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": 0.06473120367673237,
+  "epoch": 0.12946240735346473,
   "eval_steps": 5,
-  "global_step": 500,
+  "global_step": 1000,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
@@ -5123,6 +5123,5106 @@
       "eval_samples_per_second": 2.178,
       "eval_steps_per_second": 0.149,
       "step": 500
+    },
+    {
+      "epoch": 0.06486066608408583,
+      "grad_norm": 0.15533986164701288,
+      "learning_rate": 8.101552393272963e-07,
+      "loss": 1.2524,
+      "step": 501
+    },
+    {
+      "epoch": 0.0649901284914393,
+      "grad_norm": 0.13272793025742835,
+      "learning_rate": 8.117723156532989e-07,
+      "loss": 1.2069,
+      "step": 502
+    },
+    {
+      "epoch": 0.06511959089879277,
+      "grad_norm": 0.14307161885062755,
+      "learning_rate": 8.133893919793015e-07,
+      "loss": 1.1619,
+      "step": 503
+    },
+    {
+      "epoch": 0.06524905330614623,
+      "grad_norm": 0.2049652065365769,
+      "learning_rate": 8.150064683053041e-07,
+      "loss": 1.4006,
+      "step": 504
+    },
+    {
+      "epoch": 0.06537851571349969,
+      "grad_norm": 0.16082439520482802,
+      "learning_rate": 8.166235446313067e-07,
+      "loss": 1.2555,
+      "step": 505
+    },
+    {
+      "epoch": 0.06537851571349969,
+      "eval_PRM Accuracy": 0.41509433962264153,
+      "eval_PRM F1": 0.45614035087719296,
+      "eval_PRM F1 AUC": 0.5479308538501834,
+      "eval_PRM F1 Neg": 0.3673469387755102,
+      "eval_PRM NPV": 0.24,
+      "eval_PRM Precision": 0.8387096774193549,
+      "eval_PRM Recall": 0.3132530120481928,
+      "eval_PRM Specificty": 0.782608695652174,
+      "eval_loss": 0.865234375,
+      "eval_runtime": 20.0475,
+      "eval_samples_per_second": 2.195,
+      "eval_steps_per_second": 0.15,
+      "step": 505
+    },
+    {
+      "epoch": 0.06550797812085316,
+      "grad_norm": 0.19053121423386557,
+      "learning_rate": 8.182406209573091e-07,
+      "loss": 1.3206,
+      "step": 506
+    },
+    {
+      "epoch": 0.06563744052820662,
+      "grad_norm": 0.13037440939866252,
+      "learning_rate": 8.198576972833118e-07,
+      "loss": 1.1543,
+      "step": 507
+    },
+    {
+      "epoch": 0.06576690293556009,
+      "grad_norm": 0.16469524550826165,
+      "learning_rate": 8.214747736093144e-07,
+      "loss": 1.1472,
+      "step": 508
+    },
+    {
+      "epoch": 0.06589636534291356,
+      "grad_norm": 0.15490494265034072,
+      "learning_rate": 8.23091849935317e-07,
+      "loss": 1.2928,
+      "step": 509
+    },
+    {
+      "epoch": 0.06602582775026701,
+      "grad_norm": 0.15410454528969175,
+      "learning_rate": 8.247089262613196e-07,
+      "loss": 1.1617,
+      "step": 510
+    },
+    {
+      "epoch": 0.06602582775026701,
+      "eval_PRM Accuracy": 0.41509433962264153,
+      "eval_PRM F1": 0.45614035087719296,
+      "eval_PRM F1 AUC": 0.5479308538501834,
+      "eval_PRM F1 Neg": 0.3673469387755102,
+      "eval_PRM NPV": 0.24,
+      "eval_PRM Precision": 0.8387096774193549,
+      "eval_PRM Recall": 0.3132530120481928,
+      "eval_PRM Specificty": 0.782608695652174,
+      "eval_loss": 0.8649680614471436,
+      "eval_runtime": 19.8651,
+      "eval_samples_per_second": 2.215,
+      "eval_steps_per_second": 0.151,
+      "step": 510
+    },
+    {
+      "epoch": 0.06615529015762048,
+      "grad_norm": 0.1422897460647621,
+      "learning_rate": 8.263260025873223e-07,
+      "loss": 1.1007,
+      "step": 511
+    },
+    {
+      "epoch": 0.06628475256497394,
+      "grad_norm": 0.12994141541611715,
+      "learning_rate": 8.279430789133249e-07,
+      "loss": 1.231,
+      "step": 512
+    },
+    {
+      "epoch": 0.06641421497232741,
+      "grad_norm": 0.13637163812302364,
+      "learning_rate": 8.295601552393273e-07,
+      "loss": 1.0859,
+      "step": 513
+    },
+    {
+      "epoch": 0.06654367737968088,
+      "grad_norm": 0.18845660841968817,
+      "learning_rate": 8.311772315653299e-07,
+      "loss": 1.2546,
+      "step": 514
+    },
+    {
+      "epoch": 0.06667313978703435,
+      "grad_norm": 0.19155304423346833,
+      "learning_rate": 8.327943078913325e-07,
+      "loss": 1.3252,
+      "step": 515
+    },
+    {
+      "epoch": 0.06667313978703435,
+      "eval_PRM Accuracy": 0.41509433962264153,
+      "eval_PRM F1": 0.45614035087719296,
+      "eval_PRM F1 AUC": 0.5479308538501834,
+      "eval_PRM F1 Neg": 0.3673469387755102,
+      "eval_PRM NPV": 0.24,
+      "eval_PRM Precision": 0.8387096774193549,
+      "eval_PRM Recall": 0.3132530120481928,
+      "eval_PRM Specificty": 0.782608695652174,
+      "eval_loss": 0.8631036877632141,
+      "eval_runtime": 20.4647,
+      "eval_samples_per_second": 2.15,
+      "eval_steps_per_second": 0.147,
+      "step": 515
+    },
+    {
+      "epoch": 0.0668026021943878,
+      "grad_norm": 0.1354322038132936,
+      "learning_rate": 8.344113842173351e-07,
+      "loss": 1.1366,
+      "step": 516
+    },
+    {
+      "epoch": 0.06693206460174127,
+      "grad_norm": 0.20426491263243218,
+      "learning_rate": 8.360284605433378e-07,
+      "loss": 1.3192,
+      "step": 517
+    },
+    {
+      "epoch": 0.06706152700909473,
+      "grad_norm": 0.16293582348557498,
+      "learning_rate": 8.376455368693403e-07,
+      "loss": 1.114,
+      "step": 518
+    },
+    {
+      "epoch": 0.0671909894164482,
+      "grad_norm": 0.16853350946740797,
+      "learning_rate": 8.392626131953428e-07,
+      "loss": 1.1948,
+      "step": 519
+    },
+    {
+      "epoch": 0.06732045182380167,
+      "grad_norm": 0.18334128196239635,
+      "learning_rate": 8.408796895213455e-07,
+      "loss": 1.1876,
+      "step": 520
+    },
+    {
+      "epoch": 0.06732045182380167,
+      "eval_PRM Accuracy": 0.41509433962264153,
+      "eval_PRM F1": 0.45614035087719296,
+      "eval_PRM F1 AUC": 0.5479308538501834,
+      "eval_PRM F1 Neg": 0.3673469387755102,
+      "eval_PRM NPV": 0.24,
+      "eval_PRM Precision": 0.8387096774193549,
+      "eval_PRM Recall": 0.3132530120481928,
+      "eval_PRM Specificty": 0.782608695652174,
+      "eval_loss": 0.8650568127632141,
+      "eval_runtime": 19.9449,
+      "eval_samples_per_second": 2.206,
+      "eval_steps_per_second": 0.15,
+      "step": 520
+    },
+    {
+      "epoch": 0.06744991423115512,
+      "grad_norm": 0.18358978489016634,
+      "learning_rate": 8.42496765847348e-07,
+      "loss": 1.3896,
+      "step": 521
+    },
+    {
+      "epoch": 0.06757937663850859,
+      "grad_norm": 0.2038415529520888,
+      "learning_rate": 8.441138421733506e-07,
+      "loss": 1.2625,
+      "step": 522
+    },
+    {
+      "epoch": 0.06770883904586206,
+      "grad_norm": 0.17070099910417433,
+      "learning_rate": 8.457309184993533e-07,
+      "loss": 1.1179,
+      "step": 523
+    },
+    {
+      "epoch": 0.06783830145321552,
+      "grad_norm": 0.10916369039559713,
+      "learning_rate": 8.473479948253558e-07,
+      "loss": 1.0026,
+      "step": 524
+    },
+    {
+      "epoch": 0.06796776386056899,
+      "grad_norm": 0.16343640191204784,
+      "learning_rate": 8.489650711513583e-07,
+      "loss": 1.2334,
+      "step": 525
+    },
+    {
+      "epoch": 0.06796776386056899,
+      "eval_PRM Accuracy": 0.41509433962264153,
+      "eval_PRM F1": 0.45614035087719296,
+      "eval_PRM F1 AUC": 0.5479308538501834,
+      "eval_PRM F1 Neg": 0.3673469387755102,
+      "eval_PRM NPV": 0.24,
+      "eval_PRM Precision": 0.8387096774193549,
+      "eval_PRM Recall": 0.3132530120481928,
+      "eval_PRM Specificty": 0.782608695652174,
+      "eval_loss": 0.8608842492103577,
+      "eval_runtime": 20.6875,
+      "eval_samples_per_second": 2.127,
+      "eval_steps_per_second": 0.145,
+      "step": 525
+    },
+    {
+      "epoch": 0.06809722626792246,
+      "grad_norm": 0.17444180191071568,
+      "learning_rate": 8.50582147477361e-07,
+      "loss": 1.2553,
+      "step": 526
+    },
+    {
+      "epoch": 0.06822668867527591,
+      "grad_norm": 0.18837570556921665,
+      "learning_rate": 8.521992238033635e-07,
+      "loss": 1.3142,
+      "step": 527
+    },
+    {
+      "epoch": 0.06835615108262938,
+      "grad_norm": 0.1734288492759629,
+      "learning_rate": 8.538163001293662e-07,
+      "loss": 1.324,
+      "step": 528
+    },
+    {
+      "epoch": 0.06848561348998285,
+      "grad_norm": 0.12331213871707122,
+      "learning_rate": 8.554333764553688e-07,
+      "loss": 1.1145,
+      "step": 529
+    },
+    {
+      "epoch": 0.06861507589733631,
+      "grad_norm": 0.16831465606276777,
+      "learning_rate": 8.570504527813713e-07,
+      "loss": 1.13,
+      "step": 530
+    },
+    {
+      "epoch": 0.06861507589733631,
+      "eval_PRM Accuracy": 0.41509433962264153,
+      "eval_PRM F1": 0.45614035087719296,
+      "eval_PRM F1 AUC": 0.5479308538501834,
+      "eval_PRM F1 Neg": 0.3673469387755102,
+      "eval_PRM NPV": 0.24,
+      "eval_PRM Precision": 0.8387096774193549,
+      "eval_PRM Recall": 0.3132530120481928,
+      "eval_PRM Specificty": 0.782608695652174,
+      "eval_loss": 0.8599964380264282,
+      "eval_runtime": 20.8744,
+      "eval_samples_per_second": 2.108,
+      "eval_steps_per_second": 0.144,
+      "step": 530
+    },
+    {
+      "epoch": 0.06874453830468978,
+      "grad_norm": 0.1445882247679938,
+      "learning_rate": 8.58667529107374e-07,
+      "loss": 1.158,
+      "step": 531
+    },
+    {
+      "epoch": 0.06887400071204323,
+      "grad_norm": 0.19259645345371518,
+      "learning_rate": 8.602846054333765e-07,
+      "loss": 1.3007,
+      "step": 532
+    },
+    {
+      "epoch": 0.0690034631193967,
+      "grad_norm": 0.18263105756733602,
+      "learning_rate": 8.61901681759379e-07,
+      "loss": 1.3082,
+      "step": 533
+    },
+    {
+      "epoch": 0.06913292552675017,
+      "grad_norm": 0.1919604472818187,
+      "learning_rate": 8.635187580853817e-07,
+      "loss": 1.2994,
+      "step": 534
+    },
+    {
+      "epoch": 0.06926238793410364,
+      "grad_norm": 0.16680511489009675,
+      "learning_rate": 8.651358344113843e-07,
+      "loss": 1.2211,
+      "step": 535
+    },
+    {
+      "epoch": 0.06926238793410364,
+      "eval_PRM Accuracy": 0.41509433962264153,
+      "eval_PRM F1": 0.45614035087719296,
+      "eval_PRM F1 AUC": 0.5479308538501834,
+      "eval_PRM F1 Neg": 0.3673469387755102,
+      "eval_PRM NPV": 0.24,
+      "eval_PRM Precision": 0.8387096774193549,
+      "eval_PRM Recall": 0.3132530120481928,
+      "eval_PRM Specificty": 0.782608695652174,
+      "eval_loss": 0.8592862486839294,
+      "eval_runtime": 20.7683,
+      "eval_samples_per_second": 2.119,
+      "eval_steps_per_second": 0.144,
+      "step": 535
+    },
+    {
+      "epoch": 0.0693918503414571,
+      "grad_norm": 0.11888426697223828,
+      "learning_rate": 8.667529107373869e-07,
+      "loss": 1.101,
+      "step": 536
+    },
+    {
+      "epoch": 0.06952131274881057,
+      "grad_norm": 0.13619872662400598,
+      "learning_rate": 8.683699870633895e-07,
+      "loss": 1.0791,
+      "step": 537
+    },
+    {
+      "epoch": 0.06965077515616402,
+      "grad_norm": 0.21298827859890154,
+      "learning_rate": 8.69987063389392e-07,
+      "loss": 1.3262,
+      "step": 538
+    },
+    {
+      "epoch": 0.06978023756351749,
+      "grad_norm": 0.19935232802501582,
+      "learning_rate": 8.716041397153946e-07,
+      "loss": 1.3511,
+      "step": 539
+    },
+    {
+      "epoch": 0.06990969997087096,
+      "grad_norm": 0.16116761083285033,
+      "learning_rate": 8.732212160413972e-07,
+      "loss": 1.1763,
+      "step": 540
+    },
+    {
+      "epoch": 0.06990969997087096,
+      "eval_PRM Accuracy": 0.41509433962264153,
+      "eval_PRM F1": 0.45614035087719296,
+      "eval_PRM F1 AUC": 0.5479308538501834,
+      "eval_PRM F1 Neg": 0.3673469387755102,
+      "eval_PRM NPV": 0.24,
+      "eval_PRM Precision": 0.8387096774193549,
+      "eval_PRM Recall": 0.3132530120481928,
+      "eval_PRM Specificty": 0.782608695652174,
+      "eval_loss": 0.8583984375,
+      "eval_runtime": 19.7691,
+      "eval_samples_per_second": 2.226,
+      "eval_steps_per_second": 0.152,
+      "step": 540
+    },
+    {
+      "epoch": 0.07003916237822443,
+      "grad_norm": 0.16606151506619116,
+      "learning_rate": 8.748382923673998e-07,
+      "loss": 1.248,
+      "step": 541
+    },
+    {
+      "epoch": 0.0701686247855779,
+      "grad_norm": 0.19155037350559567,
+      "learning_rate": 8.764553686934024e-07,
+      "loss": 1.2234,
+      "step": 542
+    },
+    {
+      "epoch": 0.07029808719293135,
+      "grad_norm": 0.17041558458586142,
+      "learning_rate": 8.78072445019405e-07,
+      "loss": 1.4387,
+      "step": 543
+    },
+    {
+      "epoch": 0.07042754960028481,
+      "grad_norm": 0.16917436559163487,
+      "learning_rate": 8.796895213454077e-07,
+      "loss": 1.2923,
+      "step": 544
+    },
+    {
+      "epoch": 0.07055701200763828,
+      "grad_norm": 0.1969624662563443,
+      "learning_rate": 8.813065976714101e-07,
+      "loss": 1.1927,
+      "step": 545
+    },
+    {
+      "epoch": 0.07055701200763828,
+      "eval_PRM Accuracy": 0.4056603773584906,
+      "eval_PRM F1": 0.45217391304347826,
+      "eval_PRM F1 AUC": 0.5261917234154007,
+      "eval_PRM F1 Neg": 0.35051546391752575,
+      "eval_PRM NPV": 0.22972972972972974,
+      "eval_PRM Precision": 0.8125,
+      "eval_PRM Recall": 0.3132530120481928,
+      "eval_PRM Specificty": 0.7391304347826086,
+      "eval_loss": 0.8569779992103577,
+      "eval_runtime": 20.7431,
+      "eval_samples_per_second": 2.121,
+      "eval_steps_per_second": 0.145,
+      "step": 545
+    },
+    {
+      "epoch": 0.07068647441499175,
+      "grad_norm": 0.16758299120956285,
+      "learning_rate": 8.829236739974127e-07,
+      "loss": 1.2706,
+      "step": 546
+    },
+    {
+      "epoch": 0.07081593682234522,
+      "grad_norm": 0.1711082914098508,
+      "learning_rate": 8.845407503234154e-07,
+      "loss": 1.2469,
+      "step": 547
+    },
+    {
+      "epoch": 0.07094539922969868,
+      "grad_norm": 0.20270607529372123,
+      "learning_rate": 8.861578266494179e-07,
+      "loss": 1.2977,
+      "step": 548
+    },
+    {
+      "epoch": 0.07107486163705214,
+      "grad_norm": 0.1852464253854372,
+      "learning_rate": 8.877749029754205e-07,
+      "loss": 1.3358,
+      "step": 549
+    },
+    {
+      "epoch": 0.0712043240444056,
+      "grad_norm": 0.19841295958610808,
+      "learning_rate": 8.893919793014232e-07,
+      "loss": 1.3088,
+      "step": 550
+    },
+    {
+      "epoch": 0.0712043240444056,
+      "eval_PRM Accuracy": 0.4056603773584906,
+      "eval_PRM F1": 0.45217391304347826,
+      "eval_PRM F1 AUC": 0.5261917234154007,
+      "eval_PRM F1 Neg": 0.35051546391752575,
+      "eval_PRM NPV": 0.22972972972972974,
+      "eval_PRM Precision": 0.8125,
+      "eval_PRM Recall": 0.3132530120481928,
+      "eval_PRM Specificty": 0.7391304347826086,
+      "eval_loss": 0.8559126257896423,
+      "eval_runtime": 20.8278,
+      "eval_samples_per_second": 2.113,
+      "eval_steps_per_second": 0.144,
+      "step": 550
+    },
+    {
+      "epoch": 0.07133378645175907,
+      "grad_norm": 0.18439949787344748,
+      "learning_rate": 8.910090556274256e-07,
+      "loss": 1.2147,
+      "step": 551
+    },
+    {
+      "epoch": 0.07146324885911254,
+      "grad_norm": 0.1712580730880845,
+      "learning_rate": 8.926261319534282e-07,
+      "loss": 1.1318,
+      "step": 552
+    },
+    {
+      "epoch": 0.071592711266466,
+      "grad_norm": 0.21128305636090958,
+      "learning_rate": 8.942432082794309e-07,
+      "loss": 1.3673,
+      "step": 553
+    },
+    {
+      "epoch": 0.07172217367381946,
+      "grad_norm": 0.1613265149710803,
+      "learning_rate": 8.958602846054334e-07,
+      "loss": 1.1743,
+      "step": 554
+    },
+    {
+      "epoch": 0.07185163608117293,
+      "grad_norm": 0.1457513959689151,
+      "learning_rate": 8.974773609314361e-07,
+      "loss": 1.14,
+      "step": 555
+    },
+    {
+      "epoch": 0.07185163608117293,
+      "eval_PRM Accuracy": 0.41509433962264153,
+      "eval_PRM F1": 0.45614035087719296,
+      "eval_PRM F1 AUC": 0.5479308538501834,
+      "eval_PRM F1 Neg": 0.3673469387755102,
+      "eval_PRM NPV": 0.24,
+      "eval_PRM Precision": 0.8387096774193549,
+      "eval_PRM Recall": 0.3132530120481928,
+      "eval_PRM Specificty": 0.782608695652174,
+      "eval_loss": 0.8570667505264282,
+      "eval_runtime": 19.5877,
+      "eval_samples_per_second": 2.246,
+      "eval_steps_per_second": 0.153,
+      "step": 555
+    },
+    {
+      "epoch": 0.0719810984885264,
+      "grad_norm": 0.1847152662226901,
+      "learning_rate": 8.990944372574387e-07,
+      "loss": 1.2263,
+      "step": 556
+    },
+    {
+      "epoch": 0.07211056089587986,
+      "grad_norm": 0.1155253478010054,
+      "learning_rate": 9.007115135834413e-07,
+      "loss": 1.0538,
+      "step": 557
+    },
+    {
+      "epoch": 0.07224002330323333,
+      "grad_norm": 0.1385732996983949,
+      "learning_rate": 9.023285899094438e-07,
+      "loss": 1.0852,
+      "step": 558
+    },
+    {
+      "epoch": 0.0723694857105868,
+      "grad_norm": 0.18683493876719523,
+      "learning_rate": 9.039456662354464e-07,
+      "loss": 1.2081,
+      "step": 559
+    },
+    {
+      "epoch": 0.07249894811794025,
+      "grad_norm": 0.155354811949633,
+      "learning_rate": 9.055627425614489e-07,
+      "loss": 1.2561,
+      "step": 560
+    },
+    {
+      "epoch": 0.07249894811794025,
+      "eval_PRM Accuracy": 0.4056603773584906,
+      "eval_PRM F1": 0.45217391304347826,
+      "eval_PRM F1 AUC": 0.5261917234154007,
+      "eval_PRM F1 Neg": 0.35051546391752575,
+      "eval_PRM NPV": 0.22972972972972974,
+      "eval_PRM Precision": 0.8125,
+      "eval_PRM Recall": 0.3132530120481928,
+      "eval_PRM Specificty": 0.7391304347826086,
+      "eval_loss": 0.8552024364471436,
+      "eval_runtime": 20.3271,
+      "eval_samples_per_second": 2.165,
+      "eval_steps_per_second": 0.148,
+      "step": 560
+    },
+    {
+      "epoch": 0.07262841052529372,
+      "grad_norm": 0.1892510200193733,
+      "learning_rate": 9.071798188874516e-07,
+      "loss": 1.2211,
+      "step": 561
+    },
+    {
+      "epoch": 0.07275787293264718,
+      "grad_norm": 0.21262039296717195,
+      "learning_rate": 9.087968952134542e-07,
+      "loss": 1.2981,
+      "step": 562
+    },
+    {
+      "epoch": 0.07288733534000065,
+      "grad_norm": 0.17308827453310116,
+      "learning_rate": 9.104139715394568e-07,
+      "loss": 1.2527,
+      "step": 563
+    },
+    {
+      "epoch": 0.07301679774735412,
+      "grad_norm": 0.16906490926984974,
+      "learning_rate": 9.120310478654593e-07,
+      "loss": 1.1633,
+      "step": 564
+    },
+    {
+      "epoch": 0.07314626015470757,
+      "grad_norm": 0.11988652249069812,
+      "learning_rate": 9.136481241914618e-07,
+      "loss": 1.0525,
+      "step": 565
+    },
+    {
+      "epoch": 0.07314626015470757,
+      "eval_PRM Accuracy": 0.41509433962264153,
+      "eval_PRM F1": 0.45614035087719296,
+      "eval_PRM F1 AUC": 0.5479308538501834,
+      "eval_PRM F1 Neg": 0.3673469387755102,
+      "eval_PRM NPV": 0.24,
+      "eval_PRM Precision": 0.8387096774193549,
+      "eval_PRM Recall": 0.3132530120481928,
+      "eval_PRM Specificty": 0.782608695652174,
+      "eval_loss": 0.8514737486839294,
+      "eval_runtime": 20.3487,
+      "eval_samples_per_second": 2.162,
+      "eval_steps_per_second": 0.147,
+      "step": 565
+    },
+    {
+      "epoch": 0.07327572256206104,
+      "grad_norm": 0.15296886676853855,
+      "learning_rate": 9.152652005174645e-07,
+      "loss": 1.1919,
+      "step": 566
+    },
+    {
+      "epoch": 0.0734051849694145,
+      "grad_norm": 0.2066753964853689,
+      "learning_rate": 9.168822768434671e-07,
+      "loss": 1.2742,
+      "step": 567
+    },
+    {
+      "epoch": 0.07353464737676797,
+      "grad_norm": 0.1497077319308681,
+      "learning_rate": 9.184993531694696e-07,
+      "loss": 1.0746,
+      "step": 568
+    },
+    {
+      "epoch": 0.07366410978412144,
+      "grad_norm": 0.2173177327155543,
+      "learning_rate": 9.201164294954723e-07,
+      "loss": 1.3013,
+      "step": 569
+    },
+    {
+      "epoch": 0.07379357219147491,
+      "grad_norm": 0.18223242906680462,
+      "learning_rate": 9.217335058214749e-07,
+      "loss": 1.1649,
+      "step": 570
+    },
+    {
+      "epoch": 0.07379357219147491,
+      "eval_PRM Accuracy": 0.41509433962264153,
+      "eval_PRM F1": 0.45614035087719296,
+      "eval_PRM F1 AUC": 0.5479308538501834,
+      "eval_PRM F1 Neg": 0.3673469387755102,
+      "eval_PRM NPV": 0.24,
+      "eval_PRM Precision": 0.8387096774193549,
+      "eval_PRM Recall": 0.3132530120481928,
+      "eval_PRM Specificty": 0.782608695652174,
+      "eval_loss": 0.8512961864471436,
+      "eval_runtime": 20.6008,
+      "eval_samples_per_second": 2.136,
+      "eval_steps_per_second": 0.146,
+      "step": 570
+    },
+    {
+      "epoch": 0.07392303459882836,
+      "grad_norm": 0.22343647381701554,
+      "learning_rate": 9.233505821474773e-07,
+      "loss": 1.322,
+      "step": 571
+    },
+    {
+      "epoch": 0.07405249700618183,
+      "grad_norm": 0.16184415259431117,
+      "learning_rate": 9.2496765847348e-07,
+      "loss": 1.1327,
+      "step": 572
+    },
+    {
+      "epoch": 0.0741819594135353,
+      "grad_norm": 0.15616613926206077,
+      "learning_rate": 9.265847347994826e-07,
+      "loss": 1.2274,
+      "step": 573
+    },
+    {
+      "epoch": 0.07431142182088876,
+      "grad_norm": 0.14240129436834736,
+      "learning_rate": 9.282018111254852e-07,
+      "loss": 1.2004,
+      "step": 574
+    },
+    {
+      "epoch": 0.07444088422824223,
+      "grad_norm": 0.17199606502566842,
+      "learning_rate": 9.298188874514878e-07,
+      "loss": 1.2162,
+      "step": 575
+    },
+    {
+      "epoch": 0.07444088422824223,
+      "eval_PRM Accuracy": 0.42452830188679247,
+      "eval_PRM F1": 0.46956521739130436,
+      "eval_PRM F1 AUC": 0.5539549502357256,
+      "eval_PRM F1 Neg": 0.3711340206185567,
+      "eval_PRM NPV": 0.24324324324324326,
+      "eval_PRM Precision": 0.84375,
+      "eval_PRM Recall": 0.3253012048192771,
+      "eval_PRM Specificty": 0.782608695652174,
+      "eval_loss": 0.8500532507896423,
+      "eval_runtime": 20.1227,
+      "eval_samples_per_second": 2.187,
+      "eval_steps_per_second": 0.149,
+      "step": 575
+    },
+    {
+      "epoch": 0.07457034663559568,
+      "grad_norm": 0.18013920228667082,
+      "learning_rate": 9.314359637774904e-07,
+      "loss": 1.1343,
+      "step": 576
+    },
+    {
+      "epoch": 0.07469980904294915,
+      "grad_norm": 0.22027558714483902,
+      "learning_rate": 9.330530401034929e-07,
+      "loss": 1.3381,
+      "step": 577
+    },
+    {
+      "epoch": 0.07482927145030262,
+      "grad_norm": 0.22043454340086804,
+      "learning_rate": 9.346701164294955e-07,
+      "loss": 1.2676,
+      "step": 578
+    },
+    {
+      "epoch": 0.07495873385765608,
+      "grad_norm": 0.17957710018701944,
+      "learning_rate": 9.362871927554981e-07,
+      "loss": 1.2222,
+      "step": 579
+    },
+    {
+      "epoch": 0.07508819626500955,
+      "grad_norm": 0.18252652262030405,
+      "learning_rate": 9.379042690815007e-07,
+      "loss": 1.1116,
+      "step": 580
+    },
+    {
+      "epoch": 0.07508819626500955,
+      "eval_PRM Accuracy": 0.42452830188679247,
+      "eval_PRM F1": 0.46956521739130436,
+      "eval_PRM F1 AUC": 0.5539549502357256,
+      "eval_PRM F1 Neg": 0.3711340206185567,
+      "eval_PRM NPV": 0.24324324324324326,
+      "eval_PRM Precision": 0.84375,
+      "eval_PRM Recall": 0.3253012048192771,
+      "eval_PRM Specificty": 0.782608695652174,
+      "eval_loss": 0.8490766882896423,
+      "eval_runtime": 19.6466,
+      "eval_samples_per_second": 2.24,
+      "eval_steps_per_second": 0.153,
+      "step": 580
+    },
+    {
+      "epoch": 0.07521765867236302,
+      "grad_norm": 0.20980209175127562,
+      "learning_rate": 9.395213454075033e-07,
+      "loss": 1.2555,
+      "step": 581
+    },
+    {
+      "epoch": 0.07534712107971647,
+      "grad_norm": 0.17727989607097341,
+      "learning_rate": 9.41138421733506e-07,
+      "loss": 1.2201,
+      "step": 582
+    },
+    {
+      "epoch": 0.07547658348706994,
+      "grad_norm": 0.14146353776901985,
+      "learning_rate": 9.427554980595084e-07,
+      "loss": 1.177,
+      "step": 583
+    },
+    {
+      "epoch": 0.07560604589442341,
+      "grad_norm": 0.1828278252702041,
+      "learning_rate": 9.44372574385511e-07,
+      "loss": 1.1968,
+      "step": 584
+    },
+    {
+      "epoch": 0.07573550830177687,
+      "grad_norm": 0.20763803274849194,
+      "learning_rate": 9.459896507115137e-07,
+      "loss": 1.3428,
+      "step": 585
+    },
+    {
+      "epoch": 0.07573550830177687,
+      "eval_PRM Accuracy": 0.42452830188679247,
+      "eval_PRM F1": 0.46956521739130436,
+      "eval_PRM F1 AUC": 0.5539549502357256,
+      "eval_PRM F1 Neg": 0.3711340206185567,
+      "eval_PRM NPV": 0.24324324324324326,
+      "eval_PRM Precision": 0.84375,
+      "eval_PRM Recall": 0.3253012048192771,
+      "eval_PRM Specificty": 0.782608695652174,
+      "eval_loss": 0.8464133739471436,
+      "eval_runtime": 20.5487,
+      "eval_samples_per_second": 2.141,
+      "eval_steps_per_second": 0.146,
+      "step": 585
+    },
+    {
+      "epoch": 0.07586497070913034,
+      "grad_norm": 0.18403077218642375,
+      "learning_rate": 9.476067270375162e-07,
+      "loss": 1.2616,
+      "step": 586
+    },
+    {
+      "epoch": 0.0759944331164838,
+      "grad_norm": 0.1745292206717775,
+      "learning_rate": 9.492238033635188e-07,
+      "loss": 1.2581,
+      "step": 587
+    },
+    {
+      "epoch": 0.07612389552383726,
+      "grad_norm": 0.21287042368677034,
+      "learning_rate": 9.508408796895215e-07,
+      "loss": 1.2311,
+      "step": 588
+    },
+    {
+      "epoch": 0.07625335793119073,
+      "grad_norm": 0.18562649721453459,
+      "learning_rate": 9.52457956015524e-07,
+      "loss": 1.2653,
+      "step": 589
+    },
+    {
+      "epoch": 0.0763828203385442,
+      "grad_norm": 0.15088666061786474,
+      "learning_rate": 9.540750323415266e-07,
+      "loss": 1.2665,
+      "step": 590
+    },
+    {
+      "epoch": 0.0763828203385442,
+      "eval_PRM Accuracy": 0.41509433962264153,
+      "eval_PRM F1": 0.46551724137931033,
+      "eval_PRM F1 AUC": 0.5322158198009429,
+      "eval_PRM F1 Neg": 0.3541666666666667,
+      "eval_PRM NPV": 0.2328767123287671,
+      "eval_PRM Precision": 0.8181818181818182,
+      "eval_PRM Recall": 0.3253012048192771,
+      "eval_PRM Specificty": 0.7391304347826086,
+      "eval_loss": 0.8440163135528564,
+      "eval_runtime": 20.7677,
+      "eval_samples_per_second": 2.119,
+      "eval_steps_per_second": 0.144,
+      "step": 590
+    },
+    {
+      "epoch": 0.07651228274589766,
+      "grad_norm": 0.19810884025508352,
+      "learning_rate": 9.55692108667529e-07,
+      "loss": 1.2441,
+      "step": 591
+    },
+    {
+      "epoch": 0.07664174515325113,
+      "grad_norm": 0.1538984229267891,
+      "learning_rate": 9.573091849935317e-07,
+      "loss": 1.0762,
+      "step": 592
+    },
+    {
+      "epoch": 0.07677120756060458,
+      "grad_norm": 0.227282697618542,
+      "learning_rate": 9.589262613195344e-07,
+      "loss": 1.2837,
+      "step": 593
+    },
+    {
+      "epoch": 0.07690066996795805,
+      "grad_norm": 0.14359725632220752,
+      "learning_rate": 9.60543337645537e-07,
+      "loss": 1.1511,
+      "step": 594
+    },
+    {
+      "epoch": 0.07703013237531152,
+      "grad_norm": 0.2111230098409188,
+      "learning_rate": 9.621604139715395e-07,
+      "loss": 1.2981,
+      "step": 595
+    },
+    {
+      "epoch": 0.07703013237531152,
+      "eval_PRM Accuracy": 0.41509433962264153,
+      "eval_PRM F1": 0.45614035087719296,
+      "eval_PRM F1 AUC": 0.5479308538501834,
+      "eval_PRM F1 Neg": 0.3673469387755102,
+      "eval_PRM NPV": 0.24,
+      "eval_PRM Precision": 0.8387096774193549,
+      "eval_PRM Recall": 0.3132530120481928,
+      "eval_PRM Specificty": 0.782608695652174,
+      "eval_loss": 0.8430397510528564,
+      "eval_runtime": 19.6605,
+      "eval_samples_per_second": 2.238,
+      "eval_steps_per_second": 0.153,
+      "step": 595
+    },
+    {
+      "epoch": 0.07715959478266499,
+      "grad_norm": 0.1477765175861278,
+      "learning_rate": 9.63777490297542e-07,
+      "loss": 1.0883,
+      "step": 596
+    },
+    {
+      "epoch": 0.07728905719001845,
+      "grad_norm": 0.1652592793015199,
+      "learning_rate": 9.653945666235447e-07,
+      "loss": 1.1495,
+      "step": 597
+    },
+    {
+      "epoch": 0.07741851959737191,
+      "grad_norm": 0.2434964293895327,
+      "learning_rate": 9.670116429495473e-07,
+      "loss": 1.2626,
+      "step": 598
+    },
+    {
+      "epoch": 0.07754798200472537,
+      "grad_norm": 0.17775056668002065,
+      "learning_rate": 9.686287192755498e-07,
+      "loss": 1.0856,
+      "step": 599
+    },
+    {
+      "epoch": 0.07767744441207884,
+      "grad_norm": 0.18412335237074828,
+      "learning_rate": 9.702457956015525e-07,
+      "loss": 1.1545,
+      "step": 600
+    },
+    {
+      "epoch": 0.07767744441207884,
+      "eval_PRM Accuracy": 0.41509433962264153,
+      "eval_PRM F1": 0.46551724137931033,
+      "eval_PRM F1 AUC": 0.5322158198009429,
+      "eval_PRM F1 Neg": 0.3541666666666667,
+      "eval_PRM NPV": 0.2328767123287671,
+      "eval_PRM Precision": 0.8181818181818182,
+      "eval_PRM Recall": 0.3253012048192771,
+      "eval_PRM Specificty": 0.7391304347826086,
+      "eval_loss": 0.8440163135528564,
+      "eval_runtime": 20.5536,
+      "eval_samples_per_second": 2.141,
+      "eval_steps_per_second": 0.146,
+      "step": 600
+    },
+    {
+      "epoch": 0.07780690681943231,
+      "grad_norm": 0.20451127740359148,
+      "learning_rate": 9.718628719275551e-07,
+      "loss": 1.2496,
+      "step": 601
+    },
+    {
+      "epoch": 0.07793636922678578,
+      "grad_norm": 0.134688938014961,
+      "learning_rate": 9.734799482535578e-07,
+      "loss": 1.1478,
+      "step": 602
+    },
+    {
+      "epoch": 0.07806583163413924,
+      "grad_norm": 0.18953121801844397,
+      "learning_rate": 9.750970245795603e-07,
+      "loss": 1.2623,
+      "step": 603
+    },
+    {
+      "epoch": 0.0781952940414927,
+      "grad_norm": 0.20055597187343607,
+      "learning_rate": 9.767141009055627e-07,
+      "loss": 1.2225,
+      "step": 604
+    },
+    {
+      "epoch": 0.07832475644884616,
+      "grad_norm": 0.1744283042399751,
+      "learning_rate": 9.783311772315654e-07,
+      "loss": 1.1193,
+      "step": 605
+    },
+    {
+      "epoch": 0.07832475644884616,
+      "eval_PRM Accuracy": 0.41509433962264153,
+      "eval_PRM F1": 0.46551724137931033,
+      "eval_PRM F1 AUC": 0.5322158198009429,
+      "eval_PRM F1 Neg": 0.3541666666666667,
+      "eval_PRM NPV": 0.2328767123287671,
+      "eval_PRM Precision": 0.8181818181818182,
+      "eval_PRM Recall": 0.3253012048192771,
+      "eval_PRM Specificty": 0.7391304347826086,
+      "eval_loss": 0.8406427502632141,
+      "eval_runtime": 20.005,
+      "eval_samples_per_second": 2.199,
+      "eval_steps_per_second": 0.15,
+      "step": 605
+    },
+    {
+      "epoch": 0.07845421885619963,
+      "grad_norm": 0.15709069633705608,
+      "learning_rate": 9.79948253557568e-07,
+      "loss": 1.1813,
+      "step": 606
+    },
+    {
+      "epoch": 0.0785836812635531,
+      "grad_norm": 0.12198167666990041,
+      "learning_rate": 9.815653298835705e-07,
+      "loss": 1.0535,
+      "step": 607
+    },
+    {
+      "epoch": 0.07871314367090657,
+      "grad_norm": 0.16858140050121193,
+      "learning_rate": 9.831824062095732e-07,
+      "loss": 1.0499,
+      "step": 608
+    },
+    {
+      "epoch": 0.07884260607826002,
+      "grad_norm": 0.150020677991825,
+      "learning_rate": 9.847994825355757e-07,
+      "loss": 1.2124,
+      "step": 609
+    },
+    {
+      "epoch": 0.07897206848561349,
+      "grad_norm": 0.1305125673559483,
+      "learning_rate": 9.864165588615783e-07,
+      "loss": 1.1151,
+      "step": 610
+    },
+    {
+      "epoch": 0.07897206848561349,
+      "eval_PRM Accuracy": 0.41509433962264153,
+      "eval_PRM F1": 0.46551724137931033,
+      "eval_PRM F1 AUC": 0.5322158198009429,
+      "eval_PRM F1 Neg": 0.3541666666666667,
+      "eval_PRM NPV": 0.2328767123287671,
+      "eval_PRM Precision": 0.8181818181818182,
+      "eval_PRM Recall": 0.3253012048192771,
+      "eval_PRM Specificty": 0.7391304347826086,
+      "eval_loss": 0.8380681872367859,
+      "eval_runtime": 19.4751,
+      "eval_samples_per_second": 2.259,
+      "eval_steps_per_second": 0.154,
+      "step": 610
+    },
+    {
+      "epoch": 0.07910153089296695,
+      "grad_norm": 0.19660200461069727,
+      "learning_rate": 9.88033635187581e-07,
+      "loss": 1.1732,
+      "step": 611
+    },
+    {
+      "epoch": 0.07923099330032042,
+      "grad_norm": 0.19550512634253983,
+      "learning_rate": 9.896507115135835e-07,
+      "loss": 1.2325,
+      "step": 612
+    },
+    {
+      "epoch": 0.07936045570767389,
+      "grad_norm": 0.20223104609724346,
+      "learning_rate": 9.912677878395861e-07,
+      "loss": 1.2853,
+      "step": 613
+    },
+    {
+      "epoch": 0.07948991811502736,
+      "grad_norm": 0.16229259145988997,
+      "learning_rate": 9.928848641655888e-07,
+      "loss": 1.0121,
+      "step": 614
+    },
+    {
+      "epoch": 0.07961938052238081,
+      "grad_norm": 0.17642818643864705,
+      "learning_rate": 9.945019404915913e-07,
+      "loss": 1.0804,
+      "step": 615
+    },
+    {
+      "epoch": 0.07961938052238081,
+      "eval_PRM Accuracy": 0.41509433962264153,
+      "eval_PRM F1": 0.46551724137931033,
+      "eval_PRM F1 AUC": 0.5322158198009429,
+      "eval_PRM F1 Neg": 0.3541666666666667,
+      "eval_PRM NPV": 0.2328767123287671,
+      "eval_PRM Precision": 0.8181818181818182,
+      "eval_PRM Recall": 0.3253012048192771,
+      "eval_PRM Specificty": 0.7391304347826086,
+      "eval_loss": 0.8354048132896423,
+      "eval_runtime": 20.224,
+      "eval_samples_per_second": 2.176,
+      "eval_steps_per_second": 0.148,
+      "step": 615
+    },
+    {
+      "epoch": 0.07974884292973428,
+      "grad_norm": 0.15158698152541275,
+      "learning_rate": 9.961190168175937e-07,
+      "loss": 1.1259,
+      "step": 616
+    },
+    {
+      "epoch": 0.07987830533708774,
+      "grad_norm": 0.21821063654265963,
+      "learning_rate": 9.977360931435964e-07,
+      "loss": 1.201,
+      "step": 617
+    },
+    {
+      "epoch": 0.08000776774444121,
+      "grad_norm": 0.22009196919431773,
+      "learning_rate": 9.99353169469599e-07,
+      "loss": 1.199,
+      "step": 618
+    },
+    {
+      "epoch": 0.08013723015179468,
+      "grad_norm": 0.19385089678883013,
+      "learning_rate": 1.0009702457956017e-06,
+      "loss": 1.1692,
+      "step": 619
+    },
+    {
+      "epoch": 0.08026669255914813,
+      "grad_norm": 0.1920087048864159,
+      "learning_rate": 1.0025873221216042e-06,
+      "loss": 1.3226,
+      "step": 620
+    },
+    {
+      "epoch": 0.08026669255914813,
+      "eval_PRM Accuracy": 0.41509433962264153,
+      "eval_PRM F1": 0.46551724137931033,
+      "eval_PRM F1 AUC": 0.5322158198009429,
+      "eval_PRM F1 Neg": 0.3541666666666667,
+      "eval_PRM NPV": 0.2328767123287671,
+      "eval_PRM Precision": 0.8181818181818182,
+      "eval_PRM Recall": 0.3253012048192771,
+      "eval_PRM Specificty": 0.7391304347826086,
+      "eval_loss": 0.8352272510528564,
+      "eval_runtime": 20.905,
+      "eval_samples_per_second": 2.105,
+      "eval_steps_per_second": 0.144,
+      "step": 620
+    },
+    {
+      "epoch": 0.0803961549665016,
+      "grad_norm": 0.2410325561559353,
+      "learning_rate": 1.0042043984476069e-06,
+      "loss": 1.2606,
+      "step": 621
+    },
+    {
+      "epoch": 0.08052561737385507,
+      "grad_norm": 0.21663101423396477,
+      "learning_rate": 1.0058214747736093e-06,
+      "loss": 1.2263,
+      "step": 622
+    },
+    {
+      "epoch": 0.08065507978120853,
+      "grad_norm": 0.19022487103327237,
+      "learning_rate": 1.007438551099612e-06,
+      "loss": 1.1016,
+      "step": 623
+    },
+    {
+      "epoch": 0.080784542188562,
+      "grad_norm": 0.1859061257124025,
+      "learning_rate": 1.0090556274256147e-06,
+      "loss": 1.134,
+      "step": 624
+    },
+    {
+      "epoch": 0.08091400459591547,
+      "grad_norm": 0.1750427098784179,
+      "learning_rate": 1.0106727037516171e-06,
+      "loss": 1.1778,
+      "step": 625
+    },
+    {
+      "epoch": 0.08091400459591547,
+      "eval_PRM Accuracy": 0.41509433962264153,
+      "eval_PRM F1": 0.46551724137931033,
+      "eval_PRM F1 AUC": 0.5322158198009429,
+      "eval_PRM F1 Neg": 0.3541666666666667,
+      "eval_PRM NPV": 0.2328767123287671,
+      "eval_PRM Precision": 0.8181818181818182,
+      "eval_PRM Recall": 0.3253012048192771,
+      "eval_PRM Specificty": 0.7391304347826086,
+      "eval_loss": 0.833984375,
+      "eval_runtime": 19.9809,
+      "eval_samples_per_second": 2.202,
+      "eval_steps_per_second": 0.15,
+      "step": 625
+    },
+    {
+      "epoch": 0.08104346700326892,
+      "grad_norm": 0.20373155644029953,
+      "learning_rate": 1.0122897800776198e-06,
+      "loss": 1.3265,
+      "step": 626
+    },
+    {
+      "epoch": 0.08117292941062239,
+      "grad_norm": 0.20544261412052614,
+      "learning_rate": 1.0139068564036225e-06,
+      "loss": 1.2438,
+      "step": 627
+    },
+    {
+      "epoch": 0.08130239181797586,
+      "grad_norm": 0.25213857021720304,
+      "learning_rate": 1.015523932729625e-06,
+      "loss": 1.3451,
+      "step": 628
+    },
+    {
+      "epoch": 0.08143185422532932,
+      "grad_norm": 0.1852596172036974,
+      "learning_rate": 1.0171410090556274e-06,
+      "loss": 1.2488,
+      "step": 629
+    },
+    {
+      "epoch": 0.08156131663268279,
+      "grad_norm": 0.18492490755573396,
+      "learning_rate": 1.01875808538163e-06,
+      "loss": 1.0752,
+      "step": 630
+    },
+    {
+      "epoch": 0.08156131663268279,
+      "eval_PRM Accuracy": 0.41509433962264153,
+      "eval_PRM F1": 0.46551724137931033,
+      "eval_PRM F1 AUC": 0.5322158198009429,
+      "eval_PRM F1 Neg": 0.3541666666666667,
+      "eval_PRM NPV": 0.2328767123287671,
+      "eval_PRM Precision": 0.8181818181818182,
+      "eval_PRM Recall": 0.3253012048192771,
+      "eval_PRM Specificty": 0.7391304347826086,
+      "eval_loss": 0.83203125,
+      "eval_runtime": 20.6205,
+      "eval_samples_per_second": 2.134,
+      "eval_steps_per_second": 0.145,
+      "step": 630
+    },
+    {
+      "epoch": 0.08169077904003624,
+      "grad_norm": 0.215974601433201,
+      "learning_rate": 1.0203751617076327e-06,
+      "loss": 1.2628,
+      "step": 631
+    },
+    {
+      "epoch": 0.08182024144738971,
+      "grad_norm": 0.2082029106988089,
+      "learning_rate": 1.0219922380336354e-06,
+      "loss": 1.2837,
+      "step": 632
+    },
+    {
+      "epoch": 0.08194970385474318,
+      "grad_norm": 0.24257661223325547,
+      "learning_rate": 1.0236093143596378e-06,
+      "loss": 1.2617,
+      "step": 633
+    },
+    {
+      "epoch": 0.08207916626209665,
+      "grad_norm": 0.19125457895224893,
+      "learning_rate": 1.0252263906856405e-06,
+      "loss": 1.1785,
+      "step": 634
+    },
+    {
+      "epoch": 0.08220862866945011,
+      "grad_norm": 0.18324450114320986,
+      "learning_rate": 1.026843467011643e-06,
+      "loss": 1.1764,
+      "step": 635
+    },
+    {
+      "epoch": 0.08220862866945011,
+      "eval_PRM Accuracy": 0.41509433962264153,
+      "eval_PRM F1": 0.46551724137931033,
+      "eval_PRM F1 AUC": 0.5322158198009429,
+      "eval_PRM F1 Neg": 0.3541666666666667,
+      "eval_PRM NPV": 0.2328767123287671,
+      "eval_PRM Precision": 0.8181818181818182,
+      "eval_PRM Recall": 0.3253012048192771,
+      "eval_PRM Specificty": 0.7391304347826086,
+      "eval_loss": 0.8306108117103577,
+      "eval_runtime": 19.7582,
+      "eval_samples_per_second": 2.227,
+      "eval_steps_per_second": 0.152,
+      "step": 635
+    },
+    {
+      "epoch": 0.08233809107680358,
+      "grad_norm": 0.23910022887175347,
+      "learning_rate": 1.0284605433376456e-06,
+      "loss": 1.2593,
+      "step": 636
+    },
+    {
+      "epoch": 0.08246755348415703,
+      "grad_norm": 0.17525638271870733,
+      "learning_rate": 1.030077619663648e-06,
+      "loss": 1.117,
+      "step": 637
+    },
+    {
+      "epoch": 0.0825970158915105,
+      "grad_norm": 0.18554758396414467,
+      "learning_rate": 1.0316946959896508e-06,
+      "loss": 1.1935,
+      "step": 638
+    },
+    {
+      "epoch": 0.08272647829886397,
+      "grad_norm": 0.2242274751047819,
+      "learning_rate": 1.0333117723156534e-06,
+      "loss": 1.2065,
+      "step": 639
+    },
+    {
+      "epoch": 0.08285594070621743,
+      "grad_norm": 0.18594755502473304,
+      "learning_rate": 1.0349288486416561e-06,
+      "loss": 1.0703,
+      "step": 640
+    },
+    {
+      "epoch": 0.08285594070621743,
+      "eval_PRM Accuracy": 0.42452830188679247,
+      "eval_PRM F1": 0.47863247863247865,
+      "eval_PRM F1 AUC": 0.538239916186485,
+      "eval_PRM F1 Neg": 0.35789473684210527,
+      "eval_PRM NPV": 0.2361111111111111,
+      "eval_PRM Precision": 0.8235294117647058,
+      "eval_PRM Recall": 0.3373493975903614,
+      "eval_PRM Specificty": 0.7391304347826086,
+      "eval_loss": 0.8270596861839294,
+      "eval_runtime": 19.9843,
+      "eval_samples_per_second": 2.202,
+      "eval_steps_per_second": 0.15,
+      "step": 640
+    },
+    {
+      "epoch": 0.0829854031135709,
+      "grad_norm": 0.15533079014938828,
+      "learning_rate": 1.0365459249676586e-06,
+      "loss": 1.1401,
+      "step": 641
+    },
+    {
+      "epoch": 0.08311486552092436,
+      "grad_norm": 0.16403804886050868,
+      "learning_rate": 1.038163001293661e-06,
+      "loss": 1.1161,
+      "step": 642
+    },
+    {
+      "epoch": 0.08324432792827782,
+      "grad_norm": 0.22604286299094503,
+      "learning_rate": 1.0397800776196637e-06,
+      "loss": 1.2501,
+      "step": 643
+    },
+    {
+      "epoch": 0.08337379033563129,
+      "grad_norm": 0.2725461541421229,
+      "learning_rate": 1.0413971539456664e-06,
+      "loss": 1.3757,
+      "step": 644
+    },
+    {
+      "epoch": 0.08350325274298476,
+      "grad_norm": 0.18481704246876085,
+      "learning_rate": 1.0430142302716688e-06,
+      "loss": 1.2492,
+      "step": 645
+    },
+    {
+      "epoch": 0.08350325274298476,
+      "eval_PRM Accuracy": 0.41509433962264153,
+      "eval_PRM F1": 0.46551724137931033,
+      "eval_PRM F1 AUC": 0.5322158198009429,
+      "eval_PRM F1 Neg": 0.3541666666666667,
+      "eval_PRM NPV": 0.2328767123287671,
+      "eval_PRM Precision": 0.8181818181818182,
+      "eval_PRM Recall": 0.3253012048192771,
+      "eval_PRM Specificty": 0.7391304347826086,
+      "eval_loss": 0.8249289989471436,
+      "eval_runtime": 20.1134,
+      "eval_samples_per_second": 2.188,
+      "eval_steps_per_second": 0.149,
+      "step": 645
+    },
+    {
+      "epoch": 0.08363271515033822,
+      "grad_norm": 0.15664668168743495,
+      "learning_rate": 1.0446313065976715e-06,
+      "loss": 1.0685,
+      "step": 646
+    },
+    {
+      "epoch": 0.08376217755769169,
+      "grad_norm": 0.19314453324977945,
+      "learning_rate": 1.0462483829236742e-06,
+      "loss": 1.2271,
+      "step": 647
+    },
+    {
+      "epoch": 0.08389163996504514,
+      "grad_norm": 0.22042946699958396,
+      "learning_rate": 1.0478654592496766e-06,
+      "loss": 1.1367,
+      "step": 648
+    },
+    {
+      "epoch": 0.08402110237239861,
+      "grad_norm": 0.1592804760009837,
+      "learning_rate": 1.0494825355756793e-06,
+      "loss": 1.1088,
+      "step": 649
+    },
+    {
+      "epoch": 0.08415056477975208,
+      "grad_norm": 0.21381685893169836,
+      "learning_rate": 1.0510996119016818e-06,
+      "loss": 1.181,
+      "step": 650
+    },
+    {
+      "epoch": 0.08415056477975208,
+      "eval_PRM Accuracy": 0.42452830188679247,
+      "eval_PRM F1": 0.47863247863247865,
+      "eval_PRM F1 AUC": 0.538239916186485,
+      "eval_PRM F1 Neg": 0.35789473684210527,
+      "eval_PRM NPV": 0.2361111111111111,
+      "eval_PRM Precision": 0.8235294117647058,
+      "eval_PRM Recall": 0.3373493975903614,
+      "eval_PRM Specificty": 0.7391304347826086,
+      "eval_loss": 0.8235973119735718,
+      "eval_runtime": 20.4724,
+      "eval_samples_per_second": 2.149,
+      "eval_steps_per_second": 0.147,
+      "step": 650
+    },
+    {
+      "epoch": 0.08428002718710555,
+      "grad_norm": 0.19722070171612469,
+      "learning_rate": 1.0527166882276844e-06,
+      "loss": 1.1498,
+      "step": 651
+    },
+    {
+      "epoch": 0.08440948959445901,
+      "grad_norm": 0.18638623948992936,
+      "learning_rate": 1.0543337645536871e-06,
+      "loss": 1.14,
+      "step": 652
+    },
+    {
+      "epoch": 0.08453895200181247,
+      "grad_norm": 0.17361589799056665,
+      "learning_rate": 1.0559508408796896e-06,
+      "loss": 1.0629,
+      "step": 653
+    },
+    {
+      "epoch": 0.08466841440916593,
+      "grad_norm": 0.16058232555062704,
+      "learning_rate": 1.057567917205692e-06,
+      "loss": 1.0279,
+      "step": 654
+    },
+    {
+      "epoch": 0.0847978768165194,
+      "grad_norm": 0.2594110366502841,
+      "learning_rate": 1.0591849935316947e-06,
+      "loss": 1.1725,
+      "step": 655
+    },
+    {
+      "epoch": 0.0847978768165194,
+      "eval_PRM Accuracy": 0.41509433962264153,
+      "eval_PRM F1": 0.46551724137931033,
+      "eval_PRM F1 AUC": 0.5322158198009429,
+      "eval_PRM F1 Neg": 0.3541666666666667,
+      "eval_PRM NPV": 0.2328767123287671,
+      "eval_PRM Precision": 0.8181818181818182,
+      "eval_PRM Recall": 0.3253012048192771,
+      "eval_PRM Specificty": 0.7391304347826086,
+      "eval_loss": 0.8231534361839294,
+      "eval_runtime": 20.0464,
+      "eval_samples_per_second": 2.195,
+      "eval_steps_per_second": 0.15,
+      "step": 655
+    },
+    {
+      "epoch": 0.08492733922387287,
+      "grad_norm": 0.17828931812647264,
+      "learning_rate": 1.0608020698576974e-06,
+      "loss": 1.084,
+      "step": 656
+    },
+    {
+      "epoch": 0.08505680163122634,
+      "grad_norm": 0.2292364665592309,
+      "learning_rate": 1.0624191461837e-06,
+      "loss": 1.2529,
+      "step": 657
+    },
+    {
+      "epoch": 0.0851862640385798,
+      "grad_norm": 0.2535249597619893,
+      "learning_rate": 1.0640362225097025e-06,
+      "loss": 1.3447,
+      "step": 658
+    },
+    {
+      "epoch": 0.08531572644593326,
+      "grad_norm": 0.22242542454789774,
+      "learning_rate": 1.0656532988357052e-06,
+      "loss": 1.218,
+      "step": 659
+    },
+    {
+      "epoch": 0.08544518885328672,
+      "grad_norm": 0.20438897702859776,
+      "learning_rate": 1.0672703751617078e-06,
+      "loss": 1.0831,
+      "step": 660
+    },
+    {
+      "epoch": 0.08544518885328672,
+      "eval_PRM Accuracy": 0.4339622641509434,
+      "eval_PRM F1": 0.4915254237288136,
+      "eval_PRM F1 AUC": 0.5442640125720272,
+      "eval_PRM F1 Neg": 0.3617021276595745,
+      "eval_PRM NPV": 0.23943661971830985,
+      "eval_PRM Precision": 0.8285714285714286,
+      "eval_PRM Recall": 0.3493975903614458,
+      "eval_PRM Specificty": 0.7391304347826086,
+      "eval_loss": 0.8225319385528564,
+      "eval_runtime": 20.7566,
+      "eval_samples_per_second": 2.12,
+      "eval_steps_per_second": 0.145,
+      "step": 660
+    },
+    {
+      "epoch": 0.08557465126064019,
+      "grad_norm": 0.17690767226793896,
+      "learning_rate": 1.0688874514877103e-06,
+      "loss": 1.0247,
+      "step": 661
+    },
+    {
+      "epoch": 0.08570411366799366,
+      "grad_norm": 0.20176416990123527,
+      "learning_rate": 1.070504527813713e-06,
+      "loss": 1.184,
+      "step": 662
+    },
+    {
+      "epoch": 0.08583357607534713,
+      "grad_norm": 0.18262888880229672,
+      "learning_rate": 1.0721216041397154e-06,
+      "loss": 1.2097,
+      "step": 663
+    },
+    {
+      "epoch": 0.08596303848270058,
+      "grad_norm": 0.21244129132342518,
+      "learning_rate": 1.073738680465718e-06,
+      "loss": 1.2399,
+      "step": 664
+    },
+    {
+      "epoch": 0.08609250089005405,
+      "grad_norm": 0.19529942397470798,
+      "learning_rate": 1.0753557567917208e-06,
+      "loss": 1.2333,
+      "step": 665
+    },
+    {
+      "epoch": 0.08609250089005405,
+      "eval_PRM Accuracy": 0.4339622641509434,
+      "eval_PRM F1": 0.4915254237288136,
+      "eval_PRM F1 AUC": 0.5442640125720272,
+      "eval_PRM F1 Neg": 0.3617021276595745,
+      "eval_PRM NPV": 0.23943661971830985,
+      "eval_PRM Precision": 0.8285714285714286,
+      "eval_PRM Recall": 0.3493975903614458,
+      "eval_PRM Specificty": 0.7391304347826086,
+      "eval_loss": 0.8169389367103577,
+      "eval_runtime": 20.8415,
+      "eval_samples_per_second": 2.111,
+      "eval_steps_per_second": 0.144,
+      "step": 665
+    },
+    {
+      "epoch": 0.08622196329740751,
+      "grad_norm": 0.23585250344700395,
+      "learning_rate": 1.0769728331177232e-06,
+      "loss": 1.2513,
+      "step": 666
+    },
+    {
+      "epoch": 0.08635142570476098,
+      "grad_norm": 0.14952656684145715,
+      "learning_rate": 1.0785899094437257e-06,
+      "loss": 1.0811,
+      "step": 667
+    },
+    {
+      "epoch": 0.08648088811211445,
+      "grad_norm": 0.20062848072945358,
+      "learning_rate": 1.0802069857697284e-06,
+      "loss": 1.201,
+      "step": 668
+    },
+    {
+      "epoch": 0.08661035051946792,
+      "grad_norm": 0.22367060793159224,
+      "learning_rate": 1.081824062095731e-06,
+      "loss": 1.2006,
+      "step": 669
+    },
+    {
+      "epoch": 0.08673981292682137,
+      "grad_norm": 0.2040787667759277,
+      "learning_rate": 1.0834411384217337e-06,
+      "loss": 1.2313,
+      "step": 670
+    },
+    {
+      "epoch": 0.08673981292682137,
+      "eval_PRM Accuracy": 0.42452830188679247,
+      "eval_PRM F1": 0.47863247863247865,
+      "eval_PRM F1 AUC": 0.538239916186485,
+      "eval_PRM F1 Neg": 0.35789473684210527,
+      "eval_PRM NPV": 0.2361111111111111,
+      "eval_PRM Precision": 0.8235294117647058,
+      "eval_PRM Recall": 0.3373493975903614,
+      "eval_PRM Specificty": 0.7391304347826086,
+      "eval_loss": 0.8178266882896423,
+      "eval_runtime": 20.6776,
+      "eval_samples_per_second": 2.128,
+      "eval_steps_per_second": 0.145,
+      "step": 670
+    },
+    {
+      "epoch": 0.08686927533417484,
+      "grad_norm": 0.25526686407750226,
+      "learning_rate": 1.0850582147477362e-06,
+      "loss": 1.219,
+      "step": 671
+    },
+    {
+      "epoch": 0.0869987377415283,
+      "grad_norm": 0.13101939037668023,
+      "learning_rate": 1.0866752910737388e-06,
+      "loss": 1.0942,
+      "step": 672
+    },
+    {
+      "epoch": 0.08712820014888177,
+      "grad_norm": 0.23585752002615576,
+      "learning_rate": 1.0882923673997415e-06,
+      "loss": 1.1882,
+      "step": 673
+    },
+    {
+      "epoch": 0.08725766255623524,
+      "grad_norm": 0.1339957113599109,
+      "learning_rate": 1.089909443725744e-06,
+      "loss": 1.0079,
+      "step": 674
+    },
+    {
+      "epoch": 0.08738712496358869,
+      "grad_norm": 0.1925221385017757,
+      "learning_rate": 1.0915265200517464e-06,
+      "loss": 1.0786,
+      "step": 675
+    },
+    {
+      "epoch": 0.08738712496358869,
+      "eval_PRM Accuracy": 0.44339622641509435,
+      "eval_PRM F1": 0.5042016806722689,
+      "eval_PRM F1 AUC": 0.5502881089575693,
+      "eval_PRM F1 Neg": 0.3655913978494624,
+      "eval_PRM NPV": 0.24285714285714285,
+      "eval_PRM Precision": 0.8333333333333334,
+      "eval_PRM Recall": 0.3614457831325301,
+      "eval_PRM Specificty": 0.7391304347826086,
+      "eval_loss": 0.8139204382896423,
+      "eval_runtime": 20.2163,
+      "eval_samples_per_second": 2.176,
+      "eval_steps_per_second": 0.148,
+      "step": 675
+    },
+    {
+      "epoch": 0.08751658737094216,
+      "grad_norm": 0.2267659111579892,
+      "learning_rate": 1.093143596377749e-06,
+      "loss": 1.3188,
+      "step": 676
+    },
+    {
+      "epoch": 0.08764604977829563,
+      "grad_norm": 0.26228727436253496,
+      "learning_rate": 1.0947606727037518e-06,
+      "loss": 1.3037,
+      "step": 677
+    },
+    {
+      "epoch": 0.0877755121856491,
+      "grad_norm": 0.17323508486735067,
+      "learning_rate": 1.0963777490297544e-06,
+      "loss": 1.2368,
+      "step": 678
+    },
+    {
+      "epoch": 0.08790497459300256,
+      "grad_norm": 0.2338530487729072,
+      "learning_rate": 1.0979948253557569e-06,
+      "loss": 1.2469,
+      "step": 679
+    },
+    {
+      "epoch": 0.08803443700035603,
+      "grad_norm": 0.2447686675850731,
+      "learning_rate": 1.0996119016817593e-06,
+      "loss": 1.2308,
+      "step": 680
+    },
+    {
+      "epoch": 0.08803443700035603,
+      "eval_PRM Accuracy": 0.44339622641509435,
+      "eval_PRM F1": 0.5042016806722689,
+      "eval_PRM F1 AUC": 0.5502881089575693,
+      "eval_PRM F1 Neg": 0.3655913978494624,
+      "eval_PRM NPV": 0.24285714285714285,
+      "eval_PRM Precision": 0.8333333333333334,
+      "eval_PRM Recall": 0.3614457831325301,
+      "eval_PRM Specificty": 0.7391304347826086,
+      "eval_loss": 0.8110795617103577,
+      "eval_runtime": 20.7204,
+      "eval_samples_per_second": 2.124,
+      "eval_steps_per_second": 0.145,
+      "step": 680
+    },
+    {
+      "epoch": 0.08816389940770948,
+      "grad_norm": 0.16836979000188007,
+      "learning_rate": 1.101228978007762e-06,
+      "loss": 1.1305,
+      "step": 681
+    },
+    {
+      "epoch": 0.08829336181506295,
+      "grad_norm": 0.18045528434385183,
+      "learning_rate": 1.1028460543337647e-06,
+      "loss": 1.0947,
+      "step": 682
+    },
+    {
+      "epoch": 0.08842282422241642,
+      "grad_norm": 0.14146749045737927,
+      "learning_rate": 1.1044631306597671e-06,
+      "loss": 1.0575,
+      "step": 683
+    },
+    {
+      "epoch": 0.08855228662976988,
+      "grad_norm": 0.14012250655151107,
+      "learning_rate": 1.1060802069857698e-06,
+      "loss": 1.0856,
+      "step": 684
+    },
+    {
+      "epoch": 0.08868174903712335,
+      "grad_norm": 0.2417463812326379,
+      "learning_rate": 1.1076972833117725e-06,
+      "loss": 1.2232,
+      "step": 685
+    },
+    {
+      "epoch": 0.08868174903712335,
+      "eval_PRM Accuracy": 0.44339622641509435,
+      "eval_PRM F1": 0.5042016806722689,
+      "eval_PRM F1 AUC": 0.5502881089575693,
+      "eval_PRM F1 Neg": 0.3655913978494624,
+      "eval_PRM NPV": 0.24285714285714285,
+      "eval_PRM Precision": 0.8333333333333334,
+      "eval_PRM Recall": 0.3614457831325301,
+      "eval_PRM Specificty": 0.7391304347826086,
+      "eval_loss": 0.8099254369735718,
+      "eval_runtime": 20.6724,
+      "eval_samples_per_second": 2.128,
+      "eval_steps_per_second": 0.145,
+      "step": 685
+    },
+    {
+      "epoch": 0.0888112114444768,
+      "grad_norm": 0.2575672945773416,
+      "learning_rate": 1.1093143596377752e-06,
+      "loss": 1.173,
+      "step": 686
+    },
+    {
+      "epoch": 0.08894067385183027,
+      "grad_norm": 0.1739372476969271,
+      "learning_rate": 1.1109314359637776e-06,
+      "loss": 1.1282,
+      "step": 687
+    },
+    {
+      "epoch": 0.08907013625918374,
+      "grad_norm": 0.22907972073670896,
+      "learning_rate": 1.11254851228978e-06,
+      "loss": 1.2397,
+      "step": 688
+    },
+    {
+      "epoch": 0.0891995986665372,
+      "grad_norm": 0.1586394331462787,
+      "learning_rate": 1.1141655886157827e-06,
+      "loss": 0.943,
+      "step": 689
+    },
+    {
+      "epoch": 0.08932906107389067,
+      "grad_norm": 0.17570396617790188,
+      "learning_rate": 1.1157826649417854e-06,
+      "loss": 1.1677,
+      "step": 690
+    },
+    {
+      "epoch": 0.08932906107389067,
+      "eval_PRM Accuracy": 0.4339622641509434,
+      "eval_PRM F1": 0.4915254237288136,
+      "eval_PRM F1 AUC": 0.5442640125720272,
+      "eval_PRM F1 Neg": 0.3617021276595745,
+      "eval_PRM NPV": 0.23943661971830985,
+      "eval_PRM Precision": 0.8285714285714286,
+      "eval_PRM Recall": 0.3493975903614458,
+      "eval_PRM Specificty": 0.7391304347826086,
+      "eval_loss": 0.8084161877632141,
+      "eval_runtime": 20.7112,
+      "eval_samples_per_second": 2.124,
+      "eval_steps_per_second": 0.145,
+      "step": 690
+    },
+    {
+      "epoch": 0.08945852348124414,
+      "grad_norm": 0.2023680290120114,
+      "learning_rate": 1.1173997412677879e-06,
+      "loss": 1.0775,
+      "step": 691
+    },
+    {
+      "epoch": 0.0895879858885976,
+      "grad_norm": 0.18025861010631436,
+      "learning_rate": 1.1190168175937905e-06,
+      "loss": 1.1085,
+      "step": 692
+    },
+    {
+      "epoch": 0.08971744829595106,
+      "grad_norm": 0.25113789254616725,
+      "learning_rate": 1.120633893919793e-06,
+      "loss": 1.2981,
+      "step": 693
+    },
+    {
+      "epoch": 0.08984691070330453,
+      "grad_norm": 0.22379107138037224,
+      "learning_rate": 1.1222509702457957e-06,
+      "loss": 1.1171,
+      "step": 694
+    },
+    {
+      "epoch": 0.089976373110658,
+      "grad_norm": 0.21037997112404783,
+      "learning_rate": 1.1238680465717983e-06,
+      "loss": 1.1964,
+      "step": 695
+    },
+    {
+      "epoch": 0.089976373110658,
+      "eval_PRM Accuracy": 0.4528301886792453,
+      "eval_PRM F1": 0.5166666666666667,
+      "eval_PRM F1 AUC": 0.5563122053431115,
+      "eval_PRM F1 Neg": 0.3695652173913043,
+      "eval_PRM NPV": 0.2463768115942029,
+      "eval_PRM Precision": 0.8378378378378378,
+      "eval_PRM Recall": 0.37349397590361444,
+      "eval_PRM Specificty": 0.7391304347826086,
+      "eval_loss": 0.8039772510528564,
+      "eval_runtime": 20.7496,
+      "eval_samples_per_second": 2.121,
+      "eval_steps_per_second": 0.145,
+      "step": 695
+    },
+    {
+      "epoch": 0.09010583551801146,
+      "grad_norm": 0.2606377649490679,
+      "learning_rate": 1.1254851228978008e-06,
+      "loss": 1.1665,
+      "step": 696
+    },
+    {
+      "epoch": 0.09023529792536492,
+      "grad_norm": 0.2113235011341728,
+      "learning_rate": 1.1271021992238035e-06,
+      "loss": 1.2775,
+      "step": 697
+    },
+    {
+      "epoch": 0.09036476033271838,
+      "grad_norm": 0.2486775979635233,
+      "learning_rate": 1.1287192755498062e-06,
+      "loss": 1.2161,
+      "step": 698
+    },
+    {
+      "epoch": 0.09049422274007185,
+      "grad_norm": 0.2222114752489947,
+      "learning_rate": 1.1303363518758086e-06,
+      "loss": 1.1936,
+      "step": 699
+    },
+    {
+      "epoch": 0.09062368514742532,
+      "grad_norm": 0.22642323162027564,
+      "learning_rate": 1.131953428201811e-06,
+      "loss": 1.2134,
+      "step": 700
+    },
+    {
+      "epoch": 0.09062368514742532,
+      "eval_PRM Accuracy": 0.4528301886792453,
+      "eval_PRM F1": 0.5166666666666667,
+      "eval_PRM F1 AUC": 0.5563122053431115,
+      "eval_PRM F1 Neg": 0.3695652173913043,
+      "eval_PRM NPV": 0.2463768115942029,
+      "eval_PRM Precision": 0.8378378378378378,
+      "eval_PRM Recall": 0.37349397590361444,
+      "eval_PRM Specificty": 0.7391304347826086,
+      "eval_loss": 0.8037996888160706,
+      "eval_runtime": 20.5456,
+      "eval_samples_per_second": 2.142,
+      "eval_steps_per_second": 0.146,
+      "step": 700
+    },
+    {
+      "epoch": 0.09075314755477878,
+      "grad_norm": 0.2290038933782361,
+      "learning_rate": 1.1335705045278137e-06,
+      "loss": 1.1831,
+      "step": 701
+    },
+    {
+      "epoch": 0.09088260996213225,
+      "grad_norm": 0.2025124937799278,
+      "learning_rate": 1.1351875808538164e-06,
+      "loss": 1.1666,
+      "step": 702
+    },
+    {
+      "epoch": 0.0910120723694857,
+      "grad_norm": 0.1948554749755235,
+      "learning_rate": 1.136804657179819e-06,
+      "loss": 1.0779,
+      "step": 703
+    },
+    {
+      "epoch": 0.09114153477683917,
+      "grad_norm": 0.18392465630639968,
+      "learning_rate": 1.1384217335058215e-06,
+      "loss": 1.0569,
+      "step": 704
+    },
+    {
+      "epoch": 0.09127099718419264,
+      "grad_norm": 0.17716806451257086,
+      "learning_rate": 1.1400388098318242e-06,
+      "loss": 1.1206,
+      "step": 705
+    },
+    {
+      "epoch": 0.09127099718419264,
+      "eval_PRM Accuracy": 0.44339622641509435,
+      "eval_PRM F1": 0.5042016806722689,
+      "eval_PRM F1 AUC": 0.5502881089575693,
+      "eval_PRM F1 Neg": 0.3655913978494624,
+      "eval_PRM NPV": 0.24285714285714285,
+      "eval_PRM Precision": 0.8333333333333334,
+      "eval_PRM Recall": 0.3614457831325301,
+      "eval_PRM Specificty": 0.7391304347826086,
+      "eval_loss": 0.8018465638160706,
+      "eval_runtime": 20.4005,
+      "eval_samples_per_second": 2.157,
+      "eval_steps_per_second": 0.147,
+      "step": 705
+    },
+    {
+      "epoch": 0.09140045959154611,
+      "grad_norm": 0.16953282281538468,
+      "learning_rate": 1.1416558861578267e-06,
+      "loss": 1.1445,
+      "step": 706
+    },
+    {
+      "epoch": 0.09152992199889957,
+      "grad_norm": 0.18977631120383,
+      "learning_rate": 1.1432729624838293e-06,
+      "loss": 1.0657,
+      "step": 707
+    },
+    {
+      "epoch": 0.09165938440625303,
+      "grad_norm": 0.1708875913494527,
+      "learning_rate": 1.144890038809832e-06,
+      "loss": 1.1627,
+      "step": 708
+    },
+    {
+      "epoch": 0.0917888468136065,
+      "grad_norm": 0.23252154332857472,
+      "learning_rate": 1.1465071151358345e-06,
+      "loss": 1.103,
+      "step": 709
+    },
+    {
+      "epoch": 0.09191830922095996,
+      "grad_norm": 0.2165052273751677,
+      "learning_rate": 1.1481241914618371e-06,
+      "loss": 1.1841,
+      "step": 710
+    },
+    {
+      "epoch": 0.09191830922095996,
+      "eval_PRM Accuracy": 0.46226415094339623,
+      "eval_PRM F1": 0.5289256198347108,
+      "eval_PRM F1 AUC": 0.5623363017286537,
+      "eval_PRM F1 Neg": 0.37362637362637363,
+      "eval_PRM NPV": 0.25,
+      "eval_PRM Precision": 0.8421052631578947,
+      "eval_PRM Recall": 0.3855421686746988,
+      "eval_PRM Specificty": 0.7391304347826086,
+      "eval_loss": 0.7997159361839294,
+      "eval_runtime": 20.7977,
+      "eval_samples_per_second": 2.116,
+      "eval_steps_per_second": 0.144,
+      "step": 710
+    },
+    {
+      "epoch": 0.09204777162831343,
+      "grad_norm": 0.21766521026264884,
+      "learning_rate": 1.1497412677878398e-06,
+      "loss": 1.2906,
+      "step": 711
+    },
+    {
+      "epoch": 0.0921772340356669,
+      "grad_norm": 0.22197304320077468,
+      "learning_rate": 1.1513583441138423e-06,
+      "loss": 1.089,
+      "step": 712
+    },
+    {
+      "epoch": 0.09230669644302036,
+      "grad_norm": 0.2238875762240465,
+      "learning_rate": 1.1529754204398447e-06,
+      "loss": 1.2035,
+      "step": 713
+    },
+    {
+      "epoch": 0.09243615885037382,
+      "grad_norm": 0.1954242348039915,
+      "learning_rate": 1.1545924967658474e-06,
+      "loss": 1.1066,
+      "step": 714
+    },
+    {
+      "epoch": 0.09256562125772728,
+      "grad_norm": 0.2232933614698844,
+      "learning_rate": 1.15620957309185e-06,
+      "loss": 1.1366,
+      "step": 715
+    },
+    {
+      "epoch": 0.09256562125772728,
+      "eval_PRM Accuracy": 0.46226415094339623,
+      "eval_PRM F1": 0.5289256198347108,
+      "eval_PRM F1 AUC": 0.5623363017286537,
+      "eval_PRM F1 Neg": 0.37362637362637363,
+      "eval_PRM NPV": 0.25,
+      "eval_PRM Precision": 0.8421052631578947,
+      "eval_PRM Recall": 0.3855421686746988,
+      "eval_PRM Specificty": 0.7391304347826086,
+      "eval_loss": 0.7931463122367859,
+      "eval_runtime": 20.7705,
+      "eval_samples_per_second": 2.118,
+      "eval_steps_per_second": 0.144,
+      "step": 715
+    },
+    {
+      "epoch": 0.09269508366508075,
+      "grad_norm": 0.28404114622284343,
+      "learning_rate": 1.1578266494178527e-06,
+      "loss": 1.142,
+      "step": 716
+    },
+    {
+      "epoch": 0.09282454607243422,
+      "grad_norm": 0.23132211829055413,
+      "learning_rate": 1.1594437257438552e-06,
+      "loss": 1.1134,
+      "step": 717
+    },
+    {
+      "epoch": 0.09295400847978769,
+      "grad_norm": 0.254491916302133,
+      "learning_rate": 1.1610608020698579e-06,
+      "loss": 1.1635,
+      "step": 718
+    },
+    {
+      "epoch": 0.09308347088714114,
+      "grad_norm": 0.22014051101893578,
+      "learning_rate": 1.1626778783958603e-06,
+      "loss": 1.2073,
+      "step": 719
+    },
+    {
+      "epoch": 0.09321293329449461,
+      "grad_norm": 0.26117122499873874,
+      "learning_rate": 1.164294954721863e-06,
+      "loss": 1.2126,
+      "step": 720
+    },
+    {
+      "epoch": 0.09321293329449461,
+      "eval_PRM Accuracy": 0.46226415094339623,
+      "eval_PRM F1": 0.5289256198347108,
+      "eval_PRM F1 AUC": 0.5623363017286537,
+      "eval_PRM F1 Neg": 0.37362637362637363,
+      "eval_PRM NPV": 0.25,
+      "eval_PRM Precision": 0.8421052631578947,
+      "eval_PRM Recall": 0.3855421686746988,
+      "eval_PRM Specificty": 0.7391304347826086,
+      "eval_loss": 0.7950994372367859,
+      "eval_runtime": 20.3555,
+      "eval_samples_per_second": 2.162,
+      "eval_steps_per_second": 0.147,
+      "step": 720
+    },
+    {
+      "epoch": 0.09334239570184807,
+      "grad_norm": 0.1762444227384742,
+      "learning_rate": 1.1659120310478655e-06,
+      "loss": 1.0873,
+      "step": 721
+    },
+    {
+      "epoch": 0.09347185810920154,
+      "grad_norm": 0.2185467195206546,
+      "learning_rate": 1.1675291073738681e-06,
+      "loss": 1.2135,
+      "step": 722
+    },
+    {
+      "epoch": 0.09360132051655501,
+      "grad_norm": 0.193551395954972,
+      "learning_rate": 1.1691461836998708e-06,
+      "loss": 1.0702,
+      "step": 723
+    },
+    {
+      "epoch": 0.09373078292390848,
+      "grad_norm": 0.20484451560623293,
+      "learning_rate": 1.1707632600258735e-06,
+      "loss": 1.0972,
+      "step": 724
+    },
+    {
+      "epoch": 0.09386024533126193,
+      "grad_norm": 0.20449170093654723,
+      "learning_rate": 1.172380336351876e-06,
+      "loss": 1.1573,
+      "step": 725
+    },
+    {
+      "epoch": 0.09386024533126193,
+      "eval_PRM Accuracy": 0.46226415094339623,
+      "eval_PRM F1": 0.5289256198347108,
+      "eval_PRM F1 AUC": 0.5623363017286537,
+      "eval_PRM F1 Neg": 0.37362637362637363,
+      "eval_PRM NPV": 0.25,
+      "eval_PRM Precision": 0.8421052631578947,
+      "eval_PRM Recall": 0.3855421686746988,
+      "eval_PRM Specificty": 0.7391304347826086,
+      "eval_loss": 0.7913707494735718,
+      "eval_runtime": 20.5034,
+      "eval_samples_per_second": 2.146,
+      "eval_steps_per_second": 0.146,
+      "step": 725
+    },
+    {
+      "epoch": 0.0939897077386154,
+      "grad_norm": 0.23602814734352026,
+      "learning_rate": 1.1739974126778784e-06,
+      "loss": 1.1295,
+      "step": 726
+    },
+    {
+      "epoch": 0.09411917014596886,
+      "grad_norm": 0.25020733076136553,
+      "learning_rate": 1.175614489003881e-06,
+      "loss": 1.2802,
+      "step": 727
+    },
+    {
+      "epoch": 0.09424863255332233,
+      "grad_norm": 0.21298226598414483,
+      "learning_rate": 1.1772315653298837e-06,
+      "loss": 1.0381,
+      "step": 728
+    },
+    {
+      "epoch": 0.0943780949606758,
+      "grad_norm": 0.20291897193281175,
+      "learning_rate": 1.1788486416558862e-06,
+      "loss": 1.1039,
+      "step": 729
+    },
+    {
+      "epoch": 0.09450755736802925,
+      "grad_norm": 0.21533692228539952,
+      "learning_rate": 1.1804657179818889e-06,
+      "loss": 1.1329,
+      "step": 730
+    },
+    {
+      "epoch": 0.09450755736802925,
+      "eval_PRM Accuracy": 0.46226415094339623,
+      "eval_PRM F1": 0.5289256198347108,
+      "eval_PRM F1 AUC": 0.5623363017286537,
+      "eval_PRM F1 Neg": 0.37362637362637363,
+      "eval_PRM NPV": 0.25,
+      "eval_PRM Precision": 0.8421052631578947,
+      "eval_PRM Recall": 0.3855421686746988,
+      "eval_PRM Specificty": 0.7391304347826086,
+      "eval_loss": 0.7887073755264282,
+      "eval_runtime": 20.7739,
+      "eval_samples_per_second": 2.118,
+      "eval_steps_per_second": 0.144,
+      "step": 730
+    },
+    {
+      "epoch": 0.09463701977538272,
+      "grad_norm": 0.22493749360228335,
+      "learning_rate": 1.1820827943078915e-06,
+      "loss": 1.1873,
+      "step": 731
+    },
+    {
+      "epoch": 0.09476648218273619,
+      "grad_norm": 0.25258338541134534,
+      "learning_rate": 1.183699870633894e-06,
+      "loss": 1.2369,
+      "step": 732
+    },
+    {
+      "epoch": 0.09489594459008965,
+      "grad_norm": 0.24502143906121532,
+      "learning_rate": 1.1853169469598967e-06,
+      "loss": 1.2261,
+      "step": 733
+    },
+    {
+      "epoch": 0.09502540699744312,
+      "grad_norm": 0.19867196072302493,
+      "learning_rate": 1.1869340232858991e-06,
+      "loss": 1.0807,
+      "step": 734
+    },
+    {
+      "epoch": 0.09515486940479659,
+      "grad_norm": 0.21560138528027784,
+      "learning_rate": 1.1885510996119018e-06,
+      "loss": 1.1527,
+      "step": 735
+    },
+    {
+      "epoch": 0.09515486940479659,
+      "eval_PRM Accuracy": 0.46226415094339623,
+      "eval_PRM F1": 0.5289256198347108,
+      "eval_PRM F1 AUC": 0.5623363017286537,
+      "eval_PRM F1 Neg": 0.37362637362637363,
+      "eval_PRM NPV": 0.25,
+      "eval_PRM Precision": 0.8421052631578947,
+      "eval_PRM Recall": 0.3855421686746988,
+      "eval_PRM Specificty": 0.7391304347826086,
+      "eval_loss": 0.7887073755264282,
+      "eval_runtime": 20.357,
+      "eval_samples_per_second": 2.161,
+      "eval_steps_per_second": 0.147,
+      "step": 735
+    },
+    {
+      "epoch": 0.09528433181215004,
+      "grad_norm": 0.26988956826763044,
+      "learning_rate": 1.1901681759379045e-06,
+      "loss": 1.1559,
+      "step": 736
+    },
+    {
+      "epoch": 0.09541379421950351,
+      "grad_norm": 0.23213832317126168,
+      "learning_rate": 1.191785252263907e-06,
+      "loss": 1.2421,
+      "step": 737
+    },
+    {
+      "epoch": 0.09554325662685698,
+      "grad_norm": 0.2548441415698318,
+      "learning_rate": 1.1934023285899094e-06,
+      "loss": 1.1719,
+      "step": 738
+    },
+    {
+      "epoch": 0.09567271903421044,
+      "grad_norm": 0.17831907994299862,
+      "learning_rate": 1.195019404915912e-06,
+      "loss": 1.1624,
+      "step": 739
+    },
+    {
+      "epoch": 0.09580218144156391,
+      "grad_norm": 0.2118395048352656,
+      "learning_rate": 1.1966364812419147e-06,
+      "loss": 1.1143,
+      "step": 740
+    },
+    {
+      "epoch": 0.09580218144156391,
+      "eval_PRM Accuracy": 0.46226415094339623,
+      "eval_PRM F1": 0.5289256198347108,
+      "eval_PRM F1 AUC": 0.5623363017286537,
+      "eval_PRM F1 Neg": 0.37362637362637363,
+      "eval_PRM NPV": 0.25,
+      "eval_PRM Precision": 0.8421052631578947,
+      "eval_PRM Recall": 0.3855421686746988,
+      "eval_PRM Specificty": 0.7391304347826086,
+      "eval_loss": 0.7819602489471436,
+      "eval_runtime": 20.1834,
+      "eval_samples_per_second": 2.18,
+      "eval_steps_per_second": 0.149,
+      "step": 740
+    },
+    {
+      "epoch": 0.09593164384891736,
+      "grad_norm": 0.22498359346494612,
+      "learning_rate": 1.1982535575679174e-06,
+      "loss": 1.1449,
+      "step": 741
+    },
+    {
+      "epoch": 0.09606110625627083,
+      "grad_norm": 0.25679804483242924,
+      "learning_rate": 1.1998706338939199e-06,
+      "loss": 1.1763,
+      "step": 742
+    },
+    {
+      "epoch": 0.0961905686636243,
+      "grad_norm": 0.21622985834388322,
+      "learning_rate": 1.2014877102199225e-06,
+      "loss": 1.0768,
+      "step": 743
+    },
+    {
+      "epoch": 0.09632003107097777,
+      "grad_norm": 0.28229960043745805,
+      "learning_rate": 1.2031047865459252e-06,
+      "loss": 1.073,
+      "step": 744
+    },
+    {
+      "epoch": 0.09644949347833123,
+      "grad_norm": 0.22607886227043467,
+      "learning_rate": 1.2047218628719277e-06,
+      "loss": 1.1025,
+      "step": 745
+    },
+    {
+      "epoch": 0.09644949347833123,
+      "eval_PRM Accuracy": 0.4528301886792453,
+      "eval_PRM F1": 0.5245901639344263,
+      "eval_PRM F1 AUC": 0.5405971712938711,
+      "eval_PRM F1 Neg": 0.35555555555555557,
+      "eval_PRM NPV": 0.23880597014925373,
+      "eval_PRM Precision": 0.8205128205128205,
+      "eval_PRM Recall": 0.3855421686746988,
+      "eval_PRM Specificty": 0.6956521739130435,
+      "eval_loss": 0.7828480005264282,
+      "eval_runtime": 20.8025,
+      "eval_samples_per_second": 2.115,
+      "eval_steps_per_second": 0.144,
+      "step": 745
+    },
+    {
+      "epoch": 0.0965789558856847,
+      "grad_norm": 0.21547207796355666,
+      "learning_rate": 1.2063389391979301e-06,
+      "loss": 1.0645,
+      "step": 746
+    },
+    {
+      "epoch": 0.09670841829303815,
+      "grad_norm": 0.23769599229538318,
+      "learning_rate": 1.2079560155239328e-06,
+      "loss": 1.0543,
+      "step": 747
+    },
+    {
+      "epoch": 0.09683788070039162,
+      "grad_norm": 0.19792986678409308,
+      "learning_rate": 1.2095730918499355e-06,
+      "loss": 1.0785,
+      "step": 748
+    },
+    {
+      "epoch": 0.09696734310774509,
+      "grad_norm": 0.2724932565635008,
+      "learning_rate": 1.2111901681759381e-06,
+      "loss": 1.2177,
+      "step": 749
+    },
+    {
+      "epoch": 0.09709680551509856,
+      "grad_norm": 0.2754120410788316,
+      "learning_rate": 1.2128072445019406e-06,
+      "loss": 1.2132,
+      "step": 750
+    },
+    {
+      "epoch": 0.09709680551509856,
+      "eval_PRM Accuracy": 0.46226415094339623,
+      "eval_PRM F1": 0.5289256198347108,
+      "eval_PRM F1 AUC": 0.5623363017286537,
+      "eval_PRM F1 Neg": 0.37362637362637363,
+      "eval_PRM NPV": 0.25,
+      "eval_PRM Precision": 0.8421052631578947,
+      "eval_PRM Recall": 0.3855421686746988,
+      "eval_PRM Specificty": 0.7391304347826086,
+      "eval_loss": 0.7798295617103577,
+      "eval_runtime": 20.3,
+      "eval_samples_per_second": 2.167,
+      "eval_steps_per_second": 0.148,
+      "step": 750
+    },
+    {
+      "epoch": 0.09722626792245202,
+      "grad_norm": 0.25180119882669694,
+      "learning_rate": 1.214424320827943e-06,
+      "loss": 1.1548,
+      "step": 751
+    },
+    {
+      "epoch": 0.09735573032980548,
+      "grad_norm": 0.20404790639392936,
+      "learning_rate": 1.2160413971539457e-06,
+      "loss": 1.182,
+      "step": 752
+    },
+    {
+      "epoch": 0.09748519273715894,
+      "grad_norm": 0.1735372404065468,
+      "learning_rate": 1.2176584734799484e-06,
+      "loss": 1.1332,
+      "step": 753
+    },
+    {
+      "epoch": 0.09761465514451241,
+      "grad_norm": 0.23075240046720094,
+      "learning_rate": 1.219275549805951e-06,
+      "loss": 1.0579,
+      "step": 754
+    },
+    {
+      "epoch": 0.09774411755186588,
+      "grad_norm": 0.23809237937129,
+      "learning_rate": 1.2208926261319535e-06,
+      "loss": 1.1243,
+      "step": 755
+    },
+    {
+      "epoch": 0.09774411755186588,
+      "eval_PRM Accuracy": 0.46226415094339623,
+      "eval_PRM F1": 0.5289256198347108,
+      "eval_PRM F1 AUC": 0.5623363017286537,
+      "eval_PRM F1 Neg": 0.37362637362637363,
+      "eval_PRM NPV": 0.25,
+      "eval_PRM Precision": 0.8421052631578947,
+      "eval_PRM Recall": 0.3855421686746988,
+      "eval_PRM Specificty": 0.7391304347826086,
+      "eval_loss": 0.7787641882896423,
+      "eval_runtime": 20.3263,
+      "eval_samples_per_second": 2.165,
+      "eval_steps_per_second": 0.148,
+      "step": 755
+    },
+    {
+      "epoch": 0.09787357995921935,
+      "grad_norm": 0.26256074388894685,
+      "learning_rate": 1.2225097024579562e-06,
+      "loss": 1.2206,
+      "step": 756
+    },
+    {
+      "epoch": 0.09800304236657281,
+      "grad_norm": 0.26459955124630286,
+      "learning_rate": 1.2241267787839586e-06,
+      "loss": 1.1464,
+      "step": 757
+    },
+    {
+      "epoch": 0.09813250477392627,
+      "grad_norm": 0.17619686525138678,
+      "learning_rate": 1.2257438551099613e-06,
+      "loss": 1.1153,
+      "step": 758
+    },
+    {
+      "epoch": 0.09826196718127973,
+      "grad_norm": 0.20591196202519804,
+      "learning_rate": 1.2273609314359638e-06,
+      "loss": 1.1368,
+      "step": 759
+    },
+    {
+      "epoch": 0.0983914295886332,
+      "grad_norm": 0.2758261439083429,
+      "learning_rate": 1.2289780077619664e-06,
+      "loss": 1.2095,
+      "step": 760
+    },
+    {
+      "epoch": 0.0983914295886332,
+      "eval_PRM Accuracy": 0.4716981132075472,
+      "eval_PRM F1": 0.5409836065573771,
+      "eval_PRM F1 AUC": 0.5683603981141959,
+      "eval_PRM F1 Neg": 0.37777777777777777,
+      "eval_PRM NPV": 0.2537313432835821,
+      "eval_PRM Precision": 0.8461538461538461,
+      "eval_PRM Recall": 0.39759036144578314,
+      "eval_PRM Specificty": 0.7391304347826086,
+      "eval_loss": 0.7729048132896423,
+      "eval_runtime": 21.2037,
+      "eval_samples_per_second": 2.075,
+      "eval_steps_per_second": 0.141,
+      "step": 760
+    },
+    {
+      "epoch": 0.09852089199598667,
+      "grad_norm": 0.22755765869153188,
+      "learning_rate": 1.2305950840879691e-06,
+      "loss": 1.1379,
+      "step": 761
+    },
+    {
+      "epoch": 0.09865035440334013,
+      "grad_norm": 0.26708732476154073,
+      "learning_rate": 1.2322121604139718e-06,
+      "loss": 1.1426,
+      "step": 762
+    },
+    {
+      "epoch": 0.09877981681069359,
+      "grad_norm": 0.16798937659854712,
+      "learning_rate": 1.2338292367399742e-06,
+      "loss": 1.0905,
+      "step": 763
+    },
+    {
+      "epoch": 0.09890927921804706,
+      "grad_norm": 0.23179263637874012,
+      "learning_rate": 1.2354463130659767e-06,
+      "loss": 1.1682,
+      "step": 764
+    },
+    {
+      "epoch": 0.09903874162540052,
+      "grad_norm": 0.2584907967765254,
+      "learning_rate": 1.2370633893919794e-06,
+      "loss": 1.1705,
+      "step": 765
+    },
+    {
+      "epoch": 0.09903874162540052,
+      "eval_PRM Accuracy": 0.46226415094339623,
+      "eval_PRM F1": 0.5365853658536586,
+      "eval_PRM F1 AUC": 0.5466212676794133,
+      "eval_PRM F1 Neg": 0.3595505617977528,
+      "eval_PRM NPV": 0.24242424242424243,
+      "eval_PRM Precision": 0.825,
+      "eval_PRM Recall": 0.39759036144578314,
+      "eval_PRM Specificty": 0.6956521739130435,
+      "eval_loss": 0.7745028138160706,
+      "eval_runtime": 20.6353,
+      "eval_samples_per_second": 2.132,
+      "eval_steps_per_second": 0.145,
+      "step": 765
+    },
+    {
+      "epoch": 0.09916820403275399,
+      "grad_norm": 0.18909765975689427,
+      "learning_rate": 1.238680465717982e-06,
+      "loss": 1.163,
+      "step": 766
+    },
+    {
+      "epoch": 0.09929766644010746,
+      "grad_norm": 0.21810232020049436,
+      "learning_rate": 1.2402975420439845e-06,
+      "loss": 1.0478,
+      "step": 767
+    },
+    {
+      "epoch": 0.09942712884746092,
+      "grad_norm": 0.2143941439032801,
+      "learning_rate": 1.2419146183699872e-06,
+      "loss": 1.137,
+      "step": 768
+    },
+    {
+      "epoch": 0.09955659125481438,
+      "grad_norm": 0.2206534494147366,
+      "learning_rate": 1.2435316946959898e-06,
+      "loss": 1.2112,
+      "step": 769
+    },
+    {
+      "epoch": 0.09968605366216785,
+      "grad_norm": 0.11009506674702682,
+      "learning_rate": 1.2451487710219923e-06,
+      "loss": 0.9921,
+      "step": 770
+    },
+    {
+      "epoch": 0.09968605366216785,
+      "eval_PRM Accuracy": 0.46226415094339623,
+      "eval_PRM F1": 0.5365853658536586,
+      "eval_PRM F1 AUC": 0.5466212676794133,
+      "eval_PRM F1 Neg": 0.3595505617977528,
+      "eval_PRM NPV": 0.24242424242424243,
+      "eval_PRM Precision": 0.825,
+      "eval_PRM Recall": 0.39759036144578314,
+      "eval_PRM Specificty": 0.6956521739130435,
+      "eval_loss": 0.7700639367103577,
+      "eval_runtime": 20.7333,
+      "eval_samples_per_second": 2.122,
+      "eval_steps_per_second": 0.145,
+      "step": 770
+    },
+    {
+      "epoch": 0.09981551606952131,
+      "grad_norm": 0.1726759440108128,
+      "learning_rate": 1.246765847347995e-06,
+      "loss": 1.1448,
+      "step": 771
+    },
+    {
+      "epoch": 0.09994497847687478,
+      "grad_norm": 0.18023330118750003,
+      "learning_rate": 1.2483829236739974e-06,
+      "loss": 1.0341,
+      "step": 772
+    },
+    {
+      "epoch": 0.10007444088422825,
+      "grad_norm": 0.17997573828682634,
+      "learning_rate": 1.25e-06,
+      "loss": 1.1392,
+      "step": 773
+    },
+    {
+      "epoch": 0.1002039032915817,
+      "grad_norm": 0.2947517874316338,
+      "learning_rate": 1.249999936165541e-06,
+      "loss": 1.2645,
+      "step": 774
+    },
+    {
+      "epoch": 0.10033336569893517,
+      "grad_norm": 0.1634906643517373,
+      "learning_rate": 1.2499997446621767e-06,
+      "loss": 1.1113,
+      "step": 775
+    },
+    {
+      "epoch": 0.10033336569893517,
+      "eval_PRM Accuracy": 0.4716981132075472,
+      "eval_PRM F1": 0.5483870967741935,
+      "eval_PRM F1 AUC": 0.5526453640649555,
+      "eval_PRM F1 Neg": 0.36363636363636365,
+      "eval_PRM NPV": 0.24615384615384617,
+      "eval_PRM Precision": 0.8292682926829268,
+      "eval_PRM Recall": 0.40963855421686746,
+      "eval_PRM Specificty": 0.6956521739130435,
+      "eval_loss": 0.7691761255264282,
+      "eval_runtime": 20.23,
+      "eval_samples_per_second": 2.175,
+      "eval_steps_per_second": 0.148,
+      "step": 775
+    },
+    {
+      "epoch": 0.10046282810628863,
+      "grad_norm": 0.32602478038973365,
+      "learning_rate": 1.2499994254899465e-06,
+      "loss": 1.323,
+      "step": 776
+    },
+    {
+      "epoch": 0.1005922905136421,
+      "grad_norm": 0.20179936293072492,
+      "learning_rate": 1.2499989786489155e-06,
+      "loss": 1.1453,
+      "step": 777
+    },
+    {
+      "epoch": 0.10072175292099557,
+      "grad_norm": 0.20577352481423622,
+      "learning_rate": 1.249998404139175e-06,
+      "loss": 1.181,
+      "step": 778
+    },
+    {
+      "epoch": 0.10085121532834904,
+      "grad_norm": 0.24750993568770668,
+      "learning_rate": 1.2499977019608421e-06,
+      "loss": 1.1901,
+      "step": 779
+    },
+    {
+      "epoch": 0.10098067773570249,
+      "grad_norm": 0.11606933585077482,
+      "learning_rate": 1.2499968721140606e-06,
+      "loss": 0.9203,
+      "step": 780
+    },
+    {
+      "epoch": 0.10098067773570249,
+      "eval_PRM Accuracy": 0.4716981132075472,
+      "eval_PRM F1": 0.5483870967741935,
+      "eval_PRM F1 AUC": 0.5526453640649555,
+      "eval_PRM F1 Neg": 0.36363636363636365,
+      "eval_PRM NPV": 0.24615384615384617,
+      "eval_PRM Precision": 0.8292682926829268,
+      "eval_PRM Recall": 0.40963855421686746,
+      "eval_PRM Specificty": 0.6956521739130435,
+      "eval_loss": 0.7666903138160706,
+      "eval_runtime": 20.1493,
+      "eval_samples_per_second": 2.184,
+      "eval_steps_per_second": 0.149,
+      "step": 780
+    },
+    {
+      "epoch": 0.10111014014305596,
+      "grad_norm": 0.27554214708355385,
+      "learning_rate": 1.2499959145989998e-06,
+      "loss": 1.1538,
+      "step": 781
+    },
+    {
+      "epoch": 0.10123960255040942,
+      "grad_norm": 0.29854874437792533,
+      "learning_rate": 1.2499948294158554e-06,
+      "loss": 1.1133,
+      "step": 782
+    },
+    {
+      "epoch": 0.10136906495776289,
+      "grad_norm": 0.1461415097520277,
+      "learning_rate": 1.249993616564849e-06,
+      "loss": 1.1016,
+      "step": 783
+    },
+    {
+      "epoch": 0.10149852736511636,
+      "grad_norm": 0.1899700959275549,
+      "learning_rate": 1.2499922760462283e-06,
+      "loss": 1.0471,
+      "step": 784
+    },
+    {
+      "epoch": 0.10162798977246981,
+      "grad_norm": 0.20130272206953717,
+      "learning_rate": 1.2499908078602673e-06,
+      "loss": 1.0713,
+      "step": 785
+    },
+    {
+      "epoch": 0.10162798977246981,
+      "eval_PRM Accuracy": 0.4811320754716981,
+      "eval_PRM F1": 0.56,
+      "eval_PRM F1 AUC": 0.5586694604504976,
+      "eval_PRM F1 Neg": 0.367816091954023,
+      "eval_PRM NPV": 0.25,
+      "eval_PRM Precision": 0.8333333333333334,
+      "eval_PRM Recall": 0.42168674698795183,
+      "eval_PRM Specificty": 0.6956521739130435,
+      "eval_loss": 0.7631391882896423,
+      "eval_runtime": 19.6226,
+      "eval_samples_per_second": 2.242,
+      "eval_steps_per_second": 0.153,
+      "step": 785
+    },
+    {
+      "epoch": 0.10175745217982328,
+      "grad_norm": 0.18990503731956368,
+      "learning_rate": 1.249989212007266e-06,
+      "loss": 1.0071,
+      "step": 786
+    },
+    {
+      "epoch": 0.10188691458717675,
+      "grad_norm": 0.22835499308961776,
+      "learning_rate": 1.2499874884875499e-06,
+      "loss": 1.0748,
+      "step": 787
+    },
+    {
+      "epoch": 0.10201637699453021,
+      "grad_norm": 0.21772889024471295,
+      "learning_rate": 1.2499856373014715e-06,
+      "loss": 1.1852,
+      "step": 788
+    },
+    {
+      "epoch": 0.10214583940188368,
+      "grad_norm": 0.20171428516426992,
+      "learning_rate": 1.2499836584494087e-06,
+      "loss": 1.1459,
+      "step": 789
+    },
+    {
+      "epoch": 0.10227530180923715,
+      "grad_norm": 0.26180551538176716,
+      "learning_rate": 1.2499815519317659e-06,
+      "loss": 1.0314,
+      "step": 790
+    },
+    {
+      "epoch": 0.10227530180923715,
+      "eval_PRM Accuracy": 0.46226415094339623,
+      "eval_PRM F1": 0.544,
+      "eval_PRM F1 AUC": 0.5309062336301729,
+      "eval_PRM F1 Neg": 0.3448275862068966,
+      "eval_PRM NPV": 0.234375,
+      "eval_PRM Precision": 0.8095238095238095,
+      "eval_PRM Recall": 0.40963855421686746,
+      "eval_PRM Specificty": 0.6521739130434783,
+      "eval_loss": 0.7615411877632141,
+      "eval_runtime": 19.8179,
+      "eval_samples_per_second": 2.22,
+      "eval_steps_per_second": 0.151,
+      "step": 790
+    },
+    {
+      "epoch": 0.1024047642165906,
+      "grad_norm": 0.30427271253560895,
+      "learning_rate": 1.2499793177489731e-06,
+      "loss": 1.1116,
+      "step": 791
+    },
+    {
+      "epoch": 0.10253422662394407,
+      "grad_norm": 0.2871261634016219,
+      "learning_rate": 1.249976955901487e-06,
+      "loss": 1.2562,
+      "step": 792
+    },
+    {
+      "epoch": 0.10266368903129754,
+      "grad_norm": 0.2867264385576825,
+      "learning_rate": 1.2499744663897902e-06,
+      "loss": 1.1908,
+      "step": 793
+    },
+    {
+      "epoch": 0.102793151438651,
+      "grad_norm": 0.2195857276311651,
+      "learning_rate": 1.2499718492143908e-06,
+      "loss": 1.1089,
+      "step": 794
+    },
+    {
+      "epoch": 0.10292261384600447,
+      "grad_norm": 0.18002405653809045,
+      "learning_rate": 1.2499691043758235e-06,
+      "loss": 0.9922,
+      "step": 795
+    },
+    {
+      "epoch": 0.10292261384600447,
+      "eval_PRM Accuracy": 0.4716981132075472,
+      "eval_PRM F1": 0.5555555555555556,
+      "eval_PRM F1 AUC": 0.536930330015715,
+      "eval_PRM F1 Neg": 0.3488372093023256,
+      "eval_PRM NPV": 0.23809523809523808,
+      "eval_PRM Precision": 0.813953488372093,
+      "eval_PRM Recall": 0.42168674698795183,
+      "eval_PRM Specificty": 0.6521739130434783,
+      "eval_loss": 0.755859375,
+      "eval_runtime": 20.1458,
+      "eval_samples_per_second": 2.184,
+      "eval_steps_per_second": 0.149,
+      "step": 795
+    },
+    {
+      "epoch": 0.10305207625335792,
+      "grad_norm": 0.26273441011096743,
+      "learning_rate": 1.2499662318746493e-06,
+      "loss": 1.1824,
+      "step": 796
+    },
+    {
+      "epoch": 0.10318153866071139,
+      "grad_norm": 0.2035331448092164,
+      "learning_rate": 1.2499632317114545e-06,
+      "loss": 1.1109,
+      "step": 797
+    },
+    {
+      "epoch": 0.10331100106806486,
+      "grad_norm": 0.23762720233613846,
+      "learning_rate": 1.2499601038868525e-06,
+      "loss": 1.1183,
+      "step": 798
+    },
+    {
+      "epoch": 0.10344046347541833,
+      "grad_norm": 0.22516423721188877,
+      "learning_rate": 1.2499568484014818e-06,
+      "loss": 1.1132,
+      "step": 799
+    },
+    {
+      "epoch": 0.1035699258827718,
+      "grad_norm": 0.23017321219911982,
+      "learning_rate": 1.2499534652560074e-06,
+      "loss": 1.2016,
+      "step": 800
+    },
+    {
+      "epoch": 0.1035699258827718,
+      "eval_PRM Accuracy": 0.4716981132075472,
+      "eval_PRM F1": 0.5555555555555556,
+      "eval_PRM F1 AUC": 0.536930330015715,
+      "eval_PRM F1 Neg": 0.3488372093023256,
+      "eval_PRM NPV": 0.23809523809523808,
+      "eval_PRM Precision": 0.813953488372093,
+      "eval_PRM Recall": 0.42168674698795183,
+      "eval_PRM Specificty": 0.6521739130434783,
+      "eval_loss": 0.7574573755264282,
+      "eval_runtime": 20.6817,
+      "eval_samples_per_second": 2.127,
+      "eval_steps_per_second": 0.145,
+      "step": 800
+    },
+    {
+      "epoch": 0.10369938829012526,
+      "grad_norm": 0.2661833244040786,
+      "learning_rate": 1.2499499544511205e-06,
+      "loss": 1.1356,
+      "step": 801
+    },
+    {
+      "epoch": 0.10382885069747871,
+      "grad_norm": 0.19677229037189647,
+      "learning_rate": 1.2499463159875382e-06,
+      "loss": 1.0425,
+      "step": 802
+    },
+    {
+      "epoch": 0.10395831310483218,
+      "grad_norm": 0.23335311818799653,
+      "learning_rate": 1.249942549866004e-06,
+      "loss": 1.1335,
+      "step": 803
+    },
+    {
+      "epoch": 0.10408777551218565,
+      "grad_norm": 0.16858199718321024,
+      "learning_rate": 1.2499386560872869e-06,
+      "loss": 1.028,
+      "step": 804
+    },
+    {
+      "epoch": 0.10421723791953912,
+      "grad_norm": 0.12541411018143828,
+      "learning_rate": 1.2499346346521823e-06,
+      "loss": 0.9755,
+      "step": 805
+    },
+    {
+      "epoch": 0.10421723791953912,
+      "eval_PRM Accuracy": 0.4811320754716981,
+      "eval_PRM F1": 0.5669291338582677,
+      "eval_PRM F1 AUC": 0.5429544264012572,
+      "eval_PRM F1 Neg": 0.35294117647058826,
+      "eval_PRM NPV": 0.24193548387096775,
+      "eval_PRM Precision": 0.8181818181818182,
+      "eval_PRM Recall": 0.43373493975903615,
+      "eval_PRM Specificty": 0.6521739130434783,
+      "eval_loss": 0.7537286877632141,
+      "eval_runtime": 20.1429,
+      "eval_samples_per_second": 2.184,
+      "eval_steps_per_second": 0.149,
+      "step": 805
+    },
+    {
+      "epoch": 0.10434670032689258,
+      "grad_norm": 0.27581313021921944,
+      "learning_rate": 1.249930485561512e-06,
+      "loss": 1.1312,
+      "step": 806
+    },
+    {
+      "epoch": 0.10447616273424604,
+      "grad_norm": 0.21095757866507753,
+      "learning_rate": 1.2499262088161227e-06,
+      "loss": 0.9894,
+      "step": 807
+    },
+    {
+      "epoch": 0.1046056251415995,
+      "grad_norm": 0.28978489914227834,
+      "learning_rate": 1.249921804416889e-06,
+      "loss": 1.1943,
+      "step": 808
+    },
+    {
+      "epoch": 0.10473508754895297,
+      "grad_norm": 0.29046998993172607,
+      "learning_rate": 1.24991727236471e-06,
+      "loss": 1.2877,
+      "step": 809
+    },
+    {
+      "epoch": 0.10486454995630644,
+      "grad_norm": 0.29162852816636675,
+      "learning_rate": 1.2499126126605117e-06,
+      "loss": 1.1349,
+      "step": 810
+    },
+    {
+      "epoch": 0.10486454995630644,
+      "eval_PRM Accuracy": 0.4811320754716981,
+      "eval_PRM F1": 0.5669291338582677,
+      "eval_PRM F1 AUC": 0.5429544264012572,
+      "eval_PRM F1 Neg": 0.35294117647058826,
+      "eval_PRM NPV": 0.24193548387096775,
+      "eval_PRM Precision": 0.8181818181818182,
+      "eval_PRM Recall": 0.43373493975903615,
+      "eval_PRM Specificty": 0.6521739130434783,
+      "eval_loss": 0.7535511255264282,
+      "eval_runtime": 20.7589,
+      "eval_samples_per_second": 2.12,
+      "eval_steps_per_second": 0.145,
+      "step": 810
+    },
+    {
+      "epoch": 0.1049940123636599,
+      "grad_norm": 0.2749043529240586,
+      "learning_rate": 1.2499078253052457e-06,
+      "loss": 1.0829,
+      "step": 811
+    },
+    {
+      "epoch": 0.10512347477101337,
+      "grad_norm": 0.2780284760895125,
+      "learning_rate": 1.2499029102998902e-06,
+      "loss": 1.1619,
+      "step": 812
+    },
+    {
+      "epoch": 0.10525293717836683,
+      "grad_norm": 0.2734683706584518,
+      "learning_rate": 1.2498978676454491e-06,
+      "loss": 1.1861,
+      "step": 813
+    },
+    {
+      "epoch": 0.1053823995857203,
+      "grad_norm": 0.21662488251312262,
+      "learning_rate": 1.2498926973429524e-06,
+      "loss": 1.1172,
+      "step": 814
+    },
+    {
+      "epoch": 0.10551186199307376,
+      "grad_norm": 0.19336049556245302,
+      "learning_rate": 1.2498873993934564e-06,
+      "loss": 1.1609,
+      "step": 815
+    },
+    {
+      "epoch": 0.10551186199307376,
+      "eval_PRM Accuracy": 0.4811320754716981,
+      "eval_PRM F1": 0.5669291338582677,
+      "eval_PRM F1 AUC": 0.5429544264012572,
+      "eval_PRM F1 Neg": 0.35294117647058826,
+      "eval_PRM NPV": 0.24193548387096775,
+      "eval_PRM Precision": 0.8181818181818182,
+      "eval_PRM Recall": 0.43373493975903615,
+      "eval_PRM Specificty": 0.6521739130434783,
+      "eval_loss": 0.7489346861839294,
+      "eval_runtime": 20.3461,
+      "eval_samples_per_second": 2.163,
+      "eval_steps_per_second": 0.147,
+      "step": 815
+    },
+    {
+      "epoch": 0.10564132440042723,
+      "grad_norm": 0.25441193174545484,
+      "learning_rate": 1.2498819737980431e-06,
+      "loss": 1.1665,
+      "step": 816
+    },
+    {
+      "epoch": 0.1057707868077807,
+      "grad_norm": 0.2385243915678446,
+      "learning_rate": 1.2498764205578206e-06,
+      "loss": 1.1099,
+      "step": 817
+    },
+    {
+      "epoch": 0.10590024921513416,
+      "grad_norm": 0.1927430459602009,
+      "learning_rate": 1.249870739673924e-06,
+      "loss": 0.9925,
+      "step": 818
+    },
+    {
+      "epoch": 0.10602971162248762,
+      "grad_norm": 0.1462093731654415,
+      "learning_rate": 1.249864931147513e-06,
+      "loss": 1.0054,
+      "step": 819
+    },
+    {
+      "epoch": 0.10615917402984108,
+      "grad_norm": 0.3495356342249866,
+      "learning_rate": 1.2498589949797742e-06,
+      "loss": 1.2188,
+      "step": 820
+    },
+    {
+      "epoch": 0.10615917402984108,
+      "eval_PRM Accuracy": 0.49056603773584906,
+      "eval_PRM F1": 0.578125,
+      "eval_PRM F1 AUC": 0.5489785227867994,
+      "eval_PRM F1 Neg": 0.35714285714285715,
+      "eval_PRM NPV": 0.2459016393442623,
+      "eval_PRM Precision": 0.8222222222222222,
+      "eval_PRM Recall": 0.4457831325301205,
+      "eval_PRM Specificty": 0.6521739130434783,
+      "eval_loss": 0.7455610632896423,
+      "eval_runtime": 20.5506,
+      "eval_samples_per_second": 2.141,
+      "eval_steps_per_second": 0.146,
+      "step": 820
+    },
+    {
+      "epoch": 0.10628863643719455,
+      "grad_norm": 0.27414008357139674,
+      "learning_rate": 1.2498529311719207e-06,
+      "loss": 1.0292,
+      "step": 821
+    },
+    {
+      "epoch": 0.10641809884454802,
+      "grad_norm": 0.19845022130664222,
+      "learning_rate": 1.2498467397251908e-06,
+      "loss": 1.1018,
+      "step": 822
+    },
+    {
+      "epoch": 0.10654756125190148,
+      "grad_norm": 0.272713661848405,
+      "learning_rate": 1.249840420640849e-06,
+      "loss": 1.1055,
+      "step": 823
+    },
+    {
+      "epoch": 0.10667702365925494,
+      "grad_norm": 0.09100060547831061,
+      "learning_rate": 1.2498339739201865e-06,
+      "loss": 0.9164,
+      "step": 824
+    },
+    {
+      "epoch": 0.1068064860666084,
+      "grad_norm": 0.27746205802596247,
+      "learning_rate": 1.24982739956452e-06,
+      "loss": 1.0723,
+      "step": 825
+    },
+    {
+      "epoch": 0.1068064860666084,
+      "eval_PRM Accuracy": 0.49056603773584906,
+      "eval_PRM F1": 0.578125,
+      "eval_PRM F1 AUC": 0.5489785227867994,
+      "eval_PRM F1 Neg": 0.35714285714285715,
+      "eval_PRM NPV": 0.2459016393442623,
+      "eval_PRM Precision": 0.8222222222222222,
+      "eval_PRM Recall": 0.4457831325301205,
+      "eval_PRM Specificty": 0.6521739130434783,
+      "eval_loss": 0.7446733117103577,
+      "eval_runtime": 20.9275,
+      "eval_samples_per_second": 2.102,
+      "eval_steps_per_second": 0.143,
+      "step": 825
+    },
+    {
+      "epoch": 0.10693594847396187,
+      "grad_norm": 0.2856954309225274,
+      "learning_rate": 1.2498206975751924e-06,
+      "loss": 1.1639,
+      "step": 826
+    },
+    {
+      "epoch": 0.10706541088131534,
+      "grad_norm": 0.22483660343751327,
+      "learning_rate": 1.2498138679535728e-06,
+      "loss": 1.0824,
+      "step": 827
+    },
+    {
+      "epoch": 0.10719487328866881,
+      "grad_norm": 0.28162454729901676,
+      "learning_rate": 1.2498069107010563e-06,
+      "loss": 1.0842,
+      "step": 828
+    },
+    {
+      "epoch": 0.10732433569602227,
+      "grad_norm": 0.278988184754171,
+      "learning_rate": 1.2497998258190641e-06,
+      "loss": 1.1788,
+      "step": 829
+    },
+    {
+      "epoch": 0.10745379810337573,
+      "grad_norm": 0.20258414245353767,
+      "learning_rate": 1.2497926133090432e-06,
+      "loss": 1.0306,
+      "step": 830
+    },
+    {
+      "epoch": 0.10745379810337573,
+      "eval_PRM Accuracy": 0.4811320754716981,
+      "eval_PRM F1": 0.5736434108527132,
+      "eval_PRM F1 AUC": 0.5272393923520168,
+      "eval_PRM F1 Neg": 0.3373493975903614,
+      "eval_PRM NPV": 0.23333333333333334,
+      "eval_PRM Precision": 0.8043478260869565,
+      "eval_PRM Recall": 0.4457831325301205,
+      "eval_PRM Specificty": 0.6086956521739131,
+      "eval_loss": 0.7414772510528564,
+      "eval_runtime": 20.1971,
+      "eval_samples_per_second": 2.179,
+      "eval_steps_per_second": 0.149,
+      "step": 830
+    },
+    {
+      "epoch": 0.1075832605107292,
+      "grad_norm": 0.29352504982027333,
+      "learning_rate": 1.2497852731724673e-06,
+      "loss": 1.1937,
+      "step": 831
+    },
+    {
+      "epoch": 0.10771272291808266,
+      "grad_norm": 0.1997199394712332,
+      "learning_rate": 1.2497778054108353e-06,
+      "loss": 1.0553,
+      "step": 832
+    },
+    {
+      "epoch": 0.10784218532543613,
+      "grad_norm": 0.20385573810979046,
+      "learning_rate": 1.2497702100256732e-06,
+      "loss": 1.1356,
+      "step": 833
+    },
+    {
+      "epoch": 0.1079716477327896,
+      "grad_norm": 0.2905650732021906,
+      "learning_rate": 1.2497624870185319e-06,
+      "loss": 1.1574,
+      "step": 834
+    },
+    {
+      "epoch": 0.10810111014014305,
+      "grad_norm": 0.24331090516914258,
+      "learning_rate": 1.2497546363909894e-06,
+      "loss": 1.1205,
+      "step": 835
+    },
+    {
+      "epoch": 0.10810111014014305,
+      "eval_PRM Accuracy": 0.4811320754716981,
+      "eval_PRM F1": 0.5736434108527132,
+      "eval_PRM F1 AUC": 0.5272393923520168,
+      "eval_PRM F1 Neg": 0.3373493975903614,
+      "eval_PRM NPV": 0.23333333333333334,
+      "eval_PRM Precision": 0.8043478260869565,
+      "eval_PRM Recall": 0.4457831325301205,
+      "eval_PRM Specificty": 0.6086956521739131,
+      "eval_loss": 0.7397016882896423,
+      "eval_runtime": 20.496,
+      "eval_samples_per_second": 2.147,
+      "eval_steps_per_second": 0.146,
+      "step": 835
+    },
+    {
+      "epoch": 0.10823057254749652,
+      "grad_norm": 0.22858775433161402,
+      "learning_rate": 1.249746658144649e-06,
+      "loss": 1.186,
+      "step": 836
+    },
+    {
+      "epoch": 0.10836003495484998,
+      "grad_norm": 0.21026192530249205,
+      "learning_rate": 1.2497385522811408e-06,
+      "loss": 1.179,
+      "step": 837
+    },
+    {
+      "epoch": 0.10848949736220345,
+      "grad_norm": 0.24474330888254284,
+      "learning_rate": 1.2497303188021205e-06,
+      "loss": 0.9609,
+      "step": 838
+    },
+    {
+      "epoch": 0.10861895976955692,
+      "grad_norm": 0.25352881549065953,
+      "learning_rate": 1.2497219577092696e-06,
+      "loss": 1.0745,
+      "step": 839
+    },
+    {
+      "epoch": 0.10874842217691039,
+      "grad_norm": 0.24550251550489754,
+      "learning_rate": 1.2497134690042965e-06,
+      "loss": 1.1367,
+      "step": 840
+    },
+    {
+      "epoch": 0.10874842217691039,
+      "eval_PRM Accuracy": 0.4811320754716981,
+      "eval_PRM F1": 0.5736434108527132,
+      "eval_PRM F1 AUC": 0.5272393923520168,
+      "eval_PRM F1 Neg": 0.3373493975903614,
+      "eval_PRM NPV": 0.23333333333333334,
+      "eval_PRM Precision": 0.8043478260869565,
+      "eval_PRM Recall": 0.4457831325301205,
+      "eval_PRM Specificty": 0.6086956521739131,
+      "eval_loss": 0.7350852489471436,
+      "eval_runtime": 20.6361,
+      "eval_samples_per_second": 2.132,
+      "eval_steps_per_second": 0.145,
+      "step": 840
+    },
+    {
+      "epoch": 0.10887788458426384,
+      "grad_norm": 0.26855208114368556,
+      "learning_rate": 1.249704852688935e-06,
+      "loss": 1.0919,
+      "step": 841
+    },
+    {
+      "epoch": 0.10900734699161731,
+      "grad_norm": 0.2451687749708207,
+      "learning_rate": 1.249696108764945e-06,
+      "loss": 1.0544,
+      "step": 842
+    },
+    {
+      "epoch": 0.10913680939897077,
+      "grad_norm": 0.215728239693382,
+      "learning_rate": 1.249687237234113e-06,
+      "loss": 1.0809,
+      "step": 843
+    },
+    {
+      "epoch": 0.10926627180632424,
+      "grad_norm": 0.174006739010851,
+      "learning_rate": 1.2496782380982509e-06,
+      "loss": 1.108,
+      "step": 844
+    },
+    {
+      "epoch": 0.10939573421367771,
+      "grad_norm": 0.18644207525473072,
+      "learning_rate": 1.249669111359197e-06,
+      "loss": 0.9909,
+      "step": 845
+    },
+    {
+      "epoch": 0.10939573421367771,
+      "eval_PRM Accuracy": 0.49056603773584906,
+      "eval_PRM F1": 0.5846153846153846,
+      "eval_PRM F1 AUC": 0.5332634887375589,
+      "eval_PRM F1 Neg": 0.34146341463414637,
+      "eval_PRM NPV": 0.23728813559322035,
+      "eval_PRM Precision": 0.8085106382978723,
+      "eval_PRM Recall": 0.4578313253012048,
+      "eval_PRM Specificty": 0.6086956521739131,
+      "eval_loss": 0.734375,
+      "eval_runtime": 21.009,
+      "eval_samples_per_second": 2.094,
+      "eval_steps_per_second": 0.143,
+      "step": 845
+    },
+    {
+      "epoch": 0.10952519662103116,
+      "grad_norm": 0.26419616421367637,
+      "learning_rate": 1.2496598570188156e-06,
+      "loss": 1.1373,
+      "step": 846
+    },
+    {
+      "epoch": 0.10965465902838463,
+      "grad_norm": 0.20390310198513964,
+      "learning_rate": 1.249650475078997e-06,
+      "loss": 0.9963,
+      "step": 847
+    },
+    {
+      "epoch": 0.1097841214357381,
+      "grad_norm": 0.29986196761336015,
+      "learning_rate": 1.249640965541658e-06,
+      "loss": 1.0739,
+      "step": 848
+    },
+    {
+      "epoch": 0.10991358384309156,
+      "grad_norm": 0.20405694232259175,
+      "learning_rate": 1.2496313284087408e-06,
+      "loss": 1.0385,
+      "step": 849
+    },
+    {
+      "epoch": 0.11004304625044503,
+      "grad_norm": 0.13756837749289258,
+      "learning_rate": 1.249621563682214e-06,
+      "loss": 0.9885,
+      "step": 850
+    },
+    {
+      "epoch": 0.11004304625044503,
+      "eval_PRM Accuracy": 0.49056603773584906,
+      "eval_PRM F1": 0.5846153846153846,
+      "eval_PRM F1 AUC": 0.5332634887375589,
+      "eval_PRM F1 Neg": 0.34146341463414637,
+      "eval_PRM NPV": 0.23728813559322035,
+      "eval_PRM Precision": 0.8085106382978723,
+      "eval_PRM Recall": 0.4578313253012048,
+      "eval_PRM Specificty": 0.6086956521739131,
+      "eval_loss": 0.7320667505264282,
+      "eval_runtime": 20.0761,
+      "eval_samples_per_second": 2.192,
+      "eval_steps_per_second": 0.149,
+      "step": 850
+    },
+    {
+      "epoch": 0.1101725086577985,
+      "grad_norm": 0.273805488686973,
+      "learning_rate": 1.2496116713640725e-06,
+      "loss": 1.1967,
+      "step": 851
+    },
+    {
+      "epoch": 0.11030197106515195,
+      "grad_norm": 0.3046452795490258,
+      "learning_rate": 1.2496016514563367e-06,
+      "loss": 1.1144,
+      "step": 852
+    },
+    {
+      "epoch": 0.11043143347250542,
+      "grad_norm": 0.2497664350945725,
+      "learning_rate": 1.2495915039610534e-06,
+      "loss": 1.1184,
+      "step": 853
+    },
+    {
+      "epoch": 0.11056089587985889,
+      "grad_norm": 0.18914970103280176,
+      "learning_rate": 1.2495812288802959e-06,
+      "loss": 1.0773,
+      "step": 854
+    },
+    {
+      "epoch": 0.11069035828721235,
+      "grad_norm": 0.2892537061446785,
+      "learning_rate": 1.2495708262161623e-06,
+      "loss": 1.0602,
+      "step": 855
+    },
+    {
+      "epoch": 0.11069035828721235,
+      "eval_PRM Accuracy": 0.49056603773584906,
+      "eval_PRM F1": 0.5846153846153846,
+      "eval_PRM F1 AUC": 0.5332634887375589,
+      "eval_PRM F1 Neg": 0.34146341463414637,
+      "eval_PRM NPV": 0.23728813559322035,
+      "eval_PRM Precision": 0.8085106382978723,
+      "eval_PRM Recall": 0.4578313253012048,
+      "eval_PRM Specificty": 0.6086956521739131,
+      "eval_loss": 0.7294034361839294,
+      "eval_runtime": 20.1678,
+      "eval_samples_per_second": 2.182,
+      "eval_steps_per_second": 0.149,
+      "step": 855
+    },
+    {
+      "epoch": 0.11081982069456582,
+      "grad_norm": 0.15588820046320429,
+      "learning_rate": 1.249560295970778e-06,
+      "loss": 1.0277,
+      "step": 856
+    },
+    {
+      "epoch": 0.11094928310191927,
+      "grad_norm": 0.1633100810507839,
+      "learning_rate": 1.2495496381462942e-06,
+      "loss": 1.0233,
+      "step": 857
+    },
+    {
+      "epoch": 0.11107874550927274,
+      "grad_norm": 0.22617649148387292,
+      "learning_rate": 1.2495388527448875e-06,
+      "loss": 1.0326,
+      "step": 858
+    },
+    {
+      "epoch": 0.11120820791662621,
+      "grad_norm": 0.31475706897809314,
+      "learning_rate": 1.2495279397687615e-06,
+      "loss": 1.1859,
+      "step": 859
+    },
+    {
+      "epoch": 0.11133767032397968,
+      "grad_norm": 0.18936347831859027,
+      "learning_rate": 1.2495168992201454e-06,
+      "loss": 1.0613,
+      "step": 860
+    },
+    {
+      "epoch": 0.11133767032397968,
+      "eval_PRM Accuracy": 0.49056603773584906,
+      "eval_PRM F1": 0.5846153846153846,
+      "eval_PRM F1 AUC": 0.5332634887375589,
+      "eval_PRM F1 Neg": 0.34146341463414637,
+      "eval_PRM NPV": 0.23728813559322035,
+      "eval_PRM Precision": 0.8085106382978723,
+      "eval_PRM Recall": 0.4578313253012048,
+      "eval_PRM Specificty": 0.6086956521739131,
+      "eval_loss": 0.7286931872367859,
+      "eval_runtime": 19.7793,
+      "eval_samples_per_second": 2.225,
+      "eval_steps_per_second": 0.152,
+      "step": 860
+    },
+    {
+      "epoch": 0.11146713273133314,
+      "grad_norm": 0.35337764522004383,
+      "learning_rate": 1.249505731101294e-06,
+      "loss": 1.1591,
+      "step": 861
+    },
+    {
+      "epoch": 0.11159659513868661,
+      "grad_norm": 0.14762819877439,
+      "learning_rate": 1.249494435414489e-06,
+      "loss": 1.0142,
+      "step": 862
+    },
+    {
+      "epoch": 0.11172605754604006,
+      "grad_norm": 0.13807941149101205,
+      "learning_rate": 1.2494830121620375e-06,
+      "loss": 1.2063,
+      "step": 863
+    },
+    {
+      "epoch": 0.11185551995339353,
+      "grad_norm": 0.224716573207089,
+      "learning_rate": 1.249471461346273e-06,
+      "loss": 0.9923,
+      "step": 864
+    },
+    {
+      "epoch": 0.111984982360747,
+      "grad_norm": 0.24679079063618864,
+      "learning_rate": 1.249459782969555e-06,
+      "loss": 1.0436,
+      "step": 865
+    },
+    {
+      "epoch": 0.111984982360747,
+      "eval_PRM Accuracy": 0.49056603773584906,
+      "eval_PRM F1": 0.5846153846153846,
+      "eval_PRM F1 AUC": 0.5332634887375589,
+      "eval_PRM F1 Neg": 0.34146341463414637,
+      "eval_PRM NPV": 0.23728813559322035,
+      "eval_PRM Precision": 0.8085106382978723,
+      "eval_PRM Recall": 0.4578313253012048,
+      "eval_PRM Specificty": 0.6086956521739131,
+      "eval_loss": 0.7262073755264282,
+      "eval_runtime": 20.2977,
+      "eval_samples_per_second": 2.168,
+      "eval_steps_per_second": 0.148,
+      "step": 865
+    },
+    {
+      "epoch": 0.11211444476810047,
+      "grad_norm": 0.15828680223488442,
+      "learning_rate": 1.2494479770342694e-06,
+      "loss": 0.9651,
+      "step": 866
+    },
+    {
+      "epoch": 0.11224390717545393,
+      "grad_norm": 0.23652712546802734,
+      "learning_rate": 1.2494360435428275e-06,
+      "loss": 1.0587,
+      "step": 867
+    },
+    {
+      "epoch": 0.11237336958280739,
+      "grad_norm": 0.14304318763758858,
+      "learning_rate": 1.2494239824976668e-06,
+      "loss": 0.96,
+      "step": 868
+    },
+    {
+      "epoch": 0.11250283199016085,
+      "grad_norm": 0.2542128080504395,
+      "learning_rate": 1.249411793901251e-06,
+      "loss": 1.0457,
+      "step": 869
+    },
+    {
+      "epoch": 0.11263229439751432,
+      "grad_norm": 0.2633721293475217,
+      "learning_rate": 1.2493994777560703e-06,
+      "loss": 1.0123,
+      "step": 870
+    },
+    {
+      "epoch": 0.11263229439751432,
+      "eval_PRM Accuracy": 0.5,
+      "eval_PRM F1": 0.5954198473282443,
+      "eval_PRM F1 AUC": 0.5392875851231012,
+      "eval_PRM F1 Neg": 0.345679012345679,
+      "eval_PRM NPV": 0.2413793103448276,
+      "eval_PRM Precision": 0.8125,
+      "eval_PRM Recall": 0.46987951807228917,
+      "eval_PRM Specificty": 0.6086956521739131,
+      "eval_loss": 0.7219460010528564,
+      "eval_runtime": 20.3132,
+      "eval_samples_per_second": 2.166,
+      "eval_steps_per_second": 0.148,
+      "step": 870
+    },
+    {
+      "epoch": 0.11276175680486779,
+      "grad_norm": 0.24899244204212426,
+      "learning_rate": 1.2493870340646403e-06,
+      "loss": 1.0698,
+      "step": 871
+    },
+    {
+      "epoch": 0.11289121921222126,
+      "grad_norm": 0.29978812674319827,
+      "learning_rate": 1.2493744628295027e-06,
+      "loss": 1.1271,
+      "step": 872
+    },
+    {
+      "epoch": 0.11302068161957472,
+      "grad_norm": 0.19580760823087529,
+      "learning_rate": 1.2493617640532257e-06,
+      "loss": 1.036,
+      "step": 873
+    },
+    {
+      "epoch": 0.11315014402692818,
+      "grad_norm": 0.2697965784255595,
+      "learning_rate": 1.2493489377384029e-06,
+      "loss": 1.1612,
+      "step": 874
+    },
+    {
+      "epoch": 0.11327960643428164,
+      "grad_norm": 0.2119424022794168,
+      "learning_rate": 1.249335983887655e-06,
+      "loss": 1.0673,
+      "step": 875
+    },
+    {
+      "epoch": 0.11327960643428164,
+      "eval_PRM Accuracy": 0.5,
+      "eval_PRM F1": 0.5954198473282443,
+      "eval_PRM F1 AUC": 0.5392875851231012,
+      "eval_PRM F1 Neg": 0.345679012345679,
+      "eval_PRM NPV": 0.2413793103448276,
+      "eval_PRM Precision": 0.8125,
+      "eval_PRM Recall": 0.46987951807228917,
+      "eval_PRM Specificty": 0.6086956521739131,
+      "eval_loss": 0.7215909361839294,
+      "eval_runtime": 20.7457,
+      "eval_samples_per_second": 2.121,
+      "eval_steps_per_second": 0.145,
+      "step": 875
+    },
+    {
+      "epoch": 0.11340906884163511,
+      "grad_norm": 0.24551085932339028,
+      "learning_rate": 1.2493229025036275e-06,
+      "loss": 1.0796,
+      "step": 876
+    },
+    {
+      "epoch": 0.11353853124898858,
+      "grad_norm": 0.25908531934303725,
+      "learning_rate": 1.2493096935889926e-06,
+      "loss": 1.0904,
+      "step": 877
+    },
+    {
+      "epoch": 0.11366799365634205,
+      "grad_norm": 0.28665239622251376,
+      "learning_rate": 1.2492963571464487e-06,
+      "loss": 1.1184,
+      "step": 878
+    },
+    {
+      "epoch": 0.1137974560636955,
+      "grad_norm": 0.23391033561744506,
+      "learning_rate": 1.2492828931787199e-06,
+      "loss": 1.0857,
+      "step": 879
+    },
+    {
+      "epoch": 0.11392691847104897,
+      "grad_norm": 0.21241735524506247,
+      "learning_rate": 1.2492693016885564e-06,
+      "loss": 1.024,
+      "step": 880
+    },
+    {
+      "epoch": 0.11392691847104897,
+      "eval_PRM Accuracy": 0.5094339622641509,
+      "eval_PRM F1": 0.6060606060606061,
+      "eval_PRM F1 AUC": 0.5453116815086433,
+      "eval_PRM F1 Neg": 0.35,
+      "eval_PRM NPV": 0.24561403508771928,
+      "eval_PRM Precision": 0.8163265306122449,
+      "eval_PRM Recall": 0.4819277108433735,
+      "eval_PRM Specificty": 0.6086956521739131,
+      "eval_loss": 0.71875,
+      "eval_runtime": 20.2351,
+      "eval_samples_per_second": 2.174,
+      "eval_steps_per_second": 0.148,
+      "step": 880
+    },
+    {
+      "epoch": 0.11405638087840243,
+      "grad_norm": 0.11419247987681164,
+      "learning_rate": 1.249255582678735e-06,
+      "loss": 0.8253,
+      "step": 881
+    },
+    {
+      "epoch": 0.1141858432857559,
+      "grad_norm": 0.30542636755880465,
+      "learning_rate": 1.2492417361520575e-06,
+      "loss": 1.1077,
+      "step": 882
+    },
+    {
+      "epoch": 0.11431530569310937,
+      "grad_norm": 0.2056452779904224,
+      "learning_rate": 1.2492277621113528e-06,
+      "loss": 0.9579,
+      "step": 883
+    },
+    {
+      "epoch": 0.11444476810046283,
+      "grad_norm": 0.10895236912331824,
+      "learning_rate": 1.249213660559475e-06,
+      "loss": 1.0045,
+      "step": 884
+    },
+    {
+      "epoch": 0.11457423050781629,
+      "grad_norm": 0.24844699381788002,
+      "learning_rate": 1.249199431499305e-06,
+      "loss": 1.1223,
+      "step": 885
+    },
+    {
+      "epoch": 0.11457423050781629,
+      "eval_PRM Accuracy": 0.5188679245283019,
+      "eval_PRM F1": 0.6165413533834586,
+      "eval_PRM F1 AUC": 0.5513357778941854,
+      "eval_PRM F1 Neg": 0.35443037974683544,
+      "eval_PRM NPV": 0.25,
+      "eval_PRM Precision": 0.82,
+      "eval_PRM Recall": 0.4939759036144578,
+      "eval_PRM Specificty": 0.6086956521739131,
+      "eval_loss": 0.7164417505264282,
+      "eval_runtime": 20.2647,
+      "eval_samples_per_second": 2.171,
+      "eval_steps_per_second": 0.148,
+      "step": 885
+    },
+    {
+      "epoch": 0.11470369291516976,
+      "grad_norm": 0.2318722761563911,
+      "learning_rate": 1.249185074933749e-06,
+      "loss": 0.9695,
+      "step": 886
+    },
+    {
+      "epoch": 0.11483315532252322,
+      "grad_norm": 0.2771180603843615,
+      "learning_rate": 1.2491705908657395e-06,
+      "loss": 1.1239,
+      "step": 887
+    },
+    {
+      "epoch": 0.11496261772987669,
+      "grad_norm": 0.26883115435178806,
+      "learning_rate": 1.2491559792982358e-06,
+      "loss": 1.0718,
+      "step": 888
+    },
+    {
+      "epoch": 0.11509208013723016,
+      "grad_norm": 0.2549570288540099,
+      "learning_rate": 1.2491412402342223e-06,
+      "loss": 1.038,
+      "step": 889
+    },
+    {
+      "epoch": 0.11522154254458361,
+      "grad_norm": 0.2112448201230687,
+      "learning_rate": 1.2491263736767095e-06,
+      "loss": 0.9882,
+      "step": 890
+    },
+    {
+      "epoch": 0.11522154254458361,
+      "eval_PRM Accuracy": 0.5094339622641509,
+      "eval_PRM F1": 0.6060606060606061,
+      "eval_PRM F1 AUC": 0.5453116815086433,
+      "eval_PRM F1 Neg": 0.35,
+      "eval_PRM NPV": 0.24561403508771928,
+      "eval_PRM Precision": 0.8163265306122449,
+      "eval_PRM Recall": 0.4819277108433735,
+      "eval_PRM Specificty": 0.6086956521739131,
+      "eval_loss": 0.7144886255264282,
+      "eval_runtime": 20.4399,
+      "eval_samples_per_second": 2.153,
+      "eval_steps_per_second": 0.147,
+      "step": 890
+    },
+    {
+      "epoch": 0.11535100495193708,
+      "grad_norm": 0.2594065909746897,
+      "learning_rate": 1.2491113796287347e-06,
+      "loss": 1.1175,
+      "step": 891
+    },
+    {
+      "epoch": 0.11548046735929055,
+      "grad_norm": 0.25961743714135654,
+      "learning_rate": 1.2490962580933602e-06,
+      "loss": 1.0705,
+      "step": 892
+    },
+    {
+      "epoch": 0.11560992976664401,
+      "grad_norm": 0.17908845575530666,
+      "learning_rate": 1.2490810090736752e-06,
+      "loss": 1.0261,
+      "step": 893
+    },
+    {
+      "epoch": 0.11573939217399748,
+      "grad_norm": 0.31613936347505417,
+      "learning_rate": 1.2490656325727945e-06,
+      "loss": 1.1121,
+      "step": 894
+    },
+    {
+      "epoch": 0.11586885458135095,
+      "grad_norm": 0.2857287921852463,
+      "learning_rate": 1.2490501285938592e-06,
+      "loss": 1.1281,
+      "step": 895
+    },
+    {
+      "epoch": 0.11586885458135095,
+      "eval_PRM Accuracy": 0.5188679245283019,
+      "eval_PRM F1": 0.6165413533834586,
+      "eval_PRM F1 AUC": 0.5513357778941854,
+      "eval_PRM F1 Neg": 0.35443037974683544,
+      "eval_PRM NPV": 0.25,
+      "eval_PRM Precision": 0.82,
+      "eval_PRM Recall": 0.4939759036144578,
+      "eval_PRM Specificty": 0.6086956521739131,
+      "eval_loss": 0.7132457494735718,
+      "eval_runtime": 20.9549,
+      "eval_samples_per_second": 2.1,
+      "eval_steps_per_second": 0.143,
+      "step": 895
+    },
+    {
+      "epoch": 0.1159983169887044,
+      "grad_norm": 0.30321862569801306,
+      "learning_rate": 1.2490344971400363e-06,
+      "loss": 1.1357,
+      "step": 896
+    },
+    {
+      "epoch": 0.11612777939605787,
+      "grad_norm": 0.31183746712739363,
+      "learning_rate": 1.2490187382145185e-06,
+      "loss": 1.14,
+      "step": 897
+    },
+    {
+      "epoch": 0.11625724180341133,
+      "grad_norm": 0.1755978736987123,
+      "learning_rate": 1.2490028518205256e-06,
+      "loss": 0.9612,
+      "step": 898
+    },
+    {
+      "epoch": 0.1163867042107648,
+      "grad_norm": 0.3085330783048964,
+      "learning_rate": 1.2489868379613017e-06,
+      "loss": 0.9958,
+      "step": 899
+    },
+    {
+      "epoch": 0.11651616661811827,
+      "grad_norm": 0.18416313133057224,
+      "learning_rate": 1.248970696640119e-06,
+      "loss": 1.0742,
+      "step": 900
+    },
+    {
+      "epoch": 0.11651616661811827,
+      "eval_PRM Accuracy": 0.5377358490566038,
+      "eval_PRM F1": 0.6370370370370371,
+      "eval_PRM F1 AUC": 0.5633839706652698,
+      "eval_PRM F1 Neg": 0.36363636363636365,
+      "eval_PRM NPV": 0.25925925925925924,
+      "eval_PRM Precision": 0.8269230769230769,
+      "eval_PRM Recall": 0.5180722891566265,
+      "eval_PRM Specificty": 0.6086956521739131,
+      "eval_loss": 0.7082741260528564,
+      "eval_runtime": 20.3598,
+      "eval_samples_per_second": 2.161,
+      "eval_steps_per_second": 0.147,
+      "step": 900
+    },
+    {
+      "epoch": 0.11664562902547172,
+      "grad_norm": 0.12835841830653832,
+      "learning_rate": 1.248954427860274e-06,
+      "loss": 0.963,
+      "step": 901
+    },
+    {
+      "epoch": 0.11677509143282519,
+      "grad_norm": 0.20071122181022247,
+      "learning_rate": 1.2489380316250898e-06,
+      "loss": 1.0999,
+      "step": 902
+    },
+    {
+      "epoch": 0.11690455384017866,
+      "grad_norm": 0.2924281911983938,
+      "learning_rate": 1.2489215079379164e-06,
+      "loss": 0.9619,
+      "step": 903
+    },
+    {
+      "epoch": 0.11703401624753212,
+      "grad_norm": 0.18459876402155906,
+      "learning_rate": 1.2489048568021284e-06,
+      "loss": 1.0217,
+      "step": 904
+    },
+    {
+      "epoch": 0.11716347865488559,
+      "grad_norm": 0.24870341952160963,
+      "learning_rate": 1.2488880782211277e-06,
+      "loss": 0.9905,
+      "step": 905
+    },
+    {
+      "epoch": 0.11716347865488559,
+      "eval_PRM Accuracy": 0.5377358490566038,
+      "eval_PRM F1": 0.6370370370370371,
+      "eval_PRM F1 AUC": 0.5633839706652698,
+      "eval_PRM F1 Neg": 0.36363636363636365,
+      "eval_PRM NPV": 0.25925925925925924,
+      "eval_PRM Precision": 0.8269230769230769,
+      "eval_PRM Recall": 0.5180722891566265,
+      "eval_PRM Specificty": 0.6086956521739131,
+      "eval_loss": 0.7073863744735718,
+      "eval_runtime": 20.1407,
+      "eval_samples_per_second": 2.185,
+      "eval_steps_per_second": 0.149,
+      "step": 905
+    },
+    {
+      "epoch": 0.11729294106223906,
+      "grad_norm": 0.2943022227518921,
+      "learning_rate": 1.2488711721983413e-06,
+      "loss": 1.1802,
+      "step": 906
+    },
+    {
+      "epoch": 0.11742240346959251,
+      "grad_norm": 0.20927952420906956,
+      "learning_rate": 1.2488541387372226e-06,
+      "loss": 0.9427,
+      "step": 907
+    },
+    {
+      "epoch": 0.11755186587694598,
+      "grad_norm": 0.20045053771377452,
+      "learning_rate": 1.2488369778412513e-06,
+      "loss": 1.0686,
+      "step": 908
+    },
+    {
+      "epoch": 0.11768132828429945,
+      "grad_norm": 0.21925166112671532,
+      "learning_rate": 1.2488196895139323e-06,
+      "loss": 1.0447,
+      "step": 909
+    },
+    {
+      "epoch": 0.11781079069165291,
+      "grad_norm": 0.22803737349362252,
+      "learning_rate": 1.2488022737587978e-06,
+      "loss": 1.042,
+      "step": 910
+    },
+    {
+      "epoch": 0.11781079069165291,
+      "eval_PRM Accuracy": 0.5377358490566038,
+      "eval_PRM F1": 0.6370370370370371,
+      "eval_PRM F1 AUC": 0.5633839706652698,
+      "eval_PRM F1 Neg": 0.36363636363636365,
+      "eval_PRM NPV": 0.25925925925925924,
+      "eval_PRM Precision": 0.8269230769230769,
+      "eval_PRM Recall": 0.5180722891566265,
+      "eval_PRM Specificty": 0.6086956521739131,
+      "eval_loss": 0.7080965638160706,
+      "eval_runtime": 20.2607,
+      "eval_samples_per_second": 2.172,
+      "eval_steps_per_second": 0.148,
+      "step": 910
+    },
+    {
+      "epoch": 0.11794025309900638,
+      "grad_norm": 0.26257299652913973,
+      "learning_rate": 1.248784730579405e-06,
+      "loss": 1.1315,
+      "step": 911
+    },
+    {
+      "epoch": 0.11806971550635983,
+      "grad_norm": 0.26153336312290165,
+      "learning_rate": 1.2487670599793372e-06,
+      "loss": 1.0878,
+      "step": 912
+    },
+    {
+      "epoch": 0.1181991779137133,
+      "grad_norm": 0.22446569063354554,
+      "learning_rate": 1.2487492619622045e-06,
+      "loss": 1.0463,
+      "step": 913
+    },
+    {
+      "epoch": 0.11832864032106677,
+      "grad_norm": 0.308313813282149,
+      "learning_rate": 1.2487313365316419e-06,
+      "loss": 0.9839,
+      "step": 914
+    },
+    {
+      "epoch": 0.11845810272842024,
+      "grad_norm": 0.34442116697939235,
+      "learning_rate": 1.2487132836913117e-06,
+      "loss": 1.0446,
+      "step": 915
+    },
+    {
+      "epoch": 0.11845810272842024,
+      "eval_PRM Accuracy": 0.5283018867924528,
+      "eval_PRM F1": 0.6268656716417911,
+      "eval_PRM F1 AUC": 0.5573598742797277,
+      "eval_PRM F1 Neg": 0.358974358974359,
+      "eval_PRM NPV": 0.2545454545454545,
+      "eval_PRM Precision": 0.8235294117647058,
+      "eval_PRM Recall": 0.5060240963855421,
+      "eval_PRM Specificty": 0.6086956521739131,
+      "eval_loss": 0.7052556872367859,
+      "eval_runtime": 20.2169,
+      "eval_samples_per_second": 2.176,
+      "eval_steps_per_second": 0.148,
+      "step": 915
+    },
+    {
+      "epoch": 0.1185875651357737,
+      "grad_norm": 0.20284590608333603,
+      "learning_rate": 1.248695103444901e-06,
+      "loss": 1.0165,
+      "step": 916
+    },
+    {
+      "epoch": 0.11871702754312717,
+      "grad_norm": 0.24376948009314453,
+      "learning_rate": 1.2486767957961237e-06,
+      "loss": 1.0447,
+      "step": 917
+    },
+    {
+      "epoch": 0.11884648995048062,
+      "grad_norm": 0.2675687967412616,
+      "learning_rate": 1.2486583607487197e-06,
+      "loss": 1.0378,
+      "step": 918
+    },
+    {
+      "epoch": 0.11897595235783409,
+      "grad_norm": 0.1671036375348281,
+      "learning_rate": 1.2486397983064543e-06,
+      "loss": 1.0276,
+      "step": 919
+    },
+    {
+      "epoch": 0.11910541476518756,
+      "grad_norm": 0.29111974867403917,
+      "learning_rate": 1.2486211084731197e-06,
+      "loss": 1.1477,
+      "step": 920
+    },
+    {
+      "epoch": 0.11910541476518756,
+      "eval_PRM Accuracy": 0.5377358490566038,
+      "eval_PRM F1": 0.6370370370370371,
+      "eval_PRM F1 AUC": 0.5633839706652698,
+      "eval_PRM F1 Neg": 0.36363636363636365,
+      "eval_PRM NPV": 0.25925925925925924,
+      "eval_PRM Precision": 0.8269230769230769,
+      "eval_PRM Recall": 0.5180722891566265,
+      "eval_PRM Specificty": 0.6086956521739131,
+      "eval_loss": 0.7013494372367859,
+      "eval_runtime": 20.6924,
+      "eval_samples_per_second": 2.126,
+      "eval_steps_per_second": 0.145,
+      "step": 920
+    },
+    {
+      "epoch": 0.11923487717254103,
+      "grad_norm": 0.2927476002427883,
+      "learning_rate": 1.2486022912525333e-06,
+      "loss": 1.0867,
+      "step": 921
+    },
+    {
+      "epoch": 0.1193643395798945,
+      "grad_norm": 0.30445920229169054,
+      "learning_rate": 1.2485833466485392e-06,
+      "loss": 1.1169,
+      "step": 922
+    },
+    {
+      "epoch": 0.11949380198724795,
+      "grad_norm": 0.30580982272691565,
+      "learning_rate": 1.248564274665007e-06,
+      "loss": 1.0751,
+      "step": 923
+    },
+    {
+      "epoch": 0.11962326439460141,
+      "grad_norm": 0.31166415065880815,
+      "learning_rate": 1.2485450753058326e-06,
+      "loss": 1.1058,
+      "step": 924
+    },
+    {
+      "epoch": 0.11975272680195488,
+      "grad_norm": 0.11480287425920799,
+      "learning_rate": 1.248525748574938e-06,
+      "loss": 0.9757,
+      "step": 925
+    },
+    {
+      "epoch": 0.11975272680195488,
+      "eval_PRM Accuracy": 0.5377358490566038,
+      "eval_PRM F1": 0.6370370370370371,
+      "eval_PRM F1 AUC": 0.5633839706652698,
+      "eval_PRM F1 Neg": 0.36363636363636365,
+      "eval_PRM NPV": 0.25925925925925924,
+      "eval_PRM Precision": 0.8269230769230769,
+      "eval_PRM Recall": 0.5180722891566265,
+      "eval_PRM Specificty": 0.6086956521739131,
+      "eval_loss": 0.7013494372367859,
+      "eval_runtime": 20.3504,
+      "eval_samples_per_second": 2.162,
+      "eval_steps_per_second": 0.147,
+      "step": 925
+    },
+    {
+      "epoch": 0.11988218920930835,
+      "grad_norm": 0.24314245870276266,
+      "learning_rate": 1.2485062944762707e-06,
+      "loss": 0.9986,
+      "step": 926
+    },
+    {
+      "epoch": 0.12001165161666182,
+      "grad_norm": 0.2552476338883707,
+      "learning_rate": 1.2484867130138051e-06,
+      "loss": 1.0531,
+      "step": 927
+    },
+    {
+      "epoch": 0.12014111402401528,
+      "grad_norm": 0.2359204746157435,
+      "learning_rate": 1.2484670041915408e-06,
+      "loss": 1.1035,
+      "step": 928
+    },
+    {
+      "epoch": 0.12027057643136874,
+      "grad_norm": 0.3352930236383672,
+      "learning_rate": 1.248447168013504e-06,
+      "loss": 1.1707,
+      "step": 929
+    },
+    {
+      "epoch": 0.1204000388387222,
+      "grad_norm": 0.34336983341594346,
+      "learning_rate": 1.2484272044837462e-06,
+      "loss": 1.181,
+      "step": 930
+    },
+    {
+      "epoch": 0.1204000388387222,
+      "eval_PRM Accuracy": 0.5471698113207547,
+      "eval_PRM F1": 0.6470588235294118,
+      "eval_PRM F1 AUC": 0.5694080670508119,
+      "eval_PRM F1 Neg": 0.3684210526315789,
+      "eval_PRM NPV": 0.2641509433962264,
+      "eval_PRM Precision": 0.8301886792452831,
+      "eval_PRM Recall": 0.5301204819277109,
+      "eval_PRM Specificty": 0.6086956521739131,
+      "eval_loss": 0.7004616260528564,
+      "eval_runtime": 19.7231,
+      "eval_samples_per_second": 2.231,
+      "eval_steps_per_second": 0.152,
+      "step": 930
+    },
+    {
+      "epoch": 0.12052950124607567,
+      "grad_norm": 0.26507746551746175,
+      "learning_rate": 1.2484071136063456e-06,
+      "loss": 1.0639,
+      "step": 931
+    },
+    {
+      "epoch": 0.12065896365342914,
+      "grad_norm": 0.15430756125265027,
+      "learning_rate": 1.2483868953854061e-06,
+      "loss": 0.9623,
+      "step": 932
+    },
+    {
+      "epoch": 0.1207884260607826,
+      "grad_norm": 0.24730460456392214,
+      "learning_rate": 1.248366549825058e-06,
+      "loss": 1.0803,
+      "step": 933
+    },
+    {
+      "epoch": 0.12091788846813606,
+      "grad_norm": 0.17759719029207735,
+      "learning_rate": 1.248346076929457e-06,
+      "loss": 0.9493,
+      "step": 934
+    },
+    {
+      "epoch": 0.12104735087548953,
+      "grad_norm": 0.14805744054801356,
+      "learning_rate": 1.2483254767027848e-06,
+      "loss": 1.0398,
+      "step": 935
+    },
+    {
+      "epoch": 0.12104735087548953,
+      "eval_PRM Accuracy": 0.5471698113207547,
+      "eval_PRM F1": 0.6470588235294118,
+      "eval_PRM F1 AUC": 0.5694080670508119,
+      "eval_PRM F1 Neg": 0.3684210526315789,
+      "eval_PRM NPV": 0.2641509433962264,
+      "eval_PRM Precision": 0.8301886792452831,
+      "eval_PRM Recall": 0.5301204819277109,
+      "eval_PRM Specificty": 0.6086956521739131,
+      "eval_loss": 0.6970880627632141,
+      "eval_runtime": 20.4074,
+      "eval_samples_per_second": 2.156,
+      "eval_steps_per_second": 0.147,
+      "step": 935
+    },
+    {
+      "epoch": 0.121176813282843,
+      "grad_norm": 0.0975216483398078,
+      "learning_rate": 1.2483047491492501e-06,
+      "loss": 0.8837,
+      "step": 936
+    },
+    {
+      "epoch": 0.12130627569019646,
+      "grad_norm": 0.23511993554299532,
+      "learning_rate": 1.2482838942730862e-06,
+      "loss": 1.0808,
+      "step": 937
+    },
+    {
+      "epoch": 0.12143573809754993,
+      "grad_norm": 0.10027602407600866,
+      "learning_rate": 1.248262912078554e-06,
+      "loss": 0.9457,
+      "step": 938
+    },
+    {
+      "epoch": 0.1215652005049034,
+      "grad_norm": 0.3169258597939292,
+      "learning_rate": 1.2482418025699386e-06,
+      "loss": 1.0693,
+      "step": 939
+    },
+    {
+      "epoch": 0.12169466291225685,
+      "grad_norm": 0.29996150877527894,
+      "learning_rate": 1.2482205657515528e-06,
+      "loss": 0.9786,
+      "step": 940
+    },
+    {
+      "epoch": 0.12169466291225685,
+      "eval_PRM Accuracy": 0.5566037735849056,
+      "eval_PRM F1": 0.656934306569343,
+      "eval_PRM F1 AUC": 0.5754321634363542,
+      "eval_PRM F1 Neg": 0.37333333333333335,
+      "eval_PRM NPV": 0.2692307692307692,
+      "eval_PRM Precision": 0.8333333333333334,
+      "eval_PRM Recall": 0.5421686746987951,
+      "eval_PRM Specificty": 0.6086956521739131,
+      "eval_loss": 0.6942471861839294,
+      "eval_runtime": 19.9488,
+      "eval_samples_per_second": 2.206,
+      "eval_steps_per_second": 0.15,
+      "step": 940
+    },
+    {
+      "epoch": 0.12182412531961032,
+      "grad_norm": 0.22453547614745933,
+      "learning_rate": 1.2481992016277341e-06,
+      "loss": 1.0378,
+      "step": 941
+    },
+    {
+      "epoch": 0.12195358772696378,
+      "grad_norm": 0.21006845986473938,
+      "learning_rate": 1.248177710202847e-06,
+      "loss": 1.1132,
+      "step": 942
+    },
+    {
+      "epoch": 0.12208305013431725,
+      "grad_norm": 0.15992988524757043,
+      "learning_rate": 1.2481560914812812e-06,
+      "loss": 1.101,
+      "step": 943
+    },
+    {
+      "epoch": 0.12221251254167072,
+      "grad_norm": 0.22740313189887396,
+      "learning_rate": 1.248134345467453e-06,
+      "loss": 1.1594,
+      "step": 944
+    },
+    {
+      "epoch": 0.12234197494902417,
+      "grad_norm": 0.28398594523294257,
+      "learning_rate": 1.2481124721658043e-06,
+      "loss": 0.9941,
+      "step": 945
+    },
+    {
+      "epoch": 0.12234197494902417,
+      "eval_PRM Accuracy": 0.5566037735849056,
+      "eval_PRM F1": 0.656934306569343,
+      "eval_PRM F1 AUC": 0.5754321634363542,
+      "eval_PRM F1 Neg": 0.37333333333333335,
+      "eval_PRM NPV": 0.2692307692307692,
+      "eval_PRM Precision": 0.8333333333333334,
+      "eval_PRM Recall": 0.5421686746987951,
+      "eval_PRM Specificty": 0.6086956521739131,
+      "eval_loss": 0.6922940611839294,
+      "eval_runtime": 20.4976,
+      "eval_samples_per_second": 2.147,
+      "eval_steps_per_second": 0.146,
+      "step": 945
+    },
+    {
+      "epoch": 0.12247143735637764,
+      "grad_norm": 0.28974589070997037,
+      "learning_rate": 1.2480904715808035e-06,
+      "loss": 1.1002,
+      "step": 946
+    },
+    {
+      "epoch": 0.1226008997637311,
+      "grad_norm": 0.17185736104411076,
+      "learning_rate": 1.2480683437169441e-06,
+      "loss": 0.9274,
+      "step": 947
+    },
+    {
+      "epoch": 0.12273036217108457,
+      "grad_norm": 0.3051407108944278,
+      "learning_rate": 1.2480460885787465e-06,
+      "loss": 1.0024,
+      "step": 948
+    },
+    {
+      "epoch": 0.12285982457843804,
+      "grad_norm": 0.1938359421159141,
+      "learning_rate": 1.248023706170757e-06,
+      "loss": 1.0719,
+      "step": 949
+    },
+    {
+      "epoch": 0.12298928698579151,
+      "grad_norm": 0.2642402308896572,
+      "learning_rate": 1.2480011964975472e-06,
+      "loss": 1.0945,
+      "step": 950
+    },
+    {
+      "epoch": 0.12298928698579151,
+      "eval_PRM Accuracy": 0.5566037735849056,
+      "eval_PRM F1": 0.656934306569343,
+      "eval_PRM F1 AUC": 0.5754321634363542,
+      "eval_PRM F1 Neg": 0.37333333333333335,
+      "eval_PRM NPV": 0.2692307692307692,
+      "eval_PRM Precision": 0.8333333333333334,
+      "eval_PRM Recall": 0.5421686746987951,
+      "eval_PRM Specificty": 0.6086956521739131,
+      "eval_loss": 0.6942471861839294,
+      "eval_runtime": 20.4898,
+      "eval_samples_per_second": 2.147,
+      "eval_steps_per_second": 0.146,
+      "step": 950
+    },
+    {
+      "epoch": 0.12311874939314496,
+      "grad_norm": 0.34714050386872264,
+      "learning_rate": 1.2479785595637152e-06,
+      "loss": 1.0513,
+      "step": 951
+    },
+    {
+      "epoch": 0.12324821180049843,
+      "grad_norm": 0.26637826202424403,
+      "learning_rate": 1.2479557953738854e-06,
+      "loss": 0.9675,
+      "step": 952
+    },
+    {
+      "epoch": 0.1233776742078519,
+      "grad_norm": 0.2010276621390886,
+      "learning_rate": 1.2479329039327078e-06,
+      "loss": 1.0109,
+      "step": 953
+    },
+    {
+      "epoch": 0.12350713661520536,
+      "grad_norm": 0.2287827383786289,
+      "learning_rate": 1.247909885244858e-06,
+      "loss": 1.0419,
+      "step": 954
+    },
+    {
+      "epoch": 0.12363659902255883,
+      "grad_norm": 0.1987488490791542,
+      "learning_rate": 1.2478867393150387e-06,
+      "loss": 0.8956,
+      "step": 955
+    },
+    {
+      "epoch": 0.12363659902255883,
+      "eval_PRM Accuracy": 0.5566037735849056,
+      "eval_PRM F1": 0.656934306569343,
+      "eval_PRM F1 AUC": 0.5754321634363542,
+      "eval_PRM F1 Neg": 0.37333333333333335,
+      "eval_PRM NPV": 0.2692307692307692,
+      "eval_PRM Precision": 0.8333333333333334,
+      "eval_PRM Recall": 0.5421686746987951,
+      "eval_PRM Specificty": 0.6086956521739131,
+      "eval_loss": 0.6905184388160706,
+      "eval_runtime": 20.2499,
+      "eval_samples_per_second": 2.173,
+      "eval_steps_per_second": 0.148,
+      "step": 955
+    },
+    {
+      "epoch": 0.12376606142991228,
+      "grad_norm": 0.25799023415568756,
+      "learning_rate": 1.247863466147977e-06,
+      "loss": 1.027,
+      "step": 956
+    },
+    {
+      "epoch": 0.12389552383726575,
+      "grad_norm": 0.17114997717982633,
+      "learning_rate": 1.247840065748428e-06,
+      "loss": 1.0116,
+      "step": 957
+    },
+    {
+      "epoch": 0.12402498624461922,
+      "grad_norm": 0.24403506533486202,
+      "learning_rate": 1.247816538121171e-06,
+      "loss": 1.0424,
+      "step": 958
+    },
+    {
+      "epoch": 0.12415444865197268,
+      "grad_norm": 0.2641757358370001,
+      "learning_rate": 1.247792883271012e-06,
+      "loss": 1.0206,
+      "step": 959
+    },
+    {
+      "epoch": 0.12428391105932615,
+      "grad_norm": 0.28572523781138776,
+      "learning_rate": 1.2477691012027835e-06,
+      "loss": 1.1404,
+      "step": 960
+    },
+    {
+      "epoch": 0.12428391105932615,
+      "eval_PRM Accuracy": 0.5566037735849056,
+      "eval_PRM F1": 0.656934306569343,
+      "eval_PRM F1 AUC": 0.5754321634363542,
+      "eval_PRM F1 Neg": 0.37333333333333335,
+      "eval_PRM NPV": 0.2692307692307692,
+      "eval_PRM Precision": 0.8333333333333334,
+      "eval_PRM Recall": 0.5421686746987951,
+      "eval_PRM Specificty": 0.6086956521739131,
+      "eval_loss": 0.6885653138160706,
+      "eval_runtime": 20.36,
+      "eval_samples_per_second": 2.161,
+      "eval_steps_per_second": 0.147,
+      "step": 960
+    },
+    {
+      "epoch": 0.12441337346667962,
+      "grad_norm": 0.16101329417132618,
+      "learning_rate": 1.247745191921343e-06,
+      "loss": 1.0167,
+      "step": 961
+    },
+    {
+      "epoch": 0.12454283587403307,
+      "grad_norm": 0.23178472803206407,
+      "learning_rate": 1.2477211554315744e-06,
+      "loss": 0.9841,
+      "step": 962
+    },
+    {
+      "epoch": 0.12467229828138654,
+      "grad_norm": 0.28990005171684535,
+      "learning_rate": 1.247696991738388e-06,
+      "loss": 1.0898,
+      "step": 963
+    },
+    {
+      "epoch": 0.12480176068874001,
+      "grad_norm": 0.2406609874790822,
+      "learning_rate": 1.2476727008467195e-06,
+      "loss": 1.1318,
+      "step": 964
+    },
+    {
+      "epoch": 0.12493122309609347,
+      "grad_norm": 0.23889792071952662,
+      "learning_rate": 1.2476482827615308e-06,
+      "loss": 1.1131,
+      "step": 965
+    },
+    {
+      "epoch": 0.12493122309609347,
+      "eval_PRM Accuracy": 0.5566037735849056,
+      "eval_PRM F1": 0.656934306569343,
+      "eval_PRM F1 AUC": 0.5754321634363542,
+      "eval_PRM F1 Neg": 0.37333333333333335,
+      "eval_PRM NPV": 0.2692307692307692,
+      "eval_PRM Precision": 0.8333333333333334,
+      "eval_PRM Recall": 0.5421686746987951,
+      "eval_PRM Specificty": 0.6086956521739131,
+      "eval_loss": 0.685546875,
+      "eval_runtime": 20.0248,
+      "eval_samples_per_second": 2.197,
+      "eval_steps_per_second": 0.15,
+      "step": 965
+    },
+    {
+      "epoch": 0.12506068550344693,
+      "grad_norm": 0.2867318832438925,
+      "learning_rate": 1.2476237374878098e-06,
+      "loss": 1.114,
+      "step": 966
+    },
+    {
+      "epoch": 0.1251901479108004,
+      "grad_norm": 0.24665522425255476,
+      "learning_rate": 1.2475990650305705e-06,
+      "loss": 1.0485,
+      "step": 967
+    },
+    {
+      "epoch": 0.12531961031815386,
+      "grad_norm": 0.2342570204432168,
+      "learning_rate": 1.2475742653948527e-06,
+      "loss": 1.048,
+      "step": 968
+    },
+    {
+      "epoch": 0.12544907272550734,
+      "grad_norm": 0.2524333252167745,
+      "learning_rate": 1.2475493385857222e-06,
+      "loss": 0.9614,
+      "step": 969
+    },
+    {
+      "epoch": 0.1255785351328608,
+      "grad_norm": 0.21165199025095557,
+      "learning_rate": 1.2475242846082706e-06,
+      "loss": 1.1005,
+      "step": 970
+    },
+    {
+      "epoch": 0.1255785351328608,
+      "eval_PRM Accuracy": 0.5660377358490566,
+      "eval_PRM F1": 0.6666666666666666,
+      "eval_PRM F1 AUC": 0.5814562598218963,
+      "eval_PRM F1 Neg": 0.3783783783783784,
+      "eval_PRM NPV": 0.27450980392156865,
+      "eval_PRM Precision": 0.8363636363636363,
+      "eval_PRM Recall": 0.5542168674698795,
+      "eval_PRM Specificty": 0.6086956521739131,
+      "eval_loss": 0.6850141882896423,
+      "eval_runtime": 20.3144,
+      "eval_samples_per_second": 2.166,
+      "eval_steps_per_second": 0.148,
+      "step": 970
+    },
+    {
+      "epoch": 0.12570799754021425,
+      "grad_norm": 0.13222001672625228,
+      "learning_rate": 1.247499103467616e-06,
+      "loss": 0.9894,
+      "step": 971
+    },
+    {
+      "epoch": 0.12583745994756773,
+      "grad_norm": 0.22312405691013296,
+      "learning_rate": 1.247473795168902e-06,
+      "loss": 1.0591,
+      "step": 972
+    },
+    {
+      "epoch": 0.12596692235492118,
+      "grad_norm": 0.2653192637392524,
+      "learning_rate": 1.2474483597172985e-06,
+      "loss": 0.9784,
+      "step": 973
+    },
+    {
+      "epoch": 0.12609638476227467,
+      "grad_norm": 0.24917799582714162,
+      "learning_rate": 1.247422797118001e-06,
+      "loss": 0.9749,
+      "step": 974
+    },
+    {
+      "epoch": 0.12622584716962812,
+      "grad_norm": 0.11582279268731416,
+      "learning_rate": 1.2473971073762312e-06,
+      "loss": 0.9305,
+      "step": 975
+    },
+    {
+      "epoch": 0.12622584716962812,
+      "eval_PRM Accuracy": 0.5566037735849056,
+      "eval_PRM F1": 0.656934306569343,
+      "eval_PRM F1 AUC": 0.5754321634363542,
+      "eval_PRM F1 Neg": 0.37333333333333335,
+      "eval_PRM NPV": 0.2692307692307692,
+      "eval_PRM Precision": 0.8333333333333334,
+      "eval_PRM Recall": 0.5421686746987951,
+      "eval_PRM Specificty": 0.6086956521739131,
+      "eval_loss": 0.6825284361839294,
+      "eval_runtime": 20.6381,
+      "eval_samples_per_second": 2.132,
+      "eval_steps_per_second": 0.145,
+      "step": 975
+    },
+    {
+      "epoch": 0.12635530957698157,
+      "grad_norm": 0.2042436531713094,
+      "learning_rate": 1.2473712904972369e-06,
+      "loss": 0.9087,
+      "step": 976
+    },
+    {
+      "epoch": 0.12648477198433505,
+      "grad_norm": 0.2325955386339354,
+      "learning_rate": 1.2473453464862914e-06,
+      "loss": 1.0846,
+      "step": 977
+    },
+    {
+      "epoch": 0.1266142343916885,
+      "grad_norm": 0.2818198218864742,
+      "learning_rate": 1.247319275348695e-06,
+      "loss": 1.1457,
+      "step": 978
+    },
+    {
+      "epoch": 0.126743696799042,
+      "grad_norm": 0.19345997360876083,
+      "learning_rate": 1.2472930770897722e-06,
+      "loss": 0.9977,
+      "step": 979
+    },
+    {
+      "epoch": 0.12687315920639544,
+      "grad_norm": 0.21821019923485466,
+      "learning_rate": 1.2472667517148754e-06,
+      "loss": 0.9125,
+      "step": 980
+    },
+    {
+      "epoch": 0.12687315920639544,
+      "eval_PRM Accuracy": 0.5754716981132075,
+      "eval_PRM F1": 0.6762589928057554,
+      "eval_PRM F1 AUC": 0.5874803562074384,
+      "eval_PRM F1 Neg": 0.3835616438356164,
+      "eval_PRM NPV": 0.28,
+      "eval_PRM Precision": 0.8392857142857143,
+      "eval_PRM Recall": 0.5662650602409639,
+      "eval_PRM Specificty": 0.6086956521739131,
+      "eval_loss": 0.6812855005264282,
+      "eval_runtime": 20.7751,
+      "eval_samples_per_second": 2.118,
+      "eval_steps_per_second": 0.144,
+      "step": 980
+    },
+    {
+      "epoch": 0.1270026216137489,
+      "grad_norm": 0.2541148069342123,
+      "learning_rate": 1.2472402992293817e-06,
+      "loss": 0.9821,
+      "step": 981
+    },
+    {
+      "epoch": 0.12713208402110238,
+      "grad_norm": 0.23557301504909928,
+      "learning_rate": 1.2472137196386946e-06,
+      "loss": 0.9647,
+      "step": 982
+    },
+    {
+      "epoch": 0.12726154642845583,
+      "grad_norm": 0.23328185838738508,
+      "learning_rate": 1.2471870129482436e-06,
+      "loss": 0.9785,
+      "step": 983
+    },
+    {
+      "epoch": 0.1273910088358093,
+      "grad_norm": 0.16539467124796214,
+      "learning_rate": 1.2471601791634843e-06,
+      "loss": 0.9518,
+      "step": 984
+    },
+    {
+      "epoch": 0.12752047124316276,
+      "grad_norm": 0.15360511168540106,
+      "learning_rate": 1.2471332182898976e-06,
+      "loss": 0.9041,
+      "step": 985
+    },
+    {
+      "epoch": 0.12752047124316276,
+      "eval_PRM Accuracy": 0.5754716981132075,
+      "eval_PRM F1": 0.6762589928057554,
+      "eval_PRM F1 AUC": 0.5874803562074384,
+      "eval_PRM F1 Neg": 0.3835616438356164,
+      "eval_PRM NPV": 0.28,
+      "eval_PRM Precision": 0.8392857142857143,
+      "eval_PRM Recall": 0.5662650602409639,
+      "eval_PRM Specificty": 0.6086956521739131,
+      "eval_loss": 0.6780894994735718,
+      "eval_runtime": 20.682,
+      "eval_samples_per_second": 2.127,
+      "eval_steps_per_second": 0.145,
+      "step": 985
+    },
+    {
+      "epoch": 0.12764993365051625,
+      "grad_norm": 0.23043966158517099,
+      "learning_rate": 1.247106130332991e-06,
+      "loss": 1.0165,
+      "step": 986
+    },
+    {
+      "epoch": 0.1277793960578697,
+      "grad_norm": 0.20981336403010664,
+      "learning_rate": 1.2470789152982976e-06,
+      "loss": 1.0427,
+      "step": 987
+    },
+    {
+      "epoch": 0.12790885846522315,
+      "grad_norm": 0.27582215275835326,
+      "learning_rate": 1.247051573191377e-06,
+      "loss": 1.0491,
+      "step": 988
+    },
+    {
+      "epoch": 0.12803832087257663,
+      "grad_norm": 0.2250831170100155,
+      "learning_rate": 1.247024104017814e-06,
+      "loss": 0.9861,
+      "step": 989
+    },
+    {
+      "epoch": 0.1281677832799301,
+      "grad_norm": 0.18050711983185633,
+      "learning_rate": 1.24699650778322e-06,
+      "loss": 0.8724,
+      "step": 990
+    },
+    {
+      "epoch": 0.1281677832799301,
+      "eval_PRM Accuracy": 0.5660377358490566,
+      "eval_PRM F1": 0.6666666666666666,
+      "eval_PRM F1 AUC": 0.5814562598218963,
+      "eval_PRM F1 Neg": 0.3783783783783784,
+      "eval_PRM NPV": 0.27450980392156865,
+      "eval_PRM Precision": 0.8363636363636363,
+      "eval_PRM Recall": 0.5542168674698795,
+      "eval_PRM Specificty": 0.6086956521739131,
+      "eval_loss": 0.6789772510528564,
+      "eval_runtime": 20.6872,
+      "eval_samples_per_second": 2.127,
+      "eval_steps_per_second": 0.145,
+      "step": 990
+    },
+    {
+      "epoch": 0.12829724568728357,
+      "grad_norm": 0.26754770001546774,
+      "learning_rate": 1.246968784493232e-06,
+      "loss": 1.0915,
+      "step": 991
+    },
+    {
+      "epoch": 0.12842670809463702,
+      "grad_norm": 0.16966310322097386,
+      "learning_rate": 1.2469409341535132e-06,
+      "loss": 0.9379,
+      "step": 992
+    },
+    {
+      "epoch": 0.12855617050199047,
+      "grad_norm": 0.2891491943508992,
+      "learning_rate": 1.246912956769752e-06,
+      "loss": 0.9877,
+      "step": 993
+    },
+    {
+      "epoch": 0.12868563290934396,
+      "grad_norm": 0.26344659024845823,
+      "learning_rate": 1.2468848523476642e-06,
+      "loss": 1.0109,
+      "step": 994
+    },
+    {
+      "epoch": 0.1288150953166974,
+      "grad_norm": 0.1274829380326075,
+      "learning_rate": 1.24685662089299e-06,
+      "loss": 0.9545,
+      "step": 995
+    },
+    {
+      "epoch": 0.1288150953166974,
+      "eval_PRM Accuracy": 0.5754716981132075,
+      "eval_PRM F1": 0.6762589928057554,
+      "eval_PRM F1 AUC": 0.5874803562074384,
+      "eval_PRM F1 Neg": 0.3835616438356164,
+      "eval_PRM NPV": 0.28,
+      "eval_PRM Precision": 0.8392857142857143,
+      "eval_PRM Recall": 0.5662650602409639,
+      "eval_PRM Specificty": 0.6086956521739131,
+      "eval_loss": 0.6782670617103577,
+      "eval_runtime": 20.7096,
+      "eval_samples_per_second": 2.125,
+      "eval_steps_per_second": 0.145,
+      "step": 995
+    },
+    {
+      "epoch": 0.1289445577240509,
+      "grad_norm": 0.208178081251272,
+      "learning_rate": 1.2468282624114965e-06,
+      "loss": 1.0662,
+      "step": 996
+    },
+    {
+      "epoch": 0.12907402013140434,
+      "grad_norm": 0.2814180283504166,
+      "learning_rate": 1.2467997769089768e-06,
+      "loss": 0.9773,
+      "step": 997
+    },
+    {
+      "epoch": 0.1292034825387578,
+      "grad_norm": 0.2331440924789102,
+      "learning_rate": 1.2467711643912493e-06,
+      "loss": 0.967,
+      "step": 998
+    },
+    {
+      "epoch": 0.12933294494611128,
+      "grad_norm": 0.2076851204570146,
+      "learning_rate": 1.2467424248641585e-06,
+      "loss": 1.017,
+      "step": 999
+    },
+    {
+      "epoch": 0.12946240735346473,
+      "grad_norm": 0.21775231163710648,
+      "learning_rate": 1.2467135583335757e-06,
+      "loss": 1.0594,
+      "step": 1000
+    },
+    {
+      "epoch": 0.12946240735346473,
+      "eval_PRM Accuracy": 0.5754716981132075,
+      "eval_PRM F1": 0.6808510638297872,
+      "eval_PRM F1 AUC": 0.5717653221581981,
+      "eval_PRM F1 Neg": 0.36619718309859156,
+      "eval_PRM NPV": 0.2708333333333333,
+      "eval_PRM Precision": 0.8275862068965517,
+      "eval_PRM Recall": 0.5783132530120482,
+      "eval_PRM Specificty": 0.5652173913043478,
+      "eval_loss": 0.6761363744735718,
+      "eval_runtime": 20.7606,
+      "eval_samples_per_second": 2.119,
+      "eval_steps_per_second": 0.145,
+      "step": 1000
     }
   ],
   "logging_steps": 1,
@@ -5142,7 +10242,7 @@
       "attributes": {}
     }
   },
-  "total_flos": 419680048021504.0,
+  "total_flos": 838862698414080.0,
   "train_batch_size": 4,
   "trial_name": null,
   "trial_params": null