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": 0.6473120367673237,
+  "epoch": 0.7120432404440561,
   "eval_steps": 5,
-  "global_step": 5000,
+  "global_step": 5500,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
@@ -51023,6 +51023,5106 @@
       "eval_samples_per_second": 2.996,
       "eval_steps_per_second": 0.204,
       "step": 5000
+    },
+    {
+      "epoch": 0.6474414991746772,
+      "grad_norm": 1.5423378270164758,
+      "learning_rate": 4.1651051282650565e-07,
+      "loss": 2.23,
+      "step": 5001
+    },
+    {
+      "epoch": 0.6475709615820306,
+      "grad_norm": 2.2168790713588766,
+      "learning_rate": 4.1624423752374757e-07,
+      "loss": 2.467,
+      "step": 5002
+    },
+    {
+      "epoch": 0.6477004239893841,
+      "grad_norm": 4.238044830500742,
+      "learning_rate": 4.1597800486358514e-07,
+      "loss": 2.7502,
+      "step": 5003
+    },
+    {
+      "epoch": 0.6478298863967376,
+      "grad_norm": 1.315030738518405,
+      "learning_rate": 4.1571181490040203e-07,
+      "loss": 2.2556,
+      "step": 5004
+    },
+    {
+      "epoch": 0.647959348804091,
+      "grad_norm": 2.0476048994716485,
+      "learning_rate": 4.1544566768857276e-07,
+      "loss": 2.5449,
+      "step": 5005
+    },
+    {
+      "epoch": 0.647959348804091,
+      "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": 1.5178444385528564,
+      "eval_runtime": 13.1657,
+      "eval_samples_per_second": 3.342,
+      "eval_steps_per_second": 0.228,
+      "step": 5005
+    },
+    {
+      "epoch": 0.6480888112114445,
+      "grad_norm": 4.1503996450971306,
+      "learning_rate": 4.151795632824637e-07,
+      "loss": 2.7852,
+      "step": 5006
+    },
+    {
+      "epoch": 0.6482182736187979,
+      "grad_norm": 2.3969681986438585,
+      "learning_rate": 4.149135017364313e-07,
+      "loss": 2.4185,
+      "step": 5007
+    },
+    {
+      "epoch": 0.6483477360261514,
+      "grad_norm": 5.467870955809488,
+      "learning_rate": 4.1464748310482446e-07,
+      "loss": 2.9004,
+      "step": 5008
+    },
+    {
+      "epoch": 0.6484771984335048,
+      "grad_norm": 3.302516453557103,
+      "learning_rate": 4.1438150744198285e-07,
+      "loss": 2.5051,
+      "step": 5009
+    },
+    {
+      "epoch": 0.6486066608408584,
+      "grad_norm": 1.9473029119630008,
+      "learning_rate": 4.141155748022375e-07,
+      "loss": 2.2058,
+      "step": 5010
+    },
+    {
+      "epoch": 0.6486066608408584,
+      "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": 1.5214622020721436,
+      "eval_runtime": 15.6205,
+      "eval_samples_per_second": 2.817,
+      "eval_steps_per_second": 0.192,
+      "step": 5010
+    },
+    {
+      "epoch": 0.6487361232482118,
+      "grad_norm": 1.5225694385522397,
+      "learning_rate": 4.138496852399104e-07,
+      "loss": 2.3213,
+      "step": 5011
+    },
+    {
+      "epoch": 0.6488655856555653,
+      "grad_norm": 1.29497540285883,
+      "learning_rate": 4.1358383880931497e-07,
+      "loss": 2.3738,
+      "step": 5012
+    },
+    {
+      "epoch": 0.6489950480629187,
+      "grad_norm": 2.210618806339647,
+      "learning_rate": 4.133180355647555e-07,
+      "loss": 2.4683,
+      "step": 5013
+    },
+    {
+      "epoch": 0.6491245104702722,
+      "grad_norm": 1.7513332411139912,
+      "learning_rate": 4.13052275560528e-07,
+      "loss": 2.3081,
+      "step": 5014
+    },
+    {
+      "epoch": 0.6492539728776257,
+      "grad_norm": 1.8535856629197902,
+      "learning_rate": 4.127865588509191e-07,
+      "loss": 2.4546,
+      "step": 5015
+    },
+    {
+      "epoch": 0.6492539728776257,
+      "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": 1.5193759202957153,
+      "eval_runtime": 14.0118,
+      "eval_samples_per_second": 3.14,
+      "eval_steps_per_second": 0.214,
+      "step": 5015
+    },
+    {
+      "epoch": 0.6493834352849791,
+      "grad_norm": 2.7257107635063034,
+      "learning_rate": 4.12520885490207e-07,
+      "loss": 2.5786,
+      "step": 5016
+    },
+    {
+      "epoch": 0.6495128976923326,
+      "grad_norm": 1.7039836552107441,
+      "learning_rate": 4.1225525553266086e-07,
+      "loss": 2.3533,
+      "step": 5017
+    },
+    {
+      "epoch": 0.649642360099686,
+      "grad_norm": 1.9939476435919625,
+      "learning_rate": 4.119896690325408e-07,
+      "loss": 2.5183,
+      "step": 5018
+    },
+    {
+      "epoch": 0.6497718225070395,
+      "grad_norm": 1.7527247350777142,
+      "learning_rate": 4.1172412604409836e-07,
+      "loss": 2.4221,
+      "step": 5019
+    },
+    {
+      "epoch": 0.649901284914393,
+      "grad_norm": 1.333675338909329,
+      "learning_rate": 4.114586266215762e-07,
+      "loss": 2.269,
+      "step": 5020
+    },
+    {
+      "epoch": 0.649901284914393,
+      "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": 1.5201305150985718,
+      "eval_runtime": 13.7462,
+      "eval_samples_per_second": 3.201,
+      "eval_steps_per_second": 0.218,
+      "step": 5020
+    },
+    {
+      "epoch": 0.6500307473217465,
+      "grad_norm": 1.734064725440254,
+      "learning_rate": 4.111931708192077e-07,
+      "loss": 2.5037,
+      "step": 5021
+    },
+    {
+      "epoch": 0.6501602097290999,
+      "grad_norm": 1.4368427898116118,
+      "learning_rate": 4.1092775869121764e-07,
+      "loss": 2.4485,
+      "step": 5022
+    },
+    {
+      "epoch": 0.6502896721364534,
+      "grad_norm": 1.9345469087588762,
+      "learning_rate": 4.106623902918223e-07,
+      "loss": 2.2983,
+      "step": 5023
+    },
+    {
+      "epoch": 0.6504191345438068,
+      "grad_norm": 1.3994814096798716,
+      "learning_rate": 4.103970656752279e-07,
+      "loss": 2.3792,
+      "step": 5024
+    },
+    {
+      "epoch": 0.6505485969511603,
+      "grad_norm": 1.7151358282869378,
+      "learning_rate": 4.101317848956327e-07,
+      "loss": 2.2712,
+      "step": 5025
+    },
+    {
+      "epoch": 0.6505485969511603,
+      "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": 1.5166015625,
+      "eval_runtime": 13.7232,
+      "eval_samples_per_second": 3.206,
+      "eval_steps_per_second": 0.219,
+      "step": 5025
+    },
+    {
+      "epoch": 0.6506780593585137,
+      "grad_norm": 1.343725655668395,
+      "learning_rate": 4.098665480072254e-07,
+      "loss": 2.2896,
+      "step": 5026
+    },
+    {
+      "epoch": 0.6508075217658672,
+      "grad_norm": 1.2760485238642996,
+      "learning_rate": 4.096013550641864e-07,
+      "loss": 2.2854,
+      "step": 5027
+    },
+    {
+      "epoch": 0.6509369841732207,
+      "grad_norm": 2.0562967627072375,
+      "learning_rate": 4.0933620612068683e-07,
+      "loss": 2.4966,
+      "step": 5028
+    },
+    {
+      "epoch": 0.6510664465805742,
+      "grad_norm": 2.464894842372267,
+      "learning_rate": 4.090711012308882e-07,
+      "loss": 2.509,
+      "step": 5029
+    },
+    {
+      "epoch": 0.6511959089879277,
+      "grad_norm": 1.3179228354792176,
+      "learning_rate": 4.088060404489436e-07,
+      "loss": 2.3752,
+      "step": 5030
+    },
+    {
+      "epoch": 0.6511959089879277,
+      "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": 1.5228382349014282,
+      "eval_runtime": 13.6436,
+      "eval_samples_per_second": 3.225,
+      "eval_steps_per_second": 0.22,
+      "step": 5030
+    },
+    {
+      "epoch": 0.6513253713952811,
+      "grad_norm": 3.858383770113083,
+      "learning_rate": 4.0854102382899753e-07,
+      "loss": 2.7832,
+      "step": 5031
+    },
+    {
+      "epoch": 0.6514548338026346,
+      "grad_norm": 2.513984411162413,
+      "learning_rate": 4.082760514251848e-07,
+      "loss": 2.4937,
+      "step": 5032
+    },
+    {
+      "epoch": 0.651584296209988,
+      "grad_norm": 2.5458047701558995,
+      "learning_rate": 4.0801112329163123e-07,
+      "loss": 2.2312,
+      "step": 5033
+    },
+    {
+      "epoch": 0.6517137586173415,
+      "grad_norm": 1.6480911628021269,
+      "learning_rate": 4.0774623948245414e-07,
+      "loss": 2.3281,
+      "step": 5034
+    },
+    {
+      "epoch": 0.6518432210246949,
+      "grad_norm": 1.4769260925158803,
+      "learning_rate": 4.0748140005176096e-07,
+      "loss": 2.4873,
+      "step": 5035
+    },
+    {
+      "epoch": 0.6518432210246949,
+      "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": 1.5186434984207153,
+      "eval_runtime": 14.5776,
+      "eval_samples_per_second": 3.018,
+      "eval_steps_per_second": 0.206,
+      "step": 5035
+    },
+    {
+      "epoch": 0.6519726834320484,
+      "grad_norm": 1.378453756377644,
+      "learning_rate": 4.0721660505365073e-07,
+      "loss": 2.5588,
+      "step": 5036
+    },
+    {
+      "epoch": 0.6521021458394018,
+      "grad_norm": 2.730489453808209,
+      "learning_rate": 4.069518545422132e-07,
+      "loss": 2.4067,
+      "step": 5037
+    },
+    {
+      "epoch": 0.6522316082467553,
+      "grad_norm": 3.346448500996533,
+      "learning_rate": 4.06687148571529e-07,
+      "loss": 2.5771,
+      "step": 5038
+    },
+    {
+      "epoch": 0.6523610706541089,
+      "grad_norm": 1.189941279327231,
+      "learning_rate": 4.0642248719566983e-07,
+      "loss": 2.3987,
+      "step": 5039
+    },
+    {
+      "epoch": 0.6524905330614623,
+      "grad_norm": 1.934587604175701,
+      "learning_rate": 4.06157870468698e-07,
+      "loss": 2.5571,
+      "step": 5040
+    },
+    {
+      "epoch": 0.6524905330614623,
+      "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": 1.5215731859207153,
+      "eval_runtime": 14.5379,
+      "eval_samples_per_second": 3.027,
+      "eval_steps_per_second": 0.206,
+      "step": 5040
+    },
+    {
+      "epoch": 0.6526199954688158,
+      "grad_norm": 2.6843035624478992,
+      "learning_rate": 4.058932984446667e-07,
+      "loss": 2.3965,
+      "step": 5041
+    },
+    {
+      "epoch": 0.6527494578761692,
+      "grad_norm": 2.2576884183651367,
+      "learning_rate": 4.056287711776204e-07,
+      "loss": 2.5122,
+      "step": 5042
+    },
+    {
+      "epoch": 0.6528789202835227,
+      "grad_norm": 1.3509512030833801,
+      "learning_rate": 4.053642887215939e-07,
+      "loss": 2.3586,
+      "step": 5043
+    },
+    {
+      "epoch": 0.6530083826908761,
+      "grad_norm": 1.5034457112503246,
+      "learning_rate": 4.050998511306134e-07,
+      "loss": 2.5308,
+      "step": 5044
+    },
+    {
+      "epoch": 0.6531378450982296,
+      "grad_norm": 2.443284834086762,
+      "learning_rate": 4.0483545845869556e-07,
+      "loss": 2.3098,
+      "step": 5045
+    },
+    {
+      "epoch": 0.6531378450982296,
+      "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": 1.5187100172042847,
+      "eval_runtime": 13.6516,
+      "eval_samples_per_second": 3.223,
+      "eval_steps_per_second": 0.22,
+      "step": 5045
+    },
+    {
+      "epoch": 0.653267307505583,
+      "grad_norm": 1.4599420498327067,
+      "learning_rate": 4.0457111075984767e-07,
+      "loss": 2.6724,
+      "step": 5046
+    },
+    {
+      "epoch": 0.6533967699129365,
+      "grad_norm": 5.067183238761789,
+      "learning_rate": 4.0430680808806825e-07,
+      "loss": 2.7002,
+      "step": 5047
+    },
+    {
+      "epoch": 0.65352623232029,
+      "grad_norm": 1.670059141004386,
+      "learning_rate": 4.0404255049734667e-07,
+      "loss": 2.3376,
+      "step": 5048
+    },
+    {
+      "epoch": 0.6536556947276435,
+      "grad_norm": 2.5271320683805483,
+      "learning_rate": 4.037783380416625e-07,
+      "loss": 2.2805,
+      "step": 5049
+    },
+    {
+      "epoch": 0.653785157134997,
+      "grad_norm": 2.481168647184336,
+      "learning_rate": 4.0351417077498727e-07,
+      "loss": 2.4534,
+      "step": 5050
+    },
+    {
+      "epoch": 0.653785157134997,
+      "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": 1.5179775953292847,
+      "eval_runtime": 13.5056,
+      "eval_samples_per_second": 3.258,
+      "eval_steps_per_second": 0.222,
+      "step": 5050
+    },
+    {
+      "epoch": 0.6539146195423504,
+      "grad_norm": 3.3264588868860114,
+      "learning_rate": 4.0325004875128167e-07,
+      "loss": 2.7644,
+      "step": 5051
+    },
+    {
+      "epoch": 0.6540440819497039,
+      "grad_norm": 2.2105953991858085,
+      "learning_rate": 4.0298597202449824e-07,
+      "loss": 2.4448,
+      "step": 5052
+    },
+    {
+      "epoch": 0.6541735443570573,
+      "grad_norm": 2.9751967752117374,
+      "learning_rate": 4.0272194064857995e-07,
+      "loss": 2.5493,
+      "step": 5053
+    },
+    {
+      "epoch": 0.6543030067644108,
+      "grad_norm": 1.6447415487331738,
+      "learning_rate": 4.0245795467746067e-07,
+      "loss": 2.1528,
+      "step": 5054
+    },
+    {
+      "epoch": 0.6544324691717642,
+      "grad_norm": 1.6295542111171875,
+      "learning_rate": 4.02194014165065e-07,
+      "loss": 2.4771,
+      "step": 5055
+    },
+    {
+      "epoch": 0.6544324691717642,
+      "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": 1.5200417041778564,
+      "eval_runtime": 13.9813,
+      "eval_samples_per_second": 3.147,
+      "eval_steps_per_second": 0.215,
+      "step": 5055
+    },
+    {
+      "epoch": 0.6545619315791177,
+      "grad_norm": 1.7161823327021302,
+      "learning_rate": 4.0193011916530823e-07,
+      "loss": 2.2747,
+      "step": 5056
+    },
+    {
+      "epoch": 0.6546913939864711,
+      "grad_norm": 1.8586556236181142,
+      "learning_rate": 4.016662697320956e-07,
+      "loss": 2.3313,
+      "step": 5057
+    },
+    {
+      "epoch": 0.6548208563938246,
+      "grad_norm": 2.810041862092068,
+      "learning_rate": 4.0140246591932433e-07,
+      "loss": 2.4409,
+      "step": 5058
+    },
+    {
+      "epoch": 0.6549503188011782,
+      "grad_norm": 2.674635196176011,
+      "learning_rate": 4.011387077808815e-07,
+      "loss": 2.2977,
+      "step": 5059
+    },
+    {
+      "epoch": 0.6550797812085316,
+      "grad_norm": 2.0085069906240887,
+      "learning_rate": 4.008749953706451e-07,
+      "loss": 2.606,
+      "step": 5060
+    },
+    {
+      "epoch": 0.6550797812085316,
+      "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": 1.5186434984207153,
+      "eval_runtime": 14.77,
+      "eval_samples_per_second": 2.979,
+      "eval_steps_per_second": 0.203,
+      "step": 5060
+    },
+    {
+      "epoch": 0.6552092436158851,
+      "grad_norm": 1.5851830352913399,
+      "learning_rate": 4.006113287424837e-07,
+      "loss": 2.4102,
+      "step": 5061
+    },
+    {
+      "epoch": 0.6553387060232385,
+      "grad_norm": 1.4390640008720585,
+      "learning_rate": 4.003477079502566e-07,
+      "loss": 2.3667,
+      "step": 5062
+    },
+    {
+      "epoch": 0.655468168430592,
+      "grad_norm": 1.7512951961351804,
+      "learning_rate": 4.000841330478136e-07,
+      "loss": 2.3301,
+      "step": 5063
+    },
+    {
+      "epoch": 0.6555976308379454,
+      "grad_norm": 1.3290034689401424,
+      "learning_rate": 3.998206040889952e-07,
+      "loss": 2.1821,
+      "step": 5064
+    },
+    {
+      "epoch": 0.6557270932452989,
+      "grad_norm": 2.2412698819182113,
+      "learning_rate": 3.9955712112763267e-07,
+      "loss": 2.47,
+      "step": 5065
+    },
+    {
+      "epoch": 0.6557270932452989,
+      "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": 1.5164684057235718,
+      "eval_runtime": 14.411,
+      "eval_samples_per_second": 3.053,
+      "eval_steps_per_second": 0.208,
+      "step": 5065
+    },
+    {
+      "epoch": 0.6558565556526523,
+      "grad_norm": 3.6718753992178654,
+      "learning_rate": 3.9929368421754765e-07,
+      "loss": 2.5491,
+      "step": 5066
+    },
+    {
+      "epoch": 0.6559860180600058,
+      "grad_norm": 1.228562881685534,
+      "learning_rate": 3.990302934125526e-07,
+      "loss": 2.3064,
+      "step": 5067
+    },
+    {
+      "epoch": 0.6561154804673592,
+      "grad_norm": 1.5334290980141807,
+      "learning_rate": 3.9876694876645016e-07,
+      "loss": 2.3853,
+      "step": 5068
+    },
+    {
+      "epoch": 0.6562449428747128,
+      "grad_norm": 2.1654949654858404,
+      "learning_rate": 3.9850365033303396e-07,
+      "loss": 2.4165,
+      "step": 5069
+    },
+    {
+      "epoch": 0.6563744052820663,
+      "grad_norm": 2.4019823823346673,
+      "learning_rate": 3.982403981660881e-07,
+      "loss": 2.3071,
+      "step": 5070
+    },
+    {
+      "epoch": 0.6563744052820663,
+      "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": 1.5206631422042847,
+      "eval_runtime": 15.7309,
+      "eval_samples_per_second": 2.797,
+      "eval_steps_per_second": 0.191,
+      "step": 5070
+    },
+    {
+      "epoch": 0.6565038676894197,
+      "grad_norm": 3.4869967089434932,
+      "learning_rate": 3.979771923193871e-07,
+      "loss": 2.8848,
+      "step": 5071
+    },
+    {
+      "epoch": 0.6566333300967732,
+      "grad_norm": 2.1100427911813036,
+      "learning_rate": 3.97714032846696e-07,
+      "loss": 2.8606,
+      "step": 5072
+    },
+    {
+      "epoch": 0.6567627925041266,
+      "grad_norm": 1.3733993212918951,
+      "learning_rate": 3.974509198017709e-07,
+      "loss": 2.3398,
+      "step": 5073
+    },
+    {
+      "epoch": 0.6568922549114801,
+      "grad_norm": 2.6078588587997893,
+      "learning_rate": 3.9718785323835747e-07,
+      "loss": 2.3777,
+      "step": 5074
+    },
+    {
+      "epoch": 0.6570217173188335,
+      "grad_norm": 1.8913946025395285,
+      "learning_rate": 3.969248332101925e-07,
+      "loss": 2.3518,
+      "step": 5075
+    },
+    {
+      "epoch": 0.6570217173188335,
+      "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": 1.52001953125,
+      "eval_runtime": 13.3703,
+      "eval_samples_per_second": 3.291,
+      "eval_steps_per_second": 0.224,
+      "step": 5075
+    },
+    {
+      "epoch": 0.657151179726187,
+      "grad_norm": 2.120538980944094,
+      "learning_rate": 3.9666185977100314e-07,
+      "loss": 2.5015,
+      "step": 5076
+    },
+    {
+      "epoch": 0.6572806421335404,
+      "grad_norm": 1.7462444312062881,
+      "learning_rate": 3.9639893297450714e-07,
+      "loss": 2.3416,
+      "step": 5077
+    },
+    {
+      "epoch": 0.657410104540894,
+      "grad_norm": 1.3142990010739095,
+      "learning_rate": 3.96136052874413e-07,
+      "loss": 2.4033,
+      "step": 5078
+    },
+    {
+      "epoch": 0.6575395669482474,
+      "grad_norm": 2.775288309021671,
+      "learning_rate": 3.9587321952441857e-07,
+      "loss": 2.3828,
+      "step": 5079
+    },
+    {
+      "epoch": 0.6576690293556009,
+      "grad_norm": 1.5540206084955568,
+      "learning_rate": 3.9561043297821316e-07,
+      "loss": 2.5361,
+      "step": 5080
+    },
+    {
+      "epoch": 0.6576690293556009,
+      "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": 1.5213512182235718,
+      "eval_runtime": 14.2628,
+      "eval_samples_per_second": 3.085,
+      "eval_steps_per_second": 0.21,
+      "step": 5080
+    },
+    {
+      "epoch": 0.6577984917629544,
+      "grad_norm": 1.2668582158650334,
+      "learning_rate": 3.9534769328947647e-07,
+      "loss": 2.3352,
+      "step": 5081
+    },
+    {
+      "epoch": 0.6579279541703078,
+      "grad_norm": 1.3170165531362004,
+      "learning_rate": 3.9508500051187826e-07,
+      "loss": 2.3354,
+      "step": 5082
+    },
+    {
+      "epoch": 0.6580574165776613,
+      "grad_norm": 3.303702335316497,
+      "learning_rate": 3.9482235469907884e-07,
+      "loss": 2.1265,
+      "step": 5083
+    },
+    {
+      "epoch": 0.6581868789850147,
+      "grad_norm": 1.808692046422752,
+      "learning_rate": 3.9455975590472914e-07,
+      "loss": 2.4451,
+      "step": 5084
+    },
+    {
+      "epoch": 0.6583163413923682,
+      "grad_norm": 1.3869193689002375,
+      "learning_rate": 3.942972041824699e-07,
+      "loss": 2.2463,
+      "step": 5085
+    },
+    {
+      "epoch": 0.6583163413923682,
+      "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": 1.5206853151321411,
+      "eval_runtime": 15.0582,
+      "eval_samples_per_second": 2.922,
+      "eval_steps_per_second": 0.199,
+      "step": 5085
+    },
+    {
+      "epoch": 0.6584458037997216,
+      "grad_norm": 2.0581482240865334,
+      "learning_rate": 3.9403469958593286e-07,
+      "loss": 2.51,
+      "step": 5086
+    },
+    {
+      "epoch": 0.6585752662070751,
+      "grad_norm": 1.6978150213233507,
+      "learning_rate": 3.937722421687399e-07,
+      "loss": 2.4026,
+      "step": 5087
+    },
+    {
+      "epoch": 0.6587047286144286,
+      "grad_norm": 1.5749043485762106,
+      "learning_rate": 3.935098319845032e-07,
+      "loss": 2.5483,
+      "step": 5088
+    },
+    {
+      "epoch": 0.6588341910217821,
+      "grad_norm": 1.7346630258114264,
+      "learning_rate": 3.932474690868256e-07,
+      "loss": 2.6348,
+      "step": 5089
+    },
+    {
+      "epoch": 0.6589636534291355,
+      "grad_norm": 1.8385748672661781,
+      "learning_rate": 3.9298515352929974e-07,
+      "loss": 2.396,
+      "step": 5090
+    },
+    {
+      "epoch": 0.6589636534291355,
+      "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": 1.520063877105713,
+      "eval_runtime": 14.2696,
+      "eval_samples_per_second": 3.083,
+      "eval_steps_per_second": 0.21,
+      "step": 5090
+    },
+    {
+      "epoch": 0.659093115836489,
+      "grad_norm": 1.7919133923344202,
+      "learning_rate": 3.927228853655089e-07,
+      "loss": 2.2737,
+      "step": 5091
+    },
+    {
+      "epoch": 0.6592225782438424,
+      "grad_norm": 2.472952432846999,
+      "learning_rate": 3.9246066464902685e-07,
+      "loss": 2.3416,
+      "step": 5092
+    },
+    {
+      "epoch": 0.6593520406511959,
+      "grad_norm": 1.8409635578852772,
+      "learning_rate": 3.9219849143341745e-07,
+      "loss": 2.3484,
+      "step": 5093
+    },
+    {
+      "epoch": 0.6594815030585494,
+      "grad_norm": 1.5354990759147005,
+      "learning_rate": 3.9193636577223464e-07,
+      "loss": 2.2981,
+      "step": 5094
+    },
+    {
+      "epoch": 0.6596109654659028,
+      "grad_norm": 2.8406258733537837,
+      "learning_rate": 3.916742877190235e-07,
+      "loss": 2.4692,
+      "step": 5095
+    },
+    {
+      "epoch": 0.6596109654659028,
+      "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": 1.521484375,
+      "eval_runtime": 14.1393,
+      "eval_samples_per_second": 3.112,
+      "eval_steps_per_second": 0.212,
+      "step": 5095
+    },
+    {
+      "epoch": 0.6597404278732563,
+      "grad_norm": 1.3670449957574278,
+      "learning_rate": 3.914122573273181e-07,
+      "loss": 2.2891,
+      "step": 5096
+    },
+    {
+      "epoch": 0.6598698902806097,
+      "grad_norm": 2.506823950901503,
+      "learning_rate": 3.9115027465064375e-07,
+      "loss": 2.5471,
+      "step": 5097
+    },
+    {
+      "epoch": 0.6599993526879633,
+      "grad_norm": 1.757220271206005,
+      "learning_rate": 3.908883397425157e-07,
+      "loss": 2.4419,
+      "step": 5098
+    },
+    {
+      "epoch": 0.6601288150953167,
+      "grad_norm": 1.1859617161818887,
+      "learning_rate": 3.9062645265643946e-07,
+      "loss": 2.2988,
+      "step": 5099
+    },
+    {
+      "epoch": 0.6602582775026702,
+      "grad_norm": 2.067553285362183,
+      "learning_rate": 3.903646134459112e-07,
+      "loss": 2.4841,
+      "step": 5100
+    },
+    {
+      "epoch": 0.6602582775026702,
+      "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": 1.521440029144287,
+      "eval_runtime": 14.6948,
+      "eval_samples_per_second": 2.994,
+      "eval_steps_per_second": 0.204,
+      "step": 5100
+    },
+    {
+      "epoch": 0.6603877399100236,
+      "grad_norm": 1.4590953613474575,
+      "learning_rate": 3.901028221644162e-07,
+      "loss": 2.3407,
+      "step": 5101
+    },
+    {
+      "epoch": 0.6605172023173771,
+      "grad_norm": 2.3624622050943693,
+      "learning_rate": 3.898410788654307e-07,
+      "loss": 2.4902,
+      "step": 5102
+    },
+    {
+      "epoch": 0.6606466647247305,
+      "grad_norm": 1.796253910186038,
+      "learning_rate": 3.895793836024215e-07,
+      "loss": 2.5266,
+      "step": 5103
+    },
+    {
+      "epoch": 0.660776127132084,
+      "grad_norm": 4.028303478472632,
+      "learning_rate": 3.8931773642884493e-07,
+      "loss": 2.3796,
+      "step": 5104
+    },
+    {
+      "epoch": 0.6609055895394375,
+      "grad_norm": 1.6015025071514128,
+      "learning_rate": 3.890561373981477e-07,
+      "loss": 2.5498,
+      "step": 5105
+    },
+    {
+      "epoch": 0.6609055895394375,
+      "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": 1.5191982984542847,
+      "eval_runtime": 14.5094,
+      "eval_samples_per_second": 3.033,
+      "eval_steps_per_second": 0.207,
+      "step": 5105
+    },
+    {
+      "epoch": 0.6610350519467909,
+      "grad_norm": 1.6189397534895302,
+      "learning_rate": 3.8879458656376707e-07,
+      "loss": 2.5557,
+      "step": 5106
+    },
+    {
+      "epoch": 0.6611645143541444,
+      "grad_norm": 1.7465514629863457,
+      "learning_rate": 3.885330839791296e-07,
+      "loss": 2.2998,
+      "step": 5107
+    },
+    {
+      "epoch": 0.6612939767614979,
+      "grad_norm": 2.872800723127391,
+      "learning_rate": 3.8827162969765265e-07,
+      "loss": 2.7788,
+      "step": 5108
+    },
+    {
+      "epoch": 0.6614234391688514,
+      "grad_norm": 1.1646514624490423,
+      "learning_rate": 3.8801022377274356e-07,
+      "loss": 2.3994,
+      "step": 5109
+    },
+    {
+      "epoch": 0.6615529015762048,
+      "grad_norm": 3.8993895184748006,
+      "learning_rate": 3.8774886625779985e-07,
+      "loss": 2.7563,
+      "step": 5110
+    },
+    {
+      "epoch": 0.6615529015762048,
+      "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": 1.520552158355713,
+      "eval_runtime": 14.2053,
+      "eval_samples_per_second": 3.097,
+      "eval_steps_per_second": 0.211,
+      "step": 5110
+    },
+    {
+      "epoch": 0.6616823639835583,
+      "grad_norm": 1.6761707718581333,
+      "learning_rate": 3.874875572062091e-07,
+      "loss": 2.5566,
+      "step": 5111
+    },
+    {
+      "epoch": 0.6618118263909117,
+      "grad_norm": 2.42446852769922,
+      "learning_rate": 3.872262966713491e-07,
+      "loss": 2.5781,
+      "step": 5112
+    },
+    {
+      "epoch": 0.6619412887982652,
+      "grad_norm": 1.4991038214552679,
+      "learning_rate": 3.869650847065873e-07,
+      "loss": 2.4695,
+      "step": 5113
+    },
+    {
+      "epoch": 0.6620707512056186,
+      "grad_norm": 1.3298492183229564,
+      "learning_rate": 3.867039213652817e-07,
+      "loss": 2.4062,
+      "step": 5114
+    },
+    {
+      "epoch": 0.6622002136129721,
+      "grad_norm": 2.0585475192212566,
+      "learning_rate": 3.864428067007803e-07,
+      "loss": 2.396,
+      "step": 5115
+    },
+    {
+      "epoch": 0.6622002136129721,
+      "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": 1.5179775953292847,
+      "eval_runtime": 14.5302,
+      "eval_samples_per_second": 3.028,
+      "eval_steps_per_second": 0.206,
+      "step": 5115
+    },
+    {
+      "epoch": 0.6623296760203256,
+      "grad_norm": 1.962227896635834,
+      "learning_rate": 3.8618174076642076e-07,
+      "loss": 2.3892,
+      "step": 5116
+    },
+    {
+      "epoch": 0.6624591384276791,
+      "grad_norm": 2.287359298205945,
+      "learning_rate": 3.8592072361553157e-07,
+      "loss": 2.4265,
+      "step": 5117
+    },
+    {
+      "epoch": 0.6625886008350326,
+      "grad_norm": 1.4751271525261889,
+      "learning_rate": 3.856597553014303e-07,
+      "loss": 2.3013,
+      "step": 5118
+    },
+    {
+      "epoch": 0.662718063242386,
+      "grad_norm": 1.54005879839691,
+      "learning_rate": 3.853988358774253e-07,
+      "loss": 2.3831,
+      "step": 5119
+    },
+    {
+      "epoch": 0.6628475256497395,
+      "grad_norm": 3.7889978247804117,
+      "learning_rate": 3.8513796539681444e-07,
+      "loss": 2.4536,
+      "step": 5120
+    },
+    {
+      "epoch": 0.6628475256497395,
+      "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": 1.5213955640792847,
+      "eval_runtime": 14.1483,
+      "eval_samples_per_second": 3.11,
+      "eval_steps_per_second": 0.212,
+      "step": 5120
+    },
+    {
+      "epoch": 0.6629769880570929,
+      "grad_norm": 1.4772823111355498,
+      "learning_rate": 3.8487714391288595e-07,
+      "loss": 2.3752,
+      "step": 5121
+    },
+    {
+      "epoch": 0.6631064504644464,
+      "grad_norm": 1.8540814294947863,
+      "learning_rate": 3.846163714789179e-07,
+      "loss": 2.395,
+      "step": 5122
+    },
+    {
+      "epoch": 0.6632359128717998,
+      "grad_norm": 1.5728833590556046,
+      "learning_rate": 3.843556481481786e-07,
+      "loss": 2.3787,
+      "step": 5123
+    },
+    {
+      "epoch": 0.6633653752791533,
+      "grad_norm": 1.8407201326668305,
+      "learning_rate": 3.840949739739256e-07,
+      "loss": 2.5516,
+      "step": 5124
+    },
+    {
+      "epoch": 0.6634948376865067,
+      "grad_norm": 1.1281453230800251,
+      "learning_rate": 3.8383434900940695e-07,
+      "loss": 2.1985,
+      "step": 5125
+    },
+    {
+      "epoch": 0.6634948376865067,
+      "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": 1.5184437036514282,
+      "eval_runtime": 15.7301,
+      "eval_samples_per_second": 2.797,
+      "eval_steps_per_second": 0.191,
+      "step": 5125
+    },
+    {
+      "epoch": 0.6636243000938602,
+      "grad_norm": 1.6718655606211663,
+      "learning_rate": 3.835737733078608e-07,
+      "loss": 2.6182,
+      "step": 5126
+    },
+    {
+      "epoch": 0.6637537625012138,
+      "grad_norm": 1.2695176838103426,
+      "learning_rate": 3.8331324692251494e-07,
+      "loss": 2.177,
+      "step": 5127
+    },
+    {
+      "epoch": 0.6638832249085672,
+      "grad_norm": 4.087295517944597,
+      "learning_rate": 3.8305276990658753e-07,
+      "loss": 2.4541,
+      "step": 5128
+    },
+    {
+      "epoch": 0.6640126873159207,
+      "grad_norm": 1.8454832240698456,
+      "learning_rate": 3.827923423132854e-07,
+      "loss": 2.5847,
+      "step": 5129
+    },
+    {
+      "epoch": 0.6641421497232741,
+      "grad_norm": 1.839616472821154,
+      "learning_rate": 3.825319641958068e-07,
+      "loss": 2.2156,
+      "step": 5130
+    },
+    {
+      "epoch": 0.6641421497232741,
+      "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": 1.5199307203292847,
+      "eval_runtime": 13.9844,
+      "eval_samples_per_second": 3.146,
+      "eval_steps_per_second": 0.215,
+      "step": 5130
+    },
+    {
+      "epoch": 0.6642716121306276,
+      "grad_norm": 3.2461673156177198,
+      "learning_rate": 3.8227163560733924e-07,
+      "loss": 2.7466,
+      "step": 5131
+    },
+    {
+      "epoch": 0.664401074537981,
+      "grad_norm": 1.7380332264186027,
+      "learning_rate": 3.820113566010599e-07,
+      "loss": 2.3745,
+      "step": 5132
+    },
+    {
+      "epoch": 0.6645305369453345,
+      "grad_norm": 2.344446791536525,
+      "learning_rate": 3.8175112723013615e-07,
+      "loss": 2.4751,
+      "step": 5133
+    },
+    {
+      "epoch": 0.6646599993526879,
+      "grad_norm": 1.2471046012903564,
+      "learning_rate": 3.814909475477252e-07,
+      "loss": 2.1833,
+      "step": 5134
+    },
+    {
+      "epoch": 0.6647894617600414,
+      "grad_norm": 2.8199112596985865,
+      "learning_rate": 3.812308176069739e-07,
+      "loss": 2.7944,
+      "step": 5135
+    },
+    {
+      "epoch": 0.6647894617600414,
+      "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": 1.519242763519287,
+      "eval_runtime": 14.1271,
+      "eval_samples_per_second": 3.115,
+      "eval_steps_per_second": 0.212,
+      "step": 5135
+    },
+    {
+      "epoch": 0.6649189241673948,
+      "grad_norm": 1.5752387910162489,
+      "learning_rate": 3.809707374610189e-07,
+      "loss": 2.4453,
+      "step": 5136
+    },
+    {
+      "epoch": 0.6650483865747484,
+      "grad_norm": 4.743906889161572,
+      "learning_rate": 3.807107071629871e-07,
+      "loss": 2.7966,
+      "step": 5137
+    },
+    {
+      "epoch": 0.6651778489821019,
+      "grad_norm": 1.5386285627355307,
+      "learning_rate": 3.8045072676599485e-07,
+      "loss": 2.5845,
+      "step": 5138
+    },
+    {
+      "epoch": 0.6653073113894553,
+      "grad_norm": 1.854354073363231,
+      "learning_rate": 3.801907963231487e-07,
+      "loss": 2.4751,
+      "step": 5139
+    },
+    {
+      "epoch": 0.6654367737968088,
+      "grad_norm": 2.430741100944076,
+      "learning_rate": 3.799309158875442e-07,
+      "loss": 2.4009,
+      "step": 5140
+    },
+    {
+      "epoch": 0.6654367737968088,
+      "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": 1.517777919769287,
+      "eval_runtime": 13.9566,
+      "eval_samples_per_second": 3.153,
+      "eval_steps_per_second": 0.215,
+      "step": 5140
+    },
+    {
+      "epoch": 0.6655662362041622,
+      "grad_norm": 1.680177128866981,
+      "learning_rate": 3.7967108551226745e-07,
+      "loss": 2.4585,
+      "step": 5141
+    },
+    {
+      "epoch": 0.6656956986115157,
+      "grad_norm": 2.0435515407621634,
+      "learning_rate": 3.794113052503942e-07,
+      "loss": 2.4807,
+      "step": 5142
+    },
+    {
+      "epoch": 0.6658251610188691,
+      "grad_norm": 2.2911964132494647,
+      "learning_rate": 3.7915157515498956e-07,
+      "loss": 2.6887,
+      "step": 5143
+    },
+    {
+      "epoch": 0.6659546234262226,
+      "grad_norm": 1.5835927017498963,
+      "learning_rate": 3.7889189527910876e-07,
+      "loss": 2.4379,
+      "step": 5144
+    },
+    {
+      "epoch": 0.666084085833576,
+      "grad_norm": 1.3622481158306292,
+      "learning_rate": 3.7863226567579714e-07,
+      "loss": 2.3567,
+      "step": 5145
+    },
+    {
+      "epoch": 0.666084085833576,
+      "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": 1.519975185394287,
+      "eval_runtime": 13.927,
+      "eval_samples_per_second": 3.159,
+      "eval_steps_per_second": 0.215,
+      "step": 5145
+    },
+    {
+      "epoch": 0.6662135482409295,
+      "grad_norm": 1.8280483503986014,
+      "learning_rate": 3.783726863980886e-07,
+      "loss": 2.3433,
+      "step": 5146
+    },
+    {
+      "epoch": 0.666343010648283,
+      "grad_norm": 1.8989326011540517,
+      "learning_rate": 3.781131574990078e-07,
+      "loss": 2.4128,
+      "step": 5147
+    },
+    {
+      "epoch": 0.6664724730556365,
+      "grad_norm": 1.1719131934732134,
+      "learning_rate": 3.7785367903156856e-07,
+      "loss": 2.4124,
+      "step": 5148
+    },
+    {
+      "epoch": 0.66660193546299,
+      "grad_norm": 2.6265563704726467,
+      "learning_rate": 3.7759425104877507e-07,
+      "loss": 2.3127,
+      "step": 5149
+    },
+    {
+      "epoch": 0.6667313978703434,
+      "grad_norm": 1.9314737227858774,
+      "learning_rate": 3.773348736036204e-07,
+      "loss": 2.6221,
+      "step": 5150
+    },
+    {
+      "epoch": 0.6667313978703434,
+      "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": 1.5184437036514282,
+      "eval_runtime": 14.5884,
+      "eval_samples_per_second": 3.016,
+      "eval_steps_per_second": 0.206,
+      "step": 5150
+    },
+    {
+      "epoch": 0.6668608602776969,
+      "grad_norm": 1.4916318311367764,
+      "learning_rate": 3.77075546749088e-07,
+      "loss": 2.4309,
+      "step": 5151
+    },
+    {
+      "epoch": 0.6669903226850503,
+      "grad_norm": 1.1933919085341167,
+      "learning_rate": 3.768162705381499e-07,
+      "loss": 2.3296,
+      "step": 5152
+    },
+    {
+      "epoch": 0.6671197850924038,
+      "grad_norm": 2.0277253775961017,
+      "learning_rate": 3.7655704502376934e-07,
+      "loss": 2.3223,
+      "step": 5153
+    },
+    {
+      "epoch": 0.6672492474997572,
+      "grad_norm": 1.9699504523553544,
+      "learning_rate": 3.762978702588979e-07,
+      "loss": 2.4688,
+      "step": 5154
+    },
+    {
+      "epoch": 0.6673787099071107,
+      "grad_norm": 1.2773131139319953,
+      "learning_rate": 3.7603874629647744e-07,
+      "loss": 2.2078,
+      "step": 5155
+    },
+    {
+      "epoch": 0.6673787099071107,
+      "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": 1.5199528932571411,
+      "eval_runtime": 14.8951,
+      "eval_samples_per_second": 2.954,
+      "eval_steps_per_second": 0.201,
+      "step": 5155
+    },
+    {
+      "epoch": 0.6675081723144641,
+      "grad_norm": 1.5437265689498072,
+      "learning_rate": 3.757796731894393e-07,
+      "loss": 2.3879,
+      "step": 5156
+    },
+    {
+      "epoch": 0.6676376347218177,
+      "grad_norm": 2.414722426905868,
+      "learning_rate": 3.7552065099070444e-07,
+      "loss": 2.5288,
+      "step": 5157
+    },
+    {
+      "epoch": 0.6677670971291711,
+      "grad_norm": 1.8267964633701326,
+      "learning_rate": 3.752616797531832e-07,
+      "loss": 2.4658,
+      "step": 5158
+    },
+    {
+      "epoch": 0.6678965595365246,
+      "grad_norm": 1.5019852449538464,
+      "learning_rate": 3.7500275952977587e-07,
+      "loss": 2.2144,
+      "step": 5159
+    },
+    {
+      "epoch": 0.6680260219438781,
+      "grad_norm": 1.954640115840008,
+      "learning_rate": 3.747438903733722e-07,
+      "loss": 2.4492,
+      "step": 5160
+    },
+    {
+      "epoch": 0.6680260219438781,
+      "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": 1.5178444385528564,
+      "eval_runtime": 14.0502,
+      "eval_samples_per_second": 3.132,
+      "eval_steps_per_second": 0.214,
+      "step": 5160
+    },
+    {
+      "epoch": 0.6681554843512315,
+      "grad_norm": 1.8310980985668466,
+      "learning_rate": 3.744850723368512e-07,
+      "loss": 2.5166,
+      "step": 5161
+    },
+    {
+      "epoch": 0.668284946758585,
+      "grad_norm": 1.6287100846845086,
+      "learning_rate": 3.742263054730821e-07,
+      "loss": 2.3857,
+      "step": 5162
+    },
+    {
+      "epoch": 0.6684144091659384,
+      "grad_norm": 3.895947084589344,
+      "learning_rate": 3.73967589834923e-07,
+      "loss": 2.5234,
+      "step": 5163
+    },
+    {
+      "epoch": 0.6685438715732919,
+      "grad_norm": 1.9194024118819928,
+      "learning_rate": 3.737089254752218e-07,
+      "loss": 2.2913,
+      "step": 5164
+    },
+    {
+      "epoch": 0.6686733339806453,
+      "grad_norm": 2.751060996527409,
+      "learning_rate": 3.73450312446816e-07,
+      "loss": 2.4575,
+      "step": 5165
+    },
+    {
+      "epoch": 0.6686733339806453,
+      "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": 1.5171786546707153,
+      "eval_runtime": 14.0054,
+      "eval_samples_per_second": 3.142,
+      "eval_steps_per_second": 0.214,
+      "step": 5165
+    },
+    {
+      "epoch": 0.6688027963879989,
+      "grad_norm": 1.8718206248093348,
+      "learning_rate": 3.731917508025326e-07,
+      "loss": 2.5067,
+      "step": 5166
+    },
+    {
+      "epoch": 0.6689322587953523,
+      "grad_norm": 2.315044819729028,
+      "learning_rate": 3.7293324059518813e-07,
+      "loss": 2.489,
+      "step": 5167
+    },
+    {
+      "epoch": 0.6690617212027058,
+      "grad_norm": 2.24429911689122,
+      "learning_rate": 3.7267478187758833e-07,
+      "loss": 2.333,
+      "step": 5168
+    },
+    {
+      "epoch": 0.6691911836100592,
+      "grad_norm": 1.8112099638411543,
+      "learning_rate": 3.724163747025287e-07,
+      "loss": 2.3835,
+      "step": 5169
+    },
+    {
+      "epoch": 0.6693206460174127,
+      "grad_norm": 2.940289527489837,
+      "learning_rate": 3.7215801912279413e-07,
+      "loss": 2.1499,
+      "step": 5170
+    },
+    {
+      "epoch": 0.6693206460174127,
+      "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": 1.5149592161178589,
+      "eval_runtime": 13.3265,
+      "eval_samples_per_second": 3.302,
+      "eval_steps_per_second": 0.225,
+      "step": 5170
+    },
+    {
+      "epoch": 0.6694501084247662,
+      "grad_norm": 2.1048840969972122,
+      "learning_rate": 3.7189971519115894e-07,
+      "loss": 2.3857,
+      "step": 5171
+    },
+    {
+      "epoch": 0.6695795708321196,
+      "grad_norm": 1.8588938143128404,
+      "learning_rate": 3.716414629603872e-07,
+      "loss": 2.8069,
+      "step": 5172
+    },
+    {
+      "epoch": 0.6697090332394731,
+      "grad_norm": 1.4840976175737275,
+      "learning_rate": 3.7138326248323225e-07,
+      "loss": 2.7664,
+      "step": 5173
+    },
+    {
+      "epoch": 0.6698384956468265,
+      "grad_norm": 1.4695531501696821,
+      "learning_rate": 3.7112511381243623e-07,
+      "loss": 2.4041,
+      "step": 5174
+    },
+    {
+      "epoch": 0.66996795805418,
+      "grad_norm": 1.7802913169297638,
+      "learning_rate": 3.7086701700073143e-07,
+      "loss": 2.3398,
+      "step": 5175
+    },
+    {
+      "epoch": 0.66996795805418,
+      "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": 1.5157138109207153,
+      "eval_runtime": 13.1817,
+      "eval_samples_per_second": 3.338,
+      "eval_steps_per_second": 0.228,
+      "step": 5175
+    },
+    {
+      "epoch": 0.6700974204615335,
+      "grad_norm": 4.72140639057189,
+      "learning_rate": 3.7060897210083947e-07,
+      "loss": 2.8203,
+      "step": 5176
+    },
+    {
+      "epoch": 0.670226882868887,
+      "grad_norm": 1.382085859640986,
+      "learning_rate": 3.703509791654713e-07,
+      "loss": 2.2473,
+      "step": 5177
+    },
+    {
+      "epoch": 0.6703563452762404,
+      "grad_norm": 1.8636648435792593,
+      "learning_rate": 3.700930382473272e-07,
+      "loss": 2.054,
+      "step": 5178
+    },
+    {
+      "epoch": 0.6704858076835939,
+      "grad_norm": 5.683917391002986,
+      "learning_rate": 3.6983514939909674e-07,
+      "loss": 2.9268,
+      "step": 5179
+    },
+    {
+      "epoch": 0.6706152700909473,
+      "grad_norm": 1.872971604088809,
+      "learning_rate": 3.695773126734589e-07,
+      "loss": 2.6372,
+      "step": 5180
+    },
+    {
+      "epoch": 0.6706152700909473,
+      "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": 1.5172230005264282,
+      "eval_runtime": 14.1143,
+      "eval_samples_per_second": 3.117,
+      "eval_steps_per_second": 0.213,
+      "step": 5180
+    },
+    {
+      "epoch": 0.6707447324983008,
+      "grad_norm": 3.841026394448021,
+      "learning_rate": 3.693195281230822e-07,
+      "loss": 2.7532,
+      "step": 5181
+    },
+    {
+      "epoch": 0.6708741949056543,
+      "grad_norm": 2.098262756639662,
+      "learning_rate": 3.690617958006243e-07,
+      "loss": 2.3953,
+      "step": 5182
+    },
+    {
+      "epoch": 0.6710036573130077,
+      "grad_norm": 2.6526068040980113,
+      "learning_rate": 3.6880411575873216e-07,
+      "loss": 2.1807,
+      "step": 5183
+    },
+    {
+      "epoch": 0.6711331197203612,
+      "grad_norm": 1.2352903410283615,
+      "learning_rate": 3.6854648805004236e-07,
+      "loss": 2.2712,
+      "step": 5184
+    },
+    {
+      "epoch": 0.6712625821277146,
+      "grad_norm": 1.555460958812336,
+      "learning_rate": 3.682889127271803e-07,
+      "loss": 2.2493,
+      "step": 5185
+    },
+    {
+      "epoch": 0.6712625821277146,
+      "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": 1.5150257349014282,
+      "eval_runtime": 14.3342,
+      "eval_samples_per_second": 3.07,
+      "eval_steps_per_second": 0.209,
+      "step": 5185
+    },
+    {
+      "epoch": 0.6713920445350682,
+      "grad_norm": 1.2240766574091408,
+      "learning_rate": 3.680313898427612e-07,
+      "loss": 2.3901,
+      "step": 5186
+    },
+    {
+      "epoch": 0.6715215069424216,
+      "grad_norm": 1.616171328978076,
+      "learning_rate": 3.677739194493892e-07,
+      "loss": 2.4905,
+      "step": 5187
+    },
+    {
+      "epoch": 0.6716509693497751,
+      "grad_norm": 1.9645193182919447,
+      "learning_rate": 3.675165015996579e-07,
+      "loss": 2.2568,
+      "step": 5188
+    },
+    {
+      "epoch": 0.6717804317571285,
+      "grad_norm": 1.3031903170890882,
+      "learning_rate": 3.6725913634615003e-07,
+      "loss": 2.3292,
+      "step": 5189
+    },
+    {
+      "epoch": 0.671909894164482,
+      "grad_norm": 2.869303765077146,
+      "learning_rate": 3.67001823741438e-07,
+      "loss": 2.5576,
+      "step": 5190
+    },
+    {
+      "epoch": 0.671909894164482,
+      "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": 1.5164240598678589,
+      "eval_runtime": 14.7051,
+      "eval_samples_per_second": 2.992,
+      "eval_steps_per_second": 0.204,
+      "step": 5190
+    },
+    {
+      "epoch": 0.6720393565718354,
+      "grad_norm": 2.2019988217144504,
+      "learning_rate": 3.667445638380826e-07,
+      "loss": 2.2393,
+      "step": 5191
+    },
+    {
+      "epoch": 0.6721688189791889,
+      "grad_norm": 2.028643793224081,
+      "learning_rate": 3.664873566886347e-07,
+      "loss": 2.5603,
+      "step": 5192
+    },
+    {
+      "epoch": 0.6722982813865424,
+      "grad_norm": 2.340970224544167,
+      "learning_rate": 3.662302023456339e-07,
+      "loss": 2.4185,
+      "step": 5193
+    },
+    {
+      "epoch": 0.6724277437938958,
+      "grad_norm": 2.3294127503985136,
+      "learning_rate": 3.659731008616094e-07,
+      "loss": 2.5471,
+      "step": 5194
+    },
+    {
+      "epoch": 0.6725572062012493,
+      "grad_norm": 2.0852620322255233,
+      "learning_rate": 3.657160522890794e-07,
+      "loss": 2.3655,
+      "step": 5195
+    },
+    {
+      "epoch": 0.6725572062012493,
+      "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": 1.5157581567764282,
+      "eval_runtime": 14.3533,
+      "eval_samples_per_second": 3.066,
+      "eval_steps_per_second": 0.209,
+      "step": 5195
+    },
+    {
+      "epoch": 0.6726866686086028,
+      "grad_norm": 2.6355346694446693,
+      "learning_rate": 3.654590566805511e-07,
+      "loss": 2.3953,
+      "step": 5196
+    },
+    {
+      "epoch": 0.6728161310159563,
+      "grad_norm": 3.0306316730134166,
+      "learning_rate": 3.6520211408852093e-07,
+      "loss": 2.4951,
+      "step": 5197
+    },
+    {
+      "epoch": 0.6729455934233097,
+      "grad_norm": 1.2556637892813427,
+      "learning_rate": 3.64945224565475e-07,
+      "loss": 2.2407,
+      "step": 5198
+    },
+    {
+      "epoch": 0.6730750558306632,
+      "grad_norm": 1.959171994137629,
+      "learning_rate": 3.6468838816388795e-07,
+      "loss": 2.5046,
+      "step": 5199
+    },
+    {
+      "epoch": 0.6732045182380166,
+      "grad_norm": 3.654400560571032,
+      "learning_rate": 3.6443160493622407e-07,
+      "loss": 2.697,
+      "step": 5200
+    },
+    {
+      "epoch": 0.6732045182380166,
+      "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": 1.5178000926971436,
+      "eval_runtime": 14.1994,
+      "eval_samples_per_second": 3.099,
+      "eval_steps_per_second": 0.211,
+      "step": 5200
+    },
+    {
+      "epoch": 0.6733339806453701,
+      "grad_norm": 1.4555148836636034,
+      "learning_rate": 3.6417487493493634e-07,
+      "loss": 2.2358,
+      "step": 5201
+    },
+    {
+      "epoch": 0.6734634430527235,
+      "grad_norm": 1.5211274824094976,
+      "learning_rate": 3.6391819821246685e-07,
+      "loss": 2.3491,
+      "step": 5202
+    },
+    {
+      "epoch": 0.673592905460077,
+      "grad_norm": 2.046669640509887,
+      "learning_rate": 3.6366157482124747e-07,
+      "loss": 2.3335,
+      "step": 5203
+    },
+    {
+      "epoch": 0.6737223678674304,
+      "grad_norm": 1.767659539458005,
+      "learning_rate": 3.634050048136983e-07,
+      "loss": 2.4586,
+      "step": 5204
+    },
+    {
+      "epoch": 0.673851830274784,
+      "grad_norm": 1.7714915188330898,
+      "learning_rate": 3.631484882422292e-07,
+      "loss": 2.4551,
+      "step": 5205
+    },
+    {
+      "epoch": 0.673851830274784,
+      "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": 1.5164684057235718,
+      "eval_runtime": 14.5253,
+      "eval_samples_per_second": 3.029,
+      "eval_steps_per_second": 0.207,
+      "step": 5205
+    },
+    {
+      "epoch": 0.6739812926821375,
+      "grad_norm": 2.049300693559449,
+      "learning_rate": 3.6289202515923936e-07,
+      "loss": 2.3489,
+      "step": 5206
+    },
+    {
+      "epoch": 0.6741107550894909,
+      "grad_norm": 1.7485913639195079,
+      "learning_rate": 3.6263561561711554e-07,
+      "loss": 2.4399,
+      "step": 5207
+    },
+    {
+      "epoch": 0.6742402174968444,
+      "grad_norm": 2.1131034879887527,
+      "learning_rate": 3.6237925966823547e-07,
+      "loss": 2.3411,
+      "step": 5208
+    },
+    {
+      "epoch": 0.6743696799041978,
+      "grad_norm": 1.8245187362374382,
+      "learning_rate": 3.6212295736496426e-07,
+      "loss": 2.3673,
+      "step": 5209
+    },
+    {
+      "epoch": 0.6744991423115513,
+      "grad_norm": 1.2856899072052732,
+      "learning_rate": 3.618667087596574e-07,
+      "loss": 2.249,
+      "step": 5210
+    },
+    {
+      "epoch": 0.6744991423115513,
+      "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": 1.5185546875,
+      "eval_runtime": 14.5418,
+      "eval_samples_per_second": 3.026,
+      "eval_steps_per_second": 0.206,
+      "step": 5210
+    },
+    {
+      "epoch": 0.6746286047189047,
+      "grad_norm": 1.6513603051575372,
+      "learning_rate": 3.61610513904659e-07,
+      "loss": 2.4089,
+      "step": 5211
+    },
+    {
+      "epoch": 0.6747580671262582,
+      "grad_norm": 2.258679709221032,
+      "learning_rate": 3.613543728523018e-07,
+      "loss": 2.625,
+      "step": 5212
+    },
+    {
+      "epoch": 0.6748875295336116,
+      "grad_norm": 1.426089270343243,
+      "learning_rate": 3.610982856549075e-07,
+      "loss": 2.3296,
+      "step": 5213
+    },
+    {
+      "epoch": 0.6750169919409651,
+      "grad_norm": 1.9765930465741033,
+      "learning_rate": 3.6084225236478767e-07,
+      "loss": 2.1831,
+      "step": 5214
+    },
+    {
+      "epoch": 0.6751464543483187,
+      "grad_norm": 2.63646888398566,
+      "learning_rate": 3.6058627303424174e-07,
+      "loss": 2.374,
+      "step": 5215
+    },
+    {
+      "epoch": 0.6751464543483187,
+      "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": 1.517777919769287,
+      "eval_runtime": 13.9994,
+      "eval_samples_per_second": 3.143,
+      "eval_steps_per_second": 0.214,
+      "step": 5215
+    },
+    {
+      "epoch": 0.6752759167556721,
+      "grad_norm": 1.5671852668968798,
+      "learning_rate": 3.6033034771555923e-07,
+      "loss": 2.3293,
+      "step": 5216
+    },
+    {
+      "epoch": 0.6754053791630256,
+      "grad_norm": 1.359830987904769,
+      "learning_rate": 3.6007447646101786e-07,
+      "loss": 2.405,
+      "step": 5217
+    },
+    {
+      "epoch": 0.675534841570379,
+      "grad_norm": 1.7563274816105932,
+      "learning_rate": 3.598186593228842e-07,
+      "loss": 2.3501,
+      "step": 5218
+    },
+    {
+      "epoch": 0.6756643039777325,
+      "grad_norm": 2.712288861932896,
+      "learning_rate": 3.5956289635341465e-07,
+      "loss": 2.8359,
+      "step": 5219
+    },
+    {
+      "epoch": 0.6757937663850859,
+      "grad_norm": 1.4580046995383933,
+      "learning_rate": 3.5930718760485333e-07,
+      "loss": 2.3892,
+      "step": 5220
+    },
+    {
+      "epoch": 0.6757937663850859,
+      "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": 1.5184658765792847,
+      "eval_runtime": 14.2688,
+      "eval_samples_per_second": 3.084,
+      "eval_steps_per_second": 0.21,
+      "step": 5220
+    },
+    {
+      "epoch": 0.6759232287924394,
+      "grad_norm": 2.210650569872333,
+      "learning_rate": 3.590515331294346e-07,
+      "loss": 2.5549,
+      "step": 5221
+    },
+    {
+      "epoch": 0.6760526911997928,
+      "grad_norm": 1.5490546548895017,
+      "learning_rate": 3.5879593297938055e-07,
+      "loss": 2.332,
+      "step": 5222
+    },
+    {
+      "epoch": 0.6761821536071463,
+      "grad_norm": 2.412738058317075,
+      "learning_rate": 3.5854038720690305e-07,
+      "loss": 2.5547,
+      "step": 5223
+    },
+    {
+      "epoch": 0.6763116160144997,
+      "grad_norm": 2.3987816391579657,
+      "learning_rate": 3.582848958642024e-07,
+      "loss": 2.481,
+      "step": 5224
+    },
+    {
+      "epoch": 0.6764410784218533,
+      "grad_norm": 2.3761614745260227,
+      "learning_rate": 3.5802945900346763e-07,
+      "loss": 2.5806,
+      "step": 5225
+    },
+    {
+      "epoch": 0.6764410784218533,
+      "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": 1.51708984375,
+      "eval_runtime": 15.0319,
+      "eval_samples_per_second": 2.927,
+      "eval_steps_per_second": 0.2,
+      "step": 5225
+    },
+    {
+      "epoch": 0.6765705408292068,
+      "grad_norm": 1.999288599859236,
+      "learning_rate": 3.5777407667687723e-07,
+      "loss": 2.3669,
+      "step": 5226
+    },
+    {
+      "epoch": 0.6767000032365602,
+      "grad_norm": 1.9038521953853287,
+      "learning_rate": 3.575187489365979e-07,
+      "loss": 2.5393,
+      "step": 5227
+    },
+    {
+      "epoch": 0.6768294656439137,
+      "grad_norm": 2.655885107538993,
+      "learning_rate": 3.572634758347861e-07,
+      "loss": 2.3694,
+      "step": 5228
+    },
+    {
+      "epoch": 0.6769589280512671,
+      "grad_norm": 1.2756381051525916,
+      "learning_rate": 3.5700825742358595e-07,
+      "loss": 2.2866,
+      "step": 5229
+    },
+    {
+      "epoch": 0.6770883904586206,
+      "grad_norm": 1.735971958165332,
+      "learning_rate": 3.56753093755131e-07,
+      "loss": 2.4614,
+      "step": 5230
+    },
+    {
+      "epoch": 0.6770883904586206,
+      "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": 1.5171120166778564,
+      "eval_runtime": 12.9492,
+      "eval_samples_per_second": 3.398,
+      "eval_steps_per_second": 0.232,
+      "step": 5230
+    },
+    {
+      "epoch": 0.677217852865974,
+      "grad_norm": 1.7071448357533605,
+      "learning_rate": 3.5649798488154365e-07,
+      "loss": 2.2178,
+      "step": 5231
+    },
+    {
+      "epoch": 0.6773473152733275,
+      "grad_norm": 3.4925803110194904,
+      "learning_rate": 3.5624293085493556e-07,
+      "loss": 2.7458,
+      "step": 5232
+    },
+    {
+      "epoch": 0.6774767776806809,
+      "grad_norm": 2.7030568290693404,
+      "learning_rate": 3.5598793172740597e-07,
+      "loss": 2.6646,
+      "step": 5233
+    },
+    {
+      "epoch": 0.6776062400880344,
+      "grad_norm": 2.259110266225499,
+      "learning_rate": 3.557329875510445e-07,
+      "loss": 2.3389,
+      "step": 5234
+    },
+    {
+      "epoch": 0.677735702495388,
+      "grad_norm": 2.984263218140695,
+      "learning_rate": 3.554780983779276e-07,
+      "loss": 2.645,
+      "step": 5235
+    },
+    {
+      "epoch": 0.677735702495388,
+      "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": 1.5136497020721436,
+      "eval_runtime": 15.1559,
+      "eval_samples_per_second": 2.903,
+      "eval_steps_per_second": 0.198,
+      "step": 5235
+    },
+    {
+      "epoch": 0.6778651649027414,
+      "grad_norm": 1.6363349893759631,
+      "learning_rate": 3.552232642601221e-07,
+      "loss": 2.4404,
+      "step": 5236
+    },
+    {
+      "epoch": 0.6779946273100949,
+      "grad_norm": 1.2916843377566822,
+      "learning_rate": 3.549684852496832e-07,
+      "loss": 2.3035,
+      "step": 5237
+    },
+    {
+      "epoch": 0.6781240897174483,
+      "grad_norm": 1.2762116165880084,
+      "learning_rate": 3.5471376139865427e-07,
+      "loss": 2.2085,
+      "step": 5238
+    },
+    {
+      "epoch": 0.6782535521248018,
+      "grad_norm": 2.239098902198714,
+      "learning_rate": 3.5445909275906826e-07,
+      "loss": 2.4478,
+      "step": 5239
+    },
+    {
+      "epoch": 0.6783830145321552,
+      "grad_norm": 2.362653082394167,
+      "learning_rate": 3.542044793829461e-07,
+      "loss": 2.4512,
+      "step": 5240
+    },
+    {
+      "epoch": 0.6783830145321552,
+      "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": 1.5144709348678589,
+      "eval_runtime": 14.2809,
+      "eval_samples_per_second": 3.081,
+      "eval_steps_per_second": 0.21,
+      "step": 5240
+    },
+    {
+      "epoch": 0.6785124769395087,
+      "grad_norm": 1.6466598228001548,
+      "learning_rate": 3.5394992132229767e-07,
+      "loss": 2.3201,
+      "step": 5241
+    },
+    {
+      "epoch": 0.6786419393468621,
+      "grad_norm": 1.7901975282499403,
+      "learning_rate": 3.536954186291219e-07,
+      "loss": 2.2375,
+      "step": 5242
+    },
+    {
+      "epoch": 0.6787714017542156,
+      "grad_norm": 1.4214197741100385,
+      "learning_rate": 3.534409713554057e-07,
+      "loss": 2.2954,
+      "step": 5243
+    },
+    {
+      "epoch": 0.6789008641615691,
+      "grad_norm": 2.042846127205418,
+      "learning_rate": 3.5318657955312573e-07,
+      "loss": 2.7146,
+      "step": 5244
+    },
+    {
+      "epoch": 0.6790303265689226,
+      "grad_norm": 3.5721648727622117,
+      "learning_rate": 3.5293224327424625e-07,
+      "loss": 2.3306,
+      "step": 5245
+    },
+    {
+      "epoch": 0.6790303265689226,
+      "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": 1.5171786546707153,
+      "eval_runtime": 14.7304,
+      "eval_samples_per_second": 2.987,
+      "eval_steps_per_second": 0.204,
+      "step": 5245
+    },
+    {
+      "epoch": 0.679159788976276,
+      "grad_norm": 3.0066869720647116,
+      "learning_rate": 3.526779625707204e-07,
+      "loss": 2.4009,
+      "step": 5246
+    },
+    {
+      "epoch": 0.6792892513836295,
+      "grad_norm": 2.1355296737675027,
+      "learning_rate": 3.5242373749449056e-07,
+      "loss": 2.2749,
+      "step": 5247
+    },
+    {
+      "epoch": 0.679418713790983,
+      "grad_norm": 2.1017472222072735,
+      "learning_rate": 3.52169568097487e-07,
+      "loss": 2.3696,
+      "step": 5248
+    },
+    {
+      "epoch": 0.6795481761983364,
+      "grad_norm": 3.6488145130240808,
+      "learning_rate": 3.519154544316294e-07,
+      "loss": 2.4302,
+      "step": 5249
+    },
+    {
+      "epoch": 0.6796776386056899,
+      "grad_norm": 1.880456897453718,
+      "learning_rate": 3.5166139654882525e-07,
+      "loss": 2.41,
+      "step": 5250
+    },
+    {
+      "epoch": 0.6796776386056899,
+      "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": 1.5179332494735718,
+      "eval_runtime": 15.2384,
+      "eval_samples_per_second": 2.887,
+      "eval_steps_per_second": 0.197,
+      "step": 5250
+    },
+    {
+      "epoch": 0.6798071010130433,
+      "grad_norm": 3.1613642716518733,
+      "learning_rate": 3.514073945009714e-07,
+      "loss": 2.4265,
+      "step": 5251
+    },
+    {
+      "epoch": 0.6799365634203968,
+      "grad_norm": 1.8057933957080892,
+      "learning_rate": 3.5115344833995273e-07,
+      "loss": 2.1995,
+      "step": 5252
+    },
+    {
+      "epoch": 0.6800660258277502,
+      "grad_norm": 1.4441837457786255,
+      "learning_rate": 3.508995581176425e-07,
+      "loss": 2.3928,
+      "step": 5253
+    },
+    {
+      "epoch": 0.6801954882351038,
+      "grad_norm": 3.874670791895159,
+      "learning_rate": 3.506457238859033e-07,
+      "loss": 2.4443,
+      "step": 5254
+    },
+    {
+      "epoch": 0.6803249506424572,
+      "grad_norm": 1.7297980002020237,
+      "learning_rate": 3.5039194569658625e-07,
+      "loss": 2.3359,
+      "step": 5255
+    },
+    {
+      "epoch": 0.6803249506424572,
+      "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": 1.5163795948028564,
+      "eval_runtime": 14.1876,
+      "eval_samples_per_second": 3.101,
+      "eval_steps_per_second": 0.211,
+      "step": 5255
+    },
+    {
+      "epoch": 0.6804544130498107,
+      "grad_norm": 1.3142094783061335,
+      "learning_rate": 3.5013822360153024e-07,
+      "loss": 2.3228,
+      "step": 5256
+    },
+    {
+      "epoch": 0.6805838754571641,
+      "grad_norm": 1.6543195155186603,
+      "learning_rate": 3.4988455765256326e-07,
+      "loss": 2.4092,
+      "step": 5257
+    },
+    {
+      "epoch": 0.6807133378645176,
+      "grad_norm": 1.8122925535389094,
+      "learning_rate": 3.4963094790150154e-07,
+      "loss": 2.3245,
+      "step": 5258
+    },
+    {
+      "epoch": 0.680842800271871,
+      "grad_norm": 1.600405426267812,
+      "learning_rate": 3.493773944001501e-07,
+      "loss": 2.2654,
+      "step": 5259
+    },
+    {
+      "epoch": 0.6809722626792245,
+      "grad_norm": 2.1278032642846902,
+      "learning_rate": 3.4912389720030275e-07,
+      "loss": 2.354,
+      "step": 5260
+    },
+    {
+      "epoch": 0.6809722626792245,
+      "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": 1.5129616260528564,
+      "eval_runtime": 14.4426,
+      "eval_samples_per_second": 3.047,
+      "eval_steps_per_second": 0.208,
+      "step": 5260
+    },
+    {
+      "epoch": 0.681101725086578,
+      "grad_norm": 1.9564449809840208,
+      "learning_rate": 3.4887045635374096e-07,
+      "loss": 2.5923,
+      "step": 5261
+    },
+    {
+      "epoch": 0.6812311874939314,
+      "grad_norm": 2.8503831455817883,
+      "learning_rate": 3.486170719122356e-07,
+      "loss": 2.4429,
+      "step": 5262
+    },
+    {
+      "epoch": 0.6813606499012849,
+      "grad_norm": 2.6314135290550316,
+      "learning_rate": 3.483637439275453e-07,
+      "loss": 2.4775,
+      "step": 5263
+    },
+    {
+      "epoch": 0.6814901123086384,
+      "grad_norm": 4.32835843404821,
+      "learning_rate": 3.4811047245141713e-07,
+      "loss": 2.4233,
+      "step": 5264
+    },
+    {
+      "epoch": 0.6816195747159919,
+      "grad_norm": 1.2409452199406177,
+      "learning_rate": 3.478572575355876e-07,
+      "loss": 2.2716,
+      "step": 5265
+    },
+    {
+      "epoch": 0.6816195747159919,
+      "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": 1.515669345855713,
+      "eval_runtime": 14.0382,
+      "eval_samples_per_second": 3.134,
+      "eval_steps_per_second": 0.214,
+      "step": 5265
+    },
+    {
+      "epoch": 0.6817490371233453,
+      "grad_norm": 3.1463823453908804,
+      "learning_rate": 3.4760409923178036e-07,
+      "loss": 2.7,
+      "step": 5266
+    },
+    {
+      "epoch": 0.6818784995306988,
+      "grad_norm": 1.4538289416347718,
+      "learning_rate": 3.4735099759170864e-07,
+      "loss": 2.4497,
+      "step": 5267
+    },
+    {
+      "epoch": 0.6820079619380522,
+      "grad_norm": 1.3960796627036527,
+      "learning_rate": 3.470979526670734e-07,
+      "loss": 2.2971,
+      "step": 5268
+    },
+    {
+      "epoch": 0.6821374243454057,
+      "grad_norm": 1.5680110984654894,
+      "learning_rate": 3.4684496450956386e-07,
+      "loss": 2.4592,
+      "step": 5269
+    },
+    {
+      "epoch": 0.6822668867527591,
+      "grad_norm": 1.775844244948726,
+      "learning_rate": 3.465920331708586e-07,
+      "loss": 2.4331,
+      "step": 5270
+    },
+    {
+      "epoch": 0.6822668867527591,
+      "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": 1.517045497894287,
+      "eval_runtime": 13.8252,
+      "eval_samples_per_second": 3.183,
+      "eval_steps_per_second": 0.217,
+      "step": 5270
+    },
+    {
+      "epoch": 0.6823963491601126,
+      "grad_norm": 3.0064942558009884,
+      "learning_rate": 3.4633915870262323e-07,
+      "loss": 2.4385,
+      "step": 5271
+    },
+    {
+      "epoch": 0.6825258115674661,
+      "grad_norm": 1.756906264922669,
+      "learning_rate": 3.4608634115651333e-07,
+      "loss": 2.3374,
+      "step": 5272
+    },
+    {
+      "epoch": 0.6826552739748195,
+      "grad_norm": 2.038579186850268,
+      "learning_rate": 3.4583358058417145e-07,
+      "loss": 2.3347,
+      "step": 5273
+    },
+    {
+      "epoch": 0.6827847363821731,
+      "grad_norm": 1.141392873884542,
+      "learning_rate": 3.455808770372289e-07,
+      "loss": 2.1926,
+      "step": 5274
+    },
+    {
+      "epoch": 0.6829141987895265,
+      "grad_norm": 1.6832690108620663,
+      "learning_rate": 3.453282305673061e-07,
+      "loss": 2.0496,
+      "step": 5275
+    },
+    {
+      "epoch": 0.6829141987895265,
+      "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": 1.519331455230713,
+      "eval_runtime": 13.3241,
+      "eval_samples_per_second": 3.302,
+      "eval_steps_per_second": 0.225,
+      "step": 5275
+    },
+    {
+      "epoch": 0.68304366119688,
+      "grad_norm": 1.3132853117623964,
+      "learning_rate": 3.450756412260107e-07,
+      "loss": 2.2083,
+      "step": 5276
+    },
+    {
+      "epoch": 0.6831731236042334,
+      "grad_norm": 1.8641173578625694,
+      "learning_rate": 3.4482310906493925e-07,
+      "loss": 2.4832,
+      "step": 5277
+    },
+    {
+      "epoch": 0.6833025860115869,
+      "grad_norm": 1.5046968889959005,
+      "learning_rate": 3.445706341356771e-07,
+      "loss": 2.4243,
+      "step": 5278
+    },
+    {
+      "epoch": 0.6834320484189403,
+      "grad_norm": 1.3426248272115668,
+      "learning_rate": 3.4431821648979684e-07,
+      "loss": 2.3423,
+      "step": 5279
+    },
+    {
+      "epoch": 0.6835615108262938,
+      "grad_norm": 1.7858745758616192,
+      "learning_rate": 3.4406585617886015e-07,
+      "loss": 2.4058,
+      "step": 5280
+    },
+    {
+      "epoch": 0.6835615108262938,
+      "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": 1.5186434984207153,
+      "eval_runtime": 14.1945,
+      "eval_samples_per_second": 3.1,
+      "eval_steps_per_second": 0.211,
+      "step": 5280
+    },
+    {
+      "epoch": 0.6836909732336472,
+      "grad_norm": 2.0156811300806248,
+      "learning_rate": 3.438135532544162e-07,
+      "loss": 2.4004,
+      "step": 5281
+    },
+    {
+      "epoch": 0.6838204356410007,
+      "grad_norm": 1.2292898591618622,
+      "learning_rate": 3.4356130776800347e-07,
+      "loss": 2.1897,
+      "step": 5282
+    },
+    {
+      "epoch": 0.6839498980483542,
+      "grad_norm": 1.4536210374840777,
+      "learning_rate": 3.4330911977114843e-07,
+      "loss": 2.3479,
+      "step": 5283
+    },
+    {
+      "epoch": 0.6840793604557077,
+      "grad_norm": 1.5917119251743426,
+      "learning_rate": 3.430569893153653e-07,
+      "loss": 2.321,
+      "step": 5284
+    },
+    {
+      "epoch": 0.6842088228630612,
+      "grad_norm": 1.194981712229032,
+      "learning_rate": 3.428049164521564e-07,
+      "loss": 2.2852,
+      "step": 5285
+    },
+    {
+      "epoch": 0.6842088228630612,
+      "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": 1.517134189605713,
+      "eval_runtime": 14.8292,
+      "eval_samples_per_second": 2.967,
+      "eval_steps_per_second": 0.202,
+      "step": 5285
+    },
+    {
+      "epoch": 0.6843382852704146,
+      "grad_norm": 1.4564356124850828,
+      "learning_rate": 3.4255290123301357e-07,
+      "loss": 2.3604,
+      "step": 5286
+    },
+    {
+      "epoch": 0.6844677476777681,
+      "grad_norm": 1.5239257772120258,
+      "learning_rate": 3.423009437094153e-07,
+      "loss": 2.2656,
+      "step": 5287
+    },
+    {
+      "epoch": 0.6845972100851215,
+      "grad_norm": 2.1551809473036747,
+      "learning_rate": 3.420490439328296e-07,
+      "loss": 2.6562,
+      "step": 5288
+    },
+    {
+      "epoch": 0.684726672492475,
+      "grad_norm": 1.4104229389410412,
+      "learning_rate": 3.417972019547116e-07,
+      "loss": 2.3862,
+      "step": 5289
+    },
+    {
+      "epoch": 0.6848561348998284,
+      "grad_norm": 1.780957551376093,
+      "learning_rate": 3.4154541782650556e-07,
+      "loss": 2.3049,
+      "step": 5290
+    },
+    {
+      "epoch": 0.6848561348998284,
+      "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": 1.5157803297042847,
+      "eval_runtime": 13.7832,
+      "eval_samples_per_second": 3.192,
+      "eval_steps_per_second": 0.218,
+      "step": 5290
+    },
+    {
+      "epoch": 0.6849855973071819,
+      "grad_norm": 1.5735364381034245,
+      "learning_rate": 3.4129369159964327e-07,
+      "loss": 2.3372,
+      "step": 5291
+    },
+    {
+      "epoch": 0.6851150597145353,
+      "grad_norm": 1.717616490730774,
+      "learning_rate": 3.4104202332554475e-07,
+      "loss": 2.4036,
+      "step": 5292
+    },
+    {
+      "epoch": 0.6852445221218889,
+      "grad_norm": 1.5051676450433018,
+      "learning_rate": 3.4079041305561876e-07,
+      "loss": 2.3101,
+      "step": 5293
+    },
+    {
+      "epoch": 0.6853739845292424,
+      "grad_norm": 1.3031780948838116,
+      "learning_rate": 3.405388608412613e-07,
+      "loss": 2.3641,
+      "step": 5294
+    },
+    {
+      "epoch": 0.6855034469365958,
+      "grad_norm": 1.5857000372699583,
+      "learning_rate": 3.4028736673385755e-07,
+      "loss": 2.3579,
+      "step": 5295
+    },
+    {
+      "epoch": 0.6855034469365958,
+      "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": 1.5150257349014282,
+      "eval_runtime": 15.0757,
+      "eval_samples_per_second": 2.919,
+      "eval_steps_per_second": 0.199,
+      "step": 5295
+    },
+    {
+      "epoch": 0.6856329093439493,
+      "grad_norm": 2.1342613905983585,
+      "learning_rate": 3.4003593078478e-07,
+      "loss": 2.449,
+      "step": 5296
+    },
+    {
+      "epoch": 0.6857623717513027,
+      "grad_norm": 1.4041942243521495,
+      "learning_rate": 3.397845530453893e-07,
+      "loss": 2.3462,
+      "step": 5297
+    },
+    {
+      "epoch": 0.6858918341586562,
+      "grad_norm": 1.3320730837144454,
+      "learning_rate": 3.39533233567035e-07,
+      "loss": 2.2778,
+      "step": 5298
+    },
+    {
+      "epoch": 0.6860212965660096,
+      "grad_norm": 1.7325679053807328,
+      "learning_rate": 3.392819724010535e-07,
+      "loss": 2.4946,
+      "step": 5299
+    },
+    {
+      "epoch": 0.6861507589733631,
+      "grad_norm": 1.2735286825120988,
+      "learning_rate": 3.3903076959877047e-07,
+      "loss": 2.1953,
+      "step": 5300
+    },
+    {
+      "epoch": 0.6861507589733631,
+      "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": 1.5143154859542847,
+      "eval_runtime": 14.5467,
+      "eval_samples_per_second": 3.025,
+      "eval_steps_per_second": 0.206,
+      "step": 5300
+    },
+    {
+      "epoch": 0.6862802213807165,
+      "grad_norm": 2.384971743131937,
+      "learning_rate": 3.3877962521149956e-07,
+      "loss": 2.3931,
+      "step": 5301
+    },
+    {
+      "epoch": 0.68640968378807,
+      "grad_norm": 2.147038904908983,
+      "learning_rate": 3.385285392905412e-07,
+      "loss": 2.4946,
+      "step": 5302
+    },
+    {
+      "epoch": 0.6865391461954236,
+      "grad_norm": 1.986882172640929,
+      "learning_rate": 3.3827751188718533e-07,
+      "loss": 2.2798,
+      "step": 5303
+    },
+    {
+      "epoch": 0.686668608602777,
+      "grad_norm": 1.5409879420459855,
+      "learning_rate": 3.380265430527091e-07,
+      "loss": 2.5415,
+      "step": 5304
+    },
+    {
+      "epoch": 0.6867980710101305,
+      "grad_norm": 1.8687361414984769,
+      "learning_rate": 3.3777563283837805e-07,
+      "loss": 2.2322,
+      "step": 5305
+    },
+    {
+      "epoch": 0.6867980710101305,
+      "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": 1.5142711400985718,
+      "eval_runtime": 18.8418,
+      "eval_samples_per_second": 2.335,
+      "eval_steps_per_second": 0.159,
+      "step": 5305
+    },
+    {
+      "epoch": 0.6869275334174839,
+      "grad_norm": 2.361479109078417,
+      "learning_rate": 3.375247812954463e-07,
+      "loss": 2.3794,
+      "step": 5306
+    },
+    {
+      "epoch": 0.6870569958248374,
+      "grad_norm": 1.3951456832389841,
+      "learning_rate": 3.372739884751543e-07,
+      "loss": 2.3733,
+      "step": 5307
+    },
+    {
+      "epoch": 0.6871864582321908,
+      "grad_norm": 2.028005489437002,
+      "learning_rate": 3.3702325442873217e-07,
+      "loss": 2.168,
+      "step": 5308
+    },
+    {
+      "epoch": 0.6873159206395443,
+      "grad_norm": 2.234759749833977,
+      "learning_rate": 3.367725792073976e-07,
+      "loss": 2.3535,
+      "step": 5309
+    },
+    {
+      "epoch": 0.6874453830468977,
+      "grad_norm": 1.8700754833814541,
+      "learning_rate": 3.3652196286235553e-07,
+      "loss": 2.3005,
+      "step": 5310
+    },
+    {
+      "epoch": 0.6874453830468977,
+      "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": 1.5149592161178589,
+      "eval_runtime": 14.6834,
+      "eval_samples_per_second": 2.997,
+      "eval_steps_per_second": 0.204,
+      "step": 5310
+    },
+    {
+      "epoch": 0.6875748454542512,
+      "grad_norm": 1.8876010473368234,
+      "learning_rate": 3.3627140544479997e-07,
+      "loss": 2.5183,
+      "step": 5311
+    },
+    {
+      "epoch": 0.6877043078616046,
+      "grad_norm": 1.6051176000616365,
+      "learning_rate": 3.3602090700591215e-07,
+      "loss": 2.2588,
+      "step": 5312
+    },
+    {
+      "epoch": 0.6878337702689582,
+      "grad_norm": 1.4515583689326508,
+      "learning_rate": 3.357704675968611e-07,
+      "loss": 2.2747,
+      "step": 5313
+    },
+    {
+      "epoch": 0.6879632326763117,
+      "grad_norm": 1.5844909081318954,
+      "learning_rate": 3.355200872688048e-07,
+      "loss": 2.4324,
+      "step": 5314
+    },
+    {
+      "epoch": 0.6880926950836651,
+      "grad_norm": 3.217833584365405,
+      "learning_rate": 3.3526976607288793e-07,
+      "loss": 2.5137,
+      "step": 5315
+    },
+    {
+      "epoch": 0.6880926950836651,
+      "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": 1.5150035619735718,
+      "eval_runtime": 15.5964,
+      "eval_samples_per_second": 2.821,
+      "eval_steps_per_second": 0.192,
+      "step": 5315
+    },
+    {
+      "epoch": 0.6882221574910186,
+      "grad_norm": 1.5114742460583046,
+      "learning_rate": 3.35019504060244e-07,
+      "loss": 2.4365,
+      "step": 5316
+    },
+    {
+      "epoch": 0.688351619898372,
+      "grad_norm": 3.2874319073919347,
+      "learning_rate": 3.3476930128199387e-07,
+      "loss": 2.3691,
+      "step": 5317
+    },
+    {
+      "epoch": 0.6884810823057255,
+      "grad_norm": 2.256535391542108,
+      "learning_rate": 3.345191577892469e-07,
+      "loss": 2.4149,
+      "step": 5318
+    },
+    {
+      "epoch": 0.6886105447130789,
+      "grad_norm": 3.429347989522284,
+      "learning_rate": 3.3426907363309976e-07,
+      "loss": 2.3914,
+      "step": 5319
+    },
+    {
+      "epoch": 0.6887400071204324,
+      "grad_norm": 1.0874046382361477,
+      "learning_rate": 3.3401904886463684e-07,
+      "loss": 2.1731,
+      "step": 5320
+    },
+    {
+      "epoch": 0.6887400071204324,
+      "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": 1.5172008275985718,
+      "eval_runtime": 19.1102,
+      "eval_samples_per_second": 2.302,
+      "eval_steps_per_second": 0.157,
+      "step": 5320
+    },
+    {
+      "epoch": 0.6888694695277858,
+      "grad_norm": 1.3805835847510348,
+      "learning_rate": 3.337690835349314e-07,
+      "loss": 2.3584,
+      "step": 5321
+    },
+    {
+      "epoch": 0.6889989319351393,
+      "grad_norm": 1.5954479129584558,
+      "learning_rate": 3.335191776950434e-07,
+      "loss": 2.3792,
+      "step": 5322
+    },
+    {
+      "epoch": 0.6891283943424928,
+      "grad_norm": 2.8428686669989243,
+      "learning_rate": 3.332693313960216e-07,
+      "loss": 2.4805,
+      "step": 5323
+    },
+    {
+      "epoch": 0.6892578567498463,
+      "grad_norm": 3.0839326029096332,
+      "learning_rate": 3.33019544688902e-07,
+      "loss": 2.3149,
+      "step": 5324
+    },
+    {
+      "epoch": 0.6893873191571998,
+      "grad_norm": 1.7666352164113863,
+      "learning_rate": 3.3276981762470835e-07,
+      "loss": 2.5637,
+      "step": 5325
+    },
+    {
+      "epoch": 0.6893873191571998,
+      "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": 1.514359951019287,
+      "eval_runtime": 15.8649,
+      "eval_samples_per_second": 2.773,
+      "eval_steps_per_second": 0.189,
+      "step": 5325
+    },
+    {
+      "epoch": 0.6895167815645532,
+      "grad_norm": 1.6564306111926075,
+      "learning_rate": 3.325201502544527e-07,
+      "loss": 2.3599,
+      "step": 5326
+    },
+    {
+      "epoch": 0.6896462439719067,
+      "grad_norm": 1.5945678877481886,
+      "learning_rate": 3.3227054262913484e-07,
+      "loss": 2.2998,
+      "step": 5327
+    },
+    {
+      "epoch": 0.6897757063792601,
+      "grad_norm": 1.3550227880786858,
+      "learning_rate": 3.320209947997418e-07,
+      "loss": 2.3027,
+      "step": 5328
+    },
+    {
+      "epoch": 0.6899051687866136,
+      "grad_norm": 1.3086287694967784,
+      "learning_rate": 3.3177150681724944e-07,
+      "loss": 2.499,
+      "step": 5329
+    },
+    {
+      "epoch": 0.690034631193967,
+      "grad_norm": 1.650943473562167,
+      "learning_rate": 3.3152207873261983e-07,
+      "loss": 2.1572,
+      "step": 5330
+    },
+    {
+      "epoch": 0.690034631193967,
+      "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": 1.5136497020721436,
+      "eval_runtime": 13.9794,
+      "eval_samples_per_second": 3.147,
+      "eval_steps_per_second": 0.215,
+      "step": 5330
+    },
+    {
+      "epoch": 0.6901640936013205,
+      "grad_norm": 1.2915484217800368,
+      "learning_rate": 3.312727105968042e-07,
+      "loss": 2.3193,
+      "step": 5331
+    },
+    {
+      "epoch": 0.690293556008674,
+      "grad_norm": 3.7580423761323236,
+      "learning_rate": 3.3102340246074115e-07,
+      "loss": 3.0388,
+      "step": 5332
+    },
+    {
+      "epoch": 0.6904230184160275,
+      "grad_norm": 2.8934251653665,
+      "learning_rate": 3.3077415437535664e-07,
+      "loss": 2.313,
+      "step": 5333
+    },
+    {
+      "epoch": 0.6905524808233809,
+      "grad_norm": 1.7966369994170994,
+      "learning_rate": 3.3052496639156495e-07,
+      "loss": 2.3757,
+      "step": 5334
+    },
+    {
+      "epoch": 0.6906819432307344,
+      "grad_norm": 2.068137813876219,
+      "learning_rate": 3.3027583856026767e-07,
+      "loss": 2.4111,
+      "step": 5335
+    },
+    {
+      "epoch": 0.6906819432307344,
+      "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": 1.5142933130264282,
+      "eval_runtime": 13.6879,
+      "eval_samples_per_second": 3.215,
+      "eval_steps_per_second": 0.219,
+      "step": 5335
+    },
+    {
+      "epoch": 0.6908114056380878,
+      "grad_norm": 2.2781863072102597,
+      "learning_rate": 3.3002677093235387e-07,
+      "loss": 2.6138,
+      "step": 5336
+    },
+    {
+      "epoch": 0.6909408680454413,
+      "grad_norm": 2.586761375232325,
+      "learning_rate": 3.297777635587012e-07,
+      "loss": 2.5593,
+      "step": 5337
+    },
+    {
+      "epoch": 0.6910703304527948,
+      "grad_norm": 1.746465970557761,
+      "learning_rate": 3.29528816490174e-07,
+      "loss": 2.5435,
+      "step": 5338
+    },
+    {
+      "epoch": 0.6911997928601482,
+      "grad_norm": 1.2455390194498344,
+      "learning_rate": 3.2927992977762516e-07,
+      "loss": 2.168,
+      "step": 5339
+    },
+    {
+      "epoch": 0.6913292552675017,
+      "grad_norm": 2.9018345175822784,
+      "learning_rate": 3.290311034718946e-07,
+      "loss": 2.3441,
+      "step": 5340
+    },
+    {
+      "epoch": 0.6913292552675017,
+      "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": 1.5108309984207153,
+      "eval_runtime": 15.1055,
+      "eval_samples_per_second": 2.913,
+      "eval_steps_per_second": 0.199,
+      "step": 5340
+    },
+    {
+      "epoch": 0.6914587176748551,
+      "grad_norm": 1.3597390088524353,
+      "learning_rate": 3.2878233762381003e-07,
+      "loss": 2.4189,
+      "step": 5341
+    },
+    {
+      "epoch": 0.6915881800822087,
+      "grad_norm": 1.8226028613013434,
+      "learning_rate": 3.285336322841872e-07,
+      "loss": 2.3486,
+      "step": 5342
+    },
+    {
+      "epoch": 0.6917176424895621,
+      "grad_norm": 1.808497782035811,
+      "learning_rate": 3.2828498750382887e-07,
+      "loss": 2.4976,
+      "step": 5343
+    },
+    {
+      "epoch": 0.6918471048969156,
+      "grad_norm": 3.7786067130303653,
+      "learning_rate": 3.280364033335262e-07,
+      "loss": 2.5557,
+      "step": 5344
+    },
+    {
+      "epoch": 0.691976567304269,
+      "grad_norm": 1.7876273026010536,
+      "learning_rate": 3.2778787982405737e-07,
+      "loss": 2.5051,
+      "step": 5345
+    },
+    {
+      "epoch": 0.691976567304269,
+      "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": 1.513627529144287,
+      "eval_runtime": 14.1363,
+      "eval_samples_per_second": 3.113,
+      "eval_steps_per_second": 0.212,
+      "step": 5345
+    },
+    {
+      "epoch": 0.6921060297116225,
+      "grad_norm": 1.6705348993732794,
+      "learning_rate": 3.2753941702618807e-07,
+      "loss": 2.2898,
+      "step": 5346
+    },
+    {
+      "epoch": 0.692235492118976,
+      "grad_norm": 1.9443960265234832,
+      "learning_rate": 3.272910149906723e-07,
+      "loss": 2.3772,
+      "step": 5347
+    },
+    {
+      "epoch": 0.6923649545263294,
+      "grad_norm": 1.9638856483260487,
+      "learning_rate": 3.2704267376825073e-07,
+      "loss": 2.3394,
+      "step": 5348
+    },
+    {
+      "epoch": 0.6924944169336829,
+      "grad_norm": 2.13323576349333,
+      "learning_rate": 3.267943934096525e-07,
+      "loss": 2.623,
+      "step": 5349
+    },
+    {
+      "epoch": 0.6926238793410363,
+      "grad_norm": 1.8045705918564887,
+      "learning_rate": 3.265461739655938e-07,
+      "loss": 2.4504,
+      "step": 5350
+    },
+    {
+      "epoch": 0.6926238793410363,
+      "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": 1.5150257349014282,
+      "eval_runtime": 13.9846,
+      "eval_samples_per_second": 3.146,
+      "eval_steps_per_second": 0.215,
+      "step": 5350
+    },
+    {
+      "epoch": 0.6927533417483898,
+      "grad_norm": 1.423514535967298,
+      "learning_rate": 3.262980154867786e-07,
+      "loss": 2.3643,
+      "step": 5351
+    },
+    {
+      "epoch": 0.6928828041557433,
+      "grad_norm": 2.8997005798673525,
+      "learning_rate": 3.260499180238981e-07,
+      "loss": 2.5488,
+      "step": 5352
+    },
+    {
+      "epoch": 0.6930122665630968,
+      "grad_norm": 1.958156059236238,
+      "learning_rate": 3.258018816276311e-07,
+      "loss": 2.3618,
+      "step": 5353
+    },
+    {
+      "epoch": 0.6931417289704502,
+      "grad_norm": 1.2203410520597664,
+      "learning_rate": 3.255539063486442e-07,
+      "loss": 2.1858,
+      "step": 5354
+    },
+    {
+      "epoch": 0.6932711913778037,
+      "grad_norm": 1.285710799241308,
+      "learning_rate": 3.2530599223759175e-07,
+      "loss": 2.4822,
+      "step": 5355
+    },
+    {
+      "epoch": 0.6932711913778037,
+      "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": 1.513427734375,
+      "eval_runtime": 14.9112,
+      "eval_samples_per_second": 2.951,
+      "eval_steps_per_second": 0.201,
+      "step": 5355
+    },
+    {
+      "epoch": 0.6934006537851571,
+      "grad_norm": 1.1575368038575253,
+      "learning_rate": 3.250581393451148e-07,
+      "loss": 2.321,
+      "step": 5356
+    },
+    {
+      "epoch": 0.6935301161925106,
+      "grad_norm": 2.2841893560977793,
+      "learning_rate": 3.2481034772184216e-07,
+      "loss": 2.3428,
+      "step": 5357
+    },
+    {
+      "epoch": 0.693659578599864,
+      "grad_norm": 1.434021653021914,
+      "learning_rate": 3.245626174183909e-07,
+      "loss": 2.6001,
+      "step": 5358
+    },
+    {
+      "epoch": 0.6937890410072175,
+      "grad_norm": 4.838179011618462,
+      "learning_rate": 3.243149484853643e-07,
+      "loss": 2.6089,
+      "step": 5359
+    },
+    {
+      "epoch": 0.693918503414571,
+      "grad_norm": 1.7801335770591433,
+      "learning_rate": 3.240673409733544e-07,
+      "loss": 2.5592,
+      "step": 5360
+    },
+    {
+      "epoch": 0.693918503414571,
+      "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": 1.5164240598678589,
+      "eval_runtime": 14.9471,
+      "eval_samples_per_second": 2.944,
+      "eval_steps_per_second": 0.201,
+      "step": 5360
+    },
+    {
+      "epoch": 0.6940479658219244,
+      "grad_norm": 3.8792111881073925,
+      "learning_rate": 3.238197949329395e-07,
+      "loss": 2.9326,
+      "step": 5361
+    },
+    {
+      "epoch": 0.694177428229278,
+      "grad_norm": 1.3674043055950762,
+      "learning_rate": 3.235723104146863e-07,
+      "loss": 2.3374,
+      "step": 5362
+    },
+    {
+      "epoch": 0.6943068906366314,
+      "grad_norm": 2.169764738568285,
+      "learning_rate": 3.233248874691484e-07,
+      "loss": 2.5823,
+      "step": 5363
+    },
+    {
+      "epoch": 0.6944363530439849,
+      "grad_norm": 2.28597995187258,
+      "learning_rate": 3.2307752614686665e-07,
+      "loss": 2.7053,
+      "step": 5364
+    },
+    {
+      "epoch": 0.6945658154513383,
+      "grad_norm": 1.5132922490027412,
+      "learning_rate": 3.2283022649837016e-07,
+      "loss": 2.624,
+      "step": 5365
+    },
+    {
+      "epoch": 0.6945658154513383,
+      "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": 1.513671875,
+      "eval_runtime": 14.709,
+      "eval_samples_per_second": 2.991,
+      "eval_steps_per_second": 0.204,
+      "step": 5365
+    },
+    {
+      "epoch": 0.6946952778586918,
+      "grad_norm": 2.139188625695551,
+      "learning_rate": 3.225829885741743e-07,
+      "loss": 2.168,
+      "step": 5366
+    },
+    {
+      "epoch": 0.6948247402660452,
+      "grad_norm": 1.881337685064218,
+      "learning_rate": 3.22335812424783e-07,
+      "loss": 2.2883,
+      "step": 5367
+    },
+    {
+      "epoch": 0.6949542026733987,
+      "grad_norm": 2.8039741866215344,
+      "learning_rate": 3.2208869810068674e-07,
+      "loss": 2.3616,
+      "step": 5368
+    },
+    {
+      "epoch": 0.6950836650807521,
+      "grad_norm": 1.6342965507594556,
+      "learning_rate": 3.218416456523634e-07,
+      "loss": 2.3323,
+      "step": 5369
+    },
+    {
+      "epoch": 0.6952131274881056,
+      "grad_norm": 4.108295650294959,
+      "learning_rate": 3.215946551302789e-07,
+      "loss": 2.6997,
+      "step": 5370
+    },
+    {
+      "epoch": 0.6952131274881056,
+      "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": 1.5164240598678589,
+      "eval_runtime": 15.5815,
+      "eval_samples_per_second": 2.824,
+      "eval_steps_per_second": 0.193,
+      "step": 5370
+    },
+    {
+      "epoch": 0.695342589895459,
+      "grad_norm": 1.4029174074335564,
+      "learning_rate": 3.2134772658488563e-07,
+      "loss": 2.417,
+      "step": 5371
+    },
+    {
+      "epoch": 0.6954720523028126,
+      "grad_norm": 0.9979938909333347,
+      "learning_rate": 3.2110086006662394e-07,
+      "loss": 2.1899,
+      "step": 5372
+    },
+    {
+      "epoch": 0.6956015147101661,
+      "grad_norm": 2.3507384812407146,
+      "learning_rate": 3.2085405562592184e-07,
+      "loss": 2.502,
+      "step": 5373
+    },
+    {
+      "epoch": 0.6957309771175195,
+      "grad_norm": 1.9636587874866376,
+      "learning_rate": 3.206073133131931e-07,
+      "loss": 2.4915,
+      "step": 5374
+    },
+    {
+      "epoch": 0.695860439524873,
+      "grad_norm": 1.7436191852879661,
+      "learning_rate": 3.203606331788406e-07,
+      "loss": 2.4773,
+      "step": 5375
+    },
+    {
+      "epoch": 0.695860439524873,
+      "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": 1.5128506422042847,
+      "eval_runtime": 14.8286,
+      "eval_samples_per_second": 2.967,
+      "eval_steps_per_second": 0.202,
+      "step": 5375
+    },
+    {
+      "epoch": 0.6959899019322264,
+      "grad_norm": 1.375526061002747,
+      "learning_rate": 3.2011401527325334e-07,
+      "loss": 2.3887,
+      "step": 5376
+    },
+    {
+      "epoch": 0.6961193643395799,
+      "grad_norm": 1.6880793506449154,
+      "learning_rate": 3.198674596468082e-07,
+      "loss": 2.135,
+      "step": 5377
+    },
+    {
+      "epoch": 0.6962488267469333,
+      "grad_norm": 1.9319906198231458,
+      "learning_rate": 3.196209663498694e-07,
+      "loss": 2.2468,
+      "step": 5378
+    },
+    {
+      "epoch": 0.6963782891542868,
+      "grad_norm": 1.5176458868229556,
+      "learning_rate": 3.1937453543278806e-07,
+      "loss": 2.2559,
+      "step": 5379
+    },
+    {
+      "epoch": 0.6965077515616402,
+      "grad_norm": 1.7327829117470102,
+      "learning_rate": 3.1912816694590233e-07,
+      "loss": 2.3621,
+      "step": 5380
+    },
+    {
+      "epoch": 0.6965077515616402,
+      "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": 1.5142933130264282,
+      "eval_runtime": 14.4793,
+      "eval_samples_per_second": 3.039,
+      "eval_steps_per_second": 0.207,
+      "step": 5380
+    },
+    {
+      "epoch": 0.6966372139689938,
+      "grad_norm": 1.7072638716547388,
+      "learning_rate": 3.1888186093953845e-07,
+      "loss": 2.3591,
+      "step": 5381
+    },
+    {
+      "epoch": 0.6967666763763473,
+      "grad_norm": 1.648579315286244,
+      "learning_rate": 3.186356174640089e-07,
+      "loss": 2.4902,
+      "step": 5382
+    },
+    {
+      "epoch": 0.6968961387837007,
+      "grad_norm": 1.3898330617933954,
+      "learning_rate": 3.1838943656961457e-07,
+      "loss": 2.438,
+      "step": 5383
+    },
+    {
+      "epoch": 0.6970256011910542,
+      "grad_norm": 1.8591645372883243,
+      "learning_rate": 3.181433183066425e-07,
+      "loss": 2.3237,
+      "step": 5384
+    },
+    {
+      "epoch": 0.6971550635984076,
+      "grad_norm": 3.156839388632553,
+      "learning_rate": 3.1789726272536707e-07,
+      "loss": 2.2163,
+      "step": 5385
+    },
+    {
+      "epoch": 0.6971550635984076,
+      "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": 1.5115412473678589,
+      "eval_runtime": 14.545,
+      "eval_samples_per_second": 3.025,
+      "eval_steps_per_second": 0.206,
+      "step": 5385
+    },
+    {
+      "epoch": 0.6972845260057611,
+      "grad_norm": 2.8515370536266373,
+      "learning_rate": 3.1765126987605075e-07,
+      "loss": 2.3923,
+      "step": 5386
+    },
+    {
+      "epoch": 0.6974139884131145,
+      "grad_norm": 3.513373915602973,
+      "learning_rate": 3.174053398089419e-07,
+      "loss": 2.2219,
+      "step": 5387
+    },
+    {
+      "epoch": 0.697543450820468,
+      "grad_norm": 1.5582575016486924,
+      "learning_rate": 3.171594725742773e-07,
+      "loss": 2.3403,
+      "step": 5388
+    },
+    {
+      "epoch": 0.6976729132278214,
+      "grad_norm": 1.4521040399308904,
+      "learning_rate": 3.1691366822227994e-07,
+      "loss": 2.3269,
+      "step": 5389
+    },
+    {
+      "epoch": 0.6978023756351749,
+      "grad_norm": 1.37440408682988,
+      "learning_rate": 3.1666792680316054e-07,
+      "loss": 2.4565,
+      "step": 5390
+    },
+    {
+      "epoch": 0.6978023756351749,
+      "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": 1.5149813890457153,
+      "eval_runtime": 15.1876,
+      "eval_samples_per_second": 2.897,
+      "eval_steps_per_second": 0.198,
+      "step": 5390
+    },
+    {
+      "epoch": 0.6979318380425285,
+      "grad_norm": 1.286224373879372,
+      "learning_rate": 3.1642224836711686e-07,
+      "loss": 2.3972,
+      "step": 5391
+    },
+    {
+      "epoch": 0.6980613004498819,
+      "grad_norm": 2.1602040439474774,
+      "learning_rate": 3.161766329643332e-07,
+      "loss": 2.2573,
+      "step": 5392
+    },
+    {
+      "epoch": 0.6981907628572354,
+      "grad_norm": 1.9647660843750425,
+      "learning_rate": 3.159310806449821e-07,
+      "loss": 2.2974,
+      "step": 5393
+    },
+    {
+      "epoch": 0.6983202252645888,
+      "grad_norm": 1.5469109018669407,
+      "learning_rate": 3.15685591459222e-07,
+      "loss": 2.2053,
+      "step": 5394
+    },
+    {
+      "epoch": 0.6984496876719423,
+      "grad_norm": 4.860701980807949,
+      "learning_rate": 3.154401654571998e-07,
+      "loss": 2.7876,
+      "step": 5395
+    },
+    {
+      "epoch": 0.6984496876719423,
+      "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": 1.51416015625,
+      "eval_runtime": 13.9488,
+      "eval_samples_per_second": 3.154,
+      "eval_steps_per_second": 0.215,
+      "step": 5395
+    },
+    {
+      "epoch": 0.6985791500792957,
+      "grad_norm": 1.827738356764866,
+      "learning_rate": 3.1519480268904813e-07,
+      "loss": 2.4397,
+      "step": 5396
+    },
+    {
+      "epoch": 0.6987086124866492,
+      "grad_norm": 1.866702650859283,
+      "learning_rate": 3.149495032048873e-07,
+      "loss": 2.3193,
+      "step": 5397
+    },
+    {
+      "epoch": 0.6988380748940026,
+      "grad_norm": 1.8753172545572196,
+      "learning_rate": 3.1470426705482505e-07,
+      "loss": 2.2368,
+      "step": 5398
+    },
+    {
+      "epoch": 0.6989675373013561,
+      "grad_norm": 1.6360525619899013,
+      "learning_rate": 3.1445909428895546e-07,
+      "loss": 2.2825,
+      "step": 5399
+    },
+    {
+      "epoch": 0.6990969997087095,
+      "grad_norm": 1.4791927266508855,
+      "learning_rate": 3.1421398495736026e-07,
+      "loss": 2.4946,
+      "step": 5400
+    },
+    {
+      "epoch": 0.6990969997087095,
+      "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": 1.5142489671707153,
+      "eval_runtime": 14.4992,
+      "eval_samples_per_second": 3.035,
+      "eval_steps_per_second": 0.207,
+      "step": 5400
+    },
+    {
+      "epoch": 0.6992264621160631,
+      "grad_norm": 1.186635029016169,
+      "learning_rate": 3.139689391101085e-07,
+      "loss": 2.2834,
+      "step": 5401
+    },
+    {
+      "epoch": 0.6993559245234166,
+      "grad_norm": 1.8995638273612991,
+      "learning_rate": 3.1372395679725466e-07,
+      "loss": 2.4224,
+      "step": 5402
+    },
+    {
+      "epoch": 0.69948538693077,
+      "grad_norm": 1.6429652709146856,
+      "learning_rate": 3.13479038068842e-07,
+      "loss": 2.5115,
+      "step": 5403
+    },
+    {
+      "epoch": 0.6996148493381235,
+      "grad_norm": 1.9328421132076632,
+      "learning_rate": 3.132341829749002e-07,
+      "loss": 2.3618,
+      "step": 5404
+    },
+    {
+      "epoch": 0.6997443117454769,
+      "grad_norm": 1.6013927953761928,
+      "learning_rate": 3.1298939156544554e-07,
+      "loss": 2.2534,
+      "step": 5405
+    },
+    {
+      "epoch": 0.6997443117454769,
+      "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": 1.5170010328292847,
+      "eval_runtime": 14.4779,
+      "eval_samples_per_second": 3.039,
+      "eval_steps_per_second": 0.207,
+      "step": 5405
+    },
+    {
+      "epoch": 0.6998737741528304,
+      "grad_norm": 1.734191857629449,
+      "learning_rate": 3.1274466389048214e-07,
+      "loss": 2.5435,
+      "step": 5406
+    },
+    {
+      "epoch": 0.7000032365601838,
+      "grad_norm": 1.8077362625993985,
+      "learning_rate": 3.125000000000002e-07,
+      "loss": 2.4829,
+      "step": 5407
+    },
+    {
+      "epoch": 0.7001326989675373,
+      "grad_norm": 1.1982756877128877,
+      "learning_rate": 3.1225539994397704e-07,
+      "loss": 2.0586,
+      "step": 5408
+    },
+    {
+      "epoch": 0.7002621613748907,
+      "grad_norm": 1.4034583092961634,
+      "learning_rate": 3.120108637723778e-07,
+      "loss": 2.2112,
+      "step": 5409
+    },
+    {
+      "epoch": 0.7003916237822442,
+      "grad_norm": 1.8325979482049013,
+      "learning_rate": 3.1176639153515343e-07,
+      "loss": 2.4893,
+      "step": 5410
+    },
+    {
+      "epoch": 0.7003916237822442,
+      "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": 1.5114967823028564,
+      "eval_runtime": 14.0872,
+      "eval_samples_per_second": 3.123,
+      "eval_steps_per_second": 0.213,
+      "step": 5410
+    },
+    {
+      "epoch": 0.7005210861895977,
+      "grad_norm": 1.031947826763034,
+      "learning_rate": 3.115219832822427e-07,
+      "loss": 2.335,
+      "step": 5411
+    },
+    {
+      "epoch": 0.7006505485969512,
+      "grad_norm": 1.2032347547295184,
+      "learning_rate": 3.1127763906357086e-07,
+      "loss": 2.3384,
+      "step": 5412
+    },
+    {
+      "epoch": 0.7007800110043046,
+      "grad_norm": 1.40643082208662,
+      "learning_rate": 3.1103335892904983e-07,
+      "loss": 2.3037,
+      "step": 5413
+    },
+    {
+      "epoch": 0.7009094734116581,
+      "grad_norm": 2.602429291181452,
+      "learning_rate": 3.1078914292857925e-07,
+      "loss": 2.3455,
+      "step": 5414
+    },
+    {
+      "epoch": 0.7010389358190116,
+      "grad_norm": 3.1560069744271364,
+      "learning_rate": 3.1054499111204486e-07,
+      "loss": 3.1663,
+      "step": 5415
+    },
+    {
+      "epoch": 0.7010389358190116,
+      "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": 1.5157581567764282,
+      "eval_runtime": 14.8477,
+      "eval_samples_per_second": 2.963,
+      "eval_steps_per_second": 0.202,
+      "step": 5415
+    },
+    {
+      "epoch": 0.701168398226365,
+      "grad_norm": 2.1503959903003667,
+      "learning_rate": 3.1030090352931995e-07,
+      "loss": 2.3535,
+      "step": 5416
+    },
+    {
+      "epoch": 0.7012978606337185,
+      "grad_norm": 1.1744324577985705,
+      "learning_rate": 3.1005688023026387e-07,
+      "loss": 2.2505,
+      "step": 5417
+    },
+    {
+      "epoch": 0.7014273230410719,
+      "grad_norm": 2.1089863170437857,
+      "learning_rate": 3.098129212647239e-07,
+      "loss": 2.5706,
+      "step": 5418
+    },
+    {
+      "epoch": 0.7015567854484254,
+      "grad_norm": 1.419840139890607,
+      "learning_rate": 3.0956902668253327e-07,
+      "loss": 2.27,
+      "step": 5419
+    },
+    {
+      "epoch": 0.7016862478557789,
+      "grad_norm": 2.0435726528843596,
+      "learning_rate": 3.093251965335122e-07,
+      "loss": 2.6697,
+      "step": 5420
+    },
+    {
+      "epoch": 0.7016862478557789,
+      "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": 1.515625,
+      "eval_runtime": 14.0954,
+      "eval_samples_per_second": 3.122,
+      "eval_steps_per_second": 0.213,
+      "step": 5420
+    },
+    {
+      "epoch": 0.7018157102631324,
+      "grad_norm": 1.4000930091500776,
+      "learning_rate": 3.090814308674682e-07,
+      "loss": 2.4956,
+      "step": 5421
+    },
+    {
+      "epoch": 0.7019451726704858,
+      "grad_norm": 1.4296183949185863,
+      "learning_rate": 3.0883772973419553e-07,
+      "loss": 2.2876,
+      "step": 5422
+    },
+    {
+      "epoch": 0.7020746350778393,
+      "grad_norm": 1.7022564155071238,
+      "learning_rate": 3.085940931834749e-07,
+      "loss": 2.4507,
+      "step": 5423
+    },
+    {
+      "epoch": 0.7022040974851927,
+      "grad_norm": 1.609684352223778,
+      "learning_rate": 3.08350521265074e-07,
+      "loss": 2.4968,
+      "step": 5424
+    },
+    {
+      "epoch": 0.7023335598925462,
+      "grad_norm": 1.6220631821775575,
+      "learning_rate": 3.081070140287471e-07,
+      "loss": 2.5127,
+      "step": 5425
+    },
+    {
+      "epoch": 0.7023335598925462,
+      "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": 1.5156471729278564,
+      "eval_runtime": 13.9005,
+      "eval_samples_per_second": 3.165,
+      "eval_steps_per_second": 0.216,
+      "step": 5425
+    },
+    {
+      "epoch": 0.7024630222998997,
+      "grad_norm": 1.33830444157398,
+      "learning_rate": 3.078635715242357e-07,
+      "loss": 2.29,
+      "step": 5426
+    },
+    {
+      "epoch": 0.7025924847072531,
+      "grad_norm": 1.674054403047687,
+      "learning_rate": 3.076201938012681e-07,
+      "loss": 2.406,
+      "step": 5427
+    },
+    {
+      "epoch": 0.7027219471146066,
+      "grad_norm": 1.7850975837638183,
+      "learning_rate": 3.073768809095587e-07,
+      "loss": 2.2429,
+      "step": 5428
+    },
+    {
+      "epoch": 0.70285140952196,
+      "grad_norm": 2.0379084555803524,
+      "learning_rate": 3.071336328988095e-07,
+      "loss": 2.519,
+      "step": 5429
+    },
+    {
+      "epoch": 0.7029808719293136,
+      "grad_norm": 2.0983829443582382,
+      "learning_rate": 3.0689044981870865e-07,
+      "loss": 2.4026,
+      "step": 5430
+    },
+    {
+      "epoch": 0.7029808719293136,
+      "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": 1.5149813890457153,
+      "eval_runtime": 14.4294,
+      "eval_samples_per_second": 3.049,
+      "eval_steps_per_second": 0.208,
+      "step": 5430
+    },
+    {
+      "epoch": 0.703110334336667,
+      "grad_norm": 1.4576542430247819,
+      "learning_rate": 3.06647331718931e-07,
+      "loss": 2.5098,
+      "step": 5431
+    },
+    {
+      "epoch": 0.7032397967440205,
+      "grad_norm": 1.4999026407300369,
+      "learning_rate": 3.064042786491388e-07,
+      "loss": 2.3525,
+      "step": 5432
+    },
+    {
+      "epoch": 0.7033692591513739,
+      "grad_norm": 2.3911531733057116,
+      "learning_rate": 3.0616129065898004e-07,
+      "loss": 2.447,
+      "step": 5433
+    },
+    {
+      "epoch": 0.7034987215587274,
+      "grad_norm": 1.691530032429575,
+      "learning_rate": 3.059183677980905e-07,
+      "loss": 2.3381,
+      "step": 5434
+    },
+    {
+      "epoch": 0.7036281839660808,
+      "grad_norm": 2.1022824376195604,
+      "learning_rate": 3.056755101160918e-07,
+      "loss": 2.4399,
+      "step": 5435
+    },
+    {
+      "epoch": 0.7036281839660808,
+      "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": 1.510786533355713,
+      "eval_runtime": 15.5736,
+      "eval_samples_per_second": 2.825,
+      "eval_steps_per_second": 0.193,
+      "step": 5435
+    },
+    {
+      "epoch": 0.7037576463734343,
+      "grad_norm": 2.0826754739211832,
+      "learning_rate": 3.054327176625924e-07,
+      "loss": 2.3694,
+      "step": 5436
+    },
+    {
+      "epoch": 0.7038871087807878,
+      "grad_norm": 1.552936864115532,
+      "learning_rate": 3.0518999048718795e-07,
+      "loss": 2.2969,
+      "step": 5437
+    },
+    {
+      "epoch": 0.7040165711881412,
+      "grad_norm": 2.3615324849454415,
+      "learning_rate": 3.049473286394599e-07,
+      "loss": 2.2988,
+      "step": 5438
+    },
+    {
+      "epoch": 0.7041460335954947,
+      "grad_norm": 1.9947144084490287,
+      "learning_rate": 3.047047321689775e-07,
+      "loss": 2.4597,
+      "step": 5439
+    },
+    {
+      "epoch": 0.7042754960028482,
+      "grad_norm": 2.484209298485715,
+      "learning_rate": 3.0446220112529555e-07,
+      "loss": 2.3804,
+      "step": 5440
+    },
+    {
+      "epoch": 0.7042754960028482,
+      "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": 1.515625,
+      "eval_runtime": 14.9414,
+      "eval_samples_per_second": 2.945,
+      "eval_steps_per_second": 0.201,
+      "step": 5440
+    },
+    {
+      "epoch": 0.7044049584102017,
+      "grad_norm": 1.5148496222853771,
+      "learning_rate": 3.0421973555795585e-07,
+      "loss": 2.2559,
+      "step": 5441
+    },
+    {
+      "epoch": 0.7045344208175551,
+      "grad_norm": 2.050627791832364,
+      "learning_rate": 3.0397733551648725e-07,
+      "loss": 2.2622,
+      "step": 5442
+    },
+    {
+      "epoch": 0.7046638832249086,
+      "grad_norm": 1.556620680990198,
+      "learning_rate": 3.0373500105040456e-07,
+      "loss": 2.5139,
+      "step": 5443
+    },
+    {
+      "epoch": 0.704793345632262,
+      "grad_norm": 2.105901211082287,
+      "learning_rate": 3.0349273220920955e-07,
+      "loss": 2.8171,
+      "step": 5444
+    },
+    {
+      "epoch": 0.7049228080396155,
+      "grad_norm": 1.91482161355928,
+      "learning_rate": 3.0325052904239096e-07,
+      "loss": 2.4438,
+      "step": 5445
+    },
+    {
+      "epoch": 0.7049228080396155,
+      "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": 1.5136052370071411,
+      "eval_runtime": 14.4262,
+      "eval_samples_per_second": 3.05,
+      "eval_steps_per_second": 0.208,
+      "step": 5445
+    },
+    {
+      "epoch": 0.7050522704469689,
+      "grad_norm": 2.0759088329303936,
+      "learning_rate": 3.0300839159942335e-07,
+      "loss": 2.3867,
+      "step": 5446
+    },
+    {
+      "epoch": 0.7051817328543224,
+      "grad_norm": 2.5915277587769,
+      "learning_rate": 3.027663199297682e-07,
+      "loss": 2.2825,
+      "step": 5447
+    },
+    {
+      "epoch": 0.7053111952616758,
+      "grad_norm": 1.4249593657417663,
+      "learning_rate": 3.025243140828734e-07,
+      "loss": 2.4436,
+      "step": 5448
+    },
+    {
+      "epoch": 0.7054406576690293,
+      "grad_norm": 1.8149636589990787,
+      "learning_rate": 3.022823741081737e-07,
+      "loss": 2.6514,
+      "step": 5449
+    },
+    {
+      "epoch": 0.7055701200763829,
+      "grad_norm": 2.6025866686770653,
+      "learning_rate": 3.020405000550905e-07,
+      "loss": 2.3667,
+      "step": 5450
+    },
+    {
+      "epoch": 0.7055701200763829,
+      "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": 1.5157581567764282,
+      "eval_runtime": 14.2301,
+      "eval_samples_per_second": 3.092,
+      "eval_steps_per_second": 0.211,
+      "step": 5450
+    },
+    {
+      "epoch": 0.7056995824837363,
+      "grad_norm": 1.574276657392704,
+      "learning_rate": 3.0179869197303125e-07,
+      "loss": 2.4951,
+      "step": 5451
+    },
+    {
+      "epoch": 0.7058290448910898,
+      "grad_norm": 1.8247724014621267,
+      "learning_rate": 3.0155694991139e-07,
+      "loss": 2.2301,
+      "step": 5452
+    },
+    {
+      "epoch": 0.7059585072984432,
+      "grad_norm": 1.2739726890028502,
+      "learning_rate": 3.0131527391954776e-07,
+      "loss": 2.4631,
+      "step": 5453
+    },
+    {
+      "epoch": 0.7060879697057967,
+      "grad_norm": 1.4741302234201548,
+      "learning_rate": 3.0107366404687145e-07,
+      "loss": 2.4326,
+      "step": 5454
+    },
+    {
+      "epoch": 0.7062174321131501,
+      "grad_norm": 1.7257062152662175,
+      "learning_rate": 3.008321203427151e-07,
+      "loss": 2.5139,
+      "step": 5455
+    },
+    {
+      "epoch": 0.7062174321131501,
+      "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": 1.512895107269287,
+      "eval_runtime": 14.3123,
+      "eval_samples_per_second": 3.074,
+      "eval_steps_per_second": 0.21,
+      "step": 5455
+    },
+    {
+      "epoch": 0.7063468945205036,
+      "grad_norm": 1.4845271698246092,
+      "learning_rate": 3.0059064285641857e-07,
+      "loss": 2.3633,
+      "step": 5456
+    },
+    {
+      "epoch": 0.706476356927857,
+      "grad_norm": 1.8092886527834913,
+      "learning_rate": 3.00349231637309e-07,
+      "loss": 2.4194,
+      "step": 5457
+    },
+    {
+      "epoch": 0.7066058193352105,
+      "grad_norm": 1.7936588912183051,
+      "learning_rate": 3.001078867346992e-07,
+      "loss": 2.5867,
+      "step": 5458
+    },
+    {
+      "epoch": 0.706735281742564,
+      "grad_norm": 2.0701410817647896,
+      "learning_rate": 2.9986660819788847e-07,
+      "loss": 2.2192,
+      "step": 5459
+    },
+    {
+      "epoch": 0.7068647441499175,
+      "grad_norm": 1.3732517347453355,
+      "learning_rate": 2.9962539607616353e-07,
+      "loss": 2.332,
+      "step": 5460
+    },
+    {
+      "epoch": 0.7068647441499175,
+      "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": 1.5136052370071411,
+      "eval_runtime": 13.7143,
+      "eval_samples_per_second": 3.208,
+      "eval_steps_per_second": 0.219,
+      "step": 5460
+    },
+    {
+      "epoch": 0.706994206557271,
+      "grad_norm": 1.3969195107471228,
+      "learning_rate": 2.9938425041879615e-07,
+      "loss": 2.2974,
+      "step": 5461
+    },
+    {
+      "epoch": 0.7071236689646244,
+      "grad_norm": 4.364035985519196,
+      "learning_rate": 2.9914317127504587e-07,
+      "loss": 2.5571,
+      "step": 5462
+    },
+    {
+      "epoch": 0.7072531313719779,
+      "grad_norm": 1.6150948091088442,
+      "learning_rate": 2.9890215869415764e-07,
+      "loss": 2.2448,
+      "step": 5463
+    },
+    {
+      "epoch": 0.7073825937793313,
+      "grad_norm": 1.5417784519487534,
+      "learning_rate": 2.9866121272536297e-07,
+      "loss": 2.5642,
+      "step": 5464
+    },
+    {
+      "epoch": 0.7075120561866848,
+      "grad_norm": 1.266865102094396,
+      "learning_rate": 2.984203334178804e-07,
+      "loss": 2.4429,
+      "step": 5465
+    },
+    {
+      "epoch": 0.7075120561866848,
+      "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": 1.5128728151321411,
+      "eval_runtime": 12.8483,
+      "eval_samples_per_second": 3.425,
+      "eval_steps_per_second": 0.233,
+      "step": 5465
+    },
+    {
+      "epoch": 0.7076415185940382,
+      "grad_norm": 1.7082974928030255,
+      "learning_rate": 2.98179520820914e-07,
+      "loss": 2.5574,
+      "step": 5466
+    },
+    {
+      "epoch": 0.7077709810013917,
+      "grad_norm": 1.4952069787003692,
+      "learning_rate": 2.979387749836548e-07,
+      "loss": 2.292,
+      "step": 5467
+    },
+    {
+      "epoch": 0.7079004434087451,
+      "grad_norm": 1.6882378452340079,
+      "learning_rate": 2.976980959552805e-07,
+      "loss": 2.3267,
+      "step": 5468
+    },
+    {
+      "epoch": 0.7080299058160987,
+      "grad_norm": 1.4705000244831505,
+      "learning_rate": 2.974574837849538e-07,
+      "loss": 2.6318,
+      "step": 5469
+    },
+    {
+      "epoch": 0.7081593682234522,
+      "grad_norm": 2.4697891945795516,
+      "learning_rate": 2.972169385218252e-07,
+      "loss": 2.6724,
+      "step": 5470
+    },
+    {
+      "epoch": 0.7081593682234522,
+      "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": 1.5128062963485718,
+      "eval_runtime": 14.6489,
+      "eval_samples_per_second": 3.004,
+      "eval_steps_per_second": 0.205,
+      "step": 5470
+    },
+    {
+      "epoch": 0.7082888306308056,
+      "grad_norm": 2.5365204004259683,
+      "learning_rate": 2.969764602150305e-07,
+      "loss": 2.3005,
+      "step": 5471
+    },
+    {
+      "epoch": 0.7084182930381591,
+      "grad_norm": 2.2054646464841112,
+      "learning_rate": 2.967360489136926e-07,
+      "loss": 2.4115,
+      "step": 5472
+    },
+    {
+      "epoch": 0.7085477554455125,
+      "grad_norm": 1.2705979197329882,
+      "learning_rate": 2.964957046669208e-07,
+      "loss": 2.3721,
+      "step": 5473
+    },
+    {
+      "epoch": 0.708677217852866,
+      "grad_norm": 2.7935239563139307,
+      "learning_rate": 2.9625542752380927e-07,
+      "loss": 2.4829,
+      "step": 5474
+    },
+    {
+      "epoch": 0.7088066802602194,
+      "grad_norm": 2.333047871088322,
+      "learning_rate": 2.9601521753344e-07,
+      "loss": 2.4336,
+      "step": 5475
+    },
+    {
+      "epoch": 0.7088066802602194,
+      "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": 1.5121182203292847,
+      "eval_runtime": 14.6633,
+      "eval_samples_per_second": 3.001,
+      "eval_steps_per_second": 0.205,
+      "step": 5475
+    },
+    {
+      "epoch": 0.7089361426675729,
+      "grad_norm": 1.9329422209847855,
+      "learning_rate": 2.957750747448811e-07,
+      "loss": 2.2339,
+      "step": 5476
+    },
+    {
+      "epoch": 0.7090656050749263,
+      "grad_norm": 2.3940518724987983,
+      "learning_rate": 2.9553499920718593e-07,
+      "loss": 2.2429,
+      "step": 5477
+    },
+    {
+      "epoch": 0.7091950674822798,
+      "grad_norm": 1.9642118631365526,
+      "learning_rate": 2.952949909693954e-07,
+      "loss": 2.7961,
+      "step": 5478
+    },
+    {
+      "epoch": 0.7093245298896333,
+      "grad_norm": 1.062591419130222,
+      "learning_rate": 2.9505505008053583e-07,
+      "loss": 2.1123,
+      "step": 5479
+    },
+    {
+      "epoch": 0.7094539922969868,
+      "grad_norm": 2.5608689953969916,
+      "learning_rate": 2.9481517658961965e-07,
+      "loss": 2.1968,
+      "step": 5480
+    },
+    {
+      "epoch": 0.7094539922969868,
+      "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": 1.5107200145721436,
+      "eval_runtime": 14.1735,
+      "eval_samples_per_second": 3.104,
+      "eval_steps_per_second": 0.212,
+      "step": 5480
+    },
+    {
+      "epoch": 0.7095834547043403,
+      "grad_norm": 1.337703409697096,
+      "learning_rate": 2.9457537054564647e-07,
+      "loss": 2.4028,
+      "step": 5481
+    },
+    {
+      "epoch": 0.7097129171116937,
+      "grad_norm": 1.9010823142558655,
+      "learning_rate": 2.9433563199760095e-07,
+      "loss": 2.3994,
+      "step": 5482
+    },
+    {
+      "epoch": 0.7098423795190472,
+      "grad_norm": 1.7257209889372729,
+      "learning_rate": 2.940959609944552e-07,
+      "loss": 2.3069,
+      "step": 5483
+    },
+    {
+      "epoch": 0.7099718419264006,
+      "grad_norm": 2.0761339000769263,
+      "learning_rate": 2.938563575851665e-07,
+      "loss": 2.2153,
+      "step": 5484
+    },
+    {
+      "epoch": 0.7101013043337541,
+      "grad_norm": 3.026756961222471,
+      "learning_rate": 2.936168218186785e-07,
+      "loss": 2.6416,
+      "step": 5485
+    },
+    {
+      "epoch": 0.7101013043337541,
+      "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": 1.5107200145721436,
+      "eval_runtime": 13.9592,
+      "eval_samples_per_second": 3.152,
+      "eval_steps_per_second": 0.215,
+      "step": 5485
+    },
+    {
+      "epoch": 0.7102307667411075,
+      "grad_norm": 1.4180293171233147,
+      "learning_rate": 2.933773537439217e-07,
+      "loss": 2.4192,
+      "step": 5486
+    },
+    {
+      "epoch": 0.710360229148461,
+      "grad_norm": 1.3914393537554786,
+      "learning_rate": 2.931379534098118e-07,
+      "loss": 2.303,
+      "step": 5487
+    },
+    {
+      "epoch": 0.7104896915558144,
+      "grad_norm": 1.7370934374122329,
+      "learning_rate": 2.928986208652518e-07,
+      "loss": 2.3853,
+      "step": 5488
+    },
+    {
+      "epoch": 0.710619153963168,
+      "grad_norm": 1.585344876482885,
+      "learning_rate": 2.9265935615912956e-07,
+      "loss": 2.3574,
+      "step": 5489
+    },
+    {
+      "epoch": 0.7107486163705214,
+      "grad_norm": 1.4600865931055707,
+      "learning_rate": 2.9242015934032033e-07,
+      "loss": 2.3567,
+      "step": 5490
+    },
+    {
+      "epoch": 0.7107486163705214,
+      "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": 1.511518955230713,
+      "eval_runtime": 14.7041,
+      "eval_samples_per_second": 2.992,
+      "eval_steps_per_second": 0.204,
+      "step": 5490
+    },
+    {
+      "epoch": 0.7108780787778749,
+      "grad_norm": 1.5533301614533888,
+      "learning_rate": 2.9218103045768456e-07,
+      "loss": 2.21,
+      "step": 5491
+    },
+    {
+      "epoch": 0.7110075411852284,
+      "grad_norm": 1.8222863543886278,
+      "learning_rate": 2.9194196956006914e-07,
+      "loss": 2.3228,
+      "step": 5492
+    },
+    {
+      "epoch": 0.7111370035925818,
+      "grad_norm": 1.5259869863465987,
+      "learning_rate": 2.9170297669630735e-07,
+      "loss": 2.2871,
+      "step": 5493
+    },
+    {
+      "epoch": 0.7112664659999353,
+      "grad_norm": 1.9441677128782537,
+      "learning_rate": 2.91464051915218e-07,
+      "loss": 2.3655,
+      "step": 5494
+    },
+    {
+      "epoch": 0.7113959284072887,
+      "grad_norm": 2.031245198793413,
+      "learning_rate": 2.912251952656065e-07,
+      "loss": 2.2627,
+      "step": 5495
+    },
+    {
+      "epoch": 0.7113959284072887,
+      "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": 1.5135387182235718,
+      "eval_runtime": 14.7326,
+      "eval_samples_per_second": 2.987,
+      "eval_steps_per_second": 0.204,
+      "step": 5495
+    },
+    {
+      "epoch": 0.7115253908146422,
+      "grad_norm": 1.6557814810252556,
+      "learning_rate": 2.9098640679626473e-07,
+      "loss": 2.6448,
+      "step": 5496
+    },
+    {
+      "epoch": 0.7116548532219956,
+      "grad_norm": 2.3030706547413096,
+      "learning_rate": 2.907476865559689e-07,
+      "loss": 2.6919,
+      "step": 5497
+    },
+    {
+      "epoch": 0.7117843156293491,
+      "grad_norm": 1.454468789961194,
+      "learning_rate": 2.9050903459348323e-07,
+      "loss": 2.3105,
+      "step": 5498
+    },
+    {
+      "epoch": 0.7119137780367026,
+      "grad_norm": 1.180216293314466,
+      "learning_rate": 2.902704509575572e-07,
+      "loss": 2.2432,
+      "step": 5499
+    },
+    {
+      "epoch": 0.7120432404440561,
+      "grad_norm": 0.9700997666670762,
+      "learning_rate": 2.90031935696926e-07,
+      "loss": 2.1406,
+      "step": 5500
+    },
+    {
+      "epoch": 0.7120432404440561,
+      "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": 1.510009765625,
+      "eval_runtime": 13.9422,
+      "eval_samples_per_second": 3.156,
+      "eval_steps_per_second": 0.215,
+      "step": 5500
     }
   ],
   "logging_steps": 1,
@@ -51042,7 +56142,7 @@
       "attributes": {}
     }
   },
-  "total_flos": 4202821198053376.0,
+  "total_flos": 4624798951899136.0,
   "train_batch_size": 4,
   "trial_name": null,
   "trial_params": null