diff --git "a/last-checkpoint/trainer_state.json" "b/last-checkpoint/trainer_state.json"
--- "a/last-checkpoint/trainer_state.json"
+++ "b/last-checkpoint/trainer_state.json"
@@ -1,9 +1,9 @@
 {
   "best_metric": null,
   "best_model_checkpoint": null,
-  "epoch": 0.3883872220603942,
+  "epoch": 0.4531184257371266,
   "eval_steps": 5,
-  "global_step": 3000,
+  "global_step": 3500,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
@@ -30623,6 +30623,5106 @@
       "eval_samples_per_second": 3.3,
       "eval_steps_per_second": 0.225,
       "step": 3000
+    },
+    {
+      "epoch": 0.3885166844677477,
+      "grad_norm": 2.171178650336728,
+      "learning_rate": 9.590135932037357e-07,
+      "loss": 2.7949,
+      "step": 3001
+    },
+    {
+      "epoch": 0.38864614687510113,
+      "grad_norm": 1.4455267542303654,
+      "learning_rate": 9.587748047343934e-07,
+      "loss": 2.1575,
+      "step": 3002
+    },
+    {
+      "epoch": 0.3887756092824546,
+      "grad_norm": 1.5364013753655184,
+      "learning_rate": 9.58535948084782e-07,
+      "loss": 2.4132,
+      "step": 3003
+    },
+    {
+      "epoch": 0.3889050716898081,
+      "grad_norm": 1.7077560519608992,
+      "learning_rate": 9.582970233036928e-07,
+      "loss": 2.368,
+      "step": 3004
+    },
+    {
+      "epoch": 0.38903453409716154,
+      "grad_norm": 1.5600897361404518,
+      "learning_rate": 9.580580304399311e-07,
+      "loss": 2.5352,
+      "step": 3005
+    },
+    {
+      "epoch": 0.38903453409716154,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.655961513519287,
+      "eval_runtime": 13.3538,
+      "eval_samples_per_second": 3.295,
+      "eval_steps_per_second": 0.225,
+      "step": 3005
+    },
+    {
+      "epoch": 0.389163996504515,
+      "grad_norm": 3.117618267730982,
+      "learning_rate": 9.578189695423157e-07,
+      "loss": 2.6108,
+      "step": 3006
+    },
+    {
+      "epoch": 0.38929345891186845,
+      "grad_norm": 1.5244498534377757,
+      "learning_rate": 9.575798406596797e-07,
+      "loss": 2.5015,
+      "step": 3007
+    },
+    {
+      "epoch": 0.3894229213192219,
+      "grad_norm": 2.149137809617406,
+      "learning_rate": 9.573406438408706e-07,
+      "loss": 2.5037,
+      "step": 3008
+    },
+    {
+      "epoch": 0.3895523837265754,
+      "grad_norm": 1.7104875842506515,
+      "learning_rate": 9.571013791347487e-07,
+      "loss": 2.4773,
+      "step": 3009
+    },
+    {
+      "epoch": 0.38968184613392887,
+      "grad_norm": 2.0984190124201585,
+      "learning_rate": 9.568620465901885e-07,
+      "loss": 2.7288,
+      "step": 3010
+    },
+    {
+      "epoch": 0.38968184613392887,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.6532981395721436,
+      "eval_runtime": 14.4558,
+      "eval_samples_per_second": 3.044,
+      "eval_steps_per_second": 0.208,
+      "step": 3010
+    },
+    {
+      "epoch": 0.3898113085412823,
+      "grad_norm": 3.1110778131066494,
+      "learning_rate": 9.566226462560787e-07,
+      "loss": 2.6516,
+      "step": 3011
+    },
+    {
+      "epoch": 0.3899407709486358,
+      "grad_norm": 2.8595897648197517,
+      "learning_rate": 9.563831781813216e-07,
+      "loss": 2.6724,
+      "step": 3012
+    },
+    {
+      "epoch": 0.3900702333559893,
+      "grad_norm": 3.9573390702001343,
+      "learning_rate": 9.561436424148337e-07,
+      "loss": 2.3965,
+      "step": 3013
+    },
+    {
+      "epoch": 0.39019969576334274,
+      "grad_norm": 2.556971687568334,
+      "learning_rate": 9.559040390055449e-07,
+      "loss": 2.5435,
+      "step": 3014
+    },
+    {
+      "epoch": 0.3903291581706962,
+      "grad_norm": 3.613002590592715,
+      "learning_rate": 9.556643680023992e-07,
+      "loss": 2.4116,
+      "step": 3015
+    },
+    {
+      "epoch": 0.3903291581706962,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.6519221067428589,
+      "eval_runtime": 14.7133,
+      "eval_samples_per_second": 2.99,
+      "eval_steps_per_second": 0.204,
+      "step": 3015
+    },
+    {
+      "epoch": 0.39045862057804964,
+      "grad_norm": 1.558538799388353,
+      "learning_rate": 9.554246294543536e-07,
+      "loss": 2.3438,
+      "step": 3016
+    },
+    {
+      "epoch": 0.3905880829854031,
+      "grad_norm": 1.6024890999740347,
+      "learning_rate": 9.551848234103803e-07,
+      "loss": 2.321,
+      "step": 3017
+    },
+    {
+      "epoch": 0.3907175453927566,
+      "grad_norm": 1.5739377191691255,
+      "learning_rate": 9.549449499194645e-07,
+      "loss": 2.3508,
+      "step": 3018
+    },
+    {
+      "epoch": 0.39084700780011006,
+      "grad_norm": 3.4725995709644546,
+      "learning_rate": 9.547050090306048e-07,
+      "loss": 3.0415,
+      "step": 3019
+    },
+    {
+      "epoch": 0.3909764702074635,
+      "grad_norm": 1.8670577685348317,
+      "learning_rate": 9.544650007928142e-07,
+      "loss": 2.4199,
+      "step": 3020
+    },
+    {
+      "epoch": 0.3909764702074635,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.6511008739471436,
+      "eval_runtime": 15.0301,
+      "eval_samples_per_second": 2.927,
+      "eval_steps_per_second": 0.2,
+      "step": 3020
+    },
+    {
+      "epoch": 0.39110593261481696,
+      "grad_norm": 3.1261561040662413,
+      "learning_rate": 9.542249252551191e-07,
+      "loss": 2.8362,
+      "step": 3021
+    },
+    {
+      "epoch": 0.3912353950221704,
+      "grad_norm": 2.451160499309691,
+      "learning_rate": 9.5398478246656e-07,
+      "loss": 2.6284,
+      "step": 3022
+    },
+    {
+      "epoch": 0.3913648574295239,
+      "grad_norm": 3.861183752075475,
+      "learning_rate": 9.537445724761909e-07,
+      "loss": 2.9661,
+      "step": 3023
+    },
+    {
+      "epoch": 0.3914943198368774,
+      "grad_norm": 2.485001916311851,
+      "learning_rate": 9.535042953330796e-07,
+      "loss": 2.8076,
+      "step": 3024
+    },
+    {
+      "epoch": 0.39162378224423083,
+      "grad_norm": 1.9943353752441344,
+      "learning_rate": 9.532639510863073e-07,
+      "loss": 2.3853,
+      "step": 3025
+    },
+    {
+      "epoch": 0.39162378224423083,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.6526545286178589,
+      "eval_runtime": 21.7862,
+      "eval_samples_per_second": 2.02,
+      "eval_steps_per_second": 0.138,
+      "step": 3025
+    },
+    {
+      "epoch": 0.3917532446515843,
+      "grad_norm": 1.8786943520749657,
+      "learning_rate": 9.530235397849696e-07,
+      "loss": 2.7849,
+      "step": 3026
+    },
+    {
+      "epoch": 0.39188270705893774,
+      "grad_norm": 1.395797474944203,
+      "learning_rate": 9.527830614781751e-07,
+      "loss": 2.5081,
+      "step": 3027
+    },
+    {
+      "epoch": 0.39201216946629125,
+      "grad_norm": 2.217177230705167,
+      "learning_rate": 9.525425162150465e-07,
+      "loss": 2.4597,
+      "step": 3028
+    },
+    {
+      "epoch": 0.3921416318736447,
+      "grad_norm": 1.0821303446695993,
+      "learning_rate": 9.523019040447198e-07,
+      "loss": 2.3064,
+      "step": 3029
+    },
+    {
+      "epoch": 0.39227109428099816,
+      "grad_norm": 1.3666860412646131,
+      "learning_rate": 9.520612250163451e-07,
+      "loss": 2.3789,
+      "step": 3030
+    },
+    {
+      "epoch": 0.39227109428099816,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.6512340307235718,
+      "eval_runtime": 20.8251,
+      "eval_samples_per_second": 2.113,
+      "eval_steps_per_second": 0.144,
+      "step": 3030
+    },
+    {
+      "epoch": 0.3924005566883516,
+      "grad_norm": 3.202200381584669,
+      "learning_rate": 9.518204791790861e-07,
+      "loss": 2.6785,
+      "step": 3031
+    },
+    {
+      "epoch": 0.39253001909570506,
+      "grad_norm": 6.51169773920298,
+      "learning_rate": 9.515796665821197e-07,
+      "loss": 2.7849,
+      "step": 3032
+    },
+    {
+      "epoch": 0.39265948150305857,
+      "grad_norm": 1.4196275999762826,
+      "learning_rate": 9.513387872746371e-07,
+      "loss": 2.46,
+      "step": 3033
+    },
+    {
+      "epoch": 0.392788943910412,
+      "grad_norm": 1.8440774219857519,
+      "learning_rate": 9.510978413058426e-07,
+      "loss": 2.7456,
+      "step": 3034
+    },
+    {
+      "epoch": 0.3929184063177655,
+      "grad_norm": 2.2258732752178023,
+      "learning_rate": 9.508568287249542e-07,
+      "loss": 2.7627,
+      "step": 3035
+    },
+    {
+      "epoch": 0.3929184063177655,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.6516779661178589,
+      "eval_runtime": 14.6401,
+      "eval_samples_per_second": 3.005,
+      "eval_steps_per_second": 0.205,
+      "step": 3035
+    },
+    {
+      "epoch": 0.39304786872511893,
+      "grad_norm": 4.526151102252629,
+      "learning_rate": 9.506157495812037e-07,
+      "loss": 3.0513,
+      "step": 3036
+    },
+    {
+      "epoch": 0.3931773311324724,
+      "grad_norm": 1.5664132059307387,
+      "learning_rate": 9.503746039238368e-07,
+      "loss": 2.6113,
+      "step": 3037
+    },
+    {
+      "epoch": 0.3933067935398259,
+      "grad_norm": 1.6605676456023712,
+      "learning_rate": 9.501333918021116e-07,
+      "loss": 2.3708,
+      "step": 3038
+    },
+    {
+      "epoch": 0.39343625594717935,
+      "grad_norm": 2.1685045803081646,
+      "learning_rate": 9.498921132653011e-07,
+      "loss": 2.9172,
+      "step": 3039
+    },
+    {
+      "epoch": 0.3935657183545328,
+      "grad_norm": 1.869239013237251,
+      "learning_rate": 9.496507683626913e-07,
+      "loss": 2.5569,
+      "step": 3040
+    },
+    {
+      "epoch": 0.3935657183545328,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.649702548980713,
+      "eval_runtime": 14.7449,
+      "eval_samples_per_second": 2.984,
+      "eval_steps_per_second": 0.203,
+      "step": 3040
+    },
+    {
+      "epoch": 0.39369518076188625,
+      "grad_norm": 1.9271257473699925,
+      "learning_rate": 9.494093571435814e-07,
+      "loss": 2.5493,
+      "step": 3041
+    },
+    {
+      "epoch": 0.3938246431692397,
+      "grad_norm": 2.2037140668267723,
+      "learning_rate": 9.491678796572851e-07,
+      "loss": 2.4985,
+      "step": 3042
+    },
+    {
+      "epoch": 0.3939541055765932,
+      "grad_norm": 2.7915996843622084,
+      "learning_rate": 9.489263359531286e-07,
+      "loss": 2.7314,
+      "step": 3043
+    },
+    {
+      "epoch": 0.39408356798394667,
+      "grad_norm": 5.082239798908639,
+      "learning_rate": 9.486847260804524e-07,
+      "loss": 2.7034,
+      "step": 3044
+    },
+    {
+      "epoch": 0.3942130303913001,
+      "grad_norm": 2.07919215026421,
+      "learning_rate": 9.484430500886101e-07,
+      "loss": 2.4578,
+      "step": 3045
+    },
+    {
+      "epoch": 0.3942130303913001,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.6489479541778564,
+      "eval_runtime": 16.0474,
+      "eval_samples_per_second": 2.742,
+      "eval_steps_per_second": 0.187,
+      "step": 3045
+    },
+    {
+      "epoch": 0.3943424927986536,
+      "grad_norm": 2.0770259414754824,
+      "learning_rate": 9.48201308026969e-07,
+      "loss": 2.5146,
+      "step": 3046
+    },
+    {
+      "epoch": 0.39447195520600703,
+      "grad_norm": 1.9094743532085647,
+      "learning_rate": 9.479594999449096e-07,
+      "loss": 2.6597,
+      "step": 3047
+    },
+    {
+      "epoch": 0.39460141761336054,
+      "grad_norm": 2.166069735039154,
+      "learning_rate": 9.477176258918264e-07,
+      "loss": 2.5308,
+      "step": 3048
+    },
+    {
+      "epoch": 0.394730880020714,
+      "grad_norm": 1.7517625118258298,
+      "learning_rate": 9.474756859171268e-07,
+      "loss": 2.4011,
+      "step": 3049
+    },
+    {
+      "epoch": 0.39486034242806745,
+      "grad_norm": 2.0901260404474318,
+      "learning_rate": 9.47233680070232e-07,
+      "loss": 2.5554,
+      "step": 3050
+    },
+    {
+      "epoch": 0.39486034242806745,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.648193359375,
+      "eval_runtime": 19.4171,
+      "eval_samples_per_second": 2.266,
+      "eval_steps_per_second": 0.155,
+      "step": 3050
+    },
+    {
+      "epoch": 0.3949898048354209,
+      "grad_norm": 1.4068251970442467,
+      "learning_rate": 9.469916084005771e-07,
+      "loss": 2.4116,
+      "step": 3051
+    },
+    {
+      "epoch": 0.39511926724277435,
+      "grad_norm": 1.4248039613628964,
+      "learning_rate": 9.467494709576091e-07,
+      "loss": 2.4207,
+      "step": 3052
+    },
+    {
+      "epoch": 0.39524872965012786,
+      "grad_norm": 2.0154883756342667,
+      "learning_rate": 9.465072677907903e-07,
+      "loss": 2.4954,
+      "step": 3053
+    },
+    {
+      "epoch": 0.3953781920574813,
+      "grad_norm": 2.2741005264310097,
+      "learning_rate": 9.462649989495956e-07,
+      "loss": 2.8135,
+      "step": 3054
+    },
+    {
+      "epoch": 0.39550765446483477,
+      "grad_norm": 2.1056246453063996,
+      "learning_rate": 9.46022664483513e-07,
+      "loss": 2.6316,
+      "step": 3055
+    },
+    {
+      "epoch": 0.39550765446483477,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.6426669359207153,
+      "eval_runtime": 15.5456,
+      "eval_samples_per_second": 2.83,
+      "eval_steps_per_second": 0.193,
+      "step": 3055
+    },
+    {
+      "epoch": 0.3956371168721882,
+      "grad_norm": 2.4372041345882907,
+      "learning_rate": 9.457802644420444e-07,
+      "loss": 2.3853,
+      "step": 3056
+    },
+    {
+      "epoch": 0.39576657927954173,
+      "grad_norm": 2.6901249077029696,
+      "learning_rate": 9.455377988747047e-07,
+      "loss": 2.7087,
+      "step": 3057
+    },
+    {
+      "epoch": 0.3958960416868952,
+      "grad_norm": 1.5380116052703254,
+      "learning_rate": 9.452952678310226e-07,
+      "loss": 2.4177,
+      "step": 3058
+    },
+    {
+      "epoch": 0.39602550409424864,
+      "grad_norm": 1.2958424771716672,
+      "learning_rate": 9.450526713605402e-07,
+      "loss": 2.1733,
+      "step": 3059
+    },
+    {
+      "epoch": 0.3961549665016021,
+      "grad_norm": 2.451527162775795,
+      "learning_rate": 9.448100095128124e-07,
+      "loss": 2.6582,
+      "step": 3060
+    },
+    {
+      "epoch": 0.3961549665016021,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.6427334547042847,
+      "eval_runtime": 14.6392,
+      "eval_samples_per_second": 3.006,
+      "eval_steps_per_second": 0.205,
+      "step": 3060
+    },
+    {
+      "epoch": 0.39628442890895554,
+      "grad_norm": 4.451044959814333,
+      "learning_rate": 9.445672823374078e-07,
+      "loss": 2.6443,
+      "step": 3061
+    },
+    {
+      "epoch": 0.39641389131630905,
+      "grad_norm": 2.0231162149575024,
+      "learning_rate": 9.443244898839085e-07,
+      "loss": 2.5566,
+      "step": 3062
+    },
+    {
+      "epoch": 0.3965433537236625,
+      "grad_norm": 3.6134534339377073,
+      "learning_rate": 9.440816322019097e-07,
+      "loss": 3.1411,
+      "step": 3063
+    },
+    {
+      "epoch": 0.39667281613101596,
+      "grad_norm": 2.5762787718822393,
+      "learning_rate": 9.438387093410201e-07,
+      "loss": 2.9038,
+      "step": 3064
+    },
+    {
+      "epoch": 0.3968022785383694,
+      "grad_norm": 2.222180988623937,
+      "learning_rate": 9.435957213508615e-07,
+      "loss": 2.6431,
+      "step": 3065
+    },
+    {
+      "epoch": 0.3968022785383694,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.6433993577957153,
+      "eval_runtime": 14.8405,
+      "eval_samples_per_second": 2.965,
+      "eval_steps_per_second": 0.202,
+      "step": 3065
+    },
+    {
+      "epoch": 0.39693174094572287,
+      "grad_norm": 2.062054458392921,
+      "learning_rate": 9.43352668281069e-07,
+      "loss": 2.5671,
+      "step": 3066
+    },
+    {
+      "epoch": 0.3970612033530764,
+      "grad_norm": 2.2504375012902216,
+      "learning_rate": 9.431095501812915e-07,
+      "loss": 2.5828,
+      "step": 3067
+    },
+    {
+      "epoch": 0.39719066576042983,
+      "grad_norm": 2.787612981836153,
+      "learning_rate": 9.428663671011907e-07,
+      "loss": 2.4934,
+      "step": 3068
+    },
+    {
+      "epoch": 0.3973201281677833,
+      "grad_norm": 1.6814451247588047,
+      "learning_rate": 9.426231190904414e-07,
+      "loss": 2.478,
+      "step": 3069
+    },
+    {
+      "epoch": 0.39744959057513674,
+      "grad_norm": 3.913986592034534,
+      "learning_rate": 9.423798061987321e-07,
+      "loss": 3.0737,
+      "step": 3070
+    },
+    {
+      "epoch": 0.39744959057513674,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.6434437036514282,
+      "eval_runtime": 14.0731,
+      "eval_samples_per_second": 3.127,
+      "eval_steps_per_second": 0.213,
+      "step": 3070
+    },
+    {
+      "epoch": 0.3975790529824902,
+      "grad_norm": 4.0420543144703736,
+      "learning_rate": 9.421364284757644e-07,
+      "loss": 2.5369,
+      "step": 3071
+    },
+    {
+      "epoch": 0.3977085153898437,
+      "grad_norm": 1.981399142103187,
+      "learning_rate": 9.418929859712531e-07,
+      "loss": 2.4753,
+      "step": 3072
+    },
+    {
+      "epoch": 0.39783797779719715,
+      "grad_norm": 2.6473778429723227,
+      "learning_rate": 9.416494787349265e-07,
+      "loss": 3.0818,
+      "step": 3073
+    },
+    {
+      "epoch": 0.3979674402045506,
+      "grad_norm": 2.2789585146723894,
+      "learning_rate": 9.414059068165254e-07,
+      "loss": 2.4092,
+      "step": 3074
+    },
+    {
+      "epoch": 0.39809690261190406,
+      "grad_norm": 4.9804698868183985,
+      "learning_rate": 9.411622702658045e-07,
+      "loss": 3.074,
+      "step": 3075
+    },
+    {
+      "epoch": 0.39809690261190406,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.6427112817764282,
+      "eval_runtime": 15.6032,
+      "eval_samples_per_second": 2.82,
+      "eval_steps_per_second": 0.192,
+      "step": 3075
+    },
+    {
+      "epoch": 0.3982263650192575,
+      "grad_norm": 2.0084135170836426,
+      "learning_rate": 9.409185691325317e-07,
+      "loss": 2.554,
+      "step": 3076
+    },
+    {
+      "epoch": 0.398355827426611,
+      "grad_norm": 4.293940724585163,
+      "learning_rate": 9.40674803466488e-07,
+      "loss": 2.9316,
+      "step": 3077
+    },
+    {
+      "epoch": 0.3984852898339645,
+      "grad_norm": 1.7772927181216995,
+      "learning_rate": 9.40430973317467e-07,
+      "loss": 2.3558,
+      "step": 3078
+    },
+    {
+      "epoch": 0.3986147522413179,
+      "grad_norm": 1.8421636161400678,
+      "learning_rate": 9.401870787352763e-07,
+      "loss": 2.5823,
+      "step": 3079
+    },
+    {
+      "epoch": 0.3987442146486714,
+      "grad_norm": 2.2826887420234674,
+      "learning_rate": 9.399431197697361e-07,
+      "loss": 2.7739,
+      "step": 3080
+    },
+    {
+      "epoch": 0.3987442146486714,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.6447087526321411,
+      "eval_runtime": 14.1136,
+      "eval_samples_per_second": 3.118,
+      "eval_steps_per_second": 0.213,
+      "step": 3080
+    },
+    {
+      "epoch": 0.39887367705602483,
+      "grad_norm": 2.1002790638395177,
+      "learning_rate": 9.396990964706802e-07,
+      "loss": 2.4294,
+      "step": 3081
+    },
+    {
+      "epoch": 0.39900313946337834,
+      "grad_norm": 1.6863920592929518,
+      "learning_rate": 9.394550088879554e-07,
+      "loss": 2.2727,
+      "step": 3082
+    },
+    {
+      "epoch": 0.3991326018707318,
+      "grad_norm": 2.226894359453203,
+      "learning_rate": 9.392108570714209e-07,
+      "loss": 2.3413,
+      "step": 3083
+    },
+    {
+      "epoch": 0.39926206427808525,
+      "grad_norm": 1.8551535970047062,
+      "learning_rate": 9.389666410709503e-07,
+      "loss": 2.5388,
+      "step": 3084
+    },
+    {
+      "epoch": 0.3993915266854387,
+      "grad_norm": 2.763576814244469,
+      "learning_rate": 9.387223609364295e-07,
+      "loss": 2.4541,
+      "step": 3085
+    },
+    {
+      "epoch": 0.3993915266854387,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.646928310394287,
+      "eval_runtime": 13.484,
+      "eval_samples_per_second": 3.263,
+      "eval_steps_per_second": 0.222,
+      "step": 3085
+    },
+    {
+      "epoch": 0.39952098909279216,
+      "grad_norm": 1.7293047120687537,
+      "learning_rate": 9.384780167177575e-07,
+      "loss": 2.4565,
+      "step": 3086
+    },
+    {
+      "epoch": 0.39965045150014566,
+      "grad_norm": 2.499157859208796,
+      "learning_rate": 9.382336084648467e-07,
+      "loss": 2.7568,
+      "step": 3087
+    },
+    {
+      "epoch": 0.3997799139074991,
+      "grad_norm": 2.4906710804355856,
+      "learning_rate": 9.379891362276224e-07,
+      "loss": 2.7517,
+      "step": 3088
+    },
+    {
+      "epoch": 0.39990937631485257,
+      "grad_norm": 2.6127722898284054,
+      "learning_rate": 9.377446000560231e-07,
+      "loss": 2.6042,
+      "step": 3089
+    },
+    {
+      "epoch": 0.400038838722206,
+      "grad_norm": 2.4247062554391317,
+      "learning_rate": 9.375000000000001e-07,
+      "loss": 2.7288,
+      "step": 3090
+    },
+    {
+      "epoch": 0.400038838722206,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.6405361890792847,
+      "eval_runtime": 14.954,
+      "eval_samples_per_second": 2.942,
+      "eval_steps_per_second": 0.201,
+      "step": 3090
+    },
+    {
+      "epoch": 0.4001683011295595,
+      "grad_norm": 3.302854678868893,
+      "learning_rate": 9.372553361095181e-07,
+      "loss": 2.7615,
+      "step": 3091
+    },
+    {
+      "epoch": 0.400297763536913,
+      "grad_norm": 2.477649676457822,
+      "learning_rate": 9.370106084345545e-07,
+      "loss": 2.655,
+      "step": 3092
+    },
+    {
+      "epoch": 0.40042722594426644,
+      "grad_norm": 3.4201098845011346,
+      "learning_rate": 9.367658170251e-07,
+      "loss": 2.9888,
+      "step": 3093
+    },
+    {
+      "epoch": 0.4005566883516199,
+      "grad_norm": 2.4589305440829006,
+      "learning_rate": 9.365209619311581e-07,
+      "loss": 2.6514,
+      "step": 3094
+    },
+    {
+      "epoch": 0.40068615075897335,
+      "grad_norm": 4.229198532421323,
+      "learning_rate": 9.362760432027456e-07,
+      "loss": 2.6074,
+      "step": 3095
+    },
+    {
+      "epoch": 0.40068615075897335,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.6382724046707153,
+      "eval_runtime": 16.6591,
+      "eval_samples_per_second": 2.641,
+      "eval_steps_per_second": 0.18,
+      "step": 3095
+    },
+    {
+      "epoch": 0.4008156131663268,
+      "grad_norm": 2.4787746453106436,
+      "learning_rate": 9.36031060889892e-07,
+      "loss": 2.7883,
+      "step": 3096
+    },
+    {
+      "epoch": 0.4009450755736803,
+      "grad_norm": 2.7178854594092137,
+      "learning_rate": 9.357860150426397e-07,
+      "loss": 2.728,
+      "step": 3097
+    },
+    {
+      "epoch": 0.40107453798103376,
+      "grad_norm": 1.872971907836054,
+      "learning_rate": 9.355409057110447e-07,
+      "loss": 2.5439,
+      "step": 3098
+    },
+    {
+      "epoch": 0.4012040003883872,
+      "grad_norm": 2.2670914242792937,
+      "learning_rate": 9.352957329451752e-07,
+      "loss": 2.6538,
+      "step": 3099
+    },
+    {
+      "epoch": 0.40133346279574067,
+      "grad_norm": 1.5071107420723142,
+      "learning_rate": 9.35050496795113e-07,
+      "loss": 2.3447,
+      "step": 3100
+    },
+    {
+      "epoch": 0.40133346279574067,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.64111328125,
+      "eval_runtime": 16.8423,
+      "eval_samples_per_second": 2.612,
+      "eval_steps_per_second": 0.178,
+      "step": 3100
+    },
+    {
+      "epoch": 0.4014629252030942,
+      "grad_norm": 1.8478456067846563,
+      "learning_rate": 9.348051973109522e-07,
+      "loss": 2.4612,
+      "step": 3101
+    },
+    {
+      "epoch": 0.40159238761044763,
+      "grad_norm": 1.5693193171751734,
+      "learning_rate": 9.345598345428004e-07,
+      "loss": 2.4302,
+      "step": 3102
+    },
+    {
+      "epoch": 0.4017218500178011,
+      "grad_norm": 2.025385736654082,
+      "learning_rate": 9.34314408540778e-07,
+      "loss": 2.5471,
+      "step": 3103
+    },
+    {
+      "epoch": 0.40185131242515454,
+      "grad_norm": 1.2568905976090412,
+      "learning_rate": 9.340689193550182e-07,
+      "loss": 2.4692,
+      "step": 3104
+    },
+    {
+      "epoch": 0.401980774832508,
+      "grad_norm": 2.9181681902493724,
+      "learning_rate": 9.33823367035667e-07,
+      "loss": 2.6482,
+      "step": 3105
+    },
+    {
+      "epoch": 0.401980774832508,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.638383388519287,
+      "eval_runtime": 14.5494,
+      "eval_samples_per_second": 3.024,
+      "eval_steps_per_second": 0.206,
+      "step": 3105
+    },
+    {
+      "epoch": 0.4021102372398615,
+      "grad_norm": 3.589858781874059,
+      "learning_rate": 9.335777516328833e-07,
+      "loss": 2.6055,
+      "step": 3106
+    },
+    {
+      "epoch": 0.40223969964721495,
+      "grad_norm": 2.2409088019430112,
+      "learning_rate": 9.333320731968396e-07,
+      "loss": 2.5139,
+      "step": 3107
+    },
+    {
+      "epoch": 0.4023691620545684,
+      "grad_norm": 1.2046073215318571,
+      "learning_rate": 9.330863317777201e-07,
+      "loss": 2.2661,
+      "step": 3108
+    },
+    {
+      "epoch": 0.40249862446192186,
+      "grad_norm": 1.8262955699883103,
+      "learning_rate": 9.328405274257228e-07,
+      "loss": 2.4307,
+      "step": 3109
+    },
+    {
+      "epoch": 0.4026280868692753,
+      "grad_norm": 2.300838876277139,
+      "learning_rate": 9.325946601910582e-07,
+      "loss": 2.4114,
+      "step": 3110
+    },
+    {
+      "epoch": 0.4026280868692753,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.6368962526321411,
+      "eval_runtime": 13.8464,
+      "eval_samples_per_second": 3.178,
+      "eval_steps_per_second": 0.217,
+      "step": 3110
+    },
+    {
+      "epoch": 0.4027575492766288,
+      "grad_norm": 14.40113087060862,
+      "learning_rate": 9.323487301239495e-07,
+      "loss": 3.0903,
+      "step": 3111
+    },
+    {
+      "epoch": 0.4028870116839823,
+      "grad_norm": 1.8789966606954218,
+      "learning_rate": 9.32102737274633e-07,
+      "loss": 2.479,
+      "step": 3112
+    },
+    {
+      "epoch": 0.40301647409133573,
+      "grad_norm": 3.782303856963805,
+      "learning_rate": 9.318566816933578e-07,
+      "loss": 2.623,
+      "step": 3113
+    },
+    {
+      "epoch": 0.4031459364986892,
+      "grad_norm": 1.5264590123020203,
+      "learning_rate": 9.316105634303856e-07,
+      "loss": 2.4453,
+      "step": 3114
+    },
+    {
+      "epoch": 0.40327539890604264,
+      "grad_norm": 2.663249131366837,
+      "learning_rate": 9.313643825359911e-07,
+      "loss": 2.6101,
+      "step": 3115
+    },
+    {
+      "epoch": 0.40327539890604264,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.6377174854278564,
+      "eval_runtime": 19.1181,
+      "eval_samples_per_second": 2.301,
+      "eval_steps_per_second": 0.157,
+      "step": 3115
+    },
+    {
+      "epoch": 0.40340486131339615,
+      "grad_norm": 1.3755247525337402,
+      "learning_rate": 9.311181390604618e-07,
+      "loss": 2.4829,
+      "step": 3116
+    },
+    {
+      "epoch": 0.4035343237207496,
+      "grad_norm": 1.9212891466534978,
+      "learning_rate": 9.308718330540978e-07,
+      "loss": 2.4246,
+      "step": 3117
+    },
+    {
+      "epoch": 0.40366378612810305,
+      "grad_norm": 2.544753759419078,
+      "learning_rate": 9.306254645672122e-07,
+      "loss": 2.615,
+      "step": 3118
+    },
+    {
+      "epoch": 0.4037932485354565,
+      "grad_norm": 1.9523295145421204,
+      "learning_rate": 9.303790336501307e-07,
+      "loss": 2.6646,
+      "step": 3119
+    },
+    {
+      "epoch": 0.40392271094280996,
+      "grad_norm": 1.416475618164198,
+      "learning_rate": 9.301325403531918e-07,
+      "loss": 2.3164,
+      "step": 3120
+    },
+    {
+      "epoch": 0.40392271094280996,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.6353648900985718,
+      "eval_runtime": 14.8817,
+      "eval_samples_per_second": 2.957,
+      "eval_steps_per_second": 0.202,
+      "step": 3120
+    },
+    {
+      "epoch": 0.40405217335016347,
+      "grad_norm": 2.3078047604522482,
+      "learning_rate": 9.298859847267469e-07,
+      "loss": 2.4058,
+      "step": 3121
+    },
+    {
+      "epoch": 0.4041816357575169,
+      "grad_norm": 9.388826804965841,
+      "learning_rate": 9.296393668211596e-07,
+      "loss": 2.8711,
+      "step": 3122
+    },
+    {
+      "epoch": 0.4043110981648704,
+      "grad_norm": 2.0987676802321404,
+      "learning_rate": 9.293926866868072e-07,
+      "loss": 2.5203,
+      "step": 3123
+    },
+    {
+      "epoch": 0.40444056057222383,
+      "grad_norm": 2.255919448500308,
+      "learning_rate": 9.291459443740786e-07,
+      "loss": 2.4963,
+      "step": 3124
+    },
+    {
+      "epoch": 0.4045700229795773,
+      "grad_norm": 4.102169383890809,
+      "learning_rate": 9.288991399333761e-07,
+      "loss": 2.9155,
+      "step": 3125
+    },
+    {
+      "epoch": 0.4045700229795773,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.6348544359207153,
+      "eval_runtime": 13.892,
+      "eval_samples_per_second": 3.167,
+      "eval_steps_per_second": 0.216,
+      "step": 3125
+    },
+    {
+      "epoch": 0.4046994853869308,
+      "grad_norm": 2.3625896025957704,
+      "learning_rate": 9.286522734151146e-07,
+      "loss": 2.4388,
+      "step": 3126
+    },
+    {
+      "epoch": 0.40482894779428424,
+      "grad_norm": 2.33447697730886,
+      "learning_rate": 9.284053448697213e-07,
+      "loss": 2.6108,
+      "step": 3127
+    },
+    {
+      "epoch": 0.4049584102016377,
+      "grad_norm": 4.5414991479556095,
+      "learning_rate": 9.281583543476368e-07,
+      "loss": 2.8462,
+      "step": 3128
+    },
+    {
+      "epoch": 0.40508787260899115,
+      "grad_norm": 1.8421529376703967,
+      "learning_rate": 9.279113018993136e-07,
+      "loss": 2.5676,
+      "step": 3129
+    },
+    {
+      "epoch": 0.4052173350163446,
+      "grad_norm": 1.7697452402571285,
+      "learning_rate": 9.276641875752171e-07,
+      "loss": 2.6553,
+      "step": 3130
+    },
+    {
+      "epoch": 0.4052173350163446,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.638427734375,
+      "eval_runtime": 23.0472,
+      "eval_samples_per_second": 1.909,
+      "eval_steps_per_second": 0.13,
+      "step": 3130
+    },
+    {
+      "epoch": 0.4053467974236981,
+      "grad_norm": 2.0337708679645736,
+      "learning_rate": 9.274170114258258e-07,
+      "loss": 2.4209,
+      "step": 3131
+    },
+    {
+      "epoch": 0.40547625983105157,
+      "grad_norm": 1.66956425256569,
+      "learning_rate": 9.271697735016302e-07,
+      "loss": 2.5112,
+      "step": 3132
+    },
+    {
+      "epoch": 0.405605722238405,
+      "grad_norm": 1.2457792927470352,
+      "learning_rate": 9.269224738531335e-07,
+      "loss": 2.3901,
+      "step": 3133
+    },
+    {
+      "epoch": 0.4057351846457585,
+      "grad_norm": 1.63242837213429,
+      "learning_rate": 9.266751125308518e-07,
+      "loss": 2.4534,
+      "step": 3134
+    },
+    {
+      "epoch": 0.4058646470531119,
+      "grad_norm": 1.8314878495342952,
+      "learning_rate": 9.264276895853138e-07,
+      "loss": 2.3594,
+      "step": 3135
+    },
+    {
+      "epoch": 0.4058646470531119,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.6320579051971436,
+      "eval_runtime": 14.5662,
+      "eval_samples_per_second": 3.021,
+      "eval_steps_per_second": 0.206,
+      "step": 3135
+    },
+    {
+      "epoch": 0.40599410946046544,
+      "grad_norm": 2.6779428074927383,
+      "learning_rate": 9.261802050670606e-07,
+      "loss": 2.8491,
+      "step": 3136
+    },
+    {
+      "epoch": 0.4061235718678189,
+      "grad_norm": 3.7541144768700803,
+      "learning_rate": 9.259326590266457e-07,
+      "loss": 2.5808,
+      "step": 3137
+    },
+    {
+      "epoch": 0.40625303427517234,
+      "grad_norm": 1.2850824932258837,
+      "learning_rate": 9.256850515146357e-07,
+      "loss": 2.5637,
+      "step": 3138
+    },
+    {
+      "epoch": 0.4063824966825258,
+      "grad_norm": 2.6325487961655214,
+      "learning_rate": 9.254373825816093e-07,
+      "loss": 2.5239,
+      "step": 3139
+    },
+    {
+      "epoch": 0.40651195908987925,
+      "grad_norm": 1.7155327463440555,
+      "learning_rate": 9.25189652278158e-07,
+      "loss": 2.5234,
+      "step": 3140
+    },
+    {
+      "epoch": 0.40651195908987925,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.6346768140792847,
+      "eval_runtime": 16.4256,
+      "eval_samples_per_second": 2.679,
+      "eval_steps_per_second": 0.183,
+      "step": 3140
+    },
+    {
+      "epoch": 0.40664142149723276,
+      "grad_norm": 2.6385478954307082,
+      "learning_rate": 9.249418606548855e-07,
+      "loss": 2.6528,
+      "step": 3141
+    },
+    {
+      "epoch": 0.4067708839045862,
+      "grad_norm": 2.1806615397337565,
+      "learning_rate": 9.246940077624085e-07,
+      "loss": 2.6929,
+      "step": 3142
+    },
+    {
+      "epoch": 0.40690034631193966,
+      "grad_norm": 1.9939039450824454,
+      "learning_rate": 9.244460936513557e-07,
+      "loss": 2.6274,
+      "step": 3143
+    },
+    {
+      "epoch": 0.4070298087192931,
+      "grad_norm": 1.5289750685486185,
+      "learning_rate": 9.241981183723691e-07,
+      "loss": 2.6245,
+      "step": 3144
+    },
+    {
+      "epoch": 0.4071592711266466,
+      "grad_norm": 1.8233159160424126,
+      "learning_rate": 9.239500819761022e-07,
+      "loss": 2.6013,
+      "step": 3145
+    },
+    {
+      "epoch": 0.4071592711266466,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.6326793432235718,
+      "eval_runtime": 14.4513,
+      "eval_samples_per_second": 3.045,
+      "eval_steps_per_second": 0.208,
+      "step": 3145
+    },
+    {
+      "epoch": 0.4072887335340001,
+      "grad_norm": 2.5374459117830663,
+      "learning_rate": 9.237019845132216e-07,
+      "loss": 2.6333,
+      "step": 3146
+    },
+    {
+      "epoch": 0.40741819594135353,
+      "grad_norm": 1.6299393691109445,
+      "learning_rate": 9.234538260344061e-07,
+      "loss": 2.5466,
+      "step": 3147
+    },
+    {
+      "epoch": 0.407547658348707,
+      "grad_norm": 2.676433195955329,
+      "learning_rate": 9.232056065903476e-07,
+      "loss": 2.7439,
+      "step": 3148
+    },
+    {
+      "epoch": 0.40767712075606044,
+      "grad_norm": 1.9111031935904645,
+      "learning_rate": 9.229573262317494e-07,
+      "loss": 2.6409,
+      "step": 3149
+    },
+    {
+      "epoch": 0.40780658316341395,
+      "grad_norm": 2.119199479842497,
+      "learning_rate": 9.227089850093279e-07,
+      "loss": 2.3306,
+      "step": 3150
+    },
+    {
+      "epoch": 0.40780658316341395,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.6363414525985718,
+      "eval_runtime": 14.2686,
+      "eval_samples_per_second": 3.084,
+      "eval_steps_per_second": 0.21,
+      "step": 3150
+    },
+    {
+      "epoch": 0.4079360455707674,
+      "grad_norm": 2.306589793950446,
+      "learning_rate": 9.22460582973812e-07,
+      "loss": 2.2992,
+      "step": 3151
+    },
+    {
+      "epoch": 0.40806550797812086,
+      "grad_norm": 2.5095156128091207,
+      "learning_rate": 9.222121201759429e-07,
+      "loss": 2.7454,
+      "step": 3152
+    },
+    {
+      "epoch": 0.4081949703854743,
+      "grad_norm": 1.7471666423852787,
+      "learning_rate": 9.219635966664739e-07,
+      "loss": 2.4064,
+      "step": 3153
+    },
+    {
+      "epoch": 0.40832443279282776,
+      "grad_norm": 2.346173468590584,
+      "learning_rate": 9.217150124961713e-07,
+      "loss": 2.7056,
+      "step": 3154
+    },
+    {
+      "epoch": 0.40845389520018127,
+      "grad_norm": 1.8827726403951508,
+      "learning_rate": 9.214663677158131e-07,
+      "loss": 2.3745,
+      "step": 3155
+    },
+    {
+      "epoch": 0.40845389520018127,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.6353427171707153,
+      "eval_runtime": 21.5011,
+      "eval_samples_per_second": 2.046,
+      "eval_steps_per_second": 0.14,
+      "step": 3155
+    },
+    {
+      "epoch": 0.4085833576075347,
+      "grad_norm": 3.1129255407423386,
+      "learning_rate": 9.212176623761901e-07,
+      "loss": 2.6379,
+      "step": 3156
+    },
+    {
+      "epoch": 0.4087128200148882,
+      "grad_norm": 2.6454766740203417,
+      "learning_rate": 9.209688965281057e-07,
+      "loss": 2.6465,
+      "step": 3157
+    },
+    {
+      "epoch": 0.40884228242224163,
+      "grad_norm": 3.0469970888490097,
+      "learning_rate": 9.20720070222375e-07,
+      "loss": 2.7793,
+      "step": 3158
+    },
+    {
+      "epoch": 0.4089717448295951,
+      "grad_norm": 1.6882128429732328,
+      "learning_rate": 9.20471183509826e-07,
+      "loss": 2.4279,
+      "step": 3159
+    },
+    {
+      "epoch": 0.4091012072369486,
+      "grad_norm": 2.0014236740102196,
+      "learning_rate": 9.20222236441299e-07,
+      "loss": 2.5684,
+      "step": 3160
+    },
+    {
+      "epoch": 0.4091012072369486,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.6346324682235718,
+      "eval_runtime": 15.146,
+      "eval_samples_per_second": 2.905,
+      "eval_steps_per_second": 0.198,
+      "step": 3160
+    },
+    {
+      "epoch": 0.40923066964430205,
+      "grad_norm": 1.6349466783922244,
+      "learning_rate": 9.199732290676462e-07,
+      "loss": 2.249,
+      "step": 3161
+    },
+    {
+      "epoch": 0.4093601320516555,
+      "grad_norm": 2.421945812427243,
+      "learning_rate": 9.197241614397326e-07,
+      "loss": 2.6121,
+      "step": 3162
+    },
+    {
+      "epoch": 0.40948959445900895,
+      "grad_norm": 1.6542954455696464,
+      "learning_rate": 9.194750336084351e-07,
+      "loss": 2.356,
+      "step": 3163
+    },
+    {
+      "epoch": 0.4096190568663624,
+      "grad_norm": 1.9871650766107825,
+      "learning_rate": 9.192258456246434e-07,
+      "loss": 2.6946,
+      "step": 3164
+    },
+    {
+      "epoch": 0.4097485192737159,
+      "grad_norm": 2.4313645490977867,
+      "learning_rate": 9.189765975392589e-07,
+      "loss": 2.6299,
+      "step": 3165
+    },
+    {
+      "epoch": 0.4097485192737159,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.6305264234542847,
+      "eval_runtime": 15.5697,
+      "eval_samples_per_second": 2.826,
+      "eval_steps_per_second": 0.193,
+      "step": 3165
+    },
+    {
+      "epoch": 0.40987798168106937,
+      "grad_norm": 2.2305933453543796,
+      "learning_rate": 9.187272894031959e-07,
+      "loss": 2.5015,
+      "step": 3166
+    },
+    {
+      "epoch": 0.4100074440884228,
+      "grad_norm": 2.265110689173424,
+      "learning_rate": 9.184779212673803e-07,
+      "loss": 2.5164,
+      "step": 3167
+    },
+    {
+      "epoch": 0.4101369064957763,
+      "grad_norm": 1.7082441886990973,
+      "learning_rate": 9.182284931827509e-07,
+      "loss": 2.5229,
+      "step": 3168
+    },
+    {
+      "epoch": 0.41026636890312973,
+      "grad_norm": 2.488255843895349,
+      "learning_rate": 9.179790052002582e-07,
+      "loss": 2.9529,
+      "step": 3169
+    },
+    {
+      "epoch": 0.41039583131048324,
+      "grad_norm": 4.651526710767568,
+      "learning_rate": 9.177294573708652e-07,
+      "loss": 3.5449,
+      "step": 3170
+    },
+    {
+      "epoch": 0.41039583131048324,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.6325905323028564,
+      "eval_runtime": 14.282,
+      "eval_samples_per_second": 3.081,
+      "eval_steps_per_second": 0.21,
+      "step": 3170
+    },
+    {
+      "epoch": 0.4105252937178367,
+      "grad_norm": 2.4930524696703324,
+      "learning_rate": 9.174798497455475e-07,
+      "loss": 2.5156,
+      "step": 3171
+    },
+    {
+      "epoch": 0.41065475612519015,
+      "grad_norm": 1.5528227466020093,
+      "learning_rate": 9.172301823752919e-07,
+      "loss": 2.478,
+      "step": 3172
+    },
+    {
+      "epoch": 0.4107842185325436,
+      "grad_norm": 2.677157226310644,
+      "learning_rate": 9.169804553110983e-07,
+      "loss": 2.3484,
+      "step": 3173
+    },
+    {
+      "epoch": 0.41091368093989705,
+      "grad_norm": 1.9770795272313717,
+      "learning_rate": 9.167306686039786e-07,
+      "loss": 2.7358,
+      "step": 3174
+    },
+    {
+      "epoch": 0.41104314334725056,
+      "grad_norm": 3.0029429384104036,
+      "learning_rate": 9.164808223049567e-07,
+      "loss": 3.1643,
+      "step": 3175
+    },
+    {
+      "epoch": 0.41104314334725056,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.6319690942764282,
+      "eval_runtime": 13.4563,
+      "eval_samples_per_second": 3.27,
+      "eval_steps_per_second": 0.223,
+      "step": 3175
+    },
+    {
+      "epoch": 0.411172605754604,
+      "grad_norm": 1.4529269509986145,
+      "learning_rate": 9.162309164650689e-07,
+      "loss": 2.3882,
+      "step": 3176
+    },
+    {
+      "epoch": 0.41130206816195747,
+      "grad_norm": 2.138975165999896,
+      "learning_rate": 9.159809511353633e-07,
+      "loss": 2.5444,
+      "step": 3177
+    },
+    {
+      "epoch": 0.4114315305693109,
+      "grad_norm": 1.8334174052761953,
+      "learning_rate": 9.157309263669006e-07,
+      "loss": 2.4827,
+      "step": 3178
+    },
+    {
+      "epoch": 0.4115609929766644,
+      "grad_norm": 3.53241619656035,
+      "learning_rate": 9.154808422107534e-07,
+      "loss": 2.3521,
+      "step": 3179
+    },
+    {
+      "epoch": 0.4116904553840179,
+      "grad_norm": 1.4455666561151648,
+      "learning_rate": 9.152306987180062e-07,
+      "loss": 2.6204,
+      "step": 3180
+    },
+    {
+      "epoch": 0.4116904553840179,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.6283291578292847,
+      "eval_runtime": 13.5192,
+      "eval_samples_per_second": 3.255,
+      "eval_steps_per_second": 0.222,
+      "step": 3180
+    },
+    {
+      "epoch": 0.41181991779137134,
+      "grad_norm": 2.0674732159398754,
+      "learning_rate": 9.149804959397561e-07,
+      "loss": 2.5933,
+      "step": 3181
+    },
+    {
+      "epoch": 0.4119493801987248,
+      "grad_norm": 1.6228242051521333,
+      "learning_rate": 9.147302339271124e-07,
+      "loss": 2.5283,
+      "step": 3182
+    },
+    {
+      "epoch": 0.41207884260607824,
+      "grad_norm": 4.115468897474255,
+      "learning_rate": 9.144799127311954e-07,
+      "loss": 2.7917,
+      "step": 3183
+    },
+    {
+      "epoch": 0.4122083050134317,
+      "grad_norm": 1.7892069658451697,
+      "learning_rate": 9.142295324031389e-07,
+      "loss": 2.4824,
+      "step": 3184
+    },
+    {
+      "epoch": 0.4123377674207852,
+      "grad_norm": 1.8130998555584166,
+      "learning_rate": 9.139790929940881e-07,
+      "loss": 2.4771,
+      "step": 3185
+    },
+    {
+      "epoch": 0.4123377674207852,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.6305042505264282,
+      "eval_runtime": 14.3621,
+      "eval_samples_per_second": 3.064,
+      "eval_steps_per_second": 0.209,
+      "step": 3185
+    },
+    {
+      "epoch": 0.41246722982813866,
+      "grad_norm": 1.673273409526274,
+      "learning_rate": 9.137285945552002e-07,
+      "loss": 2.5122,
+      "step": 3186
+    },
+    {
+      "epoch": 0.4125966922354921,
+      "grad_norm": 1.565240354966385,
+      "learning_rate": 9.134780371376445e-07,
+      "loss": 2.5337,
+      "step": 3187
+    },
+    {
+      "epoch": 0.41272615464284557,
+      "grad_norm": 1.4944363374687455,
+      "learning_rate": 9.132274207926025e-07,
+      "loss": 2.4297,
+      "step": 3188
+    },
+    {
+      "epoch": 0.4128556170501991,
+      "grad_norm": 1.3390616257700647,
+      "learning_rate": 9.129767455712679e-07,
+      "loss": 2.4348,
+      "step": 3189
+    },
+    {
+      "epoch": 0.41298507945755253,
+      "grad_norm": 1.8215280567863243,
+      "learning_rate": 9.127260115248459e-07,
+      "loss": 2.3762,
+      "step": 3190
+    },
+    {
+      "epoch": 0.41298507945755253,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.6277521848678589,
+      "eval_runtime": 14.4618,
+      "eval_samples_per_second": 3.042,
+      "eval_steps_per_second": 0.207,
+      "step": 3190
+    },
+    {
+      "epoch": 0.413114541864906,
+      "grad_norm": 1.7308143449323479,
+      "learning_rate": 9.124752187045541e-07,
+      "loss": 2.5685,
+      "step": 3191
+    },
+    {
+      "epoch": 0.41324400427225944,
+      "grad_norm": 2.353118060280929,
+      "learning_rate": 9.122243671616218e-07,
+      "loss": 2.8176,
+      "step": 3192
+    },
+    {
+      "epoch": 0.4133734666796129,
+      "grad_norm": 1.4677428234050347,
+      "learning_rate": 9.119734569472911e-07,
+      "loss": 2.5789,
+      "step": 3193
+    },
+    {
+      "epoch": 0.4135029290869664,
+      "grad_norm": 1.5533853174449563,
+      "learning_rate": 9.117224881128151e-07,
+      "loss": 2.5591,
+      "step": 3194
+    },
+    {
+      "epoch": 0.41363239149431985,
+      "grad_norm": 1.8359214209643377,
+      "learning_rate": 9.114714607094591e-07,
+      "loss": 2.6831,
+      "step": 3195
+    },
+    {
+      "epoch": 0.41363239149431985,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.6283957958221436,
+      "eval_runtime": 13.8051,
+      "eval_samples_per_second": 3.187,
+      "eval_steps_per_second": 0.217,
+      "step": 3195
+    },
+    {
+      "epoch": 0.4137618539016733,
+      "grad_norm": 2.439111361132348,
+      "learning_rate": 9.112203747885007e-07,
+      "loss": 2.9541,
+      "step": 3196
+    },
+    {
+      "epoch": 0.41389131630902676,
+      "grad_norm": 2.2204867672810744,
+      "learning_rate": 9.109692304012295e-07,
+      "loss": 2.6792,
+      "step": 3197
+    },
+    {
+      "epoch": 0.4140207787163802,
+      "grad_norm": 1.519237431008716,
+      "learning_rate": 9.107180275989466e-07,
+      "loss": 2.3406,
+      "step": 3198
+    },
+    {
+      "epoch": 0.4141502411237337,
+      "grad_norm": 2.9943164027139324,
+      "learning_rate": 9.104667664329654e-07,
+      "loss": 2.9312,
+      "step": 3199
+    },
+    {
+      "epoch": 0.4142797035310872,
+      "grad_norm": 3.385237029492058,
+      "learning_rate": 9.102154469546109e-07,
+      "loss": 2.8723,
+      "step": 3200
+    },
+    {
+      "epoch": 0.4142797035310872,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.6263316869735718,
+      "eval_runtime": 15.6112,
+      "eval_samples_per_second": 2.818,
+      "eval_steps_per_second": 0.192,
+      "step": 3200
+    },
+    {
+      "epoch": 0.4144091659384406,
+      "grad_norm": 2.0605726991878366,
+      "learning_rate": 9.099640692152203e-07,
+      "loss": 2.3472,
+      "step": 3201
+    },
+    {
+      "epoch": 0.4145386283457941,
+      "grad_norm": 1.9531322408622698,
+      "learning_rate": 9.097126332661427e-07,
+      "loss": 2.5063,
+      "step": 3202
+    },
+    {
+      "epoch": 0.41466809075314753,
+      "grad_norm": 1.7827777854279698,
+      "learning_rate": 9.094611391587388e-07,
+      "loss": 2.6777,
+      "step": 3203
+    },
+    {
+      "epoch": 0.41479755316050104,
+      "grad_norm": 1.6130381349912482,
+      "learning_rate": 9.092095869443814e-07,
+      "loss": 2.3628,
+      "step": 3204
+    },
+    {
+      "epoch": 0.4149270155678545,
+      "grad_norm": 1.436695709810826,
+      "learning_rate": 9.089579766744554e-07,
+      "loss": 2.3994,
+      "step": 3205
+    },
+    {
+      "epoch": 0.4149270155678545,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.6241787672042847,
+      "eval_runtime": 14.7091,
+      "eval_samples_per_second": 2.991,
+      "eval_steps_per_second": 0.204,
+      "step": 3205
+    },
+    {
+      "epoch": 0.41505647797520795,
+      "grad_norm": 2.0319122385459605,
+      "learning_rate": 9.08706308400357e-07,
+      "loss": 2.792,
+      "step": 3206
+    },
+    {
+      "epoch": 0.4151859403825614,
+      "grad_norm": 2.1304839797660797,
+      "learning_rate": 9.084545821734946e-07,
+      "loss": 2.5764,
+      "step": 3207
+    },
+    {
+      "epoch": 0.41531540278991486,
+      "grad_norm": 3.627352745893683,
+      "learning_rate": 9.082027980452884e-07,
+      "loss": 2.564,
+      "step": 3208
+    },
+    {
+      "epoch": 0.41544486519726836,
+      "grad_norm": 1.8001616268210472,
+      "learning_rate": 9.079509560671706e-07,
+      "loss": 2.4189,
+      "step": 3209
+    },
+    {
+      "epoch": 0.4155743276046218,
+      "grad_norm": 2.232693616134311,
+      "learning_rate": 9.076990562905847e-07,
+      "loss": 2.699,
+      "step": 3210
+    },
+    {
+      "epoch": 0.4155743276046218,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.6234463453292847,
+      "eval_runtime": 19.7147,
+      "eval_samples_per_second": 2.232,
+      "eval_steps_per_second": 0.152,
+      "step": 3210
+    },
+    {
+      "epoch": 0.41570379001197527,
+      "grad_norm": 1.5920193896463253,
+      "learning_rate": 9.074470987669866e-07,
+      "loss": 2.4985,
+      "step": 3211
+    },
+    {
+      "epoch": 0.4158332524193287,
+      "grad_norm": 3.34654975967097,
+      "learning_rate": 9.071950835478437e-07,
+      "loss": 2.5562,
+      "step": 3212
+    },
+    {
+      "epoch": 0.4159627148266822,
+      "grad_norm": 2.101812673255751,
+      "learning_rate": 9.069430106846349e-07,
+      "loss": 2.4448,
+      "step": 3213
+    },
+    {
+      "epoch": 0.4160921772340357,
+      "grad_norm": 1.528590510355101,
+      "learning_rate": 9.066908802288518e-07,
+      "loss": 2.2534,
+      "step": 3214
+    },
+    {
+      "epoch": 0.41622163964138914,
+      "grad_norm": 1.7738671714681336,
+      "learning_rate": 9.064386922319965e-07,
+      "loss": 2.4614,
+      "step": 3215
+    },
+    {
+      "epoch": 0.41622163964138914,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.6248224973678589,
+      "eval_runtime": 17.8083,
+      "eval_samples_per_second": 2.471,
+      "eval_steps_per_second": 0.168,
+      "step": 3215
+    },
+    {
+      "epoch": 0.4163511020487426,
+      "grad_norm": 1.8341224537735463,
+      "learning_rate": 9.061864467455839e-07,
+      "loss": 2.8481,
+      "step": 3216
+    },
+    {
+      "epoch": 0.41648056445609605,
+      "grad_norm": 1.7480148092472603,
+      "learning_rate": 9.059341438211403e-07,
+      "loss": 2.5042,
+      "step": 3217
+    },
+    {
+      "epoch": 0.4166100268634495,
+      "grad_norm": 1.3938576142248862,
+      "learning_rate": 9.056817835102035e-07,
+      "loss": 2.5427,
+      "step": 3218
+    },
+    {
+      "epoch": 0.416739489270803,
+      "grad_norm": 2.7013522583805467,
+      "learning_rate": 9.054293658643228e-07,
+      "loss": 2.6804,
+      "step": 3219
+    },
+    {
+      "epoch": 0.41686895167815646,
+      "grad_norm": 7.9317473198485935,
+      "learning_rate": 9.051768909350606e-07,
+      "loss": 2.9421,
+      "step": 3220
+    },
+    {
+      "epoch": 0.41686895167815646,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.6247780323028564,
+      "eval_runtime": 13.8241,
+      "eval_samples_per_second": 3.183,
+      "eval_steps_per_second": 0.217,
+      "step": 3220
+    },
+    {
+      "epoch": 0.4169984140855099,
+      "grad_norm": 1.9709065424588976,
+      "learning_rate": 9.049243587739894e-07,
+      "loss": 2.8203,
+      "step": 3221
+    },
+    {
+      "epoch": 0.41712787649286337,
+      "grad_norm": 1.9428330311588493,
+      "learning_rate": 9.046717694326942e-07,
+      "loss": 2.4033,
+      "step": 3222
+    },
+    {
+      "epoch": 0.4172573389002168,
+      "grad_norm": 1.6212262976038094,
+      "learning_rate": 9.044191229627712e-07,
+      "loss": 2.511,
+      "step": 3223
+    },
+    {
+      "epoch": 0.41738680130757033,
+      "grad_norm": 3.393026697002203,
+      "learning_rate": 9.041664194158288e-07,
+      "loss": 2.4211,
+      "step": 3224
+    },
+    {
+      "epoch": 0.4175162637149238,
+      "grad_norm": 2.302173837703524,
+      "learning_rate": 9.039136588434869e-07,
+      "loss": 2.4407,
+      "step": 3225
+    },
+    {
+      "epoch": 0.4175162637149238,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.6212047338485718,
+      "eval_runtime": 14.0856,
+      "eval_samples_per_second": 3.124,
+      "eval_steps_per_second": 0.213,
+      "step": 3225
+    },
+    {
+      "epoch": 0.41764572612227724,
+      "grad_norm": 1.9221139960252982,
+      "learning_rate": 9.036608412973768e-07,
+      "loss": 2.5552,
+      "step": 3226
+    },
+    {
+      "epoch": 0.4177751885296307,
+      "grad_norm": 2.476350930165937,
+      "learning_rate": 9.034079668291418e-07,
+      "loss": 2.5461,
+      "step": 3227
+    },
+    {
+      "epoch": 0.41790465093698415,
+      "grad_norm": 1.47229003898402,
+      "learning_rate": 9.031550354904362e-07,
+      "loss": 2.4373,
+      "step": 3228
+    },
+    {
+      "epoch": 0.41803411334433765,
+      "grad_norm": 2.183663206223865,
+      "learning_rate": 9.029020473329268e-07,
+      "loss": 2.8606,
+      "step": 3229
+    },
+    {
+      "epoch": 0.4181635757516911,
+      "grad_norm": 2.1246530382197535,
+      "learning_rate": 9.026490024082915e-07,
+      "loss": 2.3906,
+      "step": 3230
+    },
+    {
+      "epoch": 0.4181635757516911,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.6206942796707153,
+      "eval_runtime": 14.2383,
+      "eval_samples_per_second": 3.09,
+      "eval_steps_per_second": 0.211,
+      "step": 3230
+    },
+    {
+      "epoch": 0.41829303815904456,
+      "grad_norm": 1.9113312190773661,
+      "learning_rate": 9.023959007682197e-07,
+      "loss": 2.3535,
+      "step": 3231
+    },
+    {
+      "epoch": 0.418422500566398,
+      "grad_norm": 2.314864199281409,
+      "learning_rate": 9.021427424644127e-07,
+      "loss": 2.6824,
+      "step": 3232
+    },
+    {
+      "epoch": 0.4185519629737515,
+      "grad_norm": 1.8557837018790813,
+      "learning_rate": 9.018895275485828e-07,
+      "loss": 2.603,
+      "step": 3233
+    },
+    {
+      "epoch": 0.418681425381105,
+      "grad_norm": 1.234498423406555,
+      "learning_rate": 9.01636256072455e-07,
+      "loss": 2.4275,
+      "step": 3234
+    },
+    {
+      "epoch": 0.41881088778845843,
+      "grad_norm": 2.1039299027435256,
+      "learning_rate": 9.013829280877647e-07,
+      "loss": 2.4912,
+      "step": 3235
+    },
+    {
+      "epoch": 0.41881088778845843,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.6212269067764282,
+      "eval_runtime": 13.6099,
+      "eval_samples_per_second": 3.233,
+      "eval_steps_per_second": 0.22,
+      "step": 3235
+    },
+    {
+      "epoch": 0.4189403501958119,
+      "grad_norm": 3.4168389852644068,
+      "learning_rate": 9.011295436462592e-07,
+      "loss": 2.7859,
+      "step": 3236
+    },
+    {
+      "epoch": 0.41906981260316534,
+      "grad_norm": 2.11613200983246,
+      "learning_rate": 9.008761027996973e-07,
+      "loss": 2.5972,
+      "step": 3237
+    },
+    {
+      "epoch": 0.41919927501051885,
+      "grad_norm": 3.345025552839822,
+      "learning_rate": 9.006226055998498e-07,
+      "loss": 2.5752,
+      "step": 3238
+    },
+    {
+      "epoch": 0.4193287374178723,
+      "grad_norm": 1.633146160231083,
+      "learning_rate": 9.003690520984987e-07,
+      "loss": 2.6104,
+      "step": 3239
+    },
+    {
+      "epoch": 0.41945819982522575,
+      "grad_norm": 2.5802724005372504,
+      "learning_rate": 9.001154423474372e-07,
+      "loss": 2.5286,
+      "step": 3240
+    },
+    {
+      "epoch": 0.41945819982522575,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.622025966644287,
+      "eval_runtime": 13.7148,
+      "eval_samples_per_second": 3.208,
+      "eval_steps_per_second": 0.219,
+      "step": 3240
+    },
+    {
+      "epoch": 0.4195876622325792,
+      "grad_norm": 2.5670736306364423,
+      "learning_rate": 8.9986177639847e-07,
+      "loss": 2.5991,
+      "step": 3241
+    },
+    {
+      "epoch": 0.41971712463993266,
+      "grad_norm": 2.7100968884954866,
+      "learning_rate": 8.996080543034139e-07,
+      "loss": 2.6877,
+      "step": 3242
+    },
+    {
+      "epoch": 0.41984658704728617,
+      "grad_norm": 2.2197656405346446,
+      "learning_rate": 8.993542761140967e-07,
+      "loss": 2.7317,
+      "step": 3243
+    },
+    {
+      "epoch": 0.4199760494546396,
+      "grad_norm": 2.5189336382272995,
+      "learning_rate": 8.991004418823576e-07,
+      "loss": 2.8982,
+      "step": 3244
+    },
+    {
+      "epoch": 0.4201055118619931,
+      "grad_norm": 2.2333111446540888,
+      "learning_rate": 8.988465516600476e-07,
+      "loss": 2.8354,
+      "step": 3245
+    },
+    {
+      "epoch": 0.4201055118619931,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.6212047338485718,
+      "eval_runtime": 14.9291,
+      "eval_samples_per_second": 2.947,
+      "eval_steps_per_second": 0.201,
+      "step": 3245
+    },
+    {
+      "epoch": 0.42023497426934653,
+      "grad_norm": 3.573335572794352,
+      "learning_rate": 8.985926054990287e-07,
+      "loss": 2.5356,
+      "step": 3246
+    },
+    {
+      "epoch": 0.4203644366767,
+      "grad_norm": 1.8581797741235633,
+      "learning_rate": 8.983386034511748e-07,
+      "loss": 2.5354,
+      "step": 3247
+    },
+    {
+      "epoch": 0.4204938990840535,
+      "grad_norm": 1.490822872434631,
+      "learning_rate": 8.980845455683706e-07,
+      "loss": 2.4744,
+      "step": 3248
+    },
+    {
+      "epoch": 0.42062336149140694,
+      "grad_norm": 1.433193793646382,
+      "learning_rate": 8.978304319025131e-07,
+      "loss": 2.3569,
+      "step": 3249
+    },
+    {
+      "epoch": 0.4207528238987604,
+      "grad_norm": 1.9253861712432947,
+      "learning_rate": 8.975762625055096e-07,
+      "loss": 2.5391,
+      "step": 3250
+    },
+    {
+      "epoch": 0.4207528238987604,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.6213600635528564,
+      "eval_runtime": 13.9722,
+      "eval_samples_per_second": 3.149,
+      "eval_steps_per_second": 0.215,
+      "step": 3250
+    },
+    {
+      "epoch": 0.42088228630611385,
+      "grad_norm": 2.3023110306251238,
+      "learning_rate": 8.973220374292799e-07,
+      "loss": 2.2634,
+      "step": 3251
+    },
+    {
+      "epoch": 0.4210117487134673,
+      "grad_norm": 1.6665098237002203,
+      "learning_rate": 8.970677567257541e-07,
+      "loss": 2.5813,
+      "step": 3252
+    },
+    {
+      "epoch": 0.4211412111208208,
+      "grad_norm": 1.524505509354147,
+      "learning_rate": 8.968134204468744e-07,
+      "loss": 2.4338,
+      "step": 3253
+    },
+    {
+      "epoch": 0.42127067352817427,
+      "grad_norm": 1.4307524908907743,
+      "learning_rate": 8.965590286445943e-07,
+      "loss": 2.2358,
+      "step": 3254
+    },
+    {
+      "epoch": 0.4214001359355277,
+      "grad_norm": 2.6827791641067913,
+      "learning_rate": 8.963045813708781e-07,
+      "loss": 2.7305,
+      "step": 3255
+    },
+    {
+      "epoch": 0.4214001359355277,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.6197398900985718,
+      "eval_runtime": 13.6866,
+      "eval_samples_per_second": 3.215,
+      "eval_steps_per_second": 0.219,
+      "step": 3255
+    },
+    {
+      "epoch": 0.4215295983428812,
+      "grad_norm": 3.2374998887108175,
+      "learning_rate": 8.960500786777024e-07,
+      "loss": 2.6328,
+      "step": 3256
+    },
+    {
+      "epoch": 0.4216590607502346,
+      "grad_norm": 4.542618127048336,
+      "learning_rate": 8.957955206170542e-07,
+      "loss": 2.8135,
+      "step": 3257
+    },
+    {
+      "epoch": 0.42178852315758814,
+      "grad_norm": 1.9912255441854707,
+      "learning_rate": 8.95540907240932e-07,
+      "loss": 2.7024,
+      "step": 3258
+    },
+    {
+      "epoch": 0.4219179855649416,
+      "grad_norm": 2.6951357544798276,
+      "learning_rate": 8.952862386013458e-07,
+      "loss": 2.6274,
+      "step": 3259
+    },
+    {
+      "epoch": 0.42204744797229504,
+      "grad_norm": 2.316426134433997,
+      "learning_rate": 8.950315147503169e-07,
+      "loss": 2.4709,
+      "step": 3260
+    },
+    {
+      "epoch": 0.42204744797229504,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.6160777807235718,
+      "eval_runtime": 15.4201,
+      "eval_samples_per_second": 2.853,
+      "eval_steps_per_second": 0.195,
+      "step": 3260
+    },
+    {
+      "epoch": 0.4221769103796485,
+      "grad_norm": 1.9462753706412466,
+      "learning_rate": 8.947767357398779e-07,
+      "loss": 2.7262,
+      "step": 3261
+    },
+    {
+      "epoch": 0.42230637278700195,
+      "grad_norm": 3.678258936761399,
+      "learning_rate": 8.945219016220725e-07,
+      "loss": 2.8623,
+      "step": 3262
+    },
+    {
+      "epoch": 0.42243583519435546,
+      "grad_norm": 2.0758018068650768,
+      "learning_rate": 8.94267012448956e-07,
+      "loss": 2.6953,
+      "step": 3263
+    },
+    {
+      "epoch": 0.4225652976017089,
+      "grad_norm": 1.6535743939177947,
+      "learning_rate": 8.94012068272594e-07,
+      "loss": 2.4502,
+      "step": 3264
+    },
+    {
+      "epoch": 0.42269476000906236,
+      "grad_norm": 1.668886492377874,
+      "learning_rate": 8.937570691450646e-07,
+      "loss": 2.5532,
+      "step": 3265
+    },
+    {
+      "epoch": 0.42269476000906236,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.619140625,
+      "eval_runtime": 14.21,
+      "eval_samples_per_second": 3.096,
+      "eval_steps_per_second": 0.211,
+      "step": 3265
+    },
+    {
+      "epoch": 0.4228242224164158,
+      "grad_norm": 2.1617070471768076,
+      "learning_rate": 8.935020151184563e-07,
+      "loss": 2.5706,
+      "step": 3266
+    },
+    {
+      "epoch": 0.42295368482376927,
+      "grad_norm": 2.1647015531278186,
+      "learning_rate": 8.932469062448693e-07,
+      "loss": 2.71,
+      "step": 3267
+    },
+    {
+      "epoch": 0.4230831472311228,
+      "grad_norm": 1.962576888316768,
+      "learning_rate": 8.929917425764146e-07,
+      "loss": 2.3623,
+      "step": 3268
+    },
+    {
+      "epoch": 0.42321260963847623,
+      "grad_norm": 2.019840024360644,
+      "learning_rate": 8.927365241652142e-07,
+      "loss": 2.592,
+      "step": 3269
+    },
+    {
+      "epoch": 0.4233420720458297,
+      "grad_norm": 1.9404260348685092,
+      "learning_rate": 8.924812510634021e-07,
+      "loss": 2.5413,
+      "step": 3270
+    },
+    {
+      "epoch": 0.4233420720458297,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.6184748411178589,
+      "eval_runtime": 15.0263,
+      "eval_samples_per_second": 2.928,
+      "eval_steps_per_second": 0.2,
+      "step": 3270
+    },
+    {
+      "epoch": 0.42347153445318314,
+      "grad_norm": 2.1198821501137255,
+      "learning_rate": 8.92225923323123e-07,
+      "loss": 2.375,
+      "step": 3271
+    },
+    {
+      "epoch": 0.42360099686053665,
+      "grad_norm": 2.5545400151560305,
+      "learning_rate": 8.919705409965326e-07,
+      "loss": 2.5815,
+      "step": 3272
+    },
+    {
+      "epoch": 0.4237304592678901,
+      "grad_norm": 1.8061257156703898,
+      "learning_rate": 8.917151041357979e-07,
+      "loss": 2.5798,
+      "step": 3273
+    },
+    {
+      "epoch": 0.42385992167524356,
+      "grad_norm": 1.528558582689481,
+      "learning_rate": 8.914596127930972e-07,
+      "loss": 2.4153,
+      "step": 3274
+    },
+    {
+      "epoch": 0.423989384082597,
+      "grad_norm": 2.439987648451466,
+      "learning_rate": 8.912040670206196e-07,
+      "loss": 2.6223,
+      "step": 3275
+    },
+    {
+      "epoch": 0.423989384082597,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.6185413599014282,
+      "eval_runtime": 13.6,
+      "eval_samples_per_second": 3.235,
+      "eval_steps_per_second": 0.221,
+      "step": 3275
+    },
+    {
+      "epoch": 0.42411884648995046,
+      "grad_norm": 3.2820873926307863,
+      "learning_rate": 8.909484668705656e-07,
+      "loss": 2.8037,
+      "step": 3276
+    },
+    {
+      "epoch": 0.42424830889730397,
+      "grad_norm": 1.8007726185049484,
+      "learning_rate": 8.906928123951468e-07,
+      "loss": 2.4868,
+      "step": 3277
+    },
+    {
+      "epoch": 0.4243777713046574,
+      "grad_norm": 1.3391092303121688,
+      "learning_rate": 8.904371036465856e-07,
+      "loss": 2.3364,
+      "step": 3278
+    },
+    {
+      "epoch": 0.4245072337120109,
+      "grad_norm": 4.157233652245847,
+      "learning_rate": 8.90181340677116e-07,
+      "loss": 3.1763,
+      "step": 3279
+    },
+    {
+      "epoch": 0.42463669611936433,
+      "grad_norm": 1.4170572778663098,
+      "learning_rate": 8.899255235389824e-07,
+      "loss": 2.4294,
+      "step": 3280
+    },
+    {
+      "epoch": 0.42463669611936433,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.6171208620071411,
+      "eval_runtime": 13.1351,
+      "eval_samples_per_second": 3.35,
+      "eval_steps_per_second": 0.228,
+      "step": 3280
+    },
+    {
+      "epoch": 0.4247661585267178,
+      "grad_norm": 2.130560649596276,
+      "learning_rate": 8.896696522844408e-07,
+      "loss": 2.4863,
+      "step": 3281
+    },
+    {
+      "epoch": 0.4248956209340713,
+      "grad_norm": 1.3699061765897067,
+      "learning_rate": 8.894137269657582e-07,
+      "loss": 2.5444,
+      "step": 3282
+    },
+    {
+      "epoch": 0.42502508334142475,
+      "grad_norm": 1.5999821864135884,
+      "learning_rate": 8.891577476352126e-07,
+      "loss": 2.2563,
+      "step": 3283
+    },
+    {
+      "epoch": 0.4251545457487782,
+      "grad_norm": 1.7772694657872636,
+      "learning_rate": 8.889017143450926e-07,
+      "loss": 2.8235,
+      "step": 3284
+    },
+    {
+      "epoch": 0.42528400815613165,
+      "grad_norm": 2.937525886820933,
+      "learning_rate": 8.886456271476986e-07,
+      "loss": 2.5205,
+      "step": 3285
+    },
+    {
+      "epoch": 0.42528400815613165,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.6163440942764282,
+      "eval_runtime": 13.6553,
+      "eval_samples_per_second": 3.222,
+      "eval_steps_per_second": 0.22,
+      "step": 3285
+    },
+    {
+      "epoch": 0.4254134705634851,
+      "grad_norm": 1.5263810748228492,
+      "learning_rate": 8.883894860953413e-07,
+      "loss": 2.4785,
+      "step": 3286
+    },
+    {
+      "epoch": 0.4255429329708386,
+      "grad_norm": 1.4393148016264663,
+      "learning_rate": 8.881332912403426e-07,
+      "loss": 2.4431,
+      "step": 3287
+    },
+    {
+      "epoch": 0.42567239537819207,
+      "grad_norm": 2.9434749987556366,
+      "learning_rate": 8.878770426350358e-07,
+      "loss": 2.7056,
+      "step": 3288
+    },
+    {
+      "epoch": 0.4258018577855455,
+      "grad_norm": 4.360730596043665,
+      "learning_rate": 8.87620740331765e-07,
+      "loss": 2.6819,
+      "step": 3289
+    },
+    {
+      "epoch": 0.425931320192899,
+      "grad_norm": 1.674021010529709,
+      "learning_rate": 8.873643843828847e-07,
+      "loss": 2.5413,
+      "step": 3290
+    },
+    {
+      "epoch": 0.425931320192899,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.6112614870071411,
+      "eval_runtime": 13.7538,
+      "eval_samples_per_second": 3.199,
+      "eval_steps_per_second": 0.218,
+      "step": 3290
+    },
+    {
+      "epoch": 0.42606078260025243,
+      "grad_norm": 2.3355657717651868,
+      "learning_rate": 8.87107974840761e-07,
+      "loss": 2.4731,
+      "step": 3291
+    },
+    {
+      "epoch": 0.42619024500760594,
+      "grad_norm": 1.525437938770551,
+      "learning_rate": 8.868515117577707e-07,
+      "loss": 2.5164,
+      "step": 3292
+    },
+    {
+      "epoch": 0.4263197074149594,
+      "grad_norm": 3.8956202128910533,
+      "learning_rate": 8.865949951863019e-07,
+      "loss": 2.6938,
+      "step": 3293
+    },
+    {
+      "epoch": 0.42644916982231285,
+      "grad_norm": 2.5048905859058133,
+      "learning_rate": 8.863384251787528e-07,
+      "loss": 2.6521,
+      "step": 3294
+    },
+    {
+      "epoch": 0.4265786322296663,
+      "grad_norm": 1.746924905531486,
+      "learning_rate": 8.860818017875334e-07,
+      "loss": 2.3784,
+      "step": 3295
+    },
+    {
+      "epoch": 0.4265786322296663,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.611328125,
+      "eval_runtime": 13.4273,
+      "eval_samples_per_second": 3.277,
+      "eval_steps_per_second": 0.223,
+      "step": 3295
+    },
+    {
+      "epoch": 0.42670809463701975,
+      "grad_norm": 3.1409009993341206,
+      "learning_rate": 8.85825125065064e-07,
+      "loss": 2.4053,
+      "step": 3296
+    },
+    {
+      "epoch": 0.42683755704437326,
+      "grad_norm": 3.2568328795019545,
+      "learning_rate": 8.855683950637761e-07,
+      "loss": 2.8015,
+      "step": 3297
+    },
+    {
+      "epoch": 0.4269670194517267,
+      "grad_norm": 1.9369177785713558,
+      "learning_rate": 8.853116118361121e-07,
+      "loss": 2.4966,
+      "step": 3298
+    },
+    {
+      "epoch": 0.42709648185908017,
+      "grad_norm": 2.5107018877847325,
+      "learning_rate": 8.850547754345251e-07,
+      "loss": 2.415,
+      "step": 3299
+    },
+    {
+      "epoch": 0.4272259442664336,
+      "grad_norm": 2.0471591360427315,
+      "learning_rate": 8.847978859114792e-07,
+      "loss": 2.3945,
+      "step": 3300
+    },
+    {
+      "epoch": 0.4272259442664336,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.6119939088821411,
+      "eval_runtime": 13.967,
+      "eval_samples_per_second": 3.15,
+      "eval_steps_per_second": 0.215,
+      "step": 3300
+    },
+    {
+      "epoch": 0.4273554066737871,
+      "grad_norm": 2.3780266166788695,
+      "learning_rate": 8.845409433194491e-07,
+      "loss": 2.9751,
+      "step": 3301
+    },
+    {
+      "epoch": 0.4274848690811406,
+      "grad_norm": 2.719911936365414,
+      "learning_rate": 8.842839477109208e-07,
+      "loss": 2.4644,
+      "step": 3302
+    },
+    {
+      "epoch": 0.42761433148849404,
+      "grad_norm": 2.2835477644242106,
+      "learning_rate": 8.840268991383908e-07,
+      "loss": 2.6497,
+      "step": 3303
+    },
+    {
+      "epoch": 0.4277437938958475,
+      "grad_norm": 1.8329933799909754,
+      "learning_rate": 8.837697976543663e-07,
+      "loss": 2.5146,
+      "step": 3304
+    },
+    {
+      "epoch": 0.42787325630320094,
+      "grad_norm": 3.0023311875579655,
+      "learning_rate": 8.835126433113655e-07,
+      "loss": 2.6895,
+      "step": 3305
+    },
+    {
+      "epoch": 0.42787325630320094,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.6106178760528564,
+      "eval_runtime": 14.2485,
+      "eval_samples_per_second": 3.088,
+      "eval_steps_per_second": 0.211,
+      "step": 3305
+    },
+    {
+      "epoch": 0.4280027187105544,
+      "grad_norm": 2.471240027748569,
+      "learning_rate": 8.832554361619175e-07,
+      "loss": 2.6199,
+      "step": 3306
+    },
+    {
+      "epoch": 0.4281321811179079,
+      "grad_norm": 2.5092914791181524,
+      "learning_rate": 8.829981762585624e-07,
+      "loss": 2.4937,
+      "step": 3307
+    },
+    {
+      "epoch": 0.42826164352526136,
+      "grad_norm": 1.8282969143604033,
+      "learning_rate": 8.827408636538501e-07,
+      "loss": 2.3113,
+      "step": 3308
+    },
+    {
+      "epoch": 0.4283911059326148,
+      "grad_norm": 2.365926070476587,
+      "learning_rate": 8.824834984003422e-07,
+      "loss": 2.3408,
+      "step": 3309
+    },
+    {
+      "epoch": 0.42852056833996827,
+      "grad_norm": 1.681006429764802,
+      "learning_rate": 8.82226080550611e-07,
+      "loss": 2.6309,
+      "step": 3310
+    },
+    {
+      "epoch": 0.42852056833996827,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.6099520921707153,
+      "eval_runtime": 13.9877,
+      "eval_samples_per_second": 3.146,
+      "eval_steps_per_second": 0.214,
+      "step": 3310
+    },
+    {
+      "epoch": 0.4286500307473217,
+      "grad_norm": 1.7909116590903777,
+      "learning_rate": 8.819686101572389e-07,
+      "loss": 2.4551,
+      "step": 3311
+    },
+    {
+      "epoch": 0.42877949315467523,
+      "grad_norm": 1.8042349247125555,
+      "learning_rate": 8.817110872728199e-07,
+      "loss": 2.7805,
+      "step": 3312
+    },
+    {
+      "epoch": 0.4289089555620287,
+      "grad_norm": 1.5481426122418513,
+      "learning_rate": 8.814535119499579e-07,
+      "loss": 2.7432,
+      "step": 3313
+    },
+    {
+      "epoch": 0.42903841796938214,
+      "grad_norm": 1.961620933477069,
+      "learning_rate": 8.811958842412679e-07,
+      "loss": 2.5015,
+      "step": 3314
+    },
+    {
+      "epoch": 0.4291678803767356,
+      "grad_norm": 2.231523269839053,
+      "learning_rate": 8.80938204199376e-07,
+      "loss": 2.584,
+      "step": 3315
+    },
+    {
+      "epoch": 0.4291678803767356,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.61279296875,
+      "eval_runtime": 13.9822,
+      "eval_samples_per_second": 3.147,
+      "eval_steps_per_second": 0.215,
+      "step": 3315
+    },
+    {
+      "epoch": 0.4292973427840891,
+      "grad_norm": 1.923431514811729,
+      "learning_rate": 8.80680471876918e-07,
+      "loss": 2.5432,
+      "step": 3316
+    },
+    {
+      "epoch": 0.42942680519144255,
+      "grad_norm": 1.7322720945534793,
+      "learning_rate": 8.804226873265412e-07,
+      "loss": 2.48,
+      "step": 3317
+    },
+    {
+      "epoch": 0.429556267598796,
+      "grad_norm": 2.204274486264051,
+      "learning_rate": 8.801648506009033e-07,
+      "loss": 2.4814,
+      "step": 3318
+    },
+    {
+      "epoch": 0.42968573000614946,
+      "grad_norm": 1.6943207208619377,
+      "learning_rate": 8.799069617526728e-07,
+      "loss": 2.5107,
+      "step": 3319
+    },
+    {
+      "epoch": 0.4298151924135029,
+      "grad_norm": 2.7090515846357803,
+      "learning_rate": 8.796490208345288e-07,
+      "loss": 2.4072,
+      "step": 3320
+    },
+    {
+      "epoch": 0.4298151924135029,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.6091974973678589,
+      "eval_runtime": 14.5239,
+      "eval_samples_per_second": 3.029,
+      "eval_steps_per_second": 0.207,
+      "step": 3320
+    },
+    {
+      "epoch": 0.4299446548208564,
+      "grad_norm": 1.626923485269989,
+      "learning_rate": 8.793910278991608e-07,
+      "loss": 2.4314,
+      "step": 3321
+    },
+    {
+      "epoch": 0.4300741172282099,
+      "grad_norm": 3.1800503795137804,
+      "learning_rate": 8.791329829992688e-07,
+      "loss": 2.9187,
+      "step": 3322
+    },
+    {
+      "epoch": 0.4302035796355633,
+      "grad_norm": 1.3641566312040845,
+      "learning_rate": 8.78874886187564e-07,
+      "loss": 2.2957,
+      "step": 3323
+    },
+    {
+      "epoch": 0.4303330420429168,
+      "grad_norm": 2.7601740894497424,
+      "learning_rate": 8.786167375167681e-07,
+      "loss": 2.7703,
+      "step": 3324
+    },
+    {
+      "epoch": 0.43046250445027023,
+      "grad_norm": 2.1503164425255212,
+      "learning_rate": 8.783585370396127e-07,
+      "loss": 2.7195,
+      "step": 3325
+    },
+    {
+      "epoch": 0.43046250445027023,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.6126598119735718,
+      "eval_runtime": 14.8409,
+      "eval_samples_per_second": 2.965,
+      "eval_steps_per_second": 0.202,
+      "step": 3325
+    },
+    {
+      "epoch": 0.43059196685762374,
+      "grad_norm": 1.7714639521996063,
+      "learning_rate": 8.781002848088411e-07,
+      "loss": 2.5303,
+      "step": 3326
+    },
+    {
+      "epoch": 0.4307214292649772,
+      "grad_norm": 2.1597747392886877,
+      "learning_rate": 8.77841980877206e-07,
+      "loss": 2.666,
+      "step": 3327
+    },
+    {
+      "epoch": 0.43085089167233065,
+      "grad_norm": 1.1357448939177524,
+      "learning_rate": 8.775836252974715e-07,
+      "loss": 2.1963,
+      "step": 3328
+    },
+    {
+      "epoch": 0.4309803540796841,
+      "grad_norm": 2.134878756173607,
+      "learning_rate": 8.773252181224121e-07,
+      "loss": 2.5022,
+      "step": 3329
+    },
+    {
+      "epoch": 0.43110981648703756,
+      "grad_norm": 1.6217727308247625,
+      "learning_rate": 8.770667594048121e-07,
+      "loss": 2.2305,
+      "step": 3330
+    },
+    {
+      "epoch": 0.43110981648703756,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.612016201019287,
+      "eval_runtime": 14.5759,
+      "eval_samples_per_second": 3.019,
+      "eval_steps_per_second": 0.206,
+      "step": 3330
+    },
+    {
+      "epoch": 0.43123927889439106,
+      "grad_norm": 4.593393228292561,
+      "learning_rate": 8.768082491974675e-07,
+      "loss": 2.9822,
+      "step": 3331
+    },
+    {
+      "epoch": 0.4313687413017445,
+      "grad_norm": 2.543337754708295,
+      "learning_rate": 8.765496875531841e-07,
+      "loss": 2.6838,
+      "step": 3332
+    },
+    {
+      "epoch": 0.43149820370909797,
+      "grad_norm": 1.31875583086638,
+      "learning_rate": 8.762910745247784e-07,
+      "loss": 2.3336,
+      "step": 3333
+    },
+    {
+      "epoch": 0.4316276661164514,
+      "grad_norm": 1.8618292485888146,
+      "learning_rate": 8.760324101650772e-07,
+      "loss": 2.4878,
+      "step": 3334
+    },
+    {
+      "epoch": 0.4317571285238049,
+      "grad_norm": 1.5730626090404043,
+      "learning_rate": 8.757736945269183e-07,
+      "loss": 2.4136,
+      "step": 3335
+    },
+    {
+      "epoch": 0.4317571285238049,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.6099299192428589,
+      "eval_runtime": 15.1485,
+      "eval_samples_per_second": 2.905,
+      "eval_steps_per_second": 0.198,
+      "step": 3335
+    },
+    {
+      "epoch": 0.4318865909311584,
+      "grad_norm": 2.793919684749034,
+      "learning_rate": 8.755149276631488e-07,
+      "loss": 3.2341,
+      "step": 3336
+    },
+    {
+      "epoch": 0.43201605333851184,
+      "grad_norm": 3.027613300059181,
+      "learning_rate": 8.752561096266279e-07,
+      "loss": 2.5266,
+      "step": 3337
+    },
+    {
+      "epoch": 0.4321455157458653,
+      "grad_norm": 1.698814111646842,
+      "learning_rate": 8.749972404702242e-07,
+      "loss": 2.3486,
+      "step": 3338
+    },
+    {
+      "epoch": 0.43227497815321875,
+      "grad_norm": 2.363862095933434,
+      "learning_rate": 8.74738320246817e-07,
+      "loss": 2.6167,
+      "step": 3339
+    },
+    {
+      "epoch": 0.4324044405605722,
+      "grad_norm": 1.4915473421090697,
+      "learning_rate": 8.74479349009296e-07,
+      "loss": 2.4108,
+      "step": 3340
+    },
+    {
+      "epoch": 0.4324044405605722,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.604203701019287,
+      "eval_runtime": 14.0615,
+      "eval_samples_per_second": 3.129,
+      "eval_steps_per_second": 0.213,
+      "step": 3340
+    },
+    {
+      "epoch": 0.4325339029679257,
+      "grad_norm": 2.04943456538855,
+      "learning_rate": 8.742203268105607e-07,
+      "loss": 2.6826,
+      "step": 3341
+    },
+    {
+      "epoch": 0.43266336537527916,
+      "grad_norm": 1.533730457714011,
+      "learning_rate": 8.739612537035227e-07,
+      "loss": 2.4873,
+      "step": 3342
+    },
+    {
+      "epoch": 0.4327928277826326,
+      "grad_norm": 1.4484341318687055,
+      "learning_rate": 8.737021297411024e-07,
+      "loss": 2.6475,
+      "step": 3343
+    },
+    {
+      "epoch": 0.43292229018998607,
+      "grad_norm": 1.702869870301113,
+      "learning_rate": 8.734429549762309e-07,
+      "loss": 2.6052,
+      "step": 3344
+    },
+    {
+      "epoch": 0.4330517525973395,
+      "grad_norm": 2.859857165430935,
+      "learning_rate": 8.731837294618501e-07,
+      "loss": 2.4946,
+      "step": 3345
+    },
+    {
+      "epoch": 0.4330517525973395,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.6056462526321411,
+      "eval_runtime": 13.7967,
+      "eval_samples_per_second": 3.189,
+      "eval_steps_per_second": 0.217,
+      "step": 3345
+    },
+    {
+      "epoch": 0.43318121500469303,
+      "grad_norm": 1.4731800231537813,
+      "learning_rate": 8.729244532509124e-07,
+      "loss": 2.5166,
+      "step": 3346
+    },
+    {
+      "epoch": 0.4333106774120465,
+      "grad_norm": 2.1589361458003804,
+      "learning_rate": 8.726651263963797e-07,
+      "loss": 2.7766,
+      "step": 3347
+    },
+    {
+      "epoch": 0.43344013981939994,
+      "grad_norm": 1.3084918079653816,
+      "learning_rate": 8.72405748951225e-07,
+      "loss": 2.3594,
+      "step": 3348
+    },
+    {
+      "epoch": 0.4335696022267534,
+      "grad_norm": 1.812523510963763,
+      "learning_rate": 8.721463209684315e-07,
+      "loss": 2.6387,
+      "step": 3349
+    },
+    {
+      "epoch": 0.43369906463410685,
+      "grad_norm": 3.1303527253118726,
+      "learning_rate": 8.718868425009924e-07,
+      "loss": 2.6023,
+      "step": 3350
+    },
+    {
+      "epoch": 0.43369906463410685,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.6034934520721436,
+      "eval_runtime": 13.6155,
+      "eval_samples_per_second": 3.232,
+      "eval_steps_per_second": 0.22,
+      "step": 3350
+    },
+    {
+      "epoch": 0.43382852704146035,
+      "grad_norm": 1.8553547256261265,
+      "learning_rate": 8.716273136019116e-07,
+      "loss": 2.6775,
+      "step": 3351
+    },
+    {
+      "epoch": 0.4339579894488138,
+      "grad_norm": 2.547847038990347,
+      "learning_rate": 8.713677343242031e-07,
+      "loss": 2.4614,
+      "step": 3352
+    },
+    {
+      "epoch": 0.43408745185616726,
+      "grad_norm": 3.75109330221715,
+      "learning_rate": 8.711081047208912e-07,
+      "loss": 2.6025,
+      "step": 3353
+    },
+    {
+      "epoch": 0.4342169142635207,
+      "grad_norm": 1.9635393930900862,
+      "learning_rate": 8.708484248450105e-07,
+      "loss": 2.6328,
+      "step": 3354
+    },
+    {
+      "epoch": 0.43434637667087417,
+      "grad_norm": 1.7949588842666337,
+      "learning_rate": 8.70588694749606e-07,
+      "loss": 2.6301,
+      "step": 3355
+    },
+    {
+      "epoch": 0.43434637667087417,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.600874423980713,
+      "eval_runtime": 14.369,
+      "eval_samples_per_second": 3.062,
+      "eval_steps_per_second": 0.209,
+      "step": 3355
+    },
+    {
+      "epoch": 0.4344758390782277,
+      "grad_norm": 1.9656718842338972,
+      "learning_rate": 8.703289144877326e-07,
+      "loss": 2.7419,
+      "step": 3356
+    },
+    {
+      "epoch": 0.43460530148558113,
+      "grad_norm": 2.6642528906401544,
+      "learning_rate": 8.70069084112456e-07,
+      "loss": 2.5046,
+      "step": 3357
+    },
+    {
+      "epoch": 0.4347347638929346,
+      "grad_norm": 2.8585548146962867,
+      "learning_rate": 8.698092036768516e-07,
+      "loss": 2.8552,
+      "step": 3358
+    },
+    {
+      "epoch": 0.43486422630028804,
+      "grad_norm": 1.1999466163898747,
+      "learning_rate": 8.695492732340053e-07,
+      "loss": 2.3899,
+      "step": 3359
+    },
+    {
+      "epoch": 0.43499368870764155,
+      "grad_norm": 3.3602213170747643,
+      "learning_rate": 8.692892928370131e-07,
+      "loss": 2.5068,
+      "step": 3360
+    },
+    {
+      "epoch": 0.43499368870764155,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.6021617650985718,
+      "eval_runtime": 15.3449,
+      "eval_samples_per_second": 2.867,
+      "eval_steps_per_second": 0.196,
+      "step": 3360
+    },
+    {
+      "epoch": 0.435123151114995,
+      "grad_norm": 1.5053696365724347,
+      "learning_rate": 8.690292625389813e-07,
+      "loss": 2.3132,
+      "step": 3361
+    },
+    {
+      "epoch": 0.43525261352234845,
+      "grad_norm": 2.0258490903535473,
+      "learning_rate": 8.687691823930264e-07,
+      "loss": 2.3914,
+      "step": 3362
+    },
+    {
+      "epoch": 0.4353820759297019,
+      "grad_norm": 1.6941609582033224,
+      "learning_rate": 8.685090524522751e-07,
+      "loss": 2.2051,
+      "step": 3363
+    },
+    {
+      "epoch": 0.43551153833705536,
+      "grad_norm": 2.9855637013370764,
+      "learning_rate": 8.682488727698638e-07,
+      "loss": 2.8433,
+      "step": 3364
+    },
+    {
+      "epoch": 0.43564100074440887,
+      "grad_norm": 4.064327015181763,
+      "learning_rate": 8.679886433989401e-07,
+      "loss": 2.8965,
+      "step": 3365
+    },
+    {
+      "epoch": 0.43564100074440887,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.600053310394287,
+      "eval_runtime": 16.8051,
+      "eval_samples_per_second": 2.618,
+      "eval_steps_per_second": 0.179,
+      "step": 3365
+    },
+    {
+      "epoch": 0.4357704631517623,
+      "grad_norm": 1.4337001854329114,
+      "learning_rate": 8.677283643926609e-07,
+      "loss": 2.5332,
+      "step": 3366
+    },
+    {
+      "epoch": 0.4358999255591158,
+      "grad_norm": 1.4481161669105034,
+      "learning_rate": 8.674680358041932e-07,
+      "loss": 2.3538,
+      "step": 3367
+    },
+    {
+      "epoch": 0.43602938796646923,
+      "grad_norm": 2.7396680621904284,
+      "learning_rate": 8.672076576867147e-07,
+      "loss": 2.6411,
+      "step": 3368
+    },
+    {
+      "epoch": 0.4361588503738227,
+      "grad_norm": 1.956391892926346,
+      "learning_rate": 8.669472300934128e-07,
+      "loss": 2.4097,
+      "step": 3369
+    },
+    {
+      "epoch": 0.4362883127811762,
+      "grad_norm": 1.622880309561545,
+      "learning_rate": 8.66686753077485e-07,
+      "loss": 2.5227,
+      "step": 3370
+    },
+    {
+      "epoch": 0.4362883127811762,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.597212314605713,
+      "eval_runtime": 14.3505,
+      "eval_samples_per_second": 3.066,
+      "eval_steps_per_second": 0.209,
+      "step": 3370
+    },
+    {
+      "epoch": 0.43641777518852964,
+      "grad_norm": 3.9637736274517943,
+      "learning_rate": 8.664262266921395e-07,
+      "loss": 3.2598,
+      "step": 3371
+    },
+    {
+      "epoch": 0.4365472375958831,
+      "grad_norm": 2.2328216453439893,
+      "learning_rate": 8.661656509905933e-07,
+      "loss": 2.6431,
+      "step": 3372
+    },
+    {
+      "epoch": 0.43667670000323655,
+      "grad_norm": 1.7118936859919116,
+      "learning_rate": 8.659050260260746e-07,
+      "loss": 2.2583,
+      "step": 3373
+    },
+    {
+      "epoch": 0.43680616241059,
+      "grad_norm": 1.8773939494511314,
+      "learning_rate": 8.656443518518216e-07,
+      "loss": 2.3059,
+      "step": 3374
+    },
+    {
+      "epoch": 0.4369356248179435,
+      "grad_norm": 1.3219131010165073,
+      "learning_rate": 8.653836285210822e-07,
+      "loss": 2.4644,
+      "step": 3375
+    },
+    {
+      "epoch": 0.4369356248179435,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.5993430614471436,
+      "eval_runtime": 13.9219,
+      "eval_samples_per_second": 3.16,
+      "eval_steps_per_second": 0.215,
+      "step": 3375
+    },
+    {
+      "epoch": 0.43706508722529697,
+      "grad_norm": 1.2464528363036336,
+      "learning_rate": 8.651228560871142e-07,
+      "loss": 2.4553,
+      "step": 3376
+    },
+    {
+      "epoch": 0.4371945496326504,
+      "grad_norm": 2.9549599624428957,
+      "learning_rate": 8.648620346031856e-07,
+      "loss": 2.7131,
+      "step": 3377
+    },
+    {
+      "epoch": 0.4373240120400039,
+      "grad_norm": 1.6168513729271259,
+      "learning_rate": 8.646011641225749e-07,
+      "loss": 2.551,
+      "step": 3378
+    },
+    {
+      "epoch": 0.4374534744473573,
+      "grad_norm": 2.503684851559407,
+      "learning_rate": 8.643402446985699e-07,
+      "loss": 2.3838,
+      "step": 3379
+    },
+    {
+      "epoch": 0.43758293685471084,
+      "grad_norm": 5.204177313655293,
+      "learning_rate": 8.640792763844688e-07,
+      "loss": 3.0251,
+      "step": 3380
+    },
+    {
+      "epoch": 0.43758293685471084,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.6007634401321411,
+      "eval_runtime": 14.1301,
+      "eval_samples_per_second": 3.114,
+      "eval_steps_per_second": 0.212,
+      "step": 3380
+    },
+    {
+      "epoch": 0.4377123992620643,
+      "grad_norm": 1.6433985321588829,
+      "learning_rate": 8.638182592335793e-07,
+      "loss": 2.5015,
+      "step": 3381
+    },
+    {
+      "epoch": 0.43784186166941774,
+      "grad_norm": 1.3609592400905013,
+      "learning_rate": 8.6355719329922e-07,
+      "loss": 2.4255,
+      "step": 3382
+    },
+    {
+      "epoch": 0.4379713240767712,
+      "grad_norm": 1.8615098510039647,
+      "learning_rate": 8.632960786347184e-07,
+      "loss": 2.4766,
+      "step": 3383
+    },
+    {
+      "epoch": 0.43810078648412465,
+      "grad_norm": 2.1488978232549893,
+      "learning_rate": 8.630349152934128e-07,
+      "loss": 2.4688,
+      "step": 3384
+    },
+    {
+      "epoch": 0.43823024889147816,
+      "grad_norm": 3.941986928649056,
+      "learning_rate": 8.627737033286512e-07,
+      "loss": 2.1689,
+      "step": 3385
+    },
+    {
+      "epoch": 0.43823024889147816,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.59716796875,
+      "eval_runtime": 15.117,
+      "eval_samples_per_second": 2.911,
+      "eval_steps_per_second": 0.198,
+      "step": 3385
+    },
+    {
+      "epoch": 0.4383597112988316,
+      "grad_norm": 1.4453550552085235,
+      "learning_rate": 8.625124427937909e-07,
+      "loss": 2.532,
+      "step": 3386
+    },
+    {
+      "epoch": 0.43848917370618506,
+      "grad_norm": 2.016114918139788,
+      "learning_rate": 8.622511337422002e-07,
+      "loss": 2.2383,
+      "step": 3387
+    },
+    {
+      "epoch": 0.4386186361135385,
+      "grad_norm": 1.4919377310131778,
+      "learning_rate": 8.619897762272566e-07,
+      "loss": 2.4177,
+      "step": 3388
+    },
+    {
+      "epoch": 0.43874809852089197,
+      "grad_norm": 2.6028617708322934,
+      "learning_rate": 8.617283703023475e-07,
+      "loss": 2.3193,
+      "step": 3389
+    },
+    {
+      "epoch": 0.4388775609282455,
+      "grad_norm": 2.2683466424149334,
+      "learning_rate": 8.614669160208707e-07,
+      "loss": 2.7856,
+      "step": 3390
+    },
+    {
+      "epoch": 0.4388775609282455,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.597212314605713,
+      "eval_runtime": 15.0193,
+      "eval_samples_per_second": 2.93,
+      "eval_steps_per_second": 0.2,
+      "step": 3390
+    },
+    {
+      "epoch": 0.43900702333559893,
+      "grad_norm": 1.9721875224024048,
+      "learning_rate": 8.612054134362331e-07,
+      "loss": 2.408,
+      "step": 3391
+    },
+    {
+      "epoch": 0.4391364857429524,
+      "grad_norm": 3.7510369403129245,
+      "learning_rate": 8.609438626018522e-07,
+      "loss": 2.6289,
+      "step": 3392
+    },
+    {
+      "epoch": 0.43926594815030584,
+      "grad_norm": 1.2007011663733482,
+      "learning_rate": 8.606822635711553e-07,
+      "loss": 2.2271,
+      "step": 3393
+    },
+    {
+      "epoch": 0.4393954105576593,
+      "grad_norm": 2.2140861932290323,
+      "learning_rate": 8.604206163975787e-07,
+      "loss": 2.8462,
+      "step": 3394
+    },
+    {
+      "epoch": 0.4395248729650128,
+      "grad_norm": 1.872874280001812,
+      "learning_rate": 8.601589211345694e-07,
+      "loss": 2.366,
+      "step": 3395
+    },
+    {
+      "epoch": 0.4395248729650128,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.59716796875,
+      "eval_runtime": 13.7835,
+      "eval_samples_per_second": 3.192,
+      "eval_steps_per_second": 0.218,
+      "step": 3395
+    },
+    {
+      "epoch": 0.43965433537236626,
+      "grad_norm": 2.270875388489974,
+      "learning_rate": 8.598971778355842e-07,
+      "loss": 2.3323,
+      "step": 3396
+    },
+    {
+      "epoch": 0.4397837977797197,
+      "grad_norm": 1.8802560539137303,
+      "learning_rate": 8.596353865540891e-07,
+      "loss": 2.6616,
+      "step": 3397
+    },
+    {
+      "epoch": 0.43991326018707316,
+      "grad_norm": 1.5463882268775082,
+      "learning_rate": 8.593735473435605e-07,
+      "loss": 2.4248,
+      "step": 3398
+    },
+    {
+      "epoch": 0.4400427225944266,
+      "grad_norm": 2.5888801357490405,
+      "learning_rate": 8.591116602574845e-07,
+      "loss": 2.7466,
+      "step": 3399
+    },
+    {
+      "epoch": 0.4401721850017801,
+      "grad_norm": 2.463832441045507,
+      "learning_rate": 8.588497253493563e-07,
+      "loss": 2.604,
+      "step": 3400
+    },
+    {
+      "epoch": 0.4401721850017801,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.602095127105713,
+      "eval_runtime": 15.6022,
+      "eval_samples_per_second": 2.82,
+      "eval_steps_per_second": 0.192,
+      "step": 3400
+    },
+    {
+      "epoch": 0.4403016474091336,
+      "grad_norm": 4.191232338270609,
+      "learning_rate": 8.585877426726821e-07,
+      "loss": 3.2261,
+      "step": 3401
+    },
+    {
+      "epoch": 0.44043110981648703,
+      "grad_norm": 2.5971509161605466,
+      "learning_rate": 8.583257122809768e-07,
+      "loss": 2.5598,
+      "step": 3402
+    },
+    {
+      "epoch": 0.4405605722238405,
+      "grad_norm": 2.7262606605151185,
+      "learning_rate": 8.580636342277654e-07,
+      "loss": 2.3704,
+      "step": 3403
+    },
+    {
+      "epoch": 0.440690034631194,
+      "grad_norm": 1.575631610522882,
+      "learning_rate": 8.578015085665828e-07,
+      "loss": 2.4663,
+      "step": 3404
+    },
+    {
+      "epoch": 0.44081949703854745,
+      "grad_norm": 1.7416078489275513,
+      "learning_rate": 8.575393353509733e-07,
+      "loss": 2.437,
+      "step": 3405
+    },
+    {
+      "epoch": 0.44081949703854745,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.5992764234542847,
+      "eval_runtime": 16.736,
+      "eval_samples_per_second": 2.629,
+      "eval_steps_per_second": 0.179,
+      "step": 3405
+    },
+    {
+      "epoch": 0.4409489594459009,
+      "grad_norm": 1.858486069236183,
+      "learning_rate": 8.572771146344911e-07,
+      "loss": 2.4883,
+      "step": 3406
+    },
+    {
+      "epoch": 0.44107842185325435,
+      "grad_norm": 1.449590755098125,
+      "learning_rate": 8.570148464707005e-07,
+      "loss": 2.3585,
+      "step": 3407
+    },
+    {
+      "epoch": 0.4412078842606078,
+      "grad_norm": 1.9889072502150644,
+      "learning_rate": 8.567525309131746e-07,
+      "loss": 2.6187,
+      "step": 3408
+    },
+    {
+      "epoch": 0.4413373466679613,
+      "grad_norm": 1.7210628402362425,
+      "learning_rate": 8.564901680154968e-07,
+      "loss": 2.2261,
+      "step": 3409
+    },
+    {
+      "epoch": 0.44146680907531477,
+      "grad_norm": 2.163720413595985,
+      "learning_rate": 8.562277578312603e-07,
+      "loss": 2.3452,
+      "step": 3410
+    },
+    {
+      "epoch": 0.44146680907531477,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.6006969213485718,
+      "eval_runtime": 14.8723,
+      "eval_samples_per_second": 2.959,
+      "eval_steps_per_second": 0.202,
+      "step": 3410
+    },
+    {
+      "epoch": 0.4415962714826682,
+      "grad_norm": 2.4243626518666135,
+      "learning_rate": 8.559653004140672e-07,
+      "loss": 2.7583,
+      "step": 3411
+    },
+    {
+      "epoch": 0.4417257338900217,
+      "grad_norm": 2.0677474814978734,
+      "learning_rate": 8.557027958175303e-07,
+      "loss": 2.4653,
+      "step": 3412
+    },
+    {
+      "epoch": 0.44185519629737513,
+      "grad_norm": 2.1336206226226317,
+      "learning_rate": 8.554402440952711e-07,
+      "loss": 2.7041,
+      "step": 3413
+    },
+    {
+      "epoch": 0.44198465870472864,
+      "grad_norm": 1.2137327826622863,
+      "learning_rate": 8.551776453009212e-07,
+      "loss": 2.4465,
+      "step": 3414
+    },
+    {
+      "epoch": 0.4421141211120821,
+      "grad_norm": 1.6909561260831407,
+      "learning_rate": 8.549149994881217e-07,
+      "loss": 2.4873,
+      "step": 3415
+    },
+    {
+      "epoch": 0.4421141211120821,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.5963245630264282,
+      "eval_runtime": 14.0823,
+      "eval_samples_per_second": 3.124,
+      "eval_steps_per_second": 0.213,
+      "step": 3415
+    },
+    {
+      "epoch": 0.44224358351943555,
+      "grad_norm": 1.4637293706623524,
+      "learning_rate": 8.546523067105236e-07,
+      "loss": 2.5273,
+      "step": 3416
+    },
+    {
+      "epoch": 0.442373045926789,
+      "grad_norm": 1.9654887845983764,
+      "learning_rate": 8.543895670217869e-07,
+      "loss": 2.5505,
+      "step": 3417
+    },
+    {
+      "epoch": 0.44250250833414245,
+      "grad_norm": 2.0927922120530504,
+      "learning_rate": 8.541267804755817e-07,
+      "loss": 2.4158,
+      "step": 3418
+    },
+    {
+      "epoch": 0.44263197074149596,
+      "grad_norm": 1.7630548540773947,
+      "learning_rate": 8.538639471255873e-07,
+      "loss": 2.373,
+      "step": 3419
+    },
+    {
+      "epoch": 0.4427614331488494,
+      "grad_norm": 2.3750621974147834,
+      "learning_rate": 8.53601067025493e-07,
+      "loss": 2.7734,
+      "step": 3420
+    },
+    {
+      "epoch": 0.4427614331488494,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.6000088453292847,
+      "eval_runtime": 14.1695,
+      "eval_samples_per_second": 3.105,
+      "eval_steps_per_second": 0.212,
+      "step": 3420
+    },
+    {
+      "epoch": 0.44289089555620287,
+      "grad_norm": 2.0183413821139617,
+      "learning_rate": 8.533381402289971e-07,
+      "loss": 2.5857,
+      "step": 3421
+    },
+    {
+      "epoch": 0.4430203579635563,
+      "grad_norm": 2.25435004222512,
+      "learning_rate": 8.530751667898077e-07,
+      "loss": 2.6216,
+      "step": 3422
+    },
+    {
+      "epoch": 0.4431498203709098,
+      "grad_norm": 2.356841055509401,
+      "learning_rate": 8.528121467616426e-07,
+      "loss": 3.0442,
+      "step": 3423
+    },
+    {
+      "epoch": 0.4432792827782633,
+      "grad_norm": 1.4931724895600031,
+      "learning_rate": 8.525490801982294e-07,
+      "loss": 2.3027,
+      "step": 3424
+    },
+    {
+      "epoch": 0.44340874518561674,
+      "grad_norm": 2.283011298172669,
+      "learning_rate": 8.522859671533041e-07,
+      "loss": 2.4888,
+      "step": 3425
+    },
+    {
+      "epoch": 0.44340874518561674,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.597212314605713,
+      "eval_runtime": 13.8367,
+      "eval_samples_per_second": 3.18,
+      "eval_steps_per_second": 0.217,
+      "step": 3425
+    },
+    {
+      "epoch": 0.4435382075929702,
+      "grad_norm": 1.974450548992583,
+      "learning_rate": 8.520228076806131e-07,
+      "loss": 2.4941,
+      "step": 3426
+    },
+    {
+      "epoch": 0.44366767000032364,
+      "grad_norm": 2.048760438948378,
+      "learning_rate": 8.517596018339121e-07,
+      "loss": 2.2754,
+      "step": 3427
+    },
+    {
+      "epoch": 0.4437971324076771,
+      "grad_norm": 2.0463369374222444,
+      "learning_rate": 8.514963496669662e-07,
+      "loss": 2.3474,
+      "step": 3428
+    },
+    {
+      "epoch": 0.4439265948150306,
+      "grad_norm": 2.1441024974320495,
+      "learning_rate": 8.512330512335501e-07,
+      "loss": 2.541,
+      "step": 3429
+    },
+    {
+      "epoch": 0.44405605722238406,
+      "grad_norm": 2.274939591214134,
+      "learning_rate": 8.509697065874478e-07,
+      "loss": 2.5029,
+      "step": 3430
+    },
+    {
+      "epoch": 0.44405605722238406,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.5963689088821411,
+      "eval_runtime": 14.1581,
+      "eval_samples_per_second": 3.108,
+      "eval_steps_per_second": 0.212,
+      "step": 3430
+    },
+    {
+      "epoch": 0.4441855196297375,
+      "grad_norm": 2.0364956373765364,
+      "learning_rate": 8.507063157824526e-07,
+      "loss": 2.4841,
+      "step": 3431
+    },
+    {
+      "epoch": 0.44431498203709097,
+      "grad_norm": 3.682776302771488,
+      "learning_rate": 8.504428788723676e-07,
+      "loss": 2.8289,
+      "step": 3432
+    },
+    {
+      "epoch": 0.4444444444444444,
+      "grad_norm": 1.4349522246313382,
+      "learning_rate": 8.501793959110049e-07,
+      "loss": 2.6213,
+      "step": 3433
+    },
+    {
+      "epoch": 0.44457390685179793,
+      "grad_norm": 2.0647479134090037,
+      "learning_rate": 8.499158669521867e-07,
+      "loss": 2.6248,
+      "step": 3434
+    },
+    {
+      "epoch": 0.4447033692591514,
+      "grad_norm": 1.9191416166687798,
+      "learning_rate": 8.496522920497436e-07,
+      "loss": 2.4507,
+      "step": 3435
+    },
+    {
+      "epoch": 0.4447033692591514,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.595015048980713,
+      "eval_runtime": 14.0327,
+      "eval_samples_per_second": 3.136,
+      "eval_steps_per_second": 0.214,
+      "step": 3435
+    },
+    {
+      "epoch": 0.44483283166650484,
+      "grad_norm": 1.827599335538608,
+      "learning_rate": 8.493886712575164e-07,
+      "loss": 2.4219,
+      "step": 3436
+    },
+    {
+      "epoch": 0.4449622940738583,
+      "grad_norm": 1.7583880447580256,
+      "learning_rate": 8.491250046293549e-07,
+      "loss": 2.5254,
+      "step": 3437
+    },
+    {
+      "epoch": 0.44509175648121174,
+      "grad_norm": 3.20864614085216,
+      "learning_rate": 8.488612922191187e-07,
+      "loss": 2.5859,
+      "step": 3438
+    },
+    {
+      "epoch": 0.44522121888856525,
+      "grad_norm": 2.329617569954473,
+      "learning_rate": 8.485975340806758e-07,
+      "loss": 2.405,
+      "step": 3439
+    },
+    {
+      "epoch": 0.4453506812959187,
+      "grad_norm": 2.541178163993142,
+      "learning_rate": 8.483337302679045e-07,
+      "loss": 2.5632,
+      "step": 3440
+    },
+    {
+      "epoch": 0.4453506812959187,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.592240810394287,
+      "eval_runtime": 14.1902,
+      "eval_samples_per_second": 3.101,
+      "eval_steps_per_second": 0.211,
+      "step": 3440
+    },
+    {
+      "epoch": 0.44548014370327216,
+      "grad_norm": 9.755125786204166,
+      "learning_rate": 8.480698808346921e-07,
+      "loss": 3.2383,
+      "step": 3441
+    },
+    {
+      "epoch": 0.4456096061106256,
+      "grad_norm": 2.5351495088939333,
+      "learning_rate": 8.478059858349351e-07,
+      "loss": 2.3838,
+      "step": 3442
+    },
+    {
+      "epoch": 0.44573906851797906,
+      "grad_norm": 2.5097741971867453,
+      "learning_rate": 8.475420453225394e-07,
+      "loss": 2.6748,
+      "step": 3443
+    },
+    {
+      "epoch": 0.4458685309253326,
+      "grad_norm": 1.5619751969655316,
+      "learning_rate": 8.472780593514203e-07,
+      "loss": 2.4473,
+      "step": 3444
+    },
+    {
+      "epoch": 0.445997993332686,
+      "grad_norm": 2.106104250260571,
+      "learning_rate": 8.47014027975502e-07,
+      "loss": 2.9629,
+      "step": 3445
+    },
+    {
+      "epoch": 0.445997993332686,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.5936611890792847,
+      "eval_runtime": 14.5556,
+      "eval_samples_per_second": 3.023,
+      "eval_steps_per_second": 0.206,
+      "step": 3445
+    },
+    {
+      "epoch": 0.4461274557400395,
+      "grad_norm": 4.360322621023151,
+      "learning_rate": 8.467499512487187e-07,
+      "loss": 2.5847,
+      "step": 3446
+    },
+    {
+      "epoch": 0.44625691814739293,
+      "grad_norm": 2.042969660972691,
+      "learning_rate": 8.46485829225013e-07,
+      "loss": 2.6223,
+      "step": 3447
+    },
+    {
+      "epoch": 0.44638638055474644,
+      "grad_norm": 1.65399542988463,
+      "learning_rate": 8.462216619583375e-07,
+      "loss": 2.4766,
+      "step": 3448
+    },
+    {
+      "epoch": 0.4465158429620999,
+      "grad_norm": 1.8625609503245593,
+      "learning_rate": 8.459574495026534e-07,
+      "loss": 2.3845,
+      "step": 3449
+    },
+    {
+      "epoch": 0.44664530536945335,
+      "grad_norm": 1.543416912225978,
+      "learning_rate": 8.456931919119318e-07,
+      "loss": 2.5769,
+      "step": 3450
+    },
+    {
+      "epoch": 0.44664530536945335,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.5943936109542847,
+      "eval_runtime": 14.9721,
+      "eval_samples_per_second": 2.939,
+      "eval_steps_per_second": 0.2,
+      "step": 3450
+    },
+    {
+      "epoch": 0.4467747677768068,
+      "grad_norm": 1.7372836516407466,
+      "learning_rate": 8.454288892401525e-07,
+      "loss": 2.5654,
+      "step": 3451
+    },
+    {
+      "epoch": 0.44690423018416026,
+      "grad_norm": 1.7735550649244403,
+      "learning_rate": 8.451645415413049e-07,
+      "loss": 2.6235,
+      "step": 3452
+    },
+    {
+      "epoch": 0.44703369259151376,
+      "grad_norm": 1.8790458134274919,
+      "learning_rate": 8.449001488693868e-07,
+      "loss": 2.4988,
+      "step": 3453
+    },
+    {
+      "epoch": 0.4471631549988672,
+      "grad_norm": 2.152908518765958,
+      "learning_rate": 8.446357112784063e-07,
+      "loss": 2.8228,
+      "step": 3454
+    },
+    {
+      "epoch": 0.44729261740622067,
+      "grad_norm": 2.7746991861878727,
+      "learning_rate": 8.443712288223799e-07,
+      "loss": 2.8479,
+      "step": 3455
+    },
+    {
+      "epoch": 0.44729261740622067,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.5908424854278564,
+      "eval_runtime": 14.1587,
+      "eval_samples_per_second": 3.108,
+      "eval_steps_per_second": 0.212,
+      "step": 3455
+    },
+    {
+      "epoch": 0.4474220798135741,
+      "grad_norm": 1.212695881199798,
+      "learning_rate": 8.441067015553335e-07,
+      "loss": 2.4155,
+      "step": 3456
+    },
+    {
+      "epoch": 0.4475515422209276,
+      "grad_norm": 2.0977987197324413,
+      "learning_rate": 8.438421295313024e-07,
+      "loss": 2.5911,
+      "step": 3457
+    },
+    {
+      "epoch": 0.4476810046282811,
+      "grad_norm": 4.683650449412505,
+      "learning_rate": 8.435775128043304e-07,
+      "loss": 2.4507,
+      "step": 3458
+    },
+    {
+      "epoch": 0.44781046703563454,
+      "grad_norm": 1.7245941876593716,
+      "learning_rate": 8.43312851428471e-07,
+      "loss": 2.5325,
+      "step": 3459
+    },
+    {
+      "epoch": 0.447939929442988,
+      "grad_norm": 1.8826085030779252,
+      "learning_rate": 8.43048145457787e-07,
+      "loss": 2.4119,
+      "step": 3460
+    },
+    {
+      "epoch": 0.447939929442988,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.5928400754928589,
+      "eval_runtime": 14.1723,
+      "eval_samples_per_second": 3.105,
+      "eval_steps_per_second": 0.212,
+      "step": 3460
+    },
+    {
+      "epoch": 0.44806939185034145,
+      "grad_norm": 4.8805519417513334,
+      "learning_rate": 8.427833949463495e-07,
+      "loss": 2.9414,
+      "step": 3461
+    },
+    {
+      "epoch": 0.4481988542576949,
+      "grad_norm": 1.4515816772874621,
+      "learning_rate": 8.425185999482392e-07,
+      "loss": 2.4395,
+      "step": 3462
+    },
+    {
+      "epoch": 0.4483283166650484,
+      "grad_norm": 1.768673429111786,
+      "learning_rate": 8.422537605175461e-07,
+      "loss": 2.3357,
+      "step": 3463
+    },
+    {
+      "epoch": 0.44845777907240186,
+      "grad_norm": 4.9546667520887695,
+      "learning_rate": 8.419888767083688e-07,
+      "loss": 2.5649,
+      "step": 3464
+    },
+    {
+      "epoch": 0.4485872414797553,
+      "grad_norm": 1.5034713072550403,
+      "learning_rate": 8.417239485748154e-07,
+      "loss": 2.582,
+      "step": 3465
+    },
+    {
+      "epoch": 0.4485872414797553,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.593550205230713,
+      "eval_runtime": 14.4344,
+      "eval_samples_per_second": 3.048,
+      "eval_steps_per_second": 0.208,
+      "step": 3465
+    },
+    {
+      "epoch": 0.44871670388710877,
+      "grad_norm": 2.4904044082507673,
+      "learning_rate": 8.414589761710027e-07,
+      "loss": 2.2974,
+      "step": 3466
+    },
+    {
+      "epoch": 0.4488461662944622,
+      "grad_norm": 1.4453400295254624,
+      "learning_rate": 8.411939595510565e-07,
+      "loss": 2.3704,
+      "step": 3467
+    },
+    {
+      "epoch": 0.44897562870181573,
+      "grad_norm": 2.354240400747779,
+      "learning_rate": 8.409288987691122e-07,
+      "loss": 2.573,
+      "step": 3468
+    },
+    {
+      "epoch": 0.4491050911091692,
+      "grad_norm": 2.125029297268219,
+      "learning_rate": 8.406637938793136e-07,
+      "loss": 2.4602,
+      "step": 3469
+    },
+    {
+      "epoch": 0.44923455351652264,
+      "grad_norm": 2.653936148520379,
+      "learning_rate": 8.403986449358137e-07,
+      "loss": 2.6267,
+      "step": 3470
+    },
+    {
+      "epoch": 0.44923455351652264,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.587202548980713,
+      "eval_runtime": 14.2485,
+      "eval_samples_per_second": 3.088,
+      "eval_steps_per_second": 0.211,
+      "step": 3470
+    },
+    {
+      "epoch": 0.4493640159238761,
+      "grad_norm": 1.542150110411501,
+      "learning_rate": 8.401334519927748e-07,
+      "loss": 2.6687,
+      "step": 3471
+    },
+    {
+      "epoch": 0.44949347833122955,
+      "grad_norm": 1.970404376171036,
+      "learning_rate": 8.398682151043675e-07,
+      "loss": 2.5945,
+      "step": 3472
+    },
+    {
+      "epoch": 0.44962294073858305,
+      "grad_norm": 2.75045176495899,
+      "learning_rate": 8.396029343247722e-07,
+      "loss": 2.3618,
+      "step": 3473
+    },
+    {
+      "epoch": 0.4497524031459365,
+      "grad_norm": 1.8640981826818483,
+      "learning_rate": 8.39337609708178e-07,
+      "loss": 2.4033,
+      "step": 3474
+    },
+    {
+      "epoch": 0.44988186555328996,
+      "grad_norm": 1.8051303242980334,
+      "learning_rate": 8.390722413087823e-07,
+      "loss": 2.4114,
+      "step": 3475
+    },
+    {
+      "epoch": 0.44988186555328996,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.590775966644287,
+      "eval_runtime": 14.2357,
+      "eval_samples_per_second": 3.091,
+      "eval_steps_per_second": 0.211,
+      "step": 3475
+    },
+    {
+      "epoch": 0.4500113279606434,
+      "grad_norm": 1.5747446483541916,
+      "learning_rate": 8.388068291807924e-07,
+      "loss": 2.4559,
+      "step": 3476
+    },
+    {
+      "epoch": 0.45014079036799687,
+      "grad_norm": 1.8801924727738788,
+      "learning_rate": 8.38541373378424e-07,
+      "loss": 2.3982,
+      "step": 3477
+    },
+    {
+      "epoch": 0.4502702527753504,
+      "grad_norm": 3.0946284691450274,
+      "learning_rate": 8.382758739559017e-07,
+      "loss": 2.4312,
+      "step": 3478
+    },
+    {
+      "epoch": 0.45039971518270383,
+      "grad_norm": 1.9475762253494309,
+      "learning_rate": 8.380103309674594e-07,
+      "loss": 2.2505,
+      "step": 3479
+    },
+    {
+      "epoch": 0.4505291775900573,
+      "grad_norm": 1.952188856399258,
+      "learning_rate": 8.377447444673395e-07,
+      "loss": 2.3733,
+      "step": 3480
+    },
+    {
+      "epoch": 0.4505291775900573,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.592817783355713,
+      "eval_runtime": 13.6523,
+      "eval_samples_per_second": 3.223,
+      "eval_steps_per_second": 0.22,
+      "step": 3480
+    },
+    {
+      "epoch": 0.45065863999741074,
+      "grad_norm": 1.6809954356463555,
+      "learning_rate": 8.374791145097931e-07,
+      "loss": 2.5569,
+      "step": 3481
+    },
+    {
+      "epoch": 0.4507881024047642,
+      "grad_norm": 1.983387145091062,
+      "learning_rate": 8.37213441149081e-07,
+      "loss": 2.4927,
+      "step": 3482
+    },
+    {
+      "epoch": 0.4509175648121177,
+      "grad_norm": 1.4753342091772366,
+      "learning_rate": 8.369477244394722e-07,
+      "loss": 2.4827,
+      "step": 3483
+    },
+    {
+      "epoch": 0.45104702721947115,
+      "grad_norm": 1.7076497338199936,
+      "learning_rate": 8.366819644352446e-07,
+      "loss": 2.8013,
+      "step": 3484
+    },
+    {
+      "epoch": 0.4511764896268246,
+      "grad_norm": 1.5460583439837456,
+      "learning_rate": 8.364161611906852e-07,
+      "loss": 2.3591,
+      "step": 3485
+    },
+    {
+      "epoch": 0.4511764896268246,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.591352939605713,
+      "eval_runtime": 14.0091,
+      "eval_samples_per_second": 3.141,
+      "eval_steps_per_second": 0.214,
+      "step": 3485
+    },
+    {
+      "epoch": 0.45130595203417806,
+      "grad_norm": 1.8335786141903636,
+      "learning_rate": 8.361503147600897e-07,
+      "loss": 2.4473,
+      "step": 3486
+    },
+    {
+      "epoch": 0.4514354144415315,
+      "grad_norm": 2.754884804185985,
+      "learning_rate": 8.358844251977626e-07,
+      "loss": 2.4214,
+      "step": 3487
+    },
+    {
+      "epoch": 0.451564876848885,
+      "grad_norm": 3.5112997594448667,
+      "learning_rate": 8.356184925580173e-07,
+      "loss": 2.4131,
+      "step": 3488
+    },
+    {
+      "epoch": 0.4516943392562385,
+      "grad_norm": 1.6092942665609502,
+      "learning_rate": 8.353525168951756e-07,
+      "loss": 2.4714,
+      "step": 3489
+    },
+    {
+      "epoch": 0.45182380166359193,
+      "grad_norm": 2.125823252390096,
+      "learning_rate": 8.35086498263569e-07,
+      "loss": 2.4609,
+      "step": 3490
+    },
+    {
+      "epoch": 0.45182380166359193,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.5899990797042847,
+      "eval_runtime": 14.5263,
+      "eval_samples_per_second": 3.029,
+      "eval_steps_per_second": 0.207,
+      "step": 3490
+    },
+    {
+      "epoch": 0.4519532640709454,
+      "grad_norm": 2.159282465658603,
+      "learning_rate": 8.348204367175367e-07,
+      "loss": 2.4088,
+      "step": 3491
+    },
+    {
+      "epoch": 0.4520827264782989,
+      "grad_norm": 1.7870489868656576,
+      "learning_rate": 8.345543323114272e-07,
+      "loss": 2.3232,
+      "step": 3492
+    },
+    {
+      "epoch": 0.45221218888565234,
+      "grad_norm": 1.422530218838628,
+      "learning_rate": 8.342881850995979e-07,
+      "loss": 2.5317,
+      "step": 3493
+    },
+    {
+      "epoch": 0.4523416512930058,
+      "grad_norm": 1.7856068188226795,
+      "learning_rate": 8.340219951364151e-07,
+      "loss": 2.4795,
+      "step": 3494
+    },
+    {
+      "epoch": 0.45247111370035925,
+      "grad_norm": 1.5030901226106912,
+      "learning_rate": 8.337557624762525e-07,
+      "loss": 2.4221,
+      "step": 3495
+    },
+    {
+      "epoch": 0.45247111370035925,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.5936168432235718,
+      "eval_runtime": 14.9019,
+      "eval_samples_per_second": 2.953,
+      "eval_steps_per_second": 0.201,
+      "step": 3495
+    },
+    {
+      "epoch": 0.4526005761077127,
+      "grad_norm": 1.5523475542841818,
+      "learning_rate": 8.334894871734946e-07,
+      "loss": 2.4216,
+      "step": 3496
+    },
+    {
+      "epoch": 0.4527300385150662,
+      "grad_norm": 1.960794740924934,
+      "learning_rate": 8.332231692825329e-07,
+      "loss": 2.3936,
+      "step": 3497
+    },
+    {
+      "epoch": 0.45285950092241967,
+      "grad_norm": 2.5424394583593775,
+      "learning_rate": 8.329568088577683e-07,
+      "loss": 2.418,
+      "step": 3498
+    },
+    {
+      "epoch": 0.4529889633297731,
+      "grad_norm": 1.691201976792941,
+      "learning_rate": 8.326904059536105e-07,
+      "loss": 2.5312,
+      "step": 3499
+    },
+    {
+      "epoch": 0.4531184257371266,
+      "grad_norm": 1.3963541506015975,
+      "learning_rate": 8.324239606244775e-07,
+      "loss": 2.2056,
+      "step": 3500
+    },
+    {
+      "epoch": 0.4531184257371266,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.5943492650985718,
+      "eval_runtime": 14.3213,
+      "eval_samples_per_second": 3.072,
+      "eval_steps_per_second": 0.209,
+      "step": 3500
     }
   ],
   "logging_steps": 1,
@@ -30642,7 +35742,7 @@
       "attributes": {}
     }
   },
-  "total_flos": 2518198355329024.0,
+  "total_flos": 2941030172688384.0,
   "train_batch_size": 4,
   "trial_name": null,
   "trial_params": null