,
+      "grad_norm": 0.38677591760788566,
+      "learning_rate": 8.964763665413893e-05,
+      "loss": 0.2144,
+      "step": 128
+    },
+    {
+      "epoch": 0.2906223598986201,
+      "grad_norm": 0.5136918839465369,
+      "learning_rate": 8.940593673235962e-05,
+      "loss": 0.2276,
+      "step": 129
+    },
+    {
+      "epoch": 0.2928752464094621,
+      "grad_norm": 0.4505508024241789,
+      "learning_rate": 8.916178157594453e-05,
+      "loss": 0.2062,
+      "step": 130
+    },
+    {
+      "epoch": 0.2928752464094621,
+      "eval_PRM Accuracy": 0.9019607843137255,
+      "eval_PRM F1": 0.927536231884058,
+      "eval_PRM F1 AUC": 0.9211147850732168,
+      "eval_PRM F1 AUC (fixed)": 0.9433160132262636,
+      "eval_PRM F1 Neg": 0.8484848484848485,
+      "eval_PRM NPV": 0.7567567567567568,
+      "eval_PRM Precision": 0.9846153846153847,
+      "eval_PRM Recall": 0.8767123287671232,
+      "eval_PRM Specificty": 0.9655172413793104,
+      "eval_loss": 0.20957253873348236,
+      "eval_runtime": 13.9687,
+      "eval_samples_per_second": 3.15,
+      "eval_steps_per_second": 0.143,
+      "step": 130
+    },
+    {
+      "epoch": 0.29512813292030415,
+      "grad_norm": 0.5958359395939703,
+      "learning_rate": 8.891518639727649e-05,
+      "loss": 0.2256,
+      "step": 131
+    },
+    {
+      "epoch": 0.29738101943114614,
+      "grad_norm": 0.5334457119999583,
+      "learning_rate": 8.866616656076696e-05,
+      "loss": 0.2447,
+      "step": 132
+    },
+    {
+      "epoch": 0.2996339059419882,
+      "grad_norm": 0.45369539499139666,
+      "learning_rate": 8.841473758189854e-05,
+      "loss": 0.2295,
+      "step": 133
+    },
+    {
+      "epoch": 0.3018867924528302,
+      "grad_norm": 0.45072674002260876,
+      "learning_rate": 8.816091512625843e-05,
+      "loss": 0.2475,
+      "step": 134
+    },
+    {
+      "epoch": 0.3041396789636722,
+      "grad_norm": 0.7348790583240027,
+      "learning_rate": 8.790471500856228e-05,
+      "loss": 0.2558,
+      "step": 135
+    },
+    {
+      "epoch": 0.3041396789636722,
+      "eval_PRM Accuracy": 0.9019607843137255,
+      "eval_PRM F1": 0.927536231884058,
+      "eval_PRM F1 AUC": 0.9211147850732168,
+      "eval_PRM F1 AUC (fixed)": 0.9466225791213982,
+      "eval_PRM F1 Neg": 0.8484848484848485,
+      "eval_PRM NPV": 0.7567567567567568,
+      "eval_PRM Precision": 0.9846153846153847,
+      "eval_PRM Recall": 0.8767123287671232,
+      "eval_PRM Specificty": 0.9655172413793104,
+      "eval_loss": 0.1993463635444641,
+      "eval_runtime": 13.5833,
+      "eval_samples_per_second": 3.239,
+      "eval_steps_per_second": 0.147,
+      "step": 135
+    },
+    {
+      "epoch": 0.3063925654745142,
+      "grad_norm": 0.55546584064707,
+      "learning_rate": 8.764615319166886e-05,
+      "loss": 0.2043,
+      "step": 136
+    },
+    {
+      "epoch": 0.30864545198535626,
+      "grad_norm": 0.6027318531715503,
+      "learning_rate": 8.738524578558547e-05,
+      "loss": 0.2044,
+      "step": 137
+    },
+    {
+      "epoch": 0.31089833849619825,
+      "grad_norm": 0.4750447613537508,
+      "learning_rate": 8.712200904646416e-05,
+      "loss": 0.2443,
+      "step": 138
+    },
+    {
+      "epoch": 0.3131512250070403,
+      "grad_norm": 0.4322634772233948,
+      "learning_rate": 8.685645937558896e-05,
+      "loss": 0.2613,
+      "step": 139
+    },
+    {
+      "epoch": 0.3154041115178823,
+      "grad_norm": 0.46337845739308736,
+      "learning_rate": 8.658861331835385e-05,
+      "loss": 0.2381,
+      "step": 140
+    },
+    {
+      "epoch": 0.3154041115178823,
+      "eval_PRM Accuracy": 0.9117647058823529,
+      "eval_PRM F1": 0.9370629370629371,
+      "eval_PRM F1 AUC": 0.9071799716580067,
+      "eval_PRM F1 AUC (fixed)": 0.944733112895607,
+      "eval_PRM F1 Neg": 0.8524590163934426,
+      "eval_PRM NPV": 0.8125,
+      "eval_PRM Precision": 0.9571428571428572,
+      "eval_PRM Recall": 0.9178082191780822,
+      "eval_PRM Specificty": 0.896551724137931,
+      "eval_loss": 0.18666215240955353,
+      "eval_runtime": 13.1834,
+      "eval_samples_per_second": 3.338,
+      "eval_steps_per_second": 0.152,
+      "step": 140
+    },
+    {
+      "epoch": 0.3176569980287243,
+      "grad_norm": 0.6373616256387727,
+      "learning_rate": 8.631848756323197e-05,
+      "loss": 0.2825,
+      "step": 141
+    },
+    {
+      "epoch": 0.3199098845395663,
+      "grad_norm": 0.4818071805883264,
+      "learning_rate": 8.604609894073584e-05,
+      "loss": 0.2287,
+      "step": 142
+    },
+    {
+      "epoch": 0.32216277105040836,
+      "grad_norm": 0.5000641724080277,
+      "learning_rate": 8.577146442236857e-05,
+      "loss": 0.2072,
+      "step": 143
+    },
+    {
+      "epoch": 0.32441565756125035,
+      "grad_norm": 0.52005357056285,
+      "learning_rate": 8.549460111956664e-05,
+      "loss": 0.2393,
+      "step": 144
+    },
+    {
+      "epoch": 0.3266685440720924,
+      "grad_norm": 0.4153652123369306,
+      "learning_rate": 8.521552628263362e-05,
+      "loss": 0.2106,
+      "step": 145
+    },
+    {
+      "epoch": 0.3266685440720924,
+      "eval_PRM Accuracy": 0.9117647058823529,
+      "eval_PRM F1": 0.9361702127659575,
+      "eval_PRM F1 AUC": 0.9175720358998584,
+      "eval_PRM F1 AUC (fixed)": 0.9418989135569201,
+      "eval_PRM F1 Neg": 0.8571428571428571,
+      "eval_PRM NPV": 0.7941176470588235,
+      "eval_PRM Precision": 0.9705882352941176,
+      "eval_PRM Recall": 0.9041095890410958,
+      "eval_PRM Specificty": 0.9310344827586207,
+      "eval_loss": 0.200439453125,
+      "eval_runtime": 13.5509,
+      "eval_samples_per_second": 3.247,
+      "eval_steps_per_second": 0.148,
+      "step": 145
+    },
+    {
+      "epoch": 0.3289214305829344,
+      "grad_norm": 0.3999793628562146,
+      "learning_rate": 8.493425729966534e-05,
+      "loss": 0.2022,
+      "step": 146
+    },
+    {
+      "epoch": 0.3311743170937764,
+      "grad_norm": 0.4167206740270849,
+      "learning_rate": 8.465081169546659e-05,
+      "loss": 0.1951,
+      "step": 147
+    },
+    {
+      "epoch": 0.3334272036046184,
+      "grad_norm": 0.34588504492910277,
+      "learning_rate": 8.436520713045922e-05,
+      "loss": 0.191,
+      "step": 148
+    },
+    {
+      "epoch": 0.33568009011546046,
+      "grad_norm": 0.41266699693634523,
+      "learning_rate": 8.40774613995817e-05,
+      "loss": 0.2203,
+      "step": 149
+    },
+    {
+      "epoch": 0.33793297662630245,
+      "grad_norm": 0.516522850184383,
+      "learning_rate": 8.378759243118044e-05,
+      "loss": 0.2474,
+      "step": 150
+    },
+    {
+      "epoch": 0.33793297662630245,
+      "eval_PRM Accuracy": 0.8921568627450981,
+      "eval_PRM F1": 0.920863309352518,
+      "eval_PRM F1 AUC": 0.9038734057628719,
+      "eval_PRM F1 AUC (fixed)": 0.9428436466698157,
+      "eval_PRM F1 Neg": 0.8307692307692308,
+      "eval_PRM NPV": 0.75,
+      "eval_PRM Precision": 0.9696969696969697,
+      "eval_PRM Recall": 0.8767123287671232,
+      "eval_PRM Specificty": 0.9310344827586207,
+      "eval_loss": 0.2106267809867859,
+      "eval_runtime": 13.9955,
+      "eval_samples_per_second": 3.144,
+      "eval_steps_per_second": 0.143,
+      "step": 150
+    },
+    {
+      "epoch": 0.34018586313714444,
+      "grad_norm": 0.5879682643613816,
+      "learning_rate": 8.349561828589277e-05,
+      "loss": 0.2187,
+      "step": 151
+    },
+    {
+      "epoch": 0.3424387496479865,
+      "grad_norm": 0.47234974519199624,
+      "learning_rate": 8.320155715552155e-05,
+      "loss": 0.2049,
+      "step": 152
+    },
+    {
+      "epoch": 0.3446916361588285,
+      "grad_norm": 0.49181617910201963,
+      "learning_rate": 8.290542736190188e-05,
+      "loss": 0.2305,
+      "step": 153
+    },
+    {
+      "epoch": 0.3469445226696705,
+      "grad_norm": 0.47902037883405957,
+      "learning_rate": 8.260724735575933e-05,
+      "loss": 0.2532,
+      "step": 154
+    },
+    {
+      "epoch": 0.3491974091805125,
+      "grad_norm": 0.7801602137596034,
+      "learning_rate": 8.230703571556048e-05,
+      "loss": 0.2377,
+      "step": 155
+    },
+    {
+      "epoch": 0.3491974091805125,
+      "eval_PRM Accuracy": 0.8921568627450981,
+      "eval_PRM F1": 0.9230769230769231,
+      "eval_PRM F1 AUC": 0.8830892772791686,
+      "eval_PRM F1 AUC (fixed)": 0.9374114312706661,
+      "eval_PRM F1 Neg": 0.819672131147541,
+      "eval_PRM NPV": 0.78125,
+      "eval_PRM Precision": 0.9428571428571428,
+      "eval_PRM Recall": 0.9041095890410958,
+      "eval_PRM Specificty": 0.8620689655172413,
+      "eval_loss": 0.20374645292758942,
+      "eval_runtime": 13.6085,
+      "eval_samples_per_second": 3.233,
+      "eval_steps_per_second": 0.147,
+      "step": 155
+    },
+    {
+      "epoch": 0.35145029569135455,
+      "grad_norm": 0.530927053953448,
+      "learning_rate": 8.200481114635536e-05,
+      "loss": 0.1959,
+      "step": 156
+    },
+    {
+      "epoch": 0.35370318220219654,
+      "grad_norm": 0.4256958095148454,
+      "learning_rate": 8.170059247861194e-05,
+      "loss": 0.2117,
+      "step": 157
+    },
+    {
+      "epoch": 0.3559560687130386,
+      "grad_norm": 0.5003646952580904,
+      "learning_rate": 8.139439866704293e-05,
+      "loss": 0.2486,
+      "step": 158
+    },
+    {
+      "epoch": 0.3582089552238806,
+      "grad_norm": 0.5472711553725297,
+      "learning_rate": 8.108624878942477e-05,
+      "loss": 0.2429,
+      "step": 159
+    },
+    {
+      "epoch": 0.3604618417347226,
+      "grad_norm": 0.6955196059766148,
+      "learning_rate": 8.077616204540897e-05,
+      "loss": 0.2411,
+      "step": 160
+    },
+    {
+      "epoch": 0.3604618417347226,
+      "eval_PRM Accuracy": 0.9019607843137255,
+      "eval_PRM F1": 0.927536231884058,
+      "eval_PRM F1 AUC": 0.9211147850732168,
+      "eval_PRM F1 AUC (fixed)": 0.9485120453471895,
+      "eval_PRM F1 Neg": 0.8484848484848485,
+      "eval_PRM NPV": 0.7567567567567568,
+      "eval_PRM Precision": 0.9846153846153847,
+      "eval_PRM Recall": 0.8767123287671232,
+      "eval_PRM Specificty": 0.9655172413793104,
+      "eval_loss": 0.2217462658882141,
+      "eval_runtime": 15.1621,
+      "eval_samples_per_second": 2.902,
+      "eval_steps_per_second": 0.132,
+      "step": 160
+    },
+    {
+      "epoch": 0.3627147282455646,
+      "grad_norm": 0.4788833467687291,
+      "learning_rate": 8.046415775532585e-05,
+      "loss": 0.2344,
+      "step": 161
+    },
+    {
+      "epoch": 0.36496761475640666,
+      "grad_norm": 0.6281810800097093,
+      "learning_rate": 8.015025535898073e-05,
+      "loss": 0.2092,
+      "step": 162
+    },
+    {
+      "epoch": 0.36722050126724864,
+      "grad_norm": 0.7164346569372413,
+      "learning_rate": 7.983447441444281e-05,
+      "loss": 0.2551,
+      "step": 163
+    },
+    {
+      "epoch": 0.3694733877780907,
+      "grad_norm": 0.6461732283942793,
+      "learning_rate": 7.951683459682641e-05,
+      "loss": 0.2144,
+      "step": 164
+    },
+    {
+      "epoch": 0.3717262742889327,
+      "grad_norm": 0.46720505917887667,
+      "learning_rate": 7.919735569706533e-05,
+      "loss": 0.2158,
+      "step": 165
+    },
+    {
+      "epoch": 0.3717262742889327,
+      "eval_PRM Accuracy": 0.9019607843137255,
+      "eval_PRM F1": 0.9295774647887324,
+      "eval_PRM F1 AUC": 0.9003306565895135,
+      "eval_PRM F1 AUC (fixed)": 0.9400094473311289,
+      "eval_PRM F1 Neg": 0.8387096774193549,
+      "eval_PRM NPV": 0.7878787878787878,
+      "eval_PRM Precision": 0.9565217391304348,
+      "eval_PRM Recall": 0.9041095890410958,
+      "eval_PRM Specificty": 0.896551724137931,
+      "eval_loss": 0.20969460904598236,
+      "eval_runtime": 14.0397,
+      "eval_samples_per_second": 3.134,
+      "eval_steps_per_second": 0.142,
+      "step": 165
+    },
+    {
+      "epoch": 0.3739791607997747,
+      "grad_norm": 0.46444325422897065,
+      "learning_rate": 7.887605762067945e-05,
+      "loss": 0.2287,
+      "step": 166
+    },
+    {
+      "epoch": 0.3762320473106167,
+      "grad_norm": 0.6618484436834431,
+      "learning_rate": 7.855296038653475e-05,
+      "loss": 0.2614,
+      "step": 167
+    },
+    {
+      "epoch": 0.37848493382145876,
+      "grad_norm": 0.5521406538265332,
+      "learning_rate": 7.82280841255959e-05,
+      "loss": 0.237,
+      "step": 168
+    },
+    {
+      "epoch": 0.38073782033230075,
+      "grad_norm": 0.8558424040519867,
+      "learning_rate": 7.790144907967201e-05,
+      "loss": 0.2644,
+      "step": 169
+    },
+    {
+      "epoch": 0.3829907068431428,
+      "grad_norm": 0.7446448973715851,
+      "learning_rate": 7.757307560015538e-05,
+      "loss": 0.2489,
+      "step": 170
+    },
+    {
+      "epoch": 0.3829907068431428,
+      "eval_PRM Accuracy": 0.9019607843137255,
+      "eval_PRM F1": 0.9285714285714286,
+      "eval_PRM F1 AUC": 0.9107227208313652,
+      "eval_PRM F1 AUC (fixed)": 0.9421350968351441,
+      "eval_PRM F1 Neg": 0.84375,
+      "eval_PRM NPV": 0.7714285714285715,
+      "eval_PRM Precision": 0.9701492537313433,
+      "eval_PRM Recall": 0.8904109589041096,
+      "eval_PRM Specificty": 0.9310344827586207,
+      "eval_loss": 0.21157559752464294,
+      "eval_runtime": 13.4182,
+      "eval_samples_per_second": 3.279,
+      "eval_steps_per_second": 0.149,
+      "step": 170
+    },
+    {
+      "epoch": 0.3852435933539848,
+      "grad_norm": 0.4888177325747004,
+      "learning_rate": 7.724298414675353e-05,
+      "loss": 0.229,
+      "step": 171
+    },
+    {
+      "epoch": 0.3874964798648268,
+      "grad_norm": 0.5422603899155514,
+      "learning_rate": 7.691119528621444e-05,
+      "loss": 0.2692,
+      "step": 172
+    },
+    {
+      "epoch": 0.3897493663756688,
+      "grad_norm": 0.47128581886009085,
+      "learning_rate": 7.657772969104508e-05,
+      "loss": 0.2483,
+      "step": 173
+    },
+    {
+      "epoch": 0.39200225288651086,
+      "grad_norm": 0.8044605229083803,
+      "learning_rate": 7.624260813822342e-05,
+      "loss": 0.2689,
+      "step": 174
+    },
+    {
+      "epoch": 0.39425513939735285,
+      "grad_norm": 0.625429806813994,
+      "learning_rate": 7.590585150790389e-05,
+      "loss": 0.2406,
+      "step": 175
+    },
+    {
+      "epoch": 0.39425513939735285,
+      "eval_PRM Accuracy": 0.8921568627450981,
+      "eval_PRM F1": 0.9197080291970803,
+      "eval_PRM F1 AUC": 0.9142654700047237,
+      "eval_PRM F1 AUC (fixed)": 0.9444969296173831,
+      "eval_PRM F1 Neg": 0.835820895522388,
+      "eval_PRM NPV": 0.7368421052631579,
+      "eval_PRM Precision": 0.984375,
+      "eval_PRM Recall": 0.863013698630137,
+      "eval_PRM Specificty": 0.9655172413793104,
+      "eval_loss": 0.23708273470401764,
+      "eval_runtime": 13.7387,
+      "eval_samples_per_second": 3.203,
+      "eval_steps_per_second": 0.146,
+      "step": 175
+    },
+    {
+      "epoch": 0.3965080259081949,
+      "grad_norm": 0.5110061992764956,
+      "learning_rate": 7.556748078211635e-05,
+      "loss": 0.1949,
+      "step": 176
+    },
+    {
+      "epoch": 0.3987609124190369,
+      "grad_norm": 0.5175175569524655,
+      "learning_rate": 7.522751704345887e-05,
+      "loss": 0.2213,
+      "step": 177
+    },
+    {
+      "epoch": 0.40101379892987893,
+      "grad_norm": 0.48396227801195096,
+      "learning_rate": 7.488598147378416e-05,
+      "loss": 0.265,
+      "step": 178
+    },
+    {
+      "epoch": 0.4032666854407209,
+      "grad_norm": 0.483357564587365,
+      "learning_rate": 7.454289535287968e-05,
+      "loss": 0.209,
+      "step": 179
+    },
+    {
+      "epoch": 0.40551957195156296,
+      "grad_norm": 0.8475261300761361,
+      "learning_rate": 7.419828005714194e-05,
+      "loss": 0.2038,
+      "step": 180
+    },
+    {
+      "epoch": 0.40551957195156296,
+      "eval_PRM Accuracy": 0.9215686274509803,
+      "eval_PRM F1": 0.9428571428571428,
+      "eval_PRM F1 AUC": 0.9348134152102031,
+      "eval_PRM F1 AUC (fixed)": 0.9489844119036372,
+      "eval_PRM F1 Neg": 0.875,
+      "eval_PRM NPV": 0.8,
+      "eval_PRM Precision": 0.9850746268656716,
+      "eval_PRM Recall": 0.9041095890410958,
+      "eval_PRM Specificty": 0.9655172413793104,
+      "eval_loss": 0.20637650787830353,
+      "eval_runtime": 13.3851,
+      "eval_samples_per_second": 3.287,
+      "eval_steps_per_second": 0.149,
+      "step": 180
+    },
+    {
+      "epoch": 0.40777245846240495,
+      "grad_norm": 0.5579587691556817,
+      "learning_rate": 7.385215705824449e-05,
+      "loss": 0.1914,
+      "step": 181
+    },
+    {
+      "epoch": 0.410025344973247,
+      "grad_norm": 0.447980237776592,
+      "learning_rate": 7.350454792180016e-05,
+      "loss": 0.2246,
+      "step": 182
+    },
+    {
+      "epoch": 0.412278231484089,
+      "grad_norm": 0.5085800154515963,
+      "learning_rate": 7.31554743060174e-05,
+      "loss": 0.2263,
+      "step": 183
+    },
+    {
+      "epoch": 0.41453111799493103,
+      "grad_norm": 0.46635672749023627,
+      "learning_rate": 7.280495796035079e-05,
+      "loss": 0.248,
+      "step": 184
+    },
+    {
+      "epoch": 0.416784004505773,
+      "grad_norm": 0.6330961325795686,
+      "learning_rate": 7.245302072414601e-05,
+      "loss": 0.2539,
+      "step": 185
+    },
+    {
+      "epoch": 0.416784004505773,
+      "eval_PRM Accuracy": 0.9215686274509803,
+      "eval_PRM F1": 0.9436619718309859,
+      "eval_PRM F1 AUC": 0.9244213509683514,
+      "eval_PRM F1 AUC (fixed)": 0.9404818138875768,
+      "eval_PRM F1 Neg": 0.8709677419354839,
+      "eval_PRM NPV": 0.8181818181818182,
+      "eval_PRM Precision": 0.9710144927536232,
+      "eval_PRM Recall": 0.9178082191780822,
+      "eval_PRM Specificty": 0.9310344827586207,
+      "eval_loss": 0.18641246855258942,
+      "eval_runtime": 13.2865,
+      "eval_samples_per_second": 3.312,
+      "eval_steps_per_second": 0.151,
+      "step": 185
+    },
+    {
+      "epoch": 0.41903689101661507,
+      "grad_norm": 0.9366945778304713,
+      "learning_rate": 7.209968452527896e-05,
+      "loss": 0.2601,
+      "step": 186
+    },
+    {
+      "epoch": 0.42128977752745705,
+      "grad_norm": 0.8142239310319512,
+      "learning_rate": 7.174497137878966e-05,
+      "loss": 0.2235,
+      "step": 187
+    },
+    {
+      "epoch": 0.42354266403829904,
+      "grad_norm": 0.5719949863544413,
+      "learning_rate": 7.138890338551048e-05,
+      "loss": 0.2206,
+      "step": 188
+    },
+    {
+      "epoch": 0.4257955505491411,
+      "grad_norm": 0.515710098136613,
+      "learning_rate": 7.103150273068921e-05,
+      "loss": 0.2064,
+      "step": 189
+    },
+    {
+      "epoch": 0.4280484370599831,
+      "grad_norm": 0.6738559302560749,
+      "learning_rate": 7.067279168260671e-05,
+      "loss": 0.2583,
+      "step": 190
+    },
+    {
+      "epoch": 0.4280484370599831,
+      "eval_PRM Accuracy": 0.8921568627450981,
+      "eval_PRM F1": 0.9197080291970803,
+      "eval_PRM F1 AUC": 0.9142654700047237,
+      "eval_PRM F1 AUC (fixed)": 0.9409541804440246,
+      "eval_PRM F1 Neg": 0.835820895522388,
+      "eval_PRM NPV": 0.7368421052631579,
+      "eval_PRM Precision": 0.984375,
+      "eval_PRM Recall": 0.863013698630137,
+      "eval_PRM Specificty": 0.9655172413793104,
+      "eval_loss": 0.21796764433383942,
+      "eval_runtime": 13.5517,
+      "eval_samples_per_second": 3.247,
+      "eval_steps_per_second": 0.148,
+      "step": 190
+    },
+    {
+      "epoch": 0.4303013235708251,
+      "grad_norm": 0.5252430551668598,
+      "learning_rate": 7.031279259118946e-05,
+      "loss": 0.2116,
+      "step": 191
+    },
+    {
+      "epoch": 0.4325542100816671,
+      "grad_norm": 0.7214832747996867,
+      "learning_rate": 6.995152788661705e-05,
+      "loss": 0.2425,
+      "step": 192
+    },
+    {
+      "epoch": 0.43480709659250916,
+      "grad_norm": 0.7077700765199836,
+      "learning_rate": 6.958902007792466e-05,
+      "loss": 0.1995,
+      "step": 193
+    },
+    {
+      "epoch": 0.43705998310335115,
+      "grad_norm": 0.47270417163576717,
+      "learning_rate": 6.922529175160054e-05,
+      "loss": 0.1868,
+      "step": 194
+    },
+    {
+      "epoch": 0.4393128696141932,
+      "grad_norm": 0.518184557735015,
+      "learning_rate": 6.886036557017881e-05,
+      "loss": 0.24,
+      "step": 195
+    },
+    {
+      "epoch": 0.4393128696141932,
+      "eval_PRM Accuracy": 0.9117647058823529,
+      "eval_PRM F1": 0.9361702127659575,
+      "eval_PRM F1 AUC": 0.9175720358998584,
+      "eval_PRM F1 AUC (fixed)": 0.9338686820973074,
+      "eval_PRM F1 Neg": 0.8571428571428571,
+      "eval_PRM NPV": 0.7941176470588235,
+      "eval_PRM Precision": 0.9705882352941176,
+      "eval_PRM Recall": 0.9041095890410958,
+      "eval_PRM Specificty": 0.9310344827586207,
+      "eval_loss": 0.2053888440132141,
+      "eval_runtime": 13.5267,
+      "eval_samples_per_second": 3.253,
+      "eval_steps_per_second": 0.148,
+      "step": 195
+    },
+    {
+      "epoch": 0.4415657561250352,
+      "grad_norm": 0.7604342184453511,
+      "learning_rate": 6.849426427082735e-05,
+      "loss": 0.2439,
+      "step": 196
+    },
+    {
+      "epoch": 0.4438186426358772,
+      "grad_norm": 0.4867951826238437,
+      "learning_rate": 6.812701066393124e-05,
+      "loss": 0.2052,
+      "step": 197
+    },
+    {
+      "epoch": 0.4460715291467192,
+      "grad_norm": 0.5973881366901407,
+      "learning_rate": 6.775862763167142e-05,
+      "loss": 0.2188,
+      "step": 198
+    },
+    {
+      "epoch": 0.44832441565756126,
+      "grad_norm": 0.5284594121286487,
+      "learning_rate": 6.738913812659912e-05,
+      "loss": 0.2128,
+      "step": 199
+    },
+    {
+      "epoch": 0.45057730216840325,
+      "grad_norm": 0.48926094393682273,
+      "learning_rate": 6.701856517020565e-05,
+      "loss": 0.1977,
+      "step": 200
+    },
+    {
+      "epoch": 0.45057730216840325,
+      "eval_PRM Accuracy": 0.8823529411764706,
+      "eval_PRM F1": 0.9130434782608695,
+      "eval_PRM F1 AUC": 0.8970240906943787,
+      "eval_PRM F1 AUC (fixed)": 0.9343410486537553,
+      "eval_PRM F1 Neg": 0.8181818181818182,
+      "eval_PRM NPV": 0.7297297297297297,
+      "eval_PRM Precision": 0.9692307692307692,
+      "eval_PRM Recall": 0.863013698630137,
+      "eval_PRM Specificty": 0.9310344827586207,
+      "eval_loss": 0.22230114042758942,
+      "eval_runtime": 13.9005,
+      "eval_samples_per_second": 3.165,
+      "eval_steps_per_second": 0.144,
+      "step": 200
+    },
+    {
+      "epoch": 0.4528301886792453,
+      "grad_norm": 0.5613489000008864,
+      "learning_rate": 6.664693185148807e-05,
+      "loss": 0.2684,
+      "step": 201
+    },
+    {
+      "epoch": 0.4550830751900873,
+      "grad_norm": 0.6209828102813096,
+      "learning_rate": 6.627426132551058e-05,
+      "loss": 0.1961,
+      "step": 202
+    },
+    {
+      "epoch": 0.4573359617009293,
+      "grad_norm": 0.5349079852503461,
+      "learning_rate": 6.590057681196191e-05,
+      "loss": 0.2288,
+      "step": 203
+    },
+    {
+      "epoch": 0.4595888482117713,
+      "grad_norm": 0.47163420268942535,
+      "learning_rate": 6.552590159370844e-05,
+      "loss": 0.229,
+      "step": 204
+    },
+    {
+      "epoch": 0.46184173472261336,
+      "grad_norm": 0.5926211569990463,
+      "learning_rate": 6.515025901534364e-05,
+      "loss": 0.1992,
+      "step": 205
+    },
+    {
+      "epoch": 0.46184173472261336,
+      "eval_PRM Accuracy": 0.8921568627450981,
+      "eval_PRM F1": 0.9197080291970803,
+      "eval_PRM F1 AUC": 0.9142654700047237,
+      "eval_PRM F1 AUC (fixed)": 0.9395370807746811,
+      "eval_PRM F1 Neg": 0.835820895522388,
+      "eval_PRM NPV": 0.7368421052631579,
+      "eval_PRM Precision": 0.984375,
+      "eval_PRM Recall": 0.863013698630137,
+      "eval_PRM Specificty": 0.9655172413793104,
+      "eval_loss": 0.22206254303455353,
+      "eval_runtime": 13.8895,
+      "eval_samples_per_second": 3.168,
+      "eval_steps_per_second": 0.144,
+      "step": 205
+    },
+    {
+      "epoch": 0.46409462123345535,
+      "grad_norm": 0.5009720327114148,
+      "learning_rate": 6.477367248173352e-05,
+      "loss": 0.2467,
+      "step": 206
+    },
+    {
+      "epoch": 0.4663475077442974,
+      "grad_norm": 0.5516233806255901,
+      "learning_rate": 6.439616545655834e-05,
+      "loss": 0.2045,
+      "step": 207
+    },
+    {
+      "epoch": 0.4686003942551394,
+      "grad_norm": 0.4561387944410809,
+      "learning_rate": 6.401776146085072e-05,
+      "loss": 0.2335,
+      "step": 208
+    },
+    {
+      "epoch": 0.47085328076598143,
+      "grad_norm": 0.5547658197457285,
+      "learning_rate": 6.363848407153016e-05,
+      "loss": 0.2394,
+      "step": 209
+    },
+    {
+      "epoch": 0.4731061672768234,
+      "grad_norm": 0.5529134735473807,
+      "learning_rate": 6.325835691993394e-05,
+      "loss": 0.2732,
+      "step": 210
+    },
+    {
+      "epoch": 0.4731061672768234,
+      "eval_PRM Accuracy": 0.9215686274509803,
+      "eval_PRM F1": 0.9428571428571428,
+      "eval_PRM F1 AUC": 0.9348134152102031,
+      "eval_PRM F1 AUC (fixed)": 0.9489844119036372,
+      "eval_PRM F1 Neg": 0.875,
+      "eval_PRM NPV": 0.8,
+      "eval_PRM Precision": 0.9850746268656716,
+      "eval_PRM Recall": 0.9041095890410958,
+      "eval_PRM Specificty": 0.9655172413793104,
+      "eval_loss": 0.19468550384044647,
+      "eval_runtime": 13.2471,
+      "eval_samples_per_second": 3.321,
+      "eval_steps_per_second": 0.151,
+      "step": 210
+    },
+    {
+      "epoch": 0.47535905378766546,
+      "grad_norm": 0.4806787541228142,
+      "learning_rate": 6.287740369034485e-05,
+      "loss": 0.258,
+      "step": 211
+    },
+    {
+      "epoch": 0.47761194029850745,
+      "grad_norm": 0.43534902359210464,
+      "learning_rate": 6.249564811851543e-05,
+      "loss": 0.2087,
+      "step": 212
+    },
+    {
+      "epoch": 0.4798648268093495,
+      "grad_norm": 0.3969727011884705,
+      "learning_rate": 6.211311399018916e-05,
+      "loss": 0.1791,
+      "step": 213
+    },
+    {
+      "epoch": 0.4821177133201915,
+      "grad_norm": 0.6371736645343234,
+      "learning_rate": 6.172982513961845e-05,
+      "loss": 0.2444,
+      "step": 214
+    },
+    {
+      "epoch": 0.48437059983103353,
+      "grad_norm": 0.5365711814260565,
+      "learning_rate": 6.13458054480795e-05,
+      "loss": 0.2074,
+      "step": 215
+    },
+    {
+      "epoch": 0.48437059983103353,
+      "eval_PRM Accuracy": 0.9411764705882353,
+      "eval_PRM F1": 0.9577464788732394,
+      "eval_PRM F1 AUC": 0.9485120453471895,
+      "eval_PRM F1 AUC (fixed)": 0.949456778460085,
+      "eval_PRM F1 Neg": 0.9032258064516129,
+      "eval_PRM NPV": 0.8484848484848485,
+      "eval_PRM Precision": 0.9855072463768116,
+      "eval_PRM Recall": 0.9315068493150684,
+      "eval_PRM Specificty": 0.9655172413793104,
+      "eval_loss": 0.17953768372535706,
+      "eval_runtime": 13.593,
+      "eval_samples_per_second": 3.237,
+      "eval_steps_per_second": 0.147,
+      "step": 215
+    },
+    {
+      "epoch": 0.4866234863418755,
+      "grad_norm": 0.9204323201053622,
+      "learning_rate": 6.096107884238458e-05,
+      "loss": 0.2872,
+      "step": 216
+    },
+    {
+      "epoch": 0.48887637285271757,
+      "grad_norm": 0.9773149939692177,
+      "learning_rate": 6.0575669293390954e-05,
+      "loss": 0.25,
+      "step": 217
+    },
+    {
+      "epoch": 0.49112925936355956,
+      "grad_norm": 0.4333073062627389,
+      "learning_rate": 6.0189600814507604e-05,
+      "loss": 0.1853,
+      "step": 218
+    },
+    {
+      "epoch": 0.4933821458744016,
+      "grad_norm": 0.4477078952772944,
+      "learning_rate": 5.980289746019892e-05,
+      "loss": 0.1727,
+      "step": 219
+    },
+    {
+      "epoch": 0.4956350323852436,
+      "grad_norm": 0.6945887032793969,
+      "learning_rate": 5.941558332448589e-05,
+      "loss": 0.2161,
+      "step": 220
+    },
+    {
+      "epoch": 0.4956350323852436,
+      "eval_PRM Accuracy": 0.9019607843137255,
+      "eval_PRM F1": 0.927536231884058,
+      "eval_PRM F1 AUC": 0.9211147850732168,
+      "eval_PRM F1 AUC (fixed)": 0.954416627302787,
+      "eval_PRM F1 Neg": 0.8484848484848485,
+      "eval_PRM NPV": 0.7567567567567568,
+      "eval_PRM Precision": 0.9846153846153847,
+      "eval_PRM Recall": 0.8767123287671232,
+      "eval_PRM Specificty": 0.9655172413793104,
+      "eval_loss": 0.2091730237007141,
+      "eval_runtime": 14.4464,
+      "eval_samples_per_second": 3.046,
+      "eval_steps_per_second": 0.138,
+      "step": 220
+    },
+    {
+      "epoch": 0.49788791889608564,
+      "grad_norm": 0.729346795310851,
+      "learning_rate": 5.9027682539445104e-05,
+      "loss": 0.23,
+      "step": 221
+    },
+    {
+      "epoch": 0.5001408054069276,
+      "grad_norm": 0.4992934536711299,
+      "learning_rate": 5.863921927370498e-05,
+      "loss": 0.2147,
+      "step": 222
+    },
+    {
+      "epoch": 0.5023936919177696,
+      "grad_norm": 0.5256840527700389,
+      "learning_rate": 5.8250217730939973e-05,
+      "loss": 0.2377,
+      "step": 223
+    },
+    {
+      "epoch": 0.5046465784286117,
+      "grad_norm": 0.6737501061725462,
+      "learning_rate": 5.786070214836254e-05,
+      "loss": 0.2221,
+      "step": 224
+    },
+    {
+      "epoch": 0.5068994649394537,
+      "grad_norm": 0.7332465699903149,
+      "learning_rate": 5.747069679521305e-05,
+      "loss": 0.2013,
+      "step": 225
+    },
+    {
+      "epoch": 0.5068994649394537,
+      "eval_PRM Accuracy": 0.9019607843137255,
+      "eval_PRM F1": 0.927536231884058,
+      "eval_PRM F1 AUC": 0.9211147850732168,
+      "eval_PRM F1 AUC (fixed)": 0.9539442607463391,
+      "eval_PRM F1 Neg": 0.8484848484848485,
+      "eval_PRM NPV": 0.7567567567567568,
+      "eval_PRM Precision": 0.9846153846153847,
+      "eval_PRM Recall": 0.8767123287671232,
+      "eval_PRM Specificty": 0.9655172413793104,
+      "eval_loss": 0.20087780058383942,
+      "eval_runtime": 13.7145,
+      "eval_samples_per_second": 3.208,
+      "eval_steps_per_second": 0.146,
+      "step": 225
+    },
+    {
+      "epoch": 0.5091523514502957,
+      "grad_norm": 0.5291541122785641,
+      "learning_rate": 5.708022597124758e-05,
+      "loss": 0.2179,
+      "step": 226
+    },
+    {
+      "epoch": 0.5114052379611377,
+      "grad_norm": 0.4745324289486805,
+      "learning_rate": 5.668931400522396e-05,
+      "loss": 0.2159,
+      "step": 227
+    },
+    {
+      "epoch": 0.5136581244719797,
+      "grad_norm": 0.46645781871782865,
+      "learning_rate": 5.629798525338589e-05,
+      "loss": 0.1793,
+      "step": 228
+    },
+    {
+      "epoch": 0.5159110109828218,
+      "grad_norm": 0.47131831880693087,
+      "learning_rate": 5.5906264097945407e-05,
+      "loss": 0.1864,
+      "step": 229
+    },
+    {
+      "epoch": 0.5181638974936638,
+      "grad_norm": 0.7817091735051518,
+      "learning_rate": 5.551417494556376e-05,
+      "loss": 0.278,
+      "step": 230
+    },
+    {
+      "epoch": 0.5181638974936638,
+      "eval_PRM Accuracy": 0.9215686274509803,
+      "eval_PRM F1": 0.9436619718309859,
+      "eval_PRM F1 AUC": 0.9244213509683514,
+      "eval_PRM F1 AUC (fixed)": 0.9499291450165328,
+      "eval_PRM F1 Neg": 0.8709677419354839,
+      "eval_PRM NPV": 0.8181818181818182,
+      "eval_PRM Precision": 0.9710144927536232,
+      "eval_PRM Recall": 0.9178082191780822,
+      "eval_PRM Specificty": 0.9310344827586207,
+      "eval_loss": 0.18656782805919647,
+      "eval_runtime": 13.2612,
+      "eval_samples_per_second": 3.318,
+      "eval_steps_per_second": 0.151,
+      "step": 230
+    },
+    {
+      "epoch": 0.5204167840045058,
+      "grad_norm": 1.1720426822472467,
+      "learning_rate": 5.5121742225830665e-05,
+      "loss": 0.2711,
+      "step": 231
+    },
+    {
+      "epoch": 0.5226696705153477,
+      "grad_norm": 0.5446586207415837,
+      "learning_rate": 5.472899038974225e-05,
+      "loss": 0.222,
+      "step": 232
+    },
+    {
+      "epoch": 0.5249225570261898,
+      "grad_norm": 0.4314421350427008,
+      "learning_rate": 5.433594390817756e-05,
+      "loss": 0.2179,
+      "step": 233
+    },
+    {
+      "epoch": 0.5271754435370318,
+      "grad_norm": 0.7462368025520845,
+      "learning_rate": 5.3942627270373826e-05,
+      "loss": 0.2299,
+      "step": 234
+    },
+    {
+      "epoch": 0.5294283300478738,
+      "grad_norm": 0.6752632622707911,
+      "learning_rate": 5.35490649824008e-05,
+      "loss": 0.2042,
+      "step": 235
+    },
+    {
+      "epoch": 0.5294283300478738,
+      "eval_PRM Accuracy": 0.8823529411764706,
+      "eval_PRM F1": 0.9117647058823529,
+      "eval_PRM F1 AUC": 0.9074161549362305,
+      "eval_PRM F1 AUC (fixed)": 0.9565422768068021,
+      "eval_PRM F1 Neg": 0.8235294117647058,
+      "eval_PRM NPV": 0.717948717948718,
+      "eval_PRM Precision": 0.9841269841269841,
+      "eval_PRM Recall": 0.8493150684931506,
+      "eval_PRM Specificty": 0.9655172413793104,
+      "eval_loss": 0.2100774645805359,
+      "eval_runtime": 13.9732,
+      "eval_samples_per_second": 3.149,
+      "eval_steps_per_second": 0.143,
+      "step": 235
+    },
+    {
+      "epoch": 0.5316812165587158,
+      "grad_norm": 0.523695499678202,
+      "learning_rate": 5.315528156563367e-05,
+      "loss": 0.24,
+      "step": 236
+    },
+    {
+      "epoch": 0.5339341030695579,
+      "grad_norm": 0.523654564484976,
+      "learning_rate": 5.276130155522541e-05,
+      "loss": 0.233,
+      "step": 237
+    },
+    {
+      "epoch": 0.5361869895803999,
+      "grad_norm": 0.553830499237879,
+      "learning_rate": 5.236714949857791e-05,
+      "loss": 0.2105,
+      "step": 238
+    },
+    {
+      "epoch": 0.5384398760912419,
+      "grad_norm": 0.5110175758820199,
+      "learning_rate": 5.1972849953812644e-05,
+      "loss": 0.2206,
+      "step": 239
+    },
+    {
+      "epoch": 0.5406927626020839,
+      "grad_norm": 0.4547437022077122,
+      "learning_rate": 5.157842748824053e-05,
+      "loss": 0.2298,
+      "step": 240
+    },
+    {
+      "epoch": 0.5406927626020839,
+      "eval_PRM Accuracy": 0.8921568627450981,
+      "eval_PRM F1": 0.9197080291970803,
+      "eval_PRM F1 AUC": 0.9142654700047237,
+      "eval_PRM F1 AUC (fixed)": 0.9560699102503544,
+      "eval_PRM F1 Neg": 0.835820895522388,
+      "eval_PRM NPV": 0.7368421052631579,
+      "eval_PRM Precision": 0.984375,
+      "eval_PRM Recall": 0.863013698630137,
+      "eval_PRM Specificty": 0.9655172413793104,
+      "eval_loss": 0.20505037903785706,
+      "eval_runtime": 13.51,
+      "eval_samples_per_second": 3.257,
+      "eval_steps_per_second": 0.148,
+      "step": 240
+    },
+    {
+      "epoch": 0.542945649112926,
+      "grad_norm": 0.43784881514361,
+      "learning_rate": 5.1183906676831197e-05,
+      "loss": 0.1832,
+      "step": 241
+    },
+    {
+      "epoch": 0.545198535623768,
+      "grad_norm": 0.42941642437378513,
+      "learning_rate": 5.078931210068185e-05,
+      "loss": 0.1992,
+      "step": 242
+    },
+    {
+      "epoch": 0.54745142213461,
+      "grad_norm": 0.5917920509097113,
+      "learning_rate": 5.039466834548568e-05,
+      "loss": 0.2592,
+      "step": 243
+    },
+    {
+      "epoch": 0.549704308645452,
+      "grad_norm": 0.6091575723228944,
+      "learning_rate": 5e-05,
+      "loss": 0.2237,
+      "step": 244
+    },
+    {
+      "epoch": 0.551957195156294,
+      "grad_norm": 0.5317791532098205,
+      "learning_rate": 4.960533165451435e-05,
+      "loss": 0.1887,
+      "step": 245
+    },
+    {
+      "epoch": 0.551957195156294,
+      "eval_PRM Accuracy": 0.9117647058823529,
+      "eval_PRM F1": 0.935251798561151,
+      "eval_PRM F1 AUC": 0.92796410014171,
+      "eval_PRM F1 AUC (fixed)": 0.9499291450165328,
+      "eval_PRM F1 Neg": 0.8615384615384616,
+      "eval_PRM NPV": 0.7777777777777778,
+      "eval_PRM Precision": 0.9848484848484849,
+      "eval_PRM Recall": 0.8904109589041096,
+      "eval_PRM Specificty": 0.9655172413793104,
+      "eval_loss": 0.19755969941616058,
+      "eval_runtime": 14.0484,
+      "eval_samples_per_second": 3.132,
+      "eval_steps_per_second": 0.142,
+      "step": 245
+    },
+    {
+      "epoch": 0.554210081667136,
+      "grad_norm": 0.4482288314131815,
+      "learning_rate": 4.921068789931816e-05,
+      "loss": 0.2118,
+      "step": 246
+    },
+    {
+      "epoch": 0.556462968177978,
+      "grad_norm": 0.4931272491652408,
+      "learning_rate": 4.8816093323168815e-05,
+      "loss": 0.2629,
+      "step": 247
+    },
+    {
+      "epoch": 0.55871585468882,
+      "grad_norm": 0.6552129298212358,
+      "learning_rate": 4.842157251175947e-05,
+      "loss": 0.2515,
+      "step": 248
+    },
+    {
+      "epoch": 0.5609687411996621,
+      "grad_norm": 0.6922293962594674,
+      "learning_rate": 4.802715004618737e-05,
+      "loss": 0.226,
+      "step": 249
+    },
+    {
+      "epoch": 0.5632216277105041,
+      "grad_norm": 0.6806132163625348,
+      "learning_rate": 4.763285050142211e-05,
+      "loss": 0.2529,
+      "step": 250
+    },
+    {
+      "epoch": 0.5632216277105041,
+      "eval_PRM Accuracy": 0.9215686274509803,
+      "eval_PRM F1": 0.9436619718309859,
+      "eval_PRM F1 AUC": 0.9244213509683514,
+      "eval_PRM F1 AUC (fixed)": 0.9437883797827114,
+      "eval_PRM F1 Neg": 0.8709677419354839,
+      "eval_PRM NPV": 0.8181818181818182,
+      "eval_PRM Precision": 0.9710144927536232,
+      "eval_PRM Recall": 0.9178082191780822,
+      "eval_PRM Specificty": 0.9310344827586207,
+      "eval_loss": 0.19232177734375,
+      "eval_runtime": 13.9342,
+      "eval_samples_per_second": 3.158,
+      "eval_steps_per_second": 0.144,
+      "step": 250
+    },
+    {
+      "epoch": 0.5654745142213461,
+      "grad_norm": 0.5476128660158797,
+      "learning_rate": 4.7238698444774595e-05,
+      "loss": 0.2464,
+      "step": 251
+    },
+    {
+      "epoch": 0.5677274007321881,
+      "grad_norm": 0.4523932392398643,
+      "learning_rate": 4.6844718434366334e-05,
+      "loss": 0.1842,
+      "step": 252
+    },
+    {
+      "epoch": 0.5699802872430302,
+      "grad_norm": 0.48085849509311573,
+      "learning_rate": 4.64509350175992e-05,
+      "loss": 0.1862,
+      "step": 253
+    },
+    {
+      "epoch": 0.5722331737538722,
+      "grad_norm": 0.585088429199167,
+      "learning_rate": 4.605737272962618e-05,
+      "loss": 0.2183,
+      "step": 254
+    },
+    {
+      "epoch": 0.5744860602647142,
+      "grad_norm": 0.5016541780324376,
+      "learning_rate": 4.566405609182247e-05,
+      "loss": 0.2241,
+      "step": 255
+    },
+    {
+      "epoch": 0.5744860602647142,
+      "eval_PRM Accuracy": 0.8921568627450981,
+      "eval_PRM F1": 0.9197080291970803,
+      "eval_PRM F1 AUC": 0.9142654700047237,
+      "eval_PRM F1 AUC (fixed)": 0.9496929617383089,
+      "eval_PRM F1 Neg": 0.835820895522388,
+      "eval_PRM NPV": 0.7368421052631579,
+      "eval_PRM Precision": 0.984375,
+      "eval_PRM Recall": 0.863013698630137,
+      "eval_PRM Specificty": 0.9655172413793104,
+      "eval_loss": 0.20318603515625,
+      "eval_runtime": 13.5865,
+      "eval_samples_per_second": 3.239,
+      "eval_steps_per_second": 0.147,
+      "step": 255
+    },
+    {
+      "epoch": 0.5767389467755561,
+      "grad_norm": 0.5528907034034132,
+      "learning_rate": 4.527100961025776e-05,
+      "loss": 0.1995,
+      "step": 256
+    },
+    {
+      "epoch": 0.5789918332863982,
+      "grad_norm": 0.5846574553370062,
+      "learning_rate": 4.4878257774169346e-05,
+      "loss": 0.2274,
+      "step": 257
+    },
+    {
+      "epoch": 0.5812447197972402,
+      "grad_norm": 0.7161290279165864,
+      "learning_rate": 4.448582505443625e-05,
+      "loss": 0.227,
+      "step": 258
+    },
+    {
+      "epoch": 0.5834976063080822,
+      "grad_norm": 0.4830986221715104,
+      "learning_rate": 4.4093735902054605e-05,
+      "loss": 0.202,
+      "step": 259
+    },
+    {
+      "epoch": 0.5857504928189242,
+      "grad_norm": 0.42724659835780576,
+      "learning_rate": 4.3702014746614136e-05,
+      "loss": 0.1916,
+      "step": 260
+    },
+    {
+      "epoch": 0.5857504928189242,
+      "eval_PRM Accuracy": 0.9019607843137255,
+      "eval_PRM F1": 0.927536231884058,
+      "eval_PRM F1 AUC": 0.9211147850732168,
+      "eval_PRM F1 AUC (fixed)": 0.9499291450165328,
+      "eval_PRM F1 Neg": 0.8484848484848485,
+      "eval_PRM NPV": 0.7567567567567568,
+      "eval_PRM Precision": 0.9846153846153847,
+      "eval_PRM Recall": 0.8767123287671232,
+      "eval_PRM Specificty": 0.9655172413793104,
+      "eval_loss": 0.19813676178455353,
+      "eval_runtime": 13.8512,
+      "eval_samples_per_second": 3.177,
+      "eval_steps_per_second": 0.144,
+      "step": 260
+    },
+    {
+      "epoch": 0.5880033793297663,
+      "grad_norm": 0.7560849559943468,
+      "learning_rate": 4.331068599477605e-05,
+      "loss": 0.2197,
+      "step": 261
+    },
+    {
+      "epoch": 0.5902562658406083,
+      "grad_norm": 0.5758870137919648,
+      "learning_rate": 4.2919774028752436e-05,
+      "loss": 0.2229,
+      "step": 262
+    },
+    {
+      "epoch": 0.5925091523514503,
+      "grad_norm": 0.5122326326893871,
+      "learning_rate": 4.2529303204786953e-05,
+      "loss": 0.1795,
+      "step": 263
+    },
+    {
+      "epoch": 0.5947620388622923,
+      "grad_norm": 0.4682380634045031,
+      "learning_rate": 4.213929785163747e-05,
+      "loss": 0.1981,
+      "step": 264
+    },
+    {
+      "epoch": 0.5970149253731343,
+      "grad_norm": 0.5421765450138537,
+      "learning_rate": 4.1749782269060045e-05,
+      "loss": 0.21,
+      "step": 265
+    },
+    {
+      "epoch": 0.5970149253731343,
+      "eval_PRM Accuracy": 0.9019607843137255,
+      "eval_PRM F1": 0.927536231884058,
+      "eval_PRM F1 AUC": 0.9211147850732168,
+      "eval_PRM F1 AUC (fixed)": 0.9513462446858763,
+      "eval_PRM F1 Neg": 0.8484848484848485,
+      "eval_PRM NPV": 0.7567567567567568,
+      "eval_PRM Precision": 0.9846153846153847,
+      "eval_PRM Recall": 0.8767123287671232,
+      "eval_PRM Specificty": 0.9655172413793104,
+      "eval_loss": 0.20195423066616058,
+      "eval_runtime": 13.9529,
+      "eval_samples_per_second": 3.153,
+      "eval_steps_per_second": 0.143,
+      "step": 265
+    },
+    {
+      "epoch": 0.5992678118839764,
+      "grad_norm": 0.6282518975260062,
+      "learning_rate": 4.136078072629503e-05,
+      "loss": 0.2031,
+      "step": 266
+    },
+    {
+      "epoch": 0.6015206983948184,
+      "grad_norm": 0.5094372555440069,
+      "learning_rate": 4.097231746055491e-05,
+      "loss": 0.1993,
+      "step": 267
+    },
+    {
+      "epoch": 0.6037735849056604,
+      "grad_norm": 0.6963186816198292,
+      "learning_rate": 4.05844166755141e-05,
+      "loss": 0.2483,
+      "step": 268
+    },
+    {
+      "epoch": 0.6060264714165023,
+      "grad_norm": 0.7469941221381555,
+      "learning_rate": 4.01971025398011e-05,
+      "loss": 0.256,
+      "step": 269
+    },
+    {
+      "epoch": 0.6082793579273444,
+      "grad_norm": 0.5213114849120921,
+      "learning_rate": 3.98103991854924e-05,
+      "loss": 0.2351,
+      "step": 270
+    },
+    {
+      "epoch": 0.6082793579273444,
+      "eval_PRM Accuracy": 0.9019607843137255,
+      "eval_PRM F1": 0.927536231884058,
+      "eval_PRM F1 AUC": 0.9211147850732168,
+      "eval_PRM F1 AUC (fixed)": 0.9468587623996221,
+      "eval_PRM F1 Neg": 0.8484848484848485,
+      "eval_PRM NPV": 0.7567567567567568,
+      "eval_PRM Precision": 0.9846153846153847,
+      "eval_PRM Recall": 0.8767123287671232,
+      "eval_PRM Specificty": 0.9655172413793104,
+      "eval_loss": 0.1986139416694641,
+      "eval_runtime": 13.8117,
+      "eval_samples_per_second": 3.186,
+      "eval_steps_per_second": 0.145,
+      "step": 270
+    },
+    {
+      "epoch": 0.6105322444381864,
+      "grad_norm": 0.5299071100110285,
+      "learning_rate": 3.942433070660905e-05,
+      "loss": 0.1833,
+      "step": 271
+    },
+    {
+      "epoch": 0.6127851309490284,
+      "grad_norm": 0.6791536413163082,
+      "learning_rate": 3.9038921157615444e-05,
+      "loss": 0.2544,
+      "step": 272
+    },
+    {
+      "epoch": 0.6150380174598704,
+      "grad_norm": 0.4782693861194819,
+      "learning_rate": 3.8654194551920485e-05,
+      "loss": 0.1861,
+      "step": 273
+    },
+    {
+      "epoch": 0.6172909039707125,
+      "grad_norm": 0.6634859985494999,
+      "learning_rate": 3.827017486038157e-05,
+      "loss": 0.2249,
+      "step": 274
+    },
+    {
+      "epoch": 0.6195437904815545,
+      "grad_norm": 0.5312330005059178,
+      "learning_rate": 3.788688600981085e-05,
+      "loss": 0.1987,
+      "step": 275
+    },
+    {
+      "epoch": 0.6195437904815545,
+      "eval_PRM Accuracy": 0.9019607843137255,
+      "eval_PRM F1": 0.927536231884058,
+      "eval_PRM F1 AUC": 0.9211147850732168,
+      "eval_PRM F1 AUC (fixed)": 0.9444969296173831,
+      "eval_PRM F1 Neg": 0.8484848484848485,
+      "eval_PRM NPV": 0.7567567567567568,
+      "eval_PRM Precision": 0.9846153846153847,
+      "eval_PRM Recall": 0.8767123287671232,
+      "eval_PRM Specificty": 0.9655172413793104,
+      "eval_loss": 0.20028409361839294,
+      "eval_runtime": 13.7853,
+      "eval_samples_per_second": 3.192,
+      "eval_steps_per_second": 0.145,
+      "step": 275
+    },
+    {
+      "epoch": 0.6217966769923965,
+      "grad_norm": 0.6694178440036449,
+      "learning_rate": 3.750435188148459e-05,
+      "loss": 0.2609,
+      "step": 276
+    },
+    {
+      "epoch": 0.6240495635032385,
+      "grad_norm": 0.4836220242731773,
+      "learning_rate": 3.712259630965518e-05,
+      "loss": 0.195,
+      "step": 277
+    },
+    {
+      "epoch": 0.6263024500140806,
+      "grad_norm": 0.6029027749100382,
+      "learning_rate": 3.6741643080066065e-05,
+      "loss": 0.2365,
+      "step": 278
+    },
+    {
+      "epoch": 0.6285553365249226,
+      "grad_norm": 0.5258861381182393,
+      "learning_rate": 3.636151592846985e-05,
+      "loss": 0.1849,
+      "step": 279
+    },
+    {
+      "epoch": 0.6308082230357646,
+      "grad_norm": 0.6502615021743258,
+      "learning_rate": 3.5982238539149285e-05,
+      "loss": 0.2225,
+      "step": 280
+    },
+    {
+      "epoch": 0.6308082230357646,
+      "eval_PRM Accuracy": 0.9117647058823529,
+      "eval_PRM F1": 0.935251798561151,
+      "eval_PRM F1 AUC": 0.92796410014171,
+      "eval_PRM F1 AUC (fixed)": 0.9442607463391591,
+      "eval_PRM F1 Neg": 0.8615384615384616,
+      "eval_PRM NPV": 0.7777777777777778,
+      "eval_PRM Precision": 0.9848484848484849,
+      "eval_PRM Recall": 0.8904109589041096,
+      "eval_PRM Specificty": 0.9655172413793104,
+      "eval_loss": 0.19981245696544647,
+      "eval_runtime": 14.1782,
+      "eval_samples_per_second": 3.103,
+      "eval_steps_per_second": 0.141,
+      "step": 280
+    },
+    {
+      "epoch": 0.6330611095466065,
+      "grad_norm": 0.4637027759447034,
+      "learning_rate": 3.560383454344168e-05,
+      "loss": 0.1782,
+      "step": 281
+    },
+    {
+      "epoch": 0.6353139960574486,
+      "grad_norm": 0.4635631060680255,
+      "learning_rate": 3.522632751826651e-05,
+      "loss": 0.2146,
+      "step": 282
+    },
+    {
+      "epoch": 0.6375668825682906,
+      "grad_norm": 0.5266764589506505,
+      "learning_rate": 3.484974098465636e-05,
+      "loss": 0.2298,
+      "step": 283
+    },
+    {
+      "epoch": 0.6398197690791326,
+      "grad_norm": 0.47885834524709486,
+      "learning_rate": 3.447409840629156e-05,
+      "loss": 0.1874,
+      "step": 284
+    },
+    {
+      "epoch": 0.6420726555899746,
+      "grad_norm": 0.5158469531279536,
+      "learning_rate": 3.409942318803809e-05,
+      "loss": 0.2113,
+      "step": 285
+    },
+    {
+      "epoch": 0.6420726555899746,
+      "eval_PRM Accuracy": 0.9117647058823529,
+      "eval_PRM F1": 0.9370629370629371,
+      "eval_PRM F1 AUC": 0.9071799716580067,
+      "eval_PRM F1 AUC (fixed)": 0.9456778460085026,
+      "eval_PRM F1 Neg": 0.8524590163934426,
+      "eval_PRM NPV": 0.8125,
+      "eval_PRM Precision": 0.9571428571428572,
+      "eval_PRM Recall": 0.9178082191780822,
+      "eval_PRM Specificty": 0.896551724137931,
+      "eval_loss": 0.191650390625,
+      "eval_runtime": 13.7288,
+      "eval_samples_per_second": 3.205,
+      "eval_steps_per_second": 0.146,
+      "step": 285
+    },
+    {
+      "epoch": 0.6443255421008167,
+      "grad_norm": 0.48726069151172285,
+      "learning_rate": 3.3725738674489414e-05,
+      "loss": 0.1947,
+      "step": 286
+    },
+    {
+      "epoch": 0.6465784286116587,
+      "grad_norm": 0.6386387022941161,
+      "learning_rate": 3.335306814851196e-05,
+      "loss": 0.2098,
+      "step": 287
+    },
+    {
+      "epoch": 0.6488313151225007,
+      "grad_norm": 0.448700282205355,
+      "learning_rate": 3.298143482979436e-05,
+      "loss": 0.205,
+      "step": 288
+    },
+    {
+      "epoch": 0.6510842016333427,
+      "grad_norm": 0.675744485504258,
+      "learning_rate": 3.261086187340088e-05,
+      "loss": 0.2247,
+      "step": 289
+    },
+    {
+      "epoch": 0.6533370881441848,
+      "grad_norm": 0.5323997595627029,
+      "learning_rate": 3.224137236832859e-05,
+      "loss": 0.2216,
+      "step": 290
+    },
+    {
+      "epoch": 0.6533370881441848,
+      "eval_PRM Accuracy": 0.9019607843137255,
+      "eval_PRM F1": 0.9295774647887324,
+      "eval_PRM F1 AUC": 0.9003306565895135,
+      "eval_PRM F1 AUC (fixed)": 0.9468587623996221,
+      "eval_PRM F1 Neg": 0.8387096774193549,
+      "eval_PRM NPV": 0.7878787878787878,
+      "eval_PRM Precision": 0.9565217391304348,
+      "eval_PRM Recall": 0.9041095890410958,
+      "eval_PRM Specificty": 0.896551724137931,
+      "eval_loss": 0.1923828125,
+      "eval_runtime": 13.7883,
+      "eval_samples_per_second": 3.191,
+      "eval_steps_per_second": 0.145,
+      "step": 290
+    },
+    {
+      "epoch": 0.6555899746550268,
+      "grad_norm": 0.5663852015110561,
+      "learning_rate": 3.187298933606878e-05,
+      "loss": 0.2309,
+      "step": 291
+    },
+    {
+      "epoch": 0.6578428611658688,
+      "grad_norm": 0.6500973190347736,
+      "learning_rate": 3.150573572917267e-05,
+      "loss": 0.2236,
+      "step": 292
+    },
+    {
+      "epoch": 0.6600957476767108,
+      "grad_norm": 0.5547406940276081,
+      "learning_rate": 3.11396344298212e-05,
+      "loss": 0.2263,
+      "step": 293
+    },
+    {
+      "epoch": 0.6623486341875529,
+      "grad_norm": 0.42623036026867445,
+      "learning_rate": 3.077470824839947e-05,
+      "loss": 0.2045,
+      "step": 294
+    },
+    {
+      "epoch": 0.6646015206983948,
+      "grad_norm": 0.6015828802588874,
+      "learning_rate": 3.0410979922075343e-05,
+      "loss": 0.2501,
+      "step": 295
+    },
+    {
+      "epoch": 0.6646015206983948,
+      "eval_PRM Accuracy": 0.9117647058823529,
+      "eval_PRM F1": 0.9361702127659575,
+      "eval_PRM F1 AUC": 0.9175720358998584,
+      "eval_PRM F1 AUC (fixed)": 0.9499291450165328,
+      "eval_PRM F1 Neg": 0.8571428571428571,
+      "eval_PRM NPV": 0.7941176470588235,
+      "eval_PRM Precision": 0.9705882352941176,
+      "eval_PRM Recall": 0.9041095890410958,
+      "eval_PRM Specificty": 0.9310344827586207,
+      "eval_loss": 0.19617809355258942,
+      "eval_runtime": 14.4185,
+      "eval_samples_per_second": 3.052,
+      "eval_steps_per_second": 0.139,
+      "step": 295
+    },
+    {
+      "epoch": 0.6668544072092368,
+      "grad_norm": 0.5006156846819563,
+      "learning_rate": 3.004847211338295e-05,
+      "loss": 0.2042,
+      "step": 296
+    },
+    {
+      "epoch": 0.6691072937200788,
+      "grad_norm": 0.47412715112789594,
+      "learning_rate": 2.9687207408810557e-05,
+      "loss": 0.1507,
+      "step": 297
+    },
+    {
+      "epoch": 0.6713601802309209,
+      "grad_norm": 0.6307222335265025,
+      "learning_rate": 2.9327208317393303e-05,
+      "loss": 0.2027,
+      "step": 298
+    },
+    {
+      "epoch": 0.6736130667417629,
+      "grad_norm": 0.49278537545848705,
+      "learning_rate": 2.8968497269310803e-05,
+      "loss": 0.1767,
+      "step": 299
+    },
+    {
+      "epoch": 0.6758659532526049,
+      "grad_norm": 0.6169582670968694,
+      "learning_rate": 2.8611096614489518e-05,
+      "loss": 0.2362,
+      "step": 300
+    },
+    {
+      "epoch": 0.6758659532526049,
+      "eval_PRM Accuracy": 0.9117647058823529,
+      "eval_PRM F1": 0.935251798561151,
+      "eval_PRM F1 AUC": 0.92796410014171,
+      "eval_PRM F1 AUC (fixed)": 0.9555975436939065,
+      "eval_PRM F1 Neg": 0.8615384615384616,
+      "eval_PRM NPV": 0.7777777777777778,
+      "eval_PRM Precision": 0.9848484848484849,
+      "eval_PRM Recall": 0.8904109589041096,
+      "eval_PRM Specificty": 0.9655172413793104,
+      "eval_loss": 0.19659423828125,
+      "eval_runtime": 13.8687,
+      "eval_samples_per_second": 3.173,
+      "eval_steps_per_second": 0.144,
+      "step": 300
+    },
+    {
+      "epoch": 0.6781188397634469,
+      "grad_norm": 0.5014884232062446,
+      "learning_rate": 2.8255028621210355e-05,
+      "loss": 0.1954,
+      "step": 301
+    },
+    {
+      "epoch": 0.6803717262742889,
+      "grad_norm": 0.5383380178030245,
+      "learning_rate": 2.790031547472105e-05,
+      "loss": 0.2403,
+      "step": 302
+    },
+    {
+      "epoch": 0.682624612785131,
+      "grad_norm": 0.547598969464602,
+      "learning_rate": 2.754697927585399e-05,
+      "loss": 0.2201,
+      "step": 303
+    },
+    {
+      "epoch": 0.684877499295973,
+      "grad_norm": 0.7115982418557794,
+      "learning_rate": 2.71950420396492e-05,
+      "loss": 0.1859,
+      "step": 304
+    },
+    {
+      "epoch": 0.687130385806815,
+      "grad_norm": 0.5042783320674656,
+      "learning_rate": 2.6844525693982613e-05,
+      "loss": 0.2129,
+      "step": 305
+    },
+    {
+      "epoch": 0.687130385806815,
+      "eval_PRM Accuracy": 0.9117647058823529,
+      "eval_PRM F1": 0.935251798561151,
+      "eval_PRM F1 AUC": 0.92796410014171,
+      "eval_PRM F1 AUC (fixed)": 0.9541804440245631,
+      "eval_PRM F1 Neg": 0.8615384615384616,
+      "eval_PRM NPV": 0.7777777777777778,
+      "eval_PRM Precision": 0.9848484848484849,
+      "eval_PRM Recall": 0.8904109589041096,
+      "eval_PRM Specificty": 0.9655172413793104,
+      "eval_loss": 0.19580911099910736,
+      "eval_runtime": 13.8878,
+      "eval_samples_per_second": 3.168,
+      "eval_steps_per_second": 0.144,
+      "step": 305
+    },
+    {
+      "epoch": 0.689383272317657,
+      "grad_norm": 0.5382152312300029,
+      "learning_rate": 2.6495452078199863e-05,
+      "loss": 0.242,
+      "step": 306
+    },
+    {
+      "epoch": 0.691636158828499,
+      "grad_norm": 0.6496013284313801,
+      "learning_rate": 2.614784294175554e-05,
+      "loss": 0.2495,
+      "step": 307
+    },
+    {
+      "epoch": 0.693889045339341,
+      "grad_norm": 0.5936584477513502,
+      "learning_rate": 2.5801719942858065e-05,
+      "loss": 0.2005,
+      "step": 308
+    },
+    {
+      "epoch": 0.696141931850183,
+      "grad_norm": 0.5519137767112613,
+      "learning_rate": 2.5457104647120322e-05,
+      "loss": 0.1892,
+      "step": 309
+    },
+    {
+      "epoch": 0.698394818361025,
+      "grad_norm": 0.8236087287521814,
+      "learning_rate": 2.5114018526215844e-05,
+      "loss": 0.186,
+      "step": 310
+    },
+    {
+      "epoch": 0.698394818361025,
+      "eval_PRM Accuracy": 0.9215686274509803,
+      "eval_PRM F1": 0.9444444444444444,
+      "eval_PRM F1 AUC": 0.9140292867264997,
+      "eval_PRM F1 AUC (fixed)": 0.9537080774681153,
+      "eval_PRM F1 Neg": 0.8666666666666667,
+      "eval_PRM NPV": 0.8387096774193549,
+      "eval_PRM Precision": 0.9577464788732394,
+      "eval_PRM Recall": 0.9315068493150684,
+      "eval_PRM Specificty": 0.896551724137931,
+      "eval_loss": 0.18574662506580353,
+      "eval_runtime": 13.99,
+      "eval_samples_per_second": 3.145,
+      "eval_steps_per_second": 0.143,
+      "step": 310
+    },
+    {
+      "epoch": 0.7006477048718671,
+      "grad_norm": 0.5813467454089482,
+      "learning_rate": 2.4772482956541132e-05,
+      "loss": 0.2115,
+      "step": 311
+    },
+    {
+      "epoch": 0.7029005913827091,
+      "grad_norm": 0.6009267695934843,
+      "learning_rate": 2.4432519217883676e-05,
+      "loss": 0.1801,
+      "step": 312
+    },
+    {
+      "epoch": 0.7051534778935511,
+      "grad_norm": 0.5048420747090402,
+      "learning_rate": 2.4094148492096125e-05,
+      "loss": 0.2035,
+      "step": 313
+    },
+    {
+      "epoch": 0.7074063644043931,
+      "grad_norm": 0.5588460169769177,
+      "learning_rate": 2.3757391861776585e-05,
+      "loss": 0.2133,
+      "step": 314
+    },
+    {
+      "epoch": 0.7096592509152352,
+      "grad_norm": 0.6432652778449629,
+      "learning_rate": 2.3422270308954934e-05,
+      "loss": 0.2212,
+      "step": 315
+    },
+    {
+      "epoch": 0.7096592509152352,
+      "eval_PRM Accuracy": 0.9215686274509803,
+      "eval_PRM F1": 0.9444444444444444,
+      "eval_PRM F1 AUC": 0.9140292867264997,
+      "eval_PRM F1 AUC (fixed)": 0.9532357109116675,
+      "eval_PRM F1 Neg": 0.8666666666666667,
+      "eval_PRM NPV": 0.8387096774193549,
+      "eval_PRM Precision": 0.9577464788732394,
+      "eval_PRM Recall": 0.9315068493150684,
+      "eval_PRM Specificty": 0.896551724137931,
+      "eval_loss": 0.18171276152133942,
+      "eval_runtime": 13.9573,
+      "eval_samples_per_second": 3.152,
+      "eval_steps_per_second": 0.143,
+      "step": 315
+    },
+    {
+      "epoch": 0.7119121374260772,
+      "grad_norm": 0.5135880317683275,
+      "learning_rate": 2.3088804713785584e-05,
+      "loss": 0.2362,
+      "step": 316
+    },
+    {
+      "epoch": 0.7141650239369192,
+      "grad_norm": 0.5831693391617544,
+      "learning_rate": 2.2757015853246493e-05,
+      "loss": 0.1896,
+      "step": 317
+    },
+    {
+      "epoch": 0.7164179104477612,
+      "grad_norm": 0.5426067800893314,
+      "learning_rate": 2.242692439984463e-05,
+      "loss": 0.1927,
+      "step": 318
+    },
+    {
+      "epoch": 0.7186707969586033,
+      "grad_norm": 0.7152499695408646,
+      "learning_rate": 2.2098550920327998e-05,
+      "loss": 0.2085,
+      "step": 319
+    },
+    {
+      "epoch": 0.7209236834694452,
+      "grad_norm": 0.5290046985659368,
+      "learning_rate": 2.177191587440409e-05,
+      "loss": 0.2028,
+      "step": 320
+    },
+    {
+      "epoch": 0.7209236834694452,
+      "eval_PRM Accuracy": 0.9215686274509803,
+      "eval_PRM F1": 0.9444444444444444,
+      "eval_PRM F1 AUC": 0.9140292867264997,
+      "eval_PRM F1 AUC (fixed)": 0.9534718941898914,
+      "eval_PRM F1 Neg": 0.8666666666666667,
+      "eval_PRM NPV": 0.8387096774193549,
+      "eval_PRM Precision": 0.9577464788732394,
+      "eval_PRM Recall": 0.9315068493150684,
+      "eval_PRM Specificty": 0.896551724137931,
+      "eval_loss": 0.18126331269741058,
+      "eval_runtime": 13.4403,
+      "eval_samples_per_second": 3.274,
+      "eval_steps_per_second": 0.149,
+      "step": 320
+    },
+    {
+      "epoch": 0.7231765699802872,
+      "grad_norm": 0.512205285152136,
+      "learning_rate": 2.1447039613465265e-05,
+      "loss": 0.1944,
+      "step": 321
+    },
+    {
+      "epoch": 0.7254294564911292,
+      "grad_norm": 0.5442104287004161,
+      "learning_rate": 2.1123942379320576e-05,
+      "loss": 0.2244,
+      "step": 322
+    },
+    {
+      "epoch": 0.7276823430019713,
+      "grad_norm": 0.6781105534897924,
+      "learning_rate": 2.0802644302934683e-05,
+      "loss": 0.2371,
+      "step": 323
+    },
+    {
+      "epoch": 0.7299352295128133,
+      "grad_norm": 0.6056266618431347,
+      "learning_rate": 2.0483165403173583e-05,
+      "loss": 0.2299,
+      "step": 324
+    },
+    {
+      "epoch": 0.7321881160236553,
+      "grad_norm": 0.5541848885548317,
+      "learning_rate": 2.0165525585557204e-05,
+      "loss": 0.1866,
+      "step": 325
+    },
+    {
+      "epoch": 0.7321881160236553,
+      "eval_PRM Accuracy": 0.9313725490196079,
+      "eval_PRM F1": 0.951048951048951,
+      "eval_PRM F1 AUC": 0.9312706660368445,
+      "eval_PRM F1 AUC (fixed)": 0.9560699102503544,
+      "eval_PRM F1 Neg": 0.8852459016393442,
+      "eval_PRM NPV": 0.84375,
+      "eval_PRM Precision": 0.9714285714285714,
+      "eval_PRM Recall": 0.9315068493150684,
+      "eval_PRM Specificty": 0.9310344827586207,
+      "eval_loss": 0.18489490449428558,
+      "eval_runtime": 14.0637,
+      "eval_samples_per_second": 3.129,
+      "eval_steps_per_second": 0.142,
+      "step": 325
+    },
+    {
+      "epoch": 0.7344410025344973,
+      "grad_norm": 0.5885759838737207,
+      "learning_rate": 1.984974464101928e-05,
+      "loss": 0.2083,
+      "step": 326
+    },
+    {
+      "epoch": 0.7366938890453394,
+      "grad_norm": 0.5083836359096051,
+      "learning_rate": 1.953584224467418e-05,
+      "loss": 0.2092,
+      "step": 327
+    },
+    {
+      "epoch": 0.7389467755561814,
+      "grad_norm": 0.4914027780184102,
+      "learning_rate": 1.9223837954591046e-05,
+      "loss": 0.1924,
+      "step": 328
+    },
+    {
+      "epoch": 0.7411996620670234,
+      "grad_norm": 0.7036700096171996,
+      "learning_rate": 1.8913751210575248e-05,
+      "loss": 0.2044,
+      "step": 329
+    },
+    {
+      "epoch": 0.7434525485778654,
+      "grad_norm": 0.6854744259593741,
+      "learning_rate": 1.8605601332957077e-05,
+      "loss": 0.2064,
+      "step": 330
+    },
+    {
+      "epoch": 0.7434525485778654,
+      "eval_PRM Accuracy": 0.9313725490196079,
+      "eval_PRM F1": 0.950354609929078,
+      "eval_PRM F1 AUC": 0.9416627302786964,
+      "eval_PRM F1 AUC (fixed)": 0.9560699102503544,
+      "eval_PRM F1 Neg": 0.8888888888888888,
+      "eval_PRM NPV": 0.8235294117647058,
+      "eval_PRM Precision": 0.9852941176470589,
+      "eval_PRM Recall": 0.9178082191780822,
+      "eval_PRM Specificty": 0.9655172413793104,
+      "eval_loss": 0.1925409436225891,
+      "eval_runtime": 13.6461,
+      "eval_samples_per_second": 3.224,
+      "eval_steps_per_second": 0.147,
+      "step": 330
+    },
+    {
+      "epoch": 0.7457054350887075,
+      "grad_norm": 0.5714986548606203,
+      "learning_rate": 1.8299407521388067e-05,
+      "loss": 0.1877,
+      "step": 331
+    },
+    {
+      "epoch": 0.7479583215995494,
+      "grad_norm": 0.6978832773189276,
+      "learning_rate": 1.7995188853644646e-05,
+      "loss": 0.1596,
+      "step": 332
+    },
+    {
+      "epoch": 0.7502112081103914,
+      "grad_norm": 0.7363848410413948,
+      "learning_rate": 1.7692964284439505e-05,
+      "loss": 0.2451,
+      "step": 333
+    },
+    {
+      "epoch": 0.7524640946212334,
+      "grad_norm": 0.6317346327408855,
+      "learning_rate": 1.739275264424067e-05,
+      "loss": 0.1963,
+      "step": 334
+    },
+    {
+      "epoch": 0.7547169811320755,
+      "grad_norm": 0.6518689920383659,
+      "learning_rate": 1.7094572638098123e-05,
+      "loss": 0.2209,
+      "step": 335
+    },
+    {
+      "epoch": 0.7547169811320755,
+      "eval_PRM Accuracy": 0.9313725490196079,
+      "eval_PRM F1": 0.950354609929078,
+      "eval_PRM F1 AUC": 0.9416627302786964,
+      "eval_PRM F1 AUC (fixed)": 0.9546528105810108,
+      "eval_PRM F1 Neg": 0.8888888888888888,
+      "eval_PRM NPV": 0.8235294117647058,
+      "eval_PRM Precision": 0.9852941176470589,
+      "eval_PRM Recall": 0.9178082191780822,
+      "eval_PRM Specificty": 0.9655172413793104,
+      "eval_loss": 0.19454123079776764,
+      "eval_runtime": 13.5495,
+      "eval_samples_per_second": 3.247,
+      "eval_steps_per_second": 0.148,
+      "step": 335
+    },
+    {
+      "epoch": 0.7569698676429175,
+      "grad_norm": 0.4820066289865694,
+      "learning_rate": 1.6798442844478445e-05,
+      "loss": 0.2075,
+      "step": 336
+    },
+    {
+      "epoch": 0.7592227541537595,
+      "grad_norm": 0.7007645771561103,
+      "learning_rate": 1.6504381714107252e-05,
+      "loss": 0.1716,
+      "step": 337
+    },
+    {
+      "epoch": 0.7614756406646015,
+      "grad_norm": 0.48388003929818296,
+      "learning_rate": 1.6212407568819565e-05,
+      "loss": 0.2062,
+      "step": 338
+    },
+    {
+      "epoch": 0.7637285271754435,
+      "grad_norm": 0.6949948450463587,
+      "learning_rate": 1.5922538600418318e-05,
+      "loss": 0.1951,
+      "step": 339
+    },
+    {
+      "epoch": 0.7659814136862856,
+      "grad_norm": 1.0485720812606718,
+      "learning_rate": 1.563479286954078e-05,
+      "loss": 0.2315,
+      "step": 340
+    },
+    {
+      "epoch": 0.7659814136862856,
+      "eval_PRM Accuracy": 0.9313725490196079,
+      "eval_PRM F1": 0.951048951048951,
+      "eval_PRM F1 AUC": 0.9312706660368445,
+      "eval_PRM F1 AUC (fixed)": 0.9546528105810108,
+      "eval_PRM F1 Neg": 0.8852459016393442,
+      "eval_PRM NPV": 0.84375,
+      "eval_PRM Precision": 0.9714285714285714,
+      "eval_PRM Recall": 0.9315068493150684,
+      "eval_PRM Specificty": 0.9310344827586207,
+      "eval_loss": 0.1870061755180359,
+      "eval_runtime": 13.5772,
+      "eval_samples_per_second": 3.241,
+      "eval_steps_per_second": 0.147,
+      "step": 340
+    },
+    {
+      "epoch": 0.7682343001971276,
+      "grad_norm": 0.5959500226110623,
+      "learning_rate": 1.5349188304533413e-05,
+      "loss": 0.2019,
+      "step": 341
+    },
+    {
+      "epoch": 0.7704871867079696,
+      "grad_norm": 0.86970981661299,
+      "learning_rate": 1.5065742700334678e-05,
+      "loss": 0.2555,
+      "step": 342
+    },
+    {
+      "epoch": 0.7727400732188116,
+      "grad_norm": 0.7538872510546275,
+      "learning_rate": 1.4784473717366387e-05,
+      "loss": 0.2252,
+      "step": 343
+    },
+    {
+      "epoch": 0.7749929597296537,
+      "grad_norm": 0.5315683205778033,
+      "learning_rate": 1.4505398880433369e-05,
+      "loss": 0.1951,
+      "step": 344
+    },
+    {
+      "epoch": 0.7772458462404956,
+      "grad_norm": 0.8023613367625589,
+      "learning_rate": 1.4228535577631442e-05,
+      "loss": 0.2652,
+      "step": 345
+    },
+    {
+      "epoch": 0.7772458462404956,
+      "eval_PRM Accuracy": 0.9215686274509803,
+      "eval_PRM F1": 0.9444444444444444,
+      "eval_PRM F1 AUC": 0.9140292867264997,
+      "eval_PRM F1 AUC (fixed)": 0.9529995276334435,
+      "eval_PRM F1 Neg": 0.8666666666666667,
+      "eval_PRM NPV": 0.8387096774193549,
+      "eval_PRM Precision": 0.9577464788732394,
+      "eval_PRM Recall": 0.9315068493150684,
+      "eval_PRM Specificty": 0.896551724137931,
+      "eval_loss": 0.186492919921875,
+      "eval_runtime": 13.7076,
+      "eval_samples_per_second": 3.21,
+      "eval_steps_per_second": 0.146,
+      "step": 345
+    },
+    {
+      "epoch": 0.7794987327513376,
+      "grad_norm": 0.491916768908664,
+      "learning_rate": 1.3953901059264191e-05,
+      "loss": 0.1623,
+      "step": 346
+    },
+    {
+      "epoch": 0.7817516192621796,
+      "grad_norm": 0.5214284942340848,
+      "learning_rate": 1.3681512436768045e-05,
+      "loss": 0.1873,
+      "step": 347
+    },
+    {
+      "epoch": 0.7840045057730217,
+      "grad_norm": 0.46870784283580846,
+      "learning_rate": 1.3411386681646164e-05,
+      "loss": 0.159,
+      "step": 348
+    },
+    {
+      "epoch": 0.7862573922838637,
+      "grad_norm": 0.6313358283656494,
+      "learning_rate": 1.314354062441106e-05,
+      "loss": 0.2126,
+      "step": 349
+    },
+    {
+      "epoch": 0.7885102787947057,
+      "grad_norm": 0.8984563269548353,
+      "learning_rate": 1.287799095353584e-05,
+      "loss": 0.2637,
+      "step": 350
+    },
+    {
+      "epoch": 0.7885102787947057,
+      "eval_PRM Accuracy": 0.9313725490196079,
+      "eval_PRM F1": 0.951048951048951,
+      "eval_PRM F1 AUC": 0.9312706660368445,
+      "eval_PRM F1 AUC (fixed)": 0.9525271610769958,
+      "eval_PRM F1 Neg": 0.8852459016393442,
+      "eval_PRM NPV": 0.84375,
+      "eval_PRM Precision": 0.9714285714285714,
+      "eval_PRM Recall": 0.9315068493150684,
+      "eval_PRM Specificty": 0.9310344827586207,
+      "eval_loss": 0.18983320891857147,
+      "eval_runtime": 13.7786,
+      "eval_samples_per_second": 3.193,
+      "eval_steps_per_second": 0.145,
+      "step": 350
+    },
+    {
+      "epoch": 0.7907631653055477,
+      "grad_norm": 0.609115567198061,
+      "learning_rate": 1.2614754214414548e-05,
+      "loss": 0.2352,
+      "step": 351
+    },
+    {
+      "epoch": 0.7930160518163898,
+      "grad_norm": 0.6150640225070533,
+      "learning_rate": 1.2353846808331154e-05,
+      "loss": 0.2179,
+      "step": 352
+    },
+    {
+      "epoch": 0.7952689383272318,
+      "grad_norm": 0.659891256639334,
+      "learning_rate": 1.2095284991437733e-05,
+      "loss": 0.2824,
+      "step": 353
+    },
+    {
+      "epoch": 0.7975218248380738,
+      "grad_norm": 0.44663082105982294,
+      "learning_rate": 1.1839084873741584e-05,
+      "loss": 0.1998,
+      "step": 354
+    },
+    {
+      "epoch": 0.7997747113489158,
+      "grad_norm": 0.6940983615549994,
+      "learning_rate": 1.1585262418101467e-05,
+      "loss": 0.237,
+      "step": 355
+    },
+    {
+      "epoch": 0.7997747113489158,
+      "eval_PRM Accuracy": 0.9215686274509803,
+      "eval_PRM F1": 0.9436619718309859,
+      "eval_PRM F1 AUC": 0.9244213509683514,
+      "eval_PRM F1 AUC (fixed)": 0.9518186112423241,
+      "eval_PRM F1 Neg": 0.8709677419354839,
+      "eval_PRM NPV": 0.8181818181818182,
+      "eval_PRM Precision": 0.9710144927536232,
+      "eval_PRM Recall": 0.9178082191780822,
+      "eval_PRM Specificty": 0.9310344827586207,
+      "eval_loss": 0.19561490416526794,
+      "eval_runtime": 14.1563,
+      "eval_samples_per_second": 3.108,
+      "eval_steps_per_second": 0.141,
+      "step": 355
+    },
+    {
+      "epoch": 0.8020275978597579,
+      "grad_norm": 0.6421619660458147,
+      "learning_rate": 1.1333833439233055e-05,
+      "loss": 0.2286,
+      "step": 356
+    },
+    {
+      "epoch": 0.8042804843705998,
+      "grad_norm": 0.5380599892879142,
+      "learning_rate": 1.1084813602723515e-05,
+      "loss": 0.2027,
+      "step": 357
+    },
+    {
+      "epoch": 0.8065333708814418,
+      "grad_norm": 0.5767562377940715,
+      "learning_rate": 1.083821842405548e-05,
+      "loss": 0.2289,
+      "step": 358
+    },
+    {
+      "epoch": 0.8087862573922838,
+      "grad_norm": 0.5104292695562842,
+      "learning_rate": 1.0594063267640386e-05,
+      "loss": 0.2195,
+      "step": 359
+    },
+    {
+      "epoch": 0.8110391439031259,
+      "grad_norm": 0.4734409116884158,
+      "learning_rate": 1.0352363345861065e-05,
+      "loss": 0.1956,
+      "step": 360
+    },
+    {
+      "epoch": 0.8110391439031259,
+      "eval_PRM Accuracy": 0.9313725490196079,
+      "eval_PRM F1": 0.950354609929078,
+      "eval_PRM F1 AUC": 0.9416627302786964,
+      "eval_PRM F1 AUC (fixed)": 0.9513462446858763,
+      "eval_PRM F1 Neg": 0.8888888888888888,
+      "eval_PRM NPV": 0.8235294117647058,
+      "eval_PRM Precision": 0.9852941176470589,
+      "eval_PRM Recall": 0.9178082191780822,
+      "eval_PRM Specificty": 0.9655172413793104,
+      "eval_loss": 0.20095548033714294,
+      "eval_runtime": 13.3882,
+      "eval_samples_per_second": 3.286,
+      "eval_steps_per_second": 0.149,
+      "step": 360
+    },
+    {
+      "epoch": 0.8132920304139679,
+      "grad_norm": 0.4788121219116529,
+      "learning_rate": 1.0113133718124035e-05,
+      "loss": 0.1895,
+      "step": 361
+    },
+    {
+      "epoch": 0.8155449169248099,
+      "grad_norm": 0.5678010674218776,
+      "learning_rate": 9.876389289921106e-06,
+      "loss": 0.2086,
+      "step": 362
+    },
+    {
+      "epoch": 0.8177978034356519,
+      "grad_norm": 0.5219541280987532,
+      "learning_rate": 9.642144811900739e-06,
+      "loss": 0.2023,
+      "step": 363
+    },
+    {
+      "epoch": 0.820050689946494,
+      "grad_norm": 0.518399942023751,
+      "learning_rate": 9.410414878948975e-06,
+      "loss": 0.1654,
+      "step": 364
+    },
+    {
+      "epoch": 0.822303576457336,
+      "grad_norm": 0.6493618047994629,
+      "learning_rate": 9.181213929280046e-06,
+      "loss": 0.2379,
+      "step": 365
+    },
+    {
+      "epoch": 0.822303576457336,
+      "eval_PRM Accuracy": 0.9215686274509803,
+      "eval_PRM F1": 0.9428571428571428,
+      "eval_PRM F1 AUC": 0.9348134152102031,
+      "eval_PRM F1 AUC (fixed)": 0.9522909777987718,
+      "eval_PRM F1 Neg": 0.875,
+      "eval_PRM NPV": 0.8,
+      "eval_PRM Precision": 0.9850746268656716,
+      "eval_PRM Recall": 0.9041095890410958,
+      "eval_PRM Specificty": 0.9655172413793104,
+      "eval_loss": 0.2026727795600891,
+      "eval_runtime": 13.7197,
+      "eval_samples_per_second": 3.207,
+      "eval_steps_per_second": 0.146,
+      "step": 365
+    },
+    {
+      "epoch": 0.824556462968178,
+      "grad_norm": 0.5421377555305549,
+      "learning_rate": 8.954556243536877e-06,
+      "loss": 0.1975,
+      "step": 366
+    },
+    {
+      "epoch": 0.82680934947902,
+      "grad_norm": 0.5909787966208271,
+      "learning_rate": 8.7304559439012e-06,
+      "loss": 0.2295,
+      "step": 367
+    },
+    {
+      "epoch": 0.8290622359898621,
+      "grad_norm": 0.5835723114917231,
+      "learning_rate": 8.508926993213712e-06,
+      "loss": 0.217,
+      "step": 368
+    },
+    {
+      "epoch": 0.831315122500704,
+      "grad_norm": 0.6445638431894471,
+      "learning_rate": 8.28998319410413e-06,
+      "loss": 0.1761,
+      "step": 369
+    },
+    {
+      "epoch": 0.833568009011546,
+      "grad_norm": 0.5632550224105065,
+      "learning_rate": 8.073638188131128e-06,
+      "loss": 0.2119,
+      "step": 370
+    },
+    {
+      "epoch": 0.833568009011546,
+      "eval_PRM Accuracy": 0.9313725490196079,
+      "eval_PRM F1": 0.950354609929078,
+      "eval_PRM F1 AUC": 0.9416627302786964,
+      "eval_PRM F1 AUC (fixed)": 0.9499291450165329,
+      "eval_PRM F1 Neg": 0.8888888888888888,
+      "eval_PRM NPV": 0.8235294117647058,
+      "eval_PRM Precision": 0.9852941176470589,
+      "eval_PRM Recall": 0.9178082191780822,
+      "eval_PRM Specificty": 0.9655172413793104,
+      "eval_loss": 0.2026922106742859,
+      "eval_runtime": 14.9008,
+      "eval_samples_per_second": 2.953,
+      "eval_steps_per_second": 0.134,
+      "step": 370
+    },
+    {
+      "epoch": 0.835820895522388,
+      "grad_norm": 0.5887185521363927,
+      "learning_rate": 7.859905454932471e-06,
+      "loss": 0.1762,
+      "step": 371
+    },
+    {
+      "epoch": 0.8380737820332301,
+      "grad_norm": 0.781628792019172,
+      "learning_rate": 7.648798311385058e-06,
+      "loss": 0.2056,
+      "step": 372
+    },
+    {
+      "epoch": 0.8403266685440721,
+      "grad_norm": 0.5018239661013479,
+      "learning_rate": 7.440329910775273e-06,
+      "loss": 0.1722,
+      "step": 373
+    },
+    {
+      "epoch": 0.8425795550549141,
+      "grad_norm": 0.6355070300132148,
+      "learning_rate": 7.234513241979418e-06,
+      "loss": 0.1936,
+      "step": 374
+    },
+    {
+      "epoch": 0.8448324415657561,
+      "grad_norm": 0.5672281367427731,
+      "learning_rate": 7.031361128654401e-06,
+      "loss": 0.2032,
+      "step": 375
+    },
+    {
+      "epoch": 0.8448324415657561,
+      "eval_PRM Accuracy": 0.9215686274509803,
+      "eval_PRM F1": 0.9436619718309859,
+      "eval_PRM F1 AUC": 0.9244213509683514,
+      "eval_PRM F1 AUC (fixed)": 0.9489844119036372,
+      "eval_PRM F1 Neg": 0.8709677419354839,
+      "eval_PRM NPV": 0.8181818181818182,
+      "eval_PRM Precision": 0.9710144927536232,
+      "eval_PRM Recall": 0.9178082191780822,
+      "eval_PRM Specificty": 0.9310344827586207,
+      "eval_loss": 0.20007601380348206,
+      "eval_runtime": 13.4996,
+      "eval_samples_per_second": 3.259,
+      "eval_steps_per_second": 0.148,
+      "step": 375
+    },
+    {
+      "epoch": 0.8470853280765981,
+      "grad_norm": 0.6650570871238042,
+      "learning_rate": 6.830886228438837e-06,
+      "loss": 0.2312,
+      "step": 376
+    },
+    {
+      "epoch": 0.8493382145874402,
+      "grad_norm": 0.46766442784605805,
+      "learning_rate": 6.633101032164274e-06,
+      "loss": 0.1903,
+      "step": 377
+    },
+    {
+      "epoch": 0.8515911010982822,
+      "grad_norm": 0.6485034468573413,
+      "learning_rate": 6.4380178630770225e-06,
+      "loss": 0.255,
+      "step": 378
+    },
+    {
+      "epoch": 0.8538439876091242,
+      "grad_norm": 0.6225375400350412,
+      "learning_rate": 6.2456488760703205e-06,
+      "loss": 0.1804,
+      "step": 379
+    },
+    {
+      "epoch": 0.8560968741199662,
+      "grad_norm": 0.6049035743861589,
+      "learning_rate": 6.056006056926977e-06,
+      "loss": 0.2422,
+      "step": 380
+    },
+    {
+      "epoch": 0.8560968741199662,
+      "eval_PRM Accuracy": 0.9215686274509803,
+      "eval_PRM F1": 0.9436619718309859,
+      "eval_PRM F1 AUC": 0.9244213509683514,
+      "eval_PRM F1 AUC (fixed)": 0.9478034955125179,
+      "eval_PRM F1 Neg": 0.8709677419354839,
+      "eval_PRM NPV": 0.8181818181818182,
+      "eval_PRM Precision": 0.9710144927536232,
+      "eval_PRM Recall": 0.9178082191780822,
+      "eval_PRM Specificty": 0.9310344827586207,
+      "eval_loss": 0.19895519316196442,
+      "eval_runtime": 13.8947,
+      "eval_samples_per_second": 3.167,
+      "eval_steps_per_second": 0.144,
+      "step": 380
+    },
+    {
+      "epoch": 0.8583497606308083,
+      "grad_norm": 0.6437970091989257,
+      "learning_rate": 5.869101221572654e-06,
+      "loss": 0.2485,
+      "step": 381
+    },
+    {
+      "epoch": 0.8606026471416502,
+      "grad_norm": 0.5811548501865113,
+      "learning_rate": 5.6849460153395706e-06,
+      "loss": 0.1868,
+      "step": 382
+    },
+    {
+      "epoch": 0.8628555336524922,
+      "grad_norm": 0.47017811461422065,
+      "learning_rate": 5.5035519122409895e-06,
+      "loss": 0.1963,
+      "step": 383
+    },
+    {
+      "epoch": 0.8651084201633342,
+      "grad_norm": 0.45581906110225645,
+      "learning_rate": 5.324930214256302e-06,
+      "loss": 0.1822,
+      "step": 384
+    },
+    {
+      "epoch": 0.8673613066741763,
+      "grad_norm": 0.8034154127263667,
+      "learning_rate": 5.149092050626825e-06,
+      "loss": 0.2829,
+      "step": 385
+    },
+    {
+      "epoch": 0.8673613066741763,
+      "eval_PRM Accuracy": 0.9215686274509803,
+      "eval_PRM F1": 0.9436619718309859,
+      "eval_PRM F1 AUC": 0.9244213509683514,
+      "eval_PRM F1 AUC (fixed)": 0.9501653282947569,
+      "eval_PRM F1 Neg": 0.8709677419354839,
+      "eval_PRM NPV": 0.8181818181818182,
+      "eval_PRM Precision": 0.9710144927536232,
+      "eval_PRM Recall": 0.9178082191780822,
+      "eval_PRM Specificty": 0.9310344827586207,
+      "eval_loss": 0.19854736328125,
+      "eval_runtime": 14.1828,
+      "eval_samples_per_second": 3.102,
+      "eval_steps_per_second": 0.141,
+      "step": 385
+    },
+    {
+      "epoch": 0.8696141931850183,
+      "grad_norm": 0.6257207074165352,
+      "learning_rate": 4.9760483771624236e-06,
+      "loss": 0.2196,
+      "step": 386
+    },
+    {
+      "epoch": 0.8718670796958603,
+      "grad_norm": 0.5266023178397307,
+      "learning_rate": 4.805809975558828e-06,
+      "loss": 0.2014,
+      "step": 387
+    },
+    {
+      "epoch": 0.8741199662067023,
+      "grad_norm": 0.7199181735751551,
+      "learning_rate": 4.6383874527259345e-06,
+      "loss": 0.1744,
+      "step": 388
+    },
+    {
+      "epoch": 0.8763728527175444,
+      "grad_norm": 0.4913647305272584,
+      "learning_rate": 4.47379124012689e-06,
+      "loss": 0.1805,
+      "step": 389
+    },
+    {
+      "epoch": 0.8786257392283864,
+      "grad_norm": 0.544522432312098,
+      "learning_rate": 4.312031593128163e-06,
+      "loss": 0.2246,
+      "step": 390
+    },
+    {
+      "epoch": 0.8786257392283864,
+      "eval_PRM Accuracy": 0.9215686274509803,
+      "eval_PRM F1": 0.9436619718309859,
+      "eval_PRM F1 AUC": 0.9244213509683514,
+      "eval_PRM F1 AUC (fixed)": 0.9482758620689655,
+      "eval_PRM F1 Neg": 0.8709677419354839,
+      "eval_PRM NPV": 0.8181818181818182,
+      "eval_PRM Precision": 0.9710144927536232,
+      "eval_PRM Recall": 0.9178082191780822,
+      "eval_PRM Specificty": 0.9310344827586207,
+      "eval_loss": 0.19843639433383942,
+      "eval_runtime": 13.9566,
+      "eval_samples_per_second": 3.153,
+      "eval_steps_per_second": 0.143,
+      "step": 390
+    },
+    {
+      "epoch": 0.8808786257392284,
+      "grad_norm": 0.6671085844733642,
+      "learning_rate": 4.153118590360561e-06,
+      "loss": 0.2677,
+      "step": 391
+    },
+    {
+      "epoch": 0.8831315122500704,
+      "grad_norm": 0.4616986295556261,
+      "learning_rate": 3.997062133091284e-06,
+      "loss": 0.1857,
+      "step": 392
+    },
+    {
+      "epoch": 0.8853843987609125,
+      "grad_norm": 0.5311025932277503,
+      "learning_rate": 3.843871944606969e-06,
+      "loss": 0.2104,
+      "step": 393
+    },
+    {
+      "epoch": 0.8876372852717545,
+      "grad_norm": 0.555106383407687,
+      "learning_rate": 3.693557569607947e-06,
+      "loss": 0.2223,
+      "step": 394
+    },
+    {
+      "epoch": 0.8898901717825964,
+      "grad_norm": 0.5740248711792133,
+      "learning_rate": 3.5461283736134722e-06,
+      "loss": 0.1988,
+      "step": 395
+    },
+    {
+      "epoch": 0.8898901717825964,
+      "eval_PRM Accuracy": 0.9215686274509803,
+      "eval_PRM F1": 0.9436619718309859,
+      "eval_PRM F1 AUC": 0.9244213509683514,
+      "eval_PRM F1 AUC (fixed)": 0.9487482286254133,
+      "eval_PRM F1 Neg": 0.8709677419354839,
+      "eval_PRM NPV": 0.8181818181818182,
+      "eval_PRM Precision": 0.9710144927536232,
+      "eval_PRM Recall": 0.9178082191780822,
+      "eval_PRM Specificty": 0.9310344827586207,
+      "eval_loss": 0.19784268736839294,
+      "eval_runtime": 13.7273,
+      "eval_samples_per_second": 3.205,
+      "eval_steps_per_second": 0.146,
+      "step": 395
+    },
+    {
+      "epoch": 0.8921430582934384,
+      "grad_norm": 0.5585934081344864,
+      "learning_rate": 3.401593542378262e-06,
+      "loss": 0.2143,
+      "step": 396
+    },
+    {
+      "epoch": 0.8943959448042805,
+      "grad_norm": 0.5917230124328218,
+      "learning_rate": 3.2599620813200837e-06,
+      "loss": 0.2234,
+      "step": 397
+    },
+    {
+      "epoch": 0.8966488313151225,
+      "grad_norm": 0.6457108145180654,
+      "learning_rate": 3.121242814958747e-06,
+      "loss": 0.2259,
+      "step": 398
+    },
+    {
+      "epoch": 0.8989017178259645,
+      "grad_norm": 0.58694400368762,
+      "learning_rate": 2.9854443863662262e-06,
+      "loss": 0.26,
+      "step": 399
+    },
+    {
+      "epoch": 0.9011546043368065,
+      "grad_norm": 0.4759715421243196,
+      "learning_rate": 2.8525752566281482e-06,
+      "loss": 0.1628,
+      "step": 400
+    },
+    {
+      "epoch": 0.9011546043368065,
+      "eval_PRM Accuracy": 0.9215686274509803,
+      "eval_PRM F1": 0.9436619718309859,
+      "eval_PRM F1 AUC": 0.9244213509683514,
+      "eval_PRM F1 AUC (fixed)": 0.9504015115729807,
+      "eval_PRM F1 Neg": 0.8709677419354839,
+      "eval_PRM NPV": 0.8181818181818182,
+      "eval_PRM Precision": 0.9710144927536232,
+      "eval_PRM Recall": 0.9178082191780822,
+      "eval_PRM Specificty": 0.9310344827586207,
+      "eval_loss": 0.19778165221214294,
+      "eval_runtime": 13.8936,
+      "eval_samples_per_second": 3.167,
+      "eval_steps_per_second": 0.144,
+      "step": 400
+    },
+    {
+      "epoch": 0.9034074908476486,
+      "grad_norm": 0.5756297678218019,
+      "learning_rate": 2.722643704316652e-06,
+      "loss": 0.2027,
+      "step": 401
+    },
+    {
+      "epoch": 0.9056603773584906,
+      "grad_norm": 0.6883879694051721,
+      "learning_rate": 2.5956578249745236e-06,
+      "loss": 0.2671,
+      "step": 402
+    },
+    {
+      "epoch": 0.9079132638693326,
+      "grad_norm": 0.44568450706368234,
+      "learning_rate": 2.4716255306108605e-06,
+      "loss": 0.2108,
+      "step": 403
+    },
+    {
+      "epoch": 0.9101661503801746,
+      "grad_norm": 0.5366009940871274,
+      "learning_rate": 2.3505545492080395e-06,
+      "loss": 0.1925,
+      "step": 404
+    },
+    {
+      "epoch": 0.9124190368910167,
+      "grad_norm": 0.4612024423445205,
+      "learning_rate": 2.2324524242402613e-06,
+      "loss": 0.1933,
+      "step": 405
+    },
+    {
+      "epoch": 0.9124190368910167,
+      "eval_PRM Accuracy": 0.9215686274509803,
+      "eval_PRM F1": 0.9436619718309859,
+      "eval_PRM F1 AUC": 0.9244213509683514,
+      "eval_PRM F1 AUC (fixed)": 0.949456778460085,
+      "eval_PRM F1 Neg": 0.8709677419354839,
+      "eval_PRM NPV": 0.8181818181818182,
+      "eval_PRM Precision": 0.9710144927536232,
+      "eval_PRM Recall": 0.9178082191780822,
+      "eval_PRM Specificty": 0.9310344827586207,
+      "eval_loss": 0.19827547669410706,
+      "eval_runtime": 14.1021,
+      "eval_samples_per_second": 3.12,
+      "eval_steps_per_second": 0.142,
+      "step": 405
+    },
+    {
+      "epoch": 0.9146719234018587,
+      "grad_norm": 0.7190557505360913,
+      "learning_rate": 2.117326514203527e-06,
+      "loss": 0.186,
+      "step": 406
+    },
+    {
+      "epoch": 0.9169248099127006,
+      "grad_norm": 0.4726122825818131,
+      "learning_rate": 2.0051839921571448e-06,
+      "loss": 0.2279,
+      "step": 407
+    },
+    {
+      "epoch": 0.9191776964235426,
+      "grad_norm": 0.6710759674362287,
+      "learning_rate": 1.8960318452768577e-06,
+      "loss": 0.2342,
+      "step": 408
+    },
+    {
+      "epoch": 0.9214305829343847,
+      "grad_norm": 0.517056632541587,
+      "learning_rate": 1.7898768744194162e-06,
+      "loss": 0.196,
+      "step": 409
+    },
+    {
+      "epoch": 0.9236834694452267,
+      "grad_norm": 0.5508357797654746,
+      "learning_rate": 1.6867256936989096e-06,
+      "loss": 0.2364,
+      "step": 410
+    },
+    {
+      "epoch": 0.9236834694452267,
+      "eval_PRM Accuracy": 0.9215686274509803,
+      "eval_PRM F1": 0.9436619718309859,
+      "eval_PRM F1 AUC": 0.9244213509683514,
+      "eval_PRM F1 AUC (fixed)": 0.949692961738309,
+      "eval_PRM F1 Neg": 0.8709677419354839,
+      "eval_PRM NPV": 0.8181818181818182,
+      "eval_PRM Precision": 0.9710144927536232,
+      "eval_PRM Recall": 0.9178082191780822,
+      "eval_PRM Specificty": 0.9310344827586207,
+      "eval_loss": 0.19834761321544647,
+      "eval_runtime": 13.7997,
+      "eval_samples_per_second": 3.188,
+      "eval_steps_per_second": 0.145,
+      "step": 410
+    },
+    {
+      "epoch": 0.9259363559560687,
+      "grad_norm": 0.4569524395441974,
+      "learning_rate": 1.5865847300746417e-06,
+      "loss": 0.1799,
+      "step": 411
+    },
+    {
+      "epoch": 0.9281892424669107,
+      "grad_norm": 0.6637730948902023,
+      "learning_rate": 1.4894602229506892e-06,
+      "loss": 0.1764,
+      "step": 412
+    },
+    {
+      "epoch": 0.9304421289777527,
+      "grad_norm": 0.8886882927165042,
+      "learning_rate": 1.3953582237871521e-06,
+      "loss": 0.2291,
+      "step": 413
+    },
+    {
+      "epoch": 0.9326950154885948,
+      "grad_norm": 0.5535162792587887,
+      "learning_rate": 1.3042845957231153e-06,
+      "loss": 0.2284,
+      "step": 414
+    },
+    {
+      "epoch": 0.9349479019994368,
+      "grad_norm": 0.5892966102809842,
+      "learning_rate": 1.2162450132113201e-06,
+      "loss": 0.1937,
+      "step": 415
+    },
+    {
+      "epoch": 0.9349479019994368,
+      "eval_PRM Accuracy": 0.9215686274509803,
+      "eval_PRM F1": 0.9436619718309859,
+      "eval_PRM F1 AUC": 0.9244213509683514,
+      "eval_PRM F1 AUC (fixed)": 0.9494567784600851,
+      "eval_PRM F1 Neg": 0.8709677419354839,
+      "eval_PRM NPV": 0.8181818181818182,
+      "eval_PRM Precision": 0.9710144927536232,
+      "eval_PRM Recall": 0.9178082191780822,
+      "eval_PRM Specificty": 0.9310344827586207,
+      "eval_loss": 0.1978815197944641,
+      "eval_runtime": 13.7747,
+      "eval_samples_per_second": 3.194,
+      "eval_steps_per_second": 0.145,
+      "step": 415
+    },
+    {
+      "epoch": 0.9372007885102788,
+      "grad_norm": 0.4963992817540815,
+      "learning_rate": 1.1312449616646403e-06,
+      "loss": 0.1954,
+      "step": 416
+    },
+    {
+      "epoch": 0.9394536750211208,
+      "grad_norm": 0.5434577961556927,
+      "learning_rate": 1.049289737114273e-06,
+      "loss": 0.2491,
+      "step": 417
+    },
+    {
+      "epoch": 0.9417065615319629,
+      "grad_norm": 0.6664456336427376,
+      "learning_rate": 9.703844458797962e-07,
+      "loss": 0.2119,
+      "step": 418
+    },
+    {
+      "epoch": 0.9439594480428048,
+      "grad_norm": 0.7315806108709918,
+      "learning_rate": 8.945340042509797e-07,
+      "loss": 0.2495,
+      "step": 419
+    },
+    {
+      "epoch": 0.9462123345536468,
+      "grad_norm": 0.7619252649434921,
+      "learning_rate": 8.217431381815077e-07,
+      "loss": 0.2002,
+      "step": 420
+    },
+    {
+      "epoch": 0.9462123345536468,
+      "eval_PRM Accuracy": 0.9215686274509803,
+      "eval_PRM F1": 0.9436619718309859,
+      "eval_PRM F1 AUC": 0.9244213509683514,
+      "eval_PRM F1 AUC (fixed)": 0.949692961738309,
+      "eval_PRM F1 Neg": 0.8709677419354839,
+      "eval_PRM NPV": 0.8181818181818182,
+      "eval_PRM Precision": 0.9710144927536232,
+      "eval_PRM Recall": 0.9178082191780822,
+      "eval_PRM Specificty": 0.9310344827586207,
+      "eval_loss": 0.19802579283714294,
+      "eval_runtime": 14.6531,
+      "eval_samples_per_second": 3.003,
+      "eval_steps_per_second": 0.136,
+      "step": 420
+    },
+    {
+      "epoch": 0.9484652210644888,
+      "grad_norm": 0.6164814778360996,
+      "learning_rate": 7.520163829944804e-07,
+      "loss": 0.2729,
+      "step": 421
+    },
+    {
+      "epoch": 0.9507181075753309,
+      "grad_norm": 0.6024987672035771,
+      "learning_rate": 6.85358083099863e-07,
+      "loss": 0.2389,
+      "step": 422
+    },
+    {
+      "epoch": 0.9529709940861729,
+      "grad_norm": 0.6450109151779052,
+      "learning_rate": 6.217723917238128e-07,
+      "loss": 0.2174,
+      "step": 423
+    },
+    {
+      "epoch": 0.9552238805970149,
+      "grad_norm": 0.724571607357305,
+      "learning_rate": 5.612632706498755e-07,
+      "loss": 0.2676,
+      "step": 424
+    },
+    {
+      "epoch": 0.9574767671078569,
+      "grad_norm": 0.4812504212364875,
+      "learning_rate": 5.038344899721436e-07,
+      "loss": 0.1955,
+      "step": 425
+    },
+    {
+      "epoch": 0.9574767671078569,
+      "eval_PRM Accuracy": 0.9215686274509803,
+      "eval_PRM F1": 0.9436619718309859,
+      "eval_PRM F1 AUC": 0.9244213509683514,
+      "eval_PRM F1 AUC (fixed)": 0.9487482286254133,
+      "eval_PRM F1 Neg": 0.8709677419354839,
+      "eval_PRM NPV": 0.8181818181818182,
+      "eval_PRM Precision": 0.9710144927536232,
+      "eval_PRM Recall": 0.9178082191780822,
+      "eval_PRM Specificty": 0.9310344827586207,
+      "eval_loss": 0.19794811308383942,
+      "eval_runtime": 15.4388,
+      "eval_samples_per_second": 2.85,
+      "eval_steps_per_second": 0.13,
+      "step": 425
+    },
+    {
+      "epoch": 0.959729653618699,
+      "grad_norm": 0.45945039997665643,
+      "learning_rate": 4.4948962786039437e-07,
+      "loss": 0.2281,
+      "step": 426
+    },
+    {
+      "epoch": 0.961982540129541,
+      "grad_norm": 0.6652933534346721,
+      "learning_rate": 3.9823207033710676e-07,
+      "loss": 0.1815,
+      "step": 427
+    },
+    {
+      "epoch": 0.964235426640383,
+      "grad_norm": 0.5060340453976231,
+      "learning_rate": 3.5006501106651937e-07,
+      "loss": 0.2215,
+      "step": 428
+    },
+    {
+      "epoch": 0.966488313151225,
+      "grad_norm": 0.5195356404454302,
+      "learning_rate": 3.0499145115561176e-07,
+      "loss": 0.2085,
+      "step": 429
+    },
+    {
+      "epoch": 0.9687411996620671,
+      "grad_norm": 0.5414918359478568,
+      "learning_rate": 2.630141989671542e-07,
+      "loss": 0.2134,
+      "step": 430
+    },
+    {
+      "epoch": 0.9687411996620671,
+      "eval_PRM Accuracy": 0.9215686274509803,
+      "eval_PRM F1": 0.9436619718309859,
+      "eval_PRM F1 AUC": 0.9244213509683514,
+      "eval_PRM F1 AUC (fixed)": 0.9494567784600851,
+      "eval_PRM F1 Neg": 0.8709677419354839,
+      "eval_PRM NPV": 0.8181818181818182,
+      "eval_PRM Precision": 0.9710144927536232,
+      "eval_PRM Recall": 0.9178082191780822,
+      "eval_PRM Specificty": 0.9310344827586207,
+      "eval_loss": 0.19731001555919647,
+      "eval_runtime": 13.9311,
+      "eval_samples_per_second": 3.158,
+      "eval_steps_per_second": 0.144,
+      "step": 430
+    },
+    {
+      "epoch": 0.970994086172909,
+      "grad_norm": 0.6044015417875366,
+      "learning_rate": 2.2413586994470825e-07,
+      "loss": 0.2238,
+      "step": 431
+    },
+    {
+      "epoch": 0.973246972683751,
+      "grad_norm": 0.5346367117442473,
+      "learning_rate": 1.8835888644966325e-07,
+      "loss": 0.1723,
+      "step": 432
+    },
+    {
+      "epoch": 0.975499859194593,
+      "grad_norm": 0.4973308614809863,
+      "learning_rate": 1.5568547761034004e-07,
+      "loss": 0.2355,
+      "step": 433
+    },
+    {
+      "epoch": 0.9777527457054351,
+      "grad_norm": 0.7257391274329523,
+      "learning_rate": 1.2611767918306316e-07,
+      "loss": 0.2237,
+      "step": 434
+    },
+    {
+      "epoch": 0.9800056322162771,
+      "grad_norm": 0.6669505689077171,
+      "learning_rate": 9.965733342532924e-08,
+      "loss": 0.1779,
+      "step": 435
+    },
+    {
+      "epoch": 0.9800056322162771,
+      "eval_PRM Accuracy": 0.9215686274509803,
+      "eval_PRM F1": 0.9436619718309859,
+      "eval_PRM F1 AUC": 0.9244213509683514,
+      "eval_PRM F1 AUC (fixed)": 0.9499291450165329,
+      "eval_PRM F1 Neg": 0.8709677419354839,
+      "eval_PRM NPV": 0.8181818181818182,
+      "eval_PRM Precision": 0.9710144927536232,
+      "eval_PRM Recall": 0.9178082191780822,
+      "eval_PRM Specificty": 0.9310344827586207,
+      "eval_loss": 0.19694657623767853,
+      "eval_runtime": 14.0998,
+      "eval_samples_per_second": 3.121,
+      "eval_steps_per_second": 0.142,
+      "step": 435
+    },
+    {
+      "epoch": 0.9822585187271191,
+      "grad_norm": 0.574351981878926,
+      "learning_rate": 7.630608898105962e-08,
+      "loss": 0.2279,
+      "step": 436
+    },
+    {
+      "epoch": 0.9845114052379611,
+      "grad_norm": 0.48261887083503596,
+      "learning_rate": 5.606540077782163e-08,
+      "loss": 0.1884,
+      "step": 437
+    },
+    {
+      "epoch": 0.9867642917488032,
+      "grad_norm": 0.4934699254372814,
+      "learning_rate": 3.893652993621766e-08,
+      "loss": 0.1896,
+      "step": 438
+    },
+    {
+      "epoch": 0.9890171782596452,
+      "grad_norm": 0.5864131169227933,
+      "learning_rate": 2.4920543691309138e-08,
+      "loss": 0.1858,
+      "step": 439
+    },
+    {
+      "epoch": 0.9912700647704872,
+      "grad_norm": 0.658123197798306,
+      "learning_rate": 1.4018315326103094e-08,
+      "loss": 0.2254,
+      "step": 440
+    },
+    {
+      "epoch": 0.9912700647704872,
+      "eval_PRM Accuracy": 0.9215686274509803,
+      "eval_PRM F1": 0.9436619718309859,
+      "eval_PRM F1 AUC": 0.9244213509683514,
+      "eval_PRM F1 AUC (fixed)": 0.9501653282947566,
+      "eval_PRM F1 Neg": 0.8709677419354839,
+      "eval_PRM NPV": 0.8181818181818182,
+      "eval_PRM Precision": 0.9710144927536232,
+      "eval_PRM Recall": 0.9178082191780822,
+      "eval_PRM Specificty": 0.9310344827586207,
+      "eval_loss": 0.19855846464633942,
+      "eval_runtime": 14.0747,
+      "eval_samples_per_second": 3.126,
+      "eval_steps_per_second": 0.142,
+      "step": 440
+    },
+    {
+      "epoch": 0.9935229512813292,
+      "grad_norm": 0.5253739518491587,
+      "learning_rate": 6.2305241171345395e-09,
+      "loss": 0.2018,
+      "step": 441
+    },
+    {
+      "epoch": 0.9957758377921713,
+      "grad_norm": 0.5168516097502279,
+      "learning_rate": 1.5576552921836574e-09,
+      "loss": 0.214,
+      "step": 442
+    },
+    {
+      "epoch": 0.9980287243030133,
+      "grad_norm": 0.6939736437579103,
+      "learning_rate": 0.0,
+      "loss": 0.2332,
+      "step": 443
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 443,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 581349891178496.0,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}