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": 0.7120432404440561,
+  "epoch": 0.7767744441207884,
   "eval_steps": 5,
-  "global_step": 5500,
+  "global_step": 6000,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
@@ -56123,6 +56123,5106 @@
       "eval_samples_per_second": 3.156,
       "eval_steps_per_second": 0.215,
       "step": 5500
+    },
+    {
+      "epoch": 0.7121727028514095,
+      "grad_norm": 1.6587128474393105,
+      "learning_rate": 2.897934888603117e-07,
+      "loss": 2.4314,
+      "step": 5501
+    },
+    {
+      "epoch": 0.712302165258763,
+      "grad_norm": 1.4162562684840903,
+      "learning_rate": 2.8955511049642156e-07,
+      "loss": 2.166,
+      "step": 5502
+    },
+    {
+      "epoch": 0.7124316276661165,
+      "grad_norm": 1.6084256458162793,
+      "learning_rate": 2.8931680065394907e-07,
+      "loss": 2.478,
+      "step": 5503
+    },
+    {
+      "epoch": 0.7125610900734699,
+      "grad_norm": 2.464881038647762,
+      "learning_rate": 2.8907855938157427e-07,
+      "loss": 2.3914,
+      "step": 5504
+    },
+    {
+      "epoch": 0.7126905524808234,
+      "grad_norm": 1.2356605888206103,
+      "learning_rate": 2.888403867279623e-07,
+      "loss": 2.3491,
+      "step": 5505
+    },
+    {
+      "epoch": 0.7126905524808234,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.511474609375,
+      "eval_runtime": 13.4246,
+      "eval_samples_per_second": 3.278,
+      "eval_steps_per_second": 0.223,
+      "step": 5505
+    },
+    {
+      "epoch": 0.7128200148881768,
+      "grad_norm": 1.608224140789146,
+      "learning_rate": 2.8860228274176517e-07,
+      "loss": 2.457,
+      "step": 5506
+    },
+    {
+      "epoch": 0.7129494772955303,
+      "grad_norm": 1.2687054741524142,
+      "learning_rate": 2.8836424747162034e-07,
+      "loss": 2.1235,
+      "step": 5507
+    },
+    {
+      "epoch": 0.7130789397028838,
+      "grad_norm": 2.2563542475474856,
+      "learning_rate": 2.88126280966151e-07,
+      "loss": 2.4487,
+      "step": 5508
+    },
+    {
+      "epoch": 0.7132084021102373,
+      "grad_norm": 2.97185135449286,
+      "learning_rate": 2.878883832739671e-07,
+      "loss": 2.3892,
+      "step": 5509
+    },
+    {
+      "epoch": 0.7133378645175907,
+      "grad_norm": 0.945093829392246,
+      "learning_rate": 2.8765055444366364e-07,
+      "loss": 2.1051,
+      "step": 5510
+    },
+    {
+      "epoch": 0.7133378645175907,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.510054111480713,
+      "eval_runtime": 13.7777,
+      "eval_samples_per_second": 3.194,
+      "eval_steps_per_second": 0.218,
+      "step": 5510
+    },
+    {
+      "epoch": 0.7134673269249442,
+      "grad_norm": 1.5028758044545727,
+      "learning_rate": 2.8741279452382246e-07,
+      "loss": 2.1548,
+      "step": 5511
+    },
+    {
+      "epoch": 0.7135967893322976,
+      "grad_norm": 3.9063964122936476,
+      "learning_rate": 2.8717510356301055e-07,
+      "loss": 2.9185,
+      "step": 5512
+    },
+    {
+      "epoch": 0.7137262517396511,
+      "grad_norm": 2.330277696128063,
+      "learning_rate": 2.8693748160978095e-07,
+      "loss": 2.3647,
+      "step": 5513
+    },
+    {
+      "epoch": 0.7138557141470045,
+      "grad_norm": 2.0524267159751397,
+      "learning_rate": 2.866999287126731e-07,
+      "loss": 2.417,
+      "step": 5514
+    },
+    {
+      "epoch": 0.713985176554358,
+      "grad_norm": 1.64944727637688,
+      "learning_rate": 2.8646244492021163e-07,
+      "loss": 2.3623,
+      "step": 5515
+    },
+    {
+      "epoch": 0.713985176554358,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.507768154144287,
+      "eval_runtime": 13.7639,
+      "eval_samples_per_second": 3.197,
+      "eval_steps_per_second": 0.218,
+      "step": 5515
+    },
+    {
+      "epoch": 0.7141146389617115,
+      "grad_norm": 4.338832030132758,
+      "learning_rate": 2.8622503028090764e-07,
+      "loss": 2.4463,
+      "step": 5516
+    },
+    {
+      "epoch": 0.7142441013690649,
+      "grad_norm": 1.7430278071155016,
+      "learning_rate": 2.859876848432581e-07,
+      "loss": 2.4495,
+      "step": 5517
+    },
+    {
+      "epoch": 0.7143735637764185,
+      "grad_norm": 1.907112513554861,
+      "learning_rate": 2.857504086557454e-07,
+      "loss": 2.1299,
+      "step": 5518
+    },
+    {
+      "epoch": 0.7145030261837719,
+      "grad_norm": 2.708399660442652,
+      "learning_rate": 2.85513201766838e-07,
+      "loss": 2.5608,
+      "step": 5519
+    },
+    {
+      "epoch": 0.7146324885911254,
+      "grad_norm": 1.296584160472222,
+      "learning_rate": 2.8527606422499e-07,
+      "loss": 2.4521,
+      "step": 5520
+    },
+    {
+      "epoch": 0.7146324885911254,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.5079013109207153,
+      "eval_runtime": 13.6967,
+      "eval_samples_per_second": 3.212,
+      "eval_steps_per_second": 0.219,
+      "step": 5520
+    },
+    {
+      "epoch": 0.7147619509984788,
+      "grad_norm": 1.3739841531314605,
+      "learning_rate": 2.850389960786418e-07,
+      "loss": 2.3652,
+      "step": 5521
+    },
+    {
+      "epoch": 0.7148914134058323,
+      "grad_norm": 1.623312421708892,
+      "learning_rate": 2.848019973762195e-07,
+      "loss": 2.4771,
+      "step": 5522
+    },
+    {
+      "epoch": 0.7150208758131857,
+      "grad_norm": 1.958130293020498,
+      "learning_rate": 2.845650681661348e-07,
+      "loss": 2.5439,
+      "step": 5523
+    },
+    {
+      "epoch": 0.7151503382205392,
+      "grad_norm": 1.6339741316942522,
+      "learning_rate": 2.84328208496785e-07,
+      "loss": 2.3599,
+      "step": 5524
+    },
+    {
+      "epoch": 0.7152798006278926,
+      "grad_norm": 1.1990610677201585,
+      "learning_rate": 2.840914184165539e-07,
+      "loss": 2.4148,
+      "step": 5525
+    },
+    {
+      "epoch": 0.7152798006278926,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.512162685394287,
+      "eval_runtime": 13.4574,
+      "eval_samples_per_second": 3.27,
+      "eval_steps_per_second": 0.223,
+      "step": 5525
+    },
+    {
+      "epoch": 0.7154092630352461,
+      "grad_norm": 1.7739544361348252,
+      "learning_rate": 2.8385469797381026e-07,
+      "loss": 2.415,
+      "step": 5526
+    },
+    {
+      "epoch": 0.7155387254425996,
+      "grad_norm": 3.45287956507049,
+      "learning_rate": 2.836180472169096e-07,
+      "loss": 2.6968,
+      "step": 5527
+    },
+    {
+      "epoch": 0.7156681878499531,
+      "grad_norm": 1.2147623211617138,
+      "learning_rate": 2.8338146619419195e-07,
+      "loss": 2.1594,
+      "step": 5528
+    },
+    {
+      "epoch": 0.7157976502573066,
+      "grad_norm": 1.3240594729424797,
+      "learning_rate": 2.831449549539844e-07,
+      "loss": 2.3896,
+      "step": 5529
+    },
+    {
+      "epoch": 0.71592711266466,
+      "grad_norm": 1.476022294129693,
+      "learning_rate": 2.8290851354459887e-07,
+      "loss": 2.248,
+      "step": 5530
+    },
+    {
+      "epoch": 0.71592711266466,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.5107421875,
+      "eval_runtime": 14.0462,
+      "eval_samples_per_second": 3.133,
+      "eval_steps_per_second": 0.214,
+      "step": 5530
+    },
+    {
+      "epoch": 0.7160565750720135,
+      "grad_norm": 1.824707853093141,
+      "learning_rate": 2.826721420143331e-07,
+      "loss": 2.4666,
+      "step": 5531
+    },
+    {
+      "epoch": 0.7161860374793669,
+      "grad_norm": 1.4682217075595523,
+      "learning_rate": 2.8243584041147117e-07,
+      "loss": 2.3921,
+      "step": 5532
+    },
+    {
+      "epoch": 0.7163154998867204,
+      "grad_norm": 2.168559203650329,
+      "learning_rate": 2.821996087842822e-07,
+      "loss": 2.5039,
+      "step": 5533
+    },
+    {
+      "epoch": 0.7164449622940738,
+      "grad_norm": 1.5776966453526566,
+      "learning_rate": 2.8196344718102145e-07,
+      "loss": 2.4316,
+      "step": 5534
+    },
+    {
+      "epoch": 0.7165744247014273,
+      "grad_norm": 1.5985645007345604,
+      "learning_rate": 2.8172735564992977e-07,
+      "loss": 2.2637,
+      "step": 5535
+    },
+    {
+      "epoch": 0.7165744247014273,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.5105868577957153,
+      "eval_runtime": 14.3985,
+      "eval_samples_per_second": 3.056,
+      "eval_steps_per_second": 0.208,
+      "step": 5535
+    },
+    {
+      "epoch": 0.7167038871087807,
+      "grad_norm": 2.963478626299164,
+      "learning_rate": 2.814913342392332e-07,
+      "loss": 2.6663,
+      "step": 5536
+    },
+    {
+      "epoch": 0.7168333495161342,
+      "grad_norm": 1.0946133801481273,
+      "learning_rate": 2.8125538299714455e-07,
+      "loss": 2.3401,
+      "step": 5537
+    },
+    {
+      "epoch": 0.7169628119234878,
+      "grad_norm": 1.8524375943717375,
+      "learning_rate": 2.81019501971861e-07,
+      "loss": 2.2546,
+      "step": 5538
+    },
+    {
+      "epoch": 0.7170922743308412,
+      "grad_norm": 1.7007590761978473,
+      "learning_rate": 2.8078369121156633e-07,
+      "loss": 2.1255,
+      "step": 5539
+    },
+    {
+      "epoch": 0.7172217367381947,
+      "grad_norm": 1.7471511623985778,
+      "learning_rate": 2.805479507644302e-07,
+      "loss": 2.2898,
+      "step": 5540
+    },
+    {
+      "epoch": 0.7172217367381947,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.508589267730713,
+      "eval_runtime": 14.602,
+      "eval_samples_per_second": 3.013,
+      "eval_steps_per_second": 0.205,
+      "step": 5540
+    },
+    {
+      "epoch": 0.7173511991455481,
+      "grad_norm": 2.65354892426599,
+      "learning_rate": 2.8031228067860633e-07,
+      "loss": 2.4062,
+      "step": 5541
+    },
+    {
+      "epoch": 0.7174806615529016,
+      "grad_norm": 1.7348926744322892,
+      "learning_rate": 2.800766810022359e-07,
+      "loss": 2.3301,
+      "step": 5542
+    },
+    {
+      "epoch": 0.717610123960255,
+      "grad_norm": 2.4364528588960126,
+      "learning_rate": 2.798411517834444e-07,
+      "loss": 2.7969,
+      "step": 5543
+    },
+    {
+      "epoch": 0.7177395863676085,
+      "grad_norm": 2.3385069966979697,
+      "learning_rate": 2.7960569307034375e-07,
+      "loss": 2.8467,
+      "step": 5544
+    },
+    {
+      "epoch": 0.7178690487749619,
+      "grad_norm": 1.3253198369031496,
+      "learning_rate": 2.7937030491103125e-07,
+      "loss": 2.3762,
+      "step": 5545
+    },
+    {
+      "epoch": 0.7178690487749619,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.5106533765792847,
+      "eval_runtime": 14.6764,
+      "eval_samples_per_second": 2.998,
+      "eval_steps_per_second": 0.204,
+      "step": 5545
+    },
+    {
+      "epoch": 0.7179985111823154,
+      "grad_norm": 1.864054273156971,
+      "learning_rate": 2.7913498735358964e-07,
+      "loss": 2.3914,
+      "step": 5546
+    },
+    {
+      "epoch": 0.718127973589669,
+      "grad_norm": 1.145738931308922,
+      "learning_rate": 2.78899740446087e-07,
+      "loss": 2.2905,
+      "step": 5547
+    },
+    {
+      "epoch": 0.7182574359970224,
+      "grad_norm": 2.223617748065961,
+      "learning_rate": 2.7866456423657775e-07,
+      "loss": 2.4192,
+      "step": 5548
+    },
+    {
+      "epoch": 0.7183868984043759,
+      "grad_norm": 1.927960560650502,
+      "learning_rate": 2.7842945877310087e-07,
+      "loss": 2.3726,
+      "step": 5549
+    },
+    {
+      "epoch": 0.7185163608117293,
+      "grad_norm": 1.280723933023443,
+      "learning_rate": 2.7819442410368195e-07,
+      "loss": 2.2849,
+      "step": 5550
+    },
+    {
+      "epoch": 0.7185163608117293,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.5106533765792847,
+      "eval_runtime": 13.8223,
+      "eval_samples_per_second": 3.183,
+      "eval_steps_per_second": 0.217,
+      "step": 5550
+    },
+    {
+      "epoch": 0.7186458232190828,
+      "grad_norm": 2.69337669376921,
+      "learning_rate": 2.7795946027633135e-07,
+      "loss": 2.6292,
+      "step": 5551
+    },
+    {
+      "epoch": 0.7187752856264362,
+      "grad_norm": 1.1707863558550546,
+      "learning_rate": 2.7772456733904487e-07,
+      "loss": 2.3149,
+      "step": 5552
+    },
+    {
+      "epoch": 0.7189047480337897,
+      "grad_norm": 2.3302889611217963,
+      "learning_rate": 2.7748974533980485e-07,
+      "loss": 2.3171,
+      "step": 5553
+    },
+    {
+      "epoch": 0.7190342104411431,
+      "grad_norm": 2.858880558289602,
+      "learning_rate": 2.7725499432657766e-07,
+      "loss": 2.3002,
+      "step": 5554
+    },
+    {
+      "epoch": 0.7191636728484966,
+      "grad_norm": 1.7256317431863324,
+      "learning_rate": 2.770203143473167e-07,
+      "loss": 2.3389,
+      "step": 5555
+    },
+    {
+      "epoch": 0.7191636728484966,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.5114967823028564,
+      "eval_runtime": 14.9309,
+      "eval_samples_per_second": 2.947,
+      "eval_steps_per_second": 0.201,
+      "step": 5555
+    },
+    {
+      "epoch": 0.71929313525585,
+      "grad_norm": 1.5143480210948022,
+      "learning_rate": 2.767857054499595e-07,
+      "loss": 2.2592,
+      "step": 5556
+    },
+    {
+      "epoch": 0.7194225976632036,
+      "grad_norm": 1.295316019296011,
+      "learning_rate": 2.765511676824301e-07,
+      "loss": 2.2485,
+      "step": 5557
+    },
+    {
+      "epoch": 0.719552060070557,
+      "grad_norm": 2.2133328807686854,
+      "learning_rate": 2.763167010926376e-07,
+      "loss": 2.334,
+      "step": 5558
+    },
+    {
+      "epoch": 0.7196815224779105,
+      "grad_norm": 1.173305100737012,
+      "learning_rate": 2.7608230572847604e-07,
+      "loss": 2.1846,
+      "step": 5559
+    },
+    {
+      "epoch": 0.719810984885264,
+      "grad_norm": 1.8169553166426808,
+      "learning_rate": 2.7584798163782603e-07,
+      "loss": 2.4248,
+      "step": 5560
+    },
+    {
+      "epoch": 0.719810984885264,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.5099875926971436,
+      "eval_runtime": 14.6133,
+      "eval_samples_per_second": 3.011,
+      "eval_steps_per_second": 0.205,
+      "step": 5560
+    },
+    {
+      "epoch": 0.7199404472926174,
+      "grad_norm": 2.3093845859188873,
+      "learning_rate": 2.7561372886855244e-07,
+      "loss": 2.5293,
+      "step": 5561
+    },
+    {
+      "epoch": 0.7200699096999709,
+      "grad_norm": 1.435623281551701,
+      "learning_rate": 2.7537954746850686e-07,
+      "loss": 2.3755,
+      "step": 5562
+    },
+    {
+      "epoch": 0.7201993721073243,
+      "grad_norm": 1.1289074039455078,
+      "learning_rate": 2.751454374855251e-07,
+      "loss": 2.1689,
+      "step": 5563
+    },
+    {
+      "epoch": 0.7203288345146778,
+      "grad_norm": 2.1193161850721127,
+      "learning_rate": 2.7491139896742873e-07,
+      "loss": 2.3691,
+      "step": 5564
+    },
+    {
+      "epoch": 0.7204582969220312,
+      "grad_norm": 1.6980072454782136,
+      "learning_rate": 2.746774319620253e-07,
+      "loss": 2.4399,
+      "step": 5565
+    },
+    {
+      "epoch": 0.7204582969220312,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.50927734375,
+      "eval_runtime": 13.0915,
+      "eval_samples_per_second": 3.361,
+      "eval_steps_per_second": 0.229,
+      "step": 5565
+    },
+    {
+      "epoch": 0.7205877593293847,
+      "grad_norm": 1.5075953458803324,
+      "learning_rate": 2.744435365171069e-07,
+      "loss": 2.3982,
+      "step": 5566
+    },
+    {
+      "epoch": 0.7207172217367382,
+      "grad_norm": 1.6918291383960713,
+      "learning_rate": 2.7420971268045167e-07,
+      "loss": 2.4092,
+      "step": 5567
+    },
+    {
+      "epoch": 0.7208466841440917,
+      "grad_norm": 1.2506097152652387,
+      "learning_rate": 2.739759604998232e-07,
+      "loss": 2.2695,
+      "step": 5568
+    },
+    {
+      "epoch": 0.7209761465514452,
+      "grad_norm": 2.954445096213623,
+      "learning_rate": 2.7374228002296934e-07,
+      "loss": 2.6201,
+      "step": 5569
+    },
+    {
+      "epoch": 0.7211056089587986,
+      "grad_norm": 1.3859044433106302,
+      "learning_rate": 2.7350867129762447e-07,
+      "loss": 2.356,
+      "step": 5570
+    },
+    {
+      "epoch": 0.7211056089587986,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.5093661546707153,
+      "eval_runtime": 14.3231,
+      "eval_samples_per_second": 3.072,
+      "eval_steps_per_second": 0.209,
+      "step": 5570
+    },
+    {
+      "epoch": 0.7212350713661521,
+      "grad_norm": 1.5674237481977822,
+      "learning_rate": 2.7327513437150807e-07,
+      "loss": 2.3964,
+      "step": 5571
+    },
+    {
+      "epoch": 0.7213645337735055,
+      "grad_norm": 2.379028384116191,
+      "learning_rate": 2.730416692923244e-07,
+      "loss": 2.4958,
+      "step": 5572
+    },
+    {
+      "epoch": 0.721493996180859,
+      "grad_norm": 1.5455502487512371,
+      "learning_rate": 2.728082761077639e-07,
+      "loss": 2.5,
+      "step": 5573
+    },
+    {
+      "epoch": 0.7216234585882124,
+      "grad_norm": 1.7377203006836992,
+      "learning_rate": 2.725749548655016e-07,
+      "loss": 2.3423,
+      "step": 5574
+    },
+    {
+      "epoch": 0.7217529209955659,
+      "grad_norm": 1.6312216160984048,
+      "learning_rate": 2.723417056131978e-07,
+      "loss": 2.4758,
+      "step": 5575
+    },
+    {
+      "epoch": 0.7217529209955659,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.508589267730713,
+      "eval_runtime": 14.4312,
+      "eval_samples_per_second": 3.049,
+      "eval_steps_per_second": 0.208,
+      "step": 5575
+    },
+    {
+      "epoch": 0.7218823834029193,
+      "grad_norm": 1.6272893864648796,
+      "learning_rate": 2.7210852839849896e-07,
+      "loss": 2.4421,
+      "step": 5576
+    },
+    {
+      "epoch": 0.7220118458102729,
+      "grad_norm": 1.6615919727143313,
+      "learning_rate": 2.7187542326903576e-07,
+      "loss": 2.3574,
+      "step": 5577
+    },
+    {
+      "epoch": 0.7221413082176263,
+      "grad_norm": 2.6841609605575374,
+      "learning_rate": 2.7164239027242507e-07,
+      "loss": 2.5779,
+      "step": 5578
+    },
+    {
+      "epoch": 0.7222707706249798,
+      "grad_norm": 2.1363165319634607,
+      "learning_rate": 2.7140942945626844e-07,
+      "loss": 2.3572,
+      "step": 5579
+    },
+    {
+      "epoch": 0.7224002330323332,
+      "grad_norm": 2.1343990824011305,
+      "learning_rate": 2.7117654086815257e-07,
+      "loss": 2.2312,
+      "step": 5580
+    },
+    {
+      "epoch": 0.7224002330323332,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.5079013109207153,
+      "eval_runtime": 14.0868,
+      "eval_samples_per_second": 3.123,
+      "eval_steps_per_second": 0.213,
+      "step": 5580
+    },
+    {
+      "epoch": 0.7225296954396867,
+      "grad_norm": 1.2388845462084674,
+      "learning_rate": 2.709437245556501e-07,
+      "loss": 2.2566,
+      "step": 5581
+    },
+    {
+      "epoch": 0.7226591578470402,
+      "grad_norm": 1.5019589883987903,
+      "learning_rate": 2.7071098056631805e-07,
+      "loss": 2.4832,
+      "step": 5582
+    },
+    {
+      "epoch": 0.7227886202543936,
+      "grad_norm": 1.3062292130862347,
+      "learning_rate": 2.7047830894769956e-07,
+      "loss": 2.3442,
+      "step": 5583
+    },
+    {
+      "epoch": 0.7229180826617471,
+      "grad_norm": 1.979726227766332,
+      "learning_rate": 2.702457097473221e-07,
+      "loss": 2.3982,
+      "step": 5584
+    },
+    {
+      "epoch": 0.7230475450691005,
+      "grad_norm": 2.37744644501221,
+      "learning_rate": 2.7001318301269917e-07,
+      "loss": 2.4646,
+      "step": 5585
+    },
+    {
+      "epoch": 0.7230475450691005,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.5107200145721436,
+      "eval_runtime": 14.277,
+      "eval_samples_per_second": 3.082,
+      "eval_steps_per_second": 0.21,
+      "step": 5585
+    },
+    {
+      "epoch": 0.723177007476454,
+      "grad_norm": 1.3507179228392634,
+      "learning_rate": 2.6978072879132885e-07,
+      "loss": 2.2634,
+      "step": 5586
+    },
+    {
+      "epoch": 0.7233064698838075,
+      "grad_norm": 1.840242546852928,
+      "learning_rate": 2.695483471306945e-07,
+      "loss": 2.519,
+      "step": 5587
+    },
+    {
+      "epoch": 0.723435932291161,
+      "grad_norm": 1.5270676035336725,
+      "learning_rate": 2.6931603807826506e-07,
+      "loss": 2.3647,
+      "step": 5588
+    },
+    {
+      "epoch": 0.7235653946985144,
+      "grad_norm": 1.1749397727549062,
+      "learning_rate": 2.69083801681494e-07,
+      "loss": 2.3071,
+      "step": 5589
+    },
+    {
+      "epoch": 0.7236948571058679,
+      "grad_norm": 1.3322640689962522,
+      "learning_rate": 2.688516379878208e-07,
+      "loss": 2.3247,
+      "step": 5590
+    },
+    {
+      "epoch": 0.7236948571058679,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.508500576019287,
+      "eval_runtime": 14.9033,
+      "eval_samples_per_second": 2.952,
+      "eval_steps_per_second": 0.201,
+      "step": 5590
+    },
+    {
+      "epoch": 0.7238243195132213,
+      "grad_norm": 1.705210521195304,
+      "learning_rate": 2.686195470446693e-07,
+      "loss": 2.4436,
+      "step": 5591
+    },
+    {
+      "epoch": 0.7239537819205748,
+      "grad_norm": 2.192928496883699,
+      "learning_rate": 2.683875288994487e-07,
+      "loss": 2.4543,
+      "step": 5592
+    },
+    {
+      "epoch": 0.7240832443279283,
+      "grad_norm": 2.7342780910093536,
+      "learning_rate": 2.6815558359955343e-07,
+      "loss": 2.2744,
+      "step": 5593
+    },
+    {
+      "epoch": 0.7242127067352817,
+      "grad_norm": 1.4334541547349253,
+      "learning_rate": 2.6792371119236335e-07,
+      "loss": 2.5435,
+      "step": 5594
+    },
+    {
+      "epoch": 0.7243421691426352,
+      "grad_norm": 1.703577301806479,
+      "learning_rate": 2.6769191172524267e-07,
+      "loss": 2.3723,
+      "step": 5595
+    },
+    {
+      "epoch": 0.7243421691426352,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.5086115598678589,
+      "eval_runtime": 14.8315,
+      "eval_samples_per_second": 2.967,
+      "eval_steps_per_second": 0.202,
+      "step": 5595
+    },
+    {
+      "epoch": 0.7244716315499887,
+      "grad_norm": 2.085591545761733,
+      "learning_rate": 2.6746018524554153e-07,
+      "loss": 2.564,
+      "step": 5596
+    },
+    {
+      "epoch": 0.7246010939573422,
+      "grad_norm": 1.413747440699509,
+      "learning_rate": 2.6722853180059453e-07,
+      "loss": 2.459,
+      "step": 5597
+    },
+    {
+      "epoch": 0.7247305563646956,
+      "grad_norm": 1.3260162174318653,
+      "learning_rate": 2.6699695143772143e-07,
+      "loss": 2.3755,
+      "step": 5598
+    },
+    {
+      "epoch": 0.7248600187720491,
+      "grad_norm": 2.32683899665351,
+      "learning_rate": 2.6676544420422763e-07,
+      "loss": 2.407,
+      "step": 5599
+    },
+    {
+      "epoch": 0.7249894811794025,
+      "grad_norm": 1.7434693139780224,
+      "learning_rate": 2.6653401014740266e-07,
+      "loss": 2.4805,
+      "step": 5600
+    },
+    {
+      "epoch": 0.7249894811794025,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.5093661546707153,
+      "eval_runtime": 13.5865,
+      "eval_samples_per_second": 3.239,
+      "eval_steps_per_second": 0.221,
+      "step": 5600
+    },
+    {
+      "epoch": 0.725118943586756,
+      "grad_norm": 1.3972322914181314,
+      "learning_rate": 2.6630264931452215e-07,
+      "loss": 2.2969,
+      "step": 5601
+    },
+    {
+      "epoch": 0.7252484059941094,
+      "grad_norm": 1.2590977917336583,
+      "learning_rate": 2.6607136175284597e-07,
+      "loss": 2.3389,
+      "step": 5602
+    },
+    {
+      "epoch": 0.7253778684014629,
+      "grad_norm": 1.760524997475405,
+      "learning_rate": 2.658401475096191e-07,
+      "loss": 2.3555,
+      "step": 5603
+    },
+    {
+      "epoch": 0.7255073308088164,
+      "grad_norm": 1.6595253890090977,
+      "learning_rate": 2.6560900663207213e-07,
+      "loss": 2.5142,
+      "step": 5604
+    },
+    {
+      "epoch": 0.7256367932161698,
+      "grad_norm": 1.8617391917069044,
+      "learning_rate": 2.6537793916741987e-07,
+      "loss": 2.4392,
+      "step": 5605
+    },
+    {
+      "epoch": 0.7256367932161698,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.5079900026321411,
+      "eval_runtime": 14.0275,
+      "eval_samples_per_second": 3.137,
+      "eval_steps_per_second": 0.214,
+      "step": 5605
+    },
+    {
+      "epoch": 0.7257662556235234,
+      "grad_norm": 1.4913957516562677,
+      "learning_rate": 2.6514694516286294e-07,
+      "loss": 2.2512,
+      "step": 5606
+    },
+    {
+      "epoch": 0.7258957180308768,
+      "grad_norm": 2.006513836069113,
+      "learning_rate": 2.6491602466558636e-07,
+      "loss": 2.2847,
+      "step": 5607
+    },
+    {
+      "epoch": 0.7260251804382303,
+      "grad_norm": 1.629313339312131,
+      "learning_rate": 2.6468517772276014e-07,
+      "loss": 2.4475,
+      "step": 5608
+    },
+    {
+      "epoch": 0.7261546428455837,
+      "grad_norm": 1.5139261830156856,
+      "learning_rate": 2.6445440438153975e-07,
+      "loss": 2.4216,
+      "step": 5609
+    },
+    {
+      "epoch": 0.7262841052529372,
+      "grad_norm": 1.5199421615926034,
+      "learning_rate": 2.64223704689065e-07,
+      "loss": 2.3335,
+      "step": 5610
+    },
+    {
+      "epoch": 0.7262841052529372,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.509321689605713,
+      "eval_runtime": 14.6745,
+      "eval_samples_per_second": 2.998,
+      "eval_steps_per_second": 0.204,
+      "step": 5610
+    },
+    {
+      "epoch": 0.7264135676602906,
+      "grad_norm": 1.7194164107061336,
+      "learning_rate": 2.639930786924612e-07,
+      "loss": 2.3953,
+      "step": 5611
+    },
+    {
+      "epoch": 0.7265430300676441,
+      "grad_norm": 1.9642260942677225,
+      "learning_rate": 2.637625264388388e-07,
+      "loss": 2.4902,
+      "step": 5612
+    },
+    {
+      "epoch": 0.7266724924749975,
+      "grad_norm": 2.1264741538545775,
+      "learning_rate": 2.635320479752919e-07,
+      "loss": 2.5557,
+      "step": 5613
+    },
+    {
+      "epoch": 0.726801954882351,
+      "grad_norm": 2.5791993356191707,
+      "learning_rate": 2.63301643348901e-07,
+      "loss": 2.5876,
+      "step": 5614
+    },
+    {
+      "epoch": 0.7269314172897045,
+      "grad_norm": 2.5041629726106733,
+      "learning_rate": 2.6307131260673053e-07,
+      "loss": 2.5605,
+      "step": 5615
+    },
+    {
+      "epoch": 0.7269314172897045,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.5086559057235718,
+      "eval_runtime": 14.8714,
+      "eval_samples_per_second": 2.959,
+      "eval_steps_per_second": 0.202,
+      "step": 5615
+    },
+    {
+      "epoch": 0.727060879697058,
+      "grad_norm": 1.2063006392780222,
+      "learning_rate": 2.628410557958304e-07,
+      "loss": 2.2524,
+      "step": 5616
+    },
+    {
+      "epoch": 0.7271903421044115,
+      "grad_norm": 2.8536342445032203,
+      "learning_rate": 2.626108729632355e-07,
+      "loss": 2.2452,
+      "step": 5617
+    },
+    {
+      "epoch": 0.7273198045117649,
+      "grad_norm": 1.367484206121569,
+      "learning_rate": 2.623807641559651e-07,
+      "loss": 2.4668,
+      "step": 5618
+    },
+    {
+      "epoch": 0.7274492669191184,
+      "grad_norm": 2.4325491839380176,
+      "learning_rate": 2.6215072942102334e-07,
+      "loss": 2.6729,
+      "step": 5619
+    },
+    {
+      "epoch": 0.7275787293264718,
+      "grad_norm": 1.9967059733797172,
+      "learning_rate": 2.6192076880539994e-07,
+      "loss": 2.4453,
+      "step": 5620
+    },
+    {
+      "epoch": 0.7275787293264718,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.505859375,
+      "eval_runtime": 14.5143,
+      "eval_samples_per_second": 3.031,
+      "eval_steps_per_second": 0.207,
+      "step": 5620
+    },
+    {
+      "epoch": 0.7277081917338253,
+      "grad_norm": 2.4584002698532785,
+      "learning_rate": 2.6169088235606854e-07,
+      "loss": 2.4597,
+      "step": 5621
+    },
+    {
+      "epoch": 0.7278376541411787,
+      "grad_norm": 1.9104850311196249,
+      "learning_rate": 2.6146107011998865e-07,
+      "loss": 2.2139,
+      "step": 5622
+    },
+    {
+      "epoch": 0.7279671165485322,
+      "grad_norm": 1.9890462841262417,
+      "learning_rate": 2.6123133214410347e-07,
+      "loss": 2.2876,
+      "step": 5623
+    },
+    {
+      "epoch": 0.7280965789558856,
+      "grad_norm": 1.5827879729141614,
+      "learning_rate": 2.610016684753423e-07,
+      "loss": 2.4524,
+      "step": 5624
+    },
+    {
+      "epoch": 0.7282260413632391,
+      "grad_norm": 1.488852333731658,
+      "learning_rate": 2.607720791606181e-07,
+      "loss": 2.4172,
+      "step": 5625
+    },
+    {
+      "epoch": 0.7282260413632391,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.507279872894287,
+      "eval_runtime": 14.6041,
+      "eval_samples_per_second": 3.013,
+      "eval_steps_per_second": 0.205,
+      "step": 5625
+    },
+    {
+      "epoch": 0.7283555037705927,
+      "grad_norm": 1.7973964371338502,
+      "learning_rate": 2.605425642468292e-07,
+      "loss": 2.657,
+      "step": 5626
+    },
+    {
+      "epoch": 0.7284849661779461,
+      "grad_norm": 2.02635983372142,
+      "learning_rate": 2.603131237808589e-07,
+      "loss": 2.4758,
+      "step": 5627
+    },
+    {
+      "epoch": 0.7286144285852996,
+      "grad_norm": 2.1844991879154714,
+      "learning_rate": 2.6008375780957483e-07,
+      "loss": 2.4307,
+      "step": 5628
+    },
+    {
+      "epoch": 0.728743890992653,
+      "grad_norm": 2.090889483271363,
+      "learning_rate": 2.5985446637982984e-07,
+      "loss": 2.3623,
+      "step": 5629
+    },
+    {
+      "epoch": 0.7288733534000065,
+      "grad_norm": 1.9441238857189067,
+      "learning_rate": 2.596252495384613e-07,
+      "loss": 2.5552,
+      "step": 5630
+    },
+    {
+      "epoch": 0.7288733534000065,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.5057040452957153,
+      "eval_runtime": 14.3437,
+      "eval_samples_per_second": 3.068,
+      "eval_steps_per_second": 0.209,
+      "step": 5630
+    },
+    {
+      "epoch": 0.7290028158073599,
+      "grad_norm": 1.3746058017043443,
+      "learning_rate": 2.5939610733229115e-07,
+      "loss": 2.3572,
+      "step": 5631
+    },
+    {
+      "epoch": 0.7291322782147134,
+      "grad_norm": 1.9874513263203206,
+      "learning_rate": 2.591670398081268e-07,
+      "loss": 2.3345,
+      "step": 5632
+    },
+    {
+      "epoch": 0.7292617406220668,
+      "grad_norm": 1.7802389172942312,
+      "learning_rate": 2.5893804701275936e-07,
+      "loss": 2.3894,
+      "step": 5633
+    },
+    {
+      "epoch": 0.7293912030294203,
+      "grad_norm": 1.6169981298725404,
+      "learning_rate": 2.5870912899296553e-07,
+      "loss": 2.2141,
+      "step": 5634
+    },
+    {
+      "epoch": 0.7295206654367739,
+      "grad_norm": 1.3814381541868355,
+      "learning_rate": 2.58480285795507e-07,
+      "loss": 2.3362,
+      "step": 5635
+    },
+    {
+      "epoch": 0.7295206654367739,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.505815029144287,
+      "eval_runtime": 13.6086,
+      "eval_samples_per_second": 3.233,
+      "eval_steps_per_second": 0.22,
+      "step": 5635
+    },
+    {
+      "epoch": 0.7296501278441273,
+      "grad_norm": 1.179809472767984,
+      "learning_rate": 2.5825151746712864e-07,
+      "loss": 2.2295,
+      "step": 5636
+    },
+    {
+      "epoch": 0.7297795902514808,
+      "grad_norm": 1.8873093220526136,
+      "learning_rate": 2.5802282405456174e-07,
+      "loss": 2.3521,
+      "step": 5637
+    },
+    {
+      "epoch": 0.7299090526588342,
+      "grad_norm": 1.3724222177448588,
+      "learning_rate": 2.57794205604521e-07,
+      "loss": 2.3384,
+      "step": 5638
+    },
+    {
+      "epoch": 0.7300385150661877,
+      "grad_norm": 1.6424070725179651,
+      "learning_rate": 2.5756566216370673e-07,
+      "loss": 2.438,
+      "step": 5639
+    },
+    {
+      "epoch": 0.7301679774735411,
+      "grad_norm": 1.5747823883073737,
+      "learning_rate": 2.5733719377880405e-07,
+      "loss": 2.4412,
+      "step": 5640
+    },
+    {
+      "epoch": 0.7301679774735411,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.5072354078292847,
+      "eval_runtime": 15.4701,
+      "eval_samples_per_second": 2.844,
+      "eval_steps_per_second": 0.194,
+      "step": 5640
+    },
+    {
+      "epoch": 0.7302974398808946,
+      "grad_norm": 1.35759617611263,
+      "learning_rate": 2.571088004964812e-07,
+      "loss": 2.4189,
+      "step": 5641
+    },
+    {
+      "epoch": 0.730426902288248,
+      "grad_norm": 1.6387510116346904,
+      "learning_rate": 2.5688048236339266e-07,
+      "loss": 2.4739,
+      "step": 5642
+    },
+    {
+      "epoch": 0.7305563646956015,
+      "grad_norm": 2.3020838929599043,
+      "learning_rate": 2.566522394261773e-07,
+      "loss": 2.3151,
+      "step": 5643
+    },
+    {
+      "epoch": 0.7306858271029549,
+      "grad_norm": 1.5935686621172203,
+      "learning_rate": 2.564240717314579e-07,
+      "loss": 2.3442,
+      "step": 5644
+    },
+    {
+      "epoch": 0.7308152895103085,
+      "grad_norm": 1.6830966785197252,
+      "learning_rate": 2.5619597932584267e-07,
+      "loss": 2.3645,
+      "step": 5645
+    },
+    {
+      "epoch": 0.7308152895103085,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.5021750926971436,
+      "eval_runtime": 14.3982,
+      "eval_samples_per_second": 3.056,
+      "eval_steps_per_second": 0.208,
+      "step": 5645
+    },
+    {
+      "epoch": 0.730944751917662,
+      "grad_norm": 1.9970508536071547,
+      "learning_rate": 2.5596796225592407e-07,
+      "loss": 2.4727,
+      "step": 5646
+    },
+    {
+      "epoch": 0.7310742143250154,
+      "grad_norm": 1.9898553690333884,
+      "learning_rate": 2.557400205682789e-07,
+      "loss": 2.4949,
+      "step": 5647
+    },
+    {
+      "epoch": 0.7312036767323689,
+      "grad_norm": 1.9148069220140647,
+      "learning_rate": 2.555121543094693e-07,
+      "loss": 2.313,
+      "step": 5648
+    },
+    {
+      "epoch": 0.7313331391397223,
+      "grad_norm": 1.2900324130894611,
+      "learning_rate": 2.5528436352604107e-07,
+      "loss": 2.3911,
+      "step": 5649
+    },
+    {
+      "epoch": 0.7314626015470758,
+      "grad_norm": 1.284739442695453,
+      "learning_rate": 2.5505664826452566e-07,
+      "loss": 2.1541,
+      "step": 5650
+    },
+    {
+      "epoch": 0.7314626015470758,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.506547451019287,
+      "eval_runtime": 14.1461,
+      "eval_samples_per_second": 3.11,
+      "eval_steps_per_second": 0.212,
+      "step": 5650
+    },
+    {
+      "epoch": 0.7315920639544292,
+      "grad_norm": 1.5182764140197063,
+      "learning_rate": 2.5482900857143825e-07,
+      "loss": 2.3674,
+      "step": 5651
+    },
+    {
+      "epoch": 0.7317215263617827,
+      "grad_norm": 1.4357677924074022,
+      "learning_rate": 2.5460144449327865e-07,
+      "loss": 2.2288,
+      "step": 5652
+    },
+    {
+      "epoch": 0.7318509887691361,
+      "grad_norm": 1.378134534669576,
+      "learning_rate": 2.5437395607653183e-07,
+      "loss": 2.4429,
+      "step": 5653
+    },
+    {
+      "epoch": 0.7319804511764896,
+      "grad_norm": 1.3953457573399068,
+      "learning_rate": 2.5414654336766655e-07,
+      "loss": 2.457,
+      "step": 5654
+    },
+    {
+      "epoch": 0.7321099135838431,
+      "grad_norm": 1.418139383621082,
+      "learning_rate": 2.5391920641313684e-07,
+      "loss": 2.4565,
+      "step": 5655
+    },
+    {
+      "epoch": 0.7321099135838431,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.503662109375,
+      "eval_runtime": 15.0866,
+      "eval_samples_per_second": 2.916,
+      "eval_steps_per_second": 0.199,
+      "step": 5655
+    },
+    {
+      "epoch": 0.7322393759911966,
+      "grad_norm": 1.6689871638675509,
+      "learning_rate": 2.536919452593805e-07,
+      "loss": 2.3201,
+      "step": 5656
+    },
+    {
+      "epoch": 0.73236883839855,
+      "grad_norm": 1.457188260016672,
+      "learning_rate": 2.534647599528206e-07,
+      "loss": 2.5325,
+      "step": 5657
+    },
+    {
+      "epoch": 0.7324983008059035,
+      "grad_norm": 1.09003263851734,
+      "learning_rate": 2.5323765053986413e-07,
+      "loss": 2.1025,
+      "step": 5658
+    },
+    {
+      "epoch": 0.732627763213257,
+      "grad_norm": 1.6983701917924374,
+      "learning_rate": 2.5301061706690266e-07,
+      "loss": 2.3052,
+      "step": 5659
+    },
+    {
+      "epoch": 0.7327572256206104,
+      "grad_norm": 2.0924030152886584,
+      "learning_rate": 2.527836595803127e-07,
+      "loss": 2.584,
+      "step": 5660
+    },
+    {
+      "epoch": 0.7327572256206104,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.5094548463821411,
+      "eval_runtime": 19.5867,
+      "eval_samples_per_second": 2.246,
+      "eval_steps_per_second": 0.153,
+      "step": 5660
+    },
+    {
+      "epoch": 0.7328866880279639,
+      "grad_norm": 1.5504150261635312,
+      "learning_rate": 2.5255677812645455e-07,
+      "loss": 2.1606,
+      "step": 5661
+    },
+    {
+      "epoch": 0.7330161504353173,
+      "grad_norm": 2.4354298586938357,
+      "learning_rate": 2.5232997275167354e-07,
+      "loss": 2.4573,
+      "step": 5662
+    },
+    {
+      "epoch": 0.7331456128426708,
+      "grad_norm": 2.971163976364456,
+      "learning_rate": 2.521032435022998e-07,
+      "loss": 2.2607,
+      "step": 5663
+    },
+    {
+      "epoch": 0.7332750752500242,
+      "grad_norm": 1.9232236768873372,
+      "learning_rate": 2.5187659042464633e-07,
+      "loss": 2.6726,
+      "step": 5664
+    },
+    {
+      "epoch": 0.7334045376573778,
+      "grad_norm": 3.5001323145717405,
+      "learning_rate": 2.516500135650121e-07,
+      "loss": 2.8154,
+      "step": 5665
+    },
+    {
+      "epoch": 0.7334045376573778,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.5008655786514282,
+      "eval_runtime": 14.8822,
+      "eval_samples_per_second": 2.957,
+      "eval_steps_per_second": 0.202,
+      "step": 5665
+    },
+    {
+      "epoch": 0.7335340000647312,
+      "grad_norm": 1.7902291812424176,
+      "learning_rate": 2.514235129696803e-07,
+      "loss": 2.4338,
+      "step": 5666
+    },
+    {
+      "epoch": 0.7336634624720847,
+      "grad_norm": 2.6247102366354045,
+      "learning_rate": 2.5119708868491783e-07,
+      "loss": 2.4045,
+      "step": 5667
+    },
+    {
+      "epoch": 0.7337929248794381,
+      "grad_norm": 2.2188401294436524,
+      "learning_rate": 2.509707407569771e-07,
+      "loss": 2.3306,
+      "step": 5668
+    },
+    {
+      "epoch": 0.7339223872867916,
+      "grad_norm": 1.615794595158225,
+      "learning_rate": 2.507444692320933e-07,
+      "loss": 2.4666,
+      "step": 5669
+    },
+    {
+      "epoch": 0.734051849694145,
+      "grad_norm": 1.9520027225816075,
+      "learning_rate": 2.505182741564875e-07,
+      "loss": 2.5312,
+      "step": 5670
+    },
+    {
+      "epoch": 0.734051849694145,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.5079456567764282,
+      "eval_runtime": 14.6782,
+      "eval_samples_per_second": 2.998,
+      "eval_steps_per_second": 0.204,
+      "step": 5670
+    },
+    {
+      "epoch": 0.7341813121014985,
+      "grad_norm": 1.4703596299837818,
+      "learning_rate": 2.502921555763648e-07,
+      "loss": 2.5527,
+      "step": 5671
+    },
+    {
+      "epoch": 0.734310774508852,
+      "grad_norm": 1.3541719162793477,
+      "learning_rate": 2.5006611353791415e-07,
+      "loss": 2.4666,
+      "step": 5672
+    },
+    {
+      "epoch": 0.7344402369162054,
+      "grad_norm": 2.0854522050832447,
+      "learning_rate": 2.4984014808730954e-07,
+      "loss": 2.354,
+      "step": 5673
+    },
+    {
+      "epoch": 0.7345696993235589,
+      "grad_norm": 2.2035598476389886,
+      "learning_rate": 2.4961425927070883e-07,
+      "loss": 2.5095,
+      "step": 5674
+    },
+    {
+      "epoch": 0.7346991617309124,
+      "grad_norm": 1.2488568471900237,
+      "learning_rate": 2.493884471342542e-07,
+      "loss": 2.3457,
+      "step": 5675
+    },
+    {
+      "epoch": 0.7346991617309124,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.5079678297042847,
+      "eval_runtime": 15.4707,
+      "eval_samples_per_second": 2.844,
+      "eval_steps_per_second": 0.194,
+      "step": 5675
+    },
+    {
+      "epoch": 0.7348286241382659,
+      "grad_norm": 2.1526248325363726,
+      "learning_rate": 2.4916271172407275e-07,
+      "loss": 2.6167,
+      "step": 5676
+    },
+    {
+      "epoch": 0.7349580865456193,
+      "grad_norm": 1.7487231587904244,
+      "learning_rate": 2.48937053086275e-07,
+      "loss": 2.4727,
+      "step": 5677
+    },
+    {
+      "epoch": 0.7350875489529728,
+      "grad_norm": 2.059881484230393,
+      "learning_rate": 2.4871147126695694e-07,
+      "loss": 2.4202,
+      "step": 5678
+    },
+    {
+      "epoch": 0.7352170113603262,
+      "grad_norm": 2.0333970419315985,
+      "learning_rate": 2.4848596631219787e-07,
+      "loss": 2.2969,
+      "step": 5679
+    },
+    {
+      "epoch": 0.7353464737676797,
+      "grad_norm": 1.7041612371808403,
+      "learning_rate": 2.4826053826806154e-07,
+      "loss": 2.5162,
+      "step": 5680
+    },
+    {
+      "epoch": 0.7353464737676797,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.504438877105713,
+      "eval_runtime": 14.6295,
+      "eval_samples_per_second": 3.008,
+      "eval_steps_per_second": 0.205,
+      "step": 5680
+    },
+    {
+      "epoch": 0.7354759361750332,
+      "grad_norm": 2.3093911467602593,
+      "learning_rate": 2.480351871805966e-07,
+      "loss": 2.3008,
+      "step": 5681
+    },
+    {
+      "epoch": 0.7356053985823866,
+      "grad_norm": 2.3891438360485924,
+      "learning_rate": 2.478099130958352e-07,
+      "loss": 2.4646,
+      "step": 5682
+    },
+    {
+      "epoch": 0.7357348609897401,
+      "grad_norm": 2.5342601961660143,
+      "learning_rate": 2.4758471605979466e-07,
+      "loss": 2.5247,
+      "step": 5683
+    },
+    {
+      "epoch": 0.7358643233970936,
+      "grad_norm": 2.686969533358233,
+      "learning_rate": 2.4735959611847543e-07,
+      "loss": 2.593,
+      "step": 5684
+    },
+    {
+      "epoch": 0.7359937858044471,
+      "grad_norm": 2.404409337208305,
+      "learning_rate": 2.471345533178633e-07,
+      "loss": 2.7876,
+      "step": 5685
+    },
+    {
+      "epoch": 0.7359937858044471,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.505815029144287,
+      "eval_runtime": 14.3501,
+      "eval_samples_per_second": 3.066,
+      "eval_steps_per_second": 0.209,
+      "step": 5685
+    },
+    {
+      "epoch": 0.7361232482118005,
+      "grad_norm": 1.8538153353750857,
+      "learning_rate": 2.469095877039277e-07,
+      "loss": 2.1504,
+      "step": 5686
+    },
+    {
+      "epoch": 0.736252710619154,
+      "grad_norm": 3.4703900312985594,
+      "learning_rate": 2.466846993226221e-07,
+      "loss": 2.3835,
+      "step": 5687
+    },
+    {
+      "epoch": 0.7363821730265074,
+      "grad_norm": 1.512334269338688,
+      "learning_rate": 2.4645988821988484e-07,
+      "loss": 2.437,
+      "step": 5688
+    },
+    {
+      "epoch": 0.7365116354338609,
+      "grad_norm": 1.8137655117001923,
+      "learning_rate": 2.462351544416382e-07,
+      "loss": 2.1697,
+      "step": 5689
+    },
+    {
+      "epoch": 0.7366410978412143,
+      "grad_norm": 1.8610060580286094,
+      "learning_rate": 2.4601049803378854e-07,
+      "loss": 2.3376,
+      "step": 5690
+    },
+    {
+      "epoch": 0.7366410978412143,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.5065251588821411,
+      "eval_runtime": 14.5886,
+      "eval_samples_per_second": 3.016,
+      "eval_steps_per_second": 0.206,
+      "step": 5690
+    },
+    {
+      "epoch": 0.7367705602485678,
+      "grad_norm": 3.0868821327146914,
+      "learning_rate": 2.457859190422262e-07,
+      "loss": 2.3054,
+      "step": 5691
+    },
+    {
+      "epoch": 0.7369000226559212,
+      "grad_norm": 2.446197983909844,
+      "learning_rate": 2.455614175128264e-07,
+      "loss": 2.53,
+      "step": 5692
+    },
+    {
+      "epoch": 0.7370294850632747,
+      "grad_norm": 1.5213681017230851,
+      "learning_rate": 2.453369934914477e-07,
+      "loss": 2.1814,
+      "step": 5693
+    },
+    {
+      "epoch": 0.7371589474706283,
+      "grad_norm": 2.303615540528132,
+      "learning_rate": 2.451126470239338e-07,
+      "loss": 2.4673,
+      "step": 5694
+    },
+    {
+      "epoch": 0.7372884098779817,
+      "grad_norm": 1.4862575777019944,
+      "learning_rate": 2.448883781561114e-07,
+      "loss": 2.3198,
+      "step": 5695
+    },
+    {
+      "epoch": 0.7372884098779817,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.5035732984542847,
+      "eval_runtime": 13.9781,
+      "eval_samples_per_second": 3.148,
+      "eval_steps_per_second": 0.215,
+      "step": 5695
+    },
+    {
+      "epoch": 0.7374178722853352,
+      "grad_norm": 2.4251094151406334,
+      "learning_rate": 2.4466418693379253e-07,
+      "loss": 2.2124,
+      "step": 5696
+    },
+    {
+      "epoch": 0.7375473346926886,
+      "grad_norm": 2.588251885086974,
+      "learning_rate": 2.4444007340277257e-07,
+      "loss": 2.6187,
+      "step": 5697
+    },
+    {
+      "epoch": 0.7376767971000421,
+      "grad_norm": 1.6953634322425963,
+      "learning_rate": 2.4421603760883095e-07,
+      "loss": 2.5024,
+      "step": 5698
+    },
+    {
+      "epoch": 0.7378062595073955,
+      "grad_norm": 2.390150542524861,
+      "learning_rate": 2.4399207959773207e-07,
+      "loss": 2.2717,
+      "step": 5699
+    },
+    {
+      "epoch": 0.737935721914749,
+      "grad_norm": 3.1099283204917243,
+      "learning_rate": 2.4376819941522333e-07,
+      "loss": 2.6899,
+      "step": 5700
+    },
+    {
+      "epoch": 0.737935721914749,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.5071688890457153,
+      "eval_runtime": 13.9585,
+      "eval_samples_per_second": 3.152,
+      "eval_steps_per_second": 0.215,
+      "step": 5700
+    },
+    {
+      "epoch": 0.7380651843221024,
+      "grad_norm": 1.636977704177955,
+      "learning_rate": 2.435443971070373e-07,
+      "loss": 2.2634,
+      "step": 5701
+    },
+    {
+      "epoch": 0.7381946467294559,
+      "grad_norm": 3.7066921817979797,
+      "learning_rate": 2.433206727188899e-07,
+      "loss": 2.7783,
+      "step": 5702
+    },
+    {
+      "epoch": 0.7383241091368093,
+      "grad_norm": 1.200859903233997,
+      "learning_rate": 2.4309702629648124e-07,
+      "loss": 2.2935,
+      "step": 5703
+    },
+    {
+      "epoch": 0.7384535715441629,
+      "grad_norm": 1.3806974133034717,
+      "learning_rate": 2.428734578854959e-07,
+      "loss": 2.3254,
+      "step": 5704
+    },
+    {
+      "epoch": 0.7385830339515164,
+      "grad_norm": 1.4230174637008688,
+      "learning_rate": 2.42649967531602e-07,
+      "loss": 2.3352,
+      "step": 5705
+    },
+    {
+      "epoch": 0.7385830339515164,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.5049716234207153,
+      "eval_runtime": 14.5211,
+      "eval_samples_per_second": 3.03,
+      "eval_steps_per_second": 0.207,
+      "step": 5705
+    },
+    {
+      "epoch": 0.7387124963588698,
+      "grad_norm": 1.3485431936495211,
+      "learning_rate": 2.424265552804521e-07,
+      "loss": 2.5105,
+      "step": 5706
+    },
+    {
+      "epoch": 0.7388419587662233,
+      "grad_norm": 1.250070075372513,
+      "learning_rate": 2.4220322117768313e-07,
+      "loss": 2.2883,
+      "step": 5707
+    },
+    {
+      "epoch": 0.7389714211735767,
+      "grad_norm": 1.835187632190345,
+      "learning_rate": 2.419799652689147e-07,
+      "loss": 2.5251,
+      "step": 5708
+    },
+    {
+      "epoch": 0.7391008835809302,
+      "grad_norm": 2.011807836136752,
+      "learning_rate": 2.417567875997521e-07,
+      "loss": 2.46,
+      "step": 5709
+    },
+    {
+      "epoch": 0.7392303459882836,
+      "grad_norm": 2.07268958026009,
+      "learning_rate": 2.415336882157833e-07,
+      "loss": 2.46,
+      "step": 5710
+    },
+    {
+      "epoch": 0.7392303459882836,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.5065251588821411,
+      "eval_runtime": 14.6732,
+      "eval_samples_per_second": 2.999,
+      "eval_steps_per_second": 0.204,
+      "step": 5710
+    },
+    {
+      "epoch": 0.7393598083956371,
+      "grad_norm": 1.294458580347615,
+      "learning_rate": 2.413106671625813e-07,
+      "loss": 2.4053,
+      "step": 5711
+    },
+    {
+      "epoch": 0.7394892708029905,
+      "grad_norm": 2.7758333889794105,
+      "learning_rate": 2.410877244857027e-07,
+      "loss": 2.3513,
+      "step": 5712
+    },
+    {
+      "epoch": 0.739618733210344,
+      "grad_norm": 1.6304735267055515,
+      "learning_rate": 2.4086486023068794e-07,
+      "loss": 2.2817,
+      "step": 5713
+    },
+    {
+      "epoch": 0.7397481956176976,
+      "grad_norm": 1.6245164929996876,
+      "learning_rate": 2.406420744430614e-07,
+      "loss": 2.343,
+      "step": 5714
+    },
+    {
+      "epoch": 0.739877658025051,
+      "grad_norm": 1.8144417692251245,
+      "learning_rate": 2.404193671683319e-07,
+      "loss": 2.3185,
+      "step": 5715
+    },
+    {
+      "epoch": 0.739877658025051,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.5079013109207153,
+      "eval_runtime": 15.9937,
+      "eval_samples_per_second": 2.751,
+      "eval_steps_per_second": 0.188,
+      "step": 5715
+    },
+    {
+      "epoch": 0.7400071204324045,
+      "grad_norm": 1.5295296340283542,
+      "learning_rate": 2.4019673845199146e-07,
+      "loss": 2.4756,
+      "step": 5716
+    },
+    {
+      "epoch": 0.7401365828397579,
+      "grad_norm": 2.1146407143179546,
+      "learning_rate": 2.3997418833951706e-07,
+      "loss": 2.3105,
+      "step": 5717
+    },
+    {
+      "epoch": 0.7402660452471114,
+      "grad_norm": 1.3504357321603342,
+      "learning_rate": 2.397517168763689e-07,
+      "loss": 2.2764,
+      "step": 5718
+    },
+    {
+      "epoch": 0.7403955076544648,
+      "grad_norm": 1.6580937796607456,
+      "learning_rate": 2.395293241079909e-07,
+      "loss": 2.377,
+      "step": 5719
+    },
+    {
+      "epoch": 0.7405249700618183,
+      "grad_norm": 1.832292336307156,
+      "learning_rate": 2.393070100798118e-07,
+      "loss": 2.2739,
+      "step": 5720
+    },
+    {
+      "epoch": 0.7405249700618183,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.5051047801971436,
+      "eval_runtime": 14.2528,
+      "eval_samples_per_second": 3.087,
+      "eval_steps_per_second": 0.21,
+      "step": 5720
+    },
+    {
+      "epoch": 0.7406544324691717,
+      "grad_norm": 1.1558103016170562,
+      "learning_rate": 2.390847748372433e-07,
+      "loss": 2.2605,
+      "step": 5721
+    },
+    {
+      "epoch": 0.7407838948765252,
+      "grad_norm": 2.2592009863859515,
+      "learning_rate": 2.3886261842568194e-07,
+      "loss": 2.5537,
+      "step": 5722
+    },
+    {
+      "epoch": 0.7409133572838787,
+      "grad_norm": 1.1882650440240674,
+      "learning_rate": 2.386405408905072e-07,
+      "loss": 2.3228,
+      "step": 5723
+    },
+    {
+      "epoch": 0.7410428196912322,
+      "grad_norm": 3.134434526580684,
+      "learning_rate": 2.3841854227708337e-07,
+      "loss": 2.6421,
+      "step": 5724
+    },
+    {
+      "epoch": 0.7411722820985857,
+      "grad_norm": 2.5262998603920255,
+      "learning_rate": 2.3819662263075783e-07,
+      "loss": 2.4707,
+      "step": 5725
+    },
+    {
+      "epoch": 0.7411722820985857,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.5014870166778564,
+      "eval_runtime": 14.3866,
+      "eval_samples_per_second": 3.058,
+      "eval_steps_per_second": 0.209,
+      "step": 5725
+    },
+    {
+      "epoch": 0.7413017445059391,
+      "grad_norm": 1.9346772369502783,
+      "learning_rate": 2.379747819968621e-07,
+      "loss": 2.3423,
+      "step": 5726
+    },
+    {
+      "epoch": 0.7414312069132926,
+      "grad_norm": 1.5656660722330134,
+      "learning_rate": 2.3775302042071207e-07,
+      "loss": 2.4119,
+      "step": 5727
+    },
+    {
+      "epoch": 0.741560669320646,
+      "grad_norm": 2.888111920311494,
+      "learning_rate": 2.375313379476065e-07,
+      "loss": 2.3584,
+      "step": 5728
+    },
+    {
+      "epoch": 0.7416901317279995,
+      "grad_norm": 2.0321396189871703,
+      "learning_rate": 2.3730973462282898e-07,
+      "loss": 2.7117,
+      "step": 5729
+    },
+    {
+      "epoch": 0.7418195941353529,
+      "grad_norm": 2.09991671564426,
+      "learning_rate": 2.370882104916463e-07,
+      "loss": 2.2905,
+      "step": 5730
+    },
+    {
+      "epoch": 0.7418195941353529,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.5042835474014282,
+      "eval_runtime": 14.1895,
+      "eval_samples_per_second": 3.101,
+      "eval_steps_per_second": 0.211,
+      "step": 5730
+    },
+    {
+      "epoch": 0.7419490565427064,
+      "grad_norm": 1.149303201344051,
+      "learning_rate": 2.36866765599309e-07,
+      "loss": 2.3013,
+      "step": 5731
+    },
+    {
+      "epoch": 0.7420785189500598,
+      "grad_norm": 1.4679625395047817,
+      "learning_rate": 2.3664539999105216e-07,
+      "loss": 2.1853,
+      "step": 5732
+    },
+    {
+      "epoch": 0.7422079813574134,
+      "grad_norm": 2.24926289680569,
+      "learning_rate": 2.364241137120938e-07,
+      "loss": 2.3717,
+      "step": 5733
+    },
+    {
+      "epoch": 0.7423374437647668,
+      "grad_norm": 1.9690265007605825,
+      "learning_rate": 2.3620290680763624e-07,
+      "loss": 2.5654,
+      "step": 5734
+    },
+    {
+      "epoch": 0.7424669061721203,
+      "grad_norm": 2.437057319726628,
+      "learning_rate": 2.3598177932286595e-07,
+      "loss": 2.3335,
+      "step": 5735
+    },
+    {
+      "epoch": 0.7424669061721203,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.502974033355713,
+      "eval_runtime": 15.1362,
+      "eval_samples_per_second": 2.907,
+      "eval_steps_per_second": 0.198,
+      "step": 5735
+    },
+    {
+      "epoch": 0.7425963685794738,
+      "grad_norm": 2.2327673280717155,
+      "learning_rate": 2.3576073130295184e-07,
+      "loss": 2.4385,
+      "step": 5736
+    },
+    {
+      "epoch": 0.7427258309868272,
+      "grad_norm": 1.9250282974682684,
+      "learning_rate": 2.3553976279304787e-07,
+      "loss": 2.4343,
+      "step": 5737
+    },
+    {
+      "epoch": 0.7428552933941807,
+      "grad_norm": 1.1809759202796595,
+      "learning_rate": 2.353188738382916e-07,
+      "loss": 2.3423,
+      "step": 5738
+    },
+    {
+      "epoch": 0.7429847558015341,
+      "grad_norm": 1.6868649266849252,
+      "learning_rate": 2.3509806448380357e-07,
+      "loss": 2.4597,
+      "step": 5739
+    },
+    {
+      "epoch": 0.7431142182088876,
+      "grad_norm": 1.9366055483828428,
+      "learning_rate": 2.348773347746893e-07,
+      "loss": 2.2764,
+      "step": 5740
+    },
+    {
+      "epoch": 0.7431142182088876,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.505859375,
+      "eval_runtime": 14.3837,
+      "eval_samples_per_second": 3.059,
+      "eval_steps_per_second": 0.209,
+      "step": 5740
+    },
+    {
+      "epoch": 0.743243680616241,
+      "grad_norm": 2.4785060988254806,
+      "learning_rate": 2.3465668475603636e-07,
+      "loss": 2.3579,
+      "step": 5741
+    },
+    {
+      "epoch": 0.7433731430235945,
+      "grad_norm": 2.721171272482728,
+      "learning_rate": 2.344361144729175e-07,
+      "loss": 2.6748,
+      "step": 5742
+    },
+    {
+      "epoch": 0.743502605430948,
+      "grad_norm": 1.8367135148130516,
+      "learning_rate": 2.3421562397038884e-07,
+      "loss": 2.2817,
+      "step": 5743
+    },
+    {
+      "epoch": 0.7436320678383015,
+      "grad_norm": 2.9016193425260175,
+      "learning_rate": 2.3399521329348953e-07,
+      "loss": 2.5791,
+      "step": 5744
+    },
+    {
+      "epoch": 0.743761530245655,
+      "grad_norm": 1.5504410210589437,
+      "learning_rate": 2.337748824872435e-07,
+      "loss": 2.3179,
+      "step": 5745
+    },
+    {
+      "epoch": 0.743761530245655,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.5057483911514282,
+      "eval_runtime": 14.0921,
+      "eval_samples_per_second": 3.122,
+      "eval_steps_per_second": 0.213,
+      "step": 5745
+    },
+    {
+      "epoch": 0.7438909926530084,
+      "grad_norm": 1.885305214076558,
+      "learning_rate": 2.3355463159665749e-07,
+      "loss": 2.49,
+      "step": 5746
+    },
+    {
+      "epoch": 0.7440204550603619,
+      "grad_norm": 1.6665165486333617,
+      "learning_rate": 2.33334460666722e-07,
+      "loss": 2.5271,
+      "step": 5747
+    },
+    {
+      "epoch": 0.7441499174677153,
+      "grad_norm": 1.7026733029610266,
+      "learning_rate": 2.331143697424118e-07,
+      "loss": 2.3281,
+      "step": 5748
+    },
+    {
+      "epoch": 0.7442793798750688,
+      "grad_norm": 1.3933721389200384,
+      "learning_rate": 2.3289435886868453e-07,
+      "loss": 2.335,
+      "step": 5749
+    },
+    {
+      "epoch": 0.7444088422824222,
+      "grad_norm": 1.116011354477912,
+      "learning_rate": 2.3267442809048234e-07,
+      "loss": 2.2881,
+      "step": 5750
+    },
+    {
+      "epoch": 0.7444088422824222,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.5023082494735718,
+      "eval_runtime": 17.0839,
+      "eval_samples_per_second": 2.576,
+      "eval_steps_per_second": 0.176,
+      "step": 5750
+    },
+    {
+      "epoch": 0.7445383046897757,
+      "grad_norm": 1.4148398740897135,
+      "learning_rate": 2.3245457745273003e-07,
+      "loss": 2.3015,
+      "step": 5751
+    },
+    {
+      "epoch": 0.7446677670971291,
+      "grad_norm": 2.117948750725282,
+      "learning_rate": 2.3223480700033696e-07,
+      "loss": 2.2646,
+      "step": 5752
+    },
+    {
+      "epoch": 0.7447972295044827,
+      "grad_norm": 1.4051410784765883,
+      "learning_rate": 2.3201511677819567e-07,
+      "loss": 2.2461,
+      "step": 5753
+    },
+    {
+      "epoch": 0.7449266919118361,
+      "grad_norm": 2.6867551357303148,
+      "learning_rate": 2.3179550683118197e-07,
+      "loss": 2.5649,
+      "step": 5754
+    },
+    {
+      "epoch": 0.7450561543191896,
+      "grad_norm": 1.7292971611742591,
+      "learning_rate": 2.3157597720415607e-07,
+      "loss": 2.198,
+      "step": 5755
+    },
+    {
+      "epoch": 0.7450561543191896,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.5043723583221436,
+      "eval_runtime": 19.067,
+      "eval_samples_per_second": 2.308,
+      "eval_steps_per_second": 0.157,
+      "step": 5755
+    },
+    {
+      "epoch": 0.745185616726543,
+      "grad_norm": 2.147978560830328,
+      "learning_rate": 2.31356527941961e-07,
+      "loss": 2.4733,
+      "step": 5756
+    },
+    {
+      "epoch": 0.7453150791338965,
+      "grad_norm": 2.022867481141554,
+      "learning_rate": 2.3113715908942403e-07,
+      "loss": 2.3513,
+      "step": 5757
+    },
+    {
+      "epoch": 0.74544454154125,
+      "grad_norm": 5.079058456847697,
+      "learning_rate": 2.3091787069135563e-07,
+      "loss": 2.824,
+      "step": 5758
+    },
+    {
+      "epoch": 0.7455740039486034,
+      "grad_norm": 1.6758215541070285,
+      "learning_rate": 2.3069866279254955e-07,
+      "loss": 2.4097,
+      "step": 5759
+    },
+    {
+      "epoch": 0.7457034663559569,
+      "grad_norm": 1.7644838942104952,
+      "learning_rate": 2.3047953543778375e-07,
+      "loss": 2.3442,
+      "step": 5760
+    },
+    {
+      "epoch": 0.7457034663559569,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.4994229078292847,
+      "eval_runtime": 13.7953,
+      "eval_samples_per_second": 3.189,
+      "eval_steps_per_second": 0.217,
+      "step": 5760
+    },
+    {
+      "epoch": 0.7458329287633103,
+      "grad_norm": 1.6870904459979024,
+      "learning_rate": 2.302604886718196e-07,
+      "loss": 2.374,
+      "step": 5761
+    },
+    {
+      "epoch": 0.7459623911706638,
+      "grad_norm": 1.1533928909331745,
+      "learning_rate": 2.300415225394014e-07,
+      "loss": 2.3413,
+      "step": 5762
+    },
+    {
+      "epoch": 0.7460918535780173,
+      "grad_norm": 3.1282882856696035,
+      "learning_rate": 2.298226370852582e-07,
+      "loss": 2.4636,
+      "step": 5763
+    },
+    {
+      "epoch": 0.7462213159853708,
+      "grad_norm": 1.563479304384251,
+      "learning_rate": 2.2960383235410077e-07,
+      "loss": 2.2686,
+      "step": 5764
+    },
+    {
+      "epoch": 0.7463507783927242,
+      "grad_norm": 1.5971294582503306,
+      "learning_rate": 2.2938510839062484e-07,
+      "loss": 2.4404,
+      "step": 5765
+    },
+    {
+      "epoch": 0.7463507783927242,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.5030628442764282,
+      "eval_runtime": 14.8761,
+      "eval_samples_per_second": 2.958,
+      "eval_steps_per_second": 0.202,
+      "step": 5765
+    },
+    {
+      "epoch": 0.7464802408000777,
+      "grad_norm": 3.6200775909620138,
+      "learning_rate": 2.2916646523950952e-07,
+      "loss": 2.6165,
+      "step": 5766
+    },
+    {
+      "epoch": 0.7466097032074311,
+      "grad_norm": 1.502943780133985,
+      "learning_rate": 2.2894790294541668e-07,
+      "loss": 2.3403,
+      "step": 5767
+    },
+    {
+      "epoch": 0.7467391656147846,
+      "grad_norm": 1.9252283045414476,
+      "learning_rate": 2.2872942155299248e-07,
+      "loss": 2.1272,
+      "step": 5768
+    },
+    {
+      "epoch": 0.746868628022138,
+      "grad_norm": 1.8357036001414795,
+      "learning_rate": 2.285110211068659e-07,
+      "loss": 2.2083,
+      "step": 5769
+    },
+    {
+      "epoch": 0.7469980904294915,
+      "grad_norm": 1.402098070812646,
+      "learning_rate": 2.282927016516496e-07,
+      "loss": 2.4163,
+      "step": 5770
+    },
+    {
+      "epoch": 0.7469980904294915,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.501420497894287,
+      "eval_runtime": 15.5172,
+      "eval_samples_per_second": 2.836,
+      "eval_steps_per_second": 0.193,
+      "step": 5770
+    },
+    {
+      "epoch": 0.747127552836845,
+      "grad_norm": 1.874322282527136,
+      "learning_rate": 2.2807446323194003e-07,
+      "loss": 2.4714,
+      "step": 5771
+    },
+    {
+      "epoch": 0.7472570152441985,
+      "grad_norm": 1.714467438649989,
+      "learning_rate": 2.2785630589231646e-07,
+      "loss": 2.2888,
+      "step": 5772
+    },
+    {
+      "epoch": 0.747386477651552,
+      "grad_norm": 1.3200850637029224,
+      "learning_rate": 2.276382296773424e-07,
+      "loss": 2.3416,
+      "step": 5773
+    },
+    {
+      "epoch": 0.7475159400589054,
+      "grad_norm": 2.815673748855942,
+      "learning_rate": 2.2742023463156405e-07,
+      "loss": 2.4072,
+      "step": 5774
+    },
+    {
+      "epoch": 0.7476454024662589,
+      "grad_norm": 1.3913837637007695,
+      "learning_rate": 2.272023207995111e-07,
+      "loss": 2.3845,
+      "step": 5775
+    },
+    {
+      "epoch": 0.7476454024662589,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.5022194385528564,
+      "eval_runtime": 15.2171,
+      "eval_samples_per_second": 2.891,
+      "eval_steps_per_second": 0.197,
+      "step": 5775
+    },
+    {
+      "epoch": 0.7477748648736123,
+      "grad_norm": 2.1597954337338123,
+      "learning_rate": 2.269844882256974e-07,
+      "loss": 2.8374,
+      "step": 5776
+    },
+    {
+      "epoch": 0.7479043272809658,
+      "grad_norm": 1.5402660364256493,
+      "learning_rate": 2.2676673695461913e-07,
+      "loss": 2.4219,
+      "step": 5777
+    },
+    {
+      "epoch": 0.7480337896883192,
+      "grad_norm": 1.8661053171492508,
+      "learning_rate": 2.2654906703075664e-07,
+      "loss": 2.7046,
+      "step": 5778
+    },
+    {
+      "epoch": 0.7481632520956727,
+      "grad_norm": 1.645395126294986,
+      "learning_rate": 2.2633147849857395e-07,
+      "loss": 2.2996,
+      "step": 5779
+    },
+    {
+      "epoch": 0.7482927145030261,
+      "grad_norm": 2.2561182987384667,
+      "learning_rate": 2.2611397140251692e-07,
+      "loss": 2.4625,
+      "step": 5780
+    },
+    {
+      "epoch": 0.7482927145030261,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.4979802370071411,
+      "eval_runtime": 13.7667,
+      "eval_samples_per_second": 3.196,
+      "eval_steps_per_second": 0.218,
+      "step": 5780
+    },
+    {
+      "epoch": 0.7484221769103796,
+      "grad_norm": 1.3371922326166457,
+      "learning_rate": 2.2589654578701653e-07,
+      "loss": 2.2361,
+      "step": 5781
+    },
+    {
+      "epoch": 0.7485516393177332,
+      "grad_norm": 1.9407566309567759,
+      "learning_rate": 2.2567920169648587e-07,
+      "loss": 2.3838,
+      "step": 5782
+    },
+    {
+      "epoch": 0.7486811017250866,
+      "grad_norm": 1.83692195687607,
+      "learning_rate": 2.254619391753222e-07,
+      "loss": 2.4863,
+      "step": 5783
+    },
+    {
+      "epoch": 0.7488105641324401,
+      "grad_norm": 4.340319107503327,
+      "learning_rate": 2.2524475826790585e-07,
+      "loss": 2.6577,
+      "step": 5784
+    },
+    {
+      "epoch": 0.7489400265397935,
+      "grad_norm": 1.498664464152015,
+      "learning_rate": 2.2502765901860034e-07,
+      "loss": 2.3208,
+      "step": 5785
+    },
+    {
+      "epoch": 0.7489400265397935,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.5020862817764282,
+      "eval_runtime": 13.4527,
+      "eval_samples_per_second": 3.271,
+      "eval_steps_per_second": 0.223,
+      "step": 5785
+    },
+    {
+      "epoch": 0.749069488947147,
+      "grad_norm": 1.669561872005248,
+      "learning_rate": 2.2481064147175234e-07,
+      "loss": 2.3477,
+      "step": 5786
+    },
+    {
+      "epoch": 0.7491989513545004,
+      "grad_norm": 2.7942096472646614,
+      "learning_rate": 2.2459370567169244e-07,
+      "loss": 2.573,
+      "step": 5787
+    },
+    {
+      "epoch": 0.7493284137618539,
+      "grad_norm": 1.8587148438346652,
+      "learning_rate": 2.2437685166273384e-07,
+      "loss": 2.3375,
+      "step": 5788
+    },
+    {
+      "epoch": 0.7494578761692073,
+      "grad_norm": 1.4509462729135019,
+      "learning_rate": 2.241600794891738e-07,
+      "loss": 2.343,
+      "step": 5789
+    },
+    {
+      "epoch": 0.7495873385765608,
+      "grad_norm": 1.6681557174623465,
+      "learning_rate": 2.2394338919529199e-07,
+      "loss": 2.3665,
+      "step": 5790
+    },
+    {
+      "epoch": 0.7495873385765608,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.5028630495071411,
+      "eval_runtime": 14.0959,
+      "eval_samples_per_second": 3.121,
+      "eval_steps_per_second": 0.213,
+      "step": 5790
+    },
+    {
+      "epoch": 0.7497168009839142,
+      "grad_norm": 2.1816736864515254,
+      "learning_rate": 2.237267808253521e-07,
+      "loss": 2.3633,
+      "step": 5791
+    },
+    {
+      "epoch": 0.7498462633912678,
+      "grad_norm": 1.6088210292697507,
+      "learning_rate": 2.2351025442360072e-07,
+      "loss": 2.2981,
+      "step": 5792
+    },
+    {
+      "epoch": 0.7499757257986213,
+      "grad_norm": 2.842564527386317,
+      "learning_rate": 2.232938100342675e-07,
+      "loss": 2.5859,
+      "step": 5793
+    },
+    {
+      "epoch": 0.7501051882059747,
+      "grad_norm": 1.6004002835668634,
+      "learning_rate": 2.2307744770156598e-07,
+      "loss": 2.2358,
+      "step": 5794
+    },
+    {
+      "epoch": 0.7502346506133282,
+      "grad_norm": 2.332282601343465,
+      "learning_rate": 2.2286116746969218e-07,
+      "loss": 2.4663,
+      "step": 5795
+    },
+    {
+      "epoch": 0.7502346506133282,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.4994007349014282,
+      "eval_runtime": 13.8871,
+      "eval_samples_per_second": 3.168,
+      "eval_steps_per_second": 0.216,
+      "step": 5795
+    },
+    {
+      "epoch": 0.7503641130206816,
+      "grad_norm": 1.563404749958465,
+      "learning_rate": 2.2264496938282608e-07,
+      "loss": 2.2307,
+      "step": 5796
+    },
+    {
+      "epoch": 0.7504935754280351,
+      "grad_norm": 1.7218288638978925,
+      "learning_rate": 2.2242885348513037e-07,
+      "loss": 2.376,
+      "step": 5797
+    },
+    {
+      "epoch": 0.7506230378353885,
+      "grad_norm": 4.1520240968988364,
+      "learning_rate": 2.2221281982075085e-07,
+      "loss": 2.6885,
+      "step": 5798
+    },
+    {
+      "epoch": 0.750752500242742,
+      "grad_norm": 1.6439855781455703,
+      "learning_rate": 2.2199686843381724e-07,
+      "loss": 2.3845,
+      "step": 5799
+    },
+    {
+      "epoch": 0.7508819626500954,
+      "grad_norm": 1.2343710915623454,
+      "learning_rate": 2.2178099936844153e-07,
+      "loss": 2.3928,
+      "step": 5800
+    },
+    {
+      "epoch": 0.7508819626500954,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.499467372894287,
+      "eval_runtime": 13.9521,
+      "eval_samples_per_second": 3.154,
+      "eval_steps_per_second": 0.215,
+      "step": 5800
+    },
+    {
+      "epoch": 0.7510114250574489,
+      "grad_norm": 2.4198345859304196,
+      "learning_rate": 2.2156521266871963e-07,
+      "loss": 2.4072,
+      "step": 5801
+    },
+    {
+      "epoch": 0.7511408874648025,
+      "grad_norm": 1.5102624631034194,
+      "learning_rate": 2.2134950837873066e-07,
+      "loss": 2.4109,
+      "step": 5802
+    },
+    {
+      "epoch": 0.7512703498721559,
+      "grad_norm": 1.4704684748471535,
+      "learning_rate": 2.211338865425358e-07,
+      "loss": 2.354,
+      "step": 5803
+    },
+    {
+      "epoch": 0.7513998122795094,
+      "grad_norm": 1.3634893051480144,
+      "learning_rate": 2.2091834720418085e-07,
+      "loss": 2.4023,
+      "step": 5804
+    },
+    {
+      "epoch": 0.7515292746868628,
+      "grad_norm": 1.7659414099753081,
+      "learning_rate": 2.2070289040769362e-07,
+      "loss": 2.4258,
+      "step": 5805
+    },
+    {
+      "epoch": 0.7515292746868628,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.5008434057235718,
+      "eval_runtime": 13.7516,
+      "eval_samples_per_second": 3.2,
+      "eval_steps_per_second": 0.218,
+      "step": 5805
+    },
+    {
+      "epoch": 0.7516587370942163,
+      "grad_norm": 1.8938426151378294,
+      "learning_rate": 2.204875161970857e-07,
+      "loss": 2.532,
+      "step": 5806
+    },
+    {
+      "epoch": 0.7517881995015697,
+      "grad_norm": 2.3418997079422073,
+      "learning_rate": 2.2027222461635217e-07,
+      "loss": 2.3567,
+      "step": 5807
+    },
+    {
+      "epoch": 0.7519176619089232,
+      "grad_norm": 1.5563997643011263,
+      "learning_rate": 2.2005701570946982e-07,
+      "loss": 2.314,
+      "step": 5808
+    },
+    {
+      "epoch": 0.7520471243162766,
+      "grad_norm": 1.7858463301155232,
+      "learning_rate": 2.1984188952039977e-07,
+      "loss": 2.5083,
+      "step": 5809
+    },
+    {
+      "epoch": 0.7521765867236301,
+      "grad_norm": 1.1839448817651672,
+      "learning_rate": 2.1962684609308623e-07,
+      "loss": 2.2014,
+      "step": 5810
+    },
+    {
+      "epoch": 0.7521765867236301,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.5029963254928589,
+      "eval_runtime": 14.3925,
+      "eval_samples_per_second": 3.057,
+      "eval_steps_per_second": 0.208,
+      "step": 5810
+    },
+    {
+      "epoch": 0.7523060491309836,
+      "grad_norm": 1.77223257260354,
+      "learning_rate": 2.1941188547145567e-07,
+      "loss": 2.448,
+      "step": 5811
+    },
+    {
+      "epoch": 0.7524355115383371,
+      "grad_norm": 1.7572487436439015,
+      "learning_rate": 2.1919700769941864e-07,
+      "loss": 2.4866,
+      "step": 5812
+    },
+    {
+      "epoch": 0.7525649739456906,
+      "grad_norm": 1.4882126206999426,
+      "learning_rate": 2.1898221282086802e-07,
+      "loss": 2.459,
+      "step": 5813
+    },
+    {
+      "epoch": 0.752694436353044,
+      "grad_norm": 2.0967141929893303,
+      "learning_rate": 2.1876750087967982e-07,
+      "loss": 2.3584,
+      "step": 5814
+    },
+    {
+      "epoch": 0.7528238987603975,
+      "grad_norm": 2.580482483314053,
+      "learning_rate": 2.185528719197138e-07,
+      "loss": 2.5562,
+      "step": 5815
+    },
+    {
+      "epoch": 0.7528238987603975,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.500199794769287,
+      "eval_runtime": 14.1446,
+      "eval_samples_per_second": 3.111,
+      "eval_steps_per_second": 0.212,
+      "step": 5815
+    },
+    {
+      "epoch": 0.7529533611677509,
+      "grad_norm": 1.959816140512178,
+      "learning_rate": 2.1833832598481166e-07,
+      "loss": 2.3865,
+      "step": 5816
+    },
+    {
+      "epoch": 0.7530828235751044,
+      "grad_norm": 1.9862527337467335,
+      "learning_rate": 2.1812386311879942e-07,
+      "loss": 2.3065,
+      "step": 5817
+    },
+    {
+      "epoch": 0.7532122859824578,
+      "grad_norm": 2.4568435787390857,
+      "learning_rate": 2.179094833654851e-07,
+      "loss": 2.5728,
+      "step": 5818
+    },
+    {
+      "epoch": 0.7533417483898113,
+      "grad_norm": 2.9824604578507317,
+      "learning_rate": 2.176951867686599e-07,
+      "loss": 2.3652,
+      "step": 5819
+    },
+    {
+      "epoch": 0.7534712107971647,
+      "grad_norm": 2.854202729042815,
+      "learning_rate": 2.1748097337209874e-07,
+      "loss": 2.5706,
+      "step": 5820
+    },
+    {
+      "epoch": 0.7534712107971647,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.5008655786514282,
+      "eval_runtime": 14.4718,
+      "eval_samples_per_second": 3.04,
+      "eval_steps_per_second": 0.207,
+      "step": 5820
+    },
+    {
+      "epoch": 0.7536006732045183,
+      "grad_norm": 1.5490050823786856,
+      "learning_rate": 2.1726684321955864e-07,
+      "loss": 2.4033,
+      "step": 5821
+    },
+    {
+      "epoch": 0.7537301356118717,
+      "grad_norm": 1.7718613400586933,
+      "learning_rate": 2.1705279635478034e-07,
+      "loss": 2.6299,
+      "step": 5822
+    },
+    {
+      "epoch": 0.7538595980192252,
+      "grad_norm": 2.6252808455848338,
+      "learning_rate": 2.1683883282148688e-07,
+      "loss": 2.4365,
+      "step": 5823
+    },
+    {
+      "epoch": 0.7539890604265787,
+      "grad_norm": 1.5580361297154999,
+      "learning_rate": 2.1662495266338512e-07,
+      "loss": 2.3257,
+      "step": 5824
+    },
+    {
+      "epoch": 0.7541185228339321,
+      "grad_norm": 2.0954953948086246,
+      "learning_rate": 2.164111559241641e-07,
+      "loss": 2.2993,
+      "step": 5825
+    },
+    {
+      "epoch": 0.7541185228339321,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.5009765625,
+      "eval_runtime": 13.8884,
+      "eval_samples_per_second": 3.168,
+      "eval_steps_per_second": 0.216,
+      "step": 5825
+    },
+    {
+      "epoch": 0.7542479852412856,
+      "grad_norm": 1.8732852881099835,
+      "learning_rate": 2.161974426474961e-07,
+      "loss": 2.3396,
+      "step": 5826
+    },
+    {
+      "epoch": 0.754377447648639,
+      "grad_norm": 1.6494606523117687,
+      "learning_rate": 2.1598381287703662e-07,
+      "loss": 2.4102,
+      "step": 5827
+    },
+    {
+      "epoch": 0.7545069100559925,
+      "grad_norm": 2.183039698870735,
+      "learning_rate": 2.1577026665642355e-07,
+      "loss": 2.322,
+      "step": 5828
+    },
+    {
+      "epoch": 0.7546363724633459,
+      "grad_norm": 1.3704214102627932,
+      "learning_rate": 2.1555680402927818e-07,
+      "loss": 2.3706,
+      "step": 5829
+    },
+    {
+      "epoch": 0.7547658348706994,
+      "grad_norm": 1.5718198086108035,
+      "learning_rate": 2.1534342503920508e-07,
+      "loss": 2.3579,
+      "step": 5830
+    },
+    {
+      "epoch": 0.7547658348706994,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.5030184984207153,
+      "eval_runtime": 13.3784,
+      "eval_samples_per_second": 3.289,
+      "eval_steps_per_second": 0.224,
+      "step": 5830
+    },
+    {
+      "epoch": 0.7548952972780529,
+      "grad_norm": 1.2953292597988508,
+      "learning_rate": 2.1513012972979037e-07,
+      "loss": 2.2964,
+      "step": 5831
+    },
+    {
+      "epoch": 0.7550247596854064,
+      "grad_norm": 2.8143790455089253,
+      "learning_rate": 2.149169181446043e-07,
+      "loss": 2.5725,
+      "step": 5832
+    },
+    {
+      "epoch": 0.7551542220927598,
+      "grad_norm": 1.561761962705944,
+      "learning_rate": 2.147037903272e-07,
+      "loss": 2.4038,
+      "step": 5833
+    },
+    {
+      "epoch": 0.7552836845001133,
+      "grad_norm": 1.5135770535139144,
+      "learning_rate": 2.1449074632111254e-07,
+      "loss": 2.4106,
+      "step": 5834
+    },
+    {
+      "epoch": 0.7554131469074667,
+      "grad_norm": 1.9510422997681827,
+      "learning_rate": 2.1427778616986126e-07,
+      "loss": 2.3423,
+      "step": 5835
+    },
+    {
+      "epoch": 0.7554131469074667,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.498002529144287,
+      "eval_runtime": 14.3633,
+      "eval_samples_per_second": 3.063,
+      "eval_steps_per_second": 0.209,
+      "step": 5835
+    },
+    {
+      "epoch": 0.7555426093148202,
+      "grad_norm": 2.8136067249278183,
+      "learning_rate": 2.1406490991694672e-07,
+      "loss": 2.5547,
+      "step": 5836
+    },
+    {
+      "epoch": 0.7556720717221737,
+      "grad_norm": 2.2518835814522515,
+      "learning_rate": 2.1385211760585355e-07,
+      "loss": 2.3887,
+      "step": 5837
+    },
+    {
+      "epoch": 0.7558015341295271,
+      "grad_norm": 3.1990213138561368,
+      "learning_rate": 2.1363940928004916e-07,
+      "loss": 2.5039,
+      "step": 5838
+    },
+    {
+      "epoch": 0.7559309965368806,
+      "grad_norm": 1.9339674321589873,
+      "learning_rate": 2.1342678498298317e-07,
+      "loss": 2.3853,
+      "step": 5839
+    },
+    {
+      "epoch": 0.756060458944234,
+      "grad_norm": 3.0297434106965477,
+      "learning_rate": 2.1321424475808872e-07,
+      "loss": 2.5613,
+      "step": 5840
+    },
+    {
+      "epoch": 0.756060458944234,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.4986239671707153,
+      "eval_runtime": 13.5902,
+      "eval_samples_per_second": 3.238,
+      "eval_steps_per_second": 0.221,
+      "step": 5840
+    },
+    {
+      "epoch": 0.7561899213515876,
+      "grad_norm": 1.4276704769685569,
+      "learning_rate": 2.1300178864878133e-07,
+      "loss": 2.4243,
+      "step": 5841
+    },
+    {
+      "epoch": 0.756319383758941,
+      "grad_norm": 1.9799590651593946,
+      "learning_rate": 2.127894166984592e-07,
+      "loss": 2.3115,
+      "step": 5842
+    },
+    {
+      "epoch": 0.7564488461662945,
+      "grad_norm": 1.7754285516238946,
+      "learning_rate": 2.12577128950504e-07,
+      "loss": 2.5503,
+      "step": 5843
+    },
+    {
+      "epoch": 0.7565783085736479,
+      "grad_norm": 2.239658834594748,
+      "learning_rate": 2.123649254482795e-07,
+      "loss": 2.3618,
+      "step": 5844
+    },
+    {
+      "epoch": 0.7567077709810014,
+      "grad_norm": 1.836751532298756,
+      "learning_rate": 2.1215280623513287e-07,
+      "loss": 2.3984,
+      "step": 5845
+    },
+    {
+      "epoch": 0.7567077709810014,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.5001331567764282,
+      "eval_runtime": 14.1511,
+      "eval_samples_per_second": 3.109,
+      "eval_steps_per_second": 0.212,
+      "step": 5845
+    },
+    {
+      "epoch": 0.7568372333883548,
+      "grad_norm": 1.5455753907625613,
+      "learning_rate": 2.119407713543936e-07,
+      "loss": 2.4397,
+      "step": 5846
+    },
+    {
+      "epoch": 0.7569666957957083,
+      "grad_norm": 1.9368709358290113,
+      "learning_rate": 2.1172882084937397e-07,
+      "loss": 2.2344,
+      "step": 5847
+    },
+    {
+      "epoch": 0.7570961582030618,
+      "grad_norm": 2.436297641108228,
+      "learning_rate": 2.1151695476336948e-07,
+      "loss": 2.3818,
+      "step": 5848
+    },
+    {
+      "epoch": 0.7572256206104152,
+      "grad_norm": 2.5561262580328545,
+      "learning_rate": 2.1130517313965767e-07,
+      "loss": 2.7991,
+      "step": 5849
+    },
+    {
+      "epoch": 0.7573550830177687,
+      "grad_norm": 1.810400595822349,
+      "learning_rate": 2.110934760214997e-07,
+      "loss": 2.4092,
+      "step": 5850
+    },
+    {
+      "epoch": 0.7573550830177687,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.4994450807571411,
+      "eval_runtime": 16.0322,
+      "eval_samples_per_second": 2.744,
+      "eval_steps_per_second": 0.187,
+      "step": 5850
+    },
+    {
+      "epoch": 0.7574845454251222,
+      "grad_norm": 1.4244126918376885,
+      "learning_rate": 2.1088186345213846e-07,
+      "loss": 2.123,
+      "step": 5851
+    },
+    {
+      "epoch": 0.7576140078324757,
+      "grad_norm": 1.862321548250587,
+      "learning_rate": 2.1067033547480068e-07,
+      "loss": 2.5386,
+      "step": 5852
+    },
+    {
+      "epoch": 0.7577434702398291,
+      "grad_norm": 1.269591250880515,
+      "learning_rate": 2.10458892132695e-07,
+      "loss": 2.1313,
+      "step": 5853
+    },
+    {
+      "epoch": 0.7578729326471826,
+      "grad_norm": 1.8516166311990983,
+      "learning_rate": 2.1024753346901272e-07,
+      "loss": 2.3911,
+      "step": 5854
+    },
+    {
+      "epoch": 0.758002395054536,
+      "grad_norm": 1.8588250908961133,
+      "learning_rate": 2.1003625952692835e-07,
+      "loss": 2.4712,
+      "step": 5855
+    },
+    {
+      "epoch": 0.758002395054536,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.500044345855713,
+      "eval_runtime": 15.077,
+      "eval_samples_per_second": 2.918,
+      "eval_steps_per_second": 0.199,
+      "step": 5855
+    },
+    {
+      "epoch": 0.7581318574618895,
+      "grad_norm": 1.9009920870396213,
+      "learning_rate": 2.0982507034959917e-07,
+      "loss": 2.5229,
+      "step": 5856
+    },
+    {
+      "epoch": 0.758261319869243,
+      "grad_norm": 1.1134535998530464,
+      "learning_rate": 2.096139659801645e-07,
+      "loss": 2.1763,
+      "step": 5857
+    },
+    {
+      "epoch": 0.7583907822765964,
+      "grad_norm": 2.1086126757773753,
+      "learning_rate": 2.0940294646174683e-07,
+      "loss": 2.3394,
+      "step": 5858
+    },
+    {
+      "epoch": 0.7585202446839499,
+      "grad_norm": 1.486306292850415,
+      "learning_rate": 2.0919201183745089e-07,
+      "loss": 2.3103,
+      "step": 5859
+    },
+    {
+      "epoch": 0.7586497070913034,
+      "grad_norm": 1.2114077432887764,
+      "learning_rate": 2.0898116215036456e-07,
+      "loss": 2.2319,
+      "step": 5860
+    },
+    {
+      "epoch": 0.7586497070913034,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.4993563890457153,
+      "eval_runtime": 14.2575,
+      "eval_samples_per_second": 3.086,
+      "eval_steps_per_second": 0.21,
+      "step": 5860
+    },
+    {
+      "epoch": 0.7587791694986569,
+      "grad_norm": 1.2559508015931904,
+      "learning_rate": 2.0877039744355843e-07,
+      "loss": 2.1653,
+      "step": 5861
+    },
+    {
+      "epoch": 0.7589086319060103,
+      "grad_norm": 2.1189008374382827,
+      "learning_rate": 2.0855971776008495e-07,
+      "loss": 2.4387,
+      "step": 5862
+    },
+    {
+      "epoch": 0.7590380943133638,
+      "grad_norm": 2.3990482743751795,
+      "learning_rate": 2.0834912314298025e-07,
+      "loss": 2.4204,
+      "step": 5863
+    },
+    {
+      "epoch": 0.7591675567207172,
+      "grad_norm": 2.327636679954212,
+      "learning_rate": 2.081386136352622e-07,
+      "loss": 2.4438,
+      "step": 5864
+    },
+    {
+      "epoch": 0.7592970191280707,
+      "grad_norm": 1.391628372715472,
+      "learning_rate": 2.079281892799316e-07,
+      "loss": 2.3311,
+      "step": 5865
+    },
+    {
+      "epoch": 0.7592970191280707,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.496537685394287,
+      "eval_runtime": 13.8112,
+      "eval_samples_per_second": 3.186,
+      "eval_steps_per_second": 0.217,
+      "step": 5865
+    },
+    {
+      "epoch": 0.7594264815354241,
+      "grad_norm": 1.3662541680668019,
+      "learning_rate": 2.0771785011997212e-07,
+      "loss": 2.1653,
+      "step": 5866
+    },
+    {
+      "epoch": 0.7595559439427776,
+      "grad_norm": 1.70273889789075,
+      "learning_rate": 2.075075961983495e-07,
+      "loss": 2.4429,
+      "step": 5867
+    },
+    {
+      "epoch": 0.759685406350131,
+      "grad_norm": 2.3417248361813003,
+      "learning_rate": 2.072974275580128e-07,
+      "loss": 2.3477,
+      "step": 5868
+    },
+    {
+      "epoch": 0.7598148687574845,
+      "grad_norm": 1.4082598294894912,
+      "learning_rate": 2.070873442418929e-07,
+      "loss": 2.3438,
+      "step": 5869
+    },
+    {
+      "epoch": 0.7599443311648381,
+      "grad_norm": 1.4615806857028715,
+      "learning_rate": 2.0687734629290343e-07,
+      "loss": 2.2566,
+      "step": 5870
+    },
+    {
+      "epoch": 0.7599443311648381,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.4974254369735718,
+      "eval_runtime": 14.9405,
+      "eval_samples_per_second": 2.945,
+      "eval_steps_per_second": 0.201,
+      "step": 5870
+    },
+    {
+      "epoch": 0.7600737935721915,
+      "grad_norm": 1.2953449127370287,
+      "learning_rate": 2.0666743375394116e-07,
+      "loss": 2.4097,
+      "step": 5871
+    },
+    {
+      "epoch": 0.760203255979545,
+      "grad_norm": 1.3745094758828758,
+      "learning_rate": 2.064576066678846e-07,
+      "loss": 2.2834,
+      "step": 5872
+    },
+    {
+      "epoch": 0.7603327183868984,
+      "grad_norm": 1.9308242666352429,
+      "learning_rate": 2.062478650775953e-07,
+      "loss": 2.3555,
+      "step": 5873
+    },
+    {
+      "epoch": 0.7604621807942519,
+      "grad_norm": 3.9295688564422178,
+      "learning_rate": 2.060382090259177e-07,
+      "loss": 2.3154,
+      "step": 5874
+    },
+    {
+      "epoch": 0.7605916432016053,
+      "grad_norm": 1.414342744272351,
+      "learning_rate": 2.0582863855567743e-07,
+      "loss": 2.4702,
+      "step": 5875
+    },
+    {
+      "epoch": 0.7605916432016053,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.5001331567764282,
+      "eval_runtime": 13.5483,
+      "eval_samples_per_second": 3.248,
+      "eval_steps_per_second": 0.221,
+      "step": 5875
+    },
+    {
+      "epoch": 0.7607211056089588,
+      "grad_norm": 1.7593476608674339,
+      "learning_rate": 2.0561915370968423e-07,
+      "loss": 2.3765,
+      "step": 5876
+    },
+    {
+      "epoch": 0.7608505680163122,
+      "grad_norm": 1.3573945972191568,
+      "learning_rate": 2.054097545307291e-07,
+      "loss": 2.3594,
+      "step": 5877
+    },
+    {
+      "epoch": 0.7609800304236657,
+      "grad_norm": 1.4600431466486712,
+      "learning_rate": 2.0520044106158625e-07,
+      "loss": 2.4426,
+      "step": 5878
+    },
+    {
+      "epoch": 0.7611094928310191,
+      "grad_norm": 2.2378712758784816,
+      "learning_rate": 2.0499121334501258e-07,
+      "loss": 2.4822,
+      "step": 5879
+    },
+    {
+      "epoch": 0.7612389552383727,
+      "grad_norm": 2.4570825836503136,
+      "learning_rate": 2.0478207142374674e-07,
+      "loss": 2.3691,
+      "step": 5880
+    },
+    {
+      "epoch": 0.7612389552383727,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.4966042041778564,
+      "eval_runtime": 14.0528,
+      "eval_samples_per_second": 3.131,
+      "eval_steps_per_second": 0.213,
+      "step": 5880
+    },
+    {
+      "epoch": 0.7613684176457262,
+      "grad_norm": 1.7874177609679103,
+      "learning_rate": 2.0457301534051004e-07,
+      "loss": 2.416,
+      "step": 5881
+    },
+    {
+      "epoch": 0.7614978800530796,
+      "grad_norm": 1.5142164972034926,
+      "learning_rate": 2.0436404513800675e-07,
+      "loss": 2.6133,
+      "step": 5882
+    },
+    {
+      "epoch": 0.7616273424604331,
+      "grad_norm": 1.8254061667671044,
+      "learning_rate": 2.0415516085892303e-07,
+      "loss": 2.5039,
+      "step": 5883
+    },
+    {
+      "epoch": 0.7617568048677865,
+      "grad_norm": 1.572149682501575,
+      "learning_rate": 2.0394636254592802e-07,
+      "loss": 2.4492,
+      "step": 5884
+    },
+    {
+      "epoch": 0.76188626727514,
+      "grad_norm": 1.5063564419177435,
+      "learning_rate": 2.0373765024167284e-07,
+      "loss": 2.6472,
+      "step": 5885
+    },
+    {
+      "epoch": 0.76188626727514,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.5008877515792847,
+      "eval_runtime": 13.8895,
+      "eval_samples_per_second": 3.168,
+      "eval_steps_per_second": 0.216,
+      "step": 5885
+    },
+    {
+      "epoch": 0.7620157296824934,
+      "grad_norm": 2.1321630772537,
+      "learning_rate": 2.0352902398879098e-07,
+      "loss": 2.4194,
+      "step": 5886
+    },
+    {
+      "epoch": 0.7621451920898469,
+      "grad_norm": 1.846813322395439,
+      "learning_rate": 2.0332048382989895e-07,
+      "loss": 2.2397,
+      "step": 5887
+    },
+    {
+      "epoch": 0.7622746544972003,
+      "grad_norm": 4.134731739638366,
+      "learning_rate": 2.0311202980759504e-07,
+      "loss": 2.6807,
+      "step": 5888
+    },
+    {
+      "epoch": 0.7624041169045538,
+      "grad_norm": 3.5200971486495347,
+      "learning_rate": 2.0290366196446045e-07,
+      "loss": 2.8784,
+      "step": 5889
+    },
+    {
+      "epoch": 0.7625335793119074,
+      "grad_norm": 1.393468781608451,
+      "learning_rate": 2.0269538034305827e-07,
+      "loss": 2.3853,
+      "step": 5890
+    },
+    {
+      "epoch": 0.7625335793119074,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.500199794769287,
+      "eval_runtime": 14.4333,
+      "eval_samples_per_second": 3.048,
+      "eval_steps_per_second": 0.208,
+      "step": 5890
+    },
+    {
+      "epoch": 0.7626630417192608,
+      "grad_norm": 4.058911303706695,
+      "learning_rate": 2.0248718498593453e-07,
+      "loss": 2.9038,
+      "step": 5891
+    },
+    {
+      "epoch": 0.7627925041266143,
+      "grad_norm": 1.9489576287446886,
+      "learning_rate": 2.0227907593561718e-07,
+      "loss": 2.6946,
+      "step": 5892
+    },
+    {
+      "epoch": 0.7629219665339677,
+      "grad_norm": 1.2233543433416605,
+      "learning_rate": 2.0207105323461658e-07,
+      "loss": 2.3494,
+      "step": 5893
+    },
+    {
+      "epoch": 0.7630514289413212,
+      "grad_norm": 2.7917420980757366,
+      "learning_rate": 2.0186311692542594e-07,
+      "loss": 2.3591,
+      "step": 5894
+    },
+    {
+      "epoch": 0.7631808913486746,
+      "grad_norm": 4.952772839611828,
+      "learning_rate": 2.0165526705052007e-07,
+      "loss": 2.3567,
+      "step": 5895
+    },
+    {
+      "epoch": 0.7631808913486746,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.5015536546707153,
+      "eval_runtime": 14.1098,
+      "eval_samples_per_second": 3.118,
+      "eval_steps_per_second": 0.213,
+      "step": 5895
+    },
+    {
+      "epoch": 0.7633103537560281,
+      "grad_norm": 2.5469536401721076,
+      "learning_rate": 2.0144750365235687e-07,
+      "loss": 2.5813,
+      "step": 5896
+    },
+    {
+      "epoch": 0.7634398161633815,
+      "grad_norm": 1.4416734309857384,
+      "learning_rate": 2.012398267733761e-07,
+      "loss": 2.3984,
+      "step": 5897
+    },
+    {
+      "epoch": 0.763569278570735,
+      "grad_norm": 3.5685517982616104,
+      "learning_rate": 2.0103223645599974e-07,
+      "loss": 2.9966,
+      "step": 5898
+    },
+    {
+      "epoch": 0.7636987409780885,
+      "grad_norm": 2.833785840312419,
+      "learning_rate": 2.008247327426327e-07,
+      "loss": 2.3867,
+      "step": 5899
+    },
+    {
+      "epoch": 0.763828203385442,
+      "grad_norm": 2.6422615350344825,
+      "learning_rate": 2.006173156756614e-07,
+      "loss": 2.6487,
+      "step": 5900
+    },
+    {
+      "epoch": 0.763828203385442,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.500044345855713,
+      "eval_runtime": 13.8866,
+      "eval_samples_per_second": 3.169,
+      "eval_steps_per_second": 0.216,
+      "step": 5900
+    },
+    {
+      "epoch": 0.7639576657927954,
+      "grad_norm": 3.167058360094262,
+      "learning_rate": 2.0040998529745532e-07,
+      "loss": 2.4365,
+      "step": 5901
+    },
+    {
+      "epoch": 0.7640871282001489,
+      "grad_norm": 1.5622768032380048,
+      "learning_rate": 2.0020274165036608e-07,
+      "loss": 2.3027,
+      "step": 5902
+    },
+    {
+      "epoch": 0.7642165906075024,
+      "grad_norm": 1.5589453517866538,
+      "learning_rate": 1.9999558477672674e-07,
+      "loss": 2.1697,
+      "step": 5903
+    },
+    {
+      "epoch": 0.7643460530148558,
+      "grad_norm": 1.5635110718355771,
+      "learning_rate": 1.9978851471885373e-07,
+      "loss": 2.4434,
+      "step": 5904
+    },
+    {
+      "epoch": 0.7644755154222093,
+      "grad_norm": 3.0729043453070823,
+      "learning_rate": 1.995815315190454e-07,
+      "loss": 3.3315,
+      "step": 5905
+    },
+    {
+      "epoch": 0.7644755154222093,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.4986904859542847,
+      "eval_runtime": 14.7505,
+      "eval_samples_per_second": 2.983,
+      "eval_steps_per_second": 0.203,
+      "step": 5905
+    },
+    {
+      "epoch": 0.7646049778295627,
+      "grad_norm": 1.8095601699611552,
+      "learning_rate": 1.9937463521958183e-07,
+      "loss": 2.2881,
+      "step": 5906
+    },
+    {
+      "epoch": 0.7647344402369162,
+      "grad_norm": 1.8593368639132177,
+      "learning_rate": 1.991678258627265e-07,
+      "loss": 2.3794,
+      "step": 5907
+    },
+    {
+      "epoch": 0.7648639026442696,
+      "grad_norm": 2.770245655129175,
+      "learning_rate": 1.9896110349072357e-07,
+      "loss": 2.4385,
+      "step": 5908
+    },
+    {
+      "epoch": 0.7649933650516232,
+      "grad_norm": 1.3787170669825033,
+      "learning_rate": 1.9875446814580064e-07,
+      "loss": 2.2878,
+      "step": 5909
+    },
+    {
+      "epoch": 0.7651228274589766,
+      "grad_norm": 2.369476370607867,
+      "learning_rate": 1.9854791987016739e-07,
+      "loss": 2.459,
+      "step": 5910
+    },
+    {
+      "epoch": 0.7651228274589766,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.498734951019287,
+      "eval_runtime": 13.8554,
+      "eval_samples_per_second": 3.176,
+      "eval_steps_per_second": 0.217,
+      "step": 5910
+    },
+    {
+      "epoch": 0.7652522898663301,
+      "grad_norm": 1.2830844503907153,
+      "learning_rate": 1.9834145870601513e-07,
+      "loss": 2.2778,
+      "step": 5911
+    },
+    {
+      "epoch": 0.7653817522736835,
+      "grad_norm": 2.5302739216629244,
+      "learning_rate": 1.9813508469551808e-07,
+      "loss": 2.5352,
+      "step": 5912
+    },
+    {
+      "epoch": 0.765511214681037,
+      "grad_norm": 1.5276455432435188,
+      "learning_rate": 1.9792879788083219e-07,
+      "loss": 2.2871,
+      "step": 5913
+    },
+    {
+      "epoch": 0.7656406770883905,
+      "grad_norm": 3.3943950562170246,
+      "learning_rate": 1.9772259830409542e-07,
+      "loss": 2.9097,
+      "step": 5914
+    },
+    {
+      "epoch": 0.7657701394957439,
+      "grad_norm": 1.4990112841006813,
+      "learning_rate": 1.9751648600742863e-07,
+      "loss": 2.5159,
+      "step": 5915
+    },
+    {
+      "epoch": 0.7657701394957439,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.502241611480713,
+      "eval_runtime": 14.3836,
+      "eval_samples_per_second": 3.059,
+      "eval_steps_per_second": 0.209,
+      "step": 5915
+    },
+    {
+      "epoch": 0.7658996019030974,
+      "grad_norm": 1.0252065395231953,
+      "learning_rate": 1.973104610329341e-07,
+      "loss": 2.2339,
+      "step": 5916
+    },
+    {
+      "epoch": 0.7660290643104508,
+      "grad_norm": 1.561934321130518,
+      "learning_rate": 1.9710452342269703e-07,
+      "loss": 2.271,
+      "step": 5917
+    },
+    {
+      "epoch": 0.7661585267178043,
+      "grad_norm": 2.3183876245973787,
+      "learning_rate": 1.9689867321878387e-07,
+      "loss": 2.3923,
+      "step": 5918
+    },
+    {
+      "epoch": 0.7662879891251578,
+      "grad_norm": 1.4481714440717672,
+      "learning_rate": 1.966929104632441e-07,
+      "loss": 2.3518,
+      "step": 5919
+    },
+    {
+      "epoch": 0.7664174515325113,
+      "grad_norm": 1.7657229319959258,
+      "learning_rate": 1.9648723519810876e-07,
+      "loss": 2.3081,
+      "step": 5920
+    },
+    {
+      "epoch": 0.7664174515325113,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.5001775026321411,
+      "eval_runtime": 14.8734,
+      "eval_samples_per_second": 2.958,
+      "eval_steps_per_second": 0.202,
+      "step": 5920
+    },
+    {
+      "epoch": 0.7665469139398647,
+      "grad_norm": 3.0911986408405423,
+      "learning_rate": 1.96281647465391e-07,
+      "loss": 2.2639,
+      "step": 5921
+    },
+    {
+      "epoch": 0.7666763763472182,
+      "grad_norm": 2.1304295172481202,
+      "learning_rate": 1.9607614730708665e-07,
+      "loss": 2.3624,
+      "step": 5922
+    },
+    {
+      "epoch": 0.7668058387545716,
+      "grad_norm": 1.3855225283903845,
+      "learning_rate": 1.9587073476517289e-07,
+      "loss": 2.3735,
+      "step": 5923
+    },
+    {
+      "epoch": 0.7669353011619251,
+      "grad_norm": 2.984396702597027,
+      "learning_rate": 1.9566540988160986e-07,
+      "loss": 2.2583,
+      "step": 5924
+    },
+    {
+      "epoch": 0.7670647635692786,
+      "grad_norm": 1.7741862050664121,
+      "learning_rate": 1.9546017269833906e-07,
+      "loss": 2.478,
+      "step": 5925
+    },
+    {
+      "epoch": 0.7670647635692786,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.502152919769287,
+      "eval_runtime": 13.805,
+      "eval_samples_per_second": 3.187,
+      "eval_steps_per_second": 0.217,
+      "step": 5925
+    },
+    {
+      "epoch": 0.767194225976632,
+      "grad_norm": 2.101543329687786,
+      "learning_rate": 1.9525502325728418e-07,
+      "loss": 2.3989,
+      "step": 5926
+    },
+    {
+      "epoch": 0.7673236883839855,
+      "grad_norm": 1.4956955927310565,
+      "learning_rate": 1.9504996160035128e-07,
+      "loss": 2.3948,
+      "step": 5927
+    },
+    {
+      "epoch": 0.7674531507913389,
+      "grad_norm": 1.9052839059065405,
+      "learning_rate": 1.948449877694286e-07,
+      "loss": 2.5186,
+      "step": 5928
+    },
+    {
+      "epoch": 0.7675826131986925,
+      "grad_norm": 2.6304563370898584,
+      "learning_rate": 1.9464010180638592e-07,
+      "loss": 2.2791,
+      "step": 5929
+    },
+    {
+      "epoch": 0.7677120756060459,
+      "grad_norm": 1.1883510455228177,
+      "learning_rate": 1.9443530375307579e-07,
+      "loss": 2.2644,
+      "step": 5930
+    },
+    {
+      "epoch": 0.7677120756060459,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.5002219676971436,
+      "eval_runtime": 14.4103,
+      "eval_samples_per_second": 3.053,
+      "eval_steps_per_second": 0.208,
+      "step": 5930
+    },
+    {
+      "epoch": 0.7678415380133994,
+      "grad_norm": 1.6880308486218178,
+      "learning_rate": 1.9423059365133162e-07,
+      "loss": 2.4407,
+      "step": 5931
+    },
+    {
+      "epoch": 0.7679710004207528,
+      "grad_norm": 2.766901150411268,
+      "learning_rate": 1.9402597154296995e-07,
+      "loss": 2.5522,
+      "step": 5932
+    },
+    {
+      "epoch": 0.7681004628281063,
+      "grad_norm": 1.5873555048956454,
+      "learning_rate": 1.938214374697893e-07,
+      "loss": 2.415,
+      "step": 5933
+    },
+    {
+      "epoch": 0.7682299252354597,
+      "grad_norm": 2.7049780084981565,
+      "learning_rate": 1.9361699147356937e-07,
+      "loss": 2.2012,
+      "step": 5934
+    },
+    {
+      "epoch": 0.7683593876428132,
+      "grad_norm": 1.8552129165576046,
+      "learning_rate": 1.9341263359607295e-07,
+      "loss": 2.4868,
+      "step": 5935
+    },
+    {
+      "epoch": 0.7683593876428132,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.500932216644287,
+      "eval_runtime": 14.2395,
+      "eval_samples_per_second": 3.09,
+      "eval_steps_per_second": 0.211,
+      "step": 5935
+    },
+    {
+      "epoch": 0.7684888500501666,
+      "grad_norm": 1.6354437818331267,
+      "learning_rate": 1.9320836387904392e-07,
+      "loss": 2.3818,
+      "step": 5936
+    },
+    {
+      "epoch": 0.7686183124575201,
+      "grad_norm": 2.380509089967578,
+      "learning_rate": 1.9300418236420842e-07,
+      "loss": 2.4316,
+      "step": 5937
+    },
+    {
+      "epoch": 0.7687477748648737,
+      "grad_norm": 1.6673693466640416,
+      "learning_rate": 1.92800089093275e-07,
+      "loss": 2.5042,
+      "step": 5938
+    },
+    {
+      "epoch": 0.7688772372722271,
+      "grad_norm": 1.354707275436024,
+      "learning_rate": 1.925960841079335e-07,
+      "loss": 2.4111,
+      "step": 5939
+    },
+    {
+      "epoch": 0.7690066996795806,
+      "grad_norm": 1.884803974676278,
+      "learning_rate": 1.923921674498565e-07,
+      "loss": 2.5173,
+      "step": 5940
+    },
+    {
+      "epoch": 0.7690066996795806,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.4966042041778564,
+      "eval_runtime": 16.4031,
+      "eval_samples_per_second": 2.682,
+      "eval_steps_per_second": 0.183,
+      "step": 5940
+    },
+    {
+      "epoch": 0.769136162086934,
+      "grad_norm": 1.4961730243630353,
+      "learning_rate": 1.9218833916069775e-07,
+      "loss": 2.2083,
+      "step": 5941
+    },
+    {
+      "epoch": 0.7692656244942875,
+      "grad_norm": 2.0095951494477857,
+      "learning_rate": 1.9198459928209332e-07,
+      "loss": 2.1587,
+      "step": 5942
+    },
+    {
+      "epoch": 0.7693950869016409,
+      "grad_norm": 1.6736478695801495,
+      "learning_rate": 1.9178094785566142e-07,
+      "loss": 2.3335,
+      "step": 5943
+    },
+    {
+      "epoch": 0.7695245493089944,
+      "grad_norm": 2.192720257106963,
+      "learning_rate": 1.9157738492300176e-07,
+      "loss": 2.2224,
+      "step": 5944
+    },
+    {
+      "epoch": 0.7696540117163478,
+      "grad_norm": 2.36502245282972,
+      "learning_rate": 1.913739105256964e-07,
+      "loss": 2.3296,
+      "step": 5945
+    },
+    {
+      "epoch": 0.7696540117163478,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.503706455230713,
+      "eval_runtime": 13.9897,
+      "eval_samples_per_second": 3.145,
+      "eval_steps_per_second": 0.214,
+      "step": 5945
+    },
+    {
+      "epoch": 0.7697834741237013,
+      "grad_norm": 1.4459858726170736,
+      "learning_rate": 1.911705247053091e-07,
+      "loss": 2.4153,
+      "step": 5946
+    },
+    {
+      "epoch": 0.7699129365310547,
+      "grad_norm": 1.9920734193437843,
+      "learning_rate": 1.9096722750338524e-07,
+      "loss": 2.5237,
+      "step": 5947
+    },
+    {
+      "epoch": 0.7700423989384083,
+      "grad_norm": 1.6284989520443458,
+      "learning_rate": 1.9076401896145284e-07,
+      "loss": 2.3054,
+      "step": 5948
+    },
+    {
+      "epoch": 0.7701718613457618,
+      "grad_norm": 2.3742044914808735,
+      "learning_rate": 1.9056089912102088e-07,
+      "loss": 2.6118,
+      "step": 5949
+    },
+    {
+      "epoch": 0.7703013237531152,
+      "grad_norm": 2.4977493282187706,
+      "learning_rate": 1.903578680235809e-07,
+      "loss": 2.3774,
+      "step": 5950
+    },
+    {
+      "epoch": 0.7703013237531152,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.5000221729278564,
+      "eval_runtime": 14.7863,
+      "eval_samples_per_second": 2.976,
+      "eval_steps_per_second": 0.203,
+      "step": 5950
+    },
+    {
+      "epoch": 0.7704307861604687,
+      "grad_norm": 1.4355870507424202,
+      "learning_rate": 1.9015492571060648e-07,
+      "loss": 2.2729,
+      "step": 5951
+    },
+    {
+      "epoch": 0.7705602485678221,
+      "grad_norm": 2.0544240421967004,
+      "learning_rate": 1.8995207222355233e-07,
+      "loss": 2.5181,
+      "step": 5952
+    },
+    {
+      "epoch": 0.7706897109751756,
+      "grad_norm": 1.9853028041490168,
+      "learning_rate": 1.897493076038555e-07,
+      "loss": 2.438,
+      "step": 5953
+    },
+    {
+      "epoch": 0.770819173382529,
+      "grad_norm": 1.523101933486617,
+      "learning_rate": 1.8954663189293446e-07,
+      "loss": 2.1951,
+      "step": 5954
+    },
+    {
+      "epoch": 0.7709486357898825,
+      "grad_norm": 1.585523135004017,
+      "learning_rate": 1.8934404513219011e-07,
+      "loss": 2.4333,
+      "step": 5955
+    },
+    {
+      "epoch": 0.7709486357898825,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.4978915452957153,
+      "eval_runtime": 14.4892,
+      "eval_samples_per_second": 3.037,
+      "eval_steps_per_second": 0.207,
+      "step": 5955
+    },
+    {
+      "epoch": 0.7710780981972359,
+      "grad_norm": 1.5669963664772997,
+      "learning_rate": 1.8914154736300511e-07,
+      "loss": 2.5242,
+      "step": 5956
+    },
+    {
+      "epoch": 0.7712075606045894,
+      "grad_norm": 1.8538204540122294,
+      "learning_rate": 1.8893913862674328e-07,
+      "loss": 2.4797,
+      "step": 5957
+    },
+    {
+      "epoch": 0.771337023011943,
+      "grad_norm": 2.4622943772964843,
+      "learning_rate": 1.8873681896475113e-07,
+      "loss": 2.2961,
+      "step": 5958
+    },
+    {
+      "epoch": 0.7714664854192964,
+      "grad_norm": 1.3910046817559374,
+      "learning_rate": 1.8853458841835635e-07,
+      "loss": 2.4067,
+      "step": 5959
+    },
+    {
+      "epoch": 0.7715959478266499,
+      "grad_norm": 1.6375854849941913,
+      "learning_rate": 1.8833244702886847e-07,
+      "loss": 2.3096,
+      "step": 5960
+    },
+    {
+      "epoch": 0.7715959478266499,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.500932216644287,
+      "eval_runtime": 14.433,
+      "eval_samples_per_second": 3.049,
+      "eval_steps_per_second": 0.208,
+      "step": 5960
+    },
+    {
+      "epoch": 0.7717254102340033,
+      "grad_norm": 2.6916282684969253,
+      "learning_rate": 1.8813039483757935e-07,
+      "loss": 2.5405,
+      "step": 5961
+    },
+    {
+      "epoch": 0.7718548726413568,
+      "grad_norm": 1.5912020564879144,
+      "learning_rate": 1.879284318857618e-07,
+      "loss": 2.4434,
+      "step": 5962
+    },
+    {
+      "epoch": 0.7719843350487102,
+      "grad_norm": 2.4860316455501517,
+      "learning_rate": 1.8772655821467124e-07,
+      "loss": 2.5452,
+      "step": 5963
+    },
+    {
+      "epoch": 0.7721137974560637,
+      "grad_norm": 1.8619342726099855,
+      "learning_rate": 1.8752477386554433e-07,
+      "loss": 2.635,
+      "step": 5964
+    },
+    {
+      "epoch": 0.7722432598634171,
+      "grad_norm": 2.174473733903905,
+      "learning_rate": 1.8732307887959939e-07,
+      "loss": 2.4175,
+      "step": 5965
+    },
+    {
+      "epoch": 0.7722432598634171,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.4986461400985718,
+      "eval_runtime": 14.5514,
+      "eval_samples_per_second": 3.024,
+      "eval_steps_per_second": 0.206,
+      "step": 5965
+    },
+    {
+      "epoch": 0.7723727222707706,
+      "grad_norm": 2.2701006308701284,
+      "learning_rate": 1.871214732980371e-07,
+      "loss": 2.4016,
+      "step": 5966
+    },
+    {
+      "epoch": 0.772502184678124,
+      "grad_norm": 1.727138334459291,
+      "learning_rate": 1.8691995716203906e-07,
+      "loss": 2.2354,
+      "step": 5967
+    },
+    {
+      "epoch": 0.7726316470854776,
+      "grad_norm": 1.4155427550076716,
+      "learning_rate": 1.8671853051276924e-07,
+      "loss": 2.2651,
+      "step": 5968
+    },
+    {
+      "epoch": 0.772761109492831,
+      "grad_norm": 1.9437943650458096,
+      "learning_rate": 1.8651719339137355e-07,
+      "loss": 2.2378,
+      "step": 5969
+    },
+    {
+      "epoch": 0.7728905719001845,
+      "grad_norm": 1.3922818829841908,
+      "learning_rate": 1.8631594583897835e-07,
+      "loss": 2.3579,
+      "step": 5970
+    },
+    {
+      "epoch": 0.7728905719001845,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.5015980005264282,
+      "eval_runtime": 14.6426,
+      "eval_samples_per_second": 3.005,
+      "eval_steps_per_second": 0.205,
+      "step": 5970
+    },
+    {
+      "epoch": 0.773020034307538,
+      "grad_norm": 1.690810646949823,
+      "learning_rate": 1.861147878966931e-07,
+      "loss": 2.3079,
+      "step": 5971
+    },
+    {
+      "epoch": 0.7731494967148914,
+      "grad_norm": 2.0594655012712964,
+      "learning_rate": 1.859137196056081e-07,
+      "loss": 2.6616,
+      "step": 5972
+    },
+    {
+      "epoch": 0.7732789591222449,
+      "grad_norm": 1.735278684515017,
+      "learning_rate": 1.8571274100679562e-07,
+      "loss": 2.4963,
+      "step": 5973
+    },
+    {
+      "epoch": 0.7734084215295983,
+      "grad_norm": 1.2496522003017534,
+      "learning_rate": 1.8551185214131023e-07,
+      "loss": 2.4058,
+      "step": 5974
+    },
+    {
+      "epoch": 0.7735378839369518,
+      "grad_norm": 2.0783386478791175,
+      "learning_rate": 1.8531105305018664e-07,
+      "loss": 2.3503,
+      "step": 5975
+    },
+    {
+      "epoch": 0.7735378839369518,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.5000888109207153,
+      "eval_runtime": 14.801,
+      "eval_samples_per_second": 2.973,
+      "eval_steps_per_second": 0.203,
+      "step": 5975
+    },
+    {
+      "epoch": 0.7736673463443052,
+      "grad_norm": 2.107689541065852,
+      "learning_rate": 1.8511034377444254e-07,
+      "loss": 2.4397,
+      "step": 5976
+    },
+    {
+      "epoch": 0.7737968087516587,
+      "grad_norm": 2.0955878298144386,
+      "learning_rate": 1.849097243550771e-07,
+      "loss": 2.4419,
+      "step": 5977
+    },
+    {
+      "epoch": 0.7739262711590122,
+      "grad_norm": 1.8049458331472423,
+      "learning_rate": 1.8470919483307044e-07,
+      "loss": 2.6097,
+      "step": 5978
+    },
+    {
+      "epoch": 0.7740557335663657,
+      "grad_norm": 1.9295600186183828,
+      "learning_rate": 1.8450875524938523e-07,
+      "loss": 2.4265,
+      "step": 5979
+    },
+    {
+      "epoch": 0.7741851959737192,
+      "grad_norm": 1.314442472325798,
+      "learning_rate": 1.8430840564496516e-07,
+      "loss": 2.3301,
+      "step": 5980
+    },
+    {
+      "epoch": 0.7741851959737192,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.5035954713821411,
+      "eval_runtime": 14.0332,
+      "eval_samples_per_second": 3.135,
+      "eval_steps_per_second": 0.214,
+      "step": 5980
+    },
+    {
+      "epoch": 0.7743146583810726,
+      "grad_norm": 3.4863758435551997,
+      "learning_rate": 1.8410814606073544e-07,
+      "loss": 2.7817,
+      "step": 5981
+    },
+    {
+      "epoch": 0.7744441207884261,
+      "grad_norm": 1.2463084969887293,
+      "learning_rate": 1.8390797653760353e-07,
+      "loss": 2.4587,
+      "step": 5982
+    },
+    {
+      "epoch": 0.7745735831957795,
+      "grad_norm": 1.6988732542503902,
+      "learning_rate": 1.8370789711645778e-07,
+      "loss": 2.4956,
+      "step": 5983
+    },
+    {
+      "epoch": 0.774703045603133,
+      "grad_norm": 1.4724544971075704,
+      "learning_rate": 1.8350790783816873e-07,
+      "loss": 2.356,
+      "step": 5984
+    },
+    {
+      "epoch": 0.7748325080104864,
+      "grad_norm": 1.386543429629265,
+      "learning_rate": 1.8330800874358825e-07,
+      "loss": 2.2573,
+      "step": 5985
+    },
+    {
+      "epoch": 0.7748325080104864,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.5000221729278564,
+      "eval_runtime": 14.5141,
+      "eval_samples_per_second": 3.032,
+      "eval_steps_per_second": 0.207,
+      "step": 5985
+    },
+    {
+      "epoch": 0.7749619704178399,
+      "grad_norm": 2.7942180539683985,
+      "learning_rate": 1.831081998735494e-07,
+      "loss": 2.625,
+      "step": 5986
+    },
+    {
+      "epoch": 0.7750914328251934,
+      "grad_norm": 4.192735906577279,
+      "learning_rate": 1.8290848126886772e-07,
+      "loss": 2.7822,
+      "step": 5987
+    },
+    {
+      "epoch": 0.7752208952325469,
+      "grad_norm": 1.5906056842781113,
+      "learning_rate": 1.8270885297033928e-07,
+      "loss": 2.5168,
+      "step": 5988
+    },
+    {
+      "epoch": 0.7753503576399003,
+      "grad_norm": 1.4682108535111058,
+      "learning_rate": 1.825093150187427e-07,
+      "loss": 2.4207,
+      "step": 5989
+    },
+    {
+      "epoch": 0.7754798200472538,
+      "grad_norm": 1.42088272727761,
+      "learning_rate": 1.8230986745483715e-07,
+      "loss": 2.2549,
+      "step": 5990
+    },
+    {
+      "epoch": 0.7754798200472538,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.5007545948028564,
+      "eval_runtime": 14.0179,
+      "eval_samples_per_second": 3.139,
+      "eval_steps_per_second": 0.214,
+      "step": 5990
+    },
+    {
+      "epoch": 0.7756092824546073,
+      "grad_norm": 2.104141643266121,
+      "learning_rate": 1.8211051031936433e-07,
+      "loss": 2.7314,
+      "step": 5991
+    },
+    {
+      "epoch": 0.7757387448619607,
+      "grad_norm": 1.4343831673431897,
+      "learning_rate": 1.8191124365304673e-07,
+      "loss": 2.3268,
+      "step": 5992
+    },
+    {
+      "epoch": 0.7758682072693142,
+      "grad_norm": 4.005288644184857,
+      "learning_rate": 1.8171206749658846e-07,
+      "loss": 2.8022,
+      "step": 5993
+    },
+    {
+      "epoch": 0.7759976696766676,
+      "grad_norm": 1.8738858536186669,
+      "learning_rate": 1.8151298189067558e-07,
+      "loss": 2.3717,
+      "step": 5994
+    },
+    {
+      "epoch": 0.7761271320840211,
+      "grad_norm": 1.7372780979044173,
+      "learning_rate": 1.813139868759751e-07,
+      "loss": 2.334,
+      "step": 5995
+    },
+    {
+      "epoch": 0.7761271320840211,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.5007102489471436,
+      "eval_runtime": 13.5662,
+      "eval_samples_per_second": 3.243,
+      "eval_steps_per_second": 0.221,
+      "step": 5995
+    },
+    {
+      "epoch": 0.7762565944913745,
+      "grad_norm": 2.0536557945484213,
+      "learning_rate": 1.8111508249313596e-07,
+      "loss": 2.334,
+      "step": 5996
+    },
+    {
+      "epoch": 0.7763860568987281,
+      "grad_norm": 2.9657504393827807,
+      "learning_rate": 1.8091626878278846e-07,
+      "loss": 2.4902,
+      "step": 5997
+    },
+    {
+      "epoch": 0.7765155193060815,
+      "grad_norm": 3.226283522133443,
+      "learning_rate": 1.8071754578554403e-07,
+      "loss": 2.8318,
+      "step": 5998
+    },
+    {
+      "epoch": 0.776644981713435,
+      "grad_norm": 1.669540781358496,
+      "learning_rate": 1.8051891354199609e-07,
+      "loss": 2.3899,
+      "step": 5999
+    },
+    {
+      "epoch": 0.7767744441207884,
+      "grad_norm": 1.2407471452519037,
+      "learning_rate": 1.8032037209271942e-07,
+      "loss": 2.3147,
+      "step": 6000
+    },
+    {
+      "epoch": 0.7767744441207884,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.4986239671707153,
+      "eval_runtime": 13.5279,
+      "eval_samples_per_second": 3.253,
+      "eval_steps_per_second": 0.222,
+      "step": 6000
     }
   ],
   "logging_steps": 1,
@@ -56142,7 +61242,7 @@
       "attributes": {}
     }
   },
-  "total_flos": 4624798951899136.0,
+  "total_flos": 5045905986322432.0,
   "train_batch_size": 4,
   "trial_name": null,
   "trial_params": null