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": 2.92,
       "eval_steps_per_second": 0.199,
       "step": 500
+    },
+    {
+      "epoch": 0.06486066608408583,
+      "grad_norm": 3.0540591692146033,
+      "learning_rate": 8.101552393272963e-07,
+      "loss": 5.0771,
+      "step": 501
+    },
+    {
+      "epoch": 0.0649901284914393,
+      "grad_norm": 4.033475081295743,
+      "learning_rate": 8.117723156532989e-07,
+      "loss": 7.1152,
+      "step": 502
+    },
+    {
+      "epoch": 0.06511959089879277,
+      "grad_norm": 3.3108240258330173,
+      "learning_rate": 8.133893919793015e-07,
+      "loss": 4.3901,
+      "step": 503
+    },
+    {
+      "epoch": 0.06524905330614623,
+      "grad_norm": 1.8525932833865588,
+      "learning_rate": 8.150064683053041e-07,
+      "loss": 4.0447,
+      "step": 504
+    },
+    {
+      "epoch": 0.06537851571349969,
+      "grad_norm": 3.107982421650894,
+      "learning_rate": 8.166235446313067e-07,
+      "loss": 5.0181,
+      "step": 505
+    },
+    {
+      "epoch": 0.06537851571349969,
+      "eval_PRM Accuracy": 0.3867924528301887,
+      "eval_PRM F1": 0.4144144144144144,
+      "eval_PRM F1 AUC": 0.5298585646935569,
+      "eval_PRM F1 Neg": 0.3564356435643564,
+      "eval_PRM NPV": 0.23076923076923078,
+      "eval_PRM Precision": 0.8214285714285714,
+      "eval_PRM Recall": 0.27710843373493976,
+      "eval_PRM Specificty": 0.782608695652174,
+      "eval_loss": 0.8978160619735718,
+      "eval_runtime": 14.3945,
+      "eval_samples_per_second": 3.057,
+      "eval_steps_per_second": 0.208,
+      "step": 505
+    },
+    {
+      "epoch": 0.06550797812085316,
+      "grad_norm": 2.937715221227242,
+      "learning_rate": 8.182406209573091e-07,
+      "loss": 5.1091,
+      "step": 506
+    },
+    {
+      "epoch": 0.06563744052820662,
+      "grad_norm": 4.507810999525033,
+      "learning_rate": 8.198576972833118e-07,
+      "loss": 5.6196,
+      "step": 507
+    },
+    {
+      "epoch": 0.06576690293556009,
+      "grad_norm": 2.451316981139926,
+      "learning_rate": 8.214747736093144e-07,
+      "loss": 3.7998,
+      "step": 508
+    },
+    {
+      "epoch": 0.06589636534291356,
+      "grad_norm": 3.636112888850396,
+      "learning_rate": 8.23091849935317e-07,
+      "loss": 5.957,
+      "step": 509
+    },
+    {
+      "epoch": 0.06602582775026701,
+      "grad_norm": 3.0092481611553707,
+      "learning_rate": 8.247089262613196e-07,
+      "loss": 4.2854,
+      "step": 510
+    },
+    {
+      "epoch": 0.06602582775026701,
+      "eval_PRM Accuracy": 0.3867924528301887,
+      "eval_PRM F1": 0.4144144144144144,
+      "eval_PRM F1 AUC": 0.5298585646935569,
+      "eval_PRM F1 Neg": 0.3564356435643564,
+      "eval_PRM NPV": 0.23076923076923078,
+      "eval_PRM Precision": 0.8214285714285714,
+      "eval_PRM Recall": 0.27710843373493976,
+      "eval_PRM Specificty": 0.782608695652174,
+      "eval_loss": 0.8939098119735718,
+      "eval_runtime": 14.3544,
+      "eval_samples_per_second": 3.065,
+      "eval_steps_per_second": 0.209,
+      "step": 510
+    },
+    {
+      "epoch": 0.06615529015762048,
+      "grad_norm": 3.0288191538893163,
+      "learning_rate": 8.263260025873223e-07,
+      "loss": 4.3193,
+      "step": 511
+    },
+    {
+      "epoch": 0.06628475256497394,
+      "grad_norm": 4.736715033864446,
+      "learning_rate": 8.279430789133249e-07,
+      "loss": 6.5122,
+      "step": 512
+    },
+    {
+      "epoch": 0.06641421497232741,
+      "grad_norm": 4.116504028763779,
+      "learning_rate": 8.295601552393273e-07,
+      "loss": 5.3965,
+      "step": 513
+    },
+    {
+      "epoch": 0.06654367737968088,
+      "grad_norm": 1.9387286014118366,
+      "learning_rate": 8.311772315653299e-07,
+      "loss": 3.7812,
+      "step": 514
+    },
+    {
+      "epoch": 0.06667313978703435,
+      "grad_norm": 2.155192815384066,
+      "learning_rate": 8.327943078913325e-07,
+      "loss": 4.3325,
+      "step": 515
+    },
+    {
+      "epoch": 0.06667313978703435,
+      "eval_PRM Accuracy": 0.3867924528301887,
+      "eval_PRM F1": 0.4144144144144144,
+      "eval_PRM F1 AUC": 0.5298585646935569,
+      "eval_PRM F1 Neg": 0.3564356435643564,
+      "eval_PRM NPV": 0.23076923076923078,
+      "eval_PRM Precision": 0.8214285714285714,
+      "eval_PRM Recall": 0.27710843373493976,
+      "eval_PRM Specificty": 0.782608695652174,
+      "eval_loss": 0.8963068127632141,
+      "eval_runtime": 14.2892,
+      "eval_samples_per_second": 3.079,
+      "eval_steps_per_second": 0.21,
+      "step": 515
+    },
+    {
+      "epoch": 0.0668026021943878,
+      "grad_norm": 4.196319636548519,
+      "learning_rate": 8.344113842173351e-07,
+      "loss": 5.3833,
+      "step": 516
+    },
+    {
+      "epoch": 0.06693206460174127,
+      "grad_norm": 2.3249406032097872,
+      "learning_rate": 8.360284605433378e-07,
+      "loss": 4.3223,
+      "step": 517
+    },
+    {
+      "epoch": 0.06706152700909473,
+      "grad_norm": 3.6040502213305112,
+      "learning_rate": 8.376455368693403e-07,
+      "loss": 4.2324,
+      "step": 518
+    },
+    {
+      "epoch": 0.0671909894164482,
+      "grad_norm": 2.9002581708497632,
+      "learning_rate": 8.392626131953428e-07,
+      "loss": 5.1431,
+      "step": 519
+    },
+    {
+      "epoch": 0.06732045182380167,
+      "grad_norm": 2.440567957752614,
+      "learning_rate": 8.408796895213455e-07,
+      "loss": 4.0225,
+      "step": 520
+    },
+    {
+      "epoch": 0.06732045182380167,
+      "eval_PRM Accuracy": 0.3867924528301887,
+      "eval_PRM F1": 0.4144144144144144,
+      "eval_PRM F1 AUC": 0.5298585646935569,
+      "eval_PRM F1 Neg": 0.3564356435643564,
+      "eval_PRM NPV": 0.23076923076923078,
+      "eval_PRM Precision": 0.8214285714285714,
+      "eval_PRM Recall": 0.27710843373493976,
+      "eval_PRM Specificty": 0.782608695652174,
+      "eval_loss": 0.896484375,
+      "eval_runtime": 15.323,
+      "eval_samples_per_second": 2.872,
+      "eval_steps_per_second": 0.196,
+      "step": 520
+    },
+    {
+      "epoch": 0.06744991423115512,
+      "grad_norm": 3.3779231054013796,
+      "learning_rate": 8.42496765847348e-07,
+      "loss": 5.6074,
+      "step": 521
+    },
+    {
+      "epoch": 0.06757937663850859,
+      "grad_norm": 1.7600910136409669,
+      "learning_rate": 8.441138421733506e-07,
+      "loss": 3.5586,
+      "step": 522
+    },
+    {
+      "epoch": 0.06770883904586206,
+      "grad_norm": 2.0663228752988054,
+      "learning_rate": 8.457309184993533e-07,
+      "loss": 3.3735,
+      "step": 523
+    },
+    {
+      "epoch": 0.06783830145321552,
+      "grad_norm": 4.4520688492695815,
+      "learning_rate": 8.473479948253558e-07,
+      "loss": 6.2312,
+      "step": 524
+    },
+    {
+      "epoch": 0.06796776386056899,
+      "grad_norm": 3.3994368259801386,
+      "learning_rate": 8.489650711513583e-07,
+      "loss": 4.7607,
+      "step": 525
+    },
+    {
+      "epoch": 0.06796776386056899,
+      "eval_PRM Accuracy": 0.3867924528301887,
+      "eval_PRM F1": 0.4144144144144144,
+      "eval_PRM F1 AUC": 0.5298585646935569,
+      "eval_PRM F1 Neg": 0.3564356435643564,
+      "eval_PRM NPV": 0.23076923076923078,
+      "eval_PRM Precision": 0.8214285714285714,
+      "eval_PRM Recall": 0.27710843373493976,
+      "eval_PRM Specificty": 0.782608695652174,
+      "eval_loss": 0.8990589380264282,
+      "eval_runtime": 14.6892,
+      "eval_samples_per_second": 2.995,
+      "eval_steps_per_second": 0.204,
+      "step": 525
+    },
+    {
+      "epoch": 0.06809722626792246,
+      "grad_norm": 2.657575768338923,
+      "learning_rate": 8.50582147477361e-07,
+      "loss": 4.5498,
+      "step": 526
+    },
+    {
+      "epoch": 0.06822668867527591,
+      "grad_norm": 2.791066485016929,
+      "learning_rate": 8.521992238033635e-07,
+      "loss": 4.8328,
+      "step": 527
+    },
+    {
+      "epoch": 0.06835615108262938,
+      "grad_norm": 3.7134360308747096,
+      "learning_rate": 8.538163001293662e-07,
+      "loss": 5.7866,
+      "step": 528
+    },
+    {
+      "epoch": 0.06848561348998285,
+      "grad_norm": 4.907739684354129,
+      "learning_rate": 8.554333764553688e-07,
+      "loss": 6.2227,
+      "step": 529
+    },
+    {
+      "epoch": 0.06861507589733631,
+      "grad_norm": 2.052334601212319,
+      "learning_rate": 8.570504527813713e-07,
+      "loss": 3.3701,
+      "step": 530
+    },
+    {
+      "epoch": 0.06861507589733631,
+      "eval_PRM Accuracy": 0.3867924528301887,
+      "eval_PRM F1": 0.4144144144144144,
+      "eval_PRM F1 AUC": 0.5298585646935569,
+      "eval_PRM F1 Neg": 0.3564356435643564,
+      "eval_PRM NPV": 0.23076923076923078,
+      "eval_PRM Precision": 0.8214285714285714,
+      "eval_PRM Recall": 0.27710843373493976,
+      "eval_PRM Specificty": 0.782608695652174,
+      "eval_loss": 0.8987926244735718,
+      "eval_runtime": 13.717,
+      "eval_samples_per_second": 3.208,
+      "eval_steps_per_second": 0.219,
+      "step": 530
+    },
+    {
+      "epoch": 0.06874453830468978,
+      "grad_norm": 3.8891433361105894,
+      "learning_rate": 8.58667529107374e-07,
+      "loss": 4.9106,
+      "step": 531
+    },
+    {
+      "epoch": 0.06887400071204323,
+      "grad_norm": 2.2893042313461036,
+      "learning_rate": 8.602846054333765e-07,
+      "loss": 4.3201,
+      "step": 532
+    },
+    {
+      "epoch": 0.0690034631193967,
+      "grad_norm": 3.4573813278641876,
+      "learning_rate": 8.61901681759379e-07,
+      "loss": 5.292,
+      "step": 533
+    },
+    {
+      "epoch": 0.06913292552675017,
+      "grad_norm": 2.4345783037899076,
+      "learning_rate": 8.635187580853817e-07,
+      "loss": 4.0806,
+      "step": 534
+    },
+    {
+      "epoch": 0.06926238793410364,
+      "grad_norm": 3.1589647066428896,
+      "learning_rate": 8.651358344113843e-07,
+      "loss": 5.3013,
+      "step": 535
+    },
+    {
+      "epoch": 0.06926238793410364,
+      "eval_PRM Accuracy": 0.3867924528301887,
+      "eval_PRM F1": 0.4144144144144144,
+      "eval_PRM F1 AUC": 0.5298585646935569,
+      "eval_PRM F1 Neg": 0.3564356435643564,
+      "eval_PRM NPV": 0.23076923076923078,
+      "eval_PRM Precision": 0.8214285714285714,
+      "eval_PRM Recall": 0.27710843373493976,
+      "eval_PRM Specificty": 0.782608695652174,
+      "eval_loss": 0.900390625,
+      "eval_runtime": 14.9793,
+      "eval_samples_per_second": 2.937,
+      "eval_steps_per_second": 0.2,
+      "step": 535
+    },
+    {
+      "epoch": 0.0693918503414571,
+      "grad_norm": 4.985296325628345,
+      "learning_rate": 8.667529107373869e-07,
+      "loss": 6.3066,
+      "step": 536
+    },
+    {
+      "epoch": 0.06952131274881057,
+      "grad_norm": 3.398097220565681,
+      "learning_rate": 8.683699870633895e-07,
+      "loss": 4.2542,
+      "step": 537
+    },
+    {
+      "epoch": 0.06965077515616402,
+      "grad_norm": 1.7443168575181354,
+      "learning_rate": 8.69987063389392e-07,
+      "loss": 3.6675,
+      "step": 538
+    },
+    {
+      "epoch": 0.06978023756351749,
+      "grad_norm": 2.70134571401946,
+      "learning_rate": 8.716041397153946e-07,
+      "loss": 4.7905,
+      "step": 539
+    },
+    {
+      "epoch": 0.06990969997087096,
+      "grad_norm": 2.8409211200764792,
+      "learning_rate": 8.732212160413972e-07,
+      "loss": 4.2915,
+      "step": 540
+    },
+    {
+      "epoch": 0.06990969997087096,
+      "eval_PRM Accuracy": 0.3867924528301887,
+      "eval_PRM F1": 0.4144144144144144,
+      "eval_PRM F1 AUC": 0.5298585646935569,
+      "eval_PRM F1 Neg": 0.3564356435643564,
+      "eval_PRM NPV": 0.23076923076923078,
+      "eval_PRM Precision": 0.8214285714285714,
+      "eval_PRM Recall": 0.27710843373493976,
+      "eval_PRM Specificty": 0.782608695652174,
+      "eval_loss": 0.9003018736839294,
+      "eval_runtime": 14.4287,
+      "eval_samples_per_second": 3.049,
+      "eval_steps_per_second": 0.208,
+      "step": 540
+    },
+    {
+      "epoch": 0.07003916237822443,
+      "grad_norm": 3.3697084981128893,
+      "learning_rate": 8.748382923673998e-07,
+      "loss": 4.6543,
+      "step": 541
+    },
+    {
+      "epoch": 0.0701686247855779,
+      "grad_norm": 2.120090059916605,
+      "learning_rate": 8.764553686934024e-07,
+      "loss": 3.554,
+      "step": 542
+    },
+    {
+      "epoch": 0.07029808719293135,
+      "grad_norm": 4.873555209018028,
+      "learning_rate": 8.78072445019405e-07,
+      "loss": 7.2856,
+      "step": 543
+    },
+    {
+      "epoch": 0.07042754960028481,
+      "grad_norm": 4.094476922579132,
+      "learning_rate": 8.796895213454077e-07,
+      "loss": 6.144,
+      "step": 544
+    },
+    {
+      "epoch": 0.07055701200763828,
+      "grad_norm": 1.9128453528499718,
+      "learning_rate": 8.813065976714101e-07,
+      "loss": 3.0903,
+      "step": 545
+    },
+    {
+      "epoch": 0.07055701200763828,
+      "eval_PRM Accuracy": 0.3867924528301887,
+      "eval_PRM F1": 0.4144144144144144,
+      "eval_PRM F1 AUC": 0.5298585646935569,
+      "eval_PRM F1 Neg": 0.3564356435643564,
+      "eval_PRM NPV": 0.23076923076923078,
+      "eval_PRM Precision": 0.8214285714285714,
+      "eval_PRM Recall": 0.27710843373493976,
+      "eval_PRM Specificty": 0.782608695652174,
+      "eval_loss": 0.9029651880264282,
+      "eval_runtime": 14.4127,
+      "eval_samples_per_second": 3.053,
+      "eval_steps_per_second": 0.208,
+      "step": 545
+    },
+    {
+      "epoch": 0.07068647441499175,
+      "grad_norm": 3.088880716633396,
+      "learning_rate": 8.829236739974127e-07,
+      "loss": 5.2263,
+      "step": 546
+    },
+    {
+      "epoch": 0.07081593682234522,
+      "grad_norm": 4.0050980997038454,
+      "learning_rate": 8.845407503234154e-07,
+      "loss": 5.7109,
+      "step": 547
+    },
+    {
+      "epoch": 0.07094539922969868,
+      "grad_norm": 2.525558785824187,
+      "learning_rate": 8.861578266494179e-07,
+      "loss": 4.5991,
+      "step": 548
+    },
+    {
+      "epoch": 0.07107486163705214,
+      "grad_norm": 3.5294589158774814,
+      "learning_rate": 8.877749029754205e-07,
+      "loss": 5.3726,
+      "step": 549
+    },
+    {
+      "epoch": 0.0712043240444056,
+      "grad_norm": 1.744927169196934,
+      "learning_rate": 8.893919793014232e-07,
+      "loss": 3.8086,
+      "step": 550
+    },
+    {
+      "epoch": 0.0712043240444056,
+      "eval_PRM Accuracy": 0.3867924528301887,
+      "eval_PRM F1": 0.4144144144144144,
+      "eval_PRM F1 AUC": 0.5298585646935569,
+      "eval_PRM F1 Neg": 0.3564356435643564,
+      "eval_PRM NPV": 0.23076923076923078,
+      "eval_PRM Precision": 0.8214285714285714,
+      "eval_PRM Recall": 0.27710843373493976,
+      "eval_PRM Specificty": 0.782608695652174,
+      "eval_loss": 0.9018110632896423,
+      "eval_runtime": 14.412,
+      "eval_samples_per_second": 3.053,
+      "eval_steps_per_second": 0.208,
+      "step": 550
+    },
+    {
+      "epoch": 0.07133378645175907,
+      "grad_norm": 3.1989162668420454,
+      "learning_rate": 8.910090556274256e-07,
+      "loss": 4.9919,
+      "step": 551
+    },
+    {
+      "epoch": 0.07146324885911254,
+      "grad_norm": 3.4064242196300887,
+      "learning_rate": 8.926261319534282e-07,
+      "loss": 4.6013,
+      "step": 552
+    },
+    {
+      "epoch": 0.071592711266466,
+      "grad_norm": 3.104943466065872,
+      "learning_rate": 8.942432082794309e-07,
+      "loss": 5.2639,
+      "step": 553
+    },
+    {
+      "epoch": 0.07172217367381946,
+      "grad_norm": 3.2076788640969904,
+      "learning_rate": 8.958602846054334e-07,
+      "loss": 4.9648,
+      "step": 554
+    },
+    {
+      "epoch": 0.07185163608117293,
+      "grad_norm": 4.384616840432897,
+      "learning_rate": 8.974773609314361e-07,
+      "loss": 5.7134,
+      "step": 555
+    },
+    {
+      "epoch": 0.07185163608117293,
+      "eval_PRM Accuracy": 0.3867924528301887,
+      "eval_PRM F1": 0.4144144144144144,
+      "eval_PRM F1 AUC": 0.5298585646935569,
+      "eval_PRM F1 Neg": 0.3564356435643564,
+      "eval_PRM NPV": 0.23076923076923078,
+      "eval_PRM Precision": 0.8214285714285714,
+      "eval_PRM Recall": 0.27710843373493976,
+      "eval_PRM Specificty": 0.782608695652174,
+      "eval_loss": 0.9041193127632141,
+      "eval_runtime": 14.6811,
+      "eval_samples_per_second": 2.997,
+      "eval_steps_per_second": 0.204,
+      "step": 555
+    },
+    {
+      "epoch": 0.0719810984885264,
+      "grad_norm": 2.8606649552219032,
+      "learning_rate": 8.990944372574387e-07,
+      "loss": 4.3687,
+      "step": 556
+    },
+    {
+      "epoch": 0.07211056089587986,
+      "grad_norm": 4.6360260728371525,
+      "learning_rate": 9.007115135834413e-07,
+      "loss": 5.1851,
+      "step": 557
+    },
+    {
+      "epoch": 0.07224002330323333,
+      "grad_norm": 4.110115682068368,
+      "learning_rate": 9.023285899094438e-07,
+      "loss": 5.0361,
+      "step": 558
+    },
+    {
+      "epoch": 0.0723694857105868,
+      "grad_norm": 2.569193297039986,
+      "learning_rate": 9.039456662354464e-07,
+      "loss": 3.9204,
+      "step": 559
+    },
+    {
+      "epoch": 0.07249894811794025,
+      "grad_norm": 4.696046617483736,
+      "learning_rate": 9.055627425614489e-07,
+      "loss": 6.6973,
+      "step": 560
+    },
+    {
+      "epoch": 0.07249894811794025,
+      "eval_PRM Accuracy": 0.3867924528301887,
+      "eval_PRM F1": 0.4144144144144144,
+      "eval_PRM F1 AUC": 0.5298585646935569,
+      "eval_PRM F1 Neg": 0.3564356435643564,
+      "eval_PRM NPV": 0.23076923076923078,
+      "eval_PRM Precision": 0.8214285714285714,
+      "eval_PRM Recall": 0.27710843373493976,
+      "eval_PRM Specificty": 0.782608695652174,
+      "eval_loss": 0.9022549986839294,
+      "eval_runtime": 14.2253,
+      "eval_samples_per_second": 3.093,
+      "eval_steps_per_second": 0.211,
+      "step": 560
+    },
+    {
+      "epoch": 0.07262841052529372,
+      "grad_norm": 2.7338776532354547,
+      "learning_rate": 9.071798188874516e-07,
+      "loss": 4.1255,
+      "step": 561
+    },
+    {
+      "epoch": 0.07275787293264718,
+      "grad_norm": 1.9844796035867727,
+      "learning_rate": 9.087968952134542e-07,
+      "loss": 3.7235,
+      "step": 562
+    },
+    {
+      "epoch": 0.07288733534000065,
+      "grad_norm": 3.447358447957195,
+      "learning_rate": 9.104139715394568e-07,
+      "loss": 5.0107,
+      "step": 563
+    },
+    {
+      "epoch": 0.07301679774735412,
+      "grad_norm": 3.1047084367884197,
+      "learning_rate": 9.120310478654593e-07,
+      "loss": 4.2332,
+      "step": 564
+    },
+    {
+      "epoch": 0.07314626015470757,
+      "grad_norm": 5.146389209471009,
+      "learning_rate": 9.136481241914618e-07,
+      "loss": 5.7676,
+      "step": 565
+    },
+    {
+      "epoch": 0.07314626015470757,
+      "eval_PRM Accuracy": 0.3867924528301887,
+      "eval_PRM F1": 0.4144144144144144,
+      "eval_PRM F1 AUC": 0.5298585646935569,
+      "eval_PRM F1 Neg": 0.3564356435643564,
+      "eval_PRM NPV": 0.23076923076923078,
+      "eval_PRM Precision": 0.8214285714285714,
+      "eval_PRM Recall": 0.27710843373493976,
+      "eval_PRM Specificty": 0.782608695652174,
+      "eval_loss": 0.9055397510528564,
+      "eval_runtime": 14.2443,
+      "eval_samples_per_second": 3.089,
+      "eval_steps_per_second": 0.211,
+      "step": 565
+    },
+    {
+      "epoch": 0.07327572256206104,
+      "grad_norm": 4.501006598165547,
+      "learning_rate": 9.152652005174645e-07,
+      "loss": 5.9028,
+      "step": 566
+    },
+    {
+      "epoch": 0.0734051849694145,
+      "grad_norm": 2.566261197143029,
+      "learning_rate": 9.168822768434671e-07,
+      "loss": 4.1855,
+      "step": 567
+    },
+    {
+      "epoch": 0.07353464737676797,
+      "grad_norm": 3.318519305720438,
+      "learning_rate": 9.184993531694696e-07,
+      "loss": 4.4175,
+      "step": 568
+    },
+    {
+      "epoch": 0.07366410978412144,
+      "grad_norm": 2.2329243207688627,
+      "learning_rate": 9.201164294954723e-07,
+      "loss": 3.4346,
+      "step": 569
+    },
+    {
+      "epoch": 0.07379357219147491,
+      "grad_norm": 3.2854758216697997,
+      "learning_rate": 9.217335058214749e-07,
+      "loss": 4.0674,
+      "step": 570
+    },
+    {
+      "epoch": 0.07379357219147491,
+      "eval_PRM Accuracy": 0.3867924528301887,
+      "eval_PRM F1": 0.4144144144144144,
+      "eval_PRM F1 AUC": 0.5298585646935569,
+      "eval_PRM F1 Neg": 0.3564356435643564,
+      "eval_PRM NPV": 0.23076923076923078,
+      "eval_PRM Precision": 0.8214285714285714,
+      "eval_PRM Recall": 0.27710843373493976,
+      "eval_PRM Specificty": 0.782608695652174,
+      "eval_loss": 0.9040305614471436,
+      "eval_runtime": 14.7875,
+      "eval_samples_per_second": 2.975,
+      "eval_steps_per_second": 0.203,
+      "step": 570
+    },
+    {
+      "epoch": 0.07392303459882836,
+      "grad_norm": 1.6680723161798654,
+      "learning_rate": 9.233505821474773e-07,
+      "loss": 3.1938,
+      "step": 571
+    },
+    {
+      "epoch": 0.07405249700618183,
+      "grad_norm": 3.1843966375356128,
+      "learning_rate": 9.2496765847348e-07,
+      "loss": 5.1934,
+      "step": 572
+    },
+    {
+      "epoch": 0.0741819594135353,
+      "grad_norm": 4.506517328135267,
+      "learning_rate": 9.265847347994826e-07,
+      "loss": 6.0032,
+      "step": 573
+    },
+    {
+      "epoch": 0.07431142182088876,
+      "grad_norm": 4.845730851811751,
+      "learning_rate": 9.282018111254852e-07,
+      "loss": 6.4753,
+      "step": 574
+    },
+    {
+      "epoch": 0.07444088422824223,
+      "grad_norm": 3.6721389107272473,
+      "learning_rate": 9.298188874514878e-07,
+      "loss": 5.3804,
+      "step": 575
+    },
+    {
+      "epoch": 0.07444088422824223,
+      "eval_PRM Accuracy": 0.3867924528301887,
+      "eval_PRM F1": 0.4144144144144144,
+      "eval_PRM F1 AUC": 0.5298585646935569,
+      "eval_PRM F1 Neg": 0.3564356435643564,
+      "eval_PRM NPV": 0.23076923076923078,
+      "eval_PRM Precision": 0.8214285714285714,
+      "eval_PRM Recall": 0.27710843373493976,
+      "eval_PRM Specificty": 0.782608695652174,
+      "eval_loss": 0.9096235632896423,
+      "eval_runtime": 22.1478,
+      "eval_samples_per_second": 1.987,
+      "eval_steps_per_second": 0.135,
+      "step": 575
+    },
+    {
+      "epoch": 0.07457034663559568,
+      "grad_norm": 3.5015722139086947,
+      "learning_rate": 9.314359637774904e-07,
+      "loss": 4.3586,
+      "step": 576
+    },
+    {
+      "epoch": 0.07469980904294915,
+      "grad_norm": 2.6449535120499448,
+      "learning_rate": 9.330530401034929e-07,
+      "loss": 3.9878,
+      "step": 577
+    },
+    {
+      "epoch": 0.07482927145030262,
+      "grad_norm": 1.9122836693706646,
+      "learning_rate": 9.346701164294955e-07,
+      "loss": 3.5132,
+      "step": 578
+    },
+    {
+      "epoch": 0.07495873385765608,
+      "grad_norm": 3.460105436402456,
+      "learning_rate": 9.362871927554981e-07,
+      "loss": 4.9424,
+      "step": 579
+    },
+    {
+      "epoch": 0.07508819626500955,
+      "grad_norm": 2.3803755045750723,
+      "learning_rate": 9.379042690815007e-07,
+      "loss": 3.2925,
+      "step": 580
+    },
+    {
+      "epoch": 0.07508819626500955,
+      "eval_PRM Accuracy": 0.3867924528301887,
+      "eval_PRM F1": 0.4144144144144144,
+      "eval_PRM F1 AUC": 0.5298585646935569,
+      "eval_PRM F1 Neg": 0.3564356435643564,
+      "eval_PRM NPV": 0.23076923076923078,
+      "eval_PRM Precision": 0.8214285714285714,
+      "eval_PRM Recall": 0.27710843373493976,
+      "eval_PRM Specificty": 0.782608695652174,
+      "eval_loss": 0.9056285619735718,
+      "eval_runtime": 15.1217,
+      "eval_samples_per_second": 2.91,
+      "eval_steps_per_second": 0.198,
+      "step": 580
+    },
+    {
+      "epoch": 0.07521765867236302,
+      "grad_norm": 2.5524823901163307,
+      "learning_rate": 9.395213454075033e-07,
+      "loss": 4.2061,
+      "step": 581
+    },
+    {
+      "epoch": 0.07534712107971647,
+      "grad_norm": 3.213826628952772,
+      "learning_rate": 9.41138421733506e-07,
+      "loss": 5.3286,
+      "step": 582
+    },
+    {
+      "epoch": 0.07547658348706994,
+      "grad_norm": 5.185873619621562,
+      "learning_rate": 9.427554980595084e-07,
+      "loss": 6.583,
+      "step": 583
+    },
+    {
+      "epoch": 0.07560604589442341,
+      "grad_norm": 3.4682000183730515,
+      "learning_rate": 9.44372574385511e-07,
+      "loss": 5.1123,
+      "step": 584
+    },
+    {
+      "epoch": 0.07573550830177687,
+      "grad_norm": 2.542333651184107,
+      "learning_rate": 9.459896507115137e-07,
+      "loss": 4.5977,
+      "step": 585
+    },
+    {
+      "epoch": 0.07573550830177687,
+      "eval_PRM Accuracy": 0.37735849056603776,
+      "eval_PRM F1": 0.4,
+      "eval_PRM F1 AUC": 0.5238344683080147,
+      "eval_PRM F1 Neg": 0.35294117647058826,
+      "eval_PRM NPV": 0.22784810126582278,
+      "eval_PRM Precision": 0.8148148148148148,
+      "eval_PRM Recall": 0.26506024096385544,
+      "eval_PRM Specificty": 0.782608695652174,
+      "eval_loss": 0.9098011255264282,
+      "eval_runtime": 15.9048,
+      "eval_samples_per_second": 2.766,
+      "eval_steps_per_second": 0.189,
+      "step": 585
+    },
+    {
+      "epoch": 0.07586497070913034,
+      "grad_norm": 4.495054423452107,
+      "learning_rate": 9.476067270375162e-07,
+      "loss": 5.8911,
+      "step": 586
+    },
+    {
+      "epoch": 0.0759944331164838,
+      "grad_norm": 4.286291676009975,
+      "learning_rate": 9.492238033635188e-07,
+      "loss": 5.875,
+      "step": 587
+    },
+    {
+      "epoch": 0.07612389552383726,
+      "grad_norm": 2.2731867866477113,
+      "learning_rate": 9.508408796895215e-07,
+      "loss": 4.1794,
+      "step": 588
+    },
+    {
+      "epoch": 0.07625335793119073,
+      "grad_norm": 4.967621782770409,
+      "learning_rate": 9.52457956015524e-07,
+      "loss": 5.6804,
+      "step": 589
+    },
+    {
+      "epoch": 0.0763828203385442,
+      "grad_norm": 5.421279083915253,
+      "learning_rate": 9.540750323415266e-07,
+      "loss": 6.9507,
+      "step": 590
+    },
+    {
+      "epoch": 0.0763828203385442,
+      "eval_PRM Accuracy": 0.3867924528301887,
+      "eval_PRM F1": 0.4144144144144144,
+      "eval_PRM F1 AUC": 0.5298585646935569,
+      "eval_PRM F1 Neg": 0.3564356435643564,
+      "eval_PRM NPV": 0.23076923076923078,
+      "eval_PRM Precision": 0.8214285714285714,
+      "eval_PRM Recall": 0.27710843373493976,
+      "eval_PRM Specificty": 0.782608695652174,
+      "eval_loss": 0.9108664989471436,
+      "eval_runtime": 14.3094,
+      "eval_samples_per_second": 3.075,
+      "eval_steps_per_second": 0.21,
+      "step": 590
+    },
+    {
+      "epoch": 0.07651228274589766,
+      "grad_norm": 3.4462083989481127,
+      "learning_rate": 9.55692108667529e-07,
+      "loss": 5.4868,
+      "step": 591
+    },
+    {
+      "epoch": 0.07664174515325113,
+      "grad_norm": 3.8268458683344617,
+      "learning_rate": 9.573091849935317e-07,
+      "loss": 4.8584,
+      "step": 592
+    },
+    {
+      "epoch": 0.07677120756060458,
+      "grad_norm": 1.8740977019432772,
+      "learning_rate": 9.589262613195344e-07,
+      "loss": 3.6318,
+      "step": 593
+    },
+    {
+      "epoch": 0.07690066996795805,
+      "grad_norm": 4.548373022254427,
+      "learning_rate": 9.60543337645537e-07,
+      "loss": 6.3418,
+      "step": 594
+    },
+    {
+      "epoch": 0.07703013237531152,
+      "grad_norm": 2.886714885850772,
+      "learning_rate": 9.621604139715395e-07,
+      "loss": 4.147,
+      "step": 595
+    },
+    {
+      "epoch": 0.07703013237531152,
+      "eval_PRM Accuracy": 0.3867924528301887,
+      "eval_PRM F1": 0.4144144144144144,
+      "eval_PRM F1 AUC": 0.5298585646935569,
+      "eval_PRM F1 Neg": 0.3564356435643564,
+      "eval_PRM NPV": 0.23076923076923078,
+      "eval_PRM Precision": 0.8214285714285714,
+      "eval_PRM Recall": 0.27710843373493976,
+      "eval_PRM Specificty": 0.782608695652174,
+      "eval_loss": 0.9083806872367859,
+      "eval_runtime": 13.9448,
+      "eval_samples_per_second": 3.155,
+      "eval_steps_per_second": 0.215,
+      "step": 595
+    },
+    {
+      "epoch": 0.07715959478266499,
+      "grad_norm": 3.7486317271092116,
+      "learning_rate": 9.63777490297542e-07,
+      "loss": 5.3306,
+      "step": 596
+    },
+    {
+      "epoch": 0.07728905719001845,
+      "grad_norm": 3.8631528248660496,
+      "learning_rate": 9.653945666235447e-07,
+      "loss": 4.8525,
+      "step": 597
+    },
+    {
+      "epoch": 0.07741851959737191,
+      "grad_norm": 1.6416582456791071,
+      "learning_rate": 9.670116429495473e-07,
+      "loss": 3.3135,
+      "step": 598
+    },
+    {
+      "epoch": 0.07754798200472537,
+      "grad_norm": 2.9644891715232733,
+      "learning_rate": 9.686287192755498e-07,
+      "loss": 3.7283,
+      "step": 599
+    },
+    {
+      "epoch": 0.07767744441207884,
+      "grad_norm": 2.5950098481380723,
+      "learning_rate": 9.702457956015525e-07,
+      "loss": 3.395,
+      "step": 600
+    },
+    {
+      "epoch": 0.07767744441207884,
+      "eval_PRM Accuracy": 0.3867924528301887,
+      "eval_PRM F1": 0.4144144144144144,
+      "eval_PRM F1 AUC": 0.5298585646935569,
+      "eval_PRM F1 Neg": 0.3564356435643564,
+      "eval_PRM NPV": 0.23076923076923078,
+      "eval_PRM Precision": 0.8214285714285714,
+      "eval_PRM Recall": 0.27710843373493976,
+      "eval_PRM Specificty": 0.782608695652174,
+      "eval_loss": 0.9129083752632141,
+      "eval_runtime": 13.4266,
+      "eval_samples_per_second": 3.277,
+      "eval_steps_per_second": 0.223,
+      "step": 600
+    },
+    {
+      "epoch": 0.07780690681943231,
+      "grad_norm": 2.493656861678879,
+      "learning_rate": 9.718628719275551e-07,
+      "loss": 3.5049,
+      "step": 601
+    },
+    {
+      "epoch": 0.07793636922678578,
+      "grad_norm": 5.086132842962849,
+      "learning_rate": 9.734799482535578e-07,
+      "loss": 6.459,
+      "step": 602
+    },
+    {
+      "epoch": 0.07806583163413924,
+      "grad_norm": 3.9699953853182386,
+      "learning_rate": 9.750970245795603e-07,
+      "loss": 5.4473,
+      "step": 603
+    },
+    {
+      "epoch": 0.0781952940414927,
+      "grad_norm": 3.01379362488202,
+      "learning_rate": 9.767141009055627e-07,
+      "loss": 4.4377,
+      "step": 604
+    },
+    {
+      "epoch": 0.07832475644884616,
+      "grad_norm": 2.976840641852765,
+      "learning_rate": 9.783311772315654e-07,
+      "loss": 3.8774,
+      "step": 605
+    },
+    {
+      "epoch": 0.07832475644884616,
+      "eval_PRM Accuracy": 0.3867924528301887,
+      "eval_PRM F1": 0.4144144144144144,
+      "eval_PRM F1 AUC": 0.5298585646935569,
+      "eval_PRM F1 Neg": 0.3564356435643564,
+      "eval_PRM NPV": 0.23076923076923078,
+      "eval_PRM Precision": 0.8214285714285714,
+      "eval_PRM Recall": 0.27710843373493976,
+      "eval_PRM Specificty": 0.782608695652174,
+      "eval_loss": 0.9095348119735718,
+      "eval_runtime": 15.5441,
+      "eval_samples_per_second": 2.831,
+      "eval_steps_per_second": 0.193,
+      "step": 605
+    },
+    {
+      "epoch": 0.07845421885619963,
+      "grad_norm": 4.206832861987359,
+      "learning_rate": 9.79948253557568e-07,
+      "loss": 5.4097,
+      "step": 606
+    },
+    {
+      "epoch": 0.0785836812635531,
+      "grad_norm": 5.446122871682926,
+      "learning_rate": 9.815653298835705e-07,
+      "loss": 6.4014,
+      "step": 607
+    },
+    {
+      "epoch": 0.07871314367090657,
+      "grad_norm": 2.63538982982655,
+      "learning_rate": 9.831824062095732e-07,
+      "loss": 3.2788,
+      "step": 608
+    },
+    {
+      "epoch": 0.07884260607826002,
+      "grad_norm": 5.477882117028058,
+      "learning_rate": 9.847994825355757e-07,
+      "loss": 7.3257,
+      "step": 609
+    },
+    {
+      "epoch": 0.07897206848561349,
+      "grad_norm": 5.467805438149073,
+      "learning_rate": 9.864165588615783e-07,
+      "loss": 6.6133,
+      "step": 610
+    },
+    {
+      "epoch": 0.07897206848561349,
+      "eval_PRM Accuracy": 0.3867924528301887,
+      "eval_PRM F1": 0.4144144144144144,
+      "eval_PRM F1 AUC": 0.5298585646935569,
+      "eval_PRM F1 Neg": 0.3564356435643564,
+      "eval_PRM NPV": 0.23076923076923078,
+      "eval_PRM Precision": 0.8214285714285714,
+      "eval_PRM Recall": 0.27710843373493976,
+      "eval_PRM Specificty": 0.782608695652174,
+      "eval_loss": 0.9138849377632141,
+      "eval_runtime": 14.4609,
+      "eval_samples_per_second": 3.043,
+      "eval_steps_per_second": 0.207,
+      "step": 610
+    },
+    {
+      "epoch": 0.07910153089296695,
+      "grad_norm": 2.902241539133023,
+      "learning_rate": 9.88033635187581e-07,
+      "loss": 3.7368,
+      "step": 611
+    },
+    {
+      "epoch": 0.07923099330032042,
+      "grad_norm": 2.948470323457141,
+      "learning_rate": 9.896507115135835e-07,
+      "loss": 4.4819,
+      "step": 612
+    },
+    {
+      "epoch": 0.07936045570767389,
+      "grad_norm": 3.6804164605289014,
+      "learning_rate": 9.912677878395861e-07,
+      "loss": 5.4275,
+      "step": 613
+    },
+    {
+      "epoch": 0.07948991811502736,
+      "grad_norm": 3.308617908453641,
+      "learning_rate": 9.928848641655888e-07,
+      "loss": 3.865,
+      "step": 614
+    },
+    {
+      "epoch": 0.07961938052238081,
+      "grad_norm": 3.0972564895323926,
+      "learning_rate": 9.945019404915913e-07,
+      "loss": 4.1226,
+      "step": 615
+    },
+    {
+      "epoch": 0.07961938052238081,
+      "eval_PRM Accuracy": 0.3867924528301887,
+      "eval_PRM F1": 0.4144144144144144,
+      "eval_PRM F1 AUC": 0.5298585646935569,
+      "eval_PRM F1 Neg": 0.3564356435643564,
+      "eval_PRM NPV": 0.23076923076923078,
+      "eval_PRM Precision": 0.8214285714285714,
+      "eval_PRM Recall": 0.27710843373493976,
+      "eval_PRM Specificty": 0.782608695652174,
+      "eval_loss": 0.9142400622367859,
+      "eval_runtime": 15.1752,
+      "eval_samples_per_second": 2.899,
+      "eval_steps_per_second": 0.198,
+      "step": 615
+    },
+    {
+      "epoch": 0.07974884292973428,
+      "grad_norm": 4.431881520626385,
+      "learning_rate": 9.961190168175937e-07,
+      "loss": 4.9678,
+      "step": 616
+    },
+    {
+      "epoch": 0.07987830533708774,
+      "grad_norm": 2.0140038228607686,
+      "learning_rate": 9.977360931435964e-07,
+      "loss": 3.4004,
+      "step": 617
+    },
+    {
+      "epoch": 0.08000776774444121,
+      "grad_norm": 2.4860917383878918,
+      "learning_rate": 9.99353169469599e-07,
+      "loss": 3.9187,
+      "step": 618
+    },
+    {
+      "epoch": 0.08013723015179468,
+      "grad_norm": 3.3119505720959994,
+      "learning_rate": 1.0009702457956017e-06,
+      "loss": 3.8672,
+      "step": 619
+    },
+    {
+      "epoch": 0.08026669255914813,
+      "grad_norm": 3.718516242590131,
+      "learning_rate": 1.0025873221216042e-06,
+      "loss": 5.3223,
+      "step": 620
+    },
+    {
+      "epoch": 0.08026669255914813,
+      "eval_PRM Accuracy": 0.3867924528301887,
+      "eval_PRM F1": 0.4144144144144144,
+      "eval_PRM F1 AUC": 0.5298585646935569,
+      "eval_PRM F1 Neg": 0.3564356435643564,
+      "eval_PRM NPV": 0.23076923076923078,
+      "eval_PRM Precision": 0.8214285714285714,
+      "eval_PRM Recall": 0.27710843373493976,
+      "eval_PRM Specificty": 0.782608695652174,
+      "eval_loss": 0.9167258739471436,
+      "eval_runtime": 15.7604,
+      "eval_samples_per_second": 2.792,
+      "eval_steps_per_second": 0.19,
+      "step": 620
+    },
+    {
+      "epoch": 0.0803961549665016,
+      "grad_norm": 1.6958635648600953,
+      "learning_rate": 1.0042043984476069e-06,
+      "loss": 3.0889,
+      "step": 621
+    },
+    {
+      "epoch": 0.08052561737385507,
+      "grad_norm": 2.5357550243833566,
+      "learning_rate": 1.0058214747736093e-06,
+      "loss": 3.9617,
+      "step": 622
+    },
+    {
+      "epoch": 0.08065507978120853,
+      "grad_norm": 2.5902109060460003,
+      "learning_rate": 1.007438551099612e-06,
+      "loss": 3.4467,
+      "step": 623
+    },
+    {
+      "epoch": 0.080784542188562,
+      "grad_norm": 3.421798539664192,
+      "learning_rate": 1.0090556274256147e-06,
+      "loss": 4.6733,
+      "step": 624
+    },
+    {
+      "epoch": 0.08091400459591547,
+      "grad_norm": 3.814242406216809,
+      "learning_rate": 1.0106727037516171e-06,
+      "loss": 4.7397,
+      "step": 625
+    },
+    {
+      "epoch": 0.08091400459591547,
+      "eval_PRM Accuracy": 0.3867924528301887,
+      "eval_PRM F1": 0.4144144144144144,
+      "eval_PRM F1 AUC": 0.5298585646935569,
+      "eval_PRM F1 Neg": 0.3564356435643564,
+      "eval_PRM NPV": 0.23076923076923078,
+      "eval_PRM Precision": 0.8214285714285714,
+      "eval_PRM Recall": 0.27710843373493976,
+      "eval_PRM Specificty": 0.782608695652174,
+      "eval_loss": 0.9137073755264282,
+      "eval_runtime": 15.0257,
+      "eval_samples_per_second": 2.928,
+      "eval_steps_per_second": 0.2,
+      "step": 625
+    },
+    {
+      "epoch": 0.08104346700326892,
+      "grad_norm": 3.9540614587429306,
+      "learning_rate": 1.0122897800776198e-06,
+      "loss": 6.1907,
+      "step": 626
+    },
+    {
+      "epoch": 0.08117292941062239,
+      "grad_norm": 3.946503648694806,
+      "learning_rate": 1.0139068564036225e-06,
+      "loss": 5.1562,
+      "step": 627
+    },
+    {
+      "epoch": 0.08130239181797586,
+      "grad_norm": 2.556311376618824,
+      "learning_rate": 1.015523932729625e-06,
+      "loss": 3.7217,
+      "step": 628
+    },
+    {
+      "epoch": 0.08143185422532932,
+      "grad_norm": 4.098755670921853,
+      "learning_rate": 1.0171410090556274e-06,
+      "loss": 5.8599,
+      "step": 629
+    },
+    {
+      "epoch": 0.08156131663268279,
+      "grad_norm": 2.988489969751208,
+      "learning_rate": 1.01875808538163e-06,
+      "loss": 4.0645,
+      "step": 630
+    },
+    {
+      "epoch": 0.08156131663268279,
+      "eval_PRM Accuracy": 0.37735849056603776,
+      "eval_PRM F1": 0.4,
+      "eval_PRM F1 AUC": 0.5238344683080147,
+      "eval_PRM F1 Neg": 0.35294117647058826,
+      "eval_PRM NPV": 0.22784810126582278,
+      "eval_PRM Precision": 0.8148148148148148,
+      "eval_PRM Recall": 0.26506024096385544,
+      "eval_PRM Specificty": 0.782608695652174,
+      "eval_loss": 0.9172585010528564,
+      "eval_runtime": 14.6966,
+      "eval_samples_per_second": 2.994,
+      "eval_steps_per_second": 0.204,
+      "step": 630
+    },
+    {
+      "epoch": 0.08169077904003624,
+      "grad_norm": 3.311031159862222,
+      "learning_rate": 1.0203751617076327e-06,
+      "loss": 5.4185,
+      "step": 631
+    },
+    {
+      "epoch": 0.08182024144738971,
+      "grad_norm": 3.7274143308730396,
+      "learning_rate": 1.0219922380336354e-06,
+      "loss": 4.9485,
+      "step": 632
+    },
+    {
+      "epoch": 0.08194970385474318,
+      "grad_norm": 1.8728233992325662,
+      "learning_rate": 1.0236093143596378e-06,
+      "loss": 3.0996,
+      "step": 633
+    },
+    {
+      "epoch": 0.08207916626209665,
+      "grad_norm": 3.930862207779473,
+      "learning_rate": 1.0252263906856405e-06,
+      "loss": 5.2373,
+      "step": 634
+    },
+    {
+      "epoch": 0.08220862866945011,
+      "grad_norm": 3.3166952982896842,
+      "learning_rate": 1.026843467011643e-06,
+      "loss": 4.1846,
+      "step": 635
+    },
+    {
+      "epoch": 0.08220862866945011,
+      "eval_PRM Accuracy": 0.37735849056603776,
+      "eval_PRM F1": 0.4,
+      "eval_PRM F1 AUC": 0.5238344683080147,
+      "eval_PRM F1 Neg": 0.35294117647058826,
+      "eval_PRM NPV": 0.22784810126582278,
+      "eval_PRM Precision": 0.8148148148148148,
+      "eval_PRM Recall": 0.26506024096385544,
+      "eval_PRM Specificty": 0.782608695652174,
+      "eval_loss": 0.9192116260528564,
+      "eval_runtime": 14.9267,
+      "eval_samples_per_second": 2.948,
+      "eval_steps_per_second": 0.201,
+      "step": 635
+    },
+    {
+      "epoch": 0.08233809107680358,
+      "grad_norm": 2.1132919104041235,
+      "learning_rate": 1.0284605433376456e-06,
+      "loss": 3.5691,
+      "step": 636
+    },
+    {
+      "epoch": 0.08246755348415703,
+      "grad_norm": 4.264581045355218,
+      "learning_rate": 1.030077619663648e-06,
+      "loss": 4.8135,
+      "step": 637
+    },
+    {
+      "epoch": 0.0825970158915105,
+      "grad_norm": 3.601980431909174,
+      "learning_rate": 1.0316946959896508e-06,
+      "loss": 4.5483,
+      "step": 638
+    },
+    {
+      "epoch": 0.08272647829886397,
+      "grad_norm": 2.4651396349143675,
+      "learning_rate": 1.0333117723156534e-06,
+      "loss": 3.4932,
+      "step": 639
+    },
+    {
+      "epoch": 0.08285594070621743,
+      "grad_norm": 3.1828995893611456,
+      "learning_rate": 1.0349288486416561e-06,
+      "loss": 4.1899,
+      "step": 640
+    },
+    {
+      "epoch": 0.08285594070621743,
+      "eval_PRM Accuracy": 0.3867924528301887,
+      "eval_PRM F1": 0.4144144144144144,
+      "eval_PRM F1 AUC": 0.5298585646935569,
+      "eval_PRM F1 Neg": 0.3564356435643564,
+      "eval_PRM NPV": 0.23076923076923078,
+      "eval_PRM Precision": 0.8214285714285714,
+      "eval_PRM Recall": 0.27710843373493976,
+      "eval_PRM Specificty": 0.782608695652174,
+      "eval_loss": 0.9197443127632141,
+      "eval_runtime": 15.2649,
+      "eval_samples_per_second": 2.882,
+      "eval_steps_per_second": 0.197,
+      "step": 640
+    },
+    {
+      "epoch": 0.0829854031135709,
+      "grad_norm": 5.193147904721469,
+      "learning_rate": 1.0365459249676586e-06,
+      "loss": 6.2974,
+      "step": 641
+    },
+    {
+      "epoch": 0.08311486552092436,
+      "grad_norm": 4.983833227627457,
+      "learning_rate": 1.038163001293661e-06,
+      "loss": 5.7546,
+      "step": 642
+    },
+    {
+      "epoch": 0.08324432792827782,
+      "grad_norm": 3.6240315909900556,
+      "learning_rate": 1.0397800776196637e-06,
+      "loss": 5.4492,
+      "step": 643
+    },
+    {
+      "epoch": 0.08337379033563129,
+      "grad_norm": 1.803634512316626,
+      "learning_rate": 1.0413971539456664e-06,
+      "loss": 3.6833,
+      "step": 644
+    },
+    {
+      "epoch": 0.08350325274298476,
+      "grad_norm": 4.284586551004286,
+      "learning_rate": 1.0430142302716688e-06,
+      "loss": 5.7485,
+      "step": 645
+    },
+    {
+      "epoch": 0.08350325274298476,
+      "eval_PRM Accuracy": 0.37735849056603776,
+      "eval_PRM F1": 0.4,
+      "eval_PRM F1 AUC": 0.5238344683080147,
+      "eval_PRM F1 Neg": 0.35294117647058826,
+      "eval_PRM NPV": 0.22784810126582278,
+      "eval_PRM Precision": 0.8148148148148148,
+      "eval_PRM Recall": 0.26506024096385544,
+      "eval_PRM Specificty": 0.782608695652174,
+      "eval_loss": 0.9215198755264282,
+      "eval_runtime": 17.0676,
+      "eval_samples_per_second": 2.578,
+      "eval_steps_per_second": 0.176,
+      "step": 645
+    },
+    {
+      "epoch": 0.08363271515033822,
+      "grad_norm": 3.9553847939873887,
+      "learning_rate": 1.0446313065976715e-06,
+      "loss": 5.0854,
+      "step": 646
+    },
+    {
+      "epoch": 0.08376217755769169,
+      "grad_norm": 3.8843957975852663,
+      "learning_rate": 1.0462483829236742e-06,
+      "loss": 5.2982,
+      "step": 647
+    },
+    {
+      "epoch": 0.08389163996504514,
+      "grad_norm": 2.3196091731352615,
+      "learning_rate": 1.0478654592496766e-06,
+      "loss": 3.854,
+      "step": 648
+    },
+    {
+      "epoch": 0.08402110237239861,
+      "grad_norm": 4.835532044194964,
+      "learning_rate": 1.0494825355756793e-06,
+      "loss": 5.293,
+      "step": 649
+    },
+    {
+      "epoch": 0.08415056477975208,
+      "grad_norm": 2.9182714707524715,
+      "learning_rate": 1.0510996119016818e-06,
+      "loss": 4.1074,
+      "step": 650
+    },
+    {
+      "epoch": 0.08415056477975208,
+      "eval_PRM Accuracy": 0.3867924528301887,
+      "eval_PRM F1": 0.4144144144144144,
+      "eval_PRM F1 AUC": 0.5298585646935569,
+      "eval_PRM F1 Neg": 0.3564356435643564,
+      "eval_PRM NPV": 0.23076923076923078,
+      "eval_PRM Precision": 0.8214285714285714,
+      "eval_PRM Recall": 0.27710843373493976,
+      "eval_PRM Specificty": 0.782608695652174,
+      "eval_loss": 0.9204545617103577,
+      "eval_runtime": 14.3587,
+      "eval_samples_per_second": 3.064,
+      "eval_steps_per_second": 0.209,
+      "step": 650
+    },
+    {
+      "epoch": 0.08428002718710555,
+      "grad_norm": 3.6171615781565603,
+      "learning_rate": 1.0527166882276844e-06,
+      "loss": 4.5073,
+      "step": 651
+    },
+    {
+      "epoch": 0.08440948959445901,
+      "grad_norm": 3.1926590750818096,
+      "learning_rate": 1.0543337645536871e-06,
+      "loss": 4.1917,
+      "step": 652
+    },
+    {
+      "epoch": 0.08453895200181247,
+      "grad_norm": 3.3432764264914288,
+      "learning_rate": 1.0559508408796896e-06,
+      "loss": 4.3716,
+      "step": 653
+    },
+    {
+      "epoch": 0.08466841440916593,
+      "grad_norm": 3.71751725958284,
+      "learning_rate": 1.057567917205692e-06,
+      "loss": 4.2324,
+      "step": 654
+    },
+    {
+      "epoch": 0.0847978768165194,
+      "grad_norm": 1.3069433915514344,
+      "learning_rate": 1.0591849935316947e-06,
+      "loss": 2.7007,
+      "step": 655
+    },
+    {
+      "epoch": 0.0847978768165194,
+      "eval_PRM Accuracy": 0.37735849056603776,
+      "eval_PRM F1": 0.4,
+      "eval_PRM F1 AUC": 0.5238344683080147,
+      "eval_PRM F1 Neg": 0.35294117647058826,
+      "eval_PRM NPV": 0.22784810126582278,
+      "eval_PRM Precision": 0.8148148148148148,
+      "eval_PRM Recall": 0.26506024096385544,
+      "eval_PRM Specificty": 0.782608695652174,
+      "eval_loss": 0.9228515625,
+      "eval_runtime": 14.0529,
+      "eval_samples_per_second": 3.131,
+      "eval_steps_per_second": 0.213,
+      "step": 655
+    },
+    {
+      "epoch": 0.08492733922387287,
+      "grad_norm": 4.182457616046026,
+      "learning_rate": 1.0608020698576974e-06,
+      "loss": 4.8091,
+      "step": 656
+    },
+    {
+      "epoch": 0.08505680163122634,
+      "grad_norm": 2.5033354725795647,
+      "learning_rate": 1.0624191461837e-06,
+      "loss": 3.8281,
+      "step": 657
+    },
+    {
+      "epoch": 0.0851862640385798,
+      "grad_norm": 2.8090170748188514,
+      "learning_rate": 1.0640362225097025e-06,
+      "loss": 4.7661,
+      "step": 658
+    },
+    {
+      "epoch": 0.08531572644593326,
+      "grad_norm": 3.4595552174598208,
+      "learning_rate": 1.0656532988357052e-06,
+      "loss": 4.7891,
+      "step": 659
+    },
+    {
+      "epoch": 0.08544518885328672,
+      "grad_norm": 2.667249572922346,
+      "learning_rate": 1.0672703751617078e-06,
+      "loss": 3.4492,
+      "step": 660
+    },
+    {
+      "epoch": 0.08544518885328672,
+      "eval_PRM Accuracy": 0.37735849056603776,
+      "eval_PRM F1": 0.4,
+      "eval_PRM F1 AUC": 0.5238344683080147,
+      "eval_PRM F1 Neg": 0.35294117647058826,
+      "eval_PRM NPV": 0.22784810126582278,
+      "eval_PRM Precision": 0.8148148148148148,
+      "eval_PRM Recall": 0.26506024096385544,
+      "eval_PRM Specificty": 0.782608695652174,
+      "eval_loss": 0.9223188757896423,
+      "eval_runtime": 14.2847,
+      "eval_samples_per_second": 3.08,
+      "eval_steps_per_second": 0.21,
+      "step": 660
+    },
+    {
+      "epoch": 0.08557465126064019,
+      "grad_norm": 3.815069042013388,
+      "learning_rate": 1.0688874514877103e-06,
+      "loss": 4.5798,
+      "step": 661
+    },
+    {
+      "epoch": 0.08570411366799366,
+      "grad_norm": 3.777846261367631,
+      "learning_rate": 1.070504527813713e-06,
+      "loss": 4.9731,
+      "step": 662
+    },
+    {
+      "epoch": 0.08583357607534713,
+      "grad_norm": 5.213592986227693,
+      "learning_rate": 1.0721216041397154e-06,
+      "loss": 6.7822,
+      "step": 663
+    },
+    {
+      "epoch": 0.08596303848270058,
+      "grad_norm": 2.9773730804335647,
+      "learning_rate": 1.073738680465718e-06,
+      "loss": 3.7539,
+      "step": 664
+    },
+    {
+      "epoch": 0.08609250089005405,
+      "grad_norm": 4.015777203526543,
+      "learning_rate": 1.0753557567917208e-06,
+      "loss": 5.2158,
+      "step": 665
+    },
+    {
+      "epoch": 0.08609250089005405,
+      "eval_PRM Accuracy": 0.36792452830188677,
+      "eval_PRM F1": 0.3853211009174312,
+      "eval_PRM F1 AUC": 0.5178103719224726,
+      "eval_PRM F1 Neg": 0.34951456310679613,
+      "eval_PRM NPV": 0.225,
+      "eval_PRM Precision": 0.8076923076923077,
+      "eval_PRM Recall": 0.25301204819277107,
+      "eval_PRM Specificty": 0.782608695652174,
+      "eval_loss": 0.9247159361839294,
+      "eval_runtime": 21.0285,
+      "eval_samples_per_second": 2.092,
+      "eval_steps_per_second": 0.143,
+      "step": 665
+    },
+    {
+      "epoch": 0.08622196329740751,
+      "grad_norm": 2.65627042409324,
+      "learning_rate": 1.0769728331177232e-06,
+      "loss": 4.1765,
+      "step": 666
+    },
+    {
+      "epoch": 0.08635142570476098,
+      "grad_norm": 5.125522979331907,
+      "learning_rate": 1.0785899094437257e-06,
+      "loss": 6.1421,
+      "step": 667
+    },
+    {
+      "epoch": 0.08648088811211445,
+      "grad_norm": 3.910613713004684,
+      "learning_rate": 1.0802069857697284e-06,
+      "loss": 5.6338,
+      "step": 668
+    },
+    {
+      "epoch": 0.08661035051946792,
+      "grad_norm": 2.728668919911025,
+      "learning_rate": 1.081824062095731e-06,
+      "loss": 4.1968,
+      "step": 669
+    },
+    {
+      "epoch": 0.08673981292682137,
+      "grad_norm": 3.7978918043442955,
+      "learning_rate": 1.0834411384217337e-06,
+      "loss": 4.626,
+      "step": 670
+    },
+    {
+      "epoch": 0.08673981292682137,
+      "eval_PRM Accuracy": 0.36792452830188677,
+      "eval_PRM F1": 0.3853211009174312,
+      "eval_PRM F1 AUC": 0.5178103719224726,
+      "eval_PRM F1 Neg": 0.34951456310679613,
+      "eval_PRM NPV": 0.225,
+      "eval_PRM Precision": 0.8076923076923077,
+      "eval_PRM Recall": 0.25301204819277107,
+      "eval_PRM Specificty": 0.782608695652174,
+      "eval_loss": 0.9281782507896423,
+      "eval_runtime": 14.6199,
+      "eval_samples_per_second": 3.01,
+      "eval_steps_per_second": 0.205,
+      "step": 670
+    },
+    {
+      "epoch": 0.08686927533417484,
+      "grad_norm": 1.7110243603847934,
+      "learning_rate": 1.0850582147477362e-06,
+      "loss": 3.05,
+      "step": 671
+    },
+    {
+      "epoch": 0.0869987377415283,
+      "grad_norm": 6.078163216415924,
+      "learning_rate": 1.0866752910737388e-06,
+      "loss": 6.9224,
+      "step": 672
+    },
+    {
+      "epoch": 0.08712820014888177,
+      "grad_norm": 2.441968873569535,
+      "learning_rate": 1.0882923673997415e-06,
+      "loss": 3.5649,
+      "step": 673
+    },
+    {
+      "epoch": 0.08725766255623524,
+      "grad_norm": 5.334560482449647,
+      "learning_rate": 1.089909443725744e-06,
+      "loss": 5.2378,
+      "step": 674
+    },
+    {
+      "epoch": 0.08738712496358869,
+      "grad_norm": 3.3562273561795313,
+      "learning_rate": 1.0915265200517464e-06,
+      "loss": 4.2126,
+      "step": 675
+    },
+    {
+      "epoch": 0.08738712496358869,
+      "eval_PRM Accuracy": 0.37735849056603776,
+      "eval_PRM F1": 0.4,
+      "eval_PRM F1 AUC": 0.5238344683080147,
+      "eval_PRM F1 Neg": 0.35294117647058826,
+      "eval_PRM NPV": 0.22784810126582278,
+      "eval_PRM Precision": 0.8148148148148148,
+      "eval_PRM Recall": 0.26506024096385544,
+      "eval_PRM Specificty": 0.782608695652174,
+      "eval_loss": 0.927734375,
+      "eval_runtime": 14.9743,
+      "eval_samples_per_second": 2.938,
+      "eval_steps_per_second": 0.2,
+      "step": 675
+    },
+    {
+      "epoch": 0.08751658737094216,
+      "grad_norm": 3.839414456101029,
+      "learning_rate": 1.093143596377749e-06,
+      "loss": 5.8862,
+      "step": 676
+    },
+    {
+      "epoch": 0.08764604977829563,
+      "grad_norm": 1.8936817586298933,
+      "learning_rate": 1.0947606727037518e-06,
+      "loss": 3.3889,
+      "step": 677
+    },
+    {
+      "epoch": 0.0877755121856491,
+      "grad_norm": 5.910427212711068,
+      "learning_rate": 1.0963777490297544e-06,
+      "loss": 7.0693,
+      "step": 678
+    },
+    {
+      "epoch": 0.08790497459300256,
+      "grad_norm": 2.800044073307304,
+      "learning_rate": 1.0979948253557569e-06,
+      "loss": 4.5312,
+      "step": 679
+    },
+    {
+      "epoch": 0.08803443700035603,
+      "grad_norm": 2.4777274689541633,
+      "learning_rate": 1.0996119016817593e-06,
+      "loss": 3.7324,
+      "step": 680
+    },
+    {
+      "epoch": 0.08803443700035603,
+      "eval_PRM Accuracy": 0.3867924528301887,
+      "eval_PRM F1": 0.4144144144144144,
+      "eval_PRM F1 AUC": 0.5298585646935569,
+      "eval_PRM F1 Neg": 0.3564356435643564,
+      "eval_PRM NPV": 0.23076923076923078,
+      "eval_PRM Precision": 0.8214285714285714,
+      "eval_PRM Recall": 0.27710843373493976,
+      "eval_PRM Specificty": 0.782608695652174,
+      "eval_loss": 0.927734375,
+      "eval_runtime": 22.4193,
+      "eval_samples_per_second": 1.963,
+      "eval_steps_per_second": 0.134,
+      "step": 680
+    },
+    {
+      "epoch": 0.08816389940770948,
+      "grad_norm": 5.07762505715575,
+      "learning_rate": 1.101228978007762e-06,
+      "loss": 5.6816,
+      "step": 681
+    },
+    {
+      "epoch": 0.08829336181506295,
+      "grad_norm": 4.122952485942595,
+      "learning_rate": 1.1028460543337647e-06,
+      "loss": 4.4331,
+      "step": 682
+    },
+    {
+      "epoch": 0.08842282422241642,
+      "grad_norm": 5.731260260030666,
+      "learning_rate": 1.1044631306597671e-06,
+      "loss": 5.0352,
+      "step": 683
+    },
+    {
+      "epoch": 0.08855228662976988,
+      "grad_norm": 5.955792580882385,
+      "learning_rate": 1.1060802069857698e-06,
+      "loss": 5.5566,
+      "step": 684
+    },
+    {
+      "epoch": 0.08868174903712335,
+      "grad_norm": 2.8779771736185853,
+      "learning_rate": 1.1076972833117725e-06,
+      "loss": 4.3464,
+      "step": 685
+    },
+    {
+      "epoch": 0.08868174903712335,
+      "eval_PRM Accuracy": 0.37735849056603776,
+      "eval_PRM F1": 0.4,
+      "eval_PRM F1 AUC": 0.5238344683080147,
+      "eval_PRM F1 Neg": 0.35294117647058826,
+      "eval_PRM NPV": 0.22784810126582278,
+      "eval_PRM Precision": 0.8148148148148148,
+      "eval_PRM Recall": 0.26506024096385544,
+      "eval_PRM Specificty": 0.782608695652174,
+      "eval_loss": 0.9293323755264282,
+      "eval_runtime": 14.2059,
+      "eval_samples_per_second": 3.097,
+      "eval_steps_per_second": 0.211,
+      "step": 685
+    },
+    {
+      "epoch": 0.0888112114444768,
+      "grad_norm": 1.6034673869806007,
+      "learning_rate": 1.1093143596377752e-06,
+      "loss": 3.4072,
+      "step": 686
+    },
+    {
+      "epoch": 0.08894067385183027,
+      "grad_norm": 4.895853758787289,
+      "learning_rate": 1.1109314359637776e-06,
+      "loss": 6.3848,
+      "step": 687
+    },
+    {
+      "epoch": 0.08907013625918374,
+      "grad_norm": 3.283050375386971,
+      "learning_rate": 1.11254851228978e-06,
+      "loss": 4.6675,
+      "step": 688
+    },
+    {
+      "epoch": 0.0891995986665372,
+      "grad_norm": 3.408758314451046,
+      "learning_rate": 1.1141655886157827e-06,
+      "loss": 3.4902,
+      "step": 689
+    },
+    {
+      "epoch": 0.08932906107389067,
+      "grad_norm": 4.9801633043257905,
+      "learning_rate": 1.1157826649417854e-06,
+      "loss": 5.9888,
+      "step": 690
+    },
+    {
+      "epoch": 0.08932906107389067,
+      "eval_PRM Accuracy": 0.36792452830188677,
+      "eval_PRM F1": 0.3853211009174312,
+      "eval_PRM F1 AUC": 0.5178103719224726,
+      "eval_PRM F1 Neg": 0.34951456310679613,
+      "eval_PRM NPV": 0.225,
+      "eval_PRM Precision": 0.8076923076923077,
+      "eval_PRM Recall": 0.25301204819277107,
+      "eval_PRM Specificty": 0.782608695652174,
+      "eval_loss": 0.9329723119735718,
+      "eval_runtime": 14.6811,
+      "eval_samples_per_second": 2.997,
+      "eval_steps_per_second": 0.204,
+      "step": 690
+    },
+    {
+      "epoch": 0.08945852348124414,
+      "grad_norm": 3.2296193036231067,
+      "learning_rate": 1.1173997412677879e-06,
+      "loss": 3.9331,
+      "step": 691
+    },
+    {
+      "epoch": 0.0895879858885976,
+      "grad_norm": 4.489150987405942,
+      "learning_rate": 1.1190168175937905e-06,
+      "loss": 4.6777,
+      "step": 692
+    },
+    {
+      "epoch": 0.08971744829595106,
+      "grad_norm": 3.433086525653378,
+      "learning_rate": 1.120633893919793e-06,
+      "loss": 5.6968,
+      "step": 693
+    },
+    {
+      "epoch": 0.08984691070330453,
+      "grad_norm": 2.3814291124617575,
+      "learning_rate": 1.1222509702457957e-06,
+      "loss": 3.2441,
+      "step": 694
+    },
+    {
+      "epoch": 0.089976373110658,
+      "grad_norm": 3.4705775775257206,
+      "learning_rate": 1.1238680465717983e-06,
+      "loss": 4.6189,
+      "step": 695
+    },
+    {
+      "epoch": 0.089976373110658,
+      "eval_PRM Accuracy": 0.3584905660377358,
+      "eval_PRM F1": 0.37037037037037035,
+      "eval_PRM F1 AUC": 0.5117862755369303,
+      "eval_PRM F1 Neg": 0.34615384615384615,
+      "eval_PRM NPV": 0.2222222222222222,
+      "eval_PRM Precision": 0.8,
+      "eval_PRM Recall": 0.24096385542168675,
+      "eval_PRM Specificty": 0.782608695652174,
+      "eval_loss": 0.9323508739471436,
+      "eval_runtime": 14.3791,
+      "eval_samples_per_second": 3.06,
+      "eval_steps_per_second": 0.209,
+      "step": 695
+    },
+    {
+      "epoch": 0.09010583551801146,
+      "grad_norm": 1.4031427780005894,
+      "learning_rate": 1.1254851228978008e-06,
+      "loss": 2.8269,
+      "step": 696
+    },
+    {
+      "epoch": 0.09023529792536492,
+      "grad_norm": 4.84226040077209,
+      "learning_rate": 1.1271021992238035e-06,
+      "loss": 6.5381,
+      "step": 697
+    },
+    {
+      "epoch": 0.09036476033271838,
+      "grad_norm": 2.9830415474694476,
+      "learning_rate": 1.1287192755498062e-06,
+      "loss": 4.415,
+      "step": 698
+    },
+    {
+      "epoch": 0.09049422274007185,
+      "grad_norm": 3.3445875539537977,
+      "learning_rate": 1.1303363518758086e-06,
+      "loss": 4.6343,
+      "step": 699
+    },
+    {
+      "epoch": 0.09062368514742532,
+      "grad_norm": 3.340910486222267,
+      "learning_rate": 1.131953428201811e-06,
+      "loss": 4.5891,
+      "step": 700
+    },
+    {
+      "epoch": 0.09062368514742532,
+      "eval_PRM Accuracy": 0.36792452830188677,
+      "eval_PRM F1": 0.3853211009174312,
+      "eval_PRM F1 AUC": 0.5178103719224726,
+      "eval_PRM F1 Neg": 0.34951456310679613,
+      "eval_PRM NPV": 0.225,
+      "eval_PRM Precision": 0.8076923076923077,
+      "eval_PRM Recall": 0.25301204819277107,
+      "eval_PRM Specificty": 0.782608695652174,
+      "eval_loss": 0.9374112486839294,
+      "eval_runtime": 14.1662,
+      "eval_samples_per_second": 3.106,
+      "eval_steps_per_second": 0.212,
+      "step": 700
+    },
+    {
+      "epoch": 0.09075314755477878,
+      "grad_norm": 3.291208314334283,
+      "learning_rate": 1.1335705045278137e-06,
+      "loss": 4.3103,
+      "step": 701
+    },
+    {
+      "epoch": 0.09088260996213225,
+      "grad_norm": 3.7902054579688427,
+      "learning_rate": 1.1351875808538164e-06,
+      "loss": 4.4829,
+      "step": 702
+    },
+    {
+      "epoch": 0.0910120723694857,
+      "grad_norm": 3.8765634704855967,
+      "learning_rate": 1.136804657179819e-06,
+      "loss": 4.8525,
+      "step": 703
+    },
+    {
+      "epoch": 0.09114153477683917,
+      "grad_norm": 3.477765893551626,
+      "learning_rate": 1.1384217335058215e-06,
+      "loss": 4.377,
+      "step": 704
+    },
+    {
+      "epoch": 0.09127099718419264,
+      "grad_norm": 4.316002808739543,
+      "learning_rate": 1.1400388098318242e-06,
+      "loss": 5.2102,
+      "step": 705
+    },
+    {
+      "epoch": 0.09127099718419264,
+      "eval_PRM Accuracy": 0.3584905660377358,
+      "eval_PRM F1": 0.37037037037037035,
+      "eval_PRM F1 AUC": 0.5117862755369303,
+      "eval_PRM F1 Neg": 0.34615384615384615,
+      "eval_PRM NPV": 0.2222222222222222,
+      "eval_PRM Precision": 0.8,
+      "eval_PRM Recall": 0.24096385542168675,
+      "eval_PRM Specificty": 0.782608695652174,
+      "eval_loss": 0.9383878111839294,
+      "eval_runtime": 15.1852,
+      "eval_samples_per_second": 2.898,
+      "eval_steps_per_second": 0.198,
+      "step": 705
+    },
+    {
+      "epoch": 0.09140045959154611,
+      "grad_norm": 5.333892145532132,
+      "learning_rate": 1.1416558861578267e-06,
+      "loss": 6.0728,
+      "step": 706
+    },
+    {
+      "epoch": 0.09152992199889957,
+      "grad_norm": 4.117906116304779,
+      "learning_rate": 1.1432729624838293e-06,
+      "loss": 4.6228,
+      "step": 707
+    },
+    {
+      "epoch": 0.09165938440625303,
+      "grad_norm": 5.105444072776924,
+      "learning_rate": 1.144890038809832e-06,
+      "loss": 6.3406,
+      "step": 708
+    },
+    {
+      "epoch": 0.0917888468136065,
+      "grad_norm": 2.8326974320088505,
+      "learning_rate": 1.1465071151358345e-06,
+      "loss": 3.3999,
+      "step": 709
+    },
+    {
+      "epoch": 0.09191830922095996,
+      "grad_norm": 4.041014890890666,
+      "learning_rate": 1.1481241914618371e-06,
+      "loss": 5.6228,
+      "step": 710
+    },
+    {
+      "epoch": 0.09191830922095996,
+      "eval_PRM Accuracy": 0.3490566037735849,
+      "eval_PRM F1": 0.35514018691588783,
+      "eval_PRM F1 AUC": 0.5057621791513882,
+      "eval_PRM F1 Neg": 0.34285714285714286,
+      "eval_PRM NPV": 0.21951219512195122,
+      "eval_PRM Precision": 0.7916666666666666,
+      "eval_PRM Recall": 0.2289156626506024,
+      "eval_PRM Specificty": 0.782608695652174,
+      "eval_loss": 0.9406960010528564,
+      "eval_runtime": 15.2254,
+      "eval_samples_per_second": 2.89,
+      "eval_steps_per_second": 0.197,
+      "step": 710
+    },
+    {
+      "epoch": 0.09204777162831343,
+      "grad_norm": 4.397996167048664,
+      "learning_rate": 1.1497412677878398e-06,
+      "loss": 5.6943,
+      "step": 711
+    },
+    {
+      "epoch": 0.0921772340356669,
+      "grad_norm": 3.8179337999486886,
+      "learning_rate": 1.1513583441138423e-06,
+      "loss": 4.103,
+      "step": 712
+    },
+    {
+      "epoch": 0.09230669644302036,
+      "grad_norm": 3.8626218672791293,
+      "learning_rate": 1.1529754204398447e-06,
+      "loss": 4.5566,
+      "step": 713
+    },
+    {
+      "epoch": 0.09243615885037382,
+      "grad_norm": 3.3837870996613777,
+      "learning_rate": 1.1545924967658474e-06,
+      "loss": 4.0381,
+      "step": 714
+    },
+    {
+      "epoch": 0.09256562125772728,
+      "grad_norm": 3.5064318702253794,
+      "learning_rate": 1.15620957309185e-06,
+      "loss": 4.5518,
+      "step": 715
+    },
+    {
+      "epoch": 0.09256562125772728,
+      "eval_PRM Accuracy": 0.3584905660377358,
+      "eval_PRM F1": 0.37037037037037035,
+      "eval_PRM F1 AUC": 0.5117862755369303,
+      "eval_PRM F1 Neg": 0.34615384615384615,
+      "eval_PRM NPV": 0.2222222222222222,
+      "eval_PRM Precision": 0.8,
+      "eval_PRM Recall": 0.24096385542168675,
+      "eval_PRM Specificty": 0.782608695652174,
+      "eval_loss": 0.9413174986839294,
+      "eval_runtime": 13.7446,
+      "eval_samples_per_second": 3.201,
+      "eval_steps_per_second": 0.218,
+      "step": 715
+    },
+    {
+      "epoch": 0.09269508366508075,
+      "grad_norm": 1.4254962179750599,
+      "learning_rate": 1.1578266494178527e-06,
+      "loss": 2.6558,
+      "step": 716
+    },
+    {
+      "epoch": 0.09282454607243422,
+      "grad_norm": 2.7181580173659254,
+      "learning_rate": 1.1594437257438552e-06,
+      "loss": 3.9551,
+      "step": 717
+    },
+    {
+      "epoch": 0.09295400847978769,
+      "grad_norm": 2.1720428930117883,
+      "learning_rate": 1.1610608020698579e-06,
+      "loss": 3.2974,
+      "step": 718
+    },
+    {
+      "epoch": 0.09308347088714114,
+      "grad_norm": 3.335922023004612,
+      "learning_rate": 1.1626778783958603e-06,
+      "loss": 4.5981,
+      "step": 719
+    },
+    {
+      "epoch": 0.09321293329449461,
+      "grad_norm": 2.5720838066439913,
+      "learning_rate": 1.164294954721863e-06,
+      "loss": 3.5359,
+      "step": 720
+    },
+    {
+      "epoch": 0.09321293329449461,
+      "eval_PRM Accuracy": 0.3490566037735849,
+      "eval_PRM F1": 0.35514018691588783,
+      "eval_PRM F1 AUC": 0.5057621791513882,
+      "eval_PRM F1 Neg": 0.34285714285714286,
+      "eval_PRM NPV": 0.21951219512195122,
+      "eval_PRM Precision": 0.7916666666666666,
+      "eval_PRM Recall": 0.2289156626506024,
+      "eval_PRM Specificty": 0.782608695652174,
+      "eval_loss": 0.9451349377632141,
+      "eval_runtime": 14.817,
+      "eval_samples_per_second": 2.97,
+      "eval_steps_per_second": 0.202,
+      "step": 720
+    },
+    {
+      "epoch": 0.09334239570184807,
+      "grad_norm": 4.412096999745951,
+      "learning_rate": 1.1659120310478655e-06,
+      "loss": 5.0664,
+      "step": 721
+    },
+    {
+      "epoch": 0.09347185810920154,
+      "grad_norm": 4.043391021692804,
+      "learning_rate": 1.1675291073738681e-06,
+      "loss": 4.8105,
+      "step": 722
+    },
+    {
+      "epoch": 0.09360132051655501,
+      "grad_norm": 3.8852944956904394,
+      "learning_rate": 1.1691461836998708e-06,
+      "loss": 4.6001,
+      "step": 723
+    },
+    {
+      "epoch": 0.09373078292390848,
+      "grad_norm": 3.61702897503039,
+      "learning_rate": 1.1707632600258735e-06,
+      "loss": 4.9185,
+      "step": 724
+    },
+    {
+      "epoch": 0.09386024533126193,
+      "grad_norm": 3.770457660094383,
+      "learning_rate": 1.172380336351876e-06,
+      "loss": 4.9395,
+      "step": 725
+    },
+    {
+      "epoch": 0.09386024533126193,
+      "eval_PRM Accuracy": 0.3490566037735849,
+      "eval_PRM F1": 0.35514018691588783,
+      "eval_PRM F1 AUC": 0.5057621791513882,
+      "eval_PRM F1 Neg": 0.34285714285714286,
+      "eval_PRM NPV": 0.21951219512195122,
+      "eval_PRM Precision": 0.7916666666666666,
+      "eval_PRM Recall": 0.2289156626506024,
+      "eval_PRM Specificty": 0.782608695652174,
+      "eval_loss": 0.9439808130264282,
+      "eval_runtime": 14.3355,
+      "eval_samples_per_second": 3.069,
+      "eval_steps_per_second": 0.209,
+      "step": 725
+    },
+    {
+      "epoch": 0.0939897077386154,
+      "grad_norm": 2.422357203887003,
+      "learning_rate": 1.1739974126778784e-06,
+      "loss": 3.4121,
+      "step": 726
+    },
+    {
+      "epoch": 0.09411917014596886,
+      "grad_norm": 3.2641001532898124,
+      "learning_rate": 1.175614489003881e-06,
+      "loss": 4.0156,
+      "step": 727
+    },
+    {
+      "epoch": 0.09424863255332233,
+      "grad_norm": 3.056422956455596,
+      "learning_rate": 1.1772315653298837e-06,
+      "loss": 3.2949,
+      "step": 728
+    },
+    {
+      "epoch": 0.0943780949606758,
+      "grad_norm": 4.2993300475915355,
+      "learning_rate": 1.1788486416558862e-06,
+      "loss": 5.209,
+      "step": 729
+    },
+    {
+      "epoch": 0.09450755736802925,
+      "grad_norm": 3.7276126322748278,
+      "learning_rate": 1.1804657179818889e-06,
+      "loss": 4.8374,
+      "step": 730
+    },
+    {
+      "epoch": 0.09450755736802925,
+      "eval_PRM Accuracy": 0.3490566037735849,
+      "eval_PRM F1": 0.35514018691588783,
+      "eval_PRM F1 AUC": 0.5057621791513882,
+      "eval_PRM F1 Neg": 0.34285714285714286,
+      "eval_PRM NPV": 0.21951219512195122,
+      "eval_PRM Precision": 0.7916666666666666,
+      "eval_PRM Recall": 0.2289156626506024,
+      "eval_PRM Specificty": 0.782608695652174,
+      "eval_loss": 0.9463778138160706,
+      "eval_runtime": 13.9681,
+      "eval_samples_per_second": 3.15,
+      "eval_steps_per_second": 0.215,
+      "step": 730
+    },
+    {
+      "epoch": 0.09463701977538272,
+      "grad_norm": 3.42735558167999,
+      "learning_rate": 1.1820827943078915e-06,
+      "loss": 4.6589,
+      "step": 731
+    },
+    {
+      "epoch": 0.09476648218273619,
+      "grad_norm": 2.7507527801419918,
+      "learning_rate": 1.183699870633894e-06,
+      "loss": 3.436,
+      "step": 732
+    },
+    {
+      "epoch": 0.09489594459008965,
+      "grad_norm": 2.8666509025023625,
+      "learning_rate": 1.1853169469598967e-06,
+      "loss": 4.2251,
+      "step": 733
+    },
+    {
+      "epoch": 0.09502540699744312,
+      "grad_norm": 4.1237553399909475,
+      "learning_rate": 1.1869340232858991e-06,
+      "loss": 4.8445,
+      "step": 734
+    },
+    {
+      "epoch": 0.09515486940479659,
+      "grad_norm": 4.049456218972724,
+      "learning_rate": 1.1885510996119018e-06,
+      "loss": 5.0388,
+      "step": 735
+    },
+    {
+      "epoch": 0.09515486940479659,
+      "eval_PRM Accuracy": 0.3490566037735849,
+      "eval_PRM F1": 0.35514018691588783,
+      "eval_PRM F1 AUC": 0.5057621791513882,
+      "eval_PRM F1 Neg": 0.34285714285714286,
+      "eval_PRM NPV": 0.21951219512195122,
+      "eval_PRM Precision": 0.7916666666666666,
+      "eval_PRM Recall": 0.2289156626506024,
+      "eval_PRM Specificty": 0.782608695652174,
+      "eval_loss": 0.9477983117103577,
+      "eval_runtime": 14.6157,
+      "eval_samples_per_second": 3.01,
+      "eval_steps_per_second": 0.205,
+      "step": 735
+    },
+    {
+      "epoch": 0.09528433181215004,
+      "grad_norm": 2.8678986489506197,
+      "learning_rate": 1.1901681759379045e-06,
+      "loss": 3.9961,
+      "step": 736
+    },
+    {
+      "epoch": 0.09541379421950351,
+      "grad_norm": 4.527628480815001,
+      "learning_rate": 1.191785252263907e-06,
+      "loss": 6.0391,
+      "step": 737
+    },
+    {
+      "epoch": 0.09554325662685698,
+      "grad_norm": 2.6501271154285684,
+      "learning_rate": 1.1934023285899094e-06,
+      "loss": 3.6812,
+      "step": 738
+    },
+    {
+      "epoch": 0.09567271903421044,
+      "grad_norm": 6.202858843561828,
+      "learning_rate": 1.195019404915912e-06,
+      "loss": 6.6831,
+      "step": 739
+    },
+    {
+      "epoch": 0.09580218144156391,
+      "grad_norm": 3.4141775212994996,
+      "learning_rate": 1.1966364812419147e-06,
+      "loss": 3.7017,
+      "step": 740
+    },
+    {
+      "epoch": 0.09580218144156391,
+      "eval_PRM Accuracy": 0.3490566037735849,
+      "eval_PRM F1": 0.35514018691588783,
+      "eval_PRM F1 AUC": 0.5057621791513882,
+      "eval_PRM F1 Neg": 0.34285714285714286,
+      "eval_PRM NPV": 0.21951219512195122,
+      "eval_PRM Precision": 0.7916666666666666,
+      "eval_PRM Recall": 0.2289156626506024,
+      "eval_PRM Specificty": 0.782608695652174,
+      "eval_loss": 0.9508167505264282,
+      "eval_runtime": 14.3734,
+      "eval_samples_per_second": 3.061,
+      "eval_steps_per_second": 0.209,
+      "step": 740
+    },
+    {
+      "epoch": 0.09593164384891736,
+      "grad_norm": 3.3812200832346067,
+      "learning_rate": 1.1982535575679174e-06,
+      "loss": 4.1841,
+      "step": 741
+    },
+    {
+      "epoch": 0.09606110625627083,
+      "grad_norm": 2.839737276307249,
+      "learning_rate": 1.1998706338939199e-06,
+      "loss": 4.333,
+      "step": 742
+    },
+    {
+      "epoch": 0.0961905686636243,
+      "grad_norm": 3.7117497819259992,
+      "learning_rate": 1.2014877102199225e-06,
+      "loss": 4.0034,
+      "step": 743
+    },
+    {
+      "epoch": 0.09632003107097777,
+      "grad_norm": 1.3147220409495661,
+      "learning_rate": 1.2031047865459252e-06,
+      "loss": 2.5254,
+      "step": 744
+    },
+    {
+      "epoch": 0.09644949347833123,
+      "grad_norm": 3.06151101114409,
+      "learning_rate": 1.2047218628719277e-06,
+      "loss": 3.8535,
+      "step": 745
+    },
+    {
+      "epoch": 0.09644949347833123,
+      "eval_PRM Accuracy": 0.3490566037735849,
+      "eval_PRM F1": 0.35514018691588783,
+      "eval_PRM F1 AUC": 0.5057621791513882,
+      "eval_PRM F1 Neg": 0.34285714285714286,
+      "eval_PRM NPV": 0.21951219512195122,
+      "eval_PRM Precision": 0.7916666666666666,
+      "eval_PRM Recall": 0.2289156626506024,
+      "eval_PRM Specificty": 0.782608695652174,
+      "eval_loss": 0.9534801244735718,
+      "eval_runtime": 14.1651,
+      "eval_samples_per_second": 3.106,
+      "eval_steps_per_second": 0.212,
+      "step": 745
+    },
+    {
+      "epoch": 0.0965789558856847,
+      "grad_norm": 2.777819300257185,
+      "learning_rate": 1.2063389391979301e-06,
+      "loss": 3.231,
+      "step": 746
+    },
+    {
+      "epoch": 0.09670841829303815,
+      "grad_norm": 2.230023110319025,
+      "learning_rate": 1.2079560155239328e-06,
+      "loss": 3.4385,
+      "step": 747
+    },
+    {
+      "epoch": 0.09683788070039162,
+      "grad_norm": 4.259154743643278,
+      "learning_rate": 1.2095730918499355e-06,
+      "loss": 4.5786,
+      "step": 748
+    },
+    {
+      "epoch": 0.09696734310774509,
+      "grad_norm": 2.4180331921350446,
+      "learning_rate": 1.2111901681759381e-06,
+      "loss": 3.585,
+      "step": 749
+    },
+    {
+      "epoch": 0.09709680551509856,
+      "grad_norm": 1.8441265887393403,
+      "learning_rate": 1.2128072445019406e-06,
+      "loss": 3.1641,
+      "step": 750
+    },
+    {
+      "epoch": 0.09709680551509856,
+      "eval_PRM Accuracy": 0.33962264150943394,
+      "eval_PRM F1": 0.33962264150943394,
+      "eval_PRM F1 AUC": 0.499738082765846,
+      "eval_PRM F1 Neg": 0.33962264150943394,
+      "eval_PRM NPV": 0.21686746987951808,
+      "eval_PRM Precision": 0.782608695652174,
+      "eval_PRM Recall": 0.21686746987951808,
+      "eval_PRM Specificty": 0.782608695652174,
+      "eval_loss": 0.9540128111839294,
+      "eval_runtime": 13.4035,
+      "eval_samples_per_second": 3.283,
+      "eval_steps_per_second": 0.224,
+      "step": 750
+    },
+    {
+      "epoch": 0.09722626792245202,
+      "grad_norm": 2.5901463716558855,
+      "learning_rate": 1.214424320827943e-06,
+      "loss": 3.4058,
+      "step": 751
+    },
+    {
+      "epoch": 0.09735573032980548,
+      "grad_norm": 4.921828335737435,
+      "learning_rate": 1.2160413971539457e-06,
+      "loss": 5.6348,
+      "step": 752
+    },
+    {
+      "epoch": 0.09748519273715894,
+      "grad_norm": 5.598306921120383,
+      "learning_rate": 1.2176584734799484e-06,
+      "loss": 5.4517,
+      "step": 753
+    },
+    {
+      "epoch": 0.09761465514451241,
+      "grad_norm": 3.0004463970660735,
+      "learning_rate": 1.219275549805951e-06,
+      "loss": 3.6851,
+      "step": 754
+    },
+    {
+      "epoch": 0.09774411755186588,
+      "grad_norm": 3.023154824328046,
+      "learning_rate": 1.2208926261319535e-06,
+      "loss": 3.8838,
+      "step": 755
+    },
+    {
+      "epoch": 0.09774411755186588,
+      "eval_PRM Accuracy": 0.33962264150943394,
+      "eval_PRM F1": 0.33962264150943394,
+      "eval_PRM F1 AUC": 0.499738082765846,
+      "eval_PRM F1 Neg": 0.33962264150943394,
+      "eval_PRM NPV": 0.21686746987951808,
+      "eval_PRM Precision": 0.782608695652174,
+      "eval_PRM Recall": 0.21686746987951808,
+      "eval_PRM Specificty": 0.782608695652174,
+      "eval_loss": 0.9588068127632141,
+      "eval_runtime": 14.57,
+      "eval_samples_per_second": 3.02,
+      "eval_steps_per_second": 0.206,
+      "step": 755
+    },
+    {
+      "epoch": 0.09787357995921935,
+      "grad_norm": 3.3491400623033645,
+      "learning_rate": 1.2225097024579562e-06,
+      "loss": 4.5225,
+      "step": 756
+    },
+    {
+      "epoch": 0.09800304236657281,
+      "grad_norm": 2.4926367191547794,
+      "learning_rate": 1.2241267787839586e-06,
+      "loss": 3.6133,
+      "step": 757
+    },
+    {
+      "epoch": 0.09813250477392627,
+      "grad_norm": 5.250129071006987,
+      "learning_rate": 1.2257438551099613e-06,
+      "loss": 5.2407,
+      "step": 758
+    },
+    {
+      "epoch": 0.09826196718127973,
+      "grad_norm": 4.938303635815838,
+      "learning_rate": 1.2273609314359638e-06,
+      "loss": 5.2261,
+      "step": 759
+    },
+    {
+      "epoch": 0.0983914295886332,
+      "grad_norm": 2.892426500916422,
+      "learning_rate": 1.2289780077619664e-06,
+      "loss": 4.4419,
+      "step": 760
+    },
+    {
+      "epoch": 0.0983914295886332,
+      "eval_PRM Accuracy": 0.33962264150943394,
+      "eval_PRM F1": 0.33962264150943394,
+      "eval_PRM F1 AUC": 0.499738082765846,
+      "eval_PRM F1 Neg": 0.33962264150943394,
+      "eval_PRM NPV": 0.21686746987951808,
+      "eval_PRM Precision": 0.782608695652174,
+      "eval_PRM Recall": 0.21686746987951808,
+      "eval_PRM Specificty": 0.782608695652174,
+      "eval_loss": 0.9600496888160706,
+      "eval_runtime": 14.8324,
+      "eval_samples_per_second": 2.966,
+      "eval_steps_per_second": 0.202,
+      "step": 760
+    },
+    {
+      "epoch": 0.09852089199598667,
+      "grad_norm": 4.178198556986774,
+      "learning_rate": 1.2305950840879691e-06,
+      "loss": 5.0015,
+      "step": 761
+    },
+    {
+      "epoch": 0.09865035440334013,
+      "grad_norm": 3.0622134145854276,
+      "learning_rate": 1.2322121604139718e-06,
+      "loss": 4.0886,
+      "step": 762
+    },
+    {
+      "epoch": 0.09877981681069359,
+      "grad_norm": 5.34613239740492,
+      "learning_rate": 1.2338292367399742e-06,
+      "loss": 5.3184,
+      "step": 763
+    },
+    {
+      "epoch": 0.09890927921804706,
+      "grad_norm": 3.8116626392907333,
+      "learning_rate": 1.2354463130659767e-06,
+      "loss": 5.4507,
+      "step": 764
+    },
+    {
+      "epoch": 0.09903874162540052,
+      "grad_norm": 2.9408501223754486,
+      "learning_rate": 1.2370633893919794e-06,
+      "loss": 3.7075,
+      "step": 765
+    },
+    {
+      "epoch": 0.09903874162540052,
+      "eval_PRM Accuracy": 0.3490566037735849,
+      "eval_PRM F1": 0.35514018691588783,
+      "eval_PRM F1 AUC": 0.5057621791513882,
+      "eval_PRM F1 Neg": 0.34285714285714286,
+      "eval_PRM NPV": 0.21951219512195122,
+      "eval_PRM Precision": 0.7916666666666666,
+      "eval_PRM Recall": 0.2289156626506024,
+      "eval_PRM Specificty": 0.782608695652174,
+      "eval_loss": 0.9600496888160706,
+      "eval_runtime": 14.1306,
+      "eval_samples_per_second": 3.114,
+      "eval_steps_per_second": 0.212,
+      "step": 765
+    },
+    {
+      "epoch": 0.09916820403275399,
+      "grad_norm": 6.713956462957203,
+      "learning_rate": 1.238680465717982e-06,
+      "loss": 6.6514,
+      "step": 766
+    },
+    {
+      "epoch": 0.09929766644010746,
+      "grad_norm": 3.7341104005926904,
+      "learning_rate": 1.2402975420439845e-06,
+      "loss": 4.0283,
+      "step": 767
+    },
+    {
+      "epoch": 0.09942712884746092,
+      "grad_norm": 4.316832455260363,
+      "learning_rate": 1.2419146183699872e-06,
+      "loss": 4.8613,
+      "step": 768
+    },
+    {
+      "epoch": 0.09955659125481438,
+      "grad_norm": 4.507170174269931,
+      "learning_rate": 1.2435316946959898e-06,
+      "loss": 5.1484,
+      "step": 769
+    },
+    {
+      "epoch": 0.09968605366216785,
+      "grad_norm": 6.561329419759221,
+      "learning_rate": 1.2451487710219923e-06,
+      "loss": 6.2466,
+      "step": 770
+    },
+    {
+      "epoch": 0.09968605366216785,
+      "eval_PRM Accuracy": 0.33962264150943394,
+      "eval_PRM F1": 0.33962264150943394,
+      "eval_PRM F1 AUC": 0.499738082765846,
+      "eval_PRM F1 Neg": 0.33962264150943394,
+      "eval_PRM NPV": 0.21686746987951808,
+      "eval_PRM Precision": 0.782608695652174,
+      "eval_PRM Recall": 0.21686746987951808,
+      "eval_PRM Specificty": 0.782608695652174,
+      "eval_loss": 0.9601384997367859,
+      "eval_runtime": 14.9473,
+      "eval_samples_per_second": 2.944,
+      "eval_steps_per_second": 0.201,
+      "step": 770
+    },
+    {
+      "epoch": 0.09981551606952131,
+      "grad_norm": 6.469504466095085,
+      "learning_rate": 1.246765847347995e-06,
+      "loss": 6.5288,
+      "step": 771
+    },
+    {
+      "epoch": 0.09994497847687478,
+      "grad_norm": 4.861047041568011,
+      "learning_rate": 1.2483829236739974e-06,
+      "loss": 5.3623,
+      "step": 772
+    },
+    {
+      "epoch": 0.10007444088422825,
+      "grad_norm": 5.996185337759073,
+      "learning_rate": 1.25e-06,
+      "loss": 6.0386,
+      "step": 773
+    },
+    {
+      "epoch": 0.1002039032915817,
+      "grad_norm": 2.1163860189694,
+      "learning_rate": 1.249999936165541e-06,
+      "loss": 3.3276,
+      "step": 774
+    },
+    {
+      "epoch": 0.10033336569893517,
+      "grad_norm": 5.838219278177589,
+      "learning_rate": 1.2499997446621767e-06,
+      "loss": 5.8081,
+      "step": 775
+    },
+    {
+      "epoch": 0.10033336569893517,
+      "eval_PRM Accuracy": 0.3490566037735849,
+      "eval_PRM F1": 0.35514018691588783,
+      "eval_PRM F1 AUC": 0.5057621791513882,
+      "eval_PRM F1 Neg": 0.34285714285714286,
+      "eval_PRM NPV": 0.21951219512195122,
+      "eval_PRM Precision": 0.7916666666666666,
+      "eval_PRM Recall": 0.2289156626506024,
+      "eval_PRM Specificty": 0.782608695652174,
+      "eval_loss": 0.9632457494735718,
+      "eval_runtime": 14.1007,
+      "eval_samples_per_second": 3.12,
+      "eval_steps_per_second": 0.213,
+      "step": 775
+    },
+    {
+      "epoch": 0.10046282810628863,
+      "grad_norm": 2.2235757549701107,
+      "learning_rate": 1.2499994254899465e-06,
+      "loss": 3.8042,
+      "step": 776
+    },
+    {
+      "epoch": 0.1005922905136421,
+      "grad_norm": 4.50003441955947,
+      "learning_rate": 1.2499989786489155e-06,
+      "loss": 5.3418,
+      "step": 777
+    },
+    {
+      "epoch": 0.10072175292099557,
+      "grad_norm": 5.251162335967897,
+      "learning_rate": 1.249998404139175e-06,
+      "loss": 6.5532,
+      "step": 778
+    },
+    {
+      "epoch": 0.10085121532834904,
+      "grad_norm": 3.6217613110813027,
+      "learning_rate": 1.2499977019608421e-06,
+      "loss": 4.2666,
+      "step": 779
+    },
+    {
+      "epoch": 0.10098067773570249,
+      "grad_norm": 6.651999256435771,
+      "learning_rate": 1.2499968721140606e-06,
+      "loss": 5.873,
+      "step": 780
+    },
+    {
+      "epoch": 0.10098067773570249,
+      "eval_PRM Accuracy": 0.33962264150943394,
+      "eval_PRM F1": 0.33962264150943394,
+      "eval_PRM F1 AUC": 0.499738082765846,
+      "eval_PRM F1 Neg": 0.33962264150943394,
+      "eval_PRM NPV": 0.21686746987951808,
+      "eval_PRM Precision": 0.782608695652174,
+      "eval_PRM Recall": 0.21686746987951808,
+      "eval_PRM Specificty": 0.782608695652174,
+      "eval_loss": 0.9674183130264282,
+      "eval_runtime": 13.1967,
+      "eval_samples_per_second": 3.334,
+      "eval_steps_per_second": 0.227,
+      "step": 780
+    },
+    {
+      "epoch": 0.10111014014305596,
+      "grad_norm": 1.941792858648424,
+      "learning_rate": 1.2499959145989998e-06,
+      "loss": 3.707,
+      "step": 781
+    },
+    {
+      "epoch": 0.10123960255040942,
+      "grad_norm": 1.5217448607721415,
+      "learning_rate": 1.2499948294158554e-06,
+      "loss": 2.7529,
+      "step": 782
+    },
+    {
+      "epoch": 0.10136906495776289,
+      "grad_norm": 6.527780107406759,
+      "learning_rate": 1.249993616564849e-06,
+      "loss": 6.4536,
+      "step": 783
+    },
+    {
+      "epoch": 0.10149852736511636,
+      "grad_norm": 4.506007033579019,
+      "learning_rate": 1.2499922760462283e-06,
+      "loss": 4.9507,
+      "step": 784
+    },
+    {
+      "epoch": 0.10162798977246981,
+      "grad_norm": 4.163520430297341,
+      "learning_rate": 1.2499908078602673e-06,
+      "loss": 4.0503,
+      "step": 785
+    },
+    {
+      "epoch": 0.10162798977246981,
+      "eval_PRM Accuracy": 0.33962264150943394,
+      "eval_PRM F1": 0.33962264150943394,
+      "eval_PRM F1 AUC": 0.499738082765846,
+      "eval_PRM F1 Neg": 0.33962264150943394,
+      "eval_PRM NPV": 0.21686746987951808,
+      "eval_PRM Precision": 0.782608695652174,
+      "eval_PRM Recall": 0.21686746987951808,
+      "eval_PRM Specificty": 0.782608695652174,
+      "eval_loss": 0.9685724377632141,
+      "eval_runtime": 13.1939,
+      "eval_samples_per_second": 3.335,
+      "eval_steps_per_second": 0.227,
+      "step": 785
+    },
+    {
+      "epoch": 0.10175745217982328,
+      "grad_norm": 4.349718521857898,
+      "learning_rate": 1.249989212007266e-06,
+      "loss": 4.1675,
+      "step": 786
+    },
+    {
+      "epoch": 0.10188691458717675,
+      "grad_norm": 3.9889110504572254,
+      "learning_rate": 1.2499874884875499e-06,
+      "loss": 4.5693,
+      "step": 787
+    },
+    {
+      "epoch": 0.10201637699453021,
+      "grad_norm": 4.382411355963895,
+      "learning_rate": 1.2499856373014715e-06,
+      "loss": 5.7251,
+      "step": 788
+    },
+    {
+      "epoch": 0.10214583940188368,
+      "grad_norm": 5.396796100788135,
+      "learning_rate": 1.2499836584494087e-06,
+      "loss": 5.2334,
+      "step": 789
+    },
+    {
+      "epoch": 0.10227530180923715,
+      "grad_norm": 1.6618567265559632,
+      "learning_rate": 1.2499815519317659e-06,
+      "loss": 2.562,
+      "step": 790
+    },
+    {
+      "epoch": 0.10227530180923715,
+      "eval_PRM Accuracy": 0.330188679245283,
+      "eval_PRM F1": 0.3238095238095238,
+      "eval_PRM F1 AUC": 0.4937139863803038,
+      "eval_PRM F1 Neg": 0.3364485981308411,
+      "eval_PRM NPV": 0.21428571428571427,
+      "eval_PRM Precision": 0.7727272727272727,
+      "eval_PRM Recall": 0.20481927710843373,
+      "eval_PRM Specificty": 0.782608695652174,
+      "eval_loss": 0.9685724377632141,
+      "eval_runtime": 14.7448,
+      "eval_samples_per_second": 2.984,
+      "eval_steps_per_second": 0.203,
+      "step": 790
+    },
+    {
+      "epoch": 0.1024047642165906,
+      "grad_norm": 1.4166413772022242,
+      "learning_rate": 1.2499793177489731e-06,
+      "loss": 3.0095,
+      "step": 791
+    },
+    {
+      "epoch": 0.10253422662394407,
+      "grad_norm": 3.252997108285663,
+      "learning_rate": 1.249976955901487e-06,
+      "loss": 4.3838,
+      "step": 792
+    },
+    {
+      "epoch": 0.10266368903129754,
+      "grad_norm": 2.702674167176806,
+      "learning_rate": 1.2499744663897902e-06,
+      "loss": 4.156,
+      "step": 793
+    },
+    {
+      "epoch": 0.102793151438651,
+      "grad_norm": 3.7037228796285944,
+      "learning_rate": 1.2499718492143908e-06,
+      "loss": 4.1729,
+      "step": 794
+    },
+    {
+      "epoch": 0.10292261384600447,
+      "grad_norm": 4.174273470864699,
+      "learning_rate": 1.2499691043758235e-06,
+      "loss": 4.1172,
+      "step": 795
+    },
+    {
+      "epoch": 0.10292261384600447,
+      "eval_PRM Accuracy": 0.33962264150943394,
+      "eval_PRM F1": 0.33962264150943394,
+      "eval_PRM F1 AUC": 0.499738082765846,
+      "eval_PRM F1 Neg": 0.33962264150943394,
+      "eval_PRM NPV": 0.21686746987951808,
+      "eval_PRM Precision": 0.782608695652174,
+      "eval_PRM Recall": 0.21686746987951808,
+      "eval_PRM Specificty": 0.782608695652174,
+      "eval_loss": 0.9720348119735718,
+      "eval_runtime": 13.008,
+      "eval_samples_per_second": 3.383,
+      "eval_steps_per_second": 0.231,
+      "step": 795
+    },
+    {
+      "epoch": 0.10305207625335792,
+      "grad_norm": 3.068136111633999,
+      "learning_rate": 1.2499662318746493e-06,
+      "loss": 4.3054,
+      "step": 796
+    },
+    {
+      "epoch": 0.10318153866071139,
+      "grad_norm": 4.518625817828032,
+      "learning_rate": 1.2499632317114545e-06,
+      "loss": 4.9478,
+      "step": 797
+    },
+    {
+      "epoch": 0.10331100106806486,
+      "grad_norm": 3.2489296991270202,
+      "learning_rate": 1.2499601038868525e-06,
+      "loss": 4.1729,
+      "step": 798
+    },
+    {
+      "epoch": 0.10344046347541833,
+      "grad_norm": 4.141141218186075,
+      "learning_rate": 1.2499568484014818e-06,
+      "loss": 3.8232,
+      "step": 799
+    },
+    {
+      "epoch": 0.1035699258827718,
+      "grad_norm": 4.319817471531137,
+      "learning_rate": 1.2499534652560074e-06,
+      "loss": 5.1089,
+      "step": 800
+    },
+    {
+      "epoch": 0.1035699258827718,
+      "eval_PRM Accuracy": 0.330188679245283,
+      "eval_PRM F1": 0.3238095238095238,
+      "eval_PRM F1 AUC": 0.4937139863803038,
+      "eval_PRM F1 Neg": 0.3364485981308411,
+      "eval_PRM NPV": 0.21428571428571427,
+      "eval_PRM Precision": 0.7727272727272727,
+      "eval_PRM Recall": 0.20481927710843373,
+      "eval_PRM Specificty": 0.782608695652174,
+      "eval_loss": 0.9743430614471436,
+      "eval_runtime": 15.319,
+      "eval_samples_per_second": 2.872,
+      "eval_steps_per_second": 0.196,
+      "step": 800
+    },
+    {
+      "epoch": 0.10369938829012526,
+      "grad_norm": 2.9014865333487654,
+      "learning_rate": 1.2499499544511205e-06,
+      "loss": 4.2224,
+      "step": 801
+    },
+    {
+      "epoch": 0.10382885069747871,
+      "grad_norm": 4.725470683614037,
+      "learning_rate": 1.2499463159875382e-06,
+      "loss": 5.0999,
+      "step": 802
+    },
+    {
+      "epoch": 0.10395831310483218,
+      "grad_norm": 4.298496724704148,
+      "learning_rate": 1.249942549866004e-06,
+      "loss": 4.4946,
+      "step": 803
+    },
+    {
+      "epoch": 0.10408777551218565,
+      "grad_norm": 5.453434201047757,
+      "learning_rate": 1.2499386560872869e-06,
+      "loss": 5.1079,
+      "step": 804
+    },
+    {
+      "epoch": 0.10421723791953912,
+      "grad_norm": 7.189806185237564,
+      "learning_rate": 1.2499346346521823e-06,
+      "loss": 6.4219,
+      "step": 805
+    },
+    {
+      "epoch": 0.10421723791953912,
+      "eval_PRM Accuracy": 0.33962264150943394,
+      "eval_PRM F1": 0.33962264150943394,
+      "eval_PRM F1 AUC": 0.499738082765846,
+      "eval_PRM F1 Neg": 0.33962264150943394,
+      "eval_PRM NPV": 0.21686746987951808,
+      "eval_PRM Precision": 0.782608695652174,
+      "eval_PRM Recall": 0.21686746987951808,
+      "eval_PRM Specificty": 0.782608695652174,
+      "eval_loss": 0.9779829382896423,
+      "eval_runtime": 15.2178,
+      "eval_samples_per_second": 2.891,
+      "eval_steps_per_second": 0.197,
+      "step": 805
+    },
+    {
+      "epoch": 0.10434670032689258,
+      "grad_norm": 2.727236830674406,
+      "learning_rate": 1.249930485561512e-06,
+      "loss": 3.8376,
+      "step": 806
+    },
+    {
+      "epoch": 0.10447616273424604,
+      "grad_norm": 3.3423763458003077,
+      "learning_rate": 1.2499262088161227e-06,
+      "loss": 3.5952,
+      "step": 807
+    },
+    {
+      "epoch": 0.1046056251415995,
+      "grad_norm": 2.6734807471168778,
+      "learning_rate": 1.249921804416889e-06,
+      "loss": 3.7695,
+      "step": 808
+    },
+    {
+      "epoch": 0.10473508754895297,
+      "grad_norm": 3.7526485759162194,
+      "learning_rate": 1.24991727236471e-06,
+      "loss": 4.993,
+      "step": 809
+    },
+    {
+      "epoch": 0.10486454995630644,
+      "grad_norm": 1.8145850363033316,
+      "learning_rate": 1.2499126126605117e-06,
+      "loss": 2.9448,
+      "step": 810
+    },
+    {
+      "epoch": 0.10486454995630644,
+      "eval_PRM Accuracy": 0.330188679245283,
+      "eval_PRM F1": 0.3238095238095238,
+      "eval_PRM F1 AUC": 0.4937139863803038,
+      "eval_PRM F1 Neg": 0.3364485981308411,
+      "eval_PRM NPV": 0.21428571428571427,
+      "eval_PRM Precision": 0.7727272727272727,
+      "eval_PRM Recall": 0.20481927710843373,
+      "eval_PRM Specificty": 0.782608695652174,
+      "eval_loss": 0.9822443127632141,
+      "eval_runtime": 14.2835,
+      "eval_samples_per_second": 3.08,
+      "eval_steps_per_second": 0.21,
+      "step": 810
+    },
+    {
+      "epoch": 0.1049940123636599,
+      "grad_norm": 2.451000170082706,
+      "learning_rate": 1.2499078253052457e-06,
+      "loss": 3.1382,
+      "step": 811
+    },
+    {
+      "epoch": 0.10512347477101337,
+      "grad_norm": 2.65749940033498,
+      "learning_rate": 1.2499029102998902e-06,
+      "loss": 3.6758,
+      "step": 812
+    },
+    {
+      "epoch": 0.10525293717836683,
+      "grad_norm": 3.503890377293656,
+      "learning_rate": 1.2498978676454491e-06,
+      "loss": 4.5347,
+      "step": 813
+    },
+    {
+      "epoch": 0.1053823995857203,
+      "grad_norm": 5.026059938552501,
+      "learning_rate": 1.2498926973429524e-06,
+      "loss": 5.0806,
+      "step": 814
+    },
+    {
+      "epoch": 0.10551186199307376,
+      "grad_norm": 5.269285147372982,
+      "learning_rate": 1.2498873993934564e-06,
+      "loss": 6.0195,
+      "step": 815
+    },
+    {
+      "epoch": 0.10551186199307376,
+      "eval_PRM Accuracy": 0.33962264150943394,
+      "eval_PRM F1": 0.33962264150943394,
+      "eval_PRM F1 AUC": 0.499738082765846,
+      "eval_PRM F1 Neg": 0.33962264150943394,
+      "eval_PRM NPV": 0.21686746987951808,
+      "eval_PRM Precision": 0.782608695652174,
+      "eval_PRM Recall": 0.21686746987951808,
+      "eval_PRM Specificty": 0.782608695652174,
+      "eval_loss": 0.9808238744735718,
+      "eval_runtime": 14.1579,
+      "eval_samples_per_second": 3.108,
+      "eval_steps_per_second": 0.212,
+      "step": 815
+    },
+    {
+      "epoch": 0.10564132440042723,
+      "grad_norm": 3.753898840344595,
+      "learning_rate": 1.2498819737980431e-06,
+      "loss": 4.2593,
+      "step": 816
+    },
+    {
+      "epoch": 0.1057707868077807,
+      "grad_norm": 4.208880584724217,
+      "learning_rate": 1.2498764205578206e-06,
+      "loss": 4.4756,
+      "step": 817
+    },
+    {
+      "epoch": 0.10590024921513416,
+      "grad_norm": 3.9561165975523243,
+      "learning_rate": 1.249870739673924e-06,
+      "loss": 4.1987,
+      "step": 818
+    },
+    {
+      "epoch": 0.10602971162248762,
+      "grad_norm": 6.03267265679179,
+      "learning_rate": 1.249864931147513e-06,
+      "loss": 5.2749,
+      "step": 819
+    },
+    {
+      "epoch": 0.10615917402984108,
+      "grad_norm": 1.7585393235758906,
+      "learning_rate": 1.2498589949797742e-06,
+      "loss": 3.4028,
+      "step": 820
+    },
+    {
+      "epoch": 0.10615917402984108,
+      "eval_PRM Accuracy": 0.330188679245283,
+      "eval_PRM F1": 0.3238095238095238,
+      "eval_PRM F1 AUC": 0.4937139863803038,
+      "eval_PRM F1 Neg": 0.3364485981308411,
+      "eval_PRM NPV": 0.21428571428571427,
+      "eval_PRM Precision": 0.7727272727272727,
+      "eval_PRM Recall": 0.20481927710843373,
+      "eval_PRM Specificty": 0.782608695652174,
+      "eval_loss": 0.9831321239471436,
+      "eval_runtime": 14.5652,
+      "eval_samples_per_second": 3.021,
+      "eval_steps_per_second": 0.206,
+      "step": 820
+    },
+    {
+      "epoch": 0.10628863643719455,
+      "grad_norm": 1.9249061148750881,
+      "learning_rate": 1.2498529311719207e-06,
+      "loss": 3.2039,
+      "step": 821
+    },
+    {
+      "epoch": 0.10641809884454802,
+      "grad_norm": 4.411019319171776,
+      "learning_rate": 1.2498467397251908e-06,
+      "loss": 4.5864,
+      "step": 822
+    },
+    {
+      "epoch": 0.10654756125190148,
+      "grad_norm": 3.0029360297239838,
+      "learning_rate": 1.249840420640849e-06,
+      "loss": 3.7687,
+      "step": 823
+    },
+    {
+      "epoch": 0.10667702365925494,
+      "grad_norm": 7.898229005186057,
+      "learning_rate": 1.2498339739201865e-06,
+      "loss": 6.8633,
+      "step": 824
+    },
+    {
+      "epoch": 0.1068064860666084,
+      "grad_norm": 2.7280314836868014,
+      "learning_rate": 1.24982739956452e-06,
+      "loss": 3.6343,
+      "step": 825
+    },
+    {
+      "epoch": 0.1068064860666084,
+      "eval_PRM Accuracy": 0.330188679245283,
+      "eval_PRM F1": 0.3238095238095238,
+      "eval_PRM F1 AUC": 0.4937139863803038,
+      "eval_PRM F1 Neg": 0.3364485981308411,
+      "eval_PRM NPV": 0.21428571428571427,
+      "eval_PRM Precision": 0.7727272727272727,
+      "eval_PRM Recall": 0.20481927710843373,
+      "eval_PRM Specificty": 0.782608695652174,
+      "eval_loss": 0.9847301244735718,
+      "eval_runtime": 14.6771,
+      "eval_samples_per_second": 2.998,
+      "eval_steps_per_second": 0.204,
+      "step": 825
+    },
+    {
+      "epoch": 0.10693594847396187,
+      "grad_norm": 3.461807695266605,
+      "learning_rate": 1.2498206975751924e-06,
+      "loss": 4.4724,
+      "step": 826
+    },
+    {
+      "epoch": 0.10706541088131534,
+      "grad_norm": 4.235149535224669,
+      "learning_rate": 1.2498138679535728e-06,
+      "loss": 4.4229,
+      "step": 827
+    },
+    {
+      "epoch": 0.10719487328866881,
+      "grad_norm": 1.86838832461481,
+      "learning_rate": 1.2498069107010563e-06,
+      "loss": 2.9741,
+      "step": 828
+    },
+    {
+      "epoch": 0.10732433569602227,
+      "grad_norm": 3.1641926300180683,
+      "learning_rate": 1.2497998258190641e-06,
+      "loss": 4.3408,
+      "step": 829
+    },
+    {
+      "epoch": 0.10745379810337573,
+      "grad_norm": 3.9061723174447787,
+      "learning_rate": 1.2497926133090432e-06,
+      "loss": 4.1934,
+      "step": 830
+    },
+    {
+      "epoch": 0.10745379810337573,
+      "eval_PRM Accuracy": 0.330188679245283,
+      "eval_PRM F1": 0.3238095238095238,
+      "eval_PRM F1 AUC": 0.4937139863803038,
+      "eval_PRM F1 Neg": 0.3364485981308411,
+      "eval_PRM NPV": 0.21428571428571427,
+      "eval_PRM Precision": 0.7727272727272727,
+      "eval_PRM Recall": 0.20481927710843373,
+      "eval_PRM Specificty": 0.782608695652174,
+      "eval_loss": 0.9897016882896423,
+      "eval_runtime": 14.44,
+      "eval_samples_per_second": 3.047,
+      "eval_steps_per_second": 0.208,
+      "step": 830
+    },
+    {
+      "epoch": 0.1075832605107292,
+      "grad_norm": 2.628502672134275,
+      "learning_rate": 1.2497852731724673e-06,
+      "loss": 3.6946,
+      "step": 831
+    },
+    {
+      "epoch": 0.10771272291808266,
+      "grad_norm": 4.882329550794728,
+      "learning_rate": 1.2497778054108353e-06,
+      "loss": 5.0615,
+      "step": 832
+    },
+    {
+      "epoch": 0.10784218532543613,
+      "grad_norm": 5.284342736005965,
+      "learning_rate": 1.2497702100256732e-06,
+      "loss": 5.6089,
+      "step": 833
+    },
+    {
+      "epoch": 0.1079716477327896,
+      "grad_norm": 2.4447431854678503,
+      "learning_rate": 1.2497624870185319e-06,
+      "loss": 3.5361,
+      "step": 834
+    },
+    {
+      "epoch": 0.10810111014014305,
+      "grad_norm": 4.223731188362758,
+      "learning_rate": 1.2497546363909894e-06,
+      "loss": 3.7925,
+      "step": 835
+    },
+    {
+      "epoch": 0.10810111014014305,
+      "eval_PRM Accuracy": 0.330188679245283,
+      "eval_PRM F1": 0.3238095238095238,
+      "eval_PRM F1 AUC": 0.4937139863803038,
+      "eval_PRM F1 Neg": 0.3364485981308411,
+      "eval_PRM NPV": 0.21428571428571427,
+      "eval_PRM Precision": 0.7727272727272727,
+      "eval_PRM Recall": 0.20481927710843373,
+      "eval_PRM Specificty": 0.782608695652174,
+      "eval_loss": 0.9895241260528564,
+      "eval_runtime": 14.3143,
+      "eval_samples_per_second": 3.074,
+      "eval_steps_per_second": 0.21,
+      "step": 835
+    },
+    {
+      "epoch": 0.10823057254749652,
+      "grad_norm": 5.529833391656926,
+      "learning_rate": 1.249746658144649e-06,
+      "loss": 6.0088,
+      "step": 836
+    },
+    {
+      "epoch": 0.10836003495484998,
+      "grad_norm": 4.951873947610698,
+      "learning_rate": 1.2497385522811408e-06,
+      "loss": 5.6558,
+      "step": 837
+    },
+    {
+      "epoch": 0.10848949736220345,
+      "grad_norm": 2.973231011524131,
+      "learning_rate": 1.2497303188021205e-06,
+      "loss": 3.115,
+      "step": 838
+    },
+    {
+      "epoch": 0.10861895976955692,
+      "grad_norm": 2.4066000784751544,
+      "learning_rate": 1.2497219577092696e-06,
+      "loss": 3.5962,
+      "step": 839
+    },
+    {
+      "epoch": 0.10874842217691039,
+      "grad_norm": 3.5217087789139656,
+      "learning_rate": 1.2497134690042965e-06,
+      "loss": 4.5015,
+      "step": 840
+    },
+    {
+      "epoch": 0.10874842217691039,
+      "eval_PRM Accuracy": 0.330188679245283,
+      "eval_PRM F1": 0.3238095238095238,
+      "eval_PRM F1 AUC": 0.4937139863803038,
+      "eval_PRM F1 Neg": 0.3364485981308411,
+      "eval_PRM NPV": 0.21428571428571427,
+      "eval_PRM Precision": 0.7727272727272727,
+      "eval_PRM Recall": 0.20481927710843373,
+      "eval_PRM Specificty": 0.782608695652174,
+      "eval_loss": 0.9920987486839294,
+      "eval_runtime": 13.7161,
+      "eval_samples_per_second": 3.208,
+      "eval_steps_per_second": 0.219,
+      "step": 840
+    },
+    {
+      "epoch": 0.10887788458426384,
+      "grad_norm": 2.852173896730775,
+      "learning_rate": 1.249704852688935e-06,
+      "loss": 3.4556,
+      "step": 841
+    },
+    {
+      "epoch": 0.10900734699161731,
+      "grad_norm": 3.4229498964932374,
+      "learning_rate": 1.249696108764945e-06,
+      "loss": 3.895,
+      "step": 842
+    },
+    {
+      "epoch": 0.10913680939897077,
+      "grad_norm": 4.3670371394492795,
+      "learning_rate": 1.249687237234113e-06,
+      "loss": 3.9854,
+      "step": 843
+    },
+    {
+      "epoch": 0.10926627180632424,
+      "grad_norm": 6.42217369367119,
+      "learning_rate": 1.2496782380982509e-06,
+      "loss": 6.5903,
+      "step": 844
+    },
+    {
+      "epoch": 0.10939573421367771,
+      "grad_norm": 4.845601383962569,
+      "learning_rate": 1.249669111359197e-06,
+      "loss": 4.6348,
+      "step": 845
+    },
+    {
+      "epoch": 0.10939573421367771,
+      "eval_PRM Accuracy": 0.33962264150943394,
+      "eval_PRM F1": 0.3269230769230769,
+      "eval_PRM F1 AUC": 0.5154531168150864,
+      "eval_PRM F1 Neg": 0.35185185185185186,
+      "eval_PRM NPV": 0.2235294117647059,
+      "eval_PRM Precision": 0.8095238095238095,
+      "eval_PRM Recall": 0.20481927710843373,
+      "eval_PRM Specificty": 0.8260869565217391,
+      "eval_loss": 0.9969815611839294,
+      "eval_runtime": 14.0664,
+      "eval_samples_per_second": 3.128,
+      "eval_steps_per_second": 0.213,
+      "step": 845
+    },
+    {
+      "epoch": 0.10952519662103116,
+      "grad_norm": 3.9497601962992634,
+      "learning_rate": 1.2496598570188156e-06,
+      "loss": 4.9312,
+      "step": 846
+    },
+    {
+      "epoch": 0.10965465902838463,
+      "grad_norm": 3.731438089599192,
+      "learning_rate": 1.249650475078997e-06,
+      "loss": 3.9316,
+      "step": 847
+    },
+    {
+      "epoch": 0.1097841214357381,
+      "grad_norm": 1.7497549960975611,
+      "learning_rate": 1.249640965541658e-06,
+      "loss": 2.9502,
+      "step": 848
+    },
+    {
+      "epoch": 0.10991358384309156,
+      "grad_norm": 4.325633256350921,
+      "learning_rate": 1.2496313284087408e-06,
+      "loss": 4.395,
+      "step": 849
+    },
+    {
+      "epoch": 0.11004304625044503,
+      "grad_norm": 6.382616472343234,
+      "learning_rate": 1.249621563682214e-06,
+      "loss": 5.3101,
+      "step": 850
+    },
+    {
+      "epoch": 0.11004304625044503,
+      "eval_PRM Accuracy": 0.330188679245283,
+      "eval_PRM F1": 0.3238095238095238,
+      "eval_PRM F1 AUC": 0.4937139863803038,
+      "eval_PRM F1 Neg": 0.3364485981308411,
+      "eval_PRM NPV": 0.21428571428571427,
+      "eval_PRM Precision": 0.7727272727272727,
+      "eval_PRM Recall": 0.20481927710843373,
+      "eval_PRM Specificty": 0.782608695652174,
+      "eval_loss": 0.9966264367103577,
+      "eval_runtime": 15.1859,
+      "eval_samples_per_second": 2.897,
+      "eval_steps_per_second": 0.198,
+      "step": 850
+    },
+    {
+      "epoch": 0.1101725086577985,
+      "grad_norm": 2.8336601664198966,
+      "learning_rate": 1.2496116713640725e-06,
+      "loss": 3.7319,
+      "step": 851
+    },
+    {
+      "epoch": 0.11030197106515195,
+      "grad_norm": 1.8389531897806024,
+      "learning_rate": 1.2496016514563367e-06,
+      "loss": 3.0261,
+      "step": 852
+    },
+    {
+      "epoch": 0.11043143347250542,
+      "grad_norm": 3.6222250371246347,
+      "learning_rate": 1.2495915039610534e-06,
+      "loss": 4.6567,
+      "step": 853
+    },
+    {
+      "epoch": 0.11056089587985889,
+      "grad_norm": 5.188574916796592,
+      "learning_rate": 1.2495812288802959e-06,
+      "loss": 5.4341,
+      "step": 854
+    },
+    {
+      "epoch": 0.11069035828721235,
+      "grad_norm": 2.309330039567924,
+      "learning_rate": 1.2495708262161623e-06,
+      "loss": 3.1453,
+      "step": 855
+    },
+    {
+      "epoch": 0.11069035828721235,
+      "eval_PRM Accuracy": 0.330188679245283,
+      "eval_PRM F1": 0.3238095238095238,
+      "eval_PRM F1 AUC": 0.4937139863803038,
+      "eval_PRM F1 Neg": 0.3364485981308411,
+      "eval_PRM NPV": 0.21428571428571427,
+      "eval_PRM Precision": 0.7727272727272727,
+      "eval_PRM Recall": 0.20481927710843373,
+      "eval_PRM Specificty": 0.782608695652174,
+      "eval_loss": 1.0021306276321411,
+      "eval_runtime": 15.2091,
+      "eval_samples_per_second": 2.893,
+      "eval_steps_per_second": 0.197,
+      "step": 855
+    },
+    {
+      "epoch": 0.11081982069456582,
+      "grad_norm": 6.078198113679103,
+      "learning_rate": 1.249560295970778e-06,
+      "loss": 5.3374,
+      "step": 856
+    },
+    {
+      "epoch": 0.11094928310191927,
+      "grad_norm": 5.553502738982046,
+      "learning_rate": 1.2495496381462942e-06,
+      "loss": 4.9473,
+      "step": 857
+    },
+    {
+      "epoch": 0.11107874550927274,
+      "grad_norm": 3.997020265294788,
+      "learning_rate": 1.2495388527448875e-06,
+      "loss": 4.4836,
+      "step": 858
+    },
+    {
+      "epoch": 0.11120820791662621,
+      "grad_norm": 2.179289610827541,
+      "learning_rate": 1.2495279397687615e-06,
+      "loss": 3.5059,
+      "step": 859
+    },
+    {
+      "epoch": 0.11133767032397968,
+      "grad_norm": 5.901634015894297,
+      "learning_rate": 1.2495168992201454e-06,
+      "loss": 5.603,
+      "step": 860
+    },
+    {
+      "epoch": 0.11133767032397968,
+      "eval_PRM Accuracy": 0.33962264150943394,
+      "eval_PRM F1": 0.3269230769230769,
+      "eval_PRM F1 AUC": 0.5154531168150864,
+      "eval_PRM F1 Neg": 0.35185185185185186,
+      "eval_PRM NPV": 0.2235294117647059,
+      "eval_PRM Precision": 0.8095238095238095,
+      "eval_PRM Recall": 0.20481927710843373,
+      "eval_PRM Specificty": 0.8260869565217391,
+      "eval_loss": 0.9991121888160706,
+      "eval_runtime": 15.0122,
+      "eval_samples_per_second": 2.931,
+      "eval_steps_per_second": 0.2,
+      "step": 860
+    },
+    {
+      "epoch": 0.11146713273133314,
+      "grad_norm": 1.5009745040246112,
+      "learning_rate": 1.249505731101294e-06,
+      "loss": 3.178,
+      "step": 861
+    },
+    {
+      "epoch": 0.11159659513868661,
+      "grad_norm": 7.412442578727905,
+      "learning_rate": 1.249494435414489e-06,
+      "loss": 6.4973,
+      "step": 862
+    },
+    {
+      "epoch": 0.11172605754604006,
+      "grad_norm": 8.60535577512405,
+      "learning_rate": 1.2494830121620375e-06,
+      "loss": 7.4678,
+      "step": 863
+    },
+    {
+      "epoch": 0.11185551995339353,
+      "grad_norm": 3.926298901427391,
+      "learning_rate": 1.249471461346273e-06,
+      "loss": 4.481,
+      "step": 864
+    },
+    {
+      "epoch": 0.111984982360747,
+      "grad_norm": 3.599262243600864,
+      "learning_rate": 1.249459782969555e-06,
+      "loss": 3.9001,
+      "step": 865
+    },
+    {
+      "epoch": 0.111984982360747,
+      "eval_PRM Accuracy": 0.330188679245283,
+      "eval_PRM F1": 0.3106796116504854,
+      "eval_PRM F1 AUC": 0.5094290204295442,
+      "eval_PRM F1 Neg": 0.3486238532110092,
+      "eval_PRM NPV": 0.22093023255813954,
+      "eval_PRM Precision": 0.8,
+      "eval_PRM Recall": 0.1927710843373494,
+      "eval_PRM Specificty": 0.8260869565217391,
+      "eval_loss": 1.0030184984207153,
+      "eval_runtime": 14.7638,
+      "eval_samples_per_second": 2.98,
+      "eval_steps_per_second": 0.203,
+      "step": 865
+    },
+    {
+      "epoch": 0.11211444476810047,
+      "grad_norm": 5.375324572377989,
+      "learning_rate": 1.2494479770342694e-06,
+      "loss": 4.5283,
+      "step": 866
+    },
+    {
+      "epoch": 0.11224390717545393,
+      "grad_norm": 3.5982816431956754,
+      "learning_rate": 1.2494360435428275e-06,
+      "loss": 3.8792,
+      "step": 867
+    },
+    {
+      "epoch": 0.11237336958280739,
+      "grad_norm": 5.852933486181772,
+      "learning_rate": 1.2494239824976668e-06,
+      "loss": 4.9756,
+      "step": 868
+    },
+    {
+      "epoch": 0.11250283199016085,
+      "grad_norm": 3.426276644536621,
+      "learning_rate": 1.249411793901251e-06,
+      "loss": 3.6125,
+      "step": 869
+    },
+    {
+      "epoch": 0.11263229439751432,
+      "grad_norm": 2.4398585844201466,
+      "learning_rate": 1.2493994777560703e-06,
+      "loss": 3.0264,
+      "step": 870
+    },
+    {
+      "epoch": 0.11263229439751432,
+      "eval_PRM Accuracy": 0.3490566037735849,
+      "eval_PRM F1": 0.3300970873786408,
+      "eval_PRM F1 AUC": 0.537192247249869,
+      "eval_PRM F1 Neg": 0.3669724770642202,
+      "eval_PRM NPV": 0.23255813953488372,
+      "eval_PRM Precision": 0.85,
+      "eval_PRM Recall": 0.20481927710843373,
+      "eval_PRM Specificty": 0.8695652173913043,
+      "eval_loss": 1.0076349973678589,
+      "eval_runtime": 13.7873,
+      "eval_samples_per_second": 3.191,
+      "eval_steps_per_second": 0.218,
+      "step": 870
+    },
+    {
+      "epoch": 0.11276175680486779,
+      "grad_norm": 3.5402306806103865,
+      "learning_rate": 1.2493870340646403e-06,
+      "loss": 3.7861,
+      "step": 871
+    },
+    {
+      "epoch": 0.11289121921222126,
+      "grad_norm": 2.712796681020074,
+      "learning_rate": 1.2493744628295027e-06,
+      "loss": 3.2959,
+      "step": 872
+    },
+    {
+      "epoch": 0.11302068161957472,
+      "grad_norm": 5.453100818840841,
+      "learning_rate": 1.2493617640532257e-06,
+      "loss": 4.9917,
+      "step": 873
+    },
+    {
+      "epoch": 0.11315014402692818,
+      "grad_norm": 3.8104240445796487,
+      "learning_rate": 1.2493489377384029e-06,
+      "loss": 4.9126,
+      "step": 874
+    },
+    {
+      "epoch": 0.11327960643428164,
+      "grad_norm": 4.717319395191185,
+      "learning_rate": 1.249335983887655e-06,
+      "loss": 4.3391,
+      "step": 875
+    },
+    {
+      "epoch": 0.11327960643428164,
+      "eval_PRM Accuracy": 0.330188679245283,
+      "eval_PRM F1": 0.3106796116504854,
+      "eval_PRM F1 AUC": 0.5094290204295442,
+      "eval_PRM F1 Neg": 0.3486238532110092,
+      "eval_PRM NPV": 0.22093023255813954,
+      "eval_PRM Precision": 0.8,
+      "eval_PRM Recall": 0.1927710843373494,
+      "eval_PRM Specificty": 0.8260869565217391,
+      "eval_loss": 1.0116299390792847,
+      "eval_runtime": 13.9696,
+      "eval_samples_per_second": 3.15,
+      "eval_steps_per_second": 0.215,
+      "step": 875
+    },
+    {
+      "epoch": 0.11340906884163511,
+      "grad_norm": 4.493686327669407,
+      "learning_rate": 1.2493229025036275e-06,
+      "loss": 4.5591,
+      "step": 876
+    },
+    {
+      "epoch": 0.11353853124898858,
+      "grad_norm": 3.0504005702181916,
+      "learning_rate": 1.2493096935889926e-06,
+      "loss": 4.3931,
+      "step": 877
+    },
+    {
+      "epoch": 0.11366799365634205,
+      "grad_norm": 2.5019561657338936,
+      "learning_rate": 1.2492963571464487e-06,
+      "loss": 3.7275,
+      "step": 878
+    },
+    {
+      "epoch": 0.1137974560636955,
+      "grad_norm": 4.110635049966043,
+      "learning_rate": 1.2492828931787199e-06,
+      "loss": 4.6113,
+      "step": 879
+    },
+    {
+      "epoch": 0.11392691847104897,
+      "grad_norm": 3.9883417213824335,
+      "learning_rate": 1.2492693016885564e-06,
+      "loss": 4.3462,
+      "step": 880
+    },
+    {
+      "epoch": 0.11392691847104897,
+      "eval_PRM Accuracy": 0.330188679245283,
+      "eval_PRM F1": 0.3106796116504854,
+      "eval_PRM F1 AUC": 0.5094290204295442,
+      "eval_PRM F1 Neg": 0.3486238532110092,
+      "eval_PRM NPV": 0.22093023255813954,
+      "eval_PRM Precision": 0.8,
+      "eval_PRM Recall": 0.1927710843373494,
+      "eval_PRM Specificty": 0.8260869565217391,
+      "eval_loss": 1.0116299390792847,
+      "eval_runtime": 14.106,
+      "eval_samples_per_second": 3.119,
+      "eval_steps_per_second": 0.213,
+      "step": 880
+    },
+    {
+      "epoch": 0.11405638087840243,
+      "grad_norm": 5.739692247577143,
+      "learning_rate": 1.249255582678735e-06,
+      "loss": 4.5376,
+      "step": 881
+    },
+    {
+      "epoch": 0.1141858432857559,
+      "grad_norm": 3.062193103997161,
+      "learning_rate": 1.2492417361520575e-06,
+      "loss": 4.3318,
+      "step": 882
+    },
+    {
+      "epoch": 0.11431530569310937,
+      "grad_norm": 3.906370106428911,
+      "learning_rate": 1.2492277621113528e-06,
+      "loss": 3.2939,
+      "step": 883
+    },
+    {
+      "epoch": 0.11444476810046283,
+      "grad_norm": 7.935009829442984,
+      "learning_rate": 1.249213660559475e-06,
+      "loss": 6.3003,
+      "step": 884
+    },
+    {
+      "epoch": 0.11457423050781629,
+      "grad_norm": 4.802993612758119,
+      "learning_rate": 1.249199431499305e-06,
+      "loss": 4.7627,
+      "step": 885
+    },
+    {
+      "epoch": 0.11457423050781629,
+      "eval_PRM Accuracy": 0.33962264150943394,
+      "eval_PRM F1": 0.3,
+      "eval_PRM F1 AUC": 0.5468831849135674,
+      "eval_PRM F1 Neg": 0.375,
+      "eval_PRM NPV": 0.23595505617977527,
+      "eval_PRM Precision": 0.8823529411764706,
+      "eval_PRM Recall": 0.18072289156626506,
+      "eval_PRM Specificty": 0.9130434782608695,
+      "eval_loss": 1.0138493776321411,
+      "eval_runtime": 15.1959,
+      "eval_samples_per_second": 2.896,
+      "eval_steps_per_second": 0.197,
+      "step": 885
+    },
+    {
+      "epoch": 0.11470369291516976,
+      "grad_norm": 3.4235349694067083,
+      "learning_rate": 1.249185074933749e-06,
+      "loss": 3.2212,
+      "step": 886
+    },
+    {
+      "epoch": 0.11483315532252322,
+      "grad_norm": 3.1475045189217314,
+      "learning_rate": 1.2491705908657395e-06,
+      "loss": 3.5601,
+      "step": 887
+    },
+    {
+      "epoch": 0.11496261772987669,
+      "grad_norm": 3.3576060084887924,
+      "learning_rate": 1.2491559792982358e-06,
+      "loss": 3.9238,
+      "step": 888
+    },
+    {
+      "epoch": 0.11509208013723016,
+      "grad_norm": 3.7841363012296423,
+      "learning_rate": 1.2491412402342223e-06,
+      "loss": 3.9614,
+      "step": 889
+    },
+    {
+      "epoch": 0.11522154254458361,
+      "grad_norm": 3.601368336096151,
+      "learning_rate": 1.2491263736767095e-06,
+      "loss": 4.1152,
+      "step": 890
+    },
+    {
+      "epoch": 0.11522154254458361,
+      "eval_PRM Accuracy": 0.33962264150943394,
+      "eval_PRM F1": 0.3137254901960784,
+      "eval_PRM F1 AUC": 0.5311681508643268,
+      "eval_PRM F1 Neg": 0.36363636363636365,
+      "eval_PRM NPV": 0.22988505747126436,
+      "eval_PRM Precision": 0.8421052631578947,
+      "eval_PRM Recall": 0.1927710843373494,
+      "eval_PRM Specificty": 0.8695652173913043,
+      "eval_loss": 1.018998622894287,
+      "eval_runtime": 14.9255,
+      "eval_samples_per_second": 2.948,
+      "eval_steps_per_second": 0.201,
+      "step": 890
+    },
+    {
+      "epoch": 0.11535100495193708,
+      "grad_norm": 3.4178557242458534,
+      "learning_rate": 1.2491113796287347e-06,
+      "loss": 4.3242,
+      "step": 891
+    },
+    {
+      "epoch": 0.11548046735929055,
+      "grad_norm": 2.974190535857782,
+      "learning_rate": 1.2490962580933602e-06,
+      "loss": 3.377,
+      "step": 892
+    },
+    {
+      "epoch": 0.11560992976664401,
+      "grad_norm": 4.724263966045136,
+      "learning_rate": 1.2490810090736752e-06,
+      "loss": 4.3235,
+      "step": 893
+    },
+    {
+      "epoch": 0.11573939217399748,
+      "grad_norm": 2.242875054399141,
+      "learning_rate": 1.2490656325727945e-06,
+      "loss": 3.1953,
+      "step": 894
+    },
+    {
+      "epoch": 0.11586885458135095,
+      "grad_norm": 2.529182875396455,
+      "learning_rate": 1.2490501285938592e-06,
+      "loss": 3.3425,
+      "step": 895
+    },
+    {
+      "epoch": 0.11586885458135095,
+      "eval_PRM Accuracy": 0.33962264150943394,
+      "eval_PRM F1": 0.3,
+      "eval_PRM F1 AUC": 0.5468831849135674,
+      "eval_PRM F1 Neg": 0.375,
+      "eval_PRM NPV": 0.23595505617977527,
+      "eval_PRM Precision": 0.8823529411764706,
+      "eval_PRM Recall": 0.18072289156626506,
+      "eval_PRM Specificty": 0.9130434782608695,
+      "eval_loss": 1.0213068723678589,
+      "eval_runtime": 14.8251,
+      "eval_samples_per_second": 2.968,
+      "eval_steps_per_second": 0.202,
+      "step": 895
+    },
+    {
+      "epoch": 0.1159983169887044,
+      "grad_norm": 2.2319499551112747,
+      "learning_rate": 1.2490344971400363e-06,
+      "loss": 3.2275,
+      "step": 896
+    },
+    {
+      "epoch": 0.11612777939605787,
+      "grad_norm": 2.966351334593291,
+      "learning_rate": 1.2490187382145185e-06,
+      "loss": 3.373,
+      "step": 897
+    },
+    {
+      "epoch": 0.11625724180341133,
+      "grad_norm": 5.077339885312606,
+      "learning_rate": 1.2490028518205256e-06,
+      "loss": 4.2671,
+      "step": 898
+    },
+    {
+      "epoch": 0.1163867042107648,
+      "grad_norm": 1.444467093764416,
+      "learning_rate": 1.2489868379613017e-06,
+      "loss": 2.5872,
+      "step": 899
+    },
+    {
+      "epoch": 0.11651616661811827,
+      "grad_norm": 5.850737496333685,
+      "learning_rate": 1.248970696640119e-06,
+      "loss": 4.689,
+      "step": 900
+    },
+    {
+      "epoch": 0.11651616661811827,
+      "eval_PRM Accuracy": 0.330188679245283,
+      "eval_PRM F1": 0.2828282828282828,
+      "eval_PRM F1 AUC": 0.5408590885280251,
+      "eval_PRM F1 Neg": 0.37168141592920356,
+      "eval_PRM NPV": 0.23333333333333334,
+      "eval_PRM Precision": 0.875,
+      "eval_PRM Recall": 0.1686746987951807,
+      "eval_PRM Specificty": 0.9130434782608695,
+      "eval_loss": 1.0205966234207153,
+      "eval_runtime": 15.7606,
+      "eval_samples_per_second": 2.792,
+      "eval_steps_per_second": 0.19,
+      "step": 900
+    },
+    {
+      "epoch": 0.11664562902547172,
+      "grad_norm": 6.6209305728707974,
+      "learning_rate": 1.248954427860274e-06,
+      "loss": 5.437,
+      "step": 901
+    },
+    {
+      "epoch": 0.11677509143282519,
+      "grad_norm": 5.155904054749944,
+      "learning_rate": 1.2489380316250898e-06,
+      "loss": 5.8501,
+      "step": 902
+    },
+    {
+      "epoch": 0.11690455384017866,
+      "grad_norm": 1.8097533710532838,
+      "learning_rate": 1.2489215079379164e-06,
+      "loss": 2.7844,
+      "step": 903
+    },
+    {
+      "epoch": 0.11703401624753212,
+      "grad_norm": 5.140810409033703,
+      "learning_rate": 1.2489048568021284e-06,
+      "loss": 5.583,
+      "step": 904
+    },
+    {
+      "epoch": 0.11716347865488559,
+      "grad_norm": 2.708403766038811,
+      "learning_rate": 1.2488880782211277e-06,
+      "loss": 3.1641,
+      "step": 905
+    },
+    {
+      "epoch": 0.11716347865488559,
+      "eval_PRM Accuracy": 0.330188679245283,
+      "eval_PRM F1": 0.2828282828282828,
+      "eval_PRM F1 AUC": 0.5408590885280251,
+      "eval_PRM F1 Neg": 0.37168141592920356,
+      "eval_PRM NPV": 0.23333333333333334,
+      "eval_PRM Precision": 0.875,
+      "eval_PRM Recall": 0.1686746987951807,
+      "eval_PRM Specificty": 0.9130434782608695,
+      "eval_loss": 1.0230823755264282,
+      "eval_runtime": 14.2319,
+      "eval_samples_per_second": 3.092,
+      "eval_steps_per_second": 0.211,
+      "step": 905
+    },
+    {
+      "epoch": 0.11729294106223906,
+      "grad_norm": 2.9003673908208167,
+      "learning_rate": 1.2488711721983413e-06,
+      "loss": 4.0337,
+      "step": 906
+    },
+    {
+      "epoch": 0.11742240346959251,
+      "grad_norm": 4.41386619831515,
+      "learning_rate": 1.2488541387372226e-06,
+      "loss": 4.2759,
+      "step": 907
+    },
+    {
+      "epoch": 0.11755186587694598,
+      "grad_norm": 5.497296513095215,
+      "learning_rate": 1.2488369778412513e-06,
+      "loss": 5.5381,
+      "step": 908
+    },
+    {
+      "epoch": 0.11768132828429945,
+      "grad_norm": 4.057718708605249,
+      "learning_rate": 1.2488196895139323e-06,
+      "loss": 4.2947,
+      "step": 909
+    },
+    {
+      "epoch": 0.11781079069165291,
+      "grad_norm": 4.075251374539448,
+      "learning_rate": 1.2488022737587978e-06,
+      "loss": 4.1887,
+      "step": 910
+    },
+    {
+      "epoch": 0.11781079069165291,
+      "eval_PRM Accuracy": 0.330188679245283,
+      "eval_PRM F1": 0.26804123711340205,
+      "eval_PRM F1 AUC": 0.5565741225772656,
+      "eval_PRM F1 Neg": 0.3826086956521739,
+      "eval_PRM NPV": 0.2391304347826087,
+      "eval_PRM Precision": 0.9285714285714286,
+      "eval_PRM Recall": 0.1566265060240964,
+      "eval_PRM Specificty": 0.9565217391304348,
+      "eval_loss": 1.0245915651321411,
+      "eval_runtime": 13.6452,
+      "eval_samples_per_second": 3.225,
+      "eval_steps_per_second": 0.22,
+      "step": 910
+    },
+    {
+      "epoch": 0.11794025309900638,
+      "grad_norm": 4.546605874265857,
+      "learning_rate": 1.248784730579405e-06,
+      "loss": 4.9141,
+      "step": 911
+    },
+    {
+      "epoch": 0.11806971550635983,
+      "grad_norm": 3.5002972929514824,
+      "learning_rate": 1.2487670599793372e-06,
+      "loss": 3.5303,
+      "step": 912
+    },
+    {
+      "epoch": 0.1181991779137133,
+      "grad_norm": 4.543518740402077,
+      "learning_rate": 1.2487492619622045e-06,
+      "loss": 4.6104,
+      "step": 913
+    },
+    {
+      "epoch": 0.11832864032106677,
+      "grad_norm": 1.5433945063954684,
+      "learning_rate": 1.2487313365316419e-06,
+      "loss": 2.6094,
+      "step": 914
+    },
+    {
+      "epoch": 0.11845810272842024,
+      "grad_norm": 1.5873363221733159,
+      "learning_rate": 1.2487132836913117e-06,
+      "loss": 2.8586,
+      "step": 915
+    },
+    {
+      "epoch": 0.11845810272842024,
+      "eval_PRM Accuracy": 0.32075471698113206,
+      "eval_PRM F1": 0.25,
+      "eval_PRM F1 AUC": 0.5505500261917234,
+      "eval_PRM F1 Neg": 0.3793103448275862,
+      "eval_PRM NPV": 0.23655913978494625,
+      "eval_PRM Precision": 0.9230769230769231,
+      "eval_PRM Recall": 0.14457831325301204,
+      "eval_PRM Specificty": 0.9565217391304348,
+      "eval_loss": 1.0308948755264282,
+      "eval_runtime": 13.85,
+      "eval_samples_per_second": 3.177,
+      "eval_steps_per_second": 0.217,
+      "step": 915
+    },
+    {
+      "epoch": 0.1185875651357737,
+      "grad_norm": 4.7418148262939885,
+      "learning_rate": 1.248695103444901e-06,
+      "loss": 4.1821,
+      "step": 916
+    },
+    {
+      "epoch": 0.11871702754312717,
+      "grad_norm": 3.693319011165287,
+      "learning_rate": 1.2486767957961237e-06,
+      "loss": 3.6196,
+      "step": 917
+    },
+    {
+      "epoch": 0.11884648995048062,
+      "grad_norm": 2.649385339123106,
+      "learning_rate": 1.2486583607487197e-06,
+      "loss": 3.5439,
+      "step": 918
+    },
+    {
+      "epoch": 0.11897595235783409,
+      "grad_norm": 5.939044051126767,
+      "learning_rate": 1.2486397983064543e-06,
+      "loss": 5.5059,
+      "step": 919
+    },
+    {
+      "epoch": 0.11910541476518756,
+      "grad_norm": 2.8394658154401795,
+      "learning_rate": 1.2486211084731197e-06,
+      "loss": 3.8086,
+      "step": 920
+    },
+    {
+      "epoch": 0.11910541476518756,
+      "eval_PRM Accuracy": 0.3490566037735849,
+      "eval_PRM F1": 0.30303030303030304,
+      "eval_PRM F1 AUC": 0.56862231534835,
+      "eval_PRM F1 Neg": 0.3893805309734513,
+      "eval_PRM NPV": 0.24444444444444444,
+      "eval_PRM Precision": 0.9375,
+      "eval_PRM Recall": 0.18072289156626506,
+      "eval_PRM Specificty": 0.9565217391304348,
+      "eval_loss": 1.033647060394287,
+      "eval_runtime": 15.5497,
+      "eval_samples_per_second": 2.83,
+      "eval_steps_per_second": 0.193,
+      "step": 920
+    },
+    {
+      "epoch": 0.11923487717254103,
+      "grad_norm": 2.669443359973629,
+      "learning_rate": 1.2486022912525333e-06,
+      "loss": 3.0354,
+      "step": 921
+    },
+    {
+      "epoch": 0.1193643395798945,
+      "grad_norm": 2.6859606289086204,
+      "learning_rate": 1.2485833466485392e-06,
+      "loss": 3.7344,
+      "step": 922
+    },
+    {
+      "epoch": 0.11949380198724795,
+      "grad_norm": 2.3894788312649036,
+      "learning_rate": 1.248564274665007e-06,
+      "loss": 3.2959,
+      "step": 923
+    },
+    {
+      "epoch": 0.11962326439460141,
+      "grad_norm": 2.079663998889526,
+      "learning_rate": 1.2485450753058326e-06,
+      "loss": 3.0674,
+      "step": 924
+    },
+    {
+      "epoch": 0.11975272680195488,
+      "grad_norm": 6.620720008250604,
+      "learning_rate": 1.248525748574938e-06,
+      "loss": 5.6985,
+      "step": 925
+    },
+    {
+      "epoch": 0.11975272680195488,
+      "eval_PRM Accuracy": 0.330188679245283,
+      "eval_PRM F1": 0.26804123711340205,
+      "eval_PRM F1 AUC": 0.5565741225772656,
+      "eval_PRM F1 Neg": 0.3826086956521739,
+      "eval_PRM NPV": 0.2391304347826087,
+      "eval_PRM Precision": 0.9285714285714286,
+      "eval_PRM Recall": 0.1566265060240964,
+      "eval_PRM Specificty": 0.9565217391304348,
+      "eval_loss": 1.0389736890792847,
+      "eval_runtime": 15.1577,
+      "eval_samples_per_second": 2.903,
+      "eval_steps_per_second": 0.198,
+      "step": 925
+    },
+    {
+      "epoch": 0.11988218920930835,
+      "grad_norm": 4.121052505537082,
+      "learning_rate": 1.2485062944762707e-06,
+      "loss": 3.767,
+      "step": 926
+    },
+    {
+      "epoch": 0.12001165161666182,
+      "grad_norm": 3.949851494932497,
+      "learning_rate": 1.2484867130138051e-06,
+      "loss": 4.7007,
+      "step": 927
+    },
+    {
+      "epoch": 0.12014111402401528,
+      "grad_norm": 4.8733058805450185,
+      "learning_rate": 1.2484670041915408e-06,
+      "loss": 4.9292,
+      "step": 928
+    },
+    {
+      "epoch": 0.12027057643136874,
+      "grad_norm": 2.755096418475927,
+      "learning_rate": 1.248447168013504e-06,
+      "loss": 3.5366,
+      "step": 929
+    },
+    {
+      "epoch": 0.1204000388387222,
+      "grad_norm": 2.534427885267366,
+      "learning_rate": 1.2484272044837462e-06,
+      "loss": 3.7122,
+      "step": 930
+    },
+    {
+      "epoch": 0.1204000388387222,
+      "eval_PRM Accuracy": 0.330188679245283,
+      "eval_PRM F1": 0.25263157894736843,
+      "eval_PRM F1 AUC": 0.572289156626506,
+      "eval_PRM F1 Neg": 0.39316239316239315,
+      "eval_PRM NPV": 0.24468085106382978,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.14457831325301204,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.0392400026321411,
+      "eval_runtime": 15.12,
+      "eval_samples_per_second": 2.91,
+      "eval_steps_per_second": 0.198,
+      "step": 930
+    },
+    {
+      "epoch": 0.12052950124607567,
+      "grad_norm": 3.350680168750641,
+      "learning_rate": 1.2484071136063456e-06,
+      "loss": 4.415,
+      "step": 931
+    },
+    {
+      "epoch": 0.12065896365342914,
+      "grad_norm": 5.518309124632107,
+      "learning_rate": 1.2483868953854061e-06,
+      "loss": 4.4155,
+      "step": 932
+    },
+    {
+      "epoch": 0.1207884260607826,
+      "grad_norm": 3.6445339199367734,
+      "learning_rate": 1.248366549825058e-06,
+      "loss": 3.9683,
+      "step": 933
+    },
+    {
+      "epoch": 0.12091788846813606,
+      "grad_norm": 5.318081452162914,
+      "learning_rate": 1.248346076929457e-06,
+      "loss": 4.0686,
+      "step": 934
+    },
+    {
+      "epoch": 0.12104735087548953,
+      "grad_norm": 7.0084529244510385,
+      "learning_rate": 1.2483254767027848e-06,
+      "loss": 5.2681,
+      "step": 935
+    },
+    {
+      "epoch": 0.12104735087548953,
+      "eval_PRM Accuracy": 0.32075471698113206,
+      "eval_PRM F1": 0.25,
+      "eval_PRM F1 AUC": 0.5505500261917234,
+      "eval_PRM F1 Neg": 0.3793103448275862,
+      "eval_PRM NPV": 0.23655913978494625,
+      "eval_PRM Precision": 0.9230769230769231,
+      "eval_PRM Recall": 0.14457831325301204,
+      "eval_PRM Specificty": 0.9565217391304348,
+      "eval_loss": 1.0420809984207153,
+      "eval_runtime": 14.9876,
+      "eval_samples_per_second": 2.936,
+      "eval_steps_per_second": 0.2,
+      "step": 935
+    },
+    {
+      "epoch": 0.121176813282843,
+      "grad_norm": 8.067725993235062,
+      "learning_rate": 1.2483047491492501e-06,
+      "loss": 5.5825,
+      "step": 936
+    },
+    {
+      "epoch": 0.12130627569019646,
+      "grad_norm": 4.398388312771275,
+      "learning_rate": 1.2482838942730862e-06,
+      "loss": 4.6582,
+      "step": 937
+    },
+    {
+      "epoch": 0.12143573809754993,
+      "grad_norm": 8.748390665190263,
+      "learning_rate": 1.248262912078554e-06,
+      "loss": 5.6055,
+      "step": 938
+    },
+    {
+      "epoch": 0.1215652005049034,
+      "grad_norm": 2.396884372702812,
+      "learning_rate": 1.2482418025699386e-06,
+      "loss": 3.0403,
+      "step": 939
+    },
+    {
+      "epoch": 0.12169466291225685,
+      "grad_norm": 1.7716886889052004,
+      "learning_rate": 1.2482205657515528e-06,
+      "loss": 2.9629,
+      "step": 940
+    },
+    {
+      "epoch": 0.12169466291225685,
+      "eval_PRM Accuracy": 0.32075471698113206,
+      "eval_PRM F1": 0.25,
+      "eval_PRM F1 AUC": 0.5505500261917234,
+      "eval_PRM F1 Neg": 0.3793103448275862,
+      "eval_PRM NPV": 0.23655913978494625,
+      "eval_PRM Precision": 0.9230769230769231,
+      "eval_PRM Recall": 0.14457831325301204,
+      "eval_PRM Specificty": 0.9565217391304348,
+      "eval_loss": 1.0450993776321411,
+      "eval_runtime": 14.9449,
+      "eval_samples_per_second": 2.944,
+      "eval_steps_per_second": 0.201,
+      "step": 940
+    },
+    {
+      "epoch": 0.12182412531961032,
+      "grad_norm": 4.808822983703,
+      "learning_rate": 1.2481992016277341e-06,
+      "loss": 4.8062,
+      "step": 941
+    },
+    {
+      "epoch": 0.12195358772696378,
+      "grad_norm": 5.5255127021190775,
+      "learning_rate": 1.248177710202847e-06,
+      "loss": 4.7842,
+      "step": 942
+    },
+    {
+      "epoch": 0.12208305013431725,
+      "grad_norm": 8.477913096089036,
+      "learning_rate": 1.2481560914812812e-06,
+      "loss": 7.334,
+      "step": 943
+    },
+    {
+      "epoch": 0.12221251254167072,
+      "grad_norm": 5.035702792043305,
+      "learning_rate": 1.248134345467453e-06,
+      "loss": 4.7322,
+      "step": 944
+    },
+    {
+      "epoch": 0.12234197494902417,
+      "grad_norm": 2.471288339182732,
+      "learning_rate": 1.2481124721658043e-06,
+      "loss": 2.9038,
+      "step": 945
+    },
+    {
+      "epoch": 0.12234197494902417,
+      "eval_PRM Accuracy": 0.32075471698113206,
+      "eval_PRM F1": 0.25,
+      "eval_PRM F1 AUC": 0.5505500261917234,
+      "eval_PRM F1 Neg": 0.3793103448275862,
+      "eval_PRM NPV": 0.23655913978494625,
+      "eval_PRM Precision": 0.9230769230769231,
+      "eval_PRM Recall": 0.14457831325301204,
+      "eval_PRM Specificty": 0.9565217391304348,
+      "eval_loss": 1.0478515625,
+      "eval_runtime": 13.4994,
+      "eval_samples_per_second": 3.259,
+      "eval_steps_per_second": 0.222,
+      "step": 945
+    },
+    {
+      "epoch": 0.12247143735637764,
+      "grad_norm": 3.1523675503177246,
+      "learning_rate": 1.2480904715808035e-06,
+      "loss": 3.916,
+      "step": 946
+    },
+    {
+      "epoch": 0.1226008997637311,
+      "grad_norm": 4.728132501670559,
+      "learning_rate": 1.2480683437169441e-06,
+      "loss": 4.5342,
+      "step": 947
+    },
+    {
+      "epoch": 0.12273036217108457,
+      "grad_norm": 1.5046766742673054,
+      "learning_rate": 1.2480460885787465e-06,
+      "loss": 2.7151,
+      "step": 948
+    },
+    {
+      "epoch": 0.12285982457843804,
+      "grad_norm": 6.40134935479295,
+      "learning_rate": 1.248023706170757e-06,
+      "loss": 5.7119,
+      "step": 949
+    },
+    {
+      "epoch": 0.12298928698579151,
+      "grad_norm": 4.457082019511461,
+      "learning_rate": 1.2480011964975472e-06,
+      "loss": 4.4854,
+      "step": 950
+    },
+    {
+      "epoch": 0.12298928698579151,
+      "eval_PRM Accuracy": 0.32075471698113206,
+      "eval_PRM F1": 0.25,
+      "eval_PRM F1 AUC": 0.5505500261917234,
+      "eval_PRM F1 Neg": 0.3793103448275862,
+      "eval_PRM NPV": 0.23655913978494625,
+      "eval_PRM Precision": 0.9230769230769231,
+      "eval_PRM Recall": 0.14457831325301204,
+      "eval_PRM Specificty": 0.9565217391304348,
+      "eval_loss": 1.0516690015792847,
+      "eval_runtime": 14.6554,
+      "eval_samples_per_second": 3.002,
+      "eval_steps_per_second": 0.205,
+      "step": 950
+    },
+    {
+      "epoch": 0.12311874939314496,
+      "grad_norm": 1.6090340656282216,
+      "learning_rate": 1.2479785595637152e-06,
+      "loss": 2.9792,
+      "step": 951
+    },
+    {
+      "epoch": 0.12324821180049843,
+      "grad_norm": 2.9840250428661466,
+      "learning_rate": 1.2479557953738854e-06,
+      "loss": 3.25,
+      "step": 952
+    },
+    {
+      "epoch": 0.1233776742078519,
+      "grad_norm": 5.476687802538889,
+      "learning_rate": 1.2479329039327078e-06,
+      "loss": 4.4148,
+      "step": 953
+    },
+    {
+      "epoch": 0.12350713661520536,
+      "grad_norm": 4.3153723449268675,
+      "learning_rate": 1.247909885244858e-06,
+      "loss": 4.3208,
+      "step": 954
+    },
+    {
+      "epoch": 0.12363659902255883,
+      "grad_norm": 3.6232463768075824,
+      "learning_rate": 1.2478867393150387e-06,
+      "loss": 3.6406,
+      "step": 955
+    },
+    {
+      "epoch": 0.12363659902255883,
+      "eval_PRM Accuracy": 0.330188679245283,
+      "eval_PRM F1": 0.25263157894736843,
+      "eval_PRM F1 AUC": 0.572289156626506,
+      "eval_PRM F1 Neg": 0.39316239316239315,
+      "eval_PRM NPV": 0.24468085106382978,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.14457831325301204,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.0526455640792847,
+      "eval_runtime": 13.9618,
+      "eval_samples_per_second": 3.151,
+      "eval_steps_per_second": 0.215,
+      "step": 955
+    },
+    {
+      "epoch": 0.12376606142991228,
+      "grad_norm": 2.7932189865694106,
+      "learning_rate": 1.247863466147977e-06,
+      "loss": 3.1865,
+      "step": 956
+    },
+    {
+      "epoch": 0.12389552383726575,
+      "grad_norm": 5.355610282953668,
+      "learning_rate": 1.247840065748428e-06,
+      "loss": 4.791,
+      "step": 957
+    },
+    {
+      "epoch": 0.12402498624461922,
+      "grad_norm": 3.858147150203406,
+      "learning_rate": 1.247816538121171e-06,
+      "loss": 3.6785,
+      "step": 958
+    },
+    {
+      "epoch": 0.12415444865197268,
+      "grad_norm": 2.707026376077688,
+      "learning_rate": 1.247792883271012e-06,
+      "loss": 3.4209,
+      "step": 959
+    },
+    {
+      "epoch": 0.12428391105932615,
+      "grad_norm": 3.74142751010886,
+      "learning_rate": 1.2477691012027835e-06,
+      "loss": 4.124,
+      "step": 960
+    },
+    {
+      "epoch": 0.12428391105932615,
+      "eval_PRM Accuracy": 0.32075471698113206,
+      "eval_PRM F1": 0.23404255319148937,
+      "eval_PRM F1 AUC": 0.5662650602409639,
+      "eval_PRM F1 Neg": 0.3898305084745763,
+      "eval_PRM NPV": 0.24210526315789474,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.13253012048192772,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.0560191869735718,
+      "eval_runtime": 14.1069,
+      "eval_samples_per_second": 3.119,
+      "eval_steps_per_second": 0.213,
+      "step": 960
+    },
+    {
+      "epoch": 0.12441337346667962,
+      "grad_norm": 5.926902047536574,
+      "learning_rate": 1.247745191921343e-06,
+      "loss": 5.7275,
+      "step": 961
+    },
+    {
+      "epoch": 0.12454283587403307,
+      "grad_norm": 3.4307092732067392,
+      "learning_rate": 1.2477211554315744e-06,
+      "loss": 3.6108,
+      "step": 962
+    },
+    {
+      "epoch": 0.12467229828138654,
+      "grad_norm": 2.8287339202573696,
+      "learning_rate": 1.247696991738388e-06,
+      "loss": 3.7573,
+      "step": 963
+    },
+    {
+      "epoch": 0.12480176068874001,
+      "grad_norm": 4.414601818495286,
+      "learning_rate": 1.2476727008467195e-06,
+      "loss": 5.5747,
+      "step": 964
+    },
+    {
+      "epoch": 0.12493122309609347,
+      "grad_norm": 6.8284215950680585,
+      "learning_rate": 1.2476482827615308e-06,
+      "loss": 5.2762,
+      "step": 965
+    },
+    {
+      "epoch": 0.12493122309609347,
+      "eval_PRM Accuracy": 0.330188679245283,
+      "eval_PRM F1": 0.25263157894736843,
+      "eval_PRM F1 AUC": 0.572289156626506,
+      "eval_PRM F1 Neg": 0.39316239316239315,
+      "eval_PRM NPV": 0.24468085106382978,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.14457831325301204,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.05859375,
+      "eval_runtime": 13.9419,
+      "eval_samples_per_second": 3.156,
+      "eval_steps_per_second": 0.215,
+      "step": 965
+    },
+    {
+      "epoch": 0.12506068550344693,
+      "grad_norm": 2.8675498151650767,
+      "learning_rate": 1.2476237374878098e-06,
+      "loss": 3.4995,
+      "step": 966
+    },
+    {
+      "epoch": 0.1251901479108004,
+      "grad_norm": 4.540907752630362,
+      "learning_rate": 1.2475990650305705e-06,
+      "loss": 4.3252,
+      "step": 967
+    },
+    {
+      "epoch": 0.12531961031815386,
+      "grad_norm": 3.8781901142816855,
+      "learning_rate": 1.2475742653948527e-06,
+      "loss": 4.3799,
+      "step": 968
+    },
+    {
+      "epoch": 0.12544907272550734,
+      "grad_norm": 2.421194687012083,
+      "learning_rate": 1.2475493385857222e-06,
+      "loss": 3.0205,
+      "step": 969
+    },
+    {
+      "epoch": 0.1255785351328608,
+      "grad_norm": 5.3313143451088765,
+      "learning_rate": 1.2475242846082706e-06,
+      "loss": 4.7822,
+      "step": 970
+    },
+    {
+      "epoch": 0.1255785351328608,
+      "eval_PRM Accuracy": 0.32075471698113206,
+      "eval_PRM F1": 0.23404255319148937,
+      "eval_PRM F1 AUC": 0.5662650602409639,
+      "eval_PRM F1 Neg": 0.3898305084745763,
+      "eval_PRM NPV": 0.24210526315789474,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.13253012048192772,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.0603693723678589,
+      "eval_runtime": 15.3503,
+      "eval_samples_per_second": 2.866,
+      "eval_steps_per_second": 0.195,
+      "step": 970
+    },
+    {
+      "epoch": 0.12570799754021425,
+      "grad_norm": 8.339895406926132,
+      "learning_rate": 1.247499103467616e-06,
+      "loss": 5.8936,
+      "step": 971
+    },
+    {
+      "epoch": 0.12583745994756773,
+      "grad_norm": 5.317072011717124,
+      "learning_rate": 1.247473795168902e-06,
+      "loss": 4.4404,
+      "step": 972
+    },
+    {
+      "epoch": 0.12596692235492118,
+      "grad_norm": 2.7815267226558134,
+      "learning_rate": 1.2474483597172985e-06,
+      "loss": 3.186,
+      "step": 973
+    },
+    {
+      "epoch": 0.12609638476227467,
+      "grad_norm": 3.456682002237875,
+      "learning_rate": 1.247422797118001e-06,
+      "loss": 3.3911,
+      "step": 974
+    },
+    {
+      "epoch": 0.12622584716962812,
+      "grad_norm": 7.608816983260536,
+      "learning_rate": 1.2473971073762312e-06,
+      "loss": 5.8828,
+      "step": 975
+    },
+    {
+      "epoch": 0.12622584716962812,
+      "eval_PRM Accuracy": 0.32075471698113206,
+      "eval_PRM F1": 0.23404255319148937,
+      "eval_PRM F1 AUC": 0.5662650602409639,
+      "eval_PRM F1 Neg": 0.3898305084745763,
+      "eval_PRM NPV": 0.24210526315789474,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.13253012048192772,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.064009189605713,
+      "eval_runtime": 15.1268,
+      "eval_samples_per_second": 2.909,
+      "eval_steps_per_second": 0.198,
+      "step": 975
+    },
+    {
+      "epoch": 0.12635530957698157,
+      "grad_norm": 3.5969502525506396,
+      "learning_rate": 1.2473712904972369e-06,
+      "loss": 3.7305,
+      "step": 976
+    },
+    {
+      "epoch": 0.12648477198433505,
+      "grad_norm": 4.646756236458957,
+      "learning_rate": 1.2473453464862914e-06,
+      "loss": 4.8125,
+      "step": 977
+    },
+    {
+      "epoch": 0.1266142343916885,
+      "grad_norm": 4.720806105640083,
+      "learning_rate": 1.247319275348695e-06,
+      "loss": 5.4941,
+      "step": 978
+    },
+    {
+      "epoch": 0.126743696799042,
+      "grad_norm": 4.248075920549624,
+      "learning_rate": 1.2472930770897722e-06,
+      "loss": 3.8857,
+      "step": 979
+    },
+    {
+      "epoch": 0.12687315920639544,
+      "grad_norm": 2.8605290332286293,
+      "learning_rate": 1.2472667517148754e-06,
+      "loss": 3.0388,
+      "step": 980
+    },
+    {
+      "epoch": 0.12687315920639544,
+      "eval_PRM Accuracy": 0.32075471698113206,
+      "eval_PRM F1": 0.23404255319148937,
+      "eval_PRM F1 AUC": 0.5662650602409639,
+      "eval_PRM F1 Neg": 0.3898305084745763,
+      "eval_PRM NPV": 0.24210526315789474,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.13253012048192772,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.0687144994735718,
+      "eval_runtime": 14.0645,
+      "eval_samples_per_second": 3.128,
+      "eval_steps_per_second": 0.213,
+      "step": 980
+    },
+    {
+      "epoch": 0.1270026216137489,
+      "grad_norm": 2.854697893128177,
+      "learning_rate": 1.2472402992293817e-06,
+      "loss": 3.5352,
+      "step": 981
+    },
+    {
+      "epoch": 0.12713208402110238,
+      "grad_norm": 3.126921632202384,
+      "learning_rate": 1.2472137196386946e-06,
+      "loss": 3.4629,
+      "step": 982
+    },
+    {
+      "epoch": 0.12726154642845583,
+      "grad_norm": 3.1697296364764194,
+      "learning_rate": 1.2471870129482436e-06,
+      "loss": 3.7405,
+      "step": 983
+    },
+    {
+      "epoch": 0.1273910088358093,
+      "grad_norm": 5.887579579012297,
+      "learning_rate": 1.2471601791634843e-06,
+      "loss": 4.1284,
+      "step": 984
+    },
+    {
+      "epoch": 0.12752047124316276,
+      "grad_norm": 5.737257026895172,
+      "learning_rate": 1.2471332182898976e-06,
+      "loss": 4.3589,
+      "step": 985
+    },
+    {
+      "epoch": 0.12752047124316276,
+      "eval_PRM Accuracy": 0.32075471698113206,
+      "eval_PRM F1": 0.23404255319148937,
+      "eval_PRM F1 AUC": 0.5662650602409639,
+      "eval_PRM F1 Neg": 0.3898305084745763,
+      "eval_PRM NPV": 0.24210526315789474,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.13253012048192772,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.0690696239471436,
+      "eval_runtime": 14.6042,
+      "eval_samples_per_second": 3.013,
+      "eval_steps_per_second": 0.205,
+      "step": 985
+    },
+    {
+      "epoch": 0.12764993365051625,
+      "grad_norm": 3.954361826605491,
+      "learning_rate": 1.247106130332991e-06,
+      "loss": 4.3667,
+      "step": 986
+    },
+    {
+      "epoch": 0.1277793960578697,
+      "grad_norm": 5.238380771473893,
+      "learning_rate": 1.2470789152982976e-06,
+      "loss": 4.2905,
+      "step": 987
+    },
+    {
+      "epoch": 0.12790885846522315,
+      "grad_norm": 3.683852590339371,
+      "learning_rate": 1.247051573191377e-06,
+      "loss": 4.1904,
+      "step": 988
+    },
+    {
+      "epoch": 0.12803832087257663,
+      "grad_norm": 3.800600473229702,
+      "learning_rate": 1.247024104017814e-06,
+      "loss": 4.1104,
+      "step": 989
+    },
+    {
+      "epoch": 0.1281677832799301,
+      "grad_norm": 3.8831777476402936,
+      "learning_rate": 1.24699650778322e-06,
+      "loss": 3.3628,
+      "step": 990
+    },
+    {
+      "epoch": 0.1281677832799301,
+      "eval_PRM Accuracy": 0.32075471698113206,
+      "eval_PRM F1": 0.23404255319148937,
+      "eval_PRM F1 AUC": 0.5662650602409639,
+      "eval_PRM F1 Neg": 0.3898305084745763,
+      "eval_PRM NPV": 0.24210526315789474,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.13253012048192772,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.0739524364471436,
+      "eval_runtime": 15.813,
+      "eval_samples_per_second": 2.783,
+      "eval_steps_per_second": 0.19,
+      "step": 990
+    },
+    {
+      "epoch": 0.12829724568728357,
+      "grad_norm": 3.8936271763020236,
+      "learning_rate": 1.246968784493232e-06,
+      "loss": 4.4331,
+      "step": 991
+    },
+    {
+      "epoch": 0.12842670809463702,
+      "grad_norm": 5.1837856890403415,
+      "learning_rate": 1.2469409341535132e-06,
+      "loss": 4.3799,
+      "step": 992
+    },
+    {
+      "epoch": 0.12855617050199047,
+      "grad_norm": 2.5855666146103746,
+      "learning_rate": 1.246912956769752e-06,
+      "loss": 3.2197,
+      "step": 993
+    },
+    {
+      "epoch": 0.12868563290934396,
+      "grad_norm": 3.0354135540536085,
+      "learning_rate": 1.2468848523476642e-06,
+      "loss": 3.606,
+      "step": 994
+    },
+    {
+      "epoch": 0.1288150953166974,
+      "grad_norm": 7.425821204767459,
+      "learning_rate": 1.24685662089299e-06,
+      "loss": 4.8413,
+      "step": 995
+    },
+    {
+      "epoch": 0.1288150953166974,
+      "eval_PRM Accuracy": 0.32075471698113206,
+      "eval_PRM F1": 0.23404255319148937,
+      "eval_PRM F1 AUC": 0.5662650602409639,
+      "eval_PRM F1 Neg": 0.3898305084745763,
+      "eval_PRM NPV": 0.24210526315789474,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.13253012048192772,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.074751377105713,
+      "eval_runtime": 13.9792,
+      "eval_samples_per_second": 3.148,
+      "eval_steps_per_second": 0.215,
+      "step": 995
+    },
+    {
+      "epoch": 0.1289445577240509,
+      "grad_norm": 6.017254284461653,
+      "learning_rate": 1.2468282624114965e-06,
+      "loss": 5.498,
+      "step": 996
+    },
+    {
+      "epoch": 0.12907402013140434,
+      "grad_norm": 2.397532893122352,
+      "learning_rate": 1.2467997769089768e-06,
+      "loss": 3.0735,
+      "step": 997
+    },
+    {
+      "epoch": 0.1292034825387578,
+      "grad_norm": 3.5378089273062403,
+      "learning_rate": 1.2467711643912493e-06,
+      "loss": 3.6301,
+      "step": 998
+    },
+    {
+      "epoch": 0.12933294494611128,
+      "grad_norm": 4.768367773201599,
+      "learning_rate": 1.2467424248641585e-06,
+      "loss": 4.478,
+      "step": 999
+    },
+    {
+      "epoch": 0.12946240735346473,
+      "grad_norm": 5.105482652077706,
+      "learning_rate": 1.2467135583335757e-06,
+      "loss": 4.4482,
+      "step": 1000
+    },
+    {
+      "epoch": 0.12946240735346473,
+      "eval_PRM Accuracy": 0.32075471698113206,
+      "eval_PRM F1": 0.23404255319148937,
+      "eval_PRM F1 AUC": 0.5662650602409639,
+      "eval_PRM F1 Neg": 0.3898305084745763,
+      "eval_PRM NPV": 0.24210526315789474,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.13253012048192772,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.078657627105713,
+      "eval_runtime": 14.0803,
+      "eval_samples_per_second": 3.125,
+      "eval_steps_per_second": 0.213,
+      "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