diff --git "a/trainer_state.json" "b/trainer_state.json"
--- "a/trainer_state.json"
+++ "b/trainer_state.json"
@@ -1,1345 +1,19568 @@
 {
-  "best_metric": 0.9957906415192848,
-  "best_model_checkpoint": "FFPP-Raw_1FPS_faces\\checkpoint-2154",
-  "epoch": 2.997912317327766,
+  "best_metric": 0.99837772836593,
+  "best_model_checkpoint": "FFPP-Raw_1FPS_faces-expand-0-aligned\\checkpoint-26172",
+  "epoch": 19.99274047186933,
   "eval_steps": 500,
-  "global_step": 2154,
+  "global_step": 27540,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
     {
       "epoch": 0.01,
-      "learning_rate": 2.3148148148148148e-06,
-      "loss": 0.7001,
+      "grad_norm": 8.668222427368164,
+      "learning_rate": 1.815541031227306e-07,
+      "loss": 0.8207,
       "step": 10
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 4.6296296296296296e-06,
-      "loss": 0.5934,
+      "epoch": 0.01,
+      "grad_norm": 9.68768310546875,
+      "learning_rate": 3.631082062454612e-07,
+      "loss": 0.8143,
       "step": 20
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 6.944444444444445e-06,
-      "loss": 0.5296,
+      "epoch": 0.02,
+      "grad_norm": 7.637276649475098,
+      "learning_rate": 5.446623093681918e-07,
+      "loss": 0.7725,
       "step": 30
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 9.259259259259259e-06,
-      "loss": 0.5492,
+      "epoch": 0.03,
+      "grad_norm": 6.712704181671143,
+      "learning_rate": 7.262164124909224e-07,
+      "loss": 0.7328,
       "step": 40
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 1.1574074074074075e-05,
-      "loss": 0.5059,
+      "epoch": 0.04,
+      "grad_norm": 8.824642181396484,
+      "learning_rate": 9.077705156136529e-07,
+      "loss": 0.6809,
       "step": 50
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 1.388888888888889e-05,
-      "loss": 0.4859,
+      "epoch": 0.04,
+      "grad_norm": 5.087621688842773,
+      "learning_rate": 1.0893246187363836e-06,
+      "loss": 0.6397,
       "step": 60
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 1.6203703703703704e-05,
-      "loss": 0.514,
+      "epoch": 0.05,
+      "grad_norm": 4.494392395019531,
+      "learning_rate": 1.2708787218591142e-06,
+      "loss": 0.5765,
       "step": 70
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 1.8518518518518518e-05,
-      "loss": 0.4542,
+      "epoch": 0.06,
+      "grad_norm": 3.251326084136963,
+      "learning_rate": 1.4524328249818447e-06,
+      "loss": 0.5615,
       "step": 80
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 2.0833333333333336e-05,
-      "loss": 0.44,
+      "epoch": 0.07,
+      "grad_norm": 2.6707677841186523,
+      "learning_rate": 1.6339869281045753e-06,
+      "loss": 0.5184,
       "step": 90
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 2.314814814814815e-05,
-      "loss": 0.4303,
+      "epoch": 0.07,
+      "grad_norm": 3.2091023921966553,
+      "learning_rate": 1.8155410312273058e-06,
+      "loss": 0.5496,
       "step": 100
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 2.5462962962962965e-05,
-      "loss": 0.4103,
+      "epoch": 0.08,
+      "grad_norm": 2.7207913398742676,
+      "learning_rate": 1.9970951343500366e-06,
+      "loss": 0.531,
       "step": 110
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 2.777777777777778e-05,
-      "loss": 0.3543,
+      "epoch": 0.09,
+      "grad_norm": 2.7199337482452393,
+      "learning_rate": 2.178649237472767e-06,
+      "loss": 0.5033,
       "step": 120
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 3.0092592592592593e-05,
-      "loss": 0.3291,
+      "epoch": 0.09,
+      "grad_norm": 4.677346229553223,
+      "learning_rate": 2.3602033405954977e-06,
+      "loss": 0.5303,
       "step": 130
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 3.240740740740741e-05,
-      "loss": 0.2946,
+      "epoch": 0.1,
+      "grad_norm": 2.0803098678588867,
+      "learning_rate": 2.5417574437182283e-06,
+      "loss": 0.4915,
       "step": 140
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 3.472222222222222e-05,
-      "loss": 0.2761,
+      "epoch": 0.11,
+      "grad_norm": 3.441002130508423,
+      "learning_rate": 2.723311546840959e-06,
+      "loss": 0.5115,
       "step": 150
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 3.7037037037037037e-05,
-      "loss": 0.2661,
+      "epoch": 0.12,
+      "grad_norm": 2.644590377807617,
+      "learning_rate": 2.9048656499636894e-06,
+      "loss": 0.5185,
       "step": 160
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 3.935185185185186e-05,
-      "loss": 0.2208,
+      "epoch": 0.12,
+      "grad_norm": 3.385890245437622,
+      "learning_rate": 3.0864197530864196e-06,
+      "loss": 0.5344,
       "step": 170
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 4.166666666666667e-05,
-      "loss": 0.1985,
+      "epoch": 0.13,
+      "grad_norm": 2.4812233448028564,
+      "learning_rate": 3.2679738562091506e-06,
+      "loss": 0.5047,
       "step": 180
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 4.3981481481481486e-05,
-      "loss": 0.2291,
+      "epoch": 0.14,
+      "grad_norm": 4.330173969268799,
+      "learning_rate": 3.449527959331881e-06,
+      "loss": 0.4803,
       "step": 190
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 4.62962962962963e-05,
-      "loss": 0.1997,
+      "epoch": 0.15,
+      "grad_norm": 3.671731948852539,
+      "learning_rate": 3.6310820624546117e-06,
+      "loss": 0.4431,
       "step": 200
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 4.8611111111111115e-05,
-      "loss": 0.1806,
+      "epoch": 0.15,
+      "grad_norm": 3.001225233078003,
+      "learning_rate": 3.8126361655773422e-06,
+      "loss": 0.4784,
       "step": 210
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 4.98968008255934e-05,
-      "loss": 0.2138,
+      "epoch": 0.16,
+      "grad_norm": 6.147005558013916,
+      "learning_rate": 3.994190268700073e-06,
+      "loss": 0.4897,
       "step": 220
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 4.963880288957689e-05,
-      "loss": 0.1558,
+      "epoch": 0.17,
+      "grad_norm": 3.842711925506592,
+      "learning_rate": 4.175744371822803e-06,
+      "loss": 0.4645,
       "step": 230
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 4.9380804953560375e-05,
-      "loss": 0.1671,
+      "epoch": 0.17,
+      "grad_norm": 5.135747909545898,
+      "learning_rate": 4.357298474945534e-06,
+      "loss": 0.4478,
       "step": 240
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 4.912280701754386e-05,
-      "loss": 0.1739,
+      "epoch": 0.18,
+      "grad_norm": 5.1160359382629395,
+      "learning_rate": 4.5388525780682645e-06,
+      "loss": 0.4059,
       "step": 250
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 4.886480908152735e-05,
-      "loss": 0.1719,
+      "epoch": 0.19,
+      "grad_norm": 16.816068649291992,
+      "learning_rate": 4.7204066811909955e-06,
+      "loss": 0.3976,
       "step": 260
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 4.860681114551084e-05,
-      "loss": 0.1419,
+      "epoch": 0.2,
+      "grad_norm": 7.460231304168701,
+      "learning_rate": 4.901960784313726e-06,
+      "loss": 0.3889,
       "step": 270
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 4.8348813209494324e-05,
-      "loss": 0.1418,
+      "epoch": 0.2,
+      "grad_norm": 6.641134738922119,
+      "learning_rate": 5.083514887436457e-06,
+      "loss": 0.3661,
       "step": 280
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 4.809081527347781e-05,
-      "loss": 0.1221,
+      "epoch": 0.21,
+      "grad_norm": 26.321130752563477,
+      "learning_rate": 5.265068990559187e-06,
+      "loss": 0.387,
       "step": 290
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 4.783281733746131e-05,
-      "loss": 0.1229,
+      "epoch": 0.22,
+      "grad_norm": 19.10285186767578,
+      "learning_rate": 5.446623093681918e-06,
+      "loss": 0.3531,
       "step": 300
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 4.757481940144479e-05,
-      "loss": 0.1402,
+      "epoch": 0.23,
+      "grad_norm": 10.938743591308594,
+      "learning_rate": 5.628177196804648e-06,
+      "loss": 0.3221,
       "step": 310
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 4.731682146542828e-05,
-      "loss": 0.1225,
+      "epoch": 0.23,
+      "grad_norm": 18.0218563079834,
+      "learning_rate": 5.809731299927379e-06,
+      "loss": 0.36,
       "step": 320
     },
     {
-      "epoch": 0.46,
-      "learning_rate": 4.705882352941177e-05,
-      "loss": 0.0878,
+      "epoch": 0.24,
+      "grad_norm": 18.682737350463867,
+      "learning_rate": 5.991285403050109e-06,
+      "loss": 0.3285,
       "step": 330
     },
     {
-      "epoch": 0.47,
-      "learning_rate": 4.6800825593395256e-05,
-      "loss": 0.1161,
+      "epoch": 0.25,
+      "grad_norm": 15.782907485961914,
+      "learning_rate": 6.172839506172839e-06,
+      "loss": 0.2897,
       "step": 340
     },
     {
-      "epoch": 0.49,
-      "learning_rate": 4.6542827657378745e-05,
-      "loss": 0.1083,
+      "epoch": 0.25,
+      "grad_norm": 8.58176326751709,
+      "learning_rate": 6.354393609295569e-06,
+      "loss": 0.2722,
       "step": 350
     },
     {
-      "epoch": 0.5,
-      "learning_rate": 4.6284829721362234e-05,
-      "loss": 0.1205,
+      "epoch": 0.26,
+      "grad_norm": 30.12000846862793,
+      "learning_rate": 6.535947712418301e-06,
+      "loss": 0.2498,
       "step": 360
     },
     {
-      "epoch": 0.51,
-      "learning_rate": 4.602683178534572e-05,
-      "loss": 0.1195,
+      "epoch": 0.27,
+      "grad_norm": 18.24226188659668,
+      "learning_rate": 6.717501815541031e-06,
+      "loss": 0.2572,
       "step": 370
     },
     {
-      "epoch": 0.53,
-      "learning_rate": 4.5768833849329204e-05,
-      "loss": 0.1068,
+      "epoch": 0.28,
+      "grad_norm": 8.81354808807373,
+      "learning_rate": 6.899055918663762e-06,
+      "loss": 0.2644,
       "step": 380
     },
     {
-      "epoch": 0.54,
-      "learning_rate": 4.551083591331269e-05,
-      "loss": 0.1346,
+      "epoch": 0.28,
+      "grad_norm": 19.555673599243164,
+      "learning_rate": 7.080610021786492e-06,
+      "loss": 0.2591,
       "step": 390
     },
     {
-      "epoch": 0.56,
-      "learning_rate": 4.525283797729619e-05,
-      "loss": 0.0835,
+      "epoch": 0.29,
+      "grad_norm": 28.524112701416016,
+      "learning_rate": 7.262164124909223e-06,
+      "loss": 0.2599,
       "step": 400
     },
     {
-      "epoch": 0.57,
-      "learning_rate": 4.499484004127967e-05,
-      "loss": 0.0831,
+      "epoch": 0.3,
+      "grad_norm": 23.34917449951172,
+      "learning_rate": 7.443718228031954e-06,
+      "loss": 0.2286,
       "step": 410
     },
     {
-      "epoch": 0.58,
-      "learning_rate": 4.473684210526316e-05,
-      "loss": 0.087,
+      "epoch": 0.3,
+      "grad_norm": 14.82242202758789,
+      "learning_rate": 7.6252723311546845e-06,
+      "loss": 0.1988,
       "step": 420
     },
     {
-      "epoch": 0.6,
-      "learning_rate": 4.447884416924665e-05,
-      "loss": 0.0997,
+      "epoch": 0.31,
+      "grad_norm": 16.295961380004883,
+      "learning_rate": 7.806826434277415e-06,
+      "loss": 0.2202,
       "step": 430
     },
     {
-      "epoch": 0.61,
-      "learning_rate": 4.422084623323014e-05,
-      "loss": 0.0901,
+      "epoch": 0.32,
+      "grad_norm": 10.531784057617188,
+      "learning_rate": 7.988380537400146e-06,
+      "loss": 0.2101,
       "step": 440
     },
     {
-      "epoch": 0.63,
-      "learning_rate": 4.3962848297213626e-05,
-      "loss": 0.0917,
+      "epoch": 0.33,
+      "grad_norm": 35.081485748291016,
+      "learning_rate": 8.169934640522877e-06,
+      "loss": 0.2134,
       "step": 450
     },
     {
-      "epoch": 0.64,
-      "learning_rate": 4.3704850361197114e-05,
-      "loss": 0.0705,
+      "epoch": 0.33,
+      "grad_norm": 19.544342041015625,
+      "learning_rate": 8.351488743645607e-06,
+      "loss": 0.2354,
       "step": 460
     },
     {
-      "epoch": 0.65,
-      "learning_rate": 4.34468524251806e-05,
-      "loss": 0.0699,
+      "epoch": 0.34,
+      "grad_norm": 18.607032775878906,
+      "learning_rate": 8.533042846768337e-06,
+      "loss": 0.2113,
       "step": 470
     },
     {
-      "epoch": 0.67,
-      "learning_rate": 4.3188854489164085e-05,
-      "loss": 0.0926,
+      "epoch": 0.35,
+      "grad_norm": 27.145376205444336,
+      "learning_rate": 8.714596949891069e-06,
+      "loss": 0.1963,
       "step": 480
     },
     {
-      "epoch": 0.68,
-      "learning_rate": 4.2930856553147574e-05,
-      "loss": 0.0719,
+      "epoch": 0.36,
+      "grad_norm": 54.49102020263672,
+      "learning_rate": 8.896151053013799e-06,
+      "loss": 0.2754,
       "step": 490
     },
     {
-      "epoch": 0.7,
-      "learning_rate": 4.267285861713107e-05,
-      "loss": 0.0771,
+      "epoch": 0.36,
+      "grad_norm": 15.611857414245605,
+      "learning_rate": 9.077705156136529e-06,
+      "loss": 0.1953,
       "step": 500
     },
     {
-      "epoch": 0.71,
-      "learning_rate": 4.241486068111455e-05,
-      "loss": 0.0899,
+      "epoch": 0.37,
+      "grad_norm": 10.222444534301758,
+      "learning_rate": 9.259259259259259e-06,
+      "loss": 0.1828,
       "step": 510
     },
     {
-      "epoch": 0.72,
-      "learning_rate": 4.215686274509804e-05,
-      "loss": 0.0662,
+      "epoch": 0.38,
+      "grad_norm": 30.64798355102539,
+      "learning_rate": 9.440813362381991e-06,
+      "loss": 0.1767,
       "step": 520
     },
     {
-      "epoch": 0.74,
-      "learning_rate": 4.189886480908153e-05,
-      "loss": 0.0601,
+      "epoch": 0.38,
+      "grad_norm": 29.964618682861328,
+      "learning_rate": 9.622367465504721e-06,
+      "loss": 0.2311,
       "step": 530
     },
     {
-      "epoch": 0.75,
-      "learning_rate": 4.164086687306502e-05,
-      "loss": 0.0639,
+      "epoch": 0.39,
+      "grad_norm": 16.04256820678711,
+      "learning_rate": 9.803921568627451e-06,
+      "loss": 0.207,
       "step": 540
     },
     {
-      "epoch": 0.77,
-      "learning_rate": 4.1382868937048506e-05,
-      "loss": 0.075,
+      "epoch": 0.4,
+      "grad_norm": 12.842723846435547,
+      "learning_rate": 9.985475671750181e-06,
+      "loss": 0.2057,
       "step": 550
     },
     {
-      "epoch": 0.78,
-      "learning_rate": 4.1124871001031995e-05,
-      "loss": 0.0772,
+      "epoch": 0.41,
+      "grad_norm": 20.018783569335938,
+      "learning_rate": 1.0167029774872913e-05,
+      "loss": 0.1949,
       "step": 560
     },
     {
-      "epoch": 0.79,
-      "learning_rate": 4.0866873065015484e-05,
-      "loss": 0.074,
+      "epoch": 0.41,
+      "grad_norm": 12.877035140991211,
+      "learning_rate": 1.0348583877995643e-05,
+      "loss": 0.1891,
       "step": 570
     },
     {
-      "epoch": 0.81,
-      "learning_rate": 4.0608875128998966e-05,
-      "loss": 0.0844,
+      "epoch": 0.42,
+      "grad_norm": 40.139564514160156,
+      "learning_rate": 1.0530137981118374e-05,
+      "loss": 0.1735,
       "step": 580
     },
     {
-      "epoch": 0.82,
-      "learning_rate": 4.0350877192982455e-05,
-      "loss": 0.0727,
+      "epoch": 0.43,
+      "grad_norm": 27.097129821777344,
+      "learning_rate": 1.0711692084241104e-05,
+      "loss": 0.1776,
       "step": 590
     },
     {
-      "epoch": 0.84,
-      "learning_rate": 4.009287925696595e-05,
-      "loss": 0.0763,
+      "epoch": 0.44,
+      "grad_norm": 29.121681213378906,
+      "learning_rate": 1.0893246187363835e-05,
+      "loss": 0.1834,
       "step": 600
     },
     {
-      "epoch": 0.85,
-      "learning_rate": 3.983488132094943e-05,
-      "loss": 0.0785,
+      "epoch": 0.44,
+      "grad_norm": 12.37022876739502,
+      "learning_rate": 1.1074800290486566e-05,
+      "loss": 0.197,
       "step": 610
     },
     {
-      "epoch": 0.86,
-      "learning_rate": 3.957688338493292e-05,
-      "loss": 0.0837,
+      "epoch": 0.45,
+      "grad_norm": 12.485506057739258,
+      "learning_rate": 1.1256354393609296e-05,
+      "loss": 0.1848,
       "step": 620
     },
     {
-      "epoch": 0.88,
-      "learning_rate": 3.931888544891641e-05,
-      "loss": 0.0877,
+      "epoch": 0.46,
+      "grad_norm": 19.3685359954834,
+      "learning_rate": 1.1437908496732026e-05,
+      "loss": 0.1557,
       "step": 630
     },
     {
-      "epoch": 0.89,
-      "learning_rate": 3.90608875128999e-05,
-      "loss": 0.0729,
+      "epoch": 0.46,
+      "grad_norm": 7.632847309112549,
+      "learning_rate": 1.1619462599854758e-05,
+      "loss": 0.1862,
       "step": 640
     },
     {
-      "epoch": 0.9,
-      "learning_rate": 3.880288957688339e-05,
-      "loss": 0.0603,
+      "epoch": 0.47,
+      "grad_norm": 9.411360740661621,
+      "learning_rate": 1.1801016702977488e-05,
+      "loss": 0.1519,
       "step": 650
     },
     {
-      "epoch": 0.92,
-      "learning_rate": 3.8544891640866876e-05,
-      "loss": 0.0655,
+      "epoch": 0.48,
+      "grad_norm": 14.69303035736084,
+      "learning_rate": 1.1982570806100218e-05,
+      "loss": 0.1724,
       "step": 660
     },
     {
-      "epoch": 0.93,
-      "learning_rate": 3.8286893704850365e-05,
-      "loss": 0.0388,
+      "epoch": 0.49,
+      "grad_norm": 21.442214965820312,
+      "learning_rate": 1.2164124909222948e-05,
+      "loss": 0.178,
       "step": 670
     },
     {
-      "epoch": 0.95,
-      "learning_rate": 3.8028895768833846e-05,
-      "loss": 0.0868,
+      "epoch": 0.49,
+      "grad_norm": 26.165117263793945,
+      "learning_rate": 1.2345679012345678e-05,
+      "loss": 0.1562,
       "step": 680
     },
     {
-      "epoch": 0.96,
-      "learning_rate": 3.7770897832817335e-05,
-      "loss": 0.0798,
+      "epoch": 0.5,
+      "grad_norm": 54.52130889892578,
+      "learning_rate": 1.2527233115468408e-05,
+      "loss": 0.1784,
       "step": 690
     },
     {
-      "epoch": 0.97,
-      "learning_rate": 3.751289989680083e-05,
-      "loss": 0.0619,
+      "epoch": 0.51,
+      "grad_norm": 13.07323932647705,
+      "learning_rate": 1.2708787218591139e-05,
+      "loss": 0.1642,
       "step": 700
     },
     {
-      "epoch": 0.99,
-      "learning_rate": 3.725490196078432e-05,
-      "loss": 0.0623,
+      "epoch": 0.52,
+      "grad_norm": 10.937763214111328,
+      "learning_rate": 1.2890341321713872e-05,
+      "loss": 0.1494,
       "step": 710
     },
     {
-      "epoch": 1.0,
-      "eval_accuracy": 0.9885357523548478,
-      "eval_loss": 0.03366275876760483,
-      "eval_runtime": 442.332,
-      "eval_samples_per_second": 207.848,
-      "eval_steps_per_second": 6.497,
-      "step": 718
-    },
-    {
-      "epoch": 1.0,
-      "learning_rate": 3.69969040247678e-05,
-      "loss": 0.0452,
+      "epoch": 0.52,
+      "grad_norm": 25.107704162597656,
+      "learning_rate": 1.3071895424836602e-05,
+      "loss": 0.1698,
       "step": 720
     },
     {
-      "epoch": 1.02,
-      "learning_rate": 3.673890608875129e-05,
-      "loss": 0.0728,
+      "epoch": 0.53,
+      "grad_norm": 16.78830909729004,
+      "learning_rate": 1.3253449527959332e-05,
+      "loss": 0.2017,
       "step": 730
     },
     {
-      "epoch": 1.03,
-      "learning_rate": 3.648090815273478e-05,
-      "loss": 0.0443,
+      "epoch": 0.54,
+      "grad_norm": 22.479267120361328,
+      "learning_rate": 1.3435003631082063e-05,
+      "loss": 0.2282,
       "step": 740
     },
     {
-      "epoch": 1.04,
-      "learning_rate": 3.622291021671827e-05,
-      "loss": 0.0509,
+      "epoch": 0.54,
+      "grad_norm": 12.196043014526367,
+      "learning_rate": 1.3616557734204793e-05,
+      "loss": 0.1682,
       "step": 750
     },
     {
-      "epoch": 1.06,
-      "learning_rate": 3.5964912280701756e-05,
-      "loss": 0.0589,
+      "epoch": 0.55,
+      "grad_norm": 6.735928535461426,
+      "learning_rate": 1.3798111837327524e-05,
+      "loss": 0.1308,
       "step": 760
     },
     {
-      "epoch": 1.07,
-      "learning_rate": 3.5706914344685245e-05,
-      "loss": 0.0684,
+      "epoch": 0.56,
+      "grad_norm": 21.126880645751953,
+      "learning_rate": 1.3979665940450255e-05,
+      "loss": 0.194,
       "step": 770
     },
     {
-      "epoch": 1.09,
-      "learning_rate": 3.5448916408668734e-05,
-      "loss": 0.0577,
+      "epoch": 0.57,
+      "grad_norm": 19.850435256958008,
+      "learning_rate": 1.4161220043572985e-05,
+      "loss": 0.1497,
       "step": 780
     },
     {
-      "epoch": 1.1,
-      "learning_rate": 3.5190918472652216e-05,
-      "loss": 0.046,
+      "epoch": 0.57,
+      "grad_norm": 8.330199241638184,
+      "learning_rate": 1.4342774146695717e-05,
+      "loss": 0.1469,
       "step": 790
     },
     {
-      "epoch": 1.11,
-      "learning_rate": 3.493292053663571e-05,
-      "loss": 0.0388,
+      "epoch": 0.58,
+      "grad_norm": 25.57647705078125,
+      "learning_rate": 1.4524328249818447e-05,
+      "loss": 0.1964,
       "step": 800
     },
     {
-      "epoch": 1.13,
-      "learning_rate": 3.46749226006192e-05,
-      "loss": 0.0496,
+      "epoch": 0.59,
+      "grad_norm": 21.645137786865234,
+      "learning_rate": 1.4705882352941177e-05,
+      "loss": 0.1665,
       "step": 810
     },
     {
-      "epoch": 1.14,
-      "learning_rate": 3.441692466460268e-05,
-      "loss": 0.0539,
+      "epoch": 0.6,
+      "grad_norm": 16.67516326904297,
+      "learning_rate": 1.4887436456063909e-05,
+      "loss": 0.2308,
       "step": 820
     },
     {
-      "epoch": 1.16,
-      "learning_rate": 3.415892672858617e-05,
-      "loss": 0.0584,
+      "epoch": 0.6,
+      "grad_norm": 15.145057678222656,
+      "learning_rate": 1.5068990559186639e-05,
+      "loss": 0.148,
       "step": 830
     },
     {
-      "epoch": 1.17,
-      "learning_rate": 3.390092879256966e-05,
-      "loss": 0.0859,
+      "epoch": 0.61,
+      "grad_norm": 25.186765670776367,
+      "learning_rate": 1.5250544662309369e-05,
+      "loss": 0.1624,
       "step": 840
     },
     {
-      "epoch": 1.18,
-      "learning_rate": 3.364293085655315e-05,
-      "loss": 0.055,
+      "epoch": 0.62,
+      "grad_norm": 10.655858993530273,
+      "learning_rate": 1.54320987654321e-05,
+      "loss": 0.1919,
       "step": 850
     },
     {
-      "epoch": 1.2,
-      "learning_rate": 3.338493292053664e-05,
-      "loss": 0.055,
+      "epoch": 0.62,
+      "grad_norm": 11.99652099609375,
+      "learning_rate": 1.561365286855483e-05,
+      "loss": 0.1785,
       "step": 860
     },
     {
-      "epoch": 1.21,
-      "learning_rate": 3.3126934984520126e-05,
-      "loss": 0.0569,
+      "epoch": 0.63,
+      "grad_norm": 24.033994674682617,
+      "learning_rate": 1.5795206971677563e-05,
+      "loss": 0.1879,
       "step": 870
     },
     {
-      "epoch": 1.22,
-      "learning_rate": 3.2868937048503615e-05,
-      "loss": 0.043,
+      "epoch": 0.64,
+      "grad_norm": 33.82864761352539,
+      "learning_rate": 1.5976761074800293e-05,
+      "loss": 0.1325,
       "step": 880
     },
     {
-      "epoch": 1.24,
-      "learning_rate": 3.2610939112487103e-05,
-      "loss": 0.0488,
+      "epoch": 0.65,
+      "grad_norm": 23.036666870117188,
+      "learning_rate": 1.6158315177923023e-05,
+      "loss": 0.1517,
       "step": 890
     },
     {
-      "epoch": 1.25,
-      "learning_rate": 3.235294117647059e-05,
-      "loss": 0.0426,
+      "epoch": 0.65,
+      "grad_norm": 9.412284851074219,
+      "learning_rate": 1.6339869281045753e-05,
+      "loss": 0.1152,
       "step": 900
     },
     {
-      "epoch": 1.27,
-      "learning_rate": 3.209494324045408e-05,
-      "loss": 0.0502,
+      "epoch": 0.66,
+      "grad_norm": 10.256453514099121,
+      "learning_rate": 1.6521423384168483e-05,
+      "loss": 0.0987,
       "step": 910
     },
     {
-      "epoch": 1.28,
-      "learning_rate": 3.183694530443756e-05,
-      "loss": 0.0289,
+      "epoch": 0.67,
+      "grad_norm": 23.826547622680664,
+      "learning_rate": 1.6702977487291213e-05,
+      "loss": 0.117,
       "step": 920
     },
     {
-      "epoch": 1.29,
-      "learning_rate": 3.157894736842105e-05,
-      "loss": 0.0652,
+      "epoch": 0.68,
+      "grad_norm": 6.3597517013549805,
+      "learning_rate": 1.6884531590413944e-05,
+      "loss": 0.1716,
       "step": 930
     },
     {
-      "epoch": 1.31,
-      "learning_rate": 3.132094943240455e-05,
-      "loss": 0.0526,
+      "epoch": 0.68,
+      "grad_norm": 7.3157639503479,
+      "learning_rate": 1.7066085693536674e-05,
+      "loss": 0.1555,
       "step": 940
     },
     {
-      "epoch": 1.32,
-      "learning_rate": 3.106295149638803e-05,
-      "loss": 0.0527,
+      "epoch": 0.69,
+      "grad_norm": 20.580583572387695,
+      "learning_rate": 1.7247639796659407e-05,
+      "loss": 0.1315,
       "step": 950
     },
     {
-      "epoch": 1.34,
-      "learning_rate": 3.080495356037152e-05,
-      "loss": 0.0438,
+      "epoch": 0.7,
+      "grad_norm": 45.27377700805664,
+      "learning_rate": 1.7429193899782137e-05,
+      "loss": 0.1621,
       "step": 960
     },
     {
-      "epoch": 1.35,
-      "learning_rate": 3.054695562435501e-05,
-      "loss": 0.0404,
+      "epoch": 0.7,
+      "grad_norm": 8.26957893371582,
+      "learning_rate": 1.7610748002904868e-05,
+      "loss": 0.1007,
       "step": 970
     },
     {
-      "epoch": 1.36,
-      "learning_rate": 3.0288957688338492e-05,
-      "loss": 0.0406,
+      "epoch": 0.71,
+      "grad_norm": 15.427580833435059,
+      "learning_rate": 1.7792302106027598e-05,
+      "loss": 0.1367,
       "step": 980
     },
     {
-      "epoch": 1.38,
-      "learning_rate": 3.0030959752321984e-05,
-      "loss": 0.0384,
+      "epoch": 0.72,
+      "grad_norm": 17.598981857299805,
+      "learning_rate": 1.7973856209150328e-05,
+      "loss": 0.1406,
       "step": 990
     },
     {
-      "epoch": 1.39,
-      "learning_rate": 2.9772961816305473e-05,
-      "loss": 0.0573,
+      "epoch": 0.73,
+      "grad_norm": 28.66621208190918,
+      "learning_rate": 1.8155410312273058e-05,
+      "loss": 0.1633,
       "step": 1000
     },
     {
-      "epoch": 1.41,
-      "learning_rate": 2.9514963880288958e-05,
-      "loss": 0.0542,
+      "epoch": 0.73,
+      "grad_norm": 17.14661979675293,
+      "learning_rate": 1.8336964415395788e-05,
+      "loss": 0.1162,
       "step": 1010
     },
     {
-      "epoch": 1.42,
-      "learning_rate": 2.9256965944272447e-05,
-      "loss": 0.0411,
+      "epoch": 0.74,
+      "grad_norm": 13.592978477478027,
+      "learning_rate": 1.8518518518518518e-05,
+      "loss": 0.2155,
       "step": 1020
     },
     {
-      "epoch": 1.43,
-      "learning_rate": 2.8998968008255932e-05,
-      "loss": 0.0428,
+      "epoch": 0.75,
+      "grad_norm": 36.90546798706055,
+      "learning_rate": 1.870007262164125e-05,
+      "loss": 0.1604,
       "step": 1030
     },
     {
-      "epoch": 1.45,
-      "learning_rate": 2.8740970072239425e-05,
-      "loss": 0.0634,
+      "epoch": 0.75,
+      "grad_norm": 11.94582748413086,
+      "learning_rate": 1.8881626724763982e-05,
+      "loss": 0.1997,
       "step": 1040
     },
     {
-      "epoch": 1.46,
-      "learning_rate": 2.8482972136222913e-05,
-      "loss": 0.0396,
+      "epoch": 0.76,
+      "grad_norm": 18.860889434814453,
+      "learning_rate": 1.9063180827886712e-05,
+      "loss": 0.1189,
       "step": 1050
     },
     {
-      "epoch": 1.48,
-      "learning_rate": 2.82249742002064e-05,
-      "loss": 0.0456,
+      "epoch": 0.77,
+      "grad_norm": 13.953115463256836,
+      "learning_rate": 1.9244734931009442e-05,
+      "loss": 0.1091,
       "step": 1060
     },
     {
-      "epoch": 1.49,
-      "learning_rate": 2.7966976264189887e-05,
-      "loss": 0.0491,
+      "epoch": 0.78,
+      "grad_norm": 8.162593841552734,
+      "learning_rate": 1.9426289034132172e-05,
+      "loss": 0.1032,
       "step": 1070
     },
     {
-      "epoch": 1.5,
-      "learning_rate": 2.7708978328173373e-05,
-      "loss": 0.0407,
+      "epoch": 0.78,
+      "grad_norm": 13.563828468322754,
+      "learning_rate": 1.9607843137254903e-05,
+      "loss": 0.1666,
       "step": 1080
     },
     {
-      "epoch": 1.52,
-      "learning_rate": 2.7450980392156865e-05,
-      "loss": 0.0466,
+      "epoch": 0.79,
+      "grad_norm": 9.292993545532227,
+      "learning_rate": 1.9789397240377633e-05,
+      "loss": 0.1254,
       "step": 1090
     },
     {
-      "epoch": 1.53,
-      "learning_rate": 2.7192982456140354e-05,
-      "loss": 0.0586,
+      "epoch": 0.8,
+      "grad_norm": 12.907480239868164,
+      "learning_rate": 1.9970951343500363e-05,
+      "loss": 0.1551,
       "step": 1100
     },
     {
-      "epoch": 1.54,
-      "learning_rate": 2.693498452012384e-05,
-      "loss": 0.0379,
+      "epoch": 0.81,
+      "grad_norm": 9.786681175231934,
+      "learning_rate": 2.0152505446623093e-05,
+      "loss": 0.1087,
       "step": 1110
     },
     {
-      "epoch": 1.56,
-      "learning_rate": 2.6676986584107328e-05,
-      "loss": 0.0402,
+      "epoch": 0.81,
+      "grad_norm": 11.040635108947754,
+      "learning_rate": 2.0334059549745826e-05,
+      "loss": 0.1282,
       "step": 1120
     },
     {
-      "epoch": 1.57,
-      "learning_rate": 2.6418988648090813e-05,
-      "loss": 0.0225,
+      "epoch": 0.82,
+      "grad_norm": 12.948040008544922,
+      "learning_rate": 2.0515613652868557e-05,
+      "loss": 0.1329,
       "step": 1130
     },
     {
-      "epoch": 1.59,
-      "learning_rate": 2.616099071207431e-05,
-      "loss": 0.0268,
+      "epoch": 0.83,
+      "grad_norm": 36.504207611083984,
+      "learning_rate": 2.0697167755991287e-05,
+      "loss": 0.1145,
       "step": 1140
     },
     {
-      "epoch": 1.6,
-      "learning_rate": 2.5902992776057794e-05,
-      "loss": 0.0301,
+      "epoch": 0.83,
+      "grad_norm": 9.630166053771973,
+      "learning_rate": 2.0878721859114017e-05,
+      "loss": 0.15,
       "step": 1150
     },
     {
-      "epoch": 1.61,
-      "learning_rate": 2.564499484004128e-05,
-      "loss": 0.041,
+      "epoch": 0.84,
+      "grad_norm": 22.309511184692383,
+      "learning_rate": 2.1060275962236747e-05,
+      "loss": 0.2228,
       "step": 1160
     },
     {
-      "epoch": 1.63,
-      "learning_rate": 2.5386996904024768e-05,
-      "loss": 0.034,
+      "epoch": 0.85,
+      "grad_norm": 22.357301712036133,
+      "learning_rate": 2.1241830065359477e-05,
+      "loss": 0.149,
       "step": 1170
     },
     {
-      "epoch": 1.64,
-      "learning_rate": 2.5128998968008253e-05,
-      "loss": 0.0428,
+      "epoch": 0.86,
+      "grad_norm": 7.280608177185059,
+      "learning_rate": 2.1423384168482207e-05,
+      "loss": 0.1373,
       "step": 1180
     },
     {
-      "epoch": 1.66,
-      "learning_rate": 2.4871001031991746e-05,
-      "loss": 0.0514,
+      "epoch": 0.86,
+      "grad_norm": 6.142936706542969,
+      "learning_rate": 2.1604938271604937e-05,
+      "loss": 0.118,
       "step": 1190
     },
     {
-      "epoch": 1.67,
-      "learning_rate": 2.4613003095975234e-05,
-      "loss": 0.0399,
+      "epoch": 0.87,
+      "grad_norm": 5.754367828369141,
+      "learning_rate": 2.178649237472767e-05,
+      "loss": 0.1213,
       "step": 1200
     },
     {
-      "epoch": 1.68,
-      "learning_rate": 2.4355005159958723e-05,
-      "loss": 0.0519,
+      "epoch": 0.88,
+      "grad_norm": 10.323685646057129,
+      "learning_rate": 2.19680464778504e-05,
+      "loss": 0.1203,
       "step": 1210
     },
     {
-      "epoch": 1.7,
-      "learning_rate": 2.409700722394221e-05,
-      "loss": 0.0411,
+      "epoch": 0.89,
+      "grad_norm": 31.037113189697266,
+      "learning_rate": 2.214960058097313e-05,
+      "loss": 0.1149,
       "step": 1220
     },
     {
-      "epoch": 1.71,
-      "learning_rate": 2.3839009287925697e-05,
-      "loss": 0.0264,
+      "epoch": 0.89,
+      "grad_norm": 17.392948150634766,
+      "learning_rate": 2.233115468409586e-05,
+      "loss": 0.1326,
       "step": 1230
     },
     {
-      "epoch": 1.73,
-      "learning_rate": 2.3581011351909186e-05,
-      "loss": 0.0478,
+      "epoch": 0.9,
+      "grad_norm": 15.242566108703613,
+      "learning_rate": 2.251270878721859e-05,
+      "loss": 0.1098,
       "step": 1240
     },
     {
-      "epoch": 1.74,
-      "learning_rate": 2.3323013415892675e-05,
-      "loss": 0.0369,
+      "epoch": 0.91,
+      "grad_norm": 14.354398727416992,
+      "learning_rate": 2.269426289034132e-05,
+      "loss": 0.1288,
       "step": 1250
     },
     {
-      "epoch": 1.75,
-      "learning_rate": 2.3065015479876163e-05,
-      "loss": 0.0476,
+      "epoch": 0.91,
+      "grad_norm": 8.135055541992188,
+      "learning_rate": 2.2875816993464052e-05,
+      "loss": 0.1553,
       "step": 1260
     },
     {
-      "epoch": 1.77,
-      "learning_rate": 2.280701754385965e-05,
-      "loss": 0.0537,
+      "epoch": 0.92,
+      "grad_norm": 11.832168579101562,
+      "learning_rate": 2.3057371096586782e-05,
+      "loss": 0.1088,
       "step": 1270
     },
     {
-      "epoch": 1.78,
-      "learning_rate": 2.2549019607843138e-05,
-      "loss": 0.0387,
+      "epoch": 0.93,
+      "grad_norm": 15.753230094909668,
+      "learning_rate": 2.3238925199709515e-05,
+      "loss": 0.0886,
       "step": 1280
     },
     {
-      "epoch": 1.8,
-      "learning_rate": 2.2291021671826626e-05,
-      "loss": 0.0293,
+      "epoch": 0.94,
+      "grad_norm": 11.60594654083252,
+      "learning_rate": 2.3420479302832246e-05,
+      "loss": 0.1142,
       "step": 1290
     },
     {
-      "epoch": 1.81,
-      "learning_rate": 2.2033023735810115e-05,
-      "loss": 0.0298,
+      "epoch": 0.94,
+      "grad_norm": 9.862659454345703,
+      "learning_rate": 2.3602033405954976e-05,
+      "loss": 0.0894,
       "step": 1300
     },
     {
-      "epoch": 1.82,
-      "learning_rate": 2.1775025799793604e-05,
-      "loss": 0.0502,
+      "epoch": 0.95,
+      "grad_norm": 21.492176055908203,
+      "learning_rate": 2.3783587509077706e-05,
+      "loss": 0.1154,
       "step": 1310
     },
     {
-      "epoch": 1.84,
-      "learning_rate": 2.151702786377709e-05,
-      "loss": 0.0292,
+      "epoch": 0.96,
+      "grad_norm": 14.04377555847168,
+      "learning_rate": 2.3965141612200436e-05,
+      "loss": 0.1221,
       "step": 1320
     },
     {
-      "epoch": 1.85,
-      "learning_rate": 2.1259029927760578e-05,
-      "loss": 0.0307,
+      "epoch": 0.97,
+      "grad_norm": 7.398568153381348,
+      "learning_rate": 2.4146695715323166e-05,
+      "loss": 0.1275,
       "step": 1330
     },
     {
-      "epoch": 1.86,
-      "learning_rate": 2.1001031991744067e-05,
-      "loss": 0.0393,
+      "epoch": 0.97,
+      "grad_norm": 10.66848087310791,
+      "learning_rate": 2.4328249818445896e-05,
+      "loss": 0.0981,
       "step": 1340
     },
     {
-      "epoch": 1.88,
-      "learning_rate": 2.0743034055727555e-05,
-      "loss": 0.0363,
+      "epoch": 0.98,
+      "grad_norm": 7.866222858428955,
+      "learning_rate": 2.4509803921568626e-05,
+      "loss": 0.1048,
       "step": 1350
     },
     {
-      "epoch": 1.89,
-      "learning_rate": 2.0485036119711044e-05,
-      "loss": 0.0454,
+      "epoch": 0.99,
+      "grad_norm": 10.718284606933594,
+      "learning_rate": 2.4691358024691357e-05,
+      "loss": 0.1096,
       "step": 1360
     },
     {
-      "epoch": 1.91,
-      "learning_rate": 2.0227038183694533e-05,
-      "loss": 0.0274,
+      "epoch": 0.99,
+      "grad_norm": 38.74601745605469,
+      "learning_rate": 2.487291212781409e-05,
+      "loss": 0.0983,
       "step": 1370
     },
     {
-      "epoch": 1.92,
-      "learning_rate": 1.9969040247678018e-05,
-      "loss": 0.0207,
+      "epoch": 1.0,
+      "eval_accuracy": 0.9742818895493942,
+      "eval_f1": 0.9425071644137861,
+      "eval_loss": 0.0679459497332573,
+      "eval_precision": 0.916498150431566,
+      "eval_recall": 0.9700354980162873,
+      "eval_roc_auc": 0.9960748609284922,
+      "eval_runtime": 385.505,
+      "eval_samples_per_second": 228.656,
+      "eval_steps_per_second": 14.293,
+      "step": 1377
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 26.3389892578125,
+      "learning_rate": 2.5054466230936817e-05,
+      "loss": 0.1328,
       "step": 1380
     },
     {
-      "epoch": 1.93,
-      "learning_rate": 1.9711042311661507e-05,
-      "loss": 0.0288,
+      "epoch": 1.01,
+      "grad_norm": 7.549662113189697,
+      "learning_rate": 2.523602033405955e-05,
+      "loss": 0.1025,
       "step": 1390
     },
     {
-      "epoch": 1.95,
-      "learning_rate": 1.9453044375644996e-05,
-      "loss": 0.0328,
+      "epoch": 1.02,
+      "grad_norm": 15.942997932434082,
+      "learning_rate": 2.5417574437182277e-05,
+      "loss": 0.0898,
       "step": 1400
     },
     {
-      "epoch": 1.96,
-      "learning_rate": 1.9195046439628485e-05,
-      "loss": 0.0347,
+      "epoch": 1.02,
+      "grad_norm": 29.061594009399414,
+      "learning_rate": 2.559912854030501e-05,
+      "loss": 0.1226,
       "step": 1410
     },
     {
-      "epoch": 1.98,
-      "learning_rate": 1.8937048503611973e-05,
-      "loss": 0.0295,
+      "epoch": 1.03,
+      "grad_norm": 5.289678573608398,
+      "learning_rate": 2.5780682643427744e-05,
+      "loss": 0.0947,
       "step": 1420
     },
     {
-      "epoch": 1.99,
-      "learning_rate": 1.867905056759546e-05,
-      "loss": 0.0299,
+      "epoch": 1.04,
+      "grad_norm": 46.73320388793945,
+      "learning_rate": 2.596223674655047e-05,
+      "loss": 0.1063,
       "step": 1430
     },
     {
-      "epoch": 2.0,
-      "eval_accuracy": 0.9947464595705802,
-      "eval_loss": 0.016251688823103905,
-      "eval_runtime": 440.6692,
-      "eval_samples_per_second": 208.633,
-      "eval_steps_per_second": 6.522,
-      "step": 1437
-    },
-    {
-      "epoch": 2.0,
-      "learning_rate": 1.8421052631578947e-05,
-      "loss": 0.0374,
+      "epoch": 1.05,
+      "grad_norm": 36.990875244140625,
+      "learning_rate": 2.6143790849673204e-05,
+      "loss": 0.0916,
       "step": 1440
     },
     {
-      "epoch": 2.02,
-      "learning_rate": 1.8163054695562436e-05,
-      "loss": 0.0247,
+      "epoch": 1.05,
+      "grad_norm": 12.835508346557617,
+      "learning_rate": 2.632534495279593e-05,
+      "loss": 0.1137,
       "step": 1450
     },
     {
-      "epoch": 2.03,
-      "learning_rate": 1.7905056759545925e-05,
-      "loss": 0.0318,
+      "epoch": 1.06,
+      "grad_norm": 4.24277400970459,
+      "learning_rate": 2.6506899055918665e-05,
+      "loss": 0.1178,
       "step": 1460
     },
     {
-      "epoch": 2.05,
-      "learning_rate": 1.7647058823529414e-05,
-      "loss": 0.0253,
+      "epoch": 1.07,
+      "grad_norm": 24.840017318725586,
+      "learning_rate": 2.6688453159041395e-05,
+      "loss": 0.0862,
       "step": 1470
     },
     {
-      "epoch": 2.06,
-      "learning_rate": 1.7389060887512902e-05,
-      "loss": 0.0352,
+      "epoch": 1.07,
+      "grad_norm": 7.686561107635498,
+      "learning_rate": 2.6870007262164125e-05,
+      "loss": 0.0664,
       "step": 1480
     },
     {
-      "epoch": 2.07,
-      "learning_rate": 1.7131062951496388e-05,
-      "loss": 0.0302,
+      "epoch": 1.08,
+      "grad_norm": 9.543157577514648,
+      "learning_rate": 2.705156136528686e-05,
+      "loss": 0.1024,
       "step": 1490
     },
     {
-      "epoch": 2.09,
-      "learning_rate": 1.6873065015479876e-05,
-      "loss": 0.0149,
+      "epoch": 1.09,
+      "grad_norm": 17.78160858154297,
+      "learning_rate": 2.7233115468409585e-05,
+      "loss": 0.078,
       "step": 1500
     },
     {
-      "epoch": 2.1,
-      "learning_rate": 1.6615067079463365e-05,
-      "loss": 0.0139,
+      "epoch": 1.1,
+      "grad_norm": 43.5964241027832,
+      "learning_rate": 2.741466957153232e-05,
+      "loss": 0.1601,
       "step": 1510
     },
     {
-      "epoch": 2.12,
-      "learning_rate": 1.6357069143446854e-05,
-      "loss": 0.0289,
+      "epoch": 1.1,
+      "grad_norm": 9.158828735351562,
+      "learning_rate": 2.759622367465505e-05,
+      "loss": 0.1064,
       "step": 1520
     },
     {
-      "epoch": 2.13,
-      "learning_rate": 1.6099071207430343e-05,
-      "loss": 0.025,
+      "epoch": 1.11,
+      "grad_norm": 10.158084869384766,
+      "learning_rate": 2.777777777777778e-05,
+      "loss": 0.0888,
       "step": 1530
     },
     {
-      "epoch": 2.14,
-      "learning_rate": 1.5841073271413828e-05,
-      "loss": 0.0417,
+      "epoch": 1.12,
+      "grad_norm": 10.405439376831055,
+      "learning_rate": 2.795933188090051e-05,
+      "loss": 0.0721,
       "step": 1540
     },
     {
-      "epoch": 2.16,
-      "learning_rate": 1.5583075335397317e-05,
-      "loss": 0.0228,
+      "epoch": 1.13,
+      "grad_norm": 1.5054512023925781,
+      "learning_rate": 2.814088598402324e-05,
+      "loss": 0.0837,
       "step": 1550
     },
     {
-      "epoch": 2.17,
-      "learning_rate": 1.5325077399380806e-05,
-      "loss": 0.0187,
+      "epoch": 1.13,
+      "grad_norm": 15.655257225036621,
+      "learning_rate": 2.832244008714597e-05,
+      "loss": 0.099,
       "step": 1560
     },
     {
-      "epoch": 2.19,
-      "learning_rate": 1.5067079463364294e-05,
-      "loss": 0.0262,
+      "epoch": 1.14,
+      "grad_norm": 8.51606559753418,
+      "learning_rate": 2.85039941902687e-05,
+      "loss": 0.1474,
       "step": 1570
     },
     {
-      "epoch": 2.2,
-      "learning_rate": 1.4809081527347781e-05,
-      "loss": 0.0293,
+      "epoch": 1.15,
+      "grad_norm": 21.121200561523438,
+      "learning_rate": 2.8685548293391433e-05,
+      "loss": 0.1119,
       "step": 1580
     },
     {
-      "epoch": 2.21,
-      "learning_rate": 1.4551083591331268e-05,
-      "loss": 0.0292,
+      "epoch": 1.15,
+      "grad_norm": 7.0633673667907715,
+      "learning_rate": 2.8867102396514163e-05,
+      "loss": 0.1265,
       "step": 1590
     },
     {
-      "epoch": 2.23,
-      "learning_rate": 1.4293085655314759e-05,
-      "loss": 0.0362,
+      "epoch": 1.16,
+      "grad_norm": 35.018943786621094,
+      "learning_rate": 2.9048656499636893e-05,
+      "loss": 0.1585,
       "step": 1600
     },
     {
-      "epoch": 2.24,
-      "learning_rate": 1.4035087719298246e-05,
-      "loss": 0.0246,
+      "epoch": 1.17,
+      "grad_norm": 12.88159465789795,
+      "learning_rate": 2.9230210602759624e-05,
+      "loss": 0.1685,
       "step": 1610
     },
     {
-      "epoch": 2.25,
-      "learning_rate": 1.3777089783281735e-05,
-      "loss": 0.0212,
+      "epoch": 1.18,
+      "grad_norm": 15.151433944702148,
+      "learning_rate": 2.9411764705882354e-05,
+      "loss": 0.1472,
       "step": 1620
     },
     {
-      "epoch": 2.27,
-      "learning_rate": 1.3519091847265222e-05,
-      "loss": 0.012,
+      "epoch": 1.18,
+      "grad_norm": 4.823648452758789,
+      "learning_rate": 2.9593318809005084e-05,
+      "loss": 0.0755,
       "step": 1630
     },
     {
-      "epoch": 2.28,
-      "learning_rate": 1.3261093911248712e-05,
-      "loss": 0.0258,
+      "epoch": 1.19,
+      "grad_norm": 8.589825630187988,
+      "learning_rate": 2.9774872912127817e-05,
+      "loss": 0.1005,
       "step": 1640
     },
     {
-      "epoch": 2.3,
-      "learning_rate": 1.30030959752322e-05,
-      "loss": 0.022,
+      "epoch": 1.2,
+      "grad_norm": 11.609642028808594,
+      "learning_rate": 2.9956427015250548e-05,
+      "loss": 0.0907,
       "step": 1650
     },
     {
-      "epoch": 2.31,
-      "learning_rate": 1.2745098039215686e-05,
-      "loss": 0.0305,
+      "epoch": 1.21,
+      "grad_norm": 12.496771812438965,
+      "learning_rate": 3.0137981118373278e-05,
+      "loss": 0.1049,
       "step": 1660
     },
     {
-      "epoch": 2.32,
-      "learning_rate": 1.2487100103199175e-05,
-      "loss": 0.0262,
+      "epoch": 1.21,
+      "grad_norm": 9.315934181213379,
+      "learning_rate": 3.0319535221496008e-05,
+      "loss": 0.0993,
       "step": 1670
     },
     {
-      "epoch": 2.34,
-      "learning_rate": 1.2229102167182662e-05,
-      "loss": 0.0279,
+      "epoch": 1.22,
+      "grad_norm": 11.899771690368652,
+      "learning_rate": 3.0501089324618738e-05,
+      "loss": 0.0933,
       "step": 1680
     },
     {
-      "epoch": 2.35,
-      "learning_rate": 1.197110423116615e-05,
-      "loss": 0.026,
+      "epoch": 1.23,
+      "grad_norm": 6.324872970581055,
+      "learning_rate": 3.068264342774147e-05,
+      "loss": 0.1013,
       "step": 1690
     },
     {
-      "epoch": 2.37,
-      "learning_rate": 1.171310629514964e-05,
-      "loss": 0.0172,
+      "epoch": 1.23,
+      "grad_norm": 9.256946563720703,
+      "learning_rate": 3.08641975308642e-05,
+      "loss": 0.1089,
       "step": 1700
     },
     {
-      "epoch": 2.38,
-      "learning_rate": 1.1455108359133128e-05,
-      "loss": 0.0237,
+      "epoch": 1.24,
+      "grad_norm": 3.401277780532837,
+      "learning_rate": 3.104575163398693e-05,
+      "loss": 0.1013,
       "step": 1710
     },
     {
-      "epoch": 2.39,
-      "learning_rate": 1.1197110423116617e-05,
-      "loss": 0.0284,
+      "epoch": 1.25,
+      "grad_norm": 7.001280784606934,
+      "learning_rate": 3.122730573710966e-05,
+      "loss": 0.0752,
       "step": 1720
     },
     {
-      "epoch": 2.41,
-      "learning_rate": 1.0939112487100102e-05,
-      "loss": 0.0222,
+      "epoch": 1.26,
+      "grad_norm": 5.906143665313721,
+      "learning_rate": 3.140885984023239e-05,
+      "loss": 0.0926,
       "step": 1730
     },
     {
-      "epoch": 2.42,
-      "learning_rate": 1.0681114551083591e-05,
-      "loss": 0.0211,
+      "epoch": 1.26,
+      "grad_norm": 8.290616989135742,
+      "learning_rate": 3.1590413943355126e-05,
+      "loss": 0.0871,
       "step": 1740
     },
     {
-      "epoch": 2.44,
-      "learning_rate": 1.042311661506708e-05,
-      "loss": 0.0102,
+      "epoch": 1.27,
+      "grad_norm": 15.752880096435547,
+      "learning_rate": 3.177196804647785e-05,
+      "loss": 0.0794,
       "step": 1750
     },
     {
-      "epoch": 2.45,
-      "learning_rate": 1.0165118679050569e-05,
-      "loss": 0.0309,
+      "epoch": 1.28,
+      "grad_norm": 11.03674030303955,
+      "learning_rate": 3.1953522149600586e-05,
+      "loss": 0.0961,
       "step": 1760
     },
     {
-      "epoch": 2.46,
-      "learning_rate": 9.907120743034057e-06,
-      "loss": 0.018,
+      "epoch": 1.28,
+      "grad_norm": 13.24030876159668,
+      "learning_rate": 3.213507625272331e-05,
+      "loss": 0.089,
       "step": 1770
     },
     {
-      "epoch": 2.48,
-      "learning_rate": 9.649122807017545e-06,
-      "loss": 0.0301,
+      "epoch": 1.29,
+      "grad_norm": 22.704883575439453,
+      "learning_rate": 3.2316630355846046e-05,
+      "loss": 0.0917,
       "step": 1780
     },
     {
-      "epoch": 2.49,
-      "learning_rate": 9.391124871001032e-06,
-      "loss": 0.0337,
+      "epoch": 1.3,
+      "grad_norm": 4.330246925354004,
+      "learning_rate": 3.249818445896877e-05,
+      "loss": 0.118,
       "step": 1790
     },
     {
-      "epoch": 2.51,
-      "learning_rate": 9.13312693498452e-06,
-      "loss": 0.033,
+      "epoch": 1.31,
+      "grad_norm": 12.068599700927734,
+      "learning_rate": 3.2679738562091506e-05,
+      "loss": 0.1086,
       "step": 1800
     },
     {
-      "epoch": 2.52,
-      "learning_rate": 8.875128998968009e-06,
-      "loss": 0.034,
+      "epoch": 1.31,
+      "grad_norm": 6.073216438293457,
+      "learning_rate": 3.286129266521423e-05,
+      "loss": 0.1285,
       "step": 1810
     },
     {
-      "epoch": 2.53,
-      "learning_rate": 8.617131062951498e-06,
-      "loss": 0.0178,
+      "epoch": 1.32,
+      "grad_norm": 8.668619155883789,
+      "learning_rate": 3.304284676833697e-05,
+      "loss": 0.0784,
       "step": 1820
     },
     {
-      "epoch": 2.55,
-      "learning_rate": 8.359133126934985e-06,
-      "loss": 0.017,
+      "epoch": 1.33,
+      "grad_norm": 2.0747952461242676,
+      "learning_rate": 3.32244008714597e-05,
+      "loss": 0.0958,
       "step": 1830
     },
     {
-      "epoch": 2.56,
-      "learning_rate": 8.101135190918472e-06,
-      "loss": 0.0397,
+      "epoch": 1.34,
+      "grad_norm": 21.166549682617188,
+      "learning_rate": 3.340595497458243e-05,
+      "loss": 0.1532,
       "step": 1840
     },
     {
-      "epoch": 2.57,
-      "learning_rate": 7.84313725490196e-06,
-      "loss": 0.0221,
+      "epoch": 1.34,
+      "grad_norm": 19.43235969543457,
+      "learning_rate": 3.358750907770516e-05,
+      "loss": 0.1373,
       "step": 1850
     },
     {
-      "epoch": 2.59,
-      "learning_rate": 7.585139318885449e-06,
-      "loss": 0.0262,
+      "epoch": 1.35,
+      "grad_norm": 6.905423164367676,
+      "learning_rate": 3.376906318082789e-05,
+      "loss": 0.1291,
       "step": 1860
     },
     {
-      "epoch": 2.6,
-      "learning_rate": 7.327141382868938e-06,
-      "loss": 0.0322,
+      "epoch": 1.36,
+      "grad_norm": 17.511754989624023,
+      "learning_rate": 3.395061728395062e-05,
+      "loss": 0.1032,
       "step": 1870
     },
     {
-      "epoch": 2.62,
-      "learning_rate": 7.069143446852426e-06,
-      "loss": 0.0166,
+      "epoch": 1.36,
+      "grad_norm": 3.903677225112915,
+      "learning_rate": 3.413217138707335e-05,
+      "loss": 0.0923,
       "step": 1880
     },
     {
-      "epoch": 2.63,
-      "learning_rate": 6.811145510835913e-06,
-      "loss": 0.0235,
+      "epoch": 1.37,
+      "grad_norm": 11.573185920715332,
+      "learning_rate": 3.431372549019608e-05,
+      "loss": 0.0783,
       "step": 1890
     },
     {
-      "epoch": 2.64,
-      "learning_rate": 6.553147574819402e-06,
-      "loss": 0.0166,
+      "epoch": 1.38,
+      "grad_norm": 11.51346492767334,
+      "learning_rate": 3.4495279593318815e-05,
+      "loss": 0.0799,
       "step": 1900
     },
     {
-      "epoch": 2.66,
-      "learning_rate": 6.29514963880289e-06,
-      "loss": 0.0157,
+      "epoch": 1.39,
+      "grad_norm": 3.603184223175049,
+      "learning_rate": 3.467683369644154e-05,
+      "loss": 0.0777,
       "step": 1910
     },
     {
-      "epoch": 2.67,
-      "learning_rate": 6.0371517027863785e-06,
-      "loss": 0.0276,
+      "epoch": 1.39,
+      "grad_norm": 19.509370803833008,
+      "learning_rate": 3.4858387799564275e-05,
+      "loss": 0.1354,
       "step": 1920
     },
     {
-      "epoch": 2.69,
-      "learning_rate": 5.779153766769866e-06,
-      "loss": 0.0214,
+      "epoch": 1.4,
+      "grad_norm": 16.0355281829834,
+      "learning_rate": 3.5039941902687e-05,
+      "loss": 0.246,
       "step": 1930
     },
     {
-      "epoch": 2.7,
-      "learning_rate": 5.521155830753354e-06,
-      "loss": 0.0239,
+      "epoch": 1.41,
+      "grad_norm": 13.758955955505371,
+      "learning_rate": 3.5221496005809735e-05,
+      "loss": 0.0844,
       "step": 1940
     },
     {
-      "epoch": 2.71,
-      "learning_rate": 5.263157894736842e-06,
-      "loss": 0.0212,
+      "epoch": 1.42,
+      "grad_norm": 13.882157325744629,
+      "learning_rate": 3.540305010893246e-05,
+      "loss": 0.0778,
       "step": 1950
     },
     {
-      "epoch": 2.73,
-      "learning_rate": 5.00515995872033e-06,
-      "loss": 0.0276,
+      "epoch": 1.42,
+      "grad_norm": 15.574724197387695,
+      "learning_rate": 3.5584604212055195e-05,
+      "loss": 0.0828,
       "step": 1960
     },
     {
-      "epoch": 2.74,
-      "learning_rate": 4.747162022703819e-06,
-      "loss": 0.0323,
+      "epoch": 1.43,
+      "grad_norm": 8.679441452026367,
+      "learning_rate": 3.5766158315177926e-05,
+      "loss": 0.0776,
       "step": 1970
     },
     {
-      "epoch": 2.76,
-      "learning_rate": 4.489164086687307e-06,
-      "loss": 0.0345,
+      "epoch": 1.44,
+      "grad_norm": 53.65000534057617,
+      "learning_rate": 3.5947712418300656e-05,
+      "loss": 0.0931,
       "step": 1980
     },
     {
-      "epoch": 2.77,
-      "learning_rate": 4.231166150670795e-06,
-      "loss": 0.0214,
+      "epoch": 1.44,
+      "grad_norm": 15.00185775756836,
+      "learning_rate": 3.6129266521423386e-05,
+      "loss": 0.1215,
       "step": 1990
     },
     {
-      "epoch": 2.78,
-      "learning_rate": 3.9731682146542834e-06,
-      "loss": 0.0321,
+      "epoch": 1.45,
+      "grad_norm": 9.312851905822754,
+      "learning_rate": 3.6310820624546116e-05,
+      "loss": 0.103,
       "step": 2000
     },
     {
-      "epoch": 2.8,
-      "learning_rate": 3.715170278637771e-06,
-      "loss": 0.022,
+      "epoch": 1.46,
+      "grad_norm": 9.61681842803955,
+      "learning_rate": 3.6492374727668846e-05,
+      "loss": 0.0601,
       "step": 2010
     },
     {
-      "epoch": 2.81,
-      "learning_rate": 3.4571723426212592e-06,
-      "loss": 0.0234,
+      "epoch": 1.47,
+      "grad_norm": 7.561996936798096,
+      "learning_rate": 3.6673928830791576e-05,
+      "loss": 0.0572,
       "step": 2020
     },
     {
-      "epoch": 2.83,
-      "learning_rate": 3.199174406604747e-06,
-      "loss": 0.0288,
+      "epoch": 1.47,
+      "grad_norm": 8.18752384185791,
+      "learning_rate": 3.6855482933914306e-05,
+      "loss": 0.1414,
       "step": 2030
     },
     {
-      "epoch": 2.84,
-      "learning_rate": 2.9411764705882355e-06,
-      "loss": 0.0273,
+      "epoch": 1.48,
+      "grad_norm": 5.567200660705566,
+      "learning_rate": 3.7037037037037037e-05,
+      "loss": 0.0717,
       "step": 2040
     },
     {
-      "epoch": 2.85,
-      "learning_rate": 2.6831785345717234e-06,
-      "loss": 0.0105,
+      "epoch": 1.49,
+      "grad_norm": 2.777528762817383,
+      "learning_rate": 3.721859114015977e-05,
+      "loss": 0.0535,
       "step": 2050
     },
     {
-      "epoch": 2.87,
-      "learning_rate": 2.4251805985552117e-06,
-      "loss": 0.0236,
+      "epoch": 1.5,
+      "grad_norm": 16.293184280395508,
+      "learning_rate": 3.74001452432825e-05,
+      "loss": 0.0705,
       "step": 2060
     },
     {
-      "epoch": 2.88,
-      "learning_rate": 2.1671826625387e-06,
-      "loss": 0.0377,
+      "epoch": 1.5,
+      "grad_norm": 6.650923252105713,
+      "learning_rate": 3.758169934640523e-05,
+      "loss": 0.0741,
       "step": 2070
     },
     {
-      "epoch": 2.89,
-      "learning_rate": 1.909184726522188e-06,
-      "loss": 0.0153,
+      "epoch": 1.51,
+      "grad_norm": 11.978032112121582,
+      "learning_rate": 3.7763253449527964e-05,
+      "loss": 0.0506,
       "step": 2080
     },
     {
-      "epoch": 2.91,
-      "learning_rate": 1.6511867905056758e-06,
-      "loss": 0.0205,
+      "epoch": 1.52,
+      "grad_norm": 11.749996185302734,
+      "learning_rate": 3.7944807552650694e-05,
+      "loss": 0.0926,
       "step": 2090
     },
     {
-      "epoch": 2.92,
-      "learning_rate": 1.3931888544891641e-06,
-      "loss": 0.0179,
+      "epoch": 1.52,
+      "grad_norm": 8.854286193847656,
+      "learning_rate": 3.8126361655773424e-05,
+      "loss": 0.0767,
       "step": 2100
     },
     {
-      "epoch": 2.94,
-      "learning_rate": 1.1351909184726523e-06,
-      "loss": 0.0154,
+      "epoch": 1.53,
+      "grad_norm": 5.265554428100586,
+      "learning_rate": 3.8307915758896154e-05,
+      "loss": 0.0709,
       "step": 2110
     },
     {
-      "epoch": 2.95,
-      "learning_rate": 8.771929824561404e-07,
-      "loss": 0.0263,
+      "epoch": 1.54,
+      "grad_norm": 8.641592979431152,
+      "learning_rate": 3.8489469862018884e-05,
+      "loss": 0.0957,
       "step": 2120
     },
     {
-      "epoch": 2.96,
-      "learning_rate": 6.191950464396286e-07,
-      "loss": 0.0157,
+      "epoch": 1.55,
+      "grad_norm": 15.576934814453125,
+      "learning_rate": 3.8671023965141615e-05,
+      "loss": 0.0742,
       "step": 2130
     },
     {
-      "epoch": 2.98,
-      "learning_rate": 3.6119711042311665e-07,
-      "loss": 0.023,
+      "epoch": 1.55,
+      "grad_norm": 5.454996109008789,
+      "learning_rate": 3.8852578068264345e-05,
+      "loss": 0.0745,
       "step": 2140
     },
     {
-      "epoch": 2.99,
-      "learning_rate": 1.0319917440660474e-07,
-      "loss": 0.0164,
+      "epoch": 1.56,
+      "grad_norm": 7.110592842102051,
+      "learning_rate": 3.9034132171387075e-05,
+      "loss": 0.0785,
       "step": 2150
     },
     {
-      "epoch": 3.0,
-      "eval_accuracy": 0.9957906415192848,
-      "eval_loss": 0.012260152958333492,
-      "eval_runtime": 442.4591,
-      "eval_samples_per_second": 207.789,
-      "eval_steps_per_second": 6.496,
-      "step": 2154
+      "epoch": 1.57,
+      "grad_norm": 9.214689254760742,
+      "learning_rate": 3.9215686274509805e-05,
+      "loss": 0.0619,
+      "step": 2160
     },
     {
-      "epoch": 3.0,
-      "step": 2154,
-      "total_flos": 6.851603075929178e+18,
-      "train_loss": 0.08376573344403765,
-      "train_runtime": 4348.829,
-      "train_samples_per_second": 63.423,
-      "train_steps_per_second": 0.495
+      "epoch": 1.58,
+      "grad_norm": 3.3824822902679443,
+      "learning_rate": 3.9397240377632535e-05,
+      "loss": 0.1681,
+      "step": 2170
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 4.873193264007568,
+      "learning_rate": 3.9578794480755265e-05,
+      "loss": 0.0957,
+      "step": 2180
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 14.703625679016113,
+      "learning_rate": 3.9760348583877995e-05,
+      "loss": 0.0648,
+      "step": 2190
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 13.431695938110352,
+      "learning_rate": 3.9941902687000726e-05,
+      "loss": 0.0711,
+      "step": 2200
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 13.710531234741211,
+      "learning_rate": 4.012345679012346e-05,
+      "loss": 0.0867,
+      "step": 2210
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 9.267621994018555,
+      "learning_rate": 4.0305010893246186e-05,
+      "loss": 0.0975,
+      "step": 2220
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 7.971125602722168,
+      "learning_rate": 4.048656499636892e-05,
+      "loss": 0.0535,
+      "step": 2230
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 20.38298988342285,
+      "learning_rate": 4.066811909949165e-05,
+      "loss": 0.0907,
+      "step": 2240
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 13.68902587890625,
+      "learning_rate": 4.084967320261438e-05,
+      "loss": 0.0778,
+      "step": 2250
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 4.450131893157959,
+      "learning_rate": 4.103122730573711e-05,
+      "loss": 0.0623,
+      "step": 2260
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 3.162184238433838,
+      "learning_rate": 4.121278140885984e-05,
+      "loss": 0.0746,
+      "step": 2270
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 16.87812042236328,
+      "learning_rate": 4.1394335511982573e-05,
+      "loss": 0.083,
+      "step": 2280
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 6.631977081298828,
+      "learning_rate": 4.1575889615105304e-05,
+      "loss": 0.0634,
+      "step": 2290
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 13.702621459960938,
+      "learning_rate": 4.1757443718228034e-05,
+      "loss": 0.1152,
+      "step": 2300
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 9.709258079528809,
+      "learning_rate": 4.193899782135077e-05,
+      "loss": 0.0727,
+      "step": 2310
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 15.56152057647705,
+      "learning_rate": 4.2120551924473494e-05,
+      "loss": 0.0734,
+      "step": 2320
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 14.451997756958008,
+      "learning_rate": 4.230210602759623e-05,
+      "loss": 0.0906,
+      "step": 2330
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 3.7484872341156006,
+      "learning_rate": 4.2483660130718954e-05,
+      "loss": 0.0724,
+      "step": 2340
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 7.64939546585083,
+      "learning_rate": 4.266521423384169e-05,
+      "loss": 0.0753,
+      "step": 2350
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 5.493325710296631,
+      "learning_rate": 4.2846768336964415e-05,
+      "loss": 0.0781,
+      "step": 2360
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 23.490360260009766,
+      "learning_rate": 4.302832244008715e-05,
+      "loss": 0.0908,
+      "step": 2370
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 10.67781925201416,
+      "learning_rate": 4.3209876543209875e-05,
+      "loss": 0.0996,
+      "step": 2380
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 1.9585663080215454,
+      "learning_rate": 4.339143064633261e-05,
+      "loss": 0.0895,
+      "step": 2390
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 17.196063995361328,
+      "learning_rate": 4.357298474945534e-05,
+      "loss": 0.0667,
+      "step": 2400
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 18.92582130432129,
+      "learning_rate": 4.375453885257807e-05,
+      "loss": 0.0947,
+      "step": 2410
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 17.93027114868164,
+      "learning_rate": 4.39360929557008e-05,
+      "loss": 0.0893,
+      "step": 2420
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 4.2530670166015625,
+      "learning_rate": 4.411764705882353e-05,
+      "loss": 0.0736,
+      "step": 2430
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 10.697159767150879,
+      "learning_rate": 4.429920116194626e-05,
+      "loss": 0.0914,
+      "step": 2440
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 7.323067665100098,
+      "learning_rate": 4.448075526506899e-05,
+      "loss": 0.0936,
+      "step": 2450
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 6.918476581573486,
+      "learning_rate": 4.466230936819172e-05,
+      "loss": 0.1106,
+      "step": 2460
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 11.204655647277832,
+      "learning_rate": 4.484386347131445e-05,
+      "loss": 0.062,
+      "step": 2470
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 12.258004188537598,
+      "learning_rate": 4.502541757443718e-05,
+      "loss": 0.0533,
+      "step": 2480
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 5.304686069488525,
+      "learning_rate": 4.520697167755992e-05,
+      "loss": 0.0817,
+      "step": 2490
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 12.656023025512695,
+      "learning_rate": 4.538852578068264e-05,
+      "loss": 0.0825,
+      "step": 2500
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 2.970365047454834,
+      "learning_rate": 4.557007988380538e-05,
+      "loss": 0.0568,
+      "step": 2510
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 20.127803802490234,
+      "learning_rate": 4.5751633986928104e-05,
+      "loss": 0.0471,
+      "step": 2520
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 12.173340797424316,
+      "learning_rate": 4.593318809005084e-05,
+      "loss": 0.092,
+      "step": 2530
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 13.151833534240723,
+      "learning_rate": 4.6114742193173564e-05,
+      "loss": 0.0767,
+      "step": 2540
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 10.405213356018066,
+      "learning_rate": 4.62962962962963e-05,
+      "loss": 0.0629,
+      "step": 2550
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 3.035775899887085,
+      "learning_rate": 4.647785039941903e-05,
+      "loss": 0.0696,
+      "step": 2560
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 11.858610153198242,
+      "learning_rate": 4.665940450254176e-05,
+      "loss": 0.1054,
+      "step": 2570
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 16.282440185546875,
+      "learning_rate": 4.684095860566449e-05,
+      "loss": 0.0718,
+      "step": 2580
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 11.148844718933105,
+      "learning_rate": 4.702251270878722e-05,
+      "loss": 0.0664,
+      "step": 2590
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 6.528799057006836,
+      "learning_rate": 4.720406681190995e-05,
+      "loss": 0.0611,
+      "step": 2600
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 3.7072341442108154,
+      "learning_rate": 4.738562091503268e-05,
+      "loss": 0.043,
+      "step": 2610
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 16.07568359375,
+      "learning_rate": 4.756717501815541e-05,
+      "loss": 0.0739,
+      "step": 2620
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 13.207526206970215,
+      "learning_rate": 4.774872912127814e-05,
+      "loss": 0.0474,
+      "step": 2630
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 7.959559917449951,
+      "learning_rate": 4.793028322440087e-05,
+      "loss": 0.0883,
+      "step": 2640
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 9.581097602844238,
+      "learning_rate": 4.811183732752361e-05,
+      "loss": 0.0555,
+      "step": 2650
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 7.399395942687988,
+      "learning_rate": 4.829339143064633e-05,
+      "loss": 0.0877,
+      "step": 2660
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 1.6540353298187256,
+      "learning_rate": 4.847494553376907e-05,
+      "loss": 0.0745,
+      "step": 2670
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 8.871071815490723,
+      "learning_rate": 4.865649963689179e-05,
+      "loss": 0.0831,
+      "step": 2680
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 6.940814971923828,
+      "learning_rate": 4.883805374001453e-05,
+      "loss": 0.0736,
+      "step": 2690
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 5.259904861450195,
+      "learning_rate": 4.901960784313725e-05,
+      "loss": 0.1023,
+      "step": 2700
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 32.73249435424805,
+      "learning_rate": 4.920116194625999e-05,
+      "loss": 0.0767,
+      "step": 2710
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 13.280170440673828,
+      "learning_rate": 4.938271604938271e-05,
+      "loss": 0.1112,
+      "step": 2720
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 14.727370262145996,
+      "learning_rate": 4.956427015250545e-05,
+      "loss": 0.0511,
+      "step": 2730
+    },
+    {
+      "epoch": 1.99,
+      "grad_norm": 30.67513084411621,
+      "learning_rate": 4.974582425562818e-05,
+      "loss": 0.0814,
+      "step": 2740
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 9.064040184020996,
+      "learning_rate": 4.992737835875091e-05,
+      "loss": 0.0917,
+      "step": 2750
+    },
+    {
+      "epoch": 2.0,
+      "eval_accuracy": 0.9896310750102101,
+      "eval_f1": 0.9760394274629057,
+      "eval_loss": 0.03418319672346115,
+      "eval_precision": 0.9803054239073197,
+      "eval_recall": 0.9718103988306536,
+      "eval_roc_auc": 0.9992664428625956,
+      "eval_runtime": 387.256,
+      "eval_samples_per_second": 227.622,
+      "eval_steps_per_second": 14.228,
+      "step": 2755
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 4.060576915740967,
+      "learning_rate": 4.9987896393125154e-05,
+      "loss": 0.0802,
+      "step": 2760
+    },
+    {
+      "epoch": 2.01,
+      "grad_norm": 5.140890121459961,
+      "learning_rate": 4.996772371500041e-05,
+      "loss": 0.0407,
+      "step": 2770
+    },
+    {
+      "epoch": 2.02,
+      "grad_norm": 19.039766311645508,
+      "learning_rate": 4.9947551036875656e-05,
+      "loss": 0.0489,
+      "step": 2780
+    },
+    {
+      "epoch": 2.03,
+      "grad_norm": 12.916511535644531,
+      "learning_rate": 4.992737835875091e-05,
+      "loss": 0.0472,
+      "step": 2790
+    },
+    {
+      "epoch": 2.03,
+      "grad_norm": 1.2582281827926636,
+      "learning_rate": 4.9907205680626165e-05,
+      "loss": 0.0651,
+      "step": 2800
+    },
+    {
+      "epoch": 2.04,
+      "grad_norm": 4.840940475463867,
+      "learning_rate": 4.988703300250141e-05,
+      "loss": 0.0578,
+      "step": 2810
+    },
+    {
+      "epoch": 2.05,
+      "grad_norm": 5.674259185791016,
+      "learning_rate": 4.986686032437667e-05,
+      "loss": 0.0371,
+      "step": 2820
+    },
+    {
+      "epoch": 2.05,
+      "grad_norm": 5.847988128662109,
+      "learning_rate": 4.9846687646251915e-05,
+      "loss": 0.0404,
+      "step": 2830
+    },
+    {
+      "epoch": 2.06,
+      "grad_norm": 13.216257095336914,
+      "learning_rate": 4.982651496812717e-05,
+      "loss": 0.0904,
+      "step": 2840
+    },
+    {
+      "epoch": 2.07,
+      "grad_norm": 9.643535614013672,
+      "learning_rate": 4.9806342290002424e-05,
+      "loss": 0.0462,
+      "step": 2850
+    },
+    {
+      "epoch": 2.08,
+      "grad_norm": 10.265970230102539,
+      "learning_rate": 4.978616961187768e-05,
+      "loss": 0.0436,
+      "step": 2860
+    },
+    {
+      "epoch": 2.08,
+      "grad_norm": 16.27211570739746,
+      "learning_rate": 4.9765996933752926e-05,
+      "loss": 0.0819,
+      "step": 2870
+    },
+    {
+      "epoch": 2.09,
+      "grad_norm": 10.045124053955078,
+      "learning_rate": 4.974582425562818e-05,
+      "loss": 0.138,
+      "step": 2880
+    },
+    {
+      "epoch": 2.1,
+      "grad_norm": 6.49107551574707,
+      "learning_rate": 4.972565157750343e-05,
+      "loss": 0.0705,
+      "step": 2890
+    },
+    {
+      "epoch": 2.11,
+      "grad_norm": 7.387770175933838,
+      "learning_rate": 4.970547889937868e-05,
+      "loss": 0.0604,
+      "step": 2900
+    },
+    {
+      "epoch": 2.11,
+      "grad_norm": 7.549914360046387,
+      "learning_rate": 4.968530622125394e-05,
+      "loss": 0.0517,
+      "step": 2910
+    },
+    {
+      "epoch": 2.12,
+      "grad_norm": 16.87873077392578,
+      "learning_rate": 4.966513354312919e-05,
+      "loss": 0.0388,
+      "step": 2920
+    },
+    {
+      "epoch": 2.13,
+      "grad_norm": 16.325939178466797,
+      "learning_rate": 4.964496086500444e-05,
+      "loss": 0.1262,
+      "step": 2930
+    },
+    {
+      "epoch": 2.13,
+      "grad_norm": 5.032786846160889,
+      "learning_rate": 4.9624788186879693e-05,
+      "loss": 0.0517,
+      "step": 2940
+    },
+    {
+      "epoch": 2.14,
+      "grad_norm": 2.5657832622528076,
+      "learning_rate": 4.960461550875494e-05,
+      "loss": 0.0199,
+      "step": 2950
+    },
+    {
+      "epoch": 2.15,
+      "grad_norm": 37.460018157958984,
+      "learning_rate": 4.9584442830630196e-05,
+      "loss": 0.0367,
+      "step": 2960
+    },
+    {
+      "epoch": 2.16,
+      "grad_norm": 33.25373458862305,
+      "learning_rate": 4.956427015250545e-05,
+      "loss": 0.0636,
+      "step": 2970
+    },
+    {
+      "epoch": 2.16,
+      "grad_norm": 18.634830474853516,
+      "learning_rate": 4.9544097474380705e-05,
+      "loss": 0.0746,
+      "step": 2980
+    },
+    {
+      "epoch": 2.17,
+      "grad_norm": 13.56174087524414,
+      "learning_rate": 4.952392479625595e-05,
+      "loss": 0.1066,
+      "step": 2990
+    },
+    {
+      "epoch": 2.18,
+      "grad_norm": 9.151412963867188,
+      "learning_rate": 4.950375211813121e-05,
+      "loss": 0.0841,
+      "step": 3000
+    },
+    {
+      "epoch": 2.19,
+      "grad_norm": 0.9245597124099731,
+      "learning_rate": 4.9483579440006454e-05,
+      "loss": 0.0327,
+      "step": 3010
+    },
+    {
+      "epoch": 2.19,
+      "grad_norm": 25.459856033325195,
+      "learning_rate": 4.946340676188171e-05,
+      "loss": 0.0773,
+      "step": 3020
+    },
+    {
+      "epoch": 2.2,
+      "grad_norm": 9.643467903137207,
+      "learning_rate": 4.944323408375696e-05,
+      "loss": 0.0392,
+      "step": 3030
+    },
+    {
+      "epoch": 2.21,
+      "grad_norm": 4.589292049407959,
+      "learning_rate": 4.942306140563222e-05,
+      "loss": 0.0501,
+      "step": 3040
+    },
+    {
+      "epoch": 2.21,
+      "grad_norm": 8.616634368896484,
+      "learning_rate": 4.9402888727507465e-05,
+      "loss": 0.0514,
+      "step": 3050
+    },
+    {
+      "epoch": 2.22,
+      "grad_norm": 2.33439040184021,
+      "learning_rate": 4.938271604938271e-05,
+      "loss": 0.0411,
+      "step": 3060
+    },
+    {
+      "epoch": 2.23,
+      "grad_norm": 8.900259971618652,
+      "learning_rate": 4.936254337125797e-05,
+      "loss": 0.0449,
+      "step": 3070
+    },
+    {
+      "epoch": 2.24,
+      "grad_norm": 15.918954849243164,
+      "learning_rate": 4.934237069313322e-05,
+      "loss": 0.0539,
+      "step": 3080
+    },
+    {
+      "epoch": 2.24,
+      "grad_norm": 23.657480239868164,
+      "learning_rate": 4.9322198015008477e-05,
+      "loss": 0.0574,
+      "step": 3090
+    },
+    {
+      "epoch": 2.25,
+      "grad_norm": 13.110893249511719,
+      "learning_rate": 4.930202533688373e-05,
+      "loss": 0.0938,
+      "step": 3100
+    },
+    {
+      "epoch": 2.26,
+      "grad_norm": 1.8193144798278809,
+      "learning_rate": 4.928185265875898e-05,
+      "loss": 0.0607,
+      "step": 3110
+    },
+    {
+      "epoch": 2.26,
+      "grad_norm": 5.928867340087891,
+      "learning_rate": 4.9261679980634226e-05,
+      "loss": 0.0611,
+      "step": 3120
+    },
+    {
+      "epoch": 2.27,
+      "grad_norm": 10.63883113861084,
+      "learning_rate": 4.924150730250948e-05,
+      "loss": 0.0371,
+      "step": 3130
+    },
+    {
+      "epoch": 2.28,
+      "grad_norm": 7.217007637023926,
+      "learning_rate": 4.9221334624384735e-05,
+      "loss": 0.0453,
+      "step": 3140
+    },
+    {
+      "epoch": 2.29,
+      "grad_norm": 22.366975784301758,
+      "learning_rate": 4.920116194625999e-05,
+      "loss": 0.1012,
+      "step": 3150
+    },
+    {
+      "epoch": 2.29,
+      "grad_norm": 14.940516471862793,
+      "learning_rate": 4.9180989268135244e-05,
+      "loss": 0.0799,
+      "step": 3160
+    },
+    {
+      "epoch": 2.3,
+      "grad_norm": 3.766080617904663,
+      "learning_rate": 4.916081659001049e-05,
+      "loss": 0.0644,
+      "step": 3170
+    },
+    {
+      "epoch": 2.31,
+      "grad_norm": 17.51938819885254,
+      "learning_rate": 4.914064391188574e-05,
+      "loss": 0.0565,
+      "step": 3180
+    },
+    {
+      "epoch": 2.32,
+      "grad_norm": 3.683283567428589,
+      "learning_rate": 4.9120471233760994e-05,
+      "loss": 0.0537,
+      "step": 3190
+    },
+    {
+      "epoch": 2.32,
+      "grad_norm": 9.450156211853027,
+      "learning_rate": 4.910029855563625e-05,
+      "loss": 0.0537,
+      "step": 3200
+    },
+    {
+      "epoch": 2.33,
+      "grad_norm": 7.481932640075684,
+      "learning_rate": 4.90801258775115e-05,
+      "loss": 0.0266,
+      "step": 3210
+    },
+    {
+      "epoch": 2.34,
+      "grad_norm": 4.7421488761901855,
+      "learning_rate": 4.905995319938676e-05,
+      "loss": 0.0399,
+      "step": 3220
+    },
+    {
+      "epoch": 2.34,
+      "grad_norm": 42.04796600341797,
+      "learning_rate": 4.9039780521262005e-05,
+      "loss": 0.0667,
+      "step": 3230
+    },
+    {
+      "epoch": 2.35,
+      "grad_norm": 2.9122283458709717,
+      "learning_rate": 4.901960784313725e-05,
+      "loss": 0.0422,
+      "step": 3240
+    },
+    {
+      "epoch": 2.36,
+      "grad_norm": 8.037728309631348,
+      "learning_rate": 4.899943516501251e-05,
+      "loss": 0.0922,
+      "step": 3250
+    },
+    {
+      "epoch": 2.37,
+      "grad_norm": 1.3274216651916504,
+      "learning_rate": 4.897926248688776e-05,
+      "loss": 0.0563,
+      "step": 3260
+    },
+    {
+      "epoch": 2.37,
+      "grad_norm": 3.631781816482544,
+      "learning_rate": 4.8959089808763016e-05,
+      "loss": 0.054,
+      "step": 3270
+    },
+    {
+      "epoch": 2.38,
+      "grad_norm": 4.287731170654297,
+      "learning_rate": 4.893891713063827e-05,
+      "loss": 0.0523,
+      "step": 3280
+    },
+    {
+      "epoch": 2.39,
+      "grad_norm": 0.8176174163818359,
+      "learning_rate": 4.891874445251352e-05,
+      "loss": 0.052,
+      "step": 3290
+    },
+    {
+      "epoch": 2.4,
+      "grad_norm": 12.366915702819824,
+      "learning_rate": 4.8898571774388766e-05,
+      "loss": 0.0595,
+      "step": 3300
+    },
+    {
+      "epoch": 2.4,
+      "grad_norm": 15.8939790725708,
+      "learning_rate": 4.887839909626402e-05,
+      "loss": 0.0665,
+      "step": 3310
+    },
+    {
+      "epoch": 2.41,
+      "grad_norm": 17.355178833007812,
+      "learning_rate": 4.8858226418139275e-05,
+      "loss": 0.058,
+      "step": 3320
+    },
+    {
+      "epoch": 2.42,
+      "grad_norm": 2.637277603149414,
+      "learning_rate": 4.883805374001453e-05,
+      "loss": 0.033,
+      "step": 3330
+    },
+    {
+      "epoch": 2.42,
+      "grad_norm": 14.758194923400879,
+      "learning_rate": 4.8817881061889784e-05,
+      "loss": 0.0607,
+      "step": 3340
+    },
+    {
+      "epoch": 2.43,
+      "grad_norm": 13.589420318603516,
+      "learning_rate": 4.879770838376503e-05,
+      "loss": 0.0328,
+      "step": 3350
+    },
+    {
+      "epoch": 2.44,
+      "grad_norm": 15.758061408996582,
+      "learning_rate": 4.877753570564028e-05,
+      "loss": 0.0458,
+      "step": 3360
+    },
+    {
+      "epoch": 2.45,
+      "grad_norm": 16.480510711669922,
+      "learning_rate": 4.8757363027515534e-05,
+      "loss": 0.0584,
+      "step": 3370
+    },
+    {
+      "epoch": 2.45,
+      "grad_norm": 2.5956544876098633,
+      "learning_rate": 4.873719034939079e-05,
+      "loss": 0.0544,
+      "step": 3380
+    },
+    {
+      "epoch": 2.46,
+      "grad_norm": 13.941973686218262,
+      "learning_rate": 4.871701767126604e-05,
+      "loss": 0.0625,
+      "step": 3390
+    },
+    {
+      "epoch": 2.47,
+      "grad_norm": 9.241766929626465,
+      "learning_rate": 4.86968449931413e-05,
+      "loss": 0.0501,
+      "step": 3400
+    },
+    {
+      "epoch": 2.48,
+      "grad_norm": 1.4883322715759277,
+      "learning_rate": 4.8676672315016545e-05,
+      "loss": 0.0711,
+      "step": 3410
+    },
+    {
+      "epoch": 2.48,
+      "grad_norm": 2.9974913597106934,
+      "learning_rate": 4.865649963689179e-05,
+      "loss": 0.0285,
+      "step": 3420
+    },
+    {
+      "epoch": 2.49,
+      "grad_norm": 5.6022629737854,
+      "learning_rate": 4.863632695876705e-05,
+      "loss": 0.0886,
+      "step": 3430
+    },
+    {
+      "epoch": 2.5,
+      "grad_norm": 2.3902041912078857,
+      "learning_rate": 4.86161542806423e-05,
+      "loss": 0.051,
+      "step": 3440
+    },
+    {
+      "epoch": 2.5,
+      "grad_norm": 2.2185537815093994,
+      "learning_rate": 4.8595981602517556e-05,
+      "loss": 0.035,
+      "step": 3450
+    },
+    {
+      "epoch": 2.51,
+      "grad_norm": 24.16321563720703,
+      "learning_rate": 4.857580892439281e-05,
+      "loss": 0.0456,
+      "step": 3460
+    },
+    {
+      "epoch": 2.52,
+      "grad_norm": 10.79037857055664,
+      "learning_rate": 4.855563624626805e-05,
+      "loss": 0.053,
+      "step": 3470
+    },
+    {
+      "epoch": 2.53,
+      "grad_norm": 6.468301296234131,
+      "learning_rate": 4.8535463568143306e-05,
+      "loss": 0.0371,
+      "step": 3480
+    },
+    {
+      "epoch": 2.53,
+      "grad_norm": 4.342319488525391,
+      "learning_rate": 4.851529089001856e-05,
+      "loss": 0.0349,
+      "step": 3490
+    },
+    {
+      "epoch": 2.54,
+      "grad_norm": 1.0654453039169312,
+      "learning_rate": 4.8495118211893815e-05,
+      "loss": 0.0522,
+      "step": 3500
+    },
+    {
+      "epoch": 2.55,
+      "grad_norm": 0.6455519199371338,
+      "learning_rate": 4.847494553376907e-05,
+      "loss": 0.0501,
+      "step": 3510
+    },
+    {
+      "epoch": 2.56,
+      "grad_norm": 6.535991668701172,
+      "learning_rate": 4.8454772855644324e-05,
+      "loss": 0.0381,
+      "step": 3520
+    },
+    {
+      "epoch": 2.56,
+      "grad_norm": 15.82268238067627,
+      "learning_rate": 4.8434600177519565e-05,
+      "loss": 0.0552,
+      "step": 3530
+    },
+    {
+      "epoch": 2.57,
+      "grad_norm": 6.626350402832031,
+      "learning_rate": 4.841442749939482e-05,
+      "loss": 0.0605,
+      "step": 3540
+    },
+    {
+      "epoch": 2.58,
+      "grad_norm": 9.876846313476562,
+      "learning_rate": 4.8394254821270074e-05,
+      "loss": 0.064,
+      "step": 3550
+    },
+    {
+      "epoch": 2.58,
+      "grad_norm": 9.681201934814453,
+      "learning_rate": 4.837408214314533e-05,
+      "loss": 0.0469,
+      "step": 3560
+    },
+    {
+      "epoch": 2.59,
+      "grad_norm": 12.74880313873291,
+      "learning_rate": 4.835390946502058e-05,
+      "loss": 0.0482,
+      "step": 3570
+    },
+    {
+      "epoch": 2.6,
+      "grad_norm": 18.379024505615234,
+      "learning_rate": 4.833373678689583e-05,
+      "loss": 0.0353,
+      "step": 3580
+    },
+    {
+      "epoch": 2.61,
+      "grad_norm": 17.69583511352539,
+      "learning_rate": 4.831356410877108e-05,
+      "loss": 0.0488,
+      "step": 3590
+    },
+    {
+      "epoch": 2.61,
+      "grad_norm": 11.161755561828613,
+      "learning_rate": 4.829339143064633e-05,
+      "loss": 0.0497,
+      "step": 3600
+    },
+    {
+      "epoch": 2.62,
+      "grad_norm": 1.346876621246338,
+      "learning_rate": 4.827321875252159e-05,
+      "loss": 0.0477,
+      "step": 3610
+    },
+    {
+      "epoch": 2.63,
+      "grad_norm": 15.771431922912598,
+      "learning_rate": 4.825304607439684e-05,
+      "loss": 0.0409,
+      "step": 3620
+    },
+    {
+      "epoch": 2.64,
+      "grad_norm": 14.307646751403809,
+      "learning_rate": 4.8232873396272096e-05,
+      "loss": 0.041,
+      "step": 3630
+    },
+    {
+      "epoch": 2.64,
+      "grad_norm": 0.2704331874847412,
+      "learning_rate": 4.8212700718147343e-05,
+      "loss": 0.0552,
+      "step": 3640
+    },
+    {
+      "epoch": 2.65,
+      "grad_norm": 8.454325675964355,
+      "learning_rate": 4.819252804002259e-05,
+      "loss": 0.0527,
+      "step": 3650
+    },
+    {
+      "epoch": 2.66,
+      "grad_norm": 1.2138357162475586,
+      "learning_rate": 4.8172355361897846e-05,
+      "loss": 0.0429,
+      "step": 3660
+    },
+    {
+      "epoch": 2.66,
+      "grad_norm": 11.901808738708496,
+      "learning_rate": 4.81521826837731e-05,
+      "loss": 0.0517,
+      "step": 3670
+    },
+    {
+      "epoch": 2.67,
+      "grad_norm": 14.237427711486816,
+      "learning_rate": 4.8132010005648354e-05,
+      "loss": 0.0435,
+      "step": 3680
+    },
+    {
+      "epoch": 2.68,
+      "grad_norm": 0.18843147158622742,
+      "learning_rate": 4.811183732752361e-05,
+      "loss": 0.0335,
+      "step": 3690
+    },
+    {
+      "epoch": 2.69,
+      "grad_norm": 0.303036630153656,
+      "learning_rate": 4.809166464939886e-05,
+      "loss": 0.0432,
+      "step": 3700
+    },
+    {
+      "epoch": 2.69,
+      "grad_norm": 25.766660690307617,
+      "learning_rate": 4.8071491971274104e-05,
+      "loss": 0.0481,
+      "step": 3710
+    },
+    {
+      "epoch": 2.7,
+      "grad_norm": 16.13401222229004,
+      "learning_rate": 4.805131929314936e-05,
+      "loss": 0.0807,
+      "step": 3720
+    },
+    {
+      "epoch": 2.71,
+      "grad_norm": 16.700721740722656,
+      "learning_rate": 4.803114661502461e-05,
+      "loss": 0.0575,
+      "step": 3730
+    },
+    {
+      "epoch": 2.72,
+      "grad_norm": 1.0864169597625732,
+      "learning_rate": 4.801097393689987e-05,
+      "loss": 0.0522,
+      "step": 3740
+    },
+    {
+      "epoch": 2.72,
+      "grad_norm": 6.126308441162109,
+      "learning_rate": 4.799080125877512e-05,
+      "loss": 0.0577,
+      "step": 3750
+    },
+    {
+      "epoch": 2.73,
+      "grad_norm": 2.5702388286590576,
+      "learning_rate": 4.797062858065037e-05,
+      "loss": 0.0711,
+      "step": 3760
+    },
+    {
+      "epoch": 2.74,
+      "grad_norm": 4.559757709503174,
+      "learning_rate": 4.795045590252562e-05,
+      "loss": 0.042,
+      "step": 3770
+    },
+    {
+      "epoch": 2.74,
+      "grad_norm": 20.285289764404297,
+      "learning_rate": 4.793028322440087e-05,
+      "loss": 0.0647,
+      "step": 3780
+    },
+    {
+      "epoch": 2.75,
+      "grad_norm": 14.125584602355957,
+      "learning_rate": 4.7910110546276126e-05,
+      "loss": 0.0396,
+      "step": 3790
+    },
+    {
+      "epoch": 2.76,
+      "grad_norm": 6.698452949523926,
+      "learning_rate": 4.788993786815138e-05,
+      "loss": 0.0107,
+      "step": 3800
+    },
+    {
+      "epoch": 2.77,
+      "grad_norm": 12.645397186279297,
+      "learning_rate": 4.786976519002663e-05,
+      "loss": 0.0403,
+      "step": 3810
+    },
+    {
+      "epoch": 2.77,
+      "grad_norm": 8.991414070129395,
+      "learning_rate": 4.784959251190188e-05,
+      "loss": 0.049,
+      "step": 3820
+    },
+    {
+      "epoch": 2.78,
+      "grad_norm": 2.049801826477051,
+      "learning_rate": 4.782941983377713e-05,
+      "loss": 0.0919,
+      "step": 3830
+    },
+    {
+      "epoch": 2.79,
+      "grad_norm": 3.793942928314209,
+      "learning_rate": 4.7809247155652385e-05,
+      "loss": 0.0557,
+      "step": 3840
+    },
+    {
+      "epoch": 2.79,
+      "grad_norm": 19.353565216064453,
+      "learning_rate": 4.778907447752764e-05,
+      "loss": 0.0442,
+      "step": 3850
+    },
+    {
+      "epoch": 2.8,
+      "grad_norm": 8.632854461669922,
+      "learning_rate": 4.7768901799402894e-05,
+      "loss": 0.0458,
+      "step": 3860
+    },
+    {
+      "epoch": 2.81,
+      "grad_norm": 25.426410675048828,
+      "learning_rate": 4.774872912127814e-05,
+      "loss": 0.0329,
+      "step": 3870
+    },
+    {
+      "epoch": 2.82,
+      "grad_norm": 5.492660999298096,
+      "learning_rate": 4.7728556443153396e-05,
+      "loss": 0.0371,
+      "step": 3880
+    },
+    {
+      "epoch": 2.82,
+      "grad_norm": 17.053266525268555,
+      "learning_rate": 4.7708383765028644e-05,
+      "loss": 0.0354,
+      "step": 3890
+    },
+    {
+      "epoch": 2.83,
+      "grad_norm": 23.848909378051758,
+      "learning_rate": 4.76882110869039e-05,
+      "loss": 0.0639,
+      "step": 3900
+    },
+    {
+      "epoch": 2.84,
+      "grad_norm": 2.4269025325775146,
+      "learning_rate": 4.766803840877915e-05,
+      "loss": 0.0476,
+      "step": 3910
+    },
+    {
+      "epoch": 2.85,
+      "grad_norm": 8.225174903869629,
+      "learning_rate": 4.764786573065441e-05,
+      "loss": 0.0682,
+      "step": 3920
+    },
+    {
+      "epoch": 2.85,
+      "grad_norm": 5.187263488769531,
+      "learning_rate": 4.7627693052529655e-05,
+      "loss": 0.0318,
+      "step": 3930
+    },
+    {
+      "epoch": 2.86,
+      "grad_norm": 3.262751817703247,
+      "learning_rate": 4.760752037440491e-05,
+      "loss": 0.0415,
+      "step": 3940
+    },
+    {
+      "epoch": 2.87,
+      "grad_norm": 6.393750190734863,
+      "learning_rate": 4.758734769628016e-05,
+      "loss": 0.0454,
+      "step": 3950
+    },
+    {
+      "epoch": 2.87,
+      "grad_norm": 15.57343578338623,
+      "learning_rate": 4.756717501815541e-05,
+      "loss": 0.055,
+      "step": 3960
+    },
+    {
+      "epoch": 2.88,
+      "grad_norm": 6.782766819000244,
+      "learning_rate": 4.7547002340030666e-05,
+      "loss": 0.0253,
+      "step": 3970
+    },
+    {
+      "epoch": 2.89,
+      "grad_norm": 9.348075866699219,
+      "learning_rate": 4.752682966190592e-05,
+      "loss": 0.0761,
+      "step": 3980
+    },
+    {
+      "epoch": 2.9,
+      "grad_norm": 6.6344146728515625,
+      "learning_rate": 4.750665698378117e-05,
+      "loss": 0.0232,
+      "step": 3990
+    },
+    {
+      "epoch": 2.9,
+      "grad_norm": 9.608769416809082,
+      "learning_rate": 4.748648430565642e-05,
+      "loss": 0.0657,
+      "step": 4000
+    },
+    {
+      "epoch": 2.91,
+      "grad_norm": 8.99563980102539,
+      "learning_rate": 4.746631162753167e-05,
+      "loss": 0.0345,
+      "step": 4010
+    },
+    {
+      "epoch": 2.92,
+      "grad_norm": 33.83638381958008,
+      "learning_rate": 4.7446138949406925e-05,
+      "loss": 0.1342,
+      "step": 4020
+    },
+    {
+      "epoch": 2.93,
+      "grad_norm": 0.5134745836257935,
+      "learning_rate": 4.742596627128218e-05,
+      "loss": 0.0923,
+      "step": 4030
+    },
+    {
+      "epoch": 2.93,
+      "grad_norm": 5.171667575836182,
+      "learning_rate": 4.7405793593157434e-05,
+      "loss": 0.0384,
+      "step": 4040
+    },
+    {
+      "epoch": 2.94,
+      "grad_norm": 7.891140937805176,
+      "learning_rate": 4.738562091503268e-05,
+      "loss": 0.0369,
+      "step": 4050
+    },
+    {
+      "epoch": 2.95,
+      "grad_norm": 11.524323463439941,
+      "learning_rate": 4.7365448236907936e-05,
+      "loss": 0.041,
+      "step": 4060
+    },
+    {
+      "epoch": 2.95,
+      "grad_norm": 3.1125144958496094,
+      "learning_rate": 4.7345275558783184e-05,
+      "loss": 0.0292,
+      "step": 4070
+    },
+    {
+      "epoch": 2.96,
+      "grad_norm": 1.6844711303710938,
+      "learning_rate": 4.732510288065844e-05,
+      "loss": 0.0572,
+      "step": 4080
+    },
+    {
+      "epoch": 2.97,
+      "grad_norm": 10.969454765319824,
+      "learning_rate": 4.730493020253369e-05,
+      "loss": 0.0367,
+      "step": 4090
+    },
+    {
+      "epoch": 2.98,
+      "grad_norm": 1.6989827156066895,
+      "learning_rate": 4.728475752440894e-05,
+      "loss": 0.0284,
+      "step": 4100
+    },
+    {
+      "epoch": 2.98,
+      "grad_norm": 14.795607566833496,
+      "learning_rate": 4.7264584846284195e-05,
+      "loss": 0.0299,
+      "step": 4110
+    },
+    {
+      "epoch": 2.99,
+      "grad_norm": 0.03458476439118385,
+      "learning_rate": 4.724441216815945e-05,
+      "loss": 0.0507,
+      "step": 4120
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 3.454913377761841,
+      "learning_rate": 4.72242394900347e-05,
+      "loss": 0.0291,
+      "step": 4130
+    },
+    {
+      "epoch": 3.0,
+      "eval_accuracy": 0.9940100739665109,
+      "eval_f1": 0.9862807254586083,
+      "eval_loss": 0.016116181388497353,
+      "eval_precision": 0.9818416968442835,
+      "eval_recall": 0.9907600751722698,
+      "eval_roc_auc": 0.9997706287577176,
+      "eval_runtime": 389.701,
+      "eval_samples_per_second": 226.194,
+      "eval_steps_per_second": 14.139,
+      "step": 4132
+    },
+    {
+      "epoch": 3.01,
+      "grad_norm": 7.517953872680664,
+      "learning_rate": 4.720406681190995e-05,
+      "loss": 0.0522,
+      "step": 4140
+    },
+    {
+      "epoch": 3.01,
+      "grad_norm": 1.9929357767105103,
+      "learning_rate": 4.7183894133785206e-05,
+      "loss": 0.0338,
+      "step": 4150
+    },
+    {
+      "epoch": 3.02,
+      "grad_norm": 10.290563583374023,
+      "learning_rate": 4.7163721455660454e-05,
+      "loss": 0.055,
+      "step": 4160
+    },
+    {
+      "epoch": 3.03,
+      "grad_norm": 6.110312461853027,
+      "learning_rate": 4.714354877753571e-05,
+      "loss": 0.0607,
+      "step": 4170
+    },
+    {
+      "epoch": 3.03,
+      "grad_norm": 1.5012456178665161,
+      "learning_rate": 4.712337609941096e-05,
+      "loss": 0.0353,
+      "step": 4180
+    },
+    {
+      "epoch": 3.04,
+      "grad_norm": 0.7207418084144592,
+      "learning_rate": 4.710320342128621e-05,
+      "loss": 0.0183,
+      "step": 4190
+    },
+    {
+      "epoch": 3.05,
+      "grad_norm": 15.605572700500488,
+      "learning_rate": 4.7083030743161465e-05,
+      "loss": 0.0298,
+      "step": 4200
+    },
+    {
+      "epoch": 3.06,
+      "grad_norm": 0.7380584478378296,
+      "learning_rate": 4.706285806503672e-05,
+      "loss": 0.0274,
+      "step": 4210
+    },
+    {
+      "epoch": 3.06,
+      "grad_norm": 0.27790164947509766,
+      "learning_rate": 4.704268538691197e-05,
+      "loss": 0.0217,
+      "step": 4220
+    },
+    {
+      "epoch": 3.07,
+      "grad_norm": 15.396061897277832,
+      "learning_rate": 4.702251270878722e-05,
+      "loss": 0.0394,
+      "step": 4230
+    },
+    {
+      "epoch": 3.08,
+      "grad_norm": 7.82650899887085,
+      "learning_rate": 4.7002340030662476e-05,
+      "loss": 0.0251,
+      "step": 4240
+    },
+    {
+      "epoch": 3.09,
+      "grad_norm": 15.73935317993164,
+      "learning_rate": 4.6982167352537723e-05,
+      "loss": 0.0552,
+      "step": 4250
+    },
+    {
+      "epoch": 3.09,
+      "grad_norm": 5.134984016418457,
+      "learning_rate": 4.696199467441298e-05,
+      "loss": 0.0308,
+      "step": 4260
+    },
+    {
+      "epoch": 3.1,
+      "grad_norm": 5.871382713317871,
+      "learning_rate": 4.694182199628823e-05,
+      "loss": 0.0491,
+      "step": 4270
+    },
+    {
+      "epoch": 3.11,
+      "grad_norm": 8.34383773803711,
+      "learning_rate": 4.692164931816348e-05,
+      "loss": 0.0193,
+      "step": 4280
+    },
+    {
+      "epoch": 3.11,
+      "grad_norm": 21.052440643310547,
+      "learning_rate": 4.6901476640038735e-05,
+      "loss": 0.0225,
+      "step": 4290
+    },
+    {
+      "epoch": 3.12,
+      "grad_norm": 27.46809959411621,
+      "learning_rate": 4.688130396191399e-05,
+      "loss": 0.0557,
+      "step": 4300
+    },
+    {
+      "epoch": 3.13,
+      "grad_norm": 0.6610531806945801,
+      "learning_rate": 4.686113128378924e-05,
+      "loss": 0.0365,
+      "step": 4310
+    },
+    {
+      "epoch": 3.14,
+      "grad_norm": 0.41435083746910095,
+      "learning_rate": 4.684095860566449e-05,
+      "loss": 0.0202,
+      "step": 4320
+    },
+    {
+      "epoch": 3.14,
+      "grad_norm": 10.013864517211914,
+      "learning_rate": 4.682078592753974e-05,
+      "loss": 0.049,
+      "step": 4330
+    },
+    {
+      "epoch": 3.15,
+      "grad_norm": 22.101505279541016,
+      "learning_rate": 4.680061324941499e-05,
+      "loss": 0.0436,
+      "step": 4340
+    },
+    {
+      "epoch": 3.16,
+      "grad_norm": 6.700098037719727,
+      "learning_rate": 4.678044057129025e-05,
+      "loss": 0.1032,
+      "step": 4350
+    },
+    {
+      "epoch": 3.17,
+      "grad_norm": 8.122451782226562,
+      "learning_rate": 4.67602678931655e-05,
+      "loss": 0.049,
+      "step": 4360
+    },
+    {
+      "epoch": 3.17,
+      "grad_norm": 1.0085687637329102,
+      "learning_rate": 4.674009521504075e-05,
+      "loss": 0.0309,
+      "step": 4370
+    },
+    {
+      "epoch": 3.18,
+      "grad_norm": 0.43637529015541077,
+      "learning_rate": 4.6719922536916004e-05,
+      "loss": 0.0359,
+      "step": 4380
+    },
+    {
+      "epoch": 3.19,
+      "grad_norm": 6.261972904205322,
+      "learning_rate": 4.669974985879125e-05,
+      "loss": 0.0514,
+      "step": 4390
+    },
+    {
+      "epoch": 3.19,
+      "grad_norm": 0.32403895258903503,
+      "learning_rate": 4.6679577180666507e-05,
+      "loss": 0.0513,
+      "step": 4400
+    },
+    {
+      "epoch": 3.2,
+      "grad_norm": 3.918548583984375,
+      "learning_rate": 4.665940450254176e-05,
+      "loss": 0.0566,
+      "step": 4410
+    },
+    {
+      "epoch": 3.21,
+      "grad_norm": 17.437055587768555,
+      "learning_rate": 4.6639231824417016e-05,
+      "loss": 0.052,
+      "step": 4420
+    },
+    {
+      "epoch": 3.22,
+      "grad_norm": 0.20225393772125244,
+      "learning_rate": 4.661905914629226e-05,
+      "loss": 0.0375,
+      "step": 4430
+    },
+    {
+      "epoch": 3.22,
+      "grad_norm": 15.23760986328125,
+      "learning_rate": 4.659888646816752e-05,
+      "loss": 0.0346,
+      "step": 4440
+    },
+    {
+      "epoch": 3.23,
+      "grad_norm": 0.9687026143074036,
+      "learning_rate": 4.6578713790042765e-05,
+      "loss": 0.0222,
+      "step": 4450
+    },
+    {
+      "epoch": 3.24,
+      "grad_norm": 1.4425278902053833,
+      "learning_rate": 4.655854111191802e-05,
+      "loss": 0.0376,
+      "step": 4460
+    },
+    {
+      "epoch": 3.25,
+      "grad_norm": 7.644813537597656,
+      "learning_rate": 4.6538368433793274e-05,
+      "loss": 0.034,
+      "step": 4470
+    },
+    {
+      "epoch": 3.25,
+      "grad_norm": 9.407659530639648,
+      "learning_rate": 4.651819575566853e-05,
+      "loss": 0.0344,
+      "step": 4480
+    },
+    {
+      "epoch": 3.26,
+      "grad_norm": 9.224335670471191,
+      "learning_rate": 4.6498023077543776e-05,
+      "loss": 0.0337,
+      "step": 4490
+    },
+    {
+      "epoch": 3.27,
+      "grad_norm": 19.596595764160156,
+      "learning_rate": 4.647785039941903e-05,
+      "loss": 0.0391,
+      "step": 4500
+    },
+    {
+      "epoch": 3.27,
+      "grad_norm": 14.081114768981934,
+      "learning_rate": 4.645767772129428e-05,
+      "loss": 0.0267,
+      "step": 4510
+    },
+    {
+      "epoch": 3.28,
+      "grad_norm": 0.47743141651153564,
+      "learning_rate": 4.643750504316953e-05,
+      "loss": 0.0472,
+      "step": 4520
+    },
+    {
+      "epoch": 3.29,
+      "grad_norm": 0.03773049637675285,
+      "learning_rate": 4.641733236504479e-05,
+      "loss": 0.0164,
+      "step": 4530
+    },
+    {
+      "epoch": 3.3,
+      "grad_norm": 0.055120404809713364,
+      "learning_rate": 4.639715968692004e-05,
+      "loss": 0.0395,
+      "step": 4540
+    },
+    {
+      "epoch": 3.3,
+      "grad_norm": 14.224281311035156,
+      "learning_rate": 4.637698700879529e-05,
+      "loss": 0.0258,
+      "step": 4550
+    },
+    {
+      "epoch": 3.31,
+      "grad_norm": 19.072011947631836,
+      "learning_rate": 4.635681433067054e-05,
+      "loss": 0.0379,
+      "step": 4560
+    },
+    {
+      "epoch": 3.32,
+      "grad_norm": 0.13876385986804962,
+      "learning_rate": 4.633664165254579e-05,
+      "loss": 0.0334,
+      "step": 4570
+    },
+    {
+      "epoch": 3.32,
+      "grad_norm": 2.5720953941345215,
+      "learning_rate": 4.6316468974421046e-05,
+      "loss": 0.0297,
+      "step": 4580
+    },
+    {
+      "epoch": 3.33,
+      "grad_norm": 5.991764068603516,
+      "learning_rate": 4.62962962962963e-05,
+      "loss": 0.0555,
+      "step": 4590
+    },
+    {
+      "epoch": 3.34,
+      "grad_norm": 5.875485897064209,
+      "learning_rate": 4.627612361817155e-05,
+      "loss": 0.0323,
+      "step": 4600
+    },
+    {
+      "epoch": 3.35,
+      "grad_norm": 3.093625783920288,
+      "learning_rate": 4.62559509400468e-05,
+      "loss": 0.0449,
+      "step": 4610
+    },
+    {
+      "epoch": 3.35,
+      "grad_norm": 0.6352009773254395,
+      "learning_rate": 4.623577826192205e-05,
+      "loss": 0.0327,
+      "step": 4620
+    },
+    {
+      "epoch": 3.36,
+      "grad_norm": 0.2224309891462326,
+      "learning_rate": 4.6215605583797305e-05,
+      "loss": 0.0566,
+      "step": 4630
+    },
+    {
+      "epoch": 3.37,
+      "grad_norm": 3.2887914180755615,
+      "learning_rate": 4.619543290567256e-05,
+      "loss": 0.0494,
+      "step": 4640
+    },
+    {
+      "epoch": 3.38,
+      "grad_norm": 5.372390270233154,
+      "learning_rate": 4.6175260227547814e-05,
+      "loss": 0.034,
+      "step": 4650
+    },
+    {
+      "epoch": 3.38,
+      "grad_norm": 5.275757312774658,
+      "learning_rate": 4.615508754942306e-05,
+      "loss": 0.0348,
+      "step": 4660
+    },
+    {
+      "epoch": 3.39,
+      "grad_norm": 3.769078016281128,
+      "learning_rate": 4.6134914871298316e-05,
+      "loss": 0.025,
+      "step": 4670
+    },
+    {
+      "epoch": 3.4,
+      "grad_norm": 6.985911846160889,
+      "learning_rate": 4.6114742193173564e-05,
+      "loss": 0.0481,
+      "step": 4680
+    },
+    {
+      "epoch": 3.4,
+      "grad_norm": 3.3573648929595947,
+      "learning_rate": 4.609456951504882e-05,
+      "loss": 0.0421,
+      "step": 4690
+    },
+    {
+      "epoch": 3.41,
+      "grad_norm": 0.22904418408870697,
+      "learning_rate": 4.607439683692407e-05,
+      "loss": 0.0155,
+      "step": 4700
+    },
+    {
+      "epoch": 3.42,
+      "grad_norm": 1.3723245859146118,
+      "learning_rate": 4.605422415879933e-05,
+      "loss": 0.0151,
+      "step": 4710
+    },
+    {
+      "epoch": 3.43,
+      "grad_norm": 2.954483985900879,
+      "learning_rate": 4.6034051480674575e-05,
+      "loss": 0.0665,
+      "step": 4720
+    },
+    {
+      "epoch": 3.43,
+      "grad_norm": 24.531171798706055,
+      "learning_rate": 4.601387880254983e-05,
+      "loss": 0.06,
+      "step": 4730
+    },
+    {
+      "epoch": 3.44,
+      "grad_norm": 17.55864143371582,
+      "learning_rate": 4.599370612442508e-05,
+      "loss": 0.05,
+      "step": 4740
+    },
+    {
+      "epoch": 3.45,
+      "grad_norm": 0.5369663238525391,
+      "learning_rate": 4.597353344630033e-05,
+      "loss": 0.0139,
+      "step": 4750
+    },
+    {
+      "epoch": 3.46,
+      "grad_norm": 2.363798141479492,
+      "learning_rate": 4.5953360768175586e-05,
+      "loss": 0.013,
+      "step": 4760
+    },
+    {
+      "epoch": 3.46,
+      "grad_norm": 6.0249714851379395,
+      "learning_rate": 4.593318809005084e-05,
+      "loss": 0.066,
+      "step": 4770
+    },
+    {
+      "epoch": 3.47,
+      "grad_norm": 11.2283353805542,
+      "learning_rate": 4.591301541192609e-05,
+      "loss": 0.0614,
+      "step": 4780
+    },
+    {
+      "epoch": 3.48,
+      "grad_norm": 12.915953636169434,
+      "learning_rate": 4.5892842733801336e-05,
+      "loss": 0.0251,
+      "step": 4790
+    },
+    {
+      "epoch": 3.48,
+      "grad_norm": 2.8560121059417725,
+      "learning_rate": 4.587267005567659e-05,
+      "loss": 0.0239,
+      "step": 4800
+    },
+    {
+      "epoch": 3.49,
+      "grad_norm": 14.936617851257324,
+      "learning_rate": 4.5852497377551845e-05,
+      "loss": 0.0343,
+      "step": 4810
+    },
+    {
+      "epoch": 3.5,
+      "grad_norm": 0.29327040910720825,
+      "learning_rate": 4.58323246994271e-05,
+      "loss": 0.0474,
+      "step": 4820
+    },
+    {
+      "epoch": 3.51,
+      "grad_norm": 2.6680634021759033,
+      "learning_rate": 4.5812152021302354e-05,
+      "loss": 0.0111,
+      "step": 4830
+    },
+    {
+      "epoch": 3.51,
+      "grad_norm": 0.033980220556259155,
+      "learning_rate": 4.57919793431776e-05,
+      "loss": 0.0652,
+      "step": 4840
+    },
+    {
+      "epoch": 3.52,
+      "grad_norm": 0.2327333241701126,
+      "learning_rate": 4.577180666505285e-05,
+      "loss": 0.0188,
+      "step": 4850
+    },
+    {
+      "epoch": 3.53,
+      "grad_norm": 3.055844783782959,
+      "learning_rate": 4.5751633986928104e-05,
+      "loss": 0.0408,
+      "step": 4860
+    },
+    {
+      "epoch": 3.54,
+      "grad_norm": 4.960373878479004,
+      "learning_rate": 4.573146130880336e-05,
+      "loss": 0.0363,
+      "step": 4870
+    },
+    {
+      "epoch": 3.54,
+      "grad_norm": 1.730433702468872,
+      "learning_rate": 4.571128863067861e-05,
+      "loss": 0.0296,
+      "step": 4880
+    },
+    {
+      "epoch": 3.55,
+      "grad_norm": 12.168339729309082,
+      "learning_rate": 4.569111595255387e-05,
+      "loss": 0.0315,
+      "step": 4890
+    },
+    {
+      "epoch": 3.56,
+      "grad_norm": 11.328636169433594,
+      "learning_rate": 4.5670943274429115e-05,
+      "loss": 0.014,
+      "step": 4900
+    },
+    {
+      "epoch": 3.56,
+      "grad_norm": 0.4388793110847473,
+      "learning_rate": 4.565077059630436e-05,
+      "loss": 0.0374,
+      "step": 4910
+    },
+    {
+      "epoch": 3.57,
+      "grad_norm": 3.212505578994751,
+      "learning_rate": 4.563059791817962e-05,
+      "loss": 0.0709,
+      "step": 4920
+    },
+    {
+      "epoch": 3.58,
+      "grad_norm": 1.8563919067382812,
+      "learning_rate": 4.561042524005487e-05,
+      "loss": 0.0561,
+      "step": 4930
+    },
+    {
+      "epoch": 3.59,
+      "grad_norm": 5.985075950622559,
+      "learning_rate": 4.5590252561930126e-05,
+      "loss": 0.0382,
+      "step": 4940
+    },
+    {
+      "epoch": 3.59,
+      "grad_norm": 2.54897403717041,
+      "learning_rate": 4.557007988380538e-05,
+      "loss": 0.0399,
+      "step": 4950
+    },
+    {
+      "epoch": 3.6,
+      "grad_norm": 0.6124072074890137,
+      "learning_rate": 4.554990720568063e-05,
+      "loss": 0.0231,
+      "step": 4960
+    },
+    {
+      "epoch": 3.61,
+      "grad_norm": 7.537976264953613,
+      "learning_rate": 4.5529734527555876e-05,
+      "loss": 0.0276,
+      "step": 4970
+    },
+    {
+      "epoch": 3.62,
+      "grad_norm": 0.04458412900567055,
+      "learning_rate": 4.550956184943113e-05,
+      "loss": 0.0548,
+      "step": 4980
+    },
+    {
+      "epoch": 3.62,
+      "grad_norm": 3.02335524559021,
+      "learning_rate": 4.5489389171306385e-05,
+      "loss": 0.039,
+      "step": 4990
+    },
+    {
+      "epoch": 3.63,
+      "grad_norm": 10.046403884887695,
+      "learning_rate": 4.546921649318164e-05,
+      "loss": 0.0427,
+      "step": 5000
+    },
+    {
+      "epoch": 3.64,
+      "grad_norm": 3.8308396339416504,
+      "learning_rate": 4.5449043815056893e-05,
+      "loss": 0.0399,
+      "step": 5010
+    },
+    {
+      "epoch": 3.64,
+      "grad_norm": 0.474543958902359,
+      "learning_rate": 4.542887113693214e-05,
+      "loss": 0.0349,
+      "step": 5020
+    },
+    {
+      "epoch": 3.65,
+      "grad_norm": 0.38830775022506714,
+      "learning_rate": 4.540869845880739e-05,
+      "loss": 0.0271,
+      "step": 5030
+    },
+    {
+      "epoch": 3.66,
+      "grad_norm": 6.250176906585693,
+      "learning_rate": 4.538852578068264e-05,
+      "loss": 0.0786,
+      "step": 5040
+    },
+    {
+      "epoch": 3.67,
+      "grad_norm": 11.84267807006836,
+      "learning_rate": 4.53683531025579e-05,
+      "loss": 0.0187,
+      "step": 5050
+    },
+    {
+      "epoch": 3.67,
+      "grad_norm": 2.9608876705169678,
+      "learning_rate": 4.534818042443315e-05,
+      "loss": 0.0289,
+      "step": 5060
+    },
+    {
+      "epoch": 3.68,
+      "grad_norm": 20.82479476928711,
+      "learning_rate": 4.532800774630841e-05,
+      "loss": 0.0372,
+      "step": 5070
+    },
+    {
+      "epoch": 3.69,
+      "grad_norm": 12.964320182800293,
+      "learning_rate": 4.5307835068183654e-05,
+      "loss": 0.0439,
+      "step": 5080
+    },
+    {
+      "epoch": 3.7,
+      "grad_norm": 18.148500442504883,
+      "learning_rate": 4.52876623900589e-05,
+      "loss": 0.042,
+      "step": 5090
+    },
+    {
+      "epoch": 3.7,
+      "grad_norm": 0.2496163696050644,
+      "learning_rate": 4.5267489711934157e-05,
+      "loss": 0.0292,
+      "step": 5100
+    },
+    {
+      "epoch": 3.71,
+      "grad_norm": 1.652496576309204,
+      "learning_rate": 4.524731703380941e-05,
+      "loss": 0.0364,
+      "step": 5110
+    },
+    {
+      "epoch": 3.72,
+      "grad_norm": 6.644486427307129,
+      "learning_rate": 4.5227144355684665e-05,
+      "loss": 0.0148,
+      "step": 5120
+    },
+    {
+      "epoch": 3.72,
+      "grad_norm": 0.4298780858516693,
+      "learning_rate": 4.520697167755992e-05,
+      "loss": 0.0164,
+      "step": 5130
+    },
+    {
+      "epoch": 3.73,
+      "grad_norm": 9.257946968078613,
+      "learning_rate": 4.518679899943517e-05,
+      "loss": 0.0382,
+      "step": 5140
+    },
+    {
+      "epoch": 3.74,
+      "grad_norm": 13.765926361083984,
+      "learning_rate": 4.5166626321310415e-05,
+      "loss": 0.0378,
+      "step": 5150
+    },
+    {
+      "epoch": 3.75,
+      "grad_norm": 5.054808139801025,
+      "learning_rate": 4.514645364318567e-05,
+      "loss": 0.0532,
+      "step": 5160
+    },
+    {
+      "epoch": 3.75,
+      "grad_norm": 22.202678680419922,
+      "learning_rate": 4.5126280965060924e-05,
+      "loss": 0.0287,
+      "step": 5170
+    },
+    {
+      "epoch": 3.76,
+      "grad_norm": 3.7927277088165283,
+      "learning_rate": 4.510610828693618e-05,
+      "loss": 0.0358,
+      "step": 5180
+    },
+    {
+      "epoch": 3.77,
+      "grad_norm": 10.952911376953125,
+      "learning_rate": 4.508593560881143e-05,
+      "loss": 0.0544,
+      "step": 5190
+    },
+    {
+      "epoch": 3.77,
+      "grad_norm": 0.1568138152360916,
+      "learning_rate": 4.506576293068668e-05,
+      "loss": 0.0357,
+      "step": 5200
+    },
+    {
+      "epoch": 3.78,
+      "grad_norm": 3.809103012084961,
+      "learning_rate": 4.504559025256193e-05,
+      "loss": 0.0513,
+      "step": 5210
+    },
+    {
+      "epoch": 3.79,
+      "grad_norm": 0.44535213708877563,
+      "learning_rate": 4.502541757443718e-05,
+      "loss": 0.028,
+      "step": 5220
+    },
+    {
+      "epoch": 3.8,
+      "grad_norm": 3.141669511795044,
+      "learning_rate": 4.500524489631244e-05,
+      "loss": 0.0191,
+      "step": 5230
+    },
+    {
+      "epoch": 3.8,
+      "grad_norm": 12.636493682861328,
+      "learning_rate": 4.498507221818769e-05,
+      "loss": 0.0366,
+      "step": 5240
+    },
+    {
+      "epoch": 3.81,
+      "grad_norm": 0.021700365468859673,
+      "learning_rate": 4.4964899540062946e-05,
+      "loss": 0.0167,
+      "step": 5250
+    },
+    {
+      "epoch": 3.82,
+      "grad_norm": 25.216238021850586,
+      "learning_rate": 4.4944726861938194e-05,
+      "loss": 0.0331,
+      "step": 5260
+    },
+    {
+      "epoch": 3.83,
+      "grad_norm": 14.333684921264648,
+      "learning_rate": 4.492455418381344e-05,
+      "loss": 0.0093,
+      "step": 5270
+    },
+    {
+      "epoch": 3.83,
+      "grad_norm": 2.936802625656128,
+      "learning_rate": 4.4904381505688696e-05,
+      "loss": 0.0569,
+      "step": 5280
+    },
+    {
+      "epoch": 3.84,
+      "grad_norm": 0.08579454571008682,
+      "learning_rate": 4.488420882756395e-05,
+      "loss": 0.0667,
+      "step": 5290
+    },
+    {
+      "epoch": 3.85,
+      "grad_norm": 0.8753920793533325,
+      "learning_rate": 4.4864036149439205e-05,
+      "loss": 0.0313,
+      "step": 5300
+    },
+    {
+      "epoch": 3.85,
+      "grad_norm": 7.517170429229736,
+      "learning_rate": 4.484386347131445e-05,
+      "loss": 0.0258,
+      "step": 5310
+    },
+    {
+      "epoch": 3.86,
+      "grad_norm": 7.475151538848877,
+      "learning_rate": 4.482369079318971e-05,
+      "loss": 0.0257,
+      "step": 5320
+    },
+    {
+      "epoch": 3.87,
+      "grad_norm": 13.565512657165527,
+      "learning_rate": 4.4803518115064955e-05,
+      "loss": 0.0419,
+      "step": 5330
+    },
+    {
+      "epoch": 3.88,
+      "grad_norm": 1.0435971021652222,
+      "learning_rate": 4.478334543694021e-05,
+      "loss": 0.0623,
+      "step": 5340
+    },
+    {
+      "epoch": 3.88,
+      "grad_norm": 9.602534294128418,
+      "learning_rate": 4.4763172758815464e-05,
+      "loss": 0.0553,
+      "step": 5350
+    },
+    {
+      "epoch": 3.89,
+      "grad_norm": 0.1525646597146988,
+      "learning_rate": 4.474300008069072e-05,
+      "loss": 0.0309,
+      "step": 5360
+    },
+    {
+      "epoch": 3.9,
+      "grad_norm": 12.293474197387695,
+      "learning_rate": 4.4722827402565966e-05,
+      "loss": 0.0325,
+      "step": 5370
+    },
+    {
+      "epoch": 3.91,
+      "grad_norm": 15.566313743591309,
+      "learning_rate": 4.470265472444122e-05,
+      "loss": 0.0398,
+      "step": 5380
+    },
+    {
+      "epoch": 3.91,
+      "grad_norm": 22.519784927368164,
+      "learning_rate": 4.468248204631647e-05,
+      "loss": 0.0469,
+      "step": 5390
+    },
+    {
+      "epoch": 3.92,
+      "grad_norm": 5.122323036193848,
+      "learning_rate": 4.466230936819172e-05,
+      "loss": 0.0284,
+      "step": 5400
+    },
+    {
+      "epoch": 3.93,
+      "grad_norm": 2.0292279720306396,
+      "learning_rate": 4.464213669006698e-05,
+      "loss": 0.0123,
+      "step": 5410
+    },
+    {
+      "epoch": 3.93,
+      "grad_norm": 0.47845131158828735,
+      "learning_rate": 4.462196401194223e-05,
+      "loss": 0.0277,
+      "step": 5420
+    },
+    {
+      "epoch": 3.94,
+      "grad_norm": 0.20897838473320007,
+      "learning_rate": 4.460179133381748e-05,
+      "loss": 0.0237,
+      "step": 5430
+    },
+    {
+      "epoch": 3.95,
+      "grad_norm": 5.986893177032471,
+      "learning_rate": 4.4581618655692734e-05,
+      "loss": 0.0292,
+      "step": 5440
+    },
+    {
+      "epoch": 3.96,
+      "grad_norm": 7.410892009735107,
+      "learning_rate": 4.456144597756798e-05,
+      "loss": 0.026,
+      "step": 5450
+    },
+    {
+      "epoch": 3.96,
+      "grad_norm": 23.17868423461914,
+      "learning_rate": 4.4541273299443236e-05,
+      "loss": 0.0494,
+      "step": 5460
+    },
+    {
+      "epoch": 3.97,
+      "grad_norm": 16.8724308013916,
+      "learning_rate": 4.452110062131849e-05,
+      "loss": 0.0276,
+      "step": 5470
+    },
+    {
+      "epoch": 3.98,
+      "grad_norm": 6.479537010192871,
+      "learning_rate": 4.4500927943193745e-05,
+      "loss": 0.0244,
+      "step": 5480
+    },
+    {
+      "epoch": 3.99,
+      "grad_norm": 2.1582884788513184,
+      "learning_rate": 4.448075526506899e-05,
+      "loss": 0.0245,
+      "step": 5490
+    },
+    {
+      "epoch": 3.99,
+      "grad_norm": 1.5921260118484497,
+      "learning_rate": 4.446058258694425e-05,
+      "loss": 0.0111,
+      "step": 5500
+    },
+    {
+      "epoch": 4.0,
+      "grad_norm": 7.180184841156006,
+      "learning_rate": 4.4440409908819495e-05,
+      "loss": 0.0454,
+      "step": 5510
+    },
+    {
+      "epoch": 4.0,
+      "eval_accuracy": 0.994963016744566,
+      "eval_f1": 0.9883720930232558,
+      "eval_loss": 0.013561672531068325,
+      "eval_precision": 0.991696447340761,
+      "eval_recall": 0.9850699519732721,
+      "eval_roc_auc": 0.9998451949640312,
+      "eval_runtime": 388.167,
+      "eval_samples_per_second": 227.088,
+      "eval_steps_per_second": 14.195,
+      "step": 5510
+    },
+    {
+      "epoch": 4.01,
+      "grad_norm": 10.229007720947266,
+      "learning_rate": 4.442023723069475e-05,
+      "loss": 0.0535,
+      "step": 5520
+    },
+    {
+      "epoch": 4.01,
+      "grad_norm": 0.41513580083847046,
+      "learning_rate": 4.4400064552570004e-05,
+      "loss": 0.0197,
+      "step": 5530
+    },
+    {
+      "epoch": 4.02,
+      "grad_norm": 1.0428308248519897,
+      "learning_rate": 4.437989187444525e-05,
+      "loss": 0.0396,
+      "step": 5540
+    },
+    {
+      "epoch": 4.03,
+      "grad_norm": 21.104991912841797,
+      "learning_rate": 4.4359719196320506e-05,
+      "loss": 0.0461,
+      "step": 5550
+    },
+    {
+      "epoch": 4.04,
+      "grad_norm": 0.17535223066806793,
+      "learning_rate": 4.433954651819576e-05,
+      "loss": 0.0242,
+      "step": 5560
+    },
+    {
+      "epoch": 4.04,
+      "grad_norm": 1.7396169900894165,
+      "learning_rate": 4.431937384007101e-05,
+      "loss": 0.0127,
+      "step": 5570
+    },
+    {
+      "epoch": 4.05,
+      "grad_norm": 0.34911617636680603,
+      "learning_rate": 4.429920116194626e-05,
+      "loss": 0.0213,
+      "step": 5580
+    },
+    {
+      "epoch": 4.06,
+      "grad_norm": 26.257558822631836,
+      "learning_rate": 4.427902848382152e-05,
+      "loss": 0.045,
+      "step": 5590
+    },
+    {
+      "epoch": 4.07,
+      "grad_norm": 0.12217586487531662,
+      "learning_rate": 4.4258855805696765e-05,
+      "loss": 0.0103,
+      "step": 5600
+    },
+    {
+      "epoch": 4.07,
+      "grad_norm": 0.008990990929305553,
+      "learning_rate": 4.423868312757202e-05,
+      "loss": 0.0157,
+      "step": 5610
+    },
+    {
+      "epoch": 4.08,
+      "grad_norm": 0.0027994182892143726,
+      "learning_rate": 4.4218510449447274e-05,
+      "loss": 0.0884,
+      "step": 5620
+    },
+    {
+      "epoch": 4.09,
+      "grad_norm": 0.013051895424723625,
+      "learning_rate": 4.419833777132252e-05,
+      "loss": 0.0135,
+      "step": 5630
+    },
+    {
+      "epoch": 4.09,
+      "grad_norm": 9.522844314575195,
+      "learning_rate": 4.4178165093197776e-05,
+      "loss": 0.0197,
+      "step": 5640
+    },
+    {
+      "epoch": 4.1,
+      "grad_norm": 3.2195699214935303,
+      "learning_rate": 4.415799241507303e-05,
+      "loss": 0.02,
+      "step": 5650
+    },
+    {
+      "epoch": 4.11,
+      "grad_norm": 13.479093551635742,
+      "learning_rate": 4.413781973694828e-05,
+      "loss": 0.041,
+      "step": 5660
+    },
+    {
+      "epoch": 4.12,
+      "grad_norm": 16.785228729248047,
+      "learning_rate": 4.411764705882353e-05,
+      "loss": 0.0327,
+      "step": 5670
+    },
+    {
+      "epoch": 4.12,
+      "grad_norm": 17.039676666259766,
+      "learning_rate": 4.409747438069878e-05,
+      "loss": 0.0227,
+      "step": 5680
+    },
+    {
+      "epoch": 4.13,
+      "grad_norm": 14.65450668334961,
+      "learning_rate": 4.4077301702574034e-05,
+      "loss": 0.0299,
+      "step": 5690
+    },
+    {
+      "epoch": 4.14,
+      "grad_norm": 12.701321601867676,
+      "learning_rate": 4.405712902444929e-05,
+      "loss": 0.0146,
+      "step": 5700
+    },
+    {
+      "epoch": 4.15,
+      "grad_norm": 5.022671699523926,
+      "learning_rate": 4.4036956346324543e-05,
+      "loss": 0.01,
+      "step": 5710
+    },
+    {
+      "epoch": 4.15,
+      "grad_norm": 0.2834198474884033,
+      "learning_rate": 4.401678366819979e-05,
+      "loss": 0.0465,
+      "step": 5720
+    },
+    {
+      "epoch": 4.16,
+      "grad_norm": 3.1493167877197266,
+      "learning_rate": 4.3996610990075046e-05,
+      "loss": 0.0307,
+      "step": 5730
+    },
+    {
+      "epoch": 4.17,
+      "grad_norm": 0.09327320009469986,
+      "learning_rate": 4.397643831195029e-05,
+      "loss": 0.0339,
+      "step": 5740
+    },
+    {
+      "epoch": 4.17,
+      "grad_norm": 12.125381469726562,
+      "learning_rate": 4.395626563382555e-05,
+      "loss": 0.0245,
+      "step": 5750
+    },
+    {
+      "epoch": 4.18,
+      "grad_norm": 12.335347175598145,
+      "learning_rate": 4.39360929557008e-05,
+      "loss": 0.0333,
+      "step": 5760
+    },
+    {
+      "epoch": 4.19,
+      "grad_norm": 0.17939302325248718,
+      "learning_rate": 4.391592027757606e-05,
+      "loss": 0.0147,
+      "step": 5770
+    },
+    {
+      "epoch": 4.2,
+      "grad_norm": 2.4980413913726807,
+      "learning_rate": 4.3895747599451304e-05,
+      "loss": 0.0392,
+      "step": 5780
+    },
+    {
+      "epoch": 4.2,
+      "grad_norm": 3.9628798961639404,
+      "learning_rate": 4.387557492132656e-05,
+      "loss": 0.0202,
+      "step": 5790
+    },
+    {
+      "epoch": 4.21,
+      "grad_norm": 14.232178688049316,
+      "learning_rate": 4.3855402243201806e-05,
+      "loss": 0.0274,
+      "step": 5800
+    },
+    {
+      "epoch": 4.22,
+      "grad_norm": 3.835378646850586,
+      "learning_rate": 4.383522956507706e-05,
+      "loss": 0.0185,
+      "step": 5810
+    },
+    {
+      "epoch": 4.23,
+      "grad_norm": 3.1973910331726074,
+      "learning_rate": 4.3815056886952315e-05,
+      "loss": 0.0209,
+      "step": 5820
+    },
+    {
+      "epoch": 4.23,
+      "grad_norm": 10.52609634399414,
+      "learning_rate": 4.379488420882756e-05,
+      "loss": 0.0053,
+      "step": 5830
+    },
+    {
+      "epoch": 4.24,
+      "grad_norm": 0.010969222523272038,
+      "learning_rate": 4.377471153070282e-05,
+      "loss": 0.0307,
+      "step": 5840
+    },
+    {
+      "epoch": 4.25,
+      "grad_norm": 1.1569510698318481,
+      "learning_rate": 4.375453885257807e-05,
+      "loss": 0.0202,
+      "step": 5850
+    },
+    {
+      "epoch": 4.25,
+      "grad_norm": 11.772544860839844,
+      "learning_rate": 4.373436617445332e-05,
+      "loss": 0.0179,
+      "step": 5860
+    },
+    {
+      "epoch": 4.26,
+      "grad_norm": 29.545024871826172,
+      "learning_rate": 4.3714193496328574e-05,
+      "loss": 0.0177,
+      "step": 5870
+    },
+    {
+      "epoch": 4.27,
+      "grad_norm": 0.005079995840787888,
+      "learning_rate": 4.369402081820383e-05,
+      "loss": 0.0288,
+      "step": 5880
+    },
+    {
+      "epoch": 4.28,
+      "grad_norm": 0.520383894443512,
+      "learning_rate": 4.3673848140079076e-05,
+      "loss": 0.0416,
+      "step": 5890
+    },
+    {
+      "epoch": 4.28,
+      "grad_norm": 4.170873165130615,
+      "learning_rate": 4.365367546195433e-05,
+      "loss": 0.0407,
+      "step": 5900
+    },
+    {
+      "epoch": 4.29,
+      "grad_norm": 0.9381336569786072,
+      "learning_rate": 4.3633502783829585e-05,
+      "loss": 0.0184,
+      "step": 5910
+    },
+    {
+      "epoch": 4.3,
+      "grad_norm": 7.8422064781188965,
+      "learning_rate": 4.361333010570483e-05,
+      "loss": 0.0224,
+      "step": 5920
+    },
+    {
+      "epoch": 4.3,
+      "grad_norm": 0.09656038135290146,
+      "learning_rate": 4.359315742758009e-05,
+      "loss": 0.0224,
+      "step": 5930
+    },
+    {
+      "epoch": 4.31,
+      "grad_norm": 9.255690574645996,
+      "learning_rate": 4.357298474945534e-05,
+      "loss": 0.0532,
+      "step": 5940
+    },
+    {
+      "epoch": 4.32,
+      "grad_norm": 9.539785385131836,
+      "learning_rate": 4.355281207133059e-05,
+      "loss": 0.0523,
+      "step": 5950
+    },
+    {
+      "epoch": 4.33,
+      "grad_norm": 0.2933836877346039,
+      "learning_rate": 4.3532639393205844e-05,
+      "loss": 0.0402,
+      "step": 5960
+    },
+    {
+      "epoch": 4.33,
+      "grad_norm": 0.00884042214602232,
+      "learning_rate": 4.35124667150811e-05,
+      "loss": 0.0275,
+      "step": 5970
+    },
+    {
+      "epoch": 4.34,
+      "grad_norm": 7.750953197479248,
+      "learning_rate": 4.3492294036956346e-05,
+      "loss": 0.0352,
+      "step": 5980
+    },
+    {
+      "epoch": 4.35,
+      "grad_norm": 0.5286880135536194,
+      "learning_rate": 4.34721213588316e-05,
+      "loss": 0.0123,
+      "step": 5990
+    },
+    {
+      "epoch": 4.36,
+      "grad_norm": 2.974754571914673,
+      "learning_rate": 4.3451948680706855e-05,
+      "loss": 0.0184,
+      "step": 6000
+    },
+    {
+      "epoch": 4.36,
+      "grad_norm": 20.916255950927734,
+      "learning_rate": 4.34317760025821e-05,
+      "loss": 0.0159,
+      "step": 6010
+    },
+    {
+      "epoch": 4.37,
+      "grad_norm": 10.342968940734863,
+      "learning_rate": 4.341160332445736e-05,
+      "loss": 0.0129,
+      "step": 6020
+    },
+    {
+      "epoch": 4.38,
+      "grad_norm": 4.216011047363281,
+      "learning_rate": 4.339143064633261e-05,
+      "loss": 0.0285,
+      "step": 6030
+    },
+    {
+      "epoch": 4.38,
+      "grad_norm": 16.87735366821289,
+      "learning_rate": 4.337125796820786e-05,
+      "loss": 0.0134,
+      "step": 6040
+    },
+    {
+      "epoch": 4.39,
+      "grad_norm": 0.019013158977031708,
+      "learning_rate": 4.3351085290083114e-05,
+      "loss": 0.012,
+      "step": 6050
+    },
+    {
+      "epoch": 4.4,
+      "grad_norm": 12.512757301330566,
+      "learning_rate": 4.333091261195836e-05,
+      "loss": 0.0481,
+      "step": 6060
+    },
+    {
+      "epoch": 4.41,
+      "grad_norm": 8.553503036499023,
+      "learning_rate": 4.3310739933833616e-05,
+      "loss": 0.0139,
+      "step": 6070
+    },
+    {
+      "epoch": 4.41,
+      "grad_norm": 14.92082405090332,
+      "learning_rate": 4.329056725570887e-05,
+      "loss": 0.0212,
+      "step": 6080
+    },
+    {
+      "epoch": 4.42,
+      "grad_norm": 0.0787595734000206,
+      "learning_rate": 4.3270394577584125e-05,
+      "loss": 0.0398,
+      "step": 6090
+    },
+    {
+      "epoch": 4.43,
+      "grad_norm": 0.6032419204711914,
+      "learning_rate": 4.325022189945937e-05,
+      "loss": 0.011,
+      "step": 6100
+    },
+    {
+      "epoch": 4.44,
+      "grad_norm": 0.5409778356552124,
+      "learning_rate": 4.323004922133463e-05,
+      "loss": 0.0045,
+      "step": 6110
+    },
+    {
+      "epoch": 4.44,
+      "grad_norm": 3.50061297416687,
+      "learning_rate": 4.3209876543209875e-05,
+      "loss": 0.0704,
+      "step": 6120
+    },
+    {
+      "epoch": 4.45,
+      "grad_norm": 0.033528584986925125,
+      "learning_rate": 4.318970386508513e-05,
+      "loss": 0.0346,
+      "step": 6130
+    },
+    {
+      "epoch": 4.46,
+      "grad_norm": 13.443081855773926,
+      "learning_rate": 4.3169531186960384e-05,
+      "loss": 0.0312,
+      "step": 6140
+    },
+    {
+      "epoch": 4.46,
+      "grad_norm": 4.95137357711792,
+      "learning_rate": 4.314935850883564e-05,
+      "loss": 0.0301,
+      "step": 6150
+    },
+    {
+      "epoch": 4.47,
+      "grad_norm": 3.4258296489715576,
+      "learning_rate": 4.3129185830710886e-05,
+      "loss": 0.0081,
+      "step": 6160
+    },
+    {
+      "epoch": 4.48,
+      "grad_norm": 7.301894664764404,
+      "learning_rate": 4.310901315258614e-05,
+      "loss": 0.0255,
+      "step": 6170
+    },
+    {
+      "epoch": 4.49,
+      "grad_norm": 21.892892837524414,
+      "learning_rate": 4.308884047446139e-05,
+      "loss": 0.0221,
+      "step": 6180
+    },
+    {
+      "epoch": 4.49,
+      "grad_norm": 0.33506104350090027,
+      "learning_rate": 4.306866779633664e-05,
+      "loss": 0.0246,
+      "step": 6190
+    },
+    {
+      "epoch": 4.5,
+      "grad_norm": 0.23564085364341736,
+      "learning_rate": 4.30484951182119e-05,
+      "loss": 0.0479,
+      "step": 6200
+    },
+    {
+      "epoch": 4.51,
+      "grad_norm": 2.225267171859741,
+      "learning_rate": 4.302832244008715e-05,
+      "loss": 0.032,
+      "step": 6210
+    },
+    {
+      "epoch": 4.52,
+      "grad_norm": 6.463624477386475,
+      "learning_rate": 4.30081497619624e-05,
+      "loss": 0.0219,
+      "step": 6220
+    },
+    {
+      "epoch": 4.52,
+      "grad_norm": 0.11097526550292969,
+      "learning_rate": 4.2987977083837654e-05,
+      "loss": 0.0368,
+      "step": 6230
+    },
+    {
+      "epoch": 4.53,
+      "grad_norm": 0.5188978910446167,
+      "learning_rate": 4.29678044057129e-05,
+      "loss": 0.0233,
+      "step": 6240
+    },
+    {
+      "epoch": 4.54,
+      "grad_norm": 0.026288433000445366,
+      "learning_rate": 4.2947631727588156e-05,
+      "loss": 0.0075,
+      "step": 6250
+    },
+    {
+      "epoch": 4.54,
+      "grad_norm": 0.03321617841720581,
+      "learning_rate": 4.292745904946341e-05,
+      "loss": 0.0423,
+      "step": 6260
+    },
+    {
+      "epoch": 4.55,
+      "grad_norm": 6.4352874755859375,
+      "learning_rate": 4.2907286371338665e-05,
+      "loss": 0.0567,
+      "step": 6270
+    },
+    {
+      "epoch": 4.56,
+      "grad_norm": 0.2122620791196823,
+      "learning_rate": 4.288711369321391e-05,
+      "loss": 0.048,
+      "step": 6280
+    },
+    {
+      "epoch": 4.57,
+      "grad_norm": 1.9611496925354004,
+      "learning_rate": 4.286694101508916e-05,
+      "loss": 0.0357,
+      "step": 6290
+    },
+    {
+      "epoch": 4.57,
+      "grad_norm": 14.888888359069824,
+      "learning_rate": 4.2846768336964415e-05,
+      "loss": 0.0133,
+      "step": 6300
+    },
+    {
+      "epoch": 4.58,
+      "grad_norm": 3.209965705871582,
+      "learning_rate": 4.282659565883967e-05,
+      "loss": 0.0131,
+      "step": 6310
+    },
+    {
+      "epoch": 4.59,
+      "grad_norm": 0.4727603495121002,
+      "learning_rate": 4.2806422980714924e-05,
+      "loss": 0.0388,
+      "step": 6320
+    },
+    {
+      "epoch": 4.6,
+      "grad_norm": 20.515947341918945,
+      "learning_rate": 4.278625030259018e-05,
+      "loss": 0.0232,
+      "step": 6330
+    },
+    {
+      "epoch": 4.6,
+      "grad_norm": 13.02455997467041,
+      "learning_rate": 4.2766077624465426e-05,
+      "loss": 0.0145,
+      "step": 6340
+    },
+    {
+      "epoch": 4.61,
+      "grad_norm": 0.28295955061912537,
+      "learning_rate": 4.274590494634067e-05,
+      "loss": 0.0451,
+      "step": 6350
+    },
+    {
+      "epoch": 4.62,
+      "grad_norm": 1.2605043649673462,
+      "learning_rate": 4.272573226821593e-05,
+      "loss": 0.0298,
+      "step": 6360
+    },
+    {
+      "epoch": 4.62,
+      "grad_norm": 0.013523263856768608,
+      "learning_rate": 4.270555959009118e-05,
+      "loss": 0.0186,
+      "step": 6370
+    },
+    {
+      "epoch": 4.63,
+      "grad_norm": 0.15178602933883667,
+      "learning_rate": 4.268538691196644e-05,
+      "loss": 0.0068,
+      "step": 6380
+    },
+    {
+      "epoch": 4.64,
+      "grad_norm": 0.569879949092865,
+      "learning_rate": 4.266521423384169e-05,
+      "loss": 0.0259,
+      "step": 6390
+    },
+    {
+      "epoch": 4.65,
+      "grad_norm": 0.3333960771560669,
+      "learning_rate": 4.264504155571694e-05,
+      "loss": 0.0791,
+      "step": 6400
+    },
+    {
+      "epoch": 4.65,
+      "grad_norm": 0.40962207317352295,
+      "learning_rate": 4.2624868877592187e-05,
+      "loss": 0.0185,
+      "step": 6410
+    },
+    {
+      "epoch": 4.66,
+      "grad_norm": 0.841285228729248,
+      "learning_rate": 4.260469619946744e-05,
+      "loss": 0.0585,
+      "step": 6420
+    },
+    {
+      "epoch": 4.67,
+      "grad_norm": 0.23981308937072754,
+      "learning_rate": 4.2584523521342696e-05,
+      "loss": 0.0268,
+      "step": 6430
+    },
+    {
+      "epoch": 4.68,
+      "grad_norm": 0.7248146533966064,
+      "learning_rate": 4.256435084321795e-05,
+      "loss": 0.0131,
+      "step": 6440
+    },
+    {
+      "epoch": 4.68,
+      "grad_norm": 29.936283111572266,
+      "learning_rate": 4.2544178165093204e-05,
+      "loss": 0.0247,
+      "step": 6450
+    },
+    {
+      "epoch": 4.69,
+      "grad_norm": 15.446109771728516,
+      "learning_rate": 4.252400548696845e-05,
+      "loss": 0.044,
+      "step": 6460
+    },
+    {
+      "epoch": 4.7,
+      "grad_norm": 17.369585037231445,
+      "learning_rate": 4.25038328088437e-05,
+      "loss": 0.0611,
+      "step": 6470
+    },
+    {
+      "epoch": 4.7,
+      "grad_norm": 12.479216575622559,
+      "learning_rate": 4.2483660130718954e-05,
+      "loss": 0.0482,
+      "step": 6480
+    },
+    {
+      "epoch": 4.71,
+      "grad_norm": 7.170633316040039,
+      "learning_rate": 4.246348745259421e-05,
+      "loss": 0.0416,
+      "step": 6490
+    },
+    {
+      "epoch": 4.72,
+      "grad_norm": 0.019940435886383057,
+      "learning_rate": 4.244331477446946e-05,
+      "loss": 0.0083,
+      "step": 6500
+    },
+    {
+      "epoch": 4.73,
+      "grad_norm": 15.792586326599121,
+      "learning_rate": 4.242314209634472e-05,
+      "loss": 0.0235,
+      "step": 6510
+    },
+    {
+      "epoch": 4.73,
+      "grad_norm": 0.1848461925983429,
+      "learning_rate": 4.2402969418219965e-05,
+      "loss": 0.05,
+      "step": 6520
+    },
+    {
+      "epoch": 4.74,
+      "grad_norm": 9.763023376464844,
+      "learning_rate": 4.238279674009521e-05,
+      "loss": 0.0368,
+      "step": 6530
+    },
+    {
+      "epoch": 4.75,
+      "grad_norm": 0.3894762396812439,
+      "learning_rate": 4.236262406197047e-05,
+      "loss": 0.0259,
+      "step": 6540
+    },
+    {
+      "epoch": 4.75,
+      "grad_norm": 0.15589603781700134,
+      "learning_rate": 4.234245138384572e-05,
+      "loss": 0.0214,
+      "step": 6550
+    },
+    {
+      "epoch": 4.76,
+      "grad_norm": 10.842832565307617,
+      "learning_rate": 4.2322278705720976e-05,
+      "loss": 0.029,
+      "step": 6560
+    },
+    {
+      "epoch": 4.77,
+      "grad_norm": 0.747986376285553,
+      "learning_rate": 4.230210602759623e-05,
+      "loss": 0.0166,
+      "step": 6570
+    },
+    {
+      "epoch": 4.78,
+      "grad_norm": 6.645331382751465,
+      "learning_rate": 4.228193334947148e-05,
+      "loss": 0.0122,
+      "step": 6580
+    },
+    {
+      "epoch": 4.78,
+      "grad_norm": 0.048543334007263184,
+      "learning_rate": 4.2261760671346726e-05,
+      "loss": 0.0219,
+      "step": 6590
+    },
+    {
+      "epoch": 4.79,
+      "grad_norm": 3.219879627227783,
+      "learning_rate": 4.224158799322198e-05,
+      "loss": 0.0327,
+      "step": 6600
+    },
+    {
+      "epoch": 4.8,
+      "grad_norm": 0.23735283315181732,
+      "learning_rate": 4.2221415315097235e-05,
+      "loss": 0.0462,
+      "step": 6610
+    },
+    {
+      "epoch": 4.81,
+      "grad_norm": 0.4189997911453247,
+      "learning_rate": 4.220124263697249e-05,
+      "loss": 0.0067,
+      "step": 6620
+    },
+    {
+      "epoch": 4.81,
+      "grad_norm": 15.188400268554688,
+      "learning_rate": 4.2181069958847744e-05,
+      "loss": 0.0176,
+      "step": 6630
+    },
+    {
+      "epoch": 4.82,
+      "grad_norm": 22.570255279541016,
+      "learning_rate": 4.216089728072299e-05,
+      "loss": 0.0241,
+      "step": 6640
+    },
+    {
+      "epoch": 4.83,
+      "grad_norm": 18.151098251342773,
+      "learning_rate": 4.214072460259824e-05,
+      "loss": 0.0832,
+      "step": 6650
+    },
+    {
+      "epoch": 4.83,
+      "grad_norm": 8.57191276550293,
+      "learning_rate": 4.2120551924473494e-05,
+      "loss": 0.0193,
+      "step": 6660
+    },
+    {
+      "epoch": 4.84,
+      "grad_norm": 9.605646133422852,
+      "learning_rate": 4.210037924634875e-05,
+      "loss": 0.0168,
+      "step": 6670
+    },
+    {
+      "epoch": 4.85,
+      "grad_norm": 2.7312123775482178,
+      "learning_rate": 4.2080206568224e-05,
+      "loss": 0.0612,
+      "step": 6680
+    },
+    {
+      "epoch": 4.86,
+      "grad_norm": 1.0449771881103516,
+      "learning_rate": 4.206003389009926e-05,
+      "loss": 0.0238,
+      "step": 6690
+    },
+    {
+      "epoch": 4.86,
+      "grad_norm": 14.31949234008789,
+      "learning_rate": 4.20398612119745e-05,
+      "loss": 0.0185,
+      "step": 6700
+    },
+    {
+      "epoch": 4.87,
+      "grad_norm": 0.09144321084022522,
+      "learning_rate": 4.201968853384975e-05,
+      "loss": 0.0382,
+      "step": 6710
+    },
+    {
+      "epoch": 4.88,
+      "grad_norm": 2.1980929374694824,
+      "learning_rate": 4.199951585572501e-05,
+      "loss": 0.0169,
+      "step": 6720
+    },
+    {
+      "epoch": 4.89,
+      "grad_norm": 3.9242584705352783,
+      "learning_rate": 4.197934317760026e-05,
+      "loss": 0.0117,
+      "step": 6730
+    },
+    {
+      "epoch": 4.89,
+      "grad_norm": 0.009010029025375843,
+      "learning_rate": 4.1959170499475516e-05,
+      "loss": 0.027,
+      "step": 6740
+    },
+    {
+      "epoch": 4.9,
+      "grad_norm": 0.02251308038830757,
+      "learning_rate": 4.193899782135077e-05,
+      "loss": 0.036,
+      "step": 6750
+    },
+    {
+      "epoch": 4.91,
+      "grad_norm": 0.2532757520675659,
+      "learning_rate": 4.191882514322601e-05,
+      "loss": 0.0098,
+      "step": 6760
+    },
+    {
+      "epoch": 4.91,
+      "grad_norm": 15.302732467651367,
+      "learning_rate": 4.1898652465101266e-05,
+      "loss": 0.0105,
+      "step": 6770
+    },
+    {
+      "epoch": 4.92,
+      "grad_norm": 9.087559700012207,
+      "learning_rate": 4.187847978697652e-05,
+      "loss": 0.0313,
+      "step": 6780
+    },
+    {
+      "epoch": 4.93,
+      "grad_norm": 0.8211730718612671,
+      "learning_rate": 4.1858307108851775e-05,
+      "loss": 0.0581,
+      "step": 6790
+    },
+    {
+      "epoch": 4.94,
+      "grad_norm": 0.49661290645599365,
+      "learning_rate": 4.183813443072703e-05,
+      "loss": 0.0336,
+      "step": 6800
+    },
+    {
+      "epoch": 4.94,
+      "grad_norm": 8.808603286743164,
+      "learning_rate": 4.181796175260228e-05,
+      "loss": 0.0216,
+      "step": 6810
+    },
+    {
+      "epoch": 4.95,
+      "grad_norm": 7.822020530700684,
+      "learning_rate": 4.1797789074477525e-05,
+      "loss": 0.0087,
+      "step": 6820
+    },
+    {
+      "epoch": 4.96,
+      "grad_norm": 7.684726715087891,
+      "learning_rate": 4.177761639635278e-05,
+      "loss": 0.0207,
+      "step": 6830
+    },
+    {
+      "epoch": 4.97,
+      "grad_norm": 29.289989471435547,
+      "learning_rate": 4.1757443718228034e-05,
+      "loss": 0.0635,
+      "step": 6840
+    },
+    {
+      "epoch": 4.97,
+      "grad_norm": 1.1796934604644775,
+      "learning_rate": 4.173727104010329e-05,
+      "loss": 0.0063,
+      "step": 6850
+    },
+    {
+      "epoch": 4.98,
+      "grad_norm": 4.763782024383545,
+      "learning_rate": 4.171709836197854e-05,
+      "loss": 0.0167,
+      "step": 6860
+    },
+    {
+      "epoch": 4.99,
+      "grad_norm": 14.398111343383789,
+      "learning_rate": 4.169692568385379e-05,
+      "loss": 0.0234,
+      "step": 6870
+    },
+    {
+      "epoch": 4.99,
+      "grad_norm": 9.24953842163086,
+      "learning_rate": 4.167675300572904e-05,
+      "loss": 0.0302,
+      "step": 6880
+    },
+    {
+      "epoch": 5.0,
+      "eval_accuracy": 0.997220583564006,
+      "eval_f1": 0.993579159787195,
+      "eval_loss": 0.007504656910896301,
+      "eval_precision": 0.9976317035945477,
+      "eval_recall": 0.9895594069743161,
+      "eval_roc_auc": 0.9999509363858444,
+      "eval_runtime": 386.139,
+      "eval_samples_per_second": 228.28,
+      "eval_steps_per_second": 14.269,
+      "step": 6887
+    },
+    {
+      "epoch": 5.0,
+      "grad_norm": 0.4994419515132904,
+      "learning_rate": 4.165658032760429e-05,
+      "loss": 0.0273,
+      "step": 6890
+    },
+    {
+      "epoch": 5.01,
+      "grad_norm": 0.043419767171144485,
+      "learning_rate": 4.163640764947955e-05,
+      "loss": 0.0096,
+      "step": 6900
+    },
+    {
+      "epoch": 5.02,
+      "grad_norm": 0.07488001883029938,
+      "learning_rate": 4.16162349713548e-05,
+      "loss": 0.0067,
+      "step": 6910
+    },
+    {
+      "epoch": 5.02,
+      "grad_norm": 15.292566299438477,
+      "learning_rate": 4.1596062293230056e-05,
+      "loss": 0.0277,
+      "step": 6920
+    },
+    {
+      "epoch": 5.03,
+      "grad_norm": 1.0951852798461914,
+      "learning_rate": 4.1575889615105304e-05,
+      "loss": 0.0054,
+      "step": 6930
+    },
+    {
+      "epoch": 5.04,
+      "grad_norm": 11.888832092285156,
+      "learning_rate": 4.155571693698055e-05,
+      "loss": 0.0168,
+      "step": 6940
+    },
+    {
+      "epoch": 5.05,
+      "grad_norm": 0.0656825602054596,
+      "learning_rate": 4.1535544258855806e-05,
+      "loss": 0.0123,
+      "step": 6950
+    },
+    {
+      "epoch": 5.05,
+      "grad_norm": 22.420522689819336,
+      "learning_rate": 4.151537158073106e-05,
+      "loss": 0.0053,
+      "step": 6960
+    },
+    {
+      "epoch": 5.06,
+      "grad_norm": 1.976753830909729,
+      "learning_rate": 4.1495198902606315e-05,
+      "loss": 0.0173,
+      "step": 6970
+    },
+    {
+      "epoch": 5.07,
+      "grad_norm": 2.0210866928100586,
+      "learning_rate": 4.147502622448157e-05,
+      "loss": 0.0098,
+      "step": 6980
+    },
+    {
+      "epoch": 5.07,
+      "grad_norm": 1.1973435878753662,
+      "learning_rate": 4.145485354635682e-05,
+      "loss": 0.054,
+      "step": 6990
+    },
+    {
+      "epoch": 5.08,
+      "grad_norm": 0.7293339371681213,
+      "learning_rate": 4.1434680868232065e-05,
+      "loss": 0.0236,
+      "step": 7000
+    },
+    {
+      "epoch": 5.09,
+      "grad_norm": 0.39437663555145264,
+      "learning_rate": 4.141450819010732e-05,
+      "loss": 0.0212,
+      "step": 7010
+    },
+    {
+      "epoch": 5.1,
+      "grad_norm": 0.5972071290016174,
+      "learning_rate": 4.1394335511982573e-05,
+      "loss": 0.0189,
+      "step": 7020
+    },
+    {
+      "epoch": 5.1,
+      "grad_norm": 1.5615042448043823,
+      "learning_rate": 4.137416283385783e-05,
+      "loss": 0.0195,
+      "step": 7030
+    },
+    {
+      "epoch": 5.11,
+      "grad_norm": 0.0803157240152359,
+      "learning_rate": 4.1353990155733076e-05,
+      "loss": 0.0216,
+      "step": 7040
+    },
+    {
+      "epoch": 5.12,
+      "grad_norm": 0.17535190284252167,
+      "learning_rate": 4.133381747760833e-05,
+      "loss": 0.0232,
+      "step": 7050
+    },
+    {
+      "epoch": 5.13,
+      "grad_norm": 6.3075385093688965,
+      "learning_rate": 4.131364479948358e-05,
+      "loss": 0.0261,
+      "step": 7060
+    },
+    {
+      "epoch": 5.13,
+      "grad_norm": 0.352460652589798,
+      "learning_rate": 4.129347212135883e-05,
+      "loss": 0.0184,
+      "step": 7070
+    },
+    {
+      "epoch": 5.14,
+      "grad_norm": 0.83339524269104,
+      "learning_rate": 4.127329944323409e-05,
+      "loss": 0.019,
+      "step": 7080
+    },
+    {
+      "epoch": 5.15,
+      "grad_norm": 0.037650395184755325,
+      "learning_rate": 4.125312676510934e-05,
+      "loss": 0.0139,
+      "step": 7090
+    },
+    {
+      "epoch": 5.15,
+      "grad_norm": 0.45141324400901794,
+      "learning_rate": 4.123295408698459e-05,
+      "loss": 0.0224,
+      "step": 7100
+    },
+    {
+      "epoch": 5.16,
+      "grad_norm": 4.516534805297852,
+      "learning_rate": 4.121278140885984e-05,
+      "loss": 0.0212,
+      "step": 7110
+    },
+    {
+      "epoch": 5.17,
+      "grad_norm": 3.1175129413604736,
+      "learning_rate": 4.119260873073509e-05,
+      "loss": 0.0171,
+      "step": 7120
+    },
+    {
+      "epoch": 5.18,
+      "grad_norm": 0.03533313795924187,
+      "learning_rate": 4.1172436052610345e-05,
+      "loss": 0.0136,
+      "step": 7130
+    },
+    {
+      "epoch": 5.18,
+      "grad_norm": 0.273041307926178,
+      "learning_rate": 4.11522633744856e-05,
+      "loss": 0.0452,
+      "step": 7140
+    },
+    {
+      "epoch": 5.19,
+      "grad_norm": 0.5384491086006165,
+      "learning_rate": 4.1132090696360854e-05,
+      "loss": 0.0313,
+      "step": 7150
+    },
+    {
+      "epoch": 5.2,
+      "grad_norm": 0.5866811871528625,
+      "learning_rate": 4.11119180182361e-05,
+      "loss": 0.0235,
+      "step": 7160
+    },
+    {
+      "epoch": 5.21,
+      "grad_norm": 14.287243843078613,
+      "learning_rate": 4.1091745340111357e-05,
+      "loss": 0.0128,
+      "step": 7170
+    },
+    {
+      "epoch": 5.21,
+      "grad_norm": 13.50960636138916,
+      "learning_rate": 4.1071572661986604e-05,
+      "loss": 0.0175,
+      "step": 7180
+    },
+    {
+      "epoch": 5.22,
+      "grad_norm": 13.826930046081543,
+      "learning_rate": 4.105139998386186e-05,
+      "loss": 0.0236,
+      "step": 7190
+    },
+    {
+      "epoch": 5.23,
+      "grad_norm": 6.079420566558838,
+      "learning_rate": 4.103122730573711e-05,
+      "loss": 0.0313,
+      "step": 7200
+    },
+    {
+      "epoch": 5.23,
+      "grad_norm": 20.14948272705078,
+      "learning_rate": 4.101105462761237e-05,
+      "loss": 0.0289,
+      "step": 7210
+    },
+    {
+      "epoch": 5.24,
+      "grad_norm": 10.299389839172363,
+      "learning_rate": 4.0990881949487615e-05,
+      "loss": 0.0041,
+      "step": 7220
+    },
+    {
+      "epoch": 5.25,
+      "grad_norm": 0.27182748913764954,
+      "learning_rate": 4.097070927136287e-05,
+      "loss": 0.0194,
+      "step": 7230
+    },
+    {
+      "epoch": 5.26,
+      "grad_norm": 17.97000503540039,
+      "learning_rate": 4.095053659323812e-05,
+      "loss": 0.0186,
+      "step": 7240
+    },
+    {
+      "epoch": 5.26,
+      "grad_norm": 11.942718505859375,
+      "learning_rate": 4.093036391511337e-05,
+      "loss": 0.012,
+      "step": 7250
+    },
+    {
+      "epoch": 5.27,
+      "grad_norm": 17.377195358276367,
+      "learning_rate": 4.0910191236988626e-05,
+      "loss": 0.0375,
+      "step": 7260
+    },
+    {
+      "epoch": 5.28,
+      "grad_norm": 0.3844599425792694,
+      "learning_rate": 4.0890018558863874e-05,
+      "loss": 0.0142,
+      "step": 7270
+    },
+    {
+      "epoch": 5.28,
+      "grad_norm": 0.04491892457008362,
+      "learning_rate": 4.086984588073913e-05,
+      "loss": 0.0246,
+      "step": 7280
+    },
+    {
+      "epoch": 5.29,
+      "grad_norm": 0.2385578751564026,
+      "learning_rate": 4.084967320261438e-05,
+      "loss": 0.0163,
+      "step": 7290
+    },
+    {
+      "epoch": 5.3,
+      "grad_norm": 15.862750053405762,
+      "learning_rate": 4.082950052448963e-05,
+      "loss": 0.0127,
+      "step": 7300
+    },
+    {
+      "epoch": 5.31,
+      "grad_norm": 0.6464446783065796,
+      "learning_rate": 4.0809327846364885e-05,
+      "loss": 0.0272,
+      "step": 7310
+    },
+    {
+      "epoch": 5.31,
+      "grad_norm": 0.06587512791156769,
+      "learning_rate": 4.078915516824014e-05,
+      "loss": 0.0165,
+      "step": 7320
+    },
+    {
+      "epoch": 5.32,
+      "grad_norm": 0.5429533123970032,
+      "learning_rate": 4.076898249011539e-05,
+      "loss": 0.0168,
+      "step": 7330
+    },
+    {
+      "epoch": 5.33,
+      "grad_norm": 0.10347855091094971,
+      "learning_rate": 4.074880981199064e-05,
+      "loss": 0.0011,
+      "step": 7340
+    },
+    {
+      "epoch": 5.34,
+      "grad_norm": 1.6859651803970337,
+      "learning_rate": 4.0728637133865896e-05,
+      "loss": 0.0088,
+      "step": 7350
+    },
+    {
+      "epoch": 5.34,
+      "grad_norm": 0.5965036749839783,
+      "learning_rate": 4.0708464455741144e-05,
+      "loss": 0.0325,
+      "step": 7360
+    },
+    {
+      "epoch": 5.35,
+      "grad_norm": 0.03945688530802727,
+      "learning_rate": 4.06882917776164e-05,
+      "loss": 0.0335,
+      "step": 7370
+    },
+    {
+      "epoch": 5.36,
+      "grad_norm": 0.18698054552078247,
+      "learning_rate": 4.066811909949165e-05,
+      "loss": 0.0127,
+      "step": 7380
+    },
+    {
+      "epoch": 5.36,
+      "grad_norm": 0.028903458267450333,
+      "learning_rate": 4.06479464213669e-05,
+      "loss": 0.0129,
+      "step": 7390
+    },
+    {
+      "epoch": 5.37,
+      "grad_norm": 12.742640495300293,
+      "learning_rate": 4.0627773743242155e-05,
+      "loss": 0.0209,
+      "step": 7400
+    },
+    {
+      "epoch": 5.38,
+      "grad_norm": 0.21819807589054108,
+      "learning_rate": 4.060760106511741e-05,
+      "loss": 0.0198,
+      "step": 7410
+    },
+    {
+      "epoch": 5.39,
+      "grad_norm": 0.0063743069767951965,
+      "learning_rate": 4.058742838699266e-05,
+      "loss": 0.0225,
+      "step": 7420
+    },
+    {
+      "epoch": 5.39,
+      "grad_norm": 6.776240825653076,
+      "learning_rate": 4.056725570886791e-05,
+      "loss": 0.0232,
+      "step": 7430
+    },
+    {
+      "epoch": 5.4,
+      "grad_norm": 7.465283393859863,
+      "learning_rate": 4.0547083030743166e-05,
+      "loss": 0.0411,
+      "step": 7440
+    },
+    {
+      "epoch": 5.41,
+      "grad_norm": 0.13503609597682953,
+      "learning_rate": 4.0526910352618414e-05,
+      "loss": 0.0198,
+      "step": 7450
+    },
+    {
+      "epoch": 5.42,
+      "grad_norm": 5.632655620574951,
+      "learning_rate": 4.050673767449367e-05,
+      "loss": 0.0254,
+      "step": 7460
+    },
+    {
+      "epoch": 5.42,
+      "grad_norm": 1.882933497428894,
+      "learning_rate": 4.048656499636892e-05,
+      "loss": 0.0129,
+      "step": 7470
+    },
+    {
+      "epoch": 5.43,
+      "grad_norm": 2.3281795978546143,
+      "learning_rate": 4.046639231824417e-05,
+      "loss": 0.0061,
+      "step": 7480
+    },
+    {
+      "epoch": 5.44,
+      "grad_norm": 0.12824377417564392,
+      "learning_rate": 4.0446219640119425e-05,
+      "loss": 0.0061,
+      "step": 7490
+    },
+    {
+      "epoch": 5.44,
+      "grad_norm": 1.7312507629394531,
+      "learning_rate": 4.042604696199468e-05,
+      "loss": 0.0384,
+      "step": 7500
+    },
+    {
+      "epoch": 5.45,
+      "grad_norm": 28.485464096069336,
+      "learning_rate": 4.040587428386993e-05,
+      "loss": 0.062,
+      "step": 7510
+    },
+    {
+      "epoch": 5.46,
+      "grad_norm": 0.2245478332042694,
+      "learning_rate": 4.038570160574518e-05,
+      "loss": 0.0154,
+      "step": 7520
+    },
+    {
+      "epoch": 5.47,
+      "grad_norm": 0.02461441047489643,
+      "learning_rate": 4.0365528927620436e-05,
+      "loss": 0.0162,
+      "step": 7530
+    },
+    {
+      "epoch": 5.47,
+      "grad_norm": 0.0920153334736824,
+      "learning_rate": 4.0345356249495684e-05,
+      "loss": 0.0502,
+      "step": 7540
+    },
+    {
+      "epoch": 5.48,
+      "grad_norm": 3.1807138919830322,
+      "learning_rate": 4.032518357137094e-05,
+      "loss": 0.0179,
+      "step": 7550
+    },
+    {
+      "epoch": 5.49,
+      "grad_norm": 1.7319307327270508,
+      "learning_rate": 4.0305010893246186e-05,
+      "loss": 0.0105,
+      "step": 7560
+    },
+    {
+      "epoch": 5.5,
+      "grad_norm": 5.294769287109375,
+      "learning_rate": 4.028483821512144e-05,
+      "loss": 0.0019,
+      "step": 7570
+    },
+    {
+      "epoch": 5.5,
+      "grad_norm": 0.0027535264380276203,
+      "learning_rate": 4.0264665536996695e-05,
+      "loss": 0.0041,
+      "step": 7580
+    },
+    {
+      "epoch": 5.51,
+      "grad_norm": 40.27950668334961,
+      "learning_rate": 4.024449285887195e-05,
+      "loss": 0.0186,
+      "step": 7590
+    },
+    {
+      "epoch": 5.52,
+      "grad_norm": 0.08031009882688522,
+      "learning_rate": 4.02243201807472e-05,
+      "loss": 0.0153,
+      "step": 7600
+    },
+    {
+      "epoch": 5.52,
+      "grad_norm": 0.31836315989494324,
+      "learning_rate": 4.020414750262245e-05,
+      "loss": 0.0159,
+      "step": 7610
+    },
+    {
+      "epoch": 5.53,
+      "grad_norm": 0.3263595700263977,
+      "learning_rate": 4.01839748244977e-05,
+      "loss": 0.0286,
+      "step": 7620
+    },
+    {
+      "epoch": 5.54,
+      "grad_norm": 29.408348083496094,
+      "learning_rate": 4.0163802146372954e-05,
+      "loss": 0.0328,
+      "step": 7630
+    },
+    {
+      "epoch": 5.55,
+      "grad_norm": 0.6650347113609314,
+      "learning_rate": 4.014362946824821e-05,
+      "loss": 0.0076,
+      "step": 7640
+    },
+    {
+      "epoch": 5.55,
+      "grad_norm": 0.03056545928120613,
+      "learning_rate": 4.012345679012346e-05,
+      "loss": 0.0193,
+      "step": 7650
+    },
+    {
+      "epoch": 5.56,
+      "grad_norm": 21.62334442138672,
+      "learning_rate": 4.010328411199871e-05,
+      "loss": 0.0288,
+      "step": 7660
+    },
+    {
+      "epoch": 5.57,
+      "grad_norm": 0.035483092069625854,
+      "learning_rate": 4.0083111433873965e-05,
+      "loss": 0.0227,
+      "step": 7670
+    },
+    {
+      "epoch": 5.58,
+      "grad_norm": 0.013101032935082912,
+      "learning_rate": 4.006293875574921e-05,
+      "loss": 0.0536,
+      "step": 7680
+    },
+    {
+      "epoch": 5.58,
+      "grad_norm": 0.007710463833063841,
+      "learning_rate": 4.004276607762447e-05,
+      "loss": 0.022,
+      "step": 7690
+    },
+    {
+      "epoch": 5.59,
+      "grad_norm": 0.1644335836172104,
+      "learning_rate": 4.002259339949972e-05,
+      "loss": 0.0215,
+      "step": 7700
+    },
+    {
+      "epoch": 5.6,
+      "grad_norm": 6.9764404296875,
+      "learning_rate": 4.0002420721374976e-05,
+      "loss": 0.0314,
+      "step": 7710
+    },
+    {
+      "epoch": 5.6,
+      "grad_norm": 0.5656896233558655,
+      "learning_rate": 3.9982248043250223e-05,
+      "loss": 0.0501,
+      "step": 7720
+    },
+    {
+      "epoch": 5.61,
+      "grad_norm": 0.26260992884635925,
+      "learning_rate": 3.996207536512548e-05,
+      "loss": 0.0281,
+      "step": 7730
+    },
+    {
+      "epoch": 5.62,
+      "grad_norm": 0.4968787729740143,
+      "learning_rate": 3.9941902687000726e-05,
+      "loss": 0.016,
+      "step": 7740
+    },
+    {
+      "epoch": 5.63,
+      "grad_norm": 10.740684509277344,
+      "learning_rate": 3.992173000887598e-05,
+      "loss": 0.0176,
+      "step": 7750
+    },
+    {
+      "epoch": 5.63,
+      "grad_norm": 1.0777477025985718,
+      "learning_rate": 3.9901557330751234e-05,
+      "loss": 0.011,
+      "step": 7760
+    },
+    {
+      "epoch": 5.64,
+      "grad_norm": 0.002109379041939974,
+      "learning_rate": 3.988138465262649e-05,
+      "loss": 0.0094,
+      "step": 7770
+    },
+    {
+      "epoch": 5.65,
+      "grad_norm": 1.016344428062439,
+      "learning_rate": 3.986121197450174e-05,
+      "loss": 0.0249,
+      "step": 7780
+    },
+    {
+      "epoch": 5.66,
+      "grad_norm": 22.406539916992188,
+      "learning_rate": 3.9841039296376984e-05,
+      "loss": 0.0149,
+      "step": 7790
+    },
+    {
+      "epoch": 5.66,
+      "grad_norm": 1.52140474319458,
+      "learning_rate": 3.982086661825224e-05,
+      "loss": 0.021,
+      "step": 7800
+    },
+    {
+      "epoch": 5.67,
+      "grad_norm": 0.23657099902629852,
+      "learning_rate": 3.980069394012749e-05,
+      "loss": 0.0368,
+      "step": 7810
+    },
+    {
+      "epoch": 5.68,
+      "grad_norm": 0.6905073523521423,
+      "learning_rate": 3.978052126200275e-05,
+      "loss": 0.0199,
+      "step": 7820
+    },
+    {
+      "epoch": 5.68,
+      "grad_norm": 1.6881111860275269,
+      "learning_rate": 3.9760348583877995e-05,
+      "loss": 0.023,
+      "step": 7830
+    },
+    {
+      "epoch": 5.69,
+      "grad_norm": 0.24087966978549957,
+      "learning_rate": 3.974017590575325e-05,
+      "loss": 0.0352,
+      "step": 7840
+    },
+    {
+      "epoch": 5.7,
+      "grad_norm": 0.4764993488788605,
+      "learning_rate": 3.97200032276285e-05,
+      "loss": 0.0434,
+      "step": 7850
+    },
+    {
+      "epoch": 5.71,
+      "grad_norm": 2.340041399002075,
+      "learning_rate": 3.969983054950375e-05,
+      "loss": 0.0265,
+      "step": 7860
+    },
+    {
+      "epoch": 5.71,
+      "grad_norm": 11.737015724182129,
+      "learning_rate": 3.9679657871379007e-05,
+      "loss": 0.0176,
+      "step": 7870
+    },
+    {
+      "epoch": 5.72,
+      "grad_norm": 21.019853591918945,
+      "learning_rate": 3.965948519325426e-05,
+      "loss": 0.0234,
+      "step": 7880
+    },
+    {
+      "epoch": 5.73,
+      "grad_norm": 0.0860639363527298,
+      "learning_rate": 3.963931251512951e-05,
+      "loss": 0.0107,
+      "step": 7890
+    },
+    {
+      "epoch": 5.74,
+      "grad_norm": 0.27008432149887085,
+      "learning_rate": 3.961913983700476e-05,
+      "loss": 0.0222,
+      "step": 7900
+    },
+    {
+      "epoch": 5.74,
+      "grad_norm": 0.33239853382110596,
+      "learning_rate": 3.959896715888001e-05,
+      "loss": 0.0101,
+      "step": 7910
+    },
+    {
+      "epoch": 5.75,
+      "grad_norm": 8.389657974243164,
+      "learning_rate": 3.9578794480755265e-05,
+      "loss": 0.0255,
+      "step": 7920
+    },
+    {
+      "epoch": 5.76,
+      "grad_norm": 0.008096696808934212,
+      "learning_rate": 3.955862180263052e-05,
+      "loss": 0.0299,
+      "step": 7930
+    },
+    {
+      "epoch": 5.76,
+      "grad_norm": 0.0216965414583683,
+      "learning_rate": 3.9538449124505774e-05,
+      "loss": 0.0181,
+      "step": 7940
+    },
+    {
+      "epoch": 5.77,
+      "grad_norm": 5.718286991119385,
+      "learning_rate": 3.951827644638102e-05,
+      "loss": 0.0114,
+      "step": 7950
+    },
+    {
+      "epoch": 5.78,
+      "grad_norm": 0.2025628387928009,
+      "learning_rate": 3.9498103768256276e-05,
+      "loss": 0.0253,
+      "step": 7960
+    },
+    {
+      "epoch": 5.79,
+      "grad_norm": 8.420141220092773,
+      "learning_rate": 3.9477931090131524e-05,
+      "loss": 0.0211,
+      "step": 7970
+    },
+    {
+      "epoch": 5.79,
+      "grad_norm": 0.15034617483615875,
+      "learning_rate": 3.945775841200678e-05,
+      "loss": 0.0089,
+      "step": 7980
+    },
+    {
+      "epoch": 5.8,
+      "grad_norm": 17.66687774658203,
+      "learning_rate": 3.943758573388203e-05,
+      "loss": 0.0302,
+      "step": 7990
+    },
+    {
+      "epoch": 5.81,
+      "grad_norm": 1.483130693435669,
+      "learning_rate": 3.941741305575729e-05,
+      "loss": 0.0335,
+      "step": 8000
+    },
+    {
+      "epoch": 5.81,
+      "grad_norm": 5.107597351074219,
+      "learning_rate": 3.9397240377632535e-05,
+      "loss": 0.0318,
+      "step": 8010
+    },
+    {
+      "epoch": 5.82,
+      "grad_norm": 5.176906585693359,
+      "learning_rate": 3.937706769950778e-05,
+      "loss": 0.0152,
+      "step": 8020
+    },
+    {
+      "epoch": 5.83,
+      "grad_norm": 0.008498461917042732,
+      "learning_rate": 3.935689502138304e-05,
+      "loss": 0.0219,
+      "step": 8030
+    },
+    {
+      "epoch": 5.84,
+      "grad_norm": 0.003639570204541087,
+      "learning_rate": 3.933672234325829e-05,
+      "loss": 0.0078,
+      "step": 8040
+    },
+    {
+      "epoch": 5.84,
+      "grad_norm": 0.7268972992897034,
+      "learning_rate": 3.9316549665133546e-05,
+      "loss": 0.0246,
+      "step": 8050
+    },
+    {
+      "epoch": 5.85,
+      "grad_norm": 0.3043120205402374,
+      "learning_rate": 3.92963769870088e-05,
+      "loss": 0.03,
+      "step": 8060
+    },
+    {
+      "epoch": 5.86,
+      "grad_norm": 0.22950021922588348,
+      "learning_rate": 3.927620430888405e-05,
+      "loss": 0.0191,
+      "step": 8070
+    },
+    {
+      "epoch": 5.87,
+      "grad_norm": 19.542064666748047,
+      "learning_rate": 3.9256031630759296e-05,
+      "loss": 0.0147,
+      "step": 8080
+    },
+    {
+      "epoch": 5.87,
+      "grad_norm": 0.10597487539052963,
+      "learning_rate": 3.923585895263455e-05,
+      "loss": 0.015,
+      "step": 8090
+    },
+    {
+      "epoch": 5.88,
+      "grad_norm": 29.48893928527832,
+      "learning_rate": 3.9215686274509805e-05,
+      "loss": 0.0145,
+      "step": 8100
+    },
+    {
+      "epoch": 5.89,
+      "grad_norm": 0.14151059091091156,
+      "learning_rate": 3.919551359638506e-05,
+      "loss": 0.0663,
+      "step": 8110
+    },
+    {
+      "epoch": 5.89,
+      "grad_norm": 0.3282245695590973,
+      "learning_rate": 3.9175340918260314e-05,
+      "loss": 0.0439,
+      "step": 8120
+    },
+    {
+      "epoch": 5.9,
+      "grad_norm": 6.798530101776123,
+      "learning_rate": 3.915516824013556e-05,
+      "loss": 0.0324,
+      "step": 8130
+    },
+    {
+      "epoch": 5.91,
+      "grad_norm": 0.0029835246969014406,
+      "learning_rate": 3.913499556201081e-05,
+      "loss": 0.0056,
+      "step": 8140
+    },
+    {
+      "epoch": 5.92,
+      "grad_norm": 0.050998538732528687,
+      "learning_rate": 3.9114822883886064e-05,
+      "loss": 0.0114,
+      "step": 8150
+    },
+    {
+      "epoch": 5.92,
+      "grad_norm": 0.00737336790189147,
+      "learning_rate": 3.909465020576132e-05,
+      "loss": 0.0044,
+      "step": 8160
+    },
+    {
+      "epoch": 5.93,
+      "grad_norm": 1.9938186407089233,
+      "learning_rate": 3.907447752763657e-05,
+      "loss": 0.0131,
+      "step": 8170
+    },
+    {
+      "epoch": 5.94,
+      "grad_norm": 2.643894672393799,
+      "learning_rate": 3.905430484951183e-05,
+      "loss": 0.0151,
+      "step": 8180
+    },
+    {
+      "epoch": 5.95,
+      "grad_norm": 0.9170699715614319,
+      "learning_rate": 3.9034132171387075e-05,
+      "loss": 0.0086,
+      "step": 8190
+    },
+    {
+      "epoch": 5.95,
+      "grad_norm": 0.027857156470417976,
+      "learning_rate": 3.901395949326232e-05,
+      "loss": 0.013,
+      "step": 8200
+    },
+    {
+      "epoch": 5.96,
+      "grad_norm": 0.005472167860716581,
+      "learning_rate": 3.899378681513758e-05,
+      "loss": 0.0317,
+      "step": 8210
+    },
+    {
+      "epoch": 5.97,
+      "grad_norm": 26.97398567199707,
+      "learning_rate": 3.897361413701283e-05,
+      "loss": 0.0407,
+      "step": 8220
+    },
+    {
+      "epoch": 5.97,
+      "grad_norm": 5.671159744262695,
+      "learning_rate": 3.8953441458888086e-05,
+      "loss": 0.0463,
+      "step": 8230
+    },
+    {
+      "epoch": 5.98,
+      "grad_norm": 0.005169416777789593,
+      "learning_rate": 3.893326878076334e-05,
+      "loss": 0.0087,
+      "step": 8240
+    },
+    {
+      "epoch": 5.99,
+      "grad_norm": 0.06089169904589653,
+      "learning_rate": 3.891309610263859e-05,
+      "loss": 0.0096,
+      "step": 8250
+    },
+    {
+      "epoch": 6.0,
+      "grad_norm": 0.1659688800573349,
+      "learning_rate": 3.8892923424513836e-05,
+      "loss": 0.0073,
+      "step": 8260
+    },
+    {
+      "epoch": 6.0,
+      "eval_accuracy": 0.9975609202704543,
+      "eval_f1": 0.9943809946946136,
+      "eval_loss": 0.006377417594194412,
+      "eval_precision": 0.9956560422881666,
+      "eval_recall": 0.9931092086030486,
+      "eval_roc_auc": 0.9999651368605089,
+      "eval_runtime": 386.667,
+      "eval_samples_per_second": 227.969,
+      "eval_steps_per_second": 14.25,
+      "step": 8265
+    },
+    {
+      "epoch": 6.0,
+      "grad_norm": 0.3143859803676605,
+      "learning_rate": 3.887275074638909e-05,
+      "loss": 0.0049,
+      "step": 8270
+    },
+    {
+      "epoch": 6.01,
+      "grad_norm": 0.06847266852855682,
+      "learning_rate": 3.8852578068264345e-05,
+      "loss": 0.0059,
+      "step": 8280
+    },
+    {
+      "epoch": 6.02,
+      "grad_norm": 23.29844093322754,
+      "learning_rate": 3.88324053901396e-05,
+      "loss": 0.0419,
+      "step": 8290
+    },
+    {
+      "epoch": 6.03,
+      "grad_norm": 0.010631518438458443,
+      "learning_rate": 3.8812232712014854e-05,
+      "loss": 0.0267,
+      "step": 8300
+    },
+    {
+      "epoch": 6.03,
+      "grad_norm": 0.05696843937039375,
+      "learning_rate": 3.87920600338901e-05,
+      "loss": 0.0061,
+      "step": 8310
+    },
+    {
+      "epoch": 6.04,
+      "grad_norm": 0.07849406450986862,
+      "learning_rate": 3.877188735576535e-05,
+      "loss": 0.0147,
+      "step": 8320
+    },
+    {
+      "epoch": 6.05,
+      "grad_norm": 0.006846526637673378,
+      "learning_rate": 3.8751714677640603e-05,
+      "loss": 0.0081,
+      "step": 8330
+    },
+    {
+      "epoch": 6.05,
+      "grad_norm": 2.307755708694458,
+      "learning_rate": 3.873154199951586e-05,
+      "loss": 0.0197,
+      "step": 8340
+    },
+    {
+      "epoch": 6.06,
+      "grad_norm": 16.542884826660156,
+      "learning_rate": 3.871136932139111e-05,
+      "loss": 0.0132,
+      "step": 8350
+    },
+    {
+      "epoch": 6.07,
+      "grad_norm": 0.5160506367683411,
+      "learning_rate": 3.869119664326637e-05,
+      "loss": 0.0022,
+      "step": 8360
+    },
+    {
+      "epoch": 6.08,
+      "grad_norm": 0.0947955921292305,
+      "learning_rate": 3.8671023965141615e-05,
+      "loss": 0.0067,
+      "step": 8370
+    },
+    {
+      "epoch": 6.08,
+      "grad_norm": 2.2483770847320557,
+      "learning_rate": 3.865085128701686e-05,
+      "loss": 0.0141,
+      "step": 8380
+    },
+    {
+      "epoch": 6.09,
+      "grad_norm": 2.6705210208892822,
+      "learning_rate": 3.863067860889212e-05,
+      "loss": 0.0032,
+      "step": 8390
+    },
+    {
+      "epoch": 6.1,
+      "grad_norm": 16.716379165649414,
+      "learning_rate": 3.861050593076737e-05,
+      "loss": 0.0193,
+      "step": 8400
+    },
+    {
+      "epoch": 6.11,
+      "grad_norm": 5.897037982940674,
+      "learning_rate": 3.8590333252642626e-05,
+      "loss": 0.0197,
+      "step": 8410
+    },
+    {
+      "epoch": 6.11,
+      "grad_norm": 3.2264394760131836,
+      "learning_rate": 3.857016057451788e-05,
+      "loss": 0.0099,
+      "step": 8420
+    },
+    {
+      "epoch": 6.12,
+      "grad_norm": 3.9197843074798584,
+      "learning_rate": 3.854998789639313e-05,
+      "loss": 0.0288,
+      "step": 8430
+    },
+    {
+      "epoch": 6.13,
+      "grad_norm": 0.2298147827386856,
+      "learning_rate": 3.8529815218268376e-05,
+      "loss": 0.0182,
+      "step": 8440
+    },
+    {
+      "epoch": 6.13,
+      "grad_norm": 7.973130702972412,
+      "learning_rate": 3.850964254014363e-05,
+      "loss": 0.021,
+      "step": 8450
+    },
+    {
+      "epoch": 6.14,
+      "grad_norm": 0.713499903678894,
+      "learning_rate": 3.8489469862018884e-05,
+      "loss": 0.0219,
+      "step": 8460
+    },
+    {
+      "epoch": 6.15,
+      "grad_norm": 0.004311359953135252,
+      "learning_rate": 3.846929718389414e-05,
+      "loss": 0.0205,
+      "step": 8470
+    },
+    {
+      "epoch": 6.16,
+      "grad_norm": 0.14316757023334503,
+      "learning_rate": 3.844912450576939e-05,
+      "loss": 0.0262,
+      "step": 8480
+    },
+    {
+      "epoch": 6.16,
+      "grad_norm": 0.22620250284671783,
+      "learning_rate": 3.842895182764464e-05,
+      "loss": 0.0109,
+      "step": 8490
+    },
+    {
+      "epoch": 6.17,
+      "grad_norm": 20.04834747314453,
+      "learning_rate": 3.840877914951989e-05,
+      "loss": 0.0218,
+      "step": 8500
+    },
+    {
+      "epoch": 6.18,
+      "grad_norm": 0.3280732035636902,
+      "learning_rate": 3.838860647139514e-05,
+      "loss": 0.0102,
+      "step": 8510
+    },
+    {
+      "epoch": 6.19,
+      "grad_norm": 0.06915584206581116,
+      "learning_rate": 3.83684337932704e-05,
+      "loss": 0.0247,
+      "step": 8520
+    },
+    {
+      "epoch": 6.19,
+      "grad_norm": 0.0052139488980174065,
+      "learning_rate": 3.834826111514565e-05,
+      "loss": 0.016,
+      "step": 8530
+    },
+    {
+      "epoch": 6.2,
+      "grad_norm": 1.4115946292877197,
+      "learning_rate": 3.83280884370209e-05,
+      "loss": 0.0067,
+      "step": 8540
+    },
+    {
+      "epoch": 6.21,
+      "grad_norm": 0.0992535725235939,
+      "learning_rate": 3.8307915758896154e-05,
+      "loss": 0.0163,
+      "step": 8550
+    },
+    {
+      "epoch": 6.21,
+      "grad_norm": 0.05193591117858887,
+      "learning_rate": 3.82877430807714e-05,
+      "loss": 0.027,
+      "step": 8560
+    },
+    {
+      "epoch": 6.22,
+      "grad_norm": 0.004640494007617235,
+      "learning_rate": 3.8267570402646656e-05,
+      "loss": 0.0183,
+      "step": 8570
+    },
+    {
+      "epoch": 6.23,
+      "grad_norm": 2.3029911518096924,
+      "learning_rate": 3.824739772452191e-05,
+      "loss": 0.0091,
+      "step": 8580
+    },
+    {
+      "epoch": 6.24,
+      "grad_norm": 0.002811912214383483,
+      "learning_rate": 3.8227225046397165e-05,
+      "loss": 0.0265,
+      "step": 8590
+    },
+    {
+      "epoch": 6.24,
+      "grad_norm": 0.10390030592679977,
+      "learning_rate": 3.820705236827241e-05,
+      "loss": 0.0186,
+      "step": 8600
+    },
+    {
+      "epoch": 6.25,
+      "grad_norm": 3.3613178730010986,
+      "learning_rate": 3.818687969014767e-05,
+      "loss": 0.0144,
+      "step": 8610
+    },
+    {
+      "epoch": 6.26,
+      "grad_norm": 0.3605481684207916,
+      "learning_rate": 3.8166707012022915e-05,
+      "loss": 0.0421,
+      "step": 8620
+    },
+    {
+      "epoch": 6.26,
+      "grad_norm": 0.12366246432065964,
+      "learning_rate": 3.814653433389817e-05,
+      "loss": 0.0105,
+      "step": 8630
+    },
+    {
+      "epoch": 6.27,
+      "grad_norm": 3.998110294342041,
+      "learning_rate": 3.8126361655773424e-05,
+      "loss": 0.0449,
+      "step": 8640
+    },
+    {
+      "epoch": 6.28,
+      "grad_norm": 3.1826484203338623,
+      "learning_rate": 3.810618897764868e-05,
+      "loss": 0.0302,
+      "step": 8650
+    },
+    {
+      "epoch": 6.29,
+      "grad_norm": 0.11859409511089325,
+      "learning_rate": 3.8086016299523926e-05,
+      "loss": 0.013,
+      "step": 8660
+    },
+    {
+      "epoch": 6.29,
+      "grad_norm": 1.415103793144226,
+      "learning_rate": 3.806584362139918e-05,
+      "loss": 0.0273,
+      "step": 8670
+    },
+    {
+      "epoch": 6.3,
+      "grad_norm": 0.10443487763404846,
+      "learning_rate": 3.804567094327443e-05,
+      "loss": 0.0236,
+      "step": 8680
+    },
+    {
+      "epoch": 6.31,
+      "grad_norm": 3.090874671936035,
+      "learning_rate": 3.802549826514968e-05,
+      "loss": 0.0326,
+      "step": 8690
+    },
+    {
+      "epoch": 6.32,
+      "grad_norm": 0.018442168831825256,
+      "learning_rate": 3.800532558702494e-05,
+      "loss": 0.0238,
+      "step": 8700
+    },
+    {
+      "epoch": 6.32,
+      "grad_norm": 0.057262253016233444,
+      "learning_rate": 3.798515290890019e-05,
+      "loss": 0.0048,
+      "step": 8710
+    },
+    {
+      "epoch": 6.33,
+      "grad_norm": 0.06799819320440292,
+      "learning_rate": 3.796498023077544e-05,
+      "loss": 0.0098,
+      "step": 8720
+    },
+    {
+      "epoch": 6.34,
+      "grad_norm": 0.0023398185148835182,
+      "learning_rate": 3.7944807552650694e-05,
+      "loss": 0.0153,
+      "step": 8730
+    },
+    {
+      "epoch": 6.34,
+      "grad_norm": 0.2396043837070465,
+      "learning_rate": 3.792463487452594e-05,
+      "loss": 0.0063,
+      "step": 8740
+    },
+    {
+      "epoch": 6.35,
+      "grad_norm": 18.22637176513672,
+      "learning_rate": 3.7904462196401196e-05,
+      "loss": 0.0098,
+      "step": 8750
+    },
+    {
+      "epoch": 6.36,
+      "grad_norm": 23.354751586914062,
+      "learning_rate": 3.788428951827645e-05,
+      "loss": 0.0068,
+      "step": 8760
+    },
+    {
+      "epoch": 6.37,
+      "grad_norm": 0.0036677473690360785,
+      "learning_rate": 3.78641168401517e-05,
+      "loss": 0.0038,
+      "step": 8770
+    },
+    {
+      "epoch": 6.37,
+      "grad_norm": 0.006179885007441044,
+      "learning_rate": 3.784394416202695e-05,
+      "loss": 0.0075,
+      "step": 8780
+    },
+    {
+      "epoch": 6.38,
+      "grad_norm": 0.1622333824634552,
+      "learning_rate": 3.782377148390221e-05,
+      "loss": 0.0057,
+      "step": 8790
+    },
+    {
+      "epoch": 6.39,
+      "grad_norm": 0.019087301567196846,
+      "learning_rate": 3.7803598805777455e-05,
+      "loss": 0.0084,
+      "step": 8800
+    },
+    {
+      "epoch": 6.4,
+      "grad_norm": 0.09246546030044556,
+      "learning_rate": 3.778342612765271e-05,
+      "loss": 0.0086,
+      "step": 8810
+    },
+    {
+      "epoch": 6.4,
+      "grad_norm": 1.2068523168563843,
+      "learning_rate": 3.7763253449527964e-05,
+      "loss": 0.0308,
+      "step": 8820
+    },
+    {
+      "epoch": 6.41,
+      "grad_norm": 11.821285247802734,
+      "learning_rate": 3.774308077140321e-05,
+      "loss": 0.0085,
+      "step": 8830
+    },
+    {
+      "epoch": 6.42,
+      "grad_norm": 8.977716445922852,
+      "learning_rate": 3.7722908093278466e-05,
+      "loss": 0.0068,
+      "step": 8840
+    },
+    {
+      "epoch": 6.42,
+      "grad_norm": 0.0026208017952740192,
+      "learning_rate": 3.7702735415153714e-05,
+      "loss": 0.0109,
+      "step": 8850
+    },
+    {
+      "epoch": 6.43,
+      "grad_norm": 0.00370892439968884,
+      "learning_rate": 3.768256273702897e-05,
+      "loss": 0.0169,
+      "step": 8860
+    },
+    {
+      "epoch": 6.44,
+      "grad_norm": 0.37159231305122375,
+      "learning_rate": 3.766239005890422e-05,
+      "loss": 0.005,
+      "step": 8870
+    },
+    {
+      "epoch": 6.45,
+      "grad_norm": 0.39433690905570984,
+      "learning_rate": 3.764221738077948e-05,
+      "loss": 0.0094,
+      "step": 8880
+    },
+    {
+      "epoch": 6.45,
+      "grad_norm": 0.17372171580791473,
+      "learning_rate": 3.7622044702654725e-05,
+      "loss": 0.018,
+      "step": 8890
+    },
+    {
+      "epoch": 6.46,
+      "grad_norm": 0.02303539402782917,
+      "learning_rate": 3.760187202452998e-05,
+      "loss": 0.015,
+      "step": 8900
+    },
+    {
+      "epoch": 6.47,
+      "grad_norm": 29.997024536132812,
+      "learning_rate": 3.758169934640523e-05,
+      "loss": 0.023,
+      "step": 8910
+    },
+    {
+      "epoch": 6.48,
+      "grad_norm": 28.65287971496582,
+      "learning_rate": 3.756152666828048e-05,
+      "loss": 0.0163,
+      "step": 8920
+    },
+    {
+      "epoch": 6.48,
+      "grad_norm": 0.004773481283336878,
+      "learning_rate": 3.7541353990155736e-05,
+      "loss": 0.007,
+      "step": 8930
+    },
+    {
+      "epoch": 6.49,
+      "grad_norm": 19.97212028503418,
+      "learning_rate": 3.752118131203099e-05,
+      "loss": 0.0238,
+      "step": 8940
+    },
+    {
+      "epoch": 6.5,
+      "grad_norm": 0.3827419877052307,
+      "learning_rate": 3.750100863390624e-05,
+      "loss": 0.0245,
+      "step": 8950
+    },
+    {
+      "epoch": 6.5,
+      "grad_norm": 0.15338027477264404,
+      "learning_rate": 3.748083595578149e-05,
+      "loss": 0.0162,
+      "step": 8960
+    },
+    {
+      "epoch": 6.51,
+      "grad_norm": 16.326635360717773,
+      "learning_rate": 3.746066327765674e-05,
+      "loss": 0.0166,
+      "step": 8970
+    },
+    {
+      "epoch": 6.52,
+      "grad_norm": 0.9681591987609863,
+      "learning_rate": 3.7440490599531995e-05,
+      "loss": 0.0065,
+      "step": 8980
+    },
+    {
+      "epoch": 6.53,
+      "grad_norm": 18.14190673828125,
+      "learning_rate": 3.742031792140725e-05,
+      "loss": 0.0255,
+      "step": 8990
+    },
+    {
+      "epoch": 6.53,
+      "grad_norm": 5.85013484954834,
+      "learning_rate": 3.74001452432825e-05,
+      "loss": 0.0071,
+      "step": 9000
+    },
+    {
+      "epoch": 6.54,
+      "grad_norm": 0.15171761810779572,
+      "learning_rate": 3.737997256515775e-05,
+      "loss": 0.0121,
+      "step": 9010
+    },
+    {
+      "epoch": 6.55,
+      "grad_norm": 0.031181402504444122,
+      "learning_rate": 3.7359799887033006e-05,
+      "loss": 0.0126,
+      "step": 9020
+    },
+    {
+      "epoch": 6.56,
+      "grad_norm": 0.003250251989811659,
+      "learning_rate": 3.7339627208908253e-05,
+      "loss": 0.0045,
+      "step": 9030
+    },
+    {
+      "epoch": 6.56,
+      "grad_norm": 8.333500862121582,
+      "learning_rate": 3.731945453078351e-05,
+      "loss": 0.0225,
+      "step": 9040
+    },
+    {
+      "epoch": 6.57,
+      "grad_norm": 0.039202239364385605,
+      "learning_rate": 3.729928185265876e-05,
+      "loss": 0.0132,
+      "step": 9050
+    },
+    {
+      "epoch": 6.58,
+      "grad_norm": 11.631218910217285,
+      "learning_rate": 3.727910917453401e-05,
+      "loss": 0.0262,
+      "step": 9060
+    },
+    {
+      "epoch": 6.58,
+      "grad_norm": 6.861734390258789,
+      "learning_rate": 3.7258936496409265e-05,
+      "loss": 0.0419,
+      "step": 9070
+    },
+    {
+      "epoch": 6.59,
+      "grad_norm": 0.08070015907287598,
+      "learning_rate": 3.723876381828452e-05,
+      "loss": 0.0584,
+      "step": 9080
+    },
+    {
+      "epoch": 6.6,
+      "grad_norm": 19.176902770996094,
+      "learning_rate": 3.721859114015977e-05,
+      "loss": 0.0271,
+      "step": 9090
+    },
+    {
+      "epoch": 6.61,
+      "grad_norm": 0.048521075397729874,
+      "learning_rate": 3.719841846203502e-05,
+      "loss": 0.0117,
+      "step": 9100
+    },
+    {
+      "epoch": 6.61,
+      "grad_norm": 0.39843207597732544,
+      "learning_rate": 3.7178245783910276e-05,
+      "loss": 0.0072,
+      "step": 9110
+    },
+    {
+      "epoch": 6.62,
+      "grad_norm": 0.0638517439365387,
+      "learning_rate": 3.715807310578552e-05,
+      "loss": 0.0088,
+      "step": 9120
+    },
+    {
+      "epoch": 6.63,
+      "grad_norm": 0.6130620241165161,
+      "learning_rate": 3.713790042766078e-05,
+      "loss": 0.0039,
+      "step": 9130
+    },
+    {
+      "epoch": 6.64,
+      "grad_norm": 1.8792153596878052,
+      "learning_rate": 3.711772774953603e-05,
+      "loss": 0.0136,
+      "step": 9140
+    },
+    {
+      "epoch": 6.64,
+      "grad_norm": 3.8683571815490723,
+      "learning_rate": 3.709755507141128e-05,
+      "loss": 0.0241,
+      "step": 9150
+    },
+    {
+      "epoch": 6.65,
+      "grad_norm": 1.621826171875,
+      "learning_rate": 3.7077382393286534e-05,
+      "loss": 0.0059,
+      "step": 9160
+    },
+    {
+      "epoch": 6.66,
+      "grad_norm": 0.018028290942311287,
+      "learning_rate": 3.705720971516179e-05,
+      "loss": 0.0139,
+      "step": 9170
+    },
+    {
+      "epoch": 6.66,
+      "grad_norm": 14.395564079284668,
+      "learning_rate": 3.7037037037037037e-05,
+      "loss": 0.0075,
+      "step": 9180
+    },
+    {
+      "epoch": 6.67,
+      "grad_norm": 0.013322776183485985,
+      "learning_rate": 3.701686435891229e-05,
+      "loss": 0.0184,
+      "step": 9190
+    },
+    {
+      "epoch": 6.68,
+      "grad_norm": 6.382280349731445,
+      "learning_rate": 3.6996691680787545e-05,
+      "loss": 0.0219,
+      "step": 9200
+    },
+    {
+      "epoch": 6.69,
+      "grad_norm": 9.223612785339355,
+      "learning_rate": 3.697651900266279e-05,
+      "loss": 0.019,
+      "step": 9210
+    },
+    {
+      "epoch": 6.69,
+      "grad_norm": 0.005767362657934427,
+      "learning_rate": 3.695634632453805e-05,
+      "loss": 0.0104,
+      "step": 9220
+    },
+    {
+      "epoch": 6.7,
+      "grad_norm": 0.18802551925182343,
+      "learning_rate": 3.69361736464133e-05,
+      "loss": 0.0096,
+      "step": 9230
+    },
+    {
+      "epoch": 6.71,
+      "grad_norm": 0.18698835372924805,
+      "learning_rate": 3.691600096828855e-05,
+      "loss": 0.0207,
+      "step": 9240
+    },
+    {
+      "epoch": 6.72,
+      "grad_norm": 0.004097466357052326,
+      "learning_rate": 3.6895828290163804e-05,
+      "loss": 0.0165,
+      "step": 9250
+    },
+    {
+      "epoch": 6.72,
+      "grad_norm": 0.17292384803295135,
+      "learning_rate": 3.687565561203906e-05,
+      "loss": 0.0306,
+      "step": 9260
+    },
+    {
+      "epoch": 6.73,
+      "grad_norm": 0.30295267701148987,
+      "learning_rate": 3.6855482933914306e-05,
+      "loss": 0.01,
+      "step": 9270
+    },
+    {
+      "epoch": 6.74,
+      "grad_norm": 0.21782360970973969,
+      "learning_rate": 3.683531025578956e-05,
+      "loss": 0.0434,
+      "step": 9280
+    },
+    {
+      "epoch": 6.74,
+      "grad_norm": 14.22446346282959,
+      "learning_rate": 3.681513757766481e-05,
+      "loss": 0.0338,
+      "step": 9290
+    },
+    {
+      "epoch": 6.75,
+      "grad_norm": 0.8099685311317444,
+      "learning_rate": 3.679496489954006e-05,
+      "loss": 0.0099,
+      "step": 9300
+    },
+    {
+      "epoch": 6.76,
+      "grad_norm": 6.104836940765381,
+      "learning_rate": 3.677479222141532e-05,
+      "loss": 0.011,
+      "step": 9310
+    },
+    {
+      "epoch": 6.77,
+      "grad_norm": 0.00373910553753376,
+      "learning_rate": 3.675461954329057e-05,
+      "loss": 0.0215,
+      "step": 9320
+    },
+    {
+      "epoch": 6.77,
+      "grad_norm": 0.22883032262325287,
+      "learning_rate": 3.673444686516582e-05,
+      "loss": 0.0045,
+      "step": 9330
+    },
+    {
+      "epoch": 6.78,
+      "grad_norm": 3.783512592315674,
+      "learning_rate": 3.6714274187041074e-05,
+      "loss": 0.0276,
+      "step": 9340
+    },
+    {
+      "epoch": 6.79,
+      "grad_norm": 0.17120474576950073,
+      "learning_rate": 3.669410150891632e-05,
+      "loss": 0.022,
+      "step": 9350
+    },
+    {
+      "epoch": 6.79,
+      "grad_norm": 0.04094693809747696,
+      "learning_rate": 3.6673928830791576e-05,
+      "loss": 0.0047,
+      "step": 9360
+    },
+    {
+      "epoch": 6.8,
+      "grad_norm": 0.7232370376586914,
+      "learning_rate": 3.665375615266683e-05,
+      "loss": 0.0258,
+      "step": 9370
+    },
+    {
+      "epoch": 6.81,
+      "grad_norm": 16.711807250976562,
+      "learning_rate": 3.6633583474542085e-05,
+      "loss": 0.0124,
+      "step": 9380
+    },
+    {
+      "epoch": 6.82,
+      "grad_norm": 31.93790626525879,
+      "learning_rate": 3.661341079641733e-05,
+      "loss": 0.006,
+      "step": 9390
+    },
+    {
+      "epoch": 6.82,
+      "grad_norm": 6.95900297164917,
+      "learning_rate": 3.659323811829259e-05,
+      "loss": 0.0265,
+      "step": 9400
+    },
+    {
+      "epoch": 6.83,
+      "grad_norm": 5.2735161781311035,
+      "learning_rate": 3.6573065440167835e-05,
+      "loss": 0.0323,
+      "step": 9410
+    },
+    {
+      "epoch": 6.84,
+      "grad_norm": 0.699161171913147,
+      "learning_rate": 3.655289276204309e-05,
+      "loss": 0.0164,
+      "step": 9420
+    },
+    {
+      "epoch": 6.85,
+      "grad_norm": 16.414228439331055,
+      "learning_rate": 3.6532720083918344e-05,
+      "loss": 0.0145,
+      "step": 9430
+    },
+    {
+      "epoch": 6.85,
+      "grad_norm": 0.07426783442497253,
+      "learning_rate": 3.65125474057936e-05,
+      "loss": 0.0048,
+      "step": 9440
+    },
+    {
+      "epoch": 6.86,
+      "grad_norm": 0.013512199744582176,
+      "learning_rate": 3.6492374727668846e-05,
+      "loss": 0.0065,
+      "step": 9450
+    },
+    {
+      "epoch": 6.87,
+      "grad_norm": 0.6285327672958374,
+      "learning_rate": 3.64722020495441e-05,
+      "loss": 0.0416,
+      "step": 9460
+    },
+    {
+      "epoch": 6.87,
+      "grad_norm": 5.466975688934326,
+      "learning_rate": 3.645202937141935e-05,
+      "loss": 0.0109,
+      "step": 9470
+    },
+    {
+      "epoch": 6.88,
+      "grad_norm": 0.16771970689296722,
+      "learning_rate": 3.64318566932946e-05,
+      "loss": 0.0293,
+      "step": 9480
+    },
+    {
+      "epoch": 6.89,
+      "grad_norm": 12.002236366271973,
+      "learning_rate": 3.641168401516986e-05,
+      "loss": 0.0189,
+      "step": 9490
+    },
+    {
+      "epoch": 6.9,
+      "grad_norm": 0.3899655044078827,
+      "learning_rate": 3.639151133704511e-05,
+      "loss": 0.0376,
+      "step": 9500
+    },
+    {
+      "epoch": 6.9,
+      "grad_norm": 7.12745475769043,
+      "learning_rate": 3.637133865892036e-05,
+      "loss": 0.0093,
+      "step": 9510
+    },
+    {
+      "epoch": 6.91,
+      "grad_norm": 1.6025562286376953,
+      "learning_rate": 3.635116598079561e-05,
+      "loss": 0.0109,
+      "step": 9520
+    },
+    {
+      "epoch": 6.92,
+      "grad_norm": 0.007232175208628178,
+      "learning_rate": 3.633099330267086e-05,
+      "loss": 0.0116,
+      "step": 9530
+    },
+    {
+      "epoch": 6.93,
+      "grad_norm": 0.010008217766880989,
+      "learning_rate": 3.6310820624546116e-05,
+      "loss": 0.0203,
+      "step": 9540
+    },
+    {
+      "epoch": 6.93,
+      "grad_norm": 0.10533101111650467,
+      "learning_rate": 3.629064794642137e-05,
+      "loss": 0.0299,
+      "step": 9550
+    },
+    {
+      "epoch": 6.94,
+      "grad_norm": 0.09911059588193893,
+      "learning_rate": 3.6270475268296625e-05,
+      "loss": 0.0145,
+      "step": 9560
+    },
+    {
+      "epoch": 6.95,
+      "grad_norm": 2.048704147338867,
+      "learning_rate": 3.625030259017187e-05,
+      "loss": 0.0076,
+      "step": 9570
+    },
+    {
+      "epoch": 6.95,
+      "grad_norm": 0.05403584986925125,
+      "learning_rate": 3.623012991204712e-05,
+      "loss": 0.0086,
+      "step": 9580
+    },
+    {
+      "epoch": 6.96,
+      "grad_norm": 0.12622416019439697,
+      "learning_rate": 3.6209957233922375e-05,
+      "loss": 0.0226,
+      "step": 9590
+    },
+    {
+      "epoch": 6.97,
+      "grad_norm": 0.004092243034392595,
+      "learning_rate": 3.618978455579763e-05,
+      "loss": 0.0134,
+      "step": 9600
+    },
+    {
+      "epoch": 6.98,
+      "grad_norm": 17.107080459594727,
+      "learning_rate": 3.6169611877672884e-05,
+      "loss": 0.0195,
+      "step": 9610
+    },
+    {
+      "epoch": 6.98,
+      "grad_norm": 0.022195547819137573,
+      "learning_rate": 3.614943919954814e-05,
+      "loss": 0.0255,
+      "step": 9620
+    },
+    {
+      "epoch": 6.99,
+      "grad_norm": 0.04998145252466202,
+      "learning_rate": 3.6129266521423386e-05,
+      "loss": 0.006,
+      "step": 9630
+    },
+    {
+      "epoch": 7.0,
+      "grad_norm": 0.027100518345832825,
+      "learning_rate": 3.6109093843298634e-05,
+      "loss": 0.016,
+      "step": 9640
+    },
+    {
+      "epoch": 7.0,
+      "eval_accuracy": 0.9974588192585198,
+      "eval_f1": 0.9941486860665587,
+      "eval_loss": 0.006671547889709473,
+      "eval_precision": 0.994928369758444,
+      "eval_recall": 0.9933702234286907,
+      "eval_roc_auc": 0.9999722872260652,
+      "eval_runtime": 387.013,
+      "eval_samples_per_second": 227.765,
+      "eval_steps_per_second": 14.237,
+      "step": 9642
+    },
+    {
+      "epoch": 7.01,
+      "grad_norm": 0.15103967487812042,
+      "learning_rate": 3.608892116517389e-05,
+      "loss": 0.008,
+      "step": 9650
+    },
+    {
+      "epoch": 7.01,
+      "grad_norm": 1.6163809299468994,
+      "learning_rate": 3.606874848704914e-05,
+      "loss": 0.006,
+      "step": 9660
+    },
+    {
+      "epoch": 7.02,
+      "grad_norm": 0.4985108971595764,
+      "learning_rate": 3.60485758089244e-05,
+      "loss": 0.0162,
+      "step": 9670
+    },
+    {
+      "epoch": 7.03,
+      "grad_norm": 0.056671515107154846,
+      "learning_rate": 3.602840313079965e-05,
+      "loss": 0.0039,
+      "step": 9680
+    },
+    {
+      "epoch": 7.03,
+      "grad_norm": 12.355618476867676,
+      "learning_rate": 3.60082304526749e-05,
+      "loss": 0.0501,
+      "step": 9690
+    },
+    {
+      "epoch": 7.04,
+      "grad_norm": 11.992454528808594,
+      "learning_rate": 3.598805777455015e-05,
+      "loss": 0.045,
+      "step": 9700
+    },
+    {
+      "epoch": 7.05,
+      "grad_norm": 0.22388532757759094,
+      "learning_rate": 3.59678850964254e-05,
+      "loss": 0.0217,
+      "step": 9710
+    },
+    {
+      "epoch": 7.06,
+      "grad_norm": 1.1855632066726685,
+      "learning_rate": 3.5947712418300656e-05,
+      "loss": 0.0188,
+      "step": 9720
+    },
+    {
+      "epoch": 7.06,
+      "grad_norm": 0.43892866373062134,
+      "learning_rate": 3.592753974017591e-05,
+      "loss": 0.0112,
+      "step": 9730
+    },
+    {
+      "epoch": 7.07,
+      "grad_norm": 15.013214111328125,
+      "learning_rate": 3.5907367062051165e-05,
+      "loss": 0.0222,
+      "step": 9740
+    },
+    {
+      "epoch": 7.08,
+      "grad_norm": 1.0387877225875854,
+      "learning_rate": 3.588719438392641e-05,
+      "loss": 0.0233,
+      "step": 9750
+    },
+    {
+      "epoch": 7.09,
+      "grad_norm": 0.41390901803970337,
+      "learning_rate": 3.586702170580166e-05,
+      "loss": 0.0138,
+      "step": 9760
+    },
+    {
+      "epoch": 7.09,
+      "grad_norm": 0.1715419590473175,
+      "learning_rate": 3.5846849027676914e-05,
+      "loss": 0.0093,
+      "step": 9770
+    },
+    {
+      "epoch": 7.1,
+      "grad_norm": 0.01624373160302639,
+      "learning_rate": 3.582667634955217e-05,
+      "loss": 0.0065,
+      "step": 9780
+    },
+    {
+      "epoch": 7.11,
+      "grad_norm": 4.485403060913086,
+      "learning_rate": 3.5806503671427423e-05,
+      "loss": 0.0163,
+      "step": 9790
+    },
+    {
+      "epoch": 7.11,
+      "grad_norm": 0.5213087797164917,
+      "learning_rate": 3.578633099330268e-05,
+      "loss": 0.0152,
+      "step": 9800
+    },
+    {
+      "epoch": 7.12,
+      "grad_norm": 0.0917457863688469,
+      "learning_rate": 3.5766158315177926e-05,
+      "loss": 0.0043,
+      "step": 9810
+    },
+    {
+      "epoch": 7.13,
+      "grad_norm": 0.13996107876300812,
+      "learning_rate": 3.574598563705317e-05,
+      "loss": 0.0056,
+      "step": 9820
+    },
+    {
+      "epoch": 7.14,
+      "grad_norm": 0.2892821431159973,
+      "learning_rate": 3.572581295892843e-05,
+      "loss": 0.0249,
+      "step": 9830
+    },
+    {
+      "epoch": 7.14,
+      "grad_norm": 0.10906893014907837,
+      "learning_rate": 3.570564028080368e-05,
+      "loss": 0.0026,
+      "step": 9840
+    },
+    {
+      "epoch": 7.15,
+      "grad_norm": 0.19463422894477844,
+      "learning_rate": 3.568546760267894e-05,
+      "loss": 0.0109,
+      "step": 9850
+    },
+    {
+      "epoch": 7.16,
+      "grad_norm": 0.043046820908784866,
+      "learning_rate": 3.566529492455419e-05,
+      "loss": 0.0045,
+      "step": 9860
+    },
+    {
+      "epoch": 7.17,
+      "grad_norm": 0.0024864105507731438,
+      "learning_rate": 3.564512224642944e-05,
+      "loss": 0.0018,
+      "step": 9870
+    },
+    {
+      "epoch": 7.17,
+      "grad_norm": 0.0023697796277701855,
+      "learning_rate": 3.5624949568304687e-05,
+      "loss": 0.0214,
+      "step": 9880
+    },
+    {
+      "epoch": 7.18,
+      "grad_norm": 0.17014308273792267,
+      "learning_rate": 3.560477689017994e-05,
+      "loss": 0.0135,
+      "step": 9890
+    },
+    {
+      "epoch": 7.19,
+      "grad_norm": 0.06295622885227203,
+      "learning_rate": 3.5584604212055195e-05,
+      "loss": 0.0222,
+      "step": 9900
+    },
+    {
+      "epoch": 7.19,
+      "grad_norm": 0.06391363590955734,
+      "learning_rate": 3.556443153393045e-05,
+      "loss": 0.0054,
+      "step": 9910
+    },
+    {
+      "epoch": 7.2,
+      "grad_norm": 0.01831732876598835,
+      "learning_rate": 3.5544258855805704e-05,
+      "loss": 0.0092,
+      "step": 9920
+    },
+    {
+      "epoch": 7.21,
+      "grad_norm": 0.011373781599104404,
+      "learning_rate": 3.5524086177680945e-05,
+      "loss": 0.0033,
+      "step": 9930
+    },
+    {
+      "epoch": 7.22,
+      "grad_norm": 0.021700827404856682,
+      "learning_rate": 3.55039134995562e-05,
+      "loss": 0.042,
+      "step": 9940
+    },
+    {
+      "epoch": 7.22,
+      "grad_norm": 0.06618179380893707,
+      "learning_rate": 3.5483740821431454e-05,
+      "loss": 0.0129,
+      "step": 9950
+    },
+    {
+      "epoch": 7.23,
+      "grad_norm": 5.286431789398193,
+      "learning_rate": 3.546356814330671e-05,
+      "loss": 0.0244,
+      "step": 9960
+    },
+    {
+      "epoch": 7.24,
+      "grad_norm": 4.6873250007629395,
+      "learning_rate": 3.544339546518196e-05,
+      "loss": 0.0088,
+      "step": 9970
+    },
+    {
+      "epoch": 7.25,
+      "grad_norm": 0.43198829889297485,
+      "learning_rate": 3.542322278705721e-05,
+      "loss": 0.0122,
+      "step": 9980
+    },
+    {
+      "epoch": 7.25,
+      "grad_norm": 0.011568567715585232,
+      "learning_rate": 3.540305010893246e-05,
+      "loss": 0.0054,
+      "step": 9990
+    },
+    {
+      "epoch": 7.26,
+      "grad_norm": 0.04392976313829422,
+      "learning_rate": 3.538287743080771e-05,
+      "loss": 0.0326,
+      "step": 10000
+    },
+    {
+      "epoch": 7.27,
+      "grad_norm": 0.10357452183961868,
+      "learning_rate": 3.536270475268297e-05,
+      "loss": 0.0065,
+      "step": 10010
+    },
+    {
+      "epoch": 7.27,
+      "grad_norm": 1.8322253227233887,
+      "learning_rate": 3.534253207455822e-05,
+      "loss": 0.0231,
+      "step": 10020
+    },
+    {
+      "epoch": 7.28,
+      "grad_norm": 0.12756188213825226,
+      "learning_rate": 3.5322359396433476e-05,
+      "loss": 0.0055,
+      "step": 10030
+    },
+    {
+      "epoch": 7.29,
+      "grad_norm": 0.015612252056598663,
+      "learning_rate": 3.5302186718308724e-05,
+      "loss": 0.0117,
+      "step": 10040
+    },
+    {
+      "epoch": 7.3,
+      "grad_norm": 0.6343337297439575,
+      "learning_rate": 3.528201404018397e-05,
+      "loss": 0.0311,
+      "step": 10050
+    },
+    {
+      "epoch": 7.3,
+      "grad_norm": 0.020835332572460175,
+      "learning_rate": 3.5261841362059226e-05,
+      "loss": 0.008,
+      "step": 10060
+    },
+    {
+      "epoch": 7.31,
+      "grad_norm": 0.04921965301036835,
+      "learning_rate": 3.524166868393448e-05,
+      "loss": 0.0096,
+      "step": 10070
+    },
+    {
+      "epoch": 7.32,
+      "grad_norm": 0.031414203345775604,
+      "learning_rate": 3.5221496005809735e-05,
+      "loss": 0.0185,
+      "step": 10080
+    },
+    {
+      "epoch": 7.32,
+      "grad_norm": 0.04260152578353882,
+      "learning_rate": 3.520132332768499e-05,
+      "loss": 0.0082,
+      "step": 10090
+    },
+    {
+      "epoch": 7.33,
+      "grad_norm": 0.0031811215449124575,
+      "learning_rate": 3.518115064956024e-05,
+      "loss": 0.0101,
+      "step": 10100
+    },
+    {
+      "epoch": 7.34,
+      "grad_norm": 4.260556221008301,
+      "learning_rate": 3.5160977971435485e-05,
+      "loss": 0.0105,
+      "step": 10110
+    },
+    {
+      "epoch": 7.35,
+      "grad_norm": 0.007074627093970776,
+      "learning_rate": 3.514080529331074e-05,
+      "loss": 0.0194,
+      "step": 10120
+    },
+    {
+      "epoch": 7.35,
+      "grad_norm": 1.2988779544830322,
+      "learning_rate": 3.5120632615185994e-05,
+      "loss": 0.0295,
+      "step": 10130
+    },
+    {
+      "epoch": 7.36,
+      "grad_norm": 0.19852252304553986,
+      "learning_rate": 3.510045993706125e-05,
+      "loss": 0.0142,
+      "step": 10140
+    },
+    {
+      "epoch": 7.37,
+      "grad_norm": 0.9556043744087219,
+      "learning_rate": 3.50802872589365e-05,
+      "loss": 0.0343,
+      "step": 10150
+    },
+    {
+      "epoch": 7.38,
+      "grad_norm": 7.8187150955200195,
+      "learning_rate": 3.506011458081175e-05,
+      "loss": 0.0195,
+      "step": 10160
+    },
+    {
+      "epoch": 7.38,
+      "grad_norm": 0.00911070965230465,
+      "learning_rate": 3.5039941902687e-05,
+      "loss": 0.0091,
+      "step": 10170
+    },
+    {
+      "epoch": 7.39,
+      "grad_norm": 0.4240989089012146,
+      "learning_rate": 3.501976922456225e-05,
+      "loss": 0.0248,
+      "step": 10180
+    },
+    {
+      "epoch": 7.4,
+      "grad_norm": 0.3559081256389618,
+      "learning_rate": 3.499959654643751e-05,
+      "loss": 0.0246,
+      "step": 10190
+    },
+    {
+      "epoch": 7.4,
+      "grad_norm": 0.09215513616800308,
+      "learning_rate": 3.497942386831276e-05,
+      "loss": 0.0184,
+      "step": 10200
+    },
+    {
+      "epoch": 7.41,
+      "grad_norm": 1.6127543449401855,
+      "learning_rate": 3.4959251190188016e-05,
+      "loss": 0.027,
+      "step": 10210
+    },
+    {
+      "epoch": 7.42,
+      "grad_norm": 13.152901649475098,
+      "learning_rate": 3.4939078512063264e-05,
+      "loss": 0.0265,
+      "step": 10220
+    },
+    {
+      "epoch": 7.43,
+      "grad_norm": 0.5344623923301697,
+      "learning_rate": 3.491890583393851e-05,
+      "loss": 0.0159,
+      "step": 10230
+    },
+    {
+      "epoch": 7.43,
+      "grad_norm": 8.818399429321289,
+      "learning_rate": 3.4898733155813766e-05,
+      "loss": 0.0157,
+      "step": 10240
+    },
+    {
+      "epoch": 7.44,
+      "grad_norm": 1.7615970373153687,
+      "learning_rate": 3.487856047768902e-05,
+      "loss": 0.0187,
+      "step": 10250
+    },
+    {
+      "epoch": 7.45,
+      "grad_norm": 3.8972291946411133,
+      "learning_rate": 3.4858387799564275e-05,
+      "loss": 0.0107,
+      "step": 10260
+    },
+    {
+      "epoch": 7.46,
+      "grad_norm": 10.226646423339844,
+      "learning_rate": 3.483821512143952e-05,
+      "loss": 0.0086,
+      "step": 10270
+    },
+    {
+      "epoch": 7.46,
+      "grad_norm": 0.04844396561384201,
+      "learning_rate": 3.481804244331478e-05,
+      "loss": 0.0049,
+      "step": 10280
+    },
+    {
+      "epoch": 7.47,
+      "grad_norm": 1.5469908714294434,
+      "learning_rate": 3.4797869765190025e-05,
+      "loss": 0.0088,
+      "step": 10290
+    },
+    {
+      "epoch": 7.48,
+      "grad_norm": 0.038982976227998734,
+      "learning_rate": 3.477769708706528e-05,
+      "loss": 0.01,
+      "step": 10300
+    },
+    {
+      "epoch": 7.48,
+      "grad_norm": 0.0024217732716351748,
+      "learning_rate": 3.4757524408940534e-05,
+      "loss": 0.0012,
+      "step": 10310
+    },
+    {
+      "epoch": 7.49,
+      "grad_norm": 1.0054019689559937,
+      "learning_rate": 3.473735173081579e-05,
+      "loss": 0.0095,
+      "step": 10320
+    },
+    {
+      "epoch": 7.5,
+      "grad_norm": 22.222614288330078,
+      "learning_rate": 3.4717179052691036e-05,
+      "loss": 0.012,
+      "step": 10330
+    },
+    {
+      "epoch": 7.51,
+      "grad_norm": 2.256174087524414,
+      "learning_rate": 3.469700637456629e-05,
+      "loss": 0.014,
+      "step": 10340
+    },
+    {
+      "epoch": 7.51,
+      "grad_norm": 0.5855737924575806,
+      "learning_rate": 3.467683369644154e-05,
+      "loss": 0.0351,
+      "step": 10350
+    },
+    {
+      "epoch": 7.52,
+      "grad_norm": 0.840969443321228,
+      "learning_rate": 3.465666101831679e-05,
+      "loss": 0.0416,
+      "step": 10360
+    },
+    {
+      "epoch": 7.53,
+      "grad_norm": 0.04065997898578644,
+      "learning_rate": 3.463648834019205e-05,
+      "loss": 0.0251,
+      "step": 10370
+    },
+    {
+      "epoch": 7.54,
+      "grad_norm": 0.03526819124817848,
+      "learning_rate": 3.46163156620673e-05,
+      "loss": 0.0134,
+      "step": 10380
+    },
+    {
+      "epoch": 7.54,
+      "grad_norm": 2.399366855621338,
+      "learning_rate": 3.459614298394255e-05,
+      "loss": 0.0036,
+      "step": 10390
+    },
+    {
+      "epoch": 7.55,
+      "grad_norm": 0.23753774166107178,
+      "learning_rate": 3.4575970305817804e-05,
+      "loss": 0.0066,
+      "step": 10400
+    },
+    {
+      "epoch": 7.56,
+      "grad_norm": 0.16471104323863983,
+      "learning_rate": 3.455579762769305e-05,
+      "loss": 0.0039,
+      "step": 10410
+    },
+    {
+      "epoch": 7.56,
+      "grad_norm": 0.004009116906672716,
+      "learning_rate": 3.4535624949568306e-05,
+      "loss": 0.006,
+      "step": 10420
+    },
+    {
+      "epoch": 7.57,
+      "grad_norm": 0.005728223826736212,
+      "learning_rate": 3.451545227144356e-05,
+      "loss": 0.0044,
+      "step": 10430
+    },
+    {
+      "epoch": 7.58,
+      "grad_norm": 0.005147533491253853,
+      "learning_rate": 3.4495279593318815e-05,
+      "loss": 0.0068,
+      "step": 10440
+    },
+    {
+      "epoch": 7.59,
+      "grad_norm": 0.09112564474344254,
+      "learning_rate": 3.447510691519406e-05,
+      "loss": 0.0226,
+      "step": 10450
+    },
+    {
+      "epoch": 7.59,
+      "grad_norm": 0.001015618909150362,
+      "learning_rate": 3.445493423706932e-05,
+      "loss": 0.0263,
+      "step": 10460
+    },
+    {
+      "epoch": 7.6,
+      "grad_norm": 0.006133379880338907,
+      "learning_rate": 3.4434761558944564e-05,
+      "loss": 0.0196,
+      "step": 10470
+    },
+    {
+      "epoch": 7.61,
+      "grad_norm": 7.626928329467773,
+      "learning_rate": 3.441458888081982e-05,
+      "loss": 0.0122,
+      "step": 10480
+    },
+    {
+      "epoch": 7.62,
+      "grad_norm": 23.610570907592773,
+      "learning_rate": 3.439441620269507e-05,
+      "loss": 0.0201,
+      "step": 10490
+    },
+    {
+      "epoch": 7.62,
+      "grad_norm": 2.0458810329437256,
+      "learning_rate": 3.437424352457032e-05,
+      "loss": 0.0054,
+      "step": 10500
+    },
+    {
+      "epoch": 7.63,
+      "grad_norm": 4.983373641967773,
+      "learning_rate": 3.4354070846445576e-05,
+      "loss": 0.021,
+      "step": 10510
+    },
+    {
+      "epoch": 7.64,
+      "grad_norm": 10.006043434143066,
+      "learning_rate": 3.433389816832083e-05,
+      "loss": 0.0497,
+      "step": 10520
+    },
+    {
+      "epoch": 7.64,
+      "grad_norm": 0.06582402437925339,
+      "learning_rate": 3.431372549019608e-05,
+      "loss": 0.0136,
+      "step": 10530
+    },
+    {
+      "epoch": 7.65,
+      "grad_norm": 0.01676173321902752,
+      "learning_rate": 3.429355281207133e-05,
+      "loss": 0.0308,
+      "step": 10540
+    },
+    {
+      "epoch": 7.66,
+      "grad_norm": 0.03430357575416565,
+      "learning_rate": 3.427338013394659e-05,
+      "loss": 0.0029,
+      "step": 10550
+    },
+    {
+      "epoch": 7.67,
+      "grad_norm": 0.19467763602733612,
+      "learning_rate": 3.4253207455821834e-05,
+      "loss": 0.0191,
+      "step": 10560
+    },
+    {
+      "epoch": 7.67,
+      "grad_norm": 0.07039833068847656,
+      "learning_rate": 3.423303477769709e-05,
+      "loss": 0.008,
+      "step": 10570
+    },
+    {
+      "epoch": 7.68,
+      "grad_norm": 0.030894028022885323,
+      "learning_rate": 3.421286209957234e-05,
+      "loss": 0.0082,
+      "step": 10580
+    },
+    {
+      "epoch": 7.69,
+      "grad_norm": 0.009277657605707645,
+      "learning_rate": 3.419268942144759e-05,
+      "loss": 0.0032,
+      "step": 10590
+    },
+    {
+      "epoch": 7.7,
+      "grad_norm": 0.03260354697704315,
+      "learning_rate": 3.4172516743322845e-05,
+      "loss": 0.0145,
+      "step": 10600
+    },
+    {
+      "epoch": 7.7,
+      "grad_norm": 0.033388249576091766,
+      "learning_rate": 3.41523440651981e-05,
+      "loss": 0.0006,
+      "step": 10610
+    },
+    {
+      "epoch": 7.71,
+      "grad_norm": 0.0583641491830349,
+      "learning_rate": 3.413217138707335e-05,
+      "loss": 0.0084,
+      "step": 10620
+    },
+    {
+      "epoch": 7.72,
+      "grad_norm": 0.005505802109837532,
+      "learning_rate": 3.41119987089486e-05,
+      "loss": 0.0125,
+      "step": 10630
+    },
+    {
+      "epoch": 7.72,
+      "grad_norm": 25.258424758911133,
+      "learning_rate": 3.4091826030823856e-05,
+      "loss": 0.0247,
+      "step": 10640
+    },
+    {
+      "epoch": 7.73,
+      "grad_norm": 0.10212866216897964,
+      "learning_rate": 3.4071653352699104e-05,
+      "loss": 0.0076,
+      "step": 10650
+    },
+    {
+      "epoch": 7.74,
+      "grad_norm": 0.07812552899122238,
+      "learning_rate": 3.405148067457436e-05,
+      "loss": 0.0088,
+      "step": 10660
+    },
+    {
+      "epoch": 7.75,
+      "grad_norm": 0.017161810770630836,
+      "learning_rate": 3.403130799644961e-05,
+      "loss": 0.0131,
+      "step": 10670
+    },
+    {
+      "epoch": 7.75,
+      "grad_norm": 0.01328431349247694,
+      "learning_rate": 3.401113531832486e-05,
+      "loss": 0.0102,
+      "step": 10680
+    },
+    {
+      "epoch": 7.76,
+      "grad_norm": 0.005470677278935909,
+      "learning_rate": 3.3990962640200115e-05,
+      "loss": 0.001,
+      "step": 10690
+    },
+    {
+      "epoch": 7.77,
+      "grad_norm": 0.23102551698684692,
+      "learning_rate": 3.397078996207537e-05,
+      "loss": 0.024,
+      "step": 10700
+    },
+    {
+      "epoch": 7.77,
+      "grad_norm": 1.5232537984848022,
+      "learning_rate": 3.395061728395062e-05,
+      "loss": 0.0082,
+      "step": 10710
+    },
+    {
+      "epoch": 7.78,
+      "grad_norm": 0.37568143010139465,
+      "learning_rate": 3.393044460582587e-05,
+      "loss": 0.007,
+      "step": 10720
+    },
+    {
+      "epoch": 7.79,
+      "grad_norm": 0.005128229968249798,
+      "learning_rate": 3.391027192770112e-05,
+      "loss": 0.0153,
+      "step": 10730
+    },
+    {
+      "epoch": 7.8,
+      "grad_norm": 1.4863814115524292,
+      "learning_rate": 3.3890099249576374e-05,
+      "loss": 0.0072,
+      "step": 10740
+    },
+    {
+      "epoch": 7.8,
+      "grad_norm": 0.0014930274337530136,
+      "learning_rate": 3.386992657145163e-05,
+      "loss": 0.0154,
+      "step": 10750
+    },
+    {
+      "epoch": 7.81,
+      "grad_norm": 11.875150680541992,
+      "learning_rate": 3.384975389332688e-05,
+      "loss": 0.0089,
+      "step": 10760
+    },
+    {
+      "epoch": 7.82,
+      "grad_norm": 0.04604551941156387,
+      "learning_rate": 3.382958121520213e-05,
+      "loss": 0.0028,
+      "step": 10770
+    },
+    {
+      "epoch": 7.83,
+      "grad_norm": 1.3366645574569702,
+      "learning_rate": 3.3809408537077385e-05,
+      "loss": 0.015,
+      "step": 10780
+    },
+    {
+      "epoch": 7.83,
+      "grad_norm": 12.528473854064941,
+      "learning_rate": 3.378923585895263e-05,
+      "loss": 0.0173,
+      "step": 10790
+    },
+    {
+      "epoch": 7.84,
+      "grad_norm": 0.0017001566011458635,
+      "learning_rate": 3.376906318082789e-05,
+      "loss": 0.0087,
+      "step": 10800
+    },
+    {
+      "epoch": 7.85,
+      "grad_norm": 0.09803108125925064,
+      "learning_rate": 3.374889050270314e-05,
+      "loss": 0.0018,
+      "step": 10810
+    },
+    {
+      "epoch": 7.85,
+      "grad_norm": 0.22022363543510437,
+      "learning_rate": 3.3728717824578396e-05,
+      "loss": 0.0147,
+      "step": 10820
+    },
+    {
+      "epoch": 7.86,
+      "grad_norm": 0.052362002432346344,
+      "learning_rate": 3.3708545146453644e-05,
+      "loss": 0.0129,
+      "step": 10830
+    },
+    {
+      "epoch": 7.87,
+      "grad_norm": 0.003308740444481373,
+      "learning_rate": 3.36883724683289e-05,
+      "loss": 0.013,
+      "step": 10840
+    },
+    {
+      "epoch": 7.88,
+      "grad_norm": 0.019734688103199005,
+      "learning_rate": 3.3668199790204146e-05,
+      "loss": 0.0055,
+      "step": 10850
+    },
+    {
+      "epoch": 7.88,
+      "grad_norm": 4.014479160308838,
+      "learning_rate": 3.36480271120794e-05,
+      "loss": 0.0038,
+      "step": 10860
+    },
+    {
+      "epoch": 7.89,
+      "grad_norm": 0.002026587026193738,
+      "learning_rate": 3.3627854433954655e-05,
+      "loss": 0.0242,
+      "step": 10870
+    },
+    {
+      "epoch": 7.9,
+      "grad_norm": 1.0237423181533813,
+      "learning_rate": 3.360768175582991e-05,
+      "loss": 0.0299,
+      "step": 10880
+    },
+    {
+      "epoch": 7.91,
+      "grad_norm": 0.0015221534995362163,
+      "learning_rate": 3.358750907770516e-05,
+      "loss": 0.0091,
+      "step": 10890
+    },
+    {
+      "epoch": 7.91,
+      "grad_norm": 0.021977558732032776,
+      "learning_rate": 3.356733639958041e-05,
+      "loss": 0.0288,
+      "step": 10900
+    },
+    {
+      "epoch": 7.92,
+      "grad_norm": 12.71220588684082,
+      "learning_rate": 3.354716372145566e-05,
+      "loss": 0.0345,
+      "step": 10910
+    },
+    {
+      "epoch": 7.93,
+      "grad_norm": 0.050337791442871094,
+      "learning_rate": 3.3526991043330914e-05,
+      "loss": 0.0022,
+      "step": 10920
+    },
+    {
+      "epoch": 7.93,
+      "grad_norm": 11.188690185546875,
+      "learning_rate": 3.350681836520617e-05,
+      "loss": 0.0095,
+      "step": 10930
+    },
+    {
+      "epoch": 7.94,
+      "grad_norm": 0.6816732883453369,
+      "learning_rate": 3.348664568708142e-05,
+      "loss": 0.0196,
+      "step": 10940
+    },
+    {
+      "epoch": 7.95,
+      "grad_norm": 2.0615453720092773,
+      "learning_rate": 3.346647300895667e-05,
+      "loss": 0.0081,
+      "step": 10950
+    },
+    {
+      "epoch": 7.96,
+      "grad_norm": 6.260568618774414,
+      "learning_rate": 3.3446300330831925e-05,
+      "loss": 0.0155,
+      "step": 10960
+    },
+    {
+      "epoch": 7.96,
+      "grad_norm": 0.013660675846040249,
+      "learning_rate": 3.342612765270717e-05,
+      "loss": 0.0061,
+      "step": 10970
+    },
+    {
+      "epoch": 7.97,
+      "grad_norm": 16.132661819458008,
+      "learning_rate": 3.340595497458243e-05,
+      "loss": 0.0249,
+      "step": 10980
+    },
+    {
+      "epoch": 7.98,
+      "grad_norm": 0.0729353055357933,
+      "learning_rate": 3.338578229645768e-05,
+      "loss": 0.0021,
+      "step": 10990
+    },
+    {
+      "epoch": 7.99,
+      "grad_norm": 2.324553966522217,
+      "learning_rate": 3.336560961833293e-05,
+      "loss": 0.0111,
+      "step": 11000
+    },
+    {
+      "epoch": 7.99,
+      "grad_norm": 0.08047934621572495,
+      "learning_rate": 3.3345436940208184e-05,
+      "loss": 0.0029,
+      "step": 11010
+    },
+    {
+      "epoch": 8.0,
+      "grad_norm": 0.0027567476499825716,
+      "learning_rate": 3.332526426208343e-05,
+      "loss": 0.0054,
+      "step": 11020
+    },
+    {
+      "epoch": 8.0,
+      "eval_accuracy": 0.997799155964968,
+      "eval_f1": 0.9949188056574123,
+      "eval_loss": 0.005811151582747698,
+      "eval_precision": 0.9983704793944491,
+      "eval_recall": 0.9914909166840676,
+      "eval_roc_auc": 0.9999673239643861,
+      "eval_runtime": 387.519,
+      "eval_samples_per_second": 227.468,
+      "eval_steps_per_second": 14.219,
+      "step": 11020
+    },
+    {
+      "epoch": 8.01,
+      "grad_norm": 0.16821260750293732,
+      "learning_rate": 3.3305091583958686e-05,
+      "loss": 0.0269,
+      "step": 11030
+    },
+    {
+      "epoch": 8.01,
+      "grad_norm": 1.4832972288131714,
+      "learning_rate": 3.328491890583394e-05,
+      "loss": 0.0073,
+      "step": 11040
+    },
+    {
+      "epoch": 8.02,
+      "grad_norm": 2.289891481399536,
+      "learning_rate": 3.3264746227709195e-05,
+      "loss": 0.0065,
+      "step": 11050
+    },
+    {
+      "epoch": 8.03,
+      "grad_norm": 0.000776061206124723,
+      "learning_rate": 3.324457354958444e-05,
+      "loss": 0.0117,
+      "step": 11060
+    },
+    {
+      "epoch": 8.04,
+      "grad_norm": 0.016198599711060524,
+      "learning_rate": 3.32244008714597e-05,
+      "loss": 0.0202,
+      "step": 11070
+    },
+    {
+      "epoch": 8.04,
+      "grad_norm": 2.553449869155884,
+      "learning_rate": 3.3204228193334945e-05,
+      "loss": 0.0037,
+      "step": 11080
+    },
+    {
+      "epoch": 8.05,
+      "grad_norm": 0.055435068905353546,
+      "learning_rate": 3.31840555152102e-05,
+      "loss": 0.0372,
+      "step": 11090
+    },
+    {
+      "epoch": 8.06,
+      "grad_norm": 13.744915008544922,
+      "learning_rate": 3.3163882837085453e-05,
+      "loss": 0.031,
+      "step": 11100
+    },
+    {
+      "epoch": 8.07,
+      "grad_norm": 42.31868362426758,
+      "learning_rate": 3.314371015896071e-05,
+      "loss": 0.0065,
+      "step": 11110
+    },
+    {
+      "epoch": 8.07,
+      "grad_norm": 0.022761182859539986,
+      "learning_rate": 3.3123537480835956e-05,
+      "loss": 0.0048,
+      "step": 11120
+    },
+    {
+      "epoch": 8.08,
+      "grad_norm": 0.06478892266750336,
+      "learning_rate": 3.310336480271121e-05,
+      "loss": 0.0137,
+      "step": 11130
+    },
+    {
+      "epoch": 8.09,
+      "grad_norm": 3.4087636470794678,
+      "learning_rate": 3.308319212458646e-05,
+      "loss": 0.0112,
+      "step": 11140
+    },
+    {
+      "epoch": 8.09,
+      "grad_norm": 0.024201085790991783,
+      "learning_rate": 3.306301944646171e-05,
+      "loss": 0.0251,
+      "step": 11150
+    },
+    {
+      "epoch": 8.1,
+      "grad_norm": 0.003428951371461153,
+      "learning_rate": 3.304284676833697e-05,
+      "loss": 0.0184,
+      "step": 11160
+    },
+    {
+      "epoch": 8.11,
+      "grad_norm": 0.10663584619760513,
+      "learning_rate": 3.302267409021222e-05,
+      "loss": 0.019,
+      "step": 11170
+    },
+    {
+      "epoch": 8.12,
+      "grad_norm": 0.07480733841657639,
+      "learning_rate": 3.300250141208747e-05,
+      "loss": 0.0172,
+      "step": 11180
+    },
+    {
+      "epoch": 8.12,
+      "grad_norm": 0.0026008952409029007,
+      "learning_rate": 3.298232873396272e-05,
+      "loss": 0.0243,
+      "step": 11190
+    },
+    {
+      "epoch": 8.13,
+      "grad_norm": 30.776805877685547,
+      "learning_rate": 3.296215605583797e-05,
+      "loss": 0.0103,
+      "step": 11200
+    },
+    {
+      "epoch": 8.14,
+      "grad_norm": 0.0013422481715679169,
+      "learning_rate": 3.2941983377713225e-05,
+      "loss": 0.0033,
+      "step": 11210
+    },
+    {
+      "epoch": 8.15,
+      "grad_norm": 0.008392914198338985,
+      "learning_rate": 3.292181069958848e-05,
+      "loss": 0.0094,
+      "step": 11220
+    },
+    {
+      "epoch": 8.15,
+      "grad_norm": 0.03945827856659889,
+      "learning_rate": 3.2901638021463734e-05,
+      "loss": 0.0032,
+      "step": 11230
+    },
+    {
+      "epoch": 8.16,
+      "grad_norm": 0.002708829240873456,
+      "learning_rate": 3.288146534333898e-05,
+      "loss": 0.0006,
+      "step": 11240
+    },
+    {
+      "epoch": 8.17,
+      "grad_norm": 0.02194291725754738,
+      "learning_rate": 3.286129266521423e-05,
+      "loss": 0.0101,
+      "step": 11250
+    },
+    {
+      "epoch": 8.17,
+      "grad_norm": 5.216897487640381,
+      "learning_rate": 3.2841119987089484e-05,
+      "loss": 0.0136,
+      "step": 11260
+    },
+    {
+      "epoch": 8.18,
+      "grad_norm": 0.004165074788033962,
+      "learning_rate": 3.282094730896474e-05,
+      "loss": 0.0101,
+      "step": 11270
+    },
+    {
+      "epoch": 8.19,
+      "grad_norm": 1.777703881263733,
+      "learning_rate": 3.280077463083999e-05,
+      "loss": 0.0139,
+      "step": 11280
+    },
+    {
+      "epoch": 8.2,
+      "grad_norm": 0.010176170617341995,
+      "learning_rate": 3.278060195271525e-05,
+      "loss": 0.0141,
+      "step": 11290
+    },
+    {
+      "epoch": 8.2,
+      "grad_norm": 0.01507497113198042,
+      "learning_rate": 3.2760429274590495e-05,
+      "loss": 0.0156,
+      "step": 11300
+    },
+    {
+      "epoch": 8.21,
+      "grad_norm": 0.09720058739185333,
+      "learning_rate": 3.274025659646574e-05,
+      "loss": 0.0088,
+      "step": 11310
+    },
+    {
+      "epoch": 8.22,
+      "grad_norm": 3.414900302886963,
+      "learning_rate": 3.2720083918341e-05,
+      "loss": 0.023,
+      "step": 11320
+    },
+    {
+      "epoch": 8.23,
+      "grad_norm": 0.5592748522758484,
+      "learning_rate": 3.269991124021625e-05,
+      "loss": 0.0056,
+      "step": 11330
+    },
+    {
+      "epoch": 8.23,
+      "grad_norm": 16.837783813476562,
+      "learning_rate": 3.2679738562091506e-05,
+      "loss": 0.0319,
+      "step": 11340
+    },
+    {
+      "epoch": 8.24,
+      "grad_norm": 0.012670857831835747,
+      "learning_rate": 3.265956588396676e-05,
+      "loss": 0.0026,
+      "step": 11350
+    },
+    {
+      "epoch": 8.25,
+      "grad_norm": 0.002103676088154316,
+      "learning_rate": 3.263939320584201e-05,
+      "loss": 0.0131,
+      "step": 11360
+    },
+    {
+      "epoch": 8.25,
+      "grad_norm": 0.10566865652799606,
+      "learning_rate": 3.2619220527717256e-05,
+      "loss": 0.0063,
+      "step": 11370
+    },
+    {
+      "epoch": 8.26,
+      "grad_norm": 27.73931121826172,
+      "learning_rate": 3.259904784959251e-05,
+      "loss": 0.0048,
+      "step": 11380
+    },
+    {
+      "epoch": 8.27,
+      "grad_norm": 0.009989196434617043,
+      "learning_rate": 3.2578875171467765e-05,
+      "loss": 0.0119,
+      "step": 11390
+    },
+    {
+      "epoch": 8.28,
+      "grad_norm": 0.07179460674524307,
+      "learning_rate": 3.255870249334302e-05,
+      "loss": 0.0158,
+      "step": 11400
+    },
+    {
+      "epoch": 8.28,
+      "grad_norm": 0.017826635390520096,
+      "learning_rate": 3.2538529815218274e-05,
+      "loss": 0.014,
+      "step": 11410
+    },
+    {
+      "epoch": 8.29,
+      "grad_norm": 0.00853447150439024,
+      "learning_rate": 3.251835713709352e-05,
+      "loss": 0.0163,
+      "step": 11420
+    },
+    {
+      "epoch": 8.3,
+      "grad_norm": 0.39563441276550293,
+      "learning_rate": 3.249818445896877e-05,
+      "loss": 0.0111,
+      "step": 11430
+    },
+    {
+      "epoch": 8.3,
+      "grad_norm": 0.03939186409115791,
+      "learning_rate": 3.2478011780844024e-05,
+      "loss": 0.0224,
+      "step": 11440
+    },
+    {
+      "epoch": 8.31,
+      "grad_norm": 0.8431787490844727,
+      "learning_rate": 3.245783910271928e-05,
+      "loss": 0.0266,
+      "step": 11450
+    },
+    {
+      "epoch": 8.32,
+      "grad_norm": 0.07643716782331467,
+      "learning_rate": 3.243766642459453e-05,
+      "loss": 0.0124,
+      "step": 11460
+    },
+    {
+      "epoch": 8.33,
+      "grad_norm": 0.042288776487112045,
+      "learning_rate": 3.241749374646979e-05,
+      "loss": 0.0184,
+      "step": 11470
+    },
+    {
+      "epoch": 8.33,
+      "grad_norm": 1.1074130535125732,
+      "learning_rate": 3.2397321068345035e-05,
+      "loss": 0.0248,
+      "step": 11480
+    },
+    {
+      "epoch": 8.34,
+      "grad_norm": 0.23461972177028656,
+      "learning_rate": 3.237714839022028e-05,
+      "loss": 0.0118,
+      "step": 11490
+    },
+    {
+      "epoch": 8.35,
+      "grad_norm": 0.008372629061341286,
+      "learning_rate": 3.235697571209554e-05,
+      "loss": 0.0153,
+      "step": 11500
+    },
+    {
+      "epoch": 8.36,
+      "grad_norm": 0.011258935555815697,
+      "learning_rate": 3.233680303397079e-05,
+      "loss": 0.0105,
+      "step": 11510
+    },
+    {
+      "epoch": 8.36,
+      "grad_norm": 0.16763925552368164,
+      "learning_rate": 3.2316630355846046e-05,
+      "loss": 0.0119,
+      "step": 11520
+    },
+    {
+      "epoch": 8.37,
+      "grad_norm": 0.014371280558407307,
+      "learning_rate": 3.22964576777213e-05,
+      "loss": 0.0054,
+      "step": 11530
+    },
+    {
+      "epoch": 8.38,
+      "grad_norm": 19.372220993041992,
+      "learning_rate": 3.227628499959655e-05,
+      "loss": 0.0063,
+      "step": 11540
+    },
+    {
+      "epoch": 8.38,
+      "grad_norm": 6.232364177703857,
+      "learning_rate": 3.2256112321471796e-05,
+      "loss": 0.0048,
+      "step": 11550
+    },
+    {
+      "epoch": 8.39,
+      "grad_norm": 0.2103765457868576,
+      "learning_rate": 3.223593964334705e-05,
+      "loss": 0.0084,
+      "step": 11560
+    },
+    {
+      "epoch": 8.4,
+      "grad_norm": 0.4541609585285187,
+      "learning_rate": 3.2215766965222305e-05,
+      "loss": 0.0097,
+      "step": 11570
+    },
+    {
+      "epoch": 8.41,
+      "grad_norm": 0.05270016938447952,
+      "learning_rate": 3.219559428709756e-05,
+      "loss": 0.0023,
+      "step": 11580
+    },
+    {
+      "epoch": 8.41,
+      "grad_norm": 0.0025290350895375013,
+      "learning_rate": 3.2175421608972814e-05,
+      "loss": 0.0154,
+      "step": 11590
+    },
+    {
+      "epoch": 8.42,
+      "grad_norm": 0.0017237365245819092,
+      "learning_rate": 3.215524893084806e-05,
+      "loss": 0.0301,
+      "step": 11600
+    },
+    {
+      "epoch": 8.43,
+      "grad_norm": 0.15763328969478607,
+      "learning_rate": 3.213507625272331e-05,
+      "loss": 0.0099,
+      "step": 11610
+    },
+    {
+      "epoch": 8.44,
+      "grad_norm": 1.8129161596298218,
+      "learning_rate": 3.2114903574598564e-05,
+      "loss": 0.0114,
+      "step": 11620
+    },
+    {
+      "epoch": 8.44,
+      "grad_norm": 0.22685779631137848,
+      "learning_rate": 3.209473089647382e-05,
+      "loss": 0.0089,
+      "step": 11630
+    },
+    {
+      "epoch": 8.45,
+      "grad_norm": 0.04477900266647339,
+      "learning_rate": 3.207455821834907e-05,
+      "loss": 0.0156,
+      "step": 11640
+    },
+    {
+      "epoch": 8.46,
+      "grad_norm": 0.43626195192337036,
+      "learning_rate": 3.205438554022433e-05,
+      "loss": 0.0023,
+      "step": 11650
+    },
+    {
+      "epoch": 8.46,
+      "grad_norm": 19.21158218383789,
+      "learning_rate": 3.2034212862099575e-05,
+      "loss": 0.023,
+      "step": 11660
+    },
+    {
+      "epoch": 8.47,
+      "grad_norm": 1.7743921279907227,
+      "learning_rate": 3.201404018397482e-05,
+      "loss": 0.0072,
+      "step": 11670
+    },
+    {
+      "epoch": 8.48,
+      "grad_norm": 0.4966103434562683,
+      "learning_rate": 3.199386750585008e-05,
+      "loss": 0.0053,
+      "step": 11680
+    },
+    {
+      "epoch": 8.49,
+      "grad_norm": 0.0037831587251275778,
+      "learning_rate": 3.197369482772533e-05,
+      "loss": 0.015,
+      "step": 11690
+    },
+    {
+      "epoch": 8.49,
+      "grad_norm": 28.9018611907959,
+      "learning_rate": 3.1953522149600586e-05,
+      "loss": 0.0163,
+      "step": 11700
+    },
+    {
+      "epoch": 8.5,
+      "grad_norm": 0.0005793919553980231,
+      "learning_rate": 3.1933349471475834e-05,
+      "loss": 0.0044,
+      "step": 11710
+    },
+    {
+      "epoch": 8.51,
+      "grad_norm": 24.365251541137695,
+      "learning_rate": 3.191317679335109e-05,
+      "loss": 0.0094,
+      "step": 11720
+    },
+    {
+      "epoch": 8.52,
+      "grad_norm": 0.0010113256284967065,
+      "learning_rate": 3.1893004115226336e-05,
+      "loss": 0.0197,
+      "step": 11730
+    },
+    {
+      "epoch": 8.52,
+      "grad_norm": 0.08384249359369278,
+      "learning_rate": 3.187283143710159e-05,
+      "loss": 0.0065,
+      "step": 11740
+    },
+    {
+      "epoch": 8.53,
+      "grad_norm": 0.10842813551425934,
+      "learning_rate": 3.1852658758976845e-05,
+      "loss": 0.003,
+      "step": 11750
+    },
+    {
+      "epoch": 8.54,
+      "grad_norm": 1.8701107501983643,
+      "learning_rate": 3.18324860808521e-05,
+      "loss": 0.0206,
+      "step": 11760
+    },
+    {
+      "epoch": 8.54,
+      "grad_norm": 7.070690631866455,
+      "learning_rate": 3.181231340272735e-05,
+      "loss": 0.0176,
+      "step": 11770
+    },
+    {
+      "epoch": 8.55,
+      "grad_norm": 0.00601642020046711,
+      "learning_rate": 3.17921407246026e-05,
+      "loss": 0.0218,
+      "step": 11780
+    },
+    {
+      "epoch": 8.56,
+      "grad_norm": 2.2268316745758057,
+      "learning_rate": 3.177196804647785e-05,
+      "loss": 0.0018,
+      "step": 11790
+    },
+    {
+      "epoch": 8.57,
+      "grad_norm": 6.753377437591553,
+      "learning_rate": 3.1751795368353103e-05,
+      "loss": 0.0041,
+      "step": 11800
+    },
+    {
+      "epoch": 8.57,
+      "grad_norm": 9.824979782104492,
+      "learning_rate": 3.173162269022836e-05,
+      "loss": 0.0103,
+      "step": 11810
+    },
+    {
+      "epoch": 8.58,
+      "grad_norm": 0.2134261578321457,
+      "learning_rate": 3.171145001210361e-05,
+      "loss": 0.0044,
+      "step": 11820
+    },
+    {
+      "epoch": 8.59,
+      "grad_norm": 0.22093412280082703,
+      "learning_rate": 3.169127733397886e-05,
+      "loss": 0.0092,
+      "step": 11830
+    },
+    {
+      "epoch": 8.6,
+      "grad_norm": 4.783883571624756,
+      "learning_rate": 3.1671104655854115e-05,
+      "loss": 0.0188,
+      "step": 11840
+    },
+    {
+      "epoch": 8.6,
+      "grad_norm": 0.42982345819473267,
+      "learning_rate": 3.165093197772936e-05,
+      "loss": 0.0481,
+      "step": 11850
+    },
+    {
+      "epoch": 8.61,
+      "grad_norm": 23.18255043029785,
+      "learning_rate": 3.163075929960462e-05,
+      "loss": 0.0046,
+      "step": 11860
+    },
+    {
+      "epoch": 8.62,
+      "grad_norm": 6.138206481933594,
+      "learning_rate": 3.161058662147987e-05,
+      "loss": 0.018,
+      "step": 11870
+    },
+    {
+      "epoch": 8.62,
+      "grad_norm": 0.3428427278995514,
+      "learning_rate": 3.1590413943355126e-05,
+      "loss": 0.0143,
+      "step": 11880
+    },
+    {
+      "epoch": 8.63,
+      "grad_norm": 0.7229692339897156,
+      "learning_rate": 3.157024126523037e-05,
+      "loss": 0.0054,
+      "step": 11890
+    },
+    {
+      "epoch": 8.64,
+      "grad_norm": 0.011998855508863926,
+      "learning_rate": 3.155006858710563e-05,
+      "loss": 0.009,
+      "step": 11900
+    },
+    {
+      "epoch": 8.65,
+      "grad_norm": 0.0009840866550803185,
+      "learning_rate": 3.1529895908980875e-05,
+      "loss": 0.0202,
+      "step": 11910
+    },
+    {
+      "epoch": 8.65,
+      "grad_norm": 0.0013845226494595408,
+      "learning_rate": 3.150972323085613e-05,
+      "loss": 0.0088,
+      "step": 11920
+    },
+    {
+      "epoch": 8.66,
+      "grad_norm": 14.614180564880371,
+      "learning_rate": 3.1489550552731384e-05,
+      "loss": 0.0148,
+      "step": 11930
+    },
+    {
+      "epoch": 8.67,
+      "grad_norm": 27.177169799804688,
+      "learning_rate": 3.146937787460664e-05,
+      "loss": 0.0041,
+      "step": 11940
+    },
+    {
+      "epoch": 8.68,
+      "grad_norm": 0.06443148106336594,
+      "learning_rate": 3.1449205196481887e-05,
+      "loss": 0.0041,
+      "step": 11950
+    },
+    {
+      "epoch": 8.68,
+      "grad_norm": 0.003651421284303069,
+      "learning_rate": 3.142903251835714e-05,
+      "loss": 0.0079,
+      "step": 11960
+    },
+    {
+      "epoch": 8.69,
+      "grad_norm": 0.004412221256643534,
+      "learning_rate": 3.140885984023239e-05,
+      "loss": 0.0093,
+      "step": 11970
+    },
+    {
+      "epoch": 8.7,
+      "grad_norm": 0.5715083479881287,
+      "learning_rate": 3.138868716210764e-05,
+      "loss": 0.0066,
+      "step": 11980
+    },
+    {
+      "epoch": 8.7,
+      "grad_norm": 0.03544891998171806,
+      "learning_rate": 3.13685144839829e-05,
+      "loss": 0.0164,
+      "step": 11990
+    },
+    {
+      "epoch": 8.71,
+      "grad_norm": 0.0518750362098217,
+      "learning_rate": 3.1348341805858145e-05,
+      "loss": 0.0086,
+      "step": 12000
+    },
+    {
+      "epoch": 8.72,
+      "grad_norm": 1.9587420225143433,
+      "learning_rate": 3.13281691277334e-05,
+      "loss": 0.0286,
+      "step": 12010
+    },
+    {
+      "epoch": 8.73,
+      "grad_norm": 0.004830994643270969,
+      "learning_rate": 3.1307996449608654e-05,
+      "loss": 0.0157,
+      "step": 12020
+    },
+    {
+      "epoch": 8.73,
+      "grad_norm": 0.05058489367365837,
+      "learning_rate": 3.12878237714839e-05,
+      "loss": 0.0061,
+      "step": 12030
+    },
+    {
+      "epoch": 8.74,
+      "grad_norm": 0.13220369815826416,
+      "learning_rate": 3.1267651093359156e-05,
+      "loss": 0.0234,
+      "step": 12040
+    },
+    {
+      "epoch": 8.75,
+      "grad_norm": 0.6112781763076782,
+      "learning_rate": 3.124747841523441e-05,
+      "loss": 0.0067,
+      "step": 12050
+    },
+    {
+      "epoch": 8.75,
+      "grad_norm": 0.3486891984939575,
+      "learning_rate": 3.122730573710966e-05,
+      "loss": 0.0012,
+      "step": 12060
+    },
+    {
+      "epoch": 8.76,
+      "grad_norm": 0.04058101028203964,
+      "learning_rate": 3.120713305898491e-05,
+      "loss": 0.0242,
+      "step": 12070
+    },
+    {
+      "epoch": 8.77,
+      "grad_norm": 0.05213288590312004,
+      "learning_rate": 3.118696038086016e-05,
+      "loss": 0.0286,
+      "step": 12080
+    },
+    {
+      "epoch": 8.78,
+      "grad_norm": 0.053065940737724304,
+      "learning_rate": 3.1166787702735415e-05,
+      "loss": 0.0022,
+      "step": 12090
+    },
+    {
+      "epoch": 8.78,
+      "grad_norm": 10.85990047454834,
+      "learning_rate": 3.114661502461067e-05,
+      "loss": 0.0105,
+      "step": 12100
+    },
+    {
+      "epoch": 8.79,
+      "grad_norm": 0.009069069288671017,
+      "learning_rate": 3.1126442346485924e-05,
+      "loss": 0.0024,
+      "step": 12110
+    },
+    {
+      "epoch": 8.8,
+      "grad_norm": 0.0020809718407690525,
+      "learning_rate": 3.110626966836117e-05,
+      "loss": 0.0047,
+      "step": 12120
+    },
+    {
+      "epoch": 8.81,
+      "grad_norm": 0.2555220425128937,
+      "learning_rate": 3.1086096990236426e-05,
+      "loss": 0.0118,
+      "step": 12130
+    },
+    {
+      "epoch": 8.81,
+      "grad_norm": 0.05784722417593002,
+      "learning_rate": 3.1065924312111674e-05,
+      "loss": 0.0023,
+      "step": 12140
+    },
+    {
+      "epoch": 8.82,
+      "grad_norm": 0.035568490624427795,
+      "learning_rate": 3.104575163398693e-05,
+      "loss": 0.0307,
+      "step": 12150
+    },
+    {
+      "epoch": 8.83,
+      "grad_norm": 0.08019623160362244,
+      "learning_rate": 3.102557895586218e-05,
+      "loss": 0.017,
+      "step": 12160
+    },
+    {
+      "epoch": 8.83,
+      "grad_norm": 5.698986530303955,
+      "learning_rate": 3.100540627773744e-05,
+      "loss": 0.0115,
+      "step": 12170
+    },
+    {
+      "epoch": 8.84,
+      "grad_norm": 0.02579701505601406,
+      "learning_rate": 3.0985233599612685e-05,
+      "loss": 0.0105,
+      "step": 12180
+    },
+    {
+      "epoch": 8.85,
+      "grad_norm": 0.0029477495700120926,
+      "learning_rate": 3.096506092148794e-05,
+      "loss": 0.0027,
+      "step": 12190
+    },
+    {
+      "epoch": 8.86,
+      "grad_norm": 0.09123173356056213,
+      "learning_rate": 3.094488824336319e-05,
+      "loss": 0.0118,
+      "step": 12200
+    },
+    {
+      "epoch": 8.86,
+      "grad_norm": 0.13705745339393616,
+      "learning_rate": 3.092471556523844e-05,
+      "loss": 0.0191,
+      "step": 12210
+    },
+    {
+      "epoch": 8.87,
+      "grad_norm": 0.0693451315164566,
+      "learning_rate": 3.0904542887113696e-05,
+      "loss": 0.0141,
+      "step": 12220
+    },
+    {
+      "epoch": 8.88,
+      "grad_norm": 9.002408027648926,
+      "learning_rate": 3.0884370208988944e-05,
+      "loss": 0.014,
+      "step": 12230
+    },
+    {
+      "epoch": 8.89,
+      "grad_norm": 0.02591705694794655,
+      "learning_rate": 3.08641975308642e-05,
+      "loss": 0.0028,
+      "step": 12240
+    },
+    {
+      "epoch": 8.89,
+      "grad_norm": 0.47011977434158325,
+      "learning_rate": 3.084402485273945e-05,
+      "loss": 0.0043,
+      "step": 12250
+    },
+    {
+      "epoch": 8.9,
+      "grad_norm": 0.09690820425748825,
+      "learning_rate": 3.08238521746147e-05,
+      "loss": 0.0018,
+      "step": 12260
+    },
+    {
+      "epoch": 8.91,
+      "grad_norm": 0.21148131787776947,
+      "learning_rate": 3.0803679496489955e-05,
+      "loss": 0.0007,
+      "step": 12270
+    },
+    {
+      "epoch": 8.91,
+      "grad_norm": 0.0025974249001592398,
+      "learning_rate": 3.078350681836521e-05,
+      "loss": 0.0174,
+      "step": 12280
+    },
+    {
+      "epoch": 8.92,
+      "grad_norm": 0.03931077942252159,
+      "learning_rate": 3.076333414024046e-05,
+      "loss": 0.0039,
+      "step": 12290
+    },
+    {
+      "epoch": 8.93,
+      "grad_norm": 7.300892353057861,
+      "learning_rate": 3.074316146211571e-05,
+      "loss": 0.008,
+      "step": 12300
+    },
+    {
+      "epoch": 8.94,
+      "grad_norm": 0.007567646913230419,
+      "learning_rate": 3.0722988783990966e-05,
+      "loss": 0.0123,
+      "step": 12310
+    },
+    {
+      "epoch": 8.94,
+      "grad_norm": 38.95884323120117,
+      "learning_rate": 3.0702816105866214e-05,
+      "loss": 0.0059,
+      "step": 12320
+    },
+    {
+      "epoch": 8.95,
+      "grad_norm": 0.0021059729624539614,
+      "learning_rate": 3.068264342774147e-05,
+      "loss": 0.0362,
+      "step": 12330
+    },
+    {
+      "epoch": 8.96,
+      "grad_norm": 0.04417746886610985,
+      "learning_rate": 3.066247074961672e-05,
+      "loss": 0.0035,
+      "step": 12340
+    },
+    {
+      "epoch": 8.97,
+      "grad_norm": 0.7246804237365723,
+      "learning_rate": 3.064229807149197e-05,
+      "loss": 0.0395,
+      "step": 12350
+    },
+    {
+      "epoch": 8.97,
+      "grad_norm": 3.860435724258423,
+      "learning_rate": 3.0622125393367225e-05,
+      "loss": 0.0045,
+      "step": 12360
+    },
+    {
+      "epoch": 8.98,
+      "grad_norm": 0.16252577304840088,
+      "learning_rate": 3.060195271524248e-05,
+      "loss": 0.0015,
+      "step": 12370
+    },
+    {
+      "epoch": 8.99,
+      "grad_norm": 4.022401332855225,
+      "learning_rate": 3.058178003711773e-05,
+      "loss": 0.0086,
+      "step": 12380
+    },
+    {
+      "epoch": 8.99,
+      "grad_norm": 0.0879194512963295,
+      "learning_rate": 3.056160735899298e-05,
+      "loss": 0.0237,
+      "step": 12390
+    },
+    {
+      "epoch": 9.0,
+      "eval_accuracy": 0.9975495757135726,
+      "eval_f1": 0.994333980378784,
+      "eval_loss": 0.006345591973513365,
+      "eval_precision": 0.9993145629020352,
+      "eval_recall": 0.9894027980789308,
+      "eval_roc_auc": 0.9999658938912749,
+      "eval_runtime": 387.414,
+      "eval_samples_per_second": 227.529,
+      "eval_steps_per_second": 14.223,
+      "step": 12397
+    },
+    {
+      "epoch": 9.0,
+      "grad_norm": 0.1917732059955597,
+      "learning_rate": 3.0541434680868236e-05,
+      "loss": 0.0101,
+      "step": 12400
+    },
+    {
+      "epoch": 9.01,
+      "grad_norm": 0.033312708139419556,
+      "learning_rate": 3.0521262002743484e-05,
+      "loss": 0.0021,
+      "step": 12410
+    },
+    {
+      "epoch": 9.02,
+      "grad_norm": 0.021154897287487984,
+      "learning_rate": 3.0501089324618738e-05,
+      "loss": 0.0073,
+      "step": 12420
+    },
+    {
+      "epoch": 9.02,
+      "grad_norm": 0.05219145119190216,
+      "learning_rate": 3.048091664649399e-05,
+      "loss": 0.0023,
+      "step": 12430
+    },
+    {
+      "epoch": 9.03,
+      "grad_norm": 0.0031368627678602934,
+      "learning_rate": 3.0460743968369244e-05,
+      "loss": 0.0215,
+      "step": 12440
+    },
+    {
+      "epoch": 9.04,
+      "grad_norm": 0.0007551907910965383,
+      "learning_rate": 3.0440571290244495e-05,
+      "loss": 0.0236,
+      "step": 12450
+    },
+    {
+      "epoch": 9.05,
+      "grad_norm": 0.19673331081867218,
+      "learning_rate": 3.0420398612119742e-05,
+      "loss": 0.0061,
+      "step": 12460
+    },
+    {
+      "epoch": 9.05,
+      "grad_norm": 0.003755633719265461,
+      "learning_rate": 3.0400225933994997e-05,
+      "loss": 0.0052,
+      "step": 12470
+    },
+    {
+      "epoch": 9.06,
+      "grad_norm": 0.8004059195518494,
+      "learning_rate": 3.038005325587025e-05,
+      "loss": 0.0337,
+      "step": 12480
+    },
+    {
+      "epoch": 9.07,
+      "grad_norm": 15.695157051086426,
+      "learning_rate": 3.0359880577745502e-05,
+      "loss": 0.0083,
+      "step": 12490
+    },
+    {
+      "epoch": 9.07,
+      "grad_norm": 6.738131046295166,
+      "learning_rate": 3.0339707899620757e-05,
+      "loss": 0.0119,
+      "step": 12500
+    },
+    {
+      "epoch": 9.08,
+      "grad_norm": 0.03489381819963455,
+      "learning_rate": 3.0319535221496008e-05,
+      "loss": 0.0037,
+      "step": 12510
+    },
+    {
+      "epoch": 9.09,
+      "grad_norm": 0.353909969329834,
+      "learning_rate": 3.0299362543371256e-05,
+      "loss": 0.0159,
+      "step": 12520
+    },
+    {
+      "epoch": 9.1,
+      "grad_norm": 0.3446331024169922,
+      "learning_rate": 3.027918986524651e-05,
+      "loss": 0.0058,
+      "step": 12530
+    },
+    {
+      "epoch": 9.1,
+      "grad_norm": 1.5575968027114868,
+      "learning_rate": 3.0259017187121764e-05,
+      "loss": 0.0184,
+      "step": 12540
+    },
+    {
+      "epoch": 9.11,
+      "grad_norm": 0.026695841923356056,
+      "learning_rate": 3.0238844508997016e-05,
+      "loss": 0.0014,
+      "step": 12550
+    },
+    {
+      "epoch": 9.12,
+      "grad_norm": 0.006738999392837286,
+      "learning_rate": 3.021867183087227e-05,
+      "loss": 0.0122,
+      "step": 12560
+    },
+    {
+      "epoch": 9.13,
+      "grad_norm": 1.860974907875061,
+      "learning_rate": 3.019849915274752e-05,
+      "loss": 0.0052,
+      "step": 12570
+    },
+    {
+      "epoch": 9.13,
+      "grad_norm": 0.29735755920410156,
+      "learning_rate": 3.017832647462277e-05,
+      "loss": 0.0071,
+      "step": 12580
+    },
+    {
+      "epoch": 9.14,
+      "grad_norm": 0.08206071704626083,
+      "learning_rate": 3.0158153796498023e-05,
+      "loss": 0.0148,
+      "step": 12590
+    },
+    {
+      "epoch": 9.15,
+      "grad_norm": 1.398136854171753,
+      "learning_rate": 3.0137981118373278e-05,
+      "loss": 0.0049,
+      "step": 12600
+    },
+    {
+      "epoch": 9.15,
+      "grad_norm": 0.42362233996391296,
+      "learning_rate": 3.011780844024853e-05,
+      "loss": 0.0098,
+      "step": 12610
+    },
+    {
+      "epoch": 9.16,
+      "grad_norm": 0.08972881734371185,
+      "learning_rate": 3.0097635762123783e-05,
+      "loss": 0.0269,
+      "step": 12620
+    },
+    {
+      "epoch": 9.17,
+      "grad_norm": 6.95635461807251,
+      "learning_rate": 3.0077463083999034e-05,
+      "loss": 0.0204,
+      "step": 12630
+    },
+    {
+      "epoch": 9.18,
+      "grad_norm": 44.25766372680664,
+      "learning_rate": 3.0057290405874282e-05,
+      "loss": 0.0097,
+      "step": 12640
+    },
+    {
+      "epoch": 9.18,
+      "grad_norm": 0.01075151190161705,
+      "learning_rate": 3.0037117727749536e-05,
+      "loss": 0.0029,
+      "step": 12650
+    },
+    {
+      "epoch": 9.19,
+      "grad_norm": 2.1552395820617676,
+      "learning_rate": 3.001694504962479e-05,
+      "loss": 0.0188,
+      "step": 12660
+    },
+    {
+      "epoch": 9.2,
+      "grad_norm": 18.27303123474121,
+      "learning_rate": 2.9996772371500042e-05,
+      "loss": 0.0159,
+      "step": 12670
+    },
+    {
+      "epoch": 9.21,
+      "grad_norm": 0.42549946904182434,
+      "learning_rate": 2.9976599693375296e-05,
+      "loss": 0.0112,
+      "step": 12680
+    },
+    {
+      "epoch": 9.21,
+      "grad_norm": 0.16241183876991272,
+      "learning_rate": 2.9956427015250548e-05,
+      "loss": 0.0057,
+      "step": 12690
+    },
+    {
+      "epoch": 9.22,
+      "grad_norm": 5.789508819580078,
+      "learning_rate": 2.9936254337125795e-05,
+      "loss": 0.0115,
+      "step": 12700
+    },
+    {
+      "epoch": 9.23,
+      "grad_norm": 0.16165970265865326,
+      "learning_rate": 2.991608165900105e-05,
+      "loss": 0.0083,
+      "step": 12710
+    },
+    {
+      "epoch": 9.23,
+      "grad_norm": 0.010464129038155079,
+      "learning_rate": 2.9895908980876304e-05,
+      "loss": 0.001,
+      "step": 12720
+    },
+    {
+      "epoch": 9.24,
+      "grad_norm": 0.06300076097249985,
+      "learning_rate": 2.9875736302751555e-05,
+      "loss": 0.0098,
+      "step": 12730
+    },
+    {
+      "epoch": 9.25,
+      "grad_norm": 7.2553324699401855,
+      "learning_rate": 2.985556362462681e-05,
+      "loss": 0.0231,
+      "step": 12740
+    },
+    {
+      "epoch": 9.26,
+      "grad_norm": 0.001716342754662037,
+      "learning_rate": 2.9835390946502057e-05,
+      "loss": 0.0113,
+      "step": 12750
+    },
+    {
+      "epoch": 9.26,
+      "grad_norm": 3.9360616207122803,
+      "learning_rate": 2.981521826837731e-05,
+      "loss": 0.0246,
+      "step": 12760
+    },
+    {
+      "epoch": 9.27,
+      "grad_norm": 0.5541887283325195,
+      "learning_rate": 2.9795045590252563e-05,
+      "loss": 0.0042,
+      "step": 12770
+    },
+    {
+      "epoch": 9.28,
+      "grad_norm": 0.0037244977429509163,
+      "learning_rate": 2.9774872912127817e-05,
+      "loss": 0.0046,
+      "step": 12780
+    },
+    {
+      "epoch": 9.28,
+      "grad_norm": 0.008533765561878681,
+      "learning_rate": 2.975470023400307e-05,
+      "loss": 0.0028,
+      "step": 12790
+    },
+    {
+      "epoch": 9.29,
+      "grad_norm": 0.005289971828460693,
+      "learning_rate": 2.9734527555878323e-05,
+      "loss": 0.0027,
+      "step": 12800
+    },
+    {
+      "epoch": 9.3,
+      "grad_norm": 0.0009858094854280353,
+      "learning_rate": 2.971435487775357e-05,
+      "loss": 0.003,
+      "step": 12810
+    },
+    {
+      "epoch": 9.31,
+      "grad_norm": 0.00409921258687973,
+      "learning_rate": 2.9694182199628822e-05,
+      "loss": 0.0021,
+      "step": 12820
+    },
+    {
+      "epoch": 9.31,
+      "grad_norm": 0.0007549124420620501,
+      "learning_rate": 2.9674009521504076e-05,
+      "loss": 0.0054,
+      "step": 12830
+    },
+    {
+      "epoch": 9.32,
+      "grad_norm": 0.10273166000843048,
+      "learning_rate": 2.965383684337933e-05,
+      "loss": 0.0049,
+      "step": 12840
+    },
+    {
+      "epoch": 9.33,
+      "grad_norm": 0.1592969447374344,
+      "learning_rate": 2.9633664165254582e-05,
+      "loss": 0.0006,
+      "step": 12850
+    },
+    {
+      "epoch": 9.34,
+      "grad_norm": 0.002123448997735977,
+      "learning_rate": 2.9613491487129836e-05,
+      "loss": 0.0016,
+      "step": 12860
+    },
+    {
+      "epoch": 9.34,
+      "grad_norm": 5.097808361053467,
+      "learning_rate": 2.9593318809005084e-05,
+      "loss": 0.0085,
+      "step": 12870
+    },
+    {
+      "epoch": 9.35,
+      "grad_norm": 0.41327035427093506,
+      "learning_rate": 2.9573146130880335e-05,
+      "loss": 0.0014,
+      "step": 12880
+    },
+    {
+      "epoch": 9.36,
+      "grad_norm": 0.0021131192333996296,
+      "learning_rate": 2.955297345275559e-05,
+      "loss": 0.0093,
+      "step": 12890
+    },
+    {
+      "epoch": 9.36,
+      "grad_norm": 0.36018940806388855,
+      "learning_rate": 2.953280077463084e-05,
+      "loss": 0.0058,
+      "step": 12900
+    },
+    {
+      "epoch": 9.37,
+      "grad_norm": 0.0010047269752249122,
+      "learning_rate": 2.9512628096506095e-05,
+      "loss": 0.0024,
+      "step": 12910
+    },
+    {
+      "epoch": 9.38,
+      "grad_norm": 0.13996048271656036,
+      "learning_rate": 2.949245541838135e-05,
+      "loss": 0.0095,
+      "step": 12920
+    },
+    {
+      "epoch": 9.39,
+      "grad_norm": 0.0009951372630894184,
+      "learning_rate": 2.9472282740256597e-05,
+      "loss": 0.0017,
+      "step": 12930
+    },
+    {
+      "epoch": 9.39,
+      "grad_norm": 0.0007663732394576073,
+      "learning_rate": 2.9452110062131848e-05,
+      "loss": 0.0431,
+      "step": 12940
+    },
+    {
+      "epoch": 9.4,
+      "grad_norm": 0.0016060832422226667,
+      "learning_rate": 2.9431937384007103e-05,
+      "loss": 0.0026,
+      "step": 12950
+    },
+    {
+      "epoch": 9.41,
+      "grad_norm": 0.09539427608251572,
+      "learning_rate": 2.9411764705882354e-05,
+      "loss": 0.0056,
+      "step": 12960
+    },
+    {
+      "epoch": 9.42,
+      "grad_norm": 0.5911301970481873,
+      "learning_rate": 2.9391592027757608e-05,
+      "loss": 0.0062,
+      "step": 12970
+    },
+    {
+      "epoch": 9.42,
+      "grad_norm": 0.12249241769313812,
+      "learning_rate": 2.9371419349632856e-05,
+      "loss": 0.0049,
+      "step": 12980
+    },
+    {
+      "epoch": 9.43,
+      "grad_norm": 0.05083388090133667,
+      "learning_rate": 2.935124667150811e-05,
+      "loss": 0.0016,
+      "step": 12990
+    },
+    {
+      "epoch": 9.44,
+      "grad_norm": 0.004344270098954439,
+      "learning_rate": 2.933107399338336e-05,
+      "loss": 0.01,
+      "step": 13000
+    },
+    {
+      "epoch": 9.44,
+      "grad_norm": 0.003572734771296382,
+      "learning_rate": 2.9310901315258616e-05,
+      "loss": 0.0003,
+      "step": 13010
+    },
+    {
+      "epoch": 9.45,
+      "grad_norm": 2.558148145675659,
+      "learning_rate": 2.9290728637133867e-05,
+      "loss": 0.0101,
+      "step": 13020
+    },
+    {
+      "epoch": 9.46,
+      "grad_norm": 0.03010840341448784,
+      "learning_rate": 2.927055595900912e-05,
+      "loss": 0.0191,
+      "step": 13030
+    },
+    {
+      "epoch": 9.47,
+      "grad_norm": 0.05662524327635765,
+      "learning_rate": 2.925038328088437e-05,
+      "loss": 0.0144,
+      "step": 13040
+    },
+    {
+      "epoch": 9.47,
+      "grad_norm": 0.00602317601442337,
+      "learning_rate": 2.9230210602759624e-05,
+      "loss": 0.0081,
+      "step": 13050
+    },
+    {
+      "epoch": 9.48,
+      "grad_norm": 0.061507437378168106,
+      "learning_rate": 2.9210037924634875e-05,
+      "loss": 0.0129,
+      "step": 13060
+    },
+    {
+      "epoch": 9.49,
+      "grad_norm": 0.0022451505064964294,
+      "learning_rate": 2.918986524651013e-05,
+      "loss": 0.0063,
+      "step": 13070
+    },
+    {
+      "epoch": 9.5,
+      "grad_norm": 8.346963882446289,
+      "learning_rate": 2.916969256838538e-05,
+      "loss": 0.0078,
+      "step": 13080
+    },
+    {
+      "epoch": 9.5,
+      "grad_norm": 0.0015550401294603944,
+      "learning_rate": 2.9149519890260635e-05,
+      "loss": 0.0249,
+      "step": 13090
+    },
+    {
+      "epoch": 9.51,
+      "grad_norm": 0.03434576839208603,
+      "learning_rate": 2.9129347212135882e-05,
+      "loss": 0.0215,
+      "step": 13100
+    },
+    {
+      "epoch": 9.52,
+      "grad_norm": 24.404510498046875,
+      "learning_rate": 2.9109174534011137e-05,
+      "loss": 0.0174,
+      "step": 13110
+    },
+    {
+      "epoch": 9.52,
+      "grad_norm": 0.008947799913585186,
+      "learning_rate": 2.9089001855886388e-05,
+      "loss": 0.0039,
+      "step": 13120
+    },
+    {
+      "epoch": 9.53,
+      "grad_norm": 0.039463091641664505,
+      "learning_rate": 2.9068829177761642e-05,
+      "loss": 0.0051,
+      "step": 13130
+    },
+    {
+      "epoch": 9.54,
+      "grad_norm": 0.052592430263757706,
+      "learning_rate": 2.9048656499636893e-05,
+      "loss": 0.0052,
+      "step": 13140
+    },
+    {
+      "epoch": 9.55,
+      "grad_norm": 0.010807220824062824,
+      "learning_rate": 2.9028483821512148e-05,
+      "loss": 0.0013,
+      "step": 13150
+    },
+    {
+      "epoch": 9.55,
+      "grad_norm": 0.003055947832763195,
+      "learning_rate": 2.9008311143387396e-05,
+      "loss": 0.0137,
+      "step": 13160
+    },
+    {
+      "epoch": 9.56,
+      "grad_norm": 0.23203665018081665,
+      "learning_rate": 2.898813846526265e-05,
+      "loss": 0.0062,
+      "step": 13170
+    },
+    {
+      "epoch": 9.57,
+      "grad_norm": 0.012158505618572235,
+      "learning_rate": 2.89679657871379e-05,
+      "loss": 0.0162,
+      "step": 13180
+    },
+    {
+      "epoch": 9.58,
+      "grad_norm": 0.08189103752374649,
+      "learning_rate": 2.8947793109013156e-05,
+      "loss": 0.0207,
+      "step": 13190
+    },
+    {
+      "epoch": 9.58,
+      "grad_norm": 12.099898338317871,
+      "learning_rate": 2.8927620430888407e-05,
+      "loss": 0.0071,
+      "step": 13200
+    },
+    {
+      "epoch": 9.59,
+      "grad_norm": 0.006215301342308521,
+      "learning_rate": 2.8907447752763654e-05,
+      "loss": 0.008,
+      "step": 13210
+    },
+    {
+      "epoch": 9.6,
+      "grad_norm": 0.14043578505516052,
+      "learning_rate": 2.888727507463891e-05,
+      "loss": 0.005,
+      "step": 13220
+    },
+    {
+      "epoch": 9.6,
+      "grad_norm": 0.04191066324710846,
+      "learning_rate": 2.8867102396514163e-05,
+      "loss": 0.0046,
+      "step": 13230
+    },
+    {
+      "epoch": 9.61,
+      "grad_norm": 0.051841963082551956,
+      "learning_rate": 2.8846929718389414e-05,
+      "loss": 0.0039,
+      "step": 13240
+    },
+    {
+      "epoch": 9.62,
+      "grad_norm": 0.01983380690217018,
+      "learning_rate": 2.882675704026467e-05,
+      "loss": 0.0218,
+      "step": 13250
+    },
+    {
+      "epoch": 9.63,
+      "grad_norm": 0.001248432556167245,
+      "learning_rate": 2.880658436213992e-05,
+      "loss": 0.0089,
+      "step": 13260
+    },
+    {
+      "epoch": 9.63,
+      "grad_norm": 0.013916940428316593,
+      "learning_rate": 2.8786411684015168e-05,
+      "loss": 0.0033,
+      "step": 13270
+    },
+    {
+      "epoch": 9.64,
+      "grad_norm": 0.06915153563022614,
+      "learning_rate": 2.8766239005890422e-05,
+      "loss": 0.0101,
+      "step": 13280
+    },
+    {
+      "epoch": 9.65,
+      "grad_norm": 0.0015154919819906354,
+      "learning_rate": 2.8746066327765677e-05,
+      "loss": 0.0228,
+      "step": 13290
+    },
+    {
+      "epoch": 9.66,
+      "grad_norm": 0.0033606714569032192,
+      "learning_rate": 2.8725893649640928e-05,
+      "loss": 0.0106,
+      "step": 13300
+    },
+    {
+      "epoch": 9.66,
+      "grad_norm": 0.001192206982523203,
+      "learning_rate": 2.8705720971516182e-05,
+      "loss": 0.0038,
+      "step": 13310
+    },
+    {
+      "epoch": 9.67,
+      "grad_norm": 0.37659505009651184,
+      "learning_rate": 2.8685548293391433e-05,
+      "loss": 0.0237,
+      "step": 13320
+    },
+    {
+      "epoch": 9.68,
+      "grad_norm": 0.08516032248735428,
+      "learning_rate": 2.866537561526668e-05,
+      "loss": 0.0068,
+      "step": 13330
+    },
+    {
+      "epoch": 9.68,
+      "grad_norm": 0.053294021636247635,
+      "learning_rate": 2.8645202937141935e-05,
+      "loss": 0.0104,
+      "step": 13340
+    },
+    {
+      "epoch": 9.69,
+      "grad_norm": 0.0005857631331309676,
+      "learning_rate": 2.862503025901719e-05,
+      "loss": 0.027,
+      "step": 13350
+    },
+    {
+      "epoch": 9.7,
+      "grad_norm": 0.0786575973033905,
+      "learning_rate": 2.860485758089244e-05,
+      "loss": 0.0324,
+      "step": 13360
+    },
+    {
+      "epoch": 9.71,
+      "grad_norm": 0.013472789898514748,
+      "learning_rate": 2.8584684902767695e-05,
+      "loss": 0.0157,
+      "step": 13370
+    },
+    {
+      "epoch": 9.71,
+      "grad_norm": 0.027604950591921806,
+      "learning_rate": 2.8564512224642946e-05,
+      "loss": 0.008,
+      "step": 13380
+    },
+    {
+      "epoch": 9.72,
+      "grad_norm": 0.16292813420295715,
+      "learning_rate": 2.8544339546518194e-05,
+      "loss": 0.022,
+      "step": 13390
+    },
+    {
+      "epoch": 9.73,
+      "grad_norm": 0.2821474075317383,
+      "learning_rate": 2.852416686839345e-05,
+      "loss": 0.0104,
+      "step": 13400
+    },
+    {
+      "epoch": 9.74,
+      "grad_norm": 16.801767349243164,
+      "learning_rate": 2.85039941902687e-05,
+      "loss": 0.0101,
+      "step": 13410
+    },
+    {
+      "epoch": 9.74,
+      "grad_norm": 0.1447453647851944,
+      "learning_rate": 2.8483821512143954e-05,
+      "loss": 0.007,
+      "step": 13420
+    },
+    {
+      "epoch": 9.75,
+      "grad_norm": 0.9015370011329651,
+      "learning_rate": 2.846364883401921e-05,
+      "loss": 0.0026,
+      "step": 13430
+    },
+    {
+      "epoch": 9.76,
+      "grad_norm": 0.09575998038053513,
+      "learning_rate": 2.844347615589446e-05,
+      "loss": 0.0046,
+      "step": 13440
+    },
+    {
+      "epoch": 9.76,
+      "grad_norm": 0.8580997586250305,
+      "learning_rate": 2.8423303477769707e-05,
+      "loss": 0.0104,
+      "step": 13450
+    },
+    {
+      "epoch": 9.77,
+      "grad_norm": 0.0004182391567155719,
+      "learning_rate": 2.8403130799644962e-05,
+      "loss": 0.0054,
+      "step": 13460
+    },
+    {
+      "epoch": 9.78,
+      "grad_norm": 0.0007015741430222988,
+      "learning_rate": 2.8382958121520213e-05,
+      "loss": 0.0103,
+      "step": 13470
+    },
+    {
+      "epoch": 9.79,
+      "grad_norm": 0.1236376240849495,
+      "learning_rate": 2.8362785443395467e-05,
+      "loss": 0.0066,
+      "step": 13480
+    },
+    {
+      "epoch": 9.79,
+      "grad_norm": 0.005294484551995993,
+      "learning_rate": 2.8342612765270722e-05,
+      "loss": 0.0114,
+      "step": 13490
+    },
+    {
+      "epoch": 9.8,
+      "grad_norm": 4.20367431640625,
+      "learning_rate": 2.832244008714597e-05,
+      "loss": 0.0082,
+      "step": 13500
+    },
+    {
+      "epoch": 9.81,
+      "grad_norm": 11.44412612915039,
+      "learning_rate": 2.830226740902122e-05,
+      "loss": 0.0145,
+      "step": 13510
+    },
+    {
+      "epoch": 9.81,
+      "grad_norm": 0.02178305573761463,
+      "learning_rate": 2.8282094730896475e-05,
+      "loss": 0.0038,
+      "step": 13520
+    },
+    {
+      "epoch": 9.82,
+      "grad_norm": 0.04771624505519867,
+      "learning_rate": 2.8261922052771726e-05,
+      "loss": 0.0023,
+      "step": 13530
+    },
+    {
+      "epoch": 9.83,
+      "grad_norm": 0.006732448935508728,
+      "learning_rate": 2.824174937464698e-05,
+      "loss": 0.0058,
+      "step": 13540
+    },
+    {
+      "epoch": 9.84,
+      "grad_norm": 26.35951805114746,
+      "learning_rate": 2.8221576696522235e-05,
+      "loss": 0.0109,
+      "step": 13550
+    },
+    {
+      "epoch": 9.84,
+      "grad_norm": 0.12082793563604355,
+      "learning_rate": 2.8201404018397483e-05,
+      "loss": 0.0042,
+      "step": 13560
+    },
+    {
+      "epoch": 9.85,
+      "grad_norm": 0.001372415805235505,
+      "learning_rate": 2.8181231340272734e-05,
+      "loss": 0.0047,
+      "step": 13570
+    },
+    {
+      "epoch": 9.86,
+      "grad_norm": 0.0015574540011584759,
+      "learning_rate": 2.816105866214799e-05,
+      "loss": 0.0014,
+      "step": 13580
+    },
+    {
+      "epoch": 9.87,
+      "grad_norm": 0.002794192172586918,
+      "learning_rate": 2.814088598402324e-05,
+      "loss": 0.0011,
+      "step": 13590
+    },
+    {
+      "epoch": 9.87,
+      "grad_norm": 0.0029143195133656263,
+      "learning_rate": 2.8120713305898494e-05,
+      "loss": 0.0044,
+      "step": 13600
+    },
+    {
+      "epoch": 9.88,
+      "grad_norm": 20.12303352355957,
+      "learning_rate": 2.810054062777375e-05,
+      "loss": 0.0099,
+      "step": 13610
+    },
+    {
+      "epoch": 9.89,
+      "grad_norm": 0.29002559185028076,
+      "learning_rate": 2.8080367949648996e-05,
+      "loss": 0.003,
+      "step": 13620
+    },
+    {
+      "epoch": 9.89,
+      "grad_norm": 0.03364582359790802,
+      "learning_rate": 2.8060195271524247e-05,
+      "loss": 0.0071,
+      "step": 13630
+    },
+    {
+      "epoch": 9.9,
+      "grad_norm": 5.244564533233643,
+      "learning_rate": 2.80400225933995e-05,
+      "loss": 0.0132,
+      "step": 13640
+    },
+    {
+      "epoch": 9.91,
+      "grad_norm": 0.00039370081503875554,
+      "learning_rate": 2.8019849915274753e-05,
+      "loss": 0.0037,
+      "step": 13650
+    },
+    {
+      "epoch": 9.92,
+      "grad_norm": 0.006911866366863251,
+      "learning_rate": 2.7999677237150007e-05,
+      "loss": 0.0104,
+      "step": 13660
+    },
+    {
+      "epoch": 9.92,
+      "grad_norm": 0.1128559485077858,
+      "learning_rate": 2.797950455902526e-05,
+      "loss": 0.0097,
+      "step": 13670
+    },
+    {
+      "epoch": 9.93,
+      "grad_norm": 0.3790498375892639,
+      "learning_rate": 2.795933188090051e-05,
+      "loss": 0.0078,
+      "step": 13680
+    },
+    {
+      "epoch": 9.94,
+      "grad_norm": 19.0465087890625,
+      "learning_rate": 2.793915920277576e-05,
+      "loss": 0.004,
+      "step": 13690
+    },
+    {
+      "epoch": 9.95,
+      "grad_norm": 0.04134733974933624,
+      "learning_rate": 2.7918986524651015e-05,
+      "loss": 0.0061,
+      "step": 13700
+    },
+    {
+      "epoch": 9.95,
+      "grad_norm": 0.8866470456123352,
+      "learning_rate": 2.7898813846526266e-05,
+      "loss": 0.0113,
+      "step": 13710
+    },
+    {
+      "epoch": 9.96,
+      "grad_norm": 0.0012800481636077166,
+      "learning_rate": 2.787864116840152e-05,
+      "loss": 0.013,
+      "step": 13720
+    },
+    {
+      "epoch": 9.97,
+      "grad_norm": 0.736062228679657,
+      "learning_rate": 2.7858468490276768e-05,
+      "loss": 0.0013,
+      "step": 13730
+    },
+    {
+      "epoch": 9.97,
+      "grad_norm": 0.18687765300273895,
+      "learning_rate": 2.7838295812152022e-05,
+      "loss": 0.0049,
+      "step": 13740
+    },
+    {
+      "epoch": 9.98,
+      "grad_norm": 9.7325439453125,
+      "learning_rate": 2.7818123134027274e-05,
+      "loss": 0.03,
+      "step": 13750
+    },
+    {
+      "epoch": 9.99,
+      "grad_norm": 0.10014975816011429,
+      "learning_rate": 2.7797950455902528e-05,
+      "loss": 0.0207,
+      "step": 13760
+    },
+    {
+      "epoch": 10.0,
+      "grad_norm": 0.029998375102877617,
+      "learning_rate": 2.777777777777778e-05,
+      "loss": 0.0088,
+      "step": 13770
+    },
+    {
+      "epoch": 10.0,
+      "eval_accuracy": 0.9981508372282979,
+      "eval_f1": 0.9957295187193796,
+      "eval_loss": 0.004181100055575371,
+      "eval_precision": 0.9994740440751065,
+      "eval_recall": 0.9920129463353519,
+      "eval_roc_auc": 0.999982062040609,
+      "eval_runtime": 386.704,
+      "eval_samples_per_second": 227.947,
+      "eval_steps_per_second": 14.249,
+      "step": 13775
+    },
+    {
+      "epoch": 10.0,
+      "grad_norm": 0.07064787298440933,
+      "learning_rate": 2.7757605099653034e-05,
+      "loss": 0.0176,
+      "step": 13780
+    },
+    {
+      "epoch": 10.01,
+      "grad_norm": 0.0020518777891993523,
+      "learning_rate": 2.773743242152828e-05,
+      "loss": 0.0056,
+      "step": 13790
+    },
+    {
+      "epoch": 10.02,
+      "grad_norm": 2.226813316345215,
+      "learning_rate": 2.7717259743403536e-05,
+      "loss": 0.0137,
+      "step": 13800
+    },
+    {
+      "epoch": 10.03,
+      "grad_norm": 0.08585159480571747,
+      "learning_rate": 2.7697087065278787e-05,
+      "loss": 0.0164,
+      "step": 13810
+    },
+    {
+      "epoch": 10.03,
+      "grad_norm": 0.04419061914086342,
+      "learning_rate": 2.767691438715404e-05,
+      "loss": 0.0016,
+      "step": 13820
+    },
+    {
+      "epoch": 10.04,
+      "grad_norm": 0.11019827425479889,
+      "learning_rate": 2.7656741709029292e-05,
+      "loss": 0.0057,
+      "step": 13830
+    },
+    {
+      "epoch": 10.05,
+      "grad_norm": 0.004558782558888197,
+      "learning_rate": 2.7636569030904547e-05,
+      "loss": 0.0248,
+      "step": 13840
+    },
+    {
+      "epoch": 10.05,
+      "grad_norm": 0.20797798037528992,
+      "learning_rate": 2.7616396352779795e-05,
+      "loss": 0.004,
+      "step": 13850
+    },
+    {
+      "epoch": 10.06,
+      "grad_norm": 0.001556594274006784,
+      "learning_rate": 2.759622367465505e-05,
+      "loss": 0.0042,
+      "step": 13860
+    },
+    {
+      "epoch": 10.07,
+      "grad_norm": 2.7074778079986572,
+      "learning_rate": 2.75760509965303e-05,
+      "loss": 0.0186,
+      "step": 13870
+    },
+    {
+      "epoch": 10.08,
+      "grad_norm": 1.4041087627410889,
+      "learning_rate": 2.7555878318405555e-05,
+      "loss": 0.0321,
+      "step": 13880
+    },
+    {
+      "epoch": 10.08,
+      "grad_norm": 0.0077447472140192986,
+      "learning_rate": 2.7535705640280806e-05,
+      "loss": 0.0115,
+      "step": 13890
+    },
+    {
+      "epoch": 10.09,
+      "grad_norm": 0.2554437220096588,
+      "learning_rate": 2.751553296215606e-05,
+      "loss": 0.0112,
+      "step": 13900
+    },
+    {
+      "epoch": 10.1,
+      "grad_norm": 0.04327264055609703,
+      "learning_rate": 2.7495360284031308e-05,
+      "loss": 0.0042,
+      "step": 13910
+    },
+    {
+      "epoch": 10.11,
+      "grad_norm": 0.0026459668297320604,
+      "learning_rate": 2.747518760590656e-05,
+      "loss": 0.0129,
+      "step": 13920
+    },
+    {
+      "epoch": 10.11,
+      "grad_norm": 0.0002914105716627091,
+      "learning_rate": 2.7455014927781813e-05,
+      "loss": 0.004,
+      "step": 13930
+    },
+    {
+      "epoch": 10.12,
+      "grad_norm": 0.00024171853146981448,
+      "learning_rate": 2.7434842249657068e-05,
+      "loss": 0.0126,
+      "step": 13940
+    },
+    {
+      "epoch": 10.13,
+      "grad_norm": 0.014345817267894745,
+      "learning_rate": 2.741466957153232e-05,
+      "loss": 0.0181,
+      "step": 13950
+    },
+    {
+      "epoch": 10.13,
+      "grad_norm": 0.11584875732660294,
+      "learning_rate": 2.7394496893407567e-05,
+      "loss": 0.026,
+      "step": 13960
+    },
+    {
+      "epoch": 10.14,
+      "grad_norm": 2.401568651199341,
+      "learning_rate": 2.737432421528282e-05,
+      "loss": 0.0141,
+      "step": 13970
+    },
+    {
+      "epoch": 10.15,
+      "grad_norm": 0.01768440753221512,
+      "learning_rate": 2.7354151537158072e-05,
+      "loss": 0.018,
+      "step": 13980
+    },
+    {
+      "epoch": 10.16,
+      "grad_norm": 0.038432709872722626,
+      "learning_rate": 2.7333978859033327e-05,
+      "loss": 0.0143,
+      "step": 13990
+    },
+    {
+      "epoch": 10.16,
+      "grad_norm": 20.604793548583984,
+      "learning_rate": 2.731380618090858e-05,
+      "loss": 0.029,
+      "step": 14000
+    },
+    {
+      "epoch": 10.17,
+      "grad_norm": 0.26159873604774475,
+      "learning_rate": 2.7293633502783832e-05,
+      "loss": 0.0161,
+      "step": 14010
+    },
+    {
+      "epoch": 10.18,
+      "grad_norm": 15.196161270141602,
+      "learning_rate": 2.727346082465908e-05,
+      "loss": 0.0231,
+      "step": 14020
+    },
+    {
+      "epoch": 10.19,
+      "grad_norm": 19.31382179260254,
+      "learning_rate": 2.7253288146534334e-05,
+      "loss": 0.0157,
+      "step": 14030
+    },
+    {
+      "epoch": 10.19,
+      "grad_norm": 0.8836584091186523,
+      "learning_rate": 2.7233115468409585e-05,
+      "loss": 0.0127,
+      "step": 14040
+    },
+    {
+      "epoch": 10.2,
+      "grad_norm": 0.014301744289696217,
+      "learning_rate": 2.721294279028484e-05,
+      "loss": 0.0204,
+      "step": 14050
+    },
+    {
+      "epoch": 10.21,
+      "grad_norm": 0.010419386439025402,
+      "learning_rate": 2.7192770112160094e-05,
+      "loss": 0.0026,
+      "step": 14060
+    },
+    {
+      "epoch": 10.21,
+      "grad_norm": 0.0014298513997346163,
+      "learning_rate": 2.7172597434035345e-05,
+      "loss": 0.0048,
+      "step": 14070
+    },
+    {
+      "epoch": 10.22,
+      "grad_norm": 0.0004111882590223104,
+      "learning_rate": 2.7152424755910593e-05,
+      "loss": 0.0045,
+      "step": 14080
+    },
+    {
+      "epoch": 10.23,
+      "grad_norm": 9.619214057922363,
+      "learning_rate": 2.7132252077785847e-05,
+      "loss": 0.0135,
+      "step": 14090
+    },
+    {
+      "epoch": 10.24,
+      "grad_norm": 0.004899363964796066,
+      "learning_rate": 2.71120793996611e-05,
+      "loss": 0.0243,
+      "step": 14100
+    },
+    {
+      "epoch": 10.24,
+      "grad_norm": 0.05989596247673035,
+      "learning_rate": 2.7091906721536353e-05,
+      "loss": 0.0049,
+      "step": 14110
+    },
+    {
+      "epoch": 10.25,
+      "grad_norm": 0.08896303921937943,
+      "learning_rate": 2.7071734043411607e-05,
+      "loss": 0.0167,
+      "step": 14120
+    },
+    {
+      "epoch": 10.26,
+      "grad_norm": 0.0220908485352993,
+      "learning_rate": 2.705156136528686e-05,
+      "loss": 0.0252,
+      "step": 14130
+    },
+    {
+      "epoch": 10.26,
+      "grad_norm": 0.42359215021133423,
+      "learning_rate": 2.7031388687162106e-05,
+      "loss": 0.0049,
+      "step": 14140
+    },
+    {
+      "epoch": 10.27,
+      "grad_norm": 0.0026175114326179028,
+      "learning_rate": 2.701121600903736e-05,
+      "loss": 0.0151,
+      "step": 14150
+    },
+    {
+      "epoch": 10.28,
+      "grad_norm": 0.0033834856003522873,
+      "learning_rate": 2.6991043330912612e-05,
+      "loss": 0.0054,
+      "step": 14160
+    },
+    {
+      "epoch": 10.29,
+      "grad_norm": 13.682700157165527,
+      "learning_rate": 2.6970870652787866e-05,
+      "loss": 0.0343,
+      "step": 14170
+    },
+    {
+      "epoch": 10.29,
+      "grad_norm": 10.103180885314941,
+      "learning_rate": 2.695069797466312e-05,
+      "loss": 0.0007,
+      "step": 14180
+    },
+    {
+      "epoch": 10.3,
+      "grad_norm": 0.08334420621395111,
+      "learning_rate": 2.693052529653837e-05,
+      "loss": 0.0252,
+      "step": 14190
+    },
+    {
+      "epoch": 10.31,
+      "grad_norm": 0.08771252632141113,
+      "learning_rate": 2.691035261841362e-05,
+      "loss": 0.0061,
+      "step": 14200
+    },
+    {
+      "epoch": 10.32,
+      "grad_norm": 0.025557437911629677,
+      "learning_rate": 2.6890179940288874e-05,
+      "loss": 0.0153,
+      "step": 14210
+    },
+    {
+      "epoch": 10.32,
+      "grad_norm": 0.06524745374917984,
+      "learning_rate": 2.6870007262164125e-05,
+      "loss": 0.0018,
+      "step": 14220
+    },
+    {
+      "epoch": 10.33,
+      "grad_norm": 0.2285977602005005,
+      "learning_rate": 2.684983458403938e-05,
+      "loss": 0.0129,
+      "step": 14230
+    },
+    {
+      "epoch": 10.34,
+      "grad_norm": 0.004466556012630463,
+      "learning_rate": 2.6829661905914634e-05,
+      "loss": 0.0095,
+      "step": 14240
+    },
+    {
+      "epoch": 10.34,
+      "grad_norm": 0.238087996840477,
+      "learning_rate": 2.680948922778988e-05,
+      "loss": 0.0022,
+      "step": 14250
+    },
+    {
+      "epoch": 10.35,
+      "grad_norm": 0.004598718602210283,
+      "learning_rate": 2.6789316549665133e-05,
+      "loss": 0.009,
+      "step": 14260
+    },
+    {
+      "epoch": 10.36,
+      "grad_norm": 0.0024979293812066317,
+      "learning_rate": 2.6769143871540387e-05,
+      "loss": 0.0039,
+      "step": 14270
+    },
+    {
+      "epoch": 10.37,
+      "grad_norm": 0.17606070637702942,
+      "learning_rate": 2.6748971193415638e-05,
+      "loss": 0.0064,
+      "step": 14280
+    },
+    {
+      "epoch": 10.37,
+      "grad_norm": 0.0038365270011126995,
+      "learning_rate": 2.6728798515290893e-05,
+      "loss": 0.017,
+      "step": 14290
+    },
+    {
+      "epoch": 10.38,
+      "grad_norm": 0.42660027742385864,
+      "learning_rate": 2.6708625837166147e-05,
+      "loss": 0.002,
+      "step": 14300
+    },
+    {
+      "epoch": 10.39,
+      "grad_norm": 0.0034667763393372297,
+      "learning_rate": 2.6688453159041395e-05,
+      "loss": 0.0142,
+      "step": 14310
+    },
+    {
+      "epoch": 10.4,
+      "grad_norm": 0.26163700222969055,
+      "learning_rate": 2.6668280480916646e-05,
+      "loss": 0.0053,
+      "step": 14320
+    },
+    {
+      "epoch": 10.4,
+      "grad_norm": 0.7694606184959412,
+      "learning_rate": 2.66481078027919e-05,
+      "loss": 0.0079,
+      "step": 14330
+    },
+    {
+      "epoch": 10.41,
+      "grad_norm": 0.7763135433197021,
+      "learning_rate": 2.662793512466715e-05,
+      "loss": 0.0041,
+      "step": 14340
+    },
+    {
+      "epoch": 10.42,
+      "grad_norm": 35.8295783996582,
+      "learning_rate": 2.6607762446542406e-05,
+      "loss": 0.0264,
+      "step": 14350
+    },
+    {
+      "epoch": 10.42,
+      "grad_norm": 5.697082042694092,
+      "learning_rate": 2.658758976841766e-05,
+      "loss": 0.0429,
+      "step": 14360
+    },
+    {
+      "epoch": 10.43,
+      "grad_norm": 1.5952166318893433,
+      "learning_rate": 2.6567417090292908e-05,
+      "loss": 0.0072,
+      "step": 14370
+    },
+    {
+      "epoch": 10.44,
+      "grad_norm": 0.30308908224105835,
+      "learning_rate": 2.654724441216816e-05,
+      "loss": 0.0024,
+      "step": 14380
+    },
+    {
+      "epoch": 10.45,
+      "grad_norm": 22.062713623046875,
+      "learning_rate": 2.6527071734043414e-05,
+      "loss": 0.0301,
+      "step": 14390
+    },
+    {
+      "epoch": 10.45,
+      "grad_norm": 1.5792371034622192,
+      "learning_rate": 2.6506899055918665e-05,
+      "loss": 0.0236,
+      "step": 14400
+    },
+    {
+      "epoch": 10.46,
+      "grad_norm": 0.21174445748329163,
+      "learning_rate": 2.648672637779392e-05,
+      "loss": 0.005,
+      "step": 14410
+    },
+    {
+      "epoch": 10.47,
+      "grad_norm": 0.003805481130257249,
+      "learning_rate": 2.6466553699669174e-05,
+      "loss": 0.009,
+      "step": 14420
+    },
+    {
+      "epoch": 10.48,
+      "grad_norm": 0.05694010108709335,
+      "learning_rate": 2.6446381021544418e-05,
+      "loss": 0.0185,
+      "step": 14430
+    },
+    {
+      "epoch": 10.48,
+      "grad_norm": 0.01830306462943554,
+      "learning_rate": 2.6426208343419672e-05,
+      "loss": 0.0011,
+      "step": 14440
+    },
+    {
+      "epoch": 10.49,
+      "grad_norm": 1.060694932937622,
+      "learning_rate": 2.6406035665294927e-05,
+      "loss": 0.0131,
+      "step": 14450
+    },
+    {
+      "epoch": 10.5,
+      "grad_norm": 0.5278339982032776,
+      "learning_rate": 2.6385862987170178e-05,
+      "loss": 0.0064,
+      "step": 14460
+    },
+    {
+      "epoch": 10.5,
+      "grad_norm": 5.779145240783691,
+      "learning_rate": 2.6365690309045432e-05,
+      "loss": 0.0047,
+      "step": 14470
+    },
+    {
+      "epoch": 10.51,
+      "grad_norm": 0.07150599360466003,
+      "learning_rate": 2.634551763092068e-05,
+      "loss": 0.0034,
+      "step": 14480
+    },
+    {
+      "epoch": 10.52,
+      "grad_norm": 37.375770568847656,
+      "learning_rate": 2.632534495279593e-05,
+      "loss": 0.0171,
+      "step": 14490
+    },
+    {
+      "epoch": 10.53,
+      "grad_norm": 0.21243281662464142,
+      "learning_rate": 2.6305172274671186e-05,
+      "loss": 0.0115,
+      "step": 14500
+    },
+    {
+      "epoch": 10.53,
+      "grad_norm": 0.0017663768958300352,
+      "learning_rate": 2.628499959654644e-05,
+      "loss": 0.0041,
+      "step": 14510
+    },
+    {
+      "epoch": 10.54,
+      "grad_norm": 0.026727020740509033,
+      "learning_rate": 2.626482691842169e-05,
+      "loss": 0.0026,
+      "step": 14520
+    },
+    {
+      "epoch": 10.55,
+      "grad_norm": 0.15876881778240204,
+      "learning_rate": 2.6244654240296946e-05,
+      "loss": 0.0123,
+      "step": 14530
+    },
+    {
+      "epoch": 10.56,
+      "grad_norm": 0.006429377943277359,
+      "learning_rate": 2.6224481562172193e-05,
+      "loss": 0.001,
+      "step": 14540
+    },
+    {
+      "epoch": 10.56,
+      "grad_norm": 0.0438314713537693,
+      "learning_rate": 2.6204308884047444e-05,
+      "loss": 0.0197,
+      "step": 14550
+    },
+    {
+      "epoch": 10.57,
+      "grad_norm": 0.153500035405159,
+      "learning_rate": 2.61841362059227e-05,
+      "loss": 0.0099,
+      "step": 14560
+    },
+    {
+      "epoch": 10.58,
+      "grad_norm": 0.0012310659512877464,
+      "learning_rate": 2.6163963527797953e-05,
+      "loss": 0.0103,
+      "step": 14570
+    },
+    {
+      "epoch": 10.58,
+      "grad_norm": 14.718790054321289,
+      "learning_rate": 2.6143790849673204e-05,
+      "loss": 0.0082,
+      "step": 14580
+    },
+    {
+      "epoch": 10.59,
+      "grad_norm": 0.1377602070569992,
+      "learning_rate": 2.612361817154846e-05,
+      "loss": 0.0043,
+      "step": 14590
+    },
+    {
+      "epoch": 10.6,
+      "grad_norm": 0.01837272197008133,
+      "learning_rate": 2.6103445493423707e-05,
+      "loss": 0.0215,
+      "step": 14600
+    },
+    {
+      "epoch": 10.61,
+      "grad_norm": 0.0010429949034005404,
+      "learning_rate": 2.6083272815298958e-05,
+      "loss": 0.0145,
+      "step": 14610
+    },
+    {
+      "epoch": 10.61,
+      "grad_norm": 0.00035799731267616153,
+      "learning_rate": 2.6063100137174212e-05,
+      "loss": 0.0063,
+      "step": 14620
+    },
+    {
+      "epoch": 10.62,
+      "grad_norm": 0.17942433059215546,
+      "learning_rate": 2.6042927459049467e-05,
+      "loss": 0.0061,
+      "step": 14630
+    },
+    {
+      "epoch": 10.63,
+      "grad_norm": 8.952868461608887,
+      "learning_rate": 2.6022754780924718e-05,
+      "loss": 0.0197,
+      "step": 14640
+    },
+    {
+      "epoch": 10.64,
+      "grad_norm": 5.475463390350342,
+      "learning_rate": 2.6002582102799972e-05,
+      "loss": 0.0044,
+      "step": 14650
+    },
+    {
+      "epoch": 10.64,
+      "grad_norm": 0.020051149651408195,
+      "learning_rate": 2.598240942467522e-05,
+      "loss": 0.0021,
+      "step": 14660
+    },
+    {
+      "epoch": 10.65,
+      "grad_norm": 0.033862363547086716,
+      "learning_rate": 2.596223674655047e-05,
+      "loss": 0.0175,
+      "step": 14670
+    },
+    {
+      "epoch": 10.66,
+      "grad_norm": 0.0405866913497448,
+      "learning_rate": 2.5942064068425725e-05,
+      "loss": 0.0044,
+      "step": 14680
+    },
+    {
+      "epoch": 10.66,
+      "grad_norm": 1.4321554899215698,
+      "learning_rate": 2.592189139030098e-05,
+      "loss": 0.0085,
+      "step": 14690
+    },
+    {
+      "epoch": 10.67,
+      "grad_norm": 1.3793431520462036,
+      "learning_rate": 2.590171871217623e-05,
+      "loss": 0.0105,
+      "step": 14700
+    },
+    {
+      "epoch": 10.68,
+      "grad_norm": 0.22573919594287872,
+      "learning_rate": 2.588154603405148e-05,
+      "loss": 0.0136,
+      "step": 14710
+    },
+    {
+      "epoch": 10.69,
+      "grad_norm": 0.015440167859196663,
+      "learning_rate": 2.5861373355926733e-05,
+      "loss": 0.0457,
+      "step": 14720
+    },
+    {
+      "epoch": 10.69,
+      "grad_norm": 0.01662645861506462,
+      "learning_rate": 2.5841200677801984e-05,
+      "loss": 0.0057,
+      "step": 14730
+    },
+    {
+      "epoch": 10.7,
+      "grad_norm": 0.18805713951587677,
+      "learning_rate": 2.582102799967724e-05,
+      "loss": 0.0251,
+      "step": 14740
+    },
+    {
+      "epoch": 10.71,
+      "grad_norm": 3.959599256515503,
+      "learning_rate": 2.5800855321552493e-05,
+      "loss": 0.0078,
+      "step": 14750
+    },
+    {
+      "epoch": 10.72,
+      "grad_norm": 0.013117094524204731,
+      "learning_rate": 2.5780682643427744e-05,
+      "loss": 0.0061,
+      "step": 14760
+    },
+    {
+      "epoch": 10.72,
+      "grad_norm": 0.08931567519903183,
+      "learning_rate": 2.5760509965302992e-05,
+      "loss": 0.0076,
+      "step": 14770
+    },
+    {
+      "epoch": 10.73,
+      "grad_norm": 3.2667019367218018,
+      "learning_rate": 2.5740337287178246e-05,
+      "loss": 0.0039,
+      "step": 14780
+    },
+    {
+      "epoch": 10.74,
+      "grad_norm": 0.13932375609874725,
+      "learning_rate": 2.5720164609053497e-05,
+      "loss": 0.0079,
+      "step": 14790
+    },
+    {
+      "epoch": 10.74,
+      "grad_norm": 1.5266163349151611,
+      "learning_rate": 2.5699991930928752e-05,
+      "loss": 0.0041,
+      "step": 14800
+    },
+    {
+      "epoch": 10.75,
+      "grad_norm": 10.672917366027832,
+      "learning_rate": 2.5679819252804006e-05,
+      "loss": 0.0095,
+      "step": 14810
+    },
+    {
+      "epoch": 10.76,
+      "grad_norm": 0.008519817143678665,
+      "learning_rate": 2.5659646574679257e-05,
+      "loss": 0.0158,
+      "step": 14820
+    },
+    {
+      "epoch": 10.77,
+      "grad_norm": 0.009736945852637291,
+      "learning_rate": 2.5639473896554505e-05,
+      "loss": 0.0024,
+      "step": 14830
+    },
+    {
+      "epoch": 10.77,
+      "grad_norm": 0.02101651206612587,
+      "learning_rate": 2.561930121842976e-05,
+      "loss": 0.0281,
+      "step": 14840
+    },
+    {
+      "epoch": 10.78,
+      "grad_norm": 0.009922484867274761,
+      "learning_rate": 2.559912854030501e-05,
+      "loss": 0.0306,
+      "step": 14850
+    },
+    {
+      "epoch": 10.79,
+      "grad_norm": 1.5312800407409668,
+      "learning_rate": 2.5578955862180265e-05,
+      "loss": 0.0055,
+      "step": 14860
+    },
+    {
+      "epoch": 10.79,
+      "grad_norm": 0.24132943153381348,
+      "learning_rate": 2.555878318405552e-05,
+      "loss": 0.0061,
+      "step": 14870
+    },
+    {
+      "epoch": 10.8,
+      "grad_norm": 2.395301342010498,
+      "learning_rate": 2.553861050593077e-05,
+      "loss": 0.0133,
+      "step": 14880
+    },
+    {
+      "epoch": 10.81,
+      "grad_norm": 0.15985184907913208,
+      "learning_rate": 2.551843782780602e-05,
+      "loss": 0.0197,
+      "step": 14890
+    },
+    {
+      "epoch": 10.82,
+      "grad_norm": 1.5994820594787598,
+      "learning_rate": 2.5498265149681273e-05,
+      "loss": 0.0073,
+      "step": 14900
+    },
+    {
+      "epoch": 10.82,
+      "grad_norm": 0.009897888638079166,
+      "learning_rate": 2.5478092471556524e-05,
+      "loss": 0.0058,
+      "step": 14910
+    },
+    {
+      "epoch": 10.83,
+      "grad_norm": 0.013136355206370354,
+      "learning_rate": 2.545791979343178e-05,
+      "loss": 0.005,
+      "step": 14920
+    },
+    {
+      "epoch": 10.84,
+      "grad_norm": 0.7152259945869446,
+      "learning_rate": 2.5437747115307033e-05,
+      "loss": 0.0131,
+      "step": 14930
+    },
+    {
+      "epoch": 10.85,
+      "grad_norm": 0.002481738803908229,
+      "learning_rate": 2.5417574437182277e-05,
+      "loss": 0.0053,
+      "step": 14940
+    },
+    {
+      "epoch": 10.85,
+      "grad_norm": 0.5290198922157288,
+      "learning_rate": 2.539740175905753e-05,
+      "loss": 0.0109,
+      "step": 14950
+    },
+    {
+      "epoch": 10.86,
+      "grad_norm": 0.23663221299648285,
+      "learning_rate": 2.5377229080932786e-05,
+      "loss": 0.0159,
+      "step": 14960
+    },
+    {
+      "epoch": 10.87,
+      "grad_norm": 0.03912312537431717,
+      "learning_rate": 2.5357056402808037e-05,
+      "loss": 0.0078,
+      "step": 14970
+    },
+    {
+      "epoch": 10.87,
+      "grad_norm": 0.0030671055428683758,
+      "learning_rate": 2.533688372468329e-05,
+      "loss": 0.0033,
+      "step": 14980
+    },
+    {
+      "epoch": 10.88,
+      "grad_norm": 3.3906474113464355,
+      "learning_rate": 2.5316711046558546e-05,
+      "loss": 0.0024,
+      "step": 14990
+    },
+    {
+      "epoch": 10.89,
+      "grad_norm": 0.002407472115010023,
+      "learning_rate": 2.529653836843379e-05,
+      "loss": 0.0038,
+      "step": 15000
+    },
+    {
+      "epoch": 10.9,
+      "grad_norm": 0.9811303615570068,
+      "learning_rate": 2.5276365690309045e-05,
+      "loss": 0.0272,
+      "step": 15010
+    },
+    {
+      "epoch": 10.9,
+      "grad_norm": 0.217171773314476,
+      "learning_rate": 2.52561930121843e-05,
+      "loss": 0.013,
+      "step": 15020
+    },
+    {
+      "epoch": 10.91,
+      "grad_norm": 0.01452325563877821,
+      "learning_rate": 2.523602033405955e-05,
+      "loss": 0.0065,
+      "step": 15030
+    },
+    {
+      "epoch": 10.92,
+      "grad_norm": 0.003643118543550372,
+      "learning_rate": 2.5215847655934805e-05,
+      "loss": 0.0054,
+      "step": 15040
+    },
+    {
+      "epoch": 10.93,
+      "grad_norm": 0.1353772133588791,
+      "learning_rate": 2.5195674977810056e-05,
+      "loss": 0.0043,
+      "step": 15050
+    },
+    {
+      "epoch": 10.93,
+      "grad_norm": 0.01406155712902546,
+      "learning_rate": 2.5175502299685304e-05,
+      "loss": 0.0071,
+      "step": 15060
+    },
+    {
+      "epoch": 10.94,
+      "grad_norm": 0.0014850402949377894,
+      "learning_rate": 2.5155329621560558e-05,
+      "loss": 0.0029,
+      "step": 15070
+    },
+    {
+      "epoch": 10.95,
+      "grad_norm": 0.0004027250688523054,
+      "learning_rate": 2.5135156943435813e-05,
+      "loss": 0.0017,
+      "step": 15080
+    },
+    {
+      "epoch": 10.95,
+      "grad_norm": 0.059271667152643204,
+      "learning_rate": 2.5114984265311064e-05,
+      "loss": 0.0041,
+      "step": 15090
+    },
+    {
+      "epoch": 10.96,
+      "grad_norm": 1.4982573986053467,
+      "learning_rate": 2.5094811587186318e-05,
+      "loss": 0.0103,
+      "step": 15100
+    },
+    {
+      "epoch": 10.97,
+      "grad_norm": 11.631914138793945,
+      "learning_rate": 2.507463890906157e-05,
+      "loss": 0.0164,
+      "step": 15110
+    },
+    {
+      "epoch": 10.98,
+      "grad_norm": 0.01383855938911438,
+      "learning_rate": 2.5054466230936817e-05,
+      "loss": 0.0088,
+      "step": 15120
+    },
+    {
+      "epoch": 10.98,
+      "grad_norm": 0.0035403859801590443,
+      "learning_rate": 2.503429355281207e-05,
+      "loss": 0.0115,
+      "step": 15130
+    },
+    {
+      "epoch": 10.99,
+      "grad_norm": 0.2655543088912964,
+      "learning_rate": 2.5014120874687326e-05,
+      "loss": 0.0128,
+      "step": 15140
+    },
+    {
+      "epoch": 11.0,
+      "grad_norm": 1.0399569272994995,
+      "learning_rate": 2.4993948196562577e-05,
+      "loss": 0.0078,
+      "step": 15150
+    },
+    {
+      "epoch": 11.0,
+      "eval_accuracy": 0.9981508372282979,
+      "eval_f1": 0.995729966206481,
+      "eval_loss": 0.004333225544542074,
+      "eval_precision": 0.9993689856444234,
+      "eval_recall": 0.9921173522656087,
+      "eval_roc_auc": 0.9999826230983933,
+      "eval_runtime": 388.635,
+      "eval_samples_per_second": 226.814,
+      "eval_steps_per_second": 14.178,
+      "step": 15152
+    },
+    {
+      "epoch": 11.01,
+      "grad_norm": 0.0071834782138466835,
+      "learning_rate": 2.4973775518437828e-05,
+      "loss": 0.0167,
+      "step": 15160
+    },
+    {
+      "epoch": 11.01,
+      "grad_norm": 0.25943008065223694,
+      "learning_rate": 2.4953602840313082e-05,
+      "loss": 0.0008,
+      "step": 15170
+    },
+    {
+      "epoch": 11.02,
+      "grad_norm": 0.2013348937034607,
+      "learning_rate": 2.4933430162188333e-05,
+      "loss": 0.0068,
+      "step": 15180
+    },
+    {
+      "epoch": 11.03,
+      "grad_norm": 3.2597954273223877,
+      "learning_rate": 2.4913257484063585e-05,
+      "loss": 0.018,
+      "step": 15190
+    },
+    {
+      "epoch": 11.03,
+      "grad_norm": 0.036766473203897476,
+      "learning_rate": 2.489308480593884e-05,
+      "loss": 0.002,
+      "step": 15200
+    },
+    {
+      "epoch": 11.04,
+      "grad_norm": 0.5187474489212036,
+      "learning_rate": 2.487291212781409e-05,
+      "loss": 0.0173,
+      "step": 15210
+    },
+    {
+      "epoch": 11.05,
+      "grad_norm": 0.0028880988247692585,
+      "learning_rate": 2.485273944968934e-05,
+      "loss": 0.0036,
+      "step": 15220
+    },
+    {
+      "epoch": 11.06,
+      "grad_norm": 0.13685983419418335,
+      "learning_rate": 2.4832566771564596e-05,
+      "loss": 0.0083,
+      "step": 15230
+    },
+    {
+      "epoch": 11.06,
+      "grad_norm": 0.02378654293715954,
+      "learning_rate": 2.4812394093439847e-05,
+      "loss": 0.0243,
+      "step": 15240
+    },
+    {
+      "epoch": 11.07,
+      "grad_norm": 0.009897316806018353,
+      "learning_rate": 2.4792221415315098e-05,
+      "loss": 0.0062,
+      "step": 15250
+    },
+    {
+      "epoch": 11.08,
+      "grad_norm": 0.005875424947589636,
+      "learning_rate": 2.4772048737190352e-05,
+      "loss": 0.0028,
+      "step": 15260
+    },
+    {
+      "epoch": 11.09,
+      "grad_norm": 0.008453444577753544,
+      "learning_rate": 2.4751876059065603e-05,
+      "loss": 0.0019,
+      "step": 15270
+    },
+    {
+      "epoch": 11.09,
+      "grad_norm": 1.0412099361419678,
+      "learning_rate": 2.4731703380940854e-05,
+      "loss": 0.0115,
+      "step": 15280
+    },
+    {
+      "epoch": 11.1,
+      "grad_norm": 0.19635595381259918,
+      "learning_rate": 2.471153070281611e-05,
+      "loss": 0.0041,
+      "step": 15290
+    },
+    {
+      "epoch": 11.11,
+      "grad_norm": 0.1889576017856598,
+      "learning_rate": 2.4691358024691357e-05,
+      "loss": 0.0023,
+      "step": 15300
+    },
+    {
+      "epoch": 11.11,
+      "grad_norm": 0.0010881648631766438,
+      "learning_rate": 2.467118534656661e-05,
+      "loss": 0.0103,
+      "step": 15310
+    },
+    {
+      "epoch": 11.12,
+      "grad_norm": 0.00150064448826015,
+      "learning_rate": 2.4651012668441866e-05,
+      "loss": 0.0192,
+      "step": 15320
+    },
+    {
+      "epoch": 11.13,
+      "grad_norm": 0.2817605435848236,
+      "learning_rate": 2.4630839990317113e-05,
+      "loss": 0.0132,
+      "step": 15330
+    },
+    {
+      "epoch": 11.14,
+      "grad_norm": 5.953119277954102,
+      "learning_rate": 2.4610667312192368e-05,
+      "loss": 0.0051,
+      "step": 15340
+    },
+    {
+      "epoch": 11.14,
+      "grad_norm": 0.0015101551543921232,
+      "learning_rate": 2.4590494634067622e-05,
+      "loss": 0.0095,
+      "step": 15350
+    },
+    {
+      "epoch": 11.15,
+      "grad_norm": 0.10892230272293091,
+      "learning_rate": 2.457032195594287e-05,
+      "loss": 0.0132,
+      "step": 15360
+    },
+    {
+      "epoch": 11.16,
+      "grad_norm": 0.0027881257701665163,
+      "learning_rate": 2.4550149277818124e-05,
+      "loss": 0.002,
+      "step": 15370
+    },
+    {
+      "epoch": 11.17,
+      "grad_norm": 0.05755399912595749,
+      "learning_rate": 2.452997659969338e-05,
+      "loss": 0.0006,
+      "step": 15380
+    },
+    {
+      "epoch": 11.17,
+      "grad_norm": 0.019885243847966194,
+      "learning_rate": 2.4509803921568626e-05,
+      "loss": 0.0041,
+      "step": 15390
+    },
+    {
+      "epoch": 11.18,
+      "grad_norm": 1.7892396450042725,
+      "learning_rate": 2.448963124344388e-05,
+      "loss": 0.0077,
+      "step": 15400
+    },
+    {
+      "epoch": 11.19,
+      "grad_norm": 0.0175933800637722,
+      "learning_rate": 2.4469458565319135e-05,
+      "loss": 0.0047,
+      "step": 15410
+    },
+    {
+      "epoch": 11.19,
+      "grad_norm": 0.6241514086723328,
+      "learning_rate": 2.4449285887194383e-05,
+      "loss": 0.0067,
+      "step": 15420
+    },
+    {
+      "epoch": 11.2,
+      "grad_norm": 0.009889623150229454,
+      "learning_rate": 2.4429113209069638e-05,
+      "loss": 0.0122,
+      "step": 15430
+    },
+    {
+      "epoch": 11.21,
+      "grad_norm": 0.005964280106127262,
+      "learning_rate": 2.4408940530944892e-05,
+      "loss": 0.0056,
+      "step": 15440
+    },
+    {
+      "epoch": 11.22,
+      "grad_norm": 0.0019957488402724266,
+      "learning_rate": 2.438876785282014e-05,
+      "loss": 0.0011,
+      "step": 15450
+    },
+    {
+      "epoch": 11.22,
+      "grad_norm": 0.09828945994377136,
+      "learning_rate": 2.4368595174695394e-05,
+      "loss": 0.0047,
+      "step": 15460
+    },
+    {
+      "epoch": 11.23,
+      "grad_norm": 0.011916988529264927,
+      "learning_rate": 2.434842249657065e-05,
+      "loss": 0.005,
+      "step": 15470
+    },
+    {
+      "epoch": 11.24,
+      "grad_norm": 21.119287490844727,
+      "learning_rate": 2.4328249818445896e-05,
+      "loss": 0.0142,
+      "step": 15480
+    },
+    {
+      "epoch": 11.25,
+      "grad_norm": 0.0025653273332864046,
+      "learning_rate": 2.430807714032115e-05,
+      "loss": 0.0039,
+      "step": 15490
+    },
+    {
+      "epoch": 11.25,
+      "grad_norm": 2.179283380508423,
+      "learning_rate": 2.4287904462196405e-05,
+      "loss": 0.0019,
+      "step": 15500
+    },
+    {
+      "epoch": 11.26,
+      "grad_norm": 0.041954535990953445,
+      "learning_rate": 2.4267731784071653e-05,
+      "loss": 0.0018,
+      "step": 15510
+    },
+    {
+      "epoch": 11.27,
+      "grad_norm": 0.13100317120552063,
+      "learning_rate": 2.4247559105946907e-05,
+      "loss": 0.0206,
+      "step": 15520
+    },
+    {
+      "epoch": 11.27,
+      "grad_norm": 0.11618025600910187,
+      "learning_rate": 2.4227386427822162e-05,
+      "loss": 0.0009,
+      "step": 15530
+    },
+    {
+      "epoch": 11.28,
+      "grad_norm": 0.12495038658380508,
+      "learning_rate": 2.420721374969741e-05,
+      "loss": 0.0018,
+      "step": 15540
+    },
+    {
+      "epoch": 11.29,
+      "grad_norm": 0.1562829166650772,
+      "learning_rate": 2.4187041071572664e-05,
+      "loss": 0.0415,
+      "step": 15550
+    },
+    {
+      "epoch": 11.3,
+      "grad_norm": 0.40936851501464844,
+      "learning_rate": 2.4166868393447915e-05,
+      "loss": 0.0005,
+      "step": 15560
+    },
+    {
+      "epoch": 11.3,
+      "grad_norm": 0.7347458004951477,
+      "learning_rate": 2.4146695715323166e-05,
+      "loss": 0.0011,
+      "step": 15570
+    },
+    {
+      "epoch": 11.31,
+      "grad_norm": 0.0022218599915504456,
+      "learning_rate": 2.412652303719842e-05,
+      "loss": 0.0045,
+      "step": 15580
+    },
+    {
+      "epoch": 11.32,
+      "grad_norm": 0.26786237955093384,
+      "learning_rate": 2.4106350359073672e-05,
+      "loss": 0.0198,
+      "step": 15590
+    },
+    {
+      "epoch": 11.32,
+      "grad_norm": 0.002820109250023961,
+      "learning_rate": 2.4086177680948923e-05,
+      "loss": 0.0037,
+      "step": 15600
+    },
+    {
+      "epoch": 11.33,
+      "grad_norm": 0.006033481098711491,
+      "learning_rate": 2.4066005002824177e-05,
+      "loss": 0.0024,
+      "step": 15610
+    },
+    {
+      "epoch": 11.34,
+      "grad_norm": 0.0011767403921112418,
+      "learning_rate": 2.404583232469943e-05,
+      "loss": 0.0033,
+      "step": 15620
+    },
+    {
+      "epoch": 11.35,
+      "grad_norm": 26.068288803100586,
+      "learning_rate": 2.402565964657468e-05,
+      "loss": 0.0076,
+      "step": 15630
+    },
+    {
+      "epoch": 11.35,
+      "grad_norm": 0.006819643080234528,
+      "learning_rate": 2.4005486968449934e-05,
+      "loss": 0.0009,
+      "step": 15640
+    },
+    {
+      "epoch": 11.36,
+      "grad_norm": 0.0006057365681044757,
+      "learning_rate": 2.3985314290325185e-05,
+      "loss": 0.0,
+      "step": 15650
+    },
+    {
+      "epoch": 11.37,
+      "grad_norm": 31.54865264892578,
+      "learning_rate": 2.3965141612200436e-05,
+      "loss": 0.015,
+      "step": 15660
+    },
+    {
+      "epoch": 11.38,
+      "grad_norm": 6.902843952178955,
+      "learning_rate": 2.394496893407569e-05,
+      "loss": 0.009,
+      "step": 15670
+    },
+    {
+      "epoch": 11.38,
+      "grad_norm": 0.00346160470508039,
+      "learning_rate": 2.392479625595094e-05,
+      "loss": 0.0145,
+      "step": 15680
+    },
+    {
+      "epoch": 11.39,
+      "grad_norm": 0.0008461098768748343,
+      "learning_rate": 2.3904623577826193e-05,
+      "loss": 0.0021,
+      "step": 15690
+    },
+    {
+      "epoch": 11.4,
+      "grad_norm": 8.534778594970703,
+      "learning_rate": 2.3884450899701447e-05,
+      "loss": 0.0039,
+      "step": 15700
+    },
+    {
+      "epoch": 11.4,
+      "grad_norm": 30.421672821044922,
+      "learning_rate": 2.3864278221576698e-05,
+      "loss": 0.0148,
+      "step": 15710
+    },
+    {
+      "epoch": 11.41,
+      "grad_norm": 0.011093859560787678,
+      "learning_rate": 2.384410554345195e-05,
+      "loss": 0.0017,
+      "step": 15720
+    },
+    {
+      "epoch": 11.42,
+      "grad_norm": 0.0029349441174417734,
+      "learning_rate": 2.3823932865327204e-05,
+      "loss": 0.0215,
+      "step": 15730
+    },
+    {
+      "epoch": 11.43,
+      "grad_norm": 7.691098690032959,
+      "learning_rate": 2.3803760187202455e-05,
+      "loss": 0.0188,
+      "step": 15740
+    },
+    {
+      "epoch": 11.43,
+      "grad_norm": 0.11859267950057983,
+      "learning_rate": 2.3783587509077706e-05,
+      "loss": 0.0291,
+      "step": 15750
+    },
+    {
+      "epoch": 11.44,
+      "grad_norm": 0.039303623139858246,
+      "learning_rate": 2.376341483095296e-05,
+      "loss": 0.0058,
+      "step": 15760
+    },
+    {
+      "epoch": 11.45,
+      "grad_norm": 0.08446269482374191,
+      "learning_rate": 2.374324215282821e-05,
+      "loss": 0.0104,
+      "step": 15770
+    },
+    {
+      "epoch": 11.46,
+      "grad_norm": 0.0554688386619091,
+      "learning_rate": 2.3723069474703462e-05,
+      "loss": 0.0039,
+      "step": 15780
+    },
+    {
+      "epoch": 11.46,
+      "grad_norm": 0.057702578604221344,
+      "learning_rate": 2.3702896796578717e-05,
+      "loss": 0.004,
+      "step": 15790
+    },
+    {
+      "epoch": 11.47,
+      "grad_norm": 0.04934044927358627,
+      "learning_rate": 2.3682724118453968e-05,
+      "loss": 0.0078,
+      "step": 15800
+    },
+    {
+      "epoch": 11.48,
+      "grad_norm": 0.025431061163544655,
+      "learning_rate": 2.366255144032922e-05,
+      "loss": 0.0396,
+      "step": 15810
+    },
+    {
+      "epoch": 11.48,
+      "grad_norm": 0.8368417620658875,
+      "learning_rate": 2.364237876220447e-05,
+      "loss": 0.0046,
+      "step": 15820
+    },
+    {
+      "epoch": 11.49,
+      "grad_norm": 0.12073979526758194,
+      "learning_rate": 2.3622206084079725e-05,
+      "loss": 0.0058,
+      "step": 15830
+    },
+    {
+      "epoch": 11.5,
+      "grad_norm": 0.2134593278169632,
+      "learning_rate": 2.3602033405954976e-05,
+      "loss": 0.0023,
+      "step": 15840
+    },
+    {
+      "epoch": 11.51,
+      "grad_norm": 0.0010249282931908965,
+      "learning_rate": 2.3581860727830227e-05,
+      "loss": 0.0034,
+      "step": 15850
+    },
+    {
+      "epoch": 11.51,
+      "grad_norm": 7.7366180419921875,
+      "learning_rate": 2.356168804970548e-05,
+      "loss": 0.0234,
+      "step": 15860
+    },
+    {
+      "epoch": 11.52,
+      "grad_norm": 0.0021505900658667088,
+      "learning_rate": 2.3541515371580732e-05,
+      "loss": 0.0072,
+      "step": 15870
+    },
+    {
+      "epoch": 11.53,
+      "grad_norm": 0.06317915767431259,
+      "learning_rate": 2.3521342693455983e-05,
+      "loss": 0.0089,
+      "step": 15880
+    },
+    {
+      "epoch": 11.54,
+      "grad_norm": 0.20523525774478912,
+      "learning_rate": 2.3501170015331238e-05,
+      "loss": 0.0095,
+      "step": 15890
+    },
+    {
+      "epoch": 11.54,
+      "grad_norm": 0.0021271202713251114,
+      "learning_rate": 2.348099733720649e-05,
+      "loss": 0.0048,
+      "step": 15900
+    },
+    {
+      "epoch": 11.55,
+      "grad_norm": 0.35318243503570557,
+      "learning_rate": 2.346082465908174e-05,
+      "loss": 0.0221,
+      "step": 15910
+    },
+    {
+      "epoch": 11.56,
+      "grad_norm": 0.0006868810160085559,
+      "learning_rate": 2.3440651980956995e-05,
+      "loss": 0.0143,
+      "step": 15920
+    },
+    {
+      "epoch": 11.56,
+      "grad_norm": 8.075051307678223,
+      "learning_rate": 2.3420479302832246e-05,
+      "loss": 0.0014,
+      "step": 15930
+    },
+    {
+      "epoch": 11.57,
+      "grad_norm": 0.32641565799713135,
+      "learning_rate": 2.3400306624707497e-05,
+      "loss": 0.0079,
+      "step": 15940
+    },
+    {
+      "epoch": 11.58,
+      "grad_norm": 0.015668796375393867,
+      "learning_rate": 2.338013394658275e-05,
+      "loss": 0.0074,
+      "step": 15950
+    },
+    {
+      "epoch": 11.59,
+      "grad_norm": 0.0006520305760204792,
+      "learning_rate": 2.3359961268458002e-05,
+      "loss": 0.0031,
+      "step": 15960
+    },
+    {
+      "epoch": 11.59,
+      "grad_norm": 0.003270229557529092,
+      "learning_rate": 2.3339788590333253e-05,
+      "loss": 0.0027,
+      "step": 15970
+    },
+    {
+      "epoch": 11.6,
+      "grad_norm": 0.07816585153341293,
+      "learning_rate": 2.3319615912208508e-05,
+      "loss": 0.0081,
+      "step": 15980
+    },
+    {
+      "epoch": 11.61,
+      "grad_norm": 0.0006159089971333742,
+      "learning_rate": 2.329944323408376e-05,
+      "loss": 0.0089,
+      "step": 15990
+    },
+    {
+      "epoch": 11.62,
+      "grad_norm": 10.090989112854004,
+      "learning_rate": 2.327927055595901e-05,
+      "loss": 0.0036,
+      "step": 16000
+    },
+    {
+      "epoch": 11.62,
+      "grad_norm": 0.0013605119893327355,
+      "learning_rate": 2.3259097877834264e-05,
+      "loss": 0.0034,
+      "step": 16010
+    },
+    {
+      "epoch": 11.63,
+      "grad_norm": 0.001587074133567512,
+      "learning_rate": 2.3238925199709515e-05,
+      "loss": 0.0173,
+      "step": 16020
+    },
+    {
+      "epoch": 11.64,
+      "grad_norm": 0.042119793593883514,
+      "learning_rate": 2.3218752521584767e-05,
+      "loss": 0.0054,
+      "step": 16030
+    },
+    {
+      "epoch": 11.64,
+      "grad_norm": 0.0023439363576471806,
+      "learning_rate": 2.319857984346002e-05,
+      "loss": 0.0122,
+      "step": 16040
+    },
+    {
+      "epoch": 11.65,
+      "grad_norm": 0.0015862892614677548,
+      "learning_rate": 2.317840716533527e-05,
+      "loss": 0.0055,
+      "step": 16050
+    },
+    {
+      "epoch": 11.66,
+      "grad_norm": 0.04544699564576149,
+      "learning_rate": 2.3158234487210523e-05,
+      "loss": 0.0105,
+      "step": 16060
+    },
+    {
+      "epoch": 11.67,
+      "grad_norm": 0.005376008804887533,
+      "learning_rate": 2.3138061809085774e-05,
+      "loss": 0.0,
+      "step": 16070
+    },
+    {
+      "epoch": 11.67,
+      "grad_norm": 0.6874623894691467,
+      "learning_rate": 2.3117889130961025e-05,
+      "loss": 0.0042,
+      "step": 16080
+    },
+    {
+      "epoch": 11.68,
+      "grad_norm": 0.23841716349124908,
+      "learning_rate": 2.309771645283628e-05,
+      "loss": 0.0028,
+      "step": 16090
+    },
+    {
+      "epoch": 11.69,
+      "grad_norm": 0.046092964708805084,
+      "learning_rate": 2.307754377471153e-05,
+      "loss": 0.001,
+      "step": 16100
+    },
+    {
+      "epoch": 11.7,
+      "grad_norm": 0.0021304069086909294,
+      "learning_rate": 2.3057371096586782e-05,
+      "loss": 0.0005,
+      "step": 16110
+    },
+    {
+      "epoch": 11.7,
+      "grad_norm": 7.387868404388428,
+      "learning_rate": 2.3037198418462036e-05,
+      "loss": 0.0038,
+      "step": 16120
+    },
+    {
+      "epoch": 11.71,
+      "grad_norm": 0.0003391271748114377,
+      "learning_rate": 2.3017025740337287e-05,
+      "loss": 0.001,
+      "step": 16130
+    },
+    {
+      "epoch": 11.72,
+      "grad_norm": 0.28313395380973816,
+      "learning_rate": 2.299685306221254e-05,
+      "loss": 0.007,
+      "step": 16140
+    },
+    {
+      "epoch": 11.72,
+      "grad_norm": 0.04677537456154823,
+      "learning_rate": 2.2976680384087793e-05,
+      "loss": 0.0011,
+      "step": 16150
+    },
+    {
+      "epoch": 11.73,
+      "grad_norm": 0.15080393850803375,
+      "learning_rate": 2.2956507705963044e-05,
+      "loss": 0.0068,
+      "step": 16160
+    },
+    {
+      "epoch": 11.74,
+      "grad_norm": 0.0016375051345676184,
+      "learning_rate": 2.2936335027838295e-05,
+      "loss": 0.0045,
+      "step": 16170
+    },
+    {
+      "epoch": 11.75,
+      "grad_norm": 0.6318696141242981,
+      "learning_rate": 2.291616234971355e-05,
+      "loss": 0.015,
+      "step": 16180
+    },
+    {
+      "epoch": 11.75,
+      "grad_norm": 0.010398590005934238,
+      "learning_rate": 2.28959896715888e-05,
+      "loss": 0.0044,
+      "step": 16190
+    },
+    {
+      "epoch": 11.76,
+      "grad_norm": 6.571838855743408,
+      "learning_rate": 2.2875816993464052e-05,
+      "loss": 0.0289,
+      "step": 16200
+    },
+    {
+      "epoch": 11.77,
+      "grad_norm": 0.23911860585212708,
+      "learning_rate": 2.2855644315339306e-05,
+      "loss": 0.0108,
+      "step": 16210
+    },
+    {
+      "epoch": 11.77,
+      "grad_norm": 0.0554070845246315,
+      "learning_rate": 2.2835471637214557e-05,
+      "loss": 0.0159,
+      "step": 16220
+    },
+    {
+      "epoch": 11.78,
+      "grad_norm": 0.001894534332677722,
+      "learning_rate": 2.281529895908981e-05,
+      "loss": 0.0006,
+      "step": 16230
+    },
+    {
+      "epoch": 11.79,
+      "grad_norm": 0.11737006157636642,
+      "learning_rate": 2.2795126280965063e-05,
+      "loss": 0.003,
+      "step": 16240
+    },
+    {
+      "epoch": 11.8,
+      "grad_norm": 0.0013864507200196385,
+      "learning_rate": 2.2774953602840314e-05,
+      "loss": 0.0022,
+      "step": 16250
+    },
+    {
+      "epoch": 11.8,
+      "grad_norm": 0.004860733635723591,
+      "learning_rate": 2.2754780924715565e-05,
+      "loss": 0.0017,
+      "step": 16260
+    },
+    {
+      "epoch": 11.81,
+      "grad_norm": 0.000366124149877578,
+      "learning_rate": 2.273460824659082e-05,
+      "loss": 0.0073,
+      "step": 16270
+    },
+    {
+      "epoch": 11.82,
+      "grad_norm": 0.09364797174930573,
+      "learning_rate": 2.271443556846607e-05,
+      "loss": 0.0087,
+      "step": 16280
+    },
+    {
+      "epoch": 11.83,
+      "grad_norm": 0.02469642087817192,
+      "learning_rate": 2.269426289034132e-05,
+      "loss": 0.0099,
+      "step": 16290
+    },
+    {
+      "epoch": 11.83,
+      "grad_norm": 5.18701696395874,
+      "learning_rate": 2.2674090212216576e-05,
+      "loss": 0.0165,
+      "step": 16300
+    },
+    {
+      "epoch": 11.84,
+      "grad_norm": 0.10657285153865814,
+      "learning_rate": 2.2653917534091827e-05,
+      "loss": 0.0028,
+      "step": 16310
+    },
+    {
+      "epoch": 11.85,
+      "grad_norm": 0.03622545674443245,
+      "learning_rate": 2.2633744855967078e-05,
+      "loss": 0.0017,
+      "step": 16320
+    },
+    {
+      "epoch": 11.85,
+      "grad_norm": 0.005775143392384052,
+      "learning_rate": 2.2613572177842333e-05,
+      "loss": 0.0023,
+      "step": 16330
+    },
+    {
+      "epoch": 11.86,
+      "grad_norm": 0.07602769881486893,
+      "learning_rate": 2.2593399499717584e-05,
+      "loss": 0.0167,
+      "step": 16340
+    },
+    {
+      "epoch": 11.87,
+      "grad_norm": 0.043435435742139816,
+      "learning_rate": 2.2573226821592835e-05,
+      "loss": 0.0031,
+      "step": 16350
+    },
+    {
+      "epoch": 11.88,
+      "grad_norm": 0.002995203249156475,
+      "learning_rate": 2.255305414346809e-05,
+      "loss": 0.0255,
+      "step": 16360
+    },
+    {
+      "epoch": 11.88,
+      "grad_norm": 0.024543453007936478,
+      "learning_rate": 2.253288146534334e-05,
+      "loss": 0.0191,
+      "step": 16370
+    },
+    {
+      "epoch": 11.89,
+      "grad_norm": 0.10300853848457336,
+      "learning_rate": 2.251270878721859e-05,
+      "loss": 0.0036,
+      "step": 16380
+    },
+    {
+      "epoch": 11.9,
+      "grad_norm": 0.009681067429482937,
+      "learning_rate": 2.2492536109093846e-05,
+      "loss": 0.0027,
+      "step": 16390
+    },
+    {
+      "epoch": 11.91,
+      "grad_norm": 0.030278483405709267,
+      "learning_rate": 2.2472363430969097e-05,
+      "loss": 0.0013,
+      "step": 16400
+    },
+    {
+      "epoch": 11.91,
+      "grad_norm": 0.1338738054037094,
+      "learning_rate": 2.2452190752844348e-05,
+      "loss": 0.0014,
+      "step": 16410
+    },
+    {
+      "epoch": 11.92,
+      "grad_norm": 0.1927368938922882,
+      "learning_rate": 2.2432018074719603e-05,
+      "loss": 0.0078,
+      "step": 16420
+    },
+    {
+      "epoch": 11.93,
+      "grad_norm": 0.06984108686447144,
+      "learning_rate": 2.2411845396594854e-05,
+      "loss": 0.0078,
+      "step": 16430
+    },
+    {
+      "epoch": 11.93,
+      "grad_norm": 0.19020429253578186,
+      "learning_rate": 2.2391672718470105e-05,
+      "loss": 0.004,
+      "step": 16440
+    },
+    {
+      "epoch": 11.94,
+      "grad_norm": 0.23069943487644196,
+      "learning_rate": 2.237150004034536e-05,
+      "loss": 0.0034,
+      "step": 16450
+    },
+    {
+      "epoch": 11.95,
+      "grad_norm": 1.717404842376709,
+      "learning_rate": 2.235132736222061e-05,
+      "loss": 0.0058,
+      "step": 16460
+    },
+    {
+      "epoch": 11.96,
+      "grad_norm": 13.391761779785156,
+      "learning_rate": 2.233115468409586e-05,
+      "loss": 0.0138,
+      "step": 16470
+    },
+    {
+      "epoch": 11.96,
+      "grad_norm": 0.0007910021813586354,
+      "learning_rate": 2.2310982005971116e-05,
+      "loss": 0.0015,
+      "step": 16480
+    },
+    {
+      "epoch": 11.97,
+      "grad_norm": 0.0008350893040187657,
+      "learning_rate": 2.2290809327846367e-05,
+      "loss": 0.0077,
+      "step": 16490
+    },
+    {
+      "epoch": 11.98,
+      "grad_norm": 0.005978062283247709,
+      "learning_rate": 2.2270636649721618e-05,
+      "loss": 0.009,
+      "step": 16500
+    },
+    {
+      "epoch": 11.99,
+      "grad_norm": 0.059390950947999954,
+      "learning_rate": 2.2250463971596872e-05,
+      "loss": 0.0029,
+      "step": 16510
+    },
+    {
+      "epoch": 11.99,
+      "grad_norm": 0.384694904088974,
+      "learning_rate": 2.2230291293472124e-05,
+      "loss": 0.0059,
+      "step": 16520
+    },
+    {
+      "epoch": 12.0,
+      "grad_norm": 9.535301208496094,
+      "learning_rate": 2.2210118615347375e-05,
+      "loss": 0.0142,
+      "step": 16530
+    },
+    {
+      "epoch": 12.0,
+      "eval_accuracy": 0.9982189045695875,
+      "eval_f1": 0.9958938145678043,
+      "eval_loss": 0.003976278472691774,
+      "eval_precision": 0.997903454059437,
+      "eval_recall": 0.993892253079975,
+      "eval_roc_auc": 0.9999848726260968,
+      "eval_runtime": 386.255,
+      "eval_samples_per_second": 228.212,
+      "eval_steps_per_second": 14.265,
+      "step": 16530
+    },
+    {
+      "epoch": 12.01,
+      "grad_norm": 0.000495292479172349,
+      "learning_rate": 2.2189945937222626e-05,
+      "loss": 0.0122,
+      "step": 16540
+    },
+    {
+      "epoch": 12.01,
+      "grad_norm": 0.10421086847782135,
+      "learning_rate": 2.216977325909788e-05,
+      "loss": 0.0062,
+      "step": 16550
+    },
+    {
+      "epoch": 12.02,
+      "grad_norm": 0.003116333158686757,
+      "learning_rate": 2.214960058097313e-05,
+      "loss": 0.0042,
+      "step": 16560
+    },
+    {
+      "epoch": 12.03,
+      "grad_norm": 0.049882255494594574,
+      "learning_rate": 2.2129427902848382e-05,
+      "loss": 0.0033,
+      "step": 16570
+    },
+    {
+      "epoch": 12.04,
+      "grad_norm": 0.006924469955265522,
+      "learning_rate": 2.2109255224723637e-05,
+      "loss": 0.0145,
+      "step": 16580
+    },
+    {
+      "epoch": 12.04,
+      "grad_norm": 1.2397273778915405,
+      "learning_rate": 2.2089082546598888e-05,
+      "loss": 0.0192,
+      "step": 16590
+    },
+    {
+      "epoch": 12.05,
+      "grad_norm": 0.001038398826494813,
+      "learning_rate": 2.206890986847414e-05,
+      "loss": 0.0038,
+      "step": 16600
+    },
+    {
+      "epoch": 12.06,
+      "grad_norm": 2.1774699687957764,
+      "learning_rate": 2.204873719034939e-05,
+      "loss": 0.0145,
+      "step": 16610
+    },
+    {
+      "epoch": 12.07,
+      "grad_norm": 0.2245302051305771,
+      "learning_rate": 2.2028564512224644e-05,
+      "loss": 0.0072,
+      "step": 16620
+    },
+    {
+      "epoch": 12.07,
+      "grad_norm": 0.0035848254337906837,
+      "learning_rate": 2.2008391834099896e-05,
+      "loss": 0.0074,
+      "step": 16630
+    },
+    {
+      "epoch": 12.08,
+      "grad_norm": 0.0033830467145889997,
+      "learning_rate": 2.1988219155975147e-05,
+      "loss": 0.0019,
+      "step": 16640
+    },
+    {
+      "epoch": 12.09,
+      "grad_norm": 0.0030913010705262423,
+      "learning_rate": 2.19680464778504e-05,
+      "loss": 0.007,
+      "step": 16650
+    },
+    {
+      "epoch": 12.09,
+      "grad_norm": 11.367694854736328,
+      "learning_rate": 2.1947873799725652e-05,
+      "loss": 0.0155,
+      "step": 16660
+    },
+    {
+      "epoch": 12.1,
+      "grad_norm": 0.0011047075968235731,
+      "learning_rate": 2.1927701121600903e-05,
+      "loss": 0.0087,
+      "step": 16670
+    },
+    {
+      "epoch": 12.11,
+      "grad_norm": 0.018373142927885056,
+      "learning_rate": 2.1907528443476158e-05,
+      "loss": 0.0004,
+      "step": 16680
+    },
+    {
+      "epoch": 12.12,
+      "grad_norm": 0.0015459812711924314,
+      "learning_rate": 2.188735576535141e-05,
+      "loss": 0.0121,
+      "step": 16690
+    },
+    {
+      "epoch": 12.12,
+      "grad_norm": 0.09375672042369843,
+      "learning_rate": 2.186718308722666e-05,
+      "loss": 0.0195,
+      "step": 16700
+    },
+    {
+      "epoch": 12.13,
+      "grad_norm": 0.009158155880868435,
+      "learning_rate": 2.1847010409101914e-05,
+      "loss": 0.0024,
+      "step": 16710
+    },
+    {
+      "epoch": 12.14,
+      "grad_norm": 0.03295760229229927,
+      "learning_rate": 2.1826837730977165e-05,
+      "loss": 0.0081,
+      "step": 16720
+    },
+    {
+      "epoch": 12.15,
+      "grad_norm": 0.030867867171764374,
+      "learning_rate": 2.1806665052852416e-05,
+      "loss": 0.0065,
+      "step": 16730
+    },
+    {
+      "epoch": 12.15,
+      "grad_norm": 0.08418185263872147,
+      "learning_rate": 2.178649237472767e-05,
+      "loss": 0.0027,
+      "step": 16740
+    },
+    {
+      "epoch": 12.16,
+      "grad_norm": 0.006285299547016621,
+      "learning_rate": 2.1766319696602922e-05,
+      "loss": 0.0081,
+      "step": 16750
+    },
+    {
+      "epoch": 12.17,
+      "grad_norm": 0.0018992135301232338,
+      "learning_rate": 2.1746147018478173e-05,
+      "loss": 0.0171,
+      "step": 16760
+    },
+    {
+      "epoch": 12.17,
+      "grad_norm": 0.3470768332481384,
+      "learning_rate": 2.1725974340353428e-05,
+      "loss": 0.0004,
+      "step": 16770
+    },
+    {
+      "epoch": 12.18,
+      "grad_norm": 0.04968973249197006,
+      "learning_rate": 2.170580166222868e-05,
+      "loss": 0.0041,
+      "step": 16780
+    },
+    {
+      "epoch": 12.19,
+      "grad_norm": 0.04065666347742081,
+      "learning_rate": 2.168562898410393e-05,
+      "loss": 0.0064,
+      "step": 16790
+    },
+    {
+      "epoch": 12.2,
+      "grad_norm": 0.6985291838645935,
+      "learning_rate": 2.166545630597918e-05,
+      "loss": 0.0129,
+      "step": 16800
+    },
+    {
+      "epoch": 12.2,
+      "grad_norm": 0.01881437934935093,
+      "learning_rate": 2.1645283627854435e-05,
+      "loss": 0.0119,
+      "step": 16810
+    },
+    {
+      "epoch": 12.21,
+      "grad_norm": 0.04512902721762657,
+      "learning_rate": 2.1625110949729686e-05,
+      "loss": 0.0052,
+      "step": 16820
+    },
+    {
+      "epoch": 12.22,
+      "grad_norm": 0.0031053638085722923,
+      "learning_rate": 2.1604938271604937e-05,
+      "loss": 0.0074,
+      "step": 16830
+    },
+    {
+      "epoch": 12.23,
+      "grad_norm": 0.13527894020080566,
+      "learning_rate": 2.1584765593480192e-05,
+      "loss": 0.005,
+      "step": 16840
+    },
+    {
+      "epoch": 12.23,
+      "grad_norm": 0.027386058121919632,
+      "learning_rate": 2.1564592915355443e-05,
+      "loss": 0.0018,
+      "step": 16850
+    },
+    {
+      "epoch": 12.24,
+      "grad_norm": 0.002343888161703944,
+      "learning_rate": 2.1544420237230694e-05,
+      "loss": 0.0065,
+      "step": 16860
+    },
+    {
+      "epoch": 12.25,
+      "grad_norm": 0.002938592340797186,
+      "learning_rate": 2.152424755910595e-05,
+      "loss": 0.0,
+      "step": 16870
+    },
+    {
+      "epoch": 12.25,
+      "grad_norm": 0.0012315199710428715,
+      "learning_rate": 2.15040748809812e-05,
+      "loss": 0.0007,
+      "step": 16880
+    },
+    {
+      "epoch": 12.26,
+      "grad_norm": 0.0013336024712771177,
+      "learning_rate": 2.148390220285645e-05,
+      "loss": 0.0021,
+      "step": 16890
+    },
+    {
+      "epoch": 12.27,
+      "grad_norm": 0.11299686878919601,
+      "learning_rate": 2.1463729524731705e-05,
+      "loss": 0.01,
+      "step": 16900
+    },
+    {
+      "epoch": 12.28,
+      "grad_norm": 20.187990188598633,
+      "learning_rate": 2.1443556846606956e-05,
+      "loss": 0.0147,
+      "step": 16910
+    },
+    {
+      "epoch": 12.28,
+      "grad_norm": 0.3132416903972626,
+      "learning_rate": 2.1423384168482207e-05,
+      "loss": 0.0086,
+      "step": 16920
+    },
+    {
+      "epoch": 12.29,
+      "grad_norm": 0.0373138003051281,
+      "learning_rate": 2.1403211490357462e-05,
+      "loss": 0.0147,
+      "step": 16930
+    },
+    {
+      "epoch": 12.3,
+      "grad_norm": 0.0028076451271772385,
+      "learning_rate": 2.1383038812232713e-05,
+      "loss": 0.0013,
+      "step": 16940
+    },
+    {
+      "epoch": 12.3,
+      "grad_norm": 0.2017146497964859,
+      "learning_rate": 2.1362866134107964e-05,
+      "loss": 0.0276,
+      "step": 16950
+    },
+    {
+      "epoch": 12.31,
+      "grad_norm": 0.0041721658781170845,
+      "learning_rate": 2.134269345598322e-05,
+      "loss": 0.0106,
+      "step": 16960
+    },
+    {
+      "epoch": 12.32,
+      "grad_norm": 0.19767874479293823,
+      "learning_rate": 2.132252077785847e-05,
+      "loss": 0.0129,
+      "step": 16970
+    },
+    {
+      "epoch": 12.33,
+      "grad_norm": 0.006283420603722334,
+      "learning_rate": 2.130234809973372e-05,
+      "loss": 0.0035,
+      "step": 16980
+    },
+    {
+      "epoch": 12.33,
+      "grad_norm": 0.002427824307233095,
+      "learning_rate": 2.1282175421608975e-05,
+      "loss": 0.0032,
+      "step": 16990
+    },
+    {
+      "epoch": 12.34,
+      "grad_norm": 0.07163175940513611,
+      "learning_rate": 2.1262002743484226e-05,
+      "loss": 0.0042,
+      "step": 17000
+    },
+    {
+      "epoch": 12.35,
+      "grad_norm": 0.060925085097551346,
+      "learning_rate": 2.1241830065359477e-05,
+      "loss": 0.0082,
+      "step": 17010
+    },
+    {
+      "epoch": 12.36,
+      "grad_norm": 0.02180366963148117,
+      "learning_rate": 2.122165738723473e-05,
+      "loss": 0.0035,
+      "step": 17020
+    },
+    {
+      "epoch": 12.36,
+      "grad_norm": 11.230984687805176,
+      "learning_rate": 2.1201484709109983e-05,
+      "loss": 0.0131,
+      "step": 17030
+    },
+    {
+      "epoch": 12.37,
+      "grad_norm": 0.0020128083415329456,
+      "learning_rate": 2.1181312030985234e-05,
+      "loss": 0.002,
+      "step": 17040
+    },
+    {
+      "epoch": 12.38,
+      "grad_norm": 18.336151123046875,
+      "learning_rate": 2.1161139352860488e-05,
+      "loss": 0.0039,
+      "step": 17050
+    },
+    {
+      "epoch": 12.38,
+      "grad_norm": 0.0012726852437481284,
+      "learning_rate": 2.114096667473574e-05,
+      "loss": 0.0022,
+      "step": 17060
+    },
+    {
+      "epoch": 12.39,
+      "grad_norm": 0.005078029818832874,
+      "learning_rate": 2.112079399661099e-05,
+      "loss": 0.0,
+      "step": 17070
+    },
+    {
+      "epoch": 12.4,
+      "grad_norm": 0.0027781687676906586,
+      "learning_rate": 2.1100621318486245e-05,
+      "loss": 0.0038,
+      "step": 17080
+    },
+    {
+      "epoch": 12.41,
+      "grad_norm": 0.002532773185521364,
+      "learning_rate": 2.1080448640361496e-05,
+      "loss": 0.0079,
+      "step": 17090
+    },
+    {
+      "epoch": 12.41,
+      "grad_norm": 0.006265075411647558,
+      "learning_rate": 2.1060275962236747e-05,
+      "loss": 0.007,
+      "step": 17100
+    },
+    {
+      "epoch": 12.42,
+      "grad_norm": 14.577162742614746,
+      "learning_rate": 2.1040103284112e-05,
+      "loss": 0.007,
+      "step": 17110
+    },
+    {
+      "epoch": 12.43,
+      "grad_norm": 0.0002733594155870378,
+      "learning_rate": 2.101993060598725e-05,
+      "loss": 0.0069,
+      "step": 17120
+    },
+    {
+      "epoch": 12.44,
+      "grad_norm": 0.015362723730504513,
+      "learning_rate": 2.0999757927862504e-05,
+      "loss": 0.0023,
+      "step": 17130
+    },
+    {
+      "epoch": 12.44,
+      "grad_norm": 0.17776289582252502,
+      "learning_rate": 2.0979585249737758e-05,
+      "loss": 0.0211,
+      "step": 17140
+    },
+    {
+      "epoch": 12.45,
+      "grad_norm": 0.10553798824548721,
+      "learning_rate": 2.0959412571613006e-05,
+      "loss": 0.0032,
+      "step": 17150
+    },
+    {
+      "epoch": 12.46,
+      "grad_norm": 2.3914291858673096,
+      "learning_rate": 2.093923989348826e-05,
+      "loss": 0.0055,
+      "step": 17160
+    },
+    {
+      "epoch": 12.46,
+      "grad_norm": 0.013143391348421574,
+      "learning_rate": 2.0919067215363515e-05,
+      "loss": 0.0106,
+      "step": 17170
+    },
+    {
+      "epoch": 12.47,
+      "grad_norm": 0.07704094797372818,
+      "learning_rate": 2.0898894537238762e-05,
+      "loss": 0.0082,
+      "step": 17180
+    },
+    {
+      "epoch": 12.48,
+      "grad_norm": 0.0821012333035469,
+      "learning_rate": 2.0878721859114017e-05,
+      "loss": 0.0055,
+      "step": 17190
+    },
+    {
+      "epoch": 12.49,
+      "grad_norm": 0.0025505193043500185,
+      "learning_rate": 2.085854918098927e-05,
+      "loss": 0.0074,
+      "step": 17200
+    },
+    {
+      "epoch": 12.49,
+      "grad_norm": 0.154500812292099,
+      "learning_rate": 2.083837650286452e-05,
+      "loss": 0.0037,
+      "step": 17210
+    },
+    {
+      "epoch": 12.5,
+      "grad_norm": 0.0007235849625431001,
+      "learning_rate": 2.0818203824739773e-05,
+      "loss": 0.0025,
+      "step": 17220
+    },
+    {
+      "epoch": 12.51,
+      "grad_norm": 0.10713616758584976,
+      "learning_rate": 2.0798031146615028e-05,
+      "loss": 0.0096,
+      "step": 17230
+    },
+    {
+      "epoch": 12.52,
+      "grad_norm": 0.00021158994059078395,
+      "learning_rate": 2.0777858468490276e-05,
+      "loss": 0.0017,
+      "step": 17240
+    },
+    {
+      "epoch": 12.52,
+      "grad_norm": 0.00012255563342478126,
+      "learning_rate": 2.075768579036553e-05,
+      "loss": 0.0015,
+      "step": 17250
+    },
+    {
+      "epoch": 12.53,
+      "grad_norm": 0.04050791636109352,
+      "learning_rate": 2.0737513112240785e-05,
+      "loss": 0.0009,
+      "step": 17260
+    },
+    {
+      "epoch": 12.54,
+      "grad_norm": 0.12903136014938354,
+      "learning_rate": 2.0717340434116032e-05,
+      "loss": 0.0036,
+      "step": 17270
+    },
+    {
+      "epoch": 12.54,
+      "grad_norm": 0.0034476914443075657,
+      "learning_rate": 2.0697167755991287e-05,
+      "loss": 0.014,
+      "step": 17280
+    },
+    {
+      "epoch": 12.55,
+      "grad_norm": 1.9991146326065063,
+      "learning_rate": 2.0676995077866538e-05,
+      "loss": 0.0057,
+      "step": 17290
+    },
+    {
+      "epoch": 12.56,
+      "grad_norm": 0.5727086067199707,
+      "learning_rate": 2.065682239974179e-05,
+      "loss": 0.0048,
+      "step": 17300
+    },
+    {
+      "epoch": 12.57,
+      "grad_norm": 0.027912188321352005,
+      "learning_rate": 2.0636649721617043e-05,
+      "loss": 0.0036,
+      "step": 17310
+    },
+    {
+      "epoch": 12.57,
+      "grad_norm": 0.000993537250906229,
+      "learning_rate": 2.0616477043492294e-05,
+      "loss": 0.0001,
+      "step": 17320
+    },
+    {
+      "epoch": 12.58,
+      "grad_norm": 0.00026250266819261014,
+      "learning_rate": 2.0596304365367546e-05,
+      "loss": 0.0097,
+      "step": 17330
+    },
+    {
+      "epoch": 12.59,
+      "grad_norm": 0.00026831813738681376,
+      "learning_rate": 2.05761316872428e-05,
+      "loss": 0.0051,
+      "step": 17340
+    },
+    {
+      "epoch": 12.6,
+      "grad_norm": 0.0001645336305955425,
+      "learning_rate": 2.055595900911805e-05,
+      "loss": 0.0054,
+      "step": 17350
+    },
+    {
+      "epoch": 12.6,
+      "grad_norm": 0.0002212459803558886,
+      "learning_rate": 2.0535786330993302e-05,
+      "loss": 0.0276,
+      "step": 17360
+    },
+    {
+      "epoch": 12.61,
+      "grad_norm": 0.002810519188642502,
+      "learning_rate": 2.0515613652868557e-05,
+      "loss": 0.0324,
+      "step": 17370
+    },
+    {
+      "epoch": 12.62,
+      "grad_norm": 0.0005311826826073229,
+      "learning_rate": 2.0495440974743808e-05,
+      "loss": 0.0031,
+      "step": 17380
+    },
+    {
+      "epoch": 12.62,
+      "grad_norm": 0.14589981734752655,
+      "learning_rate": 2.047526829661906e-05,
+      "loss": 0.0081,
+      "step": 17390
+    },
+    {
+      "epoch": 12.63,
+      "grad_norm": 0.0009502907050773501,
+      "learning_rate": 2.0455095618494313e-05,
+      "loss": 0.0058,
+      "step": 17400
+    },
+    {
+      "epoch": 12.64,
+      "grad_norm": 0.054477520287036896,
+      "learning_rate": 2.0434922940369564e-05,
+      "loss": 0.0018,
+      "step": 17410
+    },
+    {
+      "epoch": 12.65,
+      "grad_norm": 0.061193205416202545,
+      "learning_rate": 2.0414750262244815e-05,
+      "loss": 0.0022,
+      "step": 17420
+    },
+    {
+      "epoch": 12.65,
+      "grad_norm": 1.0594874620437622,
+      "learning_rate": 2.039457758412007e-05,
+      "loss": 0.0064,
+      "step": 17430
+    },
+    {
+      "epoch": 12.66,
+      "grad_norm": 0.0070306709967553616,
+      "learning_rate": 2.037440490599532e-05,
+      "loss": 0.0019,
+      "step": 17440
+    },
+    {
+      "epoch": 12.67,
+      "grad_norm": 0.0007211442571133375,
+      "learning_rate": 2.0354232227870572e-05,
+      "loss": 0.0055,
+      "step": 17450
+    },
+    {
+      "epoch": 12.68,
+      "grad_norm": 0.9333264231681824,
+      "learning_rate": 2.0334059549745826e-05,
+      "loss": 0.0186,
+      "step": 17460
+    },
+    {
+      "epoch": 12.68,
+      "grad_norm": 0.0005025692516937852,
+      "learning_rate": 2.0313886871621078e-05,
+      "loss": 0.0021,
+      "step": 17470
+    },
+    {
+      "epoch": 12.69,
+      "grad_norm": 0.002434327732771635,
+      "learning_rate": 2.029371419349633e-05,
+      "loss": 0.0042,
+      "step": 17480
+    },
+    {
+      "epoch": 12.7,
+      "grad_norm": 0.030645696446299553,
+      "learning_rate": 2.0273541515371583e-05,
+      "loss": 0.0039,
+      "step": 17490
+    },
+    {
+      "epoch": 12.7,
+      "grad_norm": 0.006995252333581448,
+      "learning_rate": 2.0253368837246834e-05,
+      "loss": 0.0182,
+      "step": 17500
+    },
+    {
+      "epoch": 12.71,
+      "grad_norm": 36.959693908691406,
+      "learning_rate": 2.0233196159122085e-05,
+      "loss": 0.0159,
+      "step": 17510
+    },
+    {
+      "epoch": 12.72,
+      "grad_norm": 0.05816148966550827,
+      "learning_rate": 2.021302348099734e-05,
+      "loss": 0.0136,
+      "step": 17520
+    },
+    {
+      "epoch": 12.73,
+      "grad_norm": 0.4665600061416626,
+      "learning_rate": 2.019285080287259e-05,
+      "loss": 0.0167,
+      "step": 17530
+    },
+    {
+      "epoch": 12.73,
+      "grad_norm": 0.008300243876874447,
+      "learning_rate": 2.0172678124747842e-05,
+      "loss": 0.0069,
+      "step": 17540
+    },
+    {
+      "epoch": 12.74,
+      "grad_norm": 0.026556752622127533,
+      "learning_rate": 2.0152505446623093e-05,
+      "loss": 0.012,
+      "step": 17550
+    },
+    {
+      "epoch": 12.75,
+      "grad_norm": 0.00357626099139452,
+      "learning_rate": 2.0132332768498347e-05,
+      "loss": 0.0114,
+      "step": 17560
+    },
+    {
+      "epoch": 12.75,
+      "grad_norm": 0.22881974279880524,
+      "learning_rate": 2.01121600903736e-05,
+      "loss": 0.0039,
+      "step": 17570
+    },
+    {
+      "epoch": 12.76,
+      "grad_norm": 0.1387658715248108,
+      "learning_rate": 2.009198741224885e-05,
+      "loss": 0.001,
+      "step": 17580
+    },
+    {
+      "epoch": 12.77,
+      "grad_norm": 0.0008824353571981192,
+      "learning_rate": 2.0071814734124104e-05,
+      "loss": 0.0039,
+      "step": 17590
+    },
+    {
+      "epoch": 12.78,
+      "grad_norm": 39.46518325805664,
+      "learning_rate": 2.0051642055999355e-05,
+      "loss": 0.0112,
+      "step": 17600
+    },
+    {
+      "epoch": 12.78,
+      "grad_norm": 0.006979393772780895,
+      "learning_rate": 2.0031469377874606e-05,
+      "loss": 0.0065,
+      "step": 17610
+    },
+    {
+      "epoch": 12.79,
+      "grad_norm": 0.0024195117875933647,
+      "learning_rate": 2.001129669974986e-05,
+      "loss": 0.0006,
+      "step": 17620
+    },
+    {
+      "epoch": 12.8,
+      "grad_norm": 0.0011451839236542583,
+      "learning_rate": 1.9991124021625112e-05,
+      "loss": 0.0136,
+      "step": 17630
+    },
+    {
+      "epoch": 12.81,
+      "grad_norm": 0.06088784337043762,
+      "learning_rate": 1.9970951343500363e-05,
+      "loss": 0.0054,
+      "step": 17640
+    },
+    {
+      "epoch": 12.81,
+      "grad_norm": 0.0037643101532012224,
+      "learning_rate": 1.9950778665375617e-05,
+      "loss": 0.0045,
+      "step": 17650
+    },
+    {
+      "epoch": 12.82,
+      "grad_norm": 0.06409800797700882,
+      "learning_rate": 1.993060598725087e-05,
+      "loss": 0.0015,
+      "step": 17660
+    },
+    {
+      "epoch": 12.83,
+      "grad_norm": 0.010507245548069477,
+      "learning_rate": 1.991043330912612e-05,
+      "loss": 0.006,
+      "step": 17670
+    },
+    {
+      "epoch": 12.83,
+      "grad_norm": 3.360480785369873,
+      "learning_rate": 1.9890260631001374e-05,
+      "loss": 0.0016,
+      "step": 17680
+    },
+    {
+      "epoch": 12.84,
+      "grad_norm": 0.10210923850536346,
+      "learning_rate": 1.9870087952876625e-05,
+      "loss": 0.0131,
+      "step": 17690
+    },
+    {
+      "epoch": 12.85,
+      "grad_norm": 0.00693171750754118,
+      "learning_rate": 1.9849915274751876e-05,
+      "loss": 0.0089,
+      "step": 17700
+    },
+    {
+      "epoch": 12.86,
+      "grad_norm": 0.0036455222871154547,
+      "learning_rate": 1.982974259662713e-05,
+      "loss": 0.006,
+      "step": 17710
+    },
+    {
+      "epoch": 12.86,
+      "grad_norm": 16.66859245300293,
+      "learning_rate": 1.980956991850238e-05,
+      "loss": 0.0193,
+      "step": 17720
+    },
+    {
+      "epoch": 12.87,
+      "grad_norm": 0.0002460737305227667,
+      "learning_rate": 1.9789397240377633e-05,
+      "loss": 0.0047,
+      "step": 17730
+    },
+    {
+      "epoch": 12.88,
+      "grad_norm": 0.002102719387039542,
+      "learning_rate": 1.9769224562252887e-05,
+      "loss": 0.0202,
+      "step": 17740
+    },
+    {
+      "epoch": 12.89,
+      "grad_norm": 27.401649475097656,
+      "learning_rate": 1.9749051884128138e-05,
+      "loss": 0.0046,
+      "step": 17750
+    },
+    {
+      "epoch": 12.89,
+      "grad_norm": 0.007228259928524494,
+      "learning_rate": 1.972887920600339e-05,
+      "loss": 0.0047,
+      "step": 17760
+    },
+    {
+      "epoch": 12.9,
+      "grad_norm": 0.25575822591781616,
+      "learning_rate": 1.9708706527878644e-05,
+      "loss": 0.0009,
+      "step": 17770
+    },
+    {
+      "epoch": 12.91,
+      "grad_norm": 0.9766779541969299,
+      "learning_rate": 1.968853384975389e-05,
+      "loss": 0.0113,
+      "step": 17780
+    },
+    {
+      "epoch": 12.91,
+      "grad_norm": 0.07483760267496109,
+      "learning_rate": 1.9668361171629146e-05,
+      "loss": 0.0107,
+      "step": 17790
+    },
+    {
+      "epoch": 12.92,
+      "grad_norm": 0.0007129976293072104,
+      "learning_rate": 1.96481884935044e-05,
+      "loss": 0.0118,
+      "step": 17800
+    },
+    {
+      "epoch": 12.93,
+      "grad_norm": 0.0688149631023407,
+      "learning_rate": 1.9628015815379648e-05,
+      "loss": 0.0131,
+      "step": 17810
+    },
+    {
+      "epoch": 12.94,
+      "grad_norm": 0.04838823899626732,
+      "learning_rate": 1.9607843137254903e-05,
+      "loss": 0.0012,
+      "step": 17820
+    },
+    {
+      "epoch": 12.94,
+      "grad_norm": 0.0014652046374976635,
+      "learning_rate": 1.9587670459130157e-05,
+      "loss": 0.0106,
+      "step": 17830
+    },
+    {
+      "epoch": 12.95,
+      "grad_norm": 0.0027166057843714952,
+      "learning_rate": 1.9567497781005405e-05,
+      "loss": 0.007,
+      "step": 17840
+    },
+    {
+      "epoch": 12.96,
+      "grad_norm": 0.0015533966943621635,
+      "learning_rate": 1.954732510288066e-05,
+      "loss": 0.0032,
+      "step": 17850
+    },
+    {
+      "epoch": 12.97,
+      "grad_norm": 0.05134255439043045,
+      "learning_rate": 1.9527152424755914e-05,
+      "loss": 0.0028,
+      "step": 17860
+    },
+    {
+      "epoch": 12.97,
+      "grad_norm": 0.10609173774719238,
+      "learning_rate": 1.950697974663116e-05,
+      "loss": 0.0023,
+      "step": 17870
+    },
+    {
+      "epoch": 12.98,
+      "grad_norm": 0.17367728054523468,
+      "learning_rate": 1.9486807068506416e-05,
+      "loss": 0.0087,
+      "step": 17880
+    },
+    {
+      "epoch": 12.99,
+      "grad_norm": 0.16967636346817017,
+      "learning_rate": 1.946663439038167e-05,
+      "loss": 0.0098,
+      "step": 17890
+    },
+    {
+      "epoch": 12.99,
+      "grad_norm": 0.059492193162441254,
+      "learning_rate": 1.9446461712256918e-05,
+      "loss": 0.0058,
+      "step": 17900
+    },
+    {
+      "epoch": 13.0,
+      "eval_accuracy": 0.9983096610246404,
+      "eval_f1": 0.9960987615531642,
+      "eval_loss": 0.0034948072861880064,
+      "eval_precision": 0.9992120607238536,
+      "eval_recall": 0.9930048026727918,
+      "eval_roc_auc": 0.9999857662326282,
+      "eval_runtime": 385.943,
+      "eval_samples_per_second": 228.396,
+      "eval_steps_per_second": 14.277,
+      "step": 17907
+    },
+    {
+      "epoch": 13.0,
+      "grad_norm": 0.0003739891981240362,
+      "learning_rate": 1.9426289034132172e-05,
+      "loss": 0.013,
+      "step": 17910
+    },
+    {
+      "epoch": 13.01,
+      "grad_norm": 0.0007324207108467817,
+      "learning_rate": 1.9406116356007427e-05,
+      "loss": 0.0059,
+      "step": 17920
+    },
+    {
+      "epoch": 13.02,
+      "grad_norm": 0.20080474019050598,
+      "learning_rate": 1.9385943677882675e-05,
+      "loss": 0.0055,
+      "step": 17930
+    },
+    {
+      "epoch": 13.02,
+      "grad_norm": 0.1485166698694229,
+      "learning_rate": 1.936577099975793e-05,
+      "loss": 0.0103,
+      "step": 17940
+    },
+    {
+      "epoch": 13.03,
+      "grad_norm": 0.0007888617110438645,
+      "learning_rate": 1.9345598321633183e-05,
+      "loss": 0.0026,
+      "step": 17950
+    },
+    {
+      "epoch": 13.04,
+      "grad_norm": 1.7149608135223389,
+      "learning_rate": 1.932542564350843e-05,
+      "loss": 0.0041,
+      "step": 17960
+    },
+    {
+      "epoch": 13.05,
+      "grad_norm": 0.19555199146270752,
+      "learning_rate": 1.9305252965383686e-05,
+      "loss": 0.0126,
+      "step": 17970
+    },
+    {
+      "epoch": 13.05,
+      "grad_norm": 0.0028395382687449455,
+      "learning_rate": 1.928508028725894e-05,
+      "loss": 0.0007,
+      "step": 17980
+    },
+    {
+      "epoch": 13.06,
+      "grad_norm": 0.0003235260955989361,
+      "learning_rate": 1.9264907609134188e-05,
+      "loss": 0.0038,
+      "step": 17990
+    },
+    {
+      "epoch": 13.07,
+      "grad_norm": 0.010898897424340248,
+      "learning_rate": 1.9244734931009442e-05,
+      "loss": 0.0009,
+      "step": 18000
+    },
+    {
+      "epoch": 13.07,
+      "grad_norm": 0.29816633462905884,
+      "learning_rate": 1.9224562252884697e-05,
+      "loss": 0.0012,
+      "step": 18010
+    },
+    {
+      "epoch": 13.08,
+      "grad_norm": 0.013279824517667294,
+      "learning_rate": 1.9204389574759944e-05,
+      "loss": 0.0006,
+      "step": 18020
+    },
+    {
+      "epoch": 13.09,
+      "grad_norm": 0.0002779340138658881,
+      "learning_rate": 1.91842168966352e-05,
+      "loss": 0.0179,
+      "step": 18030
+    },
+    {
+      "epoch": 13.1,
+      "grad_norm": 30.59058952331543,
+      "learning_rate": 1.916404421851045e-05,
+      "loss": 0.0066,
+      "step": 18040
+    },
+    {
+      "epoch": 13.1,
+      "grad_norm": 0.1223394125699997,
+      "learning_rate": 1.91438715403857e-05,
+      "loss": 0.0031,
+      "step": 18050
+    },
+    {
+      "epoch": 13.11,
+      "grad_norm": 0.0008338306797668338,
+      "learning_rate": 1.9123698862260955e-05,
+      "loss": 0.0025,
+      "step": 18060
+    },
+    {
+      "epoch": 13.12,
+      "grad_norm": 0.012727318331599236,
+      "learning_rate": 1.9103526184136207e-05,
+      "loss": 0.01,
+      "step": 18070
+    },
+    {
+      "epoch": 13.13,
+      "grad_norm": 0.1314314305782318,
+      "learning_rate": 1.9083353506011458e-05,
+      "loss": 0.0077,
+      "step": 18080
+    },
+    {
+      "epoch": 13.13,
+      "grad_norm": 0.011394386179745197,
+      "learning_rate": 1.9063180827886712e-05,
+      "loss": 0.0089,
+      "step": 18090
+    },
+    {
+      "epoch": 13.14,
+      "grad_norm": 0.049658820033073425,
+      "learning_rate": 1.9043008149761963e-05,
+      "loss": 0.0001,
+      "step": 18100
+    },
+    {
+      "epoch": 13.15,
+      "grad_norm": 0.021471910178661346,
+      "learning_rate": 1.9022835471637214e-05,
+      "loss": 0.0036,
+      "step": 18110
+    },
+    {
+      "epoch": 13.15,
+      "grad_norm": 8.542447090148926,
+      "learning_rate": 1.900266279351247e-05,
+      "loss": 0.0075,
+      "step": 18120
+    },
+    {
+      "epoch": 13.16,
+      "grad_norm": 3.939650774002075,
+      "learning_rate": 1.898249011538772e-05,
+      "loss": 0.0084,
+      "step": 18130
+    },
+    {
+      "epoch": 13.17,
+      "grad_norm": 0.10803334414958954,
+      "learning_rate": 1.896231743726297e-05,
+      "loss": 0.0033,
+      "step": 18140
+    },
+    {
+      "epoch": 13.18,
+      "grad_norm": 0.0016341025475412607,
+      "learning_rate": 1.8942144759138225e-05,
+      "loss": 0.0006,
+      "step": 18150
+    },
+    {
+      "epoch": 13.18,
+      "grad_norm": 0.001856721006333828,
+      "learning_rate": 1.8921972081013476e-05,
+      "loss": 0.001,
+      "step": 18160
+    },
+    {
+      "epoch": 13.19,
+      "grad_norm": 0.0011930714827030897,
+      "learning_rate": 1.8901799402888727e-05,
+      "loss": 0.0006,
+      "step": 18170
+    },
+    {
+      "epoch": 13.2,
+      "grad_norm": 0.0006116937729530036,
+      "learning_rate": 1.8881626724763982e-05,
+      "loss": 0.0159,
+      "step": 18180
+    },
+    {
+      "epoch": 13.21,
+      "grad_norm": 0.0016800108132883906,
+      "learning_rate": 1.8861454046639233e-05,
+      "loss": 0.014,
+      "step": 18190
+    },
+    {
+      "epoch": 13.21,
+      "grad_norm": 0.23176616430282593,
+      "learning_rate": 1.8841281368514484e-05,
+      "loss": 0.0028,
+      "step": 18200
+    },
+    {
+      "epoch": 13.22,
+      "grad_norm": 0.05328686162829399,
+      "learning_rate": 1.882110869038974e-05,
+      "loss": 0.0011,
+      "step": 18210
+    },
+    {
+      "epoch": 13.23,
+      "grad_norm": 0.014226214028894901,
+      "learning_rate": 1.880093601226499e-05,
+      "loss": 0.0037,
+      "step": 18220
+    },
+    {
+      "epoch": 13.23,
+      "grad_norm": 0.002094144467264414,
+      "learning_rate": 1.878076333414024e-05,
+      "loss": 0.0023,
+      "step": 18230
+    },
+    {
+      "epoch": 13.24,
+      "grad_norm": 0.004210632294416428,
+      "learning_rate": 1.8760590656015495e-05,
+      "loss": 0.0125,
+      "step": 18240
+    },
+    {
+      "epoch": 13.25,
+      "grad_norm": 0.44930320978164673,
+      "learning_rate": 1.8740417977890746e-05,
+      "loss": 0.0025,
+      "step": 18250
+    },
+    {
+      "epoch": 13.26,
+      "grad_norm": 0.7001730799674988,
+      "learning_rate": 1.8720245299765997e-05,
+      "loss": 0.0127,
+      "step": 18260
+    },
+    {
+      "epoch": 13.26,
+      "grad_norm": 0.019284186884760857,
+      "learning_rate": 1.870007262164125e-05,
+      "loss": 0.0021,
+      "step": 18270
+    },
+    {
+      "epoch": 13.27,
+      "grad_norm": 0.0007997555076144636,
+      "learning_rate": 1.8679899943516503e-05,
+      "loss": 0.008,
+      "step": 18280
+    },
+    {
+      "epoch": 13.28,
+      "grad_norm": 0.004993034061044455,
+      "learning_rate": 1.8659727265391754e-05,
+      "loss": 0.005,
+      "step": 18290
+    },
+    {
+      "epoch": 13.28,
+      "grad_norm": 0.0015475050313398242,
+      "learning_rate": 1.8639554587267005e-05,
+      "loss": 0.0113,
+      "step": 18300
+    },
+    {
+      "epoch": 13.29,
+      "grad_norm": 0.0004995432100258768,
+      "learning_rate": 1.861938190914226e-05,
+      "loss": 0.0014,
+      "step": 18310
+    },
+    {
+      "epoch": 13.3,
+      "grad_norm": 0.0007782830507494509,
+      "learning_rate": 1.859920923101751e-05,
+      "loss": 0.0044,
+      "step": 18320
+    },
+    {
+      "epoch": 13.31,
+      "grad_norm": 0.043515272438526154,
+      "learning_rate": 1.857903655289276e-05,
+      "loss": 0.0028,
+      "step": 18330
+    },
+    {
+      "epoch": 13.31,
+      "grad_norm": 0.2453761100769043,
+      "learning_rate": 1.8558863874768016e-05,
+      "loss": 0.0087,
+      "step": 18340
+    },
+    {
+      "epoch": 13.32,
+      "grad_norm": 0.00025056241429410875,
+      "learning_rate": 1.8538691196643267e-05,
+      "loss": 0.0115,
+      "step": 18350
+    },
+    {
+      "epoch": 13.33,
+      "grad_norm": 0.003255044110119343,
+      "learning_rate": 1.8518518518518518e-05,
+      "loss": 0.0026,
+      "step": 18360
+    },
+    {
+      "epoch": 13.34,
+      "grad_norm": 0.0290953628718853,
+      "learning_rate": 1.8498345840393773e-05,
+      "loss": 0.0037,
+      "step": 18370
+    },
+    {
+      "epoch": 13.34,
+      "grad_norm": 0.0034771724604070187,
+      "learning_rate": 1.8478173162269024e-05,
+      "loss": 0.0122,
+      "step": 18380
+    },
+    {
+      "epoch": 13.35,
+      "grad_norm": 0.001946158241480589,
+      "learning_rate": 1.8458000484144275e-05,
+      "loss": 0.0146,
+      "step": 18390
+    },
+    {
+      "epoch": 13.36,
+      "grad_norm": 0.0008432241738773882,
+      "learning_rate": 1.843782780601953e-05,
+      "loss": 0.0098,
+      "step": 18400
+    },
+    {
+      "epoch": 13.36,
+      "grad_norm": 0.006002933252602816,
+      "learning_rate": 1.841765512789478e-05,
+      "loss": 0.0354,
+      "step": 18410
+    },
+    {
+      "epoch": 13.37,
+      "grad_norm": 0.0009837823454290628,
+      "learning_rate": 1.839748244977003e-05,
+      "loss": 0.0053,
+      "step": 18420
+    },
+    {
+      "epoch": 13.38,
+      "grad_norm": 0.21960854530334473,
+      "learning_rate": 1.8377309771645286e-05,
+      "loss": 0.0066,
+      "step": 18430
+    },
+    {
+      "epoch": 13.39,
+      "grad_norm": 0.0019889490213245153,
+      "learning_rate": 1.8357137093520537e-05,
+      "loss": 0.0036,
+      "step": 18440
+    },
+    {
+      "epoch": 13.39,
+      "grad_norm": 0.02662692405283451,
+      "learning_rate": 1.8336964415395788e-05,
+      "loss": 0.0007,
+      "step": 18450
+    },
+    {
+      "epoch": 13.4,
+      "grad_norm": 0.0018065626500174403,
+      "learning_rate": 1.8316791737271043e-05,
+      "loss": 0.0056,
+      "step": 18460
+    },
+    {
+      "epoch": 13.41,
+      "grad_norm": 0.061746492981910706,
+      "learning_rate": 1.8296619059146294e-05,
+      "loss": 0.0031,
+      "step": 18470
+    },
+    {
+      "epoch": 13.42,
+      "grad_norm": 0.027934031561017036,
+      "learning_rate": 1.8276446381021545e-05,
+      "loss": 0.0003,
+      "step": 18480
+    },
+    {
+      "epoch": 13.42,
+      "grad_norm": 4.120459079742432,
+      "learning_rate": 1.82562737028968e-05,
+      "loss": 0.02,
+      "step": 18490
+    },
+    {
+      "epoch": 13.43,
+      "grad_norm": 0.0340251624584198,
+      "learning_rate": 1.823610102477205e-05,
+      "loss": 0.0047,
+      "step": 18500
+    },
+    {
+      "epoch": 13.44,
+      "grad_norm": 0.11251964420080185,
+      "learning_rate": 1.82159283466473e-05,
+      "loss": 0.0012,
+      "step": 18510
+    },
+    {
+      "epoch": 13.44,
+      "grad_norm": 0.048865534365177155,
+      "learning_rate": 1.8195755668522556e-05,
+      "loss": 0.004,
+      "step": 18520
+    },
+    {
+      "epoch": 13.45,
+      "grad_norm": 0.0008931795600801706,
+      "learning_rate": 1.8175582990397804e-05,
+      "loss": 0.0104,
+      "step": 18530
+    },
+    {
+      "epoch": 13.46,
+      "grad_norm": 0.00030427967431023717,
+      "learning_rate": 1.8155410312273058e-05,
+      "loss": 0.0124,
+      "step": 18540
+    },
+    {
+      "epoch": 13.47,
+      "grad_norm": 14.171956062316895,
+      "learning_rate": 1.8135237634148312e-05,
+      "loss": 0.0032,
+      "step": 18550
+    },
+    {
+      "epoch": 13.47,
+      "grad_norm": 0.10234571248292923,
+      "learning_rate": 1.811506495602356e-05,
+      "loss": 0.0089,
+      "step": 18560
+    },
+    {
+      "epoch": 13.48,
+      "grad_norm": 0.15911656618118286,
+      "learning_rate": 1.8094892277898815e-05,
+      "loss": 0.0044,
+      "step": 18570
+    },
+    {
+      "epoch": 13.49,
+      "grad_norm": 0.0011252342956140637,
+      "learning_rate": 1.807471959977407e-05,
+      "loss": 0.0036,
+      "step": 18580
+    },
+    {
+      "epoch": 13.5,
+      "grad_norm": 0.00042991613736376166,
+      "learning_rate": 1.8054546921649317e-05,
+      "loss": 0.0082,
+      "step": 18590
+    },
+    {
+      "epoch": 13.5,
+      "grad_norm": 0.0018707435810938478,
+      "learning_rate": 1.803437424352457e-05,
+      "loss": 0.0062,
+      "step": 18600
+    },
+    {
+      "epoch": 13.51,
+      "grad_norm": 30.47728729248047,
+      "learning_rate": 1.8014201565399826e-05,
+      "loss": 0.0159,
+      "step": 18610
+    },
+    {
+      "epoch": 13.52,
+      "grad_norm": 0.1323844939470291,
+      "learning_rate": 1.7994028887275073e-05,
+      "loss": 0.0087,
+      "step": 18620
+    },
+    {
+      "epoch": 13.52,
+      "grad_norm": 0.0003529054229147732,
+      "learning_rate": 1.7973856209150328e-05,
+      "loss": 0.0002,
+      "step": 18630
+    },
+    {
+      "epoch": 13.53,
+      "grad_norm": 0.02061094157397747,
+      "learning_rate": 1.7953683531025582e-05,
+      "loss": 0.0053,
+      "step": 18640
+    },
+    {
+      "epoch": 13.54,
+      "grad_norm": 0.015290017239749432,
+      "learning_rate": 1.793351085290083e-05,
+      "loss": 0.0013,
+      "step": 18650
+    },
+    {
+      "epoch": 13.55,
+      "grad_norm": 0.0006153315771371126,
+      "learning_rate": 1.7913338174776084e-05,
+      "loss": 0.0014,
+      "step": 18660
+    },
+    {
+      "epoch": 13.55,
+      "grad_norm": 0.0002869318414013833,
+      "learning_rate": 1.789316549665134e-05,
+      "loss": 0.0099,
+      "step": 18670
+    },
+    {
+      "epoch": 13.56,
+      "grad_norm": 0.000607367604970932,
+      "learning_rate": 1.7872992818526587e-05,
+      "loss": 0.0034,
+      "step": 18680
+    },
+    {
+      "epoch": 13.57,
+      "grad_norm": 0.13318461179733276,
+      "learning_rate": 1.785282014040184e-05,
+      "loss": 0.0033,
+      "step": 18690
+    },
+    {
+      "epoch": 13.58,
+      "grad_norm": 0.0001578326482558623,
+      "learning_rate": 1.7832647462277096e-05,
+      "loss": 0.0082,
+      "step": 18700
+    },
+    {
+      "epoch": 13.58,
+      "grad_norm": 0.037437982857227325,
+      "learning_rate": 1.7812474784152343e-05,
+      "loss": 0.0061,
+      "step": 18710
+    },
+    {
+      "epoch": 13.59,
+      "grad_norm": 3.765977144241333,
+      "learning_rate": 1.7792302106027598e-05,
+      "loss": 0.0144,
+      "step": 18720
+    },
+    {
+      "epoch": 13.6,
+      "grad_norm": 0.0014689884847030044,
+      "learning_rate": 1.7772129427902852e-05,
+      "loss": 0.0011,
+      "step": 18730
+    },
+    {
+      "epoch": 13.6,
+      "grad_norm": 0.05242501571774483,
+      "learning_rate": 1.77519567497781e-05,
+      "loss": 0.0048,
+      "step": 18740
+    },
+    {
+      "epoch": 13.61,
+      "grad_norm": 0.03251238167285919,
+      "learning_rate": 1.7731784071653354e-05,
+      "loss": 0.0041,
+      "step": 18750
+    },
+    {
+      "epoch": 13.62,
+      "grad_norm": 0.0015125583158805966,
+      "learning_rate": 1.7711611393528605e-05,
+      "loss": 0.0069,
+      "step": 18760
+    },
+    {
+      "epoch": 13.63,
+      "grad_norm": 0.0005884718266315758,
+      "learning_rate": 1.7691438715403856e-05,
+      "loss": 0.0047,
+      "step": 18770
+    },
+    {
+      "epoch": 13.63,
+      "grad_norm": 0.17881019413471222,
+      "learning_rate": 1.767126603727911e-05,
+      "loss": 0.0029,
+      "step": 18780
+    },
+    {
+      "epoch": 13.64,
+      "grad_norm": 0.1586664319038391,
+      "learning_rate": 1.7651093359154362e-05,
+      "loss": 0.0007,
+      "step": 18790
+    },
+    {
+      "epoch": 13.65,
+      "grad_norm": 0.004922006744891405,
+      "learning_rate": 1.7630920681029613e-05,
+      "loss": 0.0035,
+      "step": 18800
+    },
+    {
+      "epoch": 13.66,
+      "grad_norm": 1.1970666646957397,
+      "learning_rate": 1.7610748002904868e-05,
+      "loss": 0.0009,
+      "step": 18810
+    },
+    {
+      "epoch": 13.66,
+      "grad_norm": 0.04897474870085716,
+      "learning_rate": 1.759057532478012e-05,
+      "loss": 0.0054,
+      "step": 18820
+    },
+    {
+      "epoch": 13.67,
+      "grad_norm": 0.000665748433675617,
+      "learning_rate": 1.757040264665537e-05,
+      "loss": 0.0049,
+      "step": 18830
+    },
+    {
+      "epoch": 13.68,
+      "grad_norm": 0.11206013709306717,
+      "learning_rate": 1.7550229968530624e-05,
+      "loss": 0.0063,
+      "step": 18840
+    },
+    {
+      "epoch": 13.68,
+      "grad_norm": 0.0007113315514288843,
+      "learning_rate": 1.7530057290405875e-05,
+      "loss": 0.0205,
+      "step": 18850
+    },
+    {
+      "epoch": 13.69,
+      "grad_norm": 0.0034957346506416798,
+      "learning_rate": 1.7509884612281126e-05,
+      "loss": 0.0025,
+      "step": 18860
+    },
+    {
+      "epoch": 13.7,
+      "grad_norm": 0.0020177995320409536,
+      "learning_rate": 1.748971193415638e-05,
+      "loss": 0.0015,
+      "step": 18870
+    },
+    {
+      "epoch": 13.71,
+      "grad_norm": 6.311367034912109,
+      "learning_rate": 1.7469539256031632e-05,
+      "loss": 0.0022,
+      "step": 18880
+    },
+    {
+      "epoch": 13.71,
+      "grad_norm": 0.0005880764219909906,
+      "learning_rate": 1.7449366577906883e-05,
+      "loss": 0.0015,
+      "step": 18890
+    },
+    {
+      "epoch": 13.72,
+      "grad_norm": 0.10973332077264786,
+      "learning_rate": 1.7429193899782137e-05,
+      "loss": 0.0182,
+      "step": 18900
+    },
+    {
+      "epoch": 13.73,
+      "grad_norm": 0.000982865341939032,
+      "learning_rate": 1.740902122165739e-05,
+      "loss": 0.0126,
+      "step": 18910
+    },
+    {
+      "epoch": 13.74,
+      "grad_norm": 0.509669303894043,
+      "learning_rate": 1.738884854353264e-05,
+      "loss": 0.0124,
+      "step": 18920
+    },
+    {
+      "epoch": 13.74,
+      "grad_norm": 0.05541288107633591,
+      "learning_rate": 1.7368675865407894e-05,
+      "loss": 0.0326,
+      "step": 18930
+    },
+    {
+      "epoch": 13.75,
+      "grad_norm": 0.006566929165273905,
+      "learning_rate": 1.7348503187283145e-05,
+      "loss": 0.0256,
+      "step": 18940
+    },
+    {
+      "epoch": 13.76,
+      "grad_norm": 1.1299678087234497,
+      "learning_rate": 1.7328330509158396e-05,
+      "loss": 0.014,
+      "step": 18950
+    },
+    {
+      "epoch": 13.76,
+      "grad_norm": 0.0010074771707877517,
+      "learning_rate": 1.730815783103365e-05,
+      "loss": 0.0024,
+      "step": 18960
+    },
+    {
+      "epoch": 13.77,
+      "grad_norm": 0.0006316117942333221,
+      "learning_rate": 1.7287985152908902e-05,
+      "loss": 0.0,
+      "step": 18970
+    },
+    {
+      "epoch": 13.78,
+      "grad_norm": 21.051273345947266,
+      "learning_rate": 1.7267812474784153e-05,
+      "loss": 0.015,
+      "step": 18980
+    },
+    {
+      "epoch": 13.79,
+      "grad_norm": 0.0021583428606390953,
+      "learning_rate": 1.7247639796659407e-05,
+      "loss": 0.0045,
+      "step": 18990
+    },
+    {
+      "epoch": 13.79,
+      "grad_norm": 0.13025076687335968,
+      "learning_rate": 1.722746711853466e-05,
+      "loss": 0.0054,
+      "step": 19000
+    },
+    {
+      "epoch": 13.8,
+      "grad_norm": 0.06940227001905441,
+      "learning_rate": 1.720729444040991e-05,
+      "loss": 0.0279,
+      "step": 19010
+    },
+    {
+      "epoch": 13.81,
+      "grad_norm": 1.8406766653060913,
+      "learning_rate": 1.718712176228516e-05,
+      "loss": 0.0059,
+      "step": 19020
+    },
+    {
+      "epoch": 13.81,
+      "grad_norm": 0.36772218346595764,
+      "learning_rate": 1.7166949084160415e-05,
+      "loss": 0.004,
+      "step": 19030
+    },
+    {
+      "epoch": 13.82,
+      "grad_norm": 0.2510613799095154,
+      "learning_rate": 1.7146776406035666e-05,
+      "loss": 0.0046,
+      "step": 19040
+    },
+    {
+      "epoch": 13.83,
+      "grad_norm": 0.0453202947974205,
+      "learning_rate": 1.7126603727910917e-05,
+      "loss": 0.0037,
+      "step": 19050
+    },
+    {
+      "epoch": 13.84,
+      "grad_norm": 0.002687858883291483,
+      "learning_rate": 1.710643104978617e-05,
+      "loss": 0.0062,
+      "step": 19060
+    },
+    {
+      "epoch": 13.84,
+      "grad_norm": 0.004243039526045322,
+      "learning_rate": 1.7086258371661423e-05,
+      "loss": 0.0056,
+      "step": 19070
+    },
+    {
+      "epoch": 13.85,
+      "grad_norm": 0.008846227079629898,
+      "learning_rate": 1.7066085693536674e-05,
+      "loss": 0.0083,
+      "step": 19080
+    },
+    {
+      "epoch": 13.86,
+      "grad_norm": 1.1003198623657227,
+      "learning_rate": 1.7045913015411928e-05,
+      "loss": 0.0051,
+      "step": 19090
+    },
+    {
+      "epoch": 13.87,
+      "grad_norm": 0.08140759915113449,
+      "learning_rate": 1.702574033728718e-05,
+      "loss": 0.017,
+      "step": 19100
+    },
+    {
+      "epoch": 13.87,
+      "grad_norm": 0.0035266373306512833,
+      "learning_rate": 1.700556765916243e-05,
+      "loss": 0.0051,
+      "step": 19110
+    },
+    {
+      "epoch": 13.88,
+      "grad_norm": 0.0657566487789154,
+      "learning_rate": 1.6985394981037685e-05,
+      "loss": 0.0026,
+      "step": 19120
+    },
+    {
+      "epoch": 13.89,
+      "grad_norm": 0.0039032178465276957,
+      "learning_rate": 1.6965222302912936e-05,
+      "loss": 0.0043,
+      "step": 19130
+    },
+    {
+      "epoch": 13.89,
+      "grad_norm": 0.11626137048006058,
+      "learning_rate": 1.6945049624788187e-05,
+      "loss": 0.0046,
+      "step": 19140
+    },
+    {
+      "epoch": 13.9,
+      "grad_norm": 0.21265415847301483,
+      "learning_rate": 1.692487694666344e-05,
+      "loss": 0.0026,
+      "step": 19150
+    },
+    {
+      "epoch": 13.91,
+      "grad_norm": 0.26542073488235474,
+      "learning_rate": 1.6904704268538693e-05,
+      "loss": 0.0012,
+      "step": 19160
+    },
+    {
+      "epoch": 13.92,
+      "grad_norm": 0.0003108033852186054,
+      "learning_rate": 1.6884531590413944e-05,
+      "loss": 0.0054,
+      "step": 19170
+    },
+    {
+      "epoch": 13.92,
+      "grad_norm": 0.012170134112238884,
+      "learning_rate": 1.6864358912289198e-05,
+      "loss": 0.0046,
+      "step": 19180
+    },
+    {
+      "epoch": 13.93,
+      "grad_norm": 0.17962482571601868,
+      "learning_rate": 1.684418623416445e-05,
+      "loss": 0.0032,
+      "step": 19190
+    },
+    {
+      "epoch": 13.94,
+      "grad_norm": 0.0008012820617295802,
+      "learning_rate": 1.68240135560397e-05,
+      "loss": 0.0006,
+      "step": 19200
+    },
+    {
+      "epoch": 13.95,
+      "grad_norm": 0.0011188456555828452,
+      "learning_rate": 1.6803840877914955e-05,
+      "loss": 0.012,
+      "step": 19210
+    },
+    {
+      "epoch": 13.95,
+      "grad_norm": 0.0029843682423233986,
+      "learning_rate": 1.6783668199790206e-05,
+      "loss": 0.0084,
+      "step": 19220
+    },
+    {
+      "epoch": 13.96,
+      "grad_norm": 0.006473259534686804,
+      "learning_rate": 1.6763495521665457e-05,
+      "loss": 0.0023,
+      "step": 19230
+    },
+    {
+      "epoch": 13.97,
+      "grad_norm": 0.032545410096645355,
+      "learning_rate": 1.674332284354071e-05,
+      "loss": 0.0076,
+      "step": 19240
+    },
+    {
+      "epoch": 13.97,
+      "grad_norm": 0.024446191266179085,
+      "learning_rate": 1.6723150165415962e-05,
+      "loss": 0.0106,
+      "step": 19250
+    },
+    {
+      "epoch": 13.98,
+      "grad_norm": 26.21831512451172,
+      "learning_rate": 1.6702977487291213e-05,
+      "loss": 0.0092,
+      "step": 19260
+    },
+    {
+      "epoch": 13.99,
+      "grad_norm": 0.05417324975132942,
+      "learning_rate": 1.6682804809166465e-05,
+      "loss": 0.0044,
+      "step": 19270
+    },
+    {
+      "epoch": 14.0,
+      "grad_norm": 0.12691135704517365,
+      "learning_rate": 1.6662632131041716e-05,
+      "loss": 0.0076,
+      "step": 19280
+    },
+    {
+      "epoch": 14.0,
+      "eval_accuracy": 0.9981394926714162,
+      "eval_f1": 0.9957034320146712,
+      "eval_loss": 0.004028408322483301,
+      "eval_precision": 0.9994214789102767,
+      "eval_recall": 0.9920129463353519,
+      "eval_roc_auc": 0.9999826737941067,
+      "eval_runtime": 389.171,
+      "eval_samples_per_second": 226.502,
+      "eval_steps_per_second": 14.158,
+      "step": 19285
+    },
+    {
+      "epoch": 14.0,
+      "grad_norm": 0.044279515743255615,
+      "learning_rate": 1.664245945291697e-05,
+      "loss": 0.003,
+      "step": 19290
+    },
+    {
+      "epoch": 14.01,
+      "grad_norm": 0.061287716031074524,
+      "learning_rate": 1.662228677479222e-05,
+      "loss": 0.0004,
+      "step": 19300
+    },
+    {
+      "epoch": 14.02,
+      "grad_norm": 5.015234470367432,
+      "learning_rate": 1.6602114096667472e-05,
+      "loss": 0.0062,
+      "step": 19310
+    },
+    {
+      "epoch": 14.03,
+      "grad_norm": 0.03511728346347809,
+      "learning_rate": 1.6581941418542727e-05,
+      "loss": 0.0059,
+      "step": 19320
+    },
+    {
+      "epoch": 14.03,
+      "grad_norm": 0.06877407431602478,
+      "learning_rate": 1.6561768740417978e-05,
+      "loss": 0.0008,
+      "step": 19330
+    },
+    {
+      "epoch": 14.04,
+      "grad_norm": 0.00669504189863801,
+      "learning_rate": 1.654159606229323e-05,
+      "loss": 0.0002,
+      "step": 19340
+    },
+    {
+      "epoch": 14.05,
+      "grad_norm": 1.7849411964416504,
+      "learning_rate": 1.6521423384168483e-05,
+      "loss": 0.0042,
+      "step": 19350
+    },
+    {
+      "epoch": 14.05,
+      "grad_norm": 0.0018257235642522573,
+      "learning_rate": 1.6501250706043734e-05,
+      "loss": 0.0029,
+      "step": 19360
+    },
+    {
+      "epoch": 14.06,
+      "grad_norm": 0.0007575092604383826,
+      "learning_rate": 1.6481078027918986e-05,
+      "loss": 0.0013,
+      "step": 19370
+    },
+    {
+      "epoch": 14.07,
+      "grad_norm": 0.0004143484984524548,
+      "learning_rate": 1.646090534979424e-05,
+      "loss": 0.0048,
+      "step": 19380
+    },
+    {
+      "epoch": 14.08,
+      "grad_norm": 12.652482032775879,
+      "learning_rate": 1.644073267166949e-05,
+      "loss": 0.0028,
+      "step": 19390
+    },
+    {
+      "epoch": 14.08,
+      "grad_norm": 0.00011446201096987352,
+      "learning_rate": 1.6420559993544742e-05,
+      "loss": 0.0116,
+      "step": 19400
+    },
+    {
+      "epoch": 14.09,
+      "grad_norm": 0.002371125388890505,
+      "learning_rate": 1.6400387315419997e-05,
+      "loss": 0.0141,
+      "step": 19410
+    },
+    {
+      "epoch": 14.1,
+      "grad_norm": 0.026215413585305214,
+      "learning_rate": 1.6380214637295248e-05,
+      "loss": 0.0033,
+      "step": 19420
+    },
+    {
+      "epoch": 14.11,
+      "grad_norm": 0.0014271169202402234,
+      "learning_rate": 1.63600419591705e-05,
+      "loss": 0.0014,
+      "step": 19430
+    },
+    {
+      "epoch": 14.11,
+      "grad_norm": 0.00161796307656914,
+      "learning_rate": 1.6339869281045753e-05,
+      "loss": 0.0009,
+      "step": 19440
+    },
+    {
+      "epoch": 14.12,
+      "grad_norm": 1.2255878448486328,
+      "learning_rate": 1.6319696602921004e-05,
+      "loss": 0.0022,
+      "step": 19450
+    },
+    {
+      "epoch": 14.13,
+      "grad_norm": 0.004346283618360758,
+      "learning_rate": 1.6299523924796255e-05,
+      "loss": 0.0014,
+      "step": 19460
+    },
+    {
+      "epoch": 14.13,
+      "grad_norm": 0.0006355916266329587,
+      "learning_rate": 1.627935124667151e-05,
+      "loss": 0.0143,
+      "step": 19470
+    },
+    {
+      "epoch": 14.14,
+      "grad_norm": 0.11306034028530121,
+      "learning_rate": 1.625917856854676e-05,
+      "loss": 0.0038,
+      "step": 19480
+    },
+    {
+      "epoch": 14.15,
+      "grad_norm": 0.00017098760872613639,
+      "learning_rate": 1.6239005890422012e-05,
+      "loss": 0.0028,
+      "step": 19490
+    },
+    {
+      "epoch": 14.16,
+      "grad_norm": 0.18717147409915924,
+      "learning_rate": 1.6218833212297266e-05,
+      "loss": 0.0109,
+      "step": 19500
+    },
+    {
+      "epoch": 14.16,
+      "grad_norm": 0.0023012091405689716,
+      "learning_rate": 1.6198660534172518e-05,
+      "loss": 0.0039,
+      "step": 19510
+    },
+    {
+      "epoch": 14.17,
+      "grad_norm": 0.00090291682863608,
+      "learning_rate": 1.617848785604777e-05,
+      "loss": 0.0013,
+      "step": 19520
+    },
+    {
+      "epoch": 14.18,
+      "grad_norm": 0.37925946712493896,
+      "learning_rate": 1.6158315177923023e-05,
+      "loss": 0.0113,
+      "step": 19530
+    },
+    {
+      "epoch": 14.19,
+      "grad_norm": 0.003591743763536215,
+      "learning_rate": 1.6138142499798274e-05,
+      "loss": 0.0042,
+      "step": 19540
+    },
+    {
+      "epoch": 14.19,
+      "grad_norm": 0.003399345325306058,
+      "learning_rate": 1.6117969821673525e-05,
+      "loss": 0.0001,
+      "step": 19550
+    },
+    {
+      "epoch": 14.2,
+      "grad_norm": 0.006896769627928734,
+      "learning_rate": 1.609779714354878e-05,
+      "loss": 0.0007,
+      "step": 19560
+    },
+    {
+      "epoch": 14.21,
+      "grad_norm": 0.01545551884919405,
+      "learning_rate": 1.607762446542403e-05,
+      "loss": 0.0006,
+      "step": 19570
+    },
+    {
+      "epoch": 14.21,
+      "grad_norm": 0.08701088279485703,
+      "learning_rate": 1.6057451787299282e-05,
+      "loss": 0.0055,
+      "step": 19580
+    },
+    {
+      "epoch": 14.22,
+      "grad_norm": 0.0003549981047399342,
+      "learning_rate": 1.6037279109174536e-05,
+      "loss": 0.003,
+      "step": 19590
+    },
+    {
+      "epoch": 14.23,
+      "grad_norm": 0.05535199120640755,
+      "learning_rate": 1.6017106431049787e-05,
+      "loss": 0.0125,
+      "step": 19600
+    },
+    {
+      "epoch": 14.24,
+      "grad_norm": 0.004172964952886105,
+      "learning_rate": 1.599693375292504e-05,
+      "loss": 0.0024,
+      "step": 19610
+    },
+    {
+      "epoch": 14.24,
+      "grad_norm": 0.018433071672916412,
+      "learning_rate": 1.5976761074800293e-05,
+      "loss": 0.003,
+      "step": 19620
+    },
+    {
+      "epoch": 14.25,
+      "grad_norm": 0.00013644646969623864,
+      "learning_rate": 1.5956588396675544e-05,
+      "loss": 0.0007,
+      "step": 19630
+    },
+    {
+      "epoch": 14.26,
+      "grad_norm": 0.12272168695926666,
+      "learning_rate": 1.5936415718550795e-05,
+      "loss": 0.0047,
+      "step": 19640
+    },
+    {
+      "epoch": 14.26,
+      "grad_norm": 27.803667068481445,
+      "learning_rate": 1.591624304042605e-05,
+      "loss": 0.0137,
+      "step": 19650
+    },
+    {
+      "epoch": 14.27,
+      "grad_norm": 3.5240798752056435e-05,
+      "learning_rate": 1.58960703623013e-05,
+      "loss": 0.0028,
+      "step": 19660
+    },
+    {
+      "epoch": 14.28,
+      "grad_norm": 0.00028095979359932244,
+      "learning_rate": 1.5875897684176552e-05,
+      "loss": 0.0043,
+      "step": 19670
+    },
+    {
+      "epoch": 14.29,
+      "grad_norm": 0.5708059072494507,
+      "learning_rate": 1.5855725006051806e-05,
+      "loss": 0.0063,
+      "step": 19680
+    },
+    {
+      "epoch": 14.29,
+      "grad_norm": 0.0016370153753086925,
+      "learning_rate": 1.5835552327927057e-05,
+      "loss": 0.0082,
+      "step": 19690
+    },
+    {
+      "epoch": 14.3,
+      "grad_norm": 0.00016567490820307285,
+      "learning_rate": 1.581537964980231e-05,
+      "loss": 0.0108,
+      "step": 19700
+    },
+    {
+      "epoch": 14.31,
+      "grad_norm": 0.00043105980148538947,
+      "learning_rate": 1.5795206971677563e-05,
+      "loss": 0.006,
+      "step": 19710
+    },
+    {
+      "epoch": 14.32,
+      "grad_norm": 0.00017671240493655205,
+      "learning_rate": 1.5775034293552814e-05,
+      "loss": 0.0112,
+      "step": 19720
+    },
+    {
+      "epoch": 14.32,
+      "grad_norm": 1.609751582145691,
+      "learning_rate": 1.5754861615428065e-05,
+      "loss": 0.0028,
+      "step": 19730
+    },
+    {
+      "epoch": 14.33,
+      "grad_norm": 0.00035769614623859525,
+      "learning_rate": 1.573468893730332e-05,
+      "loss": 0.0068,
+      "step": 19740
+    },
+    {
+      "epoch": 14.34,
+      "grad_norm": 0.2628963589668274,
+      "learning_rate": 1.571451625917857e-05,
+      "loss": 0.0192,
+      "step": 19750
+    },
+    {
+      "epoch": 14.34,
+      "grad_norm": 0.03742462769150734,
+      "learning_rate": 1.569434358105382e-05,
+      "loss": 0.0016,
+      "step": 19760
+    },
+    {
+      "epoch": 14.35,
+      "grad_norm": 0.000913703057449311,
+      "learning_rate": 1.5674170902929073e-05,
+      "loss": 0.0028,
+      "step": 19770
+    },
+    {
+      "epoch": 14.36,
+      "grad_norm": 0.05503058806061745,
+      "learning_rate": 1.5653998224804327e-05,
+      "loss": 0.0028,
+      "step": 19780
+    },
+    {
+      "epoch": 14.37,
+      "grad_norm": 0.028899654746055603,
+      "learning_rate": 1.5633825546679578e-05,
+      "loss": 0.0037,
+      "step": 19790
+    },
+    {
+      "epoch": 14.37,
+      "grad_norm": 0.000255098711932078,
+      "learning_rate": 1.561365286855483e-05,
+      "loss": 0.0017,
+      "step": 19800
+    },
+    {
+      "epoch": 14.38,
+      "grad_norm": 0.008871900849044323,
+      "learning_rate": 1.559348019043008e-05,
+      "loss": 0.0054,
+      "step": 19810
+    },
+    {
+      "epoch": 14.39,
+      "grad_norm": 0.030201343819499016,
+      "learning_rate": 1.5573307512305335e-05,
+      "loss": 0.0057,
+      "step": 19820
+    },
+    {
+      "epoch": 14.4,
+      "grad_norm": 0.01691788248717785,
+      "learning_rate": 1.5553134834180586e-05,
+      "loss": 0.0031,
+      "step": 19830
+    },
+    {
+      "epoch": 14.4,
+      "grad_norm": 0.0881408080458641,
+      "learning_rate": 1.5532962156055837e-05,
+      "loss": 0.0134,
+      "step": 19840
+    },
+    {
+      "epoch": 14.41,
+      "grad_norm": 0.0006498016882687807,
+      "learning_rate": 1.551278947793109e-05,
+      "loss": 0.0178,
+      "step": 19850
+    },
+    {
+      "epoch": 14.42,
+      "grad_norm": 0.000303228065604344,
+      "learning_rate": 1.5492616799806343e-05,
+      "loss": 0.0159,
+      "step": 19860
+    },
+    {
+      "epoch": 14.42,
+      "grad_norm": 0.19211918115615845,
+      "learning_rate": 1.5472444121681594e-05,
+      "loss": 0.0064,
+      "step": 19870
+    },
+    {
+      "epoch": 14.43,
+      "grad_norm": 0.0003167142567690462,
+      "learning_rate": 1.5452271443556848e-05,
+      "loss": 0.01,
+      "step": 19880
+    },
+    {
+      "epoch": 14.44,
+      "grad_norm": 34.510948181152344,
+      "learning_rate": 1.54320987654321e-05,
+      "loss": 0.0119,
+      "step": 19890
+    },
+    {
+      "epoch": 14.45,
+      "grad_norm": 0.06265545636415482,
+      "learning_rate": 1.541192608730735e-05,
+      "loss": 0.0084,
+      "step": 19900
+    },
+    {
+      "epoch": 14.45,
+      "grad_norm": 0.0007544682594016194,
+      "learning_rate": 1.5391753409182605e-05,
+      "loss": 0.0006,
+      "step": 19910
+    },
+    {
+      "epoch": 14.46,
+      "grad_norm": 1.0826385021209717,
+      "learning_rate": 1.5371580731057856e-05,
+      "loss": 0.0033,
+      "step": 19920
+    },
+    {
+      "epoch": 14.47,
+      "grad_norm": 0.031010733917355537,
+      "learning_rate": 1.5351408052933107e-05,
+      "loss": 0.0355,
+      "step": 19930
+    },
+    {
+      "epoch": 14.48,
+      "grad_norm": 0.293381005525589,
+      "learning_rate": 1.533123537480836e-05,
+      "loss": 0.0204,
+      "step": 19940
+    },
+    {
+      "epoch": 14.48,
+      "grad_norm": 0.003690192475914955,
+      "learning_rate": 1.5311062696683612e-05,
+      "loss": 0.0171,
+      "step": 19950
+    },
+    {
+      "epoch": 14.49,
+      "grad_norm": 0.0008720943587832153,
+      "learning_rate": 1.5290890018558863e-05,
+      "loss": 0.0128,
+      "step": 19960
+    },
+    {
+      "epoch": 14.5,
+      "grad_norm": 0.05105540528893471,
+      "learning_rate": 1.5270717340434118e-05,
+      "loss": 0.0033,
+      "step": 19970
+    },
+    {
+      "epoch": 14.5,
+      "grad_norm": 0.1516324281692505,
+      "learning_rate": 1.5250544662309369e-05,
+      "loss": 0.0063,
+      "step": 19980
+    },
+    {
+      "epoch": 14.51,
+      "grad_norm": 0.009380163624882698,
+      "learning_rate": 1.5230371984184622e-05,
+      "loss": 0.0109,
+      "step": 19990
+    },
+    {
+      "epoch": 14.52,
+      "grad_norm": 0.0012153387069702148,
+      "learning_rate": 1.5210199306059871e-05,
+      "loss": 0.0022,
+      "step": 20000
+    },
+    {
+      "epoch": 14.53,
+      "grad_norm": 0.0027521008159965277,
+      "learning_rate": 1.5190026627935126e-05,
+      "loss": 0.0024,
+      "step": 20010
+    },
+    {
+      "epoch": 14.53,
+      "grad_norm": 21.582521438598633,
+      "learning_rate": 1.5169853949810378e-05,
+      "loss": 0.0036,
+      "step": 20020
+    },
+    {
+      "epoch": 14.54,
+      "grad_norm": 0.003993968944996595,
+      "learning_rate": 1.5149681271685628e-05,
+      "loss": 0.0126,
+      "step": 20030
+    },
+    {
+      "epoch": 14.55,
+      "grad_norm": 0.00026681035524234176,
+      "learning_rate": 1.5129508593560882e-05,
+      "loss": 0.0022,
+      "step": 20040
+    },
+    {
+      "epoch": 14.56,
+      "grad_norm": 0.03443112596869469,
+      "learning_rate": 1.5109335915436135e-05,
+      "loss": 0.01,
+      "step": 20050
+    },
+    {
+      "epoch": 14.56,
+      "grad_norm": 0.00041420606430619955,
+      "learning_rate": 1.5089163237311384e-05,
+      "loss": 0.0173,
+      "step": 20060
+    },
+    {
+      "epoch": 14.57,
+      "grad_norm": 6.245350360870361,
+      "learning_rate": 1.5068990559186639e-05,
+      "loss": 0.0125,
+      "step": 20070
+    },
+    {
+      "epoch": 14.58,
+      "grad_norm": 0.0017795218154788017,
+      "learning_rate": 1.5048817881061892e-05,
+      "loss": 0.0115,
+      "step": 20080
+    },
+    {
+      "epoch": 14.58,
+      "grad_norm": 0.008628590032458305,
+      "learning_rate": 1.5028645202937141e-05,
+      "loss": 0.0056,
+      "step": 20090
+    },
+    {
+      "epoch": 14.59,
+      "grad_norm": 4.200293064117432,
+      "learning_rate": 1.5008472524812395e-05,
+      "loss": 0.0176,
+      "step": 20100
+    },
+    {
+      "epoch": 14.6,
+      "grad_norm": 0.03641504794359207,
+      "learning_rate": 1.4988299846687648e-05,
+      "loss": 0.0084,
+      "step": 20110
+    },
+    {
+      "epoch": 14.61,
+      "grad_norm": 0.08268699795007706,
+      "learning_rate": 1.4968127168562898e-05,
+      "loss": 0.0017,
+      "step": 20120
+    },
+    {
+      "epoch": 14.61,
+      "grad_norm": 0.0018813032656908035,
+      "learning_rate": 1.4947954490438152e-05,
+      "loss": 0.0097,
+      "step": 20130
+    },
+    {
+      "epoch": 14.62,
+      "grad_norm": 0.0015808496391400695,
+      "learning_rate": 1.4927781812313405e-05,
+      "loss": 0.0072,
+      "step": 20140
+    },
+    {
+      "epoch": 14.63,
+      "grad_norm": 0.0061777797527611256,
+      "learning_rate": 1.4907609134188654e-05,
+      "loss": 0.0026,
+      "step": 20150
+    },
+    {
+      "epoch": 14.64,
+      "grad_norm": 0.04996776953339577,
+      "learning_rate": 1.4887436456063909e-05,
+      "loss": 0.0097,
+      "step": 20160
+    },
+    {
+      "epoch": 14.64,
+      "grad_norm": 0.00472896546125412,
+      "learning_rate": 1.4867263777939161e-05,
+      "loss": 0.004,
+      "step": 20170
+    },
+    {
+      "epoch": 14.65,
+      "grad_norm": 0.014625828713178635,
+      "learning_rate": 1.4847091099814411e-05,
+      "loss": 0.0003,
+      "step": 20180
+    },
+    {
+      "epoch": 14.66,
+      "grad_norm": 0.013716181740164757,
+      "learning_rate": 1.4826918421689665e-05,
+      "loss": 0.0046,
+      "step": 20190
+    },
+    {
+      "epoch": 14.66,
+      "grad_norm": 0.0011542306747287512,
+      "learning_rate": 1.4806745743564918e-05,
+      "loss": 0.0108,
+      "step": 20200
+    },
+    {
+      "epoch": 14.67,
+      "grad_norm": 0.5403972268104553,
+      "learning_rate": 1.4786573065440167e-05,
+      "loss": 0.0004,
+      "step": 20210
+    },
+    {
+      "epoch": 14.68,
+      "grad_norm": 0.026584791019558907,
+      "learning_rate": 1.476640038731542e-05,
+      "loss": 0.0031,
+      "step": 20220
+    },
+    {
+      "epoch": 14.69,
+      "grad_norm": 0.11381633579730988,
+      "learning_rate": 1.4746227709190675e-05,
+      "loss": 0.0108,
+      "step": 20230
+    },
+    {
+      "epoch": 14.69,
+      "grad_norm": 0.28148502111434937,
+      "learning_rate": 1.4726055031065924e-05,
+      "loss": 0.0082,
+      "step": 20240
+    },
+    {
+      "epoch": 14.7,
+      "grad_norm": 0.03273105248808861,
+      "learning_rate": 1.4705882352941177e-05,
+      "loss": 0.0015,
+      "step": 20250
+    },
+    {
+      "epoch": 14.71,
+      "grad_norm": 0.10272327065467834,
+      "learning_rate": 1.4685709674816428e-05,
+      "loss": 0.0049,
+      "step": 20260
+    },
+    {
+      "epoch": 14.72,
+      "grad_norm": 0.09216368943452835,
+      "learning_rate": 1.466553699669168e-05,
+      "loss": 0.0103,
+      "step": 20270
+    },
+    {
+      "epoch": 14.72,
+      "grad_norm": 0.0008601095178164542,
+      "learning_rate": 1.4645364318566933e-05,
+      "loss": 0.0047,
+      "step": 20280
+    },
+    {
+      "epoch": 14.73,
+      "grad_norm": 0.0037641292437911034,
+      "learning_rate": 1.4625191640442185e-05,
+      "loss": 0.0013,
+      "step": 20290
+    },
+    {
+      "epoch": 14.74,
+      "grad_norm": 0.001481158658862114,
+      "learning_rate": 1.4605018962317437e-05,
+      "loss": 0.0034,
+      "step": 20300
+    },
+    {
+      "epoch": 14.74,
+      "grad_norm": 0.045567817986011505,
+      "learning_rate": 1.458484628419269e-05,
+      "loss": 0.007,
+      "step": 20310
+    },
+    {
+      "epoch": 14.75,
+      "grad_norm": 0.0007129237637855113,
+      "learning_rate": 1.4564673606067941e-05,
+      "loss": 0.008,
+      "step": 20320
+    },
+    {
+      "epoch": 14.76,
+      "grad_norm": 0.000712543202098459,
+      "learning_rate": 1.4544500927943194e-05,
+      "loss": 0.0041,
+      "step": 20330
+    },
+    {
+      "epoch": 14.77,
+      "grad_norm": 0.002881669905036688,
+      "learning_rate": 1.4524328249818447e-05,
+      "loss": 0.0083,
+      "step": 20340
+    },
+    {
+      "epoch": 14.77,
+      "grad_norm": 0.12856851518154144,
+      "learning_rate": 1.4504155571693698e-05,
+      "loss": 0.007,
+      "step": 20350
+    },
+    {
+      "epoch": 14.78,
+      "grad_norm": 0.00111017981544137,
+      "learning_rate": 1.448398289356895e-05,
+      "loss": 0.0042,
+      "step": 20360
+    },
+    {
+      "epoch": 14.79,
+      "grad_norm": 0.0010551324812695384,
+      "learning_rate": 1.4463810215444203e-05,
+      "loss": 0.0106,
+      "step": 20370
+    },
+    {
+      "epoch": 14.79,
+      "grad_norm": 0.04143055900931358,
+      "learning_rate": 1.4443637537319454e-05,
+      "loss": 0.0033,
+      "step": 20380
+    },
+    {
+      "epoch": 14.8,
+      "grad_norm": 0.005768972914665937,
+      "learning_rate": 1.4423464859194707e-05,
+      "loss": 0.0073,
+      "step": 20390
+    },
+    {
+      "epoch": 14.81,
+      "grad_norm": 6.56055417493917e-05,
+      "learning_rate": 1.440329218106996e-05,
+      "loss": 0.0173,
+      "step": 20400
+    },
+    {
+      "epoch": 14.82,
+      "grad_norm": 0.0011871436145156622,
+      "learning_rate": 1.4383119502945211e-05,
+      "loss": 0.0011,
+      "step": 20410
+    },
+    {
+      "epoch": 14.82,
+      "grad_norm": 0.0003176795144099742,
+      "learning_rate": 1.4362946824820464e-05,
+      "loss": 0.0099,
+      "step": 20420
+    },
+    {
+      "epoch": 14.83,
+      "grad_norm": 0.1483236402273178,
+      "learning_rate": 1.4342774146695717e-05,
+      "loss": 0.0035,
+      "step": 20430
+    },
+    {
+      "epoch": 14.84,
+      "grad_norm": 0.37400639057159424,
+      "learning_rate": 1.4322601468570968e-05,
+      "loss": 0.0093,
+      "step": 20440
+    },
+    {
+      "epoch": 14.85,
+      "grad_norm": 5.234686250332743e-05,
+      "learning_rate": 1.430242879044622e-05,
+      "loss": 0.0127,
+      "step": 20450
+    },
+    {
+      "epoch": 14.85,
+      "grad_norm": 0.16454139351844788,
+      "learning_rate": 1.4282256112321473e-05,
+      "loss": 0.0096,
+      "step": 20460
+    },
+    {
+      "epoch": 14.86,
+      "grad_norm": 30.7055721282959,
+      "learning_rate": 1.4262083434196724e-05,
+      "loss": 0.0079,
+      "step": 20470
+    },
+    {
+      "epoch": 14.87,
+      "grad_norm": 0.01432458683848381,
+      "learning_rate": 1.4241910756071977e-05,
+      "loss": 0.0021,
+      "step": 20480
+    },
+    {
+      "epoch": 14.87,
+      "grad_norm": 0.0014726252993568778,
+      "learning_rate": 1.422173807794723e-05,
+      "loss": 0.0036,
+      "step": 20490
+    },
+    {
+      "epoch": 14.88,
+      "grad_norm": 0.0001415656297467649,
+      "learning_rate": 1.4201565399822481e-05,
+      "loss": 0.0043,
+      "step": 20500
+    },
+    {
+      "epoch": 14.89,
+      "grad_norm": 0.9634278416633606,
+      "learning_rate": 1.4181392721697734e-05,
+      "loss": 0.0035,
+      "step": 20510
+    },
+    {
+      "epoch": 14.9,
+      "grad_norm": 0.0005977644468657672,
+      "learning_rate": 1.4161220043572985e-05,
+      "loss": 0.0017,
+      "step": 20520
+    },
+    {
+      "epoch": 14.9,
+      "grad_norm": 0.0001564495323691517,
+      "learning_rate": 1.4141047365448238e-05,
+      "loss": 0.0043,
+      "step": 20530
+    },
+    {
+      "epoch": 14.91,
+      "grad_norm": 9.115212014876306e-05,
+      "learning_rate": 1.412087468732349e-05,
+      "loss": 0.0107,
+      "step": 20540
+    },
+    {
+      "epoch": 14.92,
+      "grad_norm": 0.025375254452228546,
+      "learning_rate": 1.4100702009198741e-05,
+      "loss": 0.0135,
+      "step": 20550
+    },
+    {
+      "epoch": 14.93,
+      "grad_norm": 0.12782764434814453,
+      "learning_rate": 1.4080529331073994e-05,
+      "loss": 0.0034,
+      "step": 20560
+    },
+    {
+      "epoch": 14.93,
+      "grad_norm": 0.06092876195907593,
+      "learning_rate": 1.4060356652949247e-05,
+      "loss": 0.0066,
+      "step": 20570
+    },
+    {
+      "epoch": 14.94,
+      "grad_norm": 0.007951307110488415,
+      "learning_rate": 1.4040183974824498e-05,
+      "loss": 0.0057,
+      "step": 20580
+    },
+    {
+      "epoch": 14.95,
+      "grad_norm": 0.004196907859295607,
+      "learning_rate": 1.402001129669975e-05,
+      "loss": 0.0055,
+      "step": 20590
+    },
+    {
+      "epoch": 14.95,
+      "grad_norm": 24.28219223022461,
+      "learning_rate": 1.3999838618575004e-05,
+      "loss": 0.0047,
+      "step": 20600
+    },
+    {
+      "epoch": 14.96,
+      "grad_norm": 0.10869511216878891,
+      "learning_rate": 1.3979665940450255e-05,
+      "loss": 0.0049,
+      "step": 20610
+    },
+    {
+      "epoch": 14.97,
+      "grad_norm": 0.003674560459330678,
+      "learning_rate": 1.3959493262325507e-05,
+      "loss": 0.0203,
+      "step": 20620
+    },
+    {
+      "epoch": 14.98,
+      "grad_norm": 0.009799223393201828,
+      "learning_rate": 1.393932058420076e-05,
+      "loss": 0.0017,
+      "step": 20630
+    },
+    {
+      "epoch": 14.98,
+      "grad_norm": 0.0014988232869654894,
+      "learning_rate": 1.3919147906076011e-05,
+      "loss": 0.0011,
+      "step": 20640
+    },
+    {
+      "epoch": 14.99,
+      "grad_norm": 0.01472469512373209,
+      "learning_rate": 1.3898975227951264e-05,
+      "loss": 0.0025,
+      "step": 20650
+    },
+    {
+      "epoch": 15.0,
+      "grad_norm": 0.005289836321026087,
+      "learning_rate": 1.3878802549826517e-05,
+      "loss": 0.0032,
+      "step": 20660
+    },
+    {
+      "epoch": 15.0,
+      "eval_accuracy": 0.9982756273539956,
+      "eval_f1": 0.9960188580408591,
+      "eval_loss": 0.0035676565021276474,
+      "eval_precision": 0.9994743481917577,
+      "eval_recall": 0.9925871789517645,
+      "eval_roc_auc": 0.9999858661107502,
+      "eval_runtime": 385.486,
+      "eval_samples_per_second": 228.667,
+      "eval_steps_per_second": 14.294,
+      "step": 20662
+    },
+    {
+      "epoch": 15.01,
+      "grad_norm": 13.548345565795898,
+      "learning_rate": 1.3858629871701768e-05,
+      "loss": 0.0016,
+      "step": 20670
+    },
+    {
+      "epoch": 15.01,
+      "grad_norm": 0.04863952100276947,
+      "learning_rate": 1.383845719357702e-05,
+      "loss": 0.0031,
+      "step": 20680
+    },
+    {
+      "epoch": 15.02,
+      "grad_norm": 0.026892608031630516,
+      "learning_rate": 1.3818284515452273e-05,
+      "loss": 0.0022,
+      "step": 20690
+    },
+    {
+      "epoch": 15.03,
+      "grad_norm": 3.914630651473999,
+      "learning_rate": 1.3798111837327524e-05,
+      "loss": 0.0048,
+      "step": 20700
+    },
+    {
+      "epoch": 15.03,
+      "grad_norm": 0.031707461923360825,
+      "learning_rate": 1.3777939159202777e-05,
+      "loss": 0.004,
+      "step": 20710
+    },
+    {
+      "epoch": 15.04,
+      "grad_norm": 0.00016734382370486856,
+      "learning_rate": 1.375776648107803e-05,
+      "loss": 0.0021,
+      "step": 20720
+    },
+    {
+      "epoch": 15.05,
+      "grad_norm": 0.5772116780281067,
+      "learning_rate": 1.373759380295328e-05,
+      "loss": 0.0055,
+      "step": 20730
+    },
+    {
+      "epoch": 15.06,
+      "grad_norm": 0.10167176276445389,
+      "learning_rate": 1.3717421124828534e-05,
+      "loss": 0.0055,
+      "step": 20740
+    },
+    {
+      "epoch": 15.06,
+      "grad_norm": 0.6610731482505798,
+      "learning_rate": 1.3697248446703783e-05,
+      "loss": 0.0114,
+      "step": 20750
+    },
+    {
+      "epoch": 15.07,
+      "grad_norm": 0.001937453867867589,
+      "learning_rate": 1.3677075768579036e-05,
+      "loss": 0.0016,
+      "step": 20760
+    },
+    {
+      "epoch": 15.08,
+      "grad_norm": 0.00024962503812275827,
+      "learning_rate": 1.365690309045429e-05,
+      "loss": 0.0028,
+      "step": 20770
+    },
+    {
+      "epoch": 15.09,
+      "grad_norm": 0.017173290252685547,
+      "learning_rate": 1.363673041232954e-05,
+      "loss": 0.0053,
+      "step": 20780
+    },
+    {
+      "epoch": 15.09,
+      "grad_norm": 0.5537311434745789,
+      "learning_rate": 1.3616557734204793e-05,
+      "loss": 0.0061,
+      "step": 20790
+    },
+    {
+      "epoch": 15.1,
+      "grad_norm": 0.0027972084935754538,
+      "learning_rate": 1.3596385056080047e-05,
+      "loss": 0.0059,
+      "step": 20800
+    },
+    {
+      "epoch": 15.11,
+      "grad_norm": 0.13410428166389465,
+      "learning_rate": 1.3576212377955296e-05,
+      "loss": 0.0033,
+      "step": 20810
+    },
+    {
+      "epoch": 15.11,
+      "grad_norm": 33.71815490722656,
+      "learning_rate": 1.355603969983055e-05,
+      "loss": 0.0124,
+      "step": 20820
+    },
+    {
+      "epoch": 15.12,
+      "grad_norm": 0.003130529774352908,
+      "learning_rate": 1.3535867021705804e-05,
+      "loss": 0.005,
+      "step": 20830
+    },
+    {
+      "epoch": 15.13,
+      "grad_norm": 0.0005914736539125443,
+      "learning_rate": 1.3515694343581053e-05,
+      "loss": 0.0037,
+      "step": 20840
+    },
+    {
+      "epoch": 15.14,
+      "grad_norm": 0.43801841139793396,
+      "learning_rate": 1.3495521665456306e-05,
+      "loss": 0.0078,
+      "step": 20850
+    },
+    {
+      "epoch": 15.14,
+      "grad_norm": 0.0011702912161126733,
+      "learning_rate": 1.347534898733156e-05,
+      "loss": 0.004,
+      "step": 20860
+    },
+    {
+      "epoch": 15.15,
+      "grad_norm": 0.0029354379512369633,
+      "learning_rate": 1.345517630920681e-05,
+      "loss": 0.0,
+      "step": 20870
+    },
+    {
+      "epoch": 15.16,
+      "grad_norm": 0.11435387283563614,
+      "learning_rate": 1.3435003631082063e-05,
+      "loss": 0.0022,
+      "step": 20880
+    },
+    {
+      "epoch": 15.17,
+      "grad_norm": 43.920196533203125,
+      "learning_rate": 1.3414830952957317e-05,
+      "loss": 0.0172,
+      "step": 20890
+    },
+    {
+      "epoch": 15.17,
+      "grad_norm": 32.0137825012207,
+      "learning_rate": 1.3394658274832566e-05,
+      "loss": 0.0257,
+      "step": 20900
+    },
+    {
+      "epoch": 15.18,
+      "grad_norm": 0.0019825261551886797,
+      "learning_rate": 1.3374485596707819e-05,
+      "loss": 0.0005,
+      "step": 20910
+    },
+    {
+      "epoch": 15.19,
+      "grad_norm": 0.001471333671361208,
+      "learning_rate": 1.3354312918583074e-05,
+      "loss": 0.0105,
+      "step": 20920
+    },
+    {
+      "epoch": 15.19,
+      "grad_norm": 0.04608182981610298,
+      "learning_rate": 1.3334140240458323e-05,
+      "loss": 0.008,
+      "step": 20930
+    },
+    {
+      "epoch": 15.2,
+      "grad_norm": 0.058197587728500366,
+      "learning_rate": 1.3313967562333576e-05,
+      "loss": 0.0014,
+      "step": 20940
+    },
+    {
+      "epoch": 15.21,
+      "grad_norm": 0.00060573237715289,
+      "learning_rate": 1.329379488420883e-05,
+      "loss": 0.0039,
+      "step": 20950
+    },
+    {
+      "epoch": 15.22,
+      "grad_norm": 0.0463654026389122,
+      "learning_rate": 1.327362220608408e-05,
+      "loss": 0.0016,
+      "step": 20960
+    },
+    {
+      "epoch": 15.22,
+      "grad_norm": 0.0003432673111092299,
+      "learning_rate": 1.3253449527959332e-05,
+      "loss": 0.0022,
+      "step": 20970
+    },
+    {
+      "epoch": 15.23,
+      "grad_norm": 0.0033386244904249907,
+      "learning_rate": 1.3233276849834587e-05,
+      "loss": 0.0048,
+      "step": 20980
+    },
+    {
+      "epoch": 15.24,
+      "grad_norm": 0.0028187900315970182,
+      "learning_rate": 1.3213104171709836e-05,
+      "loss": 0.0044,
+      "step": 20990
+    },
+    {
+      "epoch": 15.25,
+      "grad_norm": 0.06616388261318207,
+      "learning_rate": 1.3192931493585089e-05,
+      "loss": 0.0025,
+      "step": 21000
+    },
+    {
+      "epoch": 15.25,
+      "grad_norm": 0.0010185488499701023,
+      "learning_rate": 1.317275881546034e-05,
+      "loss": 0.0012,
+      "step": 21010
+    },
+    {
+      "epoch": 15.26,
+      "grad_norm": 0.0004689507477451116,
+      "learning_rate": 1.3152586137335593e-05,
+      "loss": 0.0009,
+      "step": 21020
+    },
+    {
+      "epoch": 15.27,
+      "grad_norm": 0.11378660798072815,
+      "learning_rate": 1.3132413459210846e-05,
+      "loss": 0.0055,
+      "step": 21030
+    },
+    {
+      "epoch": 15.27,
+      "grad_norm": 0.00018735295452643186,
+      "learning_rate": 1.3112240781086097e-05,
+      "loss": 0.0024,
+      "step": 21040
+    },
+    {
+      "epoch": 15.28,
+      "grad_norm": 0.0002505971642676741,
+      "learning_rate": 1.309206810296135e-05,
+      "loss": 0.0048,
+      "step": 21050
+    },
+    {
+      "epoch": 15.29,
+      "grad_norm": 0.0006918599247001112,
+      "learning_rate": 1.3071895424836602e-05,
+      "loss": 0.0057,
+      "step": 21060
+    },
+    {
+      "epoch": 15.3,
+      "grad_norm": 0.0002549978089518845,
+      "learning_rate": 1.3051722746711853e-05,
+      "loss": 0.0058,
+      "step": 21070
+    },
+    {
+      "epoch": 15.3,
+      "grad_norm": 14.719676971435547,
+      "learning_rate": 1.3031550068587106e-05,
+      "loss": 0.0045,
+      "step": 21080
+    },
+    {
+      "epoch": 15.31,
+      "grad_norm": 4.1548017179593444e-05,
+      "learning_rate": 1.3011377390462359e-05,
+      "loss": 0.0027,
+      "step": 21090
+    },
+    {
+      "epoch": 15.32,
+      "grad_norm": 0.0011699338210746646,
+      "learning_rate": 1.299120471233761e-05,
+      "loss": 0.0093,
+      "step": 21100
+    },
+    {
+      "epoch": 15.32,
+      "grad_norm": 0.017013998702168465,
+      "learning_rate": 1.2971032034212863e-05,
+      "loss": 0.0046,
+      "step": 21110
+    },
+    {
+      "epoch": 15.33,
+      "grad_norm": 0.00029067235300317407,
+      "learning_rate": 1.2950859356088115e-05,
+      "loss": 0.001,
+      "step": 21120
+    },
+    {
+      "epoch": 15.34,
+      "grad_norm": 0.07131381332874298,
+      "learning_rate": 1.2930686677963367e-05,
+      "loss": 0.0158,
+      "step": 21130
+    },
+    {
+      "epoch": 15.35,
+      "grad_norm": 0.07471425831317902,
+      "learning_rate": 1.291051399983862e-05,
+      "loss": 0.0028,
+      "step": 21140
+    },
+    {
+      "epoch": 15.35,
+      "grad_norm": 26.187938690185547,
+      "learning_rate": 1.2890341321713872e-05,
+      "loss": 0.0072,
+      "step": 21150
+    },
+    {
+      "epoch": 15.36,
+      "grad_norm": 0.0010599270462989807,
+      "learning_rate": 1.2870168643589123e-05,
+      "loss": 0.0004,
+      "step": 21160
+    },
+    {
+      "epoch": 15.37,
+      "grad_norm": 0.00012274249456822872,
+      "learning_rate": 1.2849995965464376e-05,
+      "loss": 0.0012,
+      "step": 21170
+    },
+    {
+      "epoch": 15.38,
+      "grad_norm": 0.020930590108036995,
+      "learning_rate": 1.2829823287339629e-05,
+      "loss": 0.0088,
+      "step": 21180
+    },
+    {
+      "epoch": 15.38,
+      "grad_norm": 0.0004236644599586725,
+      "learning_rate": 1.280965060921488e-05,
+      "loss": 0.0085,
+      "step": 21190
+    },
+    {
+      "epoch": 15.39,
+      "grad_norm": 0.1861221194267273,
+      "learning_rate": 1.2789477931090133e-05,
+      "loss": 0.0059,
+      "step": 21200
+    },
+    {
+      "epoch": 15.4,
+      "grad_norm": 0.001298606744967401,
+      "learning_rate": 1.2769305252965385e-05,
+      "loss": 0.0005,
+      "step": 21210
+    },
+    {
+      "epoch": 15.4,
+      "grad_norm": 0.002539963461458683,
+      "learning_rate": 1.2749132574840636e-05,
+      "loss": 0.005,
+      "step": 21220
+    },
+    {
+      "epoch": 15.41,
+      "grad_norm": 0.0002390586887486279,
+      "learning_rate": 1.272895989671589e-05,
+      "loss": 0.0085,
+      "step": 21230
+    },
+    {
+      "epoch": 15.42,
+      "grad_norm": 0.003349823411554098,
+      "learning_rate": 1.2708787218591139e-05,
+      "loss": 0.0033,
+      "step": 21240
+    },
+    {
+      "epoch": 15.43,
+      "grad_norm": 9.157711029052734,
+      "learning_rate": 1.2688614540466393e-05,
+      "loss": 0.0108,
+      "step": 21250
+    },
+    {
+      "epoch": 15.43,
+      "grad_norm": 15.022656440734863,
+      "learning_rate": 1.2668441862341646e-05,
+      "loss": 0.0069,
+      "step": 21260
+    },
+    {
+      "epoch": 15.44,
+      "grad_norm": 0.014409718103706837,
+      "learning_rate": 1.2648269184216895e-05,
+      "loss": 0.002,
+      "step": 21270
+    },
+    {
+      "epoch": 15.45,
+      "grad_norm": 0.09259545803070068,
+      "learning_rate": 1.262809650609215e-05,
+      "loss": 0.0064,
+      "step": 21280
+    },
+    {
+      "epoch": 15.46,
+      "grad_norm": 0.013745992444455624,
+      "learning_rate": 1.2607923827967402e-05,
+      "loss": 0.0032,
+      "step": 21290
+    },
+    {
+      "epoch": 15.46,
+      "grad_norm": 0.3294007480144501,
+      "learning_rate": 1.2587751149842652e-05,
+      "loss": 0.005,
+      "step": 21300
+    },
+    {
+      "epoch": 15.47,
+      "grad_norm": 0.008936860598623753,
+      "learning_rate": 1.2567578471717906e-05,
+      "loss": 0.0129,
+      "step": 21310
+    },
+    {
+      "epoch": 15.48,
+      "grad_norm": 0.33121001720428467,
+      "learning_rate": 1.2547405793593159e-05,
+      "loss": 0.0022,
+      "step": 21320
+    },
+    {
+      "epoch": 15.48,
+      "grad_norm": 0.004892081022262573,
+      "learning_rate": 1.2527233115468408e-05,
+      "loss": 0.0023,
+      "step": 21330
+    },
+    {
+      "epoch": 15.49,
+      "grad_norm": 0.0006819283007644117,
+      "learning_rate": 1.2507060437343663e-05,
+      "loss": 0.003,
+      "step": 21340
+    },
+    {
+      "epoch": 15.5,
+      "grad_norm": 0.07798388600349426,
+      "learning_rate": 1.2486887759218914e-05,
+      "loss": 0.0006,
+      "step": 21350
+    },
+    {
+      "epoch": 15.51,
+      "grad_norm": 0.0054915305227041245,
+      "learning_rate": 1.2466715081094167e-05,
+      "loss": 0.0068,
+      "step": 21360
+    },
+    {
+      "epoch": 15.51,
+      "grad_norm": 3.669898509979248,
+      "learning_rate": 1.244654240296942e-05,
+      "loss": 0.0025,
+      "step": 21370
+    },
+    {
+      "epoch": 15.52,
+      "grad_norm": 0.00045484190923161805,
+      "learning_rate": 1.242636972484467e-05,
+      "loss": 0.003,
+      "step": 21380
+    },
+    {
+      "epoch": 15.53,
+      "grad_norm": 21.309757232666016,
+      "learning_rate": 1.2406197046719923e-05,
+      "loss": 0.008,
+      "step": 21390
+    },
+    {
+      "epoch": 15.54,
+      "grad_norm": 0.001877294504083693,
+      "learning_rate": 1.2386024368595176e-05,
+      "loss": 0.0058,
+      "step": 21400
+    },
+    {
+      "epoch": 15.54,
+      "grad_norm": 0.034363534301519394,
+      "learning_rate": 1.2365851690470427e-05,
+      "loss": 0.0159,
+      "step": 21410
+    },
+    {
+      "epoch": 15.55,
+      "grad_norm": 0.6154148578643799,
+      "learning_rate": 1.2345679012345678e-05,
+      "loss": 0.0019,
+      "step": 21420
+    },
+    {
+      "epoch": 15.56,
+      "grad_norm": 0.0009509364608675241,
+      "learning_rate": 1.2325506334220933e-05,
+      "loss": 0.0031,
+      "step": 21430
+    },
+    {
+      "epoch": 15.56,
+      "grad_norm": 0.00043885348713956773,
+      "learning_rate": 1.2305333656096184e-05,
+      "loss": 0.0127,
+      "step": 21440
+    },
+    {
+      "epoch": 15.57,
+      "grad_norm": 0.0027128588408231735,
+      "learning_rate": 1.2285160977971435e-05,
+      "loss": 0.0013,
+      "step": 21450
+    },
+    {
+      "epoch": 15.58,
+      "grad_norm": 2.2161917686462402,
+      "learning_rate": 1.226498829984669e-05,
+      "loss": 0.0108,
+      "step": 21460
+    },
+    {
+      "epoch": 15.59,
+      "grad_norm": 0.13706153631210327,
+      "learning_rate": 1.224481562172194e-05,
+      "loss": 0.0075,
+      "step": 21470
+    },
+    {
+      "epoch": 15.59,
+      "grad_norm": 0.009210462681949139,
+      "learning_rate": 1.2224642943597192e-05,
+      "loss": 0.0047,
+      "step": 21480
+    },
+    {
+      "epoch": 15.6,
+      "grad_norm": 0.3337669372558594,
+      "learning_rate": 1.2204470265472446e-05,
+      "loss": 0.0065,
+      "step": 21490
+    },
+    {
+      "epoch": 15.61,
+      "grad_norm": 0.042242661118507385,
+      "learning_rate": 1.2184297587347697e-05,
+      "loss": 0.0056,
+      "step": 21500
+    },
+    {
+      "epoch": 15.62,
+      "grad_norm": 1.992792010307312,
+      "learning_rate": 1.2164124909222948e-05,
+      "loss": 0.0096,
+      "step": 21510
+    },
+    {
+      "epoch": 15.62,
+      "grad_norm": 0.028501229360699654,
+      "learning_rate": 1.2143952231098203e-05,
+      "loss": 0.0053,
+      "step": 21520
+    },
+    {
+      "epoch": 15.63,
+      "grad_norm": 54.20029830932617,
+      "learning_rate": 1.2123779552973454e-05,
+      "loss": 0.0046,
+      "step": 21530
+    },
+    {
+      "epoch": 15.64,
+      "grad_norm": 0.05367405712604523,
+      "learning_rate": 1.2103606874848705e-05,
+      "loss": 0.0044,
+      "step": 21540
+    },
+    {
+      "epoch": 15.64,
+      "grad_norm": 0.0021568976808339357,
+      "learning_rate": 1.2083434196723958e-05,
+      "loss": 0.0035,
+      "step": 21550
+    },
+    {
+      "epoch": 15.65,
+      "grad_norm": 0.04927918314933777,
+      "learning_rate": 1.206326151859921e-05,
+      "loss": 0.0012,
+      "step": 21560
+    },
+    {
+      "epoch": 15.66,
+      "grad_norm": 0.7703726291656494,
+      "learning_rate": 1.2043088840474461e-05,
+      "loss": 0.0036,
+      "step": 21570
+    },
+    {
+      "epoch": 15.67,
+      "grad_norm": 0.03503729775547981,
+      "learning_rate": 1.2022916162349714e-05,
+      "loss": 0.0018,
+      "step": 21580
+    },
+    {
+      "epoch": 15.67,
+      "grad_norm": 0.01794448494911194,
+      "learning_rate": 1.2002743484224967e-05,
+      "loss": 0.0059,
+      "step": 21590
+    },
+    {
+      "epoch": 15.68,
+      "grad_norm": 0.09404520690441132,
+      "learning_rate": 1.1982570806100218e-05,
+      "loss": 0.0041,
+      "step": 21600
+    },
+    {
+      "epoch": 15.69,
+      "grad_norm": 0.14694754779338837,
+      "learning_rate": 1.196239812797547e-05,
+      "loss": 0.0031,
+      "step": 21610
+    },
+    {
+      "epoch": 15.7,
+      "grad_norm": 0.003776370780542493,
+      "learning_rate": 1.1942225449850724e-05,
+      "loss": 0.007,
+      "step": 21620
+    },
+    {
+      "epoch": 15.7,
+      "grad_norm": 0.0026016957126557827,
+      "learning_rate": 1.1922052771725975e-05,
+      "loss": 0.0078,
+      "step": 21630
+    },
+    {
+      "epoch": 15.71,
+      "grad_norm": 0.011103923432528973,
+      "learning_rate": 1.1901880093601227e-05,
+      "loss": 0.0102,
+      "step": 21640
+    },
+    {
+      "epoch": 15.72,
+      "grad_norm": 0.13180720806121826,
+      "learning_rate": 1.188170741547648e-05,
+      "loss": 0.0035,
+      "step": 21650
+    },
+    {
+      "epoch": 15.72,
+      "grad_norm": 16.478391647338867,
+      "learning_rate": 1.1861534737351731e-05,
+      "loss": 0.0131,
+      "step": 21660
+    },
+    {
+      "epoch": 15.73,
+      "grad_norm": 0.001228134031407535,
+      "learning_rate": 1.1841362059226984e-05,
+      "loss": 0.0045,
+      "step": 21670
+    },
+    {
+      "epoch": 15.74,
+      "grad_norm": 0.0005743647925555706,
+      "learning_rate": 1.1821189381102235e-05,
+      "loss": 0.0022,
+      "step": 21680
+    },
+    {
+      "epoch": 15.75,
+      "grad_norm": 0.07435394078493118,
+      "learning_rate": 1.1801016702977488e-05,
+      "loss": 0.004,
+      "step": 21690
+    },
+    {
+      "epoch": 15.75,
+      "grad_norm": 0.0004962944076396525,
+      "learning_rate": 1.178084402485274e-05,
+      "loss": 0.006,
+      "step": 21700
+    },
+    {
+      "epoch": 15.76,
+      "grad_norm": 0.008445637300610542,
+      "learning_rate": 1.1760671346727992e-05,
+      "loss": 0.0014,
+      "step": 21710
+    },
+    {
+      "epoch": 15.77,
+      "grad_norm": 0.08228937536478043,
+      "learning_rate": 1.1740498668603244e-05,
+      "loss": 0.0041,
+      "step": 21720
+    },
+    {
+      "epoch": 15.77,
+      "grad_norm": 0.001986768562346697,
+      "learning_rate": 1.1720325990478497e-05,
+      "loss": 0.0006,
+      "step": 21730
+    },
+    {
+      "epoch": 15.78,
+      "grad_norm": 0.030242083594202995,
+      "learning_rate": 1.1700153312353748e-05,
+      "loss": 0.0026,
+      "step": 21740
+    },
+    {
+      "epoch": 15.79,
+      "grad_norm": 0.000717415998224169,
+      "learning_rate": 1.1679980634229001e-05,
+      "loss": 0.0036,
+      "step": 21750
+    },
+    {
+      "epoch": 15.8,
+      "grad_norm": 0.001572469249367714,
+      "learning_rate": 1.1659807956104254e-05,
+      "loss": 0.0053,
+      "step": 21760
+    },
+    {
+      "epoch": 15.8,
+      "grad_norm": 0.025059890002012253,
+      "learning_rate": 1.1639635277979505e-05,
+      "loss": 0.0017,
+      "step": 21770
+    },
+    {
+      "epoch": 15.81,
+      "grad_norm": 0.0006473218672908843,
+      "learning_rate": 1.1619462599854758e-05,
+      "loss": 0.0061,
+      "step": 21780
+    },
+    {
+      "epoch": 15.82,
+      "grad_norm": 0.12124957889318466,
+      "learning_rate": 1.159928992173001e-05,
+      "loss": 0.0038,
+      "step": 21790
+    },
+    {
+      "epoch": 15.83,
+      "grad_norm": 0.2481914907693863,
+      "learning_rate": 1.1579117243605262e-05,
+      "loss": 0.0028,
+      "step": 21800
+    },
+    {
+      "epoch": 15.83,
+      "grad_norm": 0.0032148726750165224,
+      "learning_rate": 1.1558944565480513e-05,
+      "loss": 0.0047,
+      "step": 21810
+    },
+    {
+      "epoch": 15.84,
+      "grad_norm": 0.00942063145339489,
+      "learning_rate": 1.1538771887355765e-05,
+      "loss": 0.001,
+      "step": 21820
+    },
+    {
+      "epoch": 15.85,
+      "grad_norm": 0.04141150042414665,
+      "learning_rate": 1.1518599209231018e-05,
+      "loss": 0.0036,
+      "step": 21830
+    },
+    {
+      "epoch": 15.85,
+      "grad_norm": 0.03869690001010895,
+      "learning_rate": 1.149842653110627e-05,
+      "loss": 0.0027,
+      "step": 21840
+    },
+    {
+      "epoch": 15.86,
+      "grad_norm": 0.0011927615851163864,
+      "learning_rate": 1.1478253852981522e-05,
+      "loss": 0.0017,
+      "step": 21850
+    },
+    {
+      "epoch": 15.87,
+      "grad_norm": 0.008258639834821224,
+      "learning_rate": 1.1458081174856775e-05,
+      "loss": 0.0005,
+      "step": 21860
+    },
+    {
+      "epoch": 15.88,
+      "grad_norm": 0.006110721733421087,
+      "learning_rate": 1.1437908496732026e-05,
+      "loss": 0.0054,
+      "step": 21870
+    },
+    {
+      "epoch": 15.88,
+      "grad_norm": 0.15199817717075348,
+      "learning_rate": 1.1417735818607279e-05,
+      "loss": 0.0043,
+      "step": 21880
+    },
+    {
+      "epoch": 15.89,
+      "grad_norm": 0.0010181881953030825,
+      "learning_rate": 1.1397563140482531e-05,
+      "loss": 0.0037,
+      "step": 21890
+    },
+    {
+      "epoch": 15.9,
+      "grad_norm": 0.0009028404019773006,
+      "learning_rate": 1.1377390462357783e-05,
+      "loss": 0.0042,
+      "step": 21900
+    },
+    {
+      "epoch": 15.91,
+      "grad_norm": 14.625401496887207,
+      "learning_rate": 1.1357217784233035e-05,
+      "loss": 0.016,
+      "step": 21910
+    },
+    {
+      "epoch": 15.91,
+      "grad_norm": 0.0001988597068702802,
+      "learning_rate": 1.1337045106108288e-05,
+      "loss": 0.0017,
+      "step": 21920
+    },
+    {
+      "epoch": 15.92,
+      "grad_norm": 0.000211259801289998,
+      "learning_rate": 1.1316872427983539e-05,
+      "loss": 0.0051,
+      "step": 21930
+    },
+    {
+      "epoch": 15.93,
+      "grad_norm": 3.789830952882767e-05,
+      "learning_rate": 1.1296699749858792e-05,
+      "loss": 0.001,
+      "step": 21940
+    },
+    {
+      "epoch": 15.93,
+      "grad_norm": 3.52205170202069e-05,
+      "learning_rate": 1.1276527071734045e-05,
+      "loss": 0.0026,
+      "step": 21950
+    },
+    {
+      "epoch": 15.94,
+      "grad_norm": 0.0002518637338653207,
+      "learning_rate": 1.1256354393609296e-05,
+      "loss": 0.004,
+      "step": 21960
+    },
+    {
+      "epoch": 15.95,
+      "grad_norm": 0.0853363573551178,
+      "learning_rate": 1.1236181715484549e-05,
+      "loss": 0.0112,
+      "step": 21970
+    },
+    {
+      "epoch": 15.96,
+      "grad_norm": 0.0005988592747598886,
+      "learning_rate": 1.1216009037359801e-05,
+      "loss": 0.0028,
+      "step": 21980
+    },
+    {
+      "epoch": 15.96,
+      "grad_norm": 0.07639417797327042,
+      "learning_rate": 1.1195836359235052e-05,
+      "loss": 0.0199,
+      "step": 21990
+    },
+    {
+      "epoch": 15.97,
+      "grad_norm": 0.0889679491519928,
+      "learning_rate": 1.1175663681110305e-05,
+      "loss": 0.0067,
+      "step": 22000
+    },
+    {
+      "epoch": 15.98,
+      "grad_norm": 0.0020072232000529766,
+      "learning_rate": 1.1155491002985558e-05,
+      "loss": 0.0003,
+      "step": 22010
+    },
+    {
+      "epoch": 15.99,
+      "grad_norm": 0.05244208872318268,
+      "learning_rate": 1.1135318324860809e-05,
+      "loss": 0.0046,
+      "step": 22020
+    },
+    {
+      "epoch": 15.99,
+      "grad_norm": 0.0003801613347604871,
+      "learning_rate": 1.1115145646736062e-05,
+      "loss": 0.0049,
+      "step": 22030
+    },
+    {
+      "epoch": 16.0,
+      "grad_norm": 0.0038860926870256662,
+      "learning_rate": 1.1094972968611313e-05,
+      "loss": 0.0154,
+      "step": 22040
+    },
+    {
+      "epoch": 16.0,
+      "eval_accuracy": 0.9983436946952852,
+      "eval_f1": 0.9961762086847206,
+      "eval_loss": 0.003294909605756402,
+      "eval_precision": 0.9995795227583307,
+      "eval_recall": 0.9927959908122781,
+      "eval_roc_auc": 0.9999867430709279,
+      "eval_runtime": 387.006,
+      "eval_samples_per_second": 227.769,
+      "eval_steps_per_second": 14.238,
+      "step": 22040
+    },
+    {
+      "epoch": 16.01,
+      "grad_norm": 0.0014847785932943225,
+      "learning_rate": 1.1074800290486566e-05,
+      "loss": 0.0051,
+      "step": 22050
+    },
+    {
+      "epoch": 16.01,
+      "grad_norm": 2.5662457942962646,
+      "learning_rate": 1.1054627612361818e-05,
+      "loss": 0.0037,
+      "step": 22060
+    },
+    {
+      "epoch": 16.02,
+      "grad_norm": 0.0007208075257949531,
+      "learning_rate": 1.103445493423707e-05,
+      "loss": 0.004,
+      "step": 22070
+    },
+    {
+      "epoch": 16.03,
+      "grad_norm": 0.0007924524252302945,
+      "learning_rate": 1.1014282256112322e-05,
+      "loss": 0.0034,
+      "step": 22080
+    },
+    {
+      "epoch": 16.04,
+      "grad_norm": 0.15768921375274658,
+      "learning_rate": 1.0994109577987573e-05,
+      "loss": 0.0037,
+      "step": 22090
+    },
+    {
+      "epoch": 16.04,
+      "grad_norm": 0.0002586398331914097,
+      "learning_rate": 1.0973936899862826e-05,
+      "loss": 0.0021,
+      "step": 22100
+    },
+    {
+      "epoch": 16.05,
+      "grad_norm": 0.005661542993038893,
+      "learning_rate": 1.0953764221738079e-05,
+      "loss": 0.0031,
+      "step": 22110
+    },
+    {
+      "epoch": 16.06,
+      "grad_norm": 0.0002936258970294148,
+      "learning_rate": 1.093359154361333e-05,
+      "loss": 0.0038,
+      "step": 22120
+    },
+    {
+      "epoch": 16.07,
+      "grad_norm": 0.001346628530882299,
+      "learning_rate": 1.0913418865488583e-05,
+      "loss": 0.0014,
+      "step": 22130
+    },
+    {
+      "epoch": 16.07,
+      "grad_norm": 0.08555719256401062,
+      "learning_rate": 1.0893246187363835e-05,
+      "loss": 0.0173,
+      "step": 22140
+    },
+    {
+      "epoch": 16.08,
+      "grad_norm": 0.00019549914577510208,
+      "learning_rate": 1.0873073509239087e-05,
+      "loss": 0.0,
+      "step": 22150
+    },
+    {
+      "epoch": 16.09,
+      "grad_norm": 0.0027101049199700356,
+      "learning_rate": 1.085290083111434e-05,
+      "loss": 0.0052,
+      "step": 22160
+    },
+    {
+      "epoch": 16.09,
+      "grad_norm": 0.00028255765209905803,
+      "learning_rate": 1.083272815298959e-05,
+      "loss": 0.0046,
+      "step": 22170
+    },
+    {
+      "epoch": 16.1,
+      "grad_norm": 0.0008104601292870939,
+      "learning_rate": 1.0812555474864843e-05,
+      "loss": 0.001,
+      "step": 22180
+    },
+    {
+      "epoch": 16.11,
+      "grad_norm": 0.09307300299406052,
+      "learning_rate": 1.0792382796740096e-05,
+      "loss": 0.0097,
+      "step": 22190
+    },
+    {
+      "epoch": 16.12,
+      "grad_norm": 0.00042667603702284396,
+      "learning_rate": 1.0772210118615347e-05,
+      "loss": 0.0029,
+      "step": 22200
+    },
+    {
+      "epoch": 16.12,
+      "grad_norm": 0.03504578396677971,
+      "learning_rate": 1.07520374404906e-05,
+      "loss": 0.0032,
+      "step": 22210
+    },
+    {
+      "epoch": 16.13,
+      "grad_norm": 0.007888894528150558,
+      "learning_rate": 1.0731864762365853e-05,
+      "loss": 0.0021,
+      "step": 22220
+    },
+    {
+      "epoch": 16.14,
+      "grad_norm": 0.09013179689645767,
+      "learning_rate": 1.0711692084241104e-05,
+      "loss": 0.0031,
+      "step": 22230
+    },
+    {
+      "epoch": 16.15,
+      "grad_norm": 0.00033059256384149194,
+      "learning_rate": 1.0691519406116356e-05,
+      "loss": 0.0042,
+      "step": 22240
+    },
+    {
+      "epoch": 16.15,
+      "grad_norm": 0.0006427318439818919,
+      "learning_rate": 1.067134672799161e-05,
+      "loss": 0.0105,
+      "step": 22250
+    },
+    {
+      "epoch": 16.16,
+      "grad_norm": 0.0007930251304060221,
+      "learning_rate": 1.065117404986686e-05,
+      "loss": 0.0002,
+      "step": 22260
+    },
+    {
+      "epoch": 16.17,
+      "grad_norm": 0.08082820475101471,
+      "learning_rate": 1.0631001371742113e-05,
+      "loss": 0.0042,
+      "step": 22270
+    },
+    {
+      "epoch": 16.17,
+      "grad_norm": 0.10491488128900528,
+      "learning_rate": 1.0610828693617366e-05,
+      "loss": 0.0029,
+      "step": 22280
+    },
+    {
+      "epoch": 16.18,
+      "grad_norm": 0.0001469424896640703,
+      "learning_rate": 1.0590656015492617e-05,
+      "loss": 0.0028,
+      "step": 22290
+    },
+    {
+      "epoch": 16.19,
+      "grad_norm": 0.041843656450510025,
+      "learning_rate": 1.057048333736787e-05,
+      "loss": 0.0042,
+      "step": 22300
+    },
+    {
+      "epoch": 16.2,
+      "grad_norm": 0.00035109708551317453,
+      "learning_rate": 1.0550310659243122e-05,
+      "loss": 0.0024,
+      "step": 22310
+    },
+    {
+      "epoch": 16.2,
+      "grad_norm": 0.0014608385972678661,
+      "learning_rate": 1.0530137981118374e-05,
+      "loss": 0.0017,
+      "step": 22320
+    },
+    {
+      "epoch": 16.21,
+      "grad_norm": 0.00027612573467195034,
+      "learning_rate": 1.0509965302993625e-05,
+      "loss": 0.0008,
+      "step": 22330
+    },
+    {
+      "epoch": 16.22,
+      "grad_norm": 2.363631010055542,
+      "learning_rate": 1.0489792624868879e-05,
+      "loss": 0.0051,
+      "step": 22340
+    },
+    {
+      "epoch": 16.23,
+      "grad_norm": 0.0002950582420453429,
+      "learning_rate": 1.046961994674413e-05,
+      "loss": 0.0023,
+      "step": 22350
+    },
+    {
+      "epoch": 16.23,
+      "grad_norm": 0.0681629404425621,
+      "learning_rate": 1.0449447268619381e-05,
+      "loss": 0.0004,
+      "step": 22360
+    },
+    {
+      "epoch": 16.24,
+      "grad_norm": 0.00012084317131666467,
+      "learning_rate": 1.0429274590494636e-05,
+      "loss": 0.0,
+      "step": 22370
+    },
+    {
+      "epoch": 16.25,
+      "grad_norm": 0.00011871389142470434,
+      "learning_rate": 1.0409101912369887e-05,
+      "loss": 0.0026,
+      "step": 22380
+    },
+    {
+      "epoch": 16.25,
+      "grad_norm": 0.0011634239926934242,
+      "learning_rate": 1.0388929234245138e-05,
+      "loss": 0.0009,
+      "step": 22390
+    },
+    {
+      "epoch": 16.26,
+      "grad_norm": 0.20264987647533417,
+      "learning_rate": 1.0368756556120392e-05,
+      "loss": 0.0036,
+      "step": 22400
+    },
+    {
+      "epoch": 16.27,
+      "grad_norm": 0.0032964616548269987,
+      "learning_rate": 1.0348583877995643e-05,
+      "loss": 0.0023,
+      "step": 22410
+    },
+    {
+      "epoch": 16.28,
+      "grad_norm": 0.07189714908599854,
+      "learning_rate": 1.0328411199870894e-05,
+      "loss": 0.0023,
+      "step": 22420
+    },
+    {
+      "epoch": 16.28,
+      "grad_norm": 2.3079440593719482,
+      "learning_rate": 1.0308238521746147e-05,
+      "loss": 0.0012,
+      "step": 22430
+    },
+    {
+      "epoch": 16.29,
+      "grad_norm": 0.007369950879365206,
+      "learning_rate": 1.02880658436214e-05,
+      "loss": 0.0009,
+      "step": 22440
+    },
+    {
+      "epoch": 16.3,
+      "grad_norm": 0.00015242438530549407,
+      "learning_rate": 1.0267893165496651e-05,
+      "loss": 0.0009,
+      "step": 22450
+    },
+    {
+      "epoch": 16.3,
+      "grad_norm": 0.08135165274143219,
+      "learning_rate": 1.0247720487371904e-05,
+      "loss": 0.0084,
+      "step": 22460
+    },
+    {
+      "epoch": 16.31,
+      "grad_norm": 0.04192609712481499,
+      "learning_rate": 1.0227547809247157e-05,
+      "loss": 0.0042,
+      "step": 22470
+    },
+    {
+      "epoch": 16.32,
+      "grad_norm": 0.0002498167159501463,
+      "learning_rate": 1.0207375131122408e-05,
+      "loss": 0.0035,
+      "step": 22480
+    },
+    {
+      "epoch": 16.33,
+      "grad_norm": 0.005605372134596109,
+      "learning_rate": 1.018720245299766e-05,
+      "loss": 0.0018,
+      "step": 22490
+    },
+    {
+      "epoch": 16.33,
+      "grad_norm": 0.04002232104539871,
+      "learning_rate": 1.0167029774872913e-05,
+      "loss": 0.0126,
+      "step": 22500
+    },
+    {
+      "epoch": 16.34,
+      "grad_norm": 0.0337141789495945,
+      "learning_rate": 1.0146857096748164e-05,
+      "loss": 0.0093,
+      "step": 22510
+    },
+    {
+      "epoch": 16.35,
+      "grad_norm": 0.005516758654266596,
+      "learning_rate": 1.0126684418623417e-05,
+      "loss": 0.004,
+      "step": 22520
+    },
+    {
+      "epoch": 16.36,
+      "grad_norm": 0.03148741275072098,
+      "learning_rate": 1.010651174049867e-05,
+      "loss": 0.0064,
+      "step": 22530
+    },
+    {
+      "epoch": 16.36,
+      "grad_norm": 0.02057914063334465,
+      "learning_rate": 1.0086339062373921e-05,
+      "loss": 0.0183,
+      "step": 22540
+    },
+    {
+      "epoch": 16.37,
+      "grad_norm": 0.13048245012760162,
+      "learning_rate": 1.0066166384249174e-05,
+      "loss": 0.0046,
+      "step": 22550
+    },
+    {
+      "epoch": 16.38,
+      "grad_norm": 0.0004054498567711562,
+      "learning_rate": 1.0045993706124425e-05,
+      "loss": 0.0001,
+      "step": 22560
+    },
+    {
+      "epoch": 16.38,
+      "grad_norm": 0.0032589773181825876,
+      "learning_rate": 1.0025821027999678e-05,
+      "loss": 0.0005,
+      "step": 22570
+    },
+    {
+      "epoch": 16.39,
+      "grad_norm": 0.0005514941876754165,
+      "learning_rate": 1.000564834987493e-05,
+      "loss": 0.0006,
+      "step": 22580
+    },
+    {
+      "epoch": 16.4,
+      "grad_norm": 0.012021565809845924,
+      "learning_rate": 9.985475671750181e-06,
+      "loss": 0.0092,
+      "step": 22590
+    },
+    {
+      "epoch": 16.41,
+      "grad_norm": 0.0435185469686985,
+      "learning_rate": 9.965302993625434e-06,
+      "loss": 0.0035,
+      "step": 22600
+    },
+    {
+      "epoch": 16.41,
+      "grad_norm": 0.0010871135164052248,
+      "learning_rate": 9.945130315500687e-06,
+      "loss": 0.0096,
+      "step": 22610
+    },
+    {
+      "epoch": 16.42,
+      "grad_norm": 0.010872997343540192,
+      "learning_rate": 9.924957637375938e-06,
+      "loss": 0.0052,
+      "step": 22620
+    },
+    {
+      "epoch": 16.43,
+      "grad_norm": 0.0006780479452572763,
+      "learning_rate": 9.90478495925119e-06,
+      "loss": 0.001,
+      "step": 22630
+    },
+    {
+      "epoch": 16.44,
+      "grad_norm": 0.03470781072974205,
+      "learning_rate": 9.884612281126444e-06,
+      "loss": 0.0095,
+      "step": 22640
+    },
+    {
+      "epoch": 16.44,
+      "grad_norm": 0.010145553387701511,
+      "learning_rate": 9.864439603001695e-06,
+      "loss": 0.01,
+      "step": 22650
+    },
+    {
+      "epoch": 16.45,
+      "grad_norm": 0.0024723324459046125,
+      "learning_rate": 9.844266924876946e-06,
+      "loss": 0.0012,
+      "step": 22660
+    },
+    {
+      "epoch": 16.46,
+      "grad_norm": 0.19039224088191986,
+      "learning_rate": 9.8240942467522e-06,
+      "loss": 0.0057,
+      "step": 22670
+    },
+    {
+      "epoch": 16.46,
+      "grad_norm": 0.03419295325875282,
+      "learning_rate": 9.803921568627451e-06,
+      "loss": 0.0,
+      "step": 22680
+    },
+    {
+      "epoch": 16.47,
+      "grad_norm": 0.000138914241688326,
+      "learning_rate": 9.783748890502702e-06,
+      "loss": 0.0087,
+      "step": 22690
+    },
+    {
+      "epoch": 16.48,
+      "grad_norm": 0.057385221123695374,
+      "learning_rate": 9.763576212377957e-06,
+      "loss": 0.0065,
+      "step": 22700
+    },
+    {
+      "epoch": 16.49,
+      "grad_norm": 0.08868400007486343,
+      "learning_rate": 9.743403534253208e-06,
+      "loss": 0.0049,
+      "step": 22710
+    },
+    {
+      "epoch": 16.49,
+      "grad_norm": 0.0007755668484605849,
+      "learning_rate": 9.723230856128459e-06,
+      "loss": 0.0025,
+      "step": 22720
+    },
+    {
+      "epoch": 16.5,
+      "grad_norm": 0.004135680850595236,
+      "learning_rate": 9.703058178003713e-06,
+      "loss": 0.0026,
+      "step": 22730
+    },
+    {
+      "epoch": 16.51,
+      "grad_norm": 0.0001737570419209078,
+      "learning_rate": 9.682885499878964e-06,
+      "loss": 0.0047,
+      "step": 22740
+    },
+    {
+      "epoch": 16.52,
+      "grad_norm": 0.207319438457489,
+      "learning_rate": 9.662712821754216e-06,
+      "loss": 0.0057,
+      "step": 22750
+    },
+    {
+      "epoch": 16.52,
+      "grad_norm": 0.5539862513542175,
+      "learning_rate": 9.64254014362947e-06,
+      "loss": 0.0059,
+      "step": 22760
+    },
+    {
+      "epoch": 16.53,
+      "grad_norm": 0.205776184797287,
+      "learning_rate": 9.622367465504721e-06,
+      "loss": 0.0066,
+      "step": 22770
+    },
+    {
+      "epoch": 16.54,
+      "grad_norm": 0.16564740240573883,
+      "learning_rate": 9.602194787379972e-06,
+      "loss": 0.0014,
+      "step": 22780
+    },
+    {
+      "epoch": 16.54,
+      "grad_norm": 0.0013297771802172065,
+      "learning_rate": 9.582022109255225e-06,
+      "loss": 0.003,
+      "step": 22790
+    },
+    {
+      "epoch": 16.55,
+      "grad_norm": 2.7137090000906028e-05,
+      "learning_rate": 9.561849431130478e-06,
+      "loss": 0.0025,
+      "step": 22800
+    },
+    {
+      "epoch": 16.56,
+      "grad_norm": 0.00012067361240042374,
+      "learning_rate": 9.541676753005729e-06,
+      "loss": 0.0004,
+      "step": 22810
+    },
+    {
+      "epoch": 16.57,
+      "grad_norm": 0.0005142318550497293,
+      "learning_rate": 9.521504074880982e-06,
+      "loss": 0.0014,
+      "step": 22820
+    },
+    {
+      "epoch": 16.57,
+      "grad_norm": 8.987193723442033e-05,
+      "learning_rate": 9.501331396756234e-06,
+      "loss": 0.0074,
+      "step": 22830
+    },
+    {
+      "epoch": 16.58,
+      "grad_norm": 0.04224991053342819,
+      "learning_rate": 9.481158718631485e-06,
+      "loss": 0.0041,
+      "step": 22840
+    },
+    {
+      "epoch": 16.59,
+      "grad_norm": 0.0012686103582382202,
+      "learning_rate": 9.460986040506738e-06,
+      "loss": 0.0082,
+      "step": 22850
+    },
+    {
+      "epoch": 16.6,
+      "grad_norm": 1.1625761985778809,
+      "learning_rate": 9.440813362381991e-06,
+      "loss": 0.0056,
+      "step": 22860
+    },
+    {
+      "epoch": 16.6,
+      "grad_norm": 13.984100341796875,
+      "learning_rate": 9.420640684257242e-06,
+      "loss": 0.001,
+      "step": 22870
+    },
+    {
+      "epoch": 16.61,
+      "grad_norm": 0.0042472160421311855,
+      "learning_rate": 9.400468006132495e-06,
+      "loss": 0.0008,
+      "step": 22880
+    },
+    {
+      "epoch": 16.62,
+      "grad_norm": 0.013330524787306786,
+      "learning_rate": 9.380295328007748e-06,
+      "loss": 0.0023,
+      "step": 22890
+    },
+    {
+      "epoch": 16.62,
+      "grad_norm": 0.002952015260234475,
+      "learning_rate": 9.360122649882999e-06,
+      "loss": 0.0026,
+      "step": 22900
+    },
+    {
+      "epoch": 16.63,
+      "grad_norm": 6.471742381108925e-05,
+      "learning_rate": 9.339949971758251e-06,
+      "loss": 0.0049,
+      "step": 22910
+    },
+    {
+      "epoch": 16.64,
+      "grad_norm": 0.000877292244695127,
+      "learning_rate": 9.319777293633503e-06,
+      "loss": 0.0028,
+      "step": 22920
+    },
+    {
+      "epoch": 16.65,
+      "grad_norm": 0.001535005052573979,
+      "learning_rate": 9.299604615508755e-06,
+      "loss": 0.0016,
+      "step": 22930
+    },
+    {
+      "epoch": 16.65,
+      "grad_norm": 54.134674072265625,
+      "learning_rate": 9.279431937384008e-06,
+      "loss": 0.0048,
+      "step": 22940
+    },
+    {
+      "epoch": 16.66,
+      "grad_norm": 0.09877274930477142,
+      "learning_rate": 9.259259259259259e-06,
+      "loss": 0.0037,
+      "step": 22950
+    },
+    {
+      "epoch": 16.67,
+      "grad_norm": 0.03050428256392479,
+      "learning_rate": 9.239086581134512e-06,
+      "loss": 0.0016,
+      "step": 22960
+    },
+    {
+      "epoch": 16.68,
+      "grad_norm": 9.728727309266105e-05,
+      "learning_rate": 9.218913903009765e-06,
+      "loss": 0.0028,
+      "step": 22970
+    },
+    {
+      "epoch": 16.68,
+      "grad_norm": 0.10689128190279007,
+      "learning_rate": 9.198741224885016e-06,
+      "loss": 0.0013,
+      "step": 22980
+    },
+    {
+      "epoch": 16.69,
+      "grad_norm": 0.0006969768437556922,
+      "learning_rate": 9.178568546760269e-06,
+      "loss": 0.0045,
+      "step": 22990
+    },
+    {
+      "epoch": 16.7,
+      "grad_norm": 0.0004667758184950799,
+      "learning_rate": 9.158395868635521e-06,
+      "loss": 0.002,
+      "step": 23000
+    },
+    {
+      "epoch": 16.7,
+      "grad_norm": 0.3971853256225586,
+      "learning_rate": 9.138223190510772e-06,
+      "loss": 0.003,
+      "step": 23010
+    },
+    {
+      "epoch": 16.71,
+      "grad_norm": 6.46812331979163e-05,
+      "learning_rate": 9.118050512386025e-06,
+      "loss": 0.0051,
+      "step": 23020
+    },
+    {
+      "epoch": 16.72,
+      "grad_norm": 0.02988688088953495,
+      "learning_rate": 9.097877834261278e-06,
+      "loss": 0.0013,
+      "step": 23030
+    },
+    {
+      "epoch": 16.73,
+      "grad_norm": 0.0004272510705050081,
+      "learning_rate": 9.077705156136529e-06,
+      "loss": 0.0292,
+      "step": 23040
+    },
+    {
+      "epoch": 16.73,
+      "grad_norm": 0.00036017660750076175,
+      "learning_rate": 9.05753247801178e-06,
+      "loss": 0.0035,
+      "step": 23050
+    },
+    {
+      "epoch": 16.74,
+      "grad_norm": 0.00015025348693598062,
+      "learning_rate": 9.037359799887035e-06,
+      "loss": 0.0121,
+      "step": 23060
+    },
+    {
+      "epoch": 16.75,
+      "grad_norm": 0.054434388875961304,
+      "learning_rate": 9.017187121762286e-06,
+      "loss": 0.002,
+      "step": 23070
+    },
+    {
+      "epoch": 16.75,
+      "grad_norm": 0.000985904480330646,
+      "learning_rate": 8.997014443637537e-06,
+      "loss": 0.0038,
+      "step": 23080
+    },
+    {
+      "epoch": 16.76,
+      "grad_norm": 0.12436607480049133,
+      "learning_rate": 8.976841765512791e-06,
+      "loss": 0.0044,
+      "step": 23090
+    },
+    {
+      "epoch": 16.77,
+      "grad_norm": 0.03900473937392235,
+      "learning_rate": 8.956669087388042e-06,
+      "loss": 0.0127,
+      "step": 23100
+    },
+    {
+      "epoch": 16.78,
+      "grad_norm": 0.0008112489013001323,
+      "learning_rate": 8.936496409263293e-06,
+      "loss": 0.0,
+      "step": 23110
+    },
+    {
+      "epoch": 16.78,
+      "grad_norm": 0.034635029733181,
+      "learning_rate": 8.916323731138548e-06,
+      "loss": 0.0046,
+      "step": 23120
+    },
+    {
+      "epoch": 16.79,
+      "grad_norm": 0.0005446127033792436,
+      "learning_rate": 8.896151053013799e-06,
+      "loss": 0.0039,
+      "step": 23130
+    },
+    {
+      "epoch": 16.8,
+      "grad_norm": 0.02086440846323967,
+      "learning_rate": 8.87597837488905e-06,
+      "loss": 0.0056,
+      "step": 23140
+    },
+    {
+      "epoch": 16.81,
+      "grad_norm": 0.2668115198612213,
+      "learning_rate": 8.855805696764303e-06,
+      "loss": 0.0094,
+      "step": 23150
+    },
+    {
+      "epoch": 16.81,
+      "grad_norm": 0.0030102794989943504,
+      "learning_rate": 8.835633018639555e-06,
+      "loss": 0.003,
+      "step": 23160
+    },
+    {
+      "epoch": 16.82,
+      "grad_norm": 0.08962462097406387,
+      "learning_rate": 8.815460340514807e-06,
+      "loss": 0.0043,
+      "step": 23170
+    },
+    {
+      "epoch": 16.83,
+      "grad_norm": 0.0004162557306699455,
+      "learning_rate": 8.79528766239006e-06,
+      "loss": 0.0011,
+      "step": 23180
+    },
+    {
+      "epoch": 16.83,
+      "grad_norm": 0.08701346069574356,
+      "learning_rate": 8.775114984265312e-06,
+      "loss": 0.0063,
+      "step": 23190
+    },
+    {
+      "epoch": 16.84,
+      "grad_norm": 0.0005376107874326408,
+      "learning_rate": 8.754942306140563e-06,
+      "loss": 0.0008,
+      "step": 23200
+    },
+    {
+      "epoch": 16.85,
+      "grad_norm": 0.002278102096170187,
+      "learning_rate": 8.734769628015816e-06,
+      "loss": 0.0049,
+      "step": 23210
+    },
+    {
+      "epoch": 16.86,
+      "grad_norm": 0.0011193858226761222,
+      "learning_rate": 8.714596949891069e-06,
+      "loss": 0.0105,
+      "step": 23220
+    },
+    {
+      "epoch": 16.86,
+      "grad_norm": 0.03300873935222626,
+      "learning_rate": 8.69442427176632e-06,
+      "loss": 0.0038,
+      "step": 23230
+    },
+    {
+      "epoch": 16.87,
+      "grad_norm": 0.08699040114879608,
+      "learning_rate": 8.674251593641573e-06,
+      "loss": 0.0095,
+      "step": 23240
+    },
+    {
+      "epoch": 16.88,
+      "grad_norm": 0.13259164988994598,
+      "learning_rate": 8.654078915516825e-06,
+      "loss": 0.0089,
+      "step": 23250
+    },
+    {
+      "epoch": 16.89,
+      "grad_norm": 0.011855214834213257,
+      "learning_rate": 8.633906237392076e-06,
+      "loss": 0.0059,
+      "step": 23260
+    },
+    {
+      "epoch": 16.89,
+      "grad_norm": 0.001402218360453844,
+      "learning_rate": 8.61373355926733e-06,
+      "loss": 0.0048,
+      "step": 23270
+    },
+    {
+      "epoch": 16.9,
+      "grad_norm": 0.0021518440917134285,
+      "learning_rate": 8.59356088114258e-06,
+      "loss": 0.0064,
+      "step": 23280
+    },
+    {
+      "epoch": 16.91,
+      "grad_norm": 0.03953978046774864,
+      "learning_rate": 8.573388203017833e-06,
+      "loss": 0.0266,
+      "step": 23290
+    },
+    {
+      "epoch": 16.91,
+      "grad_norm": 0.002244510455057025,
+      "learning_rate": 8.553215524893086e-06,
+      "loss": 0.0006,
+      "step": 23300
+    },
+    {
+      "epoch": 16.92,
+      "grad_norm": 0.03700876981019974,
+      "learning_rate": 8.533042846768337e-06,
+      "loss": 0.0026,
+      "step": 23310
+    },
+    {
+      "epoch": 16.93,
+      "grad_norm": 0.005215387791395187,
+      "learning_rate": 8.51287016864359e-06,
+      "loss": 0.0039,
+      "step": 23320
+    },
+    {
+      "epoch": 16.94,
+      "grad_norm": 0.0015378224197775126,
+      "learning_rate": 8.492697490518842e-06,
+      "loss": 0.0042,
+      "step": 23330
+    },
+    {
+      "epoch": 16.94,
+      "grad_norm": 0.05736195296049118,
+      "learning_rate": 8.472524812394094e-06,
+      "loss": 0.0039,
+      "step": 23340
+    },
+    {
+      "epoch": 16.95,
+      "grad_norm": 0.03182898834347725,
+      "learning_rate": 8.452352134269346e-06,
+      "loss": 0.0086,
+      "step": 23350
+    },
+    {
+      "epoch": 16.96,
+      "grad_norm": 7.0105366706848145,
+      "learning_rate": 8.432179456144599e-06,
+      "loss": 0.006,
+      "step": 23360
+    },
+    {
+      "epoch": 16.97,
+      "grad_norm": 0.006832567043602467,
+      "learning_rate": 8.41200677801985e-06,
+      "loss": 0.0043,
+      "step": 23370
+    },
+    {
+      "epoch": 16.97,
+      "grad_norm": 0.06288957595825195,
+      "learning_rate": 8.391834099895103e-06,
+      "loss": 0.0054,
+      "step": 23380
+    },
+    {
+      "epoch": 16.98,
+      "grad_norm": 0.0075091165490448475,
+      "learning_rate": 8.371661421770356e-06,
+      "loss": 0.0097,
+      "step": 23390
+    },
+    {
+      "epoch": 16.99,
+      "grad_norm": 0.0006531989201903343,
+      "learning_rate": 8.351488743645607e-06,
+      "loss": 0.0009,
+      "step": 23400
+    },
+    {
+      "epoch": 16.99,
+      "grad_norm": 0.0026475286576896906,
+      "learning_rate": 8.331316065520858e-06,
+      "loss": 0.0041,
+      "step": 23410
+    },
+    {
+      "epoch": 17.0,
+      "eval_accuracy": 0.9983663838090484,
+      "eval_f1": 0.9962274037201991,
+      "eval_loss": 0.0031724609434604645,
+      "eval_precision": 0.9999474071736615,
+      "eval_recall": 0.9925349759866361,
+      "eval_roc_auc": 0.9999873945486788,
+      "eval_runtime": 388.297,
+      "eval_samples_per_second": 227.012,
+      "eval_steps_per_second": 14.19,
+      "step": 23417
+    },
+    {
+      "epoch": 17.0,
+      "grad_norm": 0.11131101846694946,
+      "learning_rate": 8.31114338739611e-06,
+      "loss": 0.0114,
+      "step": 23420
+    },
+    {
+      "epoch": 17.01,
+      "grad_norm": 0.09979415684938431,
+      "learning_rate": 8.290970709271363e-06,
+      "loss": 0.0058,
+      "step": 23430
+    },
+    {
+      "epoch": 17.02,
+      "grad_norm": 0.05326874554157257,
+      "learning_rate": 8.270798031146614e-06,
+      "loss": 0.0027,
+      "step": 23440
+    },
+    {
+      "epoch": 17.02,
+      "grad_norm": 0.004584606271237135,
+      "learning_rate": 8.250625353021867e-06,
+      "loss": 0.0026,
+      "step": 23450
+    },
+    {
+      "epoch": 17.03,
+      "grad_norm": 0.08099240064620972,
+      "learning_rate": 8.23045267489712e-06,
+      "loss": 0.0062,
+      "step": 23460
+    },
+    {
+      "epoch": 17.04,
+      "grad_norm": 0.01665760576725006,
+      "learning_rate": 8.210279996772371e-06,
+      "loss": 0.0017,
+      "step": 23470
+    },
+    {
+      "epoch": 17.05,
+      "grad_norm": 0.0030292777810245752,
+      "learning_rate": 8.190107318647624e-06,
+      "loss": 0.0,
+      "step": 23480
+    },
+    {
+      "epoch": 17.05,
+      "grad_norm": 0.004888925235718489,
+      "learning_rate": 8.169934640522877e-06,
+      "loss": 0.0049,
+      "step": 23490
+    },
+    {
+      "epoch": 17.06,
+      "grad_norm": 0.01544449757784605,
+      "learning_rate": 8.149761962398128e-06,
+      "loss": 0.0029,
+      "step": 23500
+    },
+    {
+      "epoch": 17.07,
+      "grad_norm": 0.3572964370250702,
+      "learning_rate": 8.12958928427338e-06,
+      "loss": 0.013,
+      "step": 23510
+    },
+    {
+      "epoch": 17.07,
+      "grad_norm": 0.0019218194065615535,
+      "learning_rate": 8.109416606148633e-06,
+      "loss": 0.0005,
+      "step": 23520
+    },
+    {
+      "epoch": 17.08,
+      "grad_norm": 33.84624099731445,
+      "learning_rate": 8.089243928023884e-06,
+      "loss": 0.0251,
+      "step": 23530
+    },
+    {
+      "epoch": 17.09,
+      "grad_norm": 0.0994623526930809,
+      "learning_rate": 8.069071249899137e-06,
+      "loss": 0.0042,
+      "step": 23540
+    },
+    {
+      "epoch": 17.1,
+      "grad_norm": 0.03560971841216087,
+      "learning_rate": 8.04889857177439e-06,
+      "loss": 0.0111,
+      "step": 23550
+    },
+    {
+      "epoch": 17.1,
+      "grad_norm": 0.0007038107723928988,
+      "learning_rate": 8.028725893649641e-06,
+      "loss": 0.0067,
+      "step": 23560
+    },
+    {
+      "epoch": 17.11,
+      "grad_norm": 0.04057719558477402,
+      "learning_rate": 8.008553215524894e-06,
+      "loss": 0.0024,
+      "step": 23570
+    },
+    {
+      "epoch": 17.12,
+      "grad_norm": 0.057476069778203964,
+      "learning_rate": 7.988380537400146e-06,
+      "loss": 0.0032,
+      "step": 23580
+    },
+    {
+      "epoch": 17.13,
+      "grad_norm": 0.0002899345417972654,
+      "learning_rate": 7.968207859275398e-06,
+      "loss": 0.0029,
+      "step": 23590
+    },
+    {
+      "epoch": 17.13,
+      "grad_norm": 0.0003263648832216859,
+      "learning_rate": 7.94803518115065e-06,
+      "loss": 0.0031,
+      "step": 23600
+    },
+    {
+      "epoch": 17.14,
+      "grad_norm": 0.0024582231417298317,
+      "learning_rate": 7.927862503025903e-06,
+      "loss": 0.0061,
+      "step": 23610
+    },
+    {
+      "epoch": 17.15,
+      "grad_norm": 0.0026196760591119528,
+      "learning_rate": 7.907689824901154e-06,
+      "loss": 0.0032,
+      "step": 23620
+    },
+    {
+      "epoch": 17.15,
+      "grad_norm": 0.0027710944414138794,
+      "learning_rate": 7.887517146776407e-06,
+      "loss": 0.002,
+      "step": 23630
+    },
+    {
+      "epoch": 17.16,
+      "grad_norm": 0.0001118101499741897,
+      "learning_rate": 7.86734446865166e-06,
+      "loss": 0.0068,
+      "step": 23640
+    },
+    {
+      "epoch": 17.17,
+      "grad_norm": 0.9812076091766357,
+      "learning_rate": 7.84717179052691e-06,
+      "loss": 0.0044,
+      "step": 23650
+    },
+    {
+      "epoch": 17.18,
+      "grad_norm": 0.0007418180466629565,
+      "learning_rate": 7.826999112402164e-06,
+      "loss": 0.0042,
+      "step": 23660
+    },
+    {
+      "epoch": 17.18,
+      "grad_norm": 0.04977266862988472,
+      "learning_rate": 7.806826434277415e-06,
+      "loss": 0.0023,
+      "step": 23670
+    },
+    {
+      "epoch": 17.19,
+      "grad_norm": 0.00283865537494421,
+      "learning_rate": 7.786653756152667e-06,
+      "loss": 0.0011,
+      "step": 23680
+    },
+    {
+      "epoch": 17.2,
+      "grad_norm": 0.16779397428035736,
+      "learning_rate": 7.766481078027918e-06,
+      "loss": 0.0035,
+      "step": 23690
+    },
+    {
+      "epoch": 17.21,
+      "grad_norm": 0.001473784213885665,
+      "learning_rate": 7.746308399903171e-06,
+      "loss": 0.0036,
+      "step": 23700
+    },
+    {
+      "epoch": 17.21,
+      "grad_norm": 0.00017607476911507547,
+      "learning_rate": 7.726135721778424e-06,
+      "loss": 0.002,
+      "step": 23710
+    },
+    {
+      "epoch": 17.22,
+      "grad_norm": 0.027364423498511314,
+      "learning_rate": 7.705963043653675e-06,
+      "loss": 0.0092,
+      "step": 23720
+    },
+    {
+      "epoch": 17.23,
+      "grad_norm": 0.00015377481759060174,
+      "learning_rate": 7.685790365528928e-06,
+      "loss": 0.0023,
+      "step": 23730
+    },
+    {
+      "epoch": 17.23,
+      "grad_norm": 10.986251831054688,
+      "learning_rate": 7.66561768740418e-06,
+      "loss": 0.0128,
+      "step": 23740
+    },
+    {
+      "epoch": 17.24,
+      "grad_norm": 0.13001392781734467,
+      "learning_rate": 7.645445009279432e-06,
+      "loss": 0.0042,
+      "step": 23750
+    },
+    {
+      "epoch": 17.25,
+      "grad_norm": 0.033883240073919296,
+      "learning_rate": 7.6252723311546845e-06,
+      "loss": 0.002,
+      "step": 23760
+    },
+    {
+      "epoch": 17.26,
+      "grad_norm": 0.06336942315101624,
+      "learning_rate": 7.6050996530299356e-06,
+      "loss": 0.0008,
+      "step": 23770
+    },
+    {
+      "epoch": 17.26,
+      "grad_norm": 0.0002739182091318071,
+      "learning_rate": 7.584926974905189e-06,
+      "loss": 0.0025,
+      "step": 23780
+    },
+    {
+      "epoch": 17.27,
+      "grad_norm": 0.003048642072826624,
+      "learning_rate": 7.564754296780441e-06,
+      "loss": 0.0021,
+      "step": 23790
+    },
+    {
+      "epoch": 17.28,
+      "grad_norm": 0.002867691917344928,
+      "learning_rate": 7.544581618655692e-06,
+      "loss": 0.0004,
+      "step": 23800
+    },
+    {
+      "epoch": 17.28,
+      "grad_norm": 0.1003432348370552,
+      "learning_rate": 7.524408940530946e-06,
+      "loss": 0.0049,
+      "step": 23810
+    },
+    {
+      "epoch": 17.29,
+      "grad_norm": 0.0033953345846384764,
+      "learning_rate": 7.504236262406198e-06,
+      "loss": 0.0004,
+      "step": 23820
+    },
+    {
+      "epoch": 17.3,
+      "grad_norm": 0.012755511328577995,
+      "learning_rate": 7.484063584281449e-06,
+      "loss": 0.0054,
+      "step": 23830
+    },
+    {
+      "epoch": 17.31,
+      "grad_norm": 0.000518796790856868,
+      "learning_rate": 7.4638909061567024e-06,
+      "loss": 0.0033,
+      "step": 23840
+    },
+    {
+      "epoch": 17.31,
+      "grad_norm": 0.11989326775074005,
+      "learning_rate": 7.443718228031954e-06,
+      "loss": 0.003,
+      "step": 23850
+    },
+    {
+      "epoch": 17.32,
+      "grad_norm": 0.00038776840665377676,
+      "learning_rate": 7.4235455499072054e-06,
+      "loss": 0.0039,
+      "step": 23860
+    },
+    {
+      "epoch": 17.33,
+      "grad_norm": 0.10173983126878738,
+      "learning_rate": 7.403372871782459e-06,
+      "loss": 0.0053,
+      "step": 23870
+    },
+    {
+      "epoch": 17.34,
+      "grad_norm": 0.0036807521246373653,
+      "learning_rate": 7.38320019365771e-06,
+      "loss": 0.0016,
+      "step": 23880
+    },
+    {
+      "epoch": 17.34,
+      "grad_norm": 3.253828253946267e-05,
+      "learning_rate": 7.363027515532962e-06,
+      "loss": 0.0051,
+      "step": 23890
+    },
+    {
+      "epoch": 17.35,
+      "grad_norm": 0.00014638539869338274,
+      "learning_rate": 7.342854837408214e-06,
+      "loss": 0.0032,
+      "step": 23900
+    },
+    {
+      "epoch": 17.36,
+      "grad_norm": 0.0003724195412360132,
+      "learning_rate": 7.322682159283467e-06,
+      "loss": 0.0017,
+      "step": 23910
+    },
+    {
+      "epoch": 17.36,
+      "grad_norm": 0.0679226964712143,
+      "learning_rate": 7.302509481158719e-06,
+      "loss": 0.0016,
+      "step": 23920
+    },
+    {
+      "epoch": 17.37,
+      "grad_norm": 0.0009348354651592672,
+      "learning_rate": 7.282336803033971e-06,
+      "loss": 0.0043,
+      "step": 23930
+    },
+    {
+      "epoch": 17.38,
+      "grad_norm": 0.0015548918163403869,
+      "learning_rate": 7.262164124909223e-06,
+      "loss": 0.0044,
+      "step": 23940
+    },
+    {
+      "epoch": 17.39,
+      "grad_norm": 0.00017984594160225242,
+      "learning_rate": 7.241991446784475e-06,
+      "loss": 0.0017,
+      "step": 23950
+    },
+    {
+      "epoch": 17.39,
+      "grad_norm": 4.2534215026535094e-05,
+      "learning_rate": 7.221818768659727e-06,
+      "loss": 0.0069,
+      "step": 23960
+    },
+    {
+      "epoch": 17.4,
+      "grad_norm": 0.045961715281009674,
+      "learning_rate": 7.20164609053498e-06,
+      "loss": 0.0036,
+      "step": 23970
+    },
+    {
+      "epoch": 17.41,
+      "grad_norm": 0.1038808822631836,
+      "learning_rate": 7.181473412410232e-06,
+      "loss": 0.0028,
+      "step": 23980
+    },
+    {
+      "epoch": 17.42,
+      "grad_norm": 0.0011143676238134503,
+      "learning_rate": 7.161300734285484e-06,
+      "loss": 0.01,
+      "step": 23990
+    },
+    {
+      "epoch": 17.42,
+      "grad_norm": 0.00020528653112705797,
+      "learning_rate": 7.141128056160737e-06,
+      "loss": 0.0113,
+      "step": 24000
+    },
+    {
+      "epoch": 17.43,
+      "grad_norm": 0.1908223181962967,
+      "learning_rate": 7.1209553780359885e-06,
+      "loss": 0.0042,
+      "step": 24010
+    },
+    {
+      "epoch": 17.44,
+      "grad_norm": 0.05702119693160057,
+      "learning_rate": 7.1007826999112405e-06,
+      "loss": 0.0025,
+      "step": 24020
+    },
+    {
+      "epoch": 17.44,
+      "grad_norm": 0.1563175767660141,
+      "learning_rate": 7.080610021786492e-06,
+      "loss": 0.0039,
+      "step": 24030
+    },
+    {
+      "epoch": 17.45,
+      "grad_norm": 0.0011573946103453636,
+      "learning_rate": 7.060437343661745e-06,
+      "loss": 0.0006,
+      "step": 24040
+    },
+    {
+      "epoch": 17.46,
+      "grad_norm": 0.019460035488009453,
+      "learning_rate": 7.040264665536997e-06,
+      "loss": 0.0014,
+      "step": 24050
+    },
+    {
+      "epoch": 17.47,
+      "grad_norm": 0.0006199249182827771,
+      "learning_rate": 7.020091987412249e-06,
+      "loss": 0.0036,
+      "step": 24060
+    },
+    {
+      "epoch": 17.47,
+      "grad_norm": 0.001222978811711073,
+      "learning_rate": 6.999919309287502e-06,
+      "loss": 0.0006,
+      "step": 24070
+    },
+    {
+      "epoch": 17.48,
+      "grad_norm": 0.08943016082048416,
+      "learning_rate": 6.979746631162754e-06,
+      "loss": 0.0039,
+      "step": 24080
+    },
+    {
+      "epoch": 17.49,
+      "grad_norm": 16.41295623779297,
+      "learning_rate": 6.959573953038006e-06,
+      "loss": 0.0012,
+      "step": 24090
+    },
+    {
+      "epoch": 17.5,
+      "grad_norm": 0.029908979311585426,
+      "learning_rate": 6.939401274913258e-06,
+      "loss": 0.0067,
+      "step": 24100
+    },
+    {
+      "epoch": 17.5,
+      "grad_norm": 0.0022195398341864347,
+      "learning_rate": 6.91922859678851e-06,
+      "loss": 0.0028,
+      "step": 24110
+    },
+    {
+      "epoch": 17.51,
+      "grad_norm": 0.0004878832842223346,
+      "learning_rate": 6.899055918663762e-06,
+      "loss": 0.0047,
+      "step": 24120
+    },
+    {
+      "epoch": 17.52,
+      "grad_norm": 0.08152516931295395,
+      "learning_rate": 6.878883240539015e-06,
+      "loss": 0.0051,
+      "step": 24130
+    },
+    {
+      "epoch": 17.52,
+      "grad_norm": 6.538970774272457e-05,
+      "learning_rate": 6.858710562414267e-06,
+      "loss": 0.0002,
+      "step": 24140
+    },
+    {
+      "epoch": 17.53,
+      "grad_norm": 0.22862771153450012,
+      "learning_rate": 6.838537884289518e-06,
+      "loss": 0.0005,
+      "step": 24150
+    },
+    {
+      "epoch": 17.54,
+      "grad_norm": 8.339462280273438,
+      "learning_rate": 6.81836520616477e-06,
+      "loss": 0.0024,
+      "step": 24160
+    },
+    {
+      "epoch": 17.55,
+      "grad_norm": 0.10850653797388077,
+      "learning_rate": 6.7981925280400236e-06,
+      "loss": 0.0057,
+      "step": 24170
+    },
+    {
+      "epoch": 17.55,
+      "grad_norm": 0.007265112828463316,
+      "learning_rate": 6.778019849915275e-06,
+      "loss": 0.0012,
+      "step": 24180
+    },
+    {
+      "epoch": 17.56,
+      "grad_norm": 5.9882444475078955e-05,
+      "learning_rate": 6.7578471717905266e-06,
+      "loss": 0.0034,
+      "step": 24190
+    },
+    {
+      "epoch": 17.57,
+      "grad_norm": 8.714703290024772e-05,
+      "learning_rate": 6.73767449366578e-06,
+      "loss": 0.0036,
+      "step": 24200
+    },
+    {
+      "epoch": 17.58,
+      "grad_norm": 0.001536114257760346,
+      "learning_rate": 6.717501815541031e-06,
+      "loss": 0.0015,
+      "step": 24210
+    },
+    {
+      "epoch": 17.58,
+      "grad_norm": 1.9178543880116194e-05,
+      "learning_rate": 6.697329137416283e-06,
+      "loss": 0.0025,
+      "step": 24220
+    },
+    {
+      "epoch": 17.59,
+      "grad_norm": 0.05648133158683777,
+      "learning_rate": 6.677156459291537e-06,
+      "loss": 0.0127,
+      "step": 24230
+    },
+    {
+      "epoch": 17.6,
+      "grad_norm": 0.00012539050658233464,
+      "learning_rate": 6.656983781166788e-06,
+      "loss": 0.0045,
+      "step": 24240
+    },
+    {
+      "epoch": 17.6,
+      "grad_norm": 0.19643142819404602,
+      "learning_rate": 6.63681110304204e-06,
+      "loss": 0.0073,
+      "step": 24250
+    },
+    {
+      "epoch": 17.61,
+      "grad_norm": 0.11744473874568939,
+      "learning_rate": 6.616638424917293e-06,
+      "loss": 0.0096,
+      "step": 24260
+    },
+    {
+      "epoch": 17.62,
+      "grad_norm": 1.6758296624175273e-05,
+      "learning_rate": 6.5964657467925445e-06,
+      "loss": 0.0054,
+      "step": 24270
+    },
+    {
+      "epoch": 17.63,
+      "grad_norm": 0.0006310921744443476,
+      "learning_rate": 6.576293068667796e-06,
+      "loss": 0.0052,
+      "step": 24280
+    },
+    {
+      "epoch": 17.63,
+      "grad_norm": 0.36372512578964233,
+      "learning_rate": 6.556120390543048e-06,
+      "loss": 0.0017,
+      "step": 24290
+    },
+    {
+      "epoch": 17.64,
+      "grad_norm": 0.05171734094619751,
+      "learning_rate": 6.535947712418301e-06,
+      "loss": 0.0094,
+      "step": 24300
+    },
+    {
+      "epoch": 17.65,
+      "grad_norm": 0.0003291558241471648,
+      "learning_rate": 6.515775034293553e-06,
+      "loss": 0.0031,
+      "step": 24310
+    },
+    {
+      "epoch": 17.66,
+      "grad_norm": 0.003046097932383418,
+      "learning_rate": 6.495602356168805e-06,
+      "loss": 0.0009,
+      "step": 24320
+    },
+    {
+      "epoch": 17.66,
+      "grad_norm": 0.00018167686357628554,
+      "learning_rate": 6.475429678044058e-06,
+      "loss": 0.0075,
+      "step": 24330
+    },
+    {
+      "epoch": 17.67,
+      "grad_norm": 0.026283616200089455,
+      "learning_rate": 6.45525699991931e-06,
+      "loss": 0.002,
+      "step": 24340
+    },
+    {
+      "epoch": 17.68,
+      "grad_norm": 0.00010824885976035148,
+      "learning_rate": 6.435084321794562e-06,
+      "loss": 0.0032,
+      "step": 24350
+    },
+    {
+      "epoch": 17.68,
+      "grad_norm": 9.676727495389059e-05,
+      "learning_rate": 6.414911643669814e-06,
+      "loss": 0.0015,
+      "step": 24360
+    },
+    {
+      "epoch": 17.69,
+      "grad_norm": 3.2388255931437016e-05,
+      "learning_rate": 6.394738965545066e-06,
+      "loss": 0.002,
+      "step": 24370
+    },
+    {
+      "epoch": 17.7,
+      "grad_norm": 0.00028173986356705427,
+      "learning_rate": 6.374566287420318e-06,
+      "loss": 0.0004,
+      "step": 24380
+    },
+    {
+      "epoch": 17.71,
+      "grad_norm": 2.8843047618865967,
+      "learning_rate": 6.354393609295569e-06,
+      "loss": 0.0035,
+      "step": 24390
+    },
+    {
+      "epoch": 17.71,
+      "grad_norm": 0.00011344110680511221,
+      "learning_rate": 6.334220931170823e-06,
+      "loss": 0.0033,
+      "step": 24400
+    },
+    {
+      "epoch": 17.72,
+      "grad_norm": 0.00039801959064789116,
+      "learning_rate": 6.314048253046075e-06,
+      "loss": 0.0048,
+      "step": 24410
+    },
+    {
+      "epoch": 17.73,
+      "grad_norm": 2.9843920856365003e-05,
+      "learning_rate": 6.293875574921326e-06,
+      "loss": 0.0004,
+      "step": 24420
+    },
+    {
+      "epoch": 17.74,
+      "grad_norm": 0.00021989627566654235,
+      "learning_rate": 6.2737028967965795e-06,
+      "loss": 0.0,
+      "step": 24430
+    },
+    {
+      "epoch": 17.74,
+      "grad_norm": 0.0002266202645841986,
+      "learning_rate": 6.2535302186718314e-06,
+      "loss": 0.006,
+      "step": 24440
+    },
+    {
+      "epoch": 17.75,
+      "grad_norm": 0.00015999180322978646,
+      "learning_rate": 6.233357540547083e-06,
+      "loss": 0.0053,
+      "step": 24450
+    },
+    {
+      "epoch": 17.76,
+      "grad_norm": 0.002106759464368224,
+      "learning_rate": 6.213184862422335e-06,
+      "loss": 0.0034,
+      "step": 24460
+    },
+    {
+      "epoch": 17.76,
+      "grad_norm": 0.09708841145038605,
+      "learning_rate": 6.193012184297588e-06,
+      "loss": 0.0042,
+      "step": 24470
+    },
+    {
+      "epoch": 17.77,
+      "grad_norm": 0.00014056751388125122,
+      "learning_rate": 6.172839506172839e-06,
+      "loss": 0.0088,
+      "step": 24480
+    },
+    {
+      "epoch": 17.78,
+      "grad_norm": 0.4134717583656311,
+      "learning_rate": 6.152666828048092e-06,
+      "loss": 0.0063,
+      "step": 24490
+    },
+    {
+      "epoch": 17.79,
+      "grad_norm": 0.02683611586689949,
+      "learning_rate": 6.132494149923345e-06,
+      "loss": 0.0025,
+      "step": 24500
+    },
+    {
+      "epoch": 17.79,
+      "grad_norm": 0.00020469677110668272,
+      "learning_rate": 6.112321471798596e-06,
+      "loss": 0.0,
+      "step": 24510
+    },
+    {
+      "epoch": 17.8,
+      "grad_norm": 0.00013897515600547194,
+      "learning_rate": 6.0921487936738485e-06,
+      "loss": 0.0034,
+      "step": 24520
+    },
+    {
+      "epoch": 17.81,
+      "grad_norm": 0.0001812389527913183,
+      "learning_rate": 6.071976115549101e-06,
+      "loss": 0.0014,
+      "step": 24530
+    },
+    {
+      "epoch": 17.81,
+      "grad_norm": 9.690736624179408e-05,
+      "learning_rate": 6.051803437424352e-06,
+      "loss": 0.0037,
+      "step": 24540
+    },
+    {
+      "epoch": 17.82,
+      "grad_norm": 0.0005624780897051096,
+      "learning_rate": 6.031630759299605e-06,
+      "loss": 0.0115,
+      "step": 24550
+    },
+    {
+      "epoch": 17.83,
+      "grad_norm": 0.057843539863824844,
+      "learning_rate": 6.011458081174857e-06,
+      "loss": 0.0024,
+      "step": 24560
+    },
+    {
+      "epoch": 17.84,
+      "grad_norm": 0.00023191337822936475,
+      "learning_rate": 5.991285403050109e-06,
+      "loss": 0.0079,
+      "step": 24570
+    },
+    {
+      "epoch": 17.84,
+      "grad_norm": 0.00026599192642606795,
+      "learning_rate": 5.971112724925362e-06,
+      "loss": 0.0003,
+      "step": 24580
+    },
+    {
+      "epoch": 17.85,
+      "grad_norm": 0.0005074123037047684,
+      "learning_rate": 5.950940046800614e-06,
+      "loss": 0.0011,
+      "step": 24590
+    },
+    {
+      "epoch": 17.86,
+      "grad_norm": 0.0019474523141980171,
+      "learning_rate": 5.930767368675866e-06,
+      "loss": 0.0,
+      "step": 24600
+    },
+    {
+      "epoch": 17.87,
+      "grad_norm": 0.0010078544728457928,
+      "learning_rate": 5.9105946905511175e-06,
+      "loss": 0.0015,
+      "step": 24610
+    },
+    {
+      "epoch": 17.87,
+      "grad_norm": 0.0005764566012658179,
+      "learning_rate": 5.89042201242637e-06,
+      "loss": 0.0007,
+      "step": 24620
+    },
+    {
+      "epoch": 17.88,
+      "grad_norm": 0.00041168267489410937,
+      "learning_rate": 5.870249334301622e-06,
+      "loss": 0.0031,
+      "step": 24630
+    },
+    {
+      "epoch": 17.89,
+      "grad_norm": 0.000586585549172014,
+      "learning_rate": 5.850076656176874e-06,
+      "loss": 0.0055,
+      "step": 24640
+    },
+    {
+      "epoch": 17.89,
+      "grad_norm": 0.0007330170483328402,
+      "learning_rate": 5.829903978052127e-06,
+      "loss": 0.0013,
+      "step": 24650
+    },
+    {
+      "epoch": 17.9,
+      "grad_norm": 0.0002643170882947743,
+      "learning_rate": 5.809731299927379e-06,
+      "loss": 0.0022,
+      "step": 24660
+    },
+    {
+      "epoch": 17.91,
+      "grad_norm": 0.0039410036988556385,
+      "learning_rate": 5.789558621802631e-06,
+      "loss": 0.0055,
+      "step": 24670
+    },
+    {
+      "epoch": 17.92,
+      "grad_norm": 11.958067893981934,
+      "learning_rate": 5.769385943677883e-06,
+      "loss": 0.0058,
+      "step": 24680
+    },
+    {
+      "epoch": 17.92,
+      "grad_norm": 0.0001658492983551696,
+      "learning_rate": 5.749213265553135e-06,
+      "loss": 0.007,
+      "step": 24690
+    },
+    {
+      "epoch": 17.93,
+      "grad_norm": 0.0005535160889849067,
+      "learning_rate": 5.729040587428387e-06,
+      "loss": 0.0069,
+      "step": 24700
+    },
+    {
+      "epoch": 17.94,
+      "grad_norm": 0.002007086528465152,
+      "learning_rate": 5.708867909303639e-06,
+      "loss": 0.0068,
+      "step": 24710
+    },
+    {
+      "epoch": 17.95,
+      "grad_norm": 0.00019094701565336436,
+      "learning_rate": 5.688695231178891e-06,
+      "loss": 0.0068,
+      "step": 24720
+    },
+    {
+      "epoch": 17.95,
+      "grad_norm": 0.24637742340564728,
+      "learning_rate": 5.668522553054144e-06,
+      "loss": 0.0048,
+      "step": 24730
+    },
+    {
+      "epoch": 17.96,
+      "grad_norm": 0.00016859486640896648,
+      "learning_rate": 5.648349874929396e-06,
+      "loss": 0.0035,
+      "step": 24740
+    },
+    {
+      "epoch": 17.97,
+      "grad_norm": 0.0005759259220212698,
+      "learning_rate": 5.628177196804648e-06,
+      "loss": 0.0044,
+      "step": 24750
+    },
+    {
+      "epoch": 17.97,
+      "grad_norm": 0.007888413034379482,
+      "learning_rate": 5.608004518679901e-06,
+      "loss": 0.0041,
+      "step": 24760
+    },
+    {
+      "epoch": 17.98,
+      "grad_norm": 9.520780563354492,
+      "learning_rate": 5.5878318405551526e-06,
+      "loss": 0.0161,
+      "step": 24770
+    },
+    {
+      "epoch": 17.99,
+      "grad_norm": 0.08241154253482819,
+      "learning_rate": 5.5676591624304045e-06,
+      "loss": 0.0016,
+      "step": 24780
+    },
+    {
+      "epoch": 18.0,
+      "grad_norm": 0.044836148619651794,
+      "learning_rate": 5.547486484305656e-06,
+      "loss": 0.002,
+      "step": 24790
+    },
+    {
+      "epoch": 18.0,
+      "eval_accuracy": 0.9983550392521668,
+      "eval_f1": 0.9962040891122804,
+      "eval_loss": 0.0032196117099374533,
+      "eval_precision": 0.9991597962505908,
+      "eval_recall": 0.9932658174984339,
+      "eval_roc_auc": 0.999987405141813,
+      "eval_runtime": 386.475,
+      "eval_samples_per_second": 228.082,
+      "eval_steps_per_second": 14.257,
+      "step": 24795
+    },
+    {
+      "epoch": 18.0,
+      "grad_norm": 0.0755479708313942,
+      "learning_rate": 5.527313806180909e-06,
+      "loss": 0.0013,
+      "step": 24800
+    },
+    {
+      "epoch": 18.01,
+      "grad_norm": 0.04686315357685089,
+      "learning_rate": 5.507141128056161e-06,
+      "loss": 0.0144,
+      "step": 24810
+    },
+    {
+      "epoch": 18.02,
+      "grad_norm": 0.00485340366140008,
+      "learning_rate": 5.486968449931413e-06,
+      "loss": 0.0029,
+      "step": 24820
+    },
+    {
+      "epoch": 18.03,
+      "grad_norm": 0.00023718834563624114,
+      "learning_rate": 5.466795771806665e-06,
+      "loss": 0.0,
+      "step": 24830
+    },
+    {
+      "epoch": 18.03,
+      "grad_norm": 0.012029669247567654,
+      "learning_rate": 5.446623093681918e-06,
+      "loss": 0.004,
+      "step": 24840
+    },
+    {
+      "epoch": 18.04,
+      "grad_norm": 0.006431036163121462,
+      "learning_rate": 5.42645041555717e-06,
+      "loss": 0.0012,
+      "step": 24850
+    },
+    {
+      "epoch": 18.05,
+      "grad_norm": 0.07365299761295319,
+      "learning_rate": 5.406277737432422e-06,
+      "loss": 0.003,
+      "step": 24860
+    },
+    {
+      "epoch": 18.05,
+      "grad_norm": 0.004044768866151571,
+      "learning_rate": 5.3861050593076735e-06,
+      "loss": 0.001,
+      "step": 24870
+    },
+    {
+      "epoch": 18.06,
+      "grad_norm": 0.04818055033683777,
+      "learning_rate": 5.365932381182926e-06,
+      "loss": 0.0006,
+      "step": 24880
+    },
+    {
+      "epoch": 18.07,
+      "grad_norm": 7.61866249376908e-05,
+      "learning_rate": 5.345759703058178e-06,
+      "loss": 0.0,
+      "step": 24890
+    },
+    {
+      "epoch": 18.08,
+      "grad_norm": 0.0008586979820393026,
+      "learning_rate": 5.32558702493343e-06,
+      "loss": 0.005,
+      "step": 24900
+    },
+    {
+      "epoch": 18.08,
+      "grad_norm": 0.00014436905621550977,
+      "learning_rate": 5.305414346808683e-06,
+      "loss": 0.0027,
+      "step": 24910
+    },
+    {
+      "epoch": 18.09,
+      "grad_norm": 0.00020209125068504363,
+      "learning_rate": 5.285241668683935e-06,
+      "loss": 0.0054,
+      "step": 24920
+    },
+    {
+      "epoch": 18.1,
+      "grad_norm": 0.0005764380330219865,
+      "learning_rate": 5.265068990559187e-06,
+      "loss": 0.0023,
+      "step": 24930
+    },
+    {
+      "epoch": 18.11,
+      "grad_norm": 0.00015431991778314114,
+      "learning_rate": 5.2448963124344395e-06,
+      "loss": 0.0045,
+      "step": 24940
+    },
+    {
+      "epoch": 18.11,
+      "grad_norm": 8.73160533956252e-05,
+      "learning_rate": 5.224723634309691e-06,
+      "loss": 0.0027,
+      "step": 24950
+    },
+    {
+      "epoch": 18.12,
+      "grad_norm": 0.0016183574916794896,
+      "learning_rate": 5.204550956184943e-06,
+      "loss": 0.0022,
+      "step": 24960
+    },
+    {
+      "epoch": 18.13,
+      "grad_norm": 0.11282764375209808,
+      "learning_rate": 5.184378278060196e-06,
+      "loss": 0.0106,
+      "step": 24970
+    },
+    {
+      "epoch": 18.13,
+      "grad_norm": 0.00025519824703224003,
+      "learning_rate": 5.164205599935447e-06,
+      "loss": 0.0006,
+      "step": 24980
+    },
+    {
+      "epoch": 18.14,
+      "grad_norm": 0.0002551107027102262,
+      "learning_rate": 5.1440329218107e-06,
+      "loss": 0.0027,
+      "step": 24990
+    },
+    {
+      "epoch": 18.15,
+      "grad_norm": 5.992265505483374e-05,
+      "learning_rate": 5.123860243685952e-06,
+      "loss": 0.0026,
+      "step": 25000
+    },
+    {
+      "epoch": 18.16,
+      "grad_norm": 5.242994666332379e-05,
+      "learning_rate": 5.103687565561204e-06,
+      "loss": 0.0023,
+      "step": 25010
+    },
+    {
+      "epoch": 18.16,
+      "grad_norm": 0.004187633749097586,
+      "learning_rate": 5.083514887436457e-06,
+      "loss": 0.0039,
+      "step": 25020
+    },
+    {
+      "epoch": 18.17,
+      "grad_norm": 0.03268939256668091,
+      "learning_rate": 5.0633422093117085e-06,
+      "loss": 0.0009,
+      "step": 25030
+    },
+    {
+      "epoch": 18.18,
+      "grad_norm": 0.02059740573167801,
+      "learning_rate": 5.0431695311869605e-06,
+      "loss": 0.0058,
+      "step": 25040
+    },
+    {
+      "epoch": 18.19,
+      "grad_norm": 0.0005777775659225881,
+      "learning_rate": 5.022996853062212e-06,
+      "loss": 0.0055,
+      "step": 25050
+    },
+    {
+      "epoch": 18.19,
+      "grad_norm": 0.00028460632893256843,
+      "learning_rate": 5.002824174937465e-06,
+      "loss": 0.0072,
+      "step": 25060
+    },
+    {
+      "epoch": 18.2,
+      "grad_norm": 0.00011931911285500973,
+      "learning_rate": 4.982651496812717e-06,
+      "loss": 0.0013,
+      "step": 25070
+    },
+    {
+      "epoch": 18.21,
+      "grad_norm": 0.028543755412101746,
+      "learning_rate": 4.962478818687969e-06,
+      "loss": 0.0058,
+      "step": 25080
+    },
+    {
+      "epoch": 18.21,
+      "grad_norm": 0.06952419131994247,
+      "learning_rate": 4.942306140563222e-06,
+      "loss": 0.0012,
+      "step": 25090
+    },
+    {
+      "epoch": 18.22,
+      "grad_norm": 0.0011916140792891383,
+      "learning_rate": 4.922133462438473e-06,
+      "loss": 0.0028,
+      "step": 25100
+    },
+    {
+      "epoch": 18.23,
+      "grad_norm": 0.0002467480080667883,
+      "learning_rate": 4.901960784313726e-06,
+      "loss": 0.0053,
+      "step": 25110
+    },
+    {
+      "epoch": 18.24,
+      "grad_norm": 0.0004054481105413288,
+      "learning_rate": 4.881788106188978e-06,
+      "loss": 0.0005,
+      "step": 25120
+    },
+    {
+      "epoch": 18.24,
+      "grad_norm": 8.49825592013076e-05,
+      "learning_rate": 4.8616154280642295e-06,
+      "loss": 0.0045,
+      "step": 25130
+    },
+    {
+      "epoch": 18.25,
+      "grad_norm": 0.11147330701351166,
+      "learning_rate": 4.841442749939482e-06,
+      "loss": 0.0017,
+      "step": 25140
+    },
+    {
+      "epoch": 18.26,
+      "grad_norm": 0.0001451838470529765,
+      "learning_rate": 4.821270071814735e-06,
+      "loss": 0.0044,
+      "step": 25150
+    },
+    {
+      "epoch": 18.26,
+      "grad_norm": 0.032486509531736374,
+      "learning_rate": 4.801097393689986e-06,
+      "loss": 0.0029,
+      "step": 25160
+    },
+    {
+      "epoch": 18.27,
+      "grad_norm": 0.09991168230772018,
+      "learning_rate": 4.780924715565239e-06,
+      "loss": 0.0026,
+      "step": 25170
+    },
+    {
+      "epoch": 18.28,
+      "grad_norm": 0.0005969495978206396,
+      "learning_rate": 4.760752037440491e-06,
+      "loss": 0.0037,
+      "step": 25180
+    },
+    {
+      "epoch": 18.29,
+      "grad_norm": 0.7579265236854553,
+      "learning_rate": 4.740579359315743e-06,
+      "loss": 0.0061,
+      "step": 25190
+    },
+    {
+      "epoch": 18.29,
+      "grad_norm": 0.11252560466527939,
+      "learning_rate": 4.7204066811909955e-06,
+      "loss": 0.0023,
+      "step": 25200
+    },
+    {
+      "epoch": 18.3,
+      "grad_norm": 0.00014016269415151328,
+      "learning_rate": 4.700234003066247e-06,
+      "loss": 0.0018,
+      "step": 25210
+    },
+    {
+      "epoch": 18.31,
+      "grad_norm": 0.09807706624269485,
+      "learning_rate": 4.680061324941499e-06,
+      "loss": 0.0042,
+      "step": 25220
+    },
+    {
+      "epoch": 18.32,
+      "grad_norm": 0.1528569459915161,
+      "learning_rate": 4.659888646816751e-06,
+      "loss": 0.0058,
+      "step": 25230
+    },
+    {
+      "epoch": 18.32,
+      "grad_norm": 3.9896687667351216e-05,
+      "learning_rate": 4.639715968692004e-06,
+      "loss": 0.0052,
+      "step": 25240
+    },
+    {
+      "epoch": 18.33,
+      "grad_norm": 0.040525808930397034,
+      "learning_rate": 4.619543290567256e-06,
+      "loss": 0.0044,
+      "step": 25250
+    },
+    {
+      "epoch": 18.34,
+      "grad_norm": 0.04245885834097862,
+      "learning_rate": 4.599370612442508e-06,
+      "loss": 0.0131,
+      "step": 25260
+    },
+    {
+      "epoch": 18.34,
+      "grad_norm": 7.739640568615869e-05,
+      "learning_rate": 4.579197934317761e-06,
+      "loss": 0.0012,
+      "step": 25270
+    },
+    {
+      "epoch": 18.35,
+      "grad_norm": 0.00031783172744326293,
+      "learning_rate": 4.5590252561930126e-06,
+      "loss": 0.0037,
+      "step": 25280
+    },
+    {
+      "epoch": 18.36,
+      "grad_norm": 0.053188130259513855,
+      "learning_rate": 4.5388525780682645e-06,
+      "loss": 0.0017,
+      "step": 25290
+    },
+    {
+      "epoch": 18.37,
+      "grad_norm": 0.0014969066251069307,
+      "learning_rate": 4.518679899943517e-06,
+      "loss": 0.0066,
+      "step": 25300
+    },
+    {
+      "epoch": 18.37,
+      "grad_norm": 0.14414730668067932,
+      "learning_rate": 4.498507221818768e-06,
+      "loss": 0.0023,
+      "step": 25310
+    },
+    {
+      "epoch": 18.38,
+      "grad_norm": 0.00036431997432373464,
+      "learning_rate": 4.478334543694021e-06,
+      "loss": 0.0004,
+      "step": 25320
+    },
+    {
+      "epoch": 18.39,
+      "grad_norm": 0.0006099030142650008,
+      "learning_rate": 4.458161865569274e-06,
+      "loss": 0.0041,
+      "step": 25330
+    },
+    {
+      "epoch": 18.4,
+      "grad_norm": 0.0006587031530216336,
+      "learning_rate": 4.437989187444525e-06,
+      "loss": 0.0031,
+      "step": 25340
+    },
+    {
+      "epoch": 18.4,
+      "grad_norm": 0.13671369850635529,
+      "learning_rate": 4.417816509319778e-06,
+      "loss": 0.0041,
+      "step": 25350
+    },
+    {
+      "epoch": 18.41,
+      "grad_norm": 0.0011476201470941305,
+      "learning_rate": 4.39764383119503e-06,
+      "loss": 0.0093,
+      "step": 25360
+    },
+    {
+      "epoch": 18.42,
+      "grad_norm": 0.03173692151904106,
+      "learning_rate": 4.377471153070282e-06,
+      "loss": 0.0039,
+      "step": 25370
+    },
+    {
+      "epoch": 18.42,
+      "grad_norm": 2.042336382146459e-05,
+      "learning_rate": 4.357298474945534e-06,
+      "loss": 0.0035,
+      "step": 25380
+    },
+    {
+      "epoch": 18.43,
+      "grad_norm": 0.0008917959057725966,
+      "learning_rate": 4.337125796820786e-06,
+      "loss": 0.0042,
+      "step": 25390
+    },
+    {
+      "epoch": 18.44,
+      "grad_norm": 0.04398633912205696,
+      "learning_rate": 4.316953118696038e-06,
+      "loss": 0.0016,
+      "step": 25400
+    },
+    {
+      "epoch": 18.45,
+      "grad_norm": 0.0733947828412056,
+      "learning_rate": 4.29678044057129e-06,
+      "loss": 0.0044,
+      "step": 25410
+    },
+    {
+      "epoch": 18.45,
+      "grad_norm": 19.26169776916504,
+      "learning_rate": 4.276607762446543e-06,
+      "loss": 0.0123,
+      "step": 25420
+    },
+    {
+      "epoch": 18.46,
+      "grad_norm": 0.09142892807722092,
+      "learning_rate": 4.256435084321795e-06,
+      "loss": 0.0041,
+      "step": 25430
+    },
+    {
+      "epoch": 18.47,
+      "grad_norm": 0.006276868283748627,
+      "learning_rate": 4.236262406197047e-06,
+      "loss": 0.0035,
+      "step": 25440
+    },
+    {
+      "epoch": 18.48,
+      "grad_norm": 0.035533662885427475,
+      "learning_rate": 4.2160897280722995e-06,
+      "loss": 0.0023,
+      "step": 25450
+    },
+    {
+      "epoch": 18.48,
+      "grad_norm": 0.00048704142682254314,
+      "learning_rate": 4.1959170499475514e-06,
+      "loss": 0.0045,
+      "step": 25460
+    },
+    {
+      "epoch": 18.49,
+      "grad_norm": 1.3611451387405396,
+      "learning_rate": 4.175744371822803e-06,
+      "loss": 0.0065,
+      "step": 25470
+    },
+    {
+      "epoch": 18.5,
+      "grad_norm": 0.1825874298810959,
+      "learning_rate": 4.155571693698055e-06,
+      "loss": 0.0021,
+      "step": 25480
+    },
+    {
+      "epoch": 18.5,
+      "grad_norm": 0.11230570822954178,
+      "learning_rate": 4.135399015573307e-06,
+      "loss": 0.0077,
+      "step": 25490
+    },
+    {
+      "epoch": 18.51,
+      "grad_norm": 0.04604954272508621,
+      "learning_rate": 4.11522633744856e-06,
+      "loss": 0.0046,
+      "step": 25500
+    },
+    {
+      "epoch": 18.52,
+      "grad_norm": 0.00013642838166560978,
+      "learning_rate": 4.095053659323812e-06,
+      "loss": 0.0021,
+      "step": 25510
+    },
+    {
+      "epoch": 18.53,
+      "grad_norm": 4.127192369196564e-05,
+      "learning_rate": 4.074880981199064e-06,
+      "loss": 0.008,
+      "step": 25520
+    },
+    {
+      "epoch": 18.53,
+      "grad_norm": 0.0006564307259395719,
+      "learning_rate": 4.054708303074317e-06,
+      "loss": 0.0053,
+      "step": 25530
+    },
+    {
+      "epoch": 18.54,
+      "grad_norm": 5.794021853944287e-05,
+      "learning_rate": 4.0345356249495685e-06,
+      "loss": 0.0004,
+      "step": 25540
+    },
+    {
+      "epoch": 18.55,
+      "grad_norm": 0.003438483690842986,
+      "learning_rate": 4.0143629468248205e-06,
+      "loss": 0.005,
+      "step": 25550
+    },
+    {
+      "epoch": 18.56,
+      "grad_norm": 0.00017284188652411103,
+      "learning_rate": 3.994190268700073e-06,
+      "loss": 0.0028,
+      "step": 25560
+    },
+    {
+      "epoch": 18.56,
+      "grad_norm": 0.00043344812002032995,
+      "learning_rate": 3.974017590575325e-06,
+      "loss": 0.0009,
+      "step": 25570
+    },
+    {
+      "epoch": 18.57,
+      "grad_norm": 3.957462831749581e-05,
+      "learning_rate": 3.953844912450577e-06,
+      "loss": 0.0033,
+      "step": 25580
+    },
+    {
+      "epoch": 18.58,
+      "grad_norm": 0.08866414427757263,
+      "learning_rate": 3.93367223432583e-06,
+      "loss": 0.0011,
+      "step": 25590
+    },
+    {
+      "epoch": 18.58,
+      "grad_norm": 0.000915550219360739,
+      "learning_rate": 3.913499556201082e-06,
+      "loss": 0.0028,
+      "step": 25600
+    },
+    {
+      "epoch": 18.59,
+      "grad_norm": 5.103804141981527e-05,
+      "learning_rate": 3.893326878076334e-06,
+      "loss": 0.0009,
+      "step": 25610
+    },
+    {
+      "epoch": 18.6,
+      "grad_norm": 0.21407219767570496,
+      "learning_rate": 3.873154199951586e-06,
+      "loss": 0.0037,
+      "step": 25620
+    },
+    {
+      "epoch": 18.61,
+      "grad_norm": 0.12557661533355713,
+      "learning_rate": 3.8529815218268376e-06,
+      "loss": 0.0067,
+      "step": 25630
+    },
+    {
+      "epoch": 18.61,
+      "grad_norm": 0.004426570143550634,
+      "learning_rate": 3.83280884370209e-06,
+      "loss": 0.0013,
+      "step": 25640
+    },
+    {
+      "epoch": 18.62,
+      "grad_norm": 0.0005717293825000525,
+      "learning_rate": 3.8126361655773422e-06,
+      "loss": 0.0113,
+      "step": 25650
+    },
+    {
+      "epoch": 18.63,
+      "grad_norm": 0.1541978120803833,
+      "learning_rate": 3.7924634874525946e-06,
+      "loss": 0.0131,
+      "step": 25660
+    },
+    {
+      "epoch": 18.64,
+      "grad_norm": 0.0027115640696138144,
+      "learning_rate": 3.772290809327846e-06,
+      "loss": 0.0037,
+      "step": 25670
+    },
+    {
+      "epoch": 18.64,
+      "grad_norm": 7.32510961825028e-05,
+      "learning_rate": 3.752118131203099e-06,
+      "loss": 0.0075,
+      "step": 25680
+    },
+    {
+      "epoch": 18.65,
+      "grad_norm": 0.0003670882142614573,
+      "learning_rate": 3.7319454530783512e-06,
+      "loss": 0.0009,
+      "step": 25690
+    },
+    {
+      "epoch": 18.66,
+      "grad_norm": 3.481513704173267e-05,
+      "learning_rate": 3.7117727749536027e-06,
+      "loss": 0.0057,
+      "step": 25700
+    },
+    {
+      "epoch": 18.66,
+      "grad_norm": 0.00047917303163558245,
+      "learning_rate": 3.691600096828855e-06,
+      "loss": 0.0032,
+      "step": 25710
+    },
+    {
+      "epoch": 18.67,
+      "grad_norm": 0.0021459930576384068,
+      "learning_rate": 3.671427418704107e-06,
+      "loss": 0.003,
+      "step": 25720
+    },
+    {
+      "epoch": 18.68,
+      "grad_norm": 0.0027416530065238476,
+      "learning_rate": 3.6512547405793593e-06,
+      "loss": 0.0011,
+      "step": 25730
+    },
+    {
+      "epoch": 18.69,
+      "grad_norm": 0.05313531309366226,
+      "learning_rate": 3.6310820624546117e-06,
+      "loss": 0.0027,
+      "step": 25740
+    },
+    {
+      "epoch": 18.69,
+      "grad_norm": 0.16513313353061676,
+      "learning_rate": 3.6109093843298636e-06,
+      "loss": 0.0054,
+      "step": 25750
+    },
+    {
+      "epoch": 18.7,
+      "grad_norm": 0.003389935242012143,
+      "learning_rate": 3.590736706205116e-06,
+      "loss": 0.005,
+      "step": 25760
+    },
+    {
+      "epoch": 18.71,
+      "grad_norm": 0.0002206418284913525,
+      "learning_rate": 3.5705640280803683e-06,
+      "loss": 0.0035,
+      "step": 25770
+    },
+    {
+      "epoch": 18.72,
+      "grad_norm": 0.0001424114016117528,
+      "learning_rate": 3.5503913499556202e-06,
+      "loss": 0.0064,
+      "step": 25780
+    },
+    {
+      "epoch": 18.72,
+      "grad_norm": 0.032624099403619766,
+      "learning_rate": 3.5302186718308726e-06,
+      "loss": 0.0027,
+      "step": 25790
+    },
+    {
+      "epoch": 18.73,
+      "grad_norm": 0.0008412267197854817,
+      "learning_rate": 3.5100459937061245e-06,
+      "loss": 0.0014,
+      "step": 25800
+    },
+    {
+      "epoch": 18.74,
+      "grad_norm": 0.04843832924962044,
+      "learning_rate": 3.489873315581377e-06,
+      "loss": 0.0024,
+      "step": 25810
+    },
+    {
+      "epoch": 18.74,
+      "grad_norm": 0.056923747062683105,
+      "learning_rate": 3.469700637456629e-06,
+      "loss": 0.0022,
+      "step": 25820
+    },
+    {
+      "epoch": 18.75,
+      "grad_norm": 0.0007223181310109794,
+      "learning_rate": 3.449527959331881e-06,
+      "loss": 0.0046,
+      "step": 25830
+    },
+    {
+      "epoch": 18.76,
+      "grad_norm": 0.0008658911683596671,
+      "learning_rate": 3.4293552812071335e-06,
+      "loss": 0.0019,
+      "step": 25840
+    },
+    {
+      "epoch": 18.77,
+      "grad_norm": 0.00013374777336139232,
+      "learning_rate": 3.409182603082385e-06,
+      "loss": 0.0021,
+      "step": 25850
+    },
+    {
+      "epoch": 18.77,
+      "grad_norm": 0.0037013725377619267,
+      "learning_rate": 3.3890099249576373e-06,
+      "loss": 0.0085,
+      "step": 25860
+    },
+    {
+      "epoch": 18.78,
+      "grad_norm": 0.0005560120334848762,
+      "learning_rate": 3.36883724683289e-06,
+      "loss": 0.0059,
+      "step": 25870
+    },
+    {
+      "epoch": 18.79,
+      "grad_norm": 0.003002674551680684,
+      "learning_rate": 3.3486645687081416e-06,
+      "loss": 0.0031,
+      "step": 25880
+    },
+    {
+      "epoch": 18.79,
+      "grad_norm": 0.0007660058909095824,
+      "learning_rate": 3.328491890583394e-06,
+      "loss": 0.0029,
+      "step": 25890
+    },
+    {
+      "epoch": 18.8,
+      "grad_norm": 0.00030720618087798357,
+      "learning_rate": 3.3083192124586467e-06,
+      "loss": 0.0004,
+      "step": 25900
+    },
+    {
+      "epoch": 18.81,
+      "grad_norm": 0.00031959637999534607,
+      "learning_rate": 3.288146534333898e-06,
+      "loss": 0.0027,
+      "step": 25910
+    },
+    {
+      "epoch": 18.82,
+      "grad_norm": 0.0005406465497799218,
+      "learning_rate": 3.2679738562091506e-06,
+      "loss": 0.0016,
+      "step": 25920
+    },
+    {
+      "epoch": 18.82,
+      "grad_norm": 0.0006833134684711695,
+      "learning_rate": 3.2478011780844025e-06,
+      "loss": 0.0013,
+      "step": 25930
+    },
+    {
+      "epoch": 18.83,
+      "grad_norm": 6.237076013348997e-05,
+      "learning_rate": 3.227628499959655e-06,
+      "loss": 0.0039,
+      "step": 25940
+    },
+    {
+      "epoch": 18.84,
+      "grad_norm": 0.00028509943513199687,
+      "learning_rate": 3.207455821834907e-06,
+      "loss": 0.0013,
+      "step": 25950
+    },
+    {
+      "epoch": 18.85,
+      "grad_norm": 0.05667322129011154,
+      "learning_rate": 3.187283143710159e-06,
+      "loss": 0.0011,
+      "step": 25960
+    },
+    {
+      "epoch": 18.85,
+      "grad_norm": 0.00010201996337855235,
+      "learning_rate": 3.1671104655854115e-06,
+      "loss": 0.0021,
+      "step": 25970
+    },
+    {
+      "epoch": 18.86,
+      "grad_norm": 0.0016317203408107162,
+      "learning_rate": 3.146937787460663e-06,
+      "loss": 0.0009,
+      "step": 25980
+    },
+    {
+      "epoch": 18.87,
+      "grad_norm": 0.0036815290804952383,
+      "learning_rate": 3.1267651093359157e-06,
+      "loss": 0.0085,
+      "step": 25990
+    },
+    {
+      "epoch": 18.87,
+      "grad_norm": 0.027603503316640854,
+      "learning_rate": 3.1065924312111676e-06,
+      "loss": 0.0007,
+      "step": 26000
+    },
+    {
+      "epoch": 18.88,
+      "grad_norm": 0.02139933593571186,
+      "learning_rate": 3.0864197530864196e-06,
+      "loss": 0.0056,
+      "step": 26010
+    },
+    {
+      "epoch": 18.89,
+      "grad_norm": 8.762301149545237e-05,
+      "learning_rate": 3.0662470749616723e-06,
+      "loss": 0.0031,
+      "step": 26020
+    },
+    {
+      "epoch": 18.9,
+      "grad_norm": 0.00036692939465865493,
+      "learning_rate": 3.0460743968369243e-06,
+      "loss": 0.0017,
+      "step": 26030
+    },
+    {
+      "epoch": 18.9,
+      "grad_norm": 0.10278957337141037,
+      "learning_rate": 3.025901718712176e-06,
+      "loss": 0.0057,
+      "step": 26040
+    },
+    {
+      "epoch": 18.91,
+      "grad_norm": 5.523693471332081e-05,
+      "learning_rate": 3.0057290405874285e-06,
+      "loss": 0.0004,
+      "step": 26050
+    },
+    {
+      "epoch": 18.92,
+      "grad_norm": 0.0008847813005559146,
+      "learning_rate": 2.985556362462681e-06,
+      "loss": 0.0003,
+      "step": 26060
+    },
+    {
+      "epoch": 18.93,
+      "grad_norm": 5.930370025453158e-05,
+      "learning_rate": 2.965383684337933e-06,
+      "loss": 0.0007,
+      "step": 26070
+    },
+    {
+      "epoch": 18.93,
+      "grad_norm": 0.002461926778778434,
+      "learning_rate": 2.945211006213185e-06,
+      "loss": 0.0,
+      "step": 26080
+    },
+    {
+      "epoch": 18.94,
+      "grad_norm": 0.0001589566090842709,
+      "learning_rate": 2.925038328088437e-06,
+      "loss": 0.0038,
+      "step": 26090
+    },
+    {
+      "epoch": 18.95,
+      "grad_norm": 0.0006895341211929917,
+      "learning_rate": 2.9048656499636894e-06,
+      "loss": 0.0037,
+      "step": 26100
+    },
+    {
+      "epoch": 18.95,
+      "grad_norm": 3.800967169809155e-05,
+      "learning_rate": 2.8846929718389414e-06,
+      "loss": 0.002,
+      "step": 26110
+    },
+    {
+      "epoch": 18.96,
+      "grad_norm": 0.007250432390719652,
+      "learning_rate": 2.8645202937141937e-06,
+      "loss": 0.0004,
+      "step": 26120
+    },
+    {
+      "epoch": 18.97,
+      "grad_norm": 0.0022204366978257895,
+      "learning_rate": 2.8443476155894456e-06,
+      "loss": 0.0108,
+      "step": 26130
+    },
+    {
+      "epoch": 18.98,
+      "grad_norm": 3.95225033571478e-05,
+      "learning_rate": 2.824174937464698e-06,
+      "loss": 0.0032,
+      "step": 26140
+    },
+    {
+      "epoch": 18.98,
+      "grad_norm": 0.332366406917572,
+      "learning_rate": 2.8040022593399503e-06,
+      "loss": 0.0072,
+      "step": 26150
+    },
+    {
+      "epoch": 18.99,
+      "grad_norm": 3.236500560888089e-05,
+      "learning_rate": 2.7838295812152022e-06,
+      "loss": 0.0011,
+      "step": 26160
+    },
+    {
+      "epoch": 19.0,
+      "grad_norm": 0.00018790685862768441,
+      "learning_rate": 2.7636569030904546e-06,
+      "loss": 0.0024,
+      "step": 26170
+    },
+    {
+      "epoch": 19.0,
+      "eval_accuracy": 0.99837772836593,
+      "eval_f1": 0.9962558584033724,
+      "eval_loss": 0.0031478386372327805,
+      "eval_precision": 0.9993696485790828,
+      "eval_recall": 0.993161411568177,
+      "eval_roc_auc": 0.9999876608903375,
+      "eval_runtime": 387.535,
+      "eval_samples_per_second": 227.458,
+      "eval_steps_per_second": 14.218,
+      "step": 26172
+    },
+    {
+      "epoch": 19.01,
+      "grad_norm": 0.03183314949274063,
+      "learning_rate": 2.7434842249657065e-06,
+      "loss": 0.0072,
+      "step": 26180
+    },
+    {
+      "epoch": 19.01,
+      "grad_norm": 0.0002327613183297217,
+      "learning_rate": 2.723311546840959e-06,
+      "loss": 0.0049,
+      "step": 26190
+    },
+    {
+      "epoch": 19.02,
+      "grad_norm": 0.0012804355937987566,
+      "learning_rate": 2.703138868716211e-06,
+      "loss": 0.0102,
+      "step": 26200
+    },
+    {
+      "epoch": 19.03,
+      "grad_norm": 0.0003078650333918631,
+      "learning_rate": 2.682966190591463e-06,
+      "loss": 0.003,
+      "step": 26210
+    },
+    {
+      "epoch": 19.03,
+      "grad_norm": 0.09908101707696915,
+      "learning_rate": 2.662793512466715e-06,
+      "loss": 0.009,
+      "step": 26220
+    },
+    {
+      "epoch": 19.04,
+      "grad_norm": 0.09956044703722,
+      "learning_rate": 2.6426208343419674e-06,
+      "loss": 0.0048,
+      "step": 26230
+    },
+    {
+      "epoch": 19.05,
+      "grad_norm": 0.00046755580115132034,
+      "learning_rate": 2.6224481562172198e-06,
+      "loss": 0.0025,
+      "step": 26240
+    },
+    {
+      "epoch": 19.06,
+      "grad_norm": 3.5852412111125886e-05,
+      "learning_rate": 2.6022754780924717e-06,
+      "loss": 0.0,
+      "step": 26250
+    },
+    {
+      "epoch": 19.06,
+      "grad_norm": 0.00015138008166104555,
+      "learning_rate": 2.5821027999677236e-06,
+      "loss": 0.002,
+      "step": 26260
+    },
+    {
+      "epoch": 19.07,
+      "grad_norm": 0.00530798826366663,
+      "learning_rate": 2.561930121842976e-06,
+      "loss": 0.0015,
+      "step": 26270
+    },
+    {
+      "epoch": 19.08,
+      "grad_norm": 0.0002312654396519065,
+      "learning_rate": 2.5417574437182283e-06,
+      "loss": 0.0004,
+      "step": 26280
+    },
+    {
+      "epoch": 19.09,
+      "grad_norm": 0.0009702076204121113,
+      "learning_rate": 2.5215847655934802e-06,
+      "loss": 0.0054,
+      "step": 26290
+    },
+    {
+      "epoch": 19.09,
+      "grad_norm": 6.126202788436785e-05,
+      "learning_rate": 2.5014120874687326e-06,
+      "loss": 0.001,
+      "step": 26300
+    },
+    {
+      "epoch": 19.1,
+      "grad_norm": 1.3822760581970215,
+      "learning_rate": 2.4812394093439845e-06,
+      "loss": 0.0025,
+      "step": 26310
+    },
+    {
+      "epoch": 19.11,
+      "grad_norm": 0.049456097185611725,
+      "learning_rate": 2.4610667312192364e-06,
+      "loss": 0.0023,
+      "step": 26320
+    },
+    {
+      "epoch": 19.11,
+      "grad_norm": 0.00021031413052696735,
+      "learning_rate": 2.440894053094489e-06,
+      "loss": 0.0016,
+      "step": 26330
+    },
+    {
+      "epoch": 19.12,
+      "grad_norm": 0.04800700768828392,
+      "learning_rate": 2.420721374969741e-06,
+      "loss": 0.0021,
+      "step": 26340
+    },
+    {
+      "epoch": 19.13,
+      "grad_norm": 0.009497404098510742,
+      "learning_rate": 2.400548696844993e-06,
+      "loss": 0.003,
+      "step": 26350
+    },
+    {
+      "epoch": 19.14,
+      "grad_norm": 0.04641493409872055,
+      "learning_rate": 2.3803760187202454e-06,
+      "loss": 0.0017,
+      "step": 26360
+    },
+    {
+      "epoch": 19.14,
+      "grad_norm": 9.963675984181464e-05,
+      "learning_rate": 2.3602033405954977e-06,
+      "loss": 0.0017,
+      "step": 26370
+    },
+    {
+      "epoch": 19.15,
+      "grad_norm": 0.001153616583906114,
+      "learning_rate": 2.3400306624707497e-06,
+      "loss": 0.0041,
+      "step": 26380
+    },
+    {
+      "epoch": 19.16,
+      "grad_norm": 0.0001510605070507154,
+      "learning_rate": 2.319857984346002e-06,
+      "loss": 0.0007,
+      "step": 26390
+    },
+    {
+      "epoch": 19.17,
+      "grad_norm": 7.923934754217044e-05,
+      "learning_rate": 2.299685306221254e-06,
+      "loss": 0.0076,
+      "step": 26400
+    },
+    {
+      "epoch": 19.17,
+      "grad_norm": 0.00012218714982736856,
+      "learning_rate": 2.2795126280965063e-06,
+      "loss": 0.0021,
+      "step": 26410
+    },
+    {
+      "epoch": 19.18,
+      "grad_norm": 0.0008200127049349248,
+      "learning_rate": 2.2593399499717586e-06,
+      "loss": 0.0035,
+      "step": 26420
+    },
+    {
+      "epoch": 19.19,
+      "grad_norm": 0.05239934101700783,
+      "learning_rate": 2.2391672718470106e-06,
+      "loss": 0.0025,
+      "step": 26430
+    },
+    {
+      "epoch": 19.19,
+      "grad_norm": 3.865266262437217e-05,
+      "learning_rate": 2.2189945937222625e-06,
+      "loss": 0.0018,
+      "step": 26440
+    },
+    {
+      "epoch": 19.2,
+      "grad_norm": 8.964262815425172e-05,
+      "learning_rate": 2.198821915597515e-06,
+      "loss": 0.0041,
+      "step": 26450
+    },
+    {
+      "epoch": 19.21,
+      "grad_norm": 0.0628039538860321,
+      "learning_rate": 2.178649237472767e-06,
+      "loss": 0.0062,
+      "step": 26460
+    },
+    {
+      "epoch": 19.22,
+      "grad_norm": 0.048729509115219116,
+      "learning_rate": 2.158476559348019e-06,
+      "loss": 0.0011,
+      "step": 26470
+    },
+    {
+      "epoch": 19.22,
+      "grad_norm": 0.00014290747640188783,
+      "learning_rate": 2.1383038812232715e-06,
+      "loss": 0.0031,
+      "step": 26480
+    },
+    {
+      "epoch": 19.23,
+      "grad_norm": 0.0006844107992947102,
+      "learning_rate": 2.1181312030985234e-06,
+      "loss": 0.0025,
+      "step": 26490
+    },
+    {
+      "epoch": 19.24,
+      "grad_norm": 0.014656140469014645,
+      "learning_rate": 2.0979585249737757e-06,
+      "loss": 0.0143,
+      "step": 26500
+    },
+    {
+      "epoch": 19.25,
+      "grad_norm": 0.0003018031711690128,
+      "learning_rate": 2.0777858468490276e-06,
+      "loss": 0.0038,
+      "step": 26510
+    },
+    {
+      "epoch": 19.25,
+      "grad_norm": 6.837755790911615e-05,
+      "learning_rate": 2.05761316872428e-06,
+      "loss": 0.006,
+      "step": 26520
+    },
+    {
+      "epoch": 19.26,
+      "grad_norm": 0.09663214534521103,
+      "learning_rate": 2.037440490599532e-06,
+      "loss": 0.0061,
+      "step": 26530
+    },
+    {
+      "epoch": 19.27,
+      "grad_norm": 0.002400600351393223,
+      "learning_rate": 2.0172678124747843e-06,
+      "loss": 0.0026,
+      "step": 26540
+    },
+    {
+      "epoch": 19.27,
+      "grad_norm": 0.046237923204898834,
+      "learning_rate": 1.9970951343500366e-06,
+      "loss": 0.0031,
+      "step": 26550
+    },
+    {
+      "epoch": 19.28,
+      "grad_norm": 9.103987395064905e-05,
+      "learning_rate": 1.9769224562252885e-06,
+      "loss": 0.0008,
+      "step": 26560
+    },
+    {
+      "epoch": 19.29,
+      "grad_norm": 0.02744593285024166,
+      "learning_rate": 1.956749778100541e-06,
+      "loss": 0.0009,
+      "step": 26570
+    },
+    {
+      "epoch": 19.3,
+      "grad_norm": 0.00013158208457753062,
+      "learning_rate": 1.936577099975793e-06,
+      "loss": 0.0046,
+      "step": 26580
+    },
+    {
+      "epoch": 19.3,
+      "grad_norm": 3.296025897725485e-05,
+      "learning_rate": 1.916404421851045e-06,
+      "loss": 0.0047,
+      "step": 26590
+    },
+    {
+      "epoch": 19.31,
+      "grad_norm": 0.0005842273822054267,
+      "learning_rate": 1.8962317437262973e-06,
+      "loss": 0.0044,
+      "step": 26600
+    },
+    {
+      "epoch": 19.32,
+      "grad_norm": 0.0005419492954388261,
+      "learning_rate": 1.8760590656015494e-06,
+      "loss": 0.0014,
+      "step": 26610
+    },
+    {
+      "epoch": 19.32,
+      "grad_norm": 0.00013932572619523853,
+      "learning_rate": 1.8558863874768014e-06,
+      "loss": 0.0,
+      "step": 26620
+    },
+    {
+      "epoch": 19.33,
+      "grad_norm": 0.03754022344946861,
+      "learning_rate": 1.8357137093520535e-06,
+      "loss": 0.0092,
+      "step": 26630
+    },
+    {
+      "epoch": 19.34,
+      "grad_norm": 68.23185729980469,
+      "learning_rate": 1.8155410312273058e-06,
+      "loss": 0.0017,
+      "step": 26640
+    },
+    {
+      "epoch": 19.35,
+      "grad_norm": 0.0001436081511201337,
+      "learning_rate": 1.795368353102558e-06,
+      "loss": 0.0023,
+      "step": 26650
+    },
+    {
+      "epoch": 19.35,
+      "grad_norm": 0.002541495719924569,
+      "learning_rate": 1.7751956749778101e-06,
+      "loss": 0.0013,
+      "step": 26660
+    },
+    {
+      "epoch": 19.36,
+      "grad_norm": 0.146305650472641,
+      "learning_rate": 1.7550229968530623e-06,
+      "loss": 0.0059,
+      "step": 26670
+    },
+    {
+      "epoch": 19.37,
+      "grad_norm": 0.09827929735183716,
+      "learning_rate": 1.7348503187283146e-06,
+      "loss": 0.0024,
+      "step": 26680
+    },
+    {
+      "epoch": 19.38,
+      "grad_norm": 0.005435886327177286,
+      "learning_rate": 1.7146776406035667e-06,
+      "loss": 0.0049,
+      "step": 26690
+    },
+    {
+      "epoch": 19.38,
+      "grad_norm": 0.00013977414346300066,
+      "learning_rate": 1.6945049624788187e-06,
+      "loss": 0.0009,
+      "step": 26700
+    },
+    {
+      "epoch": 19.39,
+      "grad_norm": 0.0022497973404824734,
+      "learning_rate": 1.6743322843540708e-06,
+      "loss": 0.0031,
+      "step": 26710
+    },
+    {
+      "epoch": 19.4,
+      "grad_norm": 0.0001336606073891744,
+      "learning_rate": 1.6541596062293234e-06,
+      "loss": 0.0049,
+      "step": 26720
+    },
+    {
+      "epoch": 19.4,
+      "grad_norm": 0.1333574801683426,
+      "learning_rate": 1.6339869281045753e-06,
+      "loss": 0.0085,
+      "step": 26730
+    },
+    {
+      "epoch": 19.41,
+      "grad_norm": 0.00023709374363534153,
+      "learning_rate": 1.6138142499798274e-06,
+      "loss": 0.0055,
+      "step": 26740
+    },
+    {
+      "epoch": 19.42,
+      "grad_norm": 0.030887536704540253,
+      "learning_rate": 1.5936415718550796e-06,
+      "loss": 0.0025,
+      "step": 26750
+    },
+    {
+      "epoch": 19.43,
+      "grad_norm": 0.0002531045174691826,
+      "learning_rate": 1.5734688937303315e-06,
+      "loss": 0.0048,
+      "step": 26760
+    },
+    {
+      "epoch": 19.43,
+      "grad_norm": 9.092326217796654e-05,
+      "learning_rate": 1.5532962156055838e-06,
+      "loss": 0.0003,
+      "step": 26770
+    },
+    {
+      "epoch": 19.44,
+      "grad_norm": 0.09485254436731339,
+      "learning_rate": 1.5331235374808362e-06,
+      "loss": 0.0018,
+      "step": 26780
+    },
+    {
+      "epoch": 19.45,
+      "grad_norm": 0.00014884411939419806,
+      "learning_rate": 1.512950859356088e-06,
+      "loss": 0.0024,
+      "step": 26790
+    },
+    {
+      "epoch": 19.46,
+      "grad_norm": 0.1054830327630043,
+      "learning_rate": 1.4927781812313404e-06,
+      "loss": 0.0056,
+      "step": 26800
+    },
+    {
+      "epoch": 19.46,
+      "grad_norm": 0.00028030065004713833,
+      "learning_rate": 1.4726055031065926e-06,
+      "loss": 0.0004,
+      "step": 26810
+    },
+    {
+      "epoch": 19.47,
+      "grad_norm": 0.07415413856506348,
+      "learning_rate": 1.4524328249818447e-06,
+      "loss": 0.0062,
+      "step": 26820
+    },
+    {
+      "epoch": 19.48,
+      "grad_norm": 0.000137577997520566,
+      "learning_rate": 1.4322601468570969e-06,
+      "loss": 0.001,
+      "step": 26830
+    },
+    {
+      "epoch": 19.48,
+      "grad_norm": 0.14892277121543884,
+      "learning_rate": 1.412087468732349e-06,
+      "loss": 0.0072,
+      "step": 26840
+    },
+    {
+      "epoch": 19.49,
+      "grad_norm": 0.10421720147132874,
+      "learning_rate": 1.3919147906076011e-06,
+      "loss": 0.0066,
+      "step": 26850
+    },
+    {
+      "epoch": 19.5,
+      "grad_norm": 0.00022085083764977753,
+      "learning_rate": 1.3717421124828533e-06,
+      "loss": 0.0005,
+      "step": 26860
+    },
+    {
+      "epoch": 19.51,
+      "grad_norm": 0.05538201332092285,
+      "learning_rate": 1.3515694343581054e-06,
+      "loss": 0.0041,
+      "step": 26870
+    },
+    {
+      "epoch": 19.51,
+      "grad_norm": 0.03088965266942978,
+      "learning_rate": 1.3313967562333575e-06,
+      "loss": 0.0065,
+      "step": 26880
+    },
+    {
+      "epoch": 19.52,
+      "grad_norm": 0.00028383126482367516,
+      "learning_rate": 1.3112240781086099e-06,
+      "loss": 0.0015,
+      "step": 26890
+    },
+    {
+      "epoch": 19.53,
+      "grad_norm": 0.00039556692354381084,
+      "learning_rate": 1.2910513999838618e-06,
+      "loss": 0.0021,
+      "step": 26900
+    },
+    {
+      "epoch": 19.54,
+      "grad_norm": 0.0002842575777322054,
+      "learning_rate": 1.2708787218591142e-06,
+      "loss": 0.0041,
+      "step": 26910
+    },
+    {
+      "epoch": 19.54,
+      "grad_norm": 0.10765402764081955,
+      "learning_rate": 1.2507060437343663e-06,
+      "loss": 0.0022,
+      "step": 26920
+    },
+    {
+      "epoch": 19.55,
+      "grad_norm": 0.0026703434996306896,
+      "learning_rate": 1.2305333656096182e-06,
+      "loss": 0.0026,
+      "step": 26930
+    },
+    {
+      "epoch": 19.56,
+      "grad_norm": 0.00041996565414592624,
+      "learning_rate": 1.2103606874848706e-06,
+      "loss": 0.0023,
+      "step": 26940
+    },
+    {
+      "epoch": 19.56,
+      "grad_norm": 0.0009542067418806255,
+      "learning_rate": 1.1901880093601227e-06,
+      "loss": 0.0033,
+      "step": 26950
+    },
+    {
+      "epoch": 19.57,
+      "grad_norm": 0.00019776627596002072,
+      "learning_rate": 1.1700153312353748e-06,
+      "loss": 0.0025,
+      "step": 26960
+    },
+    {
+      "epoch": 19.58,
+      "grad_norm": 0.0005681757465936244,
+      "learning_rate": 1.149842653110627e-06,
+      "loss": 0.0045,
+      "step": 26970
+    },
+    {
+      "epoch": 19.59,
+      "grad_norm": 0.09003600478172302,
+      "learning_rate": 1.1296699749858793e-06,
+      "loss": 0.0034,
+      "step": 26980
+    },
+    {
+      "epoch": 19.59,
+      "grad_norm": 4.978038850822486e-05,
+      "learning_rate": 1.1094972968611312e-06,
+      "loss": 0.0007,
+      "step": 26990
+    },
+    {
+      "epoch": 19.6,
+      "grad_norm": 6.065259367460385e-05,
+      "learning_rate": 1.0893246187363836e-06,
+      "loss": 0.0038,
+      "step": 27000
+    },
+    {
+      "epoch": 19.61,
+      "grad_norm": 0.00032748805824667215,
+      "learning_rate": 1.0691519406116357e-06,
+      "loss": 0.0044,
+      "step": 27010
+    },
+    {
+      "epoch": 19.62,
+      "grad_norm": 0.14997267723083496,
+      "learning_rate": 1.0489792624868879e-06,
+      "loss": 0.0059,
+      "step": 27020
+    },
+    {
+      "epoch": 19.62,
+      "grad_norm": 0.0001299285504501313,
+      "learning_rate": 1.02880658436214e-06,
+      "loss": 0.0044,
+      "step": 27030
+    },
+    {
+      "epoch": 19.63,
+      "grad_norm": 0.0002876422367990017,
+      "learning_rate": 1.0086339062373921e-06,
+      "loss": 0.0017,
+      "step": 27040
+    },
+    {
+      "epoch": 19.64,
+      "grad_norm": 2.906100417021662e-05,
+      "learning_rate": 9.884612281126443e-07,
+      "loss": 0.0036,
+      "step": 27050
+    },
+    {
+      "epoch": 19.64,
+      "grad_norm": 0.000504647905472666,
+      "learning_rate": 9.682885499878964e-07,
+      "loss": 0.0021,
+      "step": 27060
+    },
+    {
+      "epoch": 19.65,
+      "grad_norm": 0.00014128659677226096,
+      "learning_rate": 9.481158718631486e-07,
+      "loss": 0.0044,
+      "step": 27070
+    },
+    {
+      "epoch": 19.66,
+      "grad_norm": 0.056433625519275665,
+      "learning_rate": 9.279431937384007e-07,
+      "loss": 0.0039,
+      "step": 27080
+    },
+    {
+      "epoch": 19.67,
+      "grad_norm": 2.6254705517203547e-05,
+      "learning_rate": 9.077705156136529e-07,
+      "loss": 0.0012,
+      "step": 27090
+    },
+    {
+      "epoch": 19.67,
+      "grad_norm": 0.0007919368799775839,
+      "learning_rate": 8.875978374889051e-07,
+      "loss": 0.0006,
+      "step": 27100
+    },
+    {
+      "epoch": 19.68,
+      "grad_norm": 0.12109678238630295,
+      "learning_rate": 8.674251593641573e-07,
+      "loss": 0.0053,
+      "step": 27110
+    },
+    {
+      "epoch": 19.69,
+      "grad_norm": 0.0010482355719432235,
+      "learning_rate": 8.472524812394093e-07,
+      "loss": 0.0039,
+      "step": 27120
+    },
+    {
+      "epoch": 19.7,
+      "grad_norm": 0.11032991856336594,
+      "learning_rate": 8.270798031146617e-07,
+      "loss": 0.0013,
+      "step": 27130
+    },
+    {
+      "epoch": 19.7,
+      "grad_norm": 0.0005860592355020344,
+      "learning_rate": 8.069071249899137e-07,
+      "loss": 0.002,
+      "step": 27140
+    },
+    {
+      "epoch": 19.71,
+      "grad_norm": 0.0007409679819829762,
+      "learning_rate": 7.867344468651657e-07,
+      "loss": 0.0012,
+      "step": 27150
+    },
+    {
+      "epoch": 19.72,
+      "grad_norm": 0.13628806173801422,
+      "learning_rate": 7.665617687404181e-07,
+      "loss": 0.0068,
+      "step": 27160
+    },
+    {
+      "epoch": 19.72,
+      "grad_norm": 4.722082303487696e-05,
+      "learning_rate": 7.463890906156702e-07,
+      "loss": 0.0014,
+      "step": 27170
+    },
+    {
+      "epoch": 19.73,
+      "grad_norm": 0.01145484484732151,
+      "learning_rate": 7.262164124909224e-07,
+      "loss": 0.0011,
+      "step": 27180
+    },
+    {
+      "epoch": 19.74,
+      "grad_norm": 3.4825185139197856e-05,
+      "learning_rate": 7.060437343661745e-07,
+      "loss": 0.0032,
+      "step": 27190
+    },
+    {
+      "epoch": 19.75,
+      "grad_norm": 0.03905890882015228,
+      "learning_rate": 6.858710562414266e-07,
+      "loss": 0.0013,
+      "step": 27200
+    },
+    {
+      "epoch": 19.75,
+      "grad_norm": 6.594491424039006e-05,
+      "learning_rate": 6.656983781166788e-07,
+      "loss": 0.0007,
+      "step": 27210
+    },
+    {
+      "epoch": 19.76,
+      "grad_norm": 0.17883096635341644,
+      "learning_rate": 6.455256999919309e-07,
+      "loss": 0.0057,
+      "step": 27220
+    },
+    {
+      "epoch": 19.77,
+      "grad_norm": 1.921937109727878e-05,
+      "learning_rate": 6.253530218671831e-07,
+      "loss": 0.0017,
+      "step": 27230
+    },
+    {
+      "epoch": 19.77,
+      "grad_norm": 0.00039736239705234766,
+      "learning_rate": 6.051803437424353e-07,
+      "loss": 0.0073,
+      "step": 27240
+    },
+    {
+      "epoch": 19.78,
+      "grad_norm": 0.006403313484042883,
+      "learning_rate": 5.850076656176874e-07,
+      "loss": 0.0032,
+      "step": 27250
+    },
+    {
+      "epoch": 19.79,
+      "grad_norm": 0.0003711440076585859,
+      "learning_rate": 5.648349874929397e-07,
+      "loss": 0.0011,
+      "step": 27260
+    },
+    {
+      "epoch": 19.8,
+      "grad_norm": 0.00021775624190922827,
+      "learning_rate": 5.446623093681918e-07,
+      "loss": 0.0016,
+      "step": 27270
+    },
+    {
+      "epoch": 19.8,
+      "grad_norm": 0.09723830223083496,
+      "learning_rate": 5.244896312434439e-07,
+      "loss": 0.0037,
+      "step": 27280
+    },
+    {
+      "epoch": 19.81,
+      "grad_norm": 3.505588392727077e-05,
+      "learning_rate": 5.043169531186961e-07,
+      "loss": 0.0043,
+      "step": 27290
+    },
+    {
+      "epoch": 19.82,
+      "grad_norm": 0.13812561333179474,
+      "learning_rate": 4.841442749939482e-07,
+      "loss": 0.0048,
+      "step": 27300
+    },
+    {
+      "epoch": 19.83,
+      "grad_norm": 0.0003846702165901661,
+      "learning_rate": 4.6397159686920034e-07,
+      "loss": 0.0018,
+      "step": 27310
+    },
+    {
+      "epoch": 19.83,
+      "grad_norm": 0.0003796774835791439,
+      "learning_rate": 4.4379891874445253e-07,
+      "loss": 0.0038,
+      "step": 27320
+    },
+    {
+      "epoch": 19.84,
+      "grad_norm": 0.026537004858255386,
+      "learning_rate": 4.2362624061970466e-07,
+      "loss": 0.0026,
+      "step": 27330
+    },
+    {
+      "epoch": 19.85,
+      "grad_norm": 0.10081913322210312,
+      "learning_rate": 4.0345356249495685e-07,
+      "loss": 0.0022,
+      "step": 27340
+    },
+    {
+      "epoch": 19.85,
+      "grad_norm": 0.13242417573928833,
+      "learning_rate": 3.8328088437020904e-07,
+      "loss": 0.007,
+      "step": 27350
+    },
+    {
+      "epoch": 19.86,
+      "grad_norm": 0.0003329771861899644,
+      "learning_rate": 3.631082062454612e-07,
+      "loss": 0.0,
+      "step": 27360
+    },
+    {
+      "epoch": 19.87,
+      "grad_norm": 0.00014159978309180588,
+      "learning_rate": 3.429355281207133e-07,
+      "loss": 0.0059,
+      "step": 27370
+    },
+    {
+      "epoch": 19.88,
+      "grad_norm": 0.0008457335061393678,
+      "learning_rate": 3.2276284999596545e-07,
+      "loss": 0.0028,
+      "step": 27380
+    },
+    {
+      "epoch": 19.88,
+      "grad_norm": 0.14093543589115143,
+      "learning_rate": 3.0259017187121764e-07,
+      "loss": 0.0057,
+      "step": 27390
+    },
+    {
+      "epoch": 19.89,
+      "grad_norm": 0.0995154157280922,
+      "learning_rate": 2.8241749374646983e-07,
+      "loss": 0.0042,
+      "step": 27400
+    },
+    {
+      "epoch": 19.9,
+      "grad_norm": 1.3960028809378855e-05,
+      "learning_rate": 2.6224481562172197e-07,
+      "loss": 0.006,
+      "step": 27410
+    },
+    {
+      "epoch": 19.91,
+      "grad_norm": 0.00027196883456781507,
+      "learning_rate": 2.420721374969741e-07,
+      "loss": 0.004,
+      "step": 27420
+    },
+    {
+      "epoch": 19.91,
+      "grad_norm": 9.932199463946745e-05,
+      "learning_rate": 2.2189945937222626e-07,
+      "loss": 0.0064,
+      "step": 27430
+    },
+    {
+      "epoch": 19.92,
+      "grad_norm": 0.0010937968036159873,
+      "learning_rate": 2.0172678124747843e-07,
+      "loss": 0.0028,
+      "step": 27440
+    },
+    {
+      "epoch": 19.93,
+      "grad_norm": 2.9746237487415783e-05,
+      "learning_rate": 1.815541031227306e-07,
+      "loss": 0.0009,
+      "step": 27450
+    },
+    {
+      "epoch": 19.93,
+      "grad_norm": 0.0013092844747006893,
+      "learning_rate": 1.6138142499798273e-07,
+      "loss": 0.0007,
+      "step": 27460
+    },
+    {
+      "epoch": 19.94,
+      "grad_norm": 0.00010451417620060965,
+      "learning_rate": 1.4120874687323491e-07,
+      "loss": 0.0011,
+      "step": 27470
+    },
+    {
+      "epoch": 19.95,
+      "grad_norm": 3.361936614965089e-05,
+      "learning_rate": 1.2103606874848705e-07,
+      "loss": 0.003,
+      "step": 27480
+    },
+    {
+      "epoch": 19.96,
+      "grad_norm": 0.0003154365695081651,
+      "learning_rate": 1.0086339062373921e-07,
+      "loss": 0.0029,
+      "step": 27490
+    },
+    {
+      "epoch": 19.96,
+      "grad_norm": 0.0006699699442833662,
+      "learning_rate": 8.069071249899136e-08,
+      "loss": 0.001,
+      "step": 27500
+    },
+    {
+      "epoch": 19.97,
+      "grad_norm": 0.00039897230453789234,
+      "learning_rate": 6.051803437424353e-08,
+      "loss": 0.0052,
+      "step": 27510
+    },
+    {
+      "epoch": 19.98,
+      "grad_norm": 0.0004209213948342949,
+      "learning_rate": 4.034535624949568e-08,
+      "loss": 0.0043,
+      "step": 27520
+    },
+    {
+      "epoch": 19.99,
+      "grad_norm": 0.10861359536647797,
+      "learning_rate": 2.017267812474784e-08,
+      "loss": 0.0044,
+      "step": 27530
+    },
+    {
+      "epoch": 19.99,
+      "grad_norm": 0.0003400925197638571,
+      "learning_rate": 0.0,
+      "loss": 0.0023,
+      "step": 27540
+    },
+    {
+      "epoch": 19.99,
+      "eval_accuracy": 0.99837772836593,
+      "eval_f1": 0.9962540929927963,
+      "eval_loss": 0.003140063723549247,
+      "eval_precision": 0.9998422630001578,
+      "eval_recall": 0.9926915848820212,
+      "eval_roc_auc": 0.9999876949396973,
+      "eval_runtime": 387.381,
+      "eval_samples_per_second": 227.549,
+      "eval_steps_per_second": 14.224,
+      "step": 27540
+    },
+    {
+      "epoch": 19.99,
+      "step": 27540,
+      "total_flos": 4.380450478142158e+19,
+      "train_loss": 0.029798476457713677,
+      "train_runtime": 23573.1101,
+      "train_samples_per_second": 74.787,
+      "train_steps_per_second": 1.168
     }
   ],
   "logging_steps": 10,
-  "max_steps": 2154,
-  "num_train_epochs": 3,
+  "max_steps": 27540,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 20,
   "save_steps": 500,
-  "total_flos": 6.851603075929178e+18,
+  "total_flos": 4.380450478142158e+19,
+  "train_batch_size": 16,
   "trial_name": null,
   "trial_params": null
 }