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": 0.8415056477975208,
+  "epoch": 0.9062368514742531,
   "eval_steps": 5,
-  "global_step": 6500,
+  "global_step": 7000,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
@@ -66323,6 +66323,5106 @@
       "eval_samples_per_second": 3.247,
       "eval_steps_per_second": 0.221,
       "step": 6500
+    },
+    {
+      "epoch": 0.8416351102048742,
+      "grad_norm": 1.3259543552425912,
+      "learning_rate": 9.307267711406206e-08,
+      "loss": 2.343,
+      "step": 6501
+    },
+    {
+      "epoch": 0.8417645726122277,
+      "grad_norm": 1.4418420274137762,
+      "learning_rate": 9.292442271411635e-08,
+      "loss": 2.4102,
+      "step": 6502
+    },
+    {
+      "epoch": 0.8418940350195812,
+      "grad_norm": 2.7925919099498744,
+      "learning_rate": 9.277627700139229e-08,
+      "loss": 2.4409,
+      "step": 6503
+    },
+    {
+      "epoch": 0.8420234974269346,
+      "grad_norm": 3.2387106365662044,
+      "learning_rate": 9.262824000615102e-08,
+      "loss": 2.314,
+      "step": 6504
+    },
+    {
+      "epoch": 0.8421529598342882,
+      "grad_norm": 1.8635814648159303,
+      "learning_rate": 9.248031175863245e-08,
+      "loss": 2.3154,
+      "step": 6505
+    },
+    {
+      "epoch": 0.8421529598342882,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "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": 13.6189,
+      "eval_samples_per_second": 3.231,
+      "eval_steps_per_second": 0.22,
+      "step": 6505
+    },
+    {
+      "epoch": 0.8422824222416416,
+      "grad_norm": 2.4410969275031555,
+      "learning_rate": 9.233249228905401e-08,
+      "loss": 2.3711,
+      "step": 6506
+    },
+    {
+      "epoch": 0.8424118846489951,
+      "grad_norm": 1.4817949391045466,
+      "learning_rate": 9.21847816276105e-08,
+      "loss": 2.3257,
+      "step": 6507
+    },
+    {
+      "epoch": 0.8425413470563485,
+      "grad_norm": 2.0258687338134194,
+      "learning_rate": 9.203717980447537e-08,
+      "loss": 2.1184,
+      "step": 6508
+    },
+    {
+      "epoch": 0.842670809463702,
+      "grad_norm": 1.7264698812002046,
+      "learning_rate": 9.188968684979864e-08,
+      "loss": 2.2495,
+      "step": 6509
+    },
+    {
+      "epoch": 0.8428002718710554,
+      "grad_norm": 1.4558195482246246,
+      "learning_rate": 9.1742302793709e-08,
+      "loss": 2.2571,
+      "step": 6510
+    },
+    {
+      "epoch": 0.8428002718710554,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.4972034692764282,
+      "eval_runtime": 13.8019,
+      "eval_samples_per_second": 3.188,
+      "eval_steps_per_second": 0.217,
+      "step": 6510
+    },
+    {
+      "epoch": 0.8429297342784089,
+      "grad_norm": 1.566367849584917,
+      "learning_rate": 9.159502766631287e-08,
+      "loss": 2.2446,
+      "step": 6511
+    },
+    {
+      "epoch": 0.8430591966857623,
+      "grad_norm": 2.5926903019890455,
+      "learning_rate": 9.144786149769373e-08,
+      "loss": 2.333,
+      "step": 6512
+    },
+    {
+      "epoch": 0.8431886590931158,
+      "grad_norm": 2.114919991502529,
+      "learning_rate": 9.130080431791366e-08,
+      "loss": 2.5859,
+      "step": 6513
+    },
+    {
+      "epoch": 0.8433181215004693,
+      "grad_norm": 2.503769759007235,
+      "learning_rate": 9.11538561570119e-08,
+      "loss": 2.1353,
+      "step": 6514
+    },
+    {
+      "epoch": 0.8434475839078228,
+      "grad_norm": 2.021167943015097,
+      "learning_rate": 9.100701704500545e-08,
+      "loss": 2.3098,
+      "step": 6515
+    },
+    {
+      "epoch": 0.8434475839078228,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.49853515625,
+      "eval_runtime": 14.0454,
+      "eval_samples_per_second": 3.133,
+      "eval_steps_per_second": 0.214,
+      "step": 6515
+    },
+    {
+      "epoch": 0.8435770463151763,
+      "grad_norm": 2.16966049767526,
+      "learning_rate": 9.086028701188943e-08,
+      "loss": 2.3318,
+      "step": 6516
+    },
+    {
+      "epoch": 0.8437065087225297,
+      "grad_norm": 2.8273270822500804,
+      "learning_rate": 9.07136660876362e-08,
+      "loss": 2.6746,
+      "step": 6517
+    },
+    {
+      "epoch": 0.8438359711298832,
+      "grad_norm": 1.3446635160909532,
+      "learning_rate": 9.056715430219631e-08,
+      "loss": 2.2957,
+      "step": 6518
+    },
+    {
+      "epoch": 0.8439654335372366,
+      "grad_norm": 2.0053251293641376,
+      "learning_rate": 9.042075168549747e-08,
+      "loss": 2.4177,
+      "step": 6519
+    },
+    {
+      "epoch": 0.8440948959445901,
+      "grad_norm": 1.8918982052194862,
+      "learning_rate": 9.027445826744565e-08,
+      "loss": 2.3977,
+      "step": 6520
+    },
+    {
+      "epoch": 0.8440948959445901,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.4992897510528564,
+      "eval_runtime": 14.0505,
+      "eval_samples_per_second": 3.132,
+      "eval_steps_per_second": 0.214,
+      "step": 6520
+    },
+    {
+      "epoch": 0.8442243583519435,
+      "grad_norm": 3.2638395475813797,
+      "learning_rate": 9.012827407792413e-08,
+      "loss": 2.5859,
+      "step": 6521
+    },
+    {
+      "epoch": 0.844353820759297,
+      "grad_norm": 1.1584025231346888,
+      "learning_rate": 8.998219914679384e-08,
+      "loss": 2.2974,
+      "step": 6522
+    },
+    {
+      "epoch": 0.8444832831666504,
+      "grad_norm": 1.6002007773791291,
+      "learning_rate": 8.983623350389368e-08,
+      "loss": 2.2749,
+      "step": 6523
+    },
+    {
+      "epoch": 0.8446127455740039,
+      "grad_norm": 1.5895397898338828,
+      "learning_rate": 8.969037717904027e-08,
+      "loss": 2.3691,
+      "step": 6524
+    },
+    {
+      "epoch": 0.8447422079813575,
+      "grad_norm": 2.0815608784226822,
+      "learning_rate": 8.954463020202753e-08,
+      "loss": 2.5276,
+      "step": 6525
+    },
+    {
+      "epoch": 0.8447422079813575,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.4984685182571411,
+      "eval_runtime": 14.5154,
+      "eval_samples_per_second": 3.031,
+      "eval_steps_per_second": 0.207,
+      "step": 6525
+    },
+    {
+      "epoch": 0.8448716703887109,
+      "grad_norm": 2.3871542043188496,
+      "learning_rate": 8.93989926026273e-08,
+      "loss": 2.4087,
+      "step": 6526
+    },
+    {
+      "epoch": 0.8450011327960644,
+      "grad_norm": 2.0524388281270056,
+      "learning_rate": 8.925346441058878e-08,
+      "loss": 2.396,
+      "step": 6527
+    },
+    {
+      "epoch": 0.8451305952034178,
+      "grad_norm": 2.5308945057287793,
+      "learning_rate": 8.910804565563925e-08,
+      "loss": 2.2397,
+      "step": 6528
+    },
+    {
+      "epoch": 0.8452600576107713,
+      "grad_norm": 1.2536553032983182,
+      "learning_rate": 8.89627363674836e-08,
+      "loss": 2.1331,
+      "step": 6529
+    },
+    {
+      "epoch": 0.8453895200181247,
+      "grad_norm": 1.7762497606313208,
+      "learning_rate": 8.881753657580386e-08,
+      "loss": 2.2786,
+      "step": 6530
+    },
+    {
+      "epoch": 0.8453895200181247,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.49853515625,
+      "eval_runtime": 14.0646,
+      "eval_samples_per_second": 3.128,
+      "eval_steps_per_second": 0.213,
+      "step": 6530
+    },
+    {
+      "epoch": 0.8455189824254782,
+      "grad_norm": 2.964287127326882,
+      "learning_rate": 8.867244631026034e-08,
+      "loss": 2.3127,
+      "step": 6531
+    },
+    {
+      "epoch": 0.8456484448328316,
+      "grad_norm": 3.319152351668377,
+      "learning_rate": 8.852746560049051e-08,
+      "loss": 2.4539,
+      "step": 6532
+    },
+    {
+      "epoch": 0.8457779072401851,
+      "grad_norm": 1.6225686188504977,
+      "learning_rate": 8.83825944761095e-08,
+      "loss": 2.3101,
+      "step": 6533
+    },
+    {
+      "epoch": 0.8459073696475385,
+      "grad_norm": 1.7803476185344367,
+      "learning_rate": 8.823783296671044e-08,
+      "loss": 2.3115,
+      "step": 6534
+    },
+    {
+      "epoch": 0.8460368320548921,
+      "grad_norm": 2.045672519888185,
+      "learning_rate": 8.809318110186357e-08,
+      "loss": 2.2305,
+      "step": 6535
+    },
+    {
+      "epoch": 0.8460368320548921,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.501265048980713,
+      "eval_runtime": 13.5854,
+      "eval_samples_per_second": 3.239,
+      "eval_steps_per_second": 0.221,
+      "step": 6535
+    },
+    {
+      "epoch": 0.8461662944622456,
+      "grad_norm": 1.975831045634424,
+      "learning_rate": 8.79486389111172e-08,
+      "loss": 2.4536,
+      "step": 6536
+    },
+    {
+      "epoch": 0.846295756869599,
+      "grad_norm": 1.1154771045361036,
+      "learning_rate": 8.78042064239968e-08,
+      "loss": 2.28,
+      "step": 6537
+    },
+    {
+      "epoch": 0.8464252192769525,
+      "grad_norm": 2.992747146713637,
+      "learning_rate": 8.765988367000565e-08,
+      "loss": 2.3772,
+      "step": 6538
+    },
+    {
+      "epoch": 0.8465546816843059,
+      "grad_norm": 1.3438032785430103,
+      "learning_rate": 8.751567067862475e-08,
+      "loss": 2.3877,
+      "step": 6539
+    },
+    {
+      "epoch": 0.8466841440916594,
+      "grad_norm": 1.292350606048276,
+      "learning_rate": 8.737156747931227e-08,
+      "loss": 2.3428,
+      "step": 6540
+    },
+    {
+      "epoch": 0.8466841440916594,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "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": 13.6721,
+      "eval_samples_per_second": 3.218,
+      "eval_steps_per_second": 0.219,
+      "step": 6540
+    },
+    {
+      "epoch": 0.8468136064990128,
+      "grad_norm": 1.3521108405047466,
+      "learning_rate": 8.722757410150453e-08,
+      "loss": 2.2734,
+      "step": 6541
+    },
+    {
+      "epoch": 0.8469430689063663,
+      "grad_norm": 2.4058565450452765,
+      "learning_rate": 8.708369057461488e-08,
+      "loss": 2.4351,
+      "step": 6542
+    },
+    {
+      "epoch": 0.8470725313137197,
+      "grad_norm": 1.7071696905699472,
+      "learning_rate": 8.693991692803439e-08,
+      "loss": 2.4612,
+      "step": 6543
+    },
+    {
+      "epoch": 0.8472019937210733,
+      "grad_norm": 1.8158029047366344,
+      "learning_rate": 8.679625319113205e-08,
+      "loss": 2.4351,
+      "step": 6544
+    },
+    {
+      "epoch": 0.8473314561284268,
+      "grad_norm": 2.308366482130721,
+      "learning_rate": 8.665269939325365e-08,
+      "loss": 2.2588,
+      "step": 6545
+    },
+    {
+      "epoch": 0.8473314561284268,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.497025966644287,
+      "eval_runtime": 13.4986,
+      "eval_samples_per_second": 3.26,
+      "eval_steps_per_second": 0.222,
+      "step": 6545
+    },
+    {
+      "epoch": 0.8474609185357802,
+      "grad_norm": 4.451751948532977,
+      "learning_rate": 8.650925556372328e-08,
+      "loss": 2.3818,
+      "step": 6546
+    },
+    {
+      "epoch": 0.8475903809431337,
+      "grad_norm": 2.2877676550449353,
+      "learning_rate": 8.636592173184237e-08,
+      "loss": 2.4968,
+      "step": 6547
+    },
+    {
+      "epoch": 0.8477198433504871,
+      "grad_norm": 1.5277204605513772,
+      "learning_rate": 8.622269792688938e-08,
+      "loss": 2.4919,
+      "step": 6548
+    },
+    {
+      "epoch": 0.8478493057578406,
+      "grad_norm": 1.5636627127805616,
+      "learning_rate": 8.60795841781209e-08,
+      "loss": 2.2617,
+      "step": 6549
+    },
+    {
+      "epoch": 0.847978768165194,
+      "grad_norm": 4.2296920407784375,
+      "learning_rate": 8.59365805147707e-08,
+      "loss": 2.3198,
+      "step": 6550
+    },
+    {
+      "epoch": 0.847978768165194,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "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": 14.1425,
+      "eval_samples_per_second": 3.111,
+      "eval_steps_per_second": 0.212,
+      "step": 6550
+    },
+    {
+      "epoch": 0.8481082305725475,
+      "grad_norm": 1.332680801780174,
+      "learning_rate": 8.579368696605021e-08,
+      "loss": 2.3081,
+      "step": 6551
+    },
+    {
+      "epoch": 0.8482376929799009,
+      "grad_norm": 2.779204567943503,
+      "learning_rate": 8.565090356114851e-08,
+      "loss": 2.2917,
+      "step": 6552
+    },
+    {
+      "epoch": 0.8483671553872544,
+      "grad_norm": 4.235294498534441,
+      "learning_rate": 8.550823032923191e-08,
+      "loss": 2.3481,
+      "step": 6553
+    },
+    {
+      "epoch": 0.8484966177946079,
+      "grad_norm": 2.5930879672563005,
+      "learning_rate": 8.536566729944406e-08,
+      "loss": 2.3984,
+      "step": 6554
+    },
+    {
+      "epoch": 0.8486260802019614,
+      "grad_norm": 1.6926972085102159,
+      "learning_rate": 8.522321450090673e-08,
+      "loss": 2.417,
+      "step": 6555
+    },
+    {
+      "epoch": 0.8486260802019614,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.498579502105713,
+      "eval_runtime": 15.1432,
+      "eval_samples_per_second": 2.906,
+      "eval_steps_per_second": 0.198,
+      "step": 6555
+    },
+    {
+      "epoch": 0.8487555426093149,
+      "grad_norm": 1.4755699823751198,
+      "learning_rate": 8.508087196271851e-08,
+      "loss": 2.4583,
+      "step": 6556
+    },
+    {
+      "epoch": 0.8488850050166683,
+      "grad_norm": 1.6616283466143034,
+      "learning_rate": 8.493863971395594e-08,
+      "loss": 2.3877,
+      "step": 6557
+    },
+    {
+      "epoch": 0.8490144674240218,
+      "grad_norm": 1.9845445460265752,
+      "learning_rate": 8.479651778367266e-08,
+      "loss": 2.5459,
+      "step": 6558
+    },
+    {
+      "epoch": 0.8491439298313752,
+      "grad_norm": 1.4785273182922332,
+      "learning_rate": 8.465450620090014e-08,
+      "loss": 2.2258,
+      "step": 6559
+    },
+    {
+      "epoch": 0.8492733922387287,
+      "grad_norm": 2.0323793655426,
+      "learning_rate": 8.451260499464701e-08,
+      "loss": 2.4294,
+      "step": 6560
+    },
+    {
+      "epoch": 0.8492733922387287,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "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": 13.9743,
+      "eval_samples_per_second": 3.149,
+      "eval_steps_per_second": 0.215,
+      "step": 6560
+    },
+    {
+      "epoch": 0.8494028546460821,
+      "grad_norm": 1.684013440587913,
+      "learning_rate": 8.437081419389939e-08,
+      "loss": 2.3379,
+      "step": 6561
+    },
+    {
+      "epoch": 0.8495323170534356,
+      "grad_norm": 1.6807115087927769,
+      "learning_rate": 8.422913382762119e-08,
+      "loss": 2.3931,
+      "step": 6562
+    },
+    {
+      "epoch": 0.849661779460789,
+      "grad_norm": 1.4753970675663166,
+      "learning_rate": 8.40875639247532e-08,
+      "loss": 2.1882,
+      "step": 6563
+    },
+    {
+      "epoch": 0.8497912418681426,
+      "grad_norm": 1.8688579009748316,
+      "learning_rate": 8.394610451421419e-08,
+      "loss": 2.436,
+      "step": 6564
+    },
+    {
+      "epoch": 0.849920704275496,
+      "grad_norm": 2.2596132532911466,
+      "learning_rate": 8.380475562489999e-08,
+      "loss": 2.2864,
+      "step": 6565
+    },
+    {
+      "epoch": 0.849920704275496,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.496382236480713,
+      "eval_runtime": 15.4174,
+      "eval_samples_per_second": 2.854,
+      "eval_steps_per_second": 0.195,
+      "step": 6565
+    },
+    {
+      "epoch": 0.8500501666828495,
+      "grad_norm": 2.237402806840521,
+      "learning_rate": 8.366351728568381e-08,
+      "loss": 2.79,
+      "step": 6566
+    },
+    {
+      "epoch": 0.850179629090203,
+      "grad_norm": 1.8105416369856966,
+      "learning_rate": 8.352238952541677e-08,
+      "loss": 2.344,
+      "step": 6567
+    },
+    {
+      "epoch": 0.8503090914975564,
+      "grad_norm": 1.9674130505675813,
+      "learning_rate": 8.338137237292682e-08,
+      "loss": 2.3169,
+      "step": 6568
+    },
+    {
+      "epoch": 0.8504385539049099,
+      "grad_norm": 1.4080536621367579,
+      "learning_rate": 8.324046585701961e-08,
+      "loss": 2.4047,
+      "step": 6569
+    },
+    {
+      "epoch": 0.8505680163122633,
+      "grad_norm": 1.2238358240747322,
+      "learning_rate": 8.309967000647859e-08,
+      "loss": 2.4175,
+      "step": 6570
+    },
+    {
+      "epoch": 0.8505680163122633,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.497758388519287,
+      "eval_runtime": 13.8179,
+      "eval_samples_per_second": 3.184,
+      "eval_steps_per_second": 0.217,
+      "step": 6570
+    },
+    {
+      "epoch": 0.8506974787196168,
+      "grad_norm": 1.4394485178164549,
+      "learning_rate": 8.295898485006344e-08,
+      "loss": 2.2981,
+      "step": 6571
+    },
+    {
+      "epoch": 0.8508269411269702,
+      "grad_norm": 2.8709935283050654,
+      "learning_rate": 8.281841041651244e-08,
+      "loss": 2.6245,
+      "step": 6572
+    },
+    {
+      "epoch": 0.8509564035343237,
+      "grad_norm": 2.0917170872260242,
+      "learning_rate": 8.267794673454052e-08,
+      "loss": 2.4221,
+      "step": 6573
+    },
+    {
+      "epoch": 0.8510858659416772,
+      "grad_norm": 1.6158921248350986,
+      "learning_rate": 8.253759383284037e-08,
+      "loss": 2.1921,
+      "step": 6574
+    },
+    {
+      "epoch": 0.8512153283490307,
+      "grad_norm": 1.9901127000146681,
+      "learning_rate": 8.239735174008209e-08,
+      "loss": 2.2681,
+      "step": 6575
+    },
+    {
+      "epoch": 0.8512153283490307,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.4957829713821411,
+      "eval_runtime": 14.1554,
+      "eval_samples_per_second": 3.108,
+      "eval_steps_per_second": 0.212,
+      "step": 6575
+    },
+    {
+      "epoch": 0.8513447907563841,
+      "grad_norm": 2.18364016590462,
+      "learning_rate": 8.225722048491258e-08,
+      "loss": 2.3252,
+      "step": 6576
+    },
+    {
+      "epoch": 0.8514742531637376,
+      "grad_norm": 1.4391037089945609,
+      "learning_rate": 8.211720009595663e-08,
+      "loss": 2.3042,
+      "step": 6577
+    },
+    {
+      "epoch": 0.851603715571091,
+      "grad_norm": 1.9606021614539075,
+      "learning_rate": 8.197729060181646e-08,
+      "loss": 2.4482,
+      "step": 6578
+    },
+    {
+      "epoch": 0.8517331779784445,
+      "grad_norm": 1.885092716148146,
+      "learning_rate": 8.183749203107105e-08,
+      "loss": 2.2485,
+      "step": 6579
+    },
+    {
+      "epoch": 0.851862640385798,
+      "grad_norm": 1.1966103608416532,
+      "learning_rate": 8.169780441227751e-08,
+      "loss": 2.428,
+      "step": 6580
+    },
+    {
+      "epoch": 0.851862640385798,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.4964710474014282,
+      "eval_runtime": 14.3668,
+      "eval_samples_per_second": 3.063,
+      "eval_steps_per_second": 0.209,
+      "step": 6580
+    },
+    {
+      "epoch": 0.8519921027931514,
+      "grad_norm": 1.2957432268188154,
+      "learning_rate": 8.15582277739696e-08,
+      "loss": 2.1587,
+      "step": 6581
+    },
+    {
+      "epoch": 0.8521215652005049,
+      "grad_norm": 2.176315218841233,
+      "learning_rate": 8.141876214465862e-08,
+      "loss": 2.4551,
+      "step": 6582
+    },
+    {
+      "epoch": 0.8522510276078583,
+      "grad_norm": 2.019386925982672,
+      "learning_rate": 8.12794075528335e-08,
+      "loss": 2.4011,
+      "step": 6583
+    },
+    {
+      "epoch": 0.8523804900152119,
+      "grad_norm": 3.1691583603241917,
+      "learning_rate": 8.114016402695997e-08,
+      "loss": 2.6421,
+      "step": 6584
+    },
+    {
+      "epoch": 0.8525099524225653,
+      "grad_norm": 2.134935088096232,
+      "learning_rate": 8.100103159548165e-08,
+      "loss": 2.3665,
+      "step": 6585
+    },
+    {
+      "epoch": 0.8525099524225653,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.4957829713821411,
+      "eval_runtime": 20.7121,
+      "eval_samples_per_second": 2.124,
+      "eval_steps_per_second": 0.145,
+      "step": 6585
+    },
+    {
+      "epoch": 0.8526394148299188,
+      "grad_norm": 1.4413197206552746,
+      "learning_rate": 8.086201028681892e-08,
+      "loss": 2.4827,
+      "step": 6586
+    },
+    {
+      "epoch": 0.8527688772372722,
+      "grad_norm": 1.19904193420047,
+      "learning_rate": 8.072310012936964e-08,
+      "loss": 2.291,
+      "step": 6587
+    },
+    {
+      "epoch": 0.8528983396446257,
+      "grad_norm": 1.8884830551462888,
+      "learning_rate": 8.05843011515093e-08,
+      "loss": 2.396,
+      "step": 6588
+    },
+    {
+      "epoch": 0.8530278020519791,
+      "grad_norm": 2.1765376138593675,
+      "learning_rate": 8.044561338159007e-08,
+      "loss": 2.2468,
+      "step": 6589
+    },
+    {
+      "epoch": 0.8531572644593326,
+      "grad_norm": 2.645903634177425,
+      "learning_rate": 8.030703684794208e-08,
+      "loss": 2.3191,
+      "step": 6590
+    },
+    {
+      "epoch": 0.8531572644593326,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.497270107269287,
+      "eval_runtime": 20.6437,
+      "eval_samples_per_second": 2.131,
+      "eval_steps_per_second": 0.145,
+      "step": 6590
+    },
+    {
+      "epoch": 0.853286726866686,
+      "grad_norm": 1.7761040813914157,
+      "learning_rate": 8.016857157887214e-08,
+      "loss": 2.3232,
+      "step": 6591
+    },
+    {
+      "epoch": 0.8534161892740395,
+      "grad_norm": 2.907370819672574,
+      "learning_rate": 8.003021760266482e-08,
+      "loss": 2.4331,
+      "step": 6592
+    },
+    {
+      "epoch": 0.8535456516813931,
+      "grad_norm": 1.9621918112269952,
+      "learning_rate": 7.989197494758155e-08,
+      "loss": 2.4325,
+      "step": 6593
+    },
+    {
+      "epoch": 0.8536751140887465,
+      "grad_norm": 1.1107233976492266,
+      "learning_rate": 7.975384364186108e-08,
+      "loss": 2.2961,
+      "step": 6594
+    },
+    {
+      "epoch": 0.8538045764961,
+      "grad_norm": 1.8042133984389783,
+      "learning_rate": 7.961582371371984e-08,
+      "loss": 2.2388,
+      "step": 6595
+    },
+    {
+      "epoch": 0.8538045764961,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.4979137182235718,
+      "eval_runtime": 14.1021,
+      "eval_samples_per_second": 3.12,
+      "eval_steps_per_second": 0.213,
+      "step": 6595
+    },
+    {
+      "epoch": 0.8539340389034534,
+      "grad_norm": 1.4439991292614032,
+      "learning_rate": 7.947791519135092e-08,
+      "loss": 2.3792,
+      "step": 6596
+    },
+    {
+      "epoch": 0.8540635013108069,
+      "grad_norm": 1.2676295708067427,
+      "learning_rate": 7.934011810292505e-08,
+      "loss": 2.3293,
+      "step": 6597
+    },
+    {
+      "epoch": 0.8541929637181603,
+      "grad_norm": 1.4310416848210628,
+      "learning_rate": 7.92024324765904e-08,
+      "loss": 2.4097,
+      "step": 6598
+    },
+    {
+      "epoch": 0.8543224261255138,
+      "grad_norm": 1.4984133539034448,
+      "learning_rate": 7.906485834047135e-08,
+      "loss": 2.1934,
+      "step": 6599
+    },
+    {
+      "epoch": 0.8544518885328672,
+      "grad_norm": 2.6466724781204167,
+      "learning_rate": 7.892739572267064e-08,
+      "loss": 2.5369,
+      "step": 6600
+    },
+    {
+      "epoch": 0.8544518885328672,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.49853515625,
+      "eval_runtime": 13.8267,
+      "eval_samples_per_second": 3.182,
+      "eval_steps_per_second": 0.217,
+      "step": 6600
+    },
+    {
+      "epoch": 0.8545813509402207,
+      "grad_norm": 1.91746015441227,
+      "learning_rate": 7.879004465126786e-08,
+      "loss": 2.3949,
+      "step": 6601
+    },
+    {
+      "epoch": 0.8547108133475741,
+      "grad_norm": 2.3182520639332855,
+      "learning_rate": 7.865280515431957e-08,
+      "loss": 2.27,
+      "step": 6602
+    },
+    {
+      "epoch": 0.8548402757549277,
+      "grad_norm": 1.5024747837125307,
+      "learning_rate": 7.851567725985997e-08,
+      "loss": 2.4561,
+      "step": 6603
+    },
+    {
+      "epoch": 0.8549697381622812,
+      "grad_norm": 1.4492388772804083,
+      "learning_rate": 7.837866099589974e-08,
+      "loss": 2.4192,
+      "step": 6604
+    },
+    {
+      "epoch": 0.8550992005696346,
+      "grad_norm": 3.33924252541541,
+      "learning_rate": 7.824175639042759e-08,
+      "loss": 2.3782,
+      "step": 6605
+    },
+    {
+      "epoch": 0.8550992005696346,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "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": 13.8932,
+      "eval_samples_per_second": 3.167,
+      "eval_steps_per_second": 0.216,
+      "step": 6605
+    },
+    {
+      "epoch": 0.8552286629769881,
+      "grad_norm": 3.3559383969194463,
+      "learning_rate": 7.810496347140907e-08,
+      "loss": 2.5371,
+      "step": 6606
+    },
+    {
+      "epoch": 0.8553581253843415,
+      "grad_norm": 1.3101877312417733,
+      "learning_rate": 7.796828226678668e-08,
+      "loss": 2.2766,
+      "step": 6607
+    },
+    {
+      "epoch": 0.855487587791695,
+      "grad_norm": 1.3990719601509942,
+      "learning_rate": 7.78317128044806e-08,
+      "loss": 2.4614,
+      "step": 6608
+    },
+    {
+      "epoch": 0.8556170501990484,
+      "grad_norm": 1.6327505814759302,
+      "learning_rate": 7.76952551123878e-08,
+      "loss": 2.3713,
+      "step": 6609
+    },
+    {
+      "epoch": 0.8557465126064019,
+      "grad_norm": 1.8005477359662325,
+      "learning_rate": 7.755890921838233e-08,
+      "loss": 2.479,
+      "step": 6610
+    },
+    {
+      "epoch": 0.8557465126064019,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.4964045286178589,
+      "eval_runtime": 15.0772,
+      "eval_samples_per_second": 2.918,
+      "eval_steps_per_second": 0.199,
+      "step": 6610
+    },
+    {
+      "epoch": 0.8558759750137553,
+      "grad_norm": 1.8673360100529317,
+      "learning_rate": 7.742267515031598e-08,
+      "loss": 2.2236,
+      "step": 6611
+    },
+    {
+      "epoch": 0.8560054374211088,
+      "grad_norm": 1.488865352619389,
+      "learning_rate": 7.728655293601693e-08,
+      "loss": 2.323,
+      "step": 6612
+    },
+    {
+      "epoch": 0.8561348998284624,
+      "grad_norm": 2.4792871443144997,
+      "learning_rate": 7.715054260329119e-08,
+      "loss": 2.4717,
+      "step": 6613
+    },
+    {
+      "epoch": 0.8562643622358158,
+      "grad_norm": 2.7874324571707567,
+      "learning_rate": 7.70146441799216e-08,
+      "loss": 2.3867,
+      "step": 6614
+    },
+    {
+      "epoch": 0.8563938246431693,
+      "grad_norm": 2.0496795992850987,
+      "learning_rate": 7.687885769366791e-08,
+      "loss": 2.4673,
+      "step": 6615
+    },
+    {
+      "epoch": 0.8563938246431693,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "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": 13.8288,
+      "eval_samples_per_second": 3.182,
+      "eval_steps_per_second": 0.217,
+      "step": 6615
+    },
+    {
+      "epoch": 0.8565232870505227,
+      "grad_norm": 2.2504258093393696,
+      "learning_rate": 7.674318317226759e-08,
+      "loss": 2.2751,
+      "step": 6616
+    },
+    {
+      "epoch": 0.8566527494578762,
+      "grad_norm": 1.783309652459419,
+      "learning_rate": 7.660762064343463e-08,
+      "loss": 2.2981,
+      "step": 6617
+    },
+    {
+      "epoch": 0.8567822118652296,
+      "grad_norm": 1.2477501546588274,
+      "learning_rate": 7.647217013486058e-08,
+      "loss": 2.2598,
+      "step": 6618
+    },
+    {
+      "epoch": 0.8569116742725831,
+      "grad_norm": 1.599085079126057,
+      "learning_rate": 7.633683167421406e-08,
+      "loss": 2.2395,
+      "step": 6619
+    },
+    {
+      "epoch": 0.8570411366799365,
+      "grad_norm": 1.5007528903720948,
+      "learning_rate": 7.620160528914053e-08,
+      "loss": 2.4846,
+      "step": 6620
+    },
+    {
+      "epoch": 0.8570411366799365,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.4957386255264282,
+      "eval_runtime": 14.4218,
+      "eval_samples_per_second": 3.051,
+      "eval_steps_per_second": 0.208,
+      "step": 6620
+    },
+    {
+      "epoch": 0.85717059908729,
+      "grad_norm": 2.247987149619524,
+      "learning_rate": 7.606649100726283e-08,
+      "loss": 2.4062,
+      "step": 6621
+    },
+    {
+      "epoch": 0.8573000614946434,
+      "grad_norm": 2.011566519304433,
+      "learning_rate": 7.593148885618052e-08,
+      "loss": 2.6294,
+      "step": 6622
+    },
+    {
+      "epoch": 0.857429523901997,
+      "grad_norm": 2.3806236406277876,
+      "learning_rate": 7.579659886347076e-08,
+      "loss": 2.4233,
+      "step": 6623
+    },
+    {
+      "epoch": 0.8575589863093505,
+      "grad_norm": 1.0506828312725818,
+      "learning_rate": 7.566182105668763e-08,
+      "loss": 2.071,
+      "step": 6624
+    },
+    {
+      "epoch": 0.8576884487167039,
+      "grad_norm": 1.8997382080487009,
+      "learning_rate": 7.552715546336214e-08,
+      "loss": 2.5034,
+      "step": 6625
+    },
+    {
+      "epoch": 0.8576884487167039,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "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": 14.5319,
+      "eval_samples_per_second": 3.028,
+      "eval_steps_per_second": 0.206,
+      "step": 6625
+    },
+    {
+      "epoch": 0.8578179111240574,
+      "grad_norm": 1.1481669993995705,
+      "learning_rate": 7.539260211100234e-08,
+      "loss": 2.1968,
+      "step": 6626
+    },
+    {
+      "epoch": 0.8579473735314108,
+      "grad_norm": 1.490292122150186,
+      "learning_rate": 7.525816102709367e-08,
+      "loss": 2.3159,
+      "step": 6627
+    },
+    {
+      "epoch": 0.8580768359387643,
+      "grad_norm": 1.4973001998170343,
+      "learning_rate": 7.512383223909833e-08,
+      "loss": 2.4431,
+      "step": 6628
+    },
+    {
+      "epoch": 0.8582062983461177,
+      "grad_norm": 1.6000629422857742,
+      "learning_rate": 7.498961577445577e-08,
+      "loss": 2.2461,
+      "step": 6629
+    },
+    {
+      "epoch": 0.8583357607534712,
+      "grad_norm": 2.0629435424889464,
+      "learning_rate": 7.48555116605823e-08,
+      "loss": 2.7925,
+      "step": 6630
+    },
+    {
+      "epoch": 0.8583357607534712,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.4951171875,
+      "eval_runtime": 14.859,
+      "eval_samples_per_second": 2.961,
+      "eval_steps_per_second": 0.202,
+      "step": 6630
+    },
+    {
+      "epoch": 0.8584652231608246,
+      "grad_norm": 1.8964111486105253,
+      "learning_rate": 7.472151992487166e-08,
+      "loss": 2.448,
+      "step": 6631
+    },
+    {
+      "epoch": 0.8585946855681782,
+      "grad_norm": 2.2718942043757604,
+      "learning_rate": 7.458764059469415e-08,
+      "loss": 2.2598,
+      "step": 6632
+    },
+    {
+      "epoch": 0.8587241479755316,
+      "grad_norm": 1.9176820788679079,
+      "learning_rate": 7.44538736973973e-08,
+      "loss": 2.2473,
+      "step": 6633
+    },
+    {
+      "epoch": 0.8588536103828851,
+      "grad_norm": 1.5155302775959094,
+      "learning_rate": 7.432021926030598e-08,
+      "loss": 2.3923,
+      "step": 6634
+    },
+    {
+      "epoch": 0.8589830727902386,
+      "grad_norm": 4.963229203085601,
+      "learning_rate": 7.418667731072144e-08,
+      "loss": 2.6514,
+      "step": 6635
+    },
+    {
+      "epoch": 0.8589830727902386,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.4951171875,
+      "eval_runtime": 13.5948,
+      "eval_samples_per_second": 3.237,
+      "eval_steps_per_second": 0.221,
+      "step": 6635
+    },
+    {
+      "epoch": 0.859112535197592,
+      "grad_norm": 1.919372632336787,
+      "learning_rate": 7.405324787592274e-08,
+      "loss": 2.3503,
+      "step": 6636
+    },
+    {
+      "epoch": 0.8592419976049455,
+      "grad_norm": 1.3871793180860774,
+      "learning_rate": 7.391993098316529e-08,
+      "loss": 2.1494,
+      "step": 6637
+    },
+    {
+      "epoch": 0.8593714600122989,
+      "grad_norm": 1.6644484785152112,
+      "learning_rate": 7.378672665968172e-08,
+      "loss": 2.2834,
+      "step": 6638
+    },
+    {
+      "epoch": 0.8595009224196524,
+      "grad_norm": 1.8542750543387743,
+      "learning_rate": 7.36536349326819e-08,
+      "loss": 2.4065,
+      "step": 6639
+    },
+    {
+      "epoch": 0.8596303848270058,
+      "grad_norm": 1.4092349047145702,
+      "learning_rate": 7.352065582935233e-08,
+      "loss": 2.3184,
+      "step": 6640
+    },
+    {
+      "epoch": 0.8596303848270058,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.4972034692764282,
+      "eval_runtime": 13.8724,
+      "eval_samples_per_second": 3.172,
+      "eval_steps_per_second": 0.216,
+      "step": 6640
+    },
+    {
+      "epoch": 0.8597598472343593,
+      "grad_norm": 1.5609488746223876,
+      "learning_rate": 7.338778937685677e-08,
+      "loss": 2.4194,
+      "step": 6641
+    },
+    {
+      "epoch": 0.8598893096417128,
+      "grad_norm": 1.409179470791384,
+      "learning_rate": 7.325503560233622e-08,
+      "loss": 2.1055,
+      "step": 6642
+    },
+    {
+      "epoch": 0.8600187720490663,
+      "grad_norm": 1.2215535484344948,
+      "learning_rate": 7.312239453290772e-08,
+      "loss": 2.2432,
+      "step": 6643
+    },
+    {
+      "epoch": 0.8601482344564197,
+      "grad_norm": 1.4873162733115775,
+      "learning_rate": 7.298986619566628e-08,
+      "loss": 2.3604,
+      "step": 6644
+    },
+    {
+      "epoch": 0.8602776968637732,
+      "grad_norm": 2.122053465202789,
+      "learning_rate": 7.28574506176833e-08,
+      "loss": 2.3068,
+      "step": 6645
+    },
+    {
+      "epoch": 0.8602776968637732,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.4979137182235718,
+      "eval_runtime": 14.5146,
+      "eval_samples_per_second": 3.031,
+      "eval_steps_per_second": 0.207,
+      "step": 6645
+    },
+    {
+      "epoch": 0.8604071592711267,
+      "grad_norm": 3.1898086590477694,
+      "learning_rate": 7.272514782600742e-08,
+      "loss": 2.4614,
+      "step": 6646
+    },
+    {
+      "epoch": 0.8605366216784801,
+      "grad_norm": 1.3091026421283558,
+      "learning_rate": 7.259295784766431e-08,
+      "loss": 2.3464,
+      "step": 6647
+    },
+    {
+      "epoch": 0.8606660840858336,
+      "grad_norm": 1.3026709897836444,
+      "learning_rate": 7.246088070965641e-08,
+      "loss": 2.3071,
+      "step": 6648
+    },
+    {
+      "epoch": 0.860795546493187,
+      "grad_norm": 1.72972720740264,
+      "learning_rate": 7.232891643896296e-08,
+      "loss": 2.3364,
+      "step": 6649
+    },
+    {
+      "epoch": 0.8609250089005405,
+      "grad_norm": 3.2077939199248324,
+      "learning_rate": 7.219706506254052e-08,
+      "loss": 2.2295,
+      "step": 6650
+    },
+    {
+      "epoch": 0.8609250089005405,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.4962712526321411,
+      "eval_runtime": 13.3678,
+      "eval_samples_per_second": 3.291,
+      "eval_steps_per_second": 0.224,
+      "step": 6650
+    },
+    {
+      "epoch": 0.8610544713078939,
+      "grad_norm": 1.3060665403929674,
+      "learning_rate": 7.206532660732227e-08,
+      "loss": 2.3586,
+      "step": 6651
+    },
+    {
+      "epoch": 0.8611839337152475,
+      "grad_norm": 1.7815142420066077,
+      "learning_rate": 7.193370110021865e-08,
+      "loss": 2.2949,
+      "step": 6652
+    },
+    {
+      "epoch": 0.8613133961226009,
+      "grad_norm": 1.5681106973761263,
+      "learning_rate": 7.180218856811673e-08,
+      "loss": 2.2166,
+      "step": 6653
+    },
+    {
+      "epoch": 0.8614428585299544,
+      "grad_norm": 1.3938777034923548,
+      "learning_rate": 7.167078903788045e-08,
+      "loss": 2.3921,
+      "step": 6654
+    },
+    {
+      "epoch": 0.8615723209373078,
+      "grad_norm": 4.808381169883879,
+      "learning_rate": 7.153950253635113e-08,
+      "loss": 2.6716,
+      "step": 6655
+    },
+    {
+      "epoch": 0.8615723209373078,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.498579502105713,
+      "eval_runtime": 15.08,
+      "eval_samples_per_second": 2.918,
+      "eval_steps_per_second": 0.199,
+      "step": 6655
+    },
+    {
+      "epoch": 0.8617017833446613,
+      "grad_norm": 1.5097288496512302,
+      "learning_rate": 7.140832909034639e-08,
+      "loss": 2.3577,
+      "step": 6656
+    },
+    {
+      "epoch": 0.8618312457520148,
+      "grad_norm": 2.8709418748852906,
+      "learning_rate": 7.127726872666135e-08,
+      "loss": 2.3718,
+      "step": 6657
+    },
+    {
+      "epoch": 0.8619607081593682,
+      "grad_norm": 1.9944739230712052,
+      "learning_rate": 7.114632147206758e-08,
+      "loss": 2.4968,
+      "step": 6658
+    },
+    {
+      "epoch": 0.8620901705667217,
+      "grad_norm": 2.219423702053198,
+      "learning_rate": 7.101548735331378e-08,
+      "loss": 2.2529,
+      "step": 6659
+    },
+    {
+      "epoch": 0.8622196329740751,
+      "grad_norm": 1.7419824075160113,
+      "learning_rate": 7.088476639712551e-08,
+      "loss": 2.4932,
+      "step": 6660
+    },
+    {
+      "epoch": 0.8622196329740751,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.495649814605713,
+      "eval_runtime": 13.8326,
+      "eval_samples_per_second": 3.181,
+      "eval_steps_per_second": 0.217,
+      "step": 6660
+    },
+    {
+      "epoch": 0.8623490953814286,
+      "grad_norm": 3.0449880880269626,
+      "learning_rate": 7.075415863020497e-08,
+      "loss": 2.375,
+      "step": 6661
+    },
+    {
+      "epoch": 0.8624785577887821,
+      "grad_norm": 1.7712876580479961,
+      "learning_rate": 7.062366407923168e-08,
+      "loss": 2.5535,
+      "step": 6662
+    },
+    {
+      "epoch": 0.8626080201961356,
+      "grad_norm": 1.4868857233321155,
+      "learning_rate": 7.049328277086154e-08,
+      "loss": 2.4163,
+      "step": 6663
+    },
+    {
+      "epoch": 0.862737482603489,
+      "grad_norm": 2.828166011141021,
+      "learning_rate": 7.036301473172794e-08,
+      "loss": 2.4304,
+      "step": 6664
+    },
+    {
+      "epoch": 0.8628669450108425,
+      "grad_norm": 1.9820238185600705,
+      "learning_rate": 7.023285998844054e-08,
+      "loss": 2.4917,
+      "step": 6665
+    },
+    {
+      "epoch": 0.8628669450108425,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.499267578125,
+      "eval_runtime": 13.9697,
+      "eval_samples_per_second": 3.15,
+      "eval_steps_per_second": 0.215,
+      "step": 6665
+    },
+    {
+      "epoch": 0.8629964074181959,
+      "grad_norm": 1.643282848319218,
+      "learning_rate": 7.010281856758596e-08,
+      "loss": 2.2725,
+      "step": 6666
+    },
+    {
+      "epoch": 0.8631258698255494,
+      "grad_norm": 3.094323303257011,
+      "learning_rate": 6.997289049572802e-08,
+      "loss": 2.2539,
+      "step": 6667
+    },
+    {
+      "epoch": 0.8632553322329029,
+      "grad_norm": 1.5211016580596106,
+      "learning_rate": 6.984307579940702e-08,
+      "loss": 2.3838,
+      "step": 6668
+    },
+    {
+      "epoch": 0.8633847946402563,
+      "grad_norm": 1.6884303073669678,
+      "learning_rate": 6.97133745051403e-08,
+      "loss": 2.2773,
+      "step": 6669
+    },
+    {
+      "epoch": 0.8635142570476098,
+      "grad_norm": 3.0494919365178546,
+      "learning_rate": 6.958378663942222e-08,
+      "loss": 2.2134,
+      "step": 6670
+    },
+    {
+      "epoch": 0.8635142570476098,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.4965153932571411,
+      "eval_runtime": 14.3478,
+      "eval_samples_per_second": 3.067,
+      "eval_steps_per_second": 0.209,
+      "step": 6670
+    },
+    {
+      "epoch": 0.8636437194549632,
+      "grad_norm": 3.2495713404244317,
+      "learning_rate": 6.945431222872328e-08,
+      "loss": 2.406,
+      "step": 6671
+    },
+    {
+      "epoch": 0.8637731818623168,
+      "grad_norm": 1.3206205069732337,
+      "learning_rate": 6.93249512994914e-08,
+      "loss": 2.4033,
+      "step": 6672
+    },
+    {
+      "epoch": 0.8639026442696702,
+      "grad_norm": 1.6309333651349487,
+      "learning_rate": 6.919570387815142e-08,
+      "loss": 2.4502,
+      "step": 6673
+    },
+    {
+      "epoch": 0.8640321066770237,
+      "grad_norm": 1.6215884944324817,
+      "learning_rate": 6.906656999110445e-08,
+      "loss": 2.3093,
+      "step": 6674
+    },
+    {
+      "epoch": 0.8641615690843771,
+      "grad_norm": 2.7176432304682074,
+      "learning_rate": 6.893754966472897e-08,
+      "loss": 2.2837,
+      "step": 6675
+    },
+    {
+      "epoch": 0.8641615690843771,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.498490810394287,
+      "eval_runtime": 15.4896,
+      "eval_samples_per_second": 2.841,
+      "eval_steps_per_second": 0.194,
+      "step": 6675
+    },
+    {
+      "epoch": 0.8642910314917306,
+      "grad_norm": 3.6147734963821443,
+      "learning_rate": 6.880864292537967e-08,
+      "loss": 2.3394,
+      "step": 6676
+    },
+    {
+      "epoch": 0.864420493899084,
+      "grad_norm": 1.9261856659341565,
+      "learning_rate": 6.867984979938853e-08,
+      "loss": 2.4463,
+      "step": 6677
+    },
+    {
+      "epoch": 0.8645499563064375,
+      "grad_norm": 1.2587535992701346,
+      "learning_rate": 6.855117031306431e-08,
+      "loss": 2.4731,
+      "step": 6678
+    },
+    {
+      "epoch": 0.864679418713791,
+      "grad_norm": 1.49590193668734,
+      "learning_rate": 6.842260449269209e-08,
+      "loss": 2.4614,
+      "step": 6679
+    },
+    {
+      "epoch": 0.8648088811211444,
+      "grad_norm": 2.341721144541663,
+      "learning_rate": 6.829415236453439e-08,
+      "loss": 2.6802,
+      "step": 6680
+    },
+    {
+      "epoch": 0.8648088811211444,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.4964267015457153,
+      "eval_runtime": 14.1092,
+      "eval_samples_per_second": 3.119,
+      "eval_steps_per_second": 0.213,
+      "step": 6680
+    },
+    {
+      "epoch": 0.864938343528498,
+      "grad_norm": 2.418843730853219,
+      "learning_rate": 6.81658139548299e-08,
+      "loss": 2.4199,
+      "step": 6681
+    },
+    {
+      "epoch": 0.8650678059358514,
+      "grad_norm": 6.258683298257318,
+      "learning_rate": 6.803758928979436e-08,
+      "loss": 3.4131,
+      "step": 6682
+    },
+    {
+      "epoch": 0.8651972683432049,
+      "grad_norm": 1.8592163292031774,
+      "learning_rate": 6.790947839562044e-08,
+      "loss": 2.353,
+      "step": 6683
+    },
+    {
+      "epoch": 0.8653267307505583,
+      "grad_norm": 3.149712906113063,
+      "learning_rate": 6.778148129847714e-08,
+      "loss": 2.4346,
+      "step": 6684
+    },
+    {
+      "epoch": 0.8654561931579118,
+      "grad_norm": 1.2635828765716453,
+      "learning_rate": 6.76535980245107e-08,
+      "loss": 2.2256,
+      "step": 6685
+    },
+    {
+      "epoch": 0.8654561931579118,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.4964932203292847,
+      "eval_runtime": 14.0168,
+      "eval_samples_per_second": 3.139,
+      "eval_steps_per_second": 0.214,
+      "step": 6685
+    },
+    {
+      "epoch": 0.8655856555652652,
+      "grad_norm": 2.123001861538419,
+      "learning_rate": 6.752582859984362e-08,
+      "loss": 2.4702,
+      "step": 6686
+    },
+    {
+      "epoch": 0.8657151179726187,
+      "grad_norm": 1.4411008922578532,
+      "learning_rate": 6.739817305057573e-08,
+      "loss": 2.3286,
+      "step": 6687
+    },
+    {
+      "epoch": 0.8658445803799721,
+      "grad_norm": 2.024492899125188,
+      "learning_rate": 6.727063140278302e-08,
+      "loss": 2.5676,
+      "step": 6688
+    },
+    {
+      "epoch": 0.8659740427873256,
+      "grad_norm": 1.2306431037272239,
+      "learning_rate": 6.714320368251835e-08,
+      "loss": 2.2649,
+      "step": 6689
+    },
+    {
+      "epoch": 0.866103505194679,
+      "grad_norm": 1.2588539105575596,
+      "learning_rate": 6.701588991581174e-08,
+      "loss": 2.394,
+      "step": 6690
+    },
+    {
+      "epoch": 0.866103505194679,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.4986017942428589,
+      "eval_runtime": 15.1974,
+      "eval_samples_per_second": 2.895,
+      "eval_steps_per_second": 0.197,
+      "step": 6690
+    },
+    {
+      "epoch": 0.8662329676020326,
+      "grad_norm": 1.426028810653981,
+      "learning_rate": 6.68886901286693e-08,
+      "loss": 2.4453,
+      "step": 6691
+    },
+    {
+      "epoch": 0.8663624300093861,
+      "grad_norm": 2.3146376481760207,
+      "learning_rate": 6.676160434707444e-08,
+      "loss": 2.4929,
+      "step": 6692
+    },
+    {
+      "epoch": 0.8664918924167395,
+      "grad_norm": 2.169533302902575,
+      "learning_rate": 6.663463259698688e-08,
+      "loss": 2.3918,
+      "step": 6693
+    },
+    {
+      "epoch": 0.866621354824093,
+      "grad_norm": 3.3913541751770624,
+      "learning_rate": 6.650777490434301e-08,
+      "loss": 2.6057,
+      "step": 6694
+    },
+    {
+      "epoch": 0.8667508172314464,
+      "grad_norm": 1.6259640231818908,
+      "learning_rate": 6.638103129505621e-08,
+      "loss": 2.21,
+      "step": 6695
+    },
+    {
+      "epoch": 0.8667508172314464,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.4958717823028564,
+      "eval_runtime": 14.3498,
+      "eval_samples_per_second": 3.066,
+      "eval_steps_per_second": 0.209,
+      "step": 6695
+    },
+    {
+      "epoch": 0.8668802796387999,
+      "grad_norm": 1.8977601147542396,
+      "learning_rate": 6.62544017950166e-08,
+      "loss": 2.2977,
+      "step": 6696
+    },
+    {
+      "epoch": 0.8670097420461533,
+      "grad_norm": 1.4919689917223402,
+      "learning_rate": 6.612788643009058e-08,
+      "loss": 2.541,
+      "step": 6697
+    },
+    {
+      "epoch": 0.8671392044535068,
+      "grad_norm": 1.918917568538702,
+      "learning_rate": 6.600148522612183e-08,
+      "loss": 2.395,
+      "step": 6698
+    },
+    {
+      "epoch": 0.8672686668608602,
+      "grad_norm": 2.445324326197817,
+      "learning_rate": 6.587519820892986e-08,
+      "loss": 2.5247,
+      "step": 6699
+    },
+    {
+      "epoch": 0.8673981292682137,
+      "grad_norm": 1.8177422252010311,
+      "learning_rate": 6.574902540431159e-08,
+      "loss": 2.2876,
+      "step": 6700
+    },
+    {
+      "epoch": 0.8673981292682137,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.4971369504928589,
+      "eval_runtime": 15.1819,
+      "eval_samples_per_second": 2.898,
+      "eval_steps_per_second": 0.198,
+      "step": 6700
+    },
+    {
+      "epoch": 0.8675275916755673,
+      "grad_norm": 1.4113452390545302,
+      "learning_rate": 6.56229668380405e-08,
+      "loss": 2.3835,
+      "step": 6701
+    },
+    {
+      "epoch": 0.8676570540829207,
+      "grad_norm": 2.9581728796058404,
+      "learning_rate": 6.549702253586636e-08,
+      "loss": 2.3403,
+      "step": 6702
+    },
+    {
+      "epoch": 0.8677865164902742,
+      "grad_norm": 1.6860442297937641,
+      "learning_rate": 6.537119252351613e-08,
+      "loss": 2.2998,
+      "step": 6703
+    },
+    {
+      "epoch": 0.8679159788976276,
+      "grad_norm": 3.3056083276683887,
+      "learning_rate": 6.524547682669295e-08,
+      "loss": 2.6072,
+      "step": 6704
+    },
+    {
+      "epoch": 0.8680454413049811,
+      "grad_norm": 1.3738842325250389,
+      "learning_rate": 6.511987547107669e-08,
+      "loss": 2.2642,
+      "step": 6705
+    },
+    {
+      "epoch": 0.8680454413049811,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.4956942796707153,
+      "eval_runtime": 13.7513,
+      "eval_samples_per_second": 3.2,
+      "eval_steps_per_second": 0.218,
+      "step": 6705
+    },
+    {
+      "epoch": 0.8681749037123345,
+      "grad_norm": 1.8744006896497827,
+      "learning_rate": 6.499438848232433e-08,
+      "loss": 2.4712,
+      "step": 6706
+    },
+    {
+      "epoch": 0.868304366119688,
+      "grad_norm": 1.7345172057711467,
+      "learning_rate": 6.486901588606875e-08,
+      "loss": 2.5117,
+      "step": 6707
+    },
+    {
+      "epoch": 0.8684338285270414,
+      "grad_norm": 4.480451394271401,
+      "learning_rate": 6.474375770792011e-08,
+      "loss": 2.3818,
+      "step": 6708
+    },
+    {
+      "epoch": 0.8685632909343949,
+      "grad_norm": 3.3076567599340265,
+      "learning_rate": 6.461861397346482e-08,
+      "loss": 2.334,
+      "step": 6709
+    },
+    {
+      "epoch": 0.8686927533417483,
+      "grad_norm": 2.23589586447218,
+      "learning_rate": 6.449358470826596e-08,
+      "loss": 2.2439,
+      "step": 6710
+    },
+    {
+      "epoch": 0.8686927533417483,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.4965598583221436,
+      "eval_runtime": 13.6012,
+      "eval_samples_per_second": 3.235,
+      "eval_steps_per_second": 0.221,
+      "step": 6710
+    },
+    {
+      "epoch": 0.8688222157491019,
+      "grad_norm": 1.7633218572490406,
+      "learning_rate": 6.436866993786352e-08,
+      "loss": 2.521,
+      "step": 6711
+    },
+    {
+      "epoch": 0.8689516781564554,
+      "grad_norm": 1.6903969599263293,
+      "learning_rate": 6.424386968777355e-08,
+      "loss": 2.365,
+      "step": 6712
+    },
+    {
+      "epoch": 0.8690811405638088,
+      "grad_norm": 2.494849740128448,
+      "learning_rate": 6.41191839834892e-08,
+      "loss": 2.261,
+      "step": 6713
+    },
+    {
+      "epoch": 0.8692106029711623,
+      "grad_norm": 1.7985873314679863,
+      "learning_rate": 6.399461285048037e-08,
+      "loss": 2.2898,
+      "step": 6714
+    },
+    {
+      "epoch": 0.8693400653785157,
+      "grad_norm": 1.1845342386497755,
+      "learning_rate": 6.38701563141926e-08,
+      "loss": 2.2742,
+      "step": 6715
+    },
+    {
+      "epoch": 0.8693400653785157,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.493696689605713,
+      "eval_runtime": 14.6525,
+      "eval_samples_per_second": 3.003,
+      "eval_steps_per_second": 0.205,
+      "step": 6715
+    },
+    {
+      "epoch": 0.8694695277858692,
+      "grad_norm": 2.8367812605520326,
+      "learning_rate": 6.374581440004912e-08,
+      "loss": 2.4934,
+      "step": 6716
+    },
+    {
+      "epoch": 0.8695989901932226,
+      "grad_norm": 1.297741587629704,
+      "learning_rate": 6.362158713344911e-08,
+      "loss": 2.3435,
+      "step": 6717
+    },
+    {
+      "epoch": 0.8697284526005761,
+      "grad_norm": 1.0787507772483806,
+      "learning_rate": 6.349747453976845e-08,
+      "loss": 2.2175,
+      "step": 6718
+    },
+    {
+      "epoch": 0.8698579150079295,
+      "grad_norm": 2.8228350545199863,
+      "learning_rate": 6.337347664435993e-08,
+      "loss": 2.395,
+      "step": 6719
+    },
+    {
+      "epoch": 0.8699873774152831,
+      "grad_norm": 2.1531608728093468,
+      "learning_rate": 6.324959347255244e-08,
+      "loss": 2.4441,
+      "step": 6720
+    },
+    {
+      "epoch": 0.8699873774152831,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.4956721067428589,
+      "eval_runtime": 13.705,
+      "eval_samples_per_second": 3.21,
+      "eval_steps_per_second": 0.219,
+      "step": 6720
+    },
+    {
+      "epoch": 0.8701168398226365,
+      "grad_norm": 1.5408867219279547,
+      "learning_rate": 6.312582504965158e-08,
+      "loss": 2.3599,
+      "step": 6721
+    },
+    {
+      "epoch": 0.87024630222999,
+      "grad_norm": 2.5062948685091047,
+      "learning_rate": 6.300217140093968e-08,
+      "loss": 2.4946,
+      "step": 6722
+    },
+    {
+      "epoch": 0.8703757646373435,
+      "grad_norm": 3.739691038319502,
+      "learning_rate": 6.287863255167532e-08,
+      "loss": 2.5645,
+      "step": 6723
+    },
+    {
+      "epoch": 0.8705052270446969,
+      "grad_norm": 1.9680246115938134,
+      "learning_rate": 6.27552085270941e-08,
+      "loss": 2.517,
+      "step": 6724
+    },
+    {
+      "epoch": 0.8706346894520504,
+      "grad_norm": 2.1199852540363575,
+      "learning_rate": 6.263189935240756e-08,
+      "loss": 2.2751,
+      "step": 6725
+    },
+    {
+      "epoch": 0.8706346894520504,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.4971591234207153,
+      "eval_runtime": 14.1899,
+      "eval_samples_per_second": 3.101,
+      "eval_steps_per_second": 0.211,
+      "step": 6725
+    },
+    {
+      "epoch": 0.8707641518594038,
+      "grad_norm": 1.9392810375556566,
+      "learning_rate": 6.250870505280441e-08,
+      "loss": 2.4363,
+      "step": 6726
+    },
+    {
+      "epoch": 0.8708936142667573,
+      "grad_norm": 1.470228373888008,
+      "learning_rate": 6.238562565344949e-08,
+      "loss": 2.2487,
+      "step": 6727
+    },
+    {
+      "epoch": 0.8710230766741107,
+      "grad_norm": 1.8541662871864155,
+      "learning_rate": 6.226266117948396e-08,
+      "loss": 2.5835,
+      "step": 6728
+    },
+    {
+      "epoch": 0.8711525390814642,
+      "grad_norm": 1.3166860730606136,
+      "learning_rate": 6.213981165602623e-08,
+      "loss": 2.2578,
+      "step": 6729
+    },
+    {
+      "epoch": 0.8712820014888177,
+      "grad_norm": 2.2233620269026555,
+      "learning_rate": 6.201707710817052e-08,
+      "loss": 2.2661,
+      "step": 6730
+    },
+    {
+      "epoch": 0.8712820014888177,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.500688076019287,
+      "eval_runtime": 14.6611,
+      "eval_samples_per_second": 3.001,
+      "eval_steps_per_second": 0.205,
+      "step": 6730
+    },
+    {
+      "epoch": 0.8714114638961712,
+      "grad_norm": 2.01201731499581,
+      "learning_rate": 6.189445756098811e-08,
+      "loss": 2.4124,
+      "step": 6731
+    },
+    {
+      "epoch": 0.8715409263035246,
+      "grad_norm": 1.4653296658699253,
+      "learning_rate": 6.177195303952631e-08,
+      "loss": 2.4182,
+      "step": 6732
+    },
+    {
+      "epoch": 0.8716703887108781,
+      "grad_norm": 1.295969178583342,
+      "learning_rate": 6.164956356880907e-08,
+      "loss": 2.1621,
+      "step": 6733
+    },
+    {
+      "epoch": 0.8717998511182316,
+      "grad_norm": 1.1827208234732027,
+      "learning_rate": 6.152728917383728e-08,
+      "loss": 2.2689,
+      "step": 6734
+    },
+    {
+      "epoch": 0.871929313525585,
+      "grad_norm": 1.9456854403210608,
+      "learning_rate": 6.140512987958751e-08,
+      "loss": 2.3398,
+      "step": 6735
+    },
+    {
+      "epoch": 0.871929313525585,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.4970703125,
+      "eval_runtime": 14.3118,
+      "eval_samples_per_second": 3.074,
+      "eval_steps_per_second": 0.21,
+      "step": 6735
+    },
+    {
+      "epoch": 0.8720587759329385,
+      "grad_norm": 1.4939730722222275,
+      "learning_rate": 6.12830857110136e-08,
+      "loss": 2.2427,
+      "step": 6736
+    },
+    {
+      "epoch": 0.8721882383402919,
+      "grad_norm": 1.5103520766035068,
+      "learning_rate": 6.116115669304569e-08,
+      "loss": 2.4314,
+      "step": 6737
+    },
+    {
+      "epoch": 0.8723177007476454,
+      "grad_norm": 1.8580683651981682,
+      "learning_rate": 6.103934285058975e-08,
+      "loss": 2.5381,
+      "step": 6738
+    },
+    {
+      "epoch": 0.8724471631549988,
+      "grad_norm": 2.1825185467165467,
+      "learning_rate": 6.091764420852908e-08,
+      "loss": 2.3928,
+      "step": 6739
+    },
+    {
+      "epoch": 0.8725766255623524,
+      "grad_norm": 1.994433186729432,
+      "learning_rate": 6.079606079172296e-08,
+      "loss": 2.3441,
+      "step": 6740
+    },
+    {
+      "epoch": 0.8725766255623524,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.4964932203292847,
+      "eval_runtime": 14.5621,
+      "eval_samples_per_second": 3.022,
+      "eval_steps_per_second": 0.206,
+      "step": 6740
+    },
+    {
+      "epoch": 0.8727060879697058,
+      "grad_norm": 1.641103674320105,
+      "learning_rate": 6.067459262500728e-08,
+      "loss": 2.3308,
+      "step": 6741
+    },
+    {
+      "epoch": 0.8728355503770593,
+      "grad_norm": 1.631849082225562,
+      "learning_rate": 6.055323973319469e-08,
+      "loss": 2.1851,
+      "step": 6742
+    },
+    {
+      "epoch": 0.8729650127844127,
+      "grad_norm": 1.616586959545823,
+      "learning_rate": 6.043200214107342e-08,
+      "loss": 2.2988,
+      "step": 6743
+    },
+    {
+      "epoch": 0.8730944751917662,
+      "grad_norm": 1.8153477763680517,
+      "learning_rate": 6.031087987340895e-08,
+      "loss": 2.3359,
+      "step": 6744
+    },
+    {
+      "epoch": 0.8732239375991196,
+      "grad_norm": 1.2726504952075033,
+      "learning_rate": 6.018987295494313e-08,
+      "loss": 2.3091,
+      "step": 6745
+    },
+    {
+      "epoch": 0.8732239375991196,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.4971591234207153,
+      "eval_runtime": 13.9031,
+      "eval_samples_per_second": 3.165,
+      "eval_steps_per_second": 0.216,
+      "step": 6745
+    },
+    {
+      "epoch": 0.8733534000064731,
+      "grad_norm": 4.465389909271955,
+      "learning_rate": 6.006898141039386e-08,
+      "loss": 2.4573,
+      "step": 6746
+    },
+    {
+      "epoch": 0.8734828624138266,
+      "grad_norm": 2.4611450773474592,
+      "learning_rate": 5.994820526445587e-08,
+      "loss": 2.6506,
+      "step": 6747
+    },
+    {
+      "epoch": 0.87361232482118,
+      "grad_norm": 1.7613136560798546,
+      "learning_rate": 5.982754454180007e-08,
+      "loss": 2.356,
+      "step": 6748
+    },
+    {
+      "epoch": 0.8737417872285335,
+      "grad_norm": 2.404495628018289,
+      "learning_rate": 5.970699926707361e-08,
+      "loss": 2.4065,
+      "step": 6749
+    },
+    {
+      "epoch": 0.873871249635887,
+      "grad_norm": 1.3126209330673706,
+      "learning_rate": 5.958656946490074e-08,
+      "loss": 2.3899,
+      "step": 6750
+    },
+    {
+      "epoch": 0.873871249635887,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.495161533355713,
+      "eval_runtime": 15.6843,
+      "eval_samples_per_second": 2.805,
+      "eval_steps_per_second": 0.191,
+      "step": 6750
+    },
+    {
+      "epoch": 0.8740007120432405,
+      "grad_norm": 2.2086063703509726,
+      "learning_rate": 5.946625515988125e-08,
+      "loss": 2.4753,
+      "step": 6751
+    },
+    {
+      "epoch": 0.8741301744505939,
+      "grad_norm": 1.3743774949038223,
+      "learning_rate": 5.9346056376592204e-08,
+      "loss": 2.3318,
+      "step": 6752
+    },
+    {
+      "epoch": 0.8742596368579474,
+      "grad_norm": 2.2715550593388385,
+      "learning_rate": 5.922597313958632e-08,
+      "loss": 2.533,
+      "step": 6753
+    },
+    {
+      "epoch": 0.8743890992653008,
+      "grad_norm": 1.7538062983161529,
+      "learning_rate": 5.910600547339308e-08,
+      "loss": 2.156,
+      "step": 6754
+    },
+    {
+      "epoch": 0.8745185616726543,
+      "grad_norm": 2.311886330068569,
+      "learning_rate": 5.898615340251851e-08,
+      "loss": 2.9456,
+      "step": 6755
+    },
+    {
+      "epoch": 0.8745185616726543,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.499311923980713,
+      "eval_runtime": 14.5473,
+      "eval_samples_per_second": 3.025,
+      "eval_steps_per_second": 0.206,
+      "step": 6755
+    },
+    {
+      "epoch": 0.8746480240800077,
+      "grad_norm": 1.321627514671924,
+      "learning_rate": 5.886641695144454e-08,
+      "loss": 2.3308,
+      "step": 6756
+    },
+    {
+      "epoch": 0.8747774864873612,
+      "grad_norm": 1.6868890964968488,
+      "learning_rate": 5.8746796144630026e-08,
+      "loss": 2.4585,
+      "step": 6757
+    },
+    {
+      "epoch": 0.8749069488947147,
+      "grad_norm": 1.869483961231482,
+      "learning_rate": 5.862729100650971e-08,
+      "loss": 2.387,
+      "step": 6758
+    },
+    {
+      "epoch": 0.8750364113020681,
+      "grad_norm": 1.5965096175788616,
+      "learning_rate": 5.850790156149519e-08,
+      "loss": 2.4253,
+      "step": 6759
+    },
+    {
+      "epoch": 0.8751658737094217,
+      "grad_norm": 2.01826973845739,
+      "learning_rate": 5.838862783397403e-08,
+      "loss": 2.3745,
+      "step": 6760
+    },
+    {
+      "epoch": 0.8751658737094217,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.4950062036514282,
+      "eval_runtime": 14.3968,
+      "eval_samples_per_second": 3.056,
+      "eval_steps_per_second": 0.208,
+      "step": 6760
+    },
+    {
+      "epoch": 0.8752953361167751,
+      "grad_norm": 1.2557740455679882,
+      "learning_rate": 5.8269469848310224e-08,
+      "loss": 2.2773,
+      "step": 6761
+    },
+    {
+      "epoch": 0.8754247985241286,
+      "grad_norm": 1.2105687212851428,
+      "learning_rate": 5.815042762884443e-08,
+      "loss": 2.1643,
+      "step": 6762
+    },
+    {
+      "epoch": 0.875554260931482,
+      "grad_norm": 2.3780251169723514,
+      "learning_rate": 5.8031501199893174e-08,
+      "loss": 2.6351,
+      "step": 6763
+    },
+    {
+      "epoch": 0.8756837233388355,
+      "grad_norm": 1.3686426958774904,
+      "learning_rate": 5.791269058574987e-08,
+      "loss": 2.3872,
+      "step": 6764
+    },
+    {
+      "epoch": 0.8758131857461889,
+      "grad_norm": 2.020189550657089,
+      "learning_rate": 5.779399581068379e-08,
+      "loss": 2.3906,
+      "step": 6765
+    },
+    {
+      "epoch": 0.8758131857461889,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.497114658355713,
+      "eval_runtime": 15.1632,
+      "eval_samples_per_second": 2.902,
+      "eval_steps_per_second": 0.198,
+      "step": 6765
+    },
+    {
+      "epoch": 0.8759426481535424,
+      "grad_norm": 1.972570207724478,
+      "learning_rate": 5.767541689894074e-08,
+      "loss": 2.3276,
+      "step": 6766
+    },
+    {
+      "epoch": 0.8760721105608958,
+      "grad_norm": 3.690751774115214,
+      "learning_rate": 5.755695387474294e-08,
+      "loss": 2.2214,
+      "step": 6767
+    },
+    {
+      "epoch": 0.8762015729682493,
+      "grad_norm": 2.1374501330175417,
+      "learning_rate": 5.7438606762288964e-08,
+      "loss": 2.425,
+      "step": 6768
+    },
+    {
+      "epoch": 0.8763310353756029,
+      "grad_norm": 2.070583221051381,
+      "learning_rate": 5.7320375585753416e-08,
+      "loss": 2.407,
+      "step": 6769
+    },
+    {
+      "epoch": 0.8764604977829563,
+      "grad_norm": 2.1761599122714403,
+      "learning_rate": 5.720226036928768e-08,
+      "loss": 2.5735,
+      "step": 6770
+    },
+    {
+      "epoch": 0.8764604977829563,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.498579502105713,
+      "eval_runtime": 14.1958,
+      "eval_samples_per_second": 3.1,
+      "eval_steps_per_second": 0.211,
+      "step": 6770
+    },
+    {
+      "epoch": 0.8765899601903098,
+      "grad_norm": 1.7398946188110953,
+      "learning_rate": 5.708426113701877e-08,
+      "loss": 2.405,
+      "step": 6771
+    },
+    {
+      "epoch": 0.8767194225976632,
+      "grad_norm": 1.3869566375129447,
+      "learning_rate": 5.696637791305068e-08,
+      "loss": 2.3115,
+      "step": 6772
+    },
+    {
+      "epoch": 0.8768488850050167,
+      "grad_norm": 1.8084579157259326,
+      "learning_rate": 5.684861072146351e-08,
+      "loss": 2.7219,
+      "step": 6773
+    },
+    {
+      "epoch": 0.8769783474123701,
+      "grad_norm": 6.700630691213659,
+      "learning_rate": 5.673095958631338e-08,
+      "loss": 3.1399,
+      "step": 6774
+    },
+    {
+      "epoch": 0.8771078098197236,
+      "grad_norm": 1.9381461751639362,
+      "learning_rate": 5.6613424531633197e-08,
+      "loss": 2.6365,
+      "step": 6775
+    },
+    {
+      "epoch": 0.8771078098197236,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.49853515625,
+      "eval_runtime": 13.7755,
+      "eval_samples_per_second": 3.194,
+      "eval_steps_per_second": 0.218,
+      "step": 6775
+    },
+    {
+      "epoch": 0.877237272227077,
+      "grad_norm": 1.7307214565159195,
+      "learning_rate": 5.649600558143169e-08,
+      "loss": 2.6199,
+      "step": 6776
+    },
+    {
+      "epoch": 0.8773667346344305,
+      "grad_norm": 2.3790148434202543,
+      "learning_rate": 5.6378702759693966e-08,
+      "loss": 2.4351,
+      "step": 6777
+    },
+    {
+      "epoch": 0.8774961970417839,
+      "grad_norm": 3.106152250041979,
+      "learning_rate": 5.626151609038177e-08,
+      "loss": 2.418,
+      "step": 6778
+    },
+    {
+      "epoch": 0.8776256594491375,
+      "grad_norm": 1.4183043220992424,
+      "learning_rate": 5.614444559743262e-08,
+      "loss": 2.2131,
+      "step": 6779
+    },
+    {
+      "epoch": 0.877755121856491,
+      "grad_norm": 1.2146629735113779,
+      "learning_rate": 5.602749130476072e-08,
+      "loss": 2.2922,
+      "step": 6780
+    },
+    {
+      "epoch": 0.877755121856491,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.4957386255264282,
+      "eval_runtime": 13.7841,
+      "eval_samples_per_second": 3.192,
+      "eval_steps_per_second": 0.218,
+      "step": 6780
+    },
+    {
+      "epoch": 0.8778845842638444,
+      "grad_norm": 1.4819708460149876,
+      "learning_rate": 5.591065323625626e-08,
+      "loss": 2.4458,
+      "step": 6781
+    },
+    {
+      "epoch": 0.8780140466711979,
+      "grad_norm": 3.34377513442525,
+      "learning_rate": 5.579393141578571e-08,
+      "loss": 2.4927,
+      "step": 6782
+    },
+    {
+      "epoch": 0.8781435090785513,
+      "grad_norm": 3.0271503684934227,
+      "learning_rate": 5.567732586719207e-08,
+      "loss": 2.2695,
+      "step": 6783
+    },
+    {
+      "epoch": 0.8782729714859048,
+      "grad_norm": 1.5971368293002015,
+      "learning_rate": 5.556083661429415e-08,
+      "loss": 2.394,
+      "step": 6784
+    },
+    {
+      "epoch": 0.8784024338932582,
+      "grad_norm": 2.195654431220393,
+      "learning_rate": 5.544446368088742e-08,
+      "loss": 2.2937,
+      "step": 6785
+    },
+    {
+      "epoch": 0.8784024338932582,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "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": 14.5783,
+      "eval_samples_per_second": 3.018,
+      "eval_steps_per_second": 0.206,
+      "step": 6785
+    },
+    {
+      "epoch": 0.8785318963006117,
+      "grad_norm": 1.4724710394139269,
+      "learning_rate": 5.532820709074328e-08,
+      "loss": 2.2842,
+      "step": 6786
+    },
+    {
+      "epoch": 0.8786613587079651,
+      "grad_norm": 1.2931475058509387,
+      "learning_rate": 5.5212066867609687e-08,
+      "loss": 2.2429,
+      "step": 6787
+    },
+    {
+      "epoch": 0.8787908211153186,
+      "grad_norm": 1.5520712508654095,
+      "learning_rate": 5.5096043035210435e-08,
+      "loss": 2.1504,
+      "step": 6788
+    },
+    {
+      "epoch": 0.8789202835226722,
+      "grad_norm": 1.6392619538079143,
+      "learning_rate": 5.49801356172458e-08,
+      "loss": 2.3875,
+      "step": 6789
+    },
+    {
+      "epoch": 0.8790497459300256,
+      "grad_norm": 1.202033276643043,
+      "learning_rate": 5.486434463739219e-08,
+      "loss": 2.342,
+      "step": 6790
+    },
+    {
+      "epoch": 0.8790497459300256,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.4986017942428589,
+      "eval_runtime": 14.3802,
+      "eval_samples_per_second": 3.06,
+      "eval_steps_per_second": 0.209,
+      "step": 6790
+    },
+    {
+      "epoch": 0.8791792083373791,
+      "grad_norm": 1.7293585132868499,
+      "learning_rate": 5.474867011930243e-08,
+      "loss": 2.1604,
+      "step": 6791
+    },
+    {
+      "epoch": 0.8793086707447325,
+      "grad_norm": 1.4487881241094938,
+      "learning_rate": 5.4633112086605254e-08,
+      "loss": 2.2876,
+      "step": 6792
+    },
+    {
+      "epoch": 0.879438133152086,
+      "grad_norm": 2.048277615854377,
+      "learning_rate": 5.45176705629058e-08,
+      "loss": 2.478,
+      "step": 6793
+    },
+    {
+      "epoch": 0.8795675955594394,
+      "grad_norm": 2.380244877642461,
+      "learning_rate": 5.4402345571785145e-08,
+      "loss": 2.2944,
+      "step": 6794
+    },
+    {
+      "epoch": 0.8796970579667929,
+      "grad_norm": 1.489663261776139,
+      "learning_rate": 5.4287137136800826e-08,
+      "loss": 2.2495,
+      "step": 6795
+    },
+    {
+      "epoch": 0.8796970579667929,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.493607997894287,
+      "eval_runtime": 14.571,
+      "eval_samples_per_second": 3.02,
+      "eval_steps_per_second": 0.206,
+      "step": 6795
+    },
+    {
+      "epoch": 0.8798265203741463,
+      "grad_norm": 2.9769008748587784,
+      "learning_rate": 5.41720452814868e-08,
+      "loss": 2.3701,
+      "step": 6796
+    },
+    {
+      "epoch": 0.8799559827814998,
+      "grad_norm": 2.3774783432529976,
+      "learning_rate": 5.405707002935245e-08,
+      "loss": 2.6128,
+      "step": 6797
+    },
+    {
+      "epoch": 0.8800854451888532,
+      "grad_norm": 2.796212901722403,
+      "learning_rate": 5.3942211403884214e-08,
+      "loss": 2.4675,
+      "step": 6798
+    },
+    {
+      "epoch": 0.8802149075962068,
+      "grad_norm": 1.3879262284780065,
+      "learning_rate": 5.382746942854414e-08,
+      "loss": 2.2344,
+      "step": 6799
+    },
+    {
+      "epoch": 0.8803443700035603,
+      "grad_norm": 1.2444795640800748,
+      "learning_rate": 5.371284412677043e-08,
+      "loss": 2.4526,
+      "step": 6800
+    },
+    {
+      "epoch": 0.8803443700035603,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.4963600635528564,
+      "eval_runtime": 14.5985,
+      "eval_samples_per_second": 3.014,
+      "eval_steps_per_second": 0.206,
+      "step": 6800
+    },
+    {
+      "epoch": 0.8804738324109137,
+      "grad_norm": 1.272375416587221,
+      "learning_rate": 5.3598335521977826e-08,
+      "loss": 2.25,
+      "step": 6801
+    },
+    {
+      "epoch": 0.8806032948182672,
+      "grad_norm": 1.8228716382863823,
+      "learning_rate": 5.348394363755692e-08,
+      "loss": 2.4907,
+      "step": 6802
+    },
+    {
+      "epoch": 0.8807327572256206,
+      "grad_norm": 1.4814843194662073,
+      "learning_rate": 5.336966849687472e-08,
+      "loss": 2.2861,
+      "step": 6803
+    },
+    {
+      "epoch": 0.8808622196329741,
+      "grad_norm": 2.2646095197360254,
+      "learning_rate": 5.3255510123274224e-08,
+      "loss": 2.2917,
+      "step": 6804
+    },
+    {
+      "epoch": 0.8809916820403275,
+      "grad_norm": 1.812269115029419,
+      "learning_rate": 5.314146854007433e-08,
+      "loss": 2.2893,
+      "step": 6805
+    },
+    {
+      "epoch": 0.8809916820403275,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.4956942796707153,
+      "eval_runtime": 14.3365,
+      "eval_samples_per_second": 3.069,
+      "eval_steps_per_second": 0.209,
+      "step": 6805
+    },
+    {
+      "epoch": 0.881121144447681,
+      "grad_norm": 1.4031238110066964,
+      "learning_rate": 5.302754377057066e-08,
+      "loss": 2.3474,
+      "step": 6806
+    },
+    {
+      "epoch": 0.8812506068550344,
+      "grad_norm": 2.0274924805058334,
+      "learning_rate": 5.2913735838034385e-08,
+      "loss": 2.4805,
+      "step": 6807
+    },
+    {
+      "epoch": 0.881380069262388,
+      "grad_norm": 1.7515371439213783,
+      "learning_rate": 5.2800044765713295e-08,
+      "loss": 2.1887,
+      "step": 6808
+    },
+    {
+      "epoch": 0.8815095316697414,
+      "grad_norm": 2.430880768328812,
+      "learning_rate": 5.268647057683117e-08,
+      "loss": 2.6685,
+      "step": 6809
+    },
+    {
+      "epoch": 0.8816389940770949,
+      "grad_norm": 1.3275899288501505,
+      "learning_rate": 5.2573013294587515e-08,
+      "loss": 2.4717,
+      "step": 6810
+    },
+    {
+      "epoch": 0.8816389940770949,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.4972256422042847,
+      "eval_runtime": 13.8922,
+      "eval_samples_per_second": 3.167,
+      "eval_steps_per_second": 0.216,
+      "step": 6810
+    },
+    {
+      "epoch": 0.8817684564844483,
+      "grad_norm": 1.852054907065753,
+      "learning_rate": 5.2459672942158575e-08,
+      "loss": 2.3359,
+      "step": 6811
+    },
+    {
+      "epoch": 0.8818979188918018,
+      "grad_norm": 1.697480310156964,
+      "learning_rate": 5.2346449542696185e-08,
+      "loss": 2.2881,
+      "step": 6812
+    },
+    {
+      "epoch": 0.8820273812991553,
+      "grad_norm": 2.202499180085729,
+      "learning_rate": 5.223334311932865e-08,
+      "loss": 2.4719,
+      "step": 6813
+    },
+    {
+      "epoch": 0.8821568437065087,
+      "grad_norm": 1.5761265031072722,
+      "learning_rate": 5.212035369516033e-08,
+      "loss": 2.4033,
+      "step": 6814
+    },
+    {
+      "epoch": 0.8822863061138622,
+      "grad_norm": 2.0088869742015913,
+      "learning_rate": 5.2007481293271514e-08,
+      "loss": 2.4524,
+      "step": 6815
+    },
+    {
+      "epoch": 0.8822863061138622,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.4972922801971436,
+      "eval_runtime": 14.346,
+      "eval_samples_per_second": 3.067,
+      "eval_steps_per_second": 0.209,
+      "step": 6815
+    },
+    {
+      "epoch": 0.8824157685212156,
+      "grad_norm": 2.2596026701442646,
+      "learning_rate": 5.189472593671856e-08,
+      "loss": 2.6572,
+      "step": 6816
+    },
+    {
+      "epoch": 0.8825452309285691,
+      "grad_norm": 1.9062682312606882,
+      "learning_rate": 5.1782087648534275e-08,
+      "loss": 2.3262,
+      "step": 6817
+    },
+    {
+      "epoch": 0.8826746933359226,
+      "grad_norm": 1.3008849475205415,
+      "learning_rate": 5.1669566451727005e-08,
+      "loss": 2.1953,
+      "step": 6818
+    },
+    {
+      "epoch": 0.8828041557432761,
+      "grad_norm": 1.5609962547581708,
+      "learning_rate": 5.1557162369281836e-08,
+      "loss": 2.2661,
+      "step": 6819
+    },
+    {
+      "epoch": 0.8829336181506295,
+      "grad_norm": 2.4739657629797867,
+      "learning_rate": 5.144487542415942e-08,
+      "loss": 2.4871,
+      "step": 6820
+    },
+    {
+      "epoch": 0.8829336181506295,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.4977805614471436,
+      "eval_runtime": 13.637,
+      "eval_samples_per_second": 3.227,
+      "eval_steps_per_second": 0.22,
+      "step": 6820
+    },
+    {
+      "epoch": 0.883063080557983,
+      "grad_norm": 1.6040042906766272,
+      "learning_rate": 5.1332705639296485e-08,
+      "loss": 2.4102,
+      "step": 6821
+    },
+    {
+      "epoch": 0.8831925429653364,
+      "grad_norm": 1.6887012402620727,
+      "learning_rate": 5.12206530376063e-08,
+      "loss": 2.3594,
+      "step": 6822
+    },
+    {
+      "epoch": 0.8833220053726899,
+      "grad_norm": 1.3405311912088704,
+      "learning_rate": 5.1108717641977504e-08,
+      "loss": 2.3254,
+      "step": 6823
+    },
+    {
+      "epoch": 0.8834514677800434,
+      "grad_norm": 2.5960548001675563,
+      "learning_rate": 5.099689947527554e-08,
+      "loss": 2.3899,
+      "step": 6824
+    },
+    {
+      "epoch": 0.8835809301873968,
+      "grad_norm": 1.2979289484318808,
+      "learning_rate": 5.088519856034126e-08,
+      "loss": 2.3182,
+      "step": 6825
+    },
+    {
+      "epoch": 0.8835809301873968,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.4964710474014282,
+      "eval_runtime": 14.2333,
+      "eval_samples_per_second": 3.091,
+      "eval_steps_per_second": 0.211,
+      "step": 6825
+    },
+    {
+      "epoch": 0.8837103925947503,
+      "grad_norm": 1.8178412323124833,
+      "learning_rate": 5.077361491999208e-08,
+      "loss": 2.1327,
+      "step": 6826
+    },
+    {
+      "epoch": 0.8838398550021037,
+      "grad_norm": 1.5244985575952903,
+      "learning_rate": 5.0662148577021106e-08,
+      "loss": 2.3545,
+      "step": 6827
+    },
+    {
+      "epoch": 0.8839693174094573,
+      "grad_norm": 2.362733321399371,
+      "learning_rate": 5.055079955419741e-08,
+      "loss": 2.3171,
+      "step": 6828
+    },
+    {
+      "epoch": 0.8840987798168107,
+      "grad_norm": 1.206699504106358,
+      "learning_rate": 5.043956787426661e-08,
+      "loss": 2.4153,
+      "step": 6829
+    },
+    {
+      "epoch": 0.8842282422241642,
+      "grad_norm": 1.5704394450802655,
+      "learning_rate": 5.032845355994978e-08,
+      "loss": 2.3965,
+      "step": 6830
+    },
+    {
+      "epoch": 0.8842282422241642,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.4956942796707153,
+      "eval_runtime": 13.5494,
+      "eval_samples_per_second": 3.247,
+      "eval_steps_per_second": 0.221,
+      "step": 6830
+    },
+    {
+      "epoch": 0.8843577046315176,
+      "grad_norm": 1.5112196749551587,
+      "learning_rate": 5.021745663394445e-08,
+      "loss": 2.3868,
+      "step": 6831
+    },
+    {
+      "epoch": 0.8844871670388711,
+      "grad_norm": 1.3908879176411566,
+      "learning_rate": 5.010657711892394e-08,
+      "loss": 2.3821,
+      "step": 6832
+    },
+    {
+      "epoch": 0.8846166294462245,
+      "grad_norm": 1.3499601569912798,
+      "learning_rate": 4.9995815037537433e-08,
+      "loss": 2.4922,
+      "step": 6833
+    },
+    {
+      "epoch": 0.884746091853578,
+      "grad_norm": 1.6594522414697797,
+      "learning_rate": 4.9885170412410715e-08,
+      "loss": 2.4341,
+      "step": 6834
+    },
+    {
+      "epoch": 0.8848755542609315,
+      "grad_norm": 1.2013526688641463,
+      "learning_rate": 4.9774643266144796e-08,
+      "loss": 2.3118,
+      "step": 6835
+    },
+    {
+      "epoch": 0.8848755542609315,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.4964488744735718,
+      "eval_runtime": 14.9143,
+      "eval_samples_per_second": 2.95,
+      "eval_steps_per_second": 0.201,
+      "step": 6835
+    },
+    {
+      "epoch": 0.8850050166682849,
+      "grad_norm": 1.6973190897588857,
+      "learning_rate": 4.966423362131732e-08,
+      "loss": 2.262,
+      "step": 6836
+    },
+    {
+      "epoch": 0.8851344790756384,
+      "grad_norm": 2.164161001163814,
+      "learning_rate": 4.9553941500481917e-08,
+      "loss": 2.4458,
+      "step": 6837
+    },
+    {
+      "epoch": 0.8852639414829919,
+      "grad_norm": 1.3206649933939947,
+      "learning_rate": 4.9443766926167425e-08,
+      "loss": 2.5044,
+      "step": 6838
+    },
+    {
+      "epoch": 0.8853934038903454,
+      "grad_norm": 1.5635126332503226,
+      "learning_rate": 4.933370992087953e-08,
+      "loss": 2.3807,
+      "step": 6839
+    },
+    {
+      "epoch": 0.8855228662976988,
+      "grad_norm": 2.674186711804766,
+      "learning_rate": 4.922377050709977e-08,
+      "loss": 2.4448,
+      "step": 6840
+    },
+    {
+      "epoch": 0.8855228662976988,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.4978693723678589,
+      "eval_runtime": 13.5552,
+      "eval_samples_per_second": 3.246,
+      "eval_steps_per_second": 0.221,
+      "step": 6840
+    },
+    {
+      "epoch": 0.8856523287050523,
+      "grad_norm": 1.1852061910867218,
+      "learning_rate": 4.911394870728524e-08,
+      "loss": 2.3462,
+      "step": 6841
+    },
+    {
+      "epoch": 0.8857817911124057,
+      "grad_norm": 1.6833025405061035,
+      "learning_rate": 4.9004244543869524e-08,
+      "loss": 2.561,
+      "step": 6842
+    },
+    {
+      "epoch": 0.8859112535197592,
+      "grad_norm": 1.6137981062566404,
+      "learning_rate": 4.889465803926158e-08,
+      "loss": 2.3879,
+      "step": 6843
+    },
+    {
+      "epoch": 0.8860407159271126,
+      "grad_norm": 1.4002543212038663,
+      "learning_rate": 4.8785189215846826e-08,
+      "loss": 2.4956,
+      "step": 6844
+    },
+    {
+      "epoch": 0.8861701783344661,
+      "grad_norm": 2.259311525048003,
+      "learning_rate": 4.867583809598668e-08,
+      "loss": 2.415,
+      "step": 6845
+    },
+    {
+      "epoch": 0.8861701783344661,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.4921208620071411,
+      "eval_runtime": 13.4811,
+      "eval_samples_per_second": 3.264,
+      "eval_steps_per_second": 0.223,
+      "step": 6845
+    },
+    {
+      "epoch": 0.8862996407418195,
+      "grad_norm": 1.1739139988353178,
+      "learning_rate": 4.8566604702018e-08,
+      "loss": 2.0773,
+      "step": 6846
+    },
+    {
+      "epoch": 0.8864291031491731,
+      "grad_norm": 1.6920353589218537,
+      "learning_rate": 4.845748905625419e-08,
+      "loss": 2.2161,
+      "step": 6847
+    },
+    {
+      "epoch": 0.8865585655565266,
+      "grad_norm": 2.1925823450327813,
+      "learning_rate": 4.8348491180984224e-08,
+      "loss": 2.3091,
+      "step": 6848
+    },
+    {
+      "epoch": 0.88668802796388,
+      "grad_norm": 1.7912374780015778,
+      "learning_rate": 4.823961109847307e-08,
+      "loss": 2.3779,
+      "step": 6849
+    },
+    {
+      "epoch": 0.8868174903712335,
+      "grad_norm": 1.5919658250717248,
+      "learning_rate": 4.813084883096182e-08,
+      "loss": 2.3093,
+      "step": 6850
+    },
+    {
+      "epoch": 0.8868174903712335,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.4986683130264282,
+      "eval_runtime": 14.755,
+      "eval_samples_per_second": 2.982,
+      "eval_steps_per_second": 0.203,
+      "step": 6850
+    },
+    {
+      "epoch": 0.8869469527785869,
+      "grad_norm": 2.4835319244556575,
+      "learning_rate": 4.802220440066723e-08,
+      "loss": 2.4106,
+      "step": 6851
+    },
+    {
+      "epoch": 0.8870764151859404,
+      "grad_norm": 1.9092603332812002,
+      "learning_rate": 4.791367782978222e-08,
+      "loss": 2.6543,
+      "step": 6852
+    },
+    {
+      "epoch": 0.8872058775932938,
+      "grad_norm": 1.438123227684614,
+      "learning_rate": 4.7805269140475464e-08,
+      "loss": 2.4385,
+      "step": 6853
+    },
+    {
+      "epoch": 0.8873353400006473,
+      "grad_norm": 1.409676616809633,
+      "learning_rate": 4.7696978354891816e-08,
+      "loss": 2.395,
+      "step": 6854
+    },
+    {
+      "epoch": 0.8874648024080007,
+      "grad_norm": 1.9197598074861877,
+      "learning_rate": 4.75888054951517e-08,
+      "loss": 2.2854,
+      "step": 6855
+    },
+    {
+      "epoch": 0.8874648024080007,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.4950283765792847,
+      "eval_runtime": 13.7904,
+      "eval_samples_per_second": 3.191,
+      "eval_steps_per_second": 0.218,
+      "step": 6855
+    },
+    {
+      "epoch": 0.8875942648153542,
+      "grad_norm": 1.5964177352374567,
+      "learning_rate": 4.748075058335161e-08,
+      "loss": 2.4468,
+      "step": 6856
+    },
+    {
+      "epoch": 0.8877237272227078,
+      "grad_norm": 2.18707695800178,
+      "learning_rate": 4.737281364156403e-08,
+      "loss": 2.4023,
+      "step": 6857
+    },
+    {
+      "epoch": 0.8878531896300612,
+      "grad_norm": 1.5138236784357733,
+      "learning_rate": 4.7264994691837153e-08,
+      "loss": 2.2424,
+      "step": 6858
+    },
+    {
+      "epoch": 0.8879826520374147,
+      "grad_norm": 1.9180195132690427,
+      "learning_rate": 4.7157293756195375e-08,
+      "loss": 2.314,
+      "step": 6859
+    },
+    {
+      "epoch": 0.8881121144447681,
+      "grad_norm": 1.74251428073698,
+      "learning_rate": 4.704971085663867e-08,
+      "loss": 2.4175,
+      "step": 6860
+    },
+    {
+      "epoch": 0.8881121144447681,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.496382236480713,
+      "eval_runtime": 14.1415,
+      "eval_samples_per_second": 3.111,
+      "eval_steps_per_second": 0.212,
+      "step": 6860
+    },
+    {
+      "epoch": 0.8882415768521216,
+      "grad_norm": 2.451772939019097,
+      "learning_rate": 4.694224601514301e-08,
+      "loss": 2.7568,
+      "step": 6861
+    },
+    {
+      "epoch": 0.888371039259475,
+      "grad_norm": 2.1808859934039786,
+      "learning_rate": 4.6834899253660195e-08,
+      "loss": 2.2068,
+      "step": 6862
+    },
+    {
+      "epoch": 0.8885005016668285,
+      "grad_norm": 1.9708244448495695,
+      "learning_rate": 4.672767059411824e-08,
+      "loss": 2.5842,
+      "step": 6863
+    },
+    {
+      "epoch": 0.8886299640741819,
+      "grad_norm": 1.5567832791817082,
+      "learning_rate": 4.662056005842054e-08,
+      "loss": 2.5952,
+      "step": 6864
+    },
+    {
+      "epoch": 0.8887594264815354,
+      "grad_norm": 2.109379645225532,
+      "learning_rate": 4.651356766844681e-08,
+      "loss": 2.4111,
+      "step": 6865
+    },
+    {
+      "epoch": 0.8887594264815354,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.49658203125,
+      "eval_runtime": 14.3206,
+      "eval_samples_per_second": 3.072,
+      "eval_steps_per_second": 0.209,
+      "step": 6865
+    },
+    {
+      "epoch": 0.8888888888888888,
+      "grad_norm": 1.492008547478116,
+      "learning_rate": 4.640669344605206e-08,
+      "loss": 2.3145,
+      "step": 6866
+    },
+    {
+      "epoch": 0.8890183512962424,
+      "grad_norm": 2.4863475947710816,
+      "learning_rate": 4.6299937413067794e-08,
+      "loss": 2.5022,
+      "step": 6867
+    },
+    {
+      "epoch": 0.8891478137035959,
+      "grad_norm": 1.8362858630475052,
+      "learning_rate": 4.6193299591301076e-08,
+      "loss": 2.2554,
+      "step": 6868
+    },
+    {
+      "epoch": 0.8892772761109493,
+      "grad_norm": 1.7880404505061087,
+      "learning_rate": 4.608678000253469e-08,
+      "loss": 2.3367,
+      "step": 6869
+    },
+    {
+      "epoch": 0.8894067385183028,
+      "grad_norm": 1.8306178102149797,
+      "learning_rate": 4.598037866852763e-08,
+      "loss": 2.5474,
+      "step": 6870
+    },
+    {
+      "epoch": 0.8894067385183028,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.4979358911514282,
+      "eval_runtime": 14.57,
+      "eval_samples_per_second": 3.02,
+      "eval_steps_per_second": 0.206,
+      "step": 6870
+    },
+    {
+      "epoch": 0.8895362009256562,
+      "grad_norm": 1.5091112829606663,
+      "learning_rate": 4.587409561101444e-08,
+      "loss": 2.365,
+      "step": 6871
+    },
+    {
+      "epoch": 0.8896656633330097,
+      "grad_norm": 1.3385445182413755,
+      "learning_rate": 4.576793085170547e-08,
+      "loss": 2.2312,
+      "step": 6872
+    },
+    {
+      "epoch": 0.8897951257403631,
+      "grad_norm": 2.5138482603010623,
+      "learning_rate": 4.56618844122872e-08,
+      "loss": 2.2615,
+      "step": 6873
+    },
+    {
+      "epoch": 0.8899245881477166,
+      "grad_norm": 1.880052318002391,
+      "learning_rate": 4.555595631442158e-08,
+      "loss": 2.4336,
+      "step": 6874
+    },
+    {
+      "epoch": 0.89005405055507,
+      "grad_norm": 1.424451958521749,
+      "learning_rate": 4.545014657974682e-08,
+      "loss": 2.407,
+      "step": 6875
+    },
+    {
+      "epoch": 0.89005405055507,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.4944514036178589,
+      "eval_runtime": 15.2115,
+      "eval_samples_per_second": 2.893,
+      "eval_steps_per_second": 0.197,
+      "step": 6875
+    },
+    {
+      "epoch": 0.8901835129624235,
+      "grad_norm": 1.4429270918344927,
+      "learning_rate": 4.534445522987652e-08,
+      "loss": 2.3318,
+      "step": 6876
+    },
+    {
+      "epoch": 0.890312975369777,
+      "grad_norm": 2.26301412988003,
+      "learning_rate": 4.5238882286400204e-08,
+      "loss": 2.3582,
+      "step": 6877
+    },
+    {
+      "epoch": 0.8904424377771305,
+      "grad_norm": 1.7604419801924598,
+      "learning_rate": 4.513342777088356e-08,
+      "loss": 2.4346,
+      "step": 6878
+    },
+    {
+      "epoch": 0.890571900184484,
+      "grad_norm": 1.5066813985924459,
+      "learning_rate": 4.502809170486754e-08,
+      "loss": 2.3643,
+      "step": 6879
+    },
+    {
+      "epoch": 0.8907013625918374,
+      "grad_norm": 2.55366696627522,
+      "learning_rate": 4.492287410986937e-08,
+      "loss": 2.5508,
+      "step": 6880
+    },
+    {
+      "epoch": 0.8907013625918374,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "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": 13.7504,
+      "eval_samples_per_second": 3.2,
+      "eval_steps_per_second": 0.218,
+      "step": 6880
+    },
+    {
+      "epoch": 0.8908308249991909,
+      "grad_norm": 1.5881197527451938,
+      "learning_rate": 4.4817775007381795e-08,
+      "loss": 2.2556,
+      "step": 6881
+    },
+    {
+      "epoch": 0.8909602874065443,
+      "grad_norm": 1.4310554373910798,
+      "learning_rate": 4.471279441887334e-08,
+      "loss": 2.397,
+      "step": 6882
+    },
+    {
+      "epoch": 0.8910897498138978,
+      "grad_norm": 1.9032917331131038,
+      "learning_rate": 4.460793236578859e-08,
+      "loss": 2.4518,
+      "step": 6883
+    },
+    {
+      "epoch": 0.8912192122212512,
+      "grad_norm": 2.1307513203863855,
+      "learning_rate": 4.450318886954762e-08,
+      "loss": 2.2865,
+      "step": 6884
+    },
+    {
+      "epoch": 0.8913486746286047,
+      "grad_norm": 2.7305593681769027,
+      "learning_rate": 4.439856395154637e-08,
+      "loss": 2.2349,
+      "step": 6885
+    },
+    {
+      "epoch": 0.8913486746286047,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.4957386255264282,
+      "eval_runtime": 13.6619,
+      "eval_samples_per_second": 3.221,
+      "eval_steps_per_second": 0.22,
+      "step": 6885
+    },
+    {
+      "epoch": 0.8914781370359581,
+      "grad_norm": 1.641322534487582,
+      "learning_rate": 4.4294057633156864e-08,
+      "loss": 2.1875,
+      "step": 6886
+    },
+    {
+      "epoch": 0.8916075994433117,
+      "grad_norm": 1.350728645644694,
+      "learning_rate": 4.4189669935726446e-08,
+      "loss": 2.2871,
+      "step": 6887
+    },
+    {
+      "epoch": 0.8917370618506651,
+      "grad_norm": 1.513077226970648,
+      "learning_rate": 4.408540088057842e-08,
+      "loss": 2.5142,
+      "step": 6888
+    },
+    {
+      "epoch": 0.8918665242580186,
+      "grad_norm": 1.6532640452882146,
+      "learning_rate": 4.3981250489011776e-08,
+      "loss": 2.145,
+      "step": 6889
+    },
+    {
+      "epoch": 0.891995986665372,
+      "grad_norm": 1.4154330168458797,
+      "learning_rate": 4.387721878230139e-08,
+      "loss": 2.3655,
+      "step": 6890
+    },
+    {
+      "epoch": 0.891995986665372,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.4949840307235718,
+      "eval_runtime": 14.4213,
+      "eval_samples_per_second": 3.051,
+      "eval_steps_per_second": 0.208,
+      "step": 6890
+    },
+    {
+      "epoch": 0.8921254490727255,
+      "grad_norm": 1.806321323102692,
+      "learning_rate": 4.3773305781698026e-08,
+      "loss": 2.2756,
+      "step": 6891
+    },
+    {
+      "epoch": 0.892254911480079,
+      "grad_norm": 1.2895354054768229,
+      "learning_rate": 4.3669511508427906e-08,
+      "loss": 2.4072,
+      "step": 6892
+    },
+    {
+      "epoch": 0.8923843738874324,
+      "grad_norm": 1.6770658669718612,
+      "learning_rate": 4.356583598369296e-08,
+      "loss": 2.5588,
+      "step": 6893
+    },
+    {
+      "epoch": 0.8925138362947859,
+      "grad_norm": 1.1178914600945362,
+      "learning_rate": 4.3462279228671243e-08,
+      "loss": 2.2297,
+      "step": 6894
+    },
+    {
+      "epoch": 0.8926432987021393,
+      "grad_norm": 1.658333578972004,
+      "learning_rate": 4.335884126451612e-08,
+      "loss": 2.3457,
+      "step": 6895
+    },
+    {
+      "epoch": 0.8926432987021393,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.4964932203292847,
+      "eval_runtime": 14.7287,
+      "eval_samples_per_second": 2.987,
+      "eval_steps_per_second": 0.204,
+      "step": 6895
+    },
+    {
+      "epoch": 0.8927727611094929,
+      "grad_norm": 1.533056140914135,
+      "learning_rate": 4.3255522112357134e-08,
+      "loss": 2.3196,
+      "step": 6896
+    },
+    {
+      "epoch": 0.8929022235168463,
+      "grad_norm": 1.5024013722088398,
+      "learning_rate": 4.315232179329908e-08,
+      "loss": 2.4712,
+      "step": 6897
+    },
+    {
+      "epoch": 0.8930316859241998,
+      "grad_norm": 1.811341991406139,
+      "learning_rate": 4.3049240328422865e-08,
+      "loss": 2.1892,
+      "step": 6898
+    },
+    {
+      "epoch": 0.8931611483315532,
+      "grad_norm": 3.9388349331388235,
+      "learning_rate": 4.294627773878499e-08,
+      "loss": 2.7144,
+      "step": 6899
+    },
+    {
+      "epoch": 0.8932906107389067,
+      "grad_norm": 1.403469092732384,
+      "learning_rate": 4.284343404541744e-08,
+      "loss": 2.24,
+      "step": 6900
+    },
+    {
+      "epoch": 0.8932906107389067,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "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": 14.1089,
+      "eval_samples_per_second": 3.119,
+      "eval_steps_per_second": 0.213,
+      "step": 6900
+    },
+    {
+      "epoch": 0.8934200731462602,
+      "grad_norm": 3.7223479338226206,
+      "learning_rate": 4.274070926932834e-08,
+      "loss": 2.4541,
+      "step": 6901
+    },
+    {
+      "epoch": 0.8935495355536136,
+      "grad_norm": 1.3644230410435525,
+      "learning_rate": 4.2638103431501126e-08,
+      "loss": 2.2668,
+      "step": 6902
+    },
+    {
+      "epoch": 0.8936789979609671,
+      "grad_norm": 1.609142453978087,
+      "learning_rate": 4.253561655289527e-08,
+      "loss": 2.2925,
+      "step": 6903
+    },
+    {
+      "epoch": 0.8938084603683205,
+      "grad_norm": 2.4635030186402402,
+      "learning_rate": 4.243324865444591e-08,
+      "loss": 2.1685,
+      "step": 6904
+    },
+    {
+      "epoch": 0.893937922775674,
+      "grad_norm": 3.4963444964201673,
+      "learning_rate": 4.2330999757063463e-08,
+      "loss": 2.6826,
+      "step": 6905
+    },
+    {
+      "epoch": 0.893937922775674,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.4964932203292847,
+      "eval_runtime": 14.3624,
+      "eval_samples_per_second": 3.064,
+      "eval_steps_per_second": 0.209,
+      "step": 6905
+    },
+    {
+      "epoch": 0.8940673851830275,
+      "grad_norm": 1.8002477917651825,
+      "learning_rate": 4.2228869881634496e-08,
+      "loss": 2.2915,
+      "step": 6906
+    },
+    {
+      "epoch": 0.894196847590381,
+      "grad_norm": 3.193912068537218,
+      "learning_rate": 4.2126859049021054e-08,
+      "loss": 2.4922,
+      "step": 6907
+    },
+    {
+      "epoch": 0.8943263099977344,
+      "grad_norm": 2.1680841385996215,
+      "learning_rate": 4.2024967280060936e-08,
+      "loss": 2.4824,
+      "step": 6908
+    },
+    {
+      "epoch": 0.8944557724050879,
+      "grad_norm": 1.4891514868125184,
+      "learning_rate": 4.192319459556775e-08,
+      "loss": 2.3562,
+      "step": 6909
+    },
+    {
+      "epoch": 0.8945852348124413,
+      "grad_norm": 2.4653715109882235,
+      "learning_rate": 4.182154101633036e-08,
+      "loss": 2.4199,
+      "step": 6910
+    },
+    {
+      "epoch": 0.8945852348124413,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.4964488744735718,
+      "eval_runtime": 14.8241,
+      "eval_samples_per_second": 2.968,
+      "eval_steps_per_second": 0.202,
+      "step": 6910
+    },
+    {
+      "epoch": 0.8947146972197948,
+      "grad_norm": 2.069641679533272,
+      "learning_rate": 4.1720006563113685e-08,
+      "loss": 2.4431,
+      "step": 6911
+    },
+    {
+      "epoch": 0.8948441596271483,
+      "grad_norm": 1.3878531424713867,
+      "learning_rate": 4.1618591256658364e-08,
+      "loss": 2.3965,
+      "step": 6912
+    },
+    {
+      "epoch": 0.8949736220345017,
+      "grad_norm": 2.441516934084074,
+      "learning_rate": 4.151729511768024e-08,
+      "loss": 2.3274,
+      "step": 6913
+    },
+    {
+      "epoch": 0.8951030844418552,
+      "grad_norm": 1.2444497277156765,
+      "learning_rate": 4.141611816687138e-08,
+      "loss": 2.1733,
+      "step": 6914
+    },
+    {
+      "epoch": 0.8952325468492086,
+      "grad_norm": 1.25034331459268,
+      "learning_rate": 4.1315060424899006e-08,
+      "loss": 2.3225,
+      "step": 6915
+    },
+    {
+      "epoch": 0.8952325468492086,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.4957829713821411,
+      "eval_runtime": 13.5391,
+      "eval_samples_per_second": 3.25,
+      "eval_steps_per_second": 0.222,
+      "step": 6915
+    },
+    {
+      "epoch": 0.8953620092565622,
+      "grad_norm": 1.4956653083155558,
+      "learning_rate": 4.121412191240626e-08,
+      "loss": 2.4302,
+      "step": 6916
+    },
+    {
+      "epoch": 0.8954914716639156,
+      "grad_norm": 3.695456243006164,
+      "learning_rate": 4.1113302650012054e-08,
+      "loss": 2.373,
+      "step": 6917
+    },
+    {
+      "epoch": 0.8956209340712691,
+      "grad_norm": 1.5775926229543367,
+      "learning_rate": 4.1012602658310476e-08,
+      "loss": 2.1853,
+      "step": 6918
+    },
+    {
+      "epoch": 0.8957503964786225,
+      "grad_norm": 1.755700282843993,
+      "learning_rate": 4.0912021957871734e-08,
+      "loss": 2.3325,
+      "step": 6919
+    },
+    {
+      "epoch": 0.895879858885976,
+      "grad_norm": 1.227436237800927,
+      "learning_rate": 4.0811560569241485e-08,
+      "loss": 2.3989,
+      "step": 6920
+    },
+    {
+      "epoch": 0.895879858885976,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.4963600635528564,
+      "eval_runtime": 14.1878,
+      "eval_samples_per_second": 3.101,
+      "eval_steps_per_second": 0.211,
+      "step": 6920
+    },
+    {
+      "epoch": 0.8960093212933294,
+      "grad_norm": 1.688768588163638,
+      "learning_rate": 4.071121851294088e-08,
+      "loss": 2.3564,
+      "step": 6921
+    },
+    {
+      "epoch": 0.8961387837006829,
+      "grad_norm": 1.5718793231134163,
+      "learning_rate": 4.0610995809466924e-08,
+      "loss": 2.3838,
+      "step": 6922
+    },
+    {
+      "epoch": 0.8962682461080363,
+      "grad_norm": 3.762872273849478,
+      "learning_rate": 4.051089247929206e-08,
+      "loss": 2.259,
+      "step": 6923
+    },
+    {
+      "epoch": 0.8963977085153898,
+      "grad_norm": 1.6658417124342157,
+      "learning_rate": 4.0410908542864515e-08,
+      "loss": 2.3289,
+      "step": 6924
+    },
+    {
+      "epoch": 0.8965271709227433,
+      "grad_norm": 2.6997947261750572,
+      "learning_rate": 4.031104402060795e-08,
+      "loss": 2.2056,
+      "step": 6925
+    },
+    {
+      "epoch": 0.8965271709227433,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.4957386255264282,
+      "eval_runtime": 15.2393,
+      "eval_samples_per_second": 2.887,
+      "eval_steps_per_second": 0.197,
+      "step": 6925
+    },
+    {
+      "epoch": 0.8966566333300968,
+      "grad_norm": 1.6703939407210462,
+      "learning_rate": 4.021129893292182e-08,
+      "loss": 2.2632,
+      "step": 6926
+    },
+    {
+      "epoch": 0.8967860957374503,
+      "grad_norm": 2.1294872006307854,
+      "learning_rate": 4.0111673300180926e-08,
+      "loss": 2.3159,
+      "step": 6927
+    },
+    {
+      "epoch": 0.8969155581448037,
+      "grad_norm": 1.7415914614987027,
+      "learning_rate": 4.0012167142735936e-08,
+      "loss": 2.2163,
+      "step": 6928
+    },
+    {
+      "epoch": 0.8970450205521572,
+      "grad_norm": 1.1887525851054148,
+      "learning_rate": 3.991278048091296e-08,
+      "loss": 2.1575,
+      "step": 6929
+    },
+    {
+      "epoch": 0.8971744829595106,
+      "grad_norm": 1.4394557523763398,
+      "learning_rate": 3.9813513335013735e-08,
+      "loss": 2.1777,
+      "step": 6930
+    },
+    {
+      "epoch": 0.8971744829595106,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "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": 13.6703,
+      "eval_samples_per_second": 3.219,
+      "eval_steps_per_second": 0.219,
+      "step": 6930
+    },
+    {
+      "epoch": 0.8973039453668641,
+      "grad_norm": 1.2593171368973932,
+      "learning_rate": 3.971436572531566e-08,
+      "loss": 2.2441,
+      "step": 6931
+    },
+    {
+      "epoch": 0.8974334077742175,
+      "grad_norm": 1.2891194530636572,
+      "learning_rate": 3.9615337672071555e-08,
+      "loss": 2.296,
+      "step": 6932
+    },
+    {
+      "epoch": 0.897562870181571,
+      "grad_norm": 1.2017006243417034,
+      "learning_rate": 3.95164291955099e-08,
+      "loss": 2.3989,
+      "step": 6933
+    },
+    {
+      "epoch": 0.8976923325889244,
+      "grad_norm": 1.1455584727164558,
+      "learning_rate": 3.9417640315834724e-08,
+      "loss": 2.2739,
+      "step": 6934
+    },
+    {
+      "epoch": 0.897821794996278,
+      "grad_norm": 1.1819944504299127,
+      "learning_rate": 3.931897105322588e-08,
+      "loss": 2.2979,
+      "step": 6935
+    },
+    {
+      "epoch": 0.897821794996278,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.4935635328292847,
+      "eval_runtime": 14.1786,
+      "eval_samples_per_second": 3.103,
+      "eval_steps_per_second": 0.212,
+      "step": 6935
+    },
+    {
+      "epoch": 0.8979512574036315,
+      "grad_norm": 2.983504529933777,
+      "learning_rate": 3.9220421427838276e-08,
+      "loss": 2.3772,
+      "step": 6936
+    },
+    {
+      "epoch": 0.8980807198109849,
+      "grad_norm": 1.6038312588693155,
+      "learning_rate": 3.9121991459803026e-08,
+      "loss": 2.3679,
+      "step": 6937
+    },
+    {
+      "epoch": 0.8982101822183384,
+      "grad_norm": 2.8079810843264,
+      "learning_rate": 3.902368116922607e-08,
+      "loss": 2.2349,
+      "step": 6938
+    },
+    {
+      "epoch": 0.8983396446256918,
+      "grad_norm": 1.9090172639941165,
+      "learning_rate": 3.89254905761894e-08,
+      "loss": 2.4524,
+      "step": 6939
+    },
+    {
+      "epoch": 0.8984691070330453,
+      "grad_norm": 2.200047488020452,
+      "learning_rate": 3.8827419700750574e-08,
+      "loss": 2.4019,
+      "step": 6940
+    },
+    {
+      "epoch": 0.8984691070330453,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.4957386255264282,
+      "eval_runtime": 14.3267,
+      "eval_samples_per_second": 3.071,
+      "eval_steps_per_second": 0.209,
+      "step": 6940
+    },
+    {
+      "epoch": 0.8985985694403987,
+      "grad_norm": 1.5384715793458743,
+      "learning_rate": 3.8729468562942325e-08,
+      "loss": 2.3403,
+      "step": 6941
+    },
+    {
+      "epoch": 0.8987280318477522,
+      "grad_norm": 2.1449745506697586,
+      "learning_rate": 3.863163718277336e-08,
+      "loss": 2.3335,
+      "step": 6942
+    },
+    {
+      "epoch": 0.8988574942551056,
+      "grad_norm": 4.3114234142983285,
+      "learning_rate": 3.853392558022771e-08,
+      "loss": 2.5251,
+      "step": 6943
+    },
+    {
+      "epoch": 0.8989869566624591,
+      "grad_norm": 2.59681552414949,
+      "learning_rate": 3.843633377526475e-08,
+      "loss": 2.3013,
+      "step": 6944
+    },
+    {
+      "epoch": 0.8991164190698127,
+      "grad_norm": 1.7149763388676593,
+      "learning_rate": 3.833886178781984e-08,
+      "loss": 2.4336,
+      "step": 6945
+    },
+    {
+      "epoch": 0.8991164190698127,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.4972034692764282,
+      "eval_runtime": 13.9669,
+      "eval_samples_per_second": 3.15,
+      "eval_steps_per_second": 0.215,
+      "step": 6945
+    },
+    {
+      "epoch": 0.8992458814771661,
+      "grad_norm": 1.3751559712862687,
+      "learning_rate": 3.8241509637803387e-08,
+      "loss": 2.4021,
+      "step": 6946
+    },
+    {
+      "epoch": 0.8993753438845196,
+      "grad_norm": 1.907702766922529,
+      "learning_rate": 3.81442773451017e-08,
+      "loss": 2.4231,
+      "step": 6947
+    },
+    {
+      "epoch": 0.899504806291873,
+      "grad_norm": 2.153565100561712,
+      "learning_rate": 3.804716492957647e-08,
+      "loss": 2.3845,
+      "step": 6948
+    },
+    {
+      "epoch": 0.8996342686992265,
+      "grad_norm": 1.7111247405700598,
+      "learning_rate": 3.7950172411064594e-08,
+      "loss": 2.3345,
+      "step": 6949
+    },
+    {
+      "epoch": 0.8997637311065799,
+      "grad_norm": 1.503382015285647,
+      "learning_rate": 3.785329980937912e-08,
+      "loss": 2.2749,
+      "step": 6950
+    },
+    {
+      "epoch": 0.8997637311065799,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.4972256422042847,
+      "eval_runtime": 14.1392,
+      "eval_samples_per_second": 3.112,
+      "eval_steps_per_second": 0.212,
+      "step": 6950
+    },
+    {
+      "epoch": 0.8998931935139334,
+      "grad_norm": 1.6561190580163554,
+      "learning_rate": 3.775654714430803e-08,
+      "loss": 2.5015,
+      "step": 6951
+    },
+    {
+      "epoch": 0.9000226559212868,
+      "grad_norm": 1.4122607964748026,
+      "learning_rate": 3.765991443561511e-08,
+      "loss": 2.4363,
+      "step": 6952
+    },
+    {
+      "epoch": 0.9001521183286403,
+      "grad_norm": 1.4949728824649122,
+      "learning_rate": 3.756340170303942e-08,
+      "loss": 2.3779,
+      "step": 6953
+    },
+    {
+      "epoch": 0.9002815807359937,
+      "grad_norm": 2.0285896518359925,
+      "learning_rate": 3.746700896629582e-08,
+      "loss": 2.4622,
+      "step": 6954
+    },
+    {
+      "epoch": 0.9004110431433473,
+      "grad_norm": 1.8488577650294844,
+      "learning_rate": 3.73707362450744e-08,
+      "loss": 2.4185,
+      "step": 6955
+    },
+    {
+      "epoch": 0.9004110431433473,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.4964488744735718,
+      "eval_runtime": 14.9389,
+      "eval_samples_per_second": 2.945,
+      "eval_steps_per_second": 0.201,
+      "step": 6955
+    },
+    {
+      "epoch": 0.9005405055507008,
+      "grad_norm": 1.7402328998270247,
+      "learning_rate": 3.7274583559040664e-08,
+      "loss": 2.4688,
+      "step": 6956
+    },
+    {
+      "epoch": 0.9006699679580542,
+      "grad_norm": 1.349800529381217,
+      "learning_rate": 3.717855092783593e-08,
+      "loss": 2.4836,
+      "step": 6957
+    },
+    {
+      "epoch": 0.9007994303654077,
+      "grad_norm": 1.8948832767582333,
+      "learning_rate": 3.7082638371076855e-08,
+      "loss": 2.2905,
+      "step": 6958
+    },
+    {
+      "epoch": 0.9009288927727611,
+      "grad_norm": 3.8180385902227485,
+      "learning_rate": 3.698684590835541e-08,
+      "loss": 2.5107,
+      "step": 6959
+    },
+    {
+      "epoch": 0.9010583551801146,
+      "grad_norm": 1.3492656096497837,
+      "learning_rate": 3.689117355923921e-08,
+      "loss": 2.2882,
+      "step": 6960
+    },
+    {
+      "epoch": 0.9010583551801146,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.4964710474014282,
+      "eval_runtime": 14.8863,
+      "eval_samples_per_second": 2.956,
+      "eval_steps_per_second": 0.202,
+      "step": 6960
+    },
+    {
+      "epoch": 0.901187817587468,
+      "grad_norm": 1.5259868074002851,
+      "learning_rate": 3.679562134327116e-08,
+      "loss": 2.2678,
+      "step": 6961
+    },
+    {
+      "epoch": 0.9013172799948215,
+      "grad_norm": 1.7950690981979358,
+      "learning_rate": 3.6700189279969755e-08,
+      "loss": 2.3286,
+      "step": 6962
+    },
+    {
+      "epoch": 0.9014467424021749,
+      "grad_norm": 2.053537008849433,
+      "learning_rate": 3.6604877388829104e-08,
+      "loss": 2.3889,
+      "step": 6963
+    },
+    {
+      "epoch": 0.9015762048095284,
+      "grad_norm": 1.3487691471993548,
+      "learning_rate": 3.650968568931837e-08,
+      "loss": 2.4321,
+      "step": 6964
+    },
+    {
+      "epoch": 0.901705667216882,
+      "grad_norm": 2.046988425631279,
+      "learning_rate": 3.641461420088262e-08,
+      "loss": 2.501,
+      "step": 6965
+    },
+    {
+      "epoch": 0.901705667216882,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.4957829713821411,
+      "eval_runtime": 14.4573,
+      "eval_samples_per_second": 3.043,
+      "eval_steps_per_second": 0.208,
+      "step": 6965
+    },
+    {
+      "epoch": 0.9018351296242354,
+      "grad_norm": 2.0775129527511913,
+      "learning_rate": 3.631966294294194e-08,
+      "loss": 2.468,
+      "step": 6966
+    },
+    {
+      "epoch": 0.9019645920315889,
+      "grad_norm": 1.3316849011000063,
+      "learning_rate": 3.6224831934892066e-08,
+      "loss": 2.261,
+      "step": 6967
+    },
+    {
+      "epoch": 0.9020940544389423,
+      "grad_norm": 1.1346699448886952,
+      "learning_rate": 3.613012119610425e-08,
+      "loss": 2.1799,
+      "step": 6968
+    },
+    {
+      "epoch": 0.9022235168462958,
+      "grad_norm": 1.6321728237788975,
+      "learning_rate": 3.603553074592495e-08,
+      "loss": 2.4453,
+      "step": 6969
+    },
+    {
+      "epoch": 0.9023529792536492,
+      "grad_norm": 1.4179594472025552,
+      "learning_rate": 3.594106060367635e-08,
+      "loss": 2.251,
+      "step": 6970
+    },
+    {
+      "epoch": 0.9023529792536492,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.4964045286178589,
+      "eval_runtime": 13.8209,
+      "eval_samples_per_second": 3.184,
+      "eval_steps_per_second": 0.217,
+      "step": 6970
+    },
+    {
+      "epoch": 0.9024824416610027,
+      "grad_norm": 1.3813409505729468,
+      "learning_rate": 3.584671078865586e-08,
+      "loss": 2.2354,
+      "step": 6971
+    },
+    {
+      "epoch": 0.9026119040683561,
+      "grad_norm": 1.3871296942601798,
+      "learning_rate": 3.5752481320136115e-08,
+      "loss": 2.4346,
+      "step": 6972
+    },
+    {
+      "epoch": 0.9027413664757096,
+      "grad_norm": 1.7645533751442184,
+      "learning_rate": 3.5658372217365694e-08,
+      "loss": 2.304,
+      "step": 6973
+    },
+    {
+      "epoch": 0.902870828883063,
+      "grad_norm": 1.5786955515060197,
+      "learning_rate": 3.556438349956802e-08,
+      "loss": 2.4717,
+      "step": 6974
+    },
+    {
+      "epoch": 0.9030002912904166,
+      "grad_norm": 1.6999238827098182,
+      "learning_rate": 3.547051518594247e-08,
+      "loss": 2.3564,
+      "step": 6975
+    },
+    {
+      "epoch": 0.9030002912904166,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.4950950145721436,
+      "eval_runtime": 14.0199,
+      "eval_samples_per_second": 3.138,
+      "eval_steps_per_second": 0.214,
+      "step": 6975
+    },
+    {
+      "epoch": 0.90312975369777,
+      "grad_norm": 2.3180146528633694,
+      "learning_rate": 3.5376767295663416e-08,
+      "loss": 2.5898,
+      "step": 6976
+    },
+    {
+      "epoch": 0.9032592161051235,
+      "grad_norm": 1.2525781138706877,
+      "learning_rate": 3.528313984788069e-08,
+      "loss": 2.4712,
+      "step": 6977
+    },
+    {
+      "epoch": 0.903388678512477,
+      "grad_norm": 1.5146753675865754,
+      "learning_rate": 3.5189632861719686e-08,
+      "loss": 2.4648,
+      "step": 6978
+    },
+    {
+      "epoch": 0.9035181409198304,
+      "grad_norm": 2.022322291432427,
+      "learning_rate": 3.509624635628102e-08,
+      "loss": 2.3547,
+      "step": 6979
+    },
+    {
+      "epoch": 0.9036476033271839,
+      "grad_norm": 2.0150281599560267,
+      "learning_rate": 3.5002980350640895e-08,
+      "loss": 2.2002,
+      "step": 6980
+    },
+    {
+      "epoch": 0.9036476033271839,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.495649814605713,
+      "eval_runtime": 14.2311,
+      "eval_samples_per_second": 3.092,
+      "eval_steps_per_second": 0.211,
+      "step": 6980
+    },
+    {
+      "epoch": 0.9037770657345373,
+      "grad_norm": 1.4981461399585436,
+      "learning_rate": 3.4909834863850746e-08,
+      "loss": 2.4382,
+      "step": 6981
+    },
+    {
+      "epoch": 0.9039065281418908,
+      "grad_norm": 1.4588762847333872,
+      "learning_rate": 3.4816809914937423e-08,
+      "loss": 2.3455,
+      "step": 6982
+    },
+    {
+      "epoch": 0.9040359905492442,
+      "grad_norm": 1.3739131419699016,
+      "learning_rate": 3.47239055229031e-08,
+      "loss": 2.4976,
+      "step": 6983
+    },
+    {
+      "epoch": 0.9041654529565978,
+      "grad_norm": 1.7316968704340205,
+      "learning_rate": 3.463112170672543e-08,
+      "loss": 2.3793,
+      "step": 6984
+    },
+    {
+      "epoch": 0.9042949153639512,
+      "grad_norm": 2.545602064433286,
+      "learning_rate": 3.453845848535739e-08,
+      "loss": 2.3589,
+      "step": 6985
+    },
+    {
+      "epoch": 0.9042949153639512,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.4943625926971436,
+      "eval_runtime": 13.8958,
+      "eval_samples_per_second": 3.166,
+      "eval_steps_per_second": 0.216,
+      "step": 6985
+    },
+    {
+      "epoch": 0.9044243777713047,
+      "grad_norm": 1.3981146941723948,
+      "learning_rate": 3.4445915877727375e-08,
+      "loss": 2.1975,
+      "step": 6986
+    },
+    {
+      "epoch": 0.9045538401786581,
+      "grad_norm": 1.7952518086041092,
+      "learning_rate": 3.435349390273901e-08,
+      "loss": 2.5269,
+      "step": 6987
+    },
+    {
+      "epoch": 0.9046833025860116,
+      "grad_norm": 1.6254409535581296,
+      "learning_rate": 3.426119257927123e-08,
+      "loss": 2.3235,
+      "step": 6988
+    },
+    {
+      "epoch": 0.904812764993365,
+      "grad_norm": 1.5820971972696911,
+      "learning_rate": 3.41690119261788e-08,
+      "loss": 2.4558,
+      "step": 6989
+    },
+    {
+      "epoch": 0.9049422274007185,
+      "grad_norm": 3.7090017497336105,
+      "learning_rate": 3.407695196229112e-08,
+      "loss": 2.1719,
+      "step": 6990
+    },
+    {
+      "epoch": 0.9049422274007185,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.4948952198028564,
+      "eval_runtime": 13.8486,
+      "eval_samples_per_second": 3.177,
+      "eval_steps_per_second": 0.217,
+      "step": 6990
+    },
+    {
+      "epoch": 0.905071689808072,
+      "grad_norm": 1.8466436865878584,
+      "learning_rate": 3.3985012706413553e-08,
+      "loss": 2.2668,
+      "step": 6991
+    },
+    {
+      "epoch": 0.9052011522154254,
+      "grad_norm": 3.529990179316978,
+      "learning_rate": 3.389319417732642e-08,
+      "loss": 2.509,
+      "step": 6992
+    },
+    {
+      "epoch": 0.9053306146227789,
+      "grad_norm": 1.642571111915576,
+      "learning_rate": 3.380149639378563e-08,
+      "loss": 2.5679,
+      "step": 6993
+    },
+    {
+      "epoch": 0.9054600770301324,
+      "grad_norm": 1.5897813153183968,
+      "learning_rate": 3.370991937452224e-08,
+      "loss": 2.5044,
+      "step": 6994
+    },
+    {
+      "epoch": 0.9055895394374859,
+      "grad_norm": 2.1450771090293044,
+      "learning_rate": 3.361846313824267e-08,
+      "loss": 2.459,
+      "step": 6995
+    },
+    {
+      "epoch": 0.9055895394374859,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "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": 14.3434,
+      "eval_samples_per_second": 3.068,
+      "eval_steps_per_second": 0.209,
+      "step": 6995
+    },
+    {
+      "epoch": 0.9057190018448393,
+      "grad_norm": 1.90263647944974,
+      "learning_rate": 3.3527127703628784e-08,
+      "loss": 2.4326,
+      "step": 6996
+    },
+    {
+      "epoch": 0.9058484642521928,
+      "grad_norm": 1.9344058518450322,
+      "learning_rate": 3.343591308933761e-08,
+      "loss": 2.5249,
+      "step": 6997
+    },
+    {
+      "epoch": 0.9059779266595462,
+      "grad_norm": 2.1295819523747905,
+      "learning_rate": 3.3344819314001794e-08,
+      "loss": 2.4021,
+      "step": 6998
+    },
+    {
+      "epoch": 0.9061073890668997,
+      "grad_norm": 1.4863903839298132,
+      "learning_rate": 3.32538463962289e-08,
+      "loss": 2.4209,
+      "step": 6999
+    },
+    {
+      "epoch": 0.9062368514742531,
+      "grad_norm": 1.3075175734582196,
+      "learning_rate": 3.316299435460191e-08,
+      "loss": 2.3711,
+      "step": 7000
+    },
+    {
+      "epoch": 0.9062368514742531,
+      "eval_PRM Accuracy": 0.2358490566037736,
+      "eval_PRM F1": 0.047058823529411764,
+      "eval_PRM F1 AUC": 0.5120481927710844,
+      "eval_PRM F1 Neg": 0.36220472440944884,
+      "eval_PRM NPV": 0.22115384615384615,
+      "eval_PRM Precision": 1.0,
+      "eval_PRM Recall": 0.024096385542168676,
+      "eval_PRM Specificty": 1.0,
+      "eval_loss": 1.4944069385528564,
+      "eval_runtime": 14.6662,
+      "eval_samples_per_second": 3.0,
+      "eval_steps_per_second": 0.205,
+      "step": 7000
     }
   ],
   "logging_steps": 1,
@@ -66342,7 +71442,7 @@
       "attributes": {}
     }
   },
-  "total_flos": 5466858348052480.0,
+  "total_flos": 5886714695548928.0,
   "train_batch_size": 4,
   "trial_name": null,
   "trial_params": null