diff --git "a/trainer_state.json" "b/trainer_state.json"
--- "a/trainer_state.json"
+++ "b/trainer_state.json"
@@ -1,6 +1,6 @@
 {
-  "best_metric": 0.8306,
-  "best_model_checkpoint": "checkpoint/resnet-50/checkpoint-93240",
+  "best_metric": 0.8328,
+  "best_model_checkpoint": "checkpoint/cifar100/checkpoint-99567",
   "epoch": 300.0,
   "eval_steps": 500,
   "global_step": 99900,
@@ -10,72642 +10,72642 @@
   "log_history": [
     {
       "epoch": 0.03,
-      "grad_norm": 0.5662223696708679,
+      "grad_norm": 0.5893929600715637,
       "learning_rate": 9.998998998999e-06,
-      "loss": 4.6049,
+      "loss": 4.6054,
       "step": 10
     },
     {
       "epoch": 0.06,
-      "grad_norm": 0.5612590312957764,
+      "grad_norm": 1.0497225522994995,
       "learning_rate": 9.997997997998e-06,
-      "loss": 4.6045,
+      "loss": 4.6042,
       "step": 20
     },
     {
       "epoch": 0.09,
-      "grad_norm": 0.5509904026985168,
+      "grad_norm": 2.149851083755493,
       "learning_rate": 9.996996996996998e-06,
-      "loss": 4.6035,
+      "loss": 4.6038,
       "step": 30
     },
     {
       "epoch": 0.12,
-      "grad_norm": 0.5701844692230225,
+      "grad_norm": 0.7253397703170776,
       "learning_rate": 9.995995995995997e-06,
-      "loss": 4.6028,
+      "loss": 4.6022,
       "step": 40
     },
     {
       "epoch": 0.15,
-      "grad_norm": 0.5322362184524536,
+      "grad_norm": 0.9531165361404419,
       "learning_rate": 9.994994994994995e-06,
-      "loss": 4.6022,
+      "loss": 4.6019,
       "step": 50
     },
     {
       "epoch": 0.18,
-      "grad_norm": 0.6232250928878784,
+      "grad_norm": 0.5869887471199036,
       "learning_rate": 9.993993993993994e-06,
-      "loss": 4.6017,
+      "loss": 4.6026,
       "step": 60
     },
     {
       "epoch": 0.21,
-      "grad_norm": 0.5879863500595093,
+      "grad_norm": 0.8752070069313049,
       "learning_rate": 9.992992992992994e-06,
-      "loss": 4.6007,
+      "loss": 4.6004,
       "step": 70
     },
     {
       "epoch": 0.24,
-      "grad_norm": 0.5930666923522949,
+      "grad_norm": 1.3148916959762573,
       "learning_rate": 9.991991991991993e-06,
       "loss": 4.6002,
       "step": 80
     },
     {
       "epoch": 0.27,
-      "grad_norm": 0.6648198962211609,
+      "grad_norm": 1.710907220840454,
       "learning_rate": 9.990990990990992e-06,
-      "loss": 4.5994,
+      "loss": 4.6,
       "step": 90
     },
     {
       "epoch": 0.3,
-      "grad_norm": 0.5807310342788696,
+      "grad_norm": 0.5566960573196411,
       "learning_rate": 9.989989989989992e-06,
-      "loss": 4.5983,
+      "loss": 4.5992,
       "step": 100
     },
     {
       "epoch": 0.33,
-      "grad_norm": 0.6038708686828613,
+      "grad_norm": 1.0048409700393677,
       "learning_rate": 9.98898898898899e-06,
-      "loss": 4.5971,
+      "loss": 4.598,
       "step": 110
     },
     {
       "epoch": 0.36,
-      "grad_norm": 0.6390262842178345,
+      "grad_norm": 3.1871283054351807,
       "learning_rate": 9.987987987987989e-06,
-      "loss": 4.5969,
+      "loss": 4.5982,
       "step": 120
     },
     {
       "epoch": 0.39,
-      "grad_norm": 0.6634562015533447,
+      "grad_norm": 1.6046234369277954,
       "learning_rate": 9.986986986986988e-06,
-      "loss": 4.5948,
+      "loss": 4.5961,
       "step": 130
     },
     {
       "epoch": 0.42,
-      "grad_norm": 0.7678673267364502,
+      "grad_norm": 0.9088203310966492,
       "learning_rate": 9.985985985985986e-06,
-      "loss": 4.594,
+      "loss": 4.5955,
       "step": 140
     },
     {
       "epoch": 0.45,
-      "grad_norm": 0.6390228271484375,
+      "grad_norm": 1.368189811706543,
       "learning_rate": 9.984984984984985e-06,
-      "loss": 4.5932,
+      "loss": 4.5949,
       "step": 150
     },
     {
       "epoch": 0.48,
-      "grad_norm": 0.6680023670196533,
+      "grad_norm": 1.1708967685699463,
       "learning_rate": 9.983983983983985e-06,
-      "loss": 4.5924,
+      "loss": 4.5953,
       "step": 160
     },
     {
       "epoch": 0.51,
-      "grad_norm": 0.813166618347168,
+      "grad_norm": 2.170832395553589,
       "learning_rate": 9.982982982982984e-06,
-      "loss": 4.5911,
+      "loss": 4.595,
       "step": 170
     },
     {
       "epoch": 0.54,
-      "grad_norm": 0.7710539698600769,
+      "grad_norm": 0.9115569591522217,
       "learning_rate": 9.981981981981982e-06,
-      "loss": 4.5898,
+      "loss": 4.5929,
       "step": 180
     },
     {
       "epoch": 0.57,
-      "grad_norm": 0.7087071537971497,
+      "grad_norm": 1.8154083490371704,
       "learning_rate": 9.980980980980983e-06,
-      "loss": 4.5888,
+      "loss": 4.5918,
       "step": 190
     },
     {
       "epoch": 0.6,
-      "grad_norm": 0.7266985774040222,
+      "grad_norm": 0.8806125521659851,
       "learning_rate": 9.979979979979981e-06,
-      "loss": 4.5865,
+      "loss": 4.5909,
       "step": 200
     },
     {
       "epoch": 0.63,
-      "grad_norm": 0.830901563167572,
+      "grad_norm": 5.837784767150879,
       "learning_rate": 9.97897897897898e-06,
-      "loss": 4.5847,
+      "loss": 4.5895,
       "step": 210
     },
     {
       "epoch": 0.66,
-      "grad_norm": 0.7523330450057983,
+      "grad_norm": 1.5387754440307617,
       "learning_rate": 9.977977977977978e-06,
-      "loss": 4.5828,
+      "loss": 4.5902,
       "step": 220
     },
     {
       "epoch": 0.69,
-      "grad_norm": 1.183030366897583,
+      "grad_norm": 1.0322996377944946,
       "learning_rate": 9.976976976976977e-06,
-      "loss": 4.5824,
+      "loss": 4.5888,
       "step": 230
     },
     {
       "epoch": 0.72,
-      "grad_norm": 0.7077327966690063,
+      "grad_norm": 2.0640599727630615,
       "learning_rate": 9.975975975975977e-06,
-      "loss": 4.5788,
+      "loss": 4.5861,
       "step": 240
     },
     {
       "epoch": 0.75,
-      "grad_norm": 0.9782766103744507,
+      "grad_norm": 0.7117034792900085,
       "learning_rate": 9.974974974974976e-06,
-      "loss": 4.5778,
+      "loss": 4.586,
       "step": 250
     },
     {
       "epoch": 0.78,
-      "grad_norm": 0.9299131035804749,
+      "grad_norm": 1.6061571836471558,
       "learning_rate": 9.973973973973974e-06,
-      "loss": 4.5751,
+      "loss": 4.5852,
       "step": 260
     },
     {
       "epoch": 0.81,
-      "grad_norm": 1.1334943771362305,
+      "grad_norm": 1.6375962495803833,
       "learning_rate": 9.972972972972975e-06,
-      "loss": 4.5742,
+      "loss": 4.5858,
       "step": 270
     },
     {
       "epoch": 0.84,
-      "grad_norm": 0.9779238104820251,
+      "grad_norm": 1.495666265487671,
       "learning_rate": 9.971971971971973e-06,
-      "loss": 4.5708,
+      "loss": 4.5829,
       "step": 280
     },
     {
       "epoch": 0.87,
-      "grad_norm": 0.9815706014633179,
+      "grad_norm": 4.255990505218506,
       "learning_rate": 9.970970970970972e-06,
-      "loss": 4.5678,
+      "loss": 4.5823,
       "step": 290
     },
     {
       "epoch": 0.9,
-      "grad_norm": 1.2638413906097412,
+      "grad_norm": 0.8907470107078552,
       "learning_rate": 9.96996996996997e-06,
-      "loss": 4.5649,
+      "loss": 4.5792,
       "step": 300
     },
     {
       "epoch": 0.93,
-      "grad_norm": 0.9004982709884644,
+      "grad_norm": 0.8158383965492249,
       "learning_rate": 9.968968968968969e-06,
-      "loss": 4.5634,
+      "loss": 4.5792,
       "step": 310
     },
     {
       "epoch": 0.96,
-      "grad_norm": 1.2020589113235474,
+      "grad_norm": 0.8726849555969238,
       "learning_rate": 9.96796796796797e-06,
-      "loss": 4.5616,
+      "loss": 4.5798,
       "step": 320
     },
     {
       "epoch": 0.99,
-      "grad_norm": 1.0714811086654663,
+      "grad_norm": 1.0344123840332031,
       "learning_rate": 9.966966966966968e-06,
-      "loss": 4.5562,
+      "loss": 4.5769,
       "step": 330
     },
     {
       "epoch": 1.0,
-      "eval_accuracy": 0.1308,
-      "eval_loss": 4.554624080657959,
-      "eval_runtime": 7.7178,
-      "eval_samples_per_second": 1295.699,
-      "eval_steps_per_second": 5.183,
+      "eval_accuracy": 0.1152,
+      "eval_loss": 4.575807094573975,
+      "eval_runtime": 8.0851,
+      "eval_samples_per_second": 1236.849,
+      "eval_steps_per_second": 4.947,
       "step": 333
     },
     {
       "epoch": 1.02,
-      "grad_norm": 1.3365260362625122,
+      "grad_norm": 4.456799507141113,
       "learning_rate": 9.965965965965967e-06,
-      "loss": 4.5579,
+      "loss": 4.5785,
       "step": 340
     },
     {
       "epoch": 1.05,
-      "grad_norm": 1.6138296127319336,
+      "grad_norm": 3.2076544761657715,
       "learning_rate": 9.964964964964965e-06,
-      "loss": 4.5527,
+      "loss": 4.5747,
       "step": 350
     },
     {
       "epoch": 1.08,
-      "grad_norm": 1.505331039428711,
+      "grad_norm": 2.183274030685425,
       "learning_rate": 9.963963963963965e-06,
-      "loss": 4.549,
+      "loss": 4.5701,
       "step": 360
     },
     {
       "epoch": 1.11,
-      "grad_norm": 2.3446741104125977,
+      "grad_norm": 2.1832613945007324,
       "learning_rate": 9.962962962962964e-06,
-      "loss": 4.5469,
+      "loss": 4.5694,
       "step": 370
     },
     {
       "epoch": 1.14,
-      "grad_norm": 1.6814932823181152,
+      "grad_norm": 1.4050533771514893,
       "learning_rate": 9.961961961961963e-06,
-      "loss": 4.5433,
+      "loss": 4.5713,
       "step": 380
     },
     {
       "epoch": 1.17,
-      "grad_norm": 1.3763940334320068,
+      "grad_norm": 0.9986792802810669,
       "learning_rate": 9.960960960960961e-06,
-      "loss": 4.5388,
+      "loss": 4.5665,
       "step": 390
     },
     {
       "epoch": 1.2,
-      "grad_norm": 2.0994813442230225,
+      "grad_norm": 1.8422985076904297,
       "learning_rate": 9.95995995995996e-06,
-      "loss": 4.5355,
+      "loss": 4.5624,
       "step": 400
     },
     {
       "epoch": 1.23,
-      "grad_norm": 1.1807478666305542,
+      "grad_norm": 0.9850733280181885,
       "learning_rate": 9.95895895895896e-06,
-      "loss": 4.5323,
+      "loss": 4.5637,
       "step": 410
     },
     {
       "epoch": 1.26,
-      "grad_norm": 1.673993468284607,
+      "grad_norm": 1.059990644454956,
       "learning_rate": 9.957957957957959e-06,
-      "loss": 4.528,
+      "loss": 4.5624,
       "step": 420
     },
     {
       "epoch": 1.29,
-      "grad_norm": 1.316047191619873,
+      "grad_norm": 2.2468533515930176,
       "learning_rate": 9.956956956956957e-06,
-      "loss": 4.5249,
+      "loss": 4.5566,
       "step": 430
     },
     {
       "epoch": 1.32,
-      "grad_norm": 1.6580400466918945,
+      "grad_norm": 1.9284751415252686,
       "learning_rate": 9.955955955955958e-06,
-      "loss": 4.5198,
+      "loss": 4.5601,
       "step": 440
     },
     {
       "epoch": 1.35,
-      "grad_norm": 2.112293243408203,
+      "grad_norm": 1.556959867477417,
       "learning_rate": 9.954954954954956e-06,
-      "loss": 4.5122,
+      "loss": 4.5542,
       "step": 450
     },
     {
       "epoch": 1.38,
-      "grad_norm": 2.1887693405151367,
+      "grad_norm": 1.6963975429534912,
       "learning_rate": 9.953953953953955e-06,
-      "loss": 4.5131,
+      "loss": 4.5517,
       "step": 460
     },
     {
       "epoch": 1.41,
-      "grad_norm": 2.187678337097168,
+      "grad_norm": 1.2878575325012207,
       "learning_rate": 9.952952952952953e-06,
-      "loss": 4.5101,
+      "loss": 4.5518,
       "step": 470
     },
     {
       "epoch": 1.44,
-      "grad_norm": 1.6725214719772339,
+      "grad_norm": 1.4770427942276,
       "learning_rate": 9.951951951951952e-06,
-      "loss": 4.5085,
+      "loss": 4.5523,
       "step": 480
     },
     {
       "epoch": 1.47,
-      "grad_norm": 2.1836774349212646,
+      "grad_norm": 1.7538787126541138,
       "learning_rate": 9.950950950950952e-06,
-      "loss": 4.4954,
+      "loss": 4.5373,
       "step": 490
     },
     {
       "epoch": 1.5,
-      "grad_norm": 2.6754326820373535,
+      "grad_norm": 1.3768036365509033,
       "learning_rate": 9.949949949949951e-06,
-      "loss": 4.4926,
+      "loss": 4.5406,
       "step": 500
     },
     {
       "epoch": 1.53,
-      "grad_norm": 1.4669157266616821,
+      "grad_norm": 1.0017095804214478,
       "learning_rate": 9.94894894894895e-06,
-      "loss": 4.4971,
+      "loss": 4.5418,
       "step": 510
     },
     {
       "epoch": 1.56,
-      "grad_norm": 1.840080738067627,
+      "grad_norm": 1.4403549432754517,
       "learning_rate": 9.94794794794795e-06,
-      "loss": 4.4893,
+      "loss": 4.5426,
       "step": 520
     },
     {
       "epoch": 1.59,
-      "grad_norm": 1.796227216720581,
+      "grad_norm": 2.033220052719116,
       "learning_rate": 9.946946946946948e-06,
-      "loss": 4.4822,
+      "loss": 4.5346,
       "step": 530
     },
     {
       "epoch": 1.62,
-      "grad_norm": 3.022962808609009,
+      "grad_norm": 1.9720423221588135,
       "learning_rate": 9.945945945945947e-06,
-      "loss": 4.4818,
+      "loss": 4.5359,
       "step": 540
     },
     {
       "epoch": 1.65,
-      "grad_norm": 2.6112163066864014,
+      "grad_norm": 2.671513795852661,
       "learning_rate": 9.944944944944946e-06,
-      "loss": 4.4782,
+      "loss": 4.5359,
       "step": 550
     },
     {
       "epoch": 1.68,
-      "grad_norm": 1.706602692604065,
+      "grad_norm": 2.097236156463623,
       "learning_rate": 9.943943943943944e-06,
-      "loss": 4.4779,
+      "loss": 4.5308,
       "step": 560
     },
     {
       "epoch": 1.71,
-      "grad_norm": 2.2898976802825928,
+      "grad_norm": 2.562105894088745,
       "learning_rate": 9.942942942942944e-06,
-      "loss": 4.4731,
+      "loss": 4.5298,
       "step": 570
     },
     {
       "epoch": 1.74,
-      "grad_norm": 3.610862970352173,
+      "grad_norm": 2.040905714035034,
       "learning_rate": 9.941941941941943e-06,
-      "loss": 4.4746,
+      "loss": 4.5317,
       "step": 580
     },
     {
       "epoch": 1.77,
-      "grad_norm": 3.357612371444702,
+      "grad_norm": 2.713381767272949,
       "learning_rate": 9.940940940940942e-06,
-      "loss": 4.4656,
+      "loss": 4.523,
       "step": 590
     },
     {
       "epoch": 1.8,
-      "grad_norm": 3.0623044967651367,
+      "grad_norm": 1.4767674207687378,
       "learning_rate": 9.93993993993994e-06,
-      "loss": 4.4706,
+      "loss": 4.5262,
       "step": 600
     },
     {
       "epoch": 1.83,
-      "grad_norm": 2.096788167953491,
+      "grad_norm": 2.0011942386627197,
       "learning_rate": 9.93893893893894e-06,
-      "loss": 4.4609,
+      "loss": 4.5215,
       "step": 610
     },
     {
       "epoch": 1.86,
-      "grad_norm": 3.7277462482452393,
+      "grad_norm": 2.7474653720855713,
       "learning_rate": 9.937937937937939e-06,
-      "loss": 4.4521,
+      "loss": 4.5153,
       "step": 620
     },
     {
       "epoch": 1.89,
-      "grad_norm": 2.9841678142547607,
+      "grad_norm": 1.7028028964996338,
       "learning_rate": 9.936936936936938e-06,
-      "loss": 4.4596,
+      "loss": 4.5174,
       "step": 630
     },
     {
       "epoch": 1.92,
-      "grad_norm": 2.544673204421997,
+      "grad_norm": 2.0432889461517334,
       "learning_rate": 9.935935935935936e-06,
-      "loss": 4.4431,
+      "loss": 4.5085,
       "step": 640
     },
     {
       "epoch": 1.95,
-      "grad_norm": 1.9371894598007202,
+      "grad_norm": 2.1324310302734375,
       "learning_rate": 9.934934934934935e-06,
-      "loss": 4.4498,
+      "loss": 4.5187,
       "step": 650
     },
     {
       "epoch": 1.98,
-      "grad_norm": 2.023080348968506,
+      "grad_norm": 3.151634693145752,
       "learning_rate": 9.933933933933935e-06,
-      "loss": 4.4468,
+      "loss": 4.5036,
       "step": 660
     },
     {
       "epoch": 2.0,
-      "eval_accuracy": 0.145,
-      "eval_loss": 4.436723709106445,
-      "eval_runtime": 5.7282,
-      "eval_samples_per_second": 1745.757,
-      "eval_steps_per_second": 6.983,
+      "eval_accuracy": 0.1428,
+      "eval_loss": 4.500322341918945,
+      "eval_runtime": 5.1042,
+      "eval_samples_per_second": 1959.159,
+      "eval_steps_per_second": 7.837,
       "step": 666
     },
     {
       "epoch": 2.01,
-      "grad_norm": 2.1556339263916016,
+      "grad_norm": 2.1811535358428955,
       "learning_rate": 9.932932932932934e-06,
-      "loss": 4.4334,
+      "loss": 4.5033,
       "step": 670
     },
     {
       "epoch": 2.04,
-      "grad_norm": 1.6516185998916626,
+      "grad_norm": 1.3013170957565308,
       "learning_rate": 9.931931931931932e-06,
-      "loss": 4.4358,
+      "loss": 4.4985,
       "step": 680
     },
     {
       "epoch": 2.07,
-      "grad_norm": 3.211136817932129,
+      "grad_norm": 1.9128952026367188,
       "learning_rate": 9.930930930930933e-06,
-      "loss": 4.4382,
+      "loss": 4.5055,
       "step": 690
     },
     {
       "epoch": 2.1,
-      "grad_norm": 3.0844428539276123,
+      "grad_norm": 2.6640326976776123,
       "learning_rate": 9.929929929929931e-06,
-      "loss": 4.4279,
+      "loss": 4.4959,
       "step": 700
     },
     {
       "epoch": 2.13,
-      "grad_norm": 2.327069044113159,
+      "grad_norm": 1.6650125980377197,
       "learning_rate": 9.92892892892893e-06,
-      "loss": 4.4323,
+      "loss": 4.4939,
       "step": 710
     },
     {
       "epoch": 2.16,
-      "grad_norm": 4.18211555480957,
+      "grad_norm": 1.438989520072937,
       "learning_rate": 9.927927927927928e-06,
-      "loss": 4.4219,
+      "loss": 4.4871,
       "step": 720
     },
     {
       "epoch": 2.19,
-      "grad_norm": 3.7693631649017334,
+      "grad_norm": 3.0065155029296875,
       "learning_rate": 9.926926926926927e-06,
-      "loss": 4.4191,
+      "loss": 4.4881,
       "step": 730
     },
     {
       "epoch": 2.22,
-      "grad_norm": 2.327669858932495,
+      "grad_norm": 3.245229482650757,
       "learning_rate": 9.925925925925927e-06,
-      "loss": 4.4217,
+      "loss": 4.4842,
       "step": 740
     },
     {
       "epoch": 2.25,
-      "grad_norm": 4.02529239654541,
+      "grad_norm": 2.8599746227264404,
       "learning_rate": 9.924924924924926e-06,
-      "loss": 4.4201,
+      "loss": 4.4797,
       "step": 750
     },
     {
       "epoch": 2.28,
-      "grad_norm": 3.6580870151519775,
+      "grad_norm": 2.936551809310913,
       "learning_rate": 9.923923923923925e-06,
-      "loss": 4.4193,
+      "loss": 4.4768,
       "step": 760
     },
     {
       "epoch": 2.31,
-      "grad_norm": 3.334617853164673,
+      "grad_norm": 2.607294797897339,
       "learning_rate": 9.922922922922925e-06,
-      "loss": 4.4005,
+      "loss": 4.4736,
       "step": 770
     },
     {
       "epoch": 2.34,
-      "grad_norm": 2.5071446895599365,
+      "grad_norm": 3.685732841491699,
       "learning_rate": 9.921921921921923e-06,
-      "loss": 4.398,
+      "loss": 4.4673,
       "step": 780
     },
     {
       "epoch": 2.37,
-      "grad_norm": 3.2208921909332275,
+      "grad_norm": 3.6327199935913086,
       "learning_rate": 9.920920920920922e-06,
-      "loss": 4.3949,
+      "loss": 4.4553,
       "step": 790
     },
     {
       "epoch": 2.4,
-      "grad_norm": 4.86293888092041,
+      "grad_norm": 2.2524008750915527,
       "learning_rate": 9.91991991991992e-06,
-      "loss": 4.3992,
+      "loss": 4.4717,
       "step": 800
     },
     {
       "epoch": 2.43,
-      "grad_norm": 3.8809378147125244,
+      "grad_norm": 3.0974678993225098,
       "learning_rate": 9.91891891891892e-06,
-      "loss": 4.4176,
+      "loss": 4.4702,
       "step": 810
     },
     {
       "epoch": 2.46,
-      "grad_norm": 2.973187208175659,
+      "grad_norm": 4.572568893432617,
       "learning_rate": 9.917917917917918e-06,
-      "loss": 4.4091,
+      "loss": 4.455,
       "step": 820
     },
     {
       "epoch": 2.49,
-      "grad_norm": 2.8818836212158203,
+      "grad_norm": 3.0259478092193604,
       "learning_rate": 9.916916916916918e-06,
-      "loss": 4.4146,
+      "loss": 4.4555,
       "step": 830
     },
     {
       "epoch": 2.52,
-      "grad_norm": 2.97062087059021,
+      "grad_norm": 4.0475945472717285,
       "learning_rate": 9.915915915915917e-06,
-      "loss": 4.3909,
+      "loss": 4.4529,
       "step": 840
     },
     {
       "epoch": 2.55,
-      "grad_norm": 3.1629340648651123,
+      "grad_norm": 4.355692386627197,
       "learning_rate": 9.914914914914915e-06,
-      "loss": 4.3909,
+      "loss": 4.4537,
       "step": 850
     },
     {
       "epoch": 2.58,
-      "grad_norm": 2.2132112979888916,
+      "grad_norm": 2.8458850383758545,
       "learning_rate": 9.913913913913916e-06,
-      "loss": 4.3864,
+      "loss": 4.4456,
       "step": 860
     },
     {
       "epoch": 2.61,
-      "grad_norm": 3.493638277053833,
+      "grad_norm": 3.0762367248535156,
       "learning_rate": 9.912912912912914e-06,
-      "loss": 4.3796,
+      "loss": 4.4339,
       "step": 870
     },
     {
       "epoch": 2.64,
-      "grad_norm": 3.384003162384033,
+      "grad_norm": 2.5200531482696533,
       "learning_rate": 9.911911911911913e-06,
-      "loss": 4.389,
+      "loss": 4.4403,
       "step": 880
     },
     {
       "epoch": 2.67,
-      "grad_norm": 2.344482183456421,
+      "grad_norm": 4.2043938636779785,
       "learning_rate": 9.910910910910911e-06,
-      "loss": 4.3743,
+      "loss": 4.4274,
       "step": 890
     },
     {
       "epoch": 2.7,
-      "grad_norm": 5.703094005584717,
+      "grad_norm": 4.316439151763916,
       "learning_rate": 9.90990990990991e-06,
-      "loss": 4.3715,
+      "loss": 4.4226,
       "step": 900
     },
     {
       "epoch": 2.73,
-      "grad_norm": 3.333181619644165,
+      "grad_norm": 1.7869346141815186,
       "learning_rate": 9.90890890890891e-06,
-      "loss": 4.384,
+      "loss": 4.426,
       "step": 910
     },
     {
       "epoch": 2.76,
-      "grad_norm": 2.483485221862793,
+      "grad_norm": 2.6213884353637695,
       "learning_rate": 9.907907907907909e-06,
-      "loss": 4.3677,
+      "loss": 4.4179,
       "step": 920
     },
     {
       "epoch": 2.79,
-      "grad_norm": 3.4019031524658203,
+      "grad_norm": 3.125615119934082,
       "learning_rate": 9.906906906906907e-06,
-      "loss": 4.3656,
+      "loss": 4.4282,
       "step": 930
     },
     {
       "epoch": 2.82,
-      "grad_norm": 2.6729891300201416,
+      "grad_norm": 4.113916873931885,
       "learning_rate": 9.905905905905908e-06,
-      "loss": 4.3588,
+      "loss": 4.3989,
       "step": 940
     },
     {
       "epoch": 2.85,
-      "grad_norm": 2.548947811126709,
+      "grad_norm": 4.200078010559082,
       "learning_rate": 9.904904904904906e-06,
-      "loss": 4.3602,
+      "loss": 4.4103,
       "step": 950
     },
     {
       "epoch": 2.88,
-      "grad_norm": 3.6089022159576416,
+      "grad_norm": 3.0239779949188232,
       "learning_rate": 9.903903903903905e-06,
-      "loss": 4.3533,
+      "loss": 4.4051,
       "step": 960
     },
     {
       "epoch": 2.91,
-      "grad_norm": 2.5302388668060303,
+      "grad_norm": 3.8236331939697266,
       "learning_rate": 9.902902902902903e-06,
-      "loss": 4.357,
+      "loss": 4.4038,
       "step": 970
     },
     {
       "epoch": 2.94,
-      "grad_norm": 2.4023447036743164,
+      "grad_norm": 5.516148567199707,
       "learning_rate": 9.901901901901902e-06,
-      "loss": 4.355,
+      "loss": 4.4066,
       "step": 980
     },
     {
       "epoch": 2.97,
-      "grad_norm": 4.4992804527282715,
+      "grad_norm": 4.682029724121094,
       "learning_rate": 9.900900900900902e-06,
-      "loss": 4.3538,
+      "loss": 4.4013,
       "step": 990
     },
     {
       "epoch": 3.0,
-      "eval_accuracy": 0.1859,
-      "eval_loss": 4.3131256103515625,
-      "eval_runtime": 5.3799,
-      "eval_samples_per_second": 1858.785,
-      "eval_steps_per_second": 7.435,
+      "eval_accuracy": 0.1533,
+      "eval_loss": 4.369607925415039,
+      "eval_runtime": 5.5301,
+      "eval_samples_per_second": 1808.293,
+      "eval_steps_per_second": 7.233,
       "step": 999
     },
     {
       "epoch": 3.0,
-      "grad_norm": 2.955087661743164,
+      "grad_norm": 4.833372116088867,
       "learning_rate": 9.899899899899901e-06,
-      "loss": 4.3421,
+      "loss": 4.3998,
       "step": 1000
     },
     {
       "epoch": 3.03,
-      "grad_norm": 3.368352174758911,
+      "grad_norm": 4.458285331726074,
       "learning_rate": 9.8988988988989e-06,
-      "loss": 4.3244,
+      "loss": 4.379,
       "step": 1010
     },
     {
       "epoch": 3.06,
-      "grad_norm": 3.0585217475891113,
+      "grad_norm": 3.7412867546081543,
       "learning_rate": 9.8978978978979e-06,
-      "loss": 4.3388,
+      "loss": 4.393,
       "step": 1020
     },
     {
       "epoch": 3.09,
-      "grad_norm": 5.0030903816223145,
+      "grad_norm": 3.271860361099243,
       "learning_rate": 9.896896896896898e-06,
-      "loss": 4.33,
+      "loss": 4.3715,
       "step": 1030
     },
     {
       "epoch": 3.12,
-      "grad_norm": 2.289600133895874,
+      "grad_norm": 3.362905979156494,
       "learning_rate": 9.895895895895895e-06,
-      "loss": 4.3329,
+      "loss": 4.3807,
       "step": 1040
     },
     {
       "epoch": 3.15,
-      "grad_norm": 3.614248037338257,
+      "grad_norm": 6.021758556365967,
       "learning_rate": 9.894894894894896e-06,
-      "loss": 4.3048,
+      "loss": 4.371,
       "step": 1050
     },
     {
       "epoch": 3.18,
-      "grad_norm": 3.0962393283843994,
+      "grad_norm": 3.420862913131714,
       "learning_rate": 9.893893893893894e-06,
-      "loss": 4.345,
+      "loss": 4.3805,
       "step": 1060
     },
     {
       "epoch": 3.21,
-      "grad_norm": 2.918386220932007,
+      "grad_norm": 2.756922483444214,
       "learning_rate": 9.892892892892893e-06,
-      "loss": 4.3301,
+      "loss": 4.3563,
       "step": 1070
     },
     {
       "epoch": 3.24,
-      "grad_norm": 4.14999532699585,
+      "grad_norm": 3.2892162799835205,
       "learning_rate": 9.891891891891893e-06,
-      "loss": 4.2985,
+      "loss": 4.3464,
       "step": 1080
     },
     {
       "epoch": 3.27,
-      "grad_norm": 4.211765289306641,
+      "grad_norm": 5.514255523681641,
       "learning_rate": 9.890890890890892e-06,
-      "loss": 4.3285,
+      "loss": 4.3732,
       "step": 1090
     },
     {
       "epoch": 3.3,
-      "grad_norm": 3.051703929901123,
+      "grad_norm": 5.192841053009033,
       "learning_rate": 9.88988988988989e-06,
-      "loss": 4.3233,
+      "loss": 4.3624,
       "step": 1100
     },
     {
       "epoch": 3.33,
-      "grad_norm": 3.3170785903930664,
+      "grad_norm": 3.1024744510650635,
       "learning_rate": 9.88888888888889e-06,
-      "loss": 4.3141,
+      "loss": 4.3649,
       "step": 1110
     },
     {
       "epoch": 3.36,
-      "grad_norm": 3.618259906768799,
+      "grad_norm": 4.612704277038574,
       "learning_rate": 9.88788788788789e-06,
-      "loss": 4.3079,
+      "loss": 4.3438,
       "step": 1120
     },
     {
       "epoch": 3.39,
-      "grad_norm": 4.114195346832275,
+      "grad_norm": 3.3861262798309326,
       "learning_rate": 9.886886886886888e-06,
-      "loss": 4.2965,
+      "loss": 4.3266,
       "step": 1130
     },
     {
       "epoch": 3.42,
-      "grad_norm": 3.5468411445617676,
+      "grad_norm": 6.676590442657471,
       "learning_rate": 9.885885885885886e-06,
-      "loss": 4.3035,
+      "loss": 4.3479,
       "step": 1140
     },
     {
       "epoch": 3.45,
-      "grad_norm": 3.6822125911712646,
+      "grad_norm": 4.572504043579102,
       "learning_rate": 9.884884884884885e-06,
-      "loss": 4.3085,
+      "loss": 4.3387,
       "step": 1150
     },
     {
       "epoch": 3.48,
-      "grad_norm": 2.403459310531616,
+      "grad_norm": 3.0916030406951904,
       "learning_rate": 9.883883883883885e-06,
-      "loss": 4.2889,
+      "loss": 4.3275,
       "step": 1160
     },
     {
       "epoch": 3.51,
-      "grad_norm": 3.4859273433685303,
+      "grad_norm": 4.608184814453125,
       "learning_rate": 9.882882882882884e-06,
-      "loss": 4.2899,
+      "loss": 4.3309,
       "step": 1170
     },
     {
       "epoch": 3.54,
-      "grad_norm": 4.09109354019165,
+      "grad_norm": 5.629223346710205,
       "learning_rate": 9.881881881881882e-06,
-      "loss": 4.2767,
+      "loss": 4.3196,
       "step": 1180
     },
     {
       "epoch": 3.57,
-      "grad_norm": 3.483778953552246,
+      "grad_norm": 5.143381595611572,
       "learning_rate": 9.880880880880883e-06,
-      "loss": 4.2804,
+      "loss": 4.3266,
       "step": 1190
     },
     {
       "epoch": 3.6,
-      "grad_norm": 2.7234344482421875,
+      "grad_norm": 3.342283010482788,
       "learning_rate": 9.879879879879881e-06,
-      "loss": 4.2677,
+      "loss": 4.3116,
       "step": 1200
     },
     {
       "epoch": 3.63,
-      "grad_norm": 5.75734806060791,
+      "grad_norm": 3.6295416355133057,
       "learning_rate": 9.87887887887888e-06,
-      "loss": 4.2548,
+      "loss": 4.3132,
       "step": 1210
     },
     {
       "epoch": 3.66,
-      "grad_norm": 3.3725943565368652,
+      "grad_norm": 4.233205795288086,
       "learning_rate": 9.877877877877879e-06,
-      "loss": 4.267,
+      "loss": 4.2901,
       "step": 1220
     },
     {
       "epoch": 3.69,
-      "grad_norm": 3.31961989402771,
+      "grad_norm": 7.189581871032715,
       "learning_rate": 9.876876876876877e-06,
-      "loss": 4.2556,
+      "loss": 4.3026,
       "step": 1230
     },
     {
       "epoch": 3.72,
-      "grad_norm": 3.086261510848999,
+      "grad_norm": 3.578017473220825,
       "learning_rate": 9.875875875875877e-06,
-      "loss": 4.2608,
+      "loss": 4.3052,
       "step": 1240
     },
     {
       "epoch": 3.75,
-      "grad_norm": 3.949026346206665,
+      "grad_norm": 5.856477737426758,
       "learning_rate": 9.874874874874876e-06,
-      "loss": 4.2419,
+      "loss": 4.2792,
       "step": 1250
     },
     {
       "epoch": 3.78,
-      "grad_norm": 3.359564781188965,
+      "grad_norm": 3.5932228565216064,
       "learning_rate": 9.873873873873875e-06,
-      "loss": 4.2464,
+      "loss": 4.2935,
       "step": 1260
     },
     {
       "epoch": 3.81,
-      "grad_norm": 2.687539577484131,
+      "grad_norm": 4.824099540710449,
       "learning_rate": 9.872872872872873e-06,
-      "loss": 4.2222,
+      "loss": 4.2701,
       "step": 1270
     },
     {
       "epoch": 3.84,
-      "grad_norm": 4.35097074508667,
+      "grad_norm": 4.46605110168457,
       "learning_rate": 9.871871871871873e-06,
-      "loss": 4.2371,
+      "loss": 4.2877,
       "step": 1280
     },
     {
       "epoch": 3.87,
-      "grad_norm": 2.964953899383545,
+      "grad_norm": 4.273735523223877,
       "learning_rate": 9.87087087087087e-06,
-      "loss": 4.2402,
+      "loss": 4.2903,
       "step": 1290
     },
     {
       "epoch": 3.9,
-      "grad_norm": 3.6431925296783447,
+      "grad_norm": 3.6705358028411865,
       "learning_rate": 9.86986986986987e-06,
-      "loss": 4.2114,
+      "loss": 4.2536,
       "step": 1300
     },
     {
       "epoch": 3.93,
-      "grad_norm": 3.4964230060577393,
+      "grad_norm": 3.710939645767212,
       "learning_rate": 9.86886886886887e-06,
-      "loss": 4.2379,
+      "loss": 4.2763,
       "step": 1310
     },
     {
       "epoch": 3.96,
-      "grad_norm": 4.928268909454346,
+      "grad_norm": 4.33703088760376,
       "learning_rate": 9.867867867867868e-06,
-      "loss": 4.2115,
+      "loss": 4.2441,
       "step": 1320
     },
     {
       "epoch": 3.99,
-      "grad_norm": 3.5851588249206543,
+      "grad_norm": 3.7469747066497803,
       "learning_rate": 9.866866866866868e-06,
-      "loss": 4.2283,
+      "loss": 4.2652,
       "step": 1330
     },
     {
       "epoch": 4.0,
-      "eval_accuracy": 0.2269,
-      "eval_loss": 4.139839172363281,
-      "eval_runtime": 5.5537,
-      "eval_samples_per_second": 1800.606,
-      "eval_steps_per_second": 7.202,
+      "eval_accuracy": 0.1868,
+      "eval_loss": 4.205499649047852,
+      "eval_runtime": 5.4796,
+      "eval_samples_per_second": 1824.964,
+      "eval_steps_per_second": 7.3,
       "step": 1332
     },
     {
       "epoch": 4.02,
-      "grad_norm": 5.1920552253723145,
+      "grad_norm": 3.984437942504883,
       "learning_rate": 9.865865865865867e-06,
-      "loss": 4.23,
+      "loss": 4.2682,
       "step": 1340
     },
     {
       "epoch": 4.05,
-      "grad_norm": 3.809221029281616,
+      "grad_norm": 4.5088653564453125,
       "learning_rate": 9.864864864864865e-06,
-      "loss": 4.1879,
+      "loss": 4.2355,
       "step": 1350
     },
     {
       "epoch": 4.08,
-      "grad_norm": 3.865419387817383,
+      "grad_norm": 6.386178016662598,
       "learning_rate": 9.863863863863866e-06,
-      "loss": 4.2112,
+      "loss": 4.2601,
       "step": 1360
     },
     {
       "epoch": 4.11,
-      "grad_norm": 3.5560433864593506,
+      "grad_norm": 5.115908622741699,
       "learning_rate": 9.862862862862864e-06,
-      "loss": 4.1882,
+      "loss": 4.2354,
       "step": 1370
     },
     {
       "epoch": 4.14,
-      "grad_norm": 5.0352325439453125,
+      "grad_norm": 6.433563709259033,
       "learning_rate": 9.861861861861863e-06,
-      "loss": 4.2121,
+      "loss": 4.2527,
       "step": 1380
     },
     {
       "epoch": 4.17,
-      "grad_norm": 3.990941286087036,
+      "grad_norm": 4.796797275543213,
       "learning_rate": 9.860860860860861e-06,
-      "loss": 4.1719,
+      "loss": 4.214,
       "step": 1390
     },
     {
       "epoch": 4.2,
-      "grad_norm": 4.546986103057861,
+      "grad_norm": 4.611423969268799,
       "learning_rate": 9.85985985985986e-06,
-      "loss": 4.1838,
+      "loss": 4.2351,
       "step": 1400
     },
     {
       "epoch": 4.23,
-      "grad_norm": 3.82480788230896,
+      "grad_norm": 6.92868709564209,
       "learning_rate": 9.85885885885886e-06,
-      "loss": 4.1793,
+      "loss": 4.2283,
       "step": 1410
     },
     {
       "epoch": 4.26,
-      "grad_norm": 4.344655513763428,
+      "grad_norm": 4.913412094116211,
       "learning_rate": 9.857857857857859e-06,
-      "loss": 4.152,
+      "loss": 4.1996,
       "step": 1420
     },
     {
       "epoch": 4.29,
-      "grad_norm": 5.877710342407227,
+      "grad_norm": 3.773505210876465,
       "learning_rate": 9.856856856856857e-06,
-      "loss": 4.1602,
+      "loss": 4.2084,
       "step": 1430
     },
     {
       "epoch": 4.32,
-      "grad_norm": 4.2201247215271,
+      "grad_norm": 5.922270774841309,
       "learning_rate": 9.855855855855858e-06,
-      "loss": 4.1595,
+      "loss": 4.2045,
       "step": 1440
     },
     {
       "epoch": 4.35,
-      "grad_norm": 4.324289321899414,
+      "grad_norm": 7.325625419616699,
       "learning_rate": 9.854854854854856e-06,
-      "loss": 4.1479,
+      "loss": 4.1915,
       "step": 1450
     },
     {
       "epoch": 4.38,
-      "grad_norm": 5.367230415344238,
+      "grad_norm": 3.512465000152588,
       "learning_rate": 9.853853853853855e-06,
-      "loss": 4.1368,
+      "loss": 4.1823,
       "step": 1460
     },
     {
       "epoch": 4.41,
-      "grad_norm": 4.7191996574401855,
+      "grad_norm": 2.834404230117798,
       "learning_rate": 9.852852852852854e-06,
-      "loss": 4.1532,
+      "loss": 4.2058,
       "step": 1470
     },
     {
       "epoch": 4.44,
-      "grad_norm": 3.5991883277893066,
+      "grad_norm": 3.1526029109954834,
       "learning_rate": 9.851851851851852e-06,
-      "loss": 4.1324,
+      "loss": 4.174,
       "step": 1480
     },
     {
       "epoch": 4.47,
-      "grad_norm": 3.0082712173461914,
+      "grad_norm": 4.999171733856201,
       "learning_rate": 9.85085085085085e-06,
-      "loss": 4.13,
+      "loss": 4.1616,
       "step": 1490
     },
     {
       "epoch": 4.5,
-      "grad_norm": 3.4249894618988037,
+      "grad_norm": 4.203711032867432,
       "learning_rate": 9.849849849849851e-06,
-      "loss": 4.1311,
+      "loss": 4.1798,
       "step": 1500
     },
     {
       "epoch": 4.53,
-      "grad_norm": 3.4547677040100098,
+      "grad_norm": 5.990175724029541,
       "learning_rate": 9.84884884884885e-06,
-      "loss": 4.1093,
+      "loss": 4.1552,
       "step": 1510
     },
     {
       "epoch": 4.56,
-      "grad_norm": 3.142854928970337,
+      "grad_norm": 7.77642822265625,
       "learning_rate": 9.847847847847848e-06,
-      "loss": 4.1074,
+      "loss": 4.1531,
       "step": 1520
     },
     {
       "epoch": 4.59,
-      "grad_norm": 4.7387237548828125,
+      "grad_norm": 4.927059650421143,
       "learning_rate": 9.846846846846849e-06,
-      "loss": 4.1289,
+      "loss": 4.1746,
       "step": 1530
     },
     {
       "epoch": 4.62,
-      "grad_norm": 3.2236664295196533,
+      "grad_norm": 4.931154727935791,
       "learning_rate": 9.845845845845845e-06,
-      "loss": 4.111,
+      "loss": 4.1723,
       "step": 1540
     },
     {
       "epoch": 4.65,
-      "grad_norm": 3.8368449211120605,
+      "grad_norm": 5.307770729064941,
       "learning_rate": 9.844844844844846e-06,
-      "loss": 4.0776,
+      "loss": 4.1192,
       "step": 1550
     },
     {
       "epoch": 4.68,
-      "grad_norm": 4.223494529724121,
+      "grad_norm": 4.554162502288818,
       "learning_rate": 9.843843843843844e-06,
-      "loss": 4.0853,
+      "loss": 4.1375,
       "step": 1560
     },
     {
       "epoch": 4.71,
-      "grad_norm": 3.6497983932495117,
+      "grad_norm": 4.879663467407227,
       "learning_rate": 9.842842842842843e-06,
-      "loss": 4.0775,
+      "loss": 4.1266,
       "step": 1570
     },
     {
       "epoch": 4.74,
-      "grad_norm": 3.4451074600219727,
+      "grad_norm": 4.82340669631958,
       "learning_rate": 9.841841841841843e-06,
-      "loss": 4.0663,
+      "loss": 4.1086,
       "step": 1580
     },
     {
       "epoch": 4.77,
-      "grad_norm": 4.336752414703369,
+      "grad_norm": 4.354609966278076,
       "learning_rate": 9.840840840840842e-06,
-      "loss": 4.0753,
+      "loss": 4.1287,
       "step": 1590
     },
     {
       "epoch": 4.8,
-      "grad_norm": 2.766197681427002,
+      "grad_norm": 8.582011222839355,
       "learning_rate": 9.83983983983984e-06,
-      "loss": 4.0934,
+      "loss": 4.1353,
       "step": 1600
     },
     {
       "epoch": 4.83,
-      "grad_norm": 5.1743950843811035,
+      "grad_norm": 5.9257988929748535,
       "learning_rate": 9.83883883883884e-06,
-      "loss": 4.0596,
+      "loss": 4.1153,
       "step": 1610
     },
     {
       "epoch": 4.86,
-      "grad_norm": 2.8578991889953613,
+      "grad_norm": 3.6178696155548096,
       "learning_rate": 9.83783783783784e-06,
-      "loss": 4.0475,
+      "loss": 4.1074,
       "step": 1620
     },
     {
       "epoch": 4.89,
-      "grad_norm": 4.761795997619629,
+      "grad_norm": 4.037405490875244,
       "learning_rate": 9.836836836836838e-06,
-      "loss": 4.0403,
+      "loss": 4.0967,
       "step": 1630
     },
     {
       "epoch": 4.92,
-      "grad_norm": 4.857820510864258,
+      "grad_norm": 9.063667297363281,
       "learning_rate": 9.835835835835836e-06,
-      "loss": 4.0496,
+      "loss": 4.095,
       "step": 1640
     },
     {
       "epoch": 4.95,
-      "grad_norm": 4.812182903289795,
+      "grad_norm": 4.213994026184082,
       "learning_rate": 9.834834834834835e-06,
-      "loss": 4.0464,
+      "loss": 4.0963,
       "step": 1650
     },
     {
       "epoch": 4.98,
-      "grad_norm": 5.4690446853637695,
+      "grad_norm": 4.659358501434326,
       "learning_rate": 9.833833833833835e-06,
-      "loss": 4.0124,
+      "loss": 4.0635,
       "step": 1660
     },
     {
       "epoch": 5.0,
-      "eval_accuracy": 0.2647,
-      "eval_loss": 3.907414436340332,
-      "eval_runtime": 5.3388,
-      "eval_samples_per_second": 1873.072,
-      "eval_steps_per_second": 7.492,
+      "eval_accuracy": 0.2296,
+      "eval_loss": 3.9880359172821045,
+      "eval_runtime": 5.2789,
+      "eval_samples_per_second": 1894.338,
+      "eval_steps_per_second": 7.577,
       "step": 1665
     },
     {
       "epoch": 5.02,
-      "grad_norm": 3.0309042930603027,
+      "grad_norm": 6.481389045715332,
       "learning_rate": 9.832832832832834e-06,
-      "loss": 4.0366,
+      "loss": 4.0769,
       "step": 1670
     },
     {
       "epoch": 5.05,
-      "grad_norm": 7.121453285217285,
+      "grad_norm": 4.547131538391113,
       "learning_rate": 9.831831831831833e-06,
-      "loss": 4.0222,
+      "loss": 4.0762,
       "step": 1680
     },
     {
       "epoch": 5.08,
-      "grad_norm": 5.54395055770874,
+      "grad_norm": 5.674590110778809,
       "learning_rate": 9.830830830830833e-06,
-      "loss": 4.0024,
+      "loss": 4.0466,
       "step": 1690
     },
     {
       "epoch": 5.11,
-      "grad_norm": 4.83833646774292,
+      "grad_norm": 7.405400276184082,
       "learning_rate": 9.829829829829831e-06,
-      "loss": 3.9965,
+      "loss": 4.0506,
       "step": 1700
     },
     {
       "epoch": 5.14,
-      "grad_norm": 5.251867294311523,
+      "grad_norm": 4.916649341583252,
       "learning_rate": 9.82882882882883e-06,
-      "loss": 4.0037,
+      "loss": 4.0631,
       "step": 1710
     },
     {
       "epoch": 5.17,
-      "grad_norm": 4.13340950012207,
+      "grad_norm": 4.581605434417725,
       "learning_rate": 9.827827827827829e-06,
-      "loss": 3.9984,
+      "loss": 4.0495,
       "step": 1720
     },
     {
       "epoch": 5.2,
-      "grad_norm": 3.9963762760162354,
+      "grad_norm": 6.1533098220825195,
       "learning_rate": 9.826826826826827e-06,
-      "loss": 3.9845,
+      "loss": 4.0401,
       "step": 1730
     },
     {
       "epoch": 5.23,
-      "grad_norm": 3.7753093242645264,
+      "grad_norm": 4.6586503982543945,
       "learning_rate": 9.825825825825826e-06,
-      "loss": 3.9556,
+      "loss": 4.0156,
       "step": 1740
     },
     {
       "epoch": 5.26,
-      "grad_norm": 4.607434272766113,
+      "grad_norm": 4.978870391845703,
       "learning_rate": 9.824824824824826e-06,
-      "loss": 3.9636,
+      "loss": 4.0354,
       "step": 1750
     },
     {
       "epoch": 5.29,
-      "grad_norm": 5.353358268737793,
+      "grad_norm": 5.291574478149414,
       "learning_rate": 9.823823823823825e-06,
-      "loss": 3.9602,
+      "loss": 4.025,
       "step": 1760
     },
     {
       "epoch": 5.32,
-      "grad_norm": 5.621310710906982,
+      "grad_norm": 4.125316143035889,
       "learning_rate": 9.822822822822823e-06,
-      "loss": 3.9856,
+      "loss": 4.0537,
       "step": 1770
     },
     {
       "epoch": 5.35,
-      "grad_norm": 3.8203694820404053,
+      "grad_norm": 4.958192348480225,
       "learning_rate": 9.821821821821824e-06,
-      "loss": 3.958,
+      "loss": 4.0213,
       "step": 1780
     },
     {
       "epoch": 5.38,
-      "grad_norm": 4.317108154296875,
+      "grad_norm": 4.504006862640381,
       "learning_rate": 9.82082082082082e-06,
-      "loss": 3.9508,
+      "loss": 4.0063,
       "step": 1790
     },
     {
       "epoch": 5.41,
-      "grad_norm": 3.898102283477783,
+      "grad_norm": 5.250901699066162,
       "learning_rate": 9.81981981981982e-06,
-      "loss": 3.9481,
+      "loss": 4.0142,
       "step": 1800
     },
     {
       "epoch": 5.44,
-      "grad_norm": 4.0414838790893555,
+      "grad_norm": 6.2618021965026855,
       "learning_rate": 9.81881881881882e-06,
-      "loss": 3.9421,
+      "loss": 4.0137,
       "step": 1810
     },
     {
       "epoch": 5.47,
-      "grad_norm": 4.4286322593688965,
+      "grad_norm": 6.268489360809326,
       "learning_rate": 9.817817817817818e-06,
-      "loss": 3.9614,
+      "loss": 4.016,
       "step": 1820
     },
     {
       "epoch": 5.5,
-      "grad_norm": 3.7750370502471924,
+      "grad_norm": 7.072010040283203,
       "learning_rate": 9.816816816816818e-06,
-      "loss": 3.8923,
+      "loss": 3.959,
       "step": 1830
     },
     {
       "epoch": 5.53,
-      "grad_norm": 4.194284439086914,
+      "grad_norm": 5.11269474029541,
       "learning_rate": 9.815815815815817e-06,
-      "loss": 3.9033,
+      "loss": 3.9835,
       "step": 1840
     },
     {
       "epoch": 5.56,
-      "grad_norm": 3.5534565448760986,
+      "grad_norm": 4.6049323081970215,
       "learning_rate": 9.814814814814815e-06,
-      "loss": 3.9173,
+      "loss": 3.992,
       "step": 1850
     },
     {
       "epoch": 5.59,
-      "grad_norm": 3.0711829662323,
+      "grad_norm": 7.966219425201416,
       "learning_rate": 9.813813813813816e-06,
-      "loss": 3.8599,
+      "loss": 3.9398,
       "step": 1860
     },
     {
       "epoch": 5.62,
-      "grad_norm": 5.571861743927002,
+      "grad_norm": 5.529684543609619,
       "learning_rate": 9.812812812812814e-06,
-      "loss": 3.8891,
+      "loss": 3.9622,
       "step": 1870
     },
     {
       "epoch": 5.65,
-      "grad_norm": 3.6509060859680176,
+      "grad_norm": 7.687751770019531,
       "learning_rate": 9.811811811811813e-06,
-      "loss": 3.9027,
+      "loss": 3.9736,
       "step": 1880
     },
     {
       "epoch": 5.68,
-      "grad_norm": 3.8952229022979736,
+      "grad_norm": 6.062169551849365,
       "learning_rate": 9.810810810810811e-06,
-      "loss": 3.8382,
+      "loss": 3.9048,
       "step": 1890
     },
     {
       "epoch": 5.71,
-      "grad_norm": 4.906396389007568,
+      "grad_norm": 6.021452903747559,
       "learning_rate": 9.80980980980981e-06,
-      "loss": 3.8726,
+      "loss": 3.9482,
       "step": 1900
     },
     {
       "epoch": 5.74,
-      "grad_norm": 3.5679867267608643,
+      "grad_norm": 4.693448543548584,
       "learning_rate": 9.80880880880881e-06,
-      "loss": 3.8333,
+      "loss": 3.9115,
       "step": 1910
     },
     {
       "epoch": 5.77,
-      "grad_norm": 6.311254501342773,
+      "grad_norm": 5.518393039703369,
       "learning_rate": 9.807807807807809e-06,
-      "loss": 3.8627,
+      "loss": 3.9316,
       "step": 1920
     },
     {
       "epoch": 5.8,
-      "grad_norm": 4.690033435821533,
+      "grad_norm": 4.771263122558594,
       "learning_rate": 9.806806806806808e-06,
-      "loss": 3.8667,
+      "loss": 3.9427,
       "step": 1930
     },
     {
       "epoch": 5.83,
-      "grad_norm": 5.286990642547607,
+      "grad_norm": 5.710914611816406,
       "learning_rate": 9.805805805805808e-06,
-      "loss": 3.8439,
+      "loss": 3.9275,
       "step": 1940
     },
     {
       "epoch": 5.86,
-      "grad_norm": 5.9812188148498535,
+      "grad_norm": 4.192609786987305,
       "learning_rate": 9.804804804804806e-06,
-      "loss": 3.8435,
+      "loss": 3.9259,
       "step": 1950
     },
     {
       "epoch": 5.89,
-      "grad_norm": 5.215433597564697,
+      "grad_norm": 8.799088478088379,
       "learning_rate": 9.803803803803803e-06,
-      "loss": 3.8439,
+      "loss": 3.9231,
       "step": 1960
     },
     {
       "epoch": 5.92,
-      "grad_norm": 9.2427396774292,
+      "grad_norm": 5.46636438369751,
       "learning_rate": 9.802802802802804e-06,
-      "loss": 3.8205,
+      "loss": 3.9079,
       "step": 1970
     },
     {
       "epoch": 5.95,
-      "grad_norm": 3.3974647521972656,
+      "grad_norm": 4.75711727142334,
       "learning_rate": 9.801801801801802e-06,
-      "loss": 3.8295,
+      "loss": 3.9132,
       "step": 1980
     },
     {
       "epoch": 5.98,
-      "grad_norm": 5.733375549316406,
+      "grad_norm": 5.972498416900635,
       "learning_rate": 9.8008008008008e-06,
-      "loss": 3.8102,
+      "loss": 3.8937,
       "step": 1990
     },
     {
       "epoch": 6.0,
-      "eval_accuracy": 0.3,
-      "eval_loss": 3.605971574783325,
-      "eval_runtime": 5.5123,
-      "eval_samples_per_second": 1814.12,
-      "eval_steps_per_second": 7.256,
+      "eval_accuracy": 0.2704,
+      "eval_loss": 3.716881275177002,
+      "eval_runtime": 5.3882,
+      "eval_samples_per_second": 1855.915,
+      "eval_steps_per_second": 7.424,
       "step": 1998
     },
     {
       "epoch": 6.01,
-      "grad_norm": 4.541661262512207,
+      "grad_norm": 6.924658298492432,
       "learning_rate": 9.799799799799801e-06,
-      "loss": 3.892,
+      "loss": 3.9717,
       "step": 2000
     },
     {
       "epoch": 6.04,
-      "grad_norm": 4.136720657348633,
+      "grad_norm": 5.908511161804199,
       "learning_rate": 9.7987987987988e-06,
-      "loss": 3.8085,
+      "loss": 3.8954,
       "step": 2010
     },
     {
       "epoch": 6.07,
-      "grad_norm": 3.9634923934936523,
+      "grad_norm": 8.730292320251465,
       "learning_rate": 9.797797797797798e-06,
-      "loss": 3.7753,
+      "loss": 3.8644,
       "step": 2020
     },
     {
       "epoch": 6.1,
-      "grad_norm": 6.353366851806641,
+      "grad_norm": 6.4170918464660645,
       "learning_rate": 9.796796796796799e-06,
-      "loss": 3.7973,
+      "loss": 3.8824,
       "step": 2030
     },
     {
       "epoch": 6.13,
-      "grad_norm": 7.301074028015137,
+      "grad_norm": 6.906913757324219,
       "learning_rate": 9.795795795795795e-06,
-      "loss": 3.7639,
+      "loss": 3.8481,
       "step": 2040
     },
     {
       "epoch": 6.16,
-      "grad_norm": 4.379009246826172,
+      "grad_norm": 7.524204254150391,
       "learning_rate": 9.794794794794796e-06,
-      "loss": 3.7622,
+      "loss": 3.8683,
       "step": 2050
     },
     {
       "epoch": 6.19,
-      "grad_norm": 6.0673017501831055,
+      "grad_norm": 5.297156810760498,
       "learning_rate": 9.793793793793794e-06,
-      "loss": 3.782,
+      "loss": 3.869,
       "step": 2060
     },
     {
       "epoch": 6.22,
-      "grad_norm": 7.272884368896484,
+      "grad_norm": 5.748040199279785,
       "learning_rate": 9.792792792792793e-06,
-      "loss": 3.7808,
+      "loss": 3.8864,
       "step": 2070
     },
     {
       "epoch": 6.25,
-      "grad_norm": 4.42779541015625,
+      "grad_norm": 7.591318607330322,
       "learning_rate": 9.791791791791793e-06,
-      "loss": 3.7739,
+      "loss": 3.8666,
       "step": 2080
     },
     {
       "epoch": 6.28,
-      "grad_norm": 8.053028106689453,
+      "grad_norm": 6.639626979827881,
       "learning_rate": 9.790790790790792e-06,
-      "loss": 3.7505,
+      "loss": 3.8461,
       "step": 2090
     },
     {
       "epoch": 6.31,
-      "grad_norm": 6.226222038269043,
+      "grad_norm": 5.813536643981934,
       "learning_rate": 9.78978978978979e-06,
-      "loss": 3.7606,
+      "loss": 3.8508,
       "step": 2100
     },
     {
       "epoch": 6.34,
-      "grad_norm": 4.875487804412842,
+      "grad_norm": 5.587143898010254,
       "learning_rate": 9.78878878878879e-06,
-      "loss": 3.7573,
+      "loss": 3.8615,
       "step": 2110
     },
     {
       "epoch": 6.37,
-      "grad_norm": 4.378990173339844,
+      "grad_norm": 5.760717868804932,
       "learning_rate": 9.787787787787788e-06,
-      "loss": 3.7284,
+      "loss": 3.8296,
       "step": 2120
     },
     {
       "epoch": 6.4,
-      "grad_norm": 6.2126264572143555,
+      "grad_norm": 5.6202521324157715,
       "learning_rate": 9.786786786786788e-06,
-      "loss": 3.7314,
+      "loss": 3.828,
       "step": 2130
     },
     {
       "epoch": 6.43,
-      "grad_norm": 4.079742431640625,
+      "grad_norm": 6.086943626403809,
       "learning_rate": 9.785785785785787e-06,
-      "loss": 3.7032,
+      "loss": 3.7952,
       "step": 2140
     },
     {
       "epoch": 6.46,
-      "grad_norm": 7.379823684692383,
+      "grad_norm": 4.865020751953125,
       "learning_rate": 9.784784784784785e-06,
-      "loss": 3.7106,
+      "loss": 3.8034,
       "step": 2150
     },
     {
       "epoch": 6.49,
-      "grad_norm": 4.514336585998535,
+      "grad_norm": 5.314583778381348,
       "learning_rate": 9.783783783783785e-06,
-      "loss": 3.6898,
+      "loss": 3.7967,
       "step": 2160
     },
     {
       "epoch": 6.52,
-      "grad_norm": 3.904522180557251,
+      "grad_norm": 7.212305068969727,
       "learning_rate": 9.782782782782784e-06,
-      "loss": 3.7002,
+      "loss": 3.8009,
       "step": 2170
     },
     {
       "epoch": 6.55,
-      "grad_norm": 4.5057172775268555,
+      "grad_norm": 5.448238372802734,
       "learning_rate": 9.781781781781783e-06,
-      "loss": 3.6705,
+      "loss": 3.7696,
       "step": 2180
     },
     {
       "epoch": 6.58,
-      "grad_norm": 3.526851177215576,
+      "grad_norm": 6.612052917480469,
       "learning_rate": 9.780780780780781e-06,
-      "loss": 3.7025,
+      "loss": 3.7999,
       "step": 2190
     },
     {
       "epoch": 6.61,
-      "grad_norm": 5.631099224090576,
+      "grad_norm": 8.778471946716309,
       "learning_rate": 9.779779779779781e-06,
-      "loss": 3.6743,
+      "loss": 3.7749,
       "step": 2200
     },
     {
       "epoch": 6.64,
-      "grad_norm": 4.578070640563965,
+      "grad_norm": 6.234766006469727,
       "learning_rate": 9.778778778778778e-06,
-      "loss": 3.6876,
+      "loss": 3.7787,
       "step": 2210
     },
     {
       "epoch": 6.67,
-      "grad_norm": 8.342512130737305,
+      "grad_norm": 8.373656272888184,
       "learning_rate": 9.777777777777779e-06,
-      "loss": 3.6695,
+      "loss": 3.7426,
       "step": 2220
     },
     {
       "epoch": 6.7,
-      "grad_norm": 8.751660346984863,
+      "grad_norm": 7.930400848388672,
       "learning_rate": 9.776776776776777e-06,
-      "loss": 3.6289,
+      "loss": 3.7318,
       "step": 2230
     },
     {
       "epoch": 6.73,
-      "grad_norm": 5.056054592132568,
+      "grad_norm": 6.091791152954102,
       "learning_rate": 9.775775775775776e-06,
-      "loss": 3.656,
+      "loss": 3.7577,
       "step": 2240
     },
     {
       "epoch": 6.76,
-      "grad_norm": 3.843867540359497,
+      "grad_norm": 8.18591594696045,
       "learning_rate": 9.774774774774776e-06,
-      "loss": 3.6576,
+      "loss": 3.7503,
       "step": 2250
     },
     {
       "epoch": 6.79,
-      "grad_norm": 6.093122959136963,
+      "grad_norm": 6.338694095611572,
       "learning_rate": 9.773773773773775e-06,
-      "loss": 3.6475,
+      "loss": 3.752,
       "step": 2260
     },
     {
       "epoch": 6.82,
-      "grad_norm": 5.330530643463135,
+      "grad_norm": 6.486820220947266,
       "learning_rate": 9.772772772772773e-06,
-      "loss": 3.6165,
+      "loss": 3.724,
       "step": 2270
     },
     {
       "epoch": 6.85,
-      "grad_norm": 5.0866522789001465,
+      "grad_norm": 7.222274303436279,
       "learning_rate": 9.771771771771774e-06,
-      "loss": 3.6244,
+      "loss": 3.7306,
       "step": 2280
     },
     {
       "epoch": 6.88,
-      "grad_norm": 4.231504917144775,
+      "grad_norm": 5.558598518371582,
       "learning_rate": 9.77077077077077e-06,
-      "loss": 3.598,
+      "loss": 3.707,
       "step": 2290
     },
     {
       "epoch": 6.91,
-      "grad_norm": 3.710031270980835,
+      "grad_norm": 5.144259929656982,
       "learning_rate": 9.76976976976977e-06,
-      "loss": 3.607,
+      "loss": 3.7176,
       "step": 2300
     },
     {
       "epoch": 6.94,
-      "grad_norm": 3.935680627822876,
+      "grad_norm": 4.85995626449585,
       "learning_rate": 9.76876876876877e-06,
-      "loss": 3.5915,
+      "loss": 3.6962,
       "step": 2310
     },
     {
       "epoch": 6.97,
-      "grad_norm": 6.9337158203125,
+      "grad_norm": 6.525197982788086,
       "learning_rate": 9.767767767767768e-06,
-      "loss": 3.5777,
+      "loss": 3.6987,
       "step": 2320
     },
     {
       "epoch": 7.0,
-      "grad_norm": 6.037449836730957,
+      "grad_norm": 7.76564359664917,
       "learning_rate": 9.766766766766768e-06,
-      "loss": 3.5657,
+      "loss": 3.6778,
       "step": 2330
     },
     {
       "epoch": 7.0,
-      "eval_accuracy": 0.3334,
-      "eval_loss": 3.3058323860168457,
-      "eval_runtime": 5.473,
-      "eval_samples_per_second": 1827.166,
-      "eval_steps_per_second": 7.309,
+      "eval_accuracy": 0.3143,
+      "eval_loss": 3.445819139480591,
+      "eval_runtime": 5.1424,
+      "eval_samples_per_second": 1944.606,
+      "eval_steps_per_second": 7.778,
       "step": 2331
     },
     {
       "epoch": 7.03,
-      "grad_norm": 5.291641712188721,
+      "grad_norm": 9.517220497131348,
       "learning_rate": 9.765765765765767e-06,
-      "loss": 3.679,
+      "loss": 3.7571,
       "step": 2340
     },
     {
       "epoch": 7.06,
-      "grad_norm": 4.0257344245910645,
+      "grad_norm": 6.669158935546875,
       "learning_rate": 9.764764764764765e-06,
-      "loss": 3.563,
+      "loss": 3.6657,
       "step": 2350
     },
     {
       "epoch": 7.09,
-      "grad_norm": 6.200382232666016,
+      "grad_norm": 7.924838542938232,
       "learning_rate": 9.763763763763766e-06,
-      "loss": 3.5908,
+      "loss": 3.6938,
       "step": 2360
     },
     {
       "epoch": 7.12,
-      "grad_norm": 5.8792195320129395,
+      "grad_norm": 5.500638961791992,
       "learning_rate": 9.762762762762763e-06,
-      "loss": 3.5365,
+      "loss": 3.6304,
       "step": 2370
     },
     {
       "epoch": 7.15,
-      "grad_norm": 4.288370609283447,
+      "grad_norm": 6.090194225311279,
       "learning_rate": 9.761761761761763e-06,
-      "loss": 3.5348,
+      "loss": 3.6601,
       "step": 2380
     },
     {
       "epoch": 7.18,
-      "grad_norm": 4.315982818603516,
+      "grad_norm": 6.263370513916016,
       "learning_rate": 9.760760760760762e-06,
-      "loss": 3.5532,
+      "loss": 3.6745,
       "step": 2390
     },
     {
       "epoch": 7.21,
-      "grad_norm": 4.326426982879639,
+      "grad_norm": 6.780274868011475,
       "learning_rate": 9.75975975975976e-06,
-      "loss": 3.5528,
+      "loss": 3.6622,
       "step": 2400
     },
     {
       "epoch": 7.24,
-      "grad_norm": 5.147789478302002,
+      "grad_norm": 7.445713520050049,
       "learning_rate": 9.758758758758759e-06,
-      "loss": 3.5382,
+      "loss": 3.6426,
       "step": 2410
     },
     {
       "epoch": 7.27,
-      "grad_norm": 3.932137966156006,
+      "grad_norm": 4.682408809661865,
       "learning_rate": 9.757757757757759e-06,
-      "loss": 3.5025,
+      "loss": 3.6141,
       "step": 2420
     },
     {
       "epoch": 7.3,
-      "grad_norm": 5.764312744140625,
+      "grad_norm": 6.9294047355651855,
       "learning_rate": 9.756756756756758e-06,
-      "loss": 3.5161,
+      "loss": 3.624,
       "step": 2430
     },
     {
       "epoch": 7.33,
-      "grad_norm": 6.128663063049316,
+      "grad_norm": 6.958638668060303,
       "learning_rate": 9.755755755755756e-06,
-      "loss": 3.5041,
+      "loss": 3.6159,
       "step": 2440
     },
     {
       "epoch": 7.36,
-      "grad_norm": 4.892031192779541,
+      "grad_norm": 5.1918463706970215,
       "learning_rate": 9.754754754754756e-06,
-      "loss": 3.5059,
+      "loss": 3.6209,
       "step": 2450
     },
     {
       "epoch": 7.39,
-      "grad_norm": 7.187922477722168,
+      "grad_norm": 10.057223320007324,
       "learning_rate": 9.753753753753753e-06,
-      "loss": 3.5016,
+      "loss": 3.6104,
       "step": 2460
     },
     {
       "epoch": 7.42,
-      "grad_norm": 11.209441184997559,
+      "grad_norm": 8.216449737548828,
       "learning_rate": 9.752752752752754e-06,
-      "loss": 3.5151,
+      "loss": 3.6395,
       "step": 2470
     },
     {
       "epoch": 7.45,
-      "grad_norm": 5.447879314422607,
+      "grad_norm": 4.761096477508545,
       "learning_rate": 9.751751751751752e-06,
-      "loss": 3.4939,
+      "loss": 3.6032,
       "step": 2480
     },
     {
       "epoch": 7.48,
-      "grad_norm": 4.490081310272217,
+      "grad_norm": 9.303138732910156,
       "learning_rate": 9.750750750750751e-06,
-      "loss": 3.4939,
+      "loss": 3.5969,
       "step": 2490
     },
     {
       "epoch": 7.51,
-      "grad_norm": 7.843050479888916,
+      "grad_norm": 10.344404220581055,
       "learning_rate": 9.749749749749751e-06,
-      "loss": 3.5062,
+      "loss": 3.6092,
       "step": 2500
     },
     {
       "epoch": 7.54,
-      "grad_norm": 6.0667033195495605,
+      "grad_norm": 7.905857086181641,
       "learning_rate": 9.74874874874875e-06,
-      "loss": 3.5231,
+      "loss": 3.6252,
       "step": 2510
     },
     {
       "epoch": 7.57,
-      "grad_norm": 8.192134857177734,
+      "grad_norm": 8.278596878051758,
       "learning_rate": 9.747747747747748e-06,
-      "loss": 3.5124,
+      "loss": 3.6091,
       "step": 2520
     },
     {
       "epoch": 7.6,
-      "grad_norm": 4.2382097244262695,
+      "grad_norm": 10.521196365356445,
       "learning_rate": 9.746746746746749e-06,
-      "loss": 3.452,
+      "loss": 3.5505,
       "step": 2530
     },
     {
       "epoch": 7.63,
-      "grad_norm": 6.560668468475342,
+      "grad_norm": 5.0142741203308105,
       "learning_rate": 9.745745745745746e-06,
-      "loss": 3.4334,
+      "loss": 3.5373,
       "step": 2540
     },
     {
       "epoch": 7.66,
-      "grad_norm": 6.222433567047119,
+      "grad_norm": 6.887248992919922,
       "learning_rate": 9.744744744744746e-06,
-      "loss": 3.4288,
+      "loss": 3.5366,
       "step": 2550
     },
     {
       "epoch": 7.69,
-      "grad_norm": 4.986079692840576,
+      "grad_norm": 7.138433456420898,
       "learning_rate": 9.743743743743744e-06,
-      "loss": 3.4335,
+      "loss": 3.5339,
       "step": 2560
     },
     {
       "epoch": 7.72,
-      "grad_norm": 5.174213886260986,
+      "grad_norm": 5.797271251678467,
       "learning_rate": 9.742742742742743e-06,
-      "loss": 3.4828,
+      "loss": 3.5964,
       "step": 2570
     },
     {
       "epoch": 7.75,
-      "grad_norm": 5.813471794128418,
+      "grad_norm": 5.385325908660889,
       "learning_rate": 9.741741741741743e-06,
-      "loss": 3.4162,
+      "loss": 3.5101,
       "step": 2580
     },
     {
       "epoch": 7.78,
-      "grad_norm": 4.290080547332764,
+      "grad_norm": 5.678484916687012,
       "learning_rate": 9.740740740740742e-06,
-      "loss": 3.4461,
+      "loss": 3.5584,
       "step": 2590
     },
     {
       "epoch": 7.81,
-      "grad_norm": 6.170722961425781,
+      "grad_norm": 7.096831798553467,
       "learning_rate": 9.73973973973974e-06,
-      "loss": 3.4619,
+      "loss": 3.568,
       "step": 2600
     },
     {
       "epoch": 7.84,
-      "grad_norm": 6.962252616882324,
+      "grad_norm": 6.322803497314453,
       "learning_rate": 9.73873873873874e-06,
-      "loss": 3.4105,
+      "loss": 3.5237,
       "step": 2610
     },
     {
       "epoch": 7.87,
-      "grad_norm": 6.705014228820801,
+      "grad_norm": 11.083927154541016,
       "learning_rate": 9.737737737737738e-06,
-      "loss": 3.4102,
+      "loss": 3.5155,
       "step": 2620
     },
     {
       "epoch": 7.9,
-      "grad_norm": 5.301889896392822,
+      "grad_norm": 5.589782238006592,
       "learning_rate": 9.736736736736738e-06,
-      "loss": 3.4097,
+      "loss": 3.5087,
       "step": 2630
     },
     {
       "epoch": 7.93,
-      "grad_norm": 5.87806510925293,
+      "grad_norm": 5.69963264465332,
       "learning_rate": 9.735735735735737e-06,
-      "loss": 3.3703,
+      "loss": 3.4768,
       "step": 2640
     },
     {
       "epoch": 7.96,
-      "grad_norm": 5.370548248291016,
+      "grad_norm": 6.673878192901611,
       "learning_rate": 9.734734734734735e-06,
-      "loss": 3.392,
+      "loss": 3.5003,
       "step": 2650
     },
     {
       "epoch": 7.99,
-      "grad_norm": 7.986278533935547,
+      "grad_norm": 6.1804280281066895,
       "learning_rate": 9.733733733733734e-06,
-      "loss": 3.3654,
+      "loss": 3.4698,
       "step": 2660
     },
     {
       "epoch": 8.0,
-      "eval_accuracy": 0.3695,
-      "eval_loss": 3.0348167419433594,
-      "eval_runtime": 5.4146,
-      "eval_samples_per_second": 1846.862,
-      "eval_steps_per_second": 7.387,
+      "eval_accuracy": 0.3563,
+      "eval_loss": 3.1671342849731445,
+      "eval_runtime": 5.3279,
+      "eval_samples_per_second": 1876.903,
+      "eval_steps_per_second": 7.508,
       "step": 2664
     },
     {
       "epoch": 8.02,
-      "grad_norm": 6.002692699432373,
+      "grad_norm": 4.99941349029541,
       "learning_rate": 9.732732732732734e-06,
-      "loss": 3.4404,
+      "loss": 3.5431,
       "step": 2670
     },
     {
       "epoch": 8.05,
-      "grad_norm": 6.926723957061768,
+      "grad_norm": 6.607128620147705,
       "learning_rate": 9.731731731731733e-06,
-      "loss": 3.3956,
+      "loss": 3.5007,
       "step": 2680
     },
     {
       "epoch": 8.08,
-      "grad_norm": 10.363128662109375,
+      "grad_norm": 7.015357494354248,
       "learning_rate": 9.730730730730731e-06,
-      "loss": 3.3887,
+      "loss": 3.5004,
       "step": 2690
     },
     {
       "epoch": 8.11,
-      "grad_norm": 4.207162857055664,
+      "grad_norm": 4.490257740020752,
       "learning_rate": 9.729729729729732e-06,
-      "loss": 3.3011,
+      "loss": 3.4203,
       "step": 2700
     },
     {
       "epoch": 8.14,
-      "grad_norm": 7.869438648223877,
+      "grad_norm": 8.291129112243652,
       "learning_rate": 9.728728728728728e-06,
-      "loss": 3.3791,
+      "loss": 3.4758,
       "step": 2710
     },
     {
       "epoch": 8.17,
-      "grad_norm": 5.768829345703125,
+      "grad_norm": 6.939655303955078,
       "learning_rate": 9.727727727727729e-06,
-      "loss": 3.373,
+      "loss": 3.4717,
       "step": 2720
     },
     {
       "epoch": 8.2,
-      "grad_norm": 8.534932136535645,
+      "grad_norm": 8.88289737701416,
       "learning_rate": 9.726726726726727e-06,
-      "loss": 3.3318,
+      "loss": 3.4523,
       "step": 2730
     },
     {
       "epoch": 8.23,
-      "grad_norm": 7.806580066680908,
+      "grad_norm": 7.005226135253906,
       "learning_rate": 9.725725725725726e-06,
-      "loss": 3.3467,
+      "loss": 3.4489,
       "step": 2740
     },
     {
       "epoch": 8.26,
-      "grad_norm": 5.99948263168335,
+      "grad_norm": 6.84315299987793,
       "learning_rate": 9.724724724724726e-06,
-      "loss": 3.3526,
+      "loss": 3.4557,
       "step": 2750
     },
     {
       "epoch": 8.29,
-      "grad_norm": 6.357593059539795,
+      "grad_norm": 7.020508766174316,
       "learning_rate": 9.723723723723725e-06,
-      "loss": 3.3829,
+      "loss": 3.4826,
       "step": 2760
     },
     {
       "epoch": 8.32,
-      "grad_norm": 5.759784698486328,
+      "grad_norm": 4.492513656616211,
       "learning_rate": 9.722722722722723e-06,
-      "loss": 3.36,
+      "loss": 3.4534,
       "step": 2770
     },
     {
       "epoch": 8.35,
-      "grad_norm": 6.890531539916992,
+      "grad_norm": 6.311222553253174,
       "learning_rate": 9.721721721721724e-06,
-      "loss": 3.2952,
+      "loss": 3.3839,
       "step": 2780
     },
     {
       "epoch": 8.38,
-      "grad_norm": 6.671658515930176,
+      "grad_norm": 6.426805019378662,
       "learning_rate": 9.72072072072072e-06,
-      "loss": 3.3099,
+      "loss": 3.4214,
       "step": 2790
     },
     {
       "epoch": 8.41,
-      "grad_norm": 5.392520904541016,
+      "grad_norm": 9.053902626037598,
       "learning_rate": 9.719719719719721e-06,
-      "loss": 3.3207,
+      "loss": 3.4299,
       "step": 2800
     },
     {
       "epoch": 8.44,
-      "grad_norm": 5.812443733215332,
+      "grad_norm": 8.069915771484375,
       "learning_rate": 9.71871871871872e-06,
-      "loss": 3.3213,
+      "loss": 3.4246,
       "step": 2810
     },
     {
       "epoch": 8.47,
-      "grad_norm": 9.847726821899414,
+      "grad_norm": 5.7368340492248535,
       "learning_rate": 9.717717717717718e-06,
-      "loss": 3.2608,
+      "loss": 3.3724,
       "step": 2820
     },
     {
       "epoch": 8.5,
-      "grad_norm": 7.5233306884765625,
+      "grad_norm": 8.907891273498535,
       "learning_rate": 9.716716716716718e-06,
-      "loss": 3.2403,
+      "loss": 3.3537,
       "step": 2830
     },
     {
       "epoch": 8.53,
-      "grad_norm": 7.249750137329102,
+      "grad_norm": 8.060093879699707,
       "learning_rate": 9.715715715715717e-06,
-      "loss": 3.3013,
+      "loss": 3.4005,
       "step": 2840
     },
     {
       "epoch": 8.56,
-      "grad_norm": 6.295549392700195,
+      "grad_norm": 9.46430492401123,
       "learning_rate": 9.714714714714716e-06,
-      "loss": 3.2893,
+      "loss": 3.3916,
       "step": 2850
     },
     {
       "epoch": 8.59,
-      "grad_norm": 7.23241662979126,
+      "grad_norm": 7.094594955444336,
       "learning_rate": 9.713713713713714e-06,
-      "loss": 3.2749,
+      "loss": 3.3743,
       "step": 2860
     },
     {
       "epoch": 8.62,
-      "grad_norm": 5.618537425994873,
+      "grad_norm": 8.086226463317871,
       "learning_rate": 9.712712712712713e-06,
-      "loss": 3.2845,
+      "loss": 3.3722,
       "step": 2870
     },
     {
       "epoch": 8.65,
-      "grad_norm": 6.242091655731201,
+      "grad_norm": 7.181630611419678,
       "learning_rate": 9.711711711711711e-06,
-      "loss": 3.2787,
+      "loss": 3.3748,
       "step": 2880
     },
     {
       "epoch": 8.68,
-      "grad_norm": 4.652316093444824,
+      "grad_norm": 7.318678379058838,
       "learning_rate": 9.710710710710712e-06,
-      "loss": 3.2639,
+      "loss": 3.3595,
       "step": 2890
     },
     {
       "epoch": 8.71,
-      "grad_norm": 4.795337677001953,
+      "grad_norm": 6.843329906463623,
       "learning_rate": 9.70970970970971e-06,
-      "loss": 3.2262,
+      "loss": 3.3251,
       "step": 2900
     },
     {
       "epoch": 8.74,
-      "grad_norm": 5.1860785484313965,
+      "grad_norm": 4.767055988311768,
       "learning_rate": 9.708708708708709e-06,
-      "loss": 3.2294,
+      "loss": 3.341,
       "step": 2910
     },
     {
       "epoch": 8.77,
-      "grad_norm": 4.808598518371582,
+      "grad_norm": 7.0615973472595215,
       "learning_rate": 9.707707707707709e-06,
-      "loss": 3.2459,
+      "loss": 3.3486,
       "step": 2920
     },
     {
       "epoch": 8.8,
-      "grad_norm": 7.304434299468994,
+      "grad_norm": 5.66889762878418,
       "learning_rate": 9.706706706706708e-06,
-      "loss": 3.2383,
+      "loss": 3.335,
       "step": 2930
     },
     {
       "epoch": 8.83,
-      "grad_norm": 5.718128204345703,
+      "grad_norm": 6.953256607055664,
       "learning_rate": 9.705705705705706e-06,
-      "loss": 3.1937,
+      "loss": 3.2937,
       "step": 2940
     },
     {
       "epoch": 8.86,
-      "grad_norm": 6.606522560119629,
+      "grad_norm": 6.755582809448242,
       "learning_rate": 9.704704704704707e-06,
-      "loss": 3.1842,
+      "loss": 3.2895,
       "step": 2950
     },
     {
       "epoch": 8.89,
-      "grad_norm": 8.326537132263184,
+      "grad_norm": 4.61521577835083,
       "learning_rate": 9.703703703703703e-06,
-      "loss": 3.2275,
+      "loss": 3.3276,
       "step": 2960
     },
     {
       "epoch": 8.92,
-      "grad_norm": 6.412814140319824,
+      "grad_norm": 8.599690437316895,
       "learning_rate": 9.702702702702704e-06,
-      "loss": 3.1738,
+      "loss": 3.2856,
       "step": 2970
     },
     {
       "epoch": 8.95,
-      "grad_norm": 6.820006847381592,
+      "grad_norm": 7.745826244354248,
       "learning_rate": 9.701701701701702e-06,
-      "loss": 3.1641,
+      "loss": 3.2584,
       "step": 2980
     },
     {
       "epoch": 8.98,
-      "grad_norm": 6.501119613647461,
+      "grad_norm": 6.4888386726379395,
       "learning_rate": 9.700700700700701e-06,
-      "loss": 3.1954,
+      "loss": 3.2876,
       "step": 2990
     },
     {
       "epoch": 9.0,
-      "eval_accuracy": 0.3996,
-      "eval_loss": 2.7789359092712402,
-      "eval_runtime": 5.3085,
-      "eval_samples_per_second": 1883.757,
-      "eval_steps_per_second": 7.535,
+      "eval_accuracy": 0.3895,
+      "eval_loss": 2.8939809799194336,
+      "eval_runtime": 5.4146,
+      "eval_samples_per_second": 1846.858,
+      "eval_steps_per_second": 7.387,
       "step": 2997
     },
     {
       "epoch": 9.01,
-      "grad_norm": 5.666793346405029,
+      "grad_norm": 10.507532119750977,
       "learning_rate": 9.699699699699701e-06,
-      "loss": 3.3258,
+      "loss": 3.4167,
       "step": 3000
     },
     {
       "epoch": 9.04,
-      "grad_norm": 7.874444961547852,
+      "grad_norm": 7.532200813293457,
       "learning_rate": 9.6986986986987e-06,
-      "loss": 3.1761,
+      "loss": 3.2686,
       "step": 3010
     },
     {
       "epoch": 9.07,
-      "grad_norm": 11.075509071350098,
+      "grad_norm": 6.850391864776611,
       "learning_rate": 9.697697697697698e-06,
-      "loss": 3.2038,
+      "loss": 3.2977,
       "step": 3020
     },
     {
       "epoch": 9.1,
-      "grad_norm": 7.179293632507324,
+      "grad_norm": 6.166021823883057,
       "learning_rate": 9.696696696696699e-06,
-      "loss": 3.1665,
+      "loss": 3.2493,
       "step": 3030
     },
     {
       "epoch": 9.13,
-      "grad_norm": 7.670932769775391,
+      "grad_norm": 6.597267150878906,
       "learning_rate": 9.695695695695696e-06,
-      "loss": 3.172,
+      "loss": 3.2661,
       "step": 3040
     },
     {
       "epoch": 9.16,
-      "grad_norm": 7.348392963409424,
+      "grad_norm": 7.0463104248046875,
       "learning_rate": 9.694694694694696e-06,
-      "loss": 3.1542,
+      "loss": 3.2438,
       "step": 3050
     },
     {
       "epoch": 9.19,
-      "grad_norm": 5.808042526245117,
+      "grad_norm": 5.287309646606445,
       "learning_rate": 9.693693693693694e-06,
-      "loss": 3.17,
+      "loss": 3.2672,
       "step": 3060
     },
     {
       "epoch": 9.22,
-      "grad_norm": 7.05147647857666,
+      "grad_norm": 8.986891746520996,
       "learning_rate": 9.692692692692693e-06,
-      "loss": 3.1652,
+      "loss": 3.2565,
       "step": 3070
     },
     {
       "epoch": 9.25,
-      "grad_norm": 5.615456581115723,
+      "grad_norm": 7.13382625579834,
       "learning_rate": 9.691691691691693e-06,
-      "loss": 3.152,
+      "loss": 3.2299,
       "step": 3080
     },
     {
       "epoch": 9.28,
-      "grad_norm": 5.297279357910156,
+      "grad_norm": 5.380059242248535,
       "learning_rate": 9.690690690690692e-06,
-      "loss": 3.1161,
+      "loss": 3.2065,
       "step": 3090
     },
     {
       "epoch": 9.31,
-      "grad_norm": 7.348149299621582,
+      "grad_norm": 9.107782363891602,
       "learning_rate": 9.68968968968969e-06,
-      "loss": 3.1439,
+      "loss": 3.2342,
       "step": 3100
     },
     {
       "epoch": 9.34,
-      "grad_norm": 9.006824493408203,
+      "grad_norm": 9.863898277282715,
       "learning_rate": 9.68868868868869e-06,
-      "loss": 3.1605,
+      "loss": 3.2487,
       "step": 3110
     },
     {
       "epoch": 9.37,
-      "grad_norm": 6.75848913192749,
+      "grad_norm": 8.917726516723633,
       "learning_rate": 9.687687687687688e-06,
-      "loss": 3.1908,
+      "loss": 3.2777,
       "step": 3120
     },
     {
       "epoch": 9.4,
-      "grad_norm": 4.358802795410156,
+      "grad_norm": 7.849170207977295,
       "learning_rate": 9.686686686686686e-06,
-      "loss": 3.1367,
+      "loss": 3.215,
       "step": 3130
     },
     {
       "epoch": 9.43,
-      "grad_norm": 8.524468421936035,
+      "grad_norm": 12.458362579345703,
       "learning_rate": 9.685685685685687e-06,
-      "loss": 3.1364,
+      "loss": 3.2329,
       "step": 3140
     },
     {
       "epoch": 9.46,
-      "grad_norm": 10.768378257751465,
+      "grad_norm": 7.917840003967285,
       "learning_rate": 9.684684684684685e-06,
-      "loss": 3.1234,
+      "loss": 3.2099,
       "step": 3150
     },
     {
       "epoch": 9.49,
-      "grad_norm": 5.764973163604736,
+      "grad_norm": 10.534679412841797,
       "learning_rate": 9.683683683683684e-06,
-      "loss": 3.0559,
+      "loss": 3.154,
       "step": 3160
     },
     {
       "epoch": 9.52,
-      "grad_norm": 5.615333080291748,
+      "grad_norm": 6.135841369628906,
       "learning_rate": 9.682682682682684e-06,
-      "loss": 3.051,
+      "loss": 3.1429,
       "step": 3170
     },
     {
       "epoch": 9.55,
-      "grad_norm": 6.663029670715332,
+      "grad_norm": 5.664760112762451,
       "learning_rate": 9.681681681681683e-06,
-      "loss": 3.1616,
+      "loss": 3.2433,
       "step": 3180
     },
     {
       "epoch": 9.58,
-      "grad_norm": 7.263551235198975,
+      "grad_norm": 7.515036106109619,
       "learning_rate": 9.680680680680681e-06,
-      "loss": 3.1268,
+      "loss": 3.2329,
       "step": 3190
     },
     {
       "epoch": 9.61,
-      "grad_norm": 5.53592586517334,
+      "grad_norm": 6.98620080947876,
       "learning_rate": 9.67967967967968e-06,
-      "loss": 3.1207,
+      "loss": 3.2043,
       "step": 3200
     },
     {
       "epoch": 9.64,
-      "grad_norm": 6.131815433502197,
+      "grad_norm": 9.919116020202637,
       "learning_rate": 9.678678678678679e-06,
-      "loss": 3.0526,
+      "loss": 3.1351,
       "step": 3210
     },
     {
       "epoch": 9.67,
-      "grad_norm": 6.56910514831543,
+      "grad_norm": 5.127571105957031,
       "learning_rate": 9.677677677677679e-06,
-      "loss": 3.0873,
+      "loss": 3.1677,
       "step": 3220
     },
     {
       "epoch": 9.7,
-      "grad_norm": 4.861180305480957,
+      "grad_norm": 6.015968322753906,
       "learning_rate": 9.676676676676677e-06,
-      "loss": 3.0744,
+      "loss": 3.1619,
       "step": 3230
     },
     {
       "epoch": 9.73,
-      "grad_norm": 7.075262069702148,
+      "grad_norm": 12.066081047058105,
       "learning_rate": 9.675675675675676e-06,
-      "loss": 3.0503,
+      "loss": 3.1398,
       "step": 3240
     },
     {
       "epoch": 9.76,
-      "grad_norm": 5.537904262542725,
+      "grad_norm": 5.146203994750977,
       "learning_rate": 9.674674674674676e-06,
-      "loss": 3.1044,
+      "loss": 3.1946,
       "step": 3250
     },
     {
       "epoch": 9.79,
-      "grad_norm": 5.602268218994141,
+      "grad_norm": 4.431717872619629,
       "learning_rate": 9.673673673673675e-06,
-      "loss": 3.0551,
+      "loss": 3.1439,
       "step": 3260
     },
     {
       "epoch": 9.82,
-      "grad_norm": 6.3274383544921875,
+      "grad_norm": 6.210360527038574,
       "learning_rate": 9.672672672672673e-06,
-      "loss": 3.0443,
+      "loss": 3.1256,
       "step": 3270
     },
     {
       "epoch": 9.85,
-      "grad_norm": 5.346724987030029,
+      "grad_norm": 6.866305828094482,
       "learning_rate": 9.671671671671674e-06,
-      "loss": 3.0897,
+      "loss": 3.1678,
       "step": 3280
     },
     {
       "epoch": 9.88,
-      "grad_norm": 7.766417503356934,
+      "grad_norm": 6.062428951263428,
       "learning_rate": 9.67067067067067e-06,
-      "loss": 3.0077,
+      "loss": 3.09,
       "step": 3290
     },
     {
       "epoch": 9.91,
-      "grad_norm": 9.037227630615234,
+      "grad_norm": 9.173726081848145,
       "learning_rate": 9.669669669669671e-06,
-      "loss": 2.9855,
+      "loss": 3.0809,
       "step": 3300
     },
     {
       "epoch": 9.94,
-      "grad_norm": 5.963006496429443,
+      "grad_norm": 5.687558174133301,
       "learning_rate": 9.66866866866867e-06,
-      "loss": 3.0066,
+      "loss": 3.1043,
       "step": 3310
     },
     {
       "epoch": 9.97,
-      "grad_norm": 5.292966365814209,
+      "grad_norm": 8.732027053833008,
       "learning_rate": 9.667667667667668e-06,
-      "loss": 3.0617,
+      "loss": 3.1362,
       "step": 3320
     },
     {
       "epoch": 10.0,
-      "grad_norm": 46.41136169433594,
+      "grad_norm": 55.31525421142578,
       "learning_rate": 9.666666666666667e-06,
-      "loss": 3.0844,
+      "loss": 3.1593,
       "step": 3330
     },
     {
       "epoch": 10.0,
-      "eval_accuracy": 0.4416,
-      "eval_loss": 2.559234857559204,
-      "eval_runtime": 5.2484,
-      "eval_samples_per_second": 1905.357,
-      "eval_steps_per_second": 7.621,
+      "eval_accuracy": 0.4262,
+      "eval_loss": 2.6568233966827393,
+      "eval_runtime": 5.2374,
+      "eval_samples_per_second": 1909.361,
+      "eval_steps_per_second": 7.637,
       "step": 3330
     },
     {
       "epoch": 10.03,
-      "grad_norm": 5.413119316101074,
+      "grad_norm": 6.48677396774292,
       "learning_rate": 9.665665665665667e-06,
-      "loss": 3.0087,
+      "loss": 3.089,
       "step": 3340
     },
     {
       "epoch": 10.06,
-      "grad_norm": 5.648168087005615,
+      "grad_norm": 6.252569675445557,
       "learning_rate": 9.664664664664666e-06,
-      "loss": 2.9811,
+      "loss": 3.0653,
       "step": 3350
     },
     {
       "epoch": 10.09,
-      "grad_norm": 6.271695137023926,
+      "grad_norm": 15.645646095275879,
       "learning_rate": 9.663663663663664e-06,
-      "loss": 2.9973,
+      "loss": 3.0828,
       "step": 3360
     },
     {
       "epoch": 10.12,
-      "grad_norm": 8.640029907226562,
+      "grad_norm": 7.632082939147949,
       "learning_rate": 9.662662662662663e-06,
-      "loss": 2.9943,
+      "loss": 3.0857,
       "step": 3370
     },
     {
       "epoch": 10.15,
-      "grad_norm": 9.350212097167969,
+      "grad_norm": 6.399557113647461,
       "learning_rate": 9.661661661661661e-06,
-      "loss": 3.0313,
+      "loss": 3.1192,
       "step": 3380
     },
     {
       "epoch": 10.18,
-      "grad_norm": 6.352114200592041,
+      "grad_norm": 7.19164514541626,
       "learning_rate": 9.660660660660662e-06,
-      "loss": 2.9949,
+      "loss": 3.0799,
       "step": 3390
     },
     {
       "epoch": 10.21,
-      "grad_norm": 6.2151265144348145,
+      "grad_norm": 6.629026412963867,
       "learning_rate": 9.65965965965966e-06,
-      "loss": 3.004,
+      "loss": 3.1019,
       "step": 3400
     },
     {
       "epoch": 10.24,
-      "grad_norm": 5.503857612609863,
+      "grad_norm": 8.625972747802734,
       "learning_rate": 9.658658658658659e-06,
-      "loss": 2.9732,
+      "loss": 3.0465,
       "step": 3410
     },
     {
       "epoch": 10.27,
-      "grad_norm": 6.430086612701416,
+      "grad_norm": 9.585230827331543,
       "learning_rate": 9.65765765765766e-06,
-      "loss": 2.9809,
+      "loss": 3.0649,
       "step": 3420
     },
     {
       "epoch": 10.3,
-      "grad_norm": 6.148303508758545,
+      "grad_norm": 10.085891723632812,
       "learning_rate": 9.656656656656658e-06,
-      "loss": 2.9221,
+      "loss": 3.01,
       "step": 3430
     },
     {
       "epoch": 10.33,
-      "grad_norm": 8.582883834838867,
+      "grad_norm": 6.38456916809082,
       "learning_rate": 9.655655655655656e-06,
-      "loss": 2.983,
+      "loss": 3.077,
       "step": 3440
     },
     {
       "epoch": 10.36,
-      "grad_norm": 11.020023345947266,
+      "grad_norm": 20.005468368530273,
       "learning_rate": 9.654654654654655e-06,
-      "loss": 2.9756,
+      "loss": 3.052,
       "step": 3450
     },
     {
       "epoch": 10.39,
-      "grad_norm": 4.530513763427734,
+      "grad_norm": 7.66636323928833,
       "learning_rate": 9.653653653653654e-06,
-      "loss": 2.9799,
+      "loss": 3.0551,
       "step": 3460
     },
     {
       "epoch": 10.42,
-      "grad_norm": 5.755140781402588,
+      "grad_norm": 11.84695053100586,
       "learning_rate": 9.652652652652654e-06,
-      "loss": 2.9681,
+      "loss": 3.0474,
       "step": 3470
     },
     {
       "epoch": 10.45,
-      "grad_norm": 6.647262096405029,
+      "grad_norm": 5.9561614990234375,
       "learning_rate": 9.651651651651652e-06,
-      "loss": 2.9987,
+      "loss": 3.0718,
       "step": 3480
     },
     {
       "epoch": 10.48,
-      "grad_norm": 5.807781219482422,
+      "grad_norm": 7.304963111877441,
       "learning_rate": 9.650650650650651e-06,
-      "loss": 2.9805,
+      "loss": 3.0623,
       "step": 3490
     },
     {
       "epoch": 10.51,
-      "grad_norm": 8.440781593322754,
+      "grad_norm": 9.812822341918945,
       "learning_rate": 9.649649649649651e-06,
-      "loss": 2.9547,
+      "loss": 3.0262,
       "step": 3500
     },
     {
       "epoch": 10.54,
-      "grad_norm": 6.510585308074951,
+      "grad_norm": 6.338140964508057,
       "learning_rate": 9.64864864864865e-06,
-      "loss": 2.8784,
+      "loss": 2.9516,
       "step": 3510
     },
     {
       "epoch": 10.57,
-      "grad_norm": 6.7697882652282715,
+      "grad_norm": 8.684576988220215,
       "learning_rate": 9.647647647647648e-06,
-      "loss": 2.9176,
+      "loss": 3.0065,
       "step": 3520
     },
     {
       "epoch": 10.6,
-      "grad_norm": 6.958645820617676,
+      "grad_norm": 8.328993797302246,
       "learning_rate": 9.646646646646649e-06,
-      "loss": 2.9219,
+      "loss": 2.9866,
       "step": 3530
     },
     {
       "epoch": 10.63,
-      "grad_norm": 9.358959197998047,
+      "grad_norm": 9.709425926208496,
       "learning_rate": 9.645645645645646e-06,
-      "loss": 2.9125,
+      "loss": 2.9823,
       "step": 3540
     },
     {
       "epoch": 10.66,
-      "grad_norm": 8.079233169555664,
+      "grad_norm": 5.012246608734131,
       "learning_rate": 9.644644644644644e-06,
-      "loss": 2.8875,
+      "loss": 2.9776,
       "step": 3550
     },
     {
       "epoch": 10.69,
-      "grad_norm": 5.118608474731445,
+      "grad_norm": 6.734545707702637,
       "learning_rate": 9.643643643643645e-06,
-      "loss": 2.9489,
+      "loss": 3.0157,
       "step": 3560
     },
     {
       "epoch": 10.72,
-      "grad_norm": 10.030895233154297,
+      "grad_norm": 10.985686302185059,
       "learning_rate": 9.642642642642643e-06,
-      "loss": 2.8895,
+      "loss": 2.9702,
       "step": 3570
     },
     {
       "epoch": 10.75,
-      "grad_norm": 7.586790561676025,
+      "grad_norm": 5.150612831115723,
       "learning_rate": 9.641641641641642e-06,
-      "loss": 2.8857,
+      "loss": 2.9677,
       "step": 3580
     },
     {
       "epoch": 10.78,
-      "grad_norm": 5.331732749938965,
+      "grad_norm": 5.08071231842041,
       "learning_rate": 9.640640640640642e-06,
-      "loss": 2.8399,
+      "loss": 2.9149,
       "step": 3590
     },
     {
       "epoch": 10.81,
-      "grad_norm": 6.730870723724365,
+      "grad_norm": 9.785350799560547,
       "learning_rate": 9.63963963963964e-06,
-      "loss": 2.9184,
+      "loss": 3.0004,
       "step": 3600
     },
     {
       "epoch": 10.84,
-      "grad_norm": 6.986204147338867,
+      "grad_norm": 6.093175888061523,
       "learning_rate": 9.63863863863864e-06,
-      "loss": 2.8743,
+      "loss": 2.9454,
       "step": 3610
     },
     {
       "epoch": 10.87,
-      "grad_norm": 7.9817094802856445,
+      "grad_norm": 8.718548774719238,
       "learning_rate": 9.637637637637638e-06,
-      "loss": 2.8864,
+      "loss": 2.9534,
       "step": 3620
     },
     {
       "epoch": 10.9,
-      "grad_norm": 5.197146892547607,
+      "grad_norm": 7.935983657836914,
       "learning_rate": 9.636636636636636e-06,
-      "loss": 2.873,
+      "loss": 2.95,
       "step": 3630
     },
     {
       "epoch": 10.93,
-      "grad_norm": 9.333887100219727,
+      "grad_norm": 7.35346794128418,
       "learning_rate": 9.635635635635637e-06,
-      "loss": 2.898,
+      "loss": 2.9618,
       "step": 3640
     },
     {
       "epoch": 10.96,
-      "grad_norm": 5.456880569458008,
+      "grad_norm": 8.22110652923584,
       "learning_rate": 9.634634634634635e-06,
-      "loss": 2.8673,
+      "loss": 2.9338,
       "step": 3650
     },
     {
       "epoch": 10.99,
-      "grad_norm": 6.91350793838501,
+      "grad_norm": 9.343371391296387,
       "learning_rate": 9.633633633633634e-06,
-      "loss": 2.9031,
+      "loss": 2.9526,
       "step": 3660
     },
     {
       "epoch": 11.0,
-      "eval_accuracy": 0.4707,
-      "eval_loss": 2.3488929271698,
-      "eval_runtime": 5.4797,
-      "eval_samples_per_second": 1824.908,
-      "eval_steps_per_second": 7.3,
+      "eval_accuracy": 0.4536,
+      "eval_loss": 2.4380807876586914,
+      "eval_runtime": 5.3656,
+      "eval_samples_per_second": 1863.73,
+      "eval_steps_per_second": 7.455,
       "step": 3663
     },
     {
       "epoch": 11.02,
-      "grad_norm": 5.921065330505371,
+      "grad_norm": 6.658874034881592,
       "learning_rate": 9.632632632632634e-06,
-      "loss": 2.966,
+      "loss": 3.0135,
       "step": 3670
     },
     {
       "epoch": 11.05,
-      "grad_norm": 6.959547519683838,
+      "grad_norm": 6.795731544494629,
       "learning_rate": 9.631631631631633e-06,
-      "loss": 2.839,
+      "loss": 2.905,
       "step": 3680
     },
     {
       "epoch": 11.08,
-      "grad_norm": 6.492772579193115,
+      "grad_norm": 8.555694580078125,
       "learning_rate": 9.630630630630631e-06,
-      "loss": 2.8586,
+      "loss": 2.9259,
       "step": 3690
     },
     {
       "epoch": 11.11,
-      "grad_norm": 9.122268676757812,
+      "grad_norm": 8.404982566833496,
       "learning_rate": 9.62962962962963e-06,
-      "loss": 2.864,
+      "loss": 2.9376,
       "step": 3700
     },
     {
       "epoch": 11.14,
-      "grad_norm": 5.102015018463135,
+      "grad_norm": 10.255926132202148,
       "learning_rate": 9.628628628628629e-06,
-      "loss": 2.8295,
+      "loss": 2.903,
       "step": 3710
     },
     {
       "epoch": 11.17,
-      "grad_norm": 5.498779773712158,
+      "grad_norm": 8.224187850952148,
       "learning_rate": 9.627627627627629e-06,
-      "loss": 2.7971,
+      "loss": 2.8626,
       "step": 3720
     },
     {
       "epoch": 11.2,
-      "grad_norm": 7.445620536804199,
+      "grad_norm": 6.741880893707275,
       "learning_rate": 9.626626626626627e-06,
-      "loss": 2.8117,
+      "loss": 2.8773,
       "step": 3730
     },
     {
       "epoch": 11.23,
-      "grad_norm": 6.615799427032471,
+      "grad_norm": 6.586903095245361,
       "learning_rate": 9.625625625625626e-06,
-      "loss": 2.8376,
+      "loss": 2.9102,
       "step": 3740
     },
     {
       "epoch": 11.26,
-      "grad_norm": 5.508298397064209,
+      "grad_norm": 8.833194732666016,
       "learning_rate": 9.624624624624626e-06,
-      "loss": 2.7868,
+      "loss": 2.8608,
       "step": 3750
     },
     {
       "epoch": 11.29,
-      "grad_norm": 7.494630813598633,
+      "grad_norm": 10.5403413772583,
       "learning_rate": 9.623623623623625e-06,
-      "loss": 2.8364,
+      "loss": 2.9035,
       "step": 3760
     },
     {
       "epoch": 11.32,
-      "grad_norm": 7.023970603942871,
+      "grad_norm": 11.01386547088623,
       "learning_rate": 9.622622622622624e-06,
-      "loss": 2.8093,
+      "loss": 2.8688,
       "step": 3770
     },
     {
       "epoch": 11.35,
-      "grad_norm": 9.165661811828613,
+      "grad_norm": 10.176712989807129,
       "learning_rate": 9.621621621621622e-06,
-      "loss": 2.8618,
+      "loss": 2.9336,
       "step": 3780
     },
     {
       "epoch": 11.38,
-      "grad_norm": 6.320451736450195,
+      "grad_norm": 11.736374855041504,
       "learning_rate": 9.62062062062062e-06,
-      "loss": 2.8398,
+      "loss": 2.9007,
       "step": 3790
     },
     {
       "epoch": 11.41,
-      "grad_norm": 4.613430023193359,
+      "grad_norm": 5.24038028717041,
       "learning_rate": 9.61961961961962e-06,
-      "loss": 2.815,
+      "loss": 2.8764,
       "step": 3800
     },
     {
       "epoch": 11.44,
-      "grad_norm": 7.975369453430176,
+      "grad_norm": 8.705958366394043,
       "learning_rate": 9.61861861861862e-06,
-      "loss": 2.8035,
+      "loss": 2.8665,
       "step": 3810
     },
     {
       "epoch": 11.47,
-      "grad_norm": 6.210441589355469,
+      "grad_norm": 8.703800201416016,
       "learning_rate": 9.617617617617618e-06,
-      "loss": 2.7855,
+      "loss": 2.856,
       "step": 3820
     },
     {
       "epoch": 11.5,
-      "grad_norm": 6.713306903839111,
+      "grad_norm": 6.525952339172363,
       "learning_rate": 9.616616616616617e-06,
-      "loss": 2.7983,
+      "loss": 2.8662,
       "step": 3830
     },
     {
       "epoch": 11.53,
-      "grad_norm": 12.956954956054688,
+      "grad_norm": 8.53999137878418,
       "learning_rate": 9.615615615615617e-06,
-      "loss": 2.7922,
+      "loss": 2.8424,
       "step": 3840
     },
     {
       "epoch": 11.56,
-      "grad_norm": 6.167584419250488,
+      "grad_norm": 7.226678371429443,
       "learning_rate": 9.614614614614616e-06,
-      "loss": 2.8004,
+      "loss": 2.8532,
       "step": 3850
     },
     {
       "epoch": 11.59,
-      "grad_norm": 10.142858505249023,
+      "grad_norm": 6.830595970153809,
       "learning_rate": 9.613613613613614e-06,
-      "loss": 2.7885,
+      "loss": 2.8532,
       "step": 3860
     },
     {
       "epoch": 11.62,
-      "grad_norm": 7.711623668670654,
+      "grad_norm": 6.560692310333252,
       "learning_rate": 9.612612612612613e-06,
-      "loss": 2.7812,
+      "loss": 2.8528,
       "step": 3870
     },
     {
       "epoch": 11.65,
-      "grad_norm": 5.236615180969238,
+      "grad_norm": 11.35714340209961,
       "learning_rate": 9.611611611611611e-06,
-      "loss": 2.8003,
+      "loss": 2.8697,
       "step": 3880
     },
     {
       "epoch": 11.68,
-      "grad_norm": 7.091507434844971,
+      "grad_norm": 11.555566787719727,
       "learning_rate": 9.610610610610612e-06,
-      "loss": 2.7735,
+      "loss": 2.8444,
       "step": 3890
     },
     {
       "epoch": 11.71,
-      "grad_norm": 9.978224754333496,
+      "grad_norm": 6.973629951477051,
       "learning_rate": 9.60960960960961e-06,
-      "loss": 2.7867,
+      "loss": 2.8528,
       "step": 3900
     },
     {
       "epoch": 11.74,
-      "grad_norm": 6.7497382164001465,
+      "grad_norm": 8.682169914245605,
       "learning_rate": 9.608608608608609e-06,
-      "loss": 2.732,
+      "loss": 2.7876,
       "step": 3910
     },
     {
       "epoch": 11.77,
-      "grad_norm": 5.517454147338867,
+      "grad_norm": 7.761390209197998,
       "learning_rate": 9.60760760760761e-06,
-      "loss": 2.741,
+      "loss": 2.811,
       "step": 3920
     },
     {
       "epoch": 11.8,
-      "grad_norm": 9.171282768249512,
+      "grad_norm": 10.123479843139648,
       "learning_rate": 9.606606606606608e-06,
-      "loss": 2.7839,
+      "loss": 2.8494,
       "step": 3930
     },
     {
       "epoch": 11.83,
-      "grad_norm": 8.752594947814941,
+      "grad_norm": 8.13418197631836,
       "learning_rate": 9.605605605605606e-06,
-      "loss": 2.7617,
+      "loss": 2.8175,
       "step": 3940
     },
     {
       "epoch": 11.86,
-      "grad_norm": 8.73362922668457,
+      "grad_norm": 5.8594584465026855,
       "learning_rate": 9.604604604604605e-06,
-      "loss": 2.7832,
+      "loss": 2.8262,
       "step": 3950
     },
     {
       "epoch": 11.89,
-      "grad_norm": 5.8772969245910645,
+      "grad_norm": 11.357876777648926,
       "learning_rate": 9.603603603603604e-06,
-      "loss": 2.7139,
+      "loss": 2.7825,
       "step": 3960
     },
     {
       "epoch": 11.92,
-      "grad_norm": 7.320891380310059,
+      "grad_norm": 6.700917720794678,
       "learning_rate": 9.602602602602604e-06,
-      "loss": 2.7683,
+      "loss": 2.827,
       "step": 3970
     },
     {
       "epoch": 11.95,
-      "grad_norm": 5.798232555389404,
+      "grad_norm": 9.411513328552246,
       "learning_rate": 9.601601601601602e-06,
-      "loss": 2.7125,
+      "loss": 2.7742,
       "step": 3980
     },
     {
       "epoch": 11.98,
-      "grad_norm": 6.342475414276123,
+      "grad_norm": 7.32675313949585,
       "learning_rate": 9.600600600600601e-06,
-      "loss": 2.7383,
+      "loss": 2.7823,
       "step": 3990
     },
     {
       "epoch": 12.0,
-      "eval_accuracy": 0.4907,
-      "eval_loss": 2.1828925609588623,
-      "eval_runtime": 5.5019,
-      "eval_samples_per_second": 1817.545,
-      "eval_steps_per_second": 7.27,
+      "eval_accuracy": 0.4763,
+      "eval_loss": 2.262507915496826,
+      "eval_runtime": 5.1717,
+      "eval_samples_per_second": 1933.592,
+      "eval_steps_per_second": 7.734,
       "step": 3996
     },
     {
       "epoch": 12.01,
-      "grad_norm": 5.237957954406738,
+      "grad_norm": 5.857390403747559,
       "learning_rate": 9.5995995995996e-06,
-      "loss": 2.9097,
+      "loss": 2.9763,
       "step": 4000
     },
     {
       "epoch": 12.04,
-      "grad_norm": 8.513269424438477,
+      "grad_norm": 8.120659828186035,
       "learning_rate": 9.5985985985986e-06,
-      "loss": 2.7328,
+      "loss": 2.7805,
       "step": 4010
     },
     {
       "epoch": 12.07,
-      "grad_norm": 5.718949794769287,
+      "grad_norm": 7.412511348724365,
       "learning_rate": 9.597597597597599e-06,
-      "loss": 2.7221,
+      "loss": 2.7722,
       "step": 4020
     },
     {
       "epoch": 12.1,
-      "grad_norm": 8.002398490905762,
+      "grad_norm": 6.2292938232421875,
       "learning_rate": 9.596596596596597e-06,
-      "loss": 2.7419,
+      "loss": 2.7922,
       "step": 4030
     },
     {
       "epoch": 12.13,
-      "grad_norm": 6.804090976715088,
+      "grad_norm": 8.190457344055176,
       "learning_rate": 9.595595595595596e-06,
-      "loss": 2.6489,
+      "loss": 2.7141,
       "step": 4040
     },
     {
       "epoch": 12.16,
-      "grad_norm": 10.463323593139648,
+      "grad_norm": 8.168481826782227,
       "learning_rate": 9.594594594594594e-06,
-      "loss": 2.716,
+      "loss": 2.7583,
       "step": 4050
     },
     {
       "epoch": 12.19,
-      "grad_norm": 5.688462257385254,
+      "grad_norm": 7.461695671081543,
       "learning_rate": 9.593593593593595e-06,
-      "loss": 2.7559,
+      "loss": 2.8191,
       "step": 4060
     },
     {
       "epoch": 12.22,
-      "grad_norm": 8.022812843322754,
+      "grad_norm": 8.524691581726074,
       "learning_rate": 9.592592592592593e-06,
-      "loss": 2.7438,
+      "loss": 2.7834,
       "step": 4070
     },
     {
       "epoch": 12.25,
-      "grad_norm": 5.698502063751221,
+      "grad_norm": 6.868655681610107,
       "learning_rate": 9.591591591591592e-06,
-      "loss": 2.6703,
+      "loss": 2.719,
       "step": 4080
     },
     {
       "epoch": 12.28,
-      "grad_norm": 10.163275718688965,
+      "grad_norm": 7.924734592437744,
       "learning_rate": 9.590590590590592e-06,
-      "loss": 2.7516,
+      "loss": 2.8053,
       "step": 4090
     },
     {
       "epoch": 12.31,
-      "grad_norm": 7.676804542541504,
+      "grad_norm": 9.868878364562988,
       "learning_rate": 9.58958958958959e-06,
-      "loss": 2.6583,
+      "loss": 2.7122,
       "step": 4100
     },
     {
       "epoch": 12.34,
-      "grad_norm": 7.096865653991699,
+      "grad_norm": 8.838659286499023,
       "learning_rate": 9.58858858858859e-06,
-      "loss": 2.6878,
+      "loss": 2.7423,
       "step": 4110
     },
     {
       "epoch": 12.37,
-      "grad_norm": 9.293025016784668,
+      "grad_norm": 10.636224746704102,
       "learning_rate": 9.587587587587588e-06,
-      "loss": 2.6779,
+      "loss": 2.7403,
       "step": 4120
     },
     {
       "epoch": 12.4,
-      "grad_norm": 9.587699890136719,
+      "grad_norm": 10.970544815063477,
       "learning_rate": 9.586586586586586e-06,
-      "loss": 2.6862,
+      "loss": 2.739,
       "step": 4130
     },
     {
       "epoch": 12.43,
-      "grad_norm": 5.810993194580078,
+      "grad_norm": 10.531293869018555,
       "learning_rate": 9.585585585585587e-06,
-      "loss": 2.6974,
+      "loss": 2.745,
       "step": 4140
     },
     {
       "epoch": 12.46,
-      "grad_norm": 8.383797645568848,
+      "grad_norm": 11.773600578308105,
       "learning_rate": 9.584584584584585e-06,
-      "loss": 2.6993,
+      "loss": 2.7551,
       "step": 4150
     },
     {
       "epoch": 12.49,
-      "grad_norm": 4.7015814781188965,
+      "grad_norm": 7.252423286437988,
       "learning_rate": 9.583583583583584e-06,
-      "loss": 2.6559,
+      "loss": 2.7164,
       "step": 4160
     },
     {
       "epoch": 12.52,
-      "grad_norm": 6.272876739501953,
+      "grad_norm": 6.621025085449219,
       "learning_rate": 9.582582582582584e-06,
-      "loss": 2.6441,
+      "loss": 2.7095,
       "step": 4170
     },
     {
       "epoch": 12.55,
-      "grad_norm": 9.866080284118652,
+      "grad_norm": 9.394515037536621,
       "learning_rate": 9.581581581581583e-06,
-      "loss": 2.6545,
+      "loss": 2.7038,
       "step": 4180
     },
     {
       "epoch": 12.58,
-      "grad_norm": 8.838409423828125,
+      "grad_norm": 12.874518394470215,
       "learning_rate": 9.580580580580581e-06,
-      "loss": 2.6319,
+      "loss": 2.6767,
       "step": 4190
     },
     {
       "epoch": 12.61,
-      "grad_norm": 5.951855659484863,
+      "grad_norm": 8.810815811157227,
       "learning_rate": 9.57957957957958e-06,
-      "loss": 2.6368,
+      "loss": 2.6928,
       "step": 4200
     },
     {
       "epoch": 12.64,
-      "grad_norm": 9.317614555358887,
+      "grad_norm": 10.79298210144043,
       "learning_rate": 9.578578578578579e-06,
-      "loss": 2.6422,
+      "loss": 2.6914,
       "step": 4210
     },
     {
       "epoch": 12.67,
-      "grad_norm": 12.077845573425293,
+      "grad_norm": 6.526676654815674,
       "learning_rate": 9.577577577577579e-06,
-      "loss": 2.6755,
+      "loss": 2.7283,
       "step": 4220
     },
     {
       "epoch": 12.7,
-      "grad_norm": 7.127326965332031,
+      "grad_norm": 9.200963973999023,
       "learning_rate": 9.576576576576578e-06,
-      "loss": 2.6166,
+      "loss": 2.6644,
       "step": 4230
     },
     {
       "epoch": 12.73,
-      "grad_norm": 6.233386039733887,
+      "grad_norm": 6.476570129394531,
       "learning_rate": 9.575575575575576e-06,
-      "loss": 2.6123,
+      "loss": 2.6617,
       "step": 4240
     },
     {
       "epoch": 12.76,
-      "grad_norm": 7.255130290985107,
+      "grad_norm": 11.802742958068848,
       "learning_rate": 9.574574574574575e-06,
-      "loss": 2.6476,
+      "loss": 2.7084,
       "step": 4250
     },
     {
       "epoch": 12.79,
-      "grad_norm": 7.743804931640625,
+      "grad_norm": 7.965548038482666,
       "learning_rate": 9.573573573573575e-06,
-      "loss": 2.5899,
+      "loss": 2.6339,
       "step": 4260
     },
     {
       "epoch": 12.82,
-      "grad_norm": 6.8271894454956055,
+      "grad_norm": 9.362634658813477,
       "learning_rate": 9.572572572572574e-06,
-      "loss": 2.6953,
+      "loss": 2.7521,
       "step": 4270
     },
     {
       "epoch": 12.85,
-      "grad_norm": 5.806545734405518,
+      "grad_norm": 6.151357650756836,
       "learning_rate": 9.571571571571572e-06,
-      "loss": 2.6908,
+      "loss": 2.7304,
       "step": 4280
     },
     {
       "epoch": 12.88,
-      "grad_norm": 6.065675735473633,
+      "grad_norm": 10.488903045654297,
       "learning_rate": 9.57057057057057e-06,
-      "loss": 2.5294,
+      "loss": 2.5854,
       "step": 4290
     },
     {
       "epoch": 12.91,
-      "grad_norm": 9.87260913848877,
+      "grad_norm": 6.467126846313477,
       "learning_rate": 9.56956956956957e-06,
-      "loss": 2.585,
+      "loss": 2.6427,
       "step": 4300
     },
     {
       "epoch": 12.94,
-      "grad_norm": 6.170048713684082,
+      "grad_norm": 8.475200653076172,
       "learning_rate": 9.56856856856857e-06,
-      "loss": 2.5926,
+      "loss": 2.6542,
       "step": 4310
     },
     {
       "epoch": 12.97,
-      "grad_norm": 8.0000581741333,
+      "grad_norm": 13.154068946838379,
       "learning_rate": 9.567567567567568e-06,
-      "loss": 2.6365,
+      "loss": 2.6816,
       "step": 4320
     },
     {
       "epoch": 13.0,
-      "eval_accuracy": 0.5173,
-      "eval_loss": 2.040445327758789,
-      "eval_runtime": 5.2866,
-      "eval_samples_per_second": 1891.579,
-      "eval_steps_per_second": 7.566,
+      "eval_accuracy": 0.5022,
+      "eval_loss": 2.095287322998047,
+      "eval_runtime": 5.2184,
+      "eval_samples_per_second": 1916.284,
+      "eval_steps_per_second": 7.665,
       "step": 4329
     },
     {
       "epoch": 13.0,
-      "grad_norm": 11.766575813293457,
+      "grad_norm": 7.427252769470215,
       "learning_rate": 9.566566566566567e-06,
-      "loss": 2.7903,
+      "loss": 2.8334,
       "step": 4330
     },
     {
       "epoch": 13.03,
-      "grad_norm": 11.979348182678223,
+      "grad_norm": 11.216113090515137,
       "learning_rate": 9.565565565565567e-06,
-      "loss": 2.5494,
+      "loss": 2.5996,
       "step": 4340
     },
     {
       "epoch": 13.06,
-      "grad_norm": 9.183178901672363,
+      "grad_norm": 11.334611892700195,
       "learning_rate": 9.564564564564566e-06,
-      "loss": 2.6042,
+      "loss": 2.6549,
       "step": 4350
     },
     {
       "epoch": 13.09,
-      "grad_norm": 5.910065650939941,
+      "grad_norm": 8.315796852111816,
       "learning_rate": 9.563563563563564e-06,
-      "loss": 2.6044,
+      "loss": 2.6638,
       "step": 4360
     },
     {
       "epoch": 13.12,
-      "grad_norm": 6.736372947692871,
+      "grad_norm": 9.011630058288574,
       "learning_rate": 9.562562562562563e-06,
-      "loss": 2.5704,
+      "loss": 2.6239,
       "step": 4370
     },
     {
       "epoch": 13.15,
-      "grad_norm": 7.387451648712158,
+      "grad_norm": 9.918655395507812,
       "learning_rate": 9.561561561561562e-06,
-      "loss": 2.5644,
+      "loss": 2.6192,
       "step": 4380
     },
     {
       "epoch": 13.18,
-      "grad_norm": 9.483909606933594,
+      "grad_norm": 7.569738864898682,
       "learning_rate": 9.560560560560562e-06,
-      "loss": 2.5851,
+      "loss": 2.6401,
       "step": 4390
     },
     {
       "epoch": 13.21,
-      "grad_norm": 12.600447654724121,
+      "grad_norm": 9.448251724243164,
       "learning_rate": 9.55955955955956e-06,
-      "loss": 2.6272,
+      "loss": 2.6669,
       "step": 4400
     },
     {
       "epoch": 13.24,
-      "grad_norm": 7.258604526519775,
+      "grad_norm": 9.664206504821777,
       "learning_rate": 9.558558558558559e-06,
-      "loss": 2.5459,
+      "loss": 2.5921,
       "step": 4410
     },
     {
       "epoch": 13.27,
-      "grad_norm": 9.401047706604004,
+      "grad_norm": 9.103913307189941,
       "learning_rate": 9.55755755755756e-06,
-      "loss": 2.6259,
+      "loss": 2.6687,
       "step": 4420
     },
     {
       "epoch": 13.3,
-      "grad_norm": 8.780267715454102,
+      "grad_norm": 10.875816345214844,
       "learning_rate": 9.556556556556558e-06,
-      "loss": 2.5441,
+      "loss": 2.594,
       "step": 4430
     },
     {
       "epoch": 13.33,
-      "grad_norm": 9.46339225769043,
+      "grad_norm": 7.2640910148620605,
       "learning_rate": 9.555555555555556e-06,
-      "loss": 2.5559,
+      "loss": 2.5997,
       "step": 4440
     },
     {
       "epoch": 13.36,
-      "grad_norm": 9.857782363891602,
+      "grad_norm": 9.848566055297852,
       "learning_rate": 9.554554554554555e-06,
-      "loss": 2.5429,
+      "loss": 2.5744,
       "step": 4450
     },
     {
       "epoch": 13.39,
-      "grad_norm": 8.497522354125977,
+      "grad_norm": 9.162285804748535,
       "learning_rate": 9.553553553553554e-06,
-      "loss": 2.6026,
+      "loss": 2.6557,
       "step": 4460
     },
     {
       "epoch": 13.42,
-      "grad_norm": 7.196791648864746,
+      "grad_norm": 6.250080585479736,
       "learning_rate": 9.552552552552552e-06,
-      "loss": 2.563,
+      "loss": 2.6127,
       "step": 4470
     },
     {
       "epoch": 13.45,
-      "grad_norm": 6.099296569824219,
+      "grad_norm": 11.269085884094238,
       "learning_rate": 9.551551551551553e-06,
-      "loss": 2.5931,
+      "loss": 2.6559,
       "step": 4480
     },
     {
       "epoch": 13.48,
-      "grad_norm": 9.927326202392578,
+      "grad_norm": 7.2637176513671875,
       "learning_rate": 9.550550550550551e-06,
-      "loss": 2.5146,
+      "loss": 2.5433,
       "step": 4490
     },
     {
       "epoch": 13.51,
-      "grad_norm": 6.045947551727295,
+      "grad_norm": 8.856646537780762,
       "learning_rate": 9.54954954954955e-06,
-      "loss": 2.5783,
+      "loss": 2.6043,
       "step": 4500
     },
     {
       "epoch": 13.54,
-      "grad_norm": 7.737361431121826,
+      "grad_norm": 9.173357009887695,
       "learning_rate": 9.54854854854855e-06,
-      "loss": 2.5914,
+      "loss": 2.6273,
       "step": 4510
     },
     {
       "epoch": 13.57,
-      "grad_norm": 8.254207611083984,
+      "grad_norm": 6.987086296081543,
       "learning_rate": 9.547547547547549e-06,
-      "loss": 2.5565,
+      "loss": 2.6144,
       "step": 4520
     },
     {
       "epoch": 13.6,
-      "grad_norm": 6.97804594039917,
+      "grad_norm": 7.6146745681762695,
       "learning_rate": 9.546546546546547e-06,
-      "loss": 2.5083,
+      "loss": 2.5451,
       "step": 4530
     },
     {
       "epoch": 13.63,
-      "grad_norm": 5.862122535705566,
+      "grad_norm": 10.84038257598877,
       "learning_rate": 9.545545545545546e-06,
-      "loss": 2.5707,
+      "loss": 2.6036,
       "step": 4540
     },
     {
       "epoch": 13.66,
-      "grad_norm": 7.2387847900390625,
+      "grad_norm": 7.272514820098877,
       "learning_rate": 9.544544544544544e-06,
-      "loss": 2.5625,
+      "loss": 2.6056,
       "step": 4550
     },
     {
       "epoch": 13.69,
-      "grad_norm": 9.292677879333496,
+      "grad_norm": 8.79472827911377,
       "learning_rate": 9.543543543543545e-06,
-      "loss": 2.4988,
+      "loss": 2.5358,
       "step": 4560
     },
     {
       "epoch": 13.72,
-      "grad_norm": 5.622021198272705,
+      "grad_norm": 7.347231388092041,
       "learning_rate": 9.542542542542543e-06,
-      "loss": 2.5066,
+      "loss": 2.566,
       "step": 4570
     },
     {
       "epoch": 13.75,
-      "grad_norm": 5.3578362464904785,
+      "grad_norm": 7.245388984680176,
       "learning_rate": 9.541541541541542e-06,
-      "loss": 2.4769,
+      "loss": 2.5289,
       "step": 4580
     },
     {
       "epoch": 13.78,
-      "grad_norm": 7.696930885314941,
+      "grad_norm": 8.159917831420898,
       "learning_rate": 9.540540540540542e-06,
-      "loss": 2.4968,
+      "loss": 2.5469,
       "step": 4590
     },
     {
       "epoch": 13.81,
-      "grad_norm": 7.834680557250977,
+      "grad_norm": 10.765388488769531,
       "learning_rate": 9.53953953953954e-06,
-      "loss": 2.5467,
+      "loss": 2.5906,
       "step": 4600
     },
     {
       "epoch": 13.84,
-      "grad_norm": 7.833067417144775,
+      "grad_norm": 8.658844947814941,
       "learning_rate": 9.53853853853854e-06,
-      "loss": 2.5277,
+      "loss": 2.5803,
       "step": 4610
     },
     {
       "epoch": 13.87,
-      "grad_norm": 11.307621955871582,
+      "grad_norm": 7.347329139709473,
       "learning_rate": 9.537537537537538e-06,
-      "loss": 2.5227,
+      "loss": 2.5821,
       "step": 4620
     },
     {
       "epoch": 13.9,
-      "grad_norm": 7.471169471740723,
+      "grad_norm": 14.07579231262207,
       "learning_rate": 9.536536536536537e-06,
-      "loss": 2.5309,
+      "loss": 2.5654,
       "step": 4630
     },
     {
       "epoch": 13.93,
-      "grad_norm": 9.578874588012695,
+      "grad_norm": 8.736212730407715,
       "learning_rate": 9.535535535535537e-06,
-      "loss": 2.5169,
+      "loss": 2.5543,
       "step": 4640
     },
     {
       "epoch": 13.96,
-      "grad_norm": 6.651345729827881,
+      "grad_norm": 8.474813461303711,
       "learning_rate": 9.534534534534535e-06,
-      "loss": 2.4457,
+      "loss": 2.4797,
       "step": 4650
     },
     {
       "epoch": 13.99,
-      "grad_norm": 8.052266120910645,
+      "grad_norm": 6.664093494415283,
       "learning_rate": 9.533533533533534e-06,
-      "loss": 2.4545,
+      "loss": 2.5017,
       "step": 4660
     },
     {
       "epoch": 14.0,
-      "eval_accuracy": 0.5402,
-      "eval_loss": 1.9062016010284424,
-      "eval_runtime": 5.3946,
-      "eval_samples_per_second": 1853.713,
-      "eval_steps_per_second": 7.415,
+      "eval_accuracy": 0.5302,
+      "eval_loss": 1.9548472166061401,
+      "eval_runtime": 5.2404,
+      "eval_samples_per_second": 1908.244,
+      "eval_steps_per_second": 7.633,
       "step": 4662
     },
     {
       "epoch": 14.02,
-      "grad_norm": 7.185265064239502,
+      "grad_norm": 7.381178855895996,
       "learning_rate": 9.532532532532534e-06,
-      "loss": 2.6056,
+      "loss": 2.6408,
       "step": 4670
     },
     {
       "epoch": 14.05,
-      "grad_norm": 12.770059585571289,
+      "grad_norm": 11.721305847167969,
       "learning_rate": 9.531531531531533e-06,
-      "loss": 2.4392,
+      "loss": 2.4856,
       "step": 4680
     },
     {
       "epoch": 14.08,
-      "grad_norm": 6.75191593170166,
+      "grad_norm": 6.465998649597168,
       "learning_rate": 9.530530530530532e-06,
-      "loss": 2.4047,
+      "loss": 2.4644,
       "step": 4690
     },
     {
       "epoch": 14.11,
-      "grad_norm": 5.679729461669922,
+      "grad_norm": 8.37296199798584,
       "learning_rate": 9.52952952952953e-06,
-      "loss": 2.4728,
+      "loss": 2.5174,
       "step": 4700
     },
     {
       "epoch": 14.14,
-      "grad_norm": 5.110313892364502,
+      "grad_norm": 6.373126029968262,
       "learning_rate": 9.528528528528529e-06,
-      "loss": 2.491,
+      "loss": 2.5312,
       "step": 4710
     },
     {
       "epoch": 14.17,
-      "grad_norm": 9.266707420349121,
+      "grad_norm": 8.110363960266113,
       "learning_rate": 9.527527527527527e-06,
-      "loss": 2.4575,
+      "loss": 2.5116,
       "step": 4720
     },
     {
       "epoch": 14.2,
-      "grad_norm": 5.694037437438965,
+      "grad_norm": 20.729318618774414,
       "learning_rate": 9.526526526526528e-06,
-      "loss": 2.4975,
+      "loss": 2.5326,
       "step": 4730
     },
     {
       "epoch": 14.23,
-      "grad_norm": 7.688876152038574,
+      "grad_norm": 11.035715103149414,
       "learning_rate": 9.525525525525526e-06,
-      "loss": 2.5477,
+      "loss": 2.5872,
       "step": 4740
     },
     {
       "epoch": 14.26,
-      "grad_norm": 5.518246173858643,
+      "grad_norm": 9.358081817626953,
       "learning_rate": 9.524524524524525e-06,
-      "loss": 2.4093,
+      "loss": 2.4545,
       "step": 4750
     },
     {
       "epoch": 14.29,
-      "grad_norm": 5.213392734527588,
+      "grad_norm": 9.555889129638672,
       "learning_rate": 9.523523523523525e-06,
-      "loss": 2.4649,
+      "loss": 2.5077,
       "step": 4760
     },
     {
       "epoch": 14.32,
-      "grad_norm": 6.395495414733887,
+      "grad_norm": 6.779592037200928,
       "learning_rate": 9.522522522522524e-06,
-      "loss": 2.4486,
+      "loss": 2.4918,
       "step": 4770
     },
     {
       "epoch": 14.35,
-      "grad_norm": 8.322802543640137,
+      "grad_norm": 9.891965866088867,
       "learning_rate": 9.521521521521522e-06,
-      "loss": 2.4938,
+      "loss": 2.5283,
       "step": 4780
     },
     {
       "epoch": 14.38,
-      "grad_norm": 6.8021159172058105,
+      "grad_norm": 8.420459747314453,
       "learning_rate": 9.520520520520521e-06,
-      "loss": 2.4622,
+      "loss": 2.4984,
       "step": 4790
     },
     {
       "epoch": 14.41,
-      "grad_norm": 6.663272380828857,
+      "grad_norm": 8.168742179870605,
       "learning_rate": 9.51951951951952e-06,
-      "loss": 2.5037,
+      "loss": 2.541,
       "step": 4800
     },
     {
       "epoch": 14.44,
-      "grad_norm": 5.993630409240723,
+      "grad_norm": 7.607645511627197,
       "learning_rate": 9.51851851851852e-06,
-      "loss": 2.4807,
+      "loss": 2.525,
       "step": 4810
     },
     {
       "epoch": 14.47,
-      "grad_norm": 8.784486770629883,
+      "grad_norm": 11.1437406539917,
       "learning_rate": 9.517517517517518e-06,
-      "loss": 2.5049,
+      "loss": 2.5409,
       "step": 4820
     },
     {
       "epoch": 14.5,
-      "grad_norm": 5.584054470062256,
+      "grad_norm": 6.276561737060547,
       "learning_rate": 9.516516516516517e-06,
-      "loss": 2.4859,
+      "loss": 2.5267,
       "step": 4830
     },
     {
       "epoch": 14.53,
-      "grad_norm": 8.39543628692627,
+      "grad_norm": 9.398030281066895,
       "learning_rate": 9.515515515515517e-06,
-      "loss": 2.4442,
+      "loss": 2.4744,
       "step": 4840
     },
     {
       "epoch": 14.56,
-      "grad_norm": 5.909639835357666,
+      "grad_norm": 8.579767227172852,
       "learning_rate": 9.514514514514516e-06,
-      "loss": 2.4411,
+      "loss": 2.4826,
       "step": 4850
     },
     {
       "epoch": 14.59,
-      "grad_norm": 5.571308612823486,
+      "grad_norm": 5.987852096557617,
       "learning_rate": 9.513513513513514e-06,
-      "loss": 2.4029,
+      "loss": 2.4564,
       "step": 4860
     },
     {
       "epoch": 14.62,
-      "grad_norm": 7.9407830238342285,
+      "grad_norm": 7.89351749420166,
       "learning_rate": 9.512512512512513e-06,
-      "loss": 2.4547,
+      "loss": 2.4833,
       "step": 4870
     },
     {
       "epoch": 14.65,
-      "grad_norm": 9.955803871154785,
+      "grad_norm": 8.504823684692383,
       "learning_rate": 9.511511511511512e-06,
-      "loss": 2.4833,
+      "loss": 2.5157,
       "step": 4880
     },
     {
       "epoch": 14.68,
-      "grad_norm": 5.949026584625244,
+      "grad_norm": 9.139737129211426,
       "learning_rate": 9.510510510510512e-06,
-      "loss": 2.4653,
+      "loss": 2.5024,
       "step": 4890
     },
     {
       "epoch": 14.71,
-      "grad_norm": 9.606804847717285,
+      "grad_norm": 6.705386638641357,
       "learning_rate": 9.50950950950951e-06,
-      "loss": 2.3761,
+      "loss": 2.4188,
       "step": 4900
     },
     {
       "epoch": 14.74,
-      "grad_norm": 5.222976207733154,
+      "grad_norm": 7.481086730957031,
       "learning_rate": 9.508508508508509e-06,
-      "loss": 2.4076,
+      "loss": 2.4319,
       "step": 4910
     },
     {
       "epoch": 14.77,
-      "grad_norm": 7.917681694030762,
+      "grad_norm": 6.1258392333984375,
       "learning_rate": 9.507507507507508e-06,
-      "loss": 2.3933,
+      "loss": 2.4351,
       "step": 4920
     },
     {
       "epoch": 14.8,
-      "grad_norm": 6.634172439575195,
+      "grad_norm": 7.586241722106934,
       "learning_rate": 9.506506506506508e-06,
-      "loss": 2.4278,
+      "loss": 2.4765,
       "step": 4930
     },
     {
       "epoch": 14.83,
-      "grad_norm": 8.443916320800781,
+      "grad_norm": 5.622743606567383,
       "learning_rate": 9.505505505505507e-06,
-      "loss": 2.3765,
+      "loss": 2.4243,
       "step": 4940
     },
     {
       "epoch": 14.86,
-      "grad_norm": 8.051191329956055,
+      "grad_norm": 9.771998405456543,
       "learning_rate": 9.504504504504505e-06,
-      "loss": 2.4829,
+      "loss": 2.5277,
       "step": 4950
     },
     {
       "epoch": 14.89,
-      "grad_norm": 9.006173133850098,
+      "grad_norm": 9.455774307250977,
       "learning_rate": 9.503503503503504e-06,
-      "loss": 2.4222,
+      "loss": 2.4557,
       "step": 4960
     },
     {
       "epoch": 14.92,
-      "grad_norm": 6.763026714324951,
+      "grad_norm": 6.484610557556152,
       "learning_rate": 9.502502502502502e-06,
-      "loss": 2.3907,
+      "loss": 2.4311,
       "step": 4970
     },
     {
       "epoch": 14.95,
-      "grad_norm": 6.036118030548096,
+      "grad_norm": 9.916911125183105,
       "learning_rate": 9.501501501501503e-06,
-      "loss": 2.4189,
+      "loss": 2.4627,
       "step": 4980
     },
     {
       "epoch": 14.98,
-      "grad_norm": 5.997125148773193,
+      "grad_norm": 8.295943260192871,
       "learning_rate": 9.500500500500501e-06,
-      "loss": 2.3891,
+      "loss": 2.4221,
       "step": 4990
     },
     {
       "epoch": 15.0,
-      "eval_accuracy": 0.5673,
-      "eval_loss": 1.7760998010635376,
-      "eval_runtime": 5.2718,
-      "eval_samples_per_second": 1896.879,
-      "eval_steps_per_second": 7.588,
+      "eval_accuracy": 0.5538,
+      "eval_loss": 1.8190256357192993,
+      "eval_runtime": 5.2176,
+      "eval_samples_per_second": 1916.582,
+      "eval_steps_per_second": 7.666,
       "step": 4995
     },
     {
       "epoch": 15.02,
-      "grad_norm": 7.087186813354492,
+      "grad_norm": 8.873432159423828,
       "learning_rate": 9.4994994994995e-06,
-      "loss": 2.4401,
+      "loss": 2.4732,
       "step": 5000
     },
     {
       "epoch": 15.05,
-      "grad_norm": 9.004009246826172,
+      "grad_norm": 7.125927925109863,
       "learning_rate": 9.4984984984985e-06,
-      "loss": 2.3844,
+      "loss": 2.4299,
       "step": 5010
     },
     {
       "epoch": 15.08,
-      "grad_norm": 10.013044357299805,
+      "grad_norm": 7.490790843963623,
       "learning_rate": 9.497497497497499e-06,
-      "loss": 2.3984,
+      "loss": 2.4459,
       "step": 5020
     },
     {
       "epoch": 15.11,
-      "grad_norm": 8.390071868896484,
+      "grad_norm": 6.717255592346191,
       "learning_rate": 9.496496496496497e-06,
-      "loss": 2.409,
+      "loss": 2.4345,
       "step": 5030
     },
     {
       "epoch": 15.14,
-      "grad_norm": 13.679229736328125,
+      "grad_norm": 8.988885879516602,
       "learning_rate": 9.495495495495496e-06,
-      "loss": 2.4119,
+      "loss": 2.4534,
       "step": 5040
     },
     {
       "epoch": 15.17,
-      "grad_norm": 6.602577209472656,
+      "grad_norm": 8.598869323730469,
       "learning_rate": 9.494494494494494e-06,
-      "loss": 2.32,
+      "loss": 2.356,
       "step": 5050
     },
     {
       "epoch": 15.2,
-      "grad_norm": 7.045469760894775,
+      "grad_norm": 9.799782752990723,
       "learning_rate": 9.493493493493495e-06,
-      "loss": 2.4077,
+      "loss": 2.4376,
       "step": 5060
     },
     {
       "epoch": 15.23,
-      "grad_norm": 6.839366436004639,
+      "grad_norm": 7.557323932647705,
       "learning_rate": 9.492492492492493e-06,
-      "loss": 2.3848,
+      "loss": 2.4132,
       "step": 5070
     },
     {
       "epoch": 15.26,
-      "grad_norm": 6.932955741882324,
+      "grad_norm": 5.448122024536133,
       "learning_rate": 9.491491491491492e-06,
-      "loss": 2.3904,
+      "loss": 2.4215,
       "step": 5080
     },
     {
       "epoch": 15.29,
-      "grad_norm": 8.04763412475586,
+      "grad_norm": 8.58484935760498,
       "learning_rate": 9.490490490490492e-06,
-      "loss": 2.4206,
+      "loss": 2.4576,
       "step": 5090
     },
     {
       "epoch": 15.32,
-      "grad_norm": 7.625525951385498,
+      "grad_norm": 7.012596130371094,
       "learning_rate": 9.489489489489491e-06,
-      "loss": 2.3791,
+      "loss": 2.4122,
       "step": 5100
     },
     {
       "epoch": 15.35,
-      "grad_norm": 5.918097019195557,
+      "grad_norm": 7.188743591308594,
       "learning_rate": 9.48848848848849e-06,
-      "loss": 2.3665,
+      "loss": 2.3836,
       "step": 5110
     },
     {
       "epoch": 15.38,
-      "grad_norm": 7.602704048156738,
+      "grad_norm": 10.356210708618164,
       "learning_rate": 9.487487487487488e-06,
-      "loss": 2.3736,
+      "loss": 2.4034,
       "step": 5120
     },
     {
       "epoch": 15.41,
-      "grad_norm": 5.679454326629639,
+      "grad_norm": 6.1163434982299805,
       "learning_rate": 9.486486486486487e-06,
-      "loss": 2.3132,
+      "loss": 2.354,
       "step": 5130
     },
     {
       "epoch": 15.44,
-      "grad_norm": 6.988337516784668,
+      "grad_norm": 6.35807991027832,
       "learning_rate": 9.485485485485487e-06,
-      "loss": 2.3265,
+      "loss": 2.3538,
       "step": 5140
     },
     {
       "epoch": 15.47,
-      "grad_norm": 7.349186897277832,
+      "grad_norm": 7.396414279937744,
       "learning_rate": 9.484484484484486e-06,
-      "loss": 2.3779,
+      "loss": 2.3951,
       "step": 5150
     },
     {
       "epoch": 15.5,
-      "grad_norm": 8.78355884552002,
+      "grad_norm": 10.006298065185547,
       "learning_rate": 9.483483483483484e-06,
-      "loss": 2.3985,
+      "loss": 2.4317,
       "step": 5160
     },
     {
       "epoch": 15.53,
-      "grad_norm": 8.69871997833252,
+      "grad_norm": 10.393084526062012,
       "learning_rate": 9.482482482482483e-06,
-      "loss": 2.3352,
+      "loss": 2.3677,
       "step": 5170
     },
     {
       "epoch": 15.56,
-      "grad_norm": 6.758751392364502,
+      "grad_norm": 8.58001708984375,
       "learning_rate": 9.481481481481483e-06,
-      "loss": 2.2961,
+      "loss": 2.3335,
       "step": 5180
     },
     {
       "epoch": 15.59,
-      "grad_norm": 7.867177963256836,
+      "grad_norm": 6.8485798835754395,
       "learning_rate": 9.480480480480482e-06,
-      "loss": 2.3877,
+      "loss": 2.4167,
       "step": 5190
     },
     {
       "epoch": 15.62,
-      "grad_norm": 6.01539945602417,
+      "grad_norm": 6.350042819976807,
       "learning_rate": 9.47947947947948e-06,
-      "loss": 2.3625,
+      "loss": 2.384,
       "step": 5200
     },
     {
       "epoch": 15.65,
-      "grad_norm": 7.915289402008057,
+      "grad_norm": 6.194655895233154,
       "learning_rate": 9.478478478478479e-06,
-      "loss": 2.2537,
+      "loss": 2.2898,
       "step": 5210
     },
     {
       "epoch": 15.68,
-      "grad_norm": 6.425355911254883,
+      "grad_norm": 5.900041580200195,
       "learning_rate": 9.477477477477477e-06,
-      "loss": 2.3079,
+      "loss": 2.3247,
       "step": 5220
     },
     {
       "epoch": 15.71,
-      "grad_norm": 10.054742813110352,
+      "grad_norm": 9.351147651672363,
       "learning_rate": 9.476476476476478e-06,
-      "loss": 2.3493,
+      "loss": 2.3947,
       "step": 5230
     },
     {
       "epoch": 15.74,
-      "grad_norm": 6.709542274475098,
+      "grad_norm": 8.427316665649414,
       "learning_rate": 9.475475475475476e-06,
-      "loss": 2.3,
+      "loss": 2.3266,
       "step": 5240
     },
     {
       "epoch": 15.77,
-      "grad_norm": 8.417074203491211,
+      "grad_norm": 9.767661094665527,
       "learning_rate": 9.474474474474475e-06,
-      "loss": 2.3793,
+      "loss": 2.4046,
       "step": 5250
     },
     {
       "epoch": 15.8,
-      "grad_norm": 7.198749542236328,
+      "grad_norm": 6.926217079162598,
       "learning_rate": 9.473473473473475e-06,
-      "loss": 2.3394,
+      "loss": 2.3749,
       "step": 5260
     },
     {
       "epoch": 15.83,
-      "grad_norm": 11.865601539611816,
+      "grad_norm": 10.24268913269043,
       "learning_rate": 9.472472472472474e-06,
-      "loss": 2.2699,
+      "loss": 2.2885,
       "step": 5270
     },
     {
       "epoch": 15.86,
-      "grad_norm": 6.732371807098389,
+      "grad_norm": 9.303874969482422,
       "learning_rate": 9.471471471471472e-06,
-      "loss": 2.2618,
+      "loss": 2.2914,
       "step": 5280
     },
     {
       "epoch": 15.89,
-      "grad_norm": 6.7344183921813965,
+      "grad_norm": 7.3029961585998535,
       "learning_rate": 9.470470470470471e-06,
-      "loss": 2.3732,
+      "loss": 2.3959,
       "step": 5290
     },
     {
       "epoch": 15.92,
-      "grad_norm": 7.002033710479736,
+      "grad_norm": 7.979551315307617,
       "learning_rate": 9.46946946946947e-06,
-      "loss": 2.3463,
+      "loss": 2.3814,
       "step": 5300
     },
     {
       "epoch": 15.95,
-      "grad_norm": 6.745180606842041,
+      "grad_norm": 8.222173690795898,
       "learning_rate": 9.46846846846847e-06,
-      "loss": 2.3314,
+      "loss": 2.3408,
       "step": 5310
     },
     {
       "epoch": 15.98,
-      "grad_norm": 8.396537780761719,
+      "grad_norm": 6.876596450805664,
       "learning_rate": 9.467467467467468e-06,
-      "loss": 2.2835,
+      "loss": 2.3111,
       "step": 5320
     },
     {
       "epoch": 16.0,
-      "eval_accuracy": 0.5783,
-      "eval_loss": 1.6823691129684448,
-      "eval_runtime": 5.3969,
-      "eval_samples_per_second": 1852.915,
-      "eval_steps_per_second": 7.412,
+      "eval_accuracy": 0.5746,
+      "eval_loss": 1.7099714279174805,
+      "eval_runtime": 5.4598,
+      "eval_samples_per_second": 1831.565,
+      "eval_steps_per_second": 7.326,
       "step": 5328
     },
     {
       "epoch": 16.01,
-      "grad_norm": 7.378887176513672,
+      "grad_norm": 6.71311616897583,
       "learning_rate": 9.466466466466467e-06,
-      "loss": 2.4559,
+      "loss": 2.4711,
       "step": 5330
     },
     {
       "epoch": 16.04,
-      "grad_norm": 7.3596367835998535,
+      "grad_norm": 7.3963212966918945,
       "learning_rate": 9.465465465465467e-06,
-      "loss": 2.3242,
+      "loss": 2.3406,
       "step": 5340
     },
     {
       "epoch": 16.07,
-      "grad_norm": 7.283603191375732,
+      "grad_norm": 9.438785552978516,
       "learning_rate": 9.464464464464466e-06,
-      "loss": 2.2966,
+      "loss": 2.335,
       "step": 5350
     },
     {
       "epoch": 16.1,
-      "grad_norm": 8.8494873046875,
+      "grad_norm": 8.670868873596191,
       "learning_rate": 9.463463463463464e-06,
-      "loss": 2.2914,
+      "loss": 2.322,
       "step": 5360
     },
     {
       "epoch": 16.13,
-      "grad_norm": 12.0491304397583,
+      "grad_norm": 11.969527244567871,
       "learning_rate": 9.462462462462463e-06,
-      "loss": 2.3301,
+      "loss": 2.3644,
       "step": 5370
     },
     {
       "epoch": 16.16,
-      "grad_norm": 8.241227149963379,
+      "grad_norm": 7.7007012367248535,
       "learning_rate": 9.461461461461462e-06,
-      "loss": 2.3336,
+      "loss": 2.3576,
       "step": 5380
     },
     {
       "epoch": 16.19,
-      "grad_norm": 7.086030006408691,
+      "grad_norm": 8.905147552490234,
       "learning_rate": 9.46046046046046e-06,
-      "loss": 2.3016,
+      "loss": 2.3203,
       "step": 5390
     },
     {
       "epoch": 16.22,
-      "grad_norm": 6.14713716506958,
+      "grad_norm": 9.54321575164795,
       "learning_rate": 9.45945945945946e-06,
-      "loss": 2.2976,
+      "loss": 2.3193,
       "step": 5400
     },
     {
       "epoch": 16.25,
-      "grad_norm": 10.391640663146973,
+      "grad_norm": 7.4690680503845215,
       "learning_rate": 9.458458458458459e-06,
-      "loss": 2.314,
+      "loss": 2.3443,
       "step": 5410
     },
     {
       "epoch": 16.28,
-      "grad_norm": 6.29721212387085,
+      "grad_norm": 8.359950065612793,
       "learning_rate": 9.457457457457458e-06,
-      "loss": 2.3216,
+      "loss": 2.352,
       "step": 5420
     },
     {
       "epoch": 16.31,
-      "grad_norm": 7.678361415863037,
+      "grad_norm": 8.397698402404785,
       "learning_rate": 9.456456456456458e-06,
-      "loss": 2.2152,
+      "loss": 2.234,
       "step": 5430
     },
     {
       "epoch": 16.34,
-      "grad_norm": 9.984953880310059,
+      "grad_norm": 11.055509567260742,
       "learning_rate": 9.455455455455457e-06,
-      "loss": 2.2728,
+      "loss": 2.3048,
       "step": 5440
     },
     {
       "epoch": 16.37,
-      "grad_norm": 6.304622650146484,
+      "grad_norm": 8.035820007324219,
       "learning_rate": 9.454454454454455e-06,
-      "loss": 2.2595,
+      "loss": 2.273,
       "step": 5450
     },
     {
       "epoch": 16.4,
-      "grad_norm": 6.2005391120910645,
+      "grad_norm": 8.977294921875,
       "learning_rate": 9.453453453453454e-06,
-      "loss": 2.3119,
+      "loss": 2.3399,
       "step": 5460
     },
     {
       "epoch": 16.43,
-      "grad_norm": 7.004530429840088,
+      "grad_norm": 8.035536766052246,
       "learning_rate": 9.452452452452452e-06,
-      "loss": 2.2909,
+      "loss": 2.3202,
       "step": 5470
     },
     {
       "epoch": 16.46,
-      "grad_norm": 10.848859786987305,
+      "grad_norm": 9.306693077087402,
       "learning_rate": 9.451451451451453e-06,
-      "loss": 2.3292,
+      "loss": 2.3531,
       "step": 5480
     },
     {
       "epoch": 16.49,
-      "grad_norm": 8.08767318725586,
+      "grad_norm": 7.660901069641113,
       "learning_rate": 9.450450450450451e-06,
-      "loss": 2.2876,
+      "loss": 2.3151,
       "step": 5490
     },
     {
       "epoch": 16.52,
-      "grad_norm": 7.411654472351074,
+      "grad_norm": 6.552356719970703,
       "learning_rate": 9.44944944944945e-06,
-      "loss": 2.2169,
+      "loss": 2.2429,
       "step": 5500
     },
     {
       "epoch": 16.55,
-      "grad_norm": 7.649465084075928,
+      "grad_norm": 9.214117050170898,
       "learning_rate": 9.44844844844845e-06,
-      "loss": 2.238,
+      "loss": 2.2562,
       "step": 5510
     },
     {
       "epoch": 16.58,
-      "grad_norm": 7.567043304443359,
+      "grad_norm": 8.2964448928833,
       "learning_rate": 9.447447447447449e-06,
-      "loss": 2.2463,
+      "loss": 2.2563,
       "step": 5520
     },
     {
       "epoch": 16.61,
-      "grad_norm": 9.848114967346191,
+      "grad_norm": 8.248534202575684,
       "learning_rate": 9.446446446446447e-06,
-      "loss": 2.2987,
+      "loss": 2.3121,
       "step": 5530
     },
     {
       "epoch": 16.64,
-      "grad_norm": 7.836221694946289,
+      "grad_norm": 10.03344440460205,
       "learning_rate": 9.445445445445446e-06,
-      "loss": 2.229,
+      "loss": 2.2435,
       "step": 5540
     },
     {
       "epoch": 16.67,
-      "grad_norm": 5.616601943969727,
+      "grad_norm": 8.94646167755127,
       "learning_rate": 9.444444444444445e-06,
-      "loss": 2.2253,
+      "loss": 2.2493,
       "step": 5550
     },
     {
       "epoch": 16.7,
-      "grad_norm": 9.972375869750977,
+      "grad_norm": 5.980799674987793,
       "learning_rate": 9.443443443443445e-06,
-      "loss": 2.2108,
+      "loss": 2.2368,
       "step": 5560
     },
     {
       "epoch": 16.73,
-      "grad_norm": 7.1031084060668945,
+      "grad_norm": 7.020594120025635,
       "learning_rate": 9.442442442442443e-06,
-      "loss": 2.2285,
+      "loss": 2.2578,
       "step": 5570
     },
     {
       "epoch": 16.76,
-      "grad_norm": 9.210076332092285,
+      "grad_norm": 9.757468223571777,
       "learning_rate": 9.441441441441442e-06,
-      "loss": 2.2332,
+      "loss": 2.2588,
       "step": 5580
     },
     {
       "epoch": 16.79,
-      "grad_norm": 9.152575492858887,
+      "grad_norm": 7.543290615081787,
       "learning_rate": 9.440440440440442e-06,
-      "loss": 2.2621,
+      "loss": 2.2898,
       "step": 5590
     },
     {
       "epoch": 16.82,
-      "grad_norm": 6.438522815704346,
+      "grad_norm": 9.36979866027832,
       "learning_rate": 9.439439439439441e-06,
-      "loss": 2.2539,
+      "loss": 2.2834,
       "step": 5600
     },
     {
       "epoch": 16.85,
-      "grad_norm": 12.351630210876465,
+      "grad_norm": 14.734661102294922,
       "learning_rate": 9.43843843843844e-06,
-      "loss": 2.2192,
+      "loss": 2.2453,
       "step": 5610
     },
     {
       "epoch": 16.88,
-      "grad_norm": 8.844488143920898,
+      "grad_norm": 11.429098129272461,
       "learning_rate": 9.437437437437438e-06,
-      "loss": 2.211,
+      "loss": 2.2387,
       "step": 5620
     },
     {
       "epoch": 16.91,
-      "grad_norm": 9.99282169342041,
+      "grad_norm": 7.669627666473389,
       "learning_rate": 9.436436436436437e-06,
-      "loss": 2.2067,
+      "loss": 2.2377,
       "step": 5630
     },
     {
       "epoch": 16.94,
-      "grad_norm": 8.712618827819824,
+      "grad_norm": 8.524465560913086,
       "learning_rate": 9.435435435435435e-06,
-      "loss": 2.209,
+      "loss": 2.245,
       "step": 5640
     },
     {
       "epoch": 16.97,
-      "grad_norm": 10.543987274169922,
+      "grad_norm": 8.986326217651367,
       "learning_rate": 9.434434434434436e-06,
-      "loss": 2.2399,
+      "loss": 2.2546,
       "step": 5650
     },
     {
       "epoch": 17.0,
-      "grad_norm": 9.302331924438477,
+      "grad_norm": 12.044620513916016,
       "learning_rate": 9.433433433433434e-06,
-      "loss": 2.3137,
+      "loss": 2.3419,
       "step": 5660
     },
     {
       "epoch": 17.0,
-      "eval_accuracy": 0.5932,
-      "eval_loss": 1.6064293384552002,
-      "eval_runtime": 5.5913,
-      "eval_samples_per_second": 1788.502,
-      "eval_steps_per_second": 7.154,
+      "eval_accuracy": 0.5876,
+      "eval_loss": 1.6310687065124512,
+      "eval_runtime": 5.6828,
+      "eval_samples_per_second": 1759.69,
+      "eval_steps_per_second": 7.039,
       "step": 5661
     },
     {
       "epoch": 17.03,
-      "grad_norm": 9.5619478225708,
+      "grad_norm": 9.59705924987793,
       "learning_rate": 9.432432432432433e-06,
-      "loss": 2.3417,
+      "loss": 2.367,
       "step": 5670
     },
     {
       "epoch": 17.06,
-      "grad_norm": 7.2541279792785645,
+      "grad_norm": 6.806575298309326,
       "learning_rate": 9.431431431431433e-06,
-      "loss": 2.2381,
+      "loss": 2.2572,
       "step": 5680
     },
     {
       "epoch": 17.09,
-      "grad_norm": 7.031679630279541,
+      "grad_norm": 6.294764041900635,
       "learning_rate": 9.430430430430432e-06,
-      "loss": 2.2458,
+      "loss": 2.2689,
       "step": 5690
     },
     {
       "epoch": 17.12,
-      "grad_norm": 12.70786190032959,
+      "grad_norm": 8.924921035766602,
       "learning_rate": 9.42942942942943e-06,
-      "loss": 2.1956,
+      "loss": 2.2164,
       "step": 5700
     },
     {
       "epoch": 17.15,
-      "grad_norm": 8.551438331604004,
+      "grad_norm": 6.646582126617432,
       "learning_rate": 9.428428428428429e-06,
-      "loss": 2.2298,
+      "loss": 2.2464,
       "step": 5710
     },
     {
       "epoch": 17.18,
-      "grad_norm": 10.262838363647461,
+      "grad_norm": 8.186441421508789,
       "learning_rate": 9.427427427427427e-06,
-      "loss": 2.2499,
+      "loss": 2.2663,
       "step": 5720
     },
     {
       "epoch": 17.21,
-      "grad_norm": 10.270269393920898,
+      "grad_norm": 9.534934997558594,
       "learning_rate": 9.426426426426428e-06,
-      "loss": 2.1885,
+      "loss": 2.201,
       "step": 5730
     },
     {
       "epoch": 17.24,
-      "grad_norm": 7.7437825202941895,
+      "grad_norm": 6.863450050354004,
       "learning_rate": 9.425425425425426e-06,
-      "loss": 2.2429,
+      "loss": 2.2655,
       "step": 5740
     },
     {
       "epoch": 17.27,
-      "grad_norm": 7.937880516052246,
+      "grad_norm": 5.386342525482178,
       "learning_rate": 9.424424424424425e-06,
-      "loss": 2.2288,
+      "loss": 2.2405,
       "step": 5750
     },
     {
       "epoch": 17.3,
-      "grad_norm": 7.964071750640869,
+      "grad_norm": 7.139754772186279,
       "learning_rate": 9.423423423423425e-06,
-      "loss": 2.1994,
+      "loss": 2.2142,
       "step": 5760
     },
     {
       "epoch": 17.33,
-      "grad_norm": 7.8490681648254395,
+      "grad_norm": 13.352121353149414,
       "learning_rate": 9.422422422422424e-06,
-      "loss": 2.277,
+      "loss": 2.2855,
       "step": 5770
     },
     {
       "epoch": 17.36,
-      "grad_norm": 9.511269569396973,
+      "grad_norm": 11.09267807006836,
       "learning_rate": 9.421421421421422e-06,
-      "loss": 2.2057,
+      "loss": 2.226,
       "step": 5780
     },
     {
       "epoch": 17.39,
-      "grad_norm": 7.657184600830078,
+      "grad_norm": 7.666239261627197,
       "learning_rate": 9.420420420420421e-06,
-      "loss": 2.1719,
+      "loss": 2.184,
       "step": 5790
     },
     {
       "epoch": 17.42,
-      "grad_norm": 7.351887226104736,
+      "grad_norm": 6.944879055023193,
       "learning_rate": 9.41941941941942e-06,
-      "loss": 2.2742,
+      "loss": 2.2909,
       "step": 5800
     },
     {
       "epoch": 17.45,
-      "grad_norm": 8.03100299835205,
+      "grad_norm": 11.701077461242676,
       "learning_rate": 9.41841841841842e-06,
-      "loss": 2.218,
+      "loss": 2.2179,
       "step": 5810
     },
     {
       "epoch": 17.48,
-      "grad_norm": 7.316451072692871,
+      "grad_norm": 6.413424968719482,
       "learning_rate": 9.417417417417418e-06,
-      "loss": 2.1549,
+      "loss": 2.1659,
       "step": 5820
     },
     {
       "epoch": 17.51,
-      "grad_norm": 8.929028511047363,
+      "grad_norm": 9.218364715576172,
       "learning_rate": 9.416416416416417e-06,
-      "loss": 2.1674,
+      "loss": 2.1738,
       "step": 5830
     },
     {
       "epoch": 17.54,
-      "grad_norm": 5.778770923614502,
+      "grad_norm": 7.242204666137695,
       "learning_rate": 9.415415415415416e-06,
-      "loss": 2.1955,
+      "loss": 2.2171,
       "step": 5840
     },
     {
       "epoch": 17.57,
-      "grad_norm": 8.302850723266602,
+      "grad_norm": 9.462380409240723,
       "learning_rate": 9.414414414414416e-06,
-      "loss": 2.1792,
+      "loss": 2.1756,
       "step": 5850
     },
     {
       "epoch": 17.6,
-      "grad_norm": 6.201510906219482,
+      "grad_norm": 6.639934539794922,
       "learning_rate": 9.413413413413413e-06,
-      "loss": 2.1765,
+      "loss": 2.1925,
       "step": 5860
     },
     {
       "epoch": 17.63,
-      "grad_norm": 7.91251277923584,
+      "grad_norm": 6.905362129211426,
       "learning_rate": 9.412412412412413e-06,
-      "loss": 2.2276,
+      "loss": 2.2448,
       "step": 5870
     },
     {
       "epoch": 17.66,
-      "grad_norm": 6.167596340179443,
+      "grad_norm": 8.317036628723145,
       "learning_rate": 9.411411411411412e-06,
-      "loss": 2.2183,
+      "loss": 2.2243,
       "step": 5880
     },
     {
       "epoch": 17.69,
-      "grad_norm": 6.277451515197754,
+      "grad_norm": 6.628777980804443,
       "learning_rate": 9.41041041041041e-06,
-      "loss": 2.1741,
+      "loss": 2.1982,
       "step": 5890
     },
     {
       "epoch": 17.72,
-      "grad_norm": 6.038169860839844,
+      "grad_norm": 11.43006420135498,
       "learning_rate": 9.40940940940941e-06,
-      "loss": 2.2333,
+      "loss": 2.2466,
       "step": 5900
     },
     {
       "epoch": 17.75,
-      "grad_norm": 8.004467964172363,
+      "grad_norm": 8.755187034606934,
       "learning_rate": 9.40840840840841e-06,
-      "loss": 2.2166,
+      "loss": 2.2561,
       "step": 5910
     },
     {
       "epoch": 17.78,
-      "grad_norm": 6.30347204208374,
+      "grad_norm": 6.789409637451172,
       "learning_rate": 9.407407407407408e-06,
-      "loss": 2.1607,
+      "loss": 2.1766,
       "step": 5920
     },
     {
       "epoch": 17.81,
-      "grad_norm": 8.487735748291016,
+      "grad_norm": 13.392416000366211,
       "learning_rate": 9.406406406406408e-06,
-      "loss": 2.1892,
+      "loss": 2.2019,
       "step": 5930
     },
     {
       "epoch": 17.84,
-      "grad_norm": 7.574560642242432,
+      "grad_norm": 6.540450572967529,
       "learning_rate": 9.405405405405407e-06,
-      "loss": 2.1623,
+      "loss": 2.1718,
       "step": 5940
     },
     {
       "epoch": 17.87,
-      "grad_norm": 7.5443596839904785,
+      "grad_norm": 9.144981384277344,
       "learning_rate": 9.404404404404405e-06,
-      "loss": 2.1987,
+      "loss": 2.2225,
       "step": 5950
     },
     {
       "epoch": 17.9,
-      "grad_norm": 5.321784496307373,
+      "grad_norm": 7.146183967590332,
       "learning_rate": 9.403403403403404e-06,
-      "loss": 2.1339,
+      "loss": 2.1583,
       "step": 5960
     },
     {
       "epoch": 17.93,
-      "grad_norm": 5.994297981262207,
+      "grad_norm": 7.504815101623535,
       "learning_rate": 9.402402402402402e-06,
-      "loss": 2.1864,
+      "loss": 2.2045,
       "step": 5970
     },
     {
       "epoch": 17.96,
-      "grad_norm": 5.99713659286499,
+      "grad_norm": 5.332332611083984,
       "learning_rate": 9.401401401401403e-06,
-      "loss": 2.1472,
+      "loss": 2.1505,
       "step": 5980
     },
     {
       "epoch": 17.99,
-      "grad_norm": 8.376669883728027,
+      "grad_norm": 15.102433204650879,
       "learning_rate": 9.400400400400401e-06,
-      "loss": 2.1022,
+      "loss": 2.1251,
       "step": 5990
     },
     {
       "epoch": 18.0,
-      "eval_accuracy": 0.6081,
-      "eval_loss": 1.5257073640823364,
-      "eval_runtime": 5.513,
-      "eval_samples_per_second": 1813.91,
-      "eval_steps_per_second": 7.256,
+      "eval_accuracy": 0.6058,
+      "eval_loss": 1.541817545890808,
+      "eval_runtime": 5.3599,
+      "eval_samples_per_second": 1865.7,
+      "eval_steps_per_second": 7.463,
       "step": 5994
     },
     {
       "epoch": 18.02,
-      "grad_norm": 6.291779518127441,
+      "grad_norm": 7.456717491149902,
       "learning_rate": 9.3993993993994e-06,
-      "loss": 2.2667,
+      "loss": 2.26,
       "step": 6000
     },
     {
       "epoch": 18.05,
-      "grad_norm": 10.133583068847656,
+      "grad_norm": 9.780684471130371,
       "learning_rate": 9.3983983983984e-06,
-      "loss": 2.1339,
+      "loss": 2.1451,
       "step": 6010
     },
     {
       "epoch": 18.08,
-      "grad_norm": 7.408086776733398,
+      "grad_norm": 6.914189338684082,
       "learning_rate": 9.397397397397399e-06,
-      "loss": 2.1433,
+      "loss": 2.1513,
       "step": 6020
     },
     {
       "epoch": 18.11,
-      "grad_norm": 6.190664291381836,
+      "grad_norm": 9.207704544067383,
       "learning_rate": 9.396396396396397e-06,
-      "loss": 2.1431,
+      "loss": 2.153,
       "step": 6030
     },
     {
       "epoch": 18.14,
-      "grad_norm": 7.8727707862854,
+      "grad_norm": 8.981575965881348,
       "learning_rate": 9.395395395395396e-06,
-      "loss": 2.1465,
+      "loss": 2.1631,
       "step": 6040
     },
     {
       "epoch": 18.17,
-      "grad_norm": 6.2729620933532715,
+      "grad_norm": 8.2120361328125,
       "learning_rate": 9.394394394394395e-06,
-      "loss": 2.1257,
+      "loss": 2.1458,
       "step": 6050
     },
     {
       "epoch": 18.2,
-      "grad_norm": 6.541892051696777,
+      "grad_norm": 7.836431980133057,
       "learning_rate": 9.393393393393393e-06,
-      "loss": 2.1131,
+      "loss": 2.1401,
       "step": 6060
     },
     {
       "epoch": 18.23,
-      "grad_norm": 7.56522274017334,
+      "grad_norm": 7.446945667266846,
       "learning_rate": 9.392392392392394e-06,
-      "loss": 2.1197,
+      "loss": 2.1348,
       "step": 6070
     },
     {
       "epoch": 18.26,
-      "grad_norm": 11.290907859802246,
+      "grad_norm": 9.233343124389648,
       "learning_rate": 9.391391391391392e-06,
-      "loss": 2.1798,
+      "loss": 2.2048,
       "step": 6080
     },
     {
       "epoch": 18.29,
-      "grad_norm": 6.067159652709961,
+      "grad_norm": 7.6299967765808105,
       "learning_rate": 9.39039039039039e-06,
-      "loss": 2.199,
+      "loss": 2.2014,
       "step": 6090
     },
     {
       "epoch": 18.32,
-      "grad_norm": 10.3577299118042,
+      "grad_norm": 6.300393581390381,
       "learning_rate": 9.389389389389391e-06,
-      "loss": 2.1235,
+      "loss": 2.14,
       "step": 6100
     },
     {
       "epoch": 18.35,
-      "grad_norm": 8.604345321655273,
+      "grad_norm": 11.183070182800293,
       "learning_rate": 9.388388388388388e-06,
-      "loss": 2.1699,
+      "loss": 2.1881,
       "step": 6110
     },
     {
       "epoch": 18.38,
-      "grad_norm": 6.124477386474609,
+      "grad_norm": 6.796738624572754,
       "learning_rate": 9.387387387387388e-06,
-      "loss": 2.1475,
+      "loss": 2.1491,
       "step": 6120
     },
     {
       "epoch": 18.41,
-      "grad_norm": 9.50084114074707,
+      "grad_norm": 7.341439723968506,
       "learning_rate": 9.386386386386387e-06,
-      "loss": 2.1167,
+      "loss": 2.1335,
       "step": 6130
     },
     {
       "epoch": 18.44,
-      "grad_norm": 9.012136459350586,
+      "grad_norm": 8.286240577697754,
       "learning_rate": 9.385385385385385e-06,
-      "loss": 2.0978,
+      "loss": 2.1161,
       "step": 6140
     },
     {
       "epoch": 18.47,
-      "grad_norm": 9.52843189239502,
+      "grad_norm": 9.305739402770996,
       "learning_rate": 9.384384384384386e-06,
-      "loss": 2.1205,
+      "loss": 2.1358,
       "step": 6150
     },
     {
       "epoch": 18.5,
-      "grad_norm": 7.272739887237549,
+      "grad_norm": 6.273959636688232,
       "learning_rate": 9.383383383383384e-06,
-      "loss": 2.1245,
+      "loss": 2.1419,
       "step": 6160
     },
     {
       "epoch": 18.53,
-      "grad_norm": 10.029694557189941,
+      "grad_norm": 9.141531944274902,
       "learning_rate": 9.382382382382383e-06,
-      "loss": 2.1547,
+      "loss": 2.1782,
       "step": 6170
     },
     {
       "epoch": 18.56,
-      "grad_norm": 8.667757034301758,
+      "grad_norm": 9.957012176513672,
       "learning_rate": 9.381381381381383e-06,
-      "loss": 2.1539,
+      "loss": 2.1791,
       "step": 6180
     },
     {
       "epoch": 18.59,
-      "grad_norm": 10.39034652709961,
+      "grad_norm": 7.930354595184326,
       "learning_rate": 9.380380380380382e-06,
-      "loss": 2.1335,
+      "loss": 2.1586,
       "step": 6190
     },
     {
       "epoch": 18.62,
-      "grad_norm": 7.586408615112305,
+      "grad_norm": 7.36091947555542,
       "learning_rate": 9.37937937937938e-06,
-      "loss": 2.1032,
+      "loss": 2.1168,
       "step": 6200
     },
     {
       "epoch": 18.65,
-      "grad_norm": 7.893496990203857,
+      "grad_norm": 6.480175971984863,
       "learning_rate": 9.378378378378379e-06,
-      "loss": 2.0607,
+      "loss": 2.0683,
       "step": 6210
     },
     {
       "epoch": 18.68,
-      "grad_norm": 9.066630363464355,
+      "grad_norm": 11.633779525756836,
       "learning_rate": 9.377377377377378e-06,
-      "loss": 2.0637,
+      "loss": 2.0739,
       "step": 6220
     },
     {
       "epoch": 18.71,
-      "grad_norm": 6.761674404144287,
+      "grad_norm": 5.484204292297363,
       "learning_rate": 9.376376376376378e-06,
-      "loss": 2.0694,
+      "loss": 2.0883,
       "step": 6230
     },
     {
       "epoch": 18.74,
-      "grad_norm": 9.52922534942627,
+      "grad_norm": 6.173926830291748,
       "learning_rate": 9.375375375375376e-06,
-      "loss": 2.1056,
+      "loss": 2.1306,
       "step": 6240
     },
     {
       "epoch": 18.77,
-      "grad_norm": 7.796220779418945,
+      "grad_norm": 7.839195251464844,
       "learning_rate": 9.374374374374375e-06,
-      "loss": 2.1587,
+      "loss": 2.1864,
       "step": 6250
     },
     {
       "epoch": 18.8,
-      "grad_norm": 7.404656410217285,
+      "grad_norm": 8.197251319885254,
       "learning_rate": 9.373373373373375e-06,
-      "loss": 2.168,
+      "loss": 2.1805,
       "step": 6260
     },
     {
       "epoch": 18.83,
-      "grad_norm": 6.868673801422119,
+      "grad_norm": 7.575215816497803,
       "learning_rate": 9.372372372372374e-06,
-      "loss": 2.0993,
+      "loss": 2.0929,
       "step": 6270
     },
     {
       "epoch": 18.86,
-      "grad_norm": 6.173553943634033,
+      "grad_norm": 6.643970489501953,
       "learning_rate": 9.371371371371372e-06,
-      "loss": 2.1093,
+      "loss": 2.1152,
       "step": 6280
     },
     {
       "epoch": 18.89,
-      "grad_norm": 5.088805675506592,
+      "grad_norm": 5.883133888244629,
       "learning_rate": 9.370370370370371e-06,
-      "loss": 2.063,
+      "loss": 2.0891,
       "step": 6290
     },
     {
       "epoch": 18.92,
-      "grad_norm": 13.110210418701172,
+      "grad_norm": 10.603970527648926,
       "learning_rate": 9.36936936936937e-06,
-      "loss": 2.0464,
+      "loss": 2.0594,
       "step": 6300
     },
     {
       "epoch": 18.95,
-      "grad_norm": 6.549376487731934,
+      "grad_norm": 6.693024158477783,
       "learning_rate": 9.368368368368368e-06,
-      "loss": 2.1053,
+      "loss": 2.1419,
       "step": 6310
     },
     {
       "epoch": 18.98,
-      "grad_norm": 8.602410316467285,
+      "grad_norm": 10.160317420959473,
       "learning_rate": 9.367367367367369e-06,
-      "loss": 2.0252,
+      "loss": 2.0399,
       "step": 6320
     },
     {
       "epoch": 19.0,
-      "eval_accuracy": 0.6221,
-      "eval_loss": 1.4517078399658203,
-      "eval_runtime": 5.4446,
-      "eval_samples_per_second": 1836.683,
-      "eval_steps_per_second": 7.347,
+      "eval_accuracy": 0.6183,
+      "eval_loss": 1.472111701965332,
+      "eval_runtime": 5.4928,
+      "eval_samples_per_second": 1820.555,
+      "eval_steps_per_second": 7.282,
       "step": 6327
     },
     {
       "epoch": 19.01,
-      "grad_norm": 7.538969993591309,
+      "grad_norm": 6.441681385040283,
       "learning_rate": 9.366366366366367e-06,
-      "loss": 2.2484,
+      "loss": 2.2533,
       "step": 6330
     },
     {
       "epoch": 19.04,
-      "grad_norm": 8.490106582641602,
+      "grad_norm": 7.061967372894287,
       "learning_rate": 9.365365365365366e-06,
-      "loss": 2.1128,
+      "loss": 2.125,
       "step": 6340
     },
     {
       "epoch": 19.07,
-      "grad_norm": 7.449136257171631,
+      "grad_norm": 7.693588733673096,
       "learning_rate": 9.364364364364366e-06,
-      "loss": 2.1294,
+      "loss": 2.1328,
       "step": 6350
     },
     {
       "epoch": 19.1,
-      "grad_norm": 7.482917308807373,
+      "grad_norm": 6.931527614593506,
       "learning_rate": 9.363363363363363e-06,
-      "loss": 2.1028,
+      "loss": 2.1224,
       "step": 6360
     },
     {
       "epoch": 19.13,
-      "grad_norm": 12.7097749710083,
+      "grad_norm": 10.253271102905273,
       "learning_rate": 9.362362362362363e-06,
-      "loss": 2.1111,
+      "loss": 2.1284,
       "step": 6370
     },
     {
       "epoch": 19.16,
-      "grad_norm": 8.428141593933105,
+      "grad_norm": 7.12988805770874,
       "learning_rate": 9.361361361361362e-06,
-      "loss": 2.0665,
+      "loss": 2.0751,
       "step": 6380
     },
     {
       "epoch": 19.19,
-      "grad_norm": 6.948078632354736,
+      "grad_norm": 9.496931076049805,
       "learning_rate": 9.36036036036036e-06,
-      "loss": 2.0052,
+      "loss": 2.0143,
       "step": 6390
     },
     {
       "epoch": 19.22,
-      "grad_norm": 6.275650978088379,
+      "grad_norm": 14.380762100219727,
       "learning_rate": 9.35935935935936e-06,
-      "loss": 2.0432,
+      "loss": 2.0639,
       "step": 6400
     },
     {
       "epoch": 19.25,
-      "grad_norm": 10.040605545043945,
+      "grad_norm": 9.51627254486084,
       "learning_rate": 9.35835835835836e-06,
-      "loss": 2.1098,
+      "loss": 2.1242,
       "step": 6410
     },
     {
       "epoch": 19.28,
-      "grad_norm": 7.149500846862793,
+      "grad_norm": 11.229992866516113,
       "learning_rate": 9.357357357357358e-06,
-      "loss": 2.098,
+      "loss": 2.1133,
       "step": 6420
     },
     {
       "epoch": 19.31,
-      "grad_norm": 7.478113174438477,
+      "grad_norm": 7.549458980560303,
       "learning_rate": 9.356356356356358e-06,
-      "loss": 2.0274,
+      "loss": 2.0555,
       "step": 6430
     },
     {
       "epoch": 19.34,
-      "grad_norm": 8.839607238769531,
+      "grad_norm": 7.51843786239624,
       "learning_rate": 9.355355355355357e-06,
-      "loss": 2.0976,
+      "loss": 2.1062,
       "step": 6440
     },
     {
       "epoch": 19.37,
-      "grad_norm": 6.017516613006592,
+      "grad_norm": 8.933431625366211,
       "learning_rate": 9.354354354354355e-06,
-      "loss": 2.0667,
+      "loss": 2.0807,
       "step": 6450
     },
     {
       "epoch": 19.4,
-      "grad_norm": 7.232783317565918,
+      "grad_norm": 9.124711036682129,
       "learning_rate": 9.353353353353354e-06,
-      "loss": 2.0805,
+      "loss": 2.1042,
       "step": 6460
     },
     {
       "epoch": 19.43,
-      "grad_norm": 10.285178184509277,
+      "grad_norm": 7.0775628089904785,
       "learning_rate": 9.352352352352353e-06,
-      "loss": 2.0527,
+      "loss": 2.0735,
       "step": 6470
     },
     {
       "epoch": 19.46,
-      "grad_norm": 7.526298999786377,
+      "grad_norm": 6.9851837158203125,
       "learning_rate": 9.351351351351353e-06,
-      "loss": 2.028,
+      "loss": 2.0392,
       "step": 6480
     },
     {
       "epoch": 19.49,
-      "grad_norm": 6.929281711578369,
+      "grad_norm": 7.70259952545166,
       "learning_rate": 9.350350350350351e-06,
-      "loss": 2.0597,
+      "loss": 2.0745,
       "step": 6490
     },
     {
       "epoch": 19.52,
-      "grad_norm": 6.389843463897705,
+      "grad_norm": 8.358838081359863,
       "learning_rate": 9.34934934934935e-06,
-      "loss": 2.0251,
+      "loss": 2.0477,
       "step": 6500
     },
     {
       "epoch": 19.55,
-      "grad_norm": 9.253724098205566,
+      "grad_norm": 7.844730377197266,
       "learning_rate": 9.34834834834835e-06,
-      "loss": 2.1065,
+      "loss": 2.1216,
       "step": 6510
     },
     {
       "epoch": 19.58,
-      "grad_norm": 4.837165355682373,
+      "grad_norm": 7.427902698516846,
       "learning_rate": 9.347347347347349e-06,
-      "loss": 2.0249,
+      "loss": 2.0535,
       "step": 6520
     },
     {
       "epoch": 19.61,
-      "grad_norm": 7.066263675689697,
+      "grad_norm": 7.703489303588867,
       "learning_rate": 9.346346346346346e-06,
-      "loss": 2.0509,
+      "loss": 2.0748,
       "step": 6530
     },
     {
       "epoch": 19.64,
-      "grad_norm": 7.3657426834106445,
+      "grad_norm": 9.564041137695312,
       "learning_rate": 9.345345345345346e-06,
-      "loss": 2.0847,
+      "loss": 2.0939,
       "step": 6540
     },
     {
       "epoch": 19.67,
-      "grad_norm": 7.063158988952637,
+      "grad_norm": 9.756853103637695,
       "learning_rate": 9.344344344344345e-06,
-      "loss": 2.0646,
+      "loss": 2.0739,
       "step": 6550
     },
     {
       "epoch": 19.7,
-      "grad_norm": 10.311962127685547,
+      "grad_norm": 8.079852104187012,
       "learning_rate": 9.343343343343343e-06,
-      "loss": 2.0035,
+      "loss": 2.0229,
       "step": 6560
     },
     {
       "epoch": 19.73,
-      "grad_norm": 6.503240585327148,
+      "grad_norm": 7.90424108505249,
       "learning_rate": 9.342342342342344e-06,
-      "loss": 2.0463,
+      "loss": 2.06,
       "step": 6570
     },
     {
       "epoch": 19.76,
-      "grad_norm": 7.638372898101807,
+      "grad_norm": 7.325865745544434,
       "learning_rate": 9.341341341341342e-06,
-      "loss": 2.1018,
+      "loss": 2.1094,
       "step": 6580
     },
     {
       "epoch": 19.79,
-      "grad_norm": 9.437307357788086,
+      "grad_norm": 11.61992073059082,
       "learning_rate": 9.34034034034034e-06,
-      "loss": 2.0587,
+      "loss": 2.068,
       "step": 6590
     },
     {
       "epoch": 19.82,
-      "grad_norm": 7.787014007568359,
+      "grad_norm": 8.564446449279785,
       "learning_rate": 9.339339339339341e-06,
-      "loss": 2.1038,
+      "loss": 2.1296,
       "step": 6600
     },
     {
       "epoch": 19.85,
-      "grad_norm": 10.383357048034668,
+      "grad_norm": 6.646179676055908,
       "learning_rate": 9.338338338338338e-06,
-      "loss": 2.0678,
+      "loss": 2.091,
       "step": 6610
     },
     {
       "epoch": 19.88,
-      "grad_norm": 11.84603214263916,
+      "grad_norm": 8.81992244720459,
       "learning_rate": 9.337337337337338e-06,
-      "loss": 1.9969,
+      "loss": 2.0247,
       "step": 6620
     },
     {
       "epoch": 19.91,
-      "grad_norm": 8.996867179870605,
+      "grad_norm": 9.360638618469238,
       "learning_rate": 9.336336336336337e-06,
-      "loss": 2.0529,
+      "loss": 2.0644,
       "step": 6630
     },
     {
       "epoch": 19.94,
-      "grad_norm": 8.659820556640625,
+      "grad_norm": 11.25341796875,
       "learning_rate": 9.335335335335335e-06,
-      "loss": 2.0639,
+      "loss": 2.0701,
       "step": 6640
     },
     {
       "epoch": 19.97,
-      "grad_norm": 5.502033233642578,
+      "grad_norm": 7.152585506439209,
       "learning_rate": 9.334334334334336e-06,
-      "loss": 2.0214,
+      "loss": 2.0227,
       "step": 6650
     },
     {
       "epoch": 20.0,
-      "grad_norm": 90.18342590332031,
+      "grad_norm": 97.39366912841797,
       "learning_rate": 9.333333333333334e-06,
-      "loss": 2.2526,
+      "loss": 2.2324,
       "step": 6660
     },
     {
       "epoch": 20.0,
-      "eval_accuracy": 0.6279,
-      "eval_loss": 1.3900591135025024,
-      "eval_runtime": 5.4693,
-      "eval_samples_per_second": 1828.403,
-      "eval_steps_per_second": 7.314,
+      "eval_accuracy": 0.6258,
+      "eval_loss": 1.402449369430542,
+      "eval_runtime": 5.3432,
+      "eval_samples_per_second": 1871.551,
+      "eval_steps_per_second": 7.486,
       "step": 6660
     },
     {
       "epoch": 20.03,
-      "grad_norm": 7.970134258270264,
+      "grad_norm": 6.425572872161865,
       "learning_rate": 9.332332332332333e-06,
-      "loss": 2.0467,
+      "loss": 2.0663,
       "step": 6670
     },
     {
       "epoch": 20.06,
-      "grad_norm": 7.970917224884033,
+      "grad_norm": 10.18450927734375,
       "learning_rate": 9.331331331331333e-06,
-      "loss": 2.0068,
+      "loss": 2.0216,
       "step": 6680
     },
     {
       "epoch": 20.09,
-      "grad_norm": 7.435835838317871,
+      "grad_norm": 8.447525978088379,
       "learning_rate": 9.330330330330332e-06,
-      "loss": 2.0453,
+      "loss": 2.0549,
       "step": 6690
     },
     {
       "epoch": 20.12,
-      "grad_norm": 8.95244026184082,
+      "grad_norm": 8.7366361618042,
       "learning_rate": 9.32932932932933e-06,
-      "loss": 2.0139,
+      "loss": 2.0357,
       "step": 6700
     },
     {
       "epoch": 20.15,
-      "grad_norm": 6.933708667755127,
+      "grad_norm": 8.066652297973633,
       "learning_rate": 9.328328328328329e-06,
-      "loss": 1.8943,
+      "loss": 1.9051,
       "step": 6710
     },
     {
       "epoch": 20.18,
-      "grad_norm": 8.392284393310547,
+      "grad_norm": 6.992068767547607,
       "learning_rate": 9.327327327327328e-06,
-      "loss": 2.0276,
+      "loss": 2.0467,
       "step": 6720
     },
     {
       "epoch": 20.21,
-      "grad_norm": 6.600712299346924,
+      "grad_norm": 8.674290657043457,
       "learning_rate": 9.326326326326328e-06,
-      "loss": 2.0124,
+      "loss": 2.0155,
       "step": 6730
     },
     {
       "epoch": 20.24,
-      "grad_norm": 7.188414573669434,
+      "grad_norm": 6.839930534362793,
       "learning_rate": 9.325325325325326e-06,
-      "loss": 2.0439,
+      "loss": 2.0585,
       "step": 6740
     },
     {
       "epoch": 20.27,
-      "grad_norm": 7.399527549743652,
+      "grad_norm": 8.476581573486328,
       "learning_rate": 9.324324324324325e-06,
-      "loss": 2.0298,
+      "loss": 2.0461,
       "step": 6750
     },
     {
       "epoch": 20.3,
-      "grad_norm": 5.995215892791748,
+      "grad_norm": 7.259369373321533,
       "learning_rate": 9.323323323323324e-06,
-      "loss": 1.9812,
+      "loss": 1.9901,
       "step": 6760
     },
     {
       "epoch": 20.33,
-      "grad_norm": 8.221821784973145,
+      "grad_norm": 11.540635108947754,
       "learning_rate": 9.322322322322324e-06,
-      "loss": 2.0536,
+      "loss": 2.0723,
       "step": 6770
     },
     {
       "epoch": 20.36,
-      "grad_norm": 6.526797771453857,
+      "grad_norm": 11.45331859588623,
       "learning_rate": 9.321321321321321e-06,
-      "loss": 2.0692,
+      "loss": 2.0971,
       "step": 6780
     },
     {
       "epoch": 20.39,
-      "grad_norm": 7.924104690551758,
+      "grad_norm": 8.628748893737793,
       "learning_rate": 9.320320320320321e-06,
-      "loss": 2.0287,
+      "loss": 2.0239,
       "step": 6790
     },
     {
       "epoch": 20.42,
-      "grad_norm": 8.257763862609863,
+      "grad_norm": 7.163269519805908,
       "learning_rate": 9.31931931931932e-06,
-      "loss": 2.0718,
+      "loss": 2.0751,
       "step": 6800
     },
     {
       "epoch": 20.45,
-      "grad_norm": 10.764581680297852,
+      "grad_norm": 24.396846771240234,
       "learning_rate": 9.318318318318318e-06,
-      "loss": 2.0209,
+      "loss": 2.0228,
       "step": 6810
     },
     {
       "epoch": 20.48,
-      "grad_norm": 6.65737247467041,
+      "grad_norm": 5.592777252197266,
       "learning_rate": 9.317317317317319e-06,
-      "loss": 2.0228,
+      "loss": 2.025,
       "step": 6820
     },
     {
       "epoch": 20.51,
-      "grad_norm": 7.482308387756348,
+      "grad_norm": 13.167409896850586,
       "learning_rate": 9.316316316316317e-06,
-      "loss": 2.0139,
+      "loss": 2.0367,
       "step": 6830
     },
     {
       "epoch": 20.54,
-      "grad_norm": 5.55311393737793,
+      "grad_norm": 9.009971618652344,
       "learning_rate": 9.315315315315316e-06,
-      "loss": 2.048,
+      "loss": 2.0685,
       "step": 6840
     },
     {
       "epoch": 20.57,
-      "grad_norm": 8.549206733703613,
+      "grad_norm": 11.416013717651367,
       "learning_rate": 9.314314314314316e-06,
-      "loss": 1.985,
+      "loss": 2.0068,
       "step": 6850
     },
     {
       "epoch": 20.6,
-      "grad_norm": 7.397112846374512,
+      "grad_norm": 8.733942985534668,
       "learning_rate": 9.313313313313313e-06,
-      "loss": 2.0926,
+      "loss": 2.0969,
       "step": 6860
     },
     {
       "epoch": 20.63,
-      "grad_norm": 6.569943904876709,
+      "grad_norm": 10.277538299560547,
       "learning_rate": 9.312312312312313e-06,
-      "loss": 1.9926,
+      "loss": 1.9884,
       "step": 6870
     },
     {
       "epoch": 20.66,
-      "grad_norm": 8.650252342224121,
+      "grad_norm": 8.732381820678711,
       "learning_rate": 9.311311311311312e-06,
-      "loss": 1.9872,
+      "loss": 2.0118,
       "step": 6880
     },
     {
       "epoch": 20.69,
-      "grad_norm": 5.311773300170898,
+      "grad_norm": 7.750004291534424,
       "learning_rate": 9.31031031031031e-06,
-      "loss": 1.973,
+      "loss": 1.9804,
       "step": 6890
     },
     {
       "epoch": 20.72,
-      "grad_norm": 7.881230354309082,
+      "grad_norm": 9.539504051208496,
       "learning_rate": 9.30930930930931e-06,
-      "loss": 1.9798,
+      "loss": 1.9885,
       "step": 6900
     },
     {
       "epoch": 20.75,
-      "grad_norm": 6.412374019622803,
+      "grad_norm": 8.85272216796875,
       "learning_rate": 9.30830830830831e-06,
-      "loss": 2.0346,
+      "loss": 2.046,
       "step": 6910
     },
     {
       "epoch": 20.78,
-      "grad_norm": 10.376585006713867,
+      "grad_norm": 9.009735107421875,
       "learning_rate": 9.307307307307308e-06,
-      "loss": 1.9556,
+      "loss": 1.9682,
       "step": 6920
     },
     {
       "epoch": 20.81,
-      "grad_norm": 6.232002258300781,
+      "grad_norm": 8.16867733001709,
       "learning_rate": 9.306306306306308e-06,
-      "loss": 1.9997,
+      "loss": 2.0226,
       "step": 6930
     },
     {
       "epoch": 20.84,
-      "grad_norm": 6.837101459503174,
+      "grad_norm": 7.038722038269043,
       "learning_rate": 9.305305305305305e-06,
-      "loss": 2.0019,
+      "loss": 2.0114,
       "step": 6940
     },
     {
       "epoch": 20.87,
-      "grad_norm": 8.245145797729492,
+      "grad_norm": 8.204200744628906,
       "learning_rate": 9.304304304304305e-06,
-      "loss": 1.9743,
+      "loss": 1.9814,
       "step": 6950
     },
     {
       "epoch": 20.9,
-      "grad_norm": 8.202388763427734,
+      "grad_norm": 7.8441925048828125,
       "learning_rate": 9.303303303303304e-06,
-      "loss": 1.9923,
+      "loss": 2.0247,
       "step": 6960
     },
     {
       "epoch": 20.93,
-      "grad_norm": 9.69970989227295,
+      "grad_norm": 9.066813468933105,
       "learning_rate": 9.302302302302303e-06,
-      "loss": 2.0067,
+      "loss": 2.0221,
       "step": 6970
     },
     {
       "epoch": 20.96,
-      "grad_norm": 6.716887950897217,
+      "grad_norm": 8.476698875427246,
       "learning_rate": 9.301301301301301e-06,
-      "loss": 1.9973,
+      "loss": 2.0093,
       "step": 6980
     },
     {
       "epoch": 20.99,
-      "grad_norm": 5.25423526763916,
+      "grad_norm": 6.6563849449157715,
       "learning_rate": 9.300300300300302e-06,
-      "loss": 1.963,
+      "loss": 1.9784,
       "step": 6990
     },
     {
       "epoch": 21.0,
-      "eval_accuracy": 0.6425,
-      "eval_loss": 1.3430200815200806,
-      "eval_runtime": 5.6163,
-      "eval_samples_per_second": 1780.525,
-      "eval_steps_per_second": 7.122,
+      "eval_accuracy": 0.6415,
+      "eval_loss": 1.353886604309082,
+      "eval_runtime": 5.2836,
+      "eval_samples_per_second": 1892.64,
+      "eval_steps_per_second": 7.571,
       "step": 6993
     },
     {
       "epoch": 21.02,
-      "grad_norm": 7.857430934906006,
+      "grad_norm": 11.158588409423828,
       "learning_rate": 9.2992992992993e-06,
-      "loss": 2.2277,
+      "loss": 2.227,
       "step": 7000
     },
     {
       "epoch": 21.05,
-      "grad_norm": 9.265414237976074,
+      "grad_norm": 8.414697647094727,
       "learning_rate": 9.298298298298299e-06,
-      "loss": 2.0095,
+      "loss": 2.0243,
       "step": 7010
     },
     {
       "epoch": 21.08,
-      "grad_norm": 8.546141624450684,
+      "grad_norm": 13.420072555541992,
       "learning_rate": 9.297297297297299e-06,
-      "loss": 2.0006,
+      "loss": 2.0097,
       "step": 7020
     },
     {
       "epoch": 21.11,
-      "grad_norm": 8.12818431854248,
+      "grad_norm": 12.241390228271484,
       "learning_rate": 9.296296296296296e-06,
-      "loss": 1.9259,
+      "loss": 1.94,
       "step": 7030
     },
     {
       "epoch": 21.14,
-      "grad_norm": 7.3910746574401855,
+      "grad_norm": 4.939540386199951,
       "learning_rate": 9.295295295295296e-06,
-      "loss": 1.9796,
+      "loss": 1.9816,
       "step": 7040
     },
     {
       "epoch": 21.17,
-      "grad_norm": 7.298797607421875,
+      "grad_norm": 8.044371604919434,
       "learning_rate": 9.294294294294295e-06,
-      "loss": 1.926,
+      "loss": 1.9351,
       "step": 7050
     },
     {
       "epoch": 21.2,
-      "grad_norm": 5.757110595703125,
+      "grad_norm": 7.417489051818848,
       "learning_rate": 9.293293293293293e-06,
-      "loss": 1.9397,
+      "loss": 1.9497,
       "step": 7060
     },
     {
       "epoch": 21.23,
-      "grad_norm": 8.715088844299316,
+      "grad_norm": 8.642778396606445,
       "learning_rate": 9.292292292292294e-06,
-      "loss": 2.0009,
+      "loss": 2.024,
       "step": 7070
     },
     {
       "epoch": 21.26,
-      "grad_norm": 6.113747596740723,
+      "grad_norm": 8.755669593811035,
       "learning_rate": 9.291291291291292e-06,
-      "loss": 2.0174,
+      "loss": 2.0244,
       "step": 7080
     },
     {
       "epoch": 21.29,
-      "grad_norm": 8.748607635498047,
+      "grad_norm": 9.779139518737793,
       "learning_rate": 9.29029029029029e-06,
-      "loss": 2.002,
+      "loss": 2.0063,
       "step": 7090
     },
     {
       "epoch": 21.32,
-      "grad_norm": 7.517253398895264,
+      "grad_norm": 12.269511222839355,
       "learning_rate": 9.289289289289291e-06,
-      "loss": 1.9141,
+      "loss": 1.9193,
       "step": 7100
     },
     {
       "epoch": 21.35,
-      "grad_norm": 7.886311054229736,
+      "grad_norm": 8.398375511169434,
       "learning_rate": 9.288288288288288e-06,
-      "loss": 2.0209,
+      "loss": 2.0483,
       "step": 7110
     },
     {
       "epoch": 21.38,
-      "grad_norm": 8.93781566619873,
+      "grad_norm": 9.294203758239746,
       "learning_rate": 9.287287287287288e-06,
-      "loss": 1.9056,
+      "loss": 1.9142,
       "step": 7120
     },
     {
       "epoch": 21.41,
-      "grad_norm": 6.5547776222229,
+      "grad_norm": 7.1016130447387695,
       "learning_rate": 9.286286286286287e-06,
-      "loss": 2.0521,
+      "loss": 2.0553,
       "step": 7130
     },
     {
       "epoch": 21.44,
-      "grad_norm": 6.648595809936523,
+      "grad_norm": 10.061352729797363,
       "learning_rate": 9.285285285285286e-06,
-      "loss": 2.0202,
+      "loss": 2.0397,
       "step": 7140
     },
     {
       "epoch": 21.47,
-      "grad_norm": 7.318085670471191,
+      "grad_norm": 7.876277923583984,
       "learning_rate": 9.284284284284286e-06,
-      "loss": 1.9119,
+      "loss": 1.9348,
       "step": 7150
     },
     {
       "epoch": 21.5,
-      "grad_norm": 8.090007781982422,
+      "grad_norm": 9.935800552368164,
       "learning_rate": 9.283283283283284e-06,
-      "loss": 1.9693,
+      "loss": 1.9749,
       "step": 7160
     },
     {
       "epoch": 21.53,
-      "grad_norm": 6.177989482879639,
+      "grad_norm": 8.76993465423584,
       "learning_rate": 9.282282282282283e-06,
-      "loss": 2.0117,
+      "loss": 2.0218,
       "step": 7170
     },
     {
       "epoch": 21.56,
-      "grad_norm": 7.02983283996582,
+      "grad_norm": 5.875413417816162,
       "learning_rate": 9.281281281281283e-06,
-      "loss": 2.0009,
+      "loss": 2.0213,
       "step": 7180
     },
     {
       "epoch": 21.59,
-      "grad_norm": 5.978540420532227,
+      "grad_norm": 7.146310806274414,
       "learning_rate": 9.28028028028028e-06,
-      "loss": 1.955,
+      "loss": 1.9632,
       "step": 7190
     },
     {
       "epoch": 21.62,
-      "grad_norm": 6.944443702697754,
+      "grad_norm": 6.622971534729004,
       "learning_rate": 9.27927927927928e-06,
-      "loss": 1.9229,
+      "loss": 1.9327,
       "step": 7200
     },
     {
       "epoch": 21.65,
-      "grad_norm": 10.240386962890625,
+      "grad_norm": 7.785336494445801,
       "learning_rate": 9.278278278278279e-06,
-      "loss": 2.0739,
+      "loss": 2.0879,
       "step": 7210
     },
     {
       "epoch": 21.68,
-      "grad_norm": 11.19546890258789,
+      "grad_norm": 9.956010818481445,
       "learning_rate": 9.277277277277278e-06,
-      "loss": 1.9823,
+      "loss": 1.9933,
       "step": 7220
     },
     {
       "epoch": 21.71,
-      "grad_norm": 4.959046840667725,
+      "grad_norm": 7.10001802444458,
       "learning_rate": 9.276276276276276e-06,
-      "loss": 1.8738,
+      "loss": 1.8681,
       "step": 7230
     },
     {
       "epoch": 21.74,
-      "grad_norm": 7.657245635986328,
+      "grad_norm": 6.468133926391602,
       "learning_rate": 9.275275275275277e-06,
-      "loss": 1.9523,
+      "loss": 1.9639,
       "step": 7240
     },
     {
       "epoch": 21.77,
-      "grad_norm": 9.604874610900879,
+      "grad_norm": 12.771150588989258,
       "learning_rate": 9.274274274274275e-06,
-      "loss": 1.9418,
+      "loss": 1.9535,
       "step": 7250
     },
     {
       "epoch": 21.8,
-      "grad_norm": 6.698229789733887,
+      "grad_norm": 7.9877543449401855,
       "learning_rate": 9.273273273273274e-06,
-      "loss": 2.0013,
+      "loss": 2.0007,
       "step": 7260
     },
     {
       "epoch": 21.83,
-      "grad_norm": 5.123570442199707,
+      "grad_norm": 6.426191806793213,
       "learning_rate": 9.272272272272274e-06,
-      "loss": 1.9391,
+      "loss": 1.9443,
       "step": 7270
     },
     {
       "epoch": 21.86,
-      "grad_norm": 10.858541488647461,
+      "grad_norm": 8.706355094909668,
       "learning_rate": 9.271271271271271e-06,
-      "loss": 1.9856,
+      "loss": 1.9894,
       "step": 7280
     },
     {
       "epoch": 21.89,
-      "grad_norm": 6.741635799407959,
+      "grad_norm": 9.114312171936035,
       "learning_rate": 9.270270270270271e-06,
-      "loss": 1.9411,
+      "loss": 1.9461,
       "step": 7290
     },
     {
       "epoch": 21.92,
-      "grad_norm": 7.840023517608643,
+      "grad_norm": 5.885374546051025,
       "learning_rate": 9.26926926926927e-06,
-      "loss": 1.9363,
+      "loss": 1.9454,
       "step": 7300
     },
     {
       "epoch": 21.95,
-      "grad_norm": 8.55334758758545,
+      "grad_norm": 9.925578117370605,
       "learning_rate": 9.268268268268268e-06,
-      "loss": 1.9416,
+      "loss": 1.9568,
       "step": 7310
     },
     {
       "epoch": 21.98,
-      "grad_norm": 6.980435848236084,
+      "grad_norm": 6.81265115737915,
       "learning_rate": 9.267267267267269e-06,
-      "loss": 1.9656,
+      "loss": 1.9792,
       "step": 7320
     },
     {
       "epoch": 22.0,
-      "eval_accuracy": 0.6413,
-      "eval_loss": 1.3013033866882324,
-      "eval_runtime": 5.4068,
-      "eval_samples_per_second": 1849.52,
-      "eval_steps_per_second": 7.398,
+      "eval_accuracy": 0.6448,
+      "eval_loss": 1.3065111637115479,
+      "eval_runtime": 5.3105,
+      "eval_samples_per_second": 1883.073,
+      "eval_steps_per_second": 7.532,
       "step": 7326
     },
     {
       "epoch": 22.01,
-      "grad_norm": 6.414568901062012,
+      "grad_norm": 8.658529281616211,
       "learning_rate": 9.266266266266267e-06,
-      "loss": 2.2209,
+      "loss": 2.2102,
       "step": 7330
     },
     {
       "epoch": 22.04,
-      "grad_norm": 6.539636611938477,
+      "grad_norm": 6.839120864868164,
       "learning_rate": 9.265265265265266e-06,
-      "loss": 1.9268,
+      "loss": 1.9491,
       "step": 7340
     },
     {
       "epoch": 22.07,
-      "grad_norm": 5.996793746948242,
+      "grad_norm": 6.542444229125977,
       "learning_rate": 9.264264264264266e-06,
-      "loss": 1.9464,
+      "loss": 1.9689,
       "step": 7350
     },
     {
       "epoch": 22.1,
-      "grad_norm": 8.042868614196777,
+      "grad_norm": 8.520700454711914,
       "learning_rate": 9.263263263263263e-06,
-      "loss": 1.9071,
+      "loss": 1.9264,
       "step": 7360
     },
     {
       "epoch": 22.13,
-      "grad_norm": 8.442831993103027,
+      "grad_norm": 9.848244667053223,
       "learning_rate": 9.262262262262263e-06,
-      "loss": 1.9255,
+      "loss": 1.9251,
       "step": 7370
     },
     {
       "epoch": 22.16,
-      "grad_norm": 6.769294261932373,
+      "grad_norm": 6.713740348815918,
       "learning_rate": 9.261261261261262e-06,
-      "loss": 1.927,
+      "loss": 1.9403,
       "step": 7380
     },
     {
       "epoch": 22.19,
-      "grad_norm": 7.13763952255249,
+      "grad_norm": 7.074524879455566,
       "learning_rate": 9.26026026026026e-06,
-      "loss": 1.9743,
+      "loss": 1.9854,
       "step": 7390
     },
     {
       "epoch": 22.22,
-      "grad_norm": 9.54120922088623,
+      "grad_norm": 7.774349689483643,
       "learning_rate": 9.25925925925926e-06,
-      "loss": 1.8936,
+      "loss": 1.8957,
       "step": 7400
     },
     {
       "epoch": 22.25,
-      "grad_norm": 8.743712425231934,
+      "grad_norm": 10.974924087524414,
       "learning_rate": 9.25825825825826e-06,
-      "loss": 1.9748,
+      "loss": 1.9787,
       "step": 7410
     },
     {
       "epoch": 22.28,
-      "grad_norm": 9.471392631530762,
+      "grad_norm": 9.373008728027344,
       "learning_rate": 9.257257257257258e-06,
-      "loss": 1.9313,
+      "loss": 1.9345,
       "step": 7420
     },
     {
       "epoch": 22.31,
-      "grad_norm": 7.2820658683776855,
+      "grad_norm": 7.439484596252441,
       "learning_rate": 9.256256256256257e-06,
-      "loss": 1.9142,
+      "loss": 1.9187,
       "step": 7430
     },
     {
       "epoch": 22.34,
-      "grad_norm": 6.653914928436279,
+      "grad_norm": 6.300307750701904,
       "learning_rate": 9.255255255255255e-06,
-      "loss": 1.8497,
+      "loss": 1.8589,
       "step": 7440
     },
     {
       "epoch": 22.37,
-      "grad_norm": 5.2832417488098145,
+      "grad_norm": 8.215286254882812,
       "learning_rate": 9.254254254254254e-06,
-      "loss": 1.9095,
+      "loss": 1.9352,
       "step": 7450
     },
     {
       "epoch": 22.4,
-      "grad_norm": 6.769468307495117,
+      "grad_norm": 9.057433128356934,
       "learning_rate": 9.253253253253254e-06,
-      "loss": 1.9077,
+      "loss": 1.9103,
       "step": 7460
     },
     {
       "epoch": 22.43,
-      "grad_norm": 6.995555400848389,
+      "grad_norm": 7.974792003631592,
       "learning_rate": 9.252252252252253e-06,
-      "loss": 1.9715,
+      "loss": 1.9876,
       "step": 7470
     },
     {
       "epoch": 22.46,
-      "grad_norm": 8.5245943069458,
+      "grad_norm": 8.436942100524902,
       "learning_rate": 9.251251251251251e-06,
-      "loss": 1.9971,
+      "loss": 2.002,
       "step": 7480
     },
     {
       "epoch": 22.49,
-      "grad_norm": 6.491293907165527,
+      "grad_norm": 7.5944437980651855,
       "learning_rate": 9.250250250250252e-06,
-      "loss": 1.9913,
+      "loss": 2.0045,
       "step": 7490
     },
     {
       "epoch": 22.52,
-      "grad_norm": 8.539793014526367,
+      "grad_norm": 10.293737411499023,
       "learning_rate": 9.24924924924925e-06,
-      "loss": 1.9125,
+      "loss": 1.9117,
       "step": 7500
     },
     {
       "epoch": 22.55,
-      "grad_norm": 6.461583614349365,
+      "grad_norm": 7.0867838859558105,
       "learning_rate": 9.248248248248249e-06,
-      "loss": 1.8666,
+      "loss": 1.8865,
       "step": 7510
     },
     {
       "epoch": 22.58,
-      "grad_norm": 7.703919410705566,
+      "grad_norm": 7.87642765045166,
       "learning_rate": 9.247247247247249e-06,
-      "loss": 1.8832,
+      "loss": 1.9072,
       "step": 7520
     },
     {
       "epoch": 22.61,
-      "grad_norm": 7.643704414367676,
+      "grad_norm": 7.695592403411865,
       "learning_rate": 9.246246246246246e-06,
-      "loss": 1.8853,
+      "loss": 1.9041,
       "step": 7530
     },
     {
       "epoch": 22.64,
-      "grad_norm": 9.403818130493164,
+      "grad_norm": 9.082470893859863,
       "learning_rate": 9.245245245245246e-06,
-      "loss": 1.9704,
+      "loss": 1.9828,
       "step": 7540
     },
     {
       "epoch": 22.67,
-      "grad_norm": 6.316361427307129,
+      "grad_norm": 6.269680500030518,
       "learning_rate": 9.244244244244245e-06,
-      "loss": 1.9216,
+      "loss": 1.9424,
       "step": 7550
     },
     {
       "epoch": 22.7,
-      "grad_norm": 7.28446102142334,
+      "grad_norm": 8.319326400756836,
       "learning_rate": 9.243243243243243e-06,
-      "loss": 1.9176,
+      "loss": 1.9287,
       "step": 7560
     },
     {
       "epoch": 22.73,
-      "grad_norm": 8.108738899230957,
+      "grad_norm": 6.1052470207214355,
       "learning_rate": 9.242242242242244e-06,
-      "loss": 1.9588,
+      "loss": 1.9535,
       "step": 7570
     },
     {
       "epoch": 22.76,
-      "grad_norm": 6.870691299438477,
+      "grad_norm": 7.557160377502441,
       "learning_rate": 9.241241241241242e-06,
-      "loss": 1.9974,
+      "loss": 2.0141,
       "step": 7580
     },
     {
       "epoch": 22.79,
-      "grad_norm": 6.8299760818481445,
+      "grad_norm": 7.388027191162109,
       "learning_rate": 9.240240240240241e-06,
-      "loss": 1.8723,
+      "loss": 1.8915,
       "step": 7590
     },
     {
       "epoch": 22.82,
-      "grad_norm": 7.338107585906982,
+      "grad_norm": 6.160009860992432,
       "learning_rate": 9.239239239239241e-06,
-      "loss": 1.9518,
+      "loss": 1.9607,
       "step": 7600
     },
     {
       "epoch": 22.85,
-      "grad_norm": 7.413039684295654,
+      "grad_norm": 8.017452239990234,
       "learning_rate": 9.238238238238238e-06,
-      "loss": 1.9124,
+      "loss": 1.9274,
       "step": 7610
     },
     {
       "epoch": 22.88,
-      "grad_norm": 6.810500621795654,
+      "grad_norm": 7.229618072509766,
       "learning_rate": 9.237237237237238e-06,
-      "loss": 1.8882,
+      "loss": 1.8885,
       "step": 7620
     },
     {
       "epoch": 22.91,
-      "grad_norm": 8.680765151977539,
+      "grad_norm": 9.62418270111084,
       "learning_rate": 9.236236236236237e-06,
-      "loss": 1.9292,
+      "loss": 1.9418,
       "step": 7630
     },
     {
       "epoch": 22.94,
-      "grad_norm": 6.226908206939697,
+      "grad_norm": 9.30985164642334,
       "learning_rate": 9.235235235235236e-06,
-      "loss": 1.9266,
+      "loss": 1.9317,
       "step": 7640
     },
     {
       "epoch": 22.97,
-      "grad_norm": 5.577744960784912,
+      "grad_norm": 11.072701454162598,
       "learning_rate": 9.234234234234236e-06,
-      "loss": 1.8864,
+      "loss": 1.8867,
       "step": 7650
     },
     {
       "epoch": 23.0,
-      "eval_accuracy": 0.6524,
-      "eval_loss": 1.2617355585098267,
-      "eval_runtime": 5.4367,
-      "eval_samples_per_second": 1839.358,
-      "eval_steps_per_second": 7.357,
+      "eval_accuracy": 0.6567,
+      "eval_loss": 1.2732000350952148,
+      "eval_runtime": 5.3615,
+      "eval_samples_per_second": 1865.16,
+      "eval_steps_per_second": 7.461,
       "step": 7659
     },
     {
       "epoch": 23.0,
-      "grad_norm": 6.34880256652832,
+      "grad_norm": 10.608221054077148,
       "learning_rate": 9.233233233233234e-06,
-      "loss": 1.9816,
+      "loss": 2.0032,
       "step": 7660
     },
     {
       "epoch": 23.03,
-      "grad_norm": 7.843090057373047,
+      "grad_norm": 10.812560081481934,
       "learning_rate": 9.232232232232233e-06,
-      "loss": 1.8844,
+      "loss": 1.8837,
       "step": 7670
     },
     {
       "epoch": 23.06,
-      "grad_norm": 8.199349403381348,
+      "grad_norm": 7.220709800720215,
       "learning_rate": 9.231231231231232e-06,
-      "loss": 1.9557,
+      "loss": 1.9678,
       "step": 7680
     },
     {
       "epoch": 23.09,
-      "grad_norm": 8.590726852416992,
+      "grad_norm": 7.438437461853027,
       "learning_rate": 9.23023023023023e-06,
-      "loss": 1.8893,
+      "loss": 1.8967,
       "step": 7690
     },
     {
       "epoch": 23.12,
-      "grad_norm": 9.105623245239258,
+      "grad_norm": 10.316338539123535,
       "learning_rate": 9.229229229229229e-06,
-      "loss": 1.8774,
+      "loss": 1.8849,
       "step": 7700
     },
     {
       "epoch": 23.15,
-      "grad_norm": 10.95720100402832,
+      "grad_norm": 11.01754379272461,
       "learning_rate": 9.228228228228229e-06,
-      "loss": 1.8669,
+      "loss": 1.8739,
       "step": 7710
     },
     {
       "epoch": 23.18,
-      "grad_norm": 9.084404945373535,
+      "grad_norm": 6.347144603729248,
       "learning_rate": 9.227227227227228e-06,
-      "loss": 1.9405,
+      "loss": 1.9513,
       "step": 7720
     },
     {
       "epoch": 23.21,
-      "grad_norm": 6.493495464324951,
+      "grad_norm": 8.680575370788574,
       "learning_rate": 9.226226226226226e-06,
-      "loss": 1.8717,
+      "loss": 1.885,
       "step": 7730
     },
     {
       "epoch": 23.24,
-      "grad_norm": 7.127620220184326,
+      "grad_norm": 7.928031921386719,
       "learning_rate": 9.225225225225227e-06,
-      "loss": 1.8262,
+      "loss": 1.84,
       "step": 7740
     },
     {
       "epoch": 23.27,
-      "grad_norm": 9.708541870117188,
+      "grad_norm": 13.25584602355957,
       "learning_rate": 9.224224224224225e-06,
-      "loss": 1.9292,
+      "loss": 1.9307,
       "step": 7750
     },
     {
       "epoch": 23.3,
-      "grad_norm": 8.01528549194336,
+      "grad_norm": 6.3097405433654785,
       "learning_rate": 9.223223223223224e-06,
-      "loss": 1.8774,
+      "loss": 1.8957,
       "step": 7760
     },
     {
       "epoch": 23.33,
-      "grad_norm": 6.762050628662109,
+      "grad_norm": 10.296032905578613,
       "learning_rate": 9.222222222222224e-06,
-      "loss": 1.8927,
+      "loss": 1.9053,
       "step": 7770
     },
     {
       "epoch": 23.36,
-      "grad_norm": 7.6017351150512695,
+      "grad_norm": 8.230875968933105,
       "learning_rate": 9.221221221221221e-06,
-      "loss": 1.9259,
+      "loss": 1.9399,
       "step": 7780
     },
     {
       "epoch": 23.39,
-      "grad_norm": 6.502980709075928,
+      "grad_norm": 6.482725620269775,
       "learning_rate": 9.220220220220221e-06,
-      "loss": 1.8982,
+      "loss": 1.9191,
       "step": 7790
     },
     {
       "epoch": 23.42,
-      "grad_norm": 9.902922630310059,
+      "grad_norm": 11.481039047241211,
       "learning_rate": 9.21921921921922e-06,
-      "loss": 1.8866,
+      "loss": 1.8968,
       "step": 7800
     },
     {
       "epoch": 23.45,
-      "grad_norm": 7.683023452758789,
+      "grad_norm": 9.079011917114258,
       "learning_rate": 9.218218218218218e-06,
-      "loss": 1.8802,
+      "loss": 1.8951,
       "step": 7810
     },
     {
       "epoch": 23.48,
-      "grad_norm": 6.332238674163818,
+      "grad_norm": 9.139366149902344,
       "learning_rate": 9.217217217217219e-06,
-      "loss": 1.8654,
+      "loss": 1.8791,
       "step": 7820
     },
     {
       "epoch": 23.51,
-      "grad_norm": 7.43429708480835,
+      "grad_norm": 11.736318588256836,
       "learning_rate": 9.216216216216217e-06,
-      "loss": 1.8794,
+      "loss": 1.8987,
       "step": 7830
     },
     {
       "epoch": 23.54,
-      "grad_norm": 7.857909202575684,
+      "grad_norm": 6.012482166290283,
       "learning_rate": 9.215215215215216e-06,
-      "loss": 1.8538,
+      "loss": 1.8559,
       "step": 7840
     },
     {
       "epoch": 23.57,
-      "grad_norm": 7.169832229614258,
+      "grad_norm": 9.67531681060791,
       "learning_rate": 9.214214214214216e-06,
-      "loss": 1.8922,
+      "loss": 1.9047,
       "step": 7850
     },
     {
       "epoch": 23.6,
-      "grad_norm": 6.449008941650391,
+      "grad_norm": 6.707828998565674,
       "learning_rate": 9.213213213213213e-06,
-      "loss": 1.8114,
+      "loss": 1.8235,
       "step": 7860
     },
     {
       "epoch": 23.63,
-      "grad_norm": 9.658852577209473,
+      "grad_norm": 9.8803129196167,
       "learning_rate": 9.212212212212213e-06,
-      "loss": 1.8663,
+      "loss": 1.8836,
       "step": 7870
     },
     {
       "epoch": 23.66,
-      "grad_norm": 8.031635284423828,
+      "grad_norm": 7.237249374389648,
       "learning_rate": 9.211211211211212e-06,
-      "loss": 1.8994,
+      "loss": 1.9056,
       "step": 7880
     },
     {
       "epoch": 23.69,
-      "grad_norm": 5.363889217376709,
+      "grad_norm": 7.4759674072265625,
       "learning_rate": 9.21021021021021e-06,
-      "loss": 1.8684,
+      "loss": 1.8724,
       "step": 7890
     },
     {
       "epoch": 23.72,
-      "grad_norm": 7.7268147468566895,
+      "grad_norm": 9.626898765563965,
       "learning_rate": 9.20920920920921e-06,
-      "loss": 1.8894,
+      "loss": 1.8916,
       "step": 7900
     },
     {
       "epoch": 23.75,
-      "grad_norm": 6.723232269287109,
+      "grad_norm": 5.858760833740234,
       "learning_rate": 9.20820820820821e-06,
-      "loss": 1.8812,
+      "loss": 1.8848,
       "step": 7910
     },
     {
       "epoch": 23.78,
-      "grad_norm": 6.823802471160889,
+      "grad_norm": 8.596174240112305,
       "learning_rate": 9.207207207207208e-06,
-      "loss": 1.8717,
+      "loss": 1.8764,
       "step": 7920
     },
     {
       "epoch": 23.81,
-      "grad_norm": 6.952330589294434,
+      "grad_norm": 7.611490726470947,
       "learning_rate": 9.206206206206207e-06,
-      "loss": 1.8487,
+      "loss": 1.877,
       "step": 7930
     },
     {
       "epoch": 23.84,
-      "grad_norm": 8.248428344726562,
+      "grad_norm": 7.905805587768555,
       "learning_rate": 9.205205205205205e-06,
-      "loss": 1.8792,
+      "loss": 1.8949,
       "step": 7940
     },
     {
       "epoch": 23.87,
-      "grad_norm": 6.262691020965576,
+      "grad_norm": 7.192712783813477,
       "learning_rate": 9.204204204204204e-06,
-      "loss": 1.8743,
+      "loss": 1.8811,
       "step": 7950
     },
     {
       "epoch": 23.9,
-      "grad_norm": 5.560808181762695,
+      "grad_norm": 8.632085800170898,
       "learning_rate": 9.203203203203204e-06,
-      "loss": 1.8427,
+      "loss": 1.8577,
       "step": 7960
     },
     {
       "epoch": 23.93,
-      "grad_norm": 7.478765487670898,
+      "grad_norm": 8.371757507324219,
       "learning_rate": 9.202202202202203e-06,
-      "loss": 1.9039,
+      "loss": 1.9154,
       "step": 7970
     },
     {
       "epoch": 23.96,
-      "grad_norm": 8.5875244140625,
+      "grad_norm": 6.734856605529785,
       "learning_rate": 9.201201201201201e-06,
-      "loss": 1.8363,
+      "loss": 1.8565,
       "step": 7980
     },
     {
       "epoch": 23.99,
-      "grad_norm": 6.773016929626465,
+      "grad_norm": 6.303768157958984,
       "learning_rate": 9.200200200200202e-06,
-      "loss": 1.887,
+      "loss": 1.8928,
       "step": 7990
     },
     {
       "epoch": 24.0,
-      "eval_accuracy": 0.6584,
-      "eval_loss": 1.2375680208206177,
-      "eval_runtime": 5.4218,
-      "eval_samples_per_second": 1844.414,
-      "eval_steps_per_second": 7.378,
+      "eval_accuracy": 0.6616,
+      "eval_loss": 1.2445435523986816,
+      "eval_runtime": 5.3197,
+      "eval_samples_per_second": 1879.795,
+      "eval_steps_per_second": 7.519,
       "step": 7992
     },
     {
       "epoch": 24.02,
-      "grad_norm": 6.8171610832214355,
+      "grad_norm": 6.737892150878906,
       "learning_rate": 9.1991991991992e-06,
-      "loss": 2.0349,
+      "loss": 2.0426,
       "step": 8000
     },
     {
       "epoch": 24.05,
-      "grad_norm": 6.51664400100708,
+      "grad_norm": 6.486743450164795,
       "learning_rate": 9.198198198198199e-06,
-      "loss": 1.8801,
+      "loss": 1.8977,
       "step": 8010
     },
     {
       "epoch": 24.08,
-      "grad_norm": 9.310213088989258,
+      "grad_norm": 10.521382331848145,
       "learning_rate": 9.197197197197199e-06,
-      "loss": 1.8268,
+      "loss": 1.8399,
       "step": 8020
     },
     {
       "epoch": 24.11,
-      "grad_norm": 6.021097660064697,
+      "grad_norm": 7.792485237121582,
       "learning_rate": 9.196196196196196e-06,
-      "loss": 1.8093,
+      "loss": 1.8263,
       "step": 8030
     },
     {
       "epoch": 24.14,
-      "grad_norm": 5.532205581665039,
+      "grad_norm": 7.585921287536621,
       "learning_rate": 9.195195195195196e-06,
-      "loss": 1.8795,
+      "loss": 1.8846,
       "step": 8040
     },
     {
       "epoch": 24.17,
-      "grad_norm": 10.052406311035156,
+      "grad_norm": 13.999678611755371,
       "learning_rate": 9.194194194194195e-06,
-      "loss": 1.8434,
+      "loss": 1.8557,
       "step": 8050
     },
     {
       "epoch": 24.2,
-      "grad_norm": 8.480722427368164,
+      "grad_norm": 10.48676872253418,
       "learning_rate": 9.193193193193194e-06,
-      "loss": 1.8248,
+      "loss": 1.8522,
       "step": 8060
     },
     {
       "epoch": 24.23,
-      "grad_norm": 6.46122407913208,
+      "grad_norm": 5.797928333282471,
       "learning_rate": 9.192192192192194e-06,
-      "loss": 1.8814,
+      "loss": 1.8875,
       "step": 8070
     },
     {
       "epoch": 24.26,
-      "grad_norm": 5.899950981140137,
+      "grad_norm": 9.056879997253418,
       "learning_rate": 9.191191191191192e-06,
-      "loss": 1.8468,
+      "loss": 1.8689,
       "step": 8080
     },
     {
       "epoch": 24.29,
-      "grad_norm": 6.817729473114014,
+      "grad_norm": 10.034852981567383,
       "learning_rate": 9.190190190190191e-06,
-      "loss": 1.8499,
+      "loss": 1.8605,
       "step": 8090
     },
     {
       "epoch": 24.32,
-      "grad_norm": 7.574281215667725,
+      "grad_norm": 9.377605438232422,
       "learning_rate": 9.189189189189191e-06,
-      "loss": 1.9022,
+      "loss": 1.9056,
       "step": 8100
     },
     {
       "epoch": 24.35,
-      "grad_norm": 8.790617942810059,
+      "grad_norm": 8.026385307312012,
       "learning_rate": 9.188188188188188e-06,
-      "loss": 1.8436,
+      "loss": 1.8605,
       "step": 8110
     },
     {
       "epoch": 24.38,
-      "grad_norm": 7.073566436767578,
+      "grad_norm": 8.613235473632812,
       "learning_rate": 9.187187187187187e-06,
-      "loss": 1.9031,
+      "loss": 1.9036,
       "step": 8120
     },
     {
       "epoch": 24.41,
-      "grad_norm": 7.444761276245117,
+      "grad_norm": 7.784913063049316,
       "learning_rate": 9.186186186186187e-06,
-      "loss": 1.8705,
+      "loss": 1.892,
       "step": 8130
     },
     {
       "epoch": 24.44,
-      "grad_norm": 7.889785289764404,
+      "grad_norm": 10.491716384887695,
       "learning_rate": 9.185185185185186e-06,
-      "loss": 1.8855,
+      "loss": 1.8874,
       "step": 8140
     },
     {
       "epoch": 24.47,
-      "grad_norm": 7.208620071411133,
+      "grad_norm": 9.618674278259277,
       "learning_rate": 9.184184184184184e-06,
-      "loss": 1.8667,
+      "loss": 1.8802,
       "step": 8150
     },
     {
       "epoch": 24.5,
-      "grad_norm": 9.727555274963379,
+      "grad_norm": 8.613658905029297,
       "learning_rate": 9.183183183183185e-06,
-      "loss": 1.8434,
+      "loss": 1.8485,
       "step": 8160
     },
     {
       "epoch": 24.53,
-      "grad_norm": 7.903265476226807,
+      "grad_norm": 8.110553741455078,
       "learning_rate": 9.182182182182183e-06,
-      "loss": 1.8487,
+      "loss": 1.8561,
       "step": 8170
     },
     {
       "epoch": 24.56,
-      "grad_norm": 6.425499439239502,
+      "grad_norm": 8.098548889160156,
       "learning_rate": 9.181181181181182e-06,
-      "loss": 1.7772,
+      "loss": 1.7876,
       "step": 8180
     },
     {
       "epoch": 24.59,
-      "grad_norm": 5.933250427246094,
+      "grad_norm": 5.697737216949463,
       "learning_rate": 9.18018018018018e-06,
-      "loss": 1.8604,
+      "loss": 1.8744,
       "step": 8190
     },
     {
       "epoch": 24.62,
-      "grad_norm": 7.56041145324707,
+      "grad_norm": 7.996273517608643,
       "learning_rate": 9.179179179179179e-06,
-      "loss": 1.7959,
+      "loss": 1.8024,
       "step": 8200
     },
     {
       "epoch": 24.65,
-      "grad_norm": 6.262991428375244,
+      "grad_norm": 7.713252067565918,
       "learning_rate": 9.17817817817818e-06,
-      "loss": 1.7827,
+      "loss": 1.7999,
       "step": 8210
     },
     {
       "epoch": 24.68,
-      "grad_norm": 8.449974060058594,
+      "grad_norm": 10.7149658203125,
       "learning_rate": 9.177177177177178e-06,
-      "loss": 1.8506,
+      "loss": 1.8595,
       "step": 8220
     },
     {
       "epoch": 24.71,
-      "grad_norm": 7.396568775177002,
+      "grad_norm": 8.200620651245117,
       "learning_rate": 9.176176176176176e-06,
-      "loss": 1.8176,
+      "loss": 1.8254,
       "step": 8230
     },
     {
       "epoch": 24.74,
-      "grad_norm": 10.236464500427246,
+      "grad_norm": 8.45687484741211,
       "learning_rate": 9.175175175175177e-06,
-      "loss": 1.7983,
+      "loss": 1.8069,
       "step": 8240
     },
     {
       "epoch": 24.77,
-      "grad_norm": 8.881173133850098,
+      "grad_norm": 8.146411895751953,
       "learning_rate": 9.174174174174175e-06,
-      "loss": 1.8714,
+      "loss": 1.8696,
       "step": 8250
     },
     {
       "epoch": 24.8,
-      "grad_norm": 6.464287281036377,
+      "grad_norm": 8.38631534576416,
       "learning_rate": 9.173173173173174e-06,
-      "loss": 1.8752,
+      "loss": 1.8807,
       "step": 8260
     },
     {
       "epoch": 24.83,
-      "grad_norm": 7.865005970001221,
+      "grad_norm": 8.0680513381958,
       "learning_rate": 9.172172172172172e-06,
-      "loss": 1.8896,
+      "loss": 1.8962,
       "step": 8270
     },
     {
       "epoch": 24.86,
-      "grad_norm": 7.50204610824585,
+      "grad_norm": 8.227824211120605,
       "learning_rate": 9.171171171171171e-06,
-      "loss": 1.7676,
+      "loss": 1.7693,
       "step": 8280
     },
     {
       "epoch": 24.89,
-      "grad_norm": 10.846785545349121,
+      "grad_norm": 13.215843200683594,
       "learning_rate": 9.170170170170171e-06,
-      "loss": 1.8572,
+      "loss": 1.8684,
       "step": 8290
     },
     {
       "epoch": 24.92,
-      "grad_norm": 9.093427658081055,
+      "grad_norm": 11.591769218444824,
       "learning_rate": 9.16916916916917e-06,
-      "loss": 1.7852,
+      "loss": 1.7984,
       "step": 8300
     },
     {
       "epoch": 24.95,
-      "grad_norm": 6.658755302429199,
+      "grad_norm": 8.797699928283691,
       "learning_rate": 9.168168168168169e-06,
-      "loss": 1.8592,
+      "loss": 1.8686,
       "step": 8310
     },
     {
       "epoch": 24.98,
-      "grad_norm": 8.736414909362793,
+      "grad_norm": 7.465150833129883,
       "learning_rate": 9.167167167167169e-06,
-      "loss": 1.7976,
+      "loss": 1.8123,
       "step": 8320
     },
     {
       "epoch": 25.0,
-      "eval_accuracy": 0.6717,
-      "eval_loss": 1.176579475402832,
-      "eval_runtime": 5.4318,
-      "eval_samples_per_second": 1841.0,
-      "eval_steps_per_second": 7.364,
+      "eval_accuracy": 0.6735,
+      "eval_loss": 1.1830896139144897,
+      "eval_runtime": 5.4966,
+      "eval_samples_per_second": 1819.31,
+      "eval_steps_per_second": 7.277,
       "step": 8325
     },
     {
       "epoch": 25.02,
-      "grad_norm": 7.31077766418457,
+      "grad_norm": 10.629810333251953,
       "learning_rate": 9.166166166166167e-06,
-      "loss": 2.0164,
+      "loss": 2.0092,
       "step": 8330
     },
     {
       "epoch": 25.05,
-      "grad_norm": 7.886953830718994,
+      "grad_norm": 9.781852722167969,
       "learning_rate": 9.165165165165166e-06,
-      "loss": 1.7927,
+      "loss": 1.8073,
       "step": 8340
     },
     {
       "epoch": 25.08,
-      "grad_norm": 8.6497802734375,
+      "grad_norm": 15.45603084564209,
       "learning_rate": 9.164164164164165e-06,
-      "loss": 1.8035,
+      "loss": 1.8063,
       "step": 8350
     },
     {
       "epoch": 25.11,
-      "grad_norm": 7.02109956741333,
+      "grad_norm": 8.895857810974121,
       "learning_rate": 9.163163163163163e-06,
-      "loss": 1.8078,
+      "loss": 1.8221,
       "step": 8360
     },
     {
       "epoch": 25.14,
-      "grad_norm": 7.3801350593566895,
+      "grad_norm": 6.265673637390137,
       "learning_rate": 9.162162162162162e-06,
-      "loss": 1.8212,
+      "loss": 1.832,
       "step": 8370
     },
     {
       "epoch": 25.17,
-      "grad_norm": 7.0469255447387695,
+      "grad_norm": 8.291892051696777,
       "learning_rate": 9.161161161161162e-06,
-      "loss": 1.8564,
+      "loss": 1.8655,
       "step": 8380
     },
     {
       "epoch": 25.2,
-      "grad_norm": 8.309500694274902,
+      "grad_norm": 10.39739990234375,
       "learning_rate": 9.16016016016016e-06,
-      "loss": 1.7984,
+      "loss": 1.8121,
       "step": 8390
     },
     {
       "epoch": 25.23,
-      "grad_norm": 6.364150524139404,
+      "grad_norm": 7.466885089874268,
       "learning_rate": 9.15915915915916e-06,
-      "loss": 1.7311,
+      "loss": 1.7373,
       "step": 8400
     },
     {
       "epoch": 25.26,
-      "grad_norm": 7.846608638763428,
+      "grad_norm": 9.492599487304688,
       "learning_rate": 9.15815815815816e-06,
-      "loss": 1.7596,
+      "loss": 1.7743,
       "step": 8410
     },
     {
       "epoch": 25.29,
-      "grad_norm": 6.912028789520264,
+      "grad_norm": 6.847721099853516,
       "learning_rate": 9.157157157157158e-06,
-      "loss": 1.8178,
+      "loss": 1.8214,
       "step": 8420
     },
     {
       "epoch": 25.32,
-      "grad_norm": 7.882516860961914,
+      "grad_norm": 12.461723327636719,
       "learning_rate": 9.156156156156157e-06,
-      "loss": 1.8298,
+      "loss": 1.8556,
       "step": 8430
     },
     {
       "epoch": 25.35,
-      "grad_norm": 6.891097068786621,
+      "grad_norm": 8.424966812133789,
       "learning_rate": 9.155155155155155e-06,
-      "loss": 1.8236,
+      "loss": 1.8245,
       "step": 8440
     },
     {
       "epoch": 25.38,
-      "grad_norm": 8.106797218322754,
+      "grad_norm": 7.22593879699707,
       "learning_rate": 9.154154154154154e-06,
-      "loss": 1.8534,
+      "loss": 1.8749,
       "step": 8450
     },
     {
       "epoch": 25.41,
-      "grad_norm": 8.248673439025879,
+      "grad_norm": 9.647261619567871,
       "learning_rate": 9.153153153153154e-06,
-      "loss": 1.8366,
+      "loss": 1.8452,
       "step": 8460
     },
     {
       "epoch": 25.44,
-      "grad_norm": 6.39265251159668,
+      "grad_norm": 5.768435955047607,
       "learning_rate": 9.152152152152153e-06,
-      "loss": 1.7796,
+      "loss": 1.7964,
       "step": 8470
     },
     {
       "epoch": 25.47,
-      "grad_norm": 6.875210285186768,
+      "grad_norm": 6.7405900955200195,
       "learning_rate": 9.151151151151151e-06,
-      "loss": 1.8025,
+      "loss": 1.8107,
       "step": 8480
     },
     {
       "epoch": 25.5,
-      "grad_norm": 7.108837604522705,
+      "grad_norm": 7.90836763381958,
       "learning_rate": 9.150150150150152e-06,
-      "loss": 1.8481,
+      "loss": 1.8505,
       "step": 8490
     },
     {
       "epoch": 25.53,
-      "grad_norm": 7.287178039550781,
+      "grad_norm": 8.313355445861816,
       "learning_rate": 9.14914914914915e-06,
-      "loss": 1.8343,
+      "loss": 1.8435,
       "step": 8500
     },
     {
       "epoch": 25.56,
-      "grad_norm": 7.0152130126953125,
+      "grad_norm": 7.732263088226318,
       "learning_rate": 9.148148148148149e-06,
-      "loss": 1.8265,
+      "loss": 1.8406,
       "step": 8510
     },
     {
       "epoch": 25.59,
-      "grad_norm": 6.357172012329102,
+      "grad_norm": 7.891409397125244,
       "learning_rate": 9.147147147147147e-06,
-      "loss": 1.7587,
+      "loss": 1.7737,
       "step": 8520
     },
     {
       "epoch": 25.62,
-      "grad_norm": 6.544309616088867,
+      "grad_norm": 7.634026050567627,
       "learning_rate": 9.146146146146146e-06,
-      "loss": 1.8137,
+      "loss": 1.8238,
       "step": 8530
     },
     {
       "epoch": 25.65,
-      "grad_norm": 7.235463619232178,
+      "grad_norm": 12.721009254455566,
       "learning_rate": 9.145145145145146e-06,
-      "loss": 1.8167,
+      "loss": 1.8365,
       "step": 8540
     },
     {
       "epoch": 25.68,
-      "grad_norm": 7.371089458465576,
+      "grad_norm": 8.244935989379883,
       "learning_rate": 9.144144144144145e-06,
-      "loss": 1.8466,
+      "loss": 1.8603,
       "step": 8550
     },
     {
       "epoch": 25.71,
-      "grad_norm": 9.439956665039062,
+      "grad_norm": 13.242118835449219,
       "learning_rate": 9.143143143143144e-06,
-      "loss": 1.7719,
+      "loss": 1.7782,
       "step": 8560
     },
     {
       "epoch": 25.74,
-      "grad_norm": 5.945064544677734,
+      "grad_norm": 9.169181823730469,
       "learning_rate": 9.142142142142144e-06,
-      "loss": 1.8384,
+      "loss": 1.8454,
       "step": 8570
     },
     {
       "epoch": 25.77,
-      "grad_norm": 10.369516372680664,
+      "grad_norm": 8.59010124206543,
       "learning_rate": 9.141141141141142e-06,
-      "loss": 1.835,
+      "loss": 1.8332,
       "step": 8580
     },
     {
       "epoch": 25.8,
-      "grad_norm": 7.482779026031494,
+      "grad_norm": 9.20497989654541,
       "learning_rate": 9.140140140140141e-06,
-      "loss": 1.8389,
+      "loss": 1.8473,
       "step": 8590
     },
     {
       "epoch": 25.83,
-      "grad_norm": 6.8010478019714355,
+      "grad_norm": 7.991240978240967,
       "learning_rate": 9.13913913913914e-06,
-      "loss": 1.8101,
+      "loss": 1.8152,
       "step": 8600
     },
     {
       "epoch": 25.86,
-      "grad_norm": 7.800801753997803,
+      "grad_norm": 6.619584560394287,
       "learning_rate": 9.138138138138138e-06,
-      "loss": 1.7968,
+      "loss": 1.8027,
       "step": 8610
     },
     {
       "epoch": 25.89,
-      "grad_norm": 6.764025688171387,
+      "grad_norm": 9.839620590209961,
       "learning_rate": 9.137137137137137e-06,
-      "loss": 1.7888,
+      "loss": 1.8088,
       "step": 8620
     },
     {
       "epoch": 25.92,
-      "grad_norm": 8.204251289367676,
+      "grad_norm": 9.542078018188477,
       "learning_rate": 9.136136136136137e-06,
-      "loss": 1.7831,
+      "loss": 1.772,
       "step": 8630
     },
     {
       "epoch": 25.95,
-      "grad_norm": 10.505181312561035,
+      "grad_norm": 11.688754081726074,
       "learning_rate": 9.135135135135136e-06,
-      "loss": 1.8094,
+      "loss": 1.8134,
       "step": 8640
     },
     {
       "epoch": 25.98,
-      "grad_norm": 7.471145153045654,
+      "grad_norm": 8.518573760986328,
       "learning_rate": 9.134134134134134e-06,
-      "loss": 1.7482,
+      "loss": 1.7624,
       "step": 8650
     },
     {
       "epoch": 26.0,
-      "eval_accuracy": 0.6758,
-      "eval_loss": 1.1570353507995605,
-      "eval_runtime": 5.7046,
-      "eval_samples_per_second": 1752.963,
-      "eval_steps_per_second": 7.012,
+      "eval_accuracy": 0.6715,
+      "eval_loss": 1.1654822826385498,
+      "eval_runtime": 5.5101,
+      "eval_samples_per_second": 1814.843,
+      "eval_steps_per_second": 7.259,
       "step": 8658
     },
     {
       "epoch": 26.01,
-      "grad_norm": 7.6260857582092285,
+      "grad_norm": 9.059782981872559,
       "learning_rate": 9.133133133133135e-06,
-      "loss": 1.9483,
+      "loss": 1.9559,
       "step": 8660
     },
     {
       "epoch": 26.04,
-      "grad_norm": 12.416500091552734,
+      "grad_norm": 15.476532936096191,
       "learning_rate": 9.132132132132133e-06,
-      "loss": 1.7676,
+      "loss": 1.7702,
       "step": 8670
     },
     {
       "epoch": 26.07,
-      "grad_norm": 9.7238130569458,
+      "grad_norm": 6.848936080932617,
       "learning_rate": 9.131131131131132e-06,
-      "loss": 1.7514,
+      "loss": 1.7626,
       "step": 8680
     },
     {
       "epoch": 26.1,
-      "grad_norm": 8.491373062133789,
+      "grad_norm": 7.2357378005981445,
       "learning_rate": 9.13013013013013e-06,
-      "loss": 1.7973,
+      "loss": 1.81,
       "step": 8690
     },
     {
       "epoch": 26.13,
-      "grad_norm": 5.7274346351623535,
+      "grad_norm": 5.496184825897217,
       "learning_rate": 9.129129129129129e-06,
-      "loss": 1.7657,
+      "loss": 1.7718,
       "step": 8700
     },
     {
       "epoch": 26.16,
-      "grad_norm": 5.849602699279785,
+      "grad_norm": 7.6372857093811035,
       "learning_rate": 9.12812812812813e-06,
-      "loss": 1.7756,
+      "loss": 1.7791,
       "step": 8710
     },
     {
       "epoch": 26.19,
-      "grad_norm": 6.986069202423096,
+      "grad_norm": 9.139528274536133,
       "learning_rate": 9.127127127127128e-06,
-      "loss": 1.8071,
+      "loss": 1.8163,
       "step": 8720
     },
     {
       "epoch": 26.22,
-      "grad_norm": 9.834226608276367,
+      "grad_norm": 10.853702545166016,
       "learning_rate": 9.126126126126126e-06,
-      "loss": 1.777,
+      "loss": 1.7744,
       "step": 8730
     },
     {
       "epoch": 26.25,
-      "grad_norm": 11.326507568359375,
+      "grad_norm": 14.611876487731934,
       "learning_rate": 9.125125125125127e-06,
-      "loss": 1.7582,
+      "loss": 1.7672,
       "step": 8740
     },
     {
       "epoch": 26.28,
-      "grad_norm": 7.117760181427002,
+      "grad_norm": 6.252352237701416,
       "learning_rate": 9.124124124124125e-06,
-      "loss": 1.7972,
+      "loss": 1.7946,
       "step": 8750
     },
     {
       "epoch": 26.31,
-      "grad_norm": 6.958147048950195,
+      "grad_norm": 7.175439834594727,
       "learning_rate": 9.123123123123124e-06,
-      "loss": 1.8023,
+      "loss": 1.8278,
       "step": 8760
     },
     {
       "epoch": 26.34,
-      "grad_norm": 8.95068645477295,
+      "grad_norm": 8.463218688964844,
       "learning_rate": 9.122122122122123e-06,
-      "loss": 1.7916,
+      "loss": 1.8079,
       "step": 8770
     },
     {
       "epoch": 26.37,
-      "grad_norm": 9.908475875854492,
+      "grad_norm": 7.664309024810791,
       "learning_rate": 9.121121121121121e-06,
-      "loss": 1.8769,
+      "loss": 1.8796,
       "step": 8780
     },
     {
       "epoch": 26.4,
-      "grad_norm": 10.984182357788086,
+      "grad_norm": 8.3771390914917,
       "learning_rate": 9.120120120120121e-06,
-      "loss": 1.8009,
+      "loss": 1.8147,
       "step": 8790
     },
     {
       "epoch": 26.43,
-      "grad_norm": 8.97208309173584,
+      "grad_norm": 9.644938468933105,
       "learning_rate": 9.11911911911912e-06,
-      "loss": 1.8602,
+      "loss": 1.8731,
       "step": 8800
     },
     {
       "epoch": 26.46,
-      "grad_norm": 9.823659896850586,
+      "grad_norm": 9.060426712036133,
       "learning_rate": 9.118118118118119e-06,
-      "loss": 1.7874,
+      "loss": 1.7954,
       "step": 8810
     },
     {
       "epoch": 26.49,
-      "grad_norm": 8.17683219909668,
+      "grad_norm": 7.978847026824951,
       "learning_rate": 9.117117117117117e-06,
-      "loss": 1.8422,
+      "loss": 1.8441,
       "step": 8820
     },
     {
       "epoch": 26.52,
-      "grad_norm": 5.831955909729004,
+      "grad_norm": 7.893728256225586,
       "learning_rate": 9.116116116116117e-06,
-      "loss": 1.8147,
+      "loss": 1.8168,
       "step": 8830
     },
     {
       "epoch": 26.55,
-      "grad_norm": 7.67328405380249,
+      "grad_norm": 6.750621318817139,
       "learning_rate": 9.115115115115116e-06,
-      "loss": 1.8093,
+      "loss": 1.8133,
       "step": 8840
     },
     {
       "epoch": 26.58,
-      "grad_norm": 9.431927680969238,
+      "grad_norm": 10.56161880493164,
       "learning_rate": 9.114114114114115e-06,
-      "loss": 1.8144,
+      "loss": 1.8186,
       "step": 8850
     },
     {
       "epoch": 26.61,
-      "grad_norm": 7.625401020050049,
+      "grad_norm": 7.856087684631348,
       "learning_rate": 9.113113113113113e-06,
-      "loss": 1.7898,
+      "loss": 1.7906,
       "step": 8860
     },
     {
       "epoch": 26.64,
-      "grad_norm": 6.7809834480285645,
+      "grad_norm": 7.637779235839844,
       "learning_rate": 9.112112112112112e-06,
-      "loss": 1.8183,
+      "loss": 1.8212,
       "step": 8870
     },
     {
       "epoch": 26.67,
-      "grad_norm": 6.861217021942139,
+      "grad_norm": 6.842888355255127,
       "learning_rate": 9.111111111111112e-06,
-      "loss": 1.8033,
+      "loss": 1.8073,
       "step": 8880
     },
     {
       "epoch": 26.7,
-      "grad_norm": 6.667482852935791,
+      "grad_norm": 5.930503845214844,
       "learning_rate": 9.11011011011011e-06,
-      "loss": 1.7059,
+      "loss": 1.7035,
       "step": 8890
     },
     {
       "epoch": 26.73,
-      "grad_norm": 7.703874588012695,
+      "grad_norm": 7.580901622772217,
       "learning_rate": 9.10910910910911e-06,
-      "loss": 1.76,
+      "loss": 1.7755,
       "step": 8900
     },
     {
       "epoch": 26.76,
-      "grad_norm": 8.709485054016113,
+      "grad_norm": 8.710724830627441,
       "learning_rate": 9.10810810810811e-06,
-      "loss": 1.7333,
+      "loss": 1.7276,
       "step": 8910
     },
     {
       "epoch": 26.79,
-      "grad_norm": 6.009936332702637,
+      "grad_norm": 9.85636043548584,
       "learning_rate": 9.107107107107108e-06,
-      "loss": 1.7712,
+      "loss": 1.7823,
       "step": 8920
     },
     {
       "epoch": 26.82,
-      "grad_norm": 6.315674781799316,
+      "grad_norm": 6.663936614990234,
       "learning_rate": 9.106106106106107e-06,
-      "loss": 1.7923,
+      "loss": 1.8038,
       "step": 8930
     },
     {
       "epoch": 26.85,
-      "grad_norm": 8.130025863647461,
+      "grad_norm": 7.114321231842041,
       "learning_rate": 9.105105105105105e-06,
-      "loss": 1.7118,
+      "loss": 1.7124,
       "step": 8940
     },
     {
       "epoch": 26.88,
-      "grad_norm": 9.546934127807617,
+      "grad_norm": 8.851607322692871,
       "learning_rate": 9.104104104104104e-06,
-      "loss": 1.7946,
+      "loss": 1.7907,
       "step": 8950
     },
     {
       "epoch": 26.91,
-      "grad_norm": 6.707799911499023,
+      "grad_norm": 6.458409309387207,
       "learning_rate": 9.103103103103104e-06,
-      "loss": 1.7493,
+      "loss": 1.7572,
       "step": 8960
     },
     {
       "epoch": 26.94,
-      "grad_norm": 7.6559739112854,
+      "grad_norm": 7.982229232788086,
       "learning_rate": 9.102102102102103e-06,
-      "loss": 1.7405,
+      "loss": 1.7487,
       "step": 8970
     },
     {
       "epoch": 26.97,
-      "grad_norm": 7.1334757804870605,
+      "grad_norm": 6.0256218910217285,
       "learning_rate": 9.101101101101101e-06,
-      "loss": 1.8102,
+      "loss": 1.819,
       "step": 8980
     },
     {
       "epoch": 27.0,
-      "grad_norm": 6.531464576721191,
+      "grad_norm": 7.734681606292725,
       "learning_rate": 9.100100100100102e-06,
-      "loss": 1.7816,
+      "loss": 1.7957,
       "step": 8990
     },
     {
       "epoch": 27.0,
-      "eval_accuracy": 0.6834,
-      "eval_loss": 1.123684048652649,
-      "eval_runtime": 5.2407,
-      "eval_samples_per_second": 1908.129,
-      "eval_steps_per_second": 7.633,
+      "eval_accuracy": 0.6856,
+      "eval_loss": 1.123692512512207,
+      "eval_runtime": 5.4927,
+      "eval_samples_per_second": 1820.588,
+      "eval_steps_per_second": 7.282,
       "step": 8991
     },
     {
       "epoch": 27.03,
-      "grad_norm": 7.02097749710083,
+      "grad_norm": 7.8650946617126465,
       "learning_rate": 9.0990990990991e-06,
-      "loss": 1.8238,
+      "loss": 1.8257,
       "step": 9000
     },
     {
       "epoch": 27.06,
-      "grad_norm": 7.801182270050049,
+      "grad_norm": 7.652089595794678,
       "learning_rate": 9.098098098098099e-06,
-      "loss": 1.659,
+      "loss": 1.6546,
       "step": 9010
     },
     {
       "epoch": 27.09,
-      "grad_norm": 7.200788497924805,
+      "grad_norm": 6.181273937225342,
       "learning_rate": 9.097097097097098e-06,
-      "loss": 1.7559,
+      "loss": 1.7649,
       "step": 9020
     },
     {
       "epoch": 27.12,
-      "grad_norm": 7.835607528686523,
+      "grad_norm": 11.663044929504395,
       "learning_rate": 9.096096096096096e-06,
-      "loss": 1.761,
+      "loss": 1.7688,
       "step": 9030
     },
     {
       "epoch": 27.15,
-      "grad_norm": 9.451446533203125,
+      "grad_norm": 12.759391784667969,
       "learning_rate": 9.095095095095095e-06,
-      "loss": 1.7847,
+      "loss": 1.7972,
       "step": 9040
     },
     {
       "epoch": 27.18,
-      "grad_norm": 9.462061882019043,
+      "grad_norm": 7.982667446136475,
       "learning_rate": 9.094094094094095e-06,
-      "loss": 1.787,
+      "loss": 1.7923,
       "step": 9050
     },
     {
       "epoch": 27.21,
-      "grad_norm": 5.129838466644287,
+      "grad_norm": 5.6103105545043945,
       "learning_rate": 9.093093093093094e-06,
-      "loss": 1.7007,
+      "loss": 1.7174,
       "step": 9060
     },
     {
       "epoch": 27.24,
-      "grad_norm": 9.388571739196777,
+      "grad_norm": 8.725183486938477,
       "learning_rate": 9.092092092092092e-06,
-      "loss": 1.7497,
+      "loss": 1.7575,
       "step": 9070
     },
     {
       "epoch": 27.27,
-      "grad_norm": 6.9084672927856445,
+      "grad_norm": 9.752557754516602,
       "learning_rate": 9.091091091091093e-06,
-      "loss": 1.6704,
+      "loss": 1.6819,
       "step": 9080
     },
     {
       "epoch": 27.3,
-      "grad_norm": 5.89355993270874,
+      "grad_norm": 7.420619964599609,
       "learning_rate": 9.090090090090091e-06,
-      "loss": 1.7472,
+      "loss": 1.7536,
       "step": 9090
     },
     {
       "epoch": 27.33,
-      "grad_norm": 8.766653060913086,
+      "grad_norm": 8.830633163452148,
       "learning_rate": 9.08908908908909e-06,
-      "loss": 1.7097,
+      "loss": 1.7123,
       "step": 9100
     },
     {
       "epoch": 27.36,
-      "grad_norm": 7.57033109664917,
+      "grad_norm": 8.225183486938477,
       "learning_rate": 9.088088088088088e-06,
-      "loss": 1.7617,
+      "loss": 1.7739,
       "step": 9110
     },
     {
       "epoch": 27.39,
-      "grad_norm": 7.167114734649658,
+      "grad_norm": 9.65100383758545,
       "learning_rate": 9.087087087087087e-06,
-      "loss": 1.7333,
+      "loss": 1.729,
       "step": 9120
     },
     {
       "epoch": 27.42,
-      "grad_norm": 6.668962001800537,
+      "grad_norm": 6.830539703369141,
       "learning_rate": 9.086086086086087e-06,
-      "loss": 1.8198,
+      "loss": 1.8197,
       "step": 9130
     },
     {
       "epoch": 27.45,
-      "grad_norm": 9.596693992614746,
+      "grad_norm": 8.972063064575195,
       "learning_rate": 9.085085085085086e-06,
-      "loss": 1.778,
+      "loss": 1.7744,
       "step": 9140
     },
     {
       "epoch": 27.48,
-      "grad_norm": 9.254670143127441,
+      "grad_norm": 10.198172569274902,
       "learning_rate": 9.084084084084084e-06,
-      "loss": 1.8224,
+      "loss": 1.837,
       "step": 9150
     },
     {
       "epoch": 27.51,
-      "grad_norm": 8.89088249206543,
+      "grad_norm": 8.129541397094727,
       "learning_rate": 9.083083083083085e-06,
-      "loss": 1.8149,
+      "loss": 1.8218,
       "step": 9160
     },
     {
       "epoch": 27.54,
-      "grad_norm": 8.225218772888184,
+      "grad_norm": 9.71297836303711,
       "learning_rate": 9.082082082082083e-06,
-      "loss": 1.76,
+      "loss": 1.7679,
       "step": 9170
     },
     {
       "epoch": 27.57,
-      "grad_norm": 8.16159725189209,
+      "grad_norm": 10.98694896697998,
       "learning_rate": 9.081081081081082e-06,
-      "loss": 1.7098,
+      "loss": 1.7133,
       "step": 9180
     },
     {
       "epoch": 27.6,
-      "grad_norm": 9.041584968566895,
+      "grad_norm": 9.306821823120117,
       "learning_rate": 9.08008008008008e-06,
-      "loss": 1.7114,
+      "loss": 1.7179,
       "step": 9190
     },
     {
       "epoch": 27.63,
-      "grad_norm": 9.658296585083008,
+      "grad_norm": 9.04599666595459,
       "learning_rate": 9.079079079079079e-06,
-      "loss": 1.7314,
+      "loss": 1.7363,
       "step": 9200
     },
     {
       "epoch": 27.66,
-      "grad_norm": 8.705323219299316,
+      "grad_norm": 6.439456939697266,
       "learning_rate": 9.07807807807808e-06,
-      "loss": 1.735,
+      "loss": 1.7511,
       "step": 9210
     },
     {
       "epoch": 27.69,
-      "grad_norm": 7.628717422485352,
+      "grad_norm": 8.105624198913574,
       "learning_rate": 9.077077077077078e-06,
-      "loss": 1.7743,
+      "loss": 1.7846,
       "step": 9220
     },
     {
       "epoch": 27.72,
-      "grad_norm": 7.3983283042907715,
+      "grad_norm": 6.390106678009033,
       "learning_rate": 9.076076076076077e-06,
-      "loss": 1.7984,
+      "loss": 1.7971,
       "step": 9230
     },
     {
       "epoch": 27.75,
-      "grad_norm": 6.301623821258545,
+      "grad_norm": 5.531813144683838,
       "learning_rate": 9.075075075075077e-06,
-      "loss": 1.735,
+      "loss": 1.7454,
       "step": 9240
     },
     {
       "epoch": 27.78,
-      "grad_norm": 7.014017581939697,
+      "grad_norm": 7.704046726226807,
       "learning_rate": 9.074074074074075e-06,
-      "loss": 1.795,
+      "loss": 1.8015,
       "step": 9250
     },
     {
       "epoch": 27.81,
-      "grad_norm": 8.455873489379883,
+      "grad_norm": 10.89915657043457,
       "learning_rate": 9.073073073073074e-06,
-      "loss": 1.7007,
+      "loss": 1.7114,
       "step": 9260
     },
     {
       "epoch": 27.84,
-      "grad_norm": 8.07751178741455,
+      "grad_norm": 12.395249366760254,
       "learning_rate": 9.072072072072073e-06,
-      "loss": 1.7545,
+      "loss": 1.7528,
       "step": 9270
     },
     {
       "epoch": 27.87,
-      "grad_norm": 7.194076061248779,
+      "grad_norm": 8.898442268371582,
       "learning_rate": 9.071071071071071e-06,
-      "loss": 1.7633,
+      "loss": 1.7819,
       "step": 9280
     },
     {
       "epoch": 27.9,
-      "grad_norm": 7.5318169593811035,
+      "grad_norm": 9.704859733581543,
       "learning_rate": 9.07007007007007e-06,
-      "loss": 1.8045,
+      "loss": 1.805,
       "step": 9290
     },
     {
       "epoch": 27.93,
-      "grad_norm": 5.108814239501953,
+      "grad_norm": 7.51887321472168,
       "learning_rate": 9.06906906906907e-06,
-      "loss": 1.7259,
+      "loss": 1.7247,
       "step": 9300
     },
     {
       "epoch": 27.96,
-      "grad_norm": 7.8756632804870605,
+      "grad_norm": 8.715901374816895,
       "learning_rate": 9.068068068068069e-06,
-      "loss": 1.7592,
+      "loss": 1.7724,
       "step": 9310
     },
     {
       "epoch": 27.99,
-      "grad_norm": 7.020995140075684,
+      "grad_norm": 9.742772102355957,
       "learning_rate": 9.067067067067067e-06,
-      "loss": 1.7477,
+      "loss": 1.7588,
       "step": 9320
     },
     {
       "epoch": 28.0,
-      "eval_accuracy": 0.6878,
-      "eval_loss": 1.1026580333709717,
-      "eval_runtime": 5.4097,
-      "eval_samples_per_second": 1848.528,
-      "eval_steps_per_second": 7.394,
+      "eval_accuracy": 0.6859,
+      "eval_loss": 1.1009823083877563,
+      "eval_runtime": 5.2276,
+      "eval_samples_per_second": 1912.917,
+      "eval_steps_per_second": 7.652,
       "step": 9324
     },
     {
       "epoch": 28.02,
-      "grad_norm": 10.079315185546875,
+      "grad_norm": 8.9668550491333,
       "learning_rate": 9.066066066066068e-06,
-      "loss": 1.9779,
+      "loss": 1.963,
       "step": 9330
     },
     {
       "epoch": 28.05,
-      "grad_norm": 7.184052467346191,
+      "grad_norm": 10.400721549987793,
       "learning_rate": 9.065065065065066e-06,
-      "loss": 1.7277,
+      "loss": 1.7265,
       "step": 9340
     },
     {
       "epoch": 28.08,
-      "grad_norm": 7.13986349105835,
+      "grad_norm": 6.193258762359619,
       "learning_rate": 9.064064064064065e-06,
-      "loss": 1.7598,
+      "loss": 1.7742,
       "step": 9350
     },
     {
       "epoch": 28.11,
-      "grad_norm": 10.388175010681152,
+      "grad_norm": 8.625391006469727,
       "learning_rate": 9.063063063063063e-06,
-      "loss": 1.7519,
+      "loss": 1.7399,
       "step": 9360
     },
     {
       "epoch": 28.14,
-      "grad_norm": 7.076479911804199,
+      "grad_norm": 8.742112159729004,
       "learning_rate": 9.062062062062062e-06,
-      "loss": 1.7311,
+      "loss": 1.7427,
       "step": 9370
     },
     {
       "epoch": 28.17,
-      "grad_norm": 8.924582481384277,
+      "grad_norm": 8.483949661254883,
       "learning_rate": 9.061061061061062e-06,
-      "loss": 1.7096,
+      "loss": 1.7104,
       "step": 9380
     },
     {
       "epoch": 28.2,
-      "grad_norm": 10.1865234375,
+      "grad_norm": 7.380753040313721,
       "learning_rate": 9.06006006006006e-06,
-      "loss": 1.7214,
+      "loss": 1.7163,
       "step": 9390
     },
     {
       "epoch": 28.23,
-      "grad_norm": 6.21671724319458,
+      "grad_norm": 5.625154972076416,
       "learning_rate": 9.05905905905906e-06,
-      "loss": 1.7267,
+      "loss": 1.7272,
       "step": 9400
     },
     {
       "epoch": 28.26,
-      "grad_norm": 6.089287281036377,
+      "grad_norm": 4.981650352478027,
       "learning_rate": 9.05805805805806e-06,
-      "loss": 1.6493,
+      "loss": 1.6482,
       "step": 9410
     },
     {
       "epoch": 28.29,
-      "grad_norm": 11.428766250610352,
+      "grad_norm": 12.720117568969727,
       "learning_rate": 9.057057057057058e-06,
-      "loss": 1.7822,
+      "loss": 1.7855,
       "step": 9420
     },
     {
       "epoch": 28.32,
-      "grad_norm": 5.123597621917725,
+      "grad_norm": 7.2715744972229,
       "learning_rate": 9.056056056056057e-06,
-      "loss": 1.7163,
+      "loss": 1.7154,
       "step": 9430
     },
     {
       "epoch": 28.35,
-      "grad_norm": 5.80772590637207,
+      "grad_norm": 6.866278171539307,
       "learning_rate": 9.055055055055055e-06,
-      "loss": 1.7095,
+      "loss": 1.707,
       "step": 9440
     },
     {
       "epoch": 28.38,
-      "grad_norm": 6.754103660583496,
+      "grad_norm": 7.82720422744751,
       "learning_rate": 9.054054054054054e-06,
-      "loss": 1.753,
+      "loss": 1.7602,
       "step": 9450
     },
     {
       "epoch": 28.41,
-      "grad_norm": 6.111235618591309,
+      "grad_norm": 6.017468452453613,
       "learning_rate": 9.053053053053054e-06,
-      "loss": 1.6978,
+      "loss": 1.6976,
       "step": 9460
     },
     {
       "epoch": 28.44,
-      "grad_norm": 7.607273101806641,
+      "grad_norm": 8.466320991516113,
       "learning_rate": 9.052052052052053e-06,
-      "loss": 1.737,
+      "loss": 1.7325,
       "step": 9470
     },
     {
       "epoch": 28.47,
-      "grad_norm": 6.403199672698975,
+      "grad_norm": 6.09275484085083,
       "learning_rate": 9.051051051051052e-06,
-      "loss": 1.725,
+      "loss": 1.7293,
       "step": 9480
     },
     {
       "epoch": 28.5,
-      "grad_norm": 7.848564147949219,
+      "grad_norm": 11.611666679382324,
       "learning_rate": 9.05005005005005e-06,
-      "loss": 1.7878,
+      "loss": 1.7928,
       "step": 9490
     },
     {
       "epoch": 28.53,
-      "grad_norm": 6.684571743011475,
+      "grad_norm": 7.774081707000732,
       "learning_rate": 9.04904904904905e-06,
-      "loss": 1.6504,
+      "loss": 1.6596,
       "step": 9500
     },
     {
       "epoch": 28.56,
-      "grad_norm": 5.80443000793457,
+      "grad_norm": 5.685336112976074,
       "learning_rate": 9.048048048048049e-06,
-      "loss": 1.7718,
+      "loss": 1.7715,
       "step": 9510
     },
     {
       "epoch": 28.59,
-      "grad_norm": 6.921514511108398,
+      "grad_norm": 11.206892967224121,
       "learning_rate": 9.047047047047048e-06,
-      "loss": 1.7323,
+      "loss": 1.7327,
       "step": 9520
     },
     {
       "epoch": 28.62,
-      "grad_norm": 9.441984176635742,
+      "grad_norm": 9.12618350982666,
       "learning_rate": 9.046046046046046e-06,
-      "loss": 1.7843,
+      "loss": 1.8031,
       "step": 9530
     },
     {
       "epoch": 28.65,
-      "grad_norm": 6.991013526916504,
+      "grad_norm": 6.53380823135376,
       "learning_rate": 9.045045045045045e-06,
-      "loss": 1.6507,
+      "loss": 1.6677,
       "step": 9540
     },
     {
       "epoch": 28.68,
-      "grad_norm": 8.442638397216797,
+      "grad_norm": 9.07605266571045,
       "learning_rate": 9.044044044044045e-06,
-      "loss": 1.7288,
+      "loss": 1.7208,
       "step": 9550
     },
     {
       "epoch": 28.71,
-      "grad_norm": 6.908164978027344,
+      "grad_norm": 5.095149040222168,
       "learning_rate": 9.043043043043044e-06,
-      "loss": 1.7845,
+      "loss": 1.7941,
       "step": 9560
     },
     {
       "epoch": 28.74,
-      "grad_norm": 7.270082950592041,
+      "grad_norm": 9.044610977172852,
       "learning_rate": 9.042042042042042e-06,
-      "loss": 1.8023,
+      "loss": 1.8125,
       "step": 9570
     },
     {
       "epoch": 28.77,
-      "grad_norm": 6.578466892242432,
+      "grad_norm": 7.095623970031738,
       "learning_rate": 9.041041041041043e-06,
-      "loss": 1.7219,
+      "loss": 1.7229,
       "step": 9580
     },
     {
       "epoch": 28.8,
-      "grad_norm": 7.386541843414307,
+      "grad_norm": 9.03903579711914,
       "learning_rate": 9.040040040040041e-06,
-      "loss": 1.81,
+      "loss": 1.8066,
       "step": 9590
     },
     {
       "epoch": 28.83,
-      "grad_norm": 5.1668701171875,
+      "grad_norm": 7.019347667694092,
       "learning_rate": 9.03903903903904e-06,
-      "loss": 1.747,
+      "loss": 1.7543,
       "step": 9600
     },
     {
       "epoch": 28.86,
-      "grad_norm": 8.899998664855957,
+      "grad_norm": 7.536007404327393,
       "learning_rate": 9.038038038038038e-06,
-      "loss": 1.7081,
+      "loss": 1.7068,
       "step": 9610
     },
     {
       "epoch": 28.89,
-      "grad_norm": 7.453086853027344,
+      "grad_norm": 6.736332893371582,
       "learning_rate": 9.037037037037037e-06,
-      "loss": 1.7242,
+      "loss": 1.7385,
       "step": 9620
     },
     {
       "epoch": 28.92,
-      "grad_norm": 7.397784233093262,
+      "grad_norm": 7.68821382522583,
       "learning_rate": 9.036036036036037e-06,
-      "loss": 1.7284,
+      "loss": 1.7424,
       "step": 9630
     },
     {
       "epoch": 28.95,
-      "grad_norm": 6.929391384124756,
+      "grad_norm": 7.129851341247559,
       "learning_rate": 9.035035035035036e-06,
-      "loss": 1.7081,
+      "loss": 1.7072,
       "step": 9640
     },
     {
       "epoch": 28.98,
-      "grad_norm": 6.636610984802246,
+      "grad_norm": 9.356767654418945,
       "learning_rate": 9.034034034034034e-06,
-      "loss": 1.7196,
+      "loss": 1.7242,
       "step": 9650
     },
     {
       "epoch": 29.0,
-      "eval_accuracy": 0.6899,
-      "eval_loss": 1.0759997367858887,
-      "eval_runtime": 5.687,
-      "eval_samples_per_second": 1758.384,
-      "eval_steps_per_second": 7.034,
+      "eval_accuracy": 0.6894,
+      "eval_loss": 1.076327919960022,
+      "eval_runtime": 5.6043,
+      "eval_samples_per_second": 1784.356,
+      "eval_steps_per_second": 7.137,
       "step": 9657
     },
     {
       "epoch": 29.01,
-      "grad_norm": 8.175017356872559,
+      "grad_norm": 7.7339277267456055,
       "learning_rate": 9.033033033033035e-06,
-      "loss": 1.8425,
+      "loss": 1.8649,
       "step": 9660
     },
     {
       "epoch": 29.04,
-      "grad_norm": 7.753376007080078,
+      "grad_norm": 8.382436752319336,
       "learning_rate": 9.032032032032033e-06,
-      "loss": 1.6944,
+      "loss": 1.6937,
       "step": 9670
     },
     {
       "epoch": 29.07,
-      "grad_norm": 7.5159149169921875,
+      "grad_norm": 7.698642253875732,
       "learning_rate": 9.031031031031032e-06,
-      "loss": 1.6812,
+      "loss": 1.6759,
       "step": 9680
     },
     {
       "epoch": 29.1,
-      "grad_norm": 7.754020690917969,
+      "grad_norm": 6.7065815925598145,
       "learning_rate": 9.03003003003003e-06,
-      "loss": 1.7137,
+      "loss": 1.7201,
       "step": 9690
     },
     {
       "epoch": 29.13,
-      "grad_norm": 7.7920145988464355,
+      "grad_norm": 11.962173461914062,
       "learning_rate": 9.029029029029029e-06,
-      "loss": 1.6665,
+      "loss": 1.6691,
       "step": 9700
     },
     {
       "epoch": 29.16,
-      "grad_norm": 5.573378562927246,
+      "grad_norm": 6.899008274078369,
       "learning_rate": 9.02802802802803e-06,
-      "loss": 1.6709,
+      "loss": 1.6746,
       "step": 9710
     },
     {
       "epoch": 29.19,
-      "grad_norm": 7.859186172485352,
+      "grad_norm": 14.638890266418457,
       "learning_rate": 9.027027027027028e-06,
-      "loss": 1.7575,
+      "loss": 1.7607,
       "step": 9720
     },
     {
       "epoch": 29.22,
-      "grad_norm": 7.525823593139648,
+      "grad_norm": 7.114095211029053,
       "learning_rate": 9.026026026026027e-06,
-      "loss": 1.7455,
+      "loss": 1.7448,
       "step": 9730
     },
     {
       "epoch": 29.25,
-      "grad_norm": 4.819455623626709,
+      "grad_norm": 6.181638240814209,
       "learning_rate": 9.025025025025025e-06,
-      "loss": 1.7617,
+      "loss": 1.774,
       "step": 9740
     },
     {
       "epoch": 29.28,
-      "grad_norm": 11.472289085388184,
+      "grad_norm": 10.128156661987305,
       "learning_rate": 9.024024024024025e-06,
-      "loss": 1.6968,
+      "loss": 1.6867,
       "step": 9750
     },
     {
       "epoch": 29.31,
-      "grad_norm": 13.124000549316406,
+      "grad_norm": 11.436808586120605,
       "learning_rate": 9.023023023023024e-06,
-      "loss": 1.7156,
+      "loss": 1.7154,
       "step": 9760
     },
     {
       "epoch": 29.34,
-      "grad_norm": 8.273530006408691,
+      "grad_norm": 7.413613796234131,
       "learning_rate": 9.022022022022023e-06,
-      "loss": 1.7732,
+      "loss": 1.7731,
       "step": 9770
     },
     {
       "epoch": 29.37,
-      "grad_norm": 7.1737494468688965,
+      "grad_norm": 6.0334553718566895,
       "learning_rate": 9.021021021021021e-06,
-      "loss": 1.7028,
+      "loss": 1.7113,
       "step": 9780
     },
     {
       "epoch": 29.4,
-      "grad_norm": 6.626444339752197,
+      "grad_norm": 5.654606342315674,
       "learning_rate": 9.02002002002002e-06,
-      "loss": 1.6927,
+      "loss": 1.7047,
       "step": 9790
     },
     {
       "epoch": 29.43,
-      "grad_norm": 9.833351135253906,
+      "grad_norm": 13.233532905578613,
       "learning_rate": 9.01901901901902e-06,
-      "loss": 1.6942,
+      "loss": 1.7027,
       "step": 9800
     },
     {
       "epoch": 29.46,
-      "grad_norm": 8.089360237121582,
+      "grad_norm": 9.217297554016113,
       "learning_rate": 9.018018018018019e-06,
-      "loss": 1.7306,
+      "loss": 1.7291,
       "step": 9810
     },
     {
       "epoch": 29.49,
-      "grad_norm": 8.657221794128418,
+      "grad_norm": 12.07960033416748,
       "learning_rate": 9.017017017017017e-06,
-      "loss": 1.6856,
+      "loss": 1.6955,
       "step": 9820
     },
     {
       "epoch": 29.52,
-      "grad_norm": 7.481504917144775,
+      "grad_norm": 7.5245680809021,
       "learning_rate": 9.016016016016018e-06,
-      "loss": 1.7027,
+      "loss": 1.7161,
       "step": 9830
     },
     {
       "epoch": 29.55,
-      "grad_norm": 7.467085838317871,
+      "grad_norm": 7.615597248077393,
       "learning_rate": 9.015015015015016e-06,
-      "loss": 1.7735,
+      "loss": 1.7709,
       "step": 9840
     },
     {
       "epoch": 29.58,
-      "grad_norm": 9.374113082885742,
+      "grad_norm": 10.543852806091309,
       "learning_rate": 9.014014014014015e-06,
-      "loss": 1.6734,
+      "loss": 1.6749,
       "step": 9850
     },
     {
       "epoch": 29.61,
-      "grad_norm": 10.505378723144531,
+      "grad_norm": 10.96261978149414,
       "learning_rate": 9.013013013013013e-06,
-      "loss": 1.7649,
+      "loss": 1.7829,
       "step": 9860
     },
     {
       "epoch": 29.64,
-      "grad_norm": 7.740830898284912,
+      "grad_norm": 8.754733085632324,
       "learning_rate": 9.012012012012012e-06,
-      "loss": 1.7053,
+      "loss": 1.7088,
       "step": 9870
     },
     {
       "epoch": 29.67,
-      "grad_norm": 7.172015190124512,
+      "grad_norm": 7.98638916015625,
       "learning_rate": 9.011011011011012e-06,
-      "loss": 1.6628,
+      "loss": 1.6718,
       "step": 9880
     },
     {
       "epoch": 29.7,
-      "grad_norm": 9.867655754089355,
+      "grad_norm": 8.909956932067871,
       "learning_rate": 9.010010010010011e-06,
-      "loss": 1.8168,
+      "loss": 1.8233,
       "step": 9890
     },
     {
       "epoch": 29.73,
-      "grad_norm": 7.0088043212890625,
+      "grad_norm": 7.951484203338623,
       "learning_rate": 9.00900900900901e-06,
-      "loss": 1.6281,
+      "loss": 1.638,
       "step": 9900
     },
     {
       "epoch": 29.76,
-      "grad_norm": 11.327157020568848,
+      "grad_norm": 11.312047004699707,
       "learning_rate": 9.00800800800801e-06,
-      "loss": 1.6838,
+      "loss": 1.6826,
       "step": 9910
     },
     {
       "epoch": 29.79,
-      "grad_norm": 7.8171467781066895,
+      "grad_norm": 8.037393569946289,
       "learning_rate": 9.007007007007008e-06,
-      "loss": 1.6922,
+      "loss": 1.7021,
       "step": 9920
     },
     {
       "epoch": 29.82,
-      "grad_norm": 9.311015129089355,
+      "grad_norm": 8.668597221374512,
       "learning_rate": 9.006006006006007e-06,
-      "loss": 1.7086,
+      "loss": 1.7158,
       "step": 9930
     },
     {
       "epoch": 29.85,
-      "grad_norm": 8.449519157409668,
+      "grad_norm": 7.284761905670166,
       "learning_rate": 9.005005005005006e-06,
-      "loss": 1.7119,
+      "loss": 1.717,
       "step": 9940
     },
     {
       "epoch": 29.88,
-      "grad_norm": 19.058582305908203,
+      "grad_norm": 13.110321998596191,
       "learning_rate": 9.004004004004004e-06,
-      "loss": 1.6787,
+      "loss": 1.6842,
       "step": 9950
     },
     {
       "epoch": 29.91,
-      "grad_norm": 7.669032573699951,
+      "grad_norm": 8.762986183166504,
       "learning_rate": 9.003003003003003e-06,
-      "loss": 1.6973,
+      "loss": 1.6949,
       "step": 9960
     },
     {
       "epoch": 29.94,
-      "grad_norm": 7.411373138427734,
+      "grad_norm": 7.450439453125,
       "learning_rate": 9.002002002002003e-06,
-      "loss": 1.6146,
+      "loss": 1.6246,
       "step": 9970
     },
     {
       "epoch": 29.97,
-      "grad_norm": 7.464860916137695,
+      "grad_norm": 6.835789680480957,
       "learning_rate": 9.001001001001002e-06,
-      "loss": 1.6238,
+      "loss": 1.6266,
       "step": 9980
     },
     {
       "epoch": 30.0,
-      "grad_norm": 46.2182502746582,
+      "grad_norm": 54.914485931396484,
       "learning_rate": 9e-06,
-      "loss": 1.7635,
+      "loss": 1.7695,
       "step": 9990
     },
     {
       "epoch": 30.0,
-      "eval_accuracy": 0.6934,
-      "eval_loss": 1.0599777698516846,
-      "eval_runtime": 5.6019,
-      "eval_samples_per_second": 1785.113,
-      "eval_steps_per_second": 7.14,
+      "eval_accuracy": 0.6954,
+      "eval_loss": 1.0613893270492554,
+      "eval_runtime": 5.338,
+      "eval_samples_per_second": 1873.374,
+      "eval_steps_per_second": 7.493,
       "step": 9990
     },
     {
       "epoch": 30.03,
-      "grad_norm": 9.159845352172852,
+      "grad_norm": 8.464838027954102,
       "learning_rate": 8.998998998999e-06,
-      "loss": 1.7071,
+      "loss": 1.7082,
       "step": 10000
     },
     {
       "epoch": 30.06,
-      "grad_norm": 7.608208656311035,
+      "grad_norm": 6.762968063354492,
       "learning_rate": 8.997997997997999e-06,
-      "loss": 1.7001,
+      "loss": 1.6914,
       "step": 10010
     },
     {
       "epoch": 30.09,
-      "grad_norm": 7.401772499084473,
+      "grad_norm": 7.149596691131592,
       "learning_rate": 8.996996996996998e-06,
-      "loss": 1.6647,
+      "loss": 1.671,
       "step": 10020
     },
     {
       "epoch": 30.12,
-      "grad_norm": 12.511927604675293,
+      "grad_norm": 10.529356956481934,
       "learning_rate": 8.995995995995996e-06,
-      "loss": 1.7322,
+      "loss": 1.7394,
       "step": 10030
     },
     {
       "epoch": 30.15,
-      "grad_norm": 7.9033918380737305,
+      "grad_norm": 8.229801177978516,
       "learning_rate": 8.994994994994995e-06,
-      "loss": 1.6786,
+      "loss": 1.6971,
       "step": 10040
     },
     {
       "epoch": 30.18,
-      "grad_norm": 5.452857494354248,
+      "grad_norm": 9.30200481414795,
       "learning_rate": 8.993993993993995e-06,
-      "loss": 1.6986,
+      "loss": 1.6966,
       "step": 10050
     },
     {
       "epoch": 30.21,
-      "grad_norm": 8.540202140808105,
+      "grad_norm": 9.633055686950684,
       "learning_rate": 8.992992992992994e-06,
-      "loss": 1.7042,
+      "loss": 1.7153,
       "step": 10060
     },
     {
       "epoch": 30.24,
-      "grad_norm": 6.853354454040527,
+      "grad_norm": 7.522928237915039,
       "learning_rate": 8.991991991991992e-06,
-      "loss": 1.5778,
+      "loss": 1.5839,
       "step": 10070
     },
     {
       "epoch": 30.27,
-      "grad_norm": 7.2142791748046875,
+      "grad_norm": 9.9769926071167,
       "learning_rate": 8.990990990990993e-06,
-      "loss": 1.7118,
+      "loss": 1.7155,
       "step": 10080
     },
     {
       "epoch": 30.3,
-      "grad_norm": 5.904212951660156,
+      "grad_norm": 8.61368465423584,
       "learning_rate": 8.989989989989991e-06,
-      "loss": 1.6281,
+      "loss": 1.6315,
       "step": 10090
     },
     {
       "epoch": 30.33,
-      "grad_norm": 12.187846183776855,
+      "grad_norm": 8.370403289794922,
       "learning_rate": 8.98898898898899e-06,
-      "loss": 1.6475,
+      "loss": 1.6478,
       "step": 10100
     },
     {
       "epoch": 30.36,
-      "grad_norm": 9.249743461608887,
+      "grad_norm": 10.865662574768066,
       "learning_rate": 8.987987987987988e-06,
-      "loss": 1.5968,
+      "loss": 1.5972,
       "step": 10110
     },
     {
       "epoch": 30.39,
-      "grad_norm": 8.156797409057617,
+      "grad_norm": 8.828994750976562,
       "learning_rate": 8.986986986986987e-06,
-      "loss": 1.662,
+      "loss": 1.6558,
       "step": 10120
     },
     {
       "epoch": 30.42,
-      "grad_norm": 11.202215194702148,
+      "grad_norm": 9.128555297851562,
       "learning_rate": 8.985985985985987e-06,
-      "loss": 1.6568,
+      "loss": 1.6598,
       "step": 10130
     },
     {
       "epoch": 30.45,
-      "grad_norm": 9.792190551757812,
+      "grad_norm": 7.645933151245117,
       "learning_rate": 8.984984984984986e-06,
-      "loss": 1.7252,
+      "loss": 1.7229,
       "step": 10140
     },
     {
       "epoch": 30.48,
-      "grad_norm": 13.672833442687988,
+      "grad_norm": 17.99295425415039,
       "learning_rate": 8.983983983983985e-06,
-      "loss": 1.6891,
+      "loss": 1.6893,
       "step": 10150
     },
     {
       "epoch": 30.51,
-      "grad_norm": 7.433216094970703,
+      "grad_norm": 10.158817291259766,
       "learning_rate": 8.982982982982985e-06,
-      "loss": 1.7303,
+      "loss": 1.7391,
       "step": 10160
     },
     {
       "epoch": 30.54,
-      "grad_norm": 6.7613115310668945,
+      "grad_norm": 8.00157642364502,
       "learning_rate": 8.981981981981983e-06,
-      "loss": 1.6612,
+      "loss": 1.6657,
       "step": 10170
     },
     {
       "epoch": 30.57,
-      "grad_norm": 9.097322463989258,
+      "grad_norm": 7.760641098022461,
       "learning_rate": 8.980980980980982e-06,
-      "loss": 1.6169,
+      "loss": 1.6173,
       "step": 10180
     },
     {
       "epoch": 30.6,
-      "grad_norm": 8.631768226623535,
+      "grad_norm": 6.671607494354248,
       "learning_rate": 8.97997997997998e-06,
-      "loss": 1.5896,
+      "loss": 1.5971,
       "step": 10190
     },
     {
       "epoch": 30.63,
-      "grad_norm": 5.965922832489014,
+      "grad_norm": 7.5528788566589355,
       "learning_rate": 8.97897897897898e-06,
-      "loss": 1.6604,
+      "loss": 1.674,
       "step": 10200
     },
     {
       "epoch": 30.66,
-      "grad_norm": 6.4168500900268555,
+      "grad_norm": 7.348338603973389,
       "learning_rate": 8.977977977977978e-06,
-      "loss": 1.5926,
+      "loss": 1.6034,
       "step": 10210
     },
     {
       "epoch": 30.69,
-      "grad_norm": 6.274549961090088,
+      "grad_norm": 8.55675983428955,
       "learning_rate": 8.976976976976978e-06,
-      "loss": 1.6257,
+      "loss": 1.6294,
       "step": 10220
     },
     {
       "epoch": 30.72,
-      "grad_norm": 6.658246040344238,
+      "grad_norm": 10.257527351379395,
       "learning_rate": 8.975975975975977e-06,
-      "loss": 1.6696,
+      "loss": 1.6771,
       "step": 10230
     },
     {
       "epoch": 30.75,
-      "grad_norm": 6.870821475982666,
+      "grad_norm": 11.894840240478516,
       "learning_rate": 8.974974974974975e-06,
-      "loss": 1.6323,
+      "loss": 1.6351,
       "step": 10240
     },
     {
       "epoch": 30.78,
-      "grad_norm": 7.907097816467285,
+      "grad_norm": 11.39207935333252,
       "learning_rate": 8.973973973973976e-06,
-      "loss": 1.6342,
+      "loss": 1.6339,
       "step": 10250
     },
     {
       "epoch": 30.81,
-      "grad_norm": 5.353482723236084,
+      "grad_norm": 5.712204456329346,
       "learning_rate": 8.972972972972974e-06,
-      "loss": 1.6722,
+      "loss": 1.6811,
       "step": 10260
     },
     {
       "epoch": 30.84,
-      "grad_norm": 11.02402114868164,
+      "grad_norm": 9.120635986328125,
       "learning_rate": 8.971971971971973e-06,
-      "loss": 1.6854,
+      "loss": 1.687,
       "step": 10270
     },
     {
       "epoch": 30.87,
-      "grad_norm": 7.559629917144775,
+      "grad_norm": 7.552114486694336,
       "learning_rate": 8.970970970970971e-06,
-      "loss": 1.6622,
+      "loss": 1.6639,
       "step": 10280
     },
     {
       "epoch": 30.9,
-      "grad_norm": 8.970190048217773,
+      "grad_norm": 6.039352893829346,
       "learning_rate": 8.96996996996997e-06,
-      "loss": 1.687,
+      "loss": 1.6831,
       "step": 10290
     },
     {
       "epoch": 30.93,
-      "grad_norm": 7.710419654846191,
+      "grad_norm": 8.572236061096191,
       "learning_rate": 8.96896896896897e-06,
-      "loss": 1.6613,
+      "loss": 1.6731,
       "step": 10300
     },
     {
       "epoch": 30.96,
-      "grad_norm": 6.102724552154541,
+      "grad_norm": 7.827875137329102,
       "learning_rate": 8.967967967967969e-06,
-      "loss": 1.6875,
+      "loss": 1.6918,
       "step": 10310
     },
     {
       "epoch": 30.99,
-      "grad_norm": 5.204246997833252,
+      "grad_norm": 9.571944236755371,
       "learning_rate": 8.966966966966967e-06,
-      "loss": 1.6424,
+      "loss": 1.6464,
       "step": 10320
     },
     {
       "epoch": 31.0,
-      "eval_accuracy": 0.6975,
-      "eval_loss": 1.038780689239502,
-      "eval_runtime": 5.3423,
-      "eval_samples_per_second": 1871.867,
-      "eval_steps_per_second": 7.487,
+      "eval_accuracy": 0.7001,
+      "eval_loss": 1.0416522026062012,
+      "eval_runtime": 5.4985,
+      "eval_samples_per_second": 1818.69,
+      "eval_steps_per_second": 7.275,
       "step": 10323
     },
     {
       "epoch": 31.02,
-      "grad_norm": 6.763729095458984,
+      "grad_norm": 8.59007453918457,
       "learning_rate": 8.965965965965968e-06,
-      "loss": 1.8201,
+      "loss": 1.8072,
       "step": 10330
     },
     {
       "epoch": 31.05,
-      "grad_norm": 9.234564781188965,
+      "grad_norm": 7.164886474609375,
       "learning_rate": 8.964964964964966e-06,
-      "loss": 1.7618,
+      "loss": 1.7792,
       "step": 10340
     },
     {
       "epoch": 31.08,
-      "grad_norm": 8.329873085021973,
+      "grad_norm": 7.189478397369385,
       "learning_rate": 8.963963963963965e-06,
-      "loss": 1.7029,
+      "loss": 1.7041,
       "step": 10350
     },
     {
       "epoch": 31.11,
-      "grad_norm": 7.258545875549316,
+      "grad_norm": 10.258389472961426,
       "learning_rate": 8.962962962962963e-06,
-      "loss": 1.6003,
+      "loss": 1.6044,
       "step": 10360
     },
     {
       "epoch": 31.14,
-      "grad_norm": 7.327395439147949,
+      "grad_norm": 7.183008193969727,
       "learning_rate": 8.961961961961962e-06,
-      "loss": 1.6035,
+      "loss": 1.6115,
       "step": 10370
     },
     {
       "epoch": 31.17,
-      "grad_norm": 5.829756736755371,
+      "grad_norm": 7.628407001495361,
       "learning_rate": 8.960960960960962e-06,
-      "loss": 1.6187,
+      "loss": 1.6284,
       "step": 10380
     },
     {
       "epoch": 31.2,
-      "grad_norm": 8.03139591217041,
+      "grad_norm": 6.033533573150635,
       "learning_rate": 8.959959959959961e-06,
-      "loss": 1.6779,
+      "loss": 1.6894,
       "step": 10390
     },
     {
       "epoch": 31.23,
-      "grad_norm": 9.696378707885742,
+      "grad_norm": 12.42297649383545,
       "learning_rate": 8.95895895895896e-06,
-      "loss": 1.6458,
+      "loss": 1.6476,
       "step": 10400
     },
     {
       "epoch": 31.26,
-      "grad_norm": 6.563789367675781,
+      "grad_norm": 8.556532859802246,
       "learning_rate": 8.957957957957958e-06,
-      "loss": 1.5695,
+      "loss": 1.575,
       "step": 10410
     },
     {
       "epoch": 31.29,
-      "grad_norm": 6.416709899902344,
+      "grad_norm": 8.145130157470703,
       "learning_rate": 8.956956956956958e-06,
-      "loss": 1.6371,
+      "loss": 1.6369,
       "step": 10420
     },
     {
       "epoch": 31.32,
-      "grad_norm": 5.5639848709106445,
+      "grad_norm": 8.304274559020996,
       "learning_rate": 8.955955955955957e-06,
-      "loss": 1.6692,
+      "loss": 1.6815,
       "step": 10430
     },
     {
       "epoch": 31.35,
-      "grad_norm": 6.054525375366211,
+      "grad_norm": 7.151530742645264,
       "learning_rate": 8.954954954954956e-06,
-      "loss": 1.6038,
+      "loss": 1.6145,
       "step": 10440
     },
     {
       "epoch": 31.38,
-      "grad_norm": 11.165546417236328,
+      "grad_norm": 9.66533088684082,
       "learning_rate": 8.953953953953954e-06,
-      "loss": 1.6377,
+      "loss": 1.6319,
       "step": 10450
     },
     {
       "epoch": 31.41,
-      "grad_norm": 5.144339084625244,
+      "grad_norm": 5.849235534667969,
       "learning_rate": 8.952952952952953e-06,
-      "loss": 1.6533,
+      "loss": 1.655,
       "step": 10460
     },
     {
       "epoch": 31.44,
-      "grad_norm": 6.876260757446289,
+      "grad_norm": 8.536874771118164,
       "learning_rate": 8.951951951951953e-06,
-      "loss": 1.5849,
+      "loss": 1.5776,
       "step": 10470
     },
     {
       "epoch": 31.47,
-      "grad_norm": 6.776674270629883,
+      "grad_norm": 7.737988471984863,
       "learning_rate": 8.950950950950952e-06,
-      "loss": 1.6576,
+      "loss": 1.6565,
       "step": 10480
     },
     {
       "epoch": 31.5,
-      "grad_norm": 7.847053050994873,
+      "grad_norm": 7.071041584014893,
       "learning_rate": 8.94994994994995e-06,
-      "loss": 1.631,
+      "loss": 1.6358,
       "step": 10490
     },
     {
       "epoch": 31.53,
-      "grad_norm": 7.8445634841918945,
+      "grad_norm": 9.08774185180664,
       "learning_rate": 8.94894894894895e-06,
-      "loss": 1.6919,
+      "loss": 1.6938,
       "step": 10500
     },
     {
       "epoch": 31.56,
-      "grad_norm": 9.430473327636719,
+      "grad_norm": 8.335801124572754,
       "learning_rate": 8.94794794794795e-06,
-      "loss": 1.665,
+      "loss": 1.65,
       "step": 10510
     },
     {
       "epoch": 31.59,
-      "grad_norm": 6.176746845245361,
+      "grad_norm": 6.143366813659668,
       "learning_rate": 8.946946946946948e-06,
-      "loss": 1.663,
+      "loss": 1.66,
       "step": 10520
     },
     {
       "epoch": 31.62,
-      "grad_norm": 8.958805084228516,
+      "grad_norm": 9.193089485168457,
       "learning_rate": 8.945945945945946e-06,
-      "loss": 1.6881,
+      "loss": 1.7005,
       "step": 10530
     },
     {
       "epoch": 31.65,
-      "grad_norm": 7.420097827911377,
+      "grad_norm": 10.510265350341797,
       "learning_rate": 8.944944944944945e-06,
-      "loss": 1.6406,
+      "loss": 1.6428,
       "step": 10540
     },
     {
       "epoch": 31.68,
-      "grad_norm": 9.44011402130127,
+      "grad_norm": 8.62302017211914,
       "learning_rate": 8.943943943943945e-06,
-      "loss": 1.6906,
+      "loss": 1.694,
       "step": 10550
     },
     {
       "epoch": 31.71,
-      "grad_norm": 9.450826644897461,
+      "grad_norm": 10.498969078063965,
       "learning_rate": 8.942942942942944e-06,
-      "loss": 1.6792,
+      "loss": 1.6872,
       "step": 10560
     },
     {
       "epoch": 31.74,
-      "grad_norm": 9.725481033325195,
+      "grad_norm": 9.295823097229004,
       "learning_rate": 8.941941941941942e-06,
-      "loss": 1.7064,
+      "loss": 1.6996,
       "step": 10570
     },
     {
       "epoch": 31.77,
-      "grad_norm": 9.447689056396484,
+      "grad_norm": 8.694743156433105,
       "learning_rate": 8.940940940940943e-06,
-      "loss": 1.6777,
+      "loss": 1.6802,
       "step": 10580
     },
     {
       "epoch": 31.8,
-      "grad_norm": 7.596888542175293,
+      "grad_norm": 8.268648147583008,
       "learning_rate": 8.939939939939941e-06,
-      "loss": 1.6535,
+      "loss": 1.6599,
       "step": 10590
     },
     {
       "epoch": 31.83,
-      "grad_norm": 7.534690856933594,
+      "grad_norm": 7.006750106811523,
       "learning_rate": 8.93893893893894e-06,
-      "loss": 1.5528,
+      "loss": 1.5517,
       "step": 10600
     },
     {
       "epoch": 31.86,
-      "grad_norm": 12.547882080078125,
+      "grad_norm": 8.097638130187988,
       "learning_rate": 8.937937937937939e-06,
-      "loss": 1.6509,
+      "loss": 1.6641,
       "step": 10610
     },
     {
       "epoch": 31.89,
-      "grad_norm": 10.777752876281738,
+      "grad_norm": 13.01207447052002,
       "learning_rate": 8.936936936936937e-06,
-      "loss": 1.6385,
+      "loss": 1.6356,
       "step": 10620
     },
     {
       "epoch": 31.92,
-      "grad_norm": 8.354254722595215,
+      "grad_norm": 9.24506950378418,
       "learning_rate": 8.935935935935937e-06,
-      "loss": 1.6336,
+      "loss": 1.636,
       "step": 10630
     },
     {
       "epoch": 31.95,
-      "grad_norm": 7.507474899291992,
+      "grad_norm": 7.071739673614502,
       "learning_rate": 8.934934934934936e-06,
-      "loss": 1.6403,
+      "loss": 1.6569,
       "step": 10640
     },
     {
       "epoch": 31.98,
-      "grad_norm": 5.957705974578857,
+      "grad_norm": 8.844524383544922,
       "learning_rate": 8.933933933933935e-06,
-      "loss": 1.6704,
+      "loss": 1.6722,
       "step": 10650
     },
     {
       "epoch": 32.0,
-      "eval_accuracy": 0.7053,
-      "eval_loss": 1.0171515941619873,
-      "eval_runtime": 5.5448,
-      "eval_samples_per_second": 1803.498,
-      "eval_steps_per_second": 7.214,
+      "eval_accuracy": 0.7093,
+      "eval_loss": 1.0193946361541748,
+      "eval_runtime": 5.5009,
+      "eval_samples_per_second": 1817.886,
+      "eval_steps_per_second": 7.272,
       "step": 10656
     },
     {
       "epoch": 32.01,
-      "grad_norm": 7.531605243682861,
+      "grad_norm": 9.714725494384766,
       "learning_rate": 8.932932932932933e-06,
-      "loss": 1.6676,
+      "loss": 1.6972,
       "step": 10660
     },
     {
       "epoch": 32.04,
-      "grad_norm": 8.791948318481445,
+      "grad_norm": 7.271981239318848,
       "learning_rate": 8.931931931931933e-06,
-      "loss": 1.6672,
+      "loss": 1.6693,
       "step": 10670
     },
     {
       "epoch": 32.07,
-      "grad_norm": 5.420367240905762,
+      "grad_norm": 7.780402183532715,
       "learning_rate": 8.93093093093093e-06,
-      "loss": 1.6044,
+      "loss": 1.5947,
       "step": 10680
     },
     {
       "epoch": 32.1,
-      "grad_norm": 8.705533981323242,
+      "grad_norm": 9.600722312927246,
       "learning_rate": 8.92992992992993e-06,
-      "loss": 1.5863,
+      "loss": 1.5862,
       "step": 10690
     },
     {
       "epoch": 32.13,
-      "grad_norm": 7.922588348388672,
+      "grad_norm": 9.35757827758789,
       "learning_rate": 8.92892892892893e-06,
-      "loss": 1.6219,
+      "loss": 1.6194,
       "step": 10700
     },
     {
       "epoch": 32.16,
-      "grad_norm": 6.989918231964111,
+      "grad_norm": 7.209897041320801,
       "learning_rate": 8.927927927927928e-06,
-      "loss": 1.696,
+      "loss": 1.6992,
       "step": 10710
     },
     {
       "epoch": 32.19,
-      "grad_norm": 10.359856605529785,
+      "grad_norm": 8.48534107208252,
       "learning_rate": 8.926926926926928e-06,
-      "loss": 1.644,
+      "loss": 1.6492,
       "step": 10720
     },
     {
       "epoch": 32.22,
-      "grad_norm": 8.0414457321167,
+      "grad_norm": 7.462686538696289,
       "learning_rate": 8.925925925925927e-06,
-      "loss": 1.6392,
+      "loss": 1.6413,
       "step": 10730
     },
     {
       "epoch": 32.25,
-      "grad_norm": 7.70808744430542,
+      "grad_norm": 7.828274250030518,
       "learning_rate": 8.924924924924925e-06,
-      "loss": 1.6815,
+      "loss": 1.6826,
       "step": 10740
     },
     {
       "epoch": 32.28,
-      "grad_norm": 7.3980231285095215,
+      "grad_norm": 8.09559440612793,
       "learning_rate": 8.923923923923926e-06,
-      "loss": 1.6125,
+      "loss": 1.612,
       "step": 10750
     },
     {
       "epoch": 32.31,
-      "grad_norm": 7.436577796936035,
+      "grad_norm": 7.762634754180908,
       "learning_rate": 8.922922922922924e-06,
-      "loss": 1.6273,
+      "loss": 1.6178,
       "step": 10760
     },
     {
       "epoch": 32.34,
-      "grad_norm": 5.661825656890869,
+      "grad_norm": 9.185091018676758,
       "learning_rate": 8.921921921921923e-06,
-      "loss": 1.5759,
+      "loss": 1.5813,
       "step": 10770
     },
     {
       "epoch": 32.37,
-      "grad_norm": 6.2902350425720215,
+      "grad_norm": 8.750696182250977,
       "learning_rate": 8.920920920920921e-06,
-      "loss": 1.6528,
+      "loss": 1.6653,
       "step": 10780
     },
     {
       "epoch": 32.4,
-      "grad_norm": 6.643796920776367,
+      "grad_norm": 12.343206405639648,
       "learning_rate": 8.91991991991992e-06,
-      "loss": 1.6404,
+      "loss": 1.6312,
       "step": 10790
     },
     {
       "epoch": 32.43,
-      "grad_norm": 9.172786712646484,
+      "grad_norm": 8.389094352722168,
       "learning_rate": 8.91891891891892e-06,
-      "loss": 1.653,
+      "loss": 1.6595,
       "step": 10800
     },
     {
       "epoch": 32.46,
-      "grad_norm": 5.795620441436768,
+      "grad_norm": 8.702826499938965,
       "learning_rate": 8.917917917917919e-06,
-      "loss": 1.6488,
+      "loss": 1.6484,
       "step": 10810
     },
     {
       "epoch": 32.49,
-      "grad_norm": 5.844393253326416,
+      "grad_norm": 7.824363708496094,
       "learning_rate": 8.916916916916917e-06,
-      "loss": 1.6127,
+      "loss": 1.6005,
       "step": 10820
     },
     {
       "epoch": 32.52,
-      "grad_norm": 7.565409183502197,
+      "grad_norm": 7.759343147277832,
       "learning_rate": 8.915915915915918e-06,
-      "loss": 1.5836,
+      "loss": 1.5863,
       "step": 10830
     },
     {
       "epoch": 32.55,
-      "grad_norm": 5.363765716552734,
+      "grad_norm": 7.809866905212402,
       "learning_rate": 8.914914914914916e-06,
-      "loss": 1.6708,
+      "loss": 1.6849,
       "step": 10840
     },
     {
       "epoch": 32.58,
-      "grad_norm": 5.0569071769714355,
+      "grad_norm": 7.715855121612549,
       "learning_rate": 8.913913913913915e-06,
-      "loss": 1.6102,
+      "loss": 1.6091,
       "step": 10850
     },
     {
       "epoch": 32.61,
-      "grad_norm": 6.834020137786865,
+      "grad_norm": 9.326638221740723,
       "learning_rate": 8.912912912912914e-06,
-      "loss": 1.6256,
+      "loss": 1.6361,
       "step": 10860
     },
     {
       "epoch": 32.64,
-      "grad_norm": 9.055957794189453,
+      "grad_norm": 9.9804048538208,
       "learning_rate": 8.911911911911912e-06,
-      "loss": 1.6216,
+      "loss": 1.6364,
       "step": 10870
     },
     {
       "epoch": 32.67,
-      "grad_norm": 7.474025726318359,
+      "grad_norm": 8.769659996032715,
       "learning_rate": 8.91091091091091e-06,
-      "loss": 1.6081,
+      "loss": 1.619,
       "step": 10880
     },
     {
       "epoch": 32.7,
-      "grad_norm": 8.588315963745117,
+      "grad_norm": 6.659862041473389,
       "learning_rate": 8.909909909909911e-06,
-      "loss": 1.5857,
+      "loss": 1.5977,
       "step": 10890
     },
     {
       "epoch": 32.73,
-      "grad_norm": 7.319324016571045,
+      "grad_norm": 6.604353904724121,
       "learning_rate": 8.90890890890891e-06,
-      "loss": 1.5416,
+      "loss": 1.5523,
       "step": 10900
     },
     {
       "epoch": 32.76,
-      "grad_norm": 6.884843349456787,
+      "grad_norm": 6.1109137535095215,
       "learning_rate": 8.907907907907908e-06,
-      "loss": 1.6229,
+      "loss": 1.628,
       "step": 10910
     },
     {
       "epoch": 32.79,
-      "grad_norm": 6.055380821228027,
+      "grad_norm": 7.75645112991333,
       "learning_rate": 8.906906906906909e-06,
-      "loss": 1.6592,
+      "loss": 1.6551,
       "step": 10920
     },
     {
       "epoch": 32.82,
-      "grad_norm": 7.92530632019043,
+      "grad_norm": 7.774423599243164,
       "learning_rate": 8.905905905905905e-06,
-      "loss": 1.6306,
+      "loss": 1.6402,
       "step": 10930
     },
     {
       "epoch": 32.85,
-      "grad_norm": 8.743660926818848,
+      "grad_norm": 11.936053276062012,
       "learning_rate": 8.904904904904906e-06,
-      "loss": 1.6201,
+      "loss": 1.6342,
       "step": 10940
     },
     {
       "epoch": 32.88,
-      "grad_norm": 8.547505378723145,
+      "grad_norm": 6.964539051055908,
       "learning_rate": 8.903903903903904e-06,
-      "loss": 1.5923,
+      "loss": 1.5914,
       "step": 10950
     },
     {
       "epoch": 32.91,
-      "grad_norm": 6.565919876098633,
+      "grad_norm": 5.757761001586914,
       "learning_rate": 8.902902902902903e-06,
-      "loss": 1.6097,
+      "loss": 1.6169,
       "step": 10960
     },
     {
       "epoch": 32.94,
-      "grad_norm": 7.652177333831787,
+      "grad_norm": 7.542760848999023,
       "learning_rate": 8.901901901901903e-06,
-      "loss": 1.6117,
+      "loss": 1.6112,
       "step": 10970
     },
     {
       "epoch": 32.97,
-      "grad_norm": 7.517940521240234,
+      "grad_norm": 8.224106788635254,
       "learning_rate": 8.900900900900902e-06,
-      "loss": 1.6393,
+      "loss": 1.638,
       "step": 10980
     },
     {
       "epoch": 33.0,
-      "eval_accuracy": 0.7106,
-      "eval_loss": 1.0008105039596558,
-      "eval_runtime": 5.377,
-      "eval_samples_per_second": 1859.779,
-      "eval_steps_per_second": 7.439,
+      "eval_accuracy": 0.712,
+      "eval_loss": 1.0023113489151,
+      "eval_runtime": 5.0978,
+      "eval_samples_per_second": 1961.639,
+      "eval_steps_per_second": 7.847,
       "step": 10989
     },
     {
       "epoch": 33.0,
-      "grad_norm": 6.397665977478027,
+      "grad_norm": 8.94605827331543,
       "learning_rate": 8.8998998998999e-06,
-      "loss": 1.6853,
+      "loss": 1.7217,
       "step": 10990
     },
     {
       "epoch": 33.03,
-      "grad_norm": 7.139054775238037,
+      "grad_norm": 7.502806663513184,
       "learning_rate": 8.8988988988989e-06,
-      "loss": 1.652,
+      "loss": 1.6532,
       "step": 11000
     },
     {
       "epoch": 33.06,
-      "grad_norm": 8.557888984680176,
+      "grad_norm": 9.421451568603516,
       "learning_rate": 8.8978978978979e-06,
-      "loss": 1.6289,
+      "loss": 1.6303,
       "step": 11010
     },
     {
       "epoch": 33.09,
-      "grad_norm": 9.052143096923828,
+      "grad_norm": 7.9836907386779785,
       "learning_rate": 8.896896896896898e-06,
-      "loss": 1.5905,
+      "loss": 1.6054,
       "step": 11020
     },
     {
       "epoch": 33.12,
-      "grad_norm": 8.642655372619629,
+      "grad_norm": 6.854736328125,
       "learning_rate": 8.895895895895896e-06,
-      "loss": 1.589,
+      "loss": 1.5904,
       "step": 11030
     },
     {
       "epoch": 33.15,
-      "grad_norm": 6.357573986053467,
+      "grad_norm": 9.445804595947266,
       "learning_rate": 8.894894894894895e-06,
-      "loss": 1.6267,
+      "loss": 1.6384,
       "step": 11040
     },
     {
       "epoch": 33.18,
-      "grad_norm": 7.736217975616455,
+      "grad_norm": 7.342143535614014,
       "learning_rate": 8.893893893893895e-06,
-      "loss": 1.5752,
+      "loss": 1.5708,
       "step": 11050
     },
     {
       "epoch": 33.21,
-      "grad_norm": 5.306316375732422,
+      "grad_norm": 6.883600234985352,
       "learning_rate": 8.892892892892894e-06,
-      "loss": 1.6232,
+      "loss": 1.6319,
       "step": 11060
     },
     {
       "epoch": 33.24,
-      "grad_norm": 8.027252197265625,
+      "grad_norm": 7.248322010040283,
       "learning_rate": 8.891891891891893e-06,
-      "loss": 1.6199,
+      "loss": 1.6321,
       "step": 11070
     },
     {
       "epoch": 33.27,
-      "grad_norm": 5.678922176361084,
+      "grad_norm": 6.786847114562988,
       "learning_rate": 8.890890890890893e-06,
-      "loss": 1.5804,
+      "loss": 1.5868,
       "step": 11080
     },
     {
       "epoch": 33.3,
-      "grad_norm": 12.014375686645508,
+      "grad_norm": 9.814334869384766,
       "learning_rate": 8.889889889889891e-06,
-      "loss": 1.6094,
+      "loss": 1.6083,
       "step": 11090
     },
     {
       "epoch": 33.33,
-      "grad_norm": 6.611203670501709,
+      "grad_norm": 6.032260417938232,
       "learning_rate": 8.888888888888888e-06,
-      "loss": 1.6031,
+      "loss": 1.6053,
       "step": 11100
     },
     {
       "epoch": 33.36,
-      "grad_norm": 7.033792495727539,
+      "grad_norm": 7.195312976837158,
       "learning_rate": 8.887887887887889e-06,
-      "loss": 1.6092,
+      "loss": 1.598,
       "step": 11110
     },
     {
       "epoch": 33.39,
-      "grad_norm": 7.033509731292725,
+      "grad_norm": 12.394963264465332,
       "learning_rate": 8.886886886886887e-06,
-      "loss": 1.5427,
+      "loss": 1.5422,
       "step": 11120
     },
     {
       "epoch": 33.42,
-      "grad_norm": 7.238529205322266,
+      "grad_norm": 7.17933988571167,
       "learning_rate": 8.885885885885886e-06,
-      "loss": 1.5885,
+      "loss": 1.6096,
       "step": 11130
     },
     {
       "epoch": 33.45,
-      "grad_norm": 6.686769485473633,
+      "grad_norm": 6.684762477874756,
       "learning_rate": 8.884884884884886e-06,
-      "loss": 1.5836,
+      "loss": 1.5939,
       "step": 11140
     },
     {
       "epoch": 33.48,
-      "grad_norm": 5.47464656829834,
+      "grad_norm": 9.544720649719238,
       "learning_rate": 8.883883883883885e-06,
-      "loss": 1.5401,
+      "loss": 1.546,
       "step": 11150
     },
     {
       "epoch": 33.51,
-      "grad_norm": 6.995031356811523,
+      "grad_norm": 7.762179851531982,
       "learning_rate": 8.882882882882883e-06,
-      "loss": 1.5958,
+      "loss": 1.592,
       "step": 11160
     },
     {
       "epoch": 33.54,
-      "grad_norm": 12.430277824401855,
+      "grad_norm": 9.200556755065918,
       "learning_rate": 8.881881881881884e-06,
-      "loss": 1.5816,
+      "loss": 1.5799,
       "step": 11170
     },
     {
       "epoch": 33.57,
-      "grad_norm": 8.481268882751465,
+      "grad_norm": 6.990026473999023,
       "learning_rate": 8.88088088088088e-06,
-      "loss": 1.6511,
+      "loss": 1.6474,
       "step": 11180
     },
     {
       "epoch": 33.6,
-      "grad_norm": 7.3181681632995605,
+      "grad_norm": 8.766324996948242,
       "learning_rate": 8.87987987987988e-06,
-      "loss": 1.5859,
+      "loss": 1.5873,
       "step": 11190
     },
     {
       "epoch": 33.63,
-      "grad_norm": 8.73676872253418,
+      "grad_norm": 5.292863368988037,
       "learning_rate": 8.87887887887888e-06,
-      "loss": 1.5576,
+      "loss": 1.5424,
       "step": 11200
     },
     {
       "epoch": 33.66,
-      "grad_norm": 5.83658504486084,
+      "grad_norm": 9.068073272705078,
       "learning_rate": 8.877877877877878e-06,
-      "loss": 1.6348,
+      "loss": 1.6331,
       "step": 11210
     },
     {
       "epoch": 33.69,
-      "grad_norm": 6.7718048095703125,
+      "grad_norm": 9.739002227783203,
       "learning_rate": 8.876876876876878e-06,
-      "loss": 1.6771,
+      "loss": 1.6797,
       "step": 11220
     },
     {
       "epoch": 33.72,
-      "grad_norm": 6.072212219238281,
+      "grad_norm": 6.184283256530762,
       "learning_rate": 8.875875875875877e-06,
-      "loss": 1.5873,
+      "loss": 1.5744,
       "step": 11230
     },
     {
       "epoch": 33.75,
-      "grad_norm": 7.569611072540283,
+      "grad_norm": 8.431503295898438,
       "learning_rate": 8.874874874874875e-06,
-      "loss": 1.6335,
+      "loss": 1.6357,
       "step": 11240
     },
     {
       "epoch": 33.78,
-      "grad_norm": 6.459864616394043,
+      "grad_norm": 8.192719459533691,
       "learning_rate": 8.873873873873876e-06,
-      "loss": 1.6882,
+      "loss": 1.6829,
       "step": 11250
     },
     {
       "epoch": 33.81,
-      "grad_norm": 9.702081680297852,
+      "grad_norm": 8.580551147460938,
       "learning_rate": 8.872872872872874e-06,
-      "loss": 1.6011,
+      "loss": 1.6109,
       "step": 11260
     },
     {
       "epoch": 33.84,
-      "grad_norm": 6.555346488952637,
+      "grad_norm": 7.063335418701172,
       "learning_rate": 8.871871871871873e-06,
-      "loss": 1.6442,
+      "loss": 1.6487,
       "step": 11270
     },
     {
       "epoch": 33.87,
-      "grad_norm": 5.975614547729492,
+      "grad_norm": 7.924859046936035,
       "learning_rate": 8.870870870870871e-06,
-      "loss": 1.5803,
+      "loss": 1.581,
       "step": 11280
     },
     {
       "epoch": 33.9,
-      "grad_norm": 9.042289733886719,
+      "grad_norm": 7.5534491539001465,
       "learning_rate": 8.86986986986987e-06,
-      "loss": 1.6145,
+      "loss": 1.6249,
       "step": 11290
     },
     {
       "epoch": 33.93,
-      "grad_norm": 9.182083129882812,
+      "grad_norm": 8.653654098510742,
       "learning_rate": 8.86886886886887e-06,
-      "loss": 1.6095,
+      "loss": 1.6078,
       "step": 11300
     },
     {
       "epoch": 33.96,
-      "grad_norm": 8.925015449523926,
+      "grad_norm": 12.056917190551758,
       "learning_rate": 8.867867867867869e-06,
-      "loss": 1.6969,
+      "loss": 1.7082,
       "step": 11310
     },
     {
       "epoch": 33.99,
-      "grad_norm": 8.728311538696289,
+      "grad_norm": 8.392271041870117,
       "learning_rate": 8.866866866866868e-06,
-      "loss": 1.5795,
+      "loss": 1.5901,
       "step": 11320
     },
     {
       "epoch": 34.0,
-      "eval_accuracy": 0.7126,
-      "eval_loss": 0.9908768534660339,
-      "eval_runtime": 5.4459,
-      "eval_samples_per_second": 1836.234,
-      "eval_steps_per_second": 7.345,
+      "eval_accuracy": 0.7136,
+      "eval_loss": 0.9909215569496155,
+      "eval_runtime": 5.5898,
+      "eval_samples_per_second": 1788.988,
+      "eval_steps_per_second": 7.156,
       "step": 11322
     },
     {
       "epoch": 34.02,
-      "grad_norm": 9.265826225280762,
+      "grad_norm": 8.403477668762207,
       "learning_rate": 8.865865865865866e-06,
-      "loss": 1.8517,
+      "loss": 1.8516,
       "step": 11330
     },
     {
       "epoch": 34.05,
-      "grad_norm": 7.498945713043213,
+      "grad_norm": 9.03896713256836,
       "learning_rate": 8.864864864864866e-06,
-      "loss": 1.638,
+      "loss": 1.6507,
       "step": 11340
     },
     {
       "epoch": 34.08,
-      "grad_norm": 9.427443504333496,
+      "grad_norm": 9.147820472717285,
       "learning_rate": 8.863863863863863e-06,
-      "loss": 1.6652,
+      "loss": 1.6517,
       "step": 11350
     },
     {
       "epoch": 34.11,
-      "grad_norm": 5.849697589874268,
+      "grad_norm": 9.787577629089355,
       "learning_rate": 8.862862862862864e-06,
-      "loss": 1.5582,
+      "loss": 1.5483,
       "step": 11360
     },
     {
       "epoch": 34.14,
-      "grad_norm": 7.90972900390625,
+      "grad_norm": 8.489692687988281,
       "learning_rate": 8.861861861861862e-06,
-      "loss": 1.5992,
+      "loss": 1.605,
       "step": 11370
     },
     {
       "epoch": 34.17,
-      "grad_norm": 5.788965702056885,
+      "grad_norm": 7.621850490570068,
       "learning_rate": 8.86086086086086e-06,
-      "loss": 1.5702,
+      "loss": 1.575,
       "step": 11380
     },
     {
       "epoch": 34.2,
-      "grad_norm": 6.500415325164795,
+      "grad_norm": 7.704878330230713,
       "learning_rate": 8.859859859859861e-06,
-      "loss": 1.6762,
+      "loss": 1.6864,
       "step": 11390
     },
     {
       "epoch": 34.23,
-      "grad_norm": 5.903830051422119,
+      "grad_norm": 6.992391586303711,
       "learning_rate": 8.85885885885886e-06,
-      "loss": 1.6184,
+      "loss": 1.6179,
       "step": 11400
     },
     {
       "epoch": 34.26,
-      "grad_norm": 6.855862617492676,
+      "grad_norm": 8.922061920166016,
       "learning_rate": 8.857857857857858e-06,
-      "loss": 1.5557,
+      "loss": 1.5607,
       "step": 11410
     },
     {
       "epoch": 34.29,
-      "grad_norm": 6.009995937347412,
+      "grad_norm": 6.738629341125488,
       "learning_rate": 8.856856856856859e-06,
-      "loss": 1.5695,
+      "loss": 1.5775,
       "step": 11420
     },
     {
       "epoch": 34.32,
-      "grad_norm": 11.386078834533691,
+      "grad_norm": 10.03793716430664,
       "learning_rate": 8.855855855855855e-06,
-      "loss": 1.6485,
+      "loss": 1.6548,
       "step": 11430
     },
     {
       "epoch": 34.35,
-      "grad_norm": 7.200382232666016,
+      "grad_norm": 6.852489471435547,
       "learning_rate": 8.854854854854856e-06,
-      "loss": 1.5083,
+      "loss": 1.5123,
       "step": 11440
     },
     {
       "epoch": 34.38,
-      "grad_norm": 8.956271171569824,
+      "grad_norm": 8.896140098571777,
       "learning_rate": 8.853853853853854e-06,
-      "loss": 1.5709,
+      "loss": 1.5669,
       "step": 11450
     },
     {
       "epoch": 34.41,
-      "grad_norm": 8.02228832244873,
+      "grad_norm": 8.041394233703613,
       "learning_rate": 8.852852852852853e-06,
-      "loss": 1.586,
+      "loss": 1.5782,
       "step": 11460
     },
     {
       "epoch": 34.44,
-      "grad_norm": 7.533289909362793,
+      "grad_norm": 9.039679527282715,
       "learning_rate": 8.851851851851853e-06,
-      "loss": 1.6586,
+      "loss": 1.6601,
       "step": 11470
     },
     {
       "epoch": 34.47,
-      "grad_norm": 11.2882080078125,
+      "grad_norm": 8.510540008544922,
       "learning_rate": 8.850850850850852e-06,
-      "loss": 1.6282,
+      "loss": 1.6393,
       "step": 11480
     },
     {
       "epoch": 34.5,
-      "grad_norm": 6.951305866241455,
+      "grad_norm": 7.647674560546875,
       "learning_rate": 8.84984984984985e-06,
-      "loss": 1.5431,
+      "loss": 1.5355,
       "step": 11490
     },
     {
       "epoch": 34.53,
-      "grad_norm": 9.120506286621094,
+      "grad_norm": 10.130387306213379,
       "learning_rate": 8.84884884884885e-06,
-      "loss": 1.5399,
+      "loss": 1.5448,
       "step": 11500
     },
     {
       "epoch": 34.56,
-      "grad_norm": 6.686323165893555,
+      "grad_norm": 5.950345039367676,
       "learning_rate": 8.84784784784785e-06,
-      "loss": 1.5668,
+      "loss": 1.5775,
       "step": 11510
     },
     {
       "epoch": 34.59,
-      "grad_norm": 8.293801307678223,
+      "grad_norm": 7.715575218200684,
       "learning_rate": 8.846846846846848e-06,
-      "loss": 1.6584,
+      "loss": 1.6502,
       "step": 11520
     },
     {
       "epoch": 34.62,
-      "grad_norm": 7.557324409484863,
+      "grad_norm": 7.2237772941589355,
       "learning_rate": 8.845845845845847e-06,
-      "loss": 1.6215,
+      "loss": 1.6376,
       "step": 11530
     },
     {
       "epoch": 34.65,
-      "grad_norm": 6.103909492492676,
+      "grad_norm": 8.132916450500488,
       "learning_rate": 8.844844844844845e-06,
-      "loss": 1.5276,
+      "loss": 1.5396,
       "step": 11540
     },
     {
       "epoch": 34.68,
-      "grad_norm": 6.289796829223633,
+      "grad_norm": 7.179383277893066,
       "learning_rate": 8.843843843843844e-06,
-      "loss": 1.585,
+      "loss": 1.5855,
       "step": 11550
     },
     {
       "epoch": 34.71,
-      "grad_norm": 8.505709648132324,
+      "grad_norm": 7.401016712188721,
       "learning_rate": 8.842842842842844e-06,
-      "loss": 1.628,
+      "loss": 1.619,
       "step": 11560
     },
     {
       "epoch": 34.74,
-      "grad_norm": 5.722997665405273,
+      "grad_norm": 6.222245693206787,
       "learning_rate": 8.841841841841843e-06,
-      "loss": 1.6228,
+      "loss": 1.6288,
       "step": 11570
     },
     {
       "epoch": 34.77,
-      "grad_norm": 7.586233615875244,
+      "grad_norm": 9.477871894836426,
       "learning_rate": 8.840840840840841e-06,
-      "loss": 1.649,
+      "loss": 1.6551,
       "step": 11580
     },
     {
       "epoch": 34.8,
-      "grad_norm": 8.902361869812012,
+      "grad_norm": 12.444064140319824,
       "learning_rate": 8.839839839839841e-06,
-      "loss": 1.6001,
+      "loss": 1.6151,
       "step": 11590
     },
     {
       "epoch": 34.83,
-      "grad_norm": 7.667283535003662,
+      "grad_norm": 4.902679920196533,
       "learning_rate": 8.838838838838838e-06,
-      "loss": 1.5912,
+      "loss": 1.5905,
       "step": 11600
     },
     {
       "epoch": 34.86,
-      "grad_norm": 8.176789283752441,
+      "grad_norm": 8.418246269226074,
       "learning_rate": 8.837837837837839e-06,
-      "loss": 1.597,
+      "loss": 1.6005,
       "step": 11610
     },
     {
       "epoch": 34.89,
-      "grad_norm": 7.643188953399658,
+      "grad_norm": 7.325955867767334,
       "learning_rate": 8.836836836836837e-06,
-      "loss": 1.5542,
+      "loss": 1.552,
       "step": 11620
     },
     {
       "epoch": 34.92,
-      "grad_norm": 7.921688556671143,
+      "grad_norm": 7.696230888366699,
       "learning_rate": 8.835835835835836e-06,
-      "loss": 1.5595,
+      "loss": 1.5717,
       "step": 11630
     },
     {
       "epoch": 34.95,
-      "grad_norm": 10.15783977508545,
+      "grad_norm": 11.0137300491333,
       "learning_rate": 8.834834834834836e-06,
-      "loss": 1.6646,
+      "loss": 1.6506,
       "step": 11640
     },
     {
       "epoch": 34.98,
-      "grad_norm": 11.063594818115234,
+      "grad_norm": 13.261284828186035,
       "learning_rate": 8.833833833833835e-06,
-      "loss": 1.6104,
+      "loss": 1.6198,
       "step": 11650
     },
     {
       "epoch": 35.0,
-      "eval_accuracy": 0.7199,
-      "eval_loss": 0.9561426043510437,
-      "eval_runtime": 5.6614,
-      "eval_samples_per_second": 1766.346,
-      "eval_steps_per_second": 7.065,
+      "eval_accuracy": 0.7206,
+      "eval_loss": 0.9546998739242554,
+      "eval_runtime": 5.2267,
+      "eval_samples_per_second": 1913.248,
+      "eval_steps_per_second": 7.653,
       "step": 11655
     },
     {
       "epoch": 35.02,
-      "grad_norm": 10.397171974182129,
+      "grad_norm": 9.964131355285645,
       "learning_rate": 8.832832832832833e-06,
-      "loss": 1.6828,
+      "loss": 1.6843,
       "step": 11660
     },
     {
       "epoch": 35.05,
-      "grad_norm": 9.858508110046387,
+      "grad_norm": 8.0030517578125,
       "learning_rate": 8.831831831831834e-06,
-      "loss": 1.5635,
+      "loss": 1.5587,
       "step": 11670
     },
     {
       "epoch": 35.08,
-      "grad_norm": 7.776301383972168,
+      "grad_norm": 7.056451797485352,
       "learning_rate": 8.83083083083083e-06,
-      "loss": 1.584,
+      "loss": 1.5907,
       "step": 11680
     },
     {
       "epoch": 35.11,
-      "grad_norm": 7.535364151000977,
+      "grad_norm": 7.766964435577393,
       "learning_rate": 8.82982982982983e-06,
-      "loss": 1.6022,
+      "loss": 1.6092,
       "step": 11690
     },
     {
       "epoch": 35.14,
-      "grad_norm": 6.790921211242676,
+      "grad_norm": 9.555071830749512,
       "learning_rate": 8.82882882882883e-06,
-      "loss": 1.5529,
+      "loss": 1.5596,
       "step": 11700
     },
     {
       "epoch": 35.17,
-      "grad_norm": 5.713064670562744,
+      "grad_norm": 8.22910213470459,
       "learning_rate": 8.827827827827828e-06,
-      "loss": 1.6039,
+      "loss": 1.6145,
       "step": 11710
     },
     {
       "epoch": 35.2,
-      "grad_norm": 7.185563564300537,
+      "grad_norm": 8.424379348754883,
       "learning_rate": 8.826826826826828e-06,
-      "loss": 1.5828,
+      "loss": 1.5857,
       "step": 11720
     },
     {
       "epoch": 35.23,
-      "grad_norm": 6.924100399017334,
+      "grad_norm": 7.350320816040039,
       "learning_rate": 8.825825825825827e-06,
-      "loss": 1.6112,
+      "loss": 1.6154,
       "step": 11730
     },
     {
       "epoch": 35.26,
-      "grad_norm": 8.085559844970703,
+      "grad_norm": 8.182185173034668,
       "learning_rate": 8.824824824824825e-06,
-      "loss": 1.5971,
+      "loss": 1.593,
       "step": 11740
     },
     {
       "epoch": 35.29,
-      "grad_norm": 6.477778434753418,
+      "grad_norm": 8.834285736083984,
       "learning_rate": 8.823823823823826e-06,
-      "loss": 1.5676,
+      "loss": 1.5711,
       "step": 11750
     },
     {
       "epoch": 35.32,
-      "grad_norm": 6.414569854736328,
+      "grad_norm": 6.129551887512207,
       "learning_rate": 8.822822822822824e-06,
-      "loss": 1.6424,
+      "loss": 1.6207,
       "step": 11760
     },
     {
       "epoch": 35.35,
-      "grad_norm": 8.24698543548584,
+      "grad_norm": 6.952534198760986,
       "learning_rate": 8.821821821821823e-06,
-      "loss": 1.6087,
+      "loss": 1.6076,
       "step": 11770
     },
     {
       "epoch": 35.38,
-      "grad_norm": 7.337802410125732,
+      "grad_norm": 6.130636692047119,
       "learning_rate": 8.820820820820822e-06,
-      "loss": 1.5751,
+      "loss": 1.579,
       "step": 11780
     },
     {
       "epoch": 35.41,
-      "grad_norm": 7.289150714874268,
+      "grad_norm": 14.919532775878906,
       "learning_rate": 8.81981981981982e-06,
-      "loss": 1.674,
+      "loss": 1.6906,
       "step": 11790
     },
     {
       "epoch": 35.44,
-      "grad_norm": 7.4793219566345215,
+      "grad_norm": 7.283972263336182,
       "learning_rate": 8.818818818818819e-06,
-      "loss": 1.6224,
+      "loss": 1.6232,
       "step": 11800
     },
     {
       "epoch": 35.47,
-      "grad_norm": 7.065758228302002,
+      "grad_norm": 7.748701095581055,
       "learning_rate": 8.817817817817819e-06,
-      "loss": 1.5758,
+      "loss": 1.5854,
       "step": 11810
     },
     {
       "epoch": 35.5,
-      "grad_norm": 7.38935661315918,
+      "grad_norm": 7.8101725578308105,
       "learning_rate": 8.816816816816818e-06,
-      "loss": 1.5778,
+      "loss": 1.5905,
       "step": 11820
     },
     {
       "epoch": 35.53,
-      "grad_norm": 9.775217056274414,
+      "grad_norm": 9.230366706848145,
       "learning_rate": 8.815815815815816e-06,
-      "loss": 1.5696,
+      "loss": 1.5703,
       "step": 11830
     },
     {
       "epoch": 35.56,
-      "grad_norm": 7.188069820404053,
+      "grad_norm": 8.415305137634277,
       "learning_rate": 8.814814814814817e-06,
-      "loss": 1.5307,
+      "loss": 1.532,
       "step": 11840
     },
     {
       "epoch": 35.59,
-      "grad_norm": 9.909000396728516,
+      "grad_norm": 8.381547927856445,
       "learning_rate": 8.813813813813813e-06,
-      "loss": 1.6444,
+      "loss": 1.639,
       "step": 11850
     },
     {
       "epoch": 35.62,
-      "grad_norm": 6.886179447174072,
+      "grad_norm": 9.476743698120117,
       "learning_rate": 8.812812812812814e-06,
-      "loss": 1.5473,
+      "loss": 1.543,
       "step": 11860
     },
     {
       "epoch": 35.65,
-      "grad_norm": 8.327747344970703,
+      "grad_norm": 7.972357749938965,
       "learning_rate": 8.811811811811812e-06,
-      "loss": 1.6558,
+      "loss": 1.6501,
       "step": 11870
     },
     {
       "epoch": 35.68,
-      "grad_norm": 6.743029594421387,
+      "grad_norm": 8.147628784179688,
       "learning_rate": 8.810810810810811e-06,
-      "loss": 1.5474,
+      "loss": 1.5517,
       "step": 11880
     },
     {
       "epoch": 35.71,
-      "grad_norm": 6.156401634216309,
+      "grad_norm": 7.758137226104736,
       "learning_rate": 8.809809809809811e-06,
-      "loss": 1.5692,
+      "loss": 1.5686,
       "step": 11890
     },
     {
       "epoch": 35.74,
-      "grad_norm": 6.105600833892822,
+      "grad_norm": 6.6250505447387695,
       "learning_rate": 8.80880880880881e-06,
-      "loss": 1.5348,
+      "loss": 1.5399,
       "step": 11900
     },
     {
       "epoch": 35.77,
-      "grad_norm": 6.247955799102783,
+      "grad_norm": 5.228471755981445,
       "learning_rate": 8.807807807807808e-06,
-      "loss": 1.5154,
+      "loss": 1.519,
       "step": 11910
     },
     {
       "epoch": 35.8,
-      "grad_norm": 9.874090194702148,
+      "grad_norm": 6.475439548492432,
       "learning_rate": 8.806806806806809e-06,
-      "loss": 1.5544,
+      "loss": 1.5608,
       "step": 11920
     },
     {
       "epoch": 35.83,
-      "grad_norm": 7.471748352050781,
+      "grad_norm": 6.7551398277282715,
       "learning_rate": 8.805805805805806e-06,
-      "loss": 1.5383,
+      "loss": 1.5688,
       "step": 11930
     },
     {
       "epoch": 35.86,
-      "grad_norm": 6.890314102172852,
+      "grad_norm": 8.8375244140625,
       "learning_rate": 8.804804804804806e-06,
-      "loss": 1.579,
+      "loss": 1.5874,
       "step": 11940
     },
     {
       "epoch": 35.89,
-      "grad_norm": 9.812318801879883,
+      "grad_norm": 8.411971092224121,
       "learning_rate": 8.803803803803804e-06,
-      "loss": 1.6073,
+      "loss": 1.606,
       "step": 11950
     },
     {
       "epoch": 35.92,
-      "grad_norm": 7.185366153717041,
+      "grad_norm": 10.43879222869873,
       "learning_rate": 8.802802802802803e-06,
-      "loss": 1.646,
+      "loss": 1.6555,
       "step": 11960
     },
     {
       "epoch": 35.95,
-      "grad_norm": 9.124354362487793,
+      "grad_norm": 7.832991600036621,
       "learning_rate": 8.801801801801803e-06,
-      "loss": 1.5148,
+      "loss": 1.5291,
       "step": 11970
     },
     {
       "epoch": 35.98,
-      "grad_norm": 6.088477611541748,
+      "grad_norm": 6.348001480102539,
       "learning_rate": 8.800800800800802e-06,
-      "loss": 1.587,
+      "loss": 1.5947,
       "step": 11980
     },
     {
       "epoch": 36.0,
-      "eval_accuracy": 0.7168,
-      "eval_loss": 0.9592888951301575,
-      "eval_runtime": 5.5701,
-      "eval_samples_per_second": 1795.298,
-      "eval_steps_per_second": 7.181,
+      "eval_accuracy": 0.7202,
+      "eval_loss": 0.9587336182594299,
+      "eval_runtime": 5.3234,
+      "eval_samples_per_second": 1878.514,
+      "eval_steps_per_second": 7.514,
       "step": 11988
     },
     {
       "epoch": 36.01,
-      "grad_norm": 9.919438362121582,
+      "grad_norm": 10.075098037719727,
       "learning_rate": 8.7997997997998e-06,
-      "loss": 1.7181,
+      "loss": 1.7059,
       "step": 11990
     },
     {
       "epoch": 36.04,
-      "grad_norm": 7.15561580657959,
+      "grad_norm": 7.402651786804199,
       "learning_rate": 8.798798798798799e-06,
-      "loss": 1.5447,
+      "loss": 1.5429,
       "step": 12000
     },
     {
       "epoch": 36.07,
-      "grad_norm": 8.456547737121582,
+      "grad_norm": 6.8465118408203125,
       "learning_rate": 8.797797797797798e-06,
-      "loss": 1.6591,
+      "loss": 1.6633,
       "step": 12010
     },
     {
       "epoch": 36.1,
-      "grad_norm": 8.252923011779785,
+      "grad_norm": 8.472245216369629,
       "learning_rate": 8.796796796796796e-06,
-      "loss": 1.5929,
+      "loss": 1.5906,
       "step": 12020
     },
     {
       "epoch": 36.13,
-      "grad_norm": 9.649632453918457,
+      "grad_norm": 14.263503074645996,
       "learning_rate": 8.795795795795797e-06,
-      "loss": 1.61,
+      "loss": 1.6106,
       "step": 12030
     },
     {
       "epoch": 36.16,
-      "grad_norm": 6.735471248626709,
+      "grad_norm": 7.341001510620117,
       "learning_rate": 8.794794794794795e-06,
-      "loss": 1.5946,
+      "loss": 1.5893,
       "step": 12040
     },
     {
       "epoch": 36.19,
-      "grad_norm": 7.51044225692749,
+      "grad_norm": 9.12683391571045,
       "learning_rate": 8.793793793793794e-06,
-      "loss": 1.5896,
+      "loss": 1.5878,
       "step": 12050
     },
     {
       "epoch": 36.22,
-      "grad_norm": 9.932256698608398,
+      "grad_norm": 13.128108978271484,
       "learning_rate": 8.792792792792794e-06,
-      "loss": 1.5481,
+      "loss": 1.5489,
       "step": 12060
     },
     {
       "epoch": 36.25,
-      "grad_norm": 9.809910774230957,
+      "grad_norm": 8.793357849121094,
       "learning_rate": 8.791791791791793e-06,
-      "loss": 1.4784,
+      "loss": 1.489,
       "step": 12070
     },
     {
       "epoch": 36.28,
-      "grad_norm": 6.632305145263672,
+      "grad_norm": 6.671650409698486,
       "learning_rate": 8.790790790790791e-06,
-      "loss": 1.5636,
+      "loss": 1.5526,
       "step": 12080
     },
     {
       "epoch": 36.31,
-      "grad_norm": 6.605998516082764,
+      "grad_norm": 8.055373191833496,
       "learning_rate": 8.789789789789792e-06,
-      "loss": 1.5755,
+      "loss": 1.5786,
       "step": 12090
     },
     {
       "epoch": 36.34,
-      "grad_norm": 7.216116905212402,
+      "grad_norm": 7.952479362487793,
       "learning_rate": 8.788788788788788e-06,
-      "loss": 1.5557,
+      "loss": 1.5629,
       "step": 12100
     },
     {
       "epoch": 36.37,
-      "grad_norm": 7.483091354370117,
+      "grad_norm": 6.009799480438232,
       "learning_rate": 8.787787787787789e-06,
-      "loss": 1.5574,
+      "loss": 1.554,
       "step": 12110
     },
     {
       "epoch": 36.4,
-      "grad_norm": 11.609368324279785,
+      "grad_norm": 8.469304084777832,
       "learning_rate": 8.786786786786787e-06,
-      "loss": 1.6505,
+      "loss": 1.6573,
       "step": 12120
     },
     {
       "epoch": 36.43,
-      "grad_norm": 8.382104873657227,
+      "grad_norm": 7.952979564666748,
       "learning_rate": 8.785785785785786e-06,
-      "loss": 1.5817,
+      "loss": 1.5881,
       "step": 12130
     },
     {
       "epoch": 36.46,
-      "grad_norm": 8.034529685974121,
+      "grad_norm": 5.807829856872559,
       "learning_rate": 8.784784784784786e-06,
-      "loss": 1.5651,
+      "loss": 1.5595,
       "step": 12140
     },
     {
       "epoch": 36.49,
-      "grad_norm": 8.693317413330078,
+      "grad_norm": 8.942174911499023,
       "learning_rate": 8.783783783783785e-06,
-      "loss": 1.5579,
+      "loss": 1.5433,
       "step": 12150
     },
     {
       "epoch": 36.52,
-      "grad_norm": 8.593585014343262,
+      "grad_norm": 10.823217391967773,
       "learning_rate": 8.782782782782783e-06,
-      "loss": 1.5415,
+      "loss": 1.5574,
       "step": 12160
     },
     {
       "epoch": 36.55,
-      "grad_norm": 8.866618156433105,
+      "grad_norm": 8.005376815795898,
       "learning_rate": 8.781781781781784e-06,
-      "loss": 1.543,
+      "loss": 1.536,
       "step": 12170
     },
     {
       "epoch": 36.58,
-      "grad_norm": 6.761986255645752,
+      "grad_norm": 5.844934463500977,
       "learning_rate": 8.78078078078078e-06,
-      "loss": 1.6012,
+      "loss": 1.6093,
       "step": 12180
     },
     {
       "epoch": 36.61,
-      "grad_norm": 7.192830562591553,
+      "grad_norm": 10.700833320617676,
       "learning_rate": 8.779779779779781e-06,
-      "loss": 1.5582,
+      "loss": 1.562,
       "step": 12190
     },
     {
       "epoch": 36.64,
-      "grad_norm": 6.593108177185059,
+      "grad_norm": 6.294002056121826,
       "learning_rate": 8.77877877877878e-06,
-      "loss": 1.5755,
+      "loss": 1.566,
       "step": 12200
     },
     {
       "epoch": 36.67,
-      "grad_norm": 5.921413421630859,
+      "grad_norm": 6.855638027191162,
       "learning_rate": 8.777777777777778e-06,
-      "loss": 1.6405,
+      "loss": 1.6408,
       "step": 12210
     },
     {
       "epoch": 36.7,
-      "grad_norm": 6.250374794006348,
+      "grad_norm": 7.067981243133545,
       "learning_rate": 8.776776776776778e-06,
-      "loss": 1.5429,
+      "loss": 1.5501,
       "step": 12220
     },
     {
       "epoch": 36.73,
-      "grad_norm": 8.645869255065918,
+      "grad_norm": 7.749777317047119,
       "learning_rate": 8.775775775775777e-06,
-      "loss": 1.5365,
+      "loss": 1.5414,
       "step": 12230
     },
     {
       "epoch": 36.76,
-      "grad_norm": 6.863344669342041,
+      "grad_norm": 8.358349800109863,
       "learning_rate": 8.774774774774776e-06,
-      "loss": 1.5128,
+      "loss": 1.5043,
       "step": 12240
     },
     {
       "epoch": 36.79,
-      "grad_norm": 8.373976707458496,
+      "grad_norm": 6.050278186798096,
       "learning_rate": 8.773773773773774e-06,
-      "loss": 1.6212,
+      "loss": 1.6188,
       "step": 12250
     },
     {
       "epoch": 36.82,
-      "grad_norm": 7.535289287567139,
+      "grad_norm": 10.843846321105957,
       "learning_rate": 8.772772772772773e-06,
-      "loss": 1.4747,
+      "loss": 1.4765,
       "step": 12260
     },
     {
       "epoch": 36.85,
-      "grad_norm": 6.260688781738281,
+      "grad_norm": 8.01616096496582,
       "learning_rate": 8.771771771771771e-06,
-      "loss": 1.5385,
+      "loss": 1.5366,
       "step": 12270
     },
     {
       "epoch": 36.88,
-      "grad_norm": 8.264459609985352,
+      "grad_norm": 6.355355262756348,
       "learning_rate": 8.770770770770772e-06,
-      "loss": 1.5662,
+      "loss": 1.561,
       "step": 12280
     },
     {
       "epoch": 36.91,
-      "grad_norm": 6.866232872009277,
+      "grad_norm": 6.262320041656494,
       "learning_rate": 8.76976976976977e-06,
-      "loss": 1.5867,
+      "loss": 1.5835,
       "step": 12290
     },
     {
       "epoch": 36.94,
-      "grad_norm": 8.39881420135498,
+      "grad_norm": 7.319380760192871,
       "learning_rate": 8.768768768768769e-06,
-      "loss": 1.5997,
+      "loss": 1.589,
       "step": 12300
     },
     {
       "epoch": 36.97,
-      "grad_norm": 5.817479610443115,
+      "grad_norm": 6.33444881439209,
       "learning_rate": 8.767767767767769e-06,
-      "loss": 1.5211,
+      "loss": 1.522,
       "step": 12310
     },
     {
       "epoch": 37.0,
-      "grad_norm": 6.376009464263916,
+      "grad_norm": 9.768819808959961,
       "learning_rate": 8.766766766766768e-06,
-      "loss": 1.6046,
+      "loss": 1.6089,
       "step": 12320
     },
     {
       "epoch": 37.0,
-      "eval_accuracy": 0.7267,
-      "eval_loss": 0.9299409985542297,
-      "eval_runtime": 5.5865,
-      "eval_samples_per_second": 1790.027,
-      "eval_steps_per_second": 7.16,
+      "eval_accuracy": 0.7286,
+      "eval_loss": 0.9266444444656372,
+      "eval_runtime": 5.4301,
+      "eval_samples_per_second": 1841.579,
+      "eval_steps_per_second": 7.366,
       "step": 12321
     },
     {
       "epoch": 37.03,
-      "grad_norm": 7.809796333312988,
+      "grad_norm": 8.823332786560059,
       "learning_rate": 8.765765765765766e-06,
-      "loss": 1.7886,
+      "loss": 1.7708,
       "step": 12330
     },
     {
       "epoch": 37.06,
-      "grad_norm": 5.629174709320068,
+      "grad_norm": 5.745283126831055,
       "learning_rate": 8.764764764764767e-06,
-      "loss": 1.5854,
+      "loss": 1.5915,
       "step": 12340
     },
     {
       "epoch": 37.09,
-      "grad_norm": 9.77220630645752,
+      "grad_norm": 7.569521427154541,
       "learning_rate": 8.763763763763763e-06,
-      "loss": 1.5367,
+      "loss": 1.55,
       "step": 12350
     },
     {
       "epoch": 37.12,
-      "grad_norm": 8.8778657913208,
+      "grad_norm": 8.707975387573242,
       "learning_rate": 8.762762762762764e-06,
-      "loss": 1.5596,
+      "loss": 1.5486,
       "step": 12360
     },
     {
       "epoch": 37.15,
-      "grad_norm": 9.631356239318848,
+      "grad_norm": 9.138187408447266,
       "learning_rate": 8.761761761761762e-06,
-      "loss": 1.522,
+      "loss": 1.5177,
       "step": 12370
     },
     {
       "epoch": 37.18,
-      "grad_norm": 7.99975061416626,
+      "grad_norm": 10.631314277648926,
       "learning_rate": 8.760760760760761e-06,
-      "loss": 1.5748,
+      "loss": 1.5863,
       "step": 12380
     },
     {
       "epoch": 37.21,
-      "grad_norm": 9.824592590332031,
+      "grad_norm": 8.2950439453125,
       "learning_rate": 8.759759759759761e-06,
-      "loss": 1.6058,
+      "loss": 1.6095,
       "step": 12390
     },
     {
       "epoch": 37.24,
-      "grad_norm": 6.616730690002441,
+      "grad_norm": 7.347188949584961,
       "learning_rate": 8.75875875875876e-06,
-      "loss": 1.4705,
+      "loss": 1.4867,
       "step": 12400
     },
     {
       "epoch": 37.27,
-      "grad_norm": 7.151388645172119,
+      "grad_norm": 7.720394611358643,
       "learning_rate": 8.757757757757758e-06,
-      "loss": 1.5369,
+      "loss": 1.5347,
       "step": 12410
     },
     {
       "epoch": 37.3,
-      "grad_norm": 9.817138671875,
+      "grad_norm": 7.7882843017578125,
       "learning_rate": 8.756756756756759e-06,
-      "loss": 1.5388,
+      "loss": 1.547,
       "step": 12420
     },
     {
       "epoch": 37.33,
-      "grad_norm": 8.400540351867676,
+      "grad_norm": 9.213582992553711,
       "learning_rate": 8.755755755755756e-06,
-      "loss": 1.5614,
+      "loss": 1.5488,
       "step": 12430
     },
     {
       "epoch": 37.36,
-      "grad_norm": 7.024816989898682,
+      "grad_norm": 9.050421714782715,
       "learning_rate": 8.754754754754756e-06,
-      "loss": 1.5564,
+      "loss": 1.5545,
       "step": 12440
     },
     {
       "epoch": 37.39,
-      "grad_norm": 8.991178512573242,
+      "grad_norm": 8.187812805175781,
       "learning_rate": 8.753753753753755e-06,
-      "loss": 1.5023,
+      "loss": 1.5103,
       "step": 12450
     },
     {
       "epoch": 37.42,
-      "grad_norm": 7.071702003479004,
+      "grad_norm": 7.797954082489014,
       "learning_rate": 8.752752752752753e-06,
-      "loss": 1.5194,
+      "loss": 1.5114,
       "step": 12460
     },
     {
       "epoch": 37.45,
-      "grad_norm": 6.410610198974609,
+      "grad_norm": 5.592929363250732,
       "learning_rate": 8.751751751751752e-06,
-      "loss": 1.5847,
+      "loss": 1.5832,
       "step": 12470
     },
     {
       "epoch": 37.48,
-      "grad_norm": 10.385780334472656,
+      "grad_norm": 6.916215896606445,
       "learning_rate": 8.750750750750752e-06,
-      "loss": 1.5743,
+      "loss": 1.5625,
       "step": 12480
     },
     {
       "epoch": 37.51,
-      "grad_norm": 7.181984901428223,
+      "grad_norm": 9.420172691345215,
       "learning_rate": 8.74974974974975e-06,
-      "loss": 1.4776,
+      "loss": 1.4756,
       "step": 12490
     },
     {
       "epoch": 37.54,
-      "grad_norm": 7.771545886993408,
+      "grad_norm": 9.815234184265137,
       "learning_rate": 8.74874874874875e-06,
-      "loss": 1.5216,
+      "loss": 1.5253,
       "step": 12500
     },
     {
       "epoch": 37.57,
-      "grad_norm": 6.137798309326172,
+      "grad_norm": 9.823161125183105,
       "learning_rate": 8.747747747747748e-06,
-      "loss": 1.4434,
+      "loss": 1.4466,
       "step": 12510
     },
     {
       "epoch": 37.6,
-      "grad_norm": 7.151047706604004,
+      "grad_norm": 9.952933311462402,
       "learning_rate": 8.746746746746746e-06,
-      "loss": 1.6001,
+      "loss": 1.6043,
       "step": 12520
     },
     {
       "epoch": 37.63,
-      "grad_norm": 9.144684791564941,
+      "grad_norm": 7.417266845703125,
       "learning_rate": 8.745745745745747e-06,
-      "loss": 1.6072,
+      "loss": 1.6079,
       "step": 12530
     },
     {
       "epoch": 37.66,
-      "grad_norm": 9.432862281799316,
+      "grad_norm": 8.509437561035156,
       "learning_rate": 8.744744744744745e-06,
-      "loss": 1.5441,
+      "loss": 1.5469,
       "step": 12540
     },
     {
       "epoch": 37.69,
-      "grad_norm": 6.939050674438477,
+      "grad_norm": 6.592541694641113,
       "learning_rate": 8.743743743743744e-06,
-      "loss": 1.5317,
+      "loss": 1.5372,
       "step": 12550
     },
     {
       "epoch": 37.72,
-      "grad_norm": 13.225080490112305,
+      "grad_norm": 9.413110733032227,
       "learning_rate": 8.742742742742744e-06,
-      "loss": 1.4877,
+      "loss": 1.4868,
       "step": 12560
     },
     {
       "epoch": 37.75,
-      "grad_norm": 7.986232757568359,
+      "grad_norm": 9.14376449584961,
       "learning_rate": 8.741741741741743e-06,
-      "loss": 1.498,
+      "loss": 1.519,
       "step": 12570
     },
     {
       "epoch": 37.78,
-      "grad_norm": 7.083442211151123,
+      "grad_norm": 9.369489669799805,
       "learning_rate": 8.740740740740741e-06,
-      "loss": 1.4667,
+      "loss": 1.479,
       "step": 12580
     },
     {
       "epoch": 37.81,
-      "grad_norm": 5.574126720428467,
+      "grad_norm": 7.318812370300293,
       "learning_rate": 8.739739739739742e-06,
-      "loss": 1.5041,
+      "loss": 1.5044,
       "step": 12590
     },
     {
       "epoch": 37.84,
-      "grad_norm": 5.589112758636475,
+      "grad_norm": 5.852060794830322,
       "learning_rate": 8.738738738738739e-06,
-      "loss": 1.5156,
+      "loss": 1.5132,
       "step": 12600
     },
     {
       "epoch": 37.87,
-      "grad_norm": 6.581096649169922,
+      "grad_norm": 13.736860275268555,
       "learning_rate": 8.737737737737739e-06,
-      "loss": 1.5712,
+      "loss": 1.5671,
       "step": 12610
     },
     {
       "epoch": 37.9,
-      "grad_norm": 6.317755222320557,
+      "grad_norm": 8.44681453704834,
       "learning_rate": 8.736736736736737e-06,
-      "loss": 1.5901,
+      "loss": 1.5926,
       "step": 12620
     },
     {
       "epoch": 37.93,
-      "grad_norm": 10.023707389831543,
+      "grad_norm": 11.455162048339844,
       "learning_rate": 8.735735735735736e-06,
-      "loss": 1.5333,
+      "loss": 1.5344,
       "step": 12630
     },
     {
       "epoch": 37.96,
-      "grad_norm": 6.231907367706299,
+      "grad_norm": 6.118809700012207,
       "learning_rate": 8.734734734734736e-06,
-      "loss": 1.5592,
+      "loss": 1.5495,
       "step": 12640
     },
     {
       "epoch": 37.99,
-      "grad_norm": 7.664279460906982,
+      "grad_norm": 8.833460807800293,
       "learning_rate": 8.733733733733735e-06,
-      "loss": 1.5859,
+      "loss": 1.5882,
       "step": 12650
     },
     {
       "epoch": 38.0,
-      "eval_accuracy": 0.7271,
-      "eval_loss": 0.916785478591919,
-      "eval_runtime": 5.6766,
-      "eval_samples_per_second": 1761.603,
-      "eval_steps_per_second": 7.046,
+      "eval_accuracy": 0.7283,
+      "eval_loss": 0.9160856604576111,
+      "eval_runtime": 5.4016,
+      "eval_samples_per_second": 1851.3,
+      "eval_steps_per_second": 7.405,
       "step": 12654
     },
     {
       "epoch": 38.02,
-      "grad_norm": 6.528322696685791,
+      "grad_norm": 8.859211921691895,
       "learning_rate": 8.732732732732733e-06,
-      "loss": 1.7042,
+      "loss": 1.6777,
       "step": 12660
     },
     {
       "epoch": 38.05,
-      "grad_norm": 5.8936944007873535,
+      "grad_norm": 7.325582981109619,
       "learning_rate": 8.731731731731734e-06,
-      "loss": 1.5106,
+      "loss": 1.5127,
       "step": 12670
     },
     {
       "epoch": 38.08,
-      "grad_norm": 8.244394302368164,
+      "grad_norm": 8.14626407623291,
       "learning_rate": 8.73073073073073e-06,
-      "loss": 1.4537,
+      "loss": 1.4614,
       "step": 12680
     },
     {
       "epoch": 38.11,
-      "grad_norm": 8.503731727600098,
+      "grad_norm": 6.7364325523376465,
       "learning_rate": 8.72972972972973e-06,
-      "loss": 1.5212,
+      "loss": 1.5277,
       "step": 12690
     },
     {
       "epoch": 38.14,
-      "grad_norm": 7.6736249923706055,
+      "grad_norm": 8.56851863861084,
       "learning_rate": 8.72872872872873e-06,
-      "loss": 1.5874,
+      "loss": 1.5902,
       "step": 12700
     },
     {
       "epoch": 38.17,
-      "grad_norm": 7.054069995880127,
+      "grad_norm": 7.063277244567871,
       "learning_rate": 8.727727727727728e-06,
-      "loss": 1.4937,
+      "loss": 1.4994,
       "step": 12710
     },
     {
       "epoch": 38.2,
-      "grad_norm": 7.354074001312256,
+      "grad_norm": 8.465609550476074,
       "learning_rate": 8.726726726726727e-06,
-      "loss": 1.6056,
+      "loss": 1.6251,
       "step": 12720
     },
     {
       "epoch": 38.23,
-      "grad_norm": 6.457119464874268,
+      "grad_norm": 7.244352340698242,
       "learning_rate": 8.725725725725727e-06,
-      "loss": 1.5895,
+      "loss": 1.5954,
       "step": 12730
     },
     {
       "epoch": 38.26,
-      "grad_norm": 9.5778226852417,
+      "grad_norm": 8.702890396118164,
       "learning_rate": 8.724724724724726e-06,
-      "loss": 1.5424,
+      "loss": 1.5464,
       "step": 12740
     },
     {
       "epoch": 38.29,
-      "grad_norm": 6.303635120391846,
+      "grad_norm": 8.866148948669434,
       "learning_rate": 8.723723723723724e-06,
-      "loss": 1.4805,
+      "loss": 1.4813,
       "step": 12750
     },
     {
       "epoch": 38.32,
-      "grad_norm": 10.407458305358887,
+      "grad_norm": 9.31981372833252,
       "learning_rate": 8.722722722722723e-06,
-      "loss": 1.5385,
+      "loss": 1.5311,
       "step": 12760
     },
     {
       "epoch": 38.35,
-      "grad_norm": 6.152017593383789,
+      "grad_norm": 10.371466636657715,
       "learning_rate": 8.721721721721721e-06,
-      "loss": 1.5084,
+      "loss": 1.5022,
       "step": 12770
     },
     {
       "epoch": 38.38,
-      "grad_norm": 7.181787967681885,
+      "grad_norm": 7.12787389755249,
       "learning_rate": 8.720720720720722e-06,
-      "loss": 1.4675,
+      "loss": 1.4804,
       "step": 12780
     },
     {
       "epoch": 38.41,
-      "grad_norm": 7.397234916687012,
+      "grad_norm": 5.9917120933532715,
       "learning_rate": 8.71971971971972e-06,
-      "loss": 1.4523,
+      "loss": 1.436,
       "step": 12790
     },
     {
       "epoch": 38.44,
-      "grad_norm": 9.433257102966309,
+      "grad_norm": 9.675634384155273,
       "learning_rate": 8.718718718718719e-06,
-      "loss": 1.4895,
+      "loss": 1.4808,
       "step": 12800
     },
     {
       "epoch": 38.47,
-      "grad_norm": 7.265789985656738,
+      "grad_norm": 9.71701431274414,
       "learning_rate": 8.71771771771772e-06,
-      "loss": 1.5299,
+      "loss": 1.5405,
       "step": 12810
     },
     {
       "epoch": 38.5,
-      "grad_norm": 8.578165054321289,
+      "grad_norm": 7.634771347045898,
       "learning_rate": 8.716716716716718e-06,
-      "loss": 1.555,
+      "loss": 1.5467,
       "step": 12820
     },
     {
       "epoch": 38.53,
-      "grad_norm": 6.036674499511719,
+      "grad_norm": 6.045835971832275,
       "learning_rate": 8.715715715715716e-06,
-      "loss": 1.5031,
+      "loss": 1.4944,
       "step": 12830
     },
     {
       "epoch": 38.56,
-      "grad_norm": 7.594533920288086,
+      "grad_norm": 7.252416610717773,
       "learning_rate": 8.714714714714717e-06,
-      "loss": 1.4688,
+      "loss": 1.4695,
       "step": 12840
     },
     {
       "epoch": 38.59,
-      "grad_norm": 9.703210830688477,
+      "grad_norm": 8.96562671661377,
       "learning_rate": 8.713713713713714e-06,
-      "loss": 1.5372,
+      "loss": 1.5422,
       "step": 12850
     },
     {
       "epoch": 38.62,
-      "grad_norm": 6.6041340827941895,
+      "grad_norm": 6.343122959136963,
       "learning_rate": 8.712712712712714e-06,
-      "loss": 1.5186,
+      "loss": 1.5126,
       "step": 12860
     },
     {
       "epoch": 38.65,
-      "grad_norm": 11.351973533630371,
+      "grad_norm": 10.314431190490723,
       "learning_rate": 8.711711711711712e-06,
-      "loss": 1.5386,
+      "loss": 1.5523,
       "step": 12870
     },
     {
       "epoch": 38.68,
-      "grad_norm": 6.620852947235107,
+      "grad_norm": 7.459563255310059,
       "learning_rate": 8.710710710710711e-06,
-      "loss": 1.5565,
+      "loss": 1.5537,
       "step": 12880
     },
     {
       "epoch": 38.71,
-      "grad_norm": 7.074520587921143,
+      "grad_norm": 8.131698608398438,
       "learning_rate": 8.709709709709711e-06,
-      "loss": 1.6248,
+      "loss": 1.622,
       "step": 12890
     },
     {
       "epoch": 38.74,
-      "grad_norm": 9.022031784057617,
+      "grad_norm": 9.756227493286133,
       "learning_rate": 8.70870870870871e-06,
-      "loss": 1.5833,
+      "loss": 1.5682,
       "step": 12900
     },
     {
       "epoch": 38.77,
-      "grad_norm": 7.814298629760742,
+      "grad_norm": 6.868837833404541,
       "learning_rate": 8.707707707707708e-06,
-      "loss": 1.5033,
+      "loss": 1.4943,
       "step": 12910
     },
     {
       "epoch": 38.8,
-      "grad_norm": 10.423232078552246,
+      "grad_norm": 9.063949584960938,
       "learning_rate": 8.706706706706707e-06,
-      "loss": 1.5147,
+      "loss": 1.5119,
       "step": 12920
     },
     {
       "epoch": 38.83,
-      "grad_norm": 5.637447357177734,
+      "grad_norm": 5.842473030090332,
       "learning_rate": 8.705705705705706e-06,
-      "loss": 1.508,
+      "loss": 1.5162,
       "step": 12930
     },
     {
       "epoch": 38.86,
-      "grad_norm": 5.528994560241699,
+      "grad_norm": 10.446511268615723,
       "learning_rate": 8.704704704704704e-06,
-      "loss": 1.5748,
+      "loss": 1.5726,
       "step": 12940
     },
     {
       "epoch": 38.89,
-      "grad_norm": 5.762948989868164,
+      "grad_norm": 7.218452453613281,
       "learning_rate": 8.703703703703705e-06,
-      "loss": 1.5502,
+      "loss": 1.5536,
       "step": 12950
     },
     {
       "epoch": 38.92,
-      "grad_norm": 8.151325225830078,
+      "grad_norm": 8.572538375854492,
       "learning_rate": 8.702702702702703e-06,
-      "loss": 1.5498,
+      "loss": 1.5604,
       "step": 12960
     },
     {
       "epoch": 38.95,
-      "grad_norm": 5.5264081954956055,
+      "grad_norm": 5.30932092666626,
       "learning_rate": 8.701701701701702e-06,
-      "loss": 1.5413,
+      "loss": 1.547,
       "step": 12970
     },
     {
       "epoch": 38.98,
-      "grad_norm": 9.976160049438477,
+      "grad_norm": 12.041853904724121,
       "learning_rate": 8.700700700700702e-06,
-      "loss": 1.5149,
+      "loss": 1.5234,
       "step": 12980
     },
     {
       "epoch": 39.0,
-      "eval_accuracy": 0.7301,
-      "eval_loss": 0.9121959209442139,
-      "eval_runtime": 5.6168,
-      "eval_samples_per_second": 1780.358,
-      "eval_steps_per_second": 7.121,
+      "eval_accuracy": 0.73,
+      "eval_loss": 0.9108250141143799,
+      "eval_runtime": 5.4376,
+      "eval_samples_per_second": 1839.036,
+      "eval_steps_per_second": 7.356,
       "step": 12987
     },
     {
       "epoch": 39.01,
-      "grad_norm": 6.584981918334961,
+      "grad_norm": 8.683141708374023,
       "learning_rate": 8.6996996996997e-06,
-      "loss": 1.7385,
+      "loss": 1.7383,
       "step": 12990
     },
     {
       "epoch": 39.04,
-      "grad_norm": 8.775403022766113,
+      "grad_norm": 15.761658668518066,
       "learning_rate": 8.6986986986987e-06,
-      "loss": 1.5522,
+      "loss": 1.5414,
       "step": 13000
     },
     {
       "epoch": 39.07,
-      "grad_norm": 7.844926357269287,
+      "grad_norm": 7.4583916664123535,
       "learning_rate": 8.697697697697698e-06,
-      "loss": 1.4973,
+      "loss": 1.4963,
       "step": 13010
     },
     {
       "epoch": 39.1,
-      "grad_norm": 8.2564697265625,
+      "grad_norm": 6.361630916595459,
       "learning_rate": 8.696696696696696e-06,
-      "loss": 1.5582,
+      "loss": 1.5525,
       "step": 13020
     },
     {
       "epoch": 39.13,
-      "grad_norm": 5.778474807739258,
+      "grad_norm": 6.261593341827393,
       "learning_rate": 8.695695695695697e-06,
-      "loss": 1.5164,
+      "loss": 1.5166,
       "step": 13030
     },
     {
       "epoch": 39.16,
-      "grad_norm": 9.902626991271973,
+      "grad_norm": 6.574051856994629,
       "learning_rate": 8.694694694694695e-06,
-      "loss": 1.5487,
+      "loss": 1.5449,
       "step": 13040
     },
     {
       "epoch": 39.19,
-      "grad_norm": 6.4814677238464355,
+      "grad_norm": 9.611030578613281,
       "learning_rate": 8.693693693693694e-06,
-      "loss": 1.4439,
+      "loss": 1.4544,
       "step": 13050
     },
     {
       "epoch": 39.22,
-      "grad_norm": 6.890933990478516,
+      "grad_norm": 6.934372901916504,
       "learning_rate": 8.692692692692694e-06,
-      "loss": 1.4788,
+      "loss": 1.4833,
       "step": 13060
     },
     {
       "epoch": 39.25,
-      "grad_norm": 7.18506383895874,
+      "grad_norm": 7.26337194442749,
       "learning_rate": 8.691691691691693e-06,
-      "loss": 1.4606,
+      "loss": 1.4623,
       "step": 13070
     },
     {
       "epoch": 39.28,
-      "grad_norm": 8.474082946777344,
+      "grad_norm": 8.203313827514648,
       "learning_rate": 8.690690690690691e-06,
-      "loss": 1.488,
+      "loss": 1.5008,
       "step": 13080
     },
     {
       "epoch": 39.31,
-      "grad_norm": 6.419044017791748,
+      "grad_norm": 10.144271850585938,
       "learning_rate": 8.689689689689692e-06,
-      "loss": 1.5475,
+      "loss": 1.5555,
       "step": 13090
     },
     {
       "epoch": 39.34,
-      "grad_norm": 9.586259841918945,
+      "grad_norm": 8.591413497924805,
       "learning_rate": 8.688688688688689e-06,
-      "loss": 1.5887,
+      "loss": 1.59,
       "step": 13100
     },
     {
       "epoch": 39.37,
-      "grad_norm": 6.768923282623291,
+      "grad_norm": 7.3462324142456055,
       "learning_rate": 8.687687687687689e-06,
-      "loss": 1.6137,
+      "loss": 1.6072,
       "step": 13110
     },
     {
       "epoch": 39.4,
-      "grad_norm": 7.809852600097656,
+      "grad_norm": 8.834155082702637,
       "learning_rate": 8.686686686686687e-06,
-      "loss": 1.5101,
+      "loss": 1.5125,
       "step": 13120
     },
     {
       "epoch": 39.43,
-      "grad_norm": 6.645540714263916,
+      "grad_norm": 6.182363033294678,
       "learning_rate": 8.685685685685686e-06,
-      "loss": 1.5789,
+      "loss": 1.5793,
       "step": 13130
     },
     {
       "epoch": 39.46,
-      "grad_norm": 9.071301460266113,
+      "grad_norm": 7.101614475250244,
       "learning_rate": 8.684684684684686e-06,
-      "loss": 1.5596,
+      "loss": 1.5425,
       "step": 13140
     },
     {
       "epoch": 39.49,
-      "grad_norm": 8.391746520996094,
+      "grad_norm": 8.596870422363281,
       "learning_rate": 8.683683683683685e-06,
-      "loss": 1.5186,
+      "loss": 1.5233,
       "step": 13150
     },
     {
       "epoch": 39.52,
-      "grad_norm": 7.995950698852539,
+      "grad_norm": 6.752140522003174,
       "learning_rate": 8.682682682682684e-06,
-      "loss": 1.4682,
+      "loss": 1.4696,
       "step": 13160
     },
     {
       "epoch": 39.55,
-      "grad_norm": 7.246039390563965,
+      "grad_norm": 7.204463481903076,
       "learning_rate": 8.681681681681682e-06,
-      "loss": 1.4901,
+      "loss": 1.5077,
       "step": 13170
     },
     {
       "epoch": 39.58,
-      "grad_norm": 8.103955268859863,
+      "grad_norm": 8.317066192626953,
       "learning_rate": 8.68068068068068e-06,
-      "loss": 1.5696,
+      "loss": 1.5816,
       "step": 13180
     },
     {
       "epoch": 39.61,
-      "grad_norm": 5.998281955718994,
+      "grad_norm": 6.120133876800537,
       "learning_rate": 8.67967967967968e-06,
-      "loss": 1.5555,
+      "loss": 1.5567,
       "step": 13190
     },
     {
       "epoch": 39.64,
-      "grad_norm": 7.516942024230957,
+      "grad_norm": 9.802016258239746,
       "learning_rate": 8.67867867867868e-06,
-      "loss": 1.5492,
+      "loss": 1.549,
       "step": 13200
     },
     {
       "epoch": 39.67,
-      "grad_norm": 9.211433410644531,
+      "grad_norm": 10.683880805969238,
       "learning_rate": 8.677677677677678e-06,
-      "loss": 1.4875,
+      "loss": 1.5014,
       "step": 13210
     },
     {
       "epoch": 39.7,
-      "grad_norm": 5.883396148681641,
+      "grad_norm": 6.350903511047363,
       "learning_rate": 8.676676676676677e-06,
-      "loss": 1.4978,
+      "loss": 1.4988,
       "step": 13220
     },
     {
       "epoch": 39.73,
-      "grad_norm": 6.711253643035889,
+      "grad_norm": 7.281722068786621,
       "learning_rate": 8.675675675675677e-06,
-      "loss": 1.5474,
+      "loss": 1.5424,
       "step": 13230
     },
     {
       "epoch": 39.76,
-      "grad_norm": 7.515738010406494,
+      "grad_norm": 9.063467979431152,
       "learning_rate": 8.674674674674676e-06,
-      "loss": 1.5446,
+      "loss": 1.5399,
       "step": 13240
     },
     {
       "epoch": 39.79,
-      "grad_norm": 6.926981449127197,
+      "grad_norm": 5.580424785614014,
       "learning_rate": 8.673673673673674e-06,
-      "loss": 1.5936,
+      "loss": 1.5848,
       "step": 13250
     },
     {
       "epoch": 39.82,
-      "grad_norm": 6.212359428405762,
+      "grad_norm": 7.584137439727783,
       "learning_rate": 8.672672672672673e-06,
-      "loss": 1.4818,
+      "loss": 1.4802,
       "step": 13260
     },
     {
       "epoch": 39.85,
-      "grad_norm": 7.678262710571289,
+      "grad_norm": 9.648362159729004,
       "learning_rate": 8.671671671671671e-06,
-      "loss": 1.5669,
+      "loss": 1.5732,
       "step": 13270
     },
     {
       "epoch": 39.88,
-      "grad_norm": 9.181349754333496,
+      "grad_norm": 9.791670799255371,
       "learning_rate": 8.670670670670672e-06,
-      "loss": 1.4599,
+      "loss": 1.4666,
       "step": 13280
     },
     {
       "epoch": 39.91,
-      "grad_norm": 6.939574718475342,
+      "grad_norm": 6.5576934814453125,
       "learning_rate": 8.66966966966967e-06,
-      "loss": 1.4228,
+      "loss": 1.4274,
       "step": 13290
     },
     {
       "epoch": 39.94,
-      "grad_norm": 6.17125940322876,
+      "grad_norm": 7.130560398101807,
       "learning_rate": 8.668668668668669e-06,
-      "loss": 1.4404,
+      "loss": 1.4538,
       "step": 13300
     },
     {
       "epoch": 39.97,
-      "grad_norm": 6.760839462280273,
+      "grad_norm": 7.128941059112549,
       "learning_rate": 8.66766766766767e-06,
-      "loss": 1.5029,
+      "loss": 1.4897,
       "step": 13310
     },
     {
       "epoch": 40.0,
-      "grad_norm": 23.396711349487305,
+      "grad_norm": 32.76634216308594,
       "learning_rate": 8.666666666666668e-06,
-      "loss": 1.6676,
+      "loss": 1.6739,
       "step": 13320
     },
     {
       "epoch": 40.0,
-      "eval_accuracy": 0.7358,
-      "eval_loss": 0.8963929414749146,
-      "eval_runtime": 5.5517,
-      "eval_samples_per_second": 1801.261,
-      "eval_steps_per_second": 7.205,
+      "eval_accuracy": 0.7356,
+      "eval_loss": 0.8964603543281555,
+      "eval_runtime": 5.4175,
+      "eval_samples_per_second": 1845.857,
+      "eval_steps_per_second": 7.383,
       "step": 13320
     },
     {
       "epoch": 40.03,
-      "grad_norm": 9.060013771057129,
+      "grad_norm": 9.37064266204834,
       "learning_rate": 8.665665665665666e-06,
-      "loss": 1.5174,
+      "loss": 1.5275,
       "step": 13330
     },
     {
       "epoch": 40.06,
-      "grad_norm": 7.199926853179932,
+      "grad_norm": 7.3911237716674805,
       "learning_rate": 8.664664664664665e-06,
-      "loss": 1.5856,
+      "loss": 1.5827,
       "step": 13340
     },
     {
       "epoch": 40.09,
-      "grad_norm": 5.5218706130981445,
+      "grad_norm": 9.1837739944458,
       "learning_rate": 8.663663663663664e-06,
-      "loss": 1.4467,
+      "loss": 1.4491,
       "step": 13350
     },
     {
       "epoch": 40.12,
-      "grad_norm": 7.374303340911865,
+      "grad_norm": 8.29172134399414,
       "learning_rate": 8.662662662662664e-06,
-      "loss": 1.4946,
+      "loss": 1.4809,
       "step": 13360
     },
     {
       "epoch": 40.15,
-      "grad_norm": 9.679436683654785,
+      "grad_norm": 13.0404691696167,
       "learning_rate": 8.661661661661662e-06,
-      "loss": 1.5272,
+      "loss": 1.5263,
       "step": 13370
     },
     {
       "epoch": 40.18,
-      "grad_norm": 9.73176097869873,
+      "grad_norm": 9.710476875305176,
       "learning_rate": 8.660660660660661e-06,
-      "loss": 1.5505,
+      "loss": 1.5514,
       "step": 13380
     },
     {
       "epoch": 40.21,
-      "grad_norm": 8.809252738952637,
+      "grad_norm": 9.380827903747559,
       "learning_rate": 8.65965965965966e-06,
-      "loss": 1.5236,
+      "loss": 1.5221,
       "step": 13390
     },
     {
       "epoch": 40.24,
-      "grad_norm": 10.67611312866211,
+      "grad_norm": 8.360639572143555,
       "learning_rate": 8.65865865865866e-06,
-      "loss": 1.5495,
+      "loss": 1.5545,
       "step": 13400
     },
     {
       "epoch": 40.27,
-      "grad_norm": 7.247917175292969,
+      "grad_norm": 7.972372055053711,
       "learning_rate": 8.657657657657659e-06,
-      "loss": 1.5663,
+      "loss": 1.5598,
       "step": 13410
     },
     {
       "epoch": 40.3,
-      "grad_norm": 7.4159932136535645,
+      "grad_norm": 6.340628147125244,
       "learning_rate": 8.656656656656657e-06,
-      "loss": 1.4605,
+      "loss": 1.4647,
       "step": 13420
     },
     {
       "epoch": 40.33,
-      "grad_norm": 9.99395751953125,
+      "grad_norm": 9.266656875610352,
       "learning_rate": 8.655655655655656e-06,
-      "loss": 1.4705,
+      "loss": 1.48,
       "step": 13430
     },
     {
       "epoch": 40.36,
-      "grad_norm": 6.809967517852783,
+      "grad_norm": 8.125178337097168,
       "learning_rate": 8.654654654654654e-06,
-      "loss": 1.4782,
+      "loss": 1.4838,
       "step": 13440
     },
     {
       "epoch": 40.39,
-      "grad_norm": 7.091662883758545,
+      "grad_norm": 8.55246639251709,
       "learning_rate": 8.653653653653655e-06,
-      "loss": 1.5485,
+      "loss": 1.5415,
       "step": 13450
     },
     {
       "epoch": 40.42,
-      "grad_norm": 7.317035675048828,
+      "grad_norm": 8.22523307800293,
       "learning_rate": 8.652652652652653e-06,
-      "loss": 1.5239,
+      "loss": 1.5223,
       "step": 13460
     },
     {
       "epoch": 40.45,
-      "grad_norm": 7.089315414428711,
+      "grad_norm": 8.86189079284668,
       "learning_rate": 8.651651651651652e-06,
-      "loss": 1.4531,
+      "loss": 1.4572,
       "step": 13470
     },
     {
       "epoch": 40.48,
-      "grad_norm": 6.383526802062988,
+      "grad_norm": 7.028423309326172,
       "learning_rate": 8.650650650650652e-06,
-      "loss": 1.5436,
+      "loss": 1.5493,
       "step": 13480
     },
     {
       "epoch": 40.51,
-      "grad_norm": 8.157090187072754,
+      "grad_norm": 9.3717679977417,
       "learning_rate": 8.64964964964965e-06,
-      "loss": 1.5154,
+      "loss": 1.5162,
       "step": 13490
     },
     {
       "epoch": 40.54,
-      "grad_norm": 8.615571975708008,
+      "grad_norm": 7.562497138977051,
       "learning_rate": 8.64864864864865e-06,
-      "loss": 1.5196,
+      "loss": 1.5314,
       "step": 13500
     },
     {
       "epoch": 40.57,
-      "grad_norm": 6.752992153167725,
+      "grad_norm": 8.831315994262695,
       "learning_rate": 8.647647647647648e-06,
-      "loss": 1.5414,
+      "loss": 1.5523,
       "step": 13510
     },
     {
       "epoch": 40.6,
-      "grad_norm": 7.4204816818237305,
+      "grad_norm": 11.634031295776367,
       "learning_rate": 8.646646646646646e-06,
-      "loss": 1.496,
+      "loss": 1.4887,
       "step": 13520
     },
     {
       "epoch": 40.63,
-      "grad_norm": 5.670519828796387,
+      "grad_norm": 7.458806037902832,
       "learning_rate": 8.645645645645647e-06,
-      "loss": 1.4399,
+      "loss": 1.4357,
       "step": 13530
     },
     {
       "epoch": 40.66,
-      "grad_norm": 7.844598770141602,
+      "grad_norm": 7.9527812004089355,
       "learning_rate": 8.644644644644645e-06,
-      "loss": 1.4629,
+      "loss": 1.4616,
       "step": 13540
     },
     {
       "epoch": 40.69,
-      "grad_norm": 10.151703834533691,
+      "grad_norm": 13.524107933044434,
       "learning_rate": 8.643643643643644e-06,
-      "loss": 1.5167,
+      "loss": 1.5235,
       "step": 13550
     },
     {
       "epoch": 40.72,
-      "grad_norm": 8.055994987487793,
+      "grad_norm": 7.419112682342529,
       "learning_rate": 8.642642642642644e-06,
-      "loss": 1.5239,
+      "loss": 1.5153,
       "step": 13560
     },
     {
       "epoch": 40.75,
-      "grad_norm": 7.450234889984131,
+      "grad_norm": 8.697877883911133,
       "learning_rate": 8.641641641641643e-06,
-      "loss": 1.4608,
+      "loss": 1.4565,
       "step": 13570
     },
     {
       "epoch": 40.78,
-      "grad_norm": 7.811600685119629,
+      "grad_norm": 10.214886665344238,
       "learning_rate": 8.640640640640641e-06,
-      "loss": 1.4933,
+      "loss": 1.5091,
       "step": 13580
     },
     {
       "epoch": 40.81,
-      "grad_norm": 8.65772819519043,
+      "grad_norm": 6.667943954467773,
       "learning_rate": 8.63963963963964e-06,
-      "loss": 1.4965,
+      "loss": 1.4928,
       "step": 13590
     },
     {
       "epoch": 40.84,
-      "grad_norm": 10.143614768981934,
+      "grad_norm": 6.128172397613525,
       "learning_rate": 8.638638638638639e-06,
-      "loss": 1.5393,
+      "loss": 1.5408,
       "step": 13600
     },
     {
       "epoch": 40.87,
-      "grad_norm": 10.834226608276367,
+      "grad_norm": 11.619234085083008,
       "learning_rate": 8.637637637637637e-06,
-      "loss": 1.4913,
+      "loss": 1.4938,
       "step": 13610
     },
     {
       "epoch": 40.9,
-      "grad_norm": 8.413848876953125,
+      "grad_norm": 9.616586685180664,
       "learning_rate": 8.636636636636638e-06,
-      "loss": 1.5072,
+      "loss": 1.4933,
       "step": 13620
     },
     {
       "epoch": 40.93,
-      "grad_norm": 7.151975154876709,
+      "grad_norm": 11.268987655639648,
       "learning_rate": 8.635635635635636e-06,
-      "loss": 1.4594,
+      "loss": 1.4603,
       "step": 13630
     },
     {
       "epoch": 40.96,
-      "grad_norm": 8.560602188110352,
+      "grad_norm": 8.737141609191895,
       "learning_rate": 8.634634634634635e-06,
-      "loss": 1.4973,
+      "loss": 1.4947,
       "step": 13640
     },
     {
       "epoch": 40.99,
-      "grad_norm": 7.324845314025879,
+      "grad_norm": 6.364629745483398,
       "learning_rate": 8.633633633633635e-06,
-      "loss": 1.4889,
+      "loss": 1.4998,
       "step": 13650
     },
     {
       "epoch": 41.0,
-      "eval_accuracy": 0.7345,
-      "eval_loss": 0.8964325189590454,
-      "eval_runtime": 5.4951,
-      "eval_samples_per_second": 1819.805,
-      "eval_steps_per_second": 7.279,
+      "eval_accuracy": 0.7349,
+      "eval_loss": 0.8947603702545166,
+      "eval_runtime": 5.1857,
+      "eval_samples_per_second": 1928.367,
+      "eval_steps_per_second": 7.713,
       "step": 13653
     },
     {
       "epoch": 41.02,
-      "grad_norm": 9.016205787658691,
+      "grad_norm": 7.38612174987793,
       "learning_rate": 8.632632632632634e-06,
-      "loss": 1.5259,
+      "loss": 1.5294,
       "step": 13660
     },
     {
       "epoch": 41.05,
-      "grad_norm": 5.718761920928955,
+      "grad_norm": 8.575469017028809,
       "learning_rate": 8.631631631631632e-06,
-      "loss": 1.4488,
+      "loss": 1.4536,
       "step": 13670
     },
     {
       "epoch": 41.08,
-      "grad_norm": 7.946920871734619,
+      "grad_norm": 10.071853637695312,
       "learning_rate": 8.63063063063063e-06,
-      "loss": 1.499,
+      "loss": 1.4983,
       "step": 13680
     },
     {
       "epoch": 41.11,
-      "grad_norm": 7.8564934730529785,
+      "grad_norm": 12.504096984863281,
       "learning_rate": 8.62962962962963e-06,
-      "loss": 1.4165,
+      "loss": 1.4166,
       "step": 13690
     },
     {
       "epoch": 41.14,
-      "grad_norm": 7.071118354797363,
+      "grad_norm": 10.655410766601562,
       "learning_rate": 8.62862862862863e-06,
-      "loss": 1.5277,
+      "loss": 1.5226,
       "step": 13700
     },
     {
       "epoch": 41.17,
-      "grad_norm": 6.548840522766113,
+      "grad_norm": 7.715792179107666,
       "learning_rate": 8.627627627627628e-06,
-      "loss": 1.4652,
+      "loss": 1.468,
       "step": 13710
     },
     {
       "epoch": 41.2,
-      "grad_norm": 7.37835168838501,
+      "grad_norm": 6.398745536804199,
       "learning_rate": 8.626626626626627e-06,
-      "loss": 1.3449,
+      "loss": 1.3461,
       "step": 13720
     },
     {
       "epoch": 41.23,
-      "grad_norm": 6.12250280380249,
+      "grad_norm": 11.171122550964355,
       "learning_rate": 8.625625625625627e-06,
-      "loss": 1.4629,
+      "loss": 1.4707,
       "step": 13730
     },
     {
       "epoch": 41.26,
-      "grad_norm": 11.735047340393066,
+      "grad_norm": 9.673612594604492,
       "learning_rate": 8.624624624624626e-06,
-      "loss": 1.4838,
+      "loss": 1.4776,
       "step": 13740
     },
     {
       "epoch": 41.29,
-      "grad_norm": 7.761935234069824,
+      "grad_norm": 10.58450698852539,
       "learning_rate": 8.623623623623624e-06,
-      "loss": 1.4958,
+      "loss": 1.496,
       "step": 13750
     },
     {
       "epoch": 41.32,
-      "grad_norm": 8.608447074890137,
+      "grad_norm": 6.657037734985352,
       "learning_rate": 8.622622622622623e-06,
-      "loss": 1.4875,
+      "loss": 1.4768,
       "step": 13760
     },
     {
       "epoch": 41.35,
-      "grad_norm": 8.121313095092773,
+      "grad_norm": 10.041502952575684,
       "learning_rate": 8.621621621621622e-06,
-      "loss": 1.4912,
+      "loss": 1.4933,
       "step": 13770
     },
     {
       "epoch": 41.38,
-      "grad_norm": 9.238100051879883,
+      "grad_norm": 7.761663913726807,
       "learning_rate": 8.620620620620622e-06,
-      "loss": 1.6172,
+      "loss": 1.6158,
       "step": 13780
     },
     {
       "epoch": 41.41,
-      "grad_norm": 7.517195224761963,
+      "grad_norm": 10.294576644897461,
       "learning_rate": 8.61961961961962e-06,
-      "loss": 1.4877,
+      "loss": 1.4901,
       "step": 13790
     },
     {
       "epoch": 41.44,
-      "grad_norm": 8.357475280761719,
+      "grad_norm": 9.070921897888184,
       "learning_rate": 8.618618618618619e-06,
-      "loss": 1.4617,
+      "loss": 1.4696,
       "step": 13800
     },
     {
       "epoch": 41.47,
-      "grad_norm": 6.834866046905518,
+      "grad_norm": 5.051883220672607,
       "learning_rate": 8.61761761761762e-06,
-      "loss": 1.4871,
+      "loss": 1.4916,
       "step": 13810
     },
     {
       "epoch": 41.5,
-      "grad_norm": 6.768181800842285,
+      "grad_norm": 7.105180740356445,
       "learning_rate": 8.616616616616618e-06,
-      "loss": 1.462,
+      "loss": 1.4648,
       "step": 13820
     },
     {
       "epoch": 41.53,
-      "grad_norm": 5.691701889038086,
+      "grad_norm": 7.705816745758057,
       "learning_rate": 8.615615615615616e-06,
-      "loss": 1.5011,
+      "loss": 1.5015,
       "step": 13830
     },
     {
       "epoch": 41.56,
-      "grad_norm": 7.510277271270752,
+      "grad_norm": 6.88606071472168,
       "learning_rate": 8.614614614614615e-06,
-      "loss": 1.4593,
+      "loss": 1.4514,
       "step": 13840
     },
     {
       "epoch": 41.59,
-      "grad_norm": 7.263909339904785,
+      "grad_norm": 8.911186218261719,
       "learning_rate": 8.613613613613614e-06,
-      "loss": 1.4626,
+      "loss": 1.4725,
       "step": 13850
     },
     {
       "epoch": 41.62,
-      "grad_norm": 8.045228004455566,
+      "grad_norm": 10.006633758544922,
       "learning_rate": 8.612612612612612e-06,
-      "loss": 1.5199,
+      "loss": 1.5211,
       "step": 13860
     },
     {
       "epoch": 41.65,
-      "grad_norm": 7.961578845977783,
+      "grad_norm": 6.939736366271973,
       "learning_rate": 8.611611611611613e-06,
-      "loss": 1.4969,
+      "loss": 1.4919,
       "step": 13870
     },
     {
       "epoch": 41.68,
-      "grad_norm": 6.891411781311035,
+      "grad_norm": 6.351618766784668,
       "learning_rate": 8.610610610610611e-06,
-      "loss": 1.4573,
+      "loss": 1.4684,
       "step": 13880
     },
     {
       "epoch": 41.71,
-      "grad_norm": 6.501956462860107,
+      "grad_norm": 10.43529224395752,
       "learning_rate": 8.60960960960961e-06,
-      "loss": 1.5013,
+      "loss": 1.501,
       "step": 13890
     },
     {
       "epoch": 41.74,
-      "grad_norm": 6.706328392028809,
+      "grad_norm": 6.531276226043701,
       "learning_rate": 8.60860860860861e-06,
-      "loss": 1.4914,
+      "loss": 1.4795,
       "step": 13900
     },
     {
       "epoch": 41.77,
-      "grad_norm": 10.50699520111084,
+      "grad_norm": 12.934110641479492,
       "learning_rate": 8.607607607607609e-06,
-      "loss": 1.5251,
+      "loss": 1.5316,
       "step": 13910
     },
     {
       "epoch": 41.8,
-      "grad_norm": 8.0604887008667,
+      "grad_norm": 9.197152137756348,
       "learning_rate": 8.606606606606607e-06,
-      "loss": 1.4779,
+      "loss": 1.482,
       "step": 13920
     },
     {
       "epoch": 41.83,
-      "grad_norm": 6.273593902587891,
+      "grad_norm": 7.474654674530029,
       "learning_rate": 8.605605605605606e-06,
-      "loss": 1.4831,
+      "loss": 1.4813,
       "step": 13930
     },
     {
       "epoch": 41.86,
-      "grad_norm": 9.171074867248535,
+      "grad_norm": 8.010626792907715,
       "learning_rate": 8.604604604604604e-06,
-      "loss": 1.4507,
+      "loss": 1.4598,
       "step": 13940
     },
     {
       "epoch": 41.89,
-      "grad_norm": 8.850640296936035,
+      "grad_norm": 6.836661338806152,
       "learning_rate": 8.603603603603605e-06,
-      "loss": 1.465,
+      "loss": 1.4769,
       "step": 13950
     },
     {
       "epoch": 41.92,
-      "grad_norm": 8.69580364227295,
+      "grad_norm": 8.198914527893066,
       "learning_rate": 8.602602602602603e-06,
-      "loss": 1.4979,
+      "loss": 1.4937,
       "step": 13960
     },
     {
       "epoch": 41.95,
-      "grad_norm": 8.949728012084961,
+      "grad_norm": 7.5826945304870605,
       "learning_rate": 8.601601601601602e-06,
-      "loss": 1.508,
+      "loss": 1.5158,
       "step": 13970
     },
     {
       "epoch": 41.98,
-      "grad_norm": 7.396697044372559,
+      "grad_norm": 7.934417247772217,
       "learning_rate": 8.600600600600602e-06,
-      "loss": 1.4958,
+      "loss": 1.4906,
       "step": 13980
     },
     {
       "epoch": 42.0,
-      "eval_accuracy": 0.7374,
-      "eval_loss": 0.8821260929107666,
-      "eval_runtime": 5.6523,
-      "eval_samples_per_second": 1769.203,
-      "eval_steps_per_second": 7.077,
+      "eval_accuracy": 0.7403,
+      "eval_loss": 0.8794697523117065,
+      "eval_runtime": 5.0758,
+      "eval_samples_per_second": 1970.139,
+      "eval_steps_per_second": 7.881,
       "step": 13986
     },
     {
       "epoch": 42.01,
-      "grad_norm": 6.3536834716796875,
+      "grad_norm": 6.116641998291016,
       "learning_rate": 8.5995995995996e-06,
-      "loss": 1.4348,
+      "loss": 1.4517,
       "step": 13990
     },
     {
       "epoch": 42.04,
-      "grad_norm": 7.912578105926514,
+      "grad_norm": 5.9460039138793945,
       "learning_rate": 8.5985985985986e-06,
       "loss": 1.4105,
       "step": 14000
     },
     {
       "epoch": 42.07,
-      "grad_norm": 6.1695556640625,
+      "grad_norm": 9.250031471252441,
       "learning_rate": 8.597597597597598e-06,
-      "loss": 1.5373,
+      "loss": 1.5395,
       "step": 14010
     },
     {
       "epoch": 42.1,
-      "grad_norm": 8.513510704040527,
+      "grad_norm": 9.806633949279785,
       "learning_rate": 8.596596596596597e-06,
-      "loss": 1.4819,
+      "loss": 1.4738,
       "step": 14020
     },
     {
       "epoch": 42.13,
-      "grad_norm": 7.013467311859131,
+      "grad_norm": 8.689554214477539,
       "learning_rate": 8.595595595595597e-06,
-      "loss": 1.5146,
+      "loss": 1.5249,
       "step": 14030
     },
     {
       "epoch": 42.16,
-      "grad_norm": 7.284332275390625,
+      "grad_norm": 7.1195902824401855,
       "learning_rate": 8.594594594594595e-06,
-      "loss": 1.4874,
+      "loss": 1.4902,
       "step": 14040
     },
     {
       "epoch": 42.19,
-      "grad_norm": 7.186245441436768,
+      "grad_norm": 9.207574844360352,
       "learning_rate": 8.593593593593594e-06,
-      "loss": 1.5844,
+      "loss": 1.5643,
       "step": 14050
     },
     {
       "epoch": 42.22,
-      "grad_norm": 7.160527229309082,
+      "grad_norm": 7.453217506408691,
       "learning_rate": 8.592592592592593e-06,
-      "loss": 1.4538,
+      "loss": 1.4609,
       "step": 14060
     },
     {
       "epoch": 42.25,
-      "grad_norm": 6.670179843902588,
+      "grad_norm": 9.47752857208252,
       "learning_rate": 8.591591591591593e-06,
-      "loss": 1.4561,
+      "loss": 1.4565,
       "step": 14070
     },
     {
       "epoch": 42.28,
-      "grad_norm": 11.691814422607422,
+      "grad_norm": 10.561897277832031,
       "learning_rate": 8.590590590590592e-06,
-      "loss": 1.4915,
+      "loss": 1.5065,
       "step": 14080
     },
     {
       "epoch": 42.31,
-      "grad_norm": 6.086369514465332,
+      "grad_norm": 6.293869495391846,
       "learning_rate": 8.58958958958959e-06,
-      "loss": 1.4979,
+      "loss": 1.4948,
       "step": 14090
     },
     {
       "epoch": 42.34,
-      "grad_norm": 7.498256683349609,
+      "grad_norm": 7.042434215545654,
       "learning_rate": 8.588588588588589e-06,
-      "loss": 1.516,
+      "loss": 1.5176,
       "step": 14100
     },
     {
       "epoch": 42.37,
-      "grad_norm": 7.703826904296875,
+      "grad_norm": 9.322961807250977,
       "learning_rate": 8.587587587587587e-06,
-      "loss": 1.4375,
+      "loss": 1.442,
       "step": 14110
     },
     {
       "epoch": 42.4,
-      "grad_norm": 6.828069686889648,
+      "grad_norm": 6.568060874938965,
       "learning_rate": 8.586586586586588e-06,
-      "loss": 1.5714,
+      "loss": 1.5674,
       "step": 14120
     },
     {
       "epoch": 42.43,
-      "grad_norm": 10.595792770385742,
+      "grad_norm": 7.796354293823242,
       "learning_rate": 8.585585585585586e-06,
-      "loss": 1.4504,
+      "loss": 1.4365,
       "step": 14130
     },
     {
       "epoch": 42.46,
-      "grad_norm": 7.735227584838867,
+      "grad_norm": 7.734197616577148,
       "learning_rate": 8.584584584584585e-06,
-      "loss": 1.4553,
+      "loss": 1.4567,
       "step": 14140
     },
     {
       "epoch": 42.49,
-      "grad_norm": 8.40160846710205,
+      "grad_norm": 10.818160057067871,
       "learning_rate": 8.583583583583585e-06,
-      "loss": 1.4462,
+      "loss": 1.4515,
       "step": 14150
     },
     {
       "epoch": 42.52,
-      "grad_norm": 5.981650352478027,
+      "grad_norm": 6.197446346282959,
       "learning_rate": 8.582582582582584e-06,
-      "loss": 1.5422,
+      "loss": 1.5419,
       "step": 14160
     },
     {
       "epoch": 42.55,
-      "grad_norm": 5.359212875366211,
+      "grad_norm": 6.741486072540283,
       "learning_rate": 8.581581581581582e-06,
-      "loss": 1.5247,
+      "loss": 1.5169,
       "step": 14170
     },
     {
       "epoch": 42.58,
-      "grad_norm": 5.738935470581055,
+      "grad_norm": 8.374690055847168,
       "learning_rate": 8.580580580580581e-06,
-      "loss": 1.4852,
+      "loss": 1.4885,
       "step": 14180
     },
     {
       "epoch": 42.61,
-      "grad_norm": 6.946876049041748,
+      "grad_norm": 8.66068172454834,
       "learning_rate": 8.57957957957958e-06,
-      "loss": 1.4326,
+      "loss": 1.4277,
       "step": 14190
     },
     {
       "epoch": 42.64,
-      "grad_norm": 8.927626609802246,
+      "grad_norm": 8.994487762451172,
       "learning_rate": 8.57857857857858e-06,
-      "loss": 1.5006,
+      "loss": 1.507,
       "step": 14200
     },
     {
       "epoch": 42.67,
-      "grad_norm": 7.488382339477539,
+      "grad_norm": 8.966557502746582,
       "learning_rate": 8.577577577577578e-06,
-      "loss": 1.5271,
+      "loss": 1.5399,
       "step": 14210
     },
     {
       "epoch": 42.7,
-      "grad_norm": 7.540319442749023,
+      "grad_norm": 9.389662742614746,
       "learning_rate": 8.576576576576577e-06,
-      "loss": 1.5066,
+      "loss": 1.5014,
       "step": 14220
     },
     {
       "epoch": 42.73,
-      "grad_norm": 7.812732219696045,
+      "grad_norm": 10.272703170776367,
       "learning_rate": 8.575575575575577e-06,
-      "loss": 1.4577,
+      "loss": 1.4553,
       "step": 14230
     },
     {
       "epoch": 42.76,
-      "grad_norm": 7.319425106048584,
+      "grad_norm": 8.250292778015137,
       "learning_rate": 8.574574574574576e-06,
-      "loss": 1.4874,
+      "loss": 1.4936,
       "step": 14240
     },
     {
       "epoch": 42.79,
-      "grad_norm": 8.889001846313477,
+      "grad_norm": 7.268491744995117,
       "learning_rate": 8.573573573573574e-06,
-      "loss": 1.4693,
+      "loss": 1.4563,
       "step": 14250
     },
     {
       "epoch": 42.82,
-      "grad_norm": 6.273212432861328,
+      "grad_norm": 8.45854377746582,
       "learning_rate": 8.572572572572573e-06,
-      "loss": 1.4655,
+      "loss": 1.464,
       "step": 14260
     },
     {
       "epoch": 42.85,
-      "grad_norm": 7.192447662353516,
+      "grad_norm": 5.903032302856445,
       "learning_rate": 8.571571571571572e-06,
-      "loss": 1.4124,
+      "loss": 1.4094,
       "step": 14270
     },
     {
       "epoch": 42.88,
-      "grad_norm": 8.508190155029297,
+      "grad_norm": 7.76492977142334,
       "learning_rate": 8.570570570570572e-06,
-      "loss": 1.374,
+      "loss": 1.3933,
       "step": 14280
     },
     {
       "epoch": 42.91,
-      "grad_norm": 6.9127116203308105,
+      "grad_norm": 8.668718338012695,
       "learning_rate": 8.56956956956957e-06,
-      "loss": 1.5834,
+      "loss": 1.5906,
       "step": 14290
     },
     {
       "epoch": 42.94,
-      "grad_norm": 7.772678852081299,
+      "grad_norm": 7.924440860748291,
       "learning_rate": 8.568568568568569e-06,
-      "loss": 1.4692,
+      "loss": 1.4861,
       "step": 14300
     },
     {
       "epoch": 42.97,
-      "grad_norm": 8.469659805297852,
+      "grad_norm": 7.3822550773620605,
       "learning_rate": 8.567567567567568e-06,
-      "loss": 1.4397,
+      "loss": 1.4338,
       "step": 14310
     },
     {
       "epoch": 43.0,
-      "eval_accuracy": 0.7441,
-      "eval_loss": 0.8732805848121643,
-      "eval_runtime": 5.4193,
-      "eval_samples_per_second": 1845.255,
-      "eval_steps_per_second": 7.381,
+      "eval_accuracy": 0.7412,
+      "eval_loss": 0.8715906143188477,
+      "eval_runtime": 5.3571,
+      "eval_samples_per_second": 1866.666,
+      "eval_steps_per_second": 7.467,
       "step": 14319
     },
     {
       "epoch": 43.0,
-      "grad_norm": 8.970520973205566,
+      "grad_norm": 8.291768074035645,
       "learning_rate": 8.566566566566568e-06,
-      "loss": 1.6106,
+      "loss": 1.6214,
       "step": 14320
     },
     {
       "epoch": 43.03,
-      "grad_norm": 7.995708465576172,
+      "grad_norm": 8.9520845413208,
       "learning_rate": 8.565565565565567e-06,
-      "loss": 1.4432,
+      "loss": 1.4468,
       "step": 14330
     },
     {
       "epoch": 43.06,
-      "grad_norm": 7.327589988708496,
+      "grad_norm": 8.546489715576172,
       "learning_rate": 8.564564564564565e-06,
-      "loss": 1.5503,
+      "loss": 1.5461,
       "step": 14340
     },
     {
       "epoch": 43.09,
-      "grad_norm": 6.406635761260986,
+      "grad_norm": 7.653972625732422,
       "learning_rate": 8.563563563563564e-06,
-      "loss": 1.5415,
+      "loss": 1.5291,
       "step": 14350
     },
     {
       "epoch": 43.12,
-      "grad_norm": 10.024909019470215,
+      "grad_norm": 8.117642402648926,
       "learning_rate": 8.562562562562562e-06,
-      "loss": 1.4196,
+      "loss": 1.4218,
       "step": 14360
     },
     {
       "epoch": 43.15,
-      "grad_norm": 7.450390815734863,
+      "grad_norm": 6.359185695648193,
       "learning_rate": 8.561561561561563e-06,
-      "loss": 1.5265,
+      "loss": 1.5304,
       "step": 14370
     },
     {
       "epoch": 43.18,
-      "grad_norm": 6.3838911056518555,
+      "grad_norm": 6.350496292114258,
       "learning_rate": 8.560560560560561e-06,
-      "loss": 1.5054,
+      "loss": 1.5106,
       "step": 14380
     },
     {
       "epoch": 43.21,
-      "grad_norm": 7.521673679351807,
+      "grad_norm": 8.748299598693848,
       "learning_rate": 8.55955955955956e-06,
-      "loss": 1.4725,
+      "loss": 1.4657,
       "step": 14390
     },
     {
       "epoch": 43.24,
-      "grad_norm": 7.370839595794678,
+      "grad_norm": 6.422264575958252,
       "learning_rate": 8.55855855855856e-06,
-      "loss": 1.3674,
+      "loss": 1.3718,
       "step": 14400
     },
     {
       "epoch": 43.27,
-      "grad_norm": 9.50501823425293,
+      "grad_norm": 9.329118728637695,
       "learning_rate": 8.557557557557559e-06,
-      "loss": 1.5461,
+      "loss": 1.5481,
       "step": 14410
     },
     {
       "epoch": 43.3,
-      "grad_norm": 5.212452411651611,
+      "grad_norm": 5.955794811248779,
       "learning_rate": 8.556556556556557e-06,
-      "loss": 1.3977,
+      "loss": 1.404,
       "step": 14420
     },
     {
       "epoch": 43.33,
-      "grad_norm": 7.384583950042725,
+      "grad_norm": 9.30413818359375,
       "learning_rate": 8.555555555555556e-06,
-      "loss": 1.4408,
+      "loss": 1.4473,
       "step": 14430
     },
     {
       "epoch": 43.36,
-      "grad_norm": 6.954019546508789,
+      "grad_norm": 9.49109172821045,
       "learning_rate": 8.554554554554554e-06,
-      "loss": 1.4754,
+      "loss": 1.4704,
       "step": 14440
     },
     {
       "epoch": 43.39,
-      "grad_norm": 7.416519641876221,
+      "grad_norm": 9.149821281433105,
       "learning_rate": 8.553553553553555e-06,
-      "loss": 1.505,
+      "loss": 1.5052,
       "step": 14450
     },
     {
       "epoch": 43.42,
-      "grad_norm": 7.7501325607299805,
+      "grad_norm": 7.728297710418701,
       "learning_rate": 8.552552552552553e-06,
-      "loss": 1.4826,
+      "loss": 1.4869,
       "step": 14460
     },
     {
       "epoch": 43.45,
-      "grad_norm": 6.9957356452941895,
+      "grad_norm": 7.6127519607543945,
       "learning_rate": 8.551551551551552e-06,
-      "loss": 1.4709,
+      "loss": 1.4808,
       "step": 14470
     },
     {
       "epoch": 43.48,
-      "grad_norm": 7.0138630867004395,
+      "grad_norm": 6.036713600158691,
       "learning_rate": 8.550550550550552e-06,
-      "loss": 1.5701,
+      "loss": 1.5613,
       "step": 14480
     },
     {
       "epoch": 43.51,
-      "grad_norm": 6.901229381561279,
+      "grad_norm": 10.07936954498291,
       "learning_rate": 8.549549549549551e-06,
-      "loss": 1.423,
+      "loss": 1.4353,
       "step": 14490
     },
     {
       "epoch": 43.54,
-      "grad_norm": 6.574093341827393,
+      "grad_norm": 7.663698673248291,
       "learning_rate": 8.54854854854855e-06,
-      "loss": 1.4485,
+      "loss": 1.4468,
       "step": 14500
     },
     {
       "epoch": 43.57,
-      "grad_norm": 7.746938705444336,
+      "grad_norm": 6.6719465255737305,
       "learning_rate": 8.547547547547548e-06,
-      "loss": 1.4395,
+      "loss": 1.4286,
       "step": 14510
     },
     {
       "epoch": 43.6,
-      "grad_norm": 6.227689266204834,
+      "grad_norm": 9.186166763305664,
       "learning_rate": 8.546546546546547e-06,
-      "loss": 1.4308,
+      "loss": 1.419,
       "step": 14520
     },
     {
       "epoch": 43.63,
-      "grad_norm": 9.893877983093262,
+      "grad_norm": 8.28736400604248,
       "learning_rate": 8.545545545545545e-06,
-      "loss": 1.4297,
+      "loss": 1.4269,
       "step": 14530
     },
     {
       "epoch": 43.66,
-      "grad_norm": 9.724652290344238,
+      "grad_norm": 9.125958442687988,
       "learning_rate": 8.544544544544546e-06,
-      "loss": 1.3809,
+      "loss": 1.3894,
       "step": 14540
     },
     {
       "epoch": 43.69,
-      "grad_norm": 8.620214462280273,
+      "grad_norm": 8.484819412231445,
       "learning_rate": 8.543543543543544e-06,
-      "loss": 1.4751,
+      "loss": 1.4766,
       "step": 14550
     },
     {
       "epoch": 43.72,
-      "grad_norm": 7.20008659362793,
+      "grad_norm": 10.133268356323242,
       "learning_rate": 8.542542542542543e-06,
-      "loss": 1.4842,
+      "loss": 1.495,
       "step": 14560
     },
     {
       "epoch": 43.75,
-      "grad_norm": 6.30487585067749,
+      "grad_norm": 7.607728958129883,
       "learning_rate": 8.541541541541543e-06,
-      "loss": 1.4931,
+      "loss": 1.4905,
       "step": 14570
     },
     {
       "epoch": 43.78,
-      "grad_norm": 5.563708782196045,
+      "grad_norm": 5.8572235107421875,
       "learning_rate": 8.540540540540542e-06,
-      "loss": 1.4675,
+      "loss": 1.4732,
       "step": 14580
     },
     {
       "epoch": 43.81,
-      "grad_norm": 6.378332138061523,
+      "grad_norm": 6.192281246185303,
       "learning_rate": 8.53953953953954e-06,
-      "loss": 1.4712,
+      "loss": 1.4662,
       "step": 14590
     },
     {
       "epoch": 43.84,
-      "grad_norm": 6.156585693359375,
+      "grad_norm": 7.178934097290039,
       "learning_rate": 8.538538538538539e-06,
-      "loss": 1.4514,
+      "loss": 1.4418,
       "step": 14600
     },
     {
       "epoch": 43.87,
-      "grad_norm": 8.458283424377441,
+      "grad_norm": 5.8541083335876465,
       "learning_rate": 8.537537537537537e-06,
-      "loss": 1.4225,
+      "loss": 1.4209,
       "step": 14610
     },
     {
       "epoch": 43.9,
-      "grad_norm": 5.715834617614746,
+      "grad_norm": 6.764065742492676,
       "learning_rate": 8.536536536536538e-06,
-      "loss": 1.462,
+      "loss": 1.4667,
       "step": 14620
     },
     {
       "epoch": 43.93,
-      "grad_norm": 7.36702299118042,
+      "grad_norm": 7.149674415588379,
       "learning_rate": 8.535535535535536e-06,
-      "loss": 1.4522,
+      "loss": 1.4377,
       "step": 14630
     },
     {
       "epoch": 43.96,
-      "grad_norm": 5.84977388381958,
+      "grad_norm": 8.634298324584961,
       "learning_rate": 8.534534534534535e-06,
-      "loss": 1.4594,
+      "loss": 1.4558,
       "step": 14640
     },
     {
       "epoch": 43.99,
-      "grad_norm": 6.436122894287109,
+      "grad_norm": 6.402073383331299,
       "learning_rate": 8.533533533533535e-06,
-      "loss": 1.4745,
+      "loss": 1.4686,
       "step": 14650
     },
     {
       "epoch": 44.0,
-      "eval_accuracy": 0.7397,
-      "eval_loss": 0.8682619333267212,
-      "eval_runtime": 5.5033,
-      "eval_samples_per_second": 1817.096,
-      "eval_steps_per_second": 7.268,
+      "eval_accuracy": 0.7435,
+      "eval_loss": 0.8667036294937134,
+      "eval_runtime": 5.3277,
+      "eval_samples_per_second": 1876.995,
+      "eval_steps_per_second": 7.508,
       "step": 14652
     },
     {
       "epoch": 44.02,
-      "grad_norm": 6.570197105407715,
+      "grad_norm": 7.00133752822876,
       "learning_rate": 8.532532532532534e-06,
-      "loss": 1.6597,
+      "loss": 1.6408,
       "step": 14660
     },
     {
       "epoch": 44.05,
-      "grad_norm": 6.310472011566162,
+      "grad_norm": 7.197926044464111,
       "learning_rate": 8.531531531531532e-06,
-      "loss": 1.4662,
+      "loss": 1.4698,
       "step": 14670
     },
     {
       "epoch": 44.08,
-      "grad_norm": 5.714624404907227,
+      "grad_norm": 8.01180362701416,
       "learning_rate": 8.530530530530531e-06,
-      "loss": 1.4034,
+      "loss": 1.4032,
       "step": 14680
     },
     {
       "epoch": 44.11,
-      "grad_norm": 7.229424476623535,
+      "grad_norm": 10.090925216674805,
       "learning_rate": 8.52952952952953e-06,
-      "loss": 1.4946,
+      "loss": 1.4903,
       "step": 14690
     },
     {
       "epoch": 44.14,
-      "grad_norm": 7.142780303955078,
+      "grad_norm": 10.828726768493652,
       "learning_rate": 8.52852852852853e-06,
-      "loss": 1.5226,
+      "loss": 1.5086,
       "step": 14700
     },
     {
       "epoch": 44.17,
-      "grad_norm": 8.141173362731934,
+      "grad_norm": 9.220129013061523,
       "learning_rate": 8.527527527527528e-06,
-      "loss": 1.4532,
+      "loss": 1.4627,
       "step": 14710
     },
     {
       "epoch": 44.2,
-      "grad_norm": 6.670751094818115,
+      "grad_norm": 9.890852928161621,
       "learning_rate": 8.526526526526527e-06,
-      "loss": 1.4141,
+      "loss": 1.4247,
       "step": 14720
     },
     {
       "epoch": 44.23,
-      "grad_norm": 7.275357246398926,
+      "grad_norm": 9.160088539123535,
       "learning_rate": 8.525525525525527e-06,
-      "loss": 1.3981,
+      "loss": 1.3976,
       "step": 14730
     },
     {
       "epoch": 44.26,
-      "grad_norm": 5.898050308227539,
+      "grad_norm": 6.865250587463379,
       "learning_rate": 8.524524524524526e-06,
-      "loss": 1.4406,
+      "loss": 1.44,
       "step": 14740
     },
     {
       "epoch": 44.29,
-      "grad_norm": 7.797832012176514,
+      "grad_norm": 9.724016189575195,
       "learning_rate": 8.523523523523524e-06,
-      "loss": 1.4237,
+      "loss": 1.417,
       "step": 14750
     },
     {
       "epoch": 44.32,
-      "grad_norm": 10.210169792175293,
+      "grad_norm": 8.665419578552246,
       "learning_rate": 8.522522522522523e-06,
-      "loss": 1.4688,
+      "loss": 1.4709,
       "step": 14760
     },
     {
       "epoch": 44.35,
-      "grad_norm": 7.094913005828857,
+      "grad_norm": 6.1924943923950195,
       "learning_rate": 8.521521521521522e-06,
-      "loss": 1.3571,
+      "loss": 1.3542,
       "step": 14770
     },
     {
       "epoch": 44.38,
-      "grad_norm": 6.03663969039917,
+      "grad_norm": 5.600772857666016,
       "learning_rate": 8.52052052052052e-06,
-      "loss": 1.4295,
+      "loss": 1.4303,
       "step": 14780
     },
     {
       "epoch": 44.41,
-      "grad_norm": 4.848151683807373,
+      "grad_norm": 11.326149940490723,
       "learning_rate": 8.51951951951952e-06,
-      "loss": 1.4041,
+      "loss": 1.3984,
       "step": 14790
     },
     {
       "epoch": 44.44,
-      "grad_norm": 7.423789024353027,
+      "grad_norm": 8.519439697265625,
       "learning_rate": 8.518518518518519e-06,
-      "loss": 1.4622,
+      "loss": 1.4706,
       "step": 14800
     },
     {
       "epoch": 44.47,
-      "grad_norm": 9.715620040893555,
+      "grad_norm": 8.554883003234863,
       "learning_rate": 8.517517517517518e-06,
-      "loss": 1.4916,
+      "loss": 1.4963,
       "step": 14810
     },
     {
       "epoch": 44.5,
-      "grad_norm": 6.749819755554199,
+      "grad_norm": 7.252227783203125,
       "learning_rate": 8.516516516516518e-06,
-      "loss": 1.3917,
+      "loss": 1.3878,
       "step": 14820
     },
     {
       "epoch": 44.53,
-      "grad_norm": 5.880351543426514,
+      "grad_norm": 8.413195610046387,
       "learning_rate": 8.515515515515517e-06,
-      "loss": 1.4889,
+      "loss": 1.476,
       "step": 14830
     },
     {
       "epoch": 44.56,
-      "grad_norm": 6.620260238647461,
+      "grad_norm": 5.621572494506836,
       "learning_rate": 8.514514514514515e-06,
-      "loss": 1.4162,
+      "loss": 1.4261,
       "step": 14840
     },
     {
       "epoch": 44.59,
-      "grad_norm": 8.002942085266113,
+      "grad_norm": 8.919646263122559,
       "learning_rate": 8.513513513513514e-06,
-      "loss": 1.435,
+      "loss": 1.4468,
       "step": 14850
     },
     {
       "epoch": 44.62,
-      "grad_norm": 7.5291595458984375,
+      "grad_norm": 6.7347798347473145,
       "learning_rate": 8.512512512512512e-06,
-      "loss": 1.399,
+      "loss": 1.4032,
       "step": 14860
     },
     {
       "epoch": 44.65,
-      "grad_norm": 5.925547122955322,
+      "grad_norm": 6.891021251678467,
       "learning_rate": 8.511511511511513e-06,
-      "loss": 1.4018,
+      "loss": 1.3905,
       "step": 14870
     },
     {
       "epoch": 44.68,
-      "grad_norm": 6.155929088592529,
+      "grad_norm": 8.915452003479004,
       "learning_rate": 8.510510510510511e-06,
-      "loss": 1.4109,
+      "loss": 1.4046,
       "step": 14880
     },
     {
       "epoch": 44.71,
-      "grad_norm": 8.189133644104004,
+      "grad_norm": 7.260479927062988,
       "learning_rate": 8.50950950950951e-06,
-      "loss": 1.4083,
+      "loss": 1.4087,
       "step": 14890
     },
     {
       "epoch": 44.74,
-      "grad_norm": 6.048114776611328,
+      "grad_norm": 6.273957252502441,
       "learning_rate": 8.50850850850851e-06,
-      "loss": 1.4866,
+      "loss": 1.4838,
       "step": 14900
     },
     {
       "epoch": 44.77,
-      "grad_norm": 7.505711555480957,
+      "grad_norm": 5.43947696685791,
       "learning_rate": 8.507507507507509e-06,
-      "loss": 1.4254,
+      "loss": 1.4275,
       "step": 14910
     },
     {
       "epoch": 44.8,
-      "grad_norm": 5.184854507446289,
+      "grad_norm": 5.321561336517334,
       "learning_rate": 8.506506506506507e-06,
-      "loss": 1.4739,
+      "loss": 1.4665,
       "step": 14920
     },
     {
       "epoch": 44.83,
-      "grad_norm": 11.227068901062012,
+      "grad_norm": 11.173468589782715,
       "learning_rate": 8.505505505505506e-06,
-      "loss": 1.4131,
+      "loss": 1.4082,
       "step": 14930
     },
     {
       "epoch": 44.86,
-      "grad_norm": 8.340692520141602,
+      "grad_norm": 10.235525131225586,
       "learning_rate": 8.504504504504505e-06,
-      "loss": 1.5162,
+      "loss": 1.5188,
       "step": 14940
     },
     {
       "epoch": 44.89,
-      "grad_norm": 6.454249382019043,
+      "grad_norm": 7.794491767883301,
       "learning_rate": 8.503503503503505e-06,
-      "loss": 1.4462,
+      "loss": 1.4352,
       "step": 14950
     },
     {
       "epoch": 44.92,
-      "grad_norm": 6.799215316772461,
+      "grad_norm": 6.854542255401611,
       "learning_rate": 8.502502502502503e-06,
-      "loss": 1.4585,
+      "loss": 1.4459,
       "step": 14960
     },
     {
       "epoch": 44.95,
-      "grad_norm": 10.143248558044434,
+      "grad_norm": 9.730775833129883,
       "learning_rate": 8.501501501501502e-06,
-      "loss": 1.5019,
+      "loss": 1.4969,
       "step": 14970
     },
     {
       "epoch": 44.98,
-      "grad_norm": 7.636873722076416,
+      "grad_norm": 7.421666145324707,
       "learning_rate": 8.5005005005005e-06,
-      "loss": 1.4804,
+      "loss": 1.4763,
       "step": 14980
     },
     {
       "epoch": 45.0,
-      "eval_accuracy": 0.7429,
-      "eval_loss": 0.861449122428894,
-      "eval_runtime": 5.4036,
-      "eval_samples_per_second": 1850.615,
-      "eval_steps_per_second": 7.402,
+      "eval_accuracy": 0.7422,
+      "eval_loss": 0.8605132102966309,
+      "eval_runtime": 5.1462,
+      "eval_samples_per_second": 1943.172,
+      "eval_steps_per_second": 7.773,
       "step": 14985
     },
     {
       "epoch": 45.02,
-      "grad_norm": 8.061476707458496,
+      "grad_norm": 10.892763137817383,
       "learning_rate": 8.499499499499501e-06,
-      "loss": 1.8269,
+      "loss": 1.8298,
       "step": 14990
     },
     {
       "epoch": 45.05,
-      "grad_norm": 5.717976093292236,
+      "grad_norm": 6.38596773147583,
       "learning_rate": 8.4984984984985e-06,
-      "loss": 1.4753,
+      "loss": 1.4629,
       "step": 15000
     },
     {
       "epoch": 45.08,
-      "grad_norm": 8.019486427307129,
+      "grad_norm": 9.494767189025879,
       "learning_rate": 8.497497497497498e-06,
-      "loss": 1.3887,
+      "loss": 1.3814,
       "step": 15010
     },
     {
       "epoch": 45.11,
-      "grad_norm": 8.41002368927002,
+      "grad_norm": 9.780126571655273,
       "learning_rate": 8.496496496496497e-06,
-      "loss": 1.4733,
+      "loss": 1.4712,
       "step": 15020
     },
     {
       "epoch": 45.14,
-      "grad_norm": 7.378839492797852,
+      "grad_norm": 8.27866268157959,
       "learning_rate": 8.495495495495495e-06,
-      "loss": 1.4139,
+      "loss": 1.3993,
       "step": 15030
     },
     {
       "epoch": 45.17,
-      "grad_norm": 7.8913164138793945,
+      "grad_norm": 8.789804458618164,
       "learning_rate": 8.494494494494496e-06,
-      "loss": 1.3613,
+      "loss": 1.3745,
       "step": 15040
     },
     {
       "epoch": 45.2,
-      "grad_norm": 6.641181468963623,
+      "grad_norm": 8.037980079650879,
       "learning_rate": 8.493493493493494e-06,
-      "loss": 1.3997,
+      "loss": 1.4133,
       "step": 15050
     },
     {
       "epoch": 45.23,
-      "grad_norm": 7.315908432006836,
+      "grad_norm": 7.7153167724609375,
       "learning_rate": 8.492492492492493e-06,
-      "loss": 1.4553,
+      "loss": 1.4418,
       "step": 15060
     },
     {
       "epoch": 45.26,
-      "grad_norm": 8.853334426879883,
+      "grad_norm": 8.515518188476562,
       "learning_rate": 8.491491491491493e-06,
-      "loss": 1.4807,
+      "loss": 1.4767,
       "step": 15070
     },
     {
       "epoch": 45.29,
-      "grad_norm": 6.201040744781494,
+      "grad_norm": 5.473116874694824,
       "learning_rate": 8.490490490490492e-06,
-      "loss": 1.4293,
+      "loss": 1.4381,
       "step": 15080
     },
     {
       "epoch": 45.32,
-      "grad_norm": 10.057747840881348,
+      "grad_norm": 7.729551315307617,
       "learning_rate": 8.48948948948949e-06,
-      "loss": 1.5282,
+      "loss": 1.5253,
       "step": 15090
     },
     {
       "epoch": 45.35,
-      "grad_norm": 7.781928062438965,
+      "grad_norm": 11.132505416870117,
       "learning_rate": 8.488488488488489e-06,
-      "loss": 1.3581,
+      "loss": 1.3595,
       "step": 15100
     },
     {
       "epoch": 45.38,
-      "grad_norm": 6.299081802368164,
+      "grad_norm": 8.16064167022705,
       "learning_rate": 8.487487487487487e-06,
-      "loss": 1.4849,
+      "loss": 1.4827,
       "step": 15110
     },
     {
       "epoch": 45.41,
-      "grad_norm": 7.964942932128906,
+      "grad_norm": 8.194117546081543,
       "learning_rate": 8.486486486486488e-06,
-      "loss": 1.4792,
+      "loss": 1.4704,
       "step": 15120
     },
     {
       "epoch": 45.44,
-      "grad_norm": 5.77691650390625,
+      "grad_norm": 6.588953971862793,
       "learning_rate": 8.485485485485486e-06,
-      "loss": 1.4383,
+      "loss": 1.4426,
       "step": 15130
     },
     {
       "epoch": 45.47,
-      "grad_norm": 5.73340368270874,
+      "grad_norm": 8.140968322753906,
       "learning_rate": 8.484484484484485e-06,
-      "loss": 1.4865,
+      "loss": 1.4801,
       "step": 15140
     },
     {
       "epoch": 45.5,
-      "grad_norm": 10.694382667541504,
+      "grad_norm": 8.366596221923828,
       "learning_rate": 8.483483483483485e-06,
-      "loss": 1.45,
+      "loss": 1.4553,
       "step": 15150
     },
     {
       "epoch": 45.53,
-      "grad_norm": 8.617444038391113,
+      "grad_norm": 7.63972806930542,
       "learning_rate": 8.482482482482484e-06,
-      "loss": 1.4234,
+      "loss": 1.415,
       "step": 15160
     },
     {
       "epoch": 45.56,
-      "grad_norm": 9.635292053222656,
+      "grad_norm": 10.53994369506836,
       "learning_rate": 8.481481481481482e-06,
-      "loss": 1.4865,
+      "loss": 1.4827,
       "step": 15170
     },
     {
       "epoch": 45.59,
-      "grad_norm": 5.6295342445373535,
+      "grad_norm": 4.895305633544922,
       "learning_rate": 8.480480480480481e-06,
-      "loss": 1.3304,
+      "loss": 1.3487,
       "step": 15180
     },
     {
       "epoch": 45.62,
-      "grad_norm": 7.559775352478027,
+      "grad_norm": 9.125795364379883,
       "learning_rate": 8.47947947947948e-06,
-      "loss": 1.4992,
+      "loss": 1.4941,
       "step": 15190
     },
     {
       "epoch": 45.65,
-      "grad_norm": 7.271817684173584,
+      "grad_norm": 6.620217323303223,
       "learning_rate": 8.47847847847848e-06,
-      "loss": 1.4432,
+      "loss": 1.4497,
       "step": 15200
     },
     {
       "epoch": 45.68,
-      "grad_norm": 8.069267272949219,
+      "grad_norm": 7.850025653839111,
       "learning_rate": 8.477477477477478e-06,
-      "loss": 1.4252,
+      "loss": 1.4237,
       "step": 15210
     },
     {
       "epoch": 45.71,
-      "grad_norm": 7.865988254547119,
+      "grad_norm": 7.140985012054443,
       "learning_rate": 8.476476476476477e-06,
-      "loss": 1.3886,
+      "loss": 1.394,
       "step": 15220
     },
     {
       "epoch": 45.74,
-      "grad_norm": 7.480466365814209,
+      "grad_norm": 7.9065704345703125,
       "learning_rate": 8.475475475475476e-06,
-      "loss": 1.3735,
+      "loss": 1.3747,
       "step": 15230
     },
     {
       "epoch": 45.77,
-      "grad_norm": 10.278926849365234,
+      "grad_norm": 11.198274612426758,
       "learning_rate": 8.474474474474476e-06,
-      "loss": 1.4792,
+      "loss": 1.4614,
       "step": 15240
     },
     {
       "epoch": 45.8,
-      "grad_norm": 10.182306289672852,
+      "grad_norm": 10.460881233215332,
       "learning_rate": 8.473473473473475e-06,
-      "loss": 1.4543,
+      "loss": 1.4528,
       "step": 15250
     },
     {
       "epoch": 45.83,
-      "grad_norm": 6.782303810119629,
+      "grad_norm": 7.999690532684326,
       "learning_rate": 8.472472472472473e-06,
-      "loss": 1.4233,
+      "loss": 1.4276,
       "step": 15260
     },
     {
       "epoch": 45.86,
-      "grad_norm": 7.620147228240967,
+      "grad_norm": 7.850228786468506,
       "learning_rate": 8.471471471471472e-06,
-      "loss": 1.3412,
+      "loss": 1.3495,
       "step": 15270
     },
     {
       "epoch": 45.89,
-      "grad_norm": 5.864634990692139,
+      "grad_norm": 7.448070526123047,
       "learning_rate": 8.47047047047047e-06,
-      "loss": 1.4738,
+      "loss": 1.4774,
       "step": 15280
     },
     {
       "epoch": 45.92,
-      "grad_norm": 9.7373628616333,
+      "grad_norm": 8.030774116516113,
       "learning_rate": 8.46946946946947e-06,
-      "loss": 1.4328,
+      "loss": 1.431,
       "step": 15290
     },
     {
       "epoch": 45.95,
-      "grad_norm": 10.586673736572266,
+      "grad_norm": 8.808553695678711,
       "learning_rate": 8.46846846846847e-06,
-      "loss": 1.3928,
+      "loss": 1.3871,
       "step": 15300
     },
     {
       "epoch": 45.98,
-      "grad_norm": 6.477640628814697,
+      "grad_norm": 7.082596778869629,
       "learning_rate": 8.467467467467468e-06,
-      "loss": 1.4372,
+      "loss": 1.4363,
       "step": 15310
     },
     {
       "epoch": 46.0,
-      "eval_accuracy": 0.7472,
-      "eval_loss": 0.844956636428833,
-      "eval_runtime": 5.2673,
-      "eval_samples_per_second": 1898.513,
-      "eval_steps_per_second": 7.594,
+      "eval_accuracy": 0.7455,
+      "eval_loss": 0.8414418697357178,
+      "eval_runtime": 5.3226,
+      "eval_samples_per_second": 1878.77,
+      "eval_steps_per_second": 7.515,
       "step": 15318
     },
     {
       "epoch": 46.01,
-      "grad_norm": 9.676457405090332,
+      "grad_norm": 6.543123722076416,
       "learning_rate": 8.466466466466468e-06,
-      "loss": 1.6216,
+      "loss": 1.6111,
       "step": 15320
     },
     {
       "epoch": 46.04,
-      "grad_norm": 8.308281898498535,
+      "grad_norm": 10.152899742126465,
       "learning_rate": 8.465465465465467e-06,
-      "loss": 1.4121,
+      "loss": 1.4165,
       "step": 15330
     },
     {
       "epoch": 46.07,
-      "grad_norm": 9.830639839172363,
+      "grad_norm": 9.214171409606934,
       "learning_rate": 8.464464464464465e-06,
-      "loss": 1.3985,
+      "loss": 1.386,
       "step": 15340
     },
     {
       "epoch": 46.1,
-      "grad_norm": 9.109746932983398,
+      "grad_norm": 11.139630317687988,
       "learning_rate": 8.463463463463464e-06,
-      "loss": 1.4203,
+      "loss": 1.411,
       "step": 15350
     },
     {
       "epoch": 46.13,
-      "grad_norm": 6.063082218170166,
+      "grad_norm": 6.958112716674805,
       "learning_rate": 8.462462462462462e-06,
-      "loss": 1.4321,
+      "loss": 1.432,
       "step": 15360
     },
     {
       "epoch": 46.16,
-      "grad_norm": 8.980171203613281,
+      "grad_norm": 11.860631942749023,
       "learning_rate": 8.461461461461463e-06,
-      "loss": 1.4548,
+      "loss": 1.4494,
       "step": 15370
     },
     {
       "epoch": 46.19,
-      "grad_norm": 7.454786777496338,
+      "grad_norm": 9.26951789855957,
       "learning_rate": 8.460460460460461e-06,
-      "loss": 1.4078,
+      "loss": 1.4079,
       "step": 15380
     },
     {
       "epoch": 46.22,
-      "grad_norm": 7.030507564544678,
+      "grad_norm": 7.332733631134033,
       "learning_rate": 8.45945945945946e-06,
-      "loss": 1.4531,
+      "loss": 1.4589,
       "step": 15390
     },
     {
       "epoch": 46.25,
-      "grad_norm": 5.823753833770752,
+      "grad_norm": 7.3712286949157715,
       "learning_rate": 8.45845845845846e-06,
-      "loss": 1.4554,
+      "loss": 1.4533,
       "step": 15400
     },
     {
       "epoch": 46.28,
-      "grad_norm": 6.350651741027832,
+      "grad_norm": 7.416656494140625,
       "learning_rate": 8.457457457457459e-06,
-      "loss": 1.4032,
+      "loss": 1.4086,
       "step": 15410
     },
     {
       "epoch": 46.31,
-      "grad_norm": 7.498602867126465,
+      "grad_norm": 9.551894187927246,
       "learning_rate": 8.456456456456457e-06,
-      "loss": 1.4658,
+      "loss": 1.4657,
       "step": 15420
     },
     {
       "epoch": 46.34,
-      "grad_norm": 7.557168483734131,
+      "grad_norm": 10.324031829833984,
       "learning_rate": 8.455455455455456e-06,
-      "loss": 1.4652,
+      "loss": 1.4709,
       "step": 15430
     },
     {
       "epoch": 46.37,
-      "grad_norm": 6.790234565734863,
+      "grad_norm": 8.53735637664795,
       "learning_rate": 8.454454454454455e-06,
-      "loss": 1.4152,
+      "loss": 1.415,
       "step": 15440
     },
     {
       "epoch": 46.4,
-      "grad_norm": 5.834433078765869,
+      "grad_norm": 7.454155921936035,
       "learning_rate": 8.453453453453453e-06,
-      "loss": 1.3746,
+      "loss": 1.376,
       "step": 15450
     },
     {
       "epoch": 46.43,
-      "grad_norm": 8.18781852722168,
+      "grad_norm": 9.392683029174805,
       "learning_rate": 8.452452452452454e-06,
-      "loss": 1.4107,
+      "loss": 1.4079,
       "step": 15460
     },
     {
       "epoch": 46.46,
-      "grad_norm": 6.943149566650391,
+      "grad_norm": 6.4322662353515625,
       "learning_rate": 8.451451451451452e-06,
-      "loss": 1.5002,
+      "loss": 1.5009,
       "step": 15470
     },
     {
       "epoch": 46.49,
-      "grad_norm": 7.7902302742004395,
+      "grad_norm": 7.907070636749268,
       "learning_rate": 8.45045045045045e-06,
-      "loss": 1.4827,
+      "loss": 1.4817,
       "step": 15480
     },
     {
       "epoch": 46.52,
-      "grad_norm": 7.823202133178711,
+      "grad_norm": 9.576949119567871,
       "learning_rate": 8.449449449449451e-06,
-      "loss": 1.4632,
+      "loss": 1.4448,
       "step": 15490
     },
     {
       "epoch": 46.55,
-      "grad_norm": 8.86640453338623,
+      "grad_norm": 9.21169376373291,
       "learning_rate": 8.44844844844845e-06,
-      "loss": 1.4124,
+      "loss": 1.4147,
       "step": 15500
     },
     {
       "epoch": 46.58,
-      "grad_norm": 6.54612922668457,
+      "grad_norm": 7.629067897796631,
       "learning_rate": 8.447447447447448e-06,
-      "loss": 1.4198,
+      "loss": 1.4222,
       "step": 15510
     },
     {
       "epoch": 46.61,
-      "grad_norm": 8.581781387329102,
+      "grad_norm": 9.623973846435547,
       "learning_rate": 8.446446446446447e-06,
-      "loss": 1.5076,
+      "loss": 1.5149,
       "step": 15520
     },
     {
       "epoch": 46.64,
-      "grad_norm": 7.01729154586792,
+      "grad_norm": 6.282039165496826,
       "learning_rate": 8.445445445445445e-06,
-      "loss": 1.4456,
+      "loss": 1.4563,
       "step": 15530
     },
     {
       "epoch": 46.67,
-      "grad_norm": 7.266882419586182,
+      "grad_norm": 6.028748989105225,
       "learning_rate": 8.444444444444446e-06,
-      "loss": 1.4264,
+      "loss": 1.4287,
       "step": 15540
     },
     {
       "epoch": 46.7,
-      "grad_norm": 7.813553810119629,
+      "grad_norm": 6.770589828491211,
       "learning_rate": 8.443443443443444e-06,
-      "loss": 1.431,
+      "loss": 1.4288,
       "step": 15550
     },
     {
       "epoch": 46.73,
-      "grad_norm": 7.085270404815674,
+      "grad_norm": 6.879647731781006,
       "learning_rate": 8.442442442442443e-06,
-      "loss": 1.4325,
+      "loss": 1.4286,
       "step": 15560
     },
     {
       "epoch": 46.76,
-      "grad_norm": 14.039443969726562,
+      "grad_norm": 9.454889297485352,
       "learning_rate": 8.441441441441443e-06,
-      "loss": 1.4351,
+      "loss": 1.4396,
       "step": 15570
     },
     {
       "epoch": 46.79,
-      "grad_norm": 10.350104331970215,
+      "grad_norm": 9.763193130493164,
       "learning_rate": 8.440440440440442e-06,
-      "loss": 1.3965,
+      "loss": 1.4049,
       "step": 15580
     },
     {
       "epoch": 46.82,
-      "grad_norm": 7.5662055015563965,
+      "grad_norm": 9.356823921203613,
       "learning_rate": 8.43943943943944e-06,
-      "loss": 1.4378,
+      "loss": 1.4321,
       "step": 15590
     },
     {
       "epoch": 46.85,
-      "grad_norm": 8.223228454589844,
+      "grad_norm": 8.431645393371582,
       "learning_rate": 8.438438438438439e-06,
-      "loss": 1.4678,
+      "loss": 1.4753,
       "step": 15600
     },
     {
       "epoch": 46.88,
-      "grad_norm": 7.859408378601074,
+      "grad_norm": 9.401679039001465,
       "learning_rate": 8.437437437437438e-06,
-      "loss": 1.4666,
+      "loss": 1.4581,
       "step": 15610
     },
     {
       "epoch": 46.91,
-      "grad_norm": 7.001174449920654,
+      "grad_norm": 9.924842834472656,
       "learning_rate": 8.436436436436438e-06,
-      "loss": 1.4153,
+      "loss": 1.4135,
       "step": 15620
     },
     {
       "epoch": 46.94,
-      "grad_norm": 9.616021156311035,
+      "grad_norm": 8.267251968383789,
       "learning_rate": 8.435435435435436e-06,
-      "loss": 1.4444,
+      "loss": 1.4451,
       "step": 15630
     },
     {
       "epoch": 46.97,
-      "grad_norm": 9.39572525024414,
+      "grad_norm": 10.816169738769531,
       "learning_rate": 8.434434434434435e-06,
-      "loss": 1.4076,
+      "loss": 1.4095,
       "step": 15640
     },
     {
       "epoch": 47.0,
-      "grad_norm": 9.589388847351074,
+      "grad_norm": 8.907186508178711,
       "learning_rate": 8.433433433433435e-06,
-      "loss": 1.4181,
+      "loss": 1.4145,
       "step": 15650
     },
     {
       "epoch": 47.0,
-      "eval_accuracy": 0.7479,
-      "eval_loss": 0.8381348848342896,
-      "eval_runtime": 5.4262,
-      "eval_samples_per_second": 1842.919,
-      "eval_steps_per_second": 7.372,
+      "eval_accuracy": 0.7485,
+      "eval_loss": 0.836327850818634,
+      "eval_runtime": 5.3468,
+      "eval_samples_per_second": 1870.269,
+      "eval_steps_per_second": 7.481,
       "step": 15651
     },
     {
       "epoch": 47.03,
-      "grad_norm": 8.014568328857422,
+      "grad_norm": 7.49144983291626,
       "learning_rate": 8.432432432432434e-06,
-      "loss": 1.5817,
+      "loss": 1.5684,
       "step": 15660
     },
     {
       "epoch": 47.06,
-      "grad_norm": 6.19480562210083,
+      "grad_norm": 8.035412788391113,
       "learning_rate": 8.43143143143143e-06,
-      "loss": 1.4126,
+      "loss": 1.4134,
       "step": 15670
     },
     {
       "epoch": 47.09,
-      "grad_norm": 7.148534297943115,
+      "grad_norm": 7.019456386566162,
       "learning_rate": 8.430430430430431e-06,
-      "loss": 1.4035,
+      "loss": 1.401,
       "step": 15680
     },
     {
       "epoch": 47.12,
-      "grad_norm": 10.386554718017578,
+      "grad_norm": 7.003774642944336,
       "learning_rate": 8.42942942942943e-06,
-      "loss": 1.3642,
+      "loss": 1.373,
       "step": 15690
     },
     {
       "epoch": 47.15,
-      "grad_norm": 9.391469955444336,
+      "grad_norm": 11.16912841796875,
       "learning_rate": 8.428428428428428e-06,
-      "loss": 1.4366,
+      "loss": 1.425,
       "step": 15700
     },
     {
       "epoch": 47.18,
-      "grad_norm": 7.055384635925293,
+      "grad_norm": 7.829858779907227,
       "learning_rate": 8.427427427427429e-06,
-      "loss": 1.3964,
+      "loss": 1.3853,
       "step": 15710
     },
     {
       "epoch": 47.21,
-      "grad_norm": 7.1285295486450195,
+      "grad_norm": 8.334274291992188,
       "learning_rate": 8.426426426426427e-06,
-      "loss": 1.4964,
+      "loss": 1.5012,
       "step": 15720
     },
     {
       "epoch": 47.24,
-      "grad_norm": 9.838037490844727,
+      "grad_norm": 10.406889915466309,
       "learning_rate": 8.425425425425426e-06,
-      "loss": 1.5118,
+      "loss": 1.5102,
       "step": 15730
     },
     {
       "epoch": 47.27,
-      "grad_norm": 6.283213138580322,
+      "grad_norm": 8.230402946472168,
       "learning_rate": 8.424424424424426e-06,
-      "loss": 1.4161,
+      "loss": 1.4213,
       "step": 15740
     },
     {
       "epoch": 47.3,
-      "grad_norm": 8.287403106689453,
+      "grad_norm": 9.859807014465332,
       "learning_rate": 8.423423423423423e-06,
-      "loss": 1.4405,
+      "loss": 1.4341,
       "step": 15750
     },
     {
       "epoch": 47.33,
-      "grad_norm": 9.93005084991455,
+      "grad_norm": 6.692809581756592,
       "learning_rate": 8.422422422422423e-06,
-      "loss": 1.3903,
+      "loss": 1.3908,
       "step": 15760
     },
     {
       "epoch": 47.36,
-      "grad_norm": 5.702488899230957,
+      "grad_norm": 6.968245983123779,
       "learning_rate": 8.421421421421422e-06,
-      "loss": 1.4369,
+      "loss": 1.4383,
       "step": 15770
     },
     {
       "epoch": 47.39,
-      "grad_norm": 11.39305305480957,
+      "grad_norm": 7.924421310424805,
       "learning_rate": 8.42042042042042e-06,
-      "loss": 1.4195,
+      "loss": 1.4228,
       "step": 15780
     },
     {
       "epoch": 47.42,
-      "grad_norm": 8.695301055908203,
+      "grad_norm": 10.217693328857422,
       "learning_rate": 8.41941941941942e-06,
-      "loss": 1.5078,
+      "loss": 1.5032,
       "step": 15790
     },
     {
       "epoch": 47.45,
-      "grad_norm": 7.13027286529541,
+      "grad_norm": 7.9779157638549805,
       "learning_rate": 8.41841841841842e-06,
-      "loss": 1.3825,
+      "loss": 1.3924,
       "step": 15800
     },
     {
       "epoch": 47.48,
-      "grad_norm": 8.34165096282959,
+      "grad_norm": 9.138983726501465,
       "learning_rate": 8.417417417417418e-06,
-      "loss": 1.4856,
+      "loss": 1.4847,
       "step": 15810
     },
     {
       "epoch": 47.51,
-      "grad_norm": 9.15456485748291,
+      "grad_norm": 8.073201179504395,
       "learning_rate": 8.416416416416418e-06,
-      "loss": 1.4396,
+      "loss": 1.4464,
       "step": 15820
     },
     {
       "epoch": 47.54,
-      "grad_norm": 7.599703788757324,
+      "grad_norm": 5.869406700134277,
       "learning_rate": 8.415415415415417e-06,
-      "loss": 1.4613,
+      "loss": 1.4642,
       "step": 15830
     },
     {
       "epoch": 47.57,
-      "grad_norm": 7.444950580596924,
+      "grad_norm": 9.273330688476562,
       "learning_rate": 8.414414414414415e-06,
-      "loss": 1.4302,
+      "loss": 1.4285,
       "step": 15840
     },
     {
       "epoch": 47.6,
-      "grad_norm": 5.778788089752197,
+      "grad_norm": 8.974254608154297,
       "learning_rate": 8.413413413413414e-06,
-      "loss": 1.3793,
+      "loss": 1.3689,
       "step": 15850
     },
     {
       "epoch": 47.63,
-      "grad_norm": 10.618995666503906,
+      "grad_norm": 7.54632568359375,
       "learning_rate": 8.412412412412413e-06,
-      "loss": 1.384,
+      "loss": 1.391,
       "step": 15860
     },
     {
       "epoch": 47.66,
-      "grad_norm": 6.3760504722595215,
+      "grad_norm": 5.783203125,
       "learning_rate": 8.411411411411413e-06,
-      "loss": 1.4147,
+      "loss": 1.4129,
       "step": 15870
     },
     {
       "epoch": 47.69,
-      "grad_norm": 7.93298864364624,
+      "grad_norm": 7.873367786407471,
       "learning_rate": 8.410410410410411e-06,
-      "loss": 1.4195,
+      "loss": 1.4012,
       "step": 15880
     },
     {
       "epoch": 47.72,
-      "grad_norm": 10.867164611816406,
+      "grad_norm": 21.1090087890625,
       "learning_rate": 8.40940940940941e-06,
-      "loss": 1.4018,
+      "loss": 1.3988,
       "step": 15890
     },
     {
       "epoch": 47.75,
-      "grad_norm": 6.306931495666504,
+      "grad_norm": 8.180758476257324,
       "learning_rate": 8.408408408408409e-06,
-      "loss": 1.3335,
+      "loss": 1.3315,
       "step": 15900
     },
     {
       "epoch": 47.78,
-      "grad_norm": 7.3455939292907715,
+      "grad_norm": 7.78203010559082,
       "learning_rate": 8.407407407407409e-06,
-      "loss": 1.4596,
+      "loss": 1.4616,
       "step": 15910
     },
     {
       "epoch": 47.81,
-      "grad_norm": 7.81700325012207,
+      "grad_norm": 7.860455513000488,
       "learning_rate": 8.406406406406406e-06,
-      "loss": 1.3425,
+      "loss": 1.3534,
       "step": 15920
     },
     {
       "epoch": 47.84,
-      "grad_norm": 6.166535377502441,
+      "grad_norm": 12.43547248840332,
       "learning_rate": 8.405405405405406e-06,
-      "loss": 1.4172,
+      "loss": 1.4174,
       "step": 15930
     },
     {
       "epoch": 47.87,
-      "grad_norm": 8.014781951904297,
+      "grad_norm": 7.454881191253662,
       "learning_rate": 8.404404404404405e-06,
-      "loss": 1.3972,
+      "loss": 1.3975,
       "step": 15940
     },
     {
       "epoch": 47.9,
-      "grad_norm": 6.930869102478027,
+      "grad_norm": 7.856265068054199,
       "learning_rate": 8.403403403403403e-06,
-      "loss": 1.4274,
+      "loss": 1.4396,
       "step": 15950
     },
     {
       "epoch": 47.93,
-      "grad_norm": 8.030455589294434,
+      "grad_norm": 8.7472505569458,
       "learning_rate": 8.402402402402404e-06,
-      "loss": 1.4425,
+      "loss": 1.4354,
       "step": 15960
     },
     {
       "epoch": 47.96,
-      "grad_norm": 7.458200454711914,
+      "grad_norm": 6.739476680755615,
       "learning_rate": 8.401401401401402e-06,
-      "loss": 1.4034,
+      "loss": 1.3961,
       "step": 15970
     },
     {
       "epoch": 47.99,
-      "grad_norm": 6.264616966247559,
+      "grad_norm": 6.6354217529296875,
       "learning_rate": 8.4004004004004e-06,
-      "loss": 1.4067,
+      "loss": 1.4002,
       "step": 15980
     },
     {
       "epoch": 48.0,
-      "eval_accuracy": 0.7533,
-      "eval_loss": 0.82379150390625,
-      "eval_runtime": 5.3526,
-      "eval_samples_per_second": 1868.235,
-      "eval_steps_per_second": 7.473,
+      "eval_accuracy": 0.757,
+      "eval_loss": 0.8185251355171204,
+      "eval_runtime": 5.3031,
+      "eval_samples_per_second": 1885.701,
+      "eval_steps_per_second": 7.543,
       "step": 15984
     },
     {
       "epoch": 48.02,
-      "grad_norm": 4.93729829788208,
+      "grad_norm": 6.29915714263916,
       "learning_rate": 8.399399399399401e-06,
-      "loss": 1.5095,
+      "loss": 1.4989,
       "step": 15990
     },
     {
       "epoch": 48.05,
-      "grad_norm": 5.5001606941223145,
+      "grad_norm": 5.697961330413818,
       "learning_rate": 8.398398398398398e-06,
-      "loss": 1.4431,
+      "loss": 1.4328,
       "step": 16000
     },
     {
       "epoch": 48.08,
-      "grad_norm": 6.6762614250183105,
+      "grad_norm": 6.163407325744629,
       "learning_rate": 8.397397397397398e-06,
-      "loss": 1.421,
+      "loss": 1.4222,
       "step": 16010
     },
     {
       "epoch": 48.11,
-      "grad_norm": 7.249762535095215,
+      "grad_norm": 7.098549842834473,
       "learning_rate": 8.396396396396397e-06,
-      "loss": 1.3722,
+      "loss": 1.3684,
       "step": 16020
     },
     {
       "epoch": 48.14,
-      "grad_norm": 8.15927791595459,
+      "grad_norm": 5.9155592918396,
       "learning_rate": 8.395395395395395e-06,
-      "loss": 1.4046,
+      "loss": 1.4082,
       "step": 16030
     },
     {
       "epoch": 48.17,
-      "grad_norm": 7.6831536293029785,
+      "grad_norm": 9.88736343383789,
       "learning_rate": 8.394394394394396e-06,
-      "loss": 1.5004,
+      "loss": 1.4948,
       "step": 16040
     },
     {
       "epoch": 48.2,
-      "grad_norm": 7.997128963470459,
+      "grad_norm": 7.806118965148926,
       "learning_rate": 8.393393393393394e-06,
-      "loss": 1.4318,
+      "loss": 1.4359,
       "step": 16050
     },
     {
       "epoch": 48.23,
-      "grad_norm": 7.910896301269531,
+      "grad_norm": 6.846301555633545,
       "learning_rate": 8.392392392392393e-06,
-      "loss": 1.3896,
+      "loss": 1.3935,
       "step": 16060
     },
     {
       "epoch": 48.26,
-      "grad_norm": 5.328464031219482,
+      "grad_norm": 6.422653675079346,
       "learning_rate": 8.391391391391393e-06,
-      "loss": 1.3685,
+      "loss": 1.3691,
       "step": 16070
     },
     {
       "epoch": 48.29,
-      "grad_norm": 7.092910289764404,
+      "grad_norm": 13.492511749267578,
       "learning_rate": 8.390390390390392e-06,
       "loss": 1.4458,
       "step": 16080
     },
     {
       "epoch": 48.32,
-      "grad_norm": 7.285274982452393,
+      "grad_norm": 8.1506929397583,
       "learning_rate": 8.38938938938939e-06,
-      "loss": 1.4087,
+      "loss": 1.4079,
       "step": 16090
     },
     {
       "epoch": 48.35,
-      "grad_norm": 7.26994514465332,
+      "grad_norm": 7.20933198928833,
       "learning_rate": 8.388388388388389e-06,
-      "loss": 1.4179,
+      "loss": 1.4208,
       "step": 16100
     },
     {
       "epoch": 48.38,
-      "grad_norm": 7.324982166290283,
+      "grad_norm": 8.173559188842773,
       "learning_rate": 8.387387387387388e-06,
-      "loss": 1.4036,
+      "loss": 1.4066,
       "step": 16110
     },
     {
       "epoch": 48.41,
-      "grad_norm": 8.98013687133789,
+      "grad_norm": 7.343977451324463,
       "learning_rate": 8.386386386386386e-06,
-      "loss": 1.4185,
+      "loss": 1.4206,
       "step": 16120
     },
     {
       "epoch": 48.44,
-      "grad_norm": 8.31724739074707,
+      "grad_norm": 6.4593071937561035,
       "learning_rate": 8.385385385385386e-06,
-      "loss": 1.4107,
+      "loss": 1.416,
       "step": 16130
     },
     {
       "epoch": 48.47,
-      "grad_norm": 6.668021202087402,
+      "grad_norm": 7.642568111419678,
       "learning_rate": 8.384384384384385e-06,
-      "loss": 1.4233,
+      "loss": 1.4238,
       "step": 16140
     },
     {
       "epoch": 48.5,
-      "grad_norm": 5.45055627822876,
+      "grad_norm": 6.7640581130981445,
       "learning_rate": 8.383383383383384e-06,
-      "loss": 1.3515,
+      "loss": 1.3381,
       "step": 16150
     },
     {
       "epoch": 48.53,
-      "grad_norm": 9.252828598022461,
+      "grad_norm": 8.40674877166748,
       "learning_rate": 8.382382382382384e-06,
-      "loss": 1.4218,
+      "loss": 1.4123,
       "step": 16160
     },
     {
       "epoch": 48.56,
-      "grad_norm": 10.81069278717041,
+      "grad_norm": 11.398109436035156,
       "learning_rate": 8.381381381381381e-06,
-      "loss": 1.3925,
+      "loss": 1.3812,
       "step": 16170
     },
     {
       "epoch": 48.59,
-      "grad_norm": 12.336777687072754,
+      "grad_norm": 10.740222930908203,
       "learning_rate": 8.380380380380381e-06,
-      "loss": 1.4731,
+      "loss": 1.4735,
       "step": 16180
     },
     {
       "epoch": 48.62,
-      "grad_norm": 8.32475757598877,
+      "grad_norm": 8.84660530090332,
       "learning_rate": 8.37937937937938e-06,
-      "loss": 1.3405,
+      "loss": 1.345,
       "step": 16190
     },
     {
       "epoch": 48.65,
-      "grad_norm": 7.7955193519592285,
+      "grad_norm": 10.77902889251709,
       "learning_rate": 8.378378378378378e-06,
-      "loss": 1.4084,
+      "loss": 1.413,
       "step": 16200
     },
     {
       "epoch": 48.68,
-      "grad_norm": 8.408472061157227,
+      "grad_norm": 9.23376178741455,
       "learning_rate": 8.377377377377379e-06,
-      "loss": 1.432,
+      "loss": 1.4347,
       "step": 16210
     },
     {
       "epoch": 48.71,
-      "grad_norm": 9.885305404663086,
+      "grad_norm": 8.02498722076416,
       "learning_rate": 8.376376376376377e-06,
-      "loss": 1.3157,
+      "loss": 1.3147,
       "step": 16220
     },
     {
       "epoch": 48.74,
-      "grad_norm": 10.875459671020508,
+      "grad_norm": 7.663610458374023,
       "learning_rate": 8.375375375375376e-06,
-      "loss": 1.3695,
+      "loss": 1.3715,
       "step": 16230
     },
     {
       "epoch": 48.77,
-      "grad_norm": 8.031193733215332,
+      "grad_norm": 7.544327735900879,
       "learning_rate": 8.374374374374376e-06,
-      "loss": 1.4479,
+      "loss": 1.4469,
       "step": 16240
     },
     {
       "epoch": 48.8,
-      "grad_norm": 8.530632019042969,
+      "grad_norm": 9.005799293518066,
       "learning_rate": 8.373373373373373e-06,
-      "loss": 1.419,
+      "loss": 1.4157,
       "step": 16250
     },
     {
       "epoch": 48.83,
-      "grad_norm": 6.733250141143799,
+      "grad_norm": 9.070554733276367,
       "learning_rate": 8.372372372372373e-06,
-      "loss": 1.3211,
+      "loss": 1.3235,
       "step": 16260
     },
     {
       "epoch": 48.86,
-      "grad_norm": 8.987641334533691,
+      "grad_norm": 11.344923973083496,
       "learning_rate": 8.371371371371372e-06,
-      "loss": 1.4532,
+      "loss": 1.4449,
       "step": 16270
     },
     {
       "epoch": 48.89,
-      "grad_norm": 5.013253688812256,
+      "grad_norm": 5.192044258117676,
       "learning_rate": 8.37037037037037e-06,
-      "loss": 1.3367,
+      "loss": 1.3249,
       "step": 16280
     },
     {
       "epoch": 48.92,
-      "grad_norm": 9.08979606628418,
+      "grad_norm": 10.901623725891113,
       "learning_rate": 8.36936936936937e-06,
-      "loss": 1.3963,
+      "loss": 1.3957,
       "step": 16290
     },
     {
       "epoch": 48.95,
-      "grad_norm": 9.40121841430664,
+      "grad_norm": 8.18507194519043,
       "learning_rate": 8.36836836836837e-06,
       "loss": 1.4128,
       "step": 16300
     },
     {
       "epoch": 48.98,
-      "grad_norm": 9.47415542602539,
+      "grad_norm": 9.09040355682373,
       "learning_rate": 8.367367367367368e-06,
-      "loss": 1.4155,
+      "loss": 1.4248,
       "step": 16310
     },
     {
       "epoch": 49.0,
-      "eval_accuracy": 0.7471,
-      "eval_loss": 0.8282808661460876,
-      "eval_runtime": 5.3834,
-      "eval_samples_per_second": 1857.568,
-      "eval_steps_per_second": 7.43,
+      "eval_accuracy": 0.7492,
+      "eval_loss": 0.8285935521125793,
+      "eval_runtime": 5.3248,
+      "eval_samples_per_second": 1878.007,
+      "eval_steps_per_second": 7.512,
       "step": 16317
     },
     {
       "epoch": 49.01,
-      "grad_norm": 7.204129219055176,
+      "grad_norm": 6.756404399871826,
       "learning_rate": 8.366366366366368e-06,
-      "loss": 1.5667,
+      "loss": 1.5422,
       "step": 16320
     },
     {
       "epoch": 49.04,
-      "grad_norm": 7.83499002456665,
+      "grad_norm": 9.052720069885254,
       "learning_rate": 8.365365365365367e-06,
-      "loss": 1.413,
+      "loss": 1.4109,
       "step": 16330
     },
     {
       "epoch": 49.07,
-      "grad_norm": 9.916139602661133,
+      "grad_norm": 10.844160079956055,
       "learning_rate": 8.364364364364365e-06,
-      "loss": 1.3926,
+      "loss": 1.3966,
       "step": 16340
     },
     {
       "epoch": 49.1,
-      "grad_norm": 9.581124305725098,
+      "grad_norm": 11.54780101776123,
       "learning_rate": 8.363363363363364e-06,
-      "loss": 1.3589,
+      "loss": 1.3645,
       "step": 16350
     },
     {
       "epoch": 49.13,
-      "grad_norm": 6.975750923156738,
+      "grad_norm": 8.275045394897461,
       "learning_rate": 8.362362362362363e-06,
-      "loss": 1.4263,
+      "loss": 1.4277,
       "step": 16360
     },
     {
       "epoch": 49.16,
-      "grad_norm": 9.338273048400879,
+      "grad_norm": 6.351825714111328,
       "learning_rate": 8.361361361361361e-06,
-      "loss": 1.3771,
+      "loss": 1.3763,
       "step": 16370
     },
     {
       "epoch": 49.19,
-      "grad_norm": 8.293461799621582,
+      "grad_norm": 9.936285972595215,
       "learning_rate": 8.360360360360362e-06,
-      "loss": 1.3741,
+      "loss": 1.3711,
       "step": 16380
     },
     {
       "epoch": 49.22,
-      "grad_norm": 6.72150993347168,
+      "grad_norm": 11.28403091430664,
       "learning_rate": 8.35935935935936e-06,
-      "loss": 1.398,
+      "loss": 1.399,
       "step": 16390
     },
     {
       "epoch": 49.25,
-      "grad_norm": 9.897619247436523,
+      "grad_norm": 10.180394172668457,
       "learning_rate": 8.358358358358359e-06,
-      "loss": 1.4134,
+      "loss": 1.4034,
       "step": 16400
     },
     {
       "epoch": 49.28,
-      "grad_norm": 6.550849914550781,
+      "grad_norm": 7.796122074127197,
       "learning_rate": 8.357357357357359e-06,
-      "loss": 1.3975,
+      "loss": 1.3946,
       "step": 16410
     },
     {
       "epoch": 49.31,
-      "grad_norm": 7.16975736618042,
+      "grad_norm": 6.908372402191162,
       "learning_rate": 8.356356356356356e-06,
-      "loss": 1.3712,
+      "loss": 1.3803,
       "step": 16420
     },
     {
       "epoch": 49.34,
-      "grad_norm": 6.953512191772461,
+      "grad_norm": 6.198023319244385,
       "learning_rate": 8.355355355355356e-06,
-      "loss": 1.4273,
+      "loss": 1.4309,
       "step": 16430
     },
     {
       "epoch": 49.37,
-      "grad_norm": 10.17813777923584,
+      "grad_norm": 8.003575325012207,
       "learning_rate": 8.354354354354355e-06,
-      "loss": 1.4531,
+      "loss": 1.4533,
       "step": 16440
     },
     {
       "epoch": 49.4,
-      "grad_norm": 7.407919406890869,
+      "grad_norm": 7.156608581542969,
       "learning_rate": 8.353353353353353e-06,
-      "loss": 1.4332,
+      "loss": 1.4357,
       "step": 16450
     },
     {
       "epoch": 49.43,
-      "grad_norm": 6.824948310852051,
+      "grad_norm": 7.664849281311035,
       "learning_rate": 8.352352352352354e-06,
-      "loss": 1.3504,
+      "loss": 1.353,
       "step": 16460
     },
     {
       "epoch": 49.46,
-      "grad_norm": 8.62364387512207,
+      "grad_norm": 8.405073165893555,
       "learning_rate": 8.351351351351352e-06,
-      "loss": 1.503,
+      "loss": 1.5095,
       "step": 16470
     },
     {
       "epoch": 49.49,
-      "grad_norm": 8.231127738952637,
+      "grad_norm": 6.380067348480225,
       "learning_rate": 8.35035035035035e-06,
-      "loss": 1.3999,
+      "loss": 1.4005,
       "step": 16480
     },
     {
       "epoch": 49.52,
-      "grad_norm": 6.75250768661499,
+      "grad_norm": 6.0467915534973145,
       "learning_rate": 8.349349349349351e-06,
-      "loss": 1.4597,
+      "loss": 1.4506,
       "step": 16490
     },
     {
       "epoch": 49.55,
-      "grad_norm": 9.240398406982422,
+      "grad_norm": 6.987295627593994,
       "learning_rate": 8.348348348348348e-06,
-      "loss": 1.3677,
+      "loss": 1.3662,
       "step": 16500
     },
     {
       "epoch": 49.58,
-      "grad_norm": 9.237610816955566,
+      "grad_norm": 9.495981216430664,
       "learning_rate": 8.347347347347348e-06,
-      "loss": 1.3243,
+      "loss": 1.3193,
       "step": 16510
     },
     {
       "epoch": 49.61,
-      "grad_norm": 6.507696151733398,
+      "grad_norm": 7.538782596588135,
       "learning_rate": 8.346346346346347e-06,
-      "loss": 1.5075,
+      "loss": 1.5154,
       "step": 16520
     },
     {
       "epoch": 49.64,
-      "grad_norm": 5.557479381561279,
+      "grad_norm": 7.226510524749756,
       "learning_rate": 8.345345345345346e-06,
-      "loss": 1.3979,
+      "loss": 1.3971,
       "step": 16530
     },
     {
       "epoch": 49.67,
-      "grad_norm": 10.215523719787598,
+      "grad_norm": 8.800800323486328,
       "learning_rate": 8.344344344344346e-06,
-      "loss": 1.4393,
+      "loss": 1.4347,
       "step": 16540
     },
     {
       "epoch": 49.7,
-      "grad_norm": 7.863548755645752,
+      "grad_norm": 7.613311767578125,
       "learning_rate": 8.343343343343344e-06,
-      "loss": 1.4773,
+      "loss": 1.4761,
       "step": 16550
     },
     {
       "epoch": 49.73,
-      "grad_norm": 7.835016250610352,
+      "grad_norm": 6.4197678565979,
       "learning_rate": 8.342342342342343e-06,
-      "loss": 1.4439,
+      "loss": 1.4399,
       "step": 16560
     },
     {
       "epoch": 49.76,
-      "grad_norm": 6.8824310302734375,
+      "grad_norm": 6.774540424346924,
       "learning_rate": 8.341341341341343e-06,
-      "loss": 1.3951,
+      "loss": 1.4008,
       "step": 16570
     },
     {
       "epoch": 49.79,
-      "grad_norm": 8.12121295928955,
+      "grad_norm": 7.2444376945495605,
       "learning_rate": 8.340340340340342e-06,
-      "loss": 1.3978,
+      "loss": 1.3942,
       "step": 16580
     },
     {
       "epoch": 49.82,
-      "grad_norm": 6.690908908843994,
+      "grad_norm": 6.42442512512207,
       "learning_rate": 8.339339339339339e-06,
-      "loss": 1.4199,
+      "loss": 1.4273,
       "step": 16590
     },
     {
       "epoch": 49.85,
-      "grad_norm": 9.7983980178833,
+      "grad_norm": 9.473259925842285,
       "learning_rate": 8.338338338338339e-06,
-      "loss": 1.3918,
+      "loss": 1.3892,
       "step": 16600
     },
     {
       "epoch": 49.88,
-      "grad_norm": 5.9173903465271,
+      "grad_norm": 7.495577335357666,
       "learning_rate": 8.337337337337338e-06,
-      "loss": 1.3826,
+      "loss": 1.3854,
       "step": 16610
     },
     {
       "epoch": 49.91,
-      "grad_norm": 6.43964958190918,
+      "grad_norm": 5.300192356109619,
       "learning_rate": 8.336336336336336e-06,
-      "loss": 1.3389,
+      "loss": 1.3403,
       "step": 16620
     },
     {
       "epoch": 49.94,
-      "grad_norm": 9.479846000671387,
+      "grad_norm": 7.829795837402344,
       "learning_rate": 8.335335335335337e-06,
-      "loss": 1.4198,
+      "loss": 1.4084,
       "step": 16630
     },
     {
       "epoch": 49.97,
-      "grad_norm": 8.055381774902344,
+      "grad_norm": 9.321260452270508,
       "learning_rate": 8.334334334334335e-06,
-      "loss": 1.3486,
+      "loss": 1.344,
       "step": 16640
     },
     {
       "epoch": 50.0,
-      "grad_norm": 25.351207733154297,
+      "grad_norm": 32.36198425292969,
       "learning_rate": 8.333333333333334e-06,
-      "loss": 1.5512,
+      "loss": 1.5606,
       "step": 16650
     },
     {
       "epoch": 50.0,
-      "eval_accuracy": 0.7546,
-      "eval_loss": 0.811343789100647,
-      "eval_runtime": 5.5756,
-      "eval_samples_per_second": 1793.543,
-      "eval_steps_per_second": 7.174,
+      "eval_accuracy": 0.7554,
+      "eval_loss": 0.8076343536376953,
+      "eval_runtime": 5.2428,
+      "eval_samples_per_second": 1907.369,
+      "eval_steps_per_second": 7.629,
       "step": 16650
     },
     {
       "epoch": 50.03,
-      "grad_norm": 12.807840347290039,
+      "grad_norm": 10.275558471679688,
       "learning_rate": 8.332332332332334e-06,
-      "loss": 1.326,
+      "loss": 1.3198,
       "step": 16660
     },
     {
       "epoch": 50.06,
-      "grad_norm": 8.203900337219238,
+      "grad_norm": 11.588313102722168,
       "learning_rate": 8.331331331331331e-06,
-      "loss": 1.4283,
+      "loss": 1.4324,
       "step": 16670
     },
     {
       "epoch": 50.09,
-      "grad_norm": 8.04151439666748,
+      "grad_norm": 7.128358364105225,
       "learning_rate": 8.330330330330331e-06,
-      "loss": 1.4388,
+      "loss": 1.4351,
       "step": 16680
     },
     {
       "epoch": 50.12,
-      "grad_norm": 7.040633201599121,
+      "grad_norm": 5.619128227233887,
       "learning_rate": 8.32932932932933e-06,
-      "loss": 1.3588,
+      "loss": 1.3621,
       "step": 16690
     },
     {
       "epoch": 50.15,
-      "grad_norm": 7.135229110717773,
+      "grad_norm": 6.719021320343018,
       "learning_rate": 8.328328328328328e-06,
-      "loss": 1.3545,
+      "loss": 1.3558,
       "step": 16700
     },
     {
       "epoch": 50.18,
-      "grad_norm": 10.40696907043457,
+      "grad_norm": 7.567909240722656,
       "learning_rate": 8.327327327327329e-06,
-      "loss": 1.4081,
+      "loss": 1.4019,
       "step": 16710
     },
     {
       "epoch": 50.21,
-      "grad_norm": 9.289224624633789,
+      "grad_norm": 7.8522515296936035,
       "learning_rate": 8.326326326326327e-06,
-      "loss": 1.4257,
+      "loss": 1.4265,
       "step": 16720
     },
     {
       "epoch": 50.24,
-      "grad_norm": 10.132648468017578,
+      "grad_norm": 10.429587364196777,
       "learning_rate": 8.325325325325326e-06,
-      "loss": 1.3762,
+      "loss": 1.3771,
       "step": 16730
     },
     {
       "epoch": 50.27,
-      "grad_norm": 7.560726642608643,
+      "grad_norm": 6.391546726226807,
       "learning_rate": 8.324324324324326e-06,
-      "loss": 1.3392,
+      "loss": 1.3417,
       "step": 16740
     },
     {
       "epoch": 50.3,
-      "grad_norm": 9.59815788269043,
+      "grad_norm": 9.621731758117676,
       "learning_rate": 8.323323323323323e-06,
-      "loss": 1.3971,
+      "loss": 1.3927,
       "step": 16750
     },
     {
       "epoch": 50.33,
-      "grad_norm": 8.09432315826416,
+      "grad_norm": 8.556849479675293,
       "learning_rate": 8.322322322322323e-06,
-      "loss": 1.4073,
+      "loss": 1.3969,
       "step": 16760
     },
     {
       "epoch": 50.36,
-      "grad_norm": 6.503231525421143,
+      "grad_norm": 8.98083782196045,
       "learning_rate": 8.321321321321322e-06,
-      "loss": 1.4382,
+      "loss": 1.4364,
       "step": 16770
     },
     {
       "epoch": 50.39,
-      "grad_norm": 7.216829299926758,
+      "grad_norm": 6.297729969024658,
       "learning_rate": 8.32032032032032e-06,
-      "loss": 1.4118,
+      "loss": 1.3993,
       "step": 16780
     },
     {
       "epoch": 50.42,
-      "grad_norm": 7.617650508880615,
+      "grad_norm": 8.285066604614258,
       "learning_rate": 8.31931931931932e-06,
-      "loss": 1.3459,
+      "loss": 1.3403,
       "step": 16790
     },
     {
       "epoch": 50.45,
-      "grad_norm": 6.58013916015625,
+      "grad_norm": 7.464366436004639,
       "learning_rate": 8.31831831831832e-06,
       "loss": 1.41,
       "step": 16800
     },
     {
       "epoch": 50.48,
-      "grad_norm": 6.68101692199707,
+      "grad_norm": 8.363178253173828,
       "learning_rate": 8.317317317317318e-06,
-      "loss": 1.3835,
+      "loss": 1.3677,
       "step": 16810
     },
     {
       "epoch": 50.51,
-      "grad_norm": 5.7715349197387695,
+      "grad_norm": 8.38349723815918,
       "learning_rate": 8.316316316316317e-06,
-      "loss": 1.379,
+      "loss": 1.3714,
       "step": 16820
     },
     {
       "epoch": 50.54,
-      "grad_norm": 7.722923278808594,
+      "grad_norm": 10.494720458984375,
       "learning_rate": 8.315315315315317e-06,
-      "loss": 1.3214,
+      "loss": 1.323,
       "step": 16830
     },
     {
       "epoch": 50.57,
-      "grad_norm": 7.550537586212158,
+      "grad_norm": 8.315917015075684,
       "learning_rate": 8.314314314314314e-06,
-      "loss": 1.3202,
+      "loss": 1.3324,
       "step": 16840
     },
     {
       "epoch": 50.6,
-      "grad_norm": 8.647178649902344,
+      "grad_norm": 10.064863204956055,
       "learning_rate": 8.313313313313314e-06,
-      "loss": 1.4001,
+      "loss": 1.3928,
       "step": 16850
     },
     {
       "epoch": 50.63,
-      "grad_norm": 4.986108303070068,
+      "grad_norm": 6.1749725341796875,
       "learning_rate": 8.312312312312313e-06,
-      "loss": 1.4096,
+      "loss": 1.4082,
       "step": 16860
     },
     {
       "epoch": 50.66,
-      "grad_norm": 8.272957801818848,
+      "grad_norm": 7.395516395568848,
       "learning_rate": 8.311311311311311e-06,
-      "loss": 1.3873,
+      "loss": 1.3938,
       "step": 16870
     },
     {
       "epoch": 50.69,
-      "grad_norm": 6.977585315704346,
+      "grad_norm": 7.748833179473877,
       "learning_rate": 8.310310310310312e-06,
-      "loss": 1.4313,
+      "loss": 1.4274,
       "step": 16880
     },
     {
       "epoch": 50.72,
-      "grad_norm": 6.907491683959961,
+      "grad_norm": 10.147880554199219,
       "learning_rate": 8.30930930930931e-06,
-      "loss": 1.3378,
+      "loss": 1.3414,
       "step": 16890
     },
     {
       "epoch": 50.75,
-      "grad_norm": 7.061301231384277,
+      "grad_norm": 7.621586322784424,
       "learning_rate": 8.308308308308309e-06,
-      "loss": 1.3791,
+      "loss": 1.3735,
       "step": 16900
     },
     {
       "epoch": 50.78,
-      "grad_norm": 5.561334133148193,
+      "grad_norm": 7.925890922546387,
       "learning_rate": 8.307307307307309e-06,
-      "loss": 1.3908,
+      "loss": 1.3891,
       "step": 16910
     },
     {
       "epoch": 50.81,
-      "grad_norm": 9.416394233703613,
+      "grad_norm": 8.277697563171387,
       "learning_rate": 8.306306306306306e-06,
-      "loss": 1.3762,
+      "loss": 1.3901,
       "step": 16920
     },
     {
       "epoch": 50.84,
-      "grad_norm": 6.735710144042969,
+      "grad_norm": 7.985469818115234,
       "learning_rate": 8.305305305305306e-06,
-      "loss": 1.417,
+      "loss": 1.4099,
       "step": 16930
     },
     {
       "epoch": 50.87,
-      "grad_norm": 8.594645500183105,
+      "grad_norm": 10.002822875976562,
       "learning_rate": 8.304304304304305e-06,
-      "loss": 1.5199,
+      "loss": 1.5213,
       "step": 16940
     },
     {
       "epoch": 50.9,
-      "grad_norm": 6.801456451416016,
+      "grad_norm": 6.559893608093262,
       "learning_rate": 8.303303303303303e-06,
-      "loss": 1.3425,
+      "loss": 1.3467,
       "step": 16950
     },
     {
       "epoch": 50.93,
-      "grad_norm": 8.027826309204102,
+      "grad_norm": 7.110071659088135,
       "learning_rate": 8.302302302302304e-06,
-      "loss": 1.3845,
+      "loss": 1.3895,
       "step": 16960
     },
     {
       "epoch": 50.96,
-      "grad_norm": 7.35542631149292,
+      "grad_norm": 6.134400367736816,
       "learning_rate": 8.301301301301302e-06,
-      "loss": 1.4452,
+      "loss": 1.4478,
       "step": 16970
     },
     {
       "epoch": 50.99,
-      "grad_norm": 9.10071849822998,
+      "grad_norm": 9.24449634552002,
       "learning_rate": 8.300300300300301e-06,
-      "loss": 1.3912,
+      "loss": 1.388,
       "step": 16980
     },
     {
       "epoch": 51.0,
-      "eval_accuracy": 0.7582,
-      "eval_loss": 0.8014225959777832,
-      "eval_runtime": 5.6202,
-      "eval_samples_per_second": 1779.294,
-      "eval_steps_per_second": 7.117,
+      "eval_accuracy": 0.757,
+      "eval_loss": 0.7997703552246094,
+      "eval_runtime": 5.2326,
+      "eval_samples_per_second": 1911.105,
+      "eval_steps_per_second": 7.644,
       "step": 16983
     },
     {
       "epoch": 51.02,
-      "grad_norm": 7.276769161224365,
+      "grad_norm": 7.511203765869141,
       "learning_rate": 8.299299299299301e-06,
-      "loss": 1.4634,
+      "loss": 1.4481,
       "step": 16990
     },
     {
       "epoch": 51.05,
-      "grad_norm": 8.341124534606934,
+      "grad_norm": 8.254359245300293,
       "learning_rate": 8.298298298298298e-06,
-      "loss": 1.4131,
+      "loss": 1.4132,
       "step": 17000
     },
     {
       "epoch": 51.08,
-      "grad_norm": 9.24995231628418,
+      "grad_norm": 6.435025215148926,
       "learning_rate": 8.297297297297298e-06,
-      "loss": 1.2992,
+      "loss": 1.2986,
       "step": 17010
     },
     {
       "epoch": 51.11,
-      "grad_norm": 6.4309163093566895,
+      "grad_norm": 7.936938762664795,
       "learning_rate": 8.296296296296297e-06,
-      "loss": 1.4272,
+      "loss": 1.4216,
       "step": 17020
     },
     {
       "epoch": 51.14,
-      "grad_norm": 9.540653228759766,
+      "grad_norm": 11.587587356567383,
       "learning_rate": 8.295295295295296e-06,
-      "loss": 1.4063,
+      "loss": 1.4001,
       "step": 17030
     },
     {
       "epoch": 51.17,
-      "grad_norm": 8.252028465270996,
+      "grad_norm": 10.622941017150879,
       "learning_rate": 8.294294294294294e-06,
-      "loss": 1.3877,
+      "loss": 1.3866,
       "step": 17040
     },
     {
       "epoch": 51.2,
-      "grad_norm": 8.484556198120117,
+      "grad_norm": 7.759365081787109,
       "learning_rate": 8.293293293293294e-06,
-      "loss": 1.4354,
+      "loss": 1.4336,
       "step": 17050
     },
     {
       "epoch": 51.23,
-      "grad_norm": 8.192554473876953,
+      "grad_norm": 10.309462547302246,
       "learning_rate": 8.292292292292293e-06,
-      "loss": 1.4634,
+      "loss": 1.4449,
       "step": 17060
     },
     {
       "epoch": 51.26,
-      "grad_norm": 6.842177867889404,
+      "grad_norm": 5.454629898071289,
       "learning_rate": 8.291291291291292e-06,
-      "loss": 1.3916,
+      "loss": 1.3915,
       "step": 17070
     },
     {
       "epoch": 51.29,
-      "grad_norm": 6.30631685256958,
+      "grad_norm": 6.110672473907471,
       "learning_rate": 8.29029029029029e-06,
-      "loss": 1.3659,
+      "loss": 1.3746,
       "step": 17080
     },
     {
       "epoch": 51.32,
-      "grad_norm": 8.465204238891602,
+      "grad_norm": 8.958422660827637,
       "learning_rate": 8.289289289289289e-06,
-      "loss": 1.4049,
+      "loss": 1.4041,
       "step": 17090
     },
     {
       "epoch": 51.35,
-      "grad_norm": 8.76944351196289,
+      "grad_norm": 12.88503646850586,
       "learning_rate": 8.288288288288289e-06,
-      "loss": 1.3616,
+      "loss": 1.3608,
       "step": 17100
     },
     {
       "epoch": 51.38,
-      "grad_norm": 8.071114540100098,
+      "grad_norm": 11.989151000976562,
       "learning_rate": 8.287287287287288e-06,
-      "loss": 1.3731,
+      "loss": 1.3677,
       "step": 17110
     },
     {
       "epoch": 51.41,
-      "grad_norm": 7.252621650695801,
+      "grad_norm": 15.845516204833984,
       "learning_rate": 8.286286286286286e-06,
-      "loss": 1.3366,
+      "loss": 1.3204,
       "step": 17120
     },
     {
       "epoch": 51.44,
-      "grad_norm": 8.387900352478027,
+      "grad_norm": 7.542552471160889,
       "learning_rate": 8.285285285285287e-06,
-      "loss": 1.4385,
+      "loss": 1.4167,
       "step": 17130
     },
     {
       "epoch": 51.47,
-      "grad_norm": 7.739916801452637,
+      "grad_norm": 8.162999153137207,
       "learning_rate": 8.284284284284285e-06,
-      "loss": 1.4466,
+      "loss": 1.4283,
       "step": 17140
     },
     {
       "epoch": 51.5,
-      "grad_norm": 6.444606304168701,
+      "grad_norm": 8.176424026489258,
       "learning_rate": 8.283283283283284e-06,
-      "loss": 1.3076,
+      "loss": 1.3199,
       "step": 17150
     },
     {
       "epoch": 51.53,
-      "grad_norm": 10.213436126708984,
+      "grad_norm": 10.48132038116455,
       "learning_rate": 8.282282282282284e-06,
-      "loss": 1.3911,
+      "loss": 1.3866,
       "step": 17160
     },
     {
       "epoch": 51.56,
-      "grad_norm": 6.119974613189697,
+      "grad_norm": 6.447178840637207,
       "learning_rate": 8.281281281281281e-06,
-      "loss": 1.4252,
+      "loss": 1.4227,
       "step": 17170
     },
     {
       "epoch": 51.59,
-      "grad_norm": 11.217130661010742,
+      "grad_norm": 9.08845329284668,
       "learning_rate": 8.280280280280281e-06,
-      "loss": 1.3261,
+      "loss": 1.326,
       "step": 17180
     },
     {
       "epoch": 51.62,
-      "grad_norm": 9.957549095153809,
+      "grad_norm": 5.844667434692383,
       "learning_rate": 8.27927927927928e-06,
-      "loss": 1.42,
+      "loss": 1.4244,
       "step": 17190
     },
     {
       "epoch": 51.65,
-      "grad_norm": 6.307145118713379,
+      "grad_norm": 7.326508045196533,
       "learning_rate": 8.278278278278278e-06,
-      "loss": 1.3709,
+      "loss": 1.3576,
       "step": 17200
     },
     {
       "epoch": 51.68,
-      "grad_norm": 8.36529541015625,
+      "grad_norm": 10.347125053405762,
       "learning_rate": 8.277277277277279e-06,
-      "loss": 1.4223,
+      "loss": 1.4147,
       "step": 17210
     },
     {
       "epoch": 51.71,
-      "grad_norm": 9.006486892700195,
+      "grad_norm": 13.477595329284668,
       "learning_rate": 8.276276276276277e-06,
-      "loss": 1.3322,
+      "loss": 1.3485,
       "step": 17220
     },
     {
       "epoch": 51.74,
-      "grad_norm": 7.973412036895752,
+      "grad_norm": 8.50554084777832,
       "learning_rate": 8.275275275275276e-06,
-      "loss": 1.3838,
+      "loss": 1.3796,
       "step": 17230
     },
     {
       "epoch": 51.77,
-      "grad_norm": 6.926713943481445,
+      "grad_norm": 6.074929714202881,
       "learning_rate": 8.274274274274276e-06,
-      "loss": 1.3433,
+      "loss": 1.3321,
       "step": 17240
     },
     {
       "epoch": 51.8,
-      "grad_norm": 8.261855125427246,
+      "grad_norm": 7.923277378082275,
       "learning_rate": 8.273273273273273e-06,
-      "loss": 1.4119,
+      "loss": 1.4046,
       "step": 17250
     },
     {
       "epoch": 51.83,
-      "grad_norm": 9.354049682617188,
+      "grad_norm": 10.157902717590332,
       "learning_rate": 8.272272272272273e-06,
-      "loss": 1.4064,
+      "loss": 1.3891,
       "step": 17260
     },
     {
       "epoch": 51.86,
-      "grad_norm": 8.52313232421875,
+      "grad_norm": 11.469818115234375,
       "learning_rate": 8.271271271271272e-06,
-      "loss": 1.3595,
+      "loss": 1.3528,
       "step": 17270
     },
     {
       "epoch": 51.89,
-      "grad_norm": 9.608983039855957,
+      "grad_norm": 9.379180908203125,
       "learning_rate": 8.27027027027027e-06,
-      "loss": 1.3492,
+      "loss": 1.3528,
       "step": 17280
     },
     {
       "epoch": 51.92,
-      "grad_norm": 5.904521942138672,
+      "grad_norm": 6.629838943481445,
       "learning_rate": 8.26926926926927e-06,
-      "loss": 1.4064,
+      "loss": 1.3952,
       "step": 17290
     },
     {
       "epoch": 51.95,
-      "grad_norm": 7.506507396697998,
+      "grad_norm": 9.450668334960938,
       "learning_rate": 8.26826826826827e-06,
-      "loss": 1.4406,
+      "loss": 1.4514,
       "step": 17300
     },
     {
       "epoch": 51.98,
-      "grad_norm": 8.973819732666016,
+      "grad_norm": 11.127699851989746,
       "learning_rate": 8.267267267267268e-06,
-      "loss": 1.4082,
+      "loss": 1.4089,
       "step": 17310
     },
     {
       "epoch": 52.0,
-      "eval_accuracy": 0.7574,
-      "eval_loss": 0.8070465922355652,
-      "eval_runtime": 5.605,
-      "eval_samples_per_second": 1784.108,
-      "eval_steps_per_second": 7.136,
+      "eval_accuracy": 0.7586,
+      "eval_loss": 0.8046875,
+      "eval_runtime": 5.5271,
+      "eval_samples_per_second": 1809.262,
+      "eval_steps_per_second": 7.237,
       "step": 17316
     },
     {
       "epoch": 52.01,
-      "grad_norm": 7.366806507110596,
+      "grad_norm": 6.0169758796691895,
       "learning_rate": 8.266266266266267e-06,
-      "loss": 1.5201,
+      "loss": 1.5024,
       "step": 17320
     },
     {
       "epoch": 52.04,
-      "grad_norm": 5.67317008972168,
+      "grad_norm": 6.408026695251465,
       "learning_rate": 8.265265265265265e-06,
-      "loss": 1.3591,
+      "loss": 1.3531,
       "step": 17330
     },
     {
       "epoch": 52.07,
-      "grad_norm": 7.651703357696533,
+      "grad_norm": 8.124414443969727,
       "learning_rate": 8.264264264264264e-06,
-      "loss": 1.489,
+      "loss": 1.4861,
       "step": 17340
     },
     {
       "epoch": 52.1,
-      "grad_norm": 6.861355304718018,
+      "grad_norm": 9.673243522644043,
       "learning_rate": 8.263263263263264e-06,
-      "loss": 1.3767,
+      "loss": 1.361,
       "step": 17350
     },
     {
       "epoch": 52.13,
-      "grad_norm": 6.706053256988525,
+      "grad_norm": 7.435117244720459,
       "learning_rate": 8.262262262262263e-06,
-      "loss": 1.3053,
+      "loss": 1.304,
       "step": 17360
     },
     {
       "epoch": 52.16,
-      "grad_norm": 8.837173461914062,
+      "grad_norm": 10.491238594055176,
       "learning_rate": 8.261261261261261e-06,
-      "loss": 1.3758,
+      "loss": 1.3838,
       "step": 17370
     },
     {
       "epoch": 52.19,
-      "grad_norm": 6.782474994659424,
+      "grad_norm": 9.130025863647461,
       "learning_rate": 8.260260260260262e-06,
-      "loss": 1.4096,
+      "loss": 1.414,
       "step": 17380
     },
     {
       "epoch": 52.22,
-      "grad_norm": 9.701000213623047,
+      "grad_norm": 8.240397453308105,
       "learning_rate": 8.25925925925926e-06,
-      "loss": 1.4288,
+      "loss": 1.4296,
       "step": 17390
     },
     {
       "epoch": 52.25,
-      "grad_norm": 7.366968631744385,
+      "grad_norm": 6.4946818351745605,
       "learning_rate": 8.258258258258259e-06,
-      "loss": 1.3578,
+      "loss": 1.3646,
       "step": 17400
     },
     {
       "epoch": 52.28,
-      "grad_norm": 6.258138179779053,
+      "grad_norm": 6.605454921722412,
       "learning_rate": 8.257257257257259e-06,
-      "loss": 1.3977,
+      "loss": 1.3948,
       "step": 17410
     },
     {
       "epoch": 52.31,
-      "grad_norm": 6.749862194061279,
+      "grad_norm": 8.687013626098633,
       "learning_rate": 8.256256256256256e-06,
-      "loss": 1.4193,
+      "loss": 1.3976,
       "step": 17420
     },
     {
       "epoch": 52.34,
-      "grad_norm": 6.9533867835998535,
+      "grad_norm": 8.383679389953613,
       "learning_rate": 8.255255255255256e-06,
-      "loss": 1.4253,
+      "loss": 1.4334,
       "step": 17430
     },
     {
       "epoch": 52.37,
-      "grad_norm": 8.452150344848633,
+      "grad_norm": 8.777721405029297,
       "learning_rate": 8.254254254254255e-06,
-      "loss": 1.3629,
+      "loss": 1.3669,
       "step": 17440
     },
     {
       "epoch": 52.4,
-      "grad_norm": 6.811629295349121,
+      "grad_norm": 9.587323188781738,
       "learning_rate": 8.253253253253254e-06,
-      "loss": 1.3846,
+      "loss": 1.3883,
       "step": 17450
     },
     {
       "epoch": 52.43,
-      "grad_norm": 8.309944152832031,
+      "grad_norm": 11.444485664367676,
       "learning_rate": 8.252252252252254e-06,
-      "loss": 1.3761,
+      "loss": 1.3695,
       "step": 17460
     },
     {
       "epoch": 52.46,
-      "grad_norm": 8.757145881652832,
+      "grad_norm": 8.051036834716797,
       "learning_rate": 8.251251251251252e-06,
-      "loss": 1.3041,
+      "loss": 1.31,
       "step": 17470
     },
     {
       "epoch": 52.49,
-      "grad_norm": 5.942065715789795,
+      "grad_norm": 7.691970348358154,
       "learning_rate": 8.250250250250251e-06,
-      "loss": 1.3707,
+      "loss": 1.3632,
       "step": 17480
     },
     {
       "epoch": 52.52,
-      "grad_norm": 6.096419811248779,
+      "grad_norm": 6.890562534332275,
       "learning_rate": 8.24924924924925e-06,
-      "loss": 1.3704,
+      "loss": 1.3625,
       "step": 17490
     },
     {
       "epoch": 52.55,
-      "grad_norm": 7.307023525238037,
+      "grad_norm": 6.858983993530273,
       "learning_rate": 8.248248248248248e-06,
-      "loss": 1.2905,
+      "loss": 1.2986,
       "step": 17500
     },
     {
       "epoch": 52.58,
-      "grad_norm": 6.923154830932617,
+      "grad_norm": 9.000014305114746,
       "learning_rate": 8.247247247247247e-06,
-      "loss": 1.4005,
+      "loss": 1.4012,
       "step": 17510
     },
     {
       "epoch": 52.61,
-      "grad_norm": 8.013869285583496,
+      "grad_norm": 8.014248847961426,
       "learning_rate": 8.246246246246247e-06,
-      "loss": 1.3451,
+      "loss": 1.3397,
       "step": 17520
     },
     {
       "epoch": 52.64,
-      "grad_norm": 5.5786662101745605,
+      "grad_norm": 6.922760963439941,
       "learning_rate": 8.245245245245246e-06,
-      "loss": 1.3899,
+      "loss": 1.3776,
       "step": 17530
     },
     {
       "epoch": 52.67,
-      "grad_norm": 7.58363676071167,
+      "grad_norm": 7.533016204833984,
       "learning_rate": 8.244244244244244e-06,
-      "loss": 1.3219,
+      "loss": 1.3159,
       "step": 17540
     },
     {
       "epoch": 52.7,
-      "grad_norm": 7.912783145904541,
+      "grad_norm": 9.856466293334961,
       "learning_rate": 8.243243243243245e-06,
-      "loss": 1.4047,
+      "loss": 1.4072,
       "step": 17550
     },
     {
       "epoch": 52.73,
-      "grad_norm": 8.99696159362793,
+      "grad_norm": 7.799111366271973,
       "learning_rate": 8.242242242242243e-06,
-      "loss": 1.4437,
+      "loss": 1.4416,
       "step": 17560
     },
     {
       "epoch": 52.76,
-      "grad_norm": 7.026357173919678,
+      "grad_norm": 7.323117256164551,
       "learning_rate": 8.241241241241242e-06,
-      "loss": 1.3339,
+      "loss": 1.3421,
       "step": 17570
     },
     {
       "epoch": 52.79,
-      "grad_norm": 6.763877868652344,
+      "grad_norm": 7.600664138793945,
       "learning_rate": 8.24024024024024e-06,
-      "loss": 1.3552,
+      "loss": 1.3497,
       "step": 17580
     },
     {
       "epoch": 52.82,
-      "grad_norm": 7.2467522621154785,
+      "grad_norm": 12.042265892028809,
       "learning_rate": 8.239239239239239e-06,
-      "loss": 1.3901,
+      "loss": 1.3996,
       "step": 17590
     },
     {
       "epoch": 52.85,
-      "grad_norm": 7.020692825317383,
+      "grad_norm": 9.86241340637207,
       "learning_rate": 8.23823823823824e-06,
-      "loss": 1.4054,
+      "loss": 1.3992,
       "step": 17600
     },
     {
       "epoch": 52.88,
-      "grad_norm": 6.873943328857422,
+      "grad_norm": 7.144519329071045,
       "learning_rate": 8.237237237237238e-06,
-      "loss": 1.3424,
+      "loss": 1.3407,
       "step": 17610
     },
     {
       "epoch": 52.91,
-      "grad_norm": 6.853343486785889,
+      "grad_norm": 7.687816143035889,
       "learning_rate": 8.236236236236236e-06,
-      "loss": 1.4079,
+      "loss": 1.414,
       "step": 17620
     },
     {
       "epoch": 52.94,
-      "grad_norm": 7.5165252685546875,
+      "grad_norm": 8.547504425048828,
       "learning_rate": 8.235235235235237e-06,
-      "loss": 1.3711,
+      "loss": 1.3772,
       "step": 17630
     },
     {
       "epoch": 52.97,
-      "grad_norm": 9.56643295288086,
+      "grad_norm": 9.070056915283203,
       "learning_rate": 8.234234234234235e-06,
-      "loss": 1.4463,
+      "loss": 1.435,
       "step": 17640
     },
     {
       "epoch": 53.0,
-      "eval_accuracy": 0.7588,
-      "eval_loss": 0.7986384630203247,
-      "eval_runtime": 5.5305,
-      "eval_samples_per_second": 1808.149,
-      "eval_steps_per_second": 7.233,
+      "eval_accuracy": 0.7579,
+      "eval_loss": 0.7961218953132629,
+      "eval_runtime": 5.3584,
+      "eval_samples_per_second": 1866.23,
+      "eval_steps_per_second": 7.465,
       "step": 17649
     },
     {
       "epoch": 53.0,
-      "grad_norm": 8.971491813659668,
+      "grad_norm": 6.878986358642578,
       "learning_rate": 8.233233233233234e-06,
-      "loss": 1.6634,
+      "loss": 1.6386,
       "step": 17650
     },
     {
       "epoch": 53.03,
-      "grad_norm": 6.6815690994262695,
+      "grad_norm": 5.8737711906433105,
       "learning_rate": 8.232232232232234e-06,
-      "loss": 1.3902,
+      "loss": 1.391,
       "step": 17660
     },
     {
       "epoch": 53.06,
-      "grad_norm": 8.435094833374023,
+      "grad_norm": 7.608234882354736,
       "learning_rate": 8.231231231231231e-06,
-      "loss": 1.3873,
+      "loss": 1.385,
       "step": 17670
     },
     {
       "epoch": 53.09,
-      "grad_norm": 7.934395790100098,
+      "grad_norm": 8.429808616638184,
       "learning_rate": 8.230230230230231e-06,
-      "loss": 1.3874,
+      "loss": 1.3907,
       "step": 17680
     },
     {
       "epoch": 53.12,
-      "grad_norm": 7.523353099822998,
+      "grad_norm": 9.321992874145508,
       "learning_rate": 8.22922922922923e-06,
-      "loss": 1.3148,
+      "loss": 1.3138,
       "step": 17690
     },
     {
       "epoch": 53.15,
-      "grad_norm": 5.527626037597656,
+      "grad_norm": 7.680136680603027,
       "learning_rate": 8.228228228228229e-06,
-      "loss": 1.426,
+      "loss": 1.415,
       "step": 17700
     },
     {
       "epoch": 53.18,
-      "grad_norm": 8.343670845031738,
+      "grad_norm": 9.681388854980469,
       "learning_rate": 8.227227227227229e-06,
-      "loss": 1.3853,
+      "loss": 1.3862,
       "step": 17710
     },
     {
       "epoch": 53.21,
-      "grad_norm": 6.073465347290039,
+      "grad_norm": 7.935455322265625,
       "learning_rate": 8.226226226226227e-06,
-      "loss": 1.3377,
+      "loss": 1.3351,
       "step": 17720
     },
     {
       "epoch": 53.24,
-      "grad_norm": 9.479571342468262,
+      "grad_norm": 10.934435844421387,
       "learning_rate": 8.225225225225226e-06,
-      "loss": 1.3856,
+      "loss": 1.3933,
       "step": 17730
     },
     {
       "epoch": 53.27,
-      "grad_norm": 12.228541374206543,
+      "grad_norm": 9.080320358276367,
       "learning_rate": 8.224224224224225e-06,
-      "loss": 1.382,
+      "loss": 1.3855,
       "step": 17740
     },
     {
       "epoch": 53.3,
-      "grad_norm": 11.362178802490234,
+      "grad_norm": 8.039287567138672,
       "learning_rate": 8.223223223223223e-06,
-      "loss": 1.3507,
+      "loss": 1.3458,
       "step": 17750
     },
     {
       "epoch": 53.33,
-      "grad_norm": 6.749588966369629,
+      "grad_norm": 9.449817657470703,
       "learning_rate": 8.222222222222222e-06,
-      "loss": 1.3516,
+      "loss": 1.3446,
       "step": 17760
     },
     {
       "epoch": 53.36,
-      "grad_norm": 5.363545894622803,
+      "grad_norm": 8.244417190551758,
       "learning_rate": 8.221221221221222e-06,
-      "loss": 1.3917,
+      "loss": 1.3941,
       "step": 17770
     },
     {
       "epoch": 53.39,
-      "grad_norm": 4.655581474304199,
+      "grad_norm": 8.668464660644531,
       "learning_rate": 8.22022022022022e-06,
-      "loss": 1.3452,
+      "loss": 1.3421,
       "step": 17780
     },
     {
       "epoch": 53.42,
-      "grad_norm": 9.101072311401367,
+      "grad_norm": 7.810549259185791,
       "learning_rate": 8.21921921921922e-06,
-      "loss": 1.3743,
+      "loss": 1.3784,
       "step": 17790
     },
     {
       "epoch": 53.45,
-      "grad_norm": 8.266436576843262,
+      "grad_norm": 8.652239799499512,
       "learning_rate": 8.21821821821822e-06,
-      "loss": 1.4155,
+      "loss": 1.4145,
       "step": 17800
     },
     {
       "epoch": 53.48,
-      "grad_norm": 10.146448135375977,
+      "grad_norm": 8.457658767700195,
       "learning_rate": 8.217217217217218e-06,
-      "loss": 1.4393,
+      "loss": 1.4166,
       "step": 17810
     },
     {
       "epoch": 53.51,
-      "grad_norm": 5.030893325805664,
+      "grad_norm": 5.916123867034912,
       "learning_rate": 8.216216216216217e-06,
-      "loss": 1.3365,
+      "loss": 1.3214,
       "step": 17820
     },
     {
       "epoch": 53.54,
-      "grad_norm": 7.185356140136719,
+      "grad_norm": 10.605713844299316,
       "learning_rate": 8.215215215215215e-06,
-      "loss": 1.3755,
+      "loss": 1.3706,
       "step": 17830
     },
     {
       "epoch": 53.57,
-      "grad_norm": 11.361291885375977,
+      "grad_norm": 9.449689865112305,
       "learning_rate": 8.214214214214214e-06,
-      "loss": 1.3975,
+      "loss": 1.3979,
       "step": 17840
     },
     {
       "epoch": 53.6,
-      "grad_norm": 7.2053422927856445,
+      "grad_norm": 12.031815528869629,
       "learning_rate": 8.213213213213214e-06,
-      "loss": 1.3577,
+      "loss": 1.3472,
       "step": 17850
     },
     {
       "epoch": 53.63,
-      "grad_norm": 8.709903717041016,
+      "grad_norm": 8.144513130187988,
       "learning_rate": 8.212212212212213e-06,
-      "loss": 1.3917,
+      "loss": 1.3809,
       "step": 17860
     },
     {
       "epoch": 53.66,
-      "grad_norm": 7.095339775085449,
+      "grad_norm": 5.552448749542236,
       "learning_rate": 8.211211211211211e-06,
-      "loss": 1.3757,
+      "loss": 1.3768,
       "step": 17870
     },
     {
       "epoch": 53.69,
-      "grad_norm": 8.816350936889648,
+      "grad_norm": 10.407093048095703,
       "learning_rate": 8.210210210210212e-06,
-      "loss": 1.3297,
+      "loss": 1.3262,
       "step": 17880
     },
     {
       "epoch": 53.72,
-      "grad_norm": 7.35960578918457,
+      "grad_norm": 6.270012855529785,
       "learning_rate": 8.20920920920921e-06,
-      "loss": 1.3314,
+      "loss": 1.3378,
       "step": 17890
     },
     {
       "epoch": 53.75,
-      "grad_norm": 8.045615196228027,
+      "grad_norm": 7.146668434143066,
       "learning_rate": 8.208208208208209e-06,
-      "loss": 1.3173,
+      "loss": 1.3158,
       "step": 17900
     },
     {
       "epoch": 53.78,
-      "grad_norm": 5.344250202178955,
+      "grad_norm": 6.238638401031494,
       "learning_rate": 8.20720720720721e-06,
-      "loss": 1.3152,
+      "loss": 1.3023,
       "step": 17910
     },
     {
       "epoch": 53.81,
-      "grad_norm": 6.8332414627075195,
+      "grad_norm": 7.412351131439209,
       "learning_rate": 8.206206206206206e-06,
-      "loss": 1.3194,
+      "loss": 1.3241,
       "step": 17920
     },
     {
       "epoch": 53.84,
-      "grad_norm": 5.906519889831543,
+      "grad_norm": 6.229217052459717,
       "learning_rate": 8.205205205205206e-06,
-      "loss": 1.3372,
+      "loss": 1.3397,
       "step": 17930
     },
     {
       "epoch": 53.87,
-      "grad_norm": 7.341153144836426,
+      "grad_norm": 10.273323059082031,
       "learning_rate": 8.204204204204205e-06,
-      "loss": 1.3527,
+      "loss": 1.3503,
       "step": 17940
     },
     {
       "epoch": 53.9,
-      "grad_norm": 8.768898963928223,
+      "grad_norm": 7.76155948638916,
       "learning_rate": 8.203203203203204e-06,
-      "loss": 1.4312,
+      "loss": 1.4304,
       "step": 17950
     },
     {
       "epoch": 53.93,
-      "grad_norm": 6.756706714630127,
+      "grad_norm": 10.559623718261719,
       "learning_rate": 8.202202202202202e-06,
-      "loss": 1.3677,
+      "loss": 1.3549,
       "step": 17960
     },
     {
       "epoch": 53.96,
-      "grad_norm": 8.818998336791992,
+      "grad_norm": 6.757030487060547,
       "learning_rate": 8.201201201201202e-06,
-      "loss": 1.4064,
+      "loss": 1.397,
       "step": 17970
     },
     {
       "epoch": 53.99,
-      "grad_norm": 6.39149808883667,
+      "grad_norm": 7.990081310272217,
       "learning_rate": 8.200200200200201e-06,
-      "loss": 1.3902,
+      "loss": 1.3824,
       "step": 17980
     },
     {
       "epoch": 54.0,
-      "eval_accuracy": 0.7629,
-      "eval_loss": 0.7865143418312073,
-      "eval_runtime": 5.4873,
-      "eval_samples_per_second": 1822.392,
-      "eval_steps_per_second": 7.29,
+      "eval_accuracy": 0.763,
+      "eval_loss": 0.7837576270103455,
+      "eval_runtime": 5.3023,
+      "eval_samples_per_second": 1885.959,
+      "eval_steps_per_second": 7.544,
       "step": 17982
     },
     {
       "epoch": 54.02,
-      "grad_norm": 4.947097301483154,
+      "grad_norm": 6.318495273590088,
       "learning_rate": 8.1991991991992e-06,
-      "loss": 1.3964,
+      "loss": 1.3722,
       "step": 17990
     },
     {
       "epoch": 54.05,
-      "grad_norm": 8.069879531860352,
+      "grad_norm": 10.135721206665039,
       "learning_rate": 8.198198198198198e-06,
-      "loss": 1.4019,
+      "loss": 1.3941,
       "step": 18000
     },
     {
       "epoch": 54.08,
-      "grad_norm": 11.314674377441406,
+      "grad_norm": 9.823216438293457,
       "learning_rate": 8.197197197197197e-06,
-      "loss": 1.3713,
+      "loss": 1.3715,
       "step": 18010
     },
     {
       "epoch": 54.11,
-      "grad_norm": 5.794673442840576,
+      "grad_norm": 7.642346382141113,
       "learning_rate": 8.196196196196197e-06,
-      "loss": 1.321,
+      "loss": 1.3164,
       "step": 18020
     },
     {
       "epoch": 54.14,
-      "grad_norm": 6.398258209228516,
+      "grad_norm": 7.514116287231445,
       "learning_rate": 8.195195195195196e-06,
-      "loss": 1.3141,
+      "loss": 1.3114,
       "step": 18030
     },
     {
       "epoch": 54.17,
-      "grad_norm": 5.902235507965088,
+      "grad_norm": 7.510837078094482,
       "learning_rate": 8.194194194194194e-06,
-      "loss": 1.2696,
+      "loss": 1.2761,
       "step": 18040
     },
     {
       "epoch": 54.2,
-      "grad_norm": 8.384127616882324,
+      "grad_norm": 8.789010047912598,
       "learning_rate": 8.193193193193195e-06,
-      "loss": 1.3857,
+      "loss": 1.3973,
       "step": 18050
     },
     {
       "epoch": 54.23,
-      "grad_norm": 8.779415130615234,
+      "grad_norm": 8.231956481933594,
       "learning_rate": 8.192192192192193e-06,
-      "loss": 1.411,
+      "loss": 1.4091,
       "step": 18060
     },
     {
       "epoch": 54.26,
-      "grad_norm": 8.73770809173584,
+      "grad_norm": 7.80242919921875,
       "learning_rate": 8.191191191191192e-06,
-      "loss": 1.2853,
+      "loss": 1.2991,
       "step": 18070
     },
     {
       "epoch": 54.29,
-      "grad_norm": 6.569752216339111,
+      "grad_norm": 8.010656356811523,
       "learning_rate": 8.19019019019019e-06,
-      "loss": 1.4387,
+      "loss": 1.4398,
       "step": 18080
     },
     {
       "epoch": 54.32,
-      "grad_norm": 6.642237186431885,
+      "grad_norm": 7.22904634475708,
       "learning_rate": 8.189189189189189e-06,
-      "loss": 1.3123,
+      "loss": 1.3113,
       "step": 18090
     },
     {
       "epoch": 54.35,
-      "grad_norm": 8.839048385620117,
+      "grad_norm": 9.058724403381348,
       "learning_rate": 8.18818818818819e-06,
-      "loss": 1.3838,
+      "loss": 1.3773,
       "step": 18100
     },
     {
       "epoch": 54.38,
-      "grad_norm": 9.56478500366211,
+      "grad_norm": 9.024547576904297,
       "learning_rate": 8.187187187187188e-06,
-      "loss": 1.3632,
+      "loss": 1.3555,
       "step": 18110
     },
     {
       "epoch": 54.41,
-      "grad_norm": 6.907779693603516,
+      "grad_norm": 10.31128978729248,
       "learning_rate": 8.186186186186186e-06,
-      "loss": 1.4148,
+      "loss": 1.4101,
       "step": 18120
     },
     {
       "epoch": 54.44,
-      "grad_norm": 7.38045072555542,
+      "grad_norm": 8.371521949768066,
       "learning_rate": 8.185185185185187e-06,
-      "loss": 1.3522,
+      "loss": 1.3434,
       "step": 18130
     },
     {
       "epoch": 54.47,
-      "grad_norm": 5.62691068649292,
+      "grad_norm": 6.718246936798096,
       "learning_rate": 8.184184184184185e-06,
-      "loss": 1.3604,
+      "loss": 1.3729,
       "step": 18140
     },
     {
       "epoch": 54.5,
-      "grad_norm": 10.508934020996094,
+      "grad_norm": 10.39156436920166,
       "learning_rate": 8.183183183183184e-06,
-      "loss": 1.3406,
+      "loss": 1.3454,
       "step": 18150
     },
     {
       "epoch": 54.53,
-      "grad_norm": 7.151280403137207,
+      "grad_norm": 5.745922565460205,
       "learning_rate": 8.182182182182183e-06,
-      "loss": 1.3341,
+      "loss": 1.3345,
       "step": 18160
     },
     {
       "epoch": 54.56,
-      "grad_norm": 5.701084613800049,
+      "grad_norm": 8.536002159118652,
       "learning_rate": 8.181181181181181e-06,
-      "loss": 1.3377,
+      "loss": 1.3314,
       "step": 18170
     },
     {
       "epoch": 54.59,
-      "grad_norm": 7.702751636505127,
+      "grad_norm": 8.092218399047852,
       "learning_rate": 8.18018018018018e-06,
-      "loss": 1.3781,
+      "loss": 1.3679,
       "step": 18180
     },
     {
       "epoch": 54.62,
-      "grad_norm": 8.199149131774902,
+      "grad_norm": 11.425930976867676,
       "learning_rate": 8.17917917917918e-06,
-      "loss": 1.3668,
+      "loss": 1.3566,
       "step": 18190
     },
     {
       "epoch": 54.65,
-      "grad_norm": 6.1144208908081055,
+      "grad_norm": 5.6689534187316895,
       "learning_rate": 8.178178178178179e-06,
-      "loss": 1.3343,
+      "loss": 1.3273,
       "step": 18200
     },
     {
       "epoch": 54.68,
-      "grad_norm": 9.707942008972168,
+      "grad_norm": 6.624577522277832,
       "learning_rate": 8.177177177177177e-06,
-      "loss": 1.4281,
+      "loss": 1.4213,
       "step": 18210
     },
     {
       "epoch": 54.71,
-      "grad_norm": 9.081879615783691,
+      "grad_norm": 13.613478660583496,
       "learning_rate": 8.176176176176177e-06,
-      "loss": 1.3161,
+      "loss": 1.3055,
       "step": 18220
     },
     {
       "epoch": 54.74,
-      "grad_norm": 10.623394966125488,
+      "grad_norm": 8.919473648071289,
       "learning_rate": 8.175175175175176e-06,
-      "loss": 1.3801,
+      "loss": 1.3814,
       "step": 18230
     },
     {
       "epoch": 54.77,
-      "grad_norm": 6.398878574371338,
+      "grad_norm": 8.353025436401367,
       "learning_rate": 8.174174174174175e-06,
-      "loss": 1.3705,
+      "loss": 1.3732,
       "step": 18240
     },
     {
       "epoch": 54.8,
-      "grad_norm": 6.405357360839844,
+      "grad_norm": 8.129802703857422,
       "learning_rate": 8.173173173173173e-06,
-      "loss": 1.3943,
+      "loss": 1.3933,
       "step": 18250
     },
     {
       "epoch": 54.83,
-      "grad_norm": 6.575839042663574,
+      "grad_norm": 6.227410316467285,
       "learning_rate": 8.172172172172172e-06,
-      "loss": 1.309,
+      "loss": 1.3036,
       "step": 18260
     },
     {
       "epoch": 54.86,
-      "grad_norm": 7.285609245300293,
+      "grad_norm": 6.846441268920898,
       "learning_rate": 8.171171171171172e-06,
-      "loss": 1.3151,
+      "loss": 1.3098,
       "step": 18270
     },
     {
       "epoch": 54.89,
-      "grad_norm": 7.8779296875,
+      "grad_norm": 7.591245651245117,
       "learning_rate": 8.17017017017017e-06,
-      "loss": 1.4072,
+      "loss": 1.4091,
       "step": 18280
     },
     {
       "epoch": 54.92,
-      "grad_norm": 6.937276363372803,
+      "grad_norm": 6.8591156005859375,
       "learning_rate": 8.16916916916917e-06,
-      "loss": 1.3636,
+      "loss": 1.3614,
       "step": 18290
     },
     {
       "epoch": 54.95,
-      "grad_norm": 7.574227333068848,
+      "grad_norm": 7.289298057556152,
       "learning_rate": 8.16816816816817e-06,
-      "loss": 1.3818,
+      "loss": 1.3825,
       "step": 18300
     },
     {
       "epoch": 54.98,
-      "grad_norm": 6.067190170288086,
+      "grad_norm": 6.791406154632568,
       "learning_rate": 8.167167167167168e-06,
-      "loss": 1.3382,
+      "loss": 1.3235,
       "step": 18310
     },
     {
       "epoch": 55.0,
-      "eval_accuracy": 0.7634,
-      "eval_loss": 0.7810105085372925,
-      "eval_runtime": 5.1606,
-      "eval_samples_per_second": 1937.746,
-      "eval_steps_per_second": 7.751,
+      "eval_accuracy": 0.7614,
+      "eval_loss": 0.7788791060447693,
+      "eval_runtime": 5.1858,
+      "eval_samples_per_second": 1928.36,
+      "eval_steps_per_second": 7.713,
       "step": 18315
     },
     {
       "epoch": 55.02,
-      "grad_norm": 6.906949520111084,
+      "grad_norm": 9.083355903625488,
       "learning_rate": 8.166166166166167e-06,
-      "loss": 1.5394,
+      "loss": 1.5562,
       "step": 18320
     },
     {
       "epoch": 55.05,
-      "grad_norm": 6.258444786071777,
+      "grad_norm": 7.735651969909668,
       "learning_rate": 8.165165165165165e-06,
-      "loss": 1.3847,
+      "loss": 1.3747,
       "step": 18330
     },
     {
       "epoch": 55.08,
-      "grad_norm": 9.339607238769531,
+      "grad_norm": 8.852681159973145,
       "learning_rate": 8.164164164164164e-06,
-      "loss": 1.401,
+      "loss": 1.3957,
       "step": 18340
     },
     {
       "epoch": 55.11,
-      "grad_norm": 7.615107536315918,
+      "grad_norm": 8.885369300842285,
       "learning_rate": 8.163163163163164e-06,
-      "loss": 1.3861,
+      "loss": 1.3849,
       "step": 18350
     },
     {
       "epoch": 55.14,
-      "grad_norm": 7.799645900726318,
+      "grad_norm": 9.84023380279541,
       "learning_rate": 8.162162162162163e-06,
-      "loss": 1.3299,
+      "loss": 1.3307,
       "step": 18360
     },
     {
       "epoch": 55.17,
-      "grad_norm": 6.153508186340332,
+      "grad_norm": 7.245664596557617,
       "learning_rate": 8.161161161161161e-06,
-      "loss": 1.3638,
+      "loss": 1.356,
       "step": 18370
     },
     {
       "epoch": 55.2,
-      "grad_norm": 5.996456623077393,
+      "grad_norm": 6.581131935119629,
       "learning_rate": 8.160160160160162e-06,
-      "loss": 1.2719,
+      "loss": 1.2845,
       "step": 18380
     },
     {
       "epoch": 55.23,
-      "grad_norm": 6.4628586769104,
+      "grad_norm": 7.241028785705566,
       "learning_rate": 8.15915915915916e-06,
-      "loss": 1.3775,
+      "loss": 1.3716,
       "step": 18390
     },
     {
       "epoch": 55.26,
-      "grad_norm": 7.092283248901367,
+      "grad_norm": 7.8357439041137695,
       "learning_rate": 8.158158158158159e-06,
-      "loss": 1.37,
+      "loss": 1.3755,
       "step": 18400
     },
     {
       "epoch": 55.29,
-      "grad_norm": 6.886444091796875,
+      "grad_norm": 8.899696350097656,
       "learning_rate": 8.157157157157158e-06,
-      "loss": 1.3778,
+      "loss": 1.3771,
       "step": 18410
     },
     {
       "epoch": 55.32,
-      "grad_norm": 7.703527450561523,
+      "grad_norm": 9.24179458618164,
       "learning_rate": 8.156156156156156e-06,
-      "loss": 1.3986,
+      "loss": 1.4091,
       "step": 18420
     },
     {
       "epoch": 55.35,
-      "grad_norm": 7.374081611633301,
+      "grad_norm": 6.163597583770752,
       "learning_rate": 8.155155155155155e-06,
-      "loss": 1.3309,
+      "loss": 1.3277,
       "step": 18430
     },
     {
       "epoch": 55.38,
-      "grad_norm": 6.13704776763916,
+      "grad_norm": 9.742182731628418,
       "learning_rate": 8.154154154154155e-06,
-      "loss": 1.4587,
+      "loss": 1.4463,
       "step": 18440
     },
     {
       "epoch": 55.41,
-      "grad_norm": 6.6250081062316895,
+      "grad_norm": 6.889835834503174,
       "learning_rate": 8.153153153153154e-06,
-      "loss": 1.3959,
+      "loss": 1.3838,
       "step": 18450
     },
     {
       "epoch": 55.44,
-      "grad_norm": 7.814140796661377,
+      "grad_norm": 8.924225807189941,
       "learning_rate": 8.152152152152152e-06,
-      "loss": 1.3402,
+      "loss": 1.3414,
       "step": 18460
     },
     {
       "epoch": 55.47,
-      "grad_norm": 6.4094624519348145,
+      "grad_norm": 7.7923665046691895,
       "learning_rate": 8.151151151151153e-06,
-      "loss": 1.2705,
+      "loss": 1.2657,
       "step": 18470
     },
     {
       "epoch": 55.5,
-      "grad_norm": 7.310944557189941,
+      "grad_norm": 6.311953544616699,
       "learning_rate": 8.150150150150151e-06,
-      "loss": 1.3408,
+      "loss": 1.3442,
       "step": 18480
     },
     {
       "epoch": 55.53,
-      "grad_norm": 7.011547088623047,
+      "grad_norm": 6.639030933380127,
       "learning_rate": 8.14914914914915e-06,
-      "loss": 1.3376,
+      "loss": 1.3423,
       "step": 18490
     },
     {
       "epoch": 55.56,
-      "grad_norm": 6.940454483032227,
+      "grad_norm": 8.569748878479004,
       "learning_rate": 8.148148148148148e-06,
-      "loss": 1.3286,
+      "loss": 1.3211,
       "step": 18500
     },
     {
       "epoch": 55.59,
-      "grad_norm": 7.6349778175354,
+      "grad_norm": 6.3366618156433105,
       "learning_rate": 8.147147147147147e-06,
-      "loss": 1.3174,
+      "loss": 1.3145,
       "step": 18510
     },
     {
       "epoch": 55.62,
-      "grad_norm": 5.832336902618408,
+      "grad_norm": 6.495047569274902,
       "learning_rate": 8.146146146146147e-06,
-      "loss": 1.289,
+      "loss": 1.2783,
       "step": 18520
     },
     {
       "epoch": 55.65,
-      "grad_norm": 10.191256523132324,
+      "grad_norm": 7.815260410308838,
       "learning_rate": 8.145145145145146e-06,
-      "loss": 1.3643,
+      "loss": 1.3589,
       "step": 18530
     },
     {
       "epoch": 55.68,
-      "grad_norm": 8.846001625061035,
+      "grad_norm": 8.474794387817383,
       "learning_rate": 8.144144144144144e-06,
-      "loss": 1.2827,
+      "loss": 1.2942,
       "step": 18540
     },
     {
       "epoch": 55.71,
-      "grad_norm": 6.929158687591553,
+      "grad_norm": 8.662571907043457,
       "learning_rate": 8.143143143143145e-06,
-      "loss": 1.3313,
+      "loss": 1.3376,
       "step": 18550
     },
     {
       "epoch": 55.74,
-      "grad_norm": 7.645487308502197,
+      "grad_norm": 7.819516658782959,
       "learning_rate": 8.142142142142143e-06,
-      "loss": 1.3721,
+      "loss": 1.3603,
       "step": 18560
     },
     {
       "epoch": 55.77,
-      "grad_norm": 8.025888442993164,
+      "grad_norm": 8.495746612548828,
       "learning_rate": 8.141141141141142e-06,
-      "loss": 1.3247,
+      "loss": 1.3236,
       "step": 18570
     },
     {
       "epoch": 55.8,
-      "grad_norm": 6.09079122543335,
+      "grad_norm": 6.466704368591309,
       "learning_rate": 8.14014014014014e-06,
-      "loss": 1.309,
+      "loss": 1.2988,
       "step": 18580
     },
     {
       "epoch": 55.83,
-      "grad_norm": 10.881485939025879,
+      "grad_norm": 7.8962202072143555,
       "learning_rate": 8.139139139139139e-06,
-      "loss": 1.3478,
+      "loss": 1.3529,
       "step": 18590
     },
     {
       "epoch": 55.86,
-      "grad_norm": 6.07639741897583,
+      "grad_norm": 10.658092498779297,
       "learning_rate": 8.13813813813814e-06,
-      "loss": 1.3365,
+      "loss": 1.3313,
       "step": 18600
     },
     {
       "epoch": 55.89,
-      "grad_norm": 7.1638994216918945,
+      "grad_norm": 9.318061828613281,
       "learning_rate": 8.137137137137138e-06,
-      "loss": 1.4181,
+      "loss": 1.4294,
       "step": 18610
     },
     {
       "epoch": 55.92,
-      "grad_norm": 8.642926216125488,
+      "grad_norm": 8.81635570526123,
       "learning_rate": 8.136136136136137e-06,
-      "loss": 1.3254,
+      "loss": 1.3303,
       "step": 18620
     },
     {
       "epoch": 55.95,
-      "grad_norm": 6.339927673339844,
+      "grad_norm": 6.078896999359131,
       "learning_rate": 8.135135135135137e-06,
-      "loss": 1.2936,
+      "loss": 1.2891,
       "step": 18630
     },
     {
       "epoch": 55.98,
-      "grad_norm": 7.997119426727295,
+      "grad_norm": 7.046920299530029,
       "learning_rate": 8.134134134134135e-06,
-      "loss": 1.3448,
+      "loss": 1.3481,
       "step": 18640
     },
     {
       "epoch": 56.0,
-      "eval_accuracy": 0.7652,
-      "eval_loss": 0.7726963758468628,
-      "eval_runtime": 5.5469,
-      "eval_samples_per_second": 1802.823,
-      "eval_steps_per_second": 7.211,
+      "eval_accuracy": 0.7658,
+      "eval_loss": 0.7723066210746765,
+      "eval_runtime": 5.4242,
+      "eval_samples_per_second": 1843.577,
+      "eval_steps_per_second": 7.374,
       "step": 18648
     },
     {
       "epoch": 56.01,
-      "grad_norm": 7.720851421356201,
+      "grad_norm": 8.278006553649902,
       "learning_rate": 8.133133133133134e-06,
-      "loss": 1.5018,
+      "loss": 1.5249,
       "step": 18650
     },
     {
       "epoch": 56.04,
-      "grad_norm": 7.296864032745361,
+      "grad_norm": 7.094216823577881,
       "learning_rate": 8.132132132132133e-06,
-      "loss": 1.3468,
+      "loss": 1.3464,
       "step": 18660
     },
     {
       "epoch": 56.07,
-      "grad_norm": 8.862027168273926,
+      "grad_norm": 7.907228469848633,
       "learning_rate": 8.131131131131131e-06,
-      "loss": 1.3979,
+      "loss": 1.3835,
       "step": 18670
     },
     {
       "epoch": 56.1,
-      "grad_norm": 6.005683898925781,
+      "grad_norm": 6.155643463134766,
       "learning_rate": 8.13013013013013e-06,
-      "loss": 1.2681,
+      "loss": 1.2603,
       "step": 18680
     },
     {
       "epoch": 56.13,
-      "grad_norm": 7.784698486328125,
+      "grad_norm": 7.705528736114502,
       "learning_rate": 8.12912912912913e-06,
-      "loss": 1.3498,
+      "loss": 1.3452,
       "step": 18690
     },
     {
       "epoch": 56.16,
-      "grad_norm": 8.858373641967773,
+      "grad_norm": 8.005882263183594,
       "learning_rate": 8.128128128128129e-06,
-      "loss": 1.3642,
+      "loss": 1.3605,
       "step": 18700
     },
     {
       "epoch": 56.19,
-      "grad_norm": 6.184598922729492,
+      "grad_norm": 8.29765796661377,
       "learning_rate": 8.127127127127127e-06,
-      "loss": 1.3222,
+      "loss": 1.3336,
       "step": 18710
     },
     {
       "epoch": 56.22,
-      "grad_norm": 7.539572715759277,
+      "grad_norm": 6.523092746734619,
       "learning_rate": 8.126126126126128e-06,
-      "loss": 1.3535,
+      "loss": 1.3455,
       "step": 18720
     },
     {
       "epoch": 56.25,
-      "grad_norm": 6.20102596282959,
+      "grad_norm": 6.295332908630371,
       "learning_rate": 8.125125125125126e-06,
-      "loss": 1.3489,
+      "loss": 1.3521,
       "step": 18730
     },
     {
       "epoch": 56.28,
-      "grad_norm": 6.979832172393799,
+      "grad_norm": 6.354363441467285,
       "learning_rate": 8.124124124124125e-06,
-      "loss": 1.3579,
+      "loss": 1.3379,
       "step": 18740
     },
     {
       "epoch": 56.31,
-      "grad_norm": 7.696981906890869,
+      "grad_norm": 5.651431083679199,
       "learning_rate": 8.123123123123123e-06,
-      "loss": 1.3416,
+      "loss": 1.3466,
       "step": 18750
     },
     {
       "epoch": 56.34,
-      "grad_norm": 5.431692123413086,
+      "grad_norm": 5.868827819824219,
       "learning_rate": 8.122122122122122e-06,
-      "loss": 1.2876,
+      "loss": 1.2785,
       "step": 18760
     },
     {
       "epoch": 56.37,
-      "grad_norm": 6.50321626663208,
+      "grad_norm": 7.759814262390137,
       "learning_rate": 8.121121121121122e-06,
-      "loss": 1.3205,
+      "loss": 1.3172,
       "step": 18770
     },
     {
       "epoch": 56.4,
-      "grad_norm": 7.381464958190918,
+      "grad_norm": 7.8524603843688965,
       "learning_rate": 8.12012012012012e-06,
-      "loss": 1.3629,
+      "loss": 1.3586,
       "step": 18780
     },
     {
       "epoch": 56.43,
-      "grad_norm": 7.814708709716797,
+      "grad_norm": 8.669193267822266,
       "learning_rate": 8.11911911911912e-06,
-      "loss": 1.3183,
+      "loss": 1.3131,
       "step": 18790
     },
     {
       "epoch": 56.46,
-      "grad_norm": 6.256265163421631,
+      "grad_norm": 7.759087085723877,
       "learning_rate": 8.11811811811812e-06,
-      "loss": 1.3179,
+      "loss": 1.3075,
       "step": 18800
     },
     {
       "epoch": 56.49,
-      "grad_norm": 6.97261381149292,
+      "grad_norm": 9.606175422668457,
       "learning_rate": 8.117117117117118e-06,
-      "loss": 1.3125,
+      "loss": 1.3243,
       "step": 18810
     },
     {
       "epoch": 56.52,
-      "grad_norm": 12.168654441833496,
+      "grad_norm": 8.537827491760254,
       "learning_rate": 8.116116116116117e-06,
-      "loss": 1.3831,
+      "loss": 1.3816,
       "step": 18820
     },
     {
       "epoch": 56.55,
-      "grad_norm": 6.955163478851318,
+      "grad_norm": 8.449309349060059,
       "learning_rate": 8.115115115115115e-06,
-      "loss": 1.3399,
+      "loss": 1.3432,
       "step": 18830
     },
     {
       "epoch": 56.58,
-      "grad_norm": 5.369374752044678,
+      "grad_norm": 7.824868202209473,
       "learning_rate": 8.114114114114114e-06,
-      "loss": 1.3376,
+      "loss": 1.3452,
       "step": 18840
     },
     {
       "epoch": 56.61,
-      "grad_norm": 7.970696926116943,
+      "grad_norm": 7.513922214508057,
       "learning_rate": 8.113113113113114e-06,
-      "loss": 1.3224,
+      "loss": 1.3221,
       "step": 18850
     },
     {
       "epoch": 56.64,
-      "grad_norm": 6.673727512359619,
+      "grad_norm": 6.441809177398682,
       "learning_rate": 8.112112112112113e-06,
-      "loss": 1.3468,
+      "loss": 1.3457,
       "step": 18860
     },
     {
       "epoch": 56.67,
-      "grad_norm": 6.3368916511535645,
+      "grad_norm": 6.55618953704834,
       "learning_rate": 8.111111111111112e-06,
-      "loss": 1.4013,
+      "loss": 1.3983,
       "step": 18870
     },
     {
       "epoch": 56.7,
-      "grad_norm": 7.489491939544678,
+      "grad_norm": 7.2881317138671875,
       "learning_rate": 8.11011011011011e-06,
-      "loss": 1.316,
+      "loss": 1.3078,
       "step": 18880
     },
     {
       "epoch": 56.73,
-      "grad_norm": 8.890159606933594,
+      "grad_norm": 7.362827301025391,
       "learning_rate": 8.10910910910911e-06,
-      "loss": 1.3792,
+      "loss": 1.3768,
       "step": 18890
     },
     {
       "epoch": 56.76,
-      "grad_norm": 5.922383785247803,
+      "grad_norm": 8.055853843688965,
       "learning_rate": 8.108108108108109e-06,
-      "loss": 1.3209,
+      "loss": 1.3146,
       "step": 18900
     },
     {
       "epoch": 56.79,
-      "grad_norm": 8.71475887298584,
+      "grad_norm": 4.884746074676514,
       "learning_rate": 8.107107107107108e-06,
-      "loss": 1.3127,
+      "loss": 1.3044,
       "step": 18910
     },
     {
       "epoch": 56.82,
-      "grad_norm": 6.1757073402404785,
+      "grad_norm": 8.874103546142578,
       "learning_rate": 8.106106106106106e-06,
-      "loss": 1.3364,
+      "loss": 1.3416,
       "step": 18920
     },
     {
       "epoch": 56.85,
-      "grad_norm": 7.054976463317871,
+      "grad_norm": 7.2240986824035645,
       "learning_rate": 8.105105105105105e-06,
-      "loss": 1.3018,
+      "loss": 1.298,
       "step": 18930
     },
     {
       "epoch": 56.88,
-      "grad_norm": 7.205659866333008,
+      "grad_norm": 6.538424968719482,
       "learning_rate": 8.104104104104105e-06,
-      "loss": 1.3489,
+      "loss": 1.3525,
       "step": 18940
     },
     {
       "epoch": 56.91,
-      "grad_norm": 9.8294095993042,
+      "grad_norm": 6.017702579498291,
       "learning_rate": 8.103103103103104e-06,
-      "loss": 1.3291,
+      "loss": 1.3229,
       "step": 18950
     },
     {
       "epoch": 56.94,
-      "grad_norm": 6.291053295135498,
+      "grad_norm": 6.727291584014893,
       "learning_rate": 8.102102102102102e-06,
-      "loss": 1.3967,
+      "loss": 1.3955,
       "step": 18960
     },
     {
       "epoch": 56.97,
-      "grad_norm": 7.177740573883057,
+      "grad_norm": 6.472428798675537,
       "learning_rate": 8.101101101101103e-06,
-      "loss": 1.3046,
+      "loss": 1.3016,
       "step": 18970
     },
     {
       "epoch": 57.0,
-      "grad_norm": 6.326959133148193,
+      "grad_norm": 6.069327354431152,
       "learning_rate": 8.100100100100101e-06,
-      "loss": 1.283,
+      "loss": 1.2645,
       "step": 18980
     },
     {
       "epoch": 57.0,
-      "eval_accuracy": 0.7664,
-      "eval_loss": 0.7681456804275513,
-      "eval_runtime": 5.5177,
-      "eval_samples_per_second": 1812.337,
-      "eval_steps_per_second": 7.249,
+      "eval_accuracy": 0.7646,
+      "eval_loss": 0.767440140247345,
+      "eval_runtime": 5.3561,
+      "eval_samples_per_second": 1867.026,
+      "eval_steps_per_second": 7.468,
       "step": 18981
     },
     {
       "epoch": 57.03,
-      "grad_norm": 7.324021816253662,
+      "grad_norm": 5.994871616363525,
       "learning_rate": 8.0990990990991e-06,
-      "loss": 1.4985,
+      "loss": 1.522,
       "step": 18990
     },
     {
       "epoch": 57.06,
-      "grad_norm": 6.775612831115723,
+      "grad_norm": 8.358731269836426,
       "learning_rate": 8.098098098098098e-06,
-      "loss": 1.3528,
+      "loss": 1.3436,
       "step": 19000
     },
     {
       "epoch": 57.09,
-      "grad_norm": 6.977473735809326,
+      "grad_norm": 10.881064414978027,
       "learning_rate": 8.097097097097097e-06,
       "loss": 1.2913,
       "step": 19010
     },
     {
       "epoch": 57.12,
-      "grad_norm": 4.752172946929932,
+      "grad_norm": 8.061651229858398,
       "learning_rate": 8.096096096096097e-06,
-      "loss": 1.3903,
+      "loss": 1.3787,
       "step": 19020
     },
     {
       "epoch": 57.15,
-      "grad_norm": 12.851792335510254,
+      "grad_norm": 8.543837547302246,
       "learning_rate": 8.095095095095096e-06,
-      "loss": 1.2756,
+      "loss": 1.276,
       "step": 19030
     },
     {
       "epoch": 57.18,
-      "grad_norm": 8.422714233398438,
+      "grad_norm": 9.101380348205566,
       "learning_rate": 8.094094094094094e-06,
-      "loss": 1.355,
+      "loss": 1.351,
       "step": 19040
     },
     {
       "epoch": 57.21,
-      "grad_norm": 4.340334415435791,
+      "grad_norm": 6.171571254730225,
       "learning_rate": 8.093093093093095e-06,
-      "loss": 1.284,
+      "loss": 1.282,
       "step": 19050
     },
     {
       "epoch": 57.24,
-      "grad_norm": 7.727368354797363,
+      "grad_norm": 10.307090759277344,
       "learning_rate": 8.092092092092093e-06,
-      "loss": 1.3565,
+      "loss": 1.352,
       "step": 19060
     },
     {
       "epoch": 57.27,
-      "grad_norm": 8.590628623962402,
+      "grad_norm": 6.5160603523254395,
       "learning_rate": 8.091091091091092e-06,
-      "loss": 1.2779,
+      "loss": 1.2688,
       "step": 19070
     },
     {
       "epoch": 57.3,
-      "grad_norm": 8.73342514038086,
+      "grad_norm": 12.671828269958496,
       "learning_rate": 8.09009009009009e-06,
-      "loss": 1.2645,
+      "loss": 1.2609,
       "step": 19080
     },
     {
       "epoch": 57.33,
-      "grad_norm": 6.139922142028809,
+      "grad_norm": 5.35656213760376,
       "learning_rate": 8.089089089089089e-06,
-      "loss": 1.3248,
+      "loss": 1.3203,
       "step": 19090
     },
     {
       "epoch": 57.36,
-      "grad_norm": 5.836198806762695,
+      "grad_norm": 6.761318683624268,
       "learning_rate": 8.088088088088088e-06,
-      "loss": 1.3492,
+      "loss": 1.3619,
       "step": 19100
     },
     {
       "epoch": 57.39,
-      "grad_norm": 8.92360782623291,
+      "grad_norm": 9.191540718078613,
       "learning_rate": 8.087087087087088e-06,
-      "loss": 1.3108,
+      "loss": 1.3021,
       "step": 19110
     },
     {
       "epoch": 57.42,
-      "grad_norm": 7.223458290100098,
+      "grad_norm": 8.931928634643555,
       "learning_rate": 8.086086086086087e-06,
-      "loss": 1.3363,
+      "loss": 1.3168,
       "step": 19120
     },
     {
       "epoch": 57.45,
-      "grad_norm": 7.151160717010498,
+      "grad_norm": 12.869669914245605,
       "learning_rate": 8.085085085085085e-06,
-      "loss": 1.3702,
+      "loss": 1.3844,
       "step": 19130
     },
     {
       "epoch": 57.48,
-      "grad_norm": 7.789914131164551,
+      "grad_norm": 9.201114654541016,
       "learning_rate": 8.084084084084085e-06,
-      "loss": 1.3354,
+      "loss": 1.318,
       "step": 19140
     },
     {
       "epoch": 57.51,
-      "grad_norm": 4.908344268798828,
+      "grad_norm": 5.446118354797363,
       "learning_rate": 8.083083083083084e-06,
-      "loss": 1.2361,
+      "loss": 1.2408,
       "step": 19150
     },
     {
       "epoch": 57.54,
-      "grad_norm": 8.43432331085205,
+      "grad_norm": 11.393574714660645,
       "learning_rate": 8.082082082082083e-06,
-      "loss": 1.3935,
+      "loss": 1.3944,
       "step": 19160
     },
     {
       "epoch": 57.57,
-      "grad_norm": 8.234566688537598,
+      "grad_norm": 8.203720092773438,
       "learning_rate": 8.081081081081081e-06,
-      "loss": 1.3125,
+      "loss": 1.3153,
       "step": 19170
     },
     {
       "epoch": 57.6,
-      "grad_norm": 6.0173659324646,
+      "grad_norm": 6.7729268074035645,
       "learning_rate": 8.08008008008008e-06,
-      "loss": 1.3581,
+      "loss": 1.3537,
       "step": 19180
     },
     {
       "epoch": 57.63,
-      "grad_norm": 7.4518656730651855,
+      "grad_norm": 7.478360176086426,
       "learning_rate": 8.07907907907908e-06,
-      "loss": 1.3774,
+      "loss": 1.3754,
       "step": 19190
     },
     {
       "epoch": 57.66,
-      "grad_norm": 5.949438095092773,
+      "grad_norm": 7.59635591506958,
       "learning_rate": 8.078078078078079e-06,
-      "loss": 1.3523,
+      "loss": 1.3539,
       "step": 19200
     },
     {
       "epoch": 57.69,
-      "grad_norm": 6.593804836273193,
+      "grad_norm": 8.404068946838379,
       "learning_rate": 8.077077077077077e-06,
-      "loss": 1.406,
+      "loss": 1.4099,
       "step": 19210
     },
     {
       "epoch": 57.72,
-      "grad_norm": 6.79438591003418,
+      "grad_norm": 6.318376064300537,
       "learning_rate": 8.076076076076078e-06,
-      "loss": 1.3723,
+      "loss": 1.3758,
       "step": 19220
     },
     {
       "epoch": 57.75,
-      "grad_norm": 7.271402359008789,
+      "grad_norm": 8.521629333496094,
       "learning_rate": 8.075075075075076e-06,
-      "loss": 1.3662,
+      "loss": 1.3667,
       "step": 19230
     },
     {
       "epoch": 57.78,
-      "grad_norm": 6.821556568145752,
+      "grad_norm": 7.291736125946045,
       "learning_rate": 8.074074074074075e-06,
-      "loss": 1.2941,
+      "loss": 1.2845,
       "step": 19240
     },
     {
       "epoch": 57.81,
-      "grad_norm": 8.509852409362793,
+      "grad_norm": 9.595659255981445,
       "learning_rate": 8.073073073073073e-06,
-      "loss": 1.3895,
+      "loss": 1.3883,
       "step": 19250
     },
     {
       "epoch": 57.84,
-      "grad_norm": 9.258196830749512,
+      "grad_norm": 10.002570152282715,
       "learning_rate": 8.072072072072072e-06,
-      "loss": 1.3084,
+      "loss": 1.2932,
       "step": 19260
     },
     {
       "epoch": 57.87,
-      "grad_norm": 6.81777286529541,
+      "grad_norm": 11.843606948852539,
       "learning_rate": 8.071071071071072e-06,
-      "loss": 1.2797,
+      "loss": 1.2811,
       "step": 19270
     },
     {
       "epoch": 57.9,
-      "grad_norm": 5.397612571716309,
+      "grad_norm": 6.448760986328125,
       "learning_rate": 8.070070070070071e-06,
       "loss": 1.3995,
       "step": 19280
     },
     {
       "epoch": 57.93,
-      "grad_norm": 7.628185272216797,
+      "grad_norm": 8.599433898925781,
       "learning_rate": 8.06906906906907e-06,
-      "loss": 1.358,
+      "loss": 1.3512,
       "step": 19290
     },
     {
       "epoch": 57.96,
-      "grad_norm": 6.4071149826049805,
+      "grad_norm": 5.656130313873291,
       "learning_rate": 8.06806806806807e-06,
-      "loss": 1.3171,
+      "loss": 1.324,
       "step": 19300
     },
     {
       "epoch": 57.99,
-      "grad_norm": 7.433269500732422,
+      "grad_norm": 8.619142532348633,
       "learning_rate": 8.067067067067068e-06,
-      "loss": 1.2979,
+      "loss": 1.2878,
       "step": 19310
     },
     {
       "epoch": 58.0,
-      "eval_accuracy": 0.7704,
-      "eval_loss": 0.763714075088501,
-      "eval_runtime": 5.2458,
-      "eval_samples_per_second": 1906.302,
-      "eval_steps_per_second": 7.625,
+      "eval_accuracy": 0.7686,
+      "eval_loss": 0.7611379623413086,
+      "eval_runtime": 5.4254,
+      "eval_samples_per_second": 1843.196,
+      "eval_steps_per_second": 7.373,
       "step": 19314
     },
     {
       "epoch": 58.02,
-      "grad_norm": 7.483592510223389,
+      "grad_norm": 7.075006484985352,
       "learning_rate": 8.066066066066067e-06,
-      "loss": 1.4666,
+      "loss": 1.4438,
       "step": 19320
     },
     {
       "epoch": 58.05,
-      "grad_norm": 7.404341697692871,
+      "grad_norm": 7.171411991119385,
       "learning_rate": 8.065065065065066e-06,
-      "loss": 1.2946,
+      "loss": 1.2926,
       "step": 19330
     },
     {
       "epoch": 58.08,
-      "grad_norm": 8.093389511108398,
+      "grad_norm": 9.094625473022461,
       "learning_rate": 8.064064064064064e-06,
-      "loss": 1.3741,
+      "loss": 1.3796,
       "step": 19340
     },
     {
       "epoch": 58.11,
-      "grad_norm": 8.196914672851562,
+      "grad_norm": 7.663345813751221,
       "learning_rate": 8.063063063063063e-06,
-      "loss": 1.3667,
+      "loss": 1.3621,
       "step": 19350
     },
     {
       "epoch": 58.14,
-      "grad_norm": 5.214630603790283,
+      "grad_norm": 6.648177146911621,
       "learning_rate": 8.062062062062063e-06,
-      "loss": 1.3098,
+      "loss": 1.3184,
       "step": 19360
     },
     {
       "epoch": 58.17,
-      "grad_norm": 8.986544609069824,
+      "grad_norm": 12.510075569152832,
       "learning_rate": 8.061061061061062e-06,
-      "loss": 1.3516,
+      "loss": 1.3452,
       "step": 19370
     },
     {
       "epoch": 58.2,
-      "grad_norm": 6.193149566650391,
+      "grad_norm": 7.129115104675293,
       "learning_rate": 8.06006006006006e-06,
-      "loss": 1.433,
+      "loss": 1.4244,
       "step": 19380
     },
     {
       "epoch": 58.23,
-      "grad_norm": 6.413369178771973,
+      "grad_norm": 6.291265487670898,
       "learning_rate": 8.05905905905906e-06,
-      "loss": 1.3752,
+      "loss": 1.3847,
       "step": 19390
     },
     {
       "epoch": 58.26,
-      "grad_norm": 7.073620319366455,
+      "grad_norm": 6.794661998748779,
       "learning_rate": 8.058058058058059e-06,
-      "loss": 1.3838,
+      "loss": 1.3877,
       "step": 19400
     },
     {
       "epoch": 58.29,
-      "grad_norm": 7.007474422454834,
+      "grad_norm": 7.117151737213135,
       "learning_rate": 8.057057057057058e-06,
-      "loss": 1.3725,
+      "loss": 1.3664,
       "step": 19410
     },
     {
       "epoch": 58.32,
-      "grad_norm": 8.348004341125488,
+      "grad_norm": 7.620694637298584,
       "learning_rate": 8.056056056056056e-06,
-      "loss": 1.2987,
+      "loss": 1.294,
       "step": 19420
     },
     {
       "epoch": 58.35,
-      "grad_norm": 7.2904887199401855,
+      "grad_norm": 5.413259029388428,
       "learning_rate": 8.055055055055055e-06,
-      "loss": 1.3293,
+      "loss": 1.3276,
       "step": 19430
     },
     {
       "epoch": 58.38,
-      "grad_norm": 6.6077656745910645,
+      "grad_norm": 8.187294006347656,
       "learning_rate": 8.054054054054055e-06,
-      "loss": 1.3317,
+      "loss": 1.3325,
       "step": 19440
     },
     {
       "epoch": 58.41,
-      "grad_norm": 5.852856159210205,
+      "grad_norm": 7.9892401695251465,
       "learning_rate": 8.053053053053054e-06,
-      "loss": 1.273,
+      "loss": 1.2738,
       "step": 19450
     },
     {
       "epoch": 58.44,
-      "grad_norm": 5.872313976287842,
+      "grad_norm": 6.118348121643066,
       "learning_rate": 8.052052052052052e-06,
-      "loss": 1.3167,
+      "loss": 1.3148,
       "step": 19460
     },
     {
       "epoch": 58.47,
-      "grad_norm": 6.656248569488525,
+      "grad_norm": 9.029107093811035,
       "learning_rate": 8.051051051051053e-06,
-      "loss": 1.2894,
+      "loss": 1.2881,
       "step": 19470
     },
     {
       "epoch": 58.5,
-      "grad_norm": 7.649667263031006,
+      "grad_norm": 8.290801048278809,
       "learning_rate": 8.050050050050051e-06,
-      "loss": 1.349,
+      "loss": 1.3337,
       "step": 19480
     },
     {
       "epoch": 58.53,
-      "grad_norm": 6.12553596496582,
+      "grad_norm": 7.805062770843506,
       "learning_rate": 8.04904904904905e-06,
-      "loss": 1.3348,
+      "loss": 1.3149,
       "step": 19490
     },
     {
       "epoch": 58.56,
-      "grad_norm": 7.334399223327637,
+      "grad_norm": 9.660841941833496,
       "learning_rate": 8.048048048048048e-06,
-      "loss": 1.3494,
+      "loss": 1.3562,
       "step": 19500
     },
     {
       "epoch": 58.59,
-      "grad_norm": 13.950333595275879,
+      "grad_norm": 9.301896095275879,
       "learning_rate": 8.047047047047047e-06,
-      "loss": 1.3174,
+      "loss": 1.3232,
       "step": 19510
     },
     {
       "epoch": 58.62,
-      "grad_norm": 9.936882019042969,
+      "grad_norm": 6.823724746704102,
       "learning_rate": 8.046046046046047e-06,
-      "loss": 1.3007,
+      "loss": 1.289,
       "step": 19520
     },
     {
       "epoch": 58.65,
-      "grad_norm": 6.340768337249756,
+      "grad_norm": 7.856751918792725,
       "learning_rate": 8.045045045045046e-06,
-      "loss": 1.3049,
+      "loss": 1.2975,
       "step": 19530
     },
     {
       "epoch": 58.68,
-      "grad_norm": 7.312600135803223,
+      "grad_norm": 7.630553722381592,
       "learning_rate": 8.044044044044045e-06,
-      "loss": 1.2965,
+      "loss": 1.3022,
       "step": 19540
     },
     {
       "epoch": 58.71,
-      "grad_norm": 6.494144916534424,
+      "grad_norm": 6.766228675842285,
       "learning_rate": 8.043043043043043e-06,
-      "loss": 1.3165,
+      "loss": 1.3157,
       "step": 19550
     },
     {
       "epoch": 58.74,
-      "grad_norm": 5.70309591293335,
+      "grad_norm": 6.327369213104248,
       "learning_rate": 8.042042042042043e-06,
-      "loss": 1.3159,
+      "loss": 1.3169,
       "step": 19560
     },
     {
       "epoch": 58.77,
-      "grad_norm": 6.781817436218262,
+      "grad_norm": 7.391269207000732,
       "learning_rate": 8.041041041041042e-06,
-      "loss": 1.3737,
+      "loss": 1.3815,
       "step": 19570
     },
     {
       "epoch": 58.8,
-      "grad_norm": 8.559160232543945,
+      "grad_norm": 6.584099769592285,
       "learning_rate": 8.04004004004004e-06,
-      "loss": 1.3062,
+      "loss": 1.333,
       "step": 19580
     },
     {
       "epoch": 58.83,
-      "grad_norm": 5.829819679260254,
+      "grad_norm": 5.971362590789795,
       "learning_rate": 8.03903903903904e-06,
-      "loss": 1.3507,
+      "loss": 1.3334,
       "step": 19590
     },
     {
       "epoch": 58.86,
-      "grad_norm": 6.325531959533691,
+      "grad_norm": 7.127225875854492,
       "learning_rate": 8.038038038038038e-06,
-      "loss": 1.3471,
+      "loss": 1.3391,
       "step": 19600
     },
     {
       "epoch": 58.89,
-      "grad_norm": 8.051213264465332,
+      "grad_norm": 6.928186893463135,
       "learning_rate": 8.037037037037038e-06,
-      "loss": 1.3327,
+      "loss": 1.3411,
       "step": 19610
     },
     {
       "epoch": 58.92,
-      "grad_norm": 5.781607151031494,
+      "grad_norm": 6.172666549682617,
       "learning_rate": 8.036036036036037e-06,
-      "loss": 1.2855,
+      "loss": 1.2897,
       "step": 19620
     },
     {
       "epoch": 58.95,
-      "grad_norm": 7.148298263549805,
+      "grad_norm": 6.599175453186035,
       "learning_rate": 8.035035035035035e-06,
-      "loss": 1.3347,
+      "loss": 1.3248,
       "step": 19630
     },
     {
       "epoch": 58.98,
-      "grad_norm": 6.538644313812256,
+      "grad_norm": 7.500275135040283,
       "learning_rate": 8.034034034034036e-06,
-      "loss": 1.3176,
+      "loss": 1.3248,
       "step": 19640
     },
     {
       "epoch": 59.0,
-      "eval_accuracy": 0.7712,
-      "eval_loss": 0.7614371180534363,
-      "eval_runtime": 5.499,
-      "eval_samples_per_second": 1818.51,
-      "eval_steps_per_second": 7.274,
+      "eval_accuracy": 0.7701,
+      "eval_loss": 0.7606554627418518,
+      "eval_runtime": 5.3142,
+      "eval_samples_per_second": 1881.733,
+      "eval_steps_per_second": 7.527,
       "step": 19647
     },
     {
       "epoch": 59.01,
-      "grad_norm": 6.491925239562988,
+      "grad_norm": 7.4956183433532715,
       "learning_rate": 8.033033033033034e-06,
-      "loss": 1.4788,
+      "loss": 1.4851,
       "step": 19650
     },
     {
       "epoch": 59.04,
-      "grad_norm": 5.318271160125732,
+      "grad_norm": 7.95383882522583,
       "learning_rate": 8.032032032032033e-06,
-      "loss": 1.2882,
+      "loss": 1.279,
       "step": 19660
     },
     {
       "epoch": 59.07,
-      "grad_norm": 8.439384460449219,
+      "grad_norm": 8.028231620788574,
       "learning_rate": 8.031031031031031e-06,
-      "loss": 1.3603,
+      "loss": 1.365,
       "step": 19670
     },
     {
       "epoch": 59.1,
-      "grad_norm": 5.660412788391113,
+      "grad_norm": 8.530021667480469,
       "learning_rate": 8.03003003003003e-06,
-      "loss": 1.3217,
+      "loss": 1.3238,
       "step": 19680
     },
     {
       "epoch": 59.13,
-      "grad_norm": 9.4834623336792,
+      "grad_norm": 7.567521572113037,
       "learning_rate": 8.02902902902903e-06,
-      "loss": 1.3011,
+      "loss": 1.2971,
       "step": 19690
     },
     {
       "epoch": 59.16,
-      "grad_norm": 6.097170352935791,
+      "grad_norm": 7.348806858062744,
       "learning_rate": 8.028028028028029e-06,
-      "loss": 1.2661,
+      "loss": 1.2594,
       "step": 19700
     },
     {
       "epoch": 59.19,
-      "grad_norm": 10.915419578552246,
+      "grad_norm": 9.809030532836914,
       "learning_rate": 8.027027027027027e-06,
-      "loss": 1.255,
+      "loss": 1.2601,
       "step": 19710
     },
     {
       "epoch": 59.22,
-      "grad_norm": 7.623536109924316,
+      "grad_norm": 9.374258995056152,
       "learning_rate": 8.026026026026028e-06,
-      "loss": 1.3372,
+      "loss": 1.343,
       "step": 19720
     },
     {
       "epoch": 59.25,
-      "grad_norm": 10.041492462158203,
+      "grad_norm": 10.97638988494873,
       "learning_rate": 8.025025025025026e-06,
-      "loss": 1.3175,
+      "loss": 1.3153,
       "step": 19730
     },
     {
       "epoch": 59.28,
-      "grad_norm": 8.099281311035156,
+      "grad_norm": 7.716129302978516,
       "learning_rate": 8.024024024024025e-06,
-      "loss": 1.2769,
+      "loss": 1.2746,
       "step": 19740
     },
     {
       "epoch": 59.31,
-      "grad_norm": 7.072831153869629,
+      "grad_norm": 9.317811012268066,
       "learning_rate": 8.023023023023023e-06,
-      "loss": 1.3615,
+      "loss": 1.3564,
       "step": 19750
     },
     {
       "epoch": 59.34,
-      "grad_norm": 7.180505752563477,
+      "grad_norm": 5.764516830444336,
       "learning_rate": 8.022022022022022e-06,
-      "loss": 1.2677,
+      "loss": 1.276,
       "step": 19760
     },
     {
       "epoch": 59.37,
-      "grad_norm": 5.295686721801758,
+      "grad_norm": 5.71919059753418,
       "learning_rate": 8.021021021021022e-06,
-      "loss": 1.3341,
+      "loss": 1.3374,
       "step": 19770
     },
     {
       "epoch": 59.4,
-      "grad_norm": 8.274558067321777,
+      "grad_norm": 6.533501625061035,
       "learning_rate": 8.020020020020021e-06,
-      "loss": 1.3716,
+      "loss": 1.3628,
       "step": 19780
     },
     {
       "epoch": 59.43,
-      "grad_norm": 6.970517158508301,
+      "grad_norm": 7.902794361114502,
       "learning_rate": 8.01901901901902e-06,
-      "loss": 1.3469,
+      "loss": 1.3506,
       "step": 19790
     },
     {
       "epoch": 59.46,
-      "grad_norm": 7.604520320892334,
+      "grad_norm": 8.325324058532715,
       "learning_rate": 8.018018018018018e-06,
-      "loss": 1.3474,
+      "loss": 1.3552,
       "step": 19800
     },
     {
       "epoch": 59.49,
-      "grad_norm": 8.736945152282715,
+      "grad_norm": 9.277702331542969,
       "learning_rate": 8.017017017017018e-06,
-      "loss": 1.3736,
+      "loss": 1.3744,
       "step": 19810
     },
     {
       "epoch": 59.52,
-      "grad_norm": 7.5100483894348145,
+      "grad_norm": 7.952363014221191,
       "learning_rate": 8.016016016016017e-06,
-      "loss": 1.338,
+      "loss": 1.3379,
       "step": 19820
     },
     {
       "epoch": 59.55,
-      "grad_norm": 8.164287567138672,
+      "grad_norm": 6.049371242523193,
       "learning_rate": 8.015015015015016e-06,
-      "loss": 1.3029,
+      "loss": 1.2997,
       "step": 19830
     },
     {
       "epoch": 59.58,
-      "grad_norm": 7.9314799308776855,
+      "grad_norm": 9.404382705688477,
       "learning_rate": 8.014014014014014e-06,
-      "loss": 1.2614,
+      "loss": 1.2667,
       "step": 19840
     },
     {
       "epoch": 59.61,
-      "grad_norm": 6.340667247772217,
+      "grad_norm": 8.090605735778809,
       "learning_rate": 8.013013013013013e-06,
-      "loss": 1.1779,
+      "loss": 1.1836,
       "step": 19850
     },
     {
       "epoch": 59.64,
-      "grad_norm": 5.737483978271484,
+      "grad_norm": 8.825246810913086,
       "learning_rate": 8.012012012012013e-06,
-      "loss": 1.2237,
+      "loss": 1.2218,
       "step": 19860
     },
     {
       "epoch": 59.67,
-      "grad_norm": 7.145668029785156,
+      "grad_norm": 7.753334045410156,
       "learning_rate": 8.011011011011012e-06,
-      "loss": 1.2576,
+      "loss": 1.2464,
       "step": 19870
     },
     {
       "epoch": 59.7,
-      "grad_norm": 11.989452362060547,
+      "grad_norm": 12.665596008300781,
       "learning_rate": 8.01001001001001e-06,
-      "loss": 1.3268,
+      "loss": 1.321,
       "step": 19880
     },
     {
       "epoch": 59.73,
-      "grad_norm": 5.8673415184021,
+      "grad_norm": 7.767548561096191,
       "learning_rate": 8.00900900900901e-06,
-      "loss": 1.3469,
+      "loss": 1.3493,
       "step": 19890
     },
     {
       "epoch": 59.76,
-      "grad_norm": 5.726264953613281,
+      "grad_norm": 6.887409687042236,
       "learning_rate": 8.00800800800801e-06,
-      "loss": 1.3911,
+      "loss": 1.3877,
       "step": 19900
     },
     {
       "epoch": 59.79,
-      "grad_norm": 9.30558967590332,
+      "grad_norm": 10.472498893737793,
       "learning_rate": 8.007007007007008e-06,
-      "loss": 1.335,
+      "loss": 1.3472,
       "step": 19910
     },
     {
       "epoch": 59.82,
-      "grad_norm": 4.739650726318359,
+      "grad_norm": 6.510967254638672,
       "learning_rate": 8.006006006006006e-06,
-      "loss": 1.3343,
+      "loss": 1.3279,
       "step": 19920
     },
     {
       "epoch": 59.85,
-      "grad_norm": 6.8755269050598145,
+      "grad_norm": 9.30593490600586,
       "learning_rate": 8.005005005005005e-06,
-      "loss": 1.3277,
+      "loss": 1.3234,
       "step": 19930
     },
     {
       "epoch": 59.88,
-      "grad_norm": 6.532683849334717,
+      "grad_norm": 7.788637638092041,
       "learning_rate": 8.004004004004005e-06,
       "loss": 1.2603,
       "step": 19940
     },
     {
       "epoch": 59.91,
-      "grad_norm": 6.29788064956665,
+      "grad_norm": 6.283743381500244,
       "learning_rate": 8.003003003003004e-06,
-      "loss": 1.3095,
+      "loss": 1.3022,
       "step": 19950
     },
     {
       "epoch": 59.94,
-      "grad_norm": 7.615232944488525,
+      "grad_norm": 6.4228997230529785,
       "learning_rate": 8.002002002002002e-06,
-      "loss": 1.2421,
+      "loss": 1.2343,
       "step": 19960
     },
     {
       "epoch": 59.97,
-      "grad_norm": 7.250647068023682,
+      "grad_norm": 9.235601425170898,
       "learning_rate": 8.001001001001003e-06,
-      "loss": 1.3393,
+      "loss": 1.3273,
       "step": 19970
     },
     {
       "epoch": 60.0,
-      "grad_norm": 35.57927703857422,
+      "grad_norm": 34.899879455566406,
       "learning_rate": 8.000000000000001e-06,
-      "loss": 1.4151,
+      "loss": 1.4243,
       "step": 19980
     },
     {
       "epoch": 60.0,
       "eval_accuracy": 0.7671,
-      "eval_loss": 0.7596631646156311,
-      "eval_runtime": 5.2986,
-      "eval_samples_per_second": 1887.283,
-      "eval_steps_per_second": 7.549,
+      "eval_loss": 0.757928729057312,
+      "eval_runtime": 5.1788,
+      "eval_samples_per_second": 1930.951,
+      "eval_steps_per_second": 7.724,
       "step": 19980
     },
     {
       "epoch": 60.03,
-      "grad_norm": 5.217360973358154,
+      "grad_norm": 10.070955276489258,
       "learning_rate": 7.998998998999e-06,
-      "loss": 1.3231,
+      "loss": 1.3179,
       "step": 19990
     },
     {
       "epoch": 60.06,
-      "grad_norm": 6.290153503417969,
+      "grad_norm": 8.259485244750977,
       "learning_rate": 7.997997997997999e-06,
-      "loss": 1.3386,
+      "loss": 1.3303,
       "step": 20000
     },
     {
       "epoch": 60.09,
-      "grad_norm": 8.492137908935547,
+      "grad_norm": 10.084449768066406,
       "learning_rate": 7.996996996996997e-06,
-      "loss": 1.3306,
+      "loss": 1.3203,
       "step": 20010
     },
     {
       "epoch": 60.12,
-      "grad_norm": 7.810229778289795,
+      "grad_norm": 8.516907691955566,
       "learning_rate": 7.995995995995996e-06,
-      "loss": 1.3326,
+      "loss": 1.3256,
       "step": 20020
     },
     {
       "epoch": 60.15,
-      "grad_norm": 6.276889324188232,
+      "grad_norm": 7.370086193084717,
       "learning_rate": 7.994994994994996e-06,
-      "loss": 1.322,
+      "loss": 1.3223,
       "step": 20030
     },
     {
       "epoch": 60.18,
-      "grad_norm": 14.628669738769531,
+      "grad_norm": 9.992877960205078,
       "learning_rate": 7.993993993993995e-06,
-      "loss": 1.3194,
+      "loss": 1.3149,
       "step": 20040
     },
     {
       "epoch": 60.21,
-      "grad_norm": 5.111385345458984,
+      "grad_norm": 6.401727199554443,
       "learning_rate": 7.992992992992993e-06,
-      "loss": 1.3113,
+      "loss": 1.3063,
       "step": 20050
     },
     {
       "epoch": 60.24,
-      "grad_norm": 8.063593864440918,
+      "grad_norm": 8.455909729003906,
       "learning_rate": 7.991991991991993e-06,
-      "loss": 1.3002,
+      "loss": 1.304,
       "step": 20060
     },
     {
       "epoch": 60.27,
-      "grad_norm": 10.713462829589844,
+      "grad_norm": 13.570226669311523,
       "learning_rate": 7.990990990990992e-06,
-      "loss": 1.3118,
+      "loss": 1.3071,
       "step": 20070
     },
     {
       "epoch": 60.3,
-      "grad_norm": 6.639123916625977,
+      "grad_norm": 8.090388298034668,
       "learning_rate": 7.98998998998999e-06,
-      "loss": 1.3193,
+      "loss": 1.3145,
       "step": 20080
     },
     {
       "epoch": 60.33,
-      "grad_norm": 8.440950393676758,
+      "grad_norm": 7.700653553009033,
       "learning_rate": 7.98898898898899e-06,
-      "loss": 1.2932,
+      "loss": 1.2949,
       "step": 20090
     },
     {
       "epoch": 60.36,
-      "grad_norm": 7.534139156341553,
+      "grad_norm": 8.69250774383545,
       "learning_rate": 7.987987987987988e-06,
-      "loss": 1.3026,
+      "loss": 1.2997,
       "step": 20100
     },
     {
       "epoch": 60.39,
-      "grad_norm": 10.147226333618164,
+      "grad_norm": 9.442805290222168,
       "learning_rate": 7.986986986986988e-06,
-      "loss": 1.2585,
+      "loss": 1.2447,
       "step": 20110
     },
     {
       "epoch": 60.42,
-      "grad_norm": 7.7242112159729,
+      "grad_norm": 5.087226867675781,
       "learning_rate": 7.985985985985987e-06,
-      "loss": 1.243,
+      "loss": 1.2543,
       "step": 20120
     },
     {
       "epoch": 60.45,
-      "grad_norm": 5.895683288574219,
+      "grad_norm": 6.626835346221924,
       "learning_rate": 7.984984984984985e-06,
-      "loss": 1.2694,
+      "loss": 1.2551,
       "step": 20130
     },
     {
       "epoch": 60.48,
-      "grad_norm": 5.6034417152404785,
+      "grad_norm": 10.057159423828125,
       "learning_rate": 7.983983983983986e-06,
-      "loss": 1.2806,
+      "loss": 1.2765,
       "step": 20140
     },
     {
       "epoch": 60.51,
-      "grad_norm": 6.811248779296875,
+      "grad_norm": 8.274714469909668,
       "learning_rate": 7.982982982982984e-06,
-      "loss": 1.2015,
+      "loss": 1.1907,
       "step": 20150
     },
     {
       "epoch": 60.54,
-      "grad_norm": 5.511872291564941,
+      "grad_norm": 7.928471088409424,
       "learning_rate": 7.981981981981983e-06,
-      "loss": 1.2957,
+      "loss": 1.3052,
       "step": 20160
     },
     {
       "epoch": 60.57,
-      "grad_norm": 6.797493934631348,
+      "grad_norm": 7.841147422790527,
       "learning_rate": 7.980980980980981e-06,
-      "loss": 1.2733,
+      "loss": 1.271,
       "step": 20170
     },
     {
       "epoch": 60.6,
-      "grad_norm": 6.288280010223389,
+      "grad_norm": 8.763837814331055,
       "learning_rate": 7.97997997997998e-06,
-      "loss": 1.312,
+      "loss": 1.3125,
       "step": 20180
     },
     {
       "epoch": 60.63,
-      "grad_norm": 7.951109886169434,
+      "grad_norm": 9.657464027404785,
       "learning_rate": 7.97897897897898e-06,
-      "loss": 1.2983,
+      "loss": 1.2856,
       "step": 20190
     },
     {
       "epoch": 60.66,
-      "grad_norm": 7.7464070320129395,
+      "grad_norm": 9.021210670471191,
       "learning_rate": 7.977977977977979e-06,
-      "loss": 1.3605,
+      "loss": 1.3554,
       "step": 20200
     },
     {
       "epoch": 60.69,
-      "grad_norm": 10.839215278625488,
+      "grad_norm": 9.897318840026855,
       "learning_rate": 7.976976976976977e-06,
-      "loss": 1.2944,
+      "loss": 1.3038,
       "step": 20210
     },
     {
       "epoch": 60.72,
-      "grad_norm": 7.027478218078613,
+      "grad_norm": 7.748193740844727,
       "learning_rate": 7.975975975975978e-06,
-      "loss": 1.307,
+      "loss": 1.2981,
       "step": 20220
     },
     {
       "epoch": 60.75,
-      "grad_norm": 8.071322441101074,
+      "grad_norm": 7.007231712341309,
       "learning_rate": 7.974974974974976e-06,
       "loss": 1.2694,
       "step": 20230
     },
     {
       "epoch": 60.78,
-      "grad_norm": 6.7838335037231445,
+      "grad_norm": 10.683167457580566,
       "learning_rate": 7.973973973973973e-06,
-      "loss": 1.3623,
+      "loss": 1.3682,
       "step": 20240
     },
     {
       "epoch": 60.81,
-      "grad_norm": 6.0882649421691895,
+      "grad_norm": 7.152737140655518,
       "learning_rate": 7.972972972972974e-06,
-      "loss": 1.3025,
+      "loss": 1.285,
       "step": 20250
     },
     {
       "epoch": 60.84,
-      "grad_norm": 5.150771141052246,
+      "grad_norm": 8.764065742492676,
       "learning_rate": 7.971971971971972e-06,
-      "loss": 1.2925,
+      "loss": 1.2946,
       "step": 20260
     },
     {
       "epoch": 60.87,
-      "grad_norm": 8.849813461303711,
+      "grad_norm": 9.880595207214355,
       "learning_rate": 7.97097097097097e-06,
-      "loss": 1.3494,
+      "loss": 1.3426,
       "step": 20270
     },
     {
       "epoch": 60.9,
-      "grad_norm": 7.056886196136475,
+      "grad_norm": 6.820680141448975,
       "learning_rate": 7.969969969969971e-06,
-      "loss": 1.3098,
+      "loss": 1.3119,
       "step": 20280
     },
     {
       "epoch": 60.93,
-      "grad_norm": 7.42539119720459,
+      "grad_norm": 8.395801544189453,
       "learning_rate": 7.96896896896897e-06,
-      "loss": 1.3161,
+      "loss": 1.3254,
       "step": 20290
     },
     {
       "epoch": 60.96,
-      "grad_norm": 6.874979496002197,
+      "grad_norm": 7.079298496246338,
       "learning_rate": 7.967967967967968e-06,
-      "loss": 1.3025,
+      "loss": 1.3004,
       "step": 20300
     },
     {
       "epoch": 60.99,
-      "grad_norm": 5.879049301147461,
+      "grad_norm": 7.492618560791016,
       "learning_rate": 7.966966966966969e-06,
-      "loss": 1.3055,
+      "loss": 1.3061,
       "step": 20310
     },
     {
       "epoch": 61.0,
-      "eval_accuracy": 0.7697,
-      "eval_loss": 0.7513387799263,
-      "eval_runtime": 5.6997,
-      "eval_samples_per_second": 1754.482,
-      "eval_steps_per_second": 7.018,
+      "eval_accuracy": 0.7711,
+      "eval_loss": 0.7504168152809143,
+      "eval_runtime": 5.2107,
+      "eval_samples_per_second": 1919.134,
+      "eval_steps_per_second": 7.677,
       "step": 20313
     },
     {
       "epoch": 61.02,
-      "grad_norm": 8.804891586303711,
+      "grad_norm": 9.342314720153809,
       "learning_rate": 7.965965965965967e-06,
-      "loss": 1.3915,
+      "loss": 1.3926,
       "step": 20320
     },
     {
       "epoch": 61.05,
-      "grad_norm": 8.954228401184082,
+      "grad_norm": 8.724595069885254,
       "learning_rate": 7.964964964964966e-06,
-      "loss": 1.2626,
+      "loss": 1.2559,
       "step": 20330
     },
     {
       "epoch": 61.08,
-      "grad_norm": 5.268406391143799,
+      "grad_norm": 6.62534761428833,
       "learning_rate": 7.963963963963964e-06,
-      "loss": 1.3582,
+      "loss": 1.3535,
       "step": 20340
     },
     {
       "epoch": 61.11,
-      "grad_norm": 6.993339538574219,
+      "grad_norm": 8.766988754272461,
       "learning_rate": 7.962962962962963e-06,
-      "loss": 1.3355,
+      "loss": 1.3216,
       "step": 20350
     },
     {
       "epoch": 61.14,
-      "grad_norm": 8.941996574401855,
+      "grad_norm": 9.382036209106445,
       "learning_rate": 7.961961961961963e-06,
-      "loss": 1.2746,
+      "loss": 1.2778,
       "step": 20360
     },
     {
       "epoch": 61.17,
-      "grad_norm": 6.089354038238525,
+      "grad_norm": 9.829968452453613,
       "learning_rate": 7.960960960960962e-06,
-      "loss": 1.2868,
+      "loss": 1.2975,
       "step": 20370
     },
     {
       "epoch": 61.2,
-      "grad_norm": 8.36475658416748,
+      "grad_norm": 9.745413780212402,
       "learning_rate": 7.95995995995996e-06,
-      "loss": 1.2926,
+      "loss": 1.2909,
       "step": 20380
     },
     {
       "epoch": 61.23,
-      "grad_norm": 6.147676467895508,
+      "grad_norm": 7.375082015991211,
       "learning_rate": 7.95895895895896e-06,
-      "loss": 1.3075,
+      "loss": 1.2989,
       "step": 20390
     },
     {
       "epoch": 61.26,
-      "grad_norm": 8.281792640686035,
+      "grad_norm": 8.023751258850098,
       "learning_rate": 7.95795795795796e-06,
-      "loss": 1.3148,
+      "loss": 1.3209,
       "step": 20400
     },
     {
       "epoch": 61.29,
-      "grad_norm": 7.665432453155518,
+      "grad_norm": 7.956850528717041,
       "learning_rate": 7.956956956956958e-06,
-      "loss": 1.2846,
+      "loss": 1.286,
       "step": 20410
     },
     {
       "epoch": 61.32,
-      "grad_norm": 9.627096176147461,
+      "grad_norm": 9.264697074890137,
       "learning_rate": 7.955955955955956e-06,
-      "loss": 1.2754,
+      "loss": 1.2658,
       "step": 20420
     },
     {
       "epoch": 61.35,
-      "grad_norm": 7.9203009605407715,
+      "grad_norm": 7.695738315582275,
       "learning_rate": 7.954954954954955e-06,
-      "loss": 1.3752,
+      "loss": 1.3781,
       "step": 20430
     },
     {
       "epoch": 61.38,
-      "grad_norm": 7.659634590148926,
+      "grad_norm": 9.000349044799805,
       "learning_rate": 7.953953953953955e-06,
       "loss": 1.3158,
       "step": 20440
     },
     {
       "epoch": 61.41,
-      "grad_norm": 6.605567932128906,
+      "grad_norm": 7.861306667327881,
       "learning_rate": 7.952952952952954e-06,
-      "loss": 1.3362,
+      "loss": 1.3313,
       "step": 20450
     },
     {
       "epoch": 61.44,
-      "grad_norm": 8.443327903747559,
+      "grad_norm": 7.626608848571777,
       "learning_rate": 7.951951951951953e-06,
-      "loss": 1.2928,
+      "loss": 1.2983,
       "step": 20460
     },
     {
       "epoch": 61.47,
-      "grad_norm": 6.368496894836426,
+      "grad_norm": 6.303131103515625,
       "learning_rate": 7.950950950950951e-06,
-      "loss": 1.2672,
+      "loss": 1.2681,
       "step": 20470
     },
     {
       "epoch": 61.5,
-      "grad_norm": 9.05487060546875,
+      "grad_norm": 7.708277225494385,
       "learning_rate": 7.949949949949951e-06,
-      "loss": 1.3906,
+      "loss": 1.3776,
       "step": 20480
     },
     {
       "epoch": 61.53,
-      "grad_norm": 5.630706787109375,
+      "grad_norm": 6.177462100982666,
       "learning_rate": 7.948948948948948e-06,
-      "loss": 1.3186,
+      "loss": 1.316,
       "step": 20490
     },
     {
       "epoch": 61.56,
-      "grad_norm": 5.518985271453857,
+      "grad_norm": 6.7039103507995605,
       "learning_rate": 7.947947947947949e-06,
-      "loss": 1.2987,
+      "loss": 1.2844,
       "step": 20500
     },
     {
       "epoch": 61.59,
-      "grad_norm": 6.891030311584473,
+      "grad_norm": 8.518594741821289,
       "learning_rate": 7.946946946946947e-06,
-      "loss": 1.2945,
+      "loss": 1.2952,
       "step": 20510
     },
     {
       "epoch": 61.62,
-      "grad_norm": 6.5515031814575195,
+      "grad_norm": 7.475151062011719,
       "learning_rate": 7.945945945945946e-06,
-      "loss": 1.3144,
+      "loss": 1.3091,
       "step": 20520
     },
     {
       "epoch": 61.65,
-      "grad_norm": 7.2461957931518555,
+      "grad_norm": 10.0471773147583,
       "learning_rate": 7.944944944944946e-06,
-      "loss": 1.3123,
+      "loss": 1.3164,
       "step": 20530
     },
     {
       "epoch": 61.68,
-      "grad_norm": 6.854095458984375,
+      "grad_norm": 8.200508117675781,
       "learning_rate": 7.943943943943945e-06,
-      "loss": 1.3218,
+      "loss": 1.3188,
       "step": 20540
     },
     {
       "epoch": 61.71,
-      "grad_norm": 7.17921257019043,
+      "grad_norm": 6.788938999176025,
       "learning_rate": 7.942942942942943e-06,
-      "loss": 1.2696,
+      "loss": 1.273,
       "step": 20550
     },
     {
       "epoch": 61.74,
-      "grad_norm": 6.277776718139648,
+      "grad_norm": 7.159660816192627,
       "learning_rate": 7.941941941941944e-06,
-      "loss": 1.2582,
+      "loss": 1.2608,
       "step": 20560
     },
     {
       "epoch": 61.77,
-      "grad_norm": 6.533068656921387,
+      "grad_norm": 8.004463195800781,
       "learning_rate": 7.940940940940942e-06,
-      "loss": 1.3232,
+      "loss": 1.3193,
       "step": 20570
     },
     {
       "epoch": 61.8,
-      "grad_norm": 4.591888904571533,
+      "grad_norm": 5.122641086578369,
       "learning_rate": 7.93993993993994e-06,
-      "loss": 1.2974,
+      "loss": 1.2995,
       "step": 20580
     },
     {
       "epoch": 61.83,
-      "grad_norm": 8.03518009185791,
+      "grad_norm": 9.461299896240234,
       "learning_rate": 7.93893893893894e-06,
-      "loss": 1.3219,
+      "loss": 1.3156,
       "step": 20590
     },
     {
       "epoch": 61.86,
-      "grad_norm": 7.953175067901611,
+      "grad_norm": 8.655679702758789,
       "learning_rate": 7.937937937937938e-06,
-      "loss": 1.3092,
+      "loss": 1.3064,
       "step": 20600
     },
     {
       "epoch": 61.89,
-      "grad_norm": 7.125205993652344,
+      "grad_norm": 7.195794105529785,
       "learning_rate": 7.936936936936938e-06,
-      "loss": 1.2654,
+      "loss": 1.2661,
       "step": 20610
     },
     {
       "epoch": 61.92,
-      "grad_norm": 7.77432107925415,
+      "grad_norm": 7.506396770477295,
       "learning_rate": 7.935935935935937e-06,
-      "loss": 1.2326,
+      "loss": 1.2446,
       "step": 20620
     },
     {
       "epoch": 61.95,
-      "grad_norm": 7.625723361968994,
+      "grad_norm": 7.402343273162842,
       "learning_rate": 7.934934934934935e-06,
-      "loss": 1.2945,
+      "loss": 1.2938,
       "step": 20630
     },
     {
       "epoch": 61.98,
-      "grad_norm": 9.399913787841797,
+      "grad_norm": 7.56856632232666,
       "learning_rate": 7.933933933933936e-06,
-      "loss": 1.3024,
+      "loss": 1.2961,
       "step": 20640
     },
     {
       "epoch": 62.0,
-      "eval_accuracy": 0.7728,
-      "eval_loss": 0.7509605288505554,
-      "eval_runtime": 5.5808,
-      "eval_samples_per_second": 1791.869,
-      "eval_steps_per_second": 7.167,
+      "eval_accuracy": 0.7713,
+      "eval_loss": 0.7511435747146606,
+      "eval_runtime": 5.564,
+      "eval_samples_per_second": 1797.259,
+      "eval_steps_per_second": 7.189,
       "step": 20646
     },
     {
       "epoch": 62.01,
-      "grad_norm": 5.953549385070801,
+      "grad_norm": 7.6172990798950195,
       "learning_rate": 7.932932932932934e-06,
-      "loss": 1.376,
+      "loss": 1.3639,
       "step": 20650
     },
     {
       "epoch": 62.04,
-      "grad_norm": 5.321707248687744,
+      "grad_norm": 8.956127166748047,
       "learning_rate": 7.931931931931933e-06,
-      "loss": 1.2554,
+      "loss": 1.2491,
       "step": 20660
     },
     {
       "epoch": 62.07,
-      "grad_norm": 8.563325881958008,
+      "grad_norm": 8.8591947555542,
       "learning_rate": 7.930930930930931e-06,
-      "loss": 1.301,
+      "loss": 1.2854,
       "step": 20670
     },
     {
       "epoch": 62.1,
-      "grad_norm": 7.113579273223877,
+      "grad_norm": 8.902877807617188,
       "learning_rate": 7.92992992992993e-06,
-      "loss": 1.3295,
+      "loss": 1.3184,
       "step": 20680
     },
     {
       "epoch": 62.13,
-      "grad_norm": 6.938343048095703,
+      "grad_norm": 7.533481597900391,
       "learning_rate": 7.928928928928929e-06,
-      "loss": 1.351,
+      "loss": 1.3454,
       "step": 20690
     },
     {
       "epoch": 62.16,
-      "grad_norm": 6.700629234313965,
+      "grad_norm": 9.840119361877441,
       "learning_rate": 7.927927927927929e-06,
-      "loss": 1.3392,
+      "loss": 1.3461,
       "step": 20700
     },
     {
       "epoch": 62.19,
-      "grad_norm": 13.614136695861816,
+      "grad_norm": 13.765921592712402,
       "learning_rate": 7.926926926926928e-06,
-      "loss": 1.3289,
+      "loss": 1.3225,
       "step": 20710
     },
     {
       "epoch": 62.22,
-      "grad_norm": 7.177995681762695,
+      "grad_norm": 7.8415374755859375,
       "learning_rate": 7.925925925925926e-06,
-      "loss": 1.2805,
+      "loss": 1.2745,
       "step": 20720
     },
     {
       "epoch": 62.25,
-      "grad_norm": 7.2353835105896,
+      "grad_norm": 9.186236381530762,
       "learning_rate": 7.924924924924926e-06,
-      "loss": 1.2979,
+      "loss": 1.3001,
       "step": 20730
     },
     {
       "epoch": 62.28,
-      "grad_norm": 6.156536102294922,
+      "grad_norm": 10.041616439819336,
       "learning_rate": 7.923923923923923e-06,
-      "loss": 1.3118,
+      "loss": 1.3208,
       "step": 20740
     },
     {
       "epoch": 62.31,
-      "grad_norm": 6.4925737380981445,
+      "grad_norm": 8.336459159851074,
       "learning_rate": 7.922922922922924e-06,
-      "loss": 1.3137,
+      "loss": 1.2958,
       "step": 20750
     },
     {
       "epoch": 62.34,
-      "grad_norm": 8.708223342895508,
+      "grad_norm": 8.116728782653809,
       "learning_rate": 7.921921921921922e-06,
-      "loss": 1.2777,
+      "loss": 1.2833,
       "step": 20760
     },
     {
       "epoch": 62.37,
-      "grad_norm": 7.131631851196289,
+      "grad_norm": 6.492475509643555,
       "learning_rate": 7.92092092092092e-06,
-      "loss": 1.2737,
+      "loss": 1.2713,
       "step": 20770
     },
     {
       "epoch": 62.4,
-      "grad_norm": 7.107925891876221,
+      "grad_norm": 7.847910404205322,
       "learning_rate": 7.919919919919921e-06,
-      "loss": 1.2739,
+      "loss": 1.285,
       "step": 20780
     },
     {
       "epoch": 62.43,
-      "grad_norm": 8.95250415802002,
+      "grad_norm": 8.123458862304688,
       "learning_rate": 7.91891891891892e-06,
       "loss": 1.3398,
       "step": 20790
     },
     {
       "epoch": 62.46,
-      "grad_norm": 5.695205211639404,
+      "grad_norm": 7.179603576660156,
       "learning_rate": 7.917917917917918e-06,
-      "loss": 1.2855,
+      "loss": 1.2828,
       "step": 20800
     },
     {
       "epoch": 62.49,
-      "grad_norm": 9.461745262145996,
+      "grad_norm": 10.120737075805664,
       "learning_rate": 7.916916916916919e-06,
-      "loss": 1.2397,
+      "loss": 1.2453,
       "step": 20810
     },
     {
       "epoch": 62.52,
-      "grad_norm": 7.9220356941223145,
+      "grad_norm": 8.784655570983887,
       "learning_rate": 7.915915915915915e-06,
-      "loss": 1.2324,
+      "loss": 1.2275,
       "step": 20820
     },
     {
       "epoch": 62.55,
-      "grad_norm": 7.228675842285156,
+      "grad_norm": 8.43416690826416,
       "learning_rate": 7.914914914914916e-06,
-      "loss": 1.304,
+      "loss": 1.3134,
       "step": 20830
     },
     {
       "epoch": 62.58,
-      "grad_norm": 4.7078118324279785,
+      "grad_norm": 6.474085807800293,
       "learning_rate": 7.913913913913914e-06,
-      "loss": 1.3772,
+      "loss": 1.3824,
       "step": 20840
     },
     {
       "epoch": 62.61,
-      "grad_norm": 6.887867450714111,
+      "grad_norm": 7.8278279304504395,
       "learning_rate": 7.912912912912913e-06,
-      "loss": 1.2473,
+      "loss": 1.2404,
       "step": 20850
     },
     {
       "epoch": 62.64,
-      "grad_norm": 6.111298561096191,
+      "grad_norm": 8.032684326171875,
       "learning_rate": 7.911911911911913e-06,
-      "loss": 1.3156,
+      "loss": 1.3216,
       "step": 20860
     },
     {
       "epoch": 62.67,
-      "grad_norm": 6.175546646118164,
+      "grad_norm": 8.206449508666992,
       "learning_rate": 7.910910910910912e-06,
-      "loss": 1.3071,
+      "loss": 1.3097,
       "step": 20870
     },
     {
       "epoch": 62.7,
-      "grad_norm": 6.714319705963135,
+      "grad_norm": 6.3854594230651855,
       "learning_rate": 7.90990990990991e-06,
-      "loss": 1.348,
+      "loss": 1.3332,
       "step": 20880
     },
     {
       "epoch": 62.73,
-      "grad_norm": 5.813368320465088,
+      "grad_norm": 7.8075971603393555,
       "learning_rate": 7.90890890890891e-06,
-      "loss": 1.2862,
+      "loss": 1.2808,
       "step": 20890
     },
     {
       "epoch": 62.76,
-      "grad_norm": 9.521125793457031,
+      "grad_norm": 9.260588645935059,
       "learning_rate": 7.90790790790791e-06,
-      "loss": 1.3106,
+      "loss": 1.289,
       "step": 20900
     },
     {
       "epoch": 62.79,
-      "grad_norm": 9.92933464050293,
+      "grad_norm": 12.30446720123291,
       "learning_rate": 7.906906906906908e-06,
-      "loss": 1.3923,
+      "loss": 1.3855,
       "step": 20910
     },
     {
       "epoch": 62.82,
-      "grad_norm": 9.619848251342773,
+      "grad_norm": 8.755667686462402,
       "learning_rate": 7.905905905905907e-06,
-      "loss": 1.2818,
+      "loss": 1.2828,
       "step": 20920
     },
     {
       "epoch": 62.85,
-      "grad_norm": 7.834268569946289,
+      "grad_norm": 9.222872734069824,
       "learning_rate": 7.904904904904905e-06,
-      "loss": 1.3698,
+      "loss": 1.3643,
       "step": 20930
     },
     {
       "epoch": 62.88,
-      "grad_norm": 6.417015552520752,
+      "grad_norm": 8.54389762878418,
       "learning_rate": 7.903903903903904e-06,
-      "loss": 1.2626,
+      "loss": 1.2677,
       "step": 20940
     },
     {
       "epoch": 62.91,
-      "grad_norm": 6.362645626068115,
+      "grad_norm": 5.985835075378418,
       "learning_rate": 7.902902902902904e-06,
-      "loss": 1.3513,
+      "loss": 1.345,
       "step": 20950
     },
     {
       "epoch": 62.94,
-      "grad_norm": 7.671140193939209,
+      "grad_norm": 6.369659423828125,
       "learning_rate": 7.901901901901903e-06,
-      "loss": 1.288,
+      "loss": 1.2946,
       "step": 20960
     },
     {
       "epoch": 62.97,
-      "grad_norm": 5.830514430999756,
+      "grad_norm": 10.018310546875,
       "learning_rate": 7.900900900900901e-06,
-      "loss": 1.3113,
+      "loss": 1.314,
       "step": 20970
     },
     {
       "epoch": 63.0,
-      "eval_accuracy": 0.7731,
-      "eval_loss": 0.7432555556297302,
-      "eval_runtime": 5.5097,
-      "eval_samples_per_second": 1814.965,
-      "eval_steps_per_second": 7.26,
+      "eval_accuracy": 0.7745,
+      "eval_loss": 0.7431166172027588,
+      "eval_runtime": 5.4257,
+      "eval_samples_per_second": 1843.079,
+      "eval_steps_per_second": 7.372,
       "step": 20979
     },
     {
       "epoch": 63.0,
-      "grad_norm": 6.015483856201172,
+      "grad_norm": 7.779452800750732,
       "learning_rate": 7.899899899899901e-06,
-      "loss": 1.3745,
+      "loss": 1.3341,
       "step": 20980
     },
     {
       "epoch": 63.03,
-      "grad_norm": 6.549117565155029,
+      "grad_norm": 5.645828723907471,
       "learning_rate": 7.898898898898898e-06,
-      "loss": 1.2343,
+      "loss": 1.2473,
       "step": 20990
     },
     {
       "epoch": 63.06,
-      "grad_norm": 6.097425937652588,
+      "grad_norm": 8.36324405670166,
       "learning_rate": 7.897897897897899e-06,
-      "loss": 1.3506,
+      "loss": 1.3433,
       "step": 21000
     },
     {
       "epoch": 63.09,
-      "grad_norm": 9.524245262145996,
+      "grad_norm": 9.949673652648926,
       "learning_rate": 7.896896896896897e-06,
-      "loss": 1.2782,
+      "loss": 1.2829,
       "step": 21010
     },
     {
       "epoch": 63.12,
-      "grad_norm": 5.165883541107178,
+      "grad_norm": 6.381639003753662,
       "learning_rate": 7.895895895895896e-06,
-      "loss": 1.2914,
+      "loss": 1.2974,
       "step": 21020
     },
     {
       "epoch": 63.15,
-      "grad_norm": 7.849681377410889,
+      "grad_norm": 9.318140029907227,
       "learning_rate": 7.894894894894896e-06,
-      "loss": 1.2319,
+      "loss": 1.2378,
       "step": 21030
     },
     {
       "epoch": 63.18,
-      "grad_norm": 5.985138416290283,
+      "grad_norm": 5.865783214569092,
       "learning_rate": 7.893893893893895e-06,
-      "loss": 1.3553,
+      "loss": 1.3591,
       "step": 21040
     },
     {
       "epoch": 63.21,
-      "grad_norm": 8.654874801635742,
+      "grad_norm": 10.829254150390625,
       "learning_rate": 7.892892892892893e-06,
-      "loss": 1.2819,
+      "loss": 1.2869,
       "step": 21050
     },
     {
       "epoch": 63.24,
-      "grad_norm": 7.863469123840332,
+      "grad_norm": 8.766711235046387,
       "learning_rate": 7.891891891891894e-06,
-      "loss": 1.2331,
+      "loss": 1.218,
       "step": 21060
     },
     {
       "epoch": 63.27,
-      "grad_norm": 6.777470111846924,
+      "grad_norm": 7.861964225769043,
       "learning_rate": 7.89089089089089e-06,
-      "loss": 1.2563,
+      "loss": 1.2469,
       "step": 21070
     },
     {
       "epoch": 63.3,
-      "grad_norm": 9.522257804870605,
+      "grad_norm": 8.132325172424316,
       "learning_rate": 7.88988988988989e-06,
-      "loss": 1.2382,
+      "loss": 1.2313,
       "step": 21080
     },
     {
       "epoch": 63.33,
-      "grad_norm": 6.856578350067139,
+      "grad_norm": 7.889732360839844,
       "learning_rate": 7.88888888888889e-06,
-      "loss": 1.3057,
+      "loss": 1.2985,
       "step": 21090
     },
     {
       "epoch": 63.36,
-      "grad_norm": 7.572615146636963,
+      "grad_norm": 8.7166166305542,
       "learning_rate": 7.887887887887888e-06,
-      "loss": 1.2251,
+      "loss": 1.2046,
       "step": 21100
     },
     {
       "epoch": 63.39,
-      "grad_norm": 6.587063312530518,
+      "grad_norm": 6.299531936645508,
       "learning_rate": 7.886886886886888e-06,
-      "loss": 1.3185,
+      "loss": 1.3137,
       "step": 21110
     },
     {
       "epoch": 63.42,
-      "grad_norm": 6.151911735534668,
+      "grad_norm": 6.8727874755859375,
       "learning_rate": 7.885885885885887e-06,
-      "loss": 1.3559,
+      "loss": 1.3522,
       "step": 21120
     },
     {
       "epoch": 63.45,
-      "grad_norm": 7.693611145019531,
+      "grad_norm": 9.370804786682129,
       "learning_rate": 7.884884884884885e-06,
-      "loss": 1.3338,
+      "loss": 1.3381,
       "step": 21130
     },
     {
       "epoch": 63.48,
-      "grad_norm": 9.689712524414062,
+      "grad_norm": 7.846403121948242,
       "learning_rate": 7.883883883883886e-06,
-      "loss": 1.3594,
+      "loss": 1.358,
       "step": 21140
     },
     {
       "epoch": 63.51,
-      "grad_norm": 7.994779109954834,
+      "grad_norm": 7.949915409088135,
       "learning_rate": 7.882882882882884e-06,
-      "loss": 1.2216,
+      "loss": 1.2196,
       "step": 21150
     },
     {
       "epoch": 63.54,
-      "grad_norm": 7.014646053314209,
+      "grad_norm": 5.884675025939941,
       "learning_rate": 7.881881881881881e-06,
-      "loss": 1.3128,
+      "loss": 1.3085,
       "step": 21160
     },
     {
       "epoch": 63.57,
-      "grad_norm": 6.589148044586182,
+      "grad_norm": 8.049286842346191,
       "learning_rate": 7.880880880880882e-06,
-      "loss": 1.3432,
+      "loss": 1.3459,
       "step": 21170
     },
     {
       "epoch": 63.6,
-      "grad_norm": 5.771283149719238,
+      "grad_norm": 7.461377143859863,
       "learning_rate": 7.87987987987988e-06,
-      "loss": 1.2848,
+      "loss": 1.2781,
       "step": 21180
     },
     {
       "epoch": 63.63,
-      "grad_norm": 5.796594142913818,
+      "grad_norm": 8.941048622131348,
       "learning_rate": 7.878878878878879e-06,
-      "loss": 1.3071,
+      "loss": 1.3144,
       "step": 21190
     },
     {
       "epoch": 63.66,
-      "grad_norm": 5.355749130249023,
+      "grad_norm": 6.7546868324279785,
       "learning_rate": 7.877877877877879e-06,
-      "loss": 1.3342,
+      "loss": 1.3246,
       "step": 21200
     },
     {
       "epoch": 63.69,
-      "grad_norm": 7.137463092803955,
+      "grad_norm": 7.739207744598389,
       "learning_rate": 7.876876876876878e-06,
-      "loss": 1.3208,
+      "loss": 1.3279,
       "step": 21210
     },
     {
       "epoch": 63.72,
-      "grad_norm": 6.895020961761475,
+      "grad_norm": 8.059503555297852,
       "learning_rate": 7.875875875875876e-06,
-      "loss": 1.2546,
+      "loss": 1.2588,
       "step": 21220
     },
     {
       "epoch": 63.75,
-      "grad_norm": 5.601885795593262,
+      "grad_norm": 6.593814373016357,
       "learning_rate": 7.874874874874877e-06,
-      "loss": 1.2112,
+      "loss": 1.2099,
       "step": 21230
     },
     {
       "epoch": 63.78,
-      "grad_norm": 5.3109354972839355,
+      "grad_norm": 7.188880920410156,
       "learning_rate": 7.873873873873873e-06,
-      "loss": 1.2285,
+      "loss": 1.2289,
       "step": 21240
     },
     {
       "epoch": 63.81,
-      "grad_norm": 8.5261869430542,
+      "grad_norm": 9.372003555297852,
       "learning_rate": 7.872872872872874e-06,
-      "loss": 1.2266,
+      "loss": 1.2254,
       "step": 21250
     },
     {
       "epoch": 63.84,
-      "grad_norm": 7.094719886779785,
+      "grad_norm": 5.324840545654297,
       "learning_rate": 7.871871871871872e-06,
-      "loss": 1.277,
+      "loss": 1.2706,
       "step": 21260
     },
     {
       "epoch": 63.87,
-      "grad_norm": 6.747025966644287,
+      "grad_norm": 7.516570091247559,
       "learning_rate": 7.870870870870871e-06,
-      "loss": 1.2795,
+      "loss": 1.267,
       "step": 21270
     },
     {
       "epoch": 63.9,
-      "grad_norm": 8.050060272216797,
+      "grad_norm": 8.119369506835938,
       "learning_rate": 7.869869869869871e-06,
-      "loss": 1.2054,
+      "loss": 1.2109,
       "step": 21280
     },
     {
       "epoch": 63.93,
-      "grad_norm": 5.304065227508545,
+      "grad_norm": 5.517627716064453,
       "learning_rate": 7.86886886886887e-06,
-      "loss": 1.2732,
+      "loss": 1.2704,
       "step": 21290
     },
     {
       "epoch": 63.96,
-      "grad_norm": 8.235851287841797,
+      "grad_norm": 6.962592124938965,
       "learning_rate": 7.867867867867868e-06,
-      "loss": 1.3161,
+      "loss": 1.3135,
       "step": 21300
     },
     {
       "epoch": 63.99,
-      "grad_norm": 7.019266605377197,
+      "grad_norm": 7.86956262588501,
       "learning_rate": 7.866866866866869e-06,
-      "loss": 1.2962,
+      "loss": 1.2919,
       "step": 21310
     },
     {
       "epoch": 64.0,
-      "eval_accuracy": 0.77,
-      "eval_loss": 0.7500496506690979,
-      "eval_runtime": 5.5648,
-      "eval_samples_per_second": 1796.994,
-      "eval_steps_per_second": 7.188,
+      "eval_accuracy": 0.7731,
+      "eval_loss": 0.7476633787155151,
+      "eval_runtime": 5.3544,
+      "eval_samples_per_second": 1867.638,
+      "eval_steps_per_second": 7.471,
       "step": 21312
     },
     {
       "epoch": 64.02,
-      "grad_norm": 7.519073009490967,
+      "grad_norm": 8.595781326293945,
       "learning_rate": 7.865865865865866e-06,
-      "loss": 1.4812,
+      "loss": 1.4844,
       "step": 21320
     },
     {
       "epoch": 64.05,
-      "grad_norm": 6.978976249694824,
+      "grad_norm": 8.669222831726074,
       "learning_rate": 7.864864864864866e-06,
-      "loss": 1.253,
+      "loss": 1.2599,
       "step": 21330
     },
     {
       "epoch": 64.08,
-      "grad_norm": 7.677917003631592,
+      "grad_norm": 8.126803398132324,
       "learning_rate": 7.863863863863864e-06,
-      "loss": 1.3177,
+      "loss": 1.3246,
       "step": 21340
     },
     {
       "epoch": 64.11,
-      "grad_norm": 6.549238681793213,
+      "grad_norm": 7.327541351318359,
       "learning_rate": 7.862862862862863e-06,
-      "loss": 1.2905,
+      "loss": 1.2969,
       "step": 21350
     },
     {
       "epoch": 64.14,
-      "grad_norm": 10.248534202575684,
+      "grad_norm": 9.474974632263184,
       "learning_rate": 7.861861861861863e-06,
-      "loss": 1.3111,
+      "loss": 1.3009,
       "step": 21360
     },
     {
       "epoch": 64.17,
-      "grad_norm": 7.4253339767456055,
+      "grad_norm": 7.740027904510498,
       "learning_rate": 7.860860860860862e-06,
-      "loss": 1.2354,
+      "loss": 1.2258,
       "step": 21370
     },
     {
       "epoch": 64.2,
-      "grad_norm": 6.039665699005127,
+      "grad_norm": 8.901143074035645,
       "learning_rate": 7.85985985985986e-06,
-      "loss": 1.2617,
+      "loss": 1.2579,
       "step": 21380
     },
     {
       "epoch": 64.23,
-      "grad_norm": 6.037467002868652,
+      "grad_norm": 6.425699710845947,
       "learning_rate": 7.858858858858859e-06,
-      "loss": 1.3122,
+      "loss": 1.3201,
       "step": 21390
     },
     {
       "epoch": 64.26,
-      "grad_norm": 7.1780829429626465,
+      "grad_norm": 7.991940498352051,
       "learning_rate": 7.85785785785786e-06,
-      "loss": 1.2647,
+      "loss": 1.2561,
       "step": 21400
     },
     {
       "epoch": 64.29,
-      "grad_norm": 8.668288230895996,
+      "grad_norm": 6.204831123352051,
       "learning_rate": 7.856856856856856e-06,
-      "loss": 1.2511,
+      "loss": 1.2519,
       "step": 21410
     },
     {
       "epoch": 64.32,
-      "grad_norm": 6.085570335388184,
+      "grad_norm": 5.215510845184326,
       "learning_rate": 7.855855855855857e-06,
-      "loss": 1.2987,
+      "loss": 1.2913,
       "step": 21420
     },
     {
       "epoch": 64.35,
-      "grad_norm": 7.871938228607178,
+      "grad_norm": 6.8644938468933105,
       "learning_rate": 7.854854854854855e-06,
-      "loss": 1.2046,
+      "loss": 1.2066,
       "step": 21430
     },
     {
       "epoch": 64.38,
-      "grad_norm": 6.556540489196777,
+      "grad_norm": 8.482439041137695,
       "learning_rate": 7.853853853853854e-06,
-      "loss": 1.2767,
+      "loss": 1.2717,
       "step": 21440
     },
     {
       "epoch": 64.41,
-      "grad_norm": 8.304003715515137,
+      "grad_norm": 9.698838233947754,
       "learning_rate": 7.852852852852854e-06,
-      "loss": 1.3291,
+      "loss": 1.3358,
       "step": 21450
     },
     {
       "epoch": 64.44,
-      "grad_norm": 7.599071979522705,
+      "grad_norm": 5.280708312988281,
       "learning_rate": 7.851851851851853e-06,
-      "loss": 1.3055,
+      "loss": 1.3125,
       "step": 21460
     },
     {
       "epoch": 64.47,
-      "grad_norm": 5.833497047424316,
+      "grad_norm": 5.0627336502075195,
       "learning_rate": 7.850850850850851e-06,
-      "loss": 1.3182,
+      "loss": 1.3243,
       "step": 21470
     },
     {
       "epoch": 64.5,
-      "grad_norm": 6.532292366027832,
+      "grad_norm": 5.275140762329102,
       "learning_rate": 7.849849849849852e-06,
-      "loss": 1.1607,
+      "loss": 1.1511,
       "step": 21480
     },
     {
       "epoch": 64.53,
-      "grad_norm": 8.404013633728027,
+      "grad_norm": 7.24367618560791,
       "learning_rate": 7.848848848848848e-06,
-      "loss": 1.2533,
+      "loss": 1.2543,
       "step": 21490
     },
     {
       "epoch": 64.56,
-      "grad_norm": 9.92132568359375,
+      "grad_norm": 7.7745842933654785,
       "learning_rate": 7.847847847847849e-06,
-      "loss": 1.316,
+      "loss": 1.3136,
       "step": 21500
     },
     {
       "epoch": 64.59,
-      "grad_norm": 8.63615894317627,
+      "grad_norm": 7.046977519989014,
       "learning_rate": 7.846846846846847e-06,
-      "loss": 1.2969,
+      "loss": 1.3009,
       "step": 21510
     },
     {
       "epoch": 64.62,
-      "grad_norm": 7.604632377624512,
+      "grad_norm": 8.183989524841309,
       "learning_rate": 7.845845845845846e-06,
-      "loss": 1.2368,
+      "loss": 1.2299,
       "step": 21520
     },
     {
       "epoch": 64.65,
-      "grad_norm": 8.739790916442871,
+      "grad_norm": 6.206587791442871,
       "learning_rate": 7.844844844844846e-06,
-      "loss": 1.306,
+      "loss": 1.2987,
       "step": 21530
     },
     {
       "epoch": 64.68,
-      "grad_norm": 7.194324970245361,
+      "grad_norm": 6.723163604736328,
       "learning_rate": 7.843843843843845e-06,
-      "loss": 1.2471,
+      "loss": 1.242,
       "step": 21540
     },
     {
       "epoch": 64.71,
-      "grad_norm": 10.142952919006348,
+      "grad_norm": 13.844172477722168,
       "learning_rate": 7.842842842842843e-06,
-      "loss": 1.2792,
+      "loss": 1.2894,
       "step": 21550
     },
     {
       "epoch": 64.74,
-      "grad_norm": 8.219758987426758,
+      "grad_norm": 10.609146118164062,
       "learning_rate": 7.841841841841844e-06,
-      "loss": 1.2518,
+      "loss": 1.2532,
       "step": 21560
     },
     {
       "epoch": 64.77,
-      "grad_norm": 6.1873369216918945,
+      "grad_norm": 7.701311111450195,
       "learning_rate": 7.84084084084084e-06,
-      "loss": 1.3366,
+      "loss": 1.3339,
       "step": 21570
     },
     {
       "epoch": 64.8,
-      "grad_norm": 8.220772743225098,
+      "grad_norm": 8.437320709228516,
       "learning_rate": 7.839839839839841e-06,
-      "loss": 1.281,
+      "loss": 1.2956,
       "step": 21580
     },
     {
       "epoch": 64.83,
-      "grad_norm": 7.328108787536621,
+      "grad_norm": 8.805789947509766,
       "learning_rate": 7.83883883883884e-06,
-      "loss": 1.3436,
+      "loss": 1.344,
       "step": 21590
     },
     {
       "epoch": 64.86,
-      "grad_norm": 7.380152702331543,
+      "grad_norm": 10.183928489685059,
       "learning_rate": 7.837837837837838e-06,
-      "loss": 1.3473,
+      "loss": 1.3348,
       "step": 21600
     },
     {
       "epoch": 64.89,
-      "grad_norm": 5.999556541442871,
+      "grad_norm": 6.227613925933838,
       "learning_rate": 7.836836836836837e-06,
-      "loss": 1.3302,
+      "loss": 1.3264,
       "step": 21610
     },
     {
       "epoch": 64.92,
-      "grad_norm": 7.146322250366211,
+      "grad_norm": 8.682738304138184,
       "learning_rate": 7.835835835835837e-06,
-      "loss": 1.2561,
+      "loss": 1.2592,
       "step": 21620
     },
     {
       "epoch": 64.95,
-      "grad_norm": 5.284373760223389,
+      "grad_norm": 7.017658233642578,
       "learning_rate": 7.834834834834836e-06,
-      "loss": 1.3056,
+      "loss": 1.2957,
       "step": 21630
     },
     {
       "epoch": 64.98,
-      "grad_norm": 6.459259986877441,
+      "grad_norm": 5.930008411407471,
       "learning_rate": 7.833833833833834e-06,
-      "loss": 1.28,
+      "loss": 1.2928,
       "step": 21640
     },
     {
       "epoch": 65.0,
-      "eval_accuracy": 0.7781,
-      "eval_loss": 0.7306812405586243,
-      "eval_runtime": 5.5331,
-      "eval_samples_per_second": 1807.32,
-      "eval_steps_per_second": 7.229,
+      "eval_accuracy": 0.7765,
+      "eval_loss": 0.7278195023536682,
+      "eval_runtime": 5.4718,
+      "eval_samples_per_second": 1827.539,
+      "eval_steps_per_second": 7.31,
       "step": 21645
     },
     {
       "epoch": 65.02,
-      "grad_norm": 10.681806564331055,
+      "grad_norm": 7.929402828216553,
       "learning_rate": 7.832832832832834e-06,
-      "loss": 1.4707,
+      "loss": 1.4899,
       "step": 21650
     },
     {
       "epoch": 65.05,
-      "grad_norm": 6.194177627563477,
+      "grad_norm": 6.314831733703613,
       "learning_rate": 7.831831831831831e-06,
-      "loss": 1.264,
+      "loss": 1.2656,
       "step": 21660
     },
     {
       "epoch": 65.08,
-      "grad_norm": 7.331558704376221,
+      "grad_norm": 5.672953128814697,
       "learning_rate": 7.830830830830832e-06,
-      "loss": 1.204,
+      "loss": 1.2011,
       "step": 21670
     },
     {
       "epoch": 65.11,
-      "grad_norm": 4.974428653717041,
+      "grad_norm": 6.026956558227539,
       "learning_rate": 7.82982982982983e-06,
-      "loss": 1.2805,
+      "loss": 1.2837,
       "step": 21680
     },
     {
       "epoch": 65.14,
-      "grad_norm": 8.394832611083984,
+      "grad_norm": 8.038647651672363,
       "learning_rate": 7.828828828828829e-06,
-      "loss": 1.2893,
+      "loss": 1.2832,
       "step": 21690
     },
     {
       "epoch": 65.17,
-      "grad_norm": 5.924606800079346,
+      "grad_norm": 6.860241889953613,
       "learning_rate": 7.827827827827829e-06,
-      "loss": 1.2803,
+      "loss": 1.2773,
       "step": 21700
     },
     {
       "epoch": 65.2,
-      "grad_norm": 8.757498741149902,
+      "grad_norm": 8.44028091430664,
       "learning_rate": 7.826826826826828e-06,
-      "loss": 1.2912,
+      "loss": 1.2866,
       "step": 21710
     },
     {
       "epoch": 65.23,
-      "grad_norm": 8.386926651000977,
+      "grad_norm": 9.039767265319824,
       "learning_rate": 7.825825825825826e-06,
-      "loss": 1.2623,
+      "loss": 1.2535,
       "step": 21720
     },
     {
       "epoch": 65.26,
-      "grad_norm": 6.683221817016602,
+      "grad_norm": 6.407773971557617,
       "learning_rate": 7.824824824824827e-06,
-      "loss": 1.3033,
+      "loss": 1.2913,
       "step": 21730
     },
     {
       "epoch": 65.29,
-      "grad_norm": 5.9955620765686035,
+      "grad_norm": 7.98794412612915,
       "learning_rate": 7.823823823823823e-06,
-      "loss": 1.2991,
+      "loss": 1.2853,
       "step": 21740
     },
     {
       "epoch": 65.32,
-      "grad_norm": 6.2723069190979,
+      "grad_norm": 7.374836444854736,
       "learning_rate": 7.822822822822824e-06,
-      "loss": 1.2855,
+      "loss": 1.2832,
       "step": 21750
     },
     {
       "epoch": 65.35,
-      "grad_norm": 7.501558303833008,
+      "grad_norm": 7.262920379638672,
       "learning_rate": 7.821821821821822e-06,
-      "loss": 1.2806,
+      "loss": 1.2825,
       "step": 21760
     },
     {
       "epoch": 65.38,
-      "grad_norm": 8.383962631225586,
+      "grad_norm": 9.482568740844727,
       "learning_rate": 7.820820820820821e-06,
-      "loss": 1.201,
+      "loss": 1.2082,
       "step": 21770
     },
     {
       "epoch": 65.41,
-      "grad_norm": 8.671431541442871,
+      "grad_norm": 8.399386405944824,
       "learning_rate": 7.819819819819821e-06,
-      "loss": 1.3127,
+      "loss": 1.3173,
       "step": 21780
     },
     {
       "epoch": 65.44,
-      "grad_norm": 5.159196376800537,
+      "grad_norm": 6.708585262298584,
       "learning_rate": 7.81881881881882e-06,
-      "loss": 1.271,
+      "loss": 1.2681,
       "step": 21790
     },
     {
       "epoch": 65.47,
-      "grad_norm": 7.192973613739014,
+      "grad_norm": 7.726068019866943,
       "learning_rate": 7.817817817817818e-06,
-      "loss": 1.1809,
+      "loss": 1.1851,
       "step": 21800
     },
     {
       "epoch": 65.5,
-      "grad_norm": 8.927165031433105,
+      "grad_norm": 7.986786842346191,
       "learning_rate": 7.816816816816819e-06,
-      "loss": 1.254,
+      "loss": 1.2515,
       "step": 21810
     },
     {
       "epoch": 65.53,
-      "grad_norm": 5.521719932556152,
+      "grad_norm": 7.431150913238525,
       "learning_rate": 7.815815815815816e-06,
-      "loss": 1.2571,
+      "loss": 1.2552,
       "step": 21820
     },
     {
       "epoch": 65.56,
-      "grad_norm": 6.399598598480225,
+      "grad_norm": 5.172964096069336,
       "learning_rate": 7.814814814814816e-06,
-      "loss": 1.2343,
+      "loss": 1.2416,
       "step": 21830
     },
     {
       "epoch": 65.59,
-      "grad_norm": 9.24201488494873,
+      "grad_norm": 9.100997924804688,
       "learning_rate": 7.813813813813815e-06,
-      "loss": 1.2715,
+      "loss": 1.2734,
       "step": 21840
     },
     {
       "epoch": 65.62,
-      "grad_norm": 6.7843499183654785,
+      "grad_norm": 7.081577777862549,
       "learning_rate": 7.812812812812813e-06,
-      "loss": 1.3365,
+      "loss": 1.3244,
       "step": 21850
     },
     {
       "epoch": 65.65,
-      "grad_norm": 6.937866687774658,
+      "grad_norm": 6.811805248260498,
       "learning_rate": 7.811811811811812e-06,
-      "loss": 1.2657,
+      "loss": 1.2664,
       "step": 21860
     },
     {
       "epoch": 65.68,
-      "grad_norm": 8.348640441894531,
+      "grad_norm": 8.451925277709961,
       "learning_rate": 7.810810810810812e-06,
-      "loss": 1.3339,
+      "loss": 1.3378,
       "step": 21870
     },
     {
       "epoch": 65.71,
-      "grad_norm": 7.515899181365967,
+      "grad_norm": 9.364296913146973,
       "learning_rate": 7.80980980980981e-06,
-      "loss": 1.2901,
+      "loss": 1.3007,
       "step": 21880
     },
     {
       "epoch": 65.74,
-      "grad_norm": 8.108311653137207,
+      "grad_norm": 6.917169094085693,
       "learning_rate": 7.80880880880881e-06,
-      "loss": 1.2529,
+      "loss": 1.2494,
       "step": 21890
     },
     {
       "epoch": 65.77,
-      "grad_norm": 8.283560752868652,
+      "grad_norm": 7.675271034240723,
       "learning_rate": 7.807807807807808e-06,
-      "loss": 1.3366,
+      "loss": 1.3331,
       "step": 21900
     },
     {
       "epoch": 65.8,
-      "grad_norm": 9.575908660888672,
+      "grad_norm": 10.724727630615234,
       "learning_rate": 7.806806806806806e-06,
-      "loss": 1.3358,
+      "loss": 1.3317,
       "step": 21910
     },
     {
       "epoch": 65.83,
-      "grad_norm": 5.408636093139648,
+      "grad_norm": 6.080071449279785,
       "learning_rate": 7.805805805805807e-06,
-      "loss": 1.2231,
+      "loss": 1.2304,
       "step": 21920
     },
     {
       "epoch": 65.86,
-      "grad_norm": 10.038664817810059,
+      "grad_norm": 11.644023895263672,
       "learning_rate": 7.804804804804805e-06,
-      "loss": 1.3213,
+      "loss": 1.3208,
       "step": 21930
     },
     {
       "epoch": 65.89,
-      "grad_norm": 4.595243453979492,
+      "grad_norm": 7.149156093597412,
       "learning_rate": 7.803803803803804e-06,
-      "loss": 1.2987,
+      "loss": 1.2909,
       "step": 21940
     },
     {
       "epoch": 65.92,
-      "grad_norm": 8.34749698638916,
+      "grad_norm": 10.533097267150879,
       "learning_rate": 7.802802802802804e-06,
-      "loss": 1.2862,
+      "loss": 1.2921,
       "step": 21950
     },
     {
       "epoch": 65.95,
-      "grad_norm": 5.760272979736328,
+      "grad_norm": 6.618061065673828,
       "learning_rate": 7.801801801801803e-06,
-      "loss": 1.2803,
+      "loss": 1.2693,
       "step": 21960
     },
     {
       "epoch": 65.98,
-      "grad_norm": 7.017726421356201,
+      "grad_norm": 9.772726058959961,
       "learning_rate": 7.800800800800801e-06,
-      "loss": 1.2518,
+      "loss": 1.2494,
       "step": 21970
     },
     {
       "epoch": 66.0,
-      "eval_accuracy": 0.7773,
-      "eval_loss": 0.7300894260406494,
-      "eval_runtime": 5.6443,
-      "eval_samples_per_second": 1771.71,
-      "eval_steps_per_second": 7.087,
+      "eval_accuracy": 0.7763,
+      "eval_loss": 0.7290955185890198,
+      "eval_runtime": 5.4266,
+      "eval_samples_per_second": 1842.777,
+      "eval_steps_per_second": 7.371,
       "step": 21978
     },
     {
       "epoch": 66.01,
-      "grad_norm": 6.955756664276123,
+      "grad_norm": 10.447281837463379,
       "learning_rate": 7.799799799799802e-06,
-      "loss": 1.5253,
+      "loss": 1.5149,
       "step": 21980
     },
     {
       "epoch": 66.04,
-      "grad_norm": 6.534083366394043,
+      "grad_norm": 6.495058059692383,
       "learning_rate": 7.798798798798799e-06,
-      "loss": 1.2996,
+      "loss": 1.2964,
       "step": 21990
     },
     {
       "epoch": 66.07,
-      "grad_norm": 8.813070297241211,
+      "grad_norm": 10.012495994567871,
       "learning_rate": 7.797797797797799e-06,
-      "loss": 1.2791,
+      "loss": 1.2734,
       "step": 22000
     },
     {
       "epoch": 66.1,
-      "grad_norm": 5.549264907836914,
+      "grad_norm": 7.621478080749512,
       "learning_rate": 7.796796796796797e-06,
-      "loss": 1.2434,
+      "loss": 1.2511,
       "step": 22010
     },
     {
       "epoch": 66.13,
-      "grad_norm": 5.769726753234863,
+      "grad_norm": 8.133270263671875,
       "learning_rate": 7.795795795795796e-06,
-      "loss": 1.2595,
+      "loss": 1.2599,
       "step": 22020
     },
     {
       "epoch": 66.16,
-      "grad_norm": 6.820769309997559,
+      "grad_norm": 16.136112213134766,
       "learning_rate": 7.794794794794796e-06,
-      "loss": 1.3086,
+      "loss": 1.3179,
       "step": 22030
     },
     {
       "epoch": 66.19,
-      "grad_norm": 6.800881862640381,
+      "grad_norm": 6.721261978149414,
       "learning_rate": 7.793793793793795e-06,
-      "loss": 1.2143,
+      "loss": 1.2078,
       "step": 22040
     },
     {
       "epoch": 66.22,
-      "grad_norm": 6.7399749755859375,
+      "grad_norm": 5.949479103088379,
       "learning_rate": 7.792792792792793e-06,
-      "loss": 1.2238,
+      "loss": 1.2241,
       "step": 22050
     },
     {
       "epoch": 66.25,
-      "grad_norm": 6.81041145324707,
+      "grad_norm": 6.572335243225098,
       "learning_rate": 7.791791791791792e-06,
-      "loss": 1.2685,
+      "loss": 1.2742,
       "step": 22060
     },
     {
       "epoch": 66.28,
-      "grad_norm": 7.510166168212891,
+      "grad_norm": 6.917099475860596,
       "learning_rate": 7.79079079079079e-06,
-      "loss": 1.3007,
+      "loss": 1.2976,
       "step": 22070
     },
     {
       "epoch": 66.31,
-      "grad_norm": 6.412529468536377,
+      "grad_norm": 8.877958297729492,
       "learning_rate": 7.78978978978979e-06,
-      "loss": 1.339,
+      "loss": 1.3394,
       "step": 22080
     },
     {
       "epoch": 66.34,
-      "grad_norm": 7.153639316558838,
+      "grad_norm": 6.275571823120117,
       "learning_rate": 7.78878878878879e-06,
-      "loss": 1.3299,
+      "loss": 1.3242,
       "step": 22090
     },
     {
       "epoch": 66.37,
-      "grad_norm": 6.0709309577941895,
+      "grad_norm": 11.076676368713379,
       "learning_rate": 7.787787787787788e-06,
-      "loss": 1.2712,
+      "loss": 1.266,
       "step": 22100
     },
     {
       "epoch": 66.4,
-      "grad_norm": 7.484030246734619,
+      "grad_norm": 6.194124221801758,
       "learning_rate": 7.786786786786787e-06,
-      "loss": 1.3002,
+      "loss": 1.2859,
       "step": 22110
     },
     {
       "epoch": 66.43,
-      "grad_norm": 7.744346618652344,
+      "grad_norm": 6.470505714416504,
       "learning_rate": 7.785785785785787e-06,
-      "loss": 1.298,
+      "loss": 1.3015,
       "step": 22120
     },
     {
       "epoch": 66.46,
-      "grad_norm": 7.8170342445373535,
+      "grad_norm": 9.16744613647461,
       "learning_rate": 7.784784784784786e-06,
-      "loss": 1.285,
+      "loss": 1.2866,
       "step": 22130
     },
     {
       "epoch": 66.49,
-      "grad_norm": 9.276078224182129,
+      "grad_norm": 10.981568336486816,
       "learning_rate": 7.783783783783784e-06,
-      "loss": 1.2904,
+      "loss": 1.2873,
       "step": 22140
     },
     {
       "epoch": 66.52,
-      "grad_norm": 6.850980281829834,
+      "grad_norm": 9.88223934173584,
       "learning_rate": 7.782782782782783e-06,
-      "loss": 1.2674,
+      "loss": 1.2754,
       "step": 22150
     },
     {
       "epoch": 66.55,
-      "grad_norm": 7.727106094360352,
+      "grad_norm": 7.3433146476745605,
       "learning_rate": 7.781781781781781e-06,
-      "loss": 1.3482,
+      "loss": 1.3455,
       "step": 22160
     },
     {
       "epoch": 66.58,
-      "grad_norm": 8.794180870056152,
+      "grad_norm": 7.969354629516602,
       "learning_rate": 7.780780780780782e-06,
-      "loss": 1.3177,
+      "loss": 1.3005,
       "step": 22170
     },
     {
       "epoch": 66.61,
-      "grad_norm": 5.408536434173584,
+      "grad_norm": 5.5983500480651855,
       "learning_rate": 7.77977977977978e-06,
-      "loss": 1.2851,
+      "loss": 1.2778,
       "step": 22180
     },
     {
       "epoch": 66.64,
-      "grad_norm": 7.817933559417725,
+      "grad_norm": 8.174105644226074,
       "learning_rate": 7.778778778778779e-06,
-      "loss": 1.2729,
+      "loss": 1.2788,
       "step": 22190
     },
     {
       "epoch": 66.67,
-      "grad_norm": 7.388306617736816,
+      "grad_norm": 7.141423225402832,
       "learning_rate": 7.77777777777778e-06,
-      "loss": 1.3328,
+      "loss": 1.3194,
       "step": 22200
     },
     {
       "epoch": 66.7,
-      "grad_norm": 5.916398525238037,
+      "grad_norm": 8.857735633850098,
       "learning_rate": 7.776776776776778e-06,
-      "loss": 1.2709,
+      "loss": 1.2692,
       "step": 22210
     },
     {
       "epoch": 66.73,
-      "grad_norm": 6.836578845977783,
+      "grad_norm": 8.74809741973877,
       "learning_rate": 7.775775775775776e-06,
-      "loss": 1.2972,
+      "loss": 1.297,
       "step": 22220
     },
     {
       "epoch": 66.76,
-      "grad_norm": 9.07907485961914,
+      "grad_norm": 9.439127922058105,
       "learning_rate": 7.774774774774777e-06,
-      "loss": 1.2775,
+      "loss": 1.2768,
       "step": 22230
     },
     {
       "epoch": 66.79,
-      "grad_norm": 8.909562110900879,
+      "grad_norm": 12.38974666595459,
       "learning_rate": 7.773773773773774e-06,
-      "loss": 1.238,
+      "loss": 1.2374,
       "step": 22240
     },
     {
       "epoch": 66.82,
-      "grad_norm": 4.802634239196777,
+      "grad_norm": 5.643285274505615,
       "learning_rate": 7.772772772772774e-06,
-      "loss": 1.2115,
+      "loss": 1.2197,
       "step": 22250
     },
     {
       "epoch": 66.85,
-      "grad_norm": 6.115462779998779,
+      "grad_norm": 9.19237995147705,
       "learning_rate": 7.771771771771772e-06,
-      "loss": 1.2801,
+      "loss": 1.2898,
       "step": 22260
     },
     {
       "epoch": 66.88,
-      "grad_norm": 8.80433464050293,
+      "grad_norm": 9.956521987915039,
       "learning_rate": 7.770770770770771e-06,
-      "loss": 1.3096,
+      "loss": 1.3089,
       "step": 22270
     },
     {
       "epoch": 66.91,
-      "grad_norm": 8.089239120483398,
+      "grad_norm": 6.728933811187744,
       "learning_rate": 7.769769769769771e-06,
-      "loss": 1.212,
+      "loss": 1.2157,
       "step": 22280
     },
     {
       "epoch": 66.94,
-      "grad_norm": 6.0377397537231445,
+      "grad_norm": 8.50819206237793,
       "learning_rate": 7.76876876876877e-06,
       "loss": 1.335,
       "step": 22290
     },
     {
       "epoch": 66.97,
-      "grad_norm": 6.762013912200928,
+      "grad_norm": 6.597423076629639,
       "learning_rate": 7.767767767767769e-06,
-      "loss": 1.2196,
+      "loss": 1.2203,
       "step": 22300
     },
     {
       "epoch": 67.0,
-      "grad_norm": 5.688404560089111,
+      "grad_norm": 5.894289016723633,
       "learning_rate": 7.766766766766767e-06,
-      "loss": 1.2792,
+      "loss": 1.2821,
       "step": 22310
     },
     {
       "epoch": 67.0,
-      "eval_accuracy": 0.7762,
-      "eval_loss": 0.7285849452018738,
-      "eval_runtime": 5.435,
-      "eval_samples_per_second": 1839.926,
-      "eval_steps_per_second": 7.36,
+      "eval_accuracy": 0.7764,
+      "eval_loss": 0.7264572978019714,
+      "eval_runtime": 5.2759,
+      "eval_samples_per_second": 1895.404,
+      "eval_steps_per_second": 7.582,
       "step": 22311
     },
     {
       "epoch": 67.03,
-      "grad_norm": 10.72032356262207,
+      "grad_norm": 12.128493309020996,
       "learning_rate": 7.765765765765766e-06,
-      "loss": 1.2792,
+      "loss": 1.2646,
       "step": 22320
     },
     {
       "epoch": 67.06,
-      "grad_norm": 7.42353630065918,
+      "grad_norm": 7.916363716125488,
       "learning_rate": 7.764764764764764e-06,
-      "loss": 1.2384,
+      "loss": 1.2429,
       "step": 22330
     },
     {
       "epoch": 67.09,
-      "grad_norm": 7.809833526611328,
+      "grad_norm": 7.653628349304199,
       "learning_rate": 7.763763763763765e-06,
-      "loss": 1.2467,
+      "loss": 1.2449,
       "step": 22340
     },
     {
       "epoch": 67.12,
-      "grad_norm": 9.162717819213867,
+      "grad_norm": 8.498823165893555,
       "learning_rate": 7.762762762762763e-06,
-      "loss": 1.2733,
+      "loss": 1.2747,
       "step": 22350
     },
     {
       "epoch": 67.15,
-      "grad_norm": 8.887195587158203,
+      "grad_norm": 13.552321434020996,
       "learning_rate": 7.761761761761762e-06,
-      "loss": 1.3153,
+      "loss": 1.3157,
       "step": 22360
     },
     {
       "epoch": 67.18,
-      "grad_norm": 6.433666229248047,
+      "grad_norm": 7.981226444244385,
       "learning_rate": 7.760760760760762e-06,
-      "loss": 1.2298,
+      "loss": 1.2302,
       "step": 22370
     },
     {
       "epoch": 67.21,
-      "grad_norm": 8.48436164855957,
+      "grad_norm": 12.545455932617188,
       "learning_rate": 7.75975975975976e-06,
-      "loss": 1.3214,
+      "loss": 1.3086,
       "step": 22380
     },
     {
       "epoch": 67.24,
-      "grad_norm": 6.2399678230285645,
+      "grad_norm": 7.099421977996826,
       "learning_rate": 7.75875875875876e-06,
-      "loss": 1.2335,
+      "loss": 1.2499,
       "step": 22390
     },
     {
       "epoch": 67.27,
-      "grad_norm": 5.863649368286133,
+      "grad_norm": 6.86795711517334,
       "learning_rate": 7.757757757757758e-06,
-      "loss": 1.2767,
+      "loss": 1.2775,
       "step": 22400
     },
     {
       "epoch": 67.3,
-      "grad_norm": 6.633460998535156,
+      "grad_norm": 6.279576778411865,
       "learning_rate": 7.756756756756756e-06,
-      "loss": 1.3067,
+      "loss": 1.2956,
       "step": 22410
     },
     {
       "epoch": 67.33,
-      "grad_norm": 6.549717426300049,
+      "grad_norm": 10.257919311523438,
       "learning_rate": 7.755755755755757e-06,
-      "loss": 1.2793,
+      "loss": 1.2824,
       "step": 22420
     },
     {
       "epoch": 67.36,
-      "grad_norm": 7.1346330642700195,
+      "grad_norm": 7.175659656524658,
       "learning_rate": 7.754754754754755e-06,
-      "loss": 1.1955,
+      "loss": 1.1949,
       "step": 22430
     },
     {
       "epoch": 67.39,
-      "grad_norm": 5.255275726318359,
+      "grad_norm": 8.426278114318848,
       "learning_rate": 7.753753753753754e-06,
-      "loss": 1.2986,
+      "loss": 1.2978,
       "step": 22440
     },
     {
       "epoch": 67.42,
-      "grad_norm": 6.6996965408325195,
+      "grad_norm": 6.454070091247559,
       "learning_rate": 7.752752752752754e-06,
-      "loss": 1.2813,
+      "loss": 1.2909,
       "step": 22450
     },
     {
       "epoch": 67.45,
-      "grad_norm": 7.33607292175293,
+      "grad_norm": 7.546976089477539,
       "learning_rate": 7.751751751751753e-06,
-      "loss": 1.2527,
+      "loss": 1.2586,
       "step": 22460
     },
     {
       "epoch": 67.48,
-      "grad_norm": 8.697004318237305,
+      "grad_norm": 7.5664448738098145,
       "learning_rate": 7.750750750750751e-06,
-      "loss": 1.3005,
+      "loss": 1.2793,
       "step": 22470
     },
     {
       "epoch": 67.51,
-      "grad_norm": 6.248418807983398,
+      "grad_norm": 7.846802711486816,
       "learning_rate": 7.749749749749752e-06,
-      "loss": 1.3143,
+      "loss": 1.3208,
       "step": 22480
     },
     {
       "epoch": 67.54,
-      "grad_norm": 7.596677303314209,
+      "grad_norm": 7.0479230880737305,
       "learning_rate": 7.748748748748749e-06,
-      "loss": 1.2804,
+      "loss": 1.2691,
       "step": 22490
     },
     {
       "epoch": 67.57,
-      "grad_norm": 6.9614105224609375,
+      "grad_norm": 10.254098892211914,
       "learning_rate": 7.747747747747749e-06,
-      "loss": 1.3121,
+      "loss": 1.3156,
       "step": 22500
     },
     {
       "epoch": 67.6,
-      "grad_norm": 4.366170406341553,
+      "grad_norm": 5.772093296051025,
       "learning_rate": 7.746746746746747e-06,
-      "loss": 1.2794,
+      "loss": 1.2859,
       "step": 22510
     },
     {
       "epoch": 67.63,
-      "grad_norm": 10.854252815246582,
+      "grad_norm": 10.386444091796875,
       "learning_rate": 7.745745745745746e-06,
-      "loss": 1.3151,
+      "loss": 1.315,
       "step": 22520
     },
     {
       "epoch": 67.66,
-      "grad_norm": 8.975493431091309,
+      "grad_norm": 6.759740829467773,
       "learning_rate": 7.744744744744745e-06,
-      "loss": 1.2091,
+      "loss": 1.2143,
       "step": 22530
     },
     {
       "epoch": 67.69,
-      "grad_norm": 6.614103317260742,
+      "grad_norm": 8.608197212219238,
       "learning_rate": 7.743743743743745e-06,
-      "loss": 1.2043,
+      "loss": 1.2045,
       "step": 22540
     },
     {
       "epoch": 67.72,
-      "grad_norm": 6.047145843505859,
+      "grad_norm": 7.37396240234375,
       "learning_rate": 7.742742742742744e-06,
-      "loss": 1.282,
+      "loss": 1.2758,
       "step": 22550
     },
     {
       "epoch": 67.75,
-      "grad_norm": 6.45681619644165,
+      "grad_norm": 5.383072376251221,
       "learning_rate": 7.741741741741742e-06,
-      "loss": 1.1711,
+      "loss": 1.1621,
       "step": 22560
     },
     {
       "epoch": 67.78,
-      "grad_norm": 6.931977272033691,
+      "grad_norm": 8.07455062866211,
       "learning_rate": 7.74074074074074e-06,
-      "loss": 1.2441,
+      "loss": 1.2351,
       "step": 22570
     },
     {
       "epoch": 67.81,
-      "grad_norm": 6.39780855178833,
+      "grad_norm": 6.5679802894592285,
       "learning_rate": 7.73973973973974e-06,
-      "loss": 1.2039,
+      "loss": 1.2084,
       "step": 22580
     },
     {
       "epoch": 67.84,
-      "grad_norm": 6.148879051208496,
+      "grad_norm": 7.870532512664795,
       "learning_rate": 7.73873873873874e-06,
-      "loss": 1.3463,
+      "loss": 1.3501,
       "step": 22590
     },
     {
       "epoch": 67.87,
-      "grad_norm": 6.509223937988281,
+      "grad_norm": 7.325404167175293,
       "learning_rate": 7.737737737737738e-06,
-      "loss": 1.2795,
+      "loss": 1.2644,
       "step": 22600
     },
     {
       "epoch": 67.9,
-      "grad_norm": 9.043903350830078,
+      "grad_norm": 9.497194290161133,
       "learning_rate": 7.736736736736737e-06,
-      "loss": 1.1854,
+      "loss": 1.1838,
       "step": 22610
     },
     {
       "epoch": 67.93,
-      "grad_norm": 9.422332763671875,
+      "grad_norm": 10.336127281188965,
       "learning_rate": 7.735735735735737e-06,
-      "loss": 1.2664,
+      "loss": 1.259,
       "step": 22620
     },
     {
       "epoch": 67.96,
-      "grad_norm": 6.229498863220215,
+      "grad_norm": 5.072948455810547,
       "learning_rate": 7.734734734734736e-06,
-      "loss": 1.2015,
+      "loss": 1.2025,
       "step": 22630
     },
     {
       "epoch": 67.99,
-      "grad_norm": 4.988348484039307,
+      "grad_norm": 5.148833751678467,
       "learning_rate": 7.733733733733734e-06,
-      "loss": 1.2137,
+      "loss": 1.1994,
       "step": 22640
     },
     {
       "epoch": 68.0,
-      "eval_accuracy": 0.7781,
-      "eval_loss": 0.7209867835044861,
-      "eval_runtime": 5.6225,
-      "eval_samples_per_second": 1778.559,
-      "eval_steps_per_second": 7.114,
+      "eval_accuracy": 0.7753,
+      "eval_loss": 0.7221060395240784,
+      "eval_runtime": 5.2394,
+      "eval_samples_per_second": 1908.613,
+      "eval_steps_per_second": 7.634,
       "step": 22644
     },
     {
       "epoch": 68.02,
-      "grad_norm": 7.649541854858398,
+      "grad_norm": 5.58429479598999,
       "learning_rate": 7.732732732732733e-06,
-      "loss": 1.5442,
+      "loss": 1.5185,
       "step": 22650
     },
     {
       "epoch": 68.05,
-      "grad_norm": 6.736617088317871,
+      "grad_norm": 8.837984085083008,
       "learning_rate": 7.731731731731731e-06,
-      "loss": 1.2868,
+      "loss": 1.2905,
       "step": 22660
     },
     {
       "epoch": 68.08,
-      "grad_norm": 8.556500434875488,
+      "grad_norm": 7.668012619018555,
       "learning_rate": 7.730730730730732e-06,
-      "loss": 1.3173,
+      "loss": 1.3206,
       "step": 22670
     },
     {
       "epoch": 68.11,
-      "grad_norm": 6.975805282592773,
+      "grad_norm": 9.192094802856445,
       "learning_rate": 7.72972972972973e-06,
-      "loss": 1.2447,
+      "loss": 1.2581,
       "step": 22680
     },
     {
       "epoch": 68.14,
-      "grad_norm": 6.671439170837402,
+      "grad_norm": 9.62798023223877,
       "learning_rate": 7.728728728728729e-06,
-      "loss": 1.2481,
+      "loss": 1.2473,
       "step": 22690
     },
     {
       "epoch": 68.17,
-      "grad_norm": 7.907862663269043,
+      "grad_norm": 8.365957260131836,
       "learning_rate": 7.72772772772773e-06,
-      "loss": 1.3045,
+      "loss": 1.3048,
       "step": 22700
     },
     {
       "epoch": 68.2,
-      "grad_norm": 5.410658359527588,
+      "grad_norm": 7.212541103363037,
       "learning_rate": 7.726726726726728e-06,
-      "loss": 1.314,
+      "loss": 1.2986,
       "step": 22710
     },
     {
       "epoch": 68.23,
-      "grad_norm": 5.608346939086914,
+      "grad_norm": 6.490209579467773,
       "learning_rate": 7.725725725725726e-06,
-      "loss": 1.277,
+      "loss": 1.2856,
       "step": 22720
     },
     {
       "epoch": 68.26,
-      "grad_norm": 6.5115180015563965,
+      "grad_norm": 7.6199164390563965,
       "learning_rate": 7.724724724724727e-06,
-      "loss": 1.2803,
+      "loss": 1.2883,
       "step": 22730
     },
     {
       "epoch": 68.29,
-      "grad_norm": 8.427542686462402,
+      "grad_norm": 10.005702018737793,
       "learning_rate": 7.723723723723724e-06,
-      "loss": 1.2702,
+      "loss": 1.2804,
       "step": 22740
     },
     {
       "epoch": 68.32,
-      "grad_norm": 7.6060099601745605,
+      "grad_norm": 6.984982490539551,
       "learning_rate": 7.722722722722722e-06,
-      "loss": 1.2862,
+      "loss": 1.2906,
       "step": 22750
     },
     {
       "epoch": 68.35,
-      "grad_norm": 6.641902923583984,
+      "grad_norm": 7.671027183532715,
       "learning_rate": 7.721721721721722e-06,
-      "loss": 1.2651,
+      "loss": 1.2549,
       "step": 22760
     },
     {
       "epoch": 68.38,
-      "grad_norm": 7.159088611602783,
+      "grad_norm": 7.260354995727539,
       "learning_rate": 7.720720720720721e-06,
-      "loss": 1.2714,
+      "loss": 1.2647,
       "step": 22770
     },
     {
       "epoch": 68.41,
-      "grad_norm": 6.374092102050781,
+      "grad_norm": 7.096216678619385,
       "learning_rate": 7.71971971971972e-06,
-      "loss": 1.2339,
+      "loss": 1.2383,
       "step": 22780
     },
     {
       "epoch": 68.44,
-      "grad_norm": 7.653698921203613,
+      "grad_norm": 7.351322650909424,
       "learning_rate": 7.71871871871872e-06,
-      "loss": 1.2283,
+      "loss": 1.2228,
       "step": 22790
     },
     {
       "epoch": 68.47,
-      "grad_norm": 6.896973609924316,
+      "grad_norm": 7.746832370758057,
       "learning_rate": 7.717717717717719e-06,
-      "loss": 1.2705,
+      "loss": 1.2677,
       "step": 22800
     },
     {
       "epoch": 68.5,
-      "grad_norm": 5.632826328277588,
+      "grad_norm": 6.4247283935546875,
       "learning_rate": 7.716716716716717e-06,
-      "loss": 1.2633,
+      "loss": 1.2592,
       "step": 22810
     },
     {
       "epoch": 68.53,
-      "grad_norm": 7.664337158203125,
+      "grad_norm": 6.014052867889404,
       "learning_rate": 7.715715715715716e-06,
-      "loss": 1.2664,
+      "loss": 1.2673,
       "step": 22820
     },
     {
       "epoch": 68.56,
-      "grad_norm": 8.904424667358398,
+      "grad_norm": 7.988062381744385,
       "learning_rate": 7.714714714714714e-06,
-      "loss": 1.2134,
+      "loss": 1.2141,
       "step": 22830
     },
     {
       "epoch": 68.59,
-      "grad_norm": 8.781312942504883,
+      "grad_norm": 7.486091136932373,
       "learning_rate": 7.713713713713715e-06,
-      "loss": 1.2972,
+      "loss": 1.2906,
       "step": 22840
     },
     {
       "epoch": 68.62,
-      "grad_norm": 6.574685096740723,
+      "grad_norm": 7.878647327423096,
       "learning_rate": 7.712712712712713e-06,
-      "loss": 1.32,
+      "loss": 1.3149,
       "step": 22850
     },
     {
       "epoch": 68.65,
-      "grad_norm": 9.825618743896484,
+      "grad_norm": 8.316675186157227,
       "learning_rate": 7.711711711711712e-06,
-      "loss": 1.2875,
+      "loss": 1.284,
       "step": 22860
     },
     {
       "epoch": 68.68,
-      "grad_norm": 6.525233268737793,
+      "grad_norm": 7.475057601928711,
       "learning_rate": 7.710710710710712e-06,
-      "loss": 1.209,
+      "loss": 1.2143,
       "step": 22870
     },
     {
       "epoch": 68.71,
-      "grad_norm": 9.75232982635498,
+      "grad_norm": 11.214247703552246,
       "learning_rate": 7.70970970970971e-06,
-      "loss": 1.3211,
+      "loss": 1.3194,
       "step": 22880
     },
     {
       "epoch": 68.74,
-      "grad_norm": 7.29796838760376,
+      "grad_norm": 7.244987964630127,
       "learning_rate": 7.70870870870871e-06,
-      "loss": 1.2068,
+      "loss": 1.2002,
       "step": 22890
     },
     {
       "epoch": 68.77,
-      "grad_norm": 5.75739860534668,
+      "grad_norm": 7.448144912719727,
       "learning_rate": 7.707707707707708e-06,
-      "loss": 1.2245,
+      "loss": 1.2221,
       "step": 22900
     },
     {
       "epoch": 68.8,
-      "grad_norm": 10.475821495056152,
+      "grad_norm": 11.211289405822754,
       "learning_rate": 7.706706706706707e-06,
-      "loss": 1.3194,
+      "loss": 1.3252,
       "step": 22910
     },
     {
       "epoch": 68.83,
-      "grad_norm": 8.361915588378906,
+      "grad_norm": 8.87155818939209,
       "learning_rate": 7.705705705705707e-06,
-      "loss": 1.2231,
+      "loss": 1.2189,
       "step": 22920
     },
     {
       "epoch": 68.86,
-      "grad_norm": 5.480363368988037,
+      "grad_norm": 8.344807624816895,
       "learning_rate": 7.704704704704705e-06,
-      "loss": 1.2781,
+      "loss": 1.2772,
       "step": 22930
     },
     {
       "epoch": 68.89,
-      "grad_norm": 7.833707809448242,
+      "grad_norm": 7.534799098968506,
       "learning_rate": 7.703703703703704e-06,
-      "loss": 1.21,
+      "loss": 1.2082,
       "step": 22940
     },
     {
       "epoch": 68.92,
-      "grad_norm": 7.144121170043945,
+      "grad_norm": 7.714147567749023,
       "learning_rate": 7.702702702702704e-06,
-      "loss": 1.2522,
+      "loss": 1.2468,
       "step": 22950
     },
     {
       "epoch": 68.95,
-      "grad_norm": 6.060871124267578,
+      "grad_norm": 6.145423889160156,
       "learning_rate": 7.701701701701703e-06,
-      "loss": 1.2599,
+      "loss": 1.252,
       "step": 22960
     },
     {
       "epoch": 68.98,
-      "grad_norm": 7.205877780914307,
+      "grad_norm": 5.99940299987793,
       "learning_rate": 7.700700700700701e-06,
-      "loss": 1.2598,
+      "loss": 1.2569,
       "step": 22970
     },
     {
       "epoch": 69.0,
-      "eval_accuracy": 0.7784,
-      "eval_loss": 0.7218632102012634,
-      "eval_runtime": 5.6717,
-      "eval_samples_per_second": 1763.129,
-      "eval_steps_per_second": 7.053,
+      "eval_accuracy": 0.7792,
+      "eval_loss": 0.7200848460197449,
+      "eval_runtime": 5.3253,
+      "eval_samples_per_second": 1877.828,
+      "eval_steps_per_second": 7.511,
       "step": 22977
     },
     {
       "epoch": 69.01,
-      "grad_norm": 5.87339973449707,
+      "grad_norm": 5.781588077545166,
       "learning_rate": 7.6996996996997e-06,
-      "loss": 1.3892,
+      "loss": 1.4056,
       "step": 22980
     },
     {
       "epoch": 69.04,
-      "grad_norm": 7.456140995025635,
+      "grad_norm": 6.200134754180908,
       "learning_rate": 7.698698698698699e-06,
-      "loss": 1.2519,
+      "loss": 1.254,
       "step": 22990
     },
     {
       "epoch": 69.07,
-      "grad_norm": 11.329538345336914,
+      "grad_norm": 7.788739204406738,
       "learning_rate": 7.697697697697697e-06,
-      "loss": 1.2398,
+      "loss": 1.2296,
       "step": 23000
     },
     {
       "epoch": 69.1,
-      "grad_norm": 5.837748050689697,
+      "grad_norm": 5.458225250244141,
       "learning_rate": 7.696696696696698e-06,
-      "loss": 1.2614,
+      "loss": 1.2516,
       "step": 23010
     },
     {
       "epoch": 69.13,
-      "grad_norm": 6.773569107055664,
+      "grad_norm": 6.000741958618164,
       "learning_rate": 7.695695695695696e-06,
-      "loss": 1.2144,
+      "loss": 1.2071,
       "step": 23020
     },
     {
       "epoch": 69.16,
-      "grad_norm": 5.665299415588379,
+      "grad_norm": 6.88495397567749,
       "learning_rate": 7.694694694694695e-06,
-      "loss": 1.2404,
+      "loss": 1.234,
       "step": 23030
     },
     {
       "epoch": 69.19,
-      "grad_norm": 10.114587783813477,
+      "grad_norm": 7.706164836883545,
       "learning_rate": 7.693693693693695e-06,
-      "loss": 1.2162,
+      "loss": 1.2164,
       "step": 23040
     },
     {
       "epoch": 69.22,
-      "grad_norm": 6.559414863586426,
+      "grad_norm": 8.46423625946045,
       "learning_rate": 7.692692692692694e-06,
-      "loss": 1.2683,
+      "loss": 1.2705,
       "step": 23050
     },
     {
       "epoch": 69.25,
-      "grad_norm": 6.207386016845703,
+      "grad_norm": 8.467294692993164,
       "learning_rate": 7.691691691691692e-06,
-      "loss": 1.1967,
+      "loss": 1.1962,
       "step": 23060
     },
     {
       "epoch": 69.28,
-      "grad_norm": 8.920671463012695,
+      "grad_norm": 9.035325050354004,
       "learning_rate": 7.69069069069069e-06,
-      "loss": 1.2384,
+      "loss": 1.2451,
       "step": 23070
     },
     {
       "epoch": 69.31,
-      "grad_norm": 10.60353946685791,
+      "grad_norm": 12.395087242126465,
       "learning_rate": 7.68968968968969e-06,
-      "loss": 1.2344,
+      "loss": 1.2214,
       "step": 23080
     },
     {
       "epoch": 69.34,
-      "grad_norm": 5.534912586212158,
+      "grad_norm": 5.5841569900512695,
       "learning_rate": 7.68868868868869e-06,
-      "loss": 1.2515,
+      "loss": 1.2495,
       "step": 23090
     },
     {
       "epoch": 69.37,
-      "grad_norm": 11.33797550201416,
+      "grad_norm": 19.23575210571289,
       "learning_rate": 7.687687687687688e-06,
-      "loss": 1.2368,
+      "loss": 1.2317,
       "step": 23100
     },
     {
       "epoch": 69.4,
-      "grad_norm": 6.697115898132324,
+      "grad_norm": 8.593292236328125,
       "learning_rate": 7.686686686686687e-06,
-      "loss": 1.2708,
+      "loss": 1.2714,
       "step": 23110
     },
     {
       "epoch": 69.43,
-      "grad_norm": 6.144682884216309,
+      "grad_norm": 7.033552169799805,
       "learning_rate": 7.685685685685687e-06,
-      "loss": 1.2374,
+      "loss": 1.2307,
       "step": 23120
     },
     {
       "epoch": 69.46,
-      "grad_norm": 4.756547927856445,
+      "grad_norm": 5.933164119720459,
       "learning_rate": 7.684684684684686e-06,
-      "loss": 1.2208,
+      "loss": 1.2121,
       "step": 23130
     },
     {
       "epoch": 69.49,
-      "grad_norm": 5.654446125030518,
+      "grad_norm": 6.551761627197266,
       "learning_rate": 7.683683683683684e-06,
-      "loss": 1.2949,
+      "loss": 1.2888,
       "step": 23140
     },
     {
       "epoch": 69.52,
-      "grad_norm": 6.753251552581787,
+      "grad_norm": 10.077223777770996,
       "learning_rate": 7.682682682682683e-06,
-      "loss": 1.2032,
+      "loss": 1.2111,
       "step": 23150
     },
     {
       "epoch": 69.55,
-      "grad_norm": 7.549383640289307,
+      "grad_norm": 8.485939025878906,
       "learning_rate": 7.681681681681682e-06,
-      "loss": 1.2338,
+      "loss": 1.2431,
       "step": 23160
     },
     {
       "epoch": 69.58,
-      "grad_norm": 6.944180965423584,
+      "grad_norm": 7.450997352600098,
       "learning_rate": 7.680680680680682e-06,
-      "loss": 1.227,
+      "loss": 1.2281,
       "step": 23170
     },
     {
       "epoch": 69.61,
-      "grad_norm": 6.4485626220703125,
+      "grad_norm": 9.021318435668945,
       "learning_rate": 7.67967967967968e-06,
-      "loss": 1.2497,
+      "loss": 1.2406,
       "step": 23180
     },
     {
       "epoch": 69.64,
-      "grad_norm": 7.488152027130127,
+      "grad_norm": 6.068024635314941,
       "learning_rate": 7.678678678678679e-06,
-      "loss": 1.2408,
+      "loss": 1.2387,
       "step": 23190
     },
     {
       "epoch": 69.67,
-      "grad_norm": 8.739192008972168,
+      "grad_norm": 7.208529472351074,
       "learning_rate": 7.67767767767768e-06,
-      "loss": 1.2504,
+      "loss": 1.2568,
       "step": 23200
     },
     {
       "epoch": 69.7,
-      "grad_norm": 5.790533065795898,
+      "grad_norm": 6.650362968444824,
       "learning_rate": 7.676676676676678e-06,
-      "loss": 1.2248,
+      "loss": 1.2303,
       "step": 23210
     },
     {
       "epoch": 69.73,
-      "grad_norm": 7.104800701141357,
+      "grad_norm": 6.014946937561035,
       "learning_rate": 7.675675675675676e-06,
-      "loss": 1.2222,
+      "loss": 1.2291,
       "step": 23220
     },
     {
       "epoch": 69.76,
-      "grad_norm": 6.279422283172607,
+      "grad_norm": 8.527204513549805,
       "learning_rate": 7.674674674674675e-06,
-      "loss": 1.2765,
+      "loss": 1.2805,
       "step": 23230
     },
     {
       "epoch": 69.79,
-      "grad_norm": 6.822751998901367,
+      "grad_norm": 6.919895172119141,
       "learning_rate": 7.673673673673674e-06,
-      "loss": 1.3009,
+      "loss": 1.2817,
       "step": 23240
     },
     {
       "epoch": 69.82,
-      "grad_norm": 8.04667854309082,
+      "grad_norm": 10.724230766296387,
       "learning_rate": 7.672672672672672e-06,
-      "loss": 1.2454,
+      "loss": 1.2479,
       "step": 23250
     },
     {
       "epoch": 69.85,
-      "grad_norm": 11.929892539978027,
+      "grad_norm": 9.855415344238281,
       "learning_rate": 7.671671671671673e-06,
-      "loss": 1.2186,
+      "loss": 1.2229,
       "step": 23260
     },
     {
       "epoch": 69.88,
-      "grad_norm": 7.23408842086792,
+      "grad_norm": 9.580161094665527,
       "learning_rate": 7.670670670670671e-06,
-      "loss": 1.2362,
+      "loss": 1.2333,
       "step": 23270
     },
     {
       "epoch": 69.91,
-      "grad_norm": 9.167101860046387,
+      "grad_norm": 9.891902923583984,
       "learning_rate": 7.66966966966967e-06,
-      "loss": 1.2669,
+      "loss": 1.2565,
       "step": 23280
     },
     {
       "epoch": 69.94,
-      "grad_norm": 6.757350921630859,
+      "grad_norm": 7.182357311248779,
       "learning_rate": 7.66866866866867e-06,
-      "loss": 1.2754,
+      "loss": 1.2817,
       "step": 23290
     },
     {
       "epoch": 69.97,
-      "grad_norm": 6.8629937171936035,
+      "grad_norm": 7.305396556854248,
       "learning_rate": 7.667667667667669e-06,
-      "loss": 1.248,
+      "loss": 1.2589,
       "step": 23300
     },
     {
       "epoch": 70.0,
-      "grad_norm": 37.53722381591797,
+      "grad_norm": 59.989463806152344,
       "learning_rate": 7.666666666666667e-06,
-      "loss": 1.4021,
+      "loss": 1.4258,
       "step": 23310
     },
     {
       "epoch": 70.0,
-      "eval_accuracy": 0.7803,
-      "eval_loss": 0.7204360961914062,
-      "eval_runtime": 5.3959,
-      "eval_samples_per_second": 1853.248,
-      "eval_steps_per_second": 7.413,
+      "eval_accuracy": 0.7782,
+      "eval_loss": 0.7183724641799927,
+      "eval_runtime": 5.3533,
+      "eval_samples_per_second": 1868.018,
+      "eval_steps_per_second": 7.472,
       "step": 23310
     },
     {
       "epoch": 70.03,
-      "grad_norm": 6.544050693511963,
+      "grad_norm": 6.946579456329346,
       "learning_rate": 7.665665665665666e-06,
-      "loss": 1.2545,
+      "loss": 1.2574,
       "step": 23320
     },
     {
       "epoch": 70.06,
-      "grad_norm": 6.240963459014893,
+      "grad_norm": 7.700075626373291,
       "learning_rate": 7.664664664664664e-06,
-      "loss": 1.2912,
+      "loss": 1.2858,
       "step": 23330
     },
     {
       "epoch": 70.09,
-      "grad_norm": 6.127787113189697,
+      "grad_norm": 5.82311487197876,
       "learning_rate": 7.663663663663665e-06,
-      "loss": 1.2028,
+      "loss": 1.2057,
       "step": 23340
     },
     {
       "epoch": 70.12,
-      "grad_norm": 9.1051607131958,
+      "grad_norm": 10.300418853759766,
       "learning_rate": 7.662662662662663e-06,
-      "loss": 1.2077,
+      "loss": 1.2075,
       "step": 23350
     },
     {
       "epoch": 70.15,
-      "grad_norm": 4.4838547706604,
+      "grad_norm": 7.484643459320068,
       "learning_rate": 7.661661661661662e-06,
-      "loss": 1.2736,
+      "loss": 1.2656,
       "step": 23360
     },
     {
       "epoch": 70.18,
-      "grad_norm": 6.984890937805176,
+      "grad_norm": 8.226691246032715,
       "learning_rate": 7.660660660660662e-06,
-      "loss": 1.2746,
+      "loss": 1.2802,
       "step": 23370
     },
     {
       "epoch": 70.21,
-      "grad_norm": 5.37226676940918,
+      "grad_norm": 9.272398948669434,
       "learning_rate": 7.65965965965966e-06,
-      "loss": 1.2355,
+      "loss": 1.2253,
       "step": 23380
     },
     {
       "epoch": 70.24,
-      "grad_norm": 7.205860614776611,
+      "grad_norm": 9.328985214233398,
       "learning_rate": 7.65865865865866e-06,
-      "loss": 1.3122,
+      "loss": 1.3121,
       "step": 23390
     },
     {
       "epoch": 70.27,
-      "grad_norm": 7.261763572692871,
+      "grad_norm": 15.029093742370605,
       "learning_rate": 7.657657657657658e-06,
-      "loss": 1.2473,
+      "loss": 1.2325,
       "step": 23400
     },
     {
       "epoch": 70.3,
-      "grad_norm": 5.397762298583984,
+      "grad_norm": 5.550683498382568,
       "learning_rate": 7.656656656656657e-06,
-      "loss": 1.2721,
+      "loss": 1.2784,
       "step": 23410
     },
     {
       "epoch": 70.33,
-      "grad_norm": 9.291464805603027,
+      "grad_norm": 8.802443504333496,
       "learning_rate": 7.655655655655657e-06,
-      "loss": 1.194,
+      "loss": 1.1961,
       "step": 23420
     },
     {
       "epoch": 70.36,
-      "grad_norm": 7.371936798095703,
+      "grad_norm": 8.523764610290527,
       "learning_rate": 7.654654654654655e-06,
-      "loss": 1.2728,
+      "loss": 1.2661,
       "step": 23430
     },
     {
       "epoch": 70.39,
-      "grad_norm": 7.1210126876831055,
+      "grad_norm": 9.373953819274902,
       "learning_rate": 7.653653653653654e-06,
-      "loss": 1.2257,
+      "loss": 1.2255,
       "step": 23440
     },
     {
       "epoch": 70.42,
-      "grad_norm": 6.572276592254639,
+      "grad_norm": 8.185187339782715,
       "learning_rate": 7.652652652652653e-06,
-      "loss": 1.2945,
+      "loss": 1.2827,
       "step": 23450
     },
     {
       "epoch": 70.45,
-      "grad_norm": 6.618112087249756,
+      "grad_norm": 6.31194543838501,
       "learning_rate": 7.651651651651653e-06,
-      "loss": 1.1788,
+      "loss": 1.1792,
       "step": 23460
     },
     {
       "epoch": 70.48,
-      "grad_norm": 7.7188496589660645,
+      "grad_norm": 9.436982154846191,
       "learning_rate": 7.650650650650652e-06,
-      "loss": 1.218,
+      "loss": 1.2231,
       "step": 23470
     },
     {
       "epoch": 70.51,
-      "grad_norm": 6.5771989822387695,
+      "grad_norm": 6.538239479064941,
       "learning_rate": 7.64964964964965e-06,
-      "loss": 1.2419,
+      "loss": 1.2365,
       "step": 23480
     },
     {
       "epoch": 70.54,
-      "grad_norm": 7.616140365600586,
+      "grad_norm": 7.834970474243164,
       "learning_rate": 7.648648648648649e-06,
-      "loss": 1.3287,
+      "loss": 1.3311,
       "step": 23490
     },
     {
       "epoch": 70.57,
-      "grad_norm": 8.49644660949707,
+      "grad_norm": 7.496759414672852,
       "learning_rate": 7.647647647647647e-06,
-      "loss": 1.2269,
+      "loss": 1.2184,
       "step": 23500
     },
     {
       "epoch": 70.6,
-      "grad_norm": 11.910633087158203,
+      "grad_norm": 7.7832794189453125,
       "learning_rate": 7.646646646646648e-06,
-      "loss": 1.2796,
+      "loss": 1.2787,
       "step": 23510
     },
     {
       "epoch": 70.63,
-      "grad_norm": 9.20281982421875,
+      "grad_norm": 8.78111457824707,
       "learning_rate": 7.645645645645646e-06,
-      "loss": 1.2092,
+      "loss": 1.2149,
       "step": 23520
     },
     {
       "epoch": 70.66,
-      "grad_norm": 6.102070331573486,
+      "grad_norm": 12.70018482208252,
       "learning_rate": 7.644644644644645e-06,
-      "loss": 1.1976,
+      "loss": 1.2009,
       "step": 23530
     },
     {
       "epoch": 70.69,
-      "grad_norm": 5.690670490264893,
+      "grad_norm": 7.184034824371338,
       "learning_rate": 7.643643643643645e-06,
-      "loss": 1.2587,
+      "loss": 1.2538,
       "step": 23540
     },
     {
       "epoch": 70.72,
-      "grad_norm": 9.146788597106934,
+      "grad_norm": 9.465507507324219,
       "learning_rate": 7.642642642642644e-06,
-      "loss": 1.2598,
+      "loss": 1.2544,
       "step": 23550
     },
     {
       "epoch": 70.75,
-      "grad_norm": 5.370345592498779,
+      "grad_norm": 8.5006103515625,
       "learning_rate": 7.641641641641642e-06,
-      "loss": 1.2153,
+      "loss": 1.2213,
       "step": 23560
     },
     {
       "epoch": 70.78,
-      "grad_norm": 6.144442081451416,
+      "grad_norm": 7.167893886566162,
       "learning_rate": 7.640640640640641e-06,
-      "loss": 1.2648,
+      "loss": 1.2655,
       "step": 23570
     },
     {
       "epoch": 70.81,
-      "grad_norm": 6.179625034332275,
+      "grad_norm": 8.374276161193848,
       "learning_rate": 7.63963963963964e-06,
-      "loss": 1.2393,
+      "loss": 1.2532,
       "step": 23580
     },
     {
       "epoch": 70.84,
-      "grad_norm": 6.869839191436768,
+      "grad_norm": 6.348972797393799,
       "learning_rate": 7.63863863863864e-06,
-      "loss": 1.2284,
+      "loss": 1.2282,
       "step": 23590
     },
     {
       "epoch": 70.87,
-      "grad_norm": 6.539061546325684,
+      "grad_norm": 5.5424885749816895,
       "learning_rate": 7.637637637637638e-06,
-      "loss": 1.2096,
+      "loss": 1.2023,
       "step": 23600
     },
     {
       "epoch": 70.9,
-      "grad_norm": 7.744644641876221,
+      "grad_norm": 8.769055366516113,
       "learning_rate": 7.636636636636637e-06,
-      "loss": 1.2741,
+      "loss": 1.2733,
       "step": 23610
     },
     {
       "epoch": 70.93,
-      "grad_norm": 12.191372871398926,
+      "grad_norm": 6.811972618103027,
       "learning_rate": 7.635635635635637e-06,
-      "loss": 1.2244,
+      "loss": 1.2281,
       "step": 23620
     },
     {
       "epoch": 70.96,
-      "grad_norm": 6.154479026794434,
+      "grad_norm": 7.422540187835693,
       "learning_rate": 7.634634634634636e-06,
-      "loss": 1.2631,
+      "loss": 1.2484,
       "step": 23630
     },
     {
       "epoch": 70.99,
-      "grad_norm": 6.325480937957764,
+      "grad_norm": 6.255156517028809,
       "learning_rate": 7.633633633633634e-06,
-      "loss": 1.3074,
+      "loss": 1.3091,
       "step": 23640
     },
     {
       "epoch": 71.0,
-      "eval_accuracy": 0.7822,
-      "eval_loss": 0.7066367268562317,
-      "eval_runtime": 5.418,
-      "eval_samples_per_second": 1845.713,
-      "eval_steps_per_second": 7.383,
+      "eval_accuracy": 0.7828,
+      "eval_loss": 0.7056503891944885,
+      "eval_runtime": 5.3194,
+      "eval_samples_per_second": 1879.904,
+      "eval_steps_per_second": 7.52,
       "step": 23643
     },
     {
       "epoch": 71.02,
-      "grad_norm": 6.198880195617676,
+      "grad_norm": 6.620068073272705,
       "learning_rate": 7.632632632632633e-06,
-      "loss": 1.3513,
+      "loss": 1.3512,
       "step": 23650
     },
     {
       "epoch": 71.05,
-      "grad_norm": 8.153656005859375,
+      "grad_norm": 6.832465648651123,
       "learning_rate": 7.631631631631632e-06,
-      "loss": 1.2315,
+      "loss": 1.2208,
       "step": 23660
     },
     {
       "epoch": 71.08,
-      "grad_norm": 9.370953559875488,
+      "grad_norm": 9.626004219055176,
       "learning_rate": 7.63063063063063e-06,
-      "loss": 1.2344,
+      "loss": 1.2289,
       "step": 23670
     },
     {
       "epoch": 71.11,
-      "grad_norm": 7.335353374481201,
+      "grad_norm": 6.885151386260986,
       "learning_rate": 7.62962962962963e-06,
-      "loss": 1.2932,
+      "loss": 1.2908,
       "step": 23680
     },
     {
       "epoch": 71.14,
-      "grad_norm": 7.1511125564575195,
+      "grad_norm": 6.231815338134766,
       "learning_rate": 7.628628628628629e-06,
-      "loss": 1.3138,
+      "loss": 1.3064,
       "step": 23690
     },
     {
       "epoch": 71.17,
-      "grad_norm": 8.526128768920898,
+      "grad_norm": 11.518723487854004,
       "learning_rate": 7.6276276276276285e-06,
-      "loss": 1.286,
+      "loss": 1.285,
       "step": 23700
     },
     {
       "epoch": 71.2,
-      "grad_norm": 5.934676647186279,
+      "grad_norm": 5.9754767417907715,
       "learning_rate": 7.626626626626628e-06,
-      "loss": 1.2078,
+      "loss": 1.2061,
       "step": 23710
     },
     {
       "epoch": 71.23,
-      "grad_norm": 6.403040409088135,
+      "grad_norm": 7.58206033706665,
       "learning_rate": 7.6256256256256266e-06,
-      "loss": 1.1999,
+      "loss": 1.2015,
       "step": 23720
     },
     {
       "epoch": 71.26,
-      "grad_norm": 5.947720527648926,
+      "grad_norm": 5.870658874511719,
       "learning_rate": 7.624624624624624e-06,
-      "loss": 1.2693,
+      "loss": 1.2857,
       "step": 23730
     },
     {
       "epoch": 71.29,
-      "grad_norm": 8.91308307647705,
+      "grad_norm": 7.281700611114502,
       "learning_rate": 7.623623623623624e-06,
-      "loss": 1.3264,
+      "loss": 1.3291,
       "step": 23740
     },
     {
       "epoch": 71.32,
-      "grad_norm": 9.095004081726074,
+      "grad_norm": 7.183254718780518,
       "learning_rate": 7.622622622622623e-06,
-      "loss": 1.1611,
+      "loss": 1.1475,
       "step": 23750
     },
     {
       "epoch": 71.35,
-      "grad_norm": 8.511629104614258,
+      "grad_norm": 6.335379600524902,
       "learning_rate": 7.621621621621622e-06,
-      "loss": 1.2297,
+      "loss": 1.2306,
       "step": 23760
     },
     {
       "epoch": 71.38,
-      "grad_norm": 7.548367023468018,
+      "grad_norm": 5.432480812072754,
       "learning_rate": 7.620620620620621e-06,
-      "loss": 1.2473,
+      "loss": 1.2505,
       "step": 23770
     },
     {
       "epoch": 71.41,
-      "grad_norm": 8.356057167053223,
+      "grad_norm": 12.751708030700684,
       "learning_rate": 7.619619619619621e-06,
-      "loss": 1.2275,
+      "loss": 1.2309,
       "step": 23780
     },
     {
       "epoch": 71.44,
-      "grad_norm": 8.700485229492188,
+      "grad_norm": 10.788970947265625,
       "learning_rate": 7.618618618618619e-06,
-      "loss": 1.2043,
+      "loss": 1.2096,
       "step": 23790
     },
     {
       "epoch": 71.47,
-      "grad_norm": 6.885606288909912,
+      "grad_norm": 10.317095756530762,
       "learning_rate": 7.617617617617619e-06,
-      "loss": 1.21,
+      "loss": 1.2147,
       "step": 23800
     },
     {
       "epoch": 71.5,
-      "grad_norm": 9.512789726257324,
+      "grad_norm": 8.2265625,
       "learning_rate": 7.616616616616618e-06,
-      "loss": 1.1922,
+      "loss": 1.1893,
       "step": 23810
     },
     {
       "epoch": 71.53,
-      "grad_norm": 7.634214401245117,
+      "grad_norm": 6.129824638366699,
       "learning_rate": 7.615615615615616e-06,
-      "loss": 1.2213,
+      "loss": 1.2079,
       "step": 23820
     },
     {
       "epoch": 71.56,
-      "grad_norm": 6.912567615509033,
+      "grad_norm": 7.846567153930664,
       "learning_rate": 7.614614614614615e-06,
-      "loss": 1.2851,
+      "loss": 1.2746,
       "step": 23830
     },
     {
       "epoch": 71.59,
-      "grad_norm": 10.59670352935791,
+      "grad_norm": 12.027969360351562,
       "learning_rate": 7.613613613613614e-06,
-      "loss": 1.3011,
+      "loss": 1.305,
       "step": 23840
     },
     {
       "epoch": 71.62,
-      "grad_norm": 7.925707817077637,
+      "grad_norm": 8.674944877624512,
       "learning_rate": 7.612612612612613e-06,
-      "loss": 1.2247,
+      "loss": 1.2328,
       "step": 23850
     },
     {
       "epoch": 71.65,
-      "grad_norm": 8.036969184875488,
+      "grad_norm": 7.800688743591309,
       "learning_rate": 7.611611611611612e-06,
-      "loss": 1.2713,
+      "loss": 1.2656,
       "step": 23860
     },
     {
       "epoch": 71.68,
-      "grad_norm": 7.118833065032959,
+      "grad_norm": 10.698493957519531,
       "learning_rate": 7.610610610610611e-06,
-      "loss": 1.2322,
+      "loss": 1.2367,
       "step": 23870
     },
     {
       "epoch": 71.71,
-      "grad_norm": 9.367517471313477,
+      "grad_norm": 12.780364036560059,
       "learning_rate": 7.609609609609611e-06,
-      "loss": 1.2648,
+      "loss": 1.2667,
       "step": 23880
     },
     {
       "epoch": 71.74,
-      "grad_norm": 6.668388843536377,
+      "grad_norm": 8.154374122619629,
       "learning_rate": 7.6086086086086095e-06,
-      "loss": 1.1867,
+      "loss": 1.1872,
       "step": 23890
     },
     {
       "epoch": 71.77,
-      "grad_norm": 7.40744686126709,
+      "grad_norm": 7.064554214477539,
       "learning_rate": 7.607607607607608e-06,
-      "loss": 1.3197,
+      "loss": 1.3269,
       "step": 23900
     },
     {
       "epoch": 71.8,
-      "grad_norm": 8.049695014953613,
+      "grad_norm": 8.133721351623535,
       "learning_rate": 7.606606606606607e-06,
-      "loss": 1.2181,
+      "loss": 1.2137,
       "step": 23910
     },
     {
       "epoch": 71.83,
-      "grad_norm": 5.819601535797119,
+      "grad_norm": 5.848204612731934,
       "learning_rate": 7.605605605605606e-06,
-      "loss": 1.2804,
+      "loss": 1.2842,
       "step": 23920
     },
     {
       "epoch": 71.86,
-      "grad_norm": 6.9818220138549805,
+      "grad_norm": 10.405204772949219,
       "learning_rate": 7.6046046046046055e-06,
-      "loss": 1.2934,
+      "loss": 1.293,
       "step": 23930
     },
     {
       "epoch": 71.89,
-      "grad_norm": 4.813801288604736,
+      "grad_norm": 6.8652167320251465,
       "learning_rate": 7.603603603603604e-06,
-      "loss": 1.1713,
+      "loss": 1.1653,
       "step": 23940
     },
     {
       "epoch": 71.92,
-      "grad_norm": 7.755620956420898,
+      "grad_norm": 11.219161033630371,
       "learning_rate": 7.6026026026026036e-06,
-      "loss": 1.2871,
+      "loss": 1.2854,
       "step": 23950
     },
     {
       "epoch": 71.95,
-      "grad_norm": 7.409787178039551,
+      "grad_norm": 6.833508014678955,
       "learning_rate": 7.601601601601602e-06,
-      "loss": 1.1837,
+      "loss": 1.1818,
       "step": 23960
     },
     {
       "epoch": 71.98,
-      "grad_norm": 10.470501899719238,
+      "grad_norm": 7.584761142730713,
       "learning_rate": 7.600600600600602e-06,
-      "loss": 1.2205,
+      "loss": 1.2214,
       "step": 23970
     },
     {
       "epoch": 72.0,
       "eval_accuracy": 0.7808,
-      "eval_loss": 0.7120937705039978,
-      "eval_runtime": 5.2976,
-      "eval_samples_per_second": 1887.644,
-      "eval_steps_per_second": 7.551,
+      "eval_loss": 0.7110925316810608,
+      "eval_runtime": 5.278,
+      "eval_samples_per_second": 1894.645,
+      "eval_steps_per_second": 7.579,
       "step": 23976
     },
     {
       "epoch": 72.01,
-      "grad_norm": 5.546104907989502,
+      "grad_norm": 5.517971038818359,
       "learning_rate": 7.599599599599599e-06,
-      "loss": 1.3655,
+      "loss": 1.3523,
       "step": 23980
     },
     {
       "epoch": 72.04,
-      "grad_norm": 6.395235538482666,
+      "grad_norm": 9.664935111999512,
       "learning_rate": 7.598598598598599e-06,
-      "loss": 1.2467,
+      "loss": 1.2503,
       "step": 23990
     },
     {
       "epoch": 72.07,
-      "grad_norm": 7.522167205810547,
+      "grad_norm": 6.187978267669678,
       "learning_rate": 7.597597597597598e-06,
-      "loss": 1.2089,
+      "loss": 1.2183,
       "step": 24000
     },
     {
       "epoch": 72.1,
-      "grad_norm": 8.732793807983398,
+      "grad_norm": 9.843351364135742,
       "learning_rate": 7.596596596596597e-06,
-      "loss": 1.2777,
+      "loss": 1.2804,
       "step": 24010
     },
     {
       "epoch": 72.13,
-      "grad_norm": 7.121822834014893,
+      "grad_norm": 8.177450180053711,
       "learning_rate": 7.595595595595596e-06,
-      "loss": 1.2775,
+      "loss": 1.2719,
       "step": 24020
     },
     {
       "epoch": 72.16,
-      "grad_norm": 6.50997257232666,
+      "grad_norm": 7.646065711975098,
       "learning_rate": 7.594594594594596e-06,
-      "loss": 1.2362,
+      "loss": 1.2434,
       "step": 24030
     },
     {
       "epoch": 72.19,
-      "grad_norm": 6.916764259338379,
+      "grad_norm": 10.118885040283203,
       "learning_rate": 7.593593593593594e-06,
-      "loss": 1.2736,
+      "loss": 1.2753,
       "step": 24040
     },
     {
       "epoch": 72.22,
-      "grad_norm": 7.954848766326904,
+      "grad_norm": 7.717072010040283,
       "learning_rate": 7.592592592592594e-06,
-      "loss": 1.2331,
+      "loss": 1.2428,
       "step": 24050
     },
     {
       "epoch": 72.25,
-      "grad_norm": 5.589578151702881,
+      "grad_norm": 8.024699211120605,
       "learning_rate": 7.591591591591592e-06,
-      "loss": 1.2161,
+      "loss": 1.221,
       "step": 24060
     },
     {
       "epoch": 72.28,
-      "grad_norm": 6.675744533538818,
+      "grad_norm": 6.549567699432373,
       "learning_rate": 7.590590590590591e-06,
-      "loss": 1.242,
+      "loss": 1.2298,
       "step": 24070
     },
     {
       "epoch": 72.31,
-      "grad_norm": 7.980341911315918,
+      "grad_norm": 6.230840682983398,
       "learning_rate": 7.5895895895895895e-06,
-      "loss": 1.246,
+      "loss": 1.2455,
       "step": 24080
     },
     {
       "epoch": 72.34,
-      "grad_norm": 8.141807556152344,
+      "grad_norm": 8.820632934570312,
       "learning_rate": 7.588588588588589e-06,
-      "loss": 1.2322,
+      "loss": 1.2289,
       "step": 24090
     },
     {
       "epoch": 72.37,
-      "grad_norm": 6.437416076660156,
+      "grad_norm": 9.636287689208984,
       "learning_rate": 7.587587587587588e-06,
-      "loss": 1.2288,
+      "loss": 1.2286,
       "step": 24100
     },
     {
       "epoch": 72.4,
-      "grad_norm": 9.296774864196777,
+      "grad_norm": 8.41147232055664,
       "learning_rate": 7.586586586586587e-06,
-      "loss": 1.2495,
+      "loss": 1.2519,
       "step": 24110
     },
     {
       "epoch": 72.43,
-      "grad_norm": 5.735087871551514,
+      "grad_norm": 7.138651371002197,
       "learning_rate": 7.5855855855855865e-06,
-      "loss": 1.2554,
+      "loss": 1.2552,
       "step": 24120
     },
     {
       "epoch": 72.46,
-      "grad_norm": 5.8101325035095215,
+      "grad_norm": 5.782553672790527,
       "learning_rate": 7.584584584584586e-06,
-      "loss": 1.2316,
+      "loss": 1.2302,
       "step": 24130
     },
     {
       "epoch": 72.49,
-      "grad_norm": 8.368491172790527,
+      "grad_norm": 7.634880065917969,
       "learning_rate": 7.5835835835835845e-06,
-      "loss": 1.2432,
+      "loss": 1.2535,
       "step": 24140
     },
     {
       "epoch": 72.52,
-      "grad_norm": 7.783821105957031,
+      "grad_norm": 6.701886177062988,
       "learning_rate": 7.582582582582583e-06,
-      "loss": 1.2391,
+      "loss": 1.2396,
       "step": 24150
     },
     {
       "epoch": 72.55,
-      "grad_norm": 7.628262042999268,
+      "grad_norm": 9.403764724731445,
       "learning_rate": 7.581581581581582e-06,
-      "loss": 1.2291,
+      "loss": 1.2255,
       "step": 24160
     },
     {
       "epoch": 72.58,
-      "grad_norm": 6.423760414123535,
+      "grad_norm": 9.187384605407715,
       "learning_rate": 7.580580580580581e-06,
-      "loss": 1.2215,
+      "loss": 1.2167,
       "step": 24170
     },
     {
       "epoch": 72.61,
-      "grad_norm": 4.60857629776001,
+      "grad_norm": 5.5967230796813965,
       "learning_rate": 7.57957957957958e-06,
-      "loss": 1.2196,
+      "loss": 1.2124,
       "step": 24180
     },
     {
       "epoch": 72.64,
-      "grad_norm": 6.469860076904297,
+      "grad_norm": 5.910060882568359,
       "learning_rate": 7.578578578578579e-06,
-      "loss": 1.2656,
+      "loss": 1.2503,
       "step": 24190
     },
     {
       "epoch": 72.67,
-      "grad_norm": 8.736680030822754,
+      "grad_norm": 7.885284900665283,
       "learning_rate": 7.577577577577579e-06,
-      "loss": 1.2548,
+      "loss": 1.24,
       "step": 24200
     },
     {
       "epoch": 72.7,
-      "grad_norm": 7.242773056030273,
+      "grad_norm": 8.46666431427002,
       "learning_rate": 7.576576576576577e-06,
-      "loss": 1.1678,
+      "loss": 1.163,
       "step": 24210
     },
     {
       "epoch": 72.73,
-      "grad_norm": 6.432340145111084,
+      "grad_norm": 7.928226947784424,
       "learning_rate": 7.575575575575577e-06,
-      "loss": 1.3011,
+      "loss": 1.2862,
       "step": 24220
     },
     {
       "epoch": 72.76,
-      "grad_norm": 5.799014568328857,
+      "grad_norm": 7.213484287261963,
       "learning_rate": 7.574574574574574e-06,
-      "loss": 1.2073,
+      "loss": 1.2119,
       "step": 24230
     },
     {
       "epoch": 72.79,
-      "grad_norm": 7.797633647918701,
+      "grad_norm": 7.509303092956543,
       "learning_rate": 7.573573573573574e-06,
-      "loss": 1.2394,
+      "loss": 1.2438,
       "step": 24240
     },
     {
       "epoch": 72.82,
-      "grad_norm": 6.635136604309082,
+      "grad_norm": 6.006824970245361,
       "learning_rate": 7.572572572572573e-06,
-      "loss": 1.2493,
+      "loss": 1.2545,
       "step": 24250
     },
     {
       "epoch": 72.85,
-      "grad_norm": 6.906064510345459,
+      "grad_norm": 7.807251453399658,
       "learning_rate": 7.571571571571572e-06,
-      "loss": 1.2424,
+      "loss": 1.2427,
       "step": 24260
     },
     {
       "epoch": 72.88,
-      "grad_norm": 5.553857803344727,
+      "grad_norm": 5.884125232696533,
       "learning_rate": 7.570570570570571e-06,
-      "loss": 1.2517,
+      "loss": 1.2488,
       "step": 24270
     },
     {
       "epoch": 72.91,
-      "grad_norm": 7.155200958251953,
+      "grad_norm": 8.524611473083496,
       "learning_rate": 7.569569569569571e-06,
-      "loss": 1.1906,
+      "loss": 1.1986,
       "step": 24280
     },
     {
       "epoch": 72.94,
-      "grad_norm": 6.202935218811035,
+      "grad_norm": 7.190827369689941,
       "learning_rate": 7.568568568568569e-06,
-      "loss": 1.2511,
+      "loss": 1.2473,
       "step": 24290
     },
     {
       "epoch": 72.97,
-      "grad_norm": 6.832364559173584,
+      "grad_norm": 6.586330413818359,
       "learning_rate": 7.567567567567569e-06,
-      "loss": 1.2696,
+      "loss": 1.2597,
       "step": 24300
     },
     {
       "epoch": 73.0,
-      "eval_accuracy": 0.7799,
-      "eval_loss": 0.7162183523178101,
-      "eval_runtime": 5.3794,
-      "eval_samples_per_second": 1858.955,
-      "eval_steps_per_second": 7.436,
+      "eval_accuracy": 0.7805,
+      "eval_loss": 0.7140112519264221,
+      "eval_runtime": 5.3819,
+      "eval_samples_per_second": 1858.075,
+      "eval_steps_per_second": 7.432,
       "step": 24309
     },
     {
       "epoch": 73.0,
-      "grad_norm": 8.379456520080566,
+      "grad_norm": 7.531033039093018,
       "learning_rate": 7.566566566566567e-06,
-      "loss": 1.339,
+      "loss": 1.325,
       "step": 24310
     },
     {
       "epoch": 73.03,
-      "grad_norm": 6.198248386383057,
+      "grad_norm": 6.838927745819092,
       "learning_rate": 7.565565565565566e-06,
-      "loss": 1.1657,
+      "loss": 1.1646,
       "step": 24320
     },
     {
       "epoch": 73.06,
-      "grad_norm": 6.7261857986450195,
+      "grad_norm": 6.7538886070251465,
       "learning_rate": 7.5645645645645646e-06,
-      "loss": 1.281,
+      "loss": 1.2702,
       "step": 24330
     },
     {
       "epoch": 73.09,
-      "grad_norm": 9.380220413208008,
+      "grad_norm": 8.836775779724121,
       "learning_rate": 7.563563563563564e-06,
-      "loss": 1.2754,
+      "loss": 1.2766,
       "step": 24340
     },
     {
       "epoch": 73.12,
-      "grad_norm": 5.889250755310059,
+      "grad_norm": 6.740044116973877,
       "learning_rate": 7.5625625625625634e-06,
-      "loss": 1.2239,
+      "loss": 1.2251,
       "step": 24350
     },
     {
       "epoch": 73.15,
-      "grad_norm": 6.678108215332031,
+      "grad_norm": 4.743236541748047,
       "learning_rate": 7.561561561561562e-06,
-      "loss": 1.2725,
+      "loss": 1.265,
       "step": 24360
     },
     {
       "epoch": 73.18,
-      "grad_norm": 6.142610549926758,
+      "grad_norm": 7.068434715270996,
       "learning_rate": 7.5605605605605615e-06,
-      "loss": 1.2784,
+      "loss": 1.2918,
       "step": 24370
     },
     {
       "epoch": 73.21,
-      "grad_norm": 8.757142066955566,
+      "grad_norm": 5.904600143432617,
       "learning_rate": 7.559559559559561e-06,
-      "loss": 1.2448,
+      "loss": 1.2372,
       "step": 24380
     },
     {
       "epoch": 73.24,
-      "grad_norm": 6.043947219848633,
+      "grad_norm": 5.9043121337890625,
       "learning_rate": 7.5585585585585595e-06,
-      "loss": 1.2538,
+      "loss": 1.2453,
       "step": 24390
     },
     {
       "epoch": 73.27,
-      "grad_norm": 7.107696056365967,
+      "grad_norm": 7.297125816345215,
       "learning_rate": 7.557557557557558e-06,
-      "loss": 1.2188,
+      "loss": 1.2221,
       "step": 24400
     },
     {
       "epoch": 73.3,
-      "grad_norm": 9.033823013305664,
+      "grad_norm": 7.870432376861572,
       "learning_rate": 7.556556556556557e-06,
-      "loss": 1.238,
+      "loss": 1.2422,
       "step": 24410
     },
     {
       "epoch": 73.33,
-      "grad_norm": 6.995325088500977,
+      "grad_norm": 7.464437961578369,
       "learning_rate": 7.555555555555556e-06,
-      "loss": 1.3047,
+      "loss": 1.3035,
       "step": 24420
     },
     {
       "epoch": 73.36,
-      "grad_norm": 8.597603797912598,
+      "grad_norm": 8.055253982543945,
       "learning_rate": 7.554554554554555e-06,
-      "loss": 1.311,
+      "loss": 1.3142,
       "step": 24430
     },
     {
       "epoch": 73.39,
-      "grad_norm": 7.90497350692749,
+      "grad_norm": 7.008742332458496,
       "learning_rate": 7.553553553553554e-06,
-      "loss": 1.2371,
+      "loss": 1.2417,
       "step": 24440
     },
     {
       "epoch": 73.42,
-      "grad_norm": 5.4672465324401855,
+      "grad_norm": 6.190035343170166,
       "learning_rate": 7.552552552552554e-06,
-      "loss": 1.214,
+      "loss": 1.216,
       "step": 24450
     },
     {
       "epoch": 73.45,
-      "grad_norm": 6.881330490112305,
+      "grad_norm": 7.5445146560668945,
       "learning_rate": 7.551551551551552e-06,
-      "loss": 1.2895,
+      "loss": 1.3049,
       "step": 24460
     },
     {
       "epoch": 73.48,
-      "grad_norm": 7.689633846282959,
+      "grad_norm": 8.126893043518066,
       "learning_rate": 7.550550550550552e-06,
-      "loss": 1.2123,
+      "loss": 1.2024,
       "step": 24470
     },
     {
       "epoch": 73.51,
-      "grad_norm": 9.734367370605469,
+      "grad_norm": 9.205645561218262,
       "learning_rate": 7.549549549549549e-06,
-      "loss": 1.2459,
+      "loss": 1.254,
       "step": 24480
     },
     {
       "epoch": 73.54,
-      "grad_norm": 6.522113800048828,
+      "grad_norm": 6.877061367034912,
       "learning_rate": 7.548548548548549e-06,
-      "loss": 1.1932,
+      "loss": 1.184,
       "step": 24490
     },
     {
       "epoch": 73.57,
-      "grad_norm": 5.06716251373291,
+      "grad_norm": 6.4929585456848145,
       "learning_rate": 7.547547547547548e-06,
-      "loss": 1.2204,
+      "loss": 1.2234,
       "step": 24500
     },
     {
       "epoch": 73.6,
-      "grad_norm": 9.443493843078613,
+      "grad_norm": 9.811019897460938,
       "learning_rate": 7.546546546546547e-06,
-      "loss": 1.1898,
+      "loss": 1.1899,
       "step": 24510
     },
     {
       "epoch": 73.63,
-      "grad_norm": 6.462500095367432,
+      "grad_norm": 7.850760459899902,
       "learning_rate": 7.545545545545546e-06,
-      "loss": 1.1907,
+      "loss": 1.1955,
       "step": 24520
     },
     {
       "epoch": 73.66,
-      "grad_norm": 6.358414173126221,
+      "grad_norm": 9.772212982177734,
       "learning_rate": 7.544544544544545e-06,
-      "loss": 1.2042,
+      "loss": 1.2066,
       "step": 24530
     },
     {
       "epoch": 73.69,
-      "grad_norm": 7.759615898132324,
+      "grad_norm": 7.324338912963867,
       "learning_rate": 7.543543543543544e-06,
-      "loss": 1.1945,
+      "loss": 1.1935,
       "step": 24540
     },
     {
       "epoch": 73.72,
-      "grad_norm": 6.954308986663818,
+      "grad_norm": 8.339109420776367,
       "learning_rate": 7.542542542542544e-06,
-      "loss": 1.2615,
+      "loss": 1.2629,
       "step": 24550
     },
     {
       "epoch": 73.75,
-      "grad_norm": 5.853650093078613,
+      "grad_norm": 8.656244277954102,
       "learning_rate": 7.5415415415415416e-06,
-      "loss": 1.2232,
+      "loss": 1.2257,
       "step": 24560
     },
     {
       "epoch": 73.78,
-      "grad_norm": 8.011367797851562,
+      "grad_norm": 7.28223180770874,
       "learning_rate": 7.540540540540541e-06,
-      "loss": 1.1836,
+      "loss": 1.1719,
       "step": 24570
     },
     {
       "epoch": 73.81,
-      "grad_norm": 6.162054061889648,
+      "grad_norm": 9.742901802062988,
       "learning_rate": 7.53953953953954e-06,
-      "loss": 1.1722,
+      "loss": 1.1815,
       "step": 24580
     },
     {
       "epoch": 73.84,
-      "grad_norm": 6.6622419357299805,
+      "grad_norm": 6.356659889221191,
       "learning_rate": 7.538538538538539e-06,
-      "loss": 1.2586,
+      "loss": 1.2528,
       "step": 24590
     },
     {
       "epoch": 73.87,
-      "grad_norm": 5.186406135559082,
+      "grad_norm": 5.296167373657227,
       "learning_rate": 7.5375375375375385e-06,
-      "loss": 1.2547,
+      "loss": 1.2537,
       "step": 24600
     },
     {
       "epoch": 73.9,
-      "grad_norm": 6.463618755340576,
+      "grad_norm": 8.62954044342041,
       "learning_rate": 7.536536536536537e-06,
-      "loss": 1.2521,
+      "loss": 1.2539,
       "step": 24610
     },
     {
       "epoch": 73.93,
-      "grad_norm": 7.929732322692871,
+      "grad_norm": 7.577630519866943,
       "learning_rate": 7.5355355355355365e-06,
-      "loss": 1.2311,
+      "loss": 1.2154,
       "step": 24620
     },
     {
       "epoch": 73.96,
-      "grad_norm": 7.456181526184082,
+      "grad_norm": 7.654656887054443,
       "learning_rate": 7.534534534534535e-06,
-      "loss": 1.273,
+      "loss": 1.2608,
       "step": 24630
     },
     {
       "epoch": 73.99,
-      "grad_norm": 8.937918663024902,
+      "grad_norm": 8.215866088867188,
       "learning_rate": 7.5335335335335346e-06,
-      "loss": 1.2083,
+      "loss": 1.2147,
       "step": 24640
     },
     {
       "epoch": 74.0,
-      "eval_accuracy": 0.786,
-      "eval_loss": 0.7030656337738037,
-      "eval_runtime": 5.1763,
-      "eval_samples_per_second": 1931.896,
-      "eval_steps_per_second": 7.728,
+      "eval_accuracy": 0.7821,
+      "eval_loss": 0.7022157311439514,
+      "eval_runtime": 5.2771,
+      "eval_samples_per_second": 1894.995,
+      "eval_steps_per_second": 7.58,
       "step": 24642
     },
     {
       "epoch": 74.02,
-      "grad_norm": 5.793099403381348,
+      "grad_norm": 6.397442817687988,
       "learning_rate": 7.532532532532532e-06,
-      "loss": 1.5204,
+      "loss": 1.5023,
       "step": 24650
     },
     {
       "epoch": 74.05,
-      "grad_norm": 6.290515899658203,
+      "grad_norm": 5.914522171020508,
       "learning_rate": 7.531531531531532e-06,
-      "loss": 1.1748,
+      "loss": 1.1759,
       "step": 24660
     },
     {
       "epoch": 74.08,
-      "grad_norm": 6.861786365509033,
+      "grad_norm": 8.997749328613281,
       "learning_rate": 7.530530530530531e-06,
-      "loss": 1.2702,
+      "loss": 1.2844,
       "step": 24670
     },
     {
       "epoch": 74.11,
-      "grad_norm": 6.580562114715576,
+      "grad_norm": 7.166990280151367,
       "learning_rate": 7.52952952952953e-06,
-      "loss": 1.2236,
+      "loss": 1.2186,
       "step": 24680
     },
     {
       "epoch": 74.14,
-      "grad_norm": 8.283895492553711,
+      "grad_norm": 11.322690963745117,
       "learning_rate": 7.528528528528529e-06,
-      "loss": 1.2292,
+      "loss": 1.2347,
       "step": 24690
     },
     {
       "epoch": 74.17,
-      "grad_norm": 11.772353172302246,
+      "grad_norm": 9.525617599487305,
       "learning_rate": 7.527527527527529e-06,
-      "loss": 1.2734,
+      "loss": 1.2674,
       "step": 24700
     },
     {
       "epoch": 74.2,
-      "grad_norm": 7.620272636413574,
+      "grad_norm": 8.674941062927246,
       "learning_rate": 7.526526526526527e-06,
-      "loss": 1.2293,
+      "loss": 1.2375,
       "step": 24710
     },
     {
       "epoch": 74.23,
-      "grad_norm": 7.075238227844238,
+      "grad_norm": 11.002622604370117,
       "learning_rate": 7.525525525525527e-06,
-      "loss": 1.217,
+      "loss": 1.203,
       "step": 24720
     },
     {
       "epoch": 74.26,
-      "grad_norm": 9.40426254272461,
+      "grad_norm": 7.2463250160217285,
       "learning_rate": 7.5245245245245245e-06,
-      "loss": 1.2706,
+      "loss": 1.2733,
       "step": 24730
     },
     {
       "epoch": 74.29,
-      "grad_norm": 7.526463508605957,
+      "grad_norm": 7.673720836639404,
       "learning_rate": 7.523523523523524e-06,
-      "loss": 1.2074,
+      "loss": 1.2084,
       "step": 24740
     },
     {
       "epoch": 74.32,
-      "grad_norm": 8.214470863342285,
+      "grad_norm": 7.007150173187256,
       "learning_rate": 7.5225225225225225e-06,
-      "loss": 1.2482,
+      "loss": 1.2442,
       "step": 24750
     },
     {
       "epoch": 74.35,
-      "grad_norm": 6.834357738494873,
+      "grad_norm": 7.601666450500488,
       "learning_rate": 7.521521521521522e-06,
-      "loss": 1.2384,
+      "loss": 1.2419,
       "step": 24760
     },
     {
       "epoch": 74.38,
-      "grad_norm": 6.4213738441467285,
+      "grad_norm": 6.963320732116699,
       "learning_rate": 7.520520520520521e-06,
-      "loss": 1.2481,
+      "loss": 1.2453,
       "step": 24770
     },
     {
       "epoch": 74.41,
-      "grad_norm": 11.627789497375488,
+      "grad_norm": 12.267460823059082,
       "learning_rate": 7.51951951951952e-06,
-      "loss": 1.207,
+      "loss": 1.2085,
       "step": 24780
     },
     {
       "epoch": 74.44,
-      "grad_norm": 4.495910167694092,
+      "grad_norm": 7.385159015655518,
       "learning_rate": 7.518518518518519e-06,
-      "loss": 1.2122,
+      "loss": 1.2179,
       "step": 24790
     },
     {
       "epoch": 74.47,
-      "grad_norm": 6.332726001739502,
+      "grad_norm": 6.007859230041504,
       "learning_rate": 7.517517517517519e-06,
-      "loss": 1.2348,
+      "loss": 1.2251,
       "step": 24800
     },
     {
       "epoch": 74.5,
-      "grad_norm": 5.893298149108887,
+      "grad_norm": 8.986599922180176,
       "learning_rate": 7.516516516516517e-06,
-      "loss": 1.2464,
+      "loss": 1.2498,
       "step": 24810
     },
     {
       "epoch": 74.53,
-      "grad_norm": 7.0014448165893555,
+      "grad_norm": 6.6137919425964355,
       "learning_rate": 7.515515515515516e-06,
-      "loss": 1.2528,
+      "loss": 1.2638,
       "step": 24820
     },
     {
       "epoch": 74.56,
-      "grad_norm": 7.39432954788208,
+      "grad_norm": 8.713010787963867,
       "learning_rate": 7.514514514514515e-06,
-      "loss": 1.2058,
+      "loss": 1.2034,
       "step": 24830
     },
     {
       "epoch": 74.59,
-      "grad_norm": 6.9833598136901855,
+      "grad_norm": 7.662017822265625,
       "learning_rate": 7.513513513513514e-06,
-      "loss": 1.2364,
+      "loss": 1.2459,
       "step": 24840
     },
     {
       "epoch": 74.62,
-      "grad_norm": 9.58503246307373,
+      "grad_norm": 9.921331405639648,
       "learning_rate": 7.5125125125125135e-06,
       "loss": 1.2676,
       "step": 24850
     },
     {
       "epoch": 74.65,
-      "grad_norm": 5.834603309631348,
+      "grad_norm": 8.14899730682373,
       "learning_rate": 7.511511511511512e-06,
-      "loss": 1.1119,
+      "loss": 1.1147,
       "step": 24860
     },
     {
       "epoch": 74.68,
-      "grad_norm": 9.252755165100098,
+      "grad_norm": 8.972559928894043,
       "learning_rate": 7.5105105105105116e-06,
-      "loss": 1.1784,
+      "loss": 1.1694,
       "step": 24870
     },
     {
       "epoch": 74.71,
-      "grad_norm": 9.062361717224121,
+      "grad_norm": 8.475200653076172,
       "learning_rate": 7.50950950950951e-06,
-      "loss": 1.1785,
+      "loss": 1.1734,
       "step": 24880
     },
     {
       "epoch": 74.74,
-      "grad_norm": 6.683990955352783,
+      "grad_norm": 10.52548599243164,
       "learning_rate": 7.50850850850851e-06,
-      "loss": 1.222,
+      "loss": 1.2228,
       "step": 24890
     },
     {
       "epoch": 74.77,
-      "grad_norm": 7.08304500579834,
+      "grad_norm": 7.199855327606201,
       "learning_rate": 7.507507507507507e-06,
-      "loss": 1.2297,
+      "loss": 1.234,
       "step": 24900
     },
     {
       "epoch": 74.8,
-      "grad_norm": 9.885730743408203,
+      "grad_norm": 9.447124481201172,
       "learning_rate": 7.506506506506507e-06,
-      "loss": 1.3441,
+      "loss": 1.3446,
       "step": 24910
     },
     {
       "epoch": 74.83,
-      "grad_norm": 5.40369176864624,
+      "grad_norm": 9.63575553894043,
       "learning_rate": 7.505505505505506e-06,
-      "loss": 1.281,
+      "loss": 1.2763,
       "step": 24920
     },
     {
       "epoch": 74.86,
-      "grad_norm": 5.514545440673828,
+      "grad_norm": 7.2716827392578125,
       "learning_rate": 7.504504504504505e-06,
-      "loss": 1.2151,
+      "loss": 1.2267,
       "step": 24930
     },
     {
       "epoch": 74.89,
-      "grad_norm": 7.9596333503723145,
+      "grad_norm": 6.141102313995361,
       "learning_rate": 7.503503503503504e-06,
-      "loss": 1.1948,
+      "loss": 1.1852,
       "step": 24940
     },
     {
       "epoch": 74.92,
-      "grad_norm": 6.651490211486816,
+      "grad_norm": 6.453799724578857,
       "learning_rate": 7.502502502502504e-06,
-      "loss": 1.1519,
+      "loss": 1.1558,
       "step": 24950
     },
     {
       "epoch": 74.95,
-      "grad_norm": 7.435955047607422,
+      "grad_norm": 5.885591506958008,
       "learning_rate": 7.501501501501502e-06,
-      "loss": 1.2008,
+      "loss": 1.1993,
       "step": 24960
     },
     {
       "epoch": 74.98,
-      "grad_norm": 6.387840747833252,
+      "grad_norm": 7.844815254211426,
       "learning_rate": 7.500500500500502e-06,
-      "loss": 1.2186,
+      "loss": 1.2105,
       "step": 24970
     },
     {
       "epoch": 75.0,
-      "eval_accuracy": 0.7876,
-      "eval_loss": 0.6934365630149841,
-      "eval_runtime": 5.7449,
-      "eval_samples_per_second": 1740.678,
-      "eval_steps_per_second": 6.963,
+      "eval_accuracy": 0.788,
+      "eval_loss": 0.6894916296005249,
+      "eval_runtime": 5.2183,
+      "eval_samples_per_second": 1916.342,
+      "eval_steps_per_second": 7.665,
       "step": 24975
     },
     {
       "epoch": 75.02,
-      "grad_norm": 5.896946430206299,
+      "grad_norm": 5.829438209533691,
       "learning_rate": 7.4994994994994995e-06,
-      "loss": 1.3074,
+      "loss": 1.3046,
       "step": 24980
     },
     {
       "epoch": 75.05,
-      "grad_norm": 6.7893385887146,
+      "grad_norm": 6.230485916137695,
       "learning_rate": 7.498498498498499e-06,
-      "loss": 1.1375,
+      "loss": 1.1399,
       "step": 24990
     },
     {
       "epoch": 75.08,
-      "grad_norm": 6.622607231140137,
+      "grad_norm": 8.583033561706543,
       "learning_rate": 7.4974974974974975e-06,
-      "loss": 1.2849,
+      "loss": 1.2866,
       "step": 25000
     },
     {
       "epoch": 75.11,
-      "grad_norm": 7.890625953674316,
+      "grad_norm": 8.56737232208252,
       "learning_rate": 7.496496496496497e-06,
-      "loss": 1.2807,
+      "loss": 1.2872,
       "step": 25010
     },
     {
       "epoch": 75.14,
-      "grad_norm": 6.72093391418457,
+      "grad_norm": 11.496731758117676,
       "learning_rate": 7.495495495495496e-06,
-      "loss": 1.2062,
+      "loss": 1.2031,
       "step": 25020
     },
     {
       "epoch": 75.17,
-      "grad_norm": 6.694521427154541,
+      "grad_norm": 6.0819315910339355,
       "learning_rate": 7.494494494494495e-06,
-      "loss": 1.2378,
+      "loss": 1.2375,
       "step": 25030
     },
     {
       "epoch": 75.2,
-      "grad_norm": 5.108382701873779,
+      "grad_norm": 6.852908134460449,
       "learning_rate": 7.4934934934934944e-06,
-      "loss": 1.1743,
+      "loss": 1.1814,
       "step": 25040
     },
     {
       "epoch": 75.23,
-      "grad_norm": 5.940899848937988,
+      "grad_norm": 6.859817028045654,
       "learning_rate": 7.492492492492494e-06,
-      "loss": 1.2635,
+      "loss": 1.2567,
       "step": 25050
     },
     {
       "epoch": 75.26,
-      "grad_norm": 4.593629360198975,
+      "grad_norm": 5.661890983581543,
       "learning_rate": 7.491491491491492e-06,
-      "loss": 1.2202,
+      "loss": 1.2168,
       "step": 25060
     },
     {
       "epoch": 75.29,
-      "grad_norm": 6.57780647277832,
+      "grad_norm": 8.073601722717285,
       "learning_rate": 7.490490490490491e-06,
-      "loss": 1.2545,
+      "loss": 1.2523,
       "step": 25070
     },
     {
       "epoch": 75.32,
-      "grad_norm": 6.144586563110352,
+      "grad_norm": 6.27846097946167,
       "learning_rate": 7.48948948948949e-06,
-      "loss": 1.2048,
+      "loss": 1.2061,
       "step": 25080
     },
     {
       "epoch": 75.35,
-      "grad_norm": 8.459406852722168,
+      "grad_norm": 7.376041889190674,
       "learning_rate": 7.488488488488489e-06,
-      "loss": 1.2555,
+      "loss": 1.2477,
       "step": 25090
     },
     {
       "epoch": 75.38,
-      "grad_norm": 5.508653163909912,
+      "grad_norm": 6.809115886688232,
       "learning_rate": 7.487487487487488e-06,
-      "loss": 1.1896,
+      "loss": 1.1939,
       "step": 25100
     },
     {
       "epoch": 75.41,
-      "grad_norm": 8.800329208374023,
+      "grad_norm": 7.034193992614746,
       "learning_rate": 7.486486486486487e-06,
-      "loss": 1.2588,
+      "loss": 1.2517,
       "step": 25110
     },
     {
       "epoch": 75.44,
-      "grad_norm": 9.387350082397461,
+      "grad_norm": 7.985945701599121,
       "learning_rate": 7.485485485485487e-06,
-      "loss": 1.1959,
+      "loss": 1.1953,
       "step": 25120
     },
     {
       "epoch": 75.47,
-      "grad_norm": 11.481096267700195,
+      "grad_norm": 7.883030891418457,
       "learning_rate": 7.484484484484485e-06,
-      "loss": 1.2771,
+      "loss": 1.2788,
       "step": 25130
     },
     {
       "epoch": 75.5,
-      "grad_norm": 7.817437648773193,
+      "grad_norm": 7.611591339111328,
       "learning_rate": 7.483483483483485e-06,
-      "loss": 1.2358,
+      "loss": 1.233,
       "step": 25140
     },
     {
       "epoch": 75.53,
-      "grad_norm": 7.453218460083008,
+      "grad_norm": 9.709419250488281,
       "learning_rate": 7.482482482482482e-06,
-      "loss": 1.1927,
+      "loss": 1.1846,
       "step": 25150
     },
     {
       "epoch": 75.56,
-      "grad_norm": 8.122724533081055,
+      "grad_norm": 7.241645336151123,
       "learning_rate": 7.481481481481482e-06,
-      "loss": 1.2426,
+      "loss": 1.2441,
       "step": 25160
     },
     {
       "epoch": 75.59,
-      "grad_norm": 5.34800910949707,
+      "grad_norm": 6.153046607971191,
       "learning_rate": 7.480480480480481e-06,
-      "loss": 1.1995,
+      "loss": 1.1923,
       "step": 25170
     },
     {
       "epoch": 75.62,
-      "grad_norm": 6.851415634155273,
+      "grad_norm": 6.6524505615234375,
       "learning_rate": 7.47947947947948e-06,
-      "loss": 1.2157,
+      "loss": 1.212,
       "step": 25180
     },
     {
       "epoch": 75.65,
-      "grad_norm": 6.510934352874756,
+      "grad_norm": 8.875775337219238,
       "learning_rate": 7.478478478478479e-06,
-      "loss": 1.1456,
+      "loss": 1.1478,
       "step": 25190
     },
     {
       "epoch": 75.68,
-      "grad_norm": 6.345443248748779,
+      "grad_norm": 5.715800762176514,
       "learning_rate": 7.477477477477479e-06,
-      "loss": 1.2249,
+      "loss": 1.2297,
       "step": 25200
     },
     {
       "epoch": 75.71,
-      "grad_norm": 6.904169082641602,
+      "grad_norm": 9.22205638885498,
       "learning_rate": 7.476476476476477e-06,
-      "loss": 1.2144,
+      "loss": 1.2188,
       "step": 25210
     },
     {
       "epoch": 75.74,
-      "grad_norm": 6.839168548583984,
+      "grad_norm": 6.412981033325195,
       "learning_rate": 7.475475475475477e-06,
-      "loss": 1.1891,
+      "loss": 1.1819,
       "step": 25220
     },
     {
       "epoch": 75.77,
-      "grad_norm": 6.9270148277282715,
+      "grad_norm": 7.71604061126709,
       "learning_rate": 7.4744744744744745e-06,
-      "loss": 1.2316,
+      "loss": 1.239,
       "step": 25230
     },
     {
       "epoch": 75.8,
-      "grad_norm": 8.121129035949707,
+      "grad_norm": 8.172898292541504,
       "learning_rate": 7.473473473473474e-06,
-      "loss": 1.2516,
+      "loss": 1.2371,
       "step": 25240
     },
     {
       "epoch": 75.83,
-      "grad_norm": 7.892098903656006,
+      "grad_norm": 9.071792602539062,
       "learning_rate": 7.4724724724724726e-06,
-      "loss": 1.1642,
+      "loss": 1.1586,
       "step": 25250
     },
     {
       "epoch": 75.86,
-      "grad_norm": 7.18553352355957,
+      "grad_norm": 8.579296112060547,
       "learning_rate": 7.471471471471472e-06,
-      "loss": 1.2031,
+      "loss": 1.1999,
       "step": 25260
     },
     {
       "epoch": 75.89,
-      "grad_norm": 6.60878324508667,
+      "grad_norm": 7.76070499420166,
       "learning_rate": 7.4704704704704714e-06,
-      "loss": 1.2324,
+      "loss": 1.2338,
       "step": 25270
     },
     {
       "epoch": 75.92,
-      "grad_norm": 9.636848449707031,
+      "grad_norm": 9.23180103302002,
       "learning_rate": 7.46946946946947e-06,
-      "loss": 1.2831,
+      "loss": 1.2851,
       "step": 25280
     },
     {
       "epoch": 75.95,
-      "grad_norm": 8.338752746582031,
+      "grad_norm": 7.984883785247803,
       "learning_rate": 7.4684684684684695e-06,
-      "loss": 1.2179,
+      "loss": 1.2012,
       "step": 25290
     },
     {
       "epoch": 75.98,
-      "grad_norm": 10.127613067626953,
+      "grad_norm": 8.25801944732666,
       "learning_rate": 7.467467467467469e-06,
-      "loss": 1.2252,
+      "loss": 1.2123,
       "step": 25300
     },
     {
       "epoch": 76.0,
-      "eval_accuracy": 0.7849,
-      "eval_loss": 0.7062305212020874,
-      "eval_runtime": 5.4316,
-      "eval_samples_per_second": 1841.072,
-      "eval_steps_per_second": 7.364,
+      "eval_accuracy": 0.7847,
+      "eval_loss": 0.7037064433097839,
+      "eval_runtime": 5.3361,
+      "eval_samples_per_second": 1874.02,
+      "eval_steps_per_second": 7.496,
       "step": 25308
     },
     {
       "epoch": 76.01,
-      "grad_norm": 8.04137897491455,
+      "grad_norm": 9.967961311340332,
       "learning_rate": 7.466466466466467e-06,
-      "loss": 1.3554,
+      "loss": 1.3689,
       "step": 25310
     },
     {
       "epoch": 76.04,
-      "grad_norm": 7.7633185386657715,
+      "grad_norm": 6.527676105499268,
       "learning_rate": 7.465465465465466e-06,
-      "loss": 1.1566,
+      "loss": 1.1607,
       "step": 25320
     },
     {
       "epoch": 76.07,
-      "grad_norm": 7.3299431800842285,
+      "grad_norm": 8.645343780517578,
       "learning_rate": 7.464464464464465e-06,
-      "loss": 1.252,
+      "loss": 1.2444,
       "step": 25330
     },
     {
       "epoch": 76.1,
-      "grad_norm": 4.845160484313965,
+      "grad_norm": 5.900119781494141,
       "learning_rate": 7.463463463463464e-06,
-      "loss": 1.3103,
+      "loss": 1.3015,
       "step": 25340
     },
     {
       "epoch": 76.13,
-      "grad_norm": 5.052088260650635,
+      "grad_norm": 5.386457443237305,
       "learning_rate": 7.462462462462463e-06,
-      "loss": 1.1521,
+      "loss": 1.1577,
       "step": 25350
     },
     {
       "epoch": 76.16,
-      "grad_norm": 6.882846355438232,
+      "grad_norm": 8.44308853149414,
       "learning_rate": 7.461461461461462e-06,
-      "loss": 1.2483,
+      "loss": 1.2515,
       "step": 25360
     },
     {
       "epoch": 76.19,
-      "grad_norm": 6.837671756744385,
+      "grad_norm": 7.33701229095459,
       "learning_rate": 7.460460460460462e-06,
-      "loss": 1.1353,
+      "loss": 1.1522,
       "step": 25370
     },
     {
       "epoch": 76.22,
-      "grad_norm": 6.489023208618164,
+      "grad_norm": 7.17464017868042,
       "learning_rate": 7.45945945945946e-06,
-      "loss": 1.246,
+      "loss": 1.2404,
       "step": 25380
     },
     {
       "epoch": 76.25,
-      "grad_norm": 6.850473403930664,
+      "grad_norm": 8.312623977661133,
       "learning_rate": 7.45845845845846e-06,
-      "loss": 1.226,
+      "loss": 1.2279,
       "step": 25390
     },
     {
       "epoch": 76.28,
-      "grad_norm": 7.4119648933410645,
+      "grad_norm": 7.305902481079102,
       "learning_rate": 7.457457457457457e-06,
-      "loss": 1.3167,
+      "loss": 1.3092,
       "step": 25400
     },
     {
       "epoch": 76.31,
-      "grad_norm": 6.283029556274414,
+      "grad_norm": 6.982364177703857,
       "learning_rate": 7.456456456456457e-06,
-      "loss": 1.2682,
+      "loss": 1.2649,
       "step": 25410
     },
     {
       "epoch": 76.34,
-      "grad_norm": 7.53995943069458,
+      "grad_norm": 8.083166122436523,
       "learning_rate": 7.455455455455456e-06,
-      "loss": 1.1287,
+      "loss": 1.1381,
       "step": 25420
     },
     {
       "epoch": 76.37,
-      "grad_norm": 7.174914836883545,
+      "grad_norm": 9.427179336547852,
       "learning_rate": 7.454454454454455e-06,
-      "loss": 1.1717,
+      "loss": 1.1723,
       "step": 25430
     },
     {
       "epoch": 76.4,
-      "grad_norm": 7.585734844207764,
+      "grad_norm": 8.316886901855469,
       "learning_rate": 7.453453453453454e-06,
-      "loss": 1.2544,
+      "loss": 1.2601,
       "step": 25440
     },
     {
       "epoch": 76.43,
-      "grad_norm": 6.0077948570251465,
+      "grad_norm": 6.8703508377075195,
       "learning_rate": 7.452452452452453e-06,
-      "loss": 1.2018,
+      "loss": 1.1975,
       "step": 25450
     },
     {
       "epoch": 76.46,
-      "grad_norm": 5.945810794830322,
+      "grad_norm": 7.136898517608643,
       "learning_rate": 7.451451451451452e-06,
-      "loss": 1.1593,
+      "loss": 1.1534,
       "step": 25460
     },
     {
       "epoch": 76.49,
-      "grad_norm": 7.585656642913818,
+      "grad_norm": 7.570375442504883,
       "learning_rate": 7.450450450450452e-06,
-      "loss": 1.2683,
+      "loss": 1.2643,
       "step": 25470
     },
     {
       "epoch": 76.52,
-      "grad_norm": 5.304036617279053,
+      "grad_norm": 8.619441032409668,
       "learning_rate": 7.4494494494494496e-06,
-      "loss": 1.1867,
+      "loss": 1.1781,
       "step": 25480
     },
     {
       "epoch": 76.55,
-      "grad_norm": 5.673671245574951,
+      "grad_norm": 6.171356201171875,
       "learning_rate": 7.448448448448449e-06,
-      "loss": 1.1496,
+      "loss": 1.1461,
       "step": 25490
     },
     {
       "epoch": 76.58,
-      "grad_norm": 5.774319648742676,
+      "grad_norm": 5.847502708435059,
       "learning_rate": 7.447447447447448e-06,
-      "loss": 1.2245,
+      "loss": 1.2269,
       "step": 25500
     },
     {
       "epoch": 76.61,
-      "grad_norm": 8.010832786560059,
+      "grad_norm": 10.270757675170898,
       "learning_rate": 7.446446446446447e-06,
-      "loss": 1.185,
+      "loss": 1.1802,
       "step": 25510
     },
     {
       "epoch": 76.64,
-      "grad_norm": 6.14829158782959,
+      "grad_norm": 8.852153778076172,
       "learning_rate": 7.4454454454454465e-06,
-      "loss": 1.1888,
+      "loss": 1.1898,
       "step": 25520
     },
     {
       "epoch": 76.67,
-      "grad_norm": 6.529942035675049,
+      "grad_norm": 7.977939605712891,
       "learning_rate": 7.444444444444445e-06,
-      "loss": 1.1848,
+      "loss": 1.1869,
       "step": 25530
     },
     {
       "epoch": 76.7,
-      "grad_norm": 6.769007205963135,
+      "grad_norm": 7.943223476409912,
       "learning_rate": 7.4434434434434445e-06,
-      "loss": 1.2805,
+      "loss": 1.2852,
       "step": 25540
     },
     {
       "epoch": 76.73,
-      "grad_norm": 7.6244378089904785,
+      "grad_norm": 5.976691246032715,
       "learning_rate": 7.442442442442443e-06,
-      "loss": 1.2953,
+      "loss": 1.2909,
       "step": 25550
     },
     {
       "epoch": 76.76,
-      "grad_norm": 7.2129292488098145,
+      "grad_norm": 7.813770771026611,
       "learning_rate": 7.441441441441442e-06,
-      "loss": 1.218,
+      "loss": 1.2174,
       "step": 25560
     },
     {
       "epoch": 76.79,
-      "grad_norm": 9.123944282531738,
+      "grad_norm": 9.890303611755371,
       "learning_rate": 7.44044044044044e-06,
-      "loss": 1.1475,
+      "loss": 1.1449,
       "step": 25570
     },
     {
       "epoch": 76.82,
-      "grad_norm": 6.663500785827637,
+      "grad_norm": 8.521355628967285,
       "learning_rate": 7.43943943943944e-06,
-      "loss": 1.2176,
+      "loss": 1.2316,
       "step": 25580
     },
     {
       "epoch": 76.85,
-      "grad_norm": 6.8810577392578125,
+      "grad_norm": 7.94570255279541,
       "learning_rate": 7.438438438438439e-06,
-      "loss": 1.2875,
+      "loss": 1.2856,
       "step": 25590
     },
     {
       "epoch": 76.88,
-      "grad_norm": 6.397066116333008,
+      "grad_norm": 5.75680685043335,
       "learning_rate": 7.437437437437438e-06,
-      "loss": 1.1453,
+      "loss": 1.1427,
       "step": 25600
     },
     {
       "epoch": 76.91,
-      "grad_norm": 9.617109298706055,
+      "grad_norm": 10.630685806274414,
       "learning_rate": 7.436436436436437e-06,
-      "loss": 1.166,
+      "loss": 1.1587,
       "step": 25610
     },
     {
       "epoch": 76.94,
-      "grad_norm": 6.588548183441162,
+      "grad_norm": 9.115632057189941,
       "learning_rate": 7.435435435435437e-06,
-      "loss": 1.2477,
+      "loss": 1.2399,
       "step": 25620
     },
     {
       "epoch": 76.97,
-      "grad_norm": 5.371367931365967,
+      "grad_norm": 7.609396934509277,
       "learning_rate": 7.434434434434435e-06,
-      "loss": 1.1567,
+      "loss": 1.153,
       "step": 25630
     },
     {
       "epoch": 77.0,
-      "grad_norm": 9.414779663085938,
+      "grad_norm": 7.026651382446289,
       "learning_rate": 7.433433433433434e-06,
-      "loss": 1.303,
+      "loss": 1.3054,
       "step": 25640
     },
     {
       "epoch": 77.0,
-      "eval_accuracy": 0.7846,
-      "eval_loss": 0.7014651894569397,
-      "eval_runtime": 5.6713,
-      "eval_samples_per_second": 1763.276,
-      "eval_steps_per_second": 7.053,
+      "eval_accuracy": 0.7836,
+      "eval_loss": 0.7009608149528503,
+      "eval_runtime": 5.4567,
+      "eval_samples_per_second": 1832.614,
+      "eval_steps_per_second": 7.33,
       "step": 25641
     },
     {
       "epoch": 77.03,
-      "grad_norm": 6.719480514526367,
+      "grad_norm": 6.275604724884033,
       "learning_rate": 7.4324324324324324e-06,
-      "loss": 1.3142,
+      "loss": 1.3411,
       "step": 25650
     },
     {
       "epoch": 77.06,
-      "grad_norm": 7.0363922119140625,
+      "grad_norm": 8.256895065307617,
       "learning_rate": 7.431431431431432e-06,
-      "loss": 1.2601,
+      "loss": 1.2633,
       "step": 25660
     },
     {
       "epoch": 77.09,
-      "grad_norm": 7.155749320983887,
+      "grad_norm": 7.020230770111084,
       "learning_rate": 7.4304304304304305e-06,
-      "loss": 1.215,
+      "loss": 1.2149,
       "step": 25670
     },
     {
       "epoch": 77.12,
-      "grad_norm": 7.8070902824401855,
+      "grad_norm": 7.762746334075928,
       "learning_rate": 7.42942942942943e-06,
-      "loss": 1.1753,
+      "loss": 1.1656,
       "step": 25680
     },
     {
       "epoch": 77.15,
-      "grad_norm": 6.912813663482666,
+      "grad_norm": 9.05882453918457,
       "learning_rate": 7.428428428428429e-06,
-      "loss": 1.1703,
+      "loss": 1.1602,
       "step": 25690
     },
     {
       "epoch": 77.18,
-      "grad_norm": 8.235419273376465,
+      "grad_norm": 7.116147994995117,
       "learning_rate": 7.427427427427428e-06,
-      "loss": 1.1802,
+      "loss": 1.1814,
       "step": 25700
     },
     {
       "epoch": 77.21,
-      "grad_norm": 10.827478408813477,
+      "grad_norm": 12.038031578063965,
       "learning_rate": 7.426426426426427e-06,
-      "loss": 1.1879,
+      "loss": 1.1684,
       "step": 25710
     },
     {
       "epoch": 77.24,
-      "grad_norm": 6.081351280212402,
+      "grad_norm": 6.054988384246826,
       "learning_rate": 7.425425425425427e-06,
-      "loss": 1.1786,
+      "loss": 1.1846,
       "step": 25720
     },
     {
       "epoch": 77.27,
-      "grad_norm": 6.425838947296143,
+      "grad_norm": 6.22188663482666,
       "learning_rate": 7.424424424424425e-06,
-      "loss": 1.1302,
+      "loss": 1.1248,
       "step": 25730
     },
     {
       "epoch": 77.3,
-      "grad_norm": 6.729597568511963,
+      "grad_norm": 8.843134880065918,
       "learning_rate": 7.423423423423424e-06,
-      "loss": 1.2344,
+      "loss": 1.2289,
       "step": 25740
     },
     {
       "epoch": 77.33,
-      "grad_norm": 8.92988395690918,
+      "grad_norm": 9.60225772857666,
       "learning_rate": 7.422422422422423e-06,
-      "loss": 1.1529,
+      "loss": 1.1533,
       "step": 25750
     },
     {
       "epoch": 77.36,
-      "grad_norm": 8.592981338500977,
+      "grad_norm": 11.120038986206055,
       "learning_rate": 7.421421421421422e-06,
-      "loss": 1.1747,
+      "loss": 1.171,
       "step": 25760
     },
     {
       "epoch": 77.39,
-      "grad_norm": 7.053458213806152,
+      "grad_norm": 8.095939636230469,
       "learning_rate": 7.4204204204204215e-06,
-      "loss": 1.2989,
+      "loss": 1.2944,
       "step": 25770
     },
     {
       "epoch": 77.42,
-      "grad_norm": 5.484196186065674,
+      "grad_norm": 7.5061469078063965,
       "learning_rate": 7.41941941941942e-06,
-      "loss": 1.2363,
+      "loss": 1.2328,
       "step": 25780
     },
     {
       "epoch": 77.45,
-      "grad_norm": 5.202413082122803,
+      "grad_norm": 6.117792129516602,
       "learning_rate": 7.4184184184184195e-06,
-      "loss": 1.1979,
+      "loss": 1.1951,
       "step": 25790
     },
     {
       "epoch": 77.48,
-      "grad_norm": 7.850091934204102,
+      "grad_norm": 6.488083839416504,
       "learning_rate": 7.417417417417418e-06,
-      "loss": 1.2429,
+      "loss": 1.2486,
       "step": 25800
     },
     {
       "epoch": 77.51,
-      "grad_norm": 7.398477077484131,
+      "grad_norm": 7.390163898468018,
       "learning_rate": 7.416416416416417e-06,
-      "loss": 1.1847,
+      "loss": 1.1849,
       "step": 25810
     },
     {
       "epoch": 77.54,
-      "grad_norm": 9.213303565979004,
+      "grad_norm": 7.278619289398193,
       "learning_rate": 7.415415415415415e-06,
-      "loss": 1.2097,
+      "loss": 1.2071,
       "step": 25820
     },
     {
       "epoch": 77.57,
-      "grad_norm": 6.633077144622803,
+      "grad_norm": 8.148453712463379,
       "learning_rate": 7.414414414414415e-06,
-      "loss": 1.2778,
+      "loss": 1.2804,
       "step": 25830
     },
     {
       "epoch": 77.6,
-      "grad_norm": 7.280057907104492,
+      "grad_norm": 7.4533915519714355,
       "learning_rate": 7.413413413413414e-06,
-      "loss": 1.2279,
+      "loss": 1.2261,
       "step": 25840
     },
     {
       "epoch": 77.63,
-      "grad_norm": 5.404929161071777,
+      "grad_norm": 8.336442947387695,
       "learning_rate": 7.412412412412413e-06,
-      "loss": 1.231,
+      "loss": 1.2407,
       "step": 25850
     },
     {
       "epoch": 77.66,
-      "grad_norm": 8.040717124938965,
+      "grad_norm": 7.224303722381592,
       "learning_rate": 7.411411411411412e-06,
-      "loss": 1.1773,
+      "loss": 1.1738,
       "step": 25860
     },
     {
       "epoch": 77.69,
-      "grad_norm": 7.01876163482666,
+      "grad_norm": 9.738655090332031,
       "learning_rate": 7.410410410410412e-06,
-      "loss": 1.3242,
+      "loss": 1.3309,
       "step": 25870
     },
     {
       "epoch": 77.72,
-      "grad_norm": 5.5880656242370605,
+      "grad_norm": 4.839864253997803,
       "learning_rate": 7.40940940940941e-06,
-      "loss": 1.1884,
+      "loss": 1.1754,
       "step": 25880
     },
     {
       "epoch": 77.75,
-      "grad_norm": 7.244921684265137,
+      "grad_norm": 8.373974800109863,
       "learning_rate": 7.408408408408409e-06,
-      "loss": 1.1853,
+      "loss": 1.1788,
       "step": 25890
     },
     {
       "epoch": 77.78,
-      "grad_norm": 8.152103424072266,
+      "grad_norm": 6.067312717437744,
       "learning_rate": 7.4074074074074075e-06,
-      "loss": 1.1745,
+      "loss": 1.1768,
       "step": 25900
     },
     {
       "epoch": 77.81,
-      "grad_norm": 6.818478584289551,
+      "grad_norm": 7.037416934967041,
       "learning_rate": 7.406406406406407e-06,
-      "loss": 1.2482,
+      "loss": 1.2443,
       "step": 25910
     },
     {
       "epoch": 77.84,
-      "grad_norm": 6.754003047943115,
+      "grad_norm": 8.015291213989258,
       "learning_rate": 7.4054054054054055e-06,
-      "loss": 1.232,
+      "loss": 1.227,
       "step": 25920
     },
     {
       "epoch": 77.87,
-      "grad_norm": 6.271683692932129,
+      "grad_norm": 7.575611591339111,
       "learning_rate": 7.404404404404405e-06,
-      "loss": 1.2047,
+      "loss": 1.2194,
       "step": 25930
     },
     {
       "epoch": 77.9,
-      "grad_norm": 6.8244547843933105,
+      "grad_norm": 5.4627909660339355,
       "learning_rate": 7.403403403403404e-06,
-      "loss": 1.2412,
+      "loss": 1.232,
       "step": 25940
     },
     {
       "epoch": 77.93,
-      "grad_norm": 8.594802856445312,
+      "grad_norm": 8.18409252166748,
       "learning_rate": 7.402402402402403e-06,
-      "loss": 1.2951,
+      "loss": 1.2936,
       "step": 25950
     },
     {
       "epoch": 77.96,
-      "grad_norm": 7.0907206535339355,
+      "grad_norm": 10.023820877075195,
       "learning_rate": 7.4014014014014024e-06,
-      "loss": 1.2057,
+      "loss": 1.2049,
       "step": 25960
     },
     {
       "epoch": 77.99,
-      "grad_norm": 7.6922807693481445,
+      "grad_norm": 8.989126205444336,
       "learning_rate": 7.400400400400402e-06,
-      "loss": 1.2131,
+      "loss": 1.2219,
       "step": 25970
     },
     {
       "epoch": 78.0,
-      "eval_accuracy": 0.7861,
-      "eval_loss": 0.6963793039321899,
-      "eval_runtime": 5.2895,
-      "eval_samples_per_second": 1890.527,
-      "eval_steps_per_second": 7.562,
+      "eval_accuracy": 0.7853,
+      "eval_loss": 0.693243682384491,
+      "eval_runtime": 5.201,
+      "eval_samples_per_second": 1922.711,
+      "eval_steps_per_second": 7.691,
       "step": 25974
     },
     {
       "epoch": 78.02,
-      "grad_norm": 7.936092376708984,
+      "grad_norm": 8.065372467041016,
       "learning_rate": 7.3993993993994e-06,
-      "loss": 1.3973,
+      "loss": 1.4072,
       "step": 25980
     },
     {
       "epoch": 78.05,
-      "grad_norm": 6.723018169403076,
+      "grad_norm": 7.5700602531433105,
       "learning_rate": 7.398398398398399e-06,
-      "loss": 1.2627,
+      "loss": 1.2678,
       "step": 25990
     },
     {
       "epoch": 78.08,
-      "grad_norm": 5.500950813293457,
+      "grad_norm": 8.14655876159668,
       "learning_rate": 7.397397397397398e-06,
-      "loss": 1.2142,
+      "loss": 1.2033,
       "step": 26000
     },
     {
       "epoch": 78.11,
-      "grad_norm": 6.991573333740234,
+      "grad_norm": 5.825113296508789,
       "learning_rate": 7.396396396396397e-06,
-      "loss": 1.1803,
+      "loss": 1.1699,
       "step": 26010
     },
     {
       "epoch": 78.14,
-      "grad_norm": 8.259876251220703,
+      "grad_norm": 8.205690383911133,
       "learning_rate": 7.395395395395396e-06,
-      "loss": 1.1389,
+      "loss": 1.1309,
       "step": 26020
     },
     {
       "epoch": 78.17,
-      "grad_norm": 7.106561660766602,
+      "grad_norm": 8.415581703186035,
       "learning_rate": 7.394394394394395e-06,
-      "loss": 1.2397,
+      "loss": 1.2283,
       "step": 26030
     },
     {
       "epoch": 78.2,
-      "grad_norm": 6.7721123695373535,
+      "grad_norm": 7.850139141082764,
       "learning_rate": 7.393393393393395e-06,
-      "loss": 1.1981,
+      "loss": 1.1953,
       "step": 26040
     },
     {
       "epoch": 78.23,
-      "grad_norm": 7.620425224304199,
+      "grad_norm": 8.605217933654785,
       "learning_rate": 7.392392392392393e-06,
-      "loss": 1.2238,
+      "loss": 1.2175,
       "step": 26050
     },
     {
       "epoch": 78.26,
-      "grad_norm": 7.755762100219727,
+      "grad_norm": 7.6903791427612305,
       "learning_rate": 7.391391391391392e-06,
-      "loss": 1.2198,
+      "loss": 1.2231,
       "step": 26060
     },
     {
       "epoch": 78.29,
-      "grad_norm": 9.32624626159668,
+      "grad_norm": 10.177104949951172,
       "learning_rate": 7.39039039039039e-06,
-      "loss": 1.2811,
+      "loss": 1.2806,
       "step": 26070
     },
     {
       "epoch": 78.32,
-      "grad_norm": 8.325776100158691,
+      "grad_norm": 6.674909591674805,
       "learning_rate": 7.38938938938939e-06,
-      "loss": 1.2518,
+      "loss": 1.2486,
       "step": 26080
     },
     {
       "epoch": 78.35,
-      "grad_norm": 9.951125144958496,
+      "grad_norm": 9.422161102294922,
       "learning_rate": 7.388388388388389e-06,
-      "loss": 1.2549,
+      "loss": 1.2538,
       "step": 26090
     },
     {
       "epoch": 78.38,
-      "grad_norm": 6.1726202964782715,
+      "grad_norm": 8.325309753417969,
       "learning_rate": 7.387387387387388e-06,
-      "loss": 1.2436,
+      "loss": 1.2475,
       "step": 26100
     },
     {
       "epoch": 78.41,
-      "grad_norm": 6.232406139373779,
+      "grad_norm": 7.4293975830078125,
       "learning_rate": 7.386386386386387e-06,
-      "loss": 1.2374,
+      "loss": 1.2367,
       "step": 26110
     },
     {
       "epoch": 78.44,
-      "grad_norm": 6.525746822357178,
+      "grad_norm": 9.157160758972168,
       "learning_rate": 7.385385385385386e-06,
-      "loss": 1.1524,
+      "loss": 1.1542,
       "step": 26120
     },
     {
       "epoch": 78.47,
-      "grad_norm": 14.140703201293945,
+      "grad_norm": 12.84088134765625,
       "learning_rate": 7.384384384384385e-06,
-      "loss": 1.1481,
+      "loss": 1.1438,
       "step": 26130
     },
     {
       "epoch": 78.5,
-      "grad_norm": 7.185839653015137,
+      "grad_norm": 7.332485198974609,
       "learning_rate": 7.383383383383383e-06,
-      "loss": 1.2237,
+      "loss": 1.2284,
       "step": 26140
     },
     {
       "epoch": 78.53,
-      "grad_norm": 8.144439697265625,
+      "grad_norm": 8.884510040283203,
       "learning_rate": 7.3823823823823825e-06,
-      "loss": 1.2069,
+      "loss": 1.2126,
       "step": 26150
     },
     {
       "epoch": 78.56,
-      "grad_norm": 8.067972183227539,
+      "grad_norm": 7.0522637367248535,
       "learning_rate": 7.381381381381382e-06,
-      "loss": 1.2116,
+      "loss": 1.2053,
       "step": 26160
     },
     {
       "epoch": 78.59,
-      "grad_norm": 8.224495887756348,
+      "grad_norm": 8.21837329864502,
       "learning_rate": 7.3803803803803806e-06,
-      "loss": 1.1865,
+      "loss": 1.1825,
       "step": 26170
     },
     {
       "epoch": 78.62,
-      "grad_norm": 6.420608997344971,
+      "grad_norm": 8.051129341125488,
       "learning_rate": 7.37937937937938e-06,
-      "loss": 1.2324,
+      "loss": 1.2268,
       "step": 26180
     },
     {
       "epoch": 78.65,
-      "grad_norm": 5.298826694488525,
+      "grad_norm": 6.556152820587158,
       "learning_rate": 7.3783783783783794e-06,
-      "loss": 1.1914,
+      "loss": 1.1847,
       "step": 26190
     },
     {
       "epoch": 78.68,
-      "grad_norm": 8.77259349822998,
+      "grad_norm": 8.506532669067383,
       "learning_rate": 7.377377377377378e-06,
-      "loss": 1.2364,
+      "loss": 1.2322,
       "step": 26200
     },
     {
       "epoch": 78.71,
-      "grad_norm": 7.1451568603515625,
+      "grad_norm": 9.658242225646973,
       "learning_rate": 7.3763763763763775e-06,
-      "loss": 1.2123,
+      "loss": 1.2077,
       "step": 26210
     },
     {
       "epoch": 78.74,
-      "grad_norm": 8.914092063903809,
+      "grad_norm": 10.453828811645508,
       "learning_rate": 7.375375375375377e-06,
-      "loss": 1.2657,
+      "loss": 1.2609,
       "step": 26220
     },
     {
       "epoch": 78.77,
-      "grad_norm": 5.20422887802124,
+      "grad_norm": 4.503804683685303,
       "learning_rate": 7.374374374374375e-06,
-      "loss": 1.1758,
+      "loss": 1.1709,
       "step": 26230
     },
     {
       "epoch": 78.8,
-      "grad_norm": 6.359464168548584,
+      "grad_norm": 5.262405872344971,
       "learning_rate": 7.373373373373373e-06,
-      "loss": 1.2321,
+      "loss": 1.2307,
       "step": 26240
     },
     {
       "epoch": 78.83,
-      "grad_norm": 7.021862983703613,
+      "grad_norm": 6.386959552764893,
       "learning_rate": 7.372372372372373e-06,
-      "loss": 1.1867,
+      "loss": 1.1958,
       "step": 26250
     },
     {
       "epoch": 78.86,
-      "grad_norm": 5.393792152404785,
+      "grad_norm": 6.592474937438965,
       "learning_rate": 7.371371371371372e-06,
-      "loss": 1.2513,
+      "loss": 1.2567,
       "step": 26260
     },
     {
       "epoch": 78.89,
-      "grad_norm": 8.970477104187012,
+      "grad_norm": 10.645689964294434,
       "learning_rate": 7.370370370370371e-06,
-      "loss": 1.1337,
+      "loss": 1.1389,
       "step": 26270
     },
     {
       "epoch": 78.92,
-      "grad_norm": 5.375817775726318,
+      "grad_norm": 5.927525043487549,
       "learning_rate": 7.36936936936937e-06,
-      "loss": 1.1786,
+      "loss": 1.171,
       "step": 26280
     },
     {
       "epoch": 78.95,
-      "grad_norm": 6.825578212738037,
+      "grad_norm": 7.881835460662842,
       "learning_rate": 7.36836836836837e-06,
-      "loss": 1.1809,
+      "loss": 1.1733,
       "step": 26290
     },
     {
       "epoch": 78.98,
-      "grad_norm": 9.20656681060791,
+      "grad_norm": 7.555628776550293,
       "learning_rate": 7.367367367367368e-06,
-      "loss": 1.1887,
+      "loss": 1.1748,
       "step": 26300
     },
     {
       "epoch": 79.0,
-      "eval_accuracy": 0.7867,
-      "eval_loss": 0.6876847743988037,
-      "eval_runtime": 5.642,
-      "eval_samples_per_second": 1772.422,
-      "eval_steps_per_second": 7.09,
+      "eval_accuracy": 0.7905,
+      "eval_loss": 0.6841825246810913,
+      "eval_runtime": 5.3332,
+      "eval_samples_per_second": 1875.059,
+      "eval_steps_per_second": 7.5,
       "step": 26307
     },
     {
       "epoch": 79.01,
-      "grad_norm": 6.517043113708496,
+      "grad_norm": 6.257315158843994,
       "learning_rate": 7.366366366366367e-06,
-      "loss": 1.2015,
+      "loss": 1.1965,
       "step": 26310
     },
     {
       "epoch": 79.04,
-      "grad_norm": 6.117551803588867,
+      "grad_norm": 6.181555271148682,
       "learning_rate": 7.365365365365365e-06,
-      "loss": 1.1902,
+      "loss": 1.1876,
       "step": 26320
     },
     {
       "epoch": 79.07,
-      "grad_norm": 9.901209831237793,
+      "grad_norm": 15.216564178466797,
       "learning_rate": 7.364364364364365e-06,
-      "loss": 1.2511,
+      "loss": 1.2557,
       "step": 26330
     },
     {
       "epoch": 79.1,
-      "grad_norm": 9.593603134155273,
+      "grad_norm": 8.49368667602539,
       "learning_rate": 7.363363363363364e-06,
-      "loss": 1.2343,
+      "loss": 1.2282,
       "step": 26340
     },
     {
       "epoch": 79.13,
-      "grad_norm": 9.56380844116211,
+      "grad_norm": 8.901236534118652,
       "learning_rate": 7.362362362362363e-06,
-      "loss": 1.1301,
+      "loss": 1.1235,
       "step": 26350
     },
     {
       "epoch": 79.16,
-      "grad_norm": 9.419458389282227,
+      "grad_norm": 7.869558811187744,
       "learning_rate": 7.361361361361362e-06,
-      "loss": 1.2725,
+      "loss": 1.267,
       "step": 26360
     },
     {
       "epoch": 79.19,
-      "grad_norm": 9.194195747375488,
+      "grad_norm": 7.684288501739502,
       "learning_rate": 7.360360360360361e-06,
-      "loss": 1.164,
+      "loss": 1.165,
       "step": 26370
     },
     {
       "epoch": 79.22,
-      "grad_norm": 6.017014026641846,
+      "grad_norm": 9.378746032714844,
       "learning_rate": 7.35935935935936e-06,
-      "loss": 1.1764,
+      "loss": 1.1765,
       "step": 26380
     },
     {
       "epoch": 79.25,
-      "grad_norm": 5.803622245788574,
+      "grad_norm": 6.878758907318115,
       "learning_rate": 7.358358358358358e-06,
-      "loss": 1.2055,
+      "loss": 1.2021,
       "step": 26390
     },
     {
       "epoch": 79.28,
-      "grad_norm": 7.884670257568359,
+      "grad_norm": 10.952702522277832,
       "learning_rate": 7.3573573573573575e-06,
-      "loss": 1.2707,
+      "loss": 1.2772,
       "step": 26400
     },
     {
       "epoch": 79.31,
-      "grad_norm": 7.625174522399902,
+      "grad_norm": 9.045029640197754,
       "learning_rate": 7.356356356356357e-06,
-      "loss": 1.1486,
+      "loss": 1.1512,
       "step": 26410
     },
     {
       "epoch": 79.34,
-      "grad_norm": 7.292159557342529,
+      "grad_norm": 9.083837509155273,
       "learning_rate": 7.355355355355356e-06,
-      "loss": 1.2266,
+      "loss": 1.2221,
       "step": 26420
     },
     {
       "epoch": 79.37,
-      "grad_norm": 7.009256839752197,
+      "grad_norm": 7.6885809898376465,
       "learning_rate": 7.354354354354355e-06,
-      "loss": 1.2175,
+      "loss": 1.2072,
       "step": 26430
     },
     {
       "epoch": 79.4,
-      "grad_norm": 6.585838794708252,
+      "grad_norm": 8.175765991210938,
       "learning_rate": 7.3533533533533545e-06,
-      "loss": 1.2097,
+      "loss": 1.2195,
       "step": 26440
     },
     {
       "epoch": 79.43,
-      "grad_norm": 7.10453987121582,
+      "grad_norm": 9.85254955291748,
       "learning_rate": 7.352352352352353e-06,
-      "loss": 1.203,
+      "loss": 1.2035,
       "step": 26450
     },
     {
       "epoch": 79.46,
-      "grad_norm": 8.235066413879395,
+      "grad_norm": 8.713233947753906,
       "learning_rate": 7.3513513513513525e-06,
-      "loss": 1.1672,
+      "loss": 1.1698,
       "step": 26460
     },
     {
       "epoch": 79.49,
-      "grad_norm": 9.144484519958496,
+      "grad_norm": 9.421059608459473,
       "learning_rate": 7.350350350350351e-06,
-      "loss": 1.1733,
+      "loss": 1.1653,
       "step": 26470
     },
     {
       "epoch": 79.52,
-      "grad_norm": 7.125712871551514,
+      "grad_norm": 7.0685319900512695,
       "learning_rate": 7.34934934934935e-06,
-      "loss": 1.2285,
+      "loss": 1.2354,
       "step": 26480
     },
     {
       "epoch": 79.55,
-      "grad_norm": 6.02527379989624,
+      "grad_norm": 7.5208916664123535,
       "learning_rate": 7.348348348348348e-06,
-      "loss": 1.1787,
+      "loss": 1.1661,
       "step": 26490
     },
     {
       "epoch": 79.58,
-      "grad_norm": 7.420890808105469,
+      "grad_norm": 7.593235015869141,
       "learning_rate": 7.347347347347348e-06,
-      "loss": 1.1537,
+      "loss": 1.1498,
       "step": 26500
     },
     {
       "epoch": 79.61,
-      "grad_norm": 5.511213779449463,
+      "grad_norm": 7.603847980499268,
       "learning_rate": 7.346346346346347e-06,
-      "loss": 1.1252,
+      "loss": 1.1087,
       "step": 26510
     },
     {
       "epoch": 79.64,
-      "grad_norm": 5.070666313171387,
+      "grad_norm": 7.372706890106201,
       "learning_rate": 7.345345345345346e-06,
-      "loss": 1.2323,
+      "loss": 1.2346,
       "step": 26520
     },
     {
       "epoch": 79.67,
-      "grad_norm": 7.078922271728516,
+      "grad_norm": 8.248793601989746,
       "learning_rate": 7.344344344344345e-06,
-      "loss": 1.1555,
+      "loss": 1.15,
       "step": 26530
     },
     {
       "epoch": 79.7,
-      "grad_norm": 9.28626823425293,
+      "grad_norm": 10.263312339782715,
       "learning_rate": 7.343343343343345e-06,
-      "loss": 1.3187,
+      "loss": 1.31,
       "step": 26540
     },
     {
       "epoch": 79.73,
-      "grad_norm": 7.898387432098389,
+      "grad_norm": 7.919864654541016,
       "learning_rate": 7.342342342342343e-06,
-      "loss": 1.1846,
+      "loss": 1.1705,
       "step": 26550
     },
     {
       "epoch": 79.76,
-      "grad_norm": 6.523582935333252,
+      "grad_norm": 7.113073348999023,
       "learning_rate": 7.341341341341342e-06,
-      "loss": 1.1682,
+      "loss": 1.1696,
       "step": 26560
     },
     {
       "epoch": 79.79,
-      "grad_norm": 8.814983367919922,
+      "grad_norm": 10.082690238952637,
       "learning_rate": 7.3403403403403404e-06,
-      "loss": 1.1705,
+      "loss": 1.168,
       "step": 26570
     },
     {
       "epoch": 79.82,
-      "grad_norm": 7.662591457366943,
+      "grad_norm": 5.612902641296387,
       "learning_rate": 7.33933933933934e-06,
-      "loss": 1.2368,
+      "loss": 1.2371,
       "step": 26580
     },
     {
       "epoch": 79.85,
-      "grad_norm": 6.9171223640441895,
+      "grad_norm": 6.750603199005127,
       "learning_rate": 7.3383383383383385e-06,
-      "loss": 1.1953,
+      "loss": 1.2061,
       "step": 26590
     },
     {
       "epoch": 79.88,
-      "grad_norm": 7.173194408416748,
+      "grad_norm": 6.587666988372803,
       "learning_rate": 7.337337337337338e-06,
-      "loss": 1.2272,
+      "loss": 1.2148,
       "step": 26600
     },
     {
       "epoch": 79.91,
-      "grad_norm": 7.0289225578308105,
+      "grad_norm": 9.434632301330566,
       "learning_rate": 7.336336336336337e-06,
-      "loss": 1.2303,
+      "loss": 1.2248,
       "step": 26610
     },
     {
       "epoch": 79.94,
-      "grad_norm": 9.116661071777344,
+      "grad_norm": 14.579792976379395,
       "learning_rate": 7.335335335335336e-06,
-      "loss": 1.2173,
+      "loss": 1.2171,
       "step": 26620
     },
     {
       "epoch": 79.97,
-      "grad_norm": 7.707890510559082,
+      "grad_norm": 7.103947162628174,
       "learning_rate": 7.334334334334335e-06,
-      "loss": 1.1442,
+      "loss": 1.151,
       "step": 26630
     },
     {
       "epoch": 80.0,
-      "grad_norm": 60.500789642333984,
+      "grad_norm": 86.515869140625,
       "learning_rate": 7.333333333333333e-06,
-      "loss": 1.6358,
+      "loss": 1.6392,
       "step": 26640
     },
     {
       "epoch": 80.0,
-      "eval_accuracy": 0.7872,
-      "eval_loss": 0.6987437009811401,
-      "eval_runtime": 5.6176,
-      "eval_samples_per_second": 1780.13,
-      "eval_steps_per_second": 7.121,
+      "eval_accuracy": 0.7856,
+      "eval_loss": 0.6997856497764587,
+      "eval_runtime": 5.3508,
+      "eval_samples_per_second": 1868.874,
+      "eval_steps_per_second": 7.475,
       "step": 26640
     },
     {
       "epoch": 80.03,
-      "grad_norm": 6.704030990600586,
+      "grad_norm": 9.57343578338623,
       "learning_rate": 7.332332332332333e-06,
-      "loss": 1.2654,
+      "loss": 1.275,
       "step": 26650
     },
     {
       "epoch": 80.06,
-      "grad_norm": 7.94455099105835,
+      "grad_norm": 10.278660774230957,
       "learning_rate": 7.331331331331332e-06,
-      "loss": 1.1894,
+      "loss": 1.1898,
       "step": 26660
     },
     {
       "epoch": 80.09,
-      "grad_norm": 7.246967792510986,
+      "grad_norm": 8.019506454467773,
       "learning_rate": 7.330330330330331e-06,
-      "loss": 1.2199,
+      "loss": 1.2144,
       "step": 26670
     },
     {
       "epoch": 80.12,
-      "grad_norm": 6.096104145050049,
+      "grad_norm": 6.206096649169922,
       "learning_rate": 7.32932932932933e-06,
-      "loss": 1.2584,
+      "loss": 1.2494,
       "step": 26680
     },
     {
       "epoch": 80.15,
-      "grad_norm": 6.896922588348389,
+      "grad_norm": 5.572656631469727,
       "learning_rate": 7.328328328328329e-06,
-      "loss": 1.2066,
+      "loss": 1.2058,
       "step": 26690
     },
     {
       "epoch": 80.18,
-      "grad_norm": 8.788232803344727,
+      "grad_norm": 11.424630165100098,
       "learning_rate": 7.327327327327328e-06,
-      "loss": 1.2576,
+      "loss": 1.2624,
       "step": 26700
     },
     {
       "epoch": 80.21,
-      "grad_norm": 5.477240562438965,
+      "grad_norm": 5.876471042633057,
       "learning_rate": 7.3263263263263275e-06,
-      "loss": 1.2226,
+      "loss": 1.2202,
       "step": 26710
     },
     {
       "epoch": 80.24,
-      "grad_norm": 7.632427215576172,
+      "grad_norm": 6.572718620300293,
       "learning_rate": 7.325325325325326e-06,
-      "loss": 1.1589,
+      "loss": 1.1621,
       "step": 26720
     },
     {
       "epoch": 80.27,
-      "grad_norm": 6.2430853843688965,
+      "grad_norm": 7.620352745056152,
       "learning_rate": 7.324324324324325e-06,
-      "loss": 1.2126,
+      "loss": 1.2095,
       "step": 26730
     },
     {
       "epoch": 80.3,
-      "grad_norm": 7.267426013946533,
+      "grad_norm": 7.459954261779785,
       "learning_rate": 7.323323323323323e-06,
-      "loss": 1.1747,
+      "loss": 1.1683,
       "step": 26740
     },
     {
       "epoch": 80.33,
-      "grad_norm": 6.667952537536621,
+      "grad_norm": 6.458949565887451,
       "learning_rate": 7.322322322322323e-06,
-      "loss": 1.2127,
+      "loss": 1.2148,
       "step": 26750
     },
     {
       "epoch": 80.36,
-      "grad_norm": 7.747252941131592,
+      "grad_norm": 9.532770156860352,
       "learning_rate": 7.321321321321322e-06,
-      "loss": 1.2638,
+      "loss": 1.2643,
       "step": 26760
     },
     {
       "epoch": 80.39,
-      "grad_norm": 7.077569007873535,
+      "grad_norm": 6.470544338226318,
       "learning_rate": 7.320320320320321e-06,
-      "loss": 1.1474,
+      "loss": 1.1503,
       "step": 26770
     },
     {
       "epoch": 80.42,
-      "grad_norm": 8.00772476196289,
+      "grad_norm": 7.889050006866455,
       "learning_rate": 7.31931931931932e-06,
-      "loss": 1.2588,
+      "loss": 1.2556,
       "step": 26780
     },
     {
       "epoch": 80.45,
-      "grad_norm": 8.191335678100586,
+      "grad_norm": 9.52310848236084,
       "learning_rate": 7.31831831831832e-06,
-      "loss": 1.2338,
+      "loss": 1.2366,
       "step": 26790
     },
     {
       "epoch": 80.48,
-      "grad_norm": 6.375512599945068,
+      "grad_norm": 6.191443920135498,
       "learning_rate": 7.317317317317318e-06,
-      "loss": 1.1563,
+      "loss": 1.173,
       "step": 26800
     },
     {
       "epoch": 80.51,
-      "grad_norm": 5.954040050506592,
+      "grad_norm": 7.5931925773620605,
       "learning_rate": 7.316316316316316e-06,
-      "loss": 1.1257,
+      "loss": 1.1178,
       "step": 26810
     },
     {
       "epoch": 80.54,
-      "grad_norm": 7.857104301452637,
+      "grad_norm": 7.464622497558594,
       "learning_rate": 7.3153153153153155e-06,
-      "loss": 1.1971,
+      "loss": 1.183,
       "step": 26820
     },
     {
       "epoch": 80.57,
-      "grad_norm": 5.774392604827881,
+      "grad_norm": 5.59458589553833,
       "learning_rate": 7.314314314314315e-06,
-      "loss": 1.1969,
+      "loss": 1.1837,
       "step": 26830
     },
     {
       "epoch": 80.6,
-      "grad_norm": 6.773274898529053,
+      "grad_norm": 8.020608901977539,
       "learning_rate": 7.3133133133133135e-06,
-      "loss": 1.2167,
+      "loss": 1.205,
       "step": 26840
     },
     {
       "epoch": 80.63,
-      "grad_norm": 8.461565017700195,
+      "grad_norm": 7.0001349449157715,
       "learning_rate": 7.312312312312313e-06,
-      "loss": 1.1499,
+      "loss": 1.1489,
       "step": 26850
     },
     {
       "epoch": 80.66,
-      "grad_norm": 6.409885883331299,
+      "grad_norm": 7.345103740692139,
       "learning_rate": 7.311311311311312e-06,
-      "loss": 1.2786,
+      "loss": 1.2798,
       "step": 26860
     },
     {
       "epoch": 80.69,
-      "grad_norm": 7.84691858291626,
+      "grad_norm": 8.526427268981934,
       "learning_rate": 7.310310310310311e-06,
-      "loss": 1.2261,
+      "loss": 1.2281,
       "step": 26870
     },
     {
       "epoch": 80.72,
-      "grad_norm": 8.167898178100586,
+      "grad_norm": 6.665166854858398,
       "learning_rate": 7.3093093093093104e-06,
-      "loss": 1.2618,
+      "loss": 1.2515,
       "step": 26880
     },
     {
       "epoch": 80.75,
-      "grad_norm": 7.3052287101745605,
+      "grad_norm": 6.685031890869141,
       "learning_rate": 7.308308308308308e-06,
-      "loss": 1.1374,
+      "loss": 1.1419,
       "step": 26890
     },
     {
       "epoch": 80.78,
-      "grad_norm": 5.6845903396606445,
+      "grad_norm": 7.00918436050415,
       "learning_rate": 7.307307307307308e-06,
-      "loss": 1.2237,
+      "loss": 1.2183,
       "step": 26900
     },
     {
       "epoch": 80.81,
-      "grad_norm": 9.162962913513184,
+      "grad_norm": 8.49411678314209,
       "learning_rate": 7.306306306306307e-06,
-      "loss": 1.1331,
+      "loss": 1.1321,
       "step": 26910
     },
     {
       "epoch": 80.84,
-      "grad_norm": 7.320842266082764,
+      "grad_norm": 7.245256423950195,
       "learning_rate": 7.305305305305306e-06,
-      "loss": 1.2124,
+      "loss": 1.2199,
       "step": 26920
     },
     {
       "epoch": 80.87,
-      "grad_norm": 6.324519157409668,
+      "grad_norm": 7.1462082862854,
       "learning_rate": 7.304304304304305e-06,
-      "loss": 1.1883,
+      "loss": 1.1885,
       "step": 26930
     },
     {
       "epoch": 80.9,
-      "grad_norm": 7.940248012542725,
+      "grad_norm": 7.885055065155029,
       "learning_rate": 7.303303303303304e-06,
-      "loss": 1.2435,
+      "loss": 1.2576,
       "step": 26940
     },
     {
       "epoch": 80.93,
-      "grad_norm": 6.932700157165527,
+      "grad_norm": 7.263689994812012,
       "learning_rate": 7.302302302302303e-06,
-      "loss": 1.1576,
+      "loss": 1.1552,
       "step": 26950
     },
     {
       "epoch": 80.96,
-      "grad_norm": 5.972362995147705,
+      "grad_norm": 8.202275276184082,
       "learning_rate": 7.3013013013013026e-06,
-      "loss": 1.0649,
+      "loss": 1.0665,
       "step": 26960
     },
     {
       "epoch": 80.99,
-      "grad_norm": 5.97263765335083,
+      "grad_norm": 8.735823631286621,
       "learning_rate": 7.3003003003003e-06,
-      "loss": 1.1976,
+      "loss": 1.2048,
       "step": 26970
     },
     {
       "epoch": 81.0,
-      "eval_accuracy": 0.7887,
-      "eval_loss": 0.6890700459480286,
-      "eval_runtime": 5.6158,
-      "eval_samples_per_second": 1780.689,
-      "eval_steps_per_second": 7.123,
+      "eval_accuracy": 0.7892,
+      "eval_loss": 0.6882181167602539,
+      "eval_runtime": 5.3875,
+      "eval_samples_per_second": 1856.163,
+      "eval_steps_per_second": 7.425,
       "step": 26973
     },
     {
       "epoch": 81.02,
-      "grad_norm": 6.917881488800049,
+      "grad_norm": 8.716523170471191,
       "learning_rate": 7.2992992992993e-06,
-      "loss": 1.3359,
+      "loss": 1.334,
       "step": 26980
     },
     {
       "epoch": 81.05,
-      "grad_norm": 9.29307746887207,
+      "grad_norm": 6.933269023895264,
       "learning_rate": 7.298298298298298e-06,
-      "loss": 1.2008,
+      "loss": 1.2,
       "step": 26990
     },
     {
       "epoch": 81.08,
-      "grad_norm": 7.962011814117432,
+      "grad_norm": 6.602358818054199,
       "learning_rate": 7.297297297297298e-06,
-      "loss": 1.1436,
+      "loss": 1.1398,
       "step": 27000
     },
     {
       "epoch": 81.11,
-      "grad_norm": 7.279539108276367,
+      "grad_norm": 9.750433921813965,
       "learning_rate": 7.296296296296297e-06,
-      "loss": 1.155,
+      "loss": 1.1511,
       "step": 27010
     },
     {
       "epoch": 81.14,
-      "grad_norm": 6.280298709869385,
+      "grad_norm": 7.003243446350098,
       "learning_rate": 7.295295295295296e-06,
-      "loss": 1.2043,
+      "loss": 1.205,
       "step": 27020
     },
     {
       "epoch": 81.17,
-      "grad_norm": 10.40195083618164,
+      "grad_norm": 6.441761016845703,
       "learning_rate": 7.294294294294295e-06,
-      "loss": 1.1878,
+      "loss": 1.1964,
       "step": 27030
     },
     {
       "epoch": 81.2,
-      "grad_norm": 6.356490135192871,
+      "grad_norm": 10.942736625671387,
       "learning_rate": 7.293293293293294e-06,
-      "loss": 1.2069,
+      "loss": 1.2257,
       "step": 27040
     },
     {
       "epoch": 81.23,
-      "grad_norm": 7.616515636444092,
+      "grad_norm": 8.863377571105957,
       "learning_rate": 7.292292292292293e-06,
-      "loss": 1.1702,
+      "loss": 1.1609,
       "step": 27050
     },
     {
       "epoch": 81.26,
-      "grad_norm": 6.792428016662598,
+      "grad_norm": 6.309885501861572,
       "learning_rate": 7.291291291291291e-06,
-      "loss": 1.1685,
+      "loss": 1.1606,
       "step": 27060
     },
     {
       "epoch": 81.29,
-      "grad_norm": 7.413721561431885,
+      "grad_norm": 7.633962631225586,
       "learning_rate": 7.2902902902902905e-06,
-      "loss": 1.1899,
+      "loss": 1.1792,
       "step": 27070
     },
     {
       "epoch": 81.32,
-      "grad_norm": 5.358495712280273,
+      "grad_norm": 5.1079511642456055,
       "learning_rate": 7.28928928928929e-06,
-      "loss": 1.1776,
+      "loss": 1.1825,
       "step": 27080
     },
     {
       "epoch": 81.35,
-      "grad_norm": 7.856200218200684,
+      "grad_norm": 7.4985480308532715,
       "learning_rate": 7.2882882882882885e-06,
-      "loss": 1.1438,
+      "loss": 1.14,
       "step": 27090
     },
     {
       "epoch": 81.38,
-      "grad_norm": 8.13717269897461,
+      "grad_norm": 6.937688827514648,
       "learning_rate": 7.287287287287288e-06,
-      "loss": 1.1496,
+      "loss": 1.1555,
       "step": 27100
     },
     {
       "epoch": 81.41,
-      "grad_norm": 7.104565620422363,
+      "grad_norm": 8.09473705291748,
       "learning_rate": 7.2862862862862874e-06,
-      "loss": 1.1699,
+      "loss": 1.1682,
       "step": 27110
     },
     {
       "epoch": 81.44,
-      "grad_norm": 7.453733444213867,
+      "grad_norm": 8.042096138000488,
       "learning_rate": 7.285285285285286e-06,
-      "loss": 1.2046,
+      "loss": 1.1976,
       "step": 27120
     },
     {
       "epoch": 81.47,
-      "grad_norm": 10.169500350952148,
+      "grad_norm": 12.881413459777832,
       "learning_rate": 7.2842842842842855e-06,
-      "loss": 1.2894,
+      "loss": 1.2851,
       "step": 27130
     },
     {
       "epoch": 81.5,
-      "grad_norm": 8.555078506469727,
+      "grad_norm": 8.788470268249512,
       "learning_rate": 7.283283283283283e-06,
-      "loss": 1.2406,
+      "loss": 1.2342,
       "step": 27140
     },
     {
       "epoch": 81.53,
-      "grad_norm": 5.583146095275879,
+      "grad_norm": 6.968081474304199,
       "learning_rate": 7.282282282282283e-06,
-      "loss": 1.1264,
+      "loss": 1.1262,
       "step": 27150
     },
     {
       "epoch": 81.56,
-      "grad_norm": 6.392098426818848,
+      "grad_norm": 8.055347442626953,
       "learning_rate": 7.281281281281281e-06,
-      "loss": 1.1578,
+      "loss": 1.1486,
       "step": 27160
     },
     {
       "epoch": 81.59,
-      "grad_norm": 7.120057582855225,
+      "grad_norm": 6.54729700088501,
       "learning_rate": 7.280280280280281e-06,
-      "loss": 1.2166,
+      "loss": 1.2133,
       "step": 27170
     },
     {
       "epoch": 81.62,
-      "grad_norm": 8.103612899780273,
+      "grad_norm": 7.944761753082275,
       "learning_rate": 7.27927927927928e-06,
-      "loss": 1.2434,
+      "loss": 1.239,
       "step": 27180
     },
     {
       "epoch": 81.65,
-      "grad_norm": 7.342371463775635,
+      "grad_norm": 8.697854995727539,
       "learning_rate": 7.278278278278279e-06,
-      "loss": 1.1839,
+      "loss": 1.1641,
       "step": 27190
     },
     {
       "epoch": 81.68,
-      "grad_norm": 6.160196304321289,
+      "grad_norm": 5.244506359100342,
       "learning_rate": 7.277277277277278e-06,
-      "loss": 1.1554,
+      "loss": 1.1567,
       "step": 27200
     },
     {
       "epoch": 81.71,
-      "grad_norm": 6.440067291259766,
+      "grad_norm": 5.64020299911499,
       "learning_rate": 7.276276276276278e-06,
-      "loss": 1.2376,
+      "loss": 1.2314,
       "step": 27210
     },
     {
       "epoch": 81.74,
-      "grad_norm": 6.1443257331848145,
+      "grad_norm": 5.906661510467529,
       "learning_rate": 7.275275275275275e-06,
-      "loss": 1.1227,
+      "loss": 1.1247,
       "step": 27220
     },
     {
       "epoch": 81.77,
-      "grad_norm": 6.73710298538208,
+      "grad_norm": 7.019298553466797,
       "learning_rate": 7.274274274274275e-06,
-      "loss": 1.2693,
+      "loss": 1.2701,
       "step": 27230
     },
     {
       "epoch": 81.8,
-      "grad_norm": 11.956660270690918,
+      "grad_norm": 14.090109825134277,
       "learning_rate": 7.273273273273273e-06,
-      "loss": 1.1949,
+      "loss": 1.1836,
       "step": 27240
     },
     {
       "epoch": 81.83,
-      "grad_norm": 6.391708850860596,
+      "grad_norm": 8.117880821228027,
       "learning_rate": 7.272272272272273e-06,
-      "loss": 1.194,
+      "loss": 1.1931,
       "step": 27250
     },
     {
       "epoch": 81.86,
-      "grad_norm": 5.890604496002197,
+      "grad_norm": 7.233791351318359,
       "learning_rate": 7.271271271271272e-06,
-      "loss": 1.1774,
+      "loss": 1.1719,
       "step": 27260
     },
     {
       "epoch": 81.89,
-      "grad_norm": 9.178711891174316,
+      "grad_norm": 8.390323638916016,
       "learning_rate": 7.270270270270271e-06,
-      "loss": 1.2299,
+      "loss": 1.2135,
       "step": 27270
     },
     {
       "epoch": 81.92,
-      "grad_norm": 6.120874881744385,
+      "grad_norm": 6.415703296661377,
       "learning_rate": 7.26926926926927e-06,
-      "loss": 1.2196,
+      "loss": 1.2053,
       "step": 27280
     },
     {
       "epoch": 81.95,
-      "grad_norm": 8.247115135192871,
+      "grad_norm": 11.033462524414062,
       "learning_rate": 7.268268268268269e-06,
-      "loss": 1.2848,
+      "loss": 1.2865,
       "step": 27290
     },
     {
       "epoch": 81.98,
-      "grad_norm": 7.056839942932129,
+      "grad_norm": 8.624213218688965,
       "learning_rate": 7.267267267267268e-06,
-      "loss": 1.1602,
+      "loss": 1.1607,
       "step": 27300
     },
     {
       "epoch": 82.0,
-      "eval_accuracy": 0.7894,
-      "eval_loss": 0.6797456741333008,
-      "eval_runtime": 5.441,
-      "eval_samples_per_second": 1837.889,
-      "eval_steps_per_second": 7.352,
+      "eval_accuracy": 0.7902,
+      "eval_loss": 0.678310215473175,
+      "eval_runtime": 5.2102,
+      "eval_samples_per_second": 1919.318,
+      "eval_steps_per_second": 7.677,
       "step": 27306
     },
     {
       "epoch": 82.01,
-      "grad_norm": 9.110605239868164,
+      "grad_norm": 7.086184978485107,
       "learning_rate": 7.266266266266266e-06,
-      "loss": 1.4751,
+      "loss": 1.4773,
       "step": 27310
     },
     {
       "epoch": 82.04,
-      "grad_norm": 8.950155258178711,
+      "grad_norm": 8.244267463684082,
       "learning_rate": 7.2652652652652655e-06,
-      "loss": 1.1747,
+      "loss": 1.177,
       "step": 27320
     },
     {
       "epoch": 82.07,
-      "grad_norm": 6.1585798263549805,
+      "grad_norm": 7.378773212432861,
       "learning_rate": 7.264264264264265e-06,
-      "loss": 1.197,
+      "loss": 1.1956,
       "step": 27330
     },
     {
       "epoch": 82.1,
-      "grad_norm": 5.883058071136475,
+      "grad_norm": 5.150636672973633,
       "learning_rate": 7.263263263263264e-06,
-      "loss": 1.2286,
+      "loss": 1.2254,
       "step": 27340
     },
     {
       "epoch": 82.13,
-      "grad_norm": 8.252819061279297,
+      "grad_norm": 11.784801483154297,
       "learning_rate": 7.262262262262263e-06,
-      "loss": 1.1928,
+      "loss": 1.1912,
       "step": 27350
     },
     {
       "epoch": 82.16,
-      "grad_norm": 6.3199381828308105,
+      "grad_norm": 5.363433837890625,
       "learning_rate": 7.2612612612612625e-06,
-      "loss": 1.1804,
+      "loss": 1.1596,
       "step": 27360
     },
     {
       "epoch": 82.19,
-      "grad_norm": 7.05450963973999,
+      "grad_norm": 9.441265106201172,
       "learning_rate": 7.260260260260261e-06,
-      "loss": 1.246,
+      "loss": 1.2442,
       "step": 27370
     },
     {
       "epoch": 82.22,
-      "grad_norm": 6.418006896972656,
+      "grad_norm": 6.971141815185547,
       "learning_rate": 7.2592592592592605e-06,
-      "loss": 1.2207,
+      "loss": 1.2252,
       "step": 27380
     },
     {
       "epoch": 82.25,
-      "grad_norm": 8.54416275024414,
+      "grad_norm": 6.490439414978027,
       "learning_rate": 7.258258258258258e-06,
-      "loss": 1.1961,
+      "loss": 1.1786,
       "step": 27390
     },
     {
       "epoch": 82.28,
-      "grad_norm": 9.437490463256836,
+      "grad_norm": 7.715824604034424,
       "learning_rate": 7.257257257257258e-06,
-      "loss": 1.2518,
+      "loss": 1.2542,
       "step": 27400
     },
     {
       "epoch": 82.31,
-      "grad_norm": 9.896111488342285,
+      "grad_norm": 9.530500411987305,
       "learning_rate": 7.256256256256256e-06,
-      "loss": 1.1885,
+      "loss": 1.1935,
       "step": 27410
     },
     {
       "epoch": 82.34,
-      "grad_norm": 6.056839466094971,
+      "grad_norm": 5.741907596588135,
       "learning_rate": 7.255255255255256e-06,
-      "loss": 1.2267,
+      "loss": 1.2155,
       "step": 27420
     },
     {
       "epoch": 82.37,
-      "grad_norm": 6.859907150268555,
+      "grad_norm": 7.172854423522949,
       "learning_rate": 7.254254254254255e-06,
-      "loss": 1.1883,
+      "loss": 1.1874,
       "step": 27430
     },
     {
       "epoch": 82.4,
-      "grad_norm": 7.937481880187988,
+      "grad_norm": 6.50814962387085,
       "learning_rate": 7.253253253253254e-06,
-      "loss": 1.2462,
+      "loss": 1.2443,
       "step": 27440
     },
     {
       "epoch": 82.43,
-      "grad_norm": 6.935020446777344,
+      "grad_norm": 9.79590129852295,
       "learning_rate": 7.252252252252253e-06,
-      "loss": 1.2388,
+      "loss": 1.2294,
       "step": 27450
     },
     {
       "epoch": 82.46,
-      "grad_norm": 6.939126014709473,
+      "grad_norm": 8.127752304077148,
       "learning_rate": 7.251251251251253e-06,
-      "loss": 1.231,
+      "loss": 1.2328,
       "step": 27460
     },
     {
       "epoch": 82.49,
-      "grad_norm": 5.691039562225342,
+      "grad_norm": 6.0983428955078125,
       "learning_rate": 7.25025025025025e-06,
-      "loss": 1.1328,
+      "loss": 1.1372,
       "step": 27470
     },
     {
       "epoch": 82.52,
-      "grad_norm": 5.712465763092041,
+      "grad_norm": 7.824089050292969,
       "learning_rate": 7.24924924924925e-06,
-      "loss": 1.1321,
+      "loss": 1.1399,
       "step": 27480
     },
     {
       "epoch": 82.55,
-      "grad_norm": 5.556812286376953,
+      "grad_norm": 7.248246669769287,
       "learning_rate": 7.2482482482482484e-06,
-      "loss": 1.255,
+      "loss": 1.2532,
       "step": 27490
     },
     {
       "epoch": 82.58,
-      "grad_norm": 5.5992350578308105,
+      "grad_norm": 5.0156731605529785,
       "learning_rate": 7.247247247247248e-06,
-      "loss": 1.2237,
+      "loss": 1.2199,
       "step": 27500
     },
     {
       "epoch": 82.61,
-      "grad_norm": 6.606954574584961,
+      "grad_norm": 8.855799674987793,
       "learning_rate": 7.2462462462462465e-06,
-      "loss": 1.2142,
+      "loss": 1.2096,
       "step": 27510
     },
     {
       "epoch": 82.64,
-      "grad_norm": 7.597163677215576,
+      "grad_norm": 8.770261764526367,
       "learning_rate": 7.245245245245246e-06,
-      "loss": 1.1546,
+      "loss": 1.1531,
       "step": 27520
     },
     {
       "epoch": 82.67,
-      "grad_norm": 4.767283916473389,
+      "grad_norm": 7.404470920562744,
       "learning_rate": 7.244244244244245e-06,
-      "loss": 1.1625,
+      "loss": 1.1586,
       "step": 27530
     },
     {
       "epoch": 82.7,
-      "grad_norm": 8.013609886169434,
+      "grad_norm": 8.54378604888916,
       "learning_rate": 7.243243243243244e-06,
-      "loss": 1.1979,
+      "loss": 1.1955,
       "step": 27540
     },
     {
       "epoch": 82.73,
-      "grad_norm": 6.634970188140869,
+      "grad_norm": 6.0917253494262695,
       "learning_rate": 7.242242242242243e-06,
-      "loss": 1.1906,
+      "loss": 1.1788,
       "step": 27550
     },
     {
       "epoch": 82.76,
-      "grad_norm": 7.673612117767334,
+      "grad_norm": 6.043922424316406,
       "learning_rate": 7.241241241241241e-06,
-      "loss": 1.1493,
+      "loss": 1.148,
       "step": 27560
     },
     {
       "epoch": 82.79,
-      "grad_norm": 7.767570495605469,
+      "grad_norm": 8.120935440063477,
       "learning_rate": 7.240240240240241e-06,
-      "loss": 1.2897,
+      "loss": 1.2883,
       "step": 27570
     },
     {
       "epoch": 82.82,
-      "grad_norm": 8.57637882232666,
+      "grad_norm": 7.265241622924805,
       "learning_rate": 7.23923923923924e-06,
-      "loss": 1.1925,
+      "loss": 1.1894,
       "step": 27580
     },
     {
       "epoch": 82.85,
-      "grad_norm": 7.437688827514648,
+      "grad_norm": 6.380127906799316,
       "learning_rate": 7.238238238238239e-06,
-      "loss": 1.2774,
+      "loss": 1.2715,
       "step": 27590
     },
     {
       "epoch": 82.88,
-      "grad_norm": 7.993193626403809,
+      "grad_norm": 7.370540618896484,
       "learning_rate": 7.237237237237238e-06,
-      "loss": 1.174,
+      "loss": 1.1756,
       "step": 27600
     },
     {
       "epoch": 82.91,
-      "grad_norm": 10.457043647766113,
+      "grad_norm": 7.725967884063721,
       "learning_rate": 7.236236236236237e-06,
-      "loss": 1.1784,
+      "loss": 1.1699,
       "step": 27610
     },
     {
       "epoch": 82.94,
-      "grad_norm": 7.46990442276001,
+      "grad_norm": 6.742214679718018,
       "learning_rate": 7.235235235235236e-06,
-      "loss": 1.1755,
+      "loss": 1.1814,
       "step": 27620
     },
     {
       "epoch": 82.97,
-      "grad_norm": 7.591980934143066,
+      "grad_norm": 7.368837356567383,
       "learning_rate": 7.2342342342342355e-06,
-      "loss": 1.2226,
+      "loss": 1.2239,
       "step": 27630
     },
     {
       "epoch": 83.0,
       "eval_accuracy": 0.7883,
-      "eval_loss": 0.6889695525169373,
-      "eval_runtime": 5.5565,
-      "eval_samples_per_second": 1799.71,
-      "eval_steps_per_second": 7.199,
+      "eval_loss": 0.6872221827507019,
+      "eval_runtime": 5.4065,
+      "eval_samples_per_second": 1849.617,
+      "eval_steps_per_second": 7.398,
       "step": 27639
     },
     {
       "epoch": 83.0,
-      "grad_norm": 6.143425941467285,
+      "grad_norm": 5.132053375244141,
       "learning_rate": 7.233233233233233e-06,
-      "loss": 1.2569,
+      "loss": 1.2506,
       "step": 27640
     },
     {
       "epoch": 83.03,
-      "grad_norm": 6.481358051300049,
+      "grad_norm": 6.8378705978393555,
       "learning_rate": 7.232232232232233e-06,
-      "loss": 1.219,
+      "loss": 1.2135,
       "step": 27650
     },
     {
       "epoch": 83.06,
-      "grad_norm": 5.960417747497559,
+      "grad_norm": 7.827085971832275,
       "learning_rate": 7.231231231231231e-06,
-      "loss": 1.1708,
+      "loss": 1.1793,
       "step": 27660
     },
     {
       "epoch": 83.09,
-      "grad_norm": 7.6334123611450195,
+      "grad_norm": 7.37404727935791,
       "learning_rate": 7.230230230230231e-06,
-      "loss": 1.1492,
+      "loss": 1.1423,
       "step": 27670
     },
     {
       "epoch": 83.12,
-      "grad_norm": 7.348234176635742,
+      "grad_norm": 9.257397651672363,
       "learning_rate": 7.22922922922923e-06,
-      "loss": 1.2214,
+      "loss": 1.2304,
       "step": 27680
     },
     {
       "epoch": 83.15,
-      "grad_norm": 7.063244819641113,
+      "grad_norm": 6.857383728027344,
       "learning_rate": 7.228228228228229e-06,
-      "loss": 1.1866,
+      "loss": 1.1776,
       "step": 27690
     },
     {
       "epoch": 83.18,
-      "grad_norm": 5.837429046630859,
+      "grad_norm": 6.942937850952148,
       "learning_rate": 7.227227227227228e-06,
-      "loss": 1.1714,
+      "loss": 1.1624,
       "step": 27700
     },
     {
       "epoch": 83.21,
-      "grad_norm": 6.989917755126953,
+      "grad_norm": 6.708613395690918,
       "learning_rate": 7.226226226226228e-06,
-      "loss": 1.2109,
+      "loss": 1.2103,
       "step": 27710
     },
     {
       "epoch": 83.24,
-      "grad_norm": 5.62252950668335,
+      "grad_norm": 5.457255840301514,
       "learning_rate": 7.2252252252252254e-06,
-      "loss": 1.2034,
+      "loss": 1.1884,
       "step": 27720
     },
     {
       "epoch": 83.27,
-      "grad_norm": 7.24932861328125,
+      "grad_norm": 6.942254066467285,
       "learning_rate": 7.224224224224224e-06,
-      "loss": 1.1463,
+      "loss": 1.15,
       "step": 27730
     },
     {
       "epoch": 83.3,
-      "grad_norm": 10.77548885345459,
+      "grad_norm": 11.013049125671387,
       "learning_rate": 7.2232232232232235e-06,
-      "loss": 1.1979,
+      "loss": 1.2003,
       "step": 27740
     },
     {
       "epoch": 83.33,
-      "grad_norm": 6.129329681396484,
+      "grad_norm": 6.9293341636657715,
       "learning_rate": 7.222222222222223e-06,
-      "loss": 1.1995,
+      "loss": 1.1963,
       "step": 27750
     },
     {
       "epoch": 83.36,
-      "grad_norm": 5.0739359855651855,
+      "grad_norm": 4.970388889312744,
       "learning_rate": 7.2212212212212215e-06,
-      "loss": 1.1473,
+      "loss": 1.1429,
       "step": 27760
     },
     {
       "epoch": 83.39,
-      "grad_norm": 9.298714637756348,
+      "grad_norm": 11.099205017089844,
       "learning_rate": 7.220220220220221e-06,
-      "loss": 1.2418,
+      "loss": 1.2458,
       "step": 27770
     },
     {
       "epoch": 83.42,
-      "grad_norm": 6.222875595092773,
+      "grad_norm": 5.525345802307129,
       "learning_rate": 7.21921921921922e-06,
-      "loss": 1.1032,
+      "loss": 1.1007,
       "step": 27780
     },
     {
       "epoch": 83.45,
-      "grad_norm": 5.63598108291626,
+      "grad_norm": 7.523680210113525,
       "learning_rate": 7.218218218218219e-06,
-      "loss": 1.1811,
+      "loss": 1.1778,
       "step": 27790
     },
     {
       "epoch": 83.48,
-      "grad_norm": 8.950194358825684,
+      "grad_norm": 8.833409309387207,
       "learning_rate": 7.217217217217218e-06,
-      "loss": 1.1961,
+      "loss": 1.1941,
       "step": 27800
     },
     {
       "epoch": 83.51,
-      "grad_norm": 6.81016731262207,
+      "grad_norm": 5.6522722244262695,
       "learning_rate": 7.216216216216216e-06,
-      "loss": 1.2022,
+      "loss": 1.2043,
       "step": 27810
     },
     {
       "epoch": 83.54,
-      "grad_norm": 6.142910480499268,
+      "grad_norm": 7.374692440032959,
       "learning_rate": 7.215215215215216e-06,
-      "loss": 1.1783,
+      "loss": 1.1755,
       "step": 27820
     },
     {
       "epoch": 83.57,
-      "grad_norm": 6.1401777267456055,
+      "grad_norm": 7.462196350097656,
       "learning_rate": 7.214214214214215e-06,
-      "loss": 1.2426,
+      "loss": 1.2266,
       "step": 27830
     },
     {
       "epoch": 83.6,
-      "grad_norm": 8.094733238220215,
+      "grad_norm": 8.523277282714844,
       "learning_rate": 7.213213213213214e-06,
-      "loss": 1.1927,
+      "loss": 1.1793,
       "step": 27840
     },
     {
       "epoch": 83.63,
-      "grad_norm": 5.702736854553223,
+      "grad_norm": 7.100970268249512,
       "learning_rate": 7.212212212212213e-06,
-      "loss": 1.1387,
+      "loss": 1.1349,
       "step": 27850
     },
     {
       "epoch": 83.66,
-      "grad_norm": 7.758513450622559,
+      "grad_norm": 7.921513080596924,
       "learning_rate": 7.211211211211212e-06,
-      "loss": 1.2116,
+      "loss": 1.2033,
       "step": 27860
     },
     {
       "epoch": 83.69,
-      "grad_norm": 5.729575157165527,
+      "grad_norm": 6.763507843017578,
       "learning_rate": 7.210210210210211e-06,
-      "loss": 1.1217,
+      "loss": 1.1186,
       "step": 27870
     },
     {
       "epoch": 83.72,
-      "grad_norm": 5.830831527709961,
+      "grad_norm": 6.412293434143066,
       "learning_rate": 7.2092092092092106e-06,
-      "loss": 1.1878,
+      "loss": 1.1868,
       "step": 27880
     },
     {
       "epoch": 83.75,
-      "grad_norm": 6.952209949493408,
+      "grad_norm": 7.10896635055542,
       "learning_rate": 7.208208208208208e-06,
-      "loss": 1.2295,
+      "loss": 1.2418,
       "step": 27890
     },
     {
       "epoch": 83.78,
-      "grad_norm": 6.263539791107178,
+      "grad_norm": 6.582788467407227,
       "learning_rate": 7.207207207207208e-06,
-      "loss": 1.1544,
+      "loss": 1.1484,
       "step": 27900
     },
     {
       "epoch": 83.81,
-      "grad_norm": 7.212370872497559,
+      "grad_norm": 6.8929524421691895,
       "learning_rate": 7.206206206206206e-06,
-      "loss": 1.2785,
+      "loss": 1.2726,
       "step": 27910
     },
     {
       "epoch": 83.84,
-      "grad_norm": 9.118874549865723,
+      "grad_norm": 8.355864524841309,
       "learning_rate": 7.205205205205206e-06,
-      "loss": 1.2114,
+      "loss": 1.2059,
       "step": 27920
     },
     {
       "epoch": 83.87,
-      "grad_norm": 5.93244743347168,
+      "grad_norm": 7.70578145980835,
       "learning_rate": 7.204204204204205e-06,
-      "loss": 1.1844,
+      "loss": 1.1932,
       "step": 27930
     },
     {
       "epoch": 83.9,
-      "grad_norm": 6.4285993576049805,
+      "grad_norm": 6.931309700012207,
       "learning_rate": 7.203203203203204e-06,
-      "loss": 1.2247,
+      "loss": 1.2208,
       "step": 27940
     },
     {
       "epoch": 83.93,
-      "grad_norm": 6.365212440490723,
+      "grad_norm": 6.115577697753906,
       "learning_rate": 7.202202202202203e-06,
-      "loss": 1.2142,
+      "loss": 1.2116,
       "step": 27950
     },
     {
       "epoch": 83.96,
-      "grad_norm": 6.358181476593018,
+      "grad_norm": 7.90662956237793,
       "learning_rate": 7.201201201201202e-06,
-      "loss": 1.1249,
+      "loss": 1.1129,
       "step": 27960
     },
     {
       "epoch": 83.99,
-      "grad_norm": 7.966670513153076,
+      "grad_norm": 8.98280143737793,
       "learning_rate": 7.2002002002002005e-06,
-      "loss": 1.2658,
+      "loss": 1.2542,
       "step": 27970
     },
     {
       "epoch": 84.0,
-      "eval_accuracy": 0.7893,
-      "eval_loss": 0.6824387907981873,
-      "eval_runtime": 5.4519,
-      "eval_samples_per_second": 1834.219,
-      "eval_steps_per_second": 7.337,
+      "eval_accuracy": 0.7919,
+      "eval_loss": 0.6810836791992188,
+      "eval_runtime": 5.472,
+      "eval_samples_per_second": 1827.49,
+      "eval_steps_per_second": 7.31,
       "step": 27972
     },
     {
       "epoch": 84.02,
-      "grad_norm": 9.322916030883789,
+      "grad_norm": 9.949172019958496,
       "learning_rate": 7.199199199199199e-06,
-      "loss": 1.3579,
+      "loss": 1.3606,
       "step": 27980
     },
     {
       "epoch": 84.05,
-      "grad_norm": 9.10340404510498,
+      "grad_norm": 8.81969928741455,
       "learning_rate": 7.1981981981981985e-06,
-      "loss": 1.1919,
+      "loss": 1.1932,
       "step": 27990
     },
     {
       "epoch": 84.08,
-      "grad_norm": 6.449718952178955,
+      "grad_norm": 8.705625534057617,
       "learning_rate": 7.197197197197198e-06,
-      "loss": 1.166,
+      "loss": 1.1612,
       "step": 28000
     },
     {
       "epoch": 84.11,
-      "grad_norm": 7.991596221923828,
+      "grad_norm": 6.812164306640625,
       "learning_rate": 7.1961961961961965e-06,
-      "loss": 1.2168,
+      "loss": 1.2199,
       "step": 28010
     },
     {
       "epoch": 84.14,
-      "grad_norm": 9.07340145111084,
+      "grad_norm": 8.756336212158203,
       "learning_rate": 7.195195195195196e-06,
-      "loss": 1.1924,
+      "loss": 1.1897,
       "step": 28020
     },
     {
       "epoch": 84.17,
-      "grad_norm": 6.521329879760742,
+      "grad_norm": 8.25122356414795,
       "learning_rate": 7.194194194194195e-06,
-      "loss": 1.2407,
+      "loss": 1.2374,
       "step": 28030
     },
     {
       "epoch": 84.2,
-      "grad_norm": 8.874615669250488,
+      "grad_norm": 10.3226900100708,
       "learning_rate": 7.193193193193194e-06,
-      "loss": 1.1922,
+      "loss": 1.199,
       "step": 28040
     },
     {
       "epoch": 84.23,
-      "grad_norm": 9.348024368286133,
+      "grad_norm": 6.769574165344238,
       "learning_rate": 7.1921921921921935e-06,
-      "loss": 1.2338,
+      "loss": 1.2339,
       "step": 28050
     },
     {
       "epoch": 84.26,
-      "grad_norm": 5.480922698974609,
+      "grad_norm": 8.461648941040039,
       "learning_rate": 7.191191191191191e-06,
-      "loss": 1.2495,
+      "loss": 1.2408,
       "step": 28060
     },
     {
       "epoch": 84.29,
-      "grad_norm": 7.803902626037598,
+      "grad_norm": 7.968683242797852,
       "learning_rate": 7.190190190190191e-06,
-      "loss": 1.1756,
+      "loss": 1.1697,
       "step": 28070
     },
     {
       "epoch": 84.32,
-      "grad_norm": 8.135506629943848,
+      "grad_norm": 8.874947547912598,
       "learning_rate": 7.189189189189189e-06,
-      "loss": 1.2321,
+      "loss": 1.2257,
       "step": 28080
     },
     {
       "epoch": 84.35,
-      "grad_norm": 8.1553316116333,
+      "grad_norm": 7.053416728973389,
       "learning_rate": 7.188188188188189e-06,
-      "loss": 1.1821,
+      "loss": 1.1788,
       "step": 28090
     },
     {
       "epoch": 84.38,
-      "grad_norm": 8.24673843383789,
+      "grad_norm": 7.364064693450928,
       "learning_rate": 7.187187187187188e-06,
-      "loss": 1.196,
+      "loss": 1.1872,
       "step": 28100
     },
     {
       "epoch": 84.41,
-      "grad_norm": 6.760404586791992,
+      "grad_norm": 7.590259075164795,
       "learning_rate": 7.186186186186187e-06,
-      "loss": 1.091,
+      "loss": 1.0843,
       "step": 28110
     },
     {
       "epoch": 84.44,
-      "grad_norm": 13.359626770019531,
+      "grad_norm": 11.351408004760742,
       "learning_rate": 7.185185185185186e-06,
-      "loss": 1.1811,
+      "loss": 1.1745,
       "step": 28120
     },
     {
       "epoch": 84.47,
-      "grad_norm": 5.800900459289551,
+      "grad_norm": 6.799036502838135,
       "learning_rate": 7.184184184184186e-06,
-      "loss": 1.158,
+      "loss": 1.157,
       "step": 28130
     },
     {
       "epoch": 84.5,
-      "grad_norm": 7.865531921386719,
+      "grad_norm": 6.969558238983154,
       "learning_rate": 7.183183183183183e-06,
-      "loss": 1.1783,
+      "loss": 1.1751,
       "step": 28140
     },
     {
       "epoch": 84.53,
-      "grad_norm": 8.52680492401123,
+      "grad_norm": 7.291007041931152,
       "learning_rate": 7.182182182182183e-06,
-      "loss": 1.212,
+      "loss": 1.2093,
       "step": 28150
     },
     {
       "epoch": 84.56,
-      "grad_norm": 5.9648284912109375,
+      "grad_norm": 6.613276481628418,
       "learning_rate": 7.181181181181181e-06,
-      "loss": 1.1591,
+      "loss": 1.1616,
       "step": 28160
     },
     {
       "epoch": 84.59,
-      "grad_norm": 6.711246967315674,
+      "grad_norm": 10.077103614807129,
       "learning_rate": 7.180180180180181e-06,
-      "loss": 1.1992,
+      "loss": 1.1971,
       "step": 28170
     },
     {
       "epoch": 84.62,
-      "grad_norm": 6.7022175788879395,
+      "grad_norm": 9.172882080078125,
       "learning_rate": 7.1791791791791794e-06,
-      "loss": 1.2024,
+      "loss": 1.2137,
       "step": 28180
     },
     {
       "epoch": 84.65,
-      "grad_norm": 6.408055305480957,
+      "grad_norm": 8.21004581451416,
       "learning_rate": 7.178178178178179e-06,
-      "loss": 1.1421,
+      "loss": 1.1357,
       "step": 28190
     },
     {
       "epoch": 84.68,
-      "grad_norm": 5.490757465362549,
+      "grad_norm": 6.705149173736572,
       "learning_rate": 7.177177177177178e-06,
-      "loss": 1.2284,
+      "loss": 1.2223,
       "step": 28200
     },
     {
       "epoch": 84.71,
-      "grad_norm": 6.237818717956543,
+      "grad_norm": 6.2703022956848145,
       "learning_rate": 7.176176176176177e-06,
-      "loss": 1.194,
+      "loss": 1.1824,
       "step": 28210
     },
     {
       "epoch": 84.74,
-      "grad_norm": 6.631192684173584,
+      "grad_norm": 8.295644760131836,
       "learning_rate": 7.1751751751751755e-06,
-      "loss": 1.1741,
+      "loss": 1.1835,
       "step": 28220
     },
     {
       "epoch": 84.77,
-      "grad_norm": 7.522990703582764,
+      "grad_norm": 8.32385540008545,
       "learning_rate": 7.174174174174174e-06,
-      "loss": 1.2356,
+      "loss": 1.2324,
       "step": 28230
     },
     {
       "epoch": 84.8,
-      "grad_norm": 7.749858856201172,
+      "grad_norm": 6.990333557128906,
       "learning_rate": 7.1731731731731735e-06,
-      "loss": 1.1619,
+      "loss": 1.1665,
       "step": 28240
     },
     {
       "epoch": 84.83,
-      "grad_norm": 8.62215518951416,
+      "grad_norm": 10.160954475402832,
       "learning_rate": 7.172172172172173e-06,
-      "loss": 1.24,
+      "loss": 1.2346,
       "step": 28250
     },
     {
       "epoch": 84.86,
-      "grad_norm": 8.369531631469727,
+      "grad_norm": 8.652544021606445,
       "learning_rate": 7.1711711711711716e-06,
-      "loss": 1.2635,
+      "loss": 1.2634,
       "step": 28260
     },
     {
       "epoch": 84.89,
-      "grad_norm": 7.269086837768555,
+      "grad_norm": 8.742996215820312,
       "learning_rate": 7.170170170170171e-06,
-      "loss": 1.1295,
+      "loss": 1.1311,
       "step": 28270
     },
     {
       "epoch": 84.92,
-      "grad_norm": 8.636741638183594,
+      "grad_norm": 9.326902389526367,
       "learning_rate": 7.1691691691691705e-06,
-      "loss": 1.2019,
+      "loss": 1.2108,
       "step": 28280
     },
     {
       "epoch": 84.95,
-      "grad_norm": 8.88403606414795,
+      "grad_norm": 7.293320178985596,
       "learning_rate": 7.168168168168169e-06,
-      "loss": 1.1462,
+      "loss": 1.1341,
       "step": 28290
     },
     {
       "epoch": 84.98,
-      "grad_norm": 6.813791275024414,
+      "grad_norm": 5.697429180145264,
       "learning_rate": 7.167167167167167e-06,
-      "loss": 1.0837,
+      "loss": 1.0815,
       "step": 28300
     },
     {
       "epoch": 85.0,
-      "eval_accuracy": 0.7902,
-      "eval_loss": 0.6840377449989319,
-      "eval_runtime": 5.5083,
-      "eval_samples_per_second": 1815.436,
-      "eval_steps_per_second": 7.262,
+      "eval_accuracy": 0.7899,
+      "eval_loss": 0.680097758769989,
+      "eval_runtime": 5.4072,
+      "eval_samples_per_second": 1849.397,
+      "eval_steps_per_second": 7.398,
       "step": 28305
     },
     {
       "epoch": 85.02,
-      "grad_norm": 5.412432670593262,
+      "grad_norm": 7.147891998291016,
       "learning_rate": 7.166166166166166e-06,
-      "loss": 1.476,
+      "loss": 1.4794,
       "step": 28310
     },
     {
       "epoch": 85.05,
-      "grad_norm": 7.474775314331055,
+      "grad_norm": 8.077577590942383,
       "learning_rate": 7.165165165165166e-06,
-      "loss": 1.2008,
+      "loss": 1.1965,
       "step": 28320
     },
     {
       "epoch": 85.08,
-      "grad_norm": 7.616730690002441,
+      "grad_norm": 7.239489555358887,
       "learning_rate": 7.164164164164164e-06,
-      "loss": 1.1503,
+      "loss": 1.1573,
       "step": 28330
     },
     {
       "epoch": 85.11,
-      "grad_norm": 7.089600563049316,
+      "grad_norm": 7.3210883140563965,
       "learning_rate": 7.163163163163164e-06,
-      "loss": 1.1943,
+      "loss": 1.1767,
       "step": 28340
     },
     {
       "epoch": 85.14,
-      "grad_norm": 8.243206977844238,
+      "grad_norm": 7.7006516456604,
       "learning_rate": 7.162162162162163e-06,
-      "loss": 1.2682,
+      "loss": 1.2716,
       "step": 28350
     },
     {
       "epoch": 85.17,
-      "grad_norm": 5.138428688049316,
+      "grad_norm": 8.388206481933594,
       "learning_rate": 7.161161161161162e-06,
-      "loss": 1.1862,
+      "loss": 1.1933,
       "step": 28360
     },
     {
       "epoch": 85.2,
-      "grad_norm": 7.216019630432129,
+      "grad_norm": 8.174551010131836,
       "learning_rate": 7.160160160160161e-06,
-      "loss": 1.1333,
+      "loss": 1.1294,
       "step": 28370
     },
     {
       "epoch": 85.23,
-      "grad_norm": 6.243071556091309,
+      "grad_norm": 6.821476936340332,
       "learning_rate": 7.159159159159161e-06,
-      "loss": 1.1785,
+      "loss": 1.1818,
       "step": 28380
     },
     {
       "epoch": 85.26,
-      "grad_norm": 7.950326919555664,
+      "grad_norm": 6.848713397979736,
       "learning_rate": 7.158158158158158e-06,
-      "loss": 1.1704,
+      "loss": 1.171,
       "step": 28390
     },
     {
       "epoch": 85.29,
-      "grad_norm": 7.800814151763916,
+      "grad_norm": 7.938762187957764,
       "learning_rate": 7.157157157157158e-06,
-      "loss": 1.182,
+      "loss": 1.1787,
       "step": 28400
     },
     {
       "epoch": 85.32,
-      "grad_norm": 8.724329948425293,
+      "grad_norm": 7.65120792388916,
       "learning_rate": 7.156156156156156e-06,
-      "loss": 1.1827,
+      "loss": 1.1908,
       "step": 28410
     },
     {
       "epoch": 85.35,
-      "grad_norm": 8.05541706085205,
+      "grad_norm": 9.41288948059082,
       "learning_rate": 7.155155155155156e-06,
-      "loss": 1.1717,
+      "loss": 1.1623,
       "step": 28420
     },
     {
       "epoch": 85.38,
-      "grad_norm": 7.889469623565674,
+      "grad_norm": 12.675919532775879,
       "learning_rate": 7.1541541541541545e-06,
-      "loss": 1.1659,
+      "loss": 1.172,
       "step": 28430
     },
     {
       "epoch": 85.41,
-      "grad_norm": 7.1153693199157715,
+      "grad_norm": 7.710938453674316,
       "learning_rate": 7.153153153153154e-06,
-      "loss": 1.1232,
+      "loss": 1.1201,
       "step": 28440
     },
     {
       "epoch": 85.44,
-      "grad_norm": 6.9101033210754395,
+      "grad_norm": 7.574905872344971,
       "learning_rate": 7.152152152152153e-06,
-      "loss": 1.2835,
+      "loss": 1.2914,
       "step": 28450
     },
     {
       "epoch": 85.47,
-      "grad_norm": 6.597909927368164,
+      "grad_norm": 5.18430233001709,
       "learning_rate": 7.151151151151152e-06,
-      "loss": 1.1448,
+      "loss": 1.1368,
       "step": 28460
     },
     {
       "epoch": 85.5,
-      "grad_norm": 6.556398391723633,
+      "grad_norm": 7.076573371887207,
       "learning_rate": 7.1501501501501505e-06,
-      "loss": 1.1915,
+      "loss": 1.1978,
       "step": 28470
     },
     {
       "epoch": 85.53,
-      "grad_norm": 9.309751510620117,
+      "grad_norm": 8.77455997467041,
       "learning_rate": 7.149149149149149e-06,
-      "loss": 1.2106,
+      "loss": 1.2079,
       "step": 28480
     },
     {
       "epoch": 85.56,
-      "grad_norm": 7.373454570770264,
+      "grad_norm": 8.84962272644043,
       "learning_rate": 7.1481481481481486e-06,
-      "loss": 1.2673,
+      "loss": 1.2733,
       "step": 28490
     },
     {
       "epoch": 85.59,
-      "grad_norm": 9.792448043823242,
+      "grad_norm": 8.847631454467773,
       "learning_rate": 7.147147147147148e-06,
-      "loss": 1.2242,
+      "loss": 1.2282,
       "step": 28500
     },
     {
       "epoch": 85.62,
-      "grad_norm": 7.08698034286499,
+      "grad_norm": 7.495229721069336,
       "learning_rate": 7.146146146146147e-06,
-      "loss": 1.1936,
+      "loss": 1.1961,
       "step": 28510
     },
     {
       "epoch": 85.65,
-      "grad_norm": 7.387067794799805,
+      "grad_norm": 6.788929462432861,
       "learning_rate": 7.145145145145146e-06,
-      "loss": 1.1996,
+      "loss": 1.2037,
       "step": 28520
     },
     {
       "epoch": 85.68,
-      "grad_norm": 7.368288993835449,
+      "grad_norm": 7.250725746154785,
       "learning_rate": 7.144144144144145e-06,
-      "loss": 1.1989,
+      "loss": 1.195,
       "step": 28530
     },
     {
       "epoch": 85.71,
-      "grad_norm": 5.707334041595459,
+      "grad_norm": 6.008976459503174,
       "learning_rate": 7.143143143143144e-06,
-      "loss": 1.1773,
+      "loss": 1.1838,
       "step": 28540
     },
     {
       "epoch": 85.74,
-      "grad_norm": 7.152153015136719,
+      "grad_norm": 7.626290321350098,
       "learning_rate": 7.142142142142142e-06,
-      "loss": 1.1651,
+      "loss": 1.1611,
       "step": 28550
     },
     {
       "epoch": 85.77,
-      "grad_norm": 5.712583541870117,
+      "grad_norm": 7.393874168395996,
       "learning_rate": 7.141141141141141e-06,
-      "loss": 1.0899,
+      "loss": 1.0816,
       "step": 28560
     },
     {
       "epoch": 85.8,
-      "grad_norm": 5.729304313659668,
+      "grad_norm": 7.931056022644043,
       "learning_rate": 7.140140140140141e-06,
-      "loss": 1.2239,
+      "loss": 1.2145,
       "step": 28570
     },
     {
       "epoch": 85.83,
-      "grad_norm": 10.14685344696045,
+      "grad_norm": 13.209464073181152,
       "learning_rate": 7.139139139139139e-06,
-      "loss": 1.1905,
+      "loss": 1.1968,
       "step": 28580
     },
     {
       "epoch": 85.86,
-      "grad_norm": 7.440260887145996,
+      "grad_norm": 9.357789039611816,
       "learning_rate": 7.138138138138139e-06,
-      "loss": 1.1843,
+      "loss": 1.175,
       "step": 28590
     },
     {
       "epoch": 85.89,
-      "grad_norm": 6.127233505249023,
+      "grad_norm": 9.333780288696289,
       "learning_rate": 7.137137137137138e-06,
-      "loss": 1.1632,
+      "loss": 1.158,
       "step": 28600
     },
     {
       "epoch": 85.92,
-      "grad_norm": 7.489003658294678,
+      "grad_norm": 10.779485702514648,
       "learning_rate": 7.136136136136137e-06,
-      "loss": 1.1748,
+      "loss": 1.17,
       "step": 28610
     },
     {
       "epoch": 85.95,
-      "grad_norm": 6.014636516571045,
+      "grad_norm": 7.3173980712890625,
       "learning_rate": 7.135135135135136e-06,
-      "loss": 1.1489,
+      "loss": 1.1378,
       "step": 28620
     },
     {
       "epoch": 85.98,
-      "grad_norm": 6.985273838043213,
+      "grad_norm": 7.751855373382568,
       "learning_rate": 7.134134134134135e-06,
-      "loss": 1.112,
+      "loss": 1.1022,
       "step": 28630
     },
     {
       "epoch": 86.0,
-      "eval_accuracy": 0.7933,
-      "eval_loss": 0.6753904223442078,
-      "eval_runtime": 5.6801,
-      "eval_samples_per_second": 1760.543,
-      "eval_steps_per_second": 7.042,
+      "eval_accuracy": 0.7932,
+      "eval_loss": 0.6736400723457336,
+      "eval_runtime": 5.3258,
+      "eval_samples_per_second": 1877.644,
+      "eval_steps_per_second": 7.511,
       "step": 28638
     },
     {
       "epoch": 86.01,
-      "grad_norm": 7.587798595428467,
+      "grad_norm": 8.500993728637695,
       "learning_rate": 7.133133133133133e-06,
-      "loss": 1.4082,
+      "loss": 1.386,
       "step": 28640
     },
     {
       "epoch": 86.04,
-      "grad_norm": 6.449240207672119,
+      "grad_norm": 6.321513652801514,
       "learning_rate": 7.132132132132132e-06,
-      "loss": 1.1683,
+      "loss": 1.17,
       "step": 28650
     },
     {
       "epoch": 86.07,
-      "grad_norm": 6.646498680114746,
+      "grad_norm": 6.809515953063965,
       "learning_rate": 7.1311311311311315e-06,
-      "loss": 1.1765,
+      "loss": 1.1857,
       "step": 28660
     },
     {
       "epoch": 86.1,
-      "grad_norm": 4.536505699157715,
+      "grad_norm": 6.8728227615356445,
       "learning_rate": 7.130130130130131e-06,
-      "loss": 1.1021,
+      "loss": 1.099,
       "step": 28670
     },
     {
       "epoch": 86.13,
-      "grad_norm": 6.5434699058532715,
+      "grad_norm": 9.80511474609375,
       "learning_rate": 7.1291291291291295e-06,
-      "loss": 1.1688,
+      "loss": 1.176,
       "step": 28680
     },
     {
       "epoch": 86.16,
-      "grad_norm": 6.568912982940674,
+      "grad_norm": 8.100757598876953,
       "learning_rate": 7.128128128128129e-06,
-      "loss": 1.143,
+      "loss": 1.1349,
       "step": 28690
     },
     {
       "epoch": 86.19,
-      "grad_norm": 7.682764053344727,
+      "grad_norm": 7.918698787689209,
       "learning_rate": 7.127127127127128e-06,
-      "loss": 1.1744,
+      "loss": 1.1814,
       "step": 28700
     },
     {
       "epoch": 86.22,
-      "grad_norm": 8.136984825134277,
+      "grad_norm": 8.248008728027344,
       "learning_rate": 7.126126126126127e-06,
-      "loss": 1.2391,
+      "loss": 1.2452,
       "step": 28710
     },
     {
       "epoch": 86.25,
-      "grad_norm": 8.810776710510254,
+      "grad_norm": 10.89544677734375,
       "learning_rate": 7.1251251251251256e-06,
-      "loss": 1.1932,
+      "loss": 1.2024,
       "step": 28720
     },
     {
       "epoch": 86.28,
-      "grad_norm": 10.47422981262207,
+      "grad_norm": 7.699671745300293,
       "learning_rate": 7.124124124124124e-06,
-      "loss": 1.1231,
+      "loss": 1.1186,
       "step": 28730
     },
     {
       "epoch": 86.31,
-      "grad_norm": 6.538118839263916,
+      "grad_norm": 6.993371486663818,
       "learning_rate": 7.123123123123124e-06,
-      "loss": 1.1773,
+      "loss": 1.1778,
       "step": 28740
     },
     {
       "epoch": 86.34,
-      "grad_norm": 6.997360706329346,
+      "grad_norm": 16.179475784301758,
       "learning_rate": 7.122122122122122e-06,
-      "loss": 1.2453,
+      "loss": 1.2522,
       "step": 28750
     },
     {
       "epoch": 86.37,
-      "grad_norm": 6.554125785827637,
+      "grad_norm": 8.886394500732422,
       "learning_rate": 7.121121121121122e-06,
-      "loss": 1.1794,
+      "loss": 1.1825,
       "step": 28760
     },
     {
       "epoch": 86.4,
-      "grad_norm": 7.136807441711426,
+      "grad_norm": 7.0296831130981445,
       "learning_rate": 7.120120120120121e-06,
-      "loss": 1.1573,
+      "loss": 1.1565,
       "step": 28770
     },
     {
       "epoch": 86.43,
-      "grad_norm": 7.951791286468506,
+      "grad_norm": 7.461602687835693,
       "learning_rate": 7.11911911911912e-06,
-      "loss": 1.2699,
+      "loss": 1.2593,
       "step": 28780
     },
     {
       "epoch": 86.46,
-      "grad_norm": 5.53377103805542,
+      "grad_norm": 5.002482891082764,
       "learning_rate": 7.118118118118119e-06,
-      "loss": 1.2635,
+      "loss": 1.2729,
       "step": 28790
     },
     {
       "epoch": 86.49,
-      "grad_norm": 5.170341491699219,
+      "grad_norm": 6.0439605712890625,
       "learning_rate": 7.117117117117117e-06,
-      "loss": 1.2146,
+      "loss": 1.2079,
       "step": 28800
     },
     {
       "epoch": 86.52,
-      "grad_norm": 9.661892890930176,
+      "grad_norm": 7.12761116027832,
       "learning_rate": 7.116116116116116e-06,
-      "loss": 1.1514,
+      "loss": 1.1642,
       "step": 28810
     },
     {
       "epoch": 86.55,
-      "grad_norm": 6.662736415863037,
+      "grad_norm": 7.895822525024414,
       "learning_rate": 7.115115115115116e-06,
-      "loss": 1.1512,
+      "loss": 1.1494,
       "step": 28820
     },
     {
       "epoch": 86.58,
-      "grad_norm": 8.532074928283691,
+      "grad_norm": 8.688102722167969,
       "learning_rate": 7.114114114114114e-06,
-      "loss": 1.1706,
+      "loss": 1.1592,
       "step": 28830
     },
     {
       "epoch": 86.61,
-      "grad_norm": 9.041383743286133,
+      "grad_norm": 6.805245876312256,
       "learning_rate": 7.113113113113114e-06,
-      "loss": 1.2439,
+      "loss": 1.2276,
       "step": 28840
     },
     {
       "epoch": 86.64,
-      "grad_norm": 6.927122116088867,
+      "grad_norm": 8.220356941223145,
       "learning_rate": 7.112112112112113e-06,
-      "loss": 1.2448,
+      "loss": 1.2522,
       "step": 28850
     },
     {
       "epoch": 86.67,
-      "grad_norm": 6.832905292510986,
+      "grad_norm": 7.8002424240112305,
       "learning_rate": 7.111111111111112e-06,
-      "loss": 1.1143,
+      "loss": 1.0999,
       "step": 28860
     },
     {
       "epoch": 86.7,
-      "grad_norm": 7.4986042976379395,
+      "grad_norm": 8.493700981140137,
       "learning_rate": 7.110110110110111e-06,
-      "loss": 1.1764,
+      "loss": 1.1642,
       "step": 28870
     },
     {
       "epoch": 86.73,
-      "grad_norm": 6.703176975250244,
+      "grad_norm": 7.19047737121582,
       "learning_rate": 7.10910910910911e-06,
-      "loss": 1.1617,
+      "loss": 1.1602,
       "step": 28880
     },
     {
       "epoch": 86.76,
-      "grad_norm": 8.00143814086914,
+      "grad_norm": 8.041542053222656,
       "learning_rate": 7.1081081081081085e-06,
-      "loss": 1.2524,
+      "loss": 1.2445,
       "step": 28890
     },
     {
       "epoch": 86.79,
-      "grad_norm": 8.135027885437012,
+      "grad_norm": 6.581404209136963,
       "learning_rate": 7.107107107107107e-06,
-      "loss": 1.2344,
+      "loss": 1.229,
       "step": 28900
     },
     {
       "epoch": 86.82,
-      "grad_norm": 8.013483047485352,
+      "grad_norm": 8.632423400878906,
       "learning_rate": 7.1061061061061065e-06,
-      "loss": 1.1634,
+      "loss": 1.1667,
       "step": 28910
     },
     {
       "epoch": 86.85,
-      "grad_norm": 7.227442741394043,
+      "grad_norm": 5.787507057189941,
       "learning_rate": 7.105105105105106e-06,
-      "loss": 1.0872,
+      "loss": 1.0877,
       "step": 28920
     },
     {
       "epoch": 86.88,
-      "grad_norm": 5.658487319946289,
+      "grad_norm": 7.817903995513916,
       "learning_rate": 7.1041041041041045e-06,
-      "loss": 1.2391,
+      "loss": 1.2313,
       "step": 28930
     },
     {
       "epoch": 86.91,
-      "grad_norm": 5.9171576499938965,
+      "grad_norm": 7.872486114501953,
       "learning_rate": 7.103103103103104e-06,
-      "loss": 1.2039,
+      "loss": 1.198,
       "step": 28940
     },
     {
       "epoch": 86.94,
-      "grad_norm": 9.849981307983398,
+      "grad_norm": 9.41792106628418,
       "learning_rate": 7.102102102102103e-06,
-      "loss": 1.1609,
+      "loss": 1.1575,
       "step": 28950
     },
     {
       "epoch": 86.97,
-      "grad_norm": 8.91883659362793,
+      "grad_norm": 6.974164962768555,
       "learning_rate": 7.101101101101102e-06,
-      "loss": 1.0993,
+      "loss": 1.0996,
       "step": 28960
     },
     {
       "epoch": 87.0,
-      "grad_norm": 8.305395126342773,
+      "grad_norm": 6.705672264099121,
       "learning_rate": 7.100100100100101e-06,
-      "loss": 1.2667,
+      "loss": 1.2601,
       "step": 28970
     },
     {
       "epoch": 87.0,
-      "eval_accuracy": 0.7964,
-      "eval_loss": 0.6649665832519531,
-      "eval_runtime": 5.7879,
-      "eval_samples_per_second": 1727.745,
-      "eval_steps_per_second": 6.911,
+      "eval_accuracy": 0.7953,
+      "eval_loss": 0.663806676864624,
+      "eval_runtime": 5.3748,
+      "eval_samples_per_second": 1860.522,
+      "eval_steps_per_second": 7.442,
       "step": 28971
     },
     {
       "epoch": 87.03,
-      "grad_norm": 9.49679946899414,
+      "grad_norm": 6.7084856033325195,
       "learning_rate": 7.099099099099099e-06,
-      "loss": 1.4187,
+      "loss": 1.404,
       "step": 28980
     },
     {
       "epoch": 87.06,
-      "grad_norm": 11.12185287475586,
+      "grad_norm": 6.265992641448975,
       "learning_rate": 7.098098098098099e-06,
-      "loss": 1.2662,
+      "loss": 1.2666,
       "step": 28990
     },
     {
       "epoch": 87.09,
-      "grad_norm": 5.997476577758789,
+      "grad_norm": 5.933598518371582,
       "learning_rate": 7.097097097097097e-06,
-      "loss": 1.1423,
+      "loss": 1.1403,
       "step": 29000
     },
     {
       "epoch": 87.12,
-      "grad_norm": 7.7088942527771,
+      "grad_norm": 9.955506324768066,
       "learning_rate": 7.096096096096097e-06,
-      "loss": 1.2047,
+      "loss": 1.2055,
       "step": 29010
     },
     {
       "epoch": 87.15,
-      "grad_norm": 10.2023344039917,
+      "grad_norm": 10.298480987548828,
       "learning_rate": 7.095095095095096e-06,
-      "loss": 1.1825,
+      "loss": 1.1778,
       "step": 29020
     },
     {
       "epoch": 87.18,
-      "grad_norm": 5.308861255645752,
+      "grad_norm": 6.0109453201293945,
       "learning_rate": 7.094094094094095e-06,
-      "loss": 1.1132,
+      "loss": 1.1111,
       "step": 29030
     },
     {
       "epoch": 87.21,
-      "grad_norm": 6.094864368438721,
+      "grad_norm": 6.712040901184082,
       "learning_rate": 7.093093093093094e-06,
-      "loss": 1.185,
+      "loss": 1.1969,
       "step": 29040
     },
     {
       "epoch": 87.24,
-      "grad_norm": 7.47301721572876,
+      "grad_norm": 6.9349751472473145,
       "learning_rate": 7.092092092092092e-06,
-      "loss": 1.1439,
+      "loss": 1.1321,
       "step": 29050
     },
     {
       "epoch": 87.27,
-      "grad_norm": 8.569295883178711,
+      "grad_norm": 10.1328125,
       "learning_rate": 7.091091091091091e-06,
-      "loss": 1.239,
+      "loss": 1.2375,
       "step": 29060
     },
     {
       "epoch": 87.3,
-      "grad_norm": 7.075590133666992,
+      "grad_norm": 7.644609451293945,
       "learning_rate": 7.090090090090091e-06,
-      "loss": 1.1906,
+      "loss": 1.1968,
       "step": 29070
     },
     {
       "epoch": 87.33,
-      "grad_norm": 7.249227523803711,
+      "grad_norm": 8.064780235290527,
       "learning_rate": 7.089089089089089e-06,
-      "loss": 1.1206,
+      "loss": 1.1168,
       "step": 29080
     },
     {
       "epoch": 87.36,
-      "grad_norm": 7.150791645050049,
+      "grad_norm": 7.142147064208984,
       "learning_rate": 7.088088088088089e-06,
-      "loss": 1.1888,
+      "loss": 1.1763,
       "step": 29090
     },
     {
       "epoch": 87.39,
-      "grad_norm": 5.535892486572266,
+      "grad_norm": 6.2222771644592285,
       "learning_rate": 7.087087087087087e-06,
-      "loss": 1.1264,
+      "loss": 1.1345,
       "step": 29100
     },
     {
       "epoch": 87.42,
-      "grad_norm": 5.041922092437744,
+      "grad_norm": 4.816227436065674,
       "learning_rate": 7.086086086086087e-06,
-      "loss": 1.1,
+      "loss": 1.1007,
       "step": 29110
     },
     {
       "epoch": 87.45,
-      "grad_norm": 9.959808349609375,
+      "grad_norm": 7.802188873291016,
       "learning_rate": 7.085085085085086e-06,
-      "loss": 1.2701,
+      "loss": 1.2731,
       "step": 29120
     },
     {
       "epoch": 87.48,
-      "grad_norm": 6.9380412101745605,
+      "grad_norm": 6.166337966918945,
       "learning_rate": 7.084084084084085e-06,
-      "loss": 1.1468,
+      "loss": 1.1467,
       "step": 29130
     },
     {
       "epoch": 87.51,
-      "grad_norm": 4.6316680908203125,
+      "grad_norm": 4.639286518096924,
       "learning_rate": 7.0830830830830835e-06,
-      "loss": 1.2041,
+      "loss": 1.208,
       "step": 29140
     },
     {
       "epoch": 87.54,
-      "grad_norm": 8.181248664855957,
+      "grad_norm": 9.535459518432617,
       "learning_rate": 7.082082082082082e-06,
-      "loss": 1.1473,
+      "loss": 1.1383,
       "step": 29150
     },
     {
       "epoch": 87.57,
-      "grad_norm": 6.855471611022949,
+      "grad_norm": 6.219590187072754,
       "learning_rate": 7.0810810810810815e-06,
-      "loss": 1.1723,
+      "loss": 1.1764,
       "step": 29160
     },
     {
       "epoch": 87.6,
-      "grad_norm": 5.756059169769287,
+      "grad_norm": 5.82999849319458,
       "learning_rate": 7.080080080080081e-06,
-      "loss": 1.2118,
+      "loss": 1.1975,
       "step": 29170
     },
     {
       "epoch": 87.63,
-      "grad_norm": 6.038612365722656,
+      "grad_norm": 7.434147357940674,
       "learning_rate": 7.0790790790790796e-06,
-      "loss": 1.1843,
+      "loss": 1.1853,
       "step": 29180
     },
     {
       "epoch": 87.66,
-      "grad_norm": 6.689416408538818,
+      "grad_norm": 8.325632095336914,
       "learning_rate": 7.078078078078079e-06,
-      "loss": 1.2128,
+      "loss": 1.2263,
       "step": 29190
     },
     {
       "epoch": 87.69,
-      "grad_norm": 7.838381767272949,
+      "grad_norm": 6.619247913360596,
       "learning_rate": 7.0770770770770784e-06,
-      "loss": 1.0891,
+      "loss": 1.096,
       "step": 29200
     },
     {
       "epoch": 87.72,
-      "grad_norm": 8.359772682189941,
+      "grad_norm": 11.217869758605957,
       "learning_rate": 7.076076076076077e-06,
-      "loss": 1.119,
+      "loss": 1.103,
       "step": 29210
     },
     {
       "epoch": 87.75,
-      "grad_norm": 6.076775074005127,
+      "grad_norm": 6.812458038330078,
       "learning_rate": 7.075075075075075e-06,
-      "loss": 1.2561,
+      "loss": 1.2467,
       "step": 29220
     },
     {
       "epoch": 87.78,
-      "grad_norm": 6.027760028839111,
+      "grad_norm": 5.263326168060303,
       "learning_rate": 7.074074074074074e-06,
-      "loss": 1.1494,
+      "loss": 1.128,
       "step": 29230
     },
     {
       "epoch": 87.81,
-      "grad_norm": 8.721056938171387,
+      "grad_norm": 8.219582557678223,
       "learning_rate": 7.073073073073074e-06,
-      "loss": 1.2218,
+      "loss": 1.213,
       "step": 29240
     },
     {
       "epoch": 87.84,
-      "grad_norm": 6.410592079162598,
+      "grad_norm": 6.772779941558838,
       "learning_rate": 7.072072072072072e-06,
-      "loss": 1.202,
+      "loss": 1.1998,
       "step": 29250
     },
     {
       "epoch": 87.87,
-      "grad_norm": 8.482304573059082,
+      "grad_norm": 5.949577808380127,
       "learning_rate": 7.071071071071072e-06,
-      "loss": 1.1764,
+      "loss": 1.1718,
       "step": 29260
     },
     {
       "epoch": 87.9,
-      "grad_norm": 7.526123046875,
+      "grad_norm": 7.969167232513428,
       "learning_rate": 7.070070070070071e-06,
-      "loss": 1.0909,
+      "loss": 1.1012,
       "step": 29270
     },
     {
       "epoch": 87.93,
-      "grad_norm": 9.038729667663574,
+      "grad_norm": 7.461268424987793,
       "learning_rate": 7.06906906906907e-06,
-      "loss": 1.1385,
+      "loss": 1.1438,
       "step": 29280
     },
     {
       "epoch": 87.96,
-      "grad_norm": 7.657118320465088,
+      "grad_norm": 7.854918003082275,
       "learning_rate": 7.068068068068069e-06,
-      "loss": 1.2444,
+      "loss": 1.2341,
       "step": 29290
     },
     {
       "epoch": 87.99,
-      "grad_norm": 16.563945770263672,
+      "grad_norm": 12.802912712097168,
       "learning_rate": 7.067067067067067e-06,
-      "loss": 1.1847,
+      "loss": 1.1799,
       "step": 29300
     },
     {
       "epoch": 88.0,
-      "eval_accuracy": 0.7926,
-      "eval_loss": 0.671636164188385,
-      "eval_runtime": 5.8652,
-      "eval_samples_per_second": 1704.961,
-      "eval_steps_per_second": 6.82,
+      "eval_accuracy": 0.7914,
+      "eval_loss": 0.6699899435043335,
+      "eval_runtime": 5.1663,
+      "eval_samples_per_second": 1935.633,
+      "eval_steps_per_second": 7.743,
       "step": 29304
     },
     {
       "epoch": 88.02,
-      "grad_norm": 7.693814277648926,
+      "grad_norm": 7.496546745300293,
       "learning_rate": 7.066066066066066e-06,
-      "loss": 1.3662,
+      "loss": 1.3452,
       "step": 29310
     },
     {
       "epoch": 88.05,
-      "grad_norm": 8.699402809143066,
+      "grad_norm": 8.052210807800293,
       "learning_rate": 7.065065065065066e-06,
-      "loss": 1.1961,
+      "loss": 1.1918,
       "step": 29320
     },
     {
       "epoch": 88.08,
-      "grad_norm": 8.828448295593262,
+      "grad_norm": 8.306015968322754,
       "learning_rate": 7.064064064064064e-06,
-      "loss": 1.1693,
+      "loss": 1.1618,
       "step": 29330
     },
     {
       "epoch": 88.11,
-      "grad_norm": 6.823010444641113,
+      "grad_norm": 7.903360366821289,
       "learning_rate": 7.063063063063064e-06,
-      "loss": 1.2227,
+      "loss": 1.218,
       "step": 29340
     },
     {
       "epoch": 88.14,
-      "grad_norm": 8.093850135803223,
+      "grad_norm": 5.930295467376709,
       "learning_rate": 7.0620620620620625e-06,
-      "loss": 1.1062,
+      "loss": 1.1153,
       "step": 29350
     },
     {
       "epoch": 88.17,
-      "grad_norm": 9.744229316711426,
+      "grad_norm": 7.574507713317871,
       "learning_rate": 7.061061061061062e-06,
-      "loss": 1.1318,
+      "loss": 1.1363,
       "step": 29360
     },
     {
       "epoch": 88.2,
-      "grad_norm": 10.28046703338623,
+      "grad_norm": 6.902871608734131,
       "learning_rate": 7.060060060060061e-06,
-      "loss": 1.1678,
+      "loss": 1.173,
       "step": 29370
     },
     {
       "epoch": 88.23,
-      "grad_norm": 5.717138290405273,
+      "grad_norm": 6.45684289932251,
       "learning_rate": 7.059059059059059e-06,
-      "loss": 1.2263,
+      "loss": 1.226,
       "step": 29380
     },
     {
       "epoch": 88.26,
-      "grad_norm": 5.3874030113220215,
+      "grad_norm": 7.891716480255127,
       "learning_rate": 7.0580580580580585e-06,
-      "loss": 1.112,
+      "loss": 1.114,
       "step": 29390
     },
     {
       "epoch": 88.29,
-      "grad_norm": 9.040179252624512,
+      "grad_norm": 8.290938377380371,
       "learning_rate": 7.057057057057057e-06,
-      "loss": 1.1507,
+      "loss": 1.1454,
       "step": 29400
     },
     {
       "epoch": 88.32,
-      "grad_norm": 7.530601978302002,
+      "grad_norm": 7.420921802520752,
       "learning_rate": 7.0560560560560566e-06,
-      "loss": 1.0732,
+      "loss": 1.0705,
       "step": 29410
     },
     {
       "epoch": 88.35,
-      "grad_norm": 6.054470539093018,
+      "grad_norm": 5.061577796936035,
       "learning_rate": 7.055055055055056e-06,
-      "loss": 1.178,
+      "loss": 1.1781,
       "step": 29420
     },
     {
       "epoch": 88.38,
-      "grad_norm": 7.699337005615234,
+      "grad_norm": 10.68574333190918,
       "learning_rate": 7.054054054054055e-06,
-      "loss": 1.1082,
+      "loss": 1.1077,
       "step": 29430
     },
     {
       "epoch": 88.41,
-      "grad_norm": 5.968991756439209,
+      "grad_norm": 5.686810493469238,
       "learning_rate": 7.053053053053054e-06,
-      "loss": 1.1842,
+      "loss": 1.1872,
       "step": 29440
     },
     {
       "epoch": 88.44,
-      "grad_norm": 10.00442123413086,
+      "grad_norm": 6.747776508331299,
       "learning_rate": 7.052052052052053e-06,
-      "loss": 1.1526,
+      "loss": 1.1468,
       "step": 29450
     },
     {
       "epoch": 88.47,
-      "grad_norm": 6.0164594650268555,
+      "grad_norm": 6.5171685218811035,
       "learning_rate": 7.051051051051052e-06,
-      "loss": 1.2261,
+      "loss": 1.2262,
       "step": 29460
     },
     {
       "epoch": 88.5,
-      "grad_norm": 9.124384880065918,
+      "grad_norm": 9.041265487670898,
       "learning_rate": 7.05005005005005e-06,
-      "loss": 1.2146,
+      "loss": 1.2159,
       "step": 29470
     },
     {
       "epoch": 88.53,
-      "grad_norm": 7.991886615753174,
+      "grad_norm": 9.477346420288086,
       "learning_rate": 7.049049049049049e-06,
-      "loss": 1.1712,
+      "loss": 1.1782,
       "step": 29480
     },
     {
       "epoch": 88.56,
-      "grad_norm": 7.5520734786987305,
+      "grad_norm": 7.059869766235352,
       "learning_rate": 7.048048048048049e-06,
-      "loss": 1.1981,
+      "loss": 1.1794,
       "step": 29490
     },
     {
       "epoch": 88.59,
-      "grad_norm": 7.487072467803955,
+      "grad_norm": 6.6953535079956055,
       "learning_rate": 7.047047047047047e-06,
-      "loss": 1.2238,
+      "loss": 1.2194,
       "step": 29500
     },
     {
       "epoch": 88.62,
-      "grad_norm": 12.922386169433594,
+      "grad_norm": 9.832620620727539,
       "learning_rate": 7.046046046046047e-06,
-      "loss": 1.2318,
+      "loss": 1.2319,
       "step": 29510
     },
     {
       "epoch": 88.65,
-      "grad_norm": 8.596291542053223,
+      "grad_norm": 6.998348712921143,
       "learning_rate": 7.045045045045046e-06,
-      "loss": 1.1402,
+      "loss": 1.1424,
       "step": 29520
     },
     {
       "epoch": 88.68,
-      "grad_norm": 5.204216003417969,
+      "grad_norm": 5.998763561248779,
       "learning_rate": 7.044044044044045e-06,
-      "loss": 1.1383,
+      "loss": 1.1481,
       "step": 29530
     },
     {
       "epoch": 88.71,
-      "grad_norm": 6.993142604827881,
+      "grad_norm": 11.329300880432129,
       "learning_rate": 7.043043043043044e-06,
-      "loss": 1.231,
+      "loss": 1.2293,
       "step": 29540
     },
     {
       "epoch": 88.74,
-      "grad_norm": 5.733518600463867,
+      "grad_norm": 11.5824613571167,
       "learning_rate": 7.042042042042042e-06,
-      "loss": 1.2144,
+      "loss": 1.2124,
       "step": 29550
     },
     {
       "epoch": 88.77,
-      "grad_norm": 5.426321506500244,
+      "grad_norm": 7.273696422576904,
       "learning_rate": 7.041041041041041e-06,
-      "loss": 1.2023,
+      "loss": 1.201,
       "step": 29560
     },
     {
       "epoch": 88.8,
-      "grad_norm": 5.527395725250244,
+      "grad_norm": 6.753218650817871,
       "learning_rate": 7.04004004004004e-06,
-      "loss": 1.1257,
+      "loss": 1.1317,
       "step": 29570
     },
     {
       "epoch": 88.83,
-      "grad_norm": 5.92045783996582,
+      "grad_norm": 5.406911849975586,
       "learning_rate": 7.0390390390390395e-06,
-      "loss": 1.1709,
+      "loss": 1.1702,
       "step": 29580
     },
     {
       "epoch": 88.86,
-      "grad_norm": 7.205174446105957,
+      "grad_norm": 6.530322074890137,
       "learning_rate": 7.038038038038039e-06,
       "loss": 1.1644,
       "step": 29590
     },
     {
       "epoch": 88.89,
-      "grad_norm": 6.68155574798584,
+      "grad_norm": 7.384870529174805,
       "learning_rate": 7.0370370370370375e-06,
-      "loss": 1.1817,
+      "loss": 1.177,
       "step": 29600
     },
     {
       "epoch": 88.92,
-      "grad_norm": 7.4664306640625,
+      "grad_norm": 13.23826789855957,
       "learning_rate": 7.036036036036037e-06,
-      "loss": 1.1943,
+      "loss": 1.1833,
       "step": 29610
     },
     {
       "epoch": 88.95,
-      "grad_norm": 6.952859878540039,
+      "grad_norm": 8.290553092956543,
       "learning_rate": 7.035035035035036e-06,
-      "loss": 1.1923,
+      "loss": 1.1926,
       "step": 29620
     },
     {
       "epoch": 88.98,
-      "grad_norm": 8.877436637878418,
+      "grad_norm": 10.962319374084473,
       "learning_rate": 7.034034034034034e-06,
-      "loss": 1.1431,
+      "loss": 1.1302,
       "step": 29630
     },
     {
       "epoch": 89.0,
-      "eval_accuracy": 0.7934,
-      "eval_loss": 0.676316499710083,
-      "eval_runtime": 5.3212,
-      "eval_samples_per_second": 1879.268,
-      "eval_steps_per_second": 7.517,
+      "eval_accuracy": 0.7924,
+      "eval_loss": 0.6756088137626648,
+      "eval_runtime": 5.3499,
+      "eval_samples_per_second": 1869.199,
+      "eval_steps_per_second": 7.477,
       "step": 29637
     },
     {
       "epoch": 89.01,
-      "grad_norm": 7.250179767608643,
+      "grad_norm": 7.222029209136963,
       "learning_rate": 7.0330330330330336e-06,
-      "loss": 1.4058,
+      "loss": 1.3789,
       "step": 29640
     },
     {
       "epoch": 89.04,
-      "grad_norm": 5.908538341522217,
+      "grad_norm": 5.904021739959717,
       "learning_rate": 7.032032032032032e-06,
-      "loss": 1.1779,
+      "loss": 1.1861,
       "step": 29650
     },
     {
       "epoch": 89.07,
-      "grad_norm": 6.347919940948486,
+      "grad_norm": 5.432807445526123,
       "learning_rate": 7.031031031031032e-06,
-      "loss": 1.097,
+      "loss": 1.1014,
       "step": 29660
     },
     {
       "epoch": 89.1,
-      "grad_norm": 5.971150875091553,
+      "grad_norm": 8.73354721069336,
       "learning_rate": 7.03003003003003e-06,
-      "loss": 1.187,
+      "loss": 1.1787,
       "step": 29670
     },
     {
       "epoch": 89.13,
-      "grad_norm": 8.224078178405762,
+      "grad_norm": 6.973397731781006,
       "learning_rate": 7.02902902902903e-06,
-      "loss": 1.2072,
+      "loss": 1.2035,
       "step": 29680
     },
     {
       "epoch": 89.16,
-      "grad_norm": 5.6268630027771,
+      "grad_norm": 4.775190830230713,
       "learning_rate": 7.028028028028029e-06,
-      "loss": 1.1354,
+      "loss": 1.1326,
       "step": 29690
     },
     {
       "epoch": 89.19,
-      "grad_norm": 6.472921848297119,
+      "grad_norm": 6.8545403480529785,
       "learning_rate": 7.027027027027028e-06,
-      "loss": 1.152,
+      "loss": 1.1456,
       "step": 29700
     },
     {
       "epoch": 89.22,
-      "grad_norm": 8.37695598602295,
+      "grad_norm": 9.47134780883789,
       "learning_rate": 7.026026026026027e-06,
-      "loss": 1.1427,
+      "loss": 1.1454,
       "step": 29710
     },
     {
       "epoch": 89.25,
-      "grad_norm": 7.0515241622924805,
+      "grad_norm": 6.424756050109863,
       "learning_rate": 7.025025025025025e-06,
-      "loss": 1.1311,
+      "loss": 1.1299,
       "step": 29720
     },
     {
       "epoch": 89.28,
-      "grad_norm": 8.603889465332031,
+      "grad_norm": 6.5139641761779785,
       "learning_rate": 7.024024024024024e-06,
-      "loss": 1.1989,
+      "loss": 1.1996,
       "step": 29730
     },
     {
       "epoch": 89.31,
-      "grad_norm": 8.445611000061035,
+      "grad_norm": 6.890460968017578,
       "learning_rate": 7.023023023023024e-06,
-      "loss": 1.1698,
+      "loss": 1.1643,
       "step": 29740
     },
     {
       "epoch": 89.34,
-      "grad_norm": 11.93207836151123,
+      "grad_norm": 6.2296905517578125,
       "learning_rate": 7.022022022022022e-06,
-      "loss": 1.2104,
+      "loss": 1.2083,
       "step": 29750
     },
     {
       "epoch": 89.37,
-      "grad_norm": 7.36000394821167,
+      "grad_norm": 9.053656578063965,
       "learning_rate": 7.021021021021022e-06,
-      "loss": 1.1282,
+      "loss": 1.1401,
       "step": 29760
     },
     {
       "epoch": 89.4,
-      "grad_norm": 7.308358192443848,
+      "grad_norm": 8.641693115234375,
       "learning_rate": 7.020020020020021e-06,
-      "loss": 1.2811,
+      "loss": 1.2648,
       "step": 29770
     },
     {
       "epoch": 89.43,
-      "grad_norm": 8.458122253417969,
+      "grad_norm": 8.004524230957031,
       "learning_rate": 7.01901901901902e-06,
-      "loss": 1.1844,
+      "loss": 1.1702,
       "step": 29780
     },
     {
       "epoch": 89.46,
-      "grad_norm": 8.380297660827637,
+      "grad_norm": 9.379302024841309,
       "learning_rate": 7.018018018018019e-06,
-      "loss": 1.1316,
+      "loss": 1.1285,
       "step": 29790
     },
     {
       "epoch": 89.49,
-      "grad_norm": 10.299762725830078,
+      "grad_norm": 10.458609580993652,
       "learning_rate": 7.017017017017017e-06,
-      "loss": 1.2161,
+      "loss": 1.2178,
       "step": 29800
     },
     {
       "epoch": 89.52,
-      "grad_norm": 9.411429405212402,
+      "grad_norm": 7.409646511077881,
       "learning_rate": 7.0160160160160164e-06,
-      "loss": 1.2271,
+      "loss": 1.2394,
       "step": 29810
     },
     {
       "epoch": 89.55,
-      "grad_norm": 5.538705825805664,
+      "grad_norm": 8.308655738830566,
       "learning_rate": 7.015015015015015e-06,
-      "loss": 1.1742,
+      "loss": 1.1707,
       "step": 29820
     },
     {
       "epoch": 89.58,
-      "grad_norm": 6.125629901885986,
+      "grad_norm": 6.165274143218994,
       "learning_rate": 7.0140140140140145e-06,
-      "loss": 1.1685,
+      "loss": 1.176,
       "step": 29830
     },
     {
       "epoch": 89.61,
-      "grad_norm": 6.32869291305542,
+      "grad_norm": 7.610963344573975,
       "learning_rate": 7.013013013013014e-06,
-      "loss": 1.1948,
+      "loss": 1.1957,
       "step": 29840
     },
     {
       "epoch": 89.64,
-      "grad_norm": 7.772653102874756,
+      "grad_norm": 8.47768783569336,
       "learning_rate": 7.0120120120120125e-06,
-      "loss": 1.1672,
+      "loss": 1.165,
       "step": 29850
     },
     {
       "epoch": 89.67,
-      "grad_norm": 8.450268745422363,
+      "grad_norm": 6.642151355743408,
       "learning_rate": 7.011011011011012e-06,
-      "loss": 1.1843,
+      "loss": 1.1799,
       "step": 29860
     },
     {
       "epoch": 89.7,
-      "grad_norm": 7.015895843505859,
+      "grad_norm": 7.205813407897949,
       "learning_rate": 7.010010010010011e-06,
-      "loss": 1.1444,
+      "loss": 1.1425,
       "step": 29870
     },
     {
       "epoch": 89.73,
-      "grad_norm": 7.125374794006348,
+      "grad_norm": 8.242076873779297,
       "learning_rate": 7.009009009009009e-06,
-      "loss": 1.2304,
+      "loss": 1.2331,
       "step": 29880
     },
     {
       "epoch": 89.76,
-      "grad_norm": 6.937643527984619,
+      "grad_norm": 6.999196529388428,
       "learning_rate": 7.008008008008009e-06,
-      "loss": 1.1046,
+      "loss": 1.1012,
       "step": 29890
     },
     {
       "epoch": 89.79,
-      "grad_norm": 9.62720012664795,
+      "grad_norm": 10.215975761413574,
       "learning_rate": 7.007007007007007e-06,
-      "loss": 1.1575,
+      "loss": 1.1633,
       "step": 29900
     },
     {
       "epoch": 89.82,
-      "grad_norm": 7.503788471221924,
+      "grad_norm": 7.662245273590088,
       "learning_rate": 7.006006006006007e-06,
-      "loss": 1.1251,
+      "loss": 1.1266,
       "step": 29910
     },
     {
       "epoch": 89.85,
-      "grad_norm": 6.970325469970703,
+      "grad_norm": 8.536480903625488,
       "learning_rate": 7.005005005005005e-06,
-      "loss": 1.1498,
+      "loss": 1.1531,
       "step": 29920
     },
     {
       "epoch": 89.88,
-      "grad_norm": 6.460524559020996,
+      "grad_norm": 5.634714126586914,
       "learning_rate": 7.004004004004005e-06,
-      "loss": 1.1644,
+      "loss": 1.1689,
       "step": 29930
     },
     {
       "epoch": 89.91,
-      "grad_norm": 4.831651210784912,
+      "grad_norm": 7.643056392669678,
       "learning_rate": 7.003003003003004e-06,
-      "loss": 1.1201,
+      "loss": 1.1189,
       "step": 29940
     },
     {
       "epoch": 89.94,
-      "grad_norm": 6.08679723739624,
+      "grad_norm": 6.799426555633545,
       "learning_rate": 7.002002002002003e-06,
-      "loss": 1.1939,
+      "loss": 1.1795,
       "step": 29950
     },
     {
       "epoch": 89.97,
-      "grad_norm": 8.646612167358398,
+      "grad_norm": 7.537627220153809,
       "learning_rate": 7.001001001001002e-06,
-      "loss": 1.1634,
+      "loss": 1.1704,
       "step": 29960
     },
     {
       "epoch": 90.0,
-      "grad_norm": 55.731197357177734,
+      "grad_norm": 71.41098022460938,
       "learning_rate": 7e-06,
-      "loss": 1.4006,
+      "loss": 1.3978,
       "step": 29970
     },
     {
       "epoch": 90.0,
-      "eval_accuracy": 0.7945,
-      "eval_loss": 0.6723083853721619,
-      "eval_runtime": 5.4026,
-      "eval_samples_per_second": 1850.97,
-      "eval_steps_per_second": 7.404,
+      "eval_accuracy": 0.7951,
+      "eval_loss": 0.6692058444023132,
+      "eval_runtime": 5.2586,
+      "eval_samples_per_second": 1901.661,
+      "eval_steps_per_second": 7.607,
       "step": 29970
     },
     {
       "epoch": 90.03,
-      "grad_norm": 4.811244964599609,
+      "grad_norm": 6.629955291748047,
       "learning_rate": 6.998998998998999e-06,
-      "loss": 1.1036,
+      "loss": 1.1238,
       "step": 29980
     },
     {
       "epoch": 90.06,
-      "grad_norm": 9.329771041870117,
+      "grad_norm": 8.756502151489258,
       "learning_rate": 6.997997997997999e-06,
-      "loss": 1.1589,
+      "loss": 1.1616,
       "step": 29990
     },
     {
       "epoch": 90.09,
-      "grad_norm": 7.471341609954834,
+      "grad_norm": 13.240763664245605,
       "learning_rate": 6.996996996996997e-06,
-      "loss": 1.2602,
+      "loss": 1.2551,
       "step": 30000
     },
     {
       "epoch": 90.12,
-      "grad_norm": 5.954415321350098,
+      "grad_norm": 6.298533916473389,
       "learning_rate": 6.995995995995997e-06,
-      "loss": 1.1834,
+      "loss": 1.1826,
       "step": 30010
     },
     {
       "epoch": 90.15,
-      "grad_norm": 6.927358627319336,
+      "grad_norm": 7.2852678298950195,
       "learning_rate": 6.994994994994995e-06,
-      "loss": 1.119,
+      "loss": 1.1256,
       "step": 30020
     },
     {
       "epoch": 90.18,
-      "grad_norm": 5.414323329925537,
+      "grad_norm": 5.816061496734619,
       "learning_rate": 6.993993993993995e-06,
-      "loss": 1.1732,
+      "loss": 1.1721,
       "step": 30030
     },
     {
       "epoch": 90.21,
-      "grad_norm": 10.485233306884766,
+      "grad_norm": 7.473488807678223,
       "learning_rate": 6.992992992992994e-06,
-      "loss": 1.19,
+      "loss": 1.1884,
       "step": 30040
     },
     {
       "epoch": 90.24,
-      "grad_norm": 6.248806953430176,
+      "grad_norm": 7.965756416320801,
       "learning_rate": 6.991991991991992e-06,
-      "loss": 1.1915,
+      "loss": 1.1783,
       "step": 30050
     },
     {
       "epoch": 90.27,
-      "grad_norm": 7.32239294052124,
+      "grad_norm": 11.726537704467773,
       "learning_rate": 6.9909909909909915e-06,
-      "loss": 1.1244,
+      "loss": 1.1236,
       "step": 30060
     },
     {
       "epoch": 90.3,
-      "grad_norm": 8.021658897399902,
+      "grad_norm": 6.3191328048706055,
       "learning_rate": 6.98998998998999e-06,
-      "loss": 1.1657,
+      "loss": 1.1829,
       "step": 30070
     },
     {
       "epoch": 90.33,
-      "grad_norm": 5.704627990722656,
+      "grad_norm": 5.989355087280273,
       "learning_rate": 6.9889889889889895e-06,
-      "loss": 1.1977,
+      "loss": 1.2,
       "step": 30080
     },
     {
       "epoch": 90.36,
-      "grad_norm": 6.780069828033447,
+      "grad_norm": 7.028436183929443,
       "learning_rate": 6.987987987987989e-06,
-      "loss": 1.2274,
+      "loss": 1.2271,
       "step": 30090
     },
     {
       "epoch": 90.39,
-      "grad_norm": 6.856760501861572,
+      "grad_norm": 7.158446311950684,
       "learning_rate": 6.9869869869869876e-06,
-      "loss": 1.0694,
+      "loss": 1.0629,
       "step": 30100
     },
     {
       "epoch": 90.42,
-      "grad_norm": 9.365646362304688,
+      "grad_norm": 9.737060546875,
       "learning_rate": 6.985985985985987e-06,
-      "loss": 1.2466,
+      "loss": 1.2427,
       "step": 30110
     },
     {
       "epoch": 90.45,
-      "grad_norm": 6.114166736602783,
+      "grad_norm": 7.225637912750244,
       "learning_rate": 6.984984984984986e-06,
-      "loss": 1.085,
+      "loss": 1.091,
       "step": 30120
     },
     {
       "epoch": 90.48,
-      "grad_norm": 7.0263495445251465,
+      "grad_norm": 7.9278998374938965,
       "learning_rate": 6.983983983983984e-06,
-      "loss": 1.1388,
+      "loss": 1.1486,
       "step": 30130
     },
     {
       "epoch": 90.51,
-      "grad_norm": 5.184991836547852,
+      "grad_norm": 7.631525993347168,
       "learning_rate": 6.982982982982983e-06,
-      "loss": 1.2327,
+      "loss": 1.2213,
       "step": 30140
     },
     {
       "epoch": 90.54,
-      "grad_norm": 5.6381683349609375,
+      "grad_norm": 5.5524420738220215,
       "learning_rate": 6.981981981981982e-06,
-      "loss": 1.1955,
+      "loss": 1.1969,
       "step": 30150
     },
     {
       "epoch": 90.57,
-      "grad_norm": 6.414748668670654,
+      "grad_norm": 7.770815372467041,
       "learning_rate": 6.980980980980982e-06,
-      "loss": 1.1173,
+      "loss": 1.1196,
       "step": 30160
     },
     {
       "epoch": 90.6,
-      "grad_norm": 7.553120136260986,
+      "grad_norm": 9.418986320495605,
       "learning_rate": 6.97997997997998e-06,
-      "loss": 1.1435,
+      "loss": 1.1392,
       "step": 30170
     },
     {
       "epoch": 90.63,
-      "grad_norm": 8.109040260314941,
+      "grad_norm": 8.278267860412598,
       "learning_rate": 6.97897897897898e-06,
-      "loss": 1.1956,
+      "loss": 1.1828,
       "step": 30180
     },
     {
       "epoch": 90.66,
-      "grad_norm": 6.306792736053467,
+      "grad_norm": 6.9556474685668945,
       "learning_rate": 6.977977977977979e-06,
-      "loss": 1.2683,
+      "loss": 1.2655,
       "step": 30190
     },
     {
       "epoch": 90.69,
-      "grad_norm": 7.1331329345703125,
+      "grad_norm": 7.39394474029541,
       "learning_rate": 6.976976976976978e-06,
-      "loss": 1.2107,
+      "loss": 1.2332,
       "step": 30200
     },
     {
       "epoch": 90.72,
-      "grad_norm": 6.9958977699279785,
+      "grad_norm": 6.9169230461120605,
       "learning_rate": 6.975975975975977e-06,
-      "loss": 1.1212,
+      "loss": 1.1193,
       "step": 30210
     },
     {
       "epoch": 90.75,
-      "grad_norm": 6.5069379806518555,
+      "grad_norm": 5.931046009063721,
       "learning_rate": 6.974974974974975e-06,
-      "loss": 1.1714,
+      "loss": 1.1786,
       "step": 30220
     },
     {
       "epoch": 90.78,
-      "grad_norm": 9.714550018310547,
+      "grad_norm": 13.64455509185791,
       "learning_rate": 6.973973973973974e-06,
-      "loss": 1.1533,
+      "loss": 1.1529,
       "step": 30230
     },
     {
       "epoch": 90.81,
-      "grad_norm": 6.377060890197754,
+      "grad_norm": 8.475553512573242,
       "learning_rate": 6.972972972972973e-06,
-      "loss": 1.2563,
+      "loss": 1.2462,
       "step": 30240
     },
     {
       "epoch": 90.84,
-      "grad_norm": 7.524052143096924,
+      "grad_norm": 8.59984302520752,
       "learning_rate": 6.971971971971972e-06,
-      "loss": 1.1362,
+      "loss": 1.1359,
       "step": 30250
     },
     {
       "epoch": 90.87,
-      "grad_norm": 7.667417526245117,
+      "grad_norm": 6.810626029968262,
       "learning_rate": 6.970970970970972e-06,
-      "loss": 1.1354,
+      "loss": 1.1182,
       "step": 30260
     },
     {
       "epoch": 90.9,
-      "grad_norm": 7.437399387359619,
+      "grad_norm": 8.246099472045898,
       "learning_rate": 6.9699699699699704e-06,
-      "loss": 1.2546,
+      "loss": 1.2482,
       "step": 30270
     },
     {
       "epoch": 90.93,
-      "grad_norm": 7.558980464935303,
+      "grad_norm": 6.350607872009277,
       "learning_rate": 6.96896896896897e-06,
-      "loss": 1.1886,
+      "loss": 1.1876,
       "step": 30280
     },
     {
       "epoch": 90.96,
-      "grad_norm": 6.427209377288818,
+      "grad_norm": 5.818161487579346,
       "learning_rate": 6.967967967967969e-06,
-      "loss": 1.1788,
+      "loss": 1.1779,
       "step": 30290
     },
     {
       "epoch": 90.99,
-      "grad_norm": 4.991384029388428,
+      "grad_norm": 7.396366119384766,
       "learning_rate": 6.966966966966967e-06,
-      "loss": 1.1093,
+      "loss": 1.1199,
       "step": 30300
     },
     {
       "epoch": 91.0,
-      "eval_accuracy": 0.7945,
-      "eval_loss": 0.6668232083320618,
-      "eval_runtime": 5.4257,
-      "eval_samples_per_second": 1843.089,
-      "eval_steps_per_second": 7.372,
+      "eval_accuracy": 0.7943,
+      "eval_loss": 0.6668897271156311,
+      "eval_runtime": 5.2541,
+      "eval_samples_per_second": 1903.285,
+      "eval_steps_per_second": 7.613,
       "step": 30303
     },
     {
       "epoch": 91.02,
-      "grad_norm": 4.637133598327637,
+      "grad_norm": 5.315195560455322,
       "learning_rate": 6.9659659659659665e-06,
-      "loss": 1.2642,
+      "loss": 1.2931,
       "step": 30310
     },
     {
       "epoch": 91.05,
-      "grad_norm": 8.46082878112793,
+      "grad_norm": 12.762721061706543,
       "learning_rate": 6.964964964964965e-06,
       "loss": 1.2464,
       "step": 30320
     },
     {
       "epoch": 91.08,
-      "grad_norm": 7.725747108459473,
+      "grad_norm": 9.02338981628418,
       "learning_rate": 6.9639639639639646e-06,
-      "loss": 1.1175,
+      "loss": 1.0976,
       "step": 30330
     },
     {
       "epoch": 91.11,
-      "grad_norm": 6.999662399291992,
+      "grad_norm": 7.152195453643799,
       "learning_rate": 6.962962962962964e-06,
-      "loss": 1.1813,
+      "loss": 1.1893,
       "step": 30340
     },
     {
       "epoch": 91.14,
-      "grad_norm": 8.217296600341797,
+      "grad_norm": 10.374801635742188,
       "learning_rate": 6.961961961961963e-06,
-      "loss": 1.1382,
+      "loss": 1.1438,
       "step": 30350
     },
     {
       "epoch": 91.17,
-      "grad_norm": 5.7958574295043945,
+      "grad_norm": 5.80914831161499,
       "learning_rate": 6.960960960960962e-06,
-      "loss": 1.1627,
+      "loss": 1.1695,
       "step": 30360
     },
     {
       "epoch": 91.2,
-      "grad_norm": 6.447906017303467,
+      "grad_norm": 9.383339881896973,
       "learning_rate": 6.959959959959961e-06,
-      "loss": 1.1428,
+      "loss": 1.149,
       "step": 30370
     },
     {
       "epoch": 91.23,
-      "grad_norm": 8.714007377624512,
+      "grad_norm": 8.061676979064941,
       "learning_rate": 6.958958958958959e-06,
-      "loss": 1.1208,
+      "loss": 1.1194,
       "step": 30380
     },
     {
       "epoch": 91.26,
-      "grad_norm": 6.201626300811768,
+      "grad_norm": 11.344412803649902,
       "learning_rate": 6.957957957957958e-06,
-      "loss": 1.1375,
+      "loss": 1.1432,
       "step": 30390
     },
     {
       "epoch": 91.29,
-      "grad_norm": 7.154749870300293,
+      "grad_norm": 5.275724411010742,
       "learning_rate": 6.956956956956957e-06,
-      "loss": 1.1744,
+      "loss": 1.1646,
       "step": 30400
     },
     {
       "epoch": 91.32,
-      "grad_norm": 7.352416515350342,
+      "grad_norm": 6.316702842712402,
       "learning_rate": 6.955955955955957e-06,
-      "loss": 1.1323,
+      "loss": 1.1241,
       "step": 30410
     },
     {
       "epoch": 91.35,
-      "grad_norm": 7.078407287597656,
+      "grad_norm": 7.244926452636719,
       "learning_rate": 6.954954954954955e-06,
-      "loss": 1.2143,
+      "loss": 1.2167,
       "step": 30420
     },
     {
       "epoch": 91.38,
-      "grad_norm": 6.858650207519531,
+      "grad_norm": 7.561052322387695,
       "learning_rate": 6.953953953953955e-06,
-      "loss": 1.2092,
+      "loss": 1.2134,
       "step": 30430
     },
     {
       "epoch": 91.41,
-      "grad_norm": 7.10488748550415,
+      "grad_norm": 7.91503381729126,
       "learning_rate": 6.952952952952954e-06,
-      "loss": 1.115,
+      "loss": 1.1084,
       "step": 30440
     },
     {
       "epoch": 91.44,
-      "grad_norm": 6.759337902069092,
+      "grad_norm": 6.19968318939209,
       "learning_rate": 6.951951951951953e-06,
-      "loss": 1.1456,
+      "loss": 1.157,
       "step": 30450
     },
     {
       "epoch": 91.47,
-      "grad_norm": 5.987698078155518,
+      "grad_norm": 8.040701866149902,
       "learning_rate": 6.950950950950952e-06,
-      "loss": 1.1965,
+      "loss": 1.2007,
       "step": 30460
     },
     {
       "epoch": 91.5,
-      "grad_norm": 7.672787666320801,
+      "grad_norm": 12.048487663269043,
       "learning_rate": 6.94994994994995e-06,
-      "loss": 1.1826,
+      "loss": 1.1731,
       "step": 30470
     },
     {
       "epoch": 91.53,
-      "grad_norm": 4.7193474769592285,
+      "grad_norm": 5.376335144042969,
       "learning_rate": 6.948948948948949e-06,
-      "loss": 1.1488,
+      "loss": 1.1468,
       "step": 30480
     },
     {
       "epoch": 91.56,
-      "grad_norm": 6.768847465515137,
+      "grad_norm": 9.183868408203125,
       "learning_rate": 6.947947947947948e-06,
-      "loss": 1.2295,
+      "loss": 1.2218,
       "step": 30490
     },
     {
       "epoch": 91.59,
-      "grad_norm": 7.4480109214782715,
+      "grad_norm": 6.872031211853027,
       "learning_rate": 6.9469469469469474e-06,
-      "loss": 1.1604,
+      "loss": 1.149,
       "step": 30500
     },
     {
       "epoch": 91.62,
-      "grad_norm": 7.380969047546387,
+      "grad_norm": 10.054756164550781,
       "learning_rate": 6.945945945945947e-06,
-      "loss": 1.2017,
+      "loss": 1.2047,
       "step": 30510
     },
     {
       "epoch": 91.65,
-      "grad_norm": 7.190296649932861,
+      "grad_norm": 7.672440052032471,
       "learning_rate": 6.9449449449449455e-06,
-      "loss": 1.1996,
+      "loss": 1.1981,
       "step": 30520
     },
     {
       "epoch": 91.68,
-      "grad_norm": 8.086702346801758,
+      "grad_norm": 7.779347896575928,
       "learning_rate": 6.943943943943945e-06,
-      "loss": 1.2139,
+      "loss": 1.2135,
       "step": 30530
     },
     {
       "epoch": 91.71,
-      "grad_norm": 8.17762279510498,
+      "grad_norm": 7.28846549987793,
       "learning_rate": 6.942942942942944e-06,
-      "loss": 1.1421,
+      "loss": 1.1455,
       "step": 30540
     },
     {
       "epoch": 91.74,
-      "grad_norm": 8.550183296203613,
+      "grad_norm": 8.885599136352539,
       "learning_rate": 6.941941941941942e-06,
-      "loss": 1.1906,
+      "loss": 1.1941,
       "step": 30550
     },
     {
       "epoch": 91.77,
-      "grad_norm": 6.372681617736816,
+      "grad_norm": 6.7873077392578125,
       "learning_rate": 6.9409409409409416e-06,
-      "loss": 1.1401,
+      "loss": 1.1303,
       "step": 30560
     },
     {
       "epoch": 91.8,
-      "grad_norm": 6.655992031097412,
+      "grad_norm": 6.972674369812012,
       "learning_rate": 6.93993993993994e-06,
-      "loss": 1.1416,
+      "loss": 1.1368,
       "step": 30570
     },
     {
       "epoch": 91.83,
-      "grad_norm": 8.381499290466309,
+      "grad_norm": 8.480287551879883,
       "learning_rate": 6.93893893893894e-06,
-      "loss": 1.1533,
+      "loss": 1.1503,
       "step": 30580
     },
     {
       "epoch": 91.86,
-      "grad_norm": 7.269525527954102,
+      "grad_norm": 6.82000732421875,
       "learning_rate": 6.937937937937938e-06,
-      "loss": 1.1327,
+      "loss": 1.1261,
       "step": 30590
     },
     {
       "epoch": 91.89,
-      "grad_norm": 5.984903812408447,
+      "grad_norm": 6.161318302154541,
       "learning_rate": 6.936936936936938e-06,
-      "loss": 1.1172,
+      "loss": 1.1131,
       "step": 30600
     },
     {
       "epoch": 91.92,
-      "grad_norm": 6.23146915435791,
+      "grad_norm": 6.63691520690918,
       "learning_rate": 6.935935935935937e-06,
-      "loss": 1.1276,
+      "loss": 1.1292,
       "step": 30610
     },
     {
       "epoch": 91.95,
-      "grad_norm": 6.560269355773926,
+      "grad_norm": 6.319340705871582,
       "learning_rate": 6.934934934934936e-06,
-      "loss": 1.1536,
+      "loss": 1.1497,
       "step": 30620
     },
     {
       "epoch": 91.98,
-      "grad_norm": 8.06259536743164,
+      "grad_norm": 8.482089042663574,
       "learning_rate": 6.933933933933934e-06,
-      "loss": 1.1468,
+      "loss": 1.136,
       "step": 30630
     },
     {
       "epoch": 92.0,
-      "eval_accuracy": 0.7958,
-      "eval_loss": 0.6611828207969666,
-      "eval_runtime": 5.4841,
-      "eval_samples_per_second": 1823.458,
-      "eval_steps_per_second": 7.294,
+      "eval_accuracy": 0.796,
+      "eval_loss": 0.6611176133155823,
+      "eval_runtime": 5.211,
+      "eval_samples_per_second": 1919.012,
+      "eval_steps_per_second": 7.676,
       "step": 30636
     },
     {
       "epoch": 92.01,
-      "grad_norm": 6.319799423217773,
+      "grad_norm": 6.031079292297363,
       "learning_rate": 6.932932932932933e-06,
-      "loss": 1.2993,
+      "loss": 1.298,
       "step": 30640
     },
     {
       "epoch": 92.04,
-      "grad_norm": 5.174209117889404,
+      "grad_norm": 8.428420066833496,
       "learning_rate": 6.931931931931932e-06,
-      "loss": 1.1048,
+      "loss": 1.1026,
       "step": 30650
     },
     {
       "epoch": 92.07,
-      "grad_norm": 8.868124008178711,
+      "grad_norm": 8.46903133392334,
       "learning_rate": 6.930930930930932e-06,
-      "loss": 1.1348,
+      "loss": 1.1329,
       "step": 30660
     },
     {
       "epoch": 92.1,
-      "grad_norm": 7.2114105224609375,
+      "grad_norm": 5.824512004852295,
       "learning_rate": 6.92992992992993e-06,
-      "loss": 1.1426,
+      "loss": 1.1405,
       "step": 30670
     },
     {
       "epoch": 92.13,
-      "grad_norm": 8.523722648620605,
+      "grad_norm": 4.961424827575684,
       "learning_rate": 6.92892892892893e-06,
-      "loss": 1.1112,
+      "loss": 1.1041,
       "step": 30680
     },
     {
       "epoch": 92.16,
-      "grad_norm": 7.208141803741455,
+      "grad_norm": 7.080151081085205,
       "learning_rate": 6.927927927927928e-06,
-      "loss": 1.1963,
+      "loss": 1.1814,
       "step": 30690
     },
     {
       "epoch": 92.19,
-      "grad_norm": 6.721158504486084,
+      "grad_norm": 9.28423023223877,
       "learning_rate": 6.926926926926928e-06,
-      "loss": 1.1798,
+      "loss": 1.1781,
       "step": 30700
     },
     {
       "epoch": 92.22,
-      "grad_norm": 8.403751373291016,
+      "grad_norm": 7.814840793609619,
       "learning_rate": 6.9259259259259256e-06,
-      "loss": 1.1714,
+      "loss": 1.1668,
       "step": 30710
     },
     {
       "epoch": 92.25,
-      "grad_norm": 9.66375732421875,
+      "grad_norm": 7.176508903503418,
       "learning_rate": 6.924924924924925e-06,
-      "loss": 1.196,
+      "loss": 1.1867,
       "step": 30720
     },
     {
       "epoch": 92.28,
-      "grad_norm": 6.40648889541626,
+      "grad_norm": 5.939242839813232,
       "learning_rate": 6.9239239239239244e-06,
-      "loss": 1.1811,
+      "loss": 1.1776,
       "step": 30730
     },
     {
       "epoch": 92.31,
-      "grad_norm": 9.143586158752441,
+      "grad_norm": 9.421391487121582,
       "learning_rate": 6.922922922922923e-06,
-      "loss": 1.1429,
+      "loss": 1.1305,
       "step": 30740
     },
     {
       "epoch": 92.34,
-      "grad_norm": 6.667135238647461,
+      "grad_norm": 7.00972318649292,
       "learning_rate": 6.9219219219219225e-06,
-      "loss": 1.183,
+      "loss": 1.1902,
       "step": 30750
     },
     {
       "epoch": 92.37,
-      "grad_norm": 7.375683307647705,
+      "grad_norm": 7.352478981018066,
       "learning_rate": 6.920920920920922e-06,
-      "loss": 1.158,
+      "loss": 1.1553,
       "step": 30760
     },
     {
       "epoch": 92.4,
-      "grad_norm": 6.7110185623168945,
+      "grad_norm": 7.195530891418457,
       "learning_rate": 6.9199199199199205e-06,
-      "loss": 1.1231,
+      "loss": 1.1132,
       "step": 30770
     },
     {
       "epoch": 92.43,
-      "grad_norm": 7.548695087432861,
+      "grad_norm": 12.048789024353027,
       "learning_rate": 6.91891891891892e-06,
-      "loss": 1.1347,
+      "loss": 1.1373,
       "step": 30780
     },
     {
       "epoch": 92.46,
-      "grad_norm": 5.977288722991943,
+      "grad_norm": 5.695720195770264,
       "learning_rate": 6.917917917917919e-06,
-      "loss": 1.1412,
+      "loss": 1.1288,
       "step": 30790
     },
     {
       "epoch": 92.49,
-      "grad_norm": 6.128280162811279,
+      "grad_norm": 6.107945442199707,
       "learning_rate": 6.916916916916917e-06,
-      "loss": 1.1492,
+      "loss": 1.1366,
       "step": 30800
     },
     {
       "epoch": 92.52,
-      "grad_norm": 6.067540168762207,
+      "grad_norm": 7.3416852951049805,
       "learning_rate": 6.915915915915916e-06,
-      "loss": 1.0951,
+      "loss": 1.0977,
       "step": 30810
     },
     {
       "epoch": 92.55,
-      "grad_norm": 5.552958965301514,
+      "grad_norm": 11.040433883666992,
       "learning_rate": 6.914914914914915e-06,
-      "loss": 1.1664,
+      "loss": 1.1773,
       "step": 30820
     },
     {
       "epoch": 92.58,
-      "grad_norm": 7.510736465454102,
+      "grad_norm": 7.179296493530273,
       "learning_rate": 6.913913913913915e-06,
-      "loss": 1.14,
+      "loss": 1.1375,
       "step": 30830
     },
     {
       "epoch": 92.61,
-      "grad_norm": 8.292641639709473,
+      "grad_norm": 8.560406684875488,
       "learning_rate": 6.912912912912913e-06,
-      "loss": 1.2147,
+      "loss": 1.2124,
       "step": 30840
     },
     {
       "epoch": 92.64,
-      "grad_norm": 7.383830547332764,
+      "grad_norm": 11.620002746582031,
       "learning_rate": 6.911911911911913e-06,
-      "loss": 1.0987,
+      "loss": 1.1035,
       "step": 30850
     },
     {
       "epoch": 92.67,
-      "grad_norm": 6.869711399078369,
+      "grad_norm": 8.418591499328613,
       "learning_rate": 6.910910910910912e-06,
-      "loss": 1.238,
+      "loss": 1.225,
       "step": 30860
     },
     {
       "epoch": 92.7,
-      "grad_norm": 7.157979488372803,
+      "grad_norm": 8.927936553955078,
       "learning_rate": 6.909909909909911e-06,
-      "loss": 1.1557,
+      "loss": 1.1681,
       "step": 30870
     },
     {
       "epoch": 92.73,
-      "grad_norm": 6.7171950340271,
+      "grad_norm": 7.460107326507568,
       "learning_rate": 6.908908908908909e-06,
-      "loss": 1.1518,
+      "loss": 1.1404,
       "step": 30880
     },
     {
       "epoch": 92.76,
-      "grad_norm": 8.966947555541992,
+      "grad_norm": 10.57024097442627,
       "learning_rate": 6.907907907907908e-06,
-      "loss": 1.1793,
+      "loss": 1.1734,
       "step": 30890
     },
     {
       "epoch": 92.79,
-      "grad_norm": 6.750990390777588,
+      "grad_norm": 8.415785789489746,
       "learning_rate": 6.906906906906907e-06,
-      "loss": 1.1623,
+      "loss": 1.1559,
       "step": 30900
     },
     {
       "epoch": 92.82,
-      "grad_norm": 7.132504940032959,
+      "grad_norm": 7.898650646209717,
       "learning_rate": 6.905905905905907e-06,
-      "loss": 1.126,
+      "loss": 1.1135,
       "step": 30910
     },
     {
       "epoch": 92.85,
-      "grad_norm": 5.424259662628174,
+      "grad_norm": 6.097274303436279,
       "learning_rate": 6.904904904904905e-06,
-      "loss": 1.1628,
+      "loss": 1.1621,
       "step": 30920
     },
     {
       "epoch": 92.88,
-      "grad_norm": 6.082546234130859,
+      "grad_norm": 6.076148986816406,
       "learning_rate": 6.903903903903905e-06,
-      "loss": 1.1057,
+      "loss": 1.1039,
       "step": 30930
     },
     {
       "epoch": 92.91,
-      "grad_norm": 6.9939446449279785,
+      "grad_norm": 6.424147605895996,
       "learning_rate": 6.902902902902903e-06,
-      "loss": 1.0774,
+      "loss": 1.0826,
       "step": 30940
     },
     {
       "epoch": 92.94,
-      "grad_norm": 6.3912177085876465,
+      "grad_norm": 6.9243950843811035,
       "learning_rate": 6.901901901901903e-06,
-      "loss": 1.1923,
+      "loss": 1.1956,
       "step": 30950
     },
     {
       "epoch": 92.97,
-      "grad_norm": 5.117764472961426,
+      "grad_norm": 6.666253566741943,
       "learning_rate": 6.900900900900901e-06,
-      "loss": 1.1783,
+      "loss": 1.1793,
       "step": 30960
     },
     {
       "epoch": 93.0,
-      "eval_accuracy": 0.795,
-      "eval_loss": 0.6684935688972473,
-      "eval_runtime": 5.5798,
-      "eval_samples_per_second": 1792.186,
-      "eval_steps_per_second": 7.169,
+      "eval_accuracy": 0.7951,
+      "eval_loss": 0.6676392555236816,
+      "eval_runtime": 5.47,
+      "eval_samples_per_second": 1828.153,
+      "eval_steps_per_second": 7.313,
       "step": 30969
     },
     {
       "epoch": 93.0,
-      "grad_norm": 8.330206871032715,
+      "grad_norm": 6.5864739418029785,
       "learning_rate": 6.8998998998999e-06,
-      "loss": 1.2901,
+      "loss": 1.3007,
       "step": 30970
     },
     {
       "epoch": 93.03,
-      "grad_norm": 7.613500595092773,
+      "grad_norm": 7.899778842926025,
       "learning_rate": 6.8988988988988995e-06,
-      "loss": 1.1276,
+      "loss": 1.124,
       "step": 30980
     },
     {
       "epoch": 93.06,
-      "grad_norm": 6.797489166259766,
+      "grad_norm": 6.938320159912109,
       "learning_rate": 6.897897897897898e-06,
-      "loss": 1.0903,
+      "loss": 1.1,
       "step": 30990
     },
     {
       "epoch": 93.09,
-      "grad_norm": 7.753300189971924,
+      "grad_norm": 7.428671360015869,
       "learning_rate": 6.8968968968968975e-06,
-      "loss": 1.149,
+      "loss": 1.1531,
       "step": 31000
     },
     {
       "epoch": 93.12,
-      "grad_norm": 6.00712251663208,
+      "grad_norm": 6.638075351715088,
       "learning_rate": 6.895895895895897e-06,
-      "loss": 1.1522,
+      "loss": 1.1593,
       "step": 31010
     },
     {
       "epoch": 93.15,
-      "grad_norm": 11.17741584777832,
+      "grad_norm": 12.125579833984375,
       "learning_rate": 6.8948948948948955e-06,
-      "loss": 1.1325,
+      "loss": 1.1316,
       "step": 31020
     },
     {
       "epoch": 93.18,
-      "grad_norm": 8.045567512512207,
+      "grad_norm": 7.045203685760498,
       "learning_rate": 6.893893893893895e-06,
-      "loss": 1.0905,
+      "loss": 1.0856,
       "step": 31030
     },
     {
       "epoch": 93.21,
-      "grad_norm": 6.880770683288574,
+      "grad_norm": 6.20102071762085,
       "learning_rate": 6.892892892892894e-06,
-      "loss": 1.1599,
+      "loss": 1.1594,
       "step": 31040
     },
     {
       "epoch": 93.24,
-      "grad_norm": 8.898472785949707,
+      "grad_norm": 11.137961387634277,
       "learning_rate": 6.891891891891892e-06,
-      "loss": 1.142,
+      "loss": 1.1303,
       "step": 31050
     },
     {
       "epoch": 93.27,
-      "grad_norm": 7.255890846252441,
+      "grad_norm": 9.675273895263672,
       "learning_rate": 6.890890890890891e-06,
-      "loss": 1.1502,
+      "loss": 1.1548,
       "step": 31060
     },
     {
       "epoch": 93.3,
-      "grad_norm": 5.260316371917725,
+      "grad_norm": 4.852452754974365,
       "learning_rate": 6.88988988988989e-06,
-      "loss": 1.095,
+      "loss": 1.1074,
       "step": 31070
     },
     {
       "epoch": 93.33,
-      "grad_norm": 8.212627410888672,
+      "grad_norm": 8.0037260055542,
       "learning_rate": 6.88888888888889e-06,
-      "loss": 1.1696,
+      "loss": 1.1798,
       "step": 31080
     },
     {
       "epoch": 93.36,
-      "grad_norm": 5.671102523803711,
+      "grad_norm": 5.425944805145264,
       "learning_rate": 6.887887887887888e-06,
-      "loss": 1.1999,
+      "loss": 1.1939,
       "step": 31090
     },
     {
       "epoch": 93.39,
-      "grad_norm": 5.882233619689941,
+      "grad_norm": 6.005474090576172,
       "learning_rate": 6.886886886886888e-06,
-      "loss": 1.119,
+      "loss": 1.1254,
       "step": 31100
     },
     {
       "epoch": 93.42,
-      "grad_norm": 5.309819221496582,
+      "grad_norm": 7.455267906188965,
       "learning_rate": 6.885885885885887e-06,
-      "loss": 1.1934,
+      "loss": 1.1897,
       "step": 31110
     },
     {
       "epoch": 93.45,
-      "grad_norm": 7.278177261352539,
+      "grad_norm": 11.734363555908203,
       "learning_rate": 6.884884884884886e-06,
-      "loss": 1.1495,
+      "loss": 1.1594,
       "step": 31120
     },
     {
       "epoch": 93.48,
-      "grad_norm": 8.552815437316895,
+      "grad_norm": 11.116238594055176,
       "learning_rate": 6.883883883883884e-06,
-      "loss": 1.1427,
+      "loss": 1.1405,
       "step": 31130
     },
     {
       "epoch": 93.51,
-      "grad_norm": 7.241766929626465,
+      "grad_norm": 6.652608871459961,
       "learning_rate": 6.882882882882883e-06,
-      "loss": 1.1661,
+      "loss": 1.1596,
       "step": 31140
     },
     {
       "epoch": 93.54,
-      "grad_norm": 6.573330879211426,
+      "grad_norm": 5.795431613922119,
       "learning_rate": 6.881881881881882e-06,
-      "loss": 1.1702,
+      "loss": 1.1727,
       "step": 31150
     },
     {
       "epoch": 93.57,
-      "grad_norm": 7.930480480194092,
+      "grad_norm": 8.914377212524414,
       "learning_rate": 6.880880880880881e-06,
-      "loss": 1.2259,
+      "loss": 1.2078,
       "step": 31160
     },
     {
       "epoch": 93.6,
-      "grad_norm": 11.412364959716797,
+      "grad_norm": 9.15998649597168,
       "learning_rate": 6.87987987987988e-06,
-      "loss": 1.1326,
+      "loss": 1.1289,
       "step": 31170
     },
     {
       "epoch": 93.63,
-      "grad_norm": 6.555348873138428,
+      "grad_norm": 15.137701988220215,
       "learning_rate": 6.87887887887888e-06,
-      "loss": 1.1576,
+      "loss": 1.1598,
       "step": 31180
     },
     {
       "epoch": 93.66,
-      "grad_norm": 7.061331272125244,
+      "grad_norm": 7.267025470733643,
       "learning_rate": 6.8778778778778784e-06,
-      "loss": 1.2076,
+      "loss": 1.2144,
       "step": 31190
     },
     {
       "epoch": 93.69,
-      "grad_norm": 8.472617149353027,
+      "grad_norm": 8.056133270263672,
       "learning_rate": 6.876876876876878e-06,
-      "loss": 1.1763,
+      "loss": 1.1755,
       "step": 31200
     },
     {
       "epoch": 93.72,
-      "grad_norm": 5.836824893951416,
+      "grad_norm": 5.211737155914307,
       "learning_rate": 6.875875875875876e-06,
-      "loss": 1.1237,
+      "loss": 1.1231,
       "step": 31210
     },
     {
       "epoch": 93.75,
-      "grad_norm": 6.149411201477051,
+      "grad_norm": 5.655396461486816,
       "learning_rate": 6.874874874874875e-06,
-      "loss": 1.1487,
+      "loss": 1.1488,
       "step": 31220
     },
     {
       "epoch": 93.78,
-      "grad_norm": 6.444913387298584,
+      "grad_norm": 7.678549289703369,
       "learning_rate": 6.8738738738738745e-06,
-      "loss": 1.1771,
+      "loss": 1.1842,
       "step": 31230
     },
     {
       "epoch": 93.81,
-      "grad_norm": 10.362194061279297,
+      "grad_norm": 9.562945365905762,
       "learning_rate": 6.872872872872873e-06,
-      "loss": 1.2315,
+      "loss": 1.2087,
       "step": 31240
     },
     {
       "epoch": 93.84,
-      "grad_norm": 7.5833258628845215,
+      "grad_norm": 8.392518043518066,
       "learning_rate": 6.8718718718718725e-06,
-      "loss": 1.221,
+      "loss": 1.2186,
       "step": 31250
     },
     {
       "epoch": 93.87,
-      "grad_norm": 10.795374870300293,
+      "grad_norm": 11.32370662689209,
       "learning_rate": 6.870870870870872e-06,
-      "loss": 1.0478,
+      "loss": 1.0463,
       "step": 31260
     },
     {
       "epoch": 93.9,
-      "grad_norm": 7.2002692222595215,
+      "grad_norm": 9.70592212677002,
       "learning_rate": 6.869869869869871e-06,
-      "loss": 1.1705,
+      "loss": 1.1694,
       "step": 31270
     },
     {
       "epoch": 93.93,
-      "grad_norm": 6.534556865692139,
+      "grad_norm": 5.9941792488098145,
       "learning_rate": 6.86886886886887e-06,
-      "loss": 1.1229,
+      "loss": 1.1222,
       "step": 31280
     },
     {
       "epoch": 93.96,
-      "grad_norm": 6.609338760375977,
+      "grad_norm": 6.824159145355225,
       "learning_rate": 6.867867867867869e-06,
-      "loss": 1.0817,
+      "loss": 1.0789,
       "step": 31290
     },
     {
       "epoch": 93.99,
-      "grad_norm": 6.139440059661865,
+      "grad_norm": 8.797876358032227,
       "learning_rate": 6.866866866866867e-06,
-      "loss": 1.1586,
+      "loss": 1.1455,
       "step": 31300
     },
     {
       "epoch": 94.0,
-      "eval_accuracy": 0.7964,
-      "eval_loss": 0.6569728851318359,
-      "eval_runtime": 5.4457,
-      "eval_samples_per_second": 1836.32,
-      "eval_steps_per_second": 7.345,
+      "eval_accuracy": 0.7947,
+      "eval_loss": 0.6575410962104797,
+      "eval_runtime": 5.2672,
+      "eval_samples_per_second": 1898.541,
+      "eval_steps_per_second": 7.594,
       "step": 31302
     },
     {
       "epoch": 94.02,
-      "grad_norm": 8.942100524902344,
+      "grad_norm": 8.379203796386719,
       "learning_rate": 6.865865865865866e-06,
-      "loss": 1.3097,
+      "loss": 1.3063,
       "step": 31310
     },
     {
       "epoch": 94.05,
-      "grad_norm": 6.0739240646362305,
+      "grad_norm": 6.4335761070251465,
       "learning_rate": 6.864864864864865e-06,
-      "loss": 1.1328,
+      "loss": 1.1331,
       "step": 31320
     },
     {
       "epoch": 94.08,
-      "grad_norm": 6.311398983001709,
+      "grad_norm": 7.168081760406494,
       "learning_rate": 6.863863863863865e-06,
-      "loss": 1.1594,
+      "loss": 1.1644,
       "step": 31330
     },
     {
       "epoch": 94.11,
-      "grad_norm": 6.879790782928467,
+      "grad_norm": 5.67033576965332,
       "learning_rate": 6.862862862862863e-06,
-      "loss": 1.1931,
+      "loss": 1.1895,
       "step": 31340
     },
     {
       "epoch": 94.14,
-      "grad_norm": 5.756179332733154,
+      "grad_norm": 7.006296157836914,
       "learning_rate": 6.861861861861863e-06,
-      "loss": 1.1302,
+      "loss": 1.1244,
       "step": 31350
     },
     {
       "epoch": 94.17,
-      "grad_norm": 8.137129783630371,
+      "grad_norm": 8.699432373046875,
       "learning_rate": 6.860860860860862e-06,
-      "loss": 1.15,
+      "loss": 1.1534,
       "step": 31360
     },
     {
       "epoch": 94.2,
-      "grad_norm": 6.815968990325928,
+      "grad_norm": 5.796480655670166,
       "learning_rate": 6.859859859859861e-06,
-      "loss": 1.1786,
+      "loss": 1.1891,
       "step": 31370
     },
     {
       "epoch": 94.23,
-      "grad_norm": 8.658731460571289,
+      "grad_norm": 6.610452651977539,
       "learning_rate": 6.8588588588588585e-06,
-      "loss": 1.1265,
+      "loss": 1.1411,
       "step": 31380
     },
     {
       "epoch": 94.26,
-      "grad_norm": 7.69881010055542,
+      "grad_norm": 7.680984020233154,
       "learning_rate": 6.857857857857858e-06,
-      "loss": 1.1722,
+      "loss": 1.1692,
       "step": 31390
     },
     {
       "epoch": 94.29,
-      "grad_norm": 6.562985897064209,
+      "grad_norm": 7.5232625007629395,
       "learning_rate": 6.856856856856857e-06,
-      "loss": 1.1724,
+      "loss": 1.163,
       "step": 31400
     },
     {
       "epoch": 94.32,
-      "grad_norm": 9.179701805114746,
+      "grad_norm": 7.576277256011963,
       "learning_rate": 6.855855855855856e-06,
-      "loss": 1.093,
+      "loss": 1.0959,
       "step": 31410
     },
     {
       "epoch": 94.35,
-      "grad_norm": 7.12808084487915,
+      "grad_norm": 6.627995014190674,
       "learning_rate": 6.8548548548548554e-06,
-      "loss": 1.1649,
+      "loss": 1.1653,
       "step": 31420
     },
     {
       "epoch": 94.38,
-      "grad_norm": 5.918236255645752,
+      "grad_norm": 8.58071517944336,
       "learning_rate": 6.853853853853855e-06,
-      "loss": 1.1178,
+      "loss": 1.121,
       "step": 31430
     },
     {
       "epoch": 94.41,
-      "grad_norm": 4.957930088043213,
+      "grad_norm": 4.904599189758301,
       "learning_rate": 6.8528528528528535e-06,
-      "loss": 1.1432,
+      "loss": 1.1407,
       "step": 31440
     },
     {
       "epoch": 94.44,
-      "grad_norm": 6.574726581573486,
+      "grad_norm": 7.362645626068115,
       "learning_rate": 6.851851851851853e-06,
-      "loss": 1.1407,
+      "loss": 1.1319,
       "step": 31450
     },
     {
       "epoch": 94.47,
-      "grad_norm": 6.782023906707764,
+      "grad_norm": 7.548234939575195,
       "learning_rate": 6.850850850850851e-06,
-      "loss": 1.1014,
+      "loss": 1.1013,
       "step": 31460
     },
     {
       "epoch": 94.5,
-      "grad_norm": 7.388020992279053,
+      "grad_norm": 5.285892486572266,
       "learning_rate": 6.84984984984985e-06,
-      "loss": 1.1714,
+      "loss": 1.179,
       "step": 31470
     },
     {
       "epoch": 94.53,
-      "grad_norm": 5.226505756378174,
+      "grad_norm": 5.58642578125,
       "learning_rate": 6.8488488488488495e-06,
-      "loss": 1.1748,
+      "loss": 1.1687,
       "step": 31480
     },
     {
       "epoch": 94.56,
-      "grad_norm": 6.045767307281494,
+      "grad_norm": 6.690946102142334,
       "learning_rate": 6.847847847847848e-06,
-      "loss": 1.1374,
+      "loss": 1.131,
       "step": 31490
     },
     {
       "epoch": 94.59,
-      "grad_norm": 6.507619380950928,
+      "grad_norm": 7.300330638885498,
       "learning_rate": 6.846846846846848e-06,
-      "loss": 1.1432,
+      "loss": 1.1276,
       "step": 31500
     },
     {
       "epoch": 94.62,
-      "grad_norm": 7.124127388000488,
+      "grad_norm": 5.972383499145508,
       "learning_rate": 6.845845845845846e-06,
-      "loss": 1.1815,
+      "loss": 1.1848,
       "step": 31510
     },
     {
       "epoch": 94.65,
-      "grad_norm": 9.199424743652344,
+      "grad_norm": 7.8304972648620605,
       "learning_rate": 6.844844844844846e-06,
-      "loss": 1.0968,
+      "loss": 1.106,
       "step": 31520
     },
     {
       "epoch": 94.68,
-      "grad_norm": 7.129841327667236,
+      "grad_norm": 9.08266830444336,
       "learning_rate": 6.843843843843845e-06,
-      "loss": 1.1192,
+      "loss": 1.1104,
       "step": 31530
     },
     {
       "epoch": 94.71,
-      "grad_norm": 6.538386821746826,
+      "grad_norm": 6.766589641571045,
       "learning_rate": 6.842842842842844e-06,
-      "loss": 1.1582,
+      "loss": 1.1405,
       "step": 31540
     },
     {
       "epoch": 94.74,
-      "grad_norm": 6.1067214012146,
+      "grad_norm": 5.635129928588867,
       "learning_rate": 6.841841841841842e-06,
-      "loss": 1.2174,
+      "loss": 1.2163,
       "step": 31550
     },
     {
       "epoch": 94.77,
-      "grad_norm": 5.283570289611816,
+      "grad_norm": 7.512295246124268,
       "learning_rate": 6.840840840840841e-06,
-      "loss": 1.1493,
+      "loss": 1.1532,
       "step": 31560
     },
     {
       "epoch": 94.8,
-      "grad_norm": 7.444109916687012,
+      "grad_norm": 11.772706031799316,
       "learning_rate": 6.83983983983984e-06,
-      "loss": 1.2032,
+      "loss": 1.1931,
       "step": 31570
     },
     {
       "epoch": 94.83,
-      "grad_norm": 8.551230430603027,
+      "grad_norm": 10.87479019165039,
       "learning_rate": 6.83883883883884e-06,
-      "loss": 1.1569,
+      "loss": 1.1594,
       "step": 31580
     },
     {
       "epoch": 94.86,
-      "grad_norm": 7.116494655609131,
+      "grad_norm": 9.105756759643555,
       "learning_rate": 6.837837837837838e-06,
-      "loss": 1.1431,
+      "loss": 1.1398,
       "step": 31590
     },
     {
       "epoch": 94.89,
-      "grad_norm": 8.593063354492188,
+      "grad_norm": 8.844403266906738,
       "learning_rate": 6.836836836836838e-06,
-      "loss": 1.2074,
+      "loss": 1.2071,
       "step": 31600
     },
     {
       "epoch": 94.92,
-      "grad_norm": 9.137843132019043,
+      "grad_norm": 8.380614280700684,
       "learning_rate": 6.835835835835836e-06,
-      "loss": 1.1742,
+      "loss": 1.1655,
       "step": 31610
     },
     {
       "epoch": 94.95,
-      "grad_norm": 5.451433181762695,
+      "grad_norm": 6.639585018157959,
       "learning_rate": 6.834834834834836e-06,
-      "loss": 1.1324,
+      "loss": 1.1314,
       "step": 31620
     },
     {
       "epoch": 94.98,
-      "grad_norm": 7.062803745269775,
+      "grad_norm": 6.987116813659668,
       "learning_rate": 6.8338338338338336e-06,
-      "loss": 1.1325,
+      "loss": 1.1295,
       "step": 31630
     },
     {
       "epoch": 95.0,
-      "eval_accuracy": 0.7946,
-      "eval_loss": 0.6604899168014526,
-      "eval_runtime": 5.7653,
-      "eval_samples_per_second": 1734.508,
-      "eval_steps_per_second": 6.938,
+      "eval_accuracy": 0.7945,
+      "eval_loss": 0.6601707339286804,
+      "eval_runtime": 5.3859,
+      "eval_samples_per_second": 1856.708,
+      "eval_steps_per_second": 7.427,
       "step": 31635
     },
     {
       "epoch": 95.02,
-      "grad_norm": 10.128588676452637,
+      "grad_norm": 7.272856712341309,
       "learning_rate": 6.832832832832833e-06,
-      "loss": 1.2753,
+      "loss": 1.2886,
       "step": 31640
     },
     {
       "epoch": 95.05,
-      "grad_norm": 8.842641830444336,
+      "grad_norm": 9.077088356018066,
       "learning_rate": 6.8318318318318324e-06,
-      "loss": 1.0567,
+      "loss": 1.0681,
       "step": 31650
     },
     {
       "epoch": 95.08,
-      "grad_norm": 5.786375522613525,
+      "grad_norm": 6.709503650665283,
       "learning_rate": 6.830830830830831e-06,
-      "loss": 1.1907,
+      "loss": 1.1846,
       "step": 31660
     },
     {
       "epoch": 95.11,
-      "grad_norm": 8.558484077453613,
+      "grad_norm": 6.38785457611084,
       "learning_rate": 6.8298298298298305e-06,
-      "loss": 1.144,
+      "loss": 1.1416,
       "step": 31670
     },
     {
       "epoch": 95.14,
-      "grad_norm": 12.37678050994873,
+      "grad_norm": 13.26954460144043,
       "learning_rate": 6.82882882882883e-06,
-      "loss": 1.1532,
+      "loss": 1.148,
       "step": 31680
     },
     {
       "epoch": 95.17,
-      "grad_norm": 5.2238593101501465,
+      "grad_norm": 5.9807329177856445,
       "learning_rate": 6.8278278278278285e-06,
-      "loss": 1.1822,
+      "loss": 1.1724,
       "step": 31690
     },
     {
       "epoch": 95.2,
-      "grad_norm": 10.18282413482666,
+      "grad_norm": 6.428393840789795,
       "learning_rate": 6.826826826826828e-06,
-      "loss": 1.1635,
+      "loss": 1.1694,
       "step": 31700
     },
     {
       "epoch": 95.23,
-      "grad_norm": 5.345583915710449,
+      "grad_norm": 6.350419998168945,
       "learning_rate": 6.825825825825826e-06,
-      "loss": 1.2096,
+      "loss": 1.2017,
       "step": 31710
     },
     {
       "epoch": 95.26,
-      "grad_norm": 6.822815418243408,
+      "grad_norm": 8.665637969970703,
       "learning_rate": 6.824824824824825e-06,
-      "loss": 1.163,
+      "loss": 1.1592,
       "step": 31720
     },
     {
       "epoch": 95.29,
-      "grad_norm": 8.481874465942383,
+      "grad_norm": 8.355578422546387,
       "learning_rate": 6.823823823823824e-06,
-      "loss": 1.1709,
+      "loss": 1.1862,
       "step": 31730
     },
     {
       "epoch": 95.32,
-      "grad_norm": 7.286296367645264,
+      "grad_norm": 6.335495948791504,
       "learning_rate": 6.822822822822823e-06,
-      "loss": 1.1037,
+      "loss": 1.1074,
       "step": 31740
     },
     {
       "epoch": 95.35,
-      "grad_norm": 9.371835708618164,
+      "grad_norm": 8.478715896606445,
       "learning_rate": 6.821821821821823e-06,
-      "loss": 1.0902,
+      "loss": 1.0888,
       "step": 31750
     },
     {
       "epoch": 95.38,
-      "grad_norm": 7.229308605194092,
+      "grad_norm": 8.672706604003906,
       "learning_rate": 6.820820820820821e-06,
-      "loss": 1.1463,
+      "loss": 1.1384,
       "step": 31760
     },
     {
       "epoch": 95.41,
-      "grad_norm": 7.176601409912109,
+      "grad_norm": 7.8936991691589355,
       "learning_rate": 6.819819819819821e-06,
-      "loss": 1.1025,
+      "loss": 1.1049,
       "step": 31770
     },
     {
       "epoch": 95.44,
-      "grad_norm": 4.95146369934082,
+      "grad_norm": 6.005762100219727,
       "learning_rate": 6.81881881881882e-06,
-      "loss": 1.1855,
+      "loss": 1.1768,
       "step": 31780
     },
     {
       "epoch": 95.47,
-      "grad_norm": 7.164148807525635,
+      "grad_norm": 6.96237850189209,
       "learning_rate": 6.817817817817819e-06,
-      "loss": 1.1511,
+      "loss": 1.1521,
       "step": 31790
     },
     {
       "epoch": 95.5,
-      "grad_norm": 7.6947407722473145,
+      "grad_norm": 5.8872222900390625,
       "learning_rate": 6.816816816816817e-06,
-      "loss": 1.1096,
+      "loss": 1.1139,
       "step": 31800
     },
     {
       "epoch": 95.53,
-      "grad_norm": 6.526983737945557,
+      "grad_norm": 6.428137302398682,
       "learning_rate": 6.815815815815816e-06,
-      "loss": 1.1515,
+      "loss": 1.1442,
       "step": 31810
     },
     {
       "epoch": 95.56,
-      "grad_norm": 7.145667552947998,
+      "grad_norm": 6.806446075439453,
       "learning_rate": 6.814814814814815e-06,
-      "loss": 1.1193,
+      "loss": 1.1183,
       "step": 31820
     },
     {
       "epoch": 95.59,
-      "grad_norm": 6.477047443389893,
+      "grad_norm": 7.085919380187988,
       "learning_rate": 6.813813813813815e-06,
-      "loss": 1.1589,
+      "loss": 1.1492,
       "step": 31830
     },
     {
       "epoch": 95.62,
-      "grad_norm": 5.943609237670898,
+      "grad_norm": 6.019573211669922,
       "learning_rate": 6.812812812812813e-06,
-      "loss": 1.1833,
+      "loss": 1.1774,
       "step": 31840
     },
     {
       "epoch": 95.65,
-      "grad_norm": 6.455084800720215,
+      "grad_norm": 5.770283222198486,
       "learning_rate": 6.811811811811813e-06,
-      "loss": 1.1569,
+      "loss": 1.1511,
       "step": 31850
     },
     {
       "epoch": 95.68,
-      "grad_norm": 7.38783597946167,
+      "grad_norm": 6.327663421630859,
       "learning_rate": 6.810810810810811e-06,
-      "loss": 1.1671,
+      "loss": 1.1645,
       "step": 31860
     },
     {
       "epoch": 95.71,
-      "grad_norm": 9.341548919677734,
+      "grad_norm": 6.966238975524902,
       "learning_rate": 6.809809809809811e-06,
-      "loss": 1.1846,
+      "loss": 1.184,
       "step": 31870
     },
     {
       "epoch": 95.74,
-      "grad_norm": 6.864021301269531,
+      "grad_norm": 6.810709476470947,
       "learning_rate": 6.808808808808809e-06,
-      "loss": 1.0866,
+      "loss": 1.0756,
       "step": 31880
     },
     {
       "epoch": 95.77,
-      "grad_norm": 7.400967597961426,
+      "grad_norm": 6.2583160400390625,
       "learning_rate": 6.807807807807808e-06,
-      "loss": 1.098,
+      "loss": 1.1027,
       "step": 31890
     },
     {
       "epoch": 95.8,
-      "grad_norm": 12.552374839782715,
+      "grad_norm": 10.668642044067383,
       "learning_rate": 6.8068068068068075e-06,
-      "loss": 1.2058,
+      "loss": 1.2082,
       "step": 31900
     },
     {
       "epoch": 95.83,
-      "grad_norm": 7.5848822593688965,
+      "grad_norm": 8.771798133850098,
       "learning_rate": 6.805805805805806e-06,
-      "loss": 1.1363,
+      "loss": 1.1343,
       "step": 31910
     },
     {
       "epoch": 95.86,
-      "grad_norm": 6.798425674438477,
+      "grad_norm": 7.004152774810791,
       "learning_rate": 6.8048048048048055e-06,
-      "loss": 1.1979,
+      "loss": 1.2002,
       "step": 31920
     },
     {
       "epoch": 95.89,
-      "grad_norm": 6.458479881286621,
+      "grad_norm": 5.05441427230835,
       "learning_rate": 6.803803803803805e-06,
-      "loss": 1.1566,
+      "loss": 1.164,
       "step": 31930
     },
     {
       "epoch": 95.92,
-      "grad_norm": 6.695705413818359,
+      "grad_norm": 8.237137794494629,
       "learning_rate": 6.8028028028028035e-06,
-      "loss": 1.1528,
+      "loss": 1.1458,
       "step": 31940
     },
     {
       "epoch": 95.95,
-      "grad_norm": 5.979049205780029,
+      "grad_norm": 6.4237189292907715,
       "learning_rate": 6.801801801801803e-06,
-      "loss": 1.1404,
+      "loss": 1.1406,
       "step": 31950
     },
     {
       "epoch": 95.98,
-      "grad_norm": 10.385017395019531,
+      "grad_norm": 11.54968547821045,
       "learning_rate": 6.800800800800801e-06,
-      "loss": 1.1619,
+      "loss": 1.1646,
       "step": 31960
     },
     {
       "epoch": 96.0,
-      "eval_accuracy": 0.7963,
-      "eval_loss": 0.6538266539573669,
-      "eval_runtime": 5.1964,
-      "eval_samples_per_second": 1924.391,
-      "eval_steps_per_second": 7.698,
+      "eval_accuracy": 0.8006,
+      "eval_loss": 0.6527626514434814,
+      "eval_runtime": 5.398,
+      "eval_samples_per_second": 1852.534,
+      "eval_steps_per_second": 7.41,
       "step": 31968
     },
     {
       "epoch": 96.01,
-      "grad_norm": 4.7939066886901855,
+      "grad_norm": 6.6915106773376465,
       "learning_rate": 6.7997997997998e-06,
-      "loss": 1.4249,
+      "loss": 1.4194,
       "step": 31970
     },
     {
       "epoch": 96.04,
-      "grad_norm": 7.545568466186523,
+      "grad_norm": 7.841568470001221,
       "learning_rate": 6.798798798798799e-06,
-      "loss": 1.1133,
+      "loss": 1.1098,
       "step": 31980
     },
     {
       "epoch": 96.07,
-      "grad_norm": 6.344456672668457,
+      "grad_norm": 9.035682678222656,
       "learning_rate": 6.797797797797798e-06,
-      "loss": 1.1766,
+      "loss": 1.1717,
       "step": 31990
     },
     {
       "epoch": 96.1,
-      "grad_norm": 4.670658111572266,
+      "grad_norm": 5.06916618347168,
       "learning_rate": 6.796796796796798e-06,
-      "loss": 1.1232,
+      "loss": 1.1092,
       "step": 32000
     },
     {
       "epoch": 96.13,
-      "grad_norm": 5.448360443115234,
+      "grad_norm": 5.53375244140625,
       "learning_rate": 6.795795795795796e-06,
-      "loss": 1.0631,
+      "loss": 1.048,
       "step": 32010
     },
     {
       "epoch": 96.16,
-      "grad_norm": 6.456271171569824,
+      "grad_norm": 6.643615245819092,
       "learning_rate": 6.794794794794796e-06,
-      "loss": 1.1178,
+      "loss": 1.126,
       "step": 32020
     },
     {
       "epoch": 96.19,
-      "grad_norm": 5.1768035888671875,
+      "grad_norm": 7.578037738800049,
       "learning_rate": 6.793793793793795e-06,
-      "loss": 1.2002,
+      "loss": 1.2055,
       "step": 32030
     },
     {
       "epoch": 96.22,
-      "grad_norm": 9.625662803649902,
+      "grad_norm": 6.794678688049316,
       "learning_rate": 6.792792792792793e-06,
-      "loss": 1.1205,
+      "loss": 1.1271,
       "step": 32040
     },
     {
       "epoch": 96.25,
-      "grad_norm": 5.286436080932617,
+      "grad_norm": 6.925400257110596,
       "learning_rate": 6.791791791791792e-06,
-      "loss": 1.0861,
+      "loss": 1.0882,
       "step": 32050
     },
     {
       "epoch": 96.28,
-      "grad_norm": 5.546974182128906,
+      "grad_norm": 6.998789310455322,
       "learning_rate": 6.790790790790791e-06,
-      "loss": 1.1035,
+      "loss": 1.1006,
       "step": 32060
     },
     {
       "epoch": 96.31,
-      "grad_norm": 6.816957950592041,
+      "grad_norm": 7.718015670776367,
       "learning_rate": 6.78978978978979e-06,
-      "loss": 1.174,
+      "loss": 1.1727,
       "step": 32070
     },
     {
       "epoch": 96.34,
-      "grad_norm": 9.290029525756836,
+      "grad_norm": 9.392635345458984,
       "learning_rate": 6.788788788788789e-06,
-      "loss": 1.2079,
+      "loss": 1.206,
       "step": 32080
     },
     {
       "epoch": 96.37,
-      "grad_norm": 9.267040252685547,
+      "grad_norm": 12.536876678466797,
       "learning_rate": 6.787787787787788e-06,
-      "loss": 1.115,
+      "loss": 1.1069,
       "step": 32090
     },
     {
       "epoch": 96.4,
-      "grad_norm": 6.354816436767578,
+      "grad_norm": 5.128821849822998,
       "learning_rate": 6.786786786786788e-06,
-      "loss": 1.1141,
+      "loss": 1.1163,
       "step": 32100
     },
     {
       "epoch": 96.43,
-      "grad_norm": 4.383440971374512,
+      "grad_norm": 4.7475666999816895,
       "learning_rate": 6.7857857857857864e-06,
-      "loss": 1.132,
+      "loss": 1.1342,
       "step": 32110
     },
     {
       "epoch": 96.46,
-      "grad_norm": 5.2222819328308105,
+      "grad_norm": 4.8382673263549805,
       "learning_rate": 6.784784784784786e-06,
-      "loss": 1.1385,
+      "loss": 1.1438,
       "step": 32120
     },
     {
       "epoch": 96.49,
-      "grad_norm": 7.875004291534424,
+      "grad_norm": 8.557880401611328,
       "learning_rate": 6.783783783783784e-06,
-      "loss": 1.1308,
+      "loss": 1.1309,
       "step": 32130
     },
     {
       "epoch": 96.52,
-      "grad_norm": 7.521064281463623,
+      "grad_norm": 7.130814075469971,
       "learning_rate": 6.782782782782783e-06,
-      "loss": 1.1443,
+      "loss": 1.1275,
       "step": 32140
     },
     {
       "epoch": 96.55,
-      "grad_norm": 8.919720649719238,
+      "grad_norm": 6.203309059143066,
       "learning_rate": 6.7817817817817825e-06,
-      "loss": 1.1281,
+      "loss": 1.1096,
       "step": 32150
     },
     {
       "epoch": 96.58,
-      "grad_norm": 5.608020782470703,
+      "grad_norm": 5.1874542236328125,
       "learning_rate": 6.780780780780781e-06,
-      "loss": 1.1158,
+      "loss": 1.1143,
       "step": 32160
     },
     {
       "epoch": 96.61,
-      "grad_norm": 8.098526954650879,
+      "grad_norm": 9.584009170532227,
       "learning_rate": 6.7797797797797805e-06,
-      "loss": 1.2099,
+      "loss": 1.2101,
       "step": 32170
     },
     {
       "epoch": 96.64,
-      "grad_norm": 7.134501934051514,
+      "grad_norm": 7.691011428833008,
       "learning_rate": 6.778778778778779e-06,
-      "loss": 1.1848,
+      "loss": 1.1917,
       "step": 32180
     },
     {
       "epoch": 96.67,
-      "grad_norm": 6.4542317390441895,
+      "grad_norm": 6.3743109703063965,
       "learning_rate": 6.777777777777779e-06,
-      "loss": 1.1214,
+      "loss": 1.1172,
       "step": 32190
     },
     {
       "epoch": 96.7,
-      "grad_norm": 6.946895122528076,
+      "grad_norm": 6.434085845947266,
       "learning_rate": 6.776776776776778e-06,
-      "loss": 1.2277,
+      "loss": 1.2111,
       "step": 32200
     },
     {
       "epoch": 96.73,
-      "grad_norm": 6.848337173461914,
+      "grad_norm": 5.961461544036865,
       "learning_rate": 6.775775775775776e-06,
-      "loss": 1.1544,
+      "loss": 1.1657,
       "step": 32210
     },
     {
       "epoch": 96.76,
-      "grad_norm": 8.165641784667969,
+      "grad_norm": 7.034841060638428,
       "learning_rate": 6.774774774774775e-06,
-      "loss": 1.1282,
+      "loss": 1.1166,
       "step": 32220
     },
     {
       "epoch": 96.79,
-      "grad_norm": 6.605316162109375,
+      "grad_norm": 7.542209625244141,
       "learning_rate": 6.773773773773774e-06,
-      "loss": 1.1652,
+      "loss": 1.1544,
       "step": 32230
     },
     {
       "epoch": 96.82,
-      "grad_norm": 6.402888774871826,
+      "grad_norm": 4.793143272399902,
       "learning_rate": 6.772772772772773e-06,
-      "loss": 1.1382,
+      "loss": 1.1472,
       "step": 32240
     },
     {
       "epoch": 96.85,
-      "grad_norm": 5.632260799407959,
+      "grad_norm": 6.976840496063232,
       "learning_rate": 6.771771771771773e-06,
-      "loss": 1.1517,
+      "loss": 1.146,
       "step": 32250
     },
     {
       "epoch": 96.88,
-      "grad_norm": 6.344549655914307,
+      "grad_norm": 6.678967475891113,
       "learning_rate": 6.770770770770771e-06,
-      "loss": 1.1387,
+      "loss": 1.1461,
       "step": 32260
     },
     {
       "epoch": 96.91,
-      "grad_norm": 6.961042404174805,
+      "grad_norm": 7.709018230438232,
       "learning_rate": 6.769769769769771e-06,
-      "loss": 1.0803,
+      "loss": 1.0843,
       "step": 32270
     },
     {
       "epoch": 96.94,
-      "grad_norm": 7.094814300537109,
+      "grad_norm": 7.341402053833008,
       "learning_rate": 6.76876876876877e-06,
-      "loss": 1.1236,
+      "loss": 1.1273,
       "step": 32280
     },
     {
       "epoch": 96.97,
-      "grad_norm": 6.4993672370910645,
+      "grad_norm": 5.737684726715088,
       "learning_rate": 6.767767767767768e-06,
-      "loss": 1.1474,
+      "loss": 1.1508,
       "step": 32290
     },
     {
       "epoch": 97.0,
-      "grad_norm": 9.764833450317383,
+      "grad_norm": 7.511504650115967,
       "learning_rate": 6.7667667667667665e-06,
-      "loss": 1.1547,
+      "loss": 1.156,
       "step": 32300
     },
     {
       "epoch": 97.0,
-      "eval_accuracy": 0.7992,
-      "eval_loss": 0.6510264873504639,
-      "eval_runtime": 5.414,
-      "eval_samples_per_second": 1847.066,
-      "eval_steps_per_second": 7.388,
+      "eval_accuracy": 0.7984,
+      "eval_loss": 0.6496440172195435,
+      "eval_runtime": 5.5004,
+      "eval_samples_per_second": 1818.059,
+      "eval_steps_per_second": 7.272,
       "step": 32301
     },
     {
       "epoch": 97.03,
-      "grad_norm": 5.562166690826416,
+      "grad_norm": 5.720500469207764,
       "learning_rate": 6.765765765765766e-06,
-      "loss": 1.2436,
+      "loss": 1.2737,
       "step": 32310
     },
     {
       "epoch": 97.06,
-      "grad_norm": 6.481811046600342,
+      "grad_norm": 8.005240440368652,
       "learning_rate": 6.764764764764765e-06,
-      "loss": 1.1273,
+      "loss": 1.1328,
       "step": 32320
     },
     {
       "epoch": 97.09,
-      "grad_norm": 5.172922611236572,
+      "grad_norm": 6.128732204437256,
       "learning_rate": 6.763763763763764e-06,
-      "loss": 1.1463,
+      "loss": 1.1515,
       "step": 32330
     },
     {
       "epoch": 97.12,
-      "grad_norm": 8.007759094238281,
+      "grad_norm": 6.843305587768555,
       "learning_rate": 6.7627627627627634e-06,
-      "loss": 1.1476,
+      "loss": 1.1397,
       "step": 32340
     },
     {
       "epoch": 97.15,
-      "grad_norm": 6.045263767242432,
+      "grad_norm": 9.21926498413086,
       "learning_rate": 6.761761761761763e-06,
-      "loss": 1.035,
+      "loss": 1.031,
       "step": 32350
     },
     {
       "epoch": 97.18,
-      "grad_norm": 8.67335033416748,
+      "grad_norm": 8.382862091064453,
       "learning_rate": 6.7607607607607615e-06,
-      "loss": 1.1851,
+      "loss": 1.1869,
       "step": 32360
     },
     {
       "epoch": 97.21,
-      "grad_norm": 6.988211154937744,
+      "grad_norm": 12.982583999633789,
       "learning_rate": 6.759759759759761e-06,
-      "loss": 1.1596,
+      "loss": 1.1564,
       "step": 32370
     },
     {
       "epoch": 97.24,
-      "grad_norm": 6.3553290367126465,
+      "grad_norm": 5.354623317718506,
       "learning_rate": 6.758758758758759e-06,
-      "loss": 1.1511,
+      "loss": 1.1443,
       "step": 32380
     },
     {
       "epoch": 97.27,
-      "grad_norm": 15.991927146911621,
+      "grad_norm": 12.456888198852539,
       "learning_rate": 6.757757757757758e-06,
-      "loss": 1.1795,
+      "loss": 1.1765,
       "step": 32390
     },
     {
       "epoch": 97.3,
-      "grad_norm": 5.016067028045654,
+      "grad_norm": 6.077402114868164,
       "learning_rate": 6.7567567567567575e-06,
-      "loss": 1.1301,
+      "loss": 1.1208,
       "step": 32400
     },
     {
       "epoch": 97.33,
-      "grad_norm": 6.1927289962768555,
+      "grad_norm": 5.745611667633057,
       "learning_rate": 6.755755755755756e-06,
-      "loss": 1.1521,
+      "loss": 1.1466,
       "step": 32410
     },
     {
       "epoch": 97.36,
-      "grad_norm": 5.217617988586426,
+      "grad_norm": 7.986681938171387,
       "learning_rate": 6.7547547547547556e-06,
-      "loss": 1.1331,
+      "loss": 1.1349,
       "step": 32420
     },
     {
       "epoch": 97.39,
-      "grad_norm": 5.906464576721191,
+      "grad_norm": 8.624505996704102,
       "learning_rate": 6.753753753753754e-06,
-      "loss": 1.1405,
+      "loss": 1.1435,
       "step": 32430
     },
     {
       "epoch": 97.42,
-      "grad_norm": 8.623823165893555,
+      "grad_norm": 9.733160018920898,
       "learning_rate": 6.752752752752754e-06,
-      "loss": 1.1828,
+      "loss": 1.1721,
       "step": 32440
     },
     {
       "epoch": 97.45,
-      "grad_norm": 8.087573051452637,
+      "grad_norm": 7.47466516494751,
       "learning_rate": 6.751751751751753e-06,
-      "loss": 1.1059,
+      "loss": 1.0993,
       "step": 32450
     },
     {
       "epoch": 97.48,
-      "grad_norm": 6.139193058013916,
+      "grad_norm": 6.87614631652832,
       "learning_rate": 6.750750750750751e-06,
-      "loss": 1.1095,
+      "loss": 1.1071,
       "step": 32460
     },
     {
       "epoch": 97.51,
-      "grad_norm": 7.098288059234619,
+      "grad_norm": 4.383477687835693,
       "learning_rate": 6.74974974974975e-06,
-      "loss": 1.1557,
+      "loss": 1.1498,
       "step": 32470
     },
     {
       "epoch": 97.54,
-      "grad_norm": 7.405278205871582,
+      "grad_norm": 6.562463283538818,
       "learning_rate": 6.748748748748749e-06,
-      "loss": 1.1148,
+      "loss": 1.1021,
       "step": 32480
     },
     {
       "epoch": 97.57,
-      "grad_norm": 5.493907928466797,
+      "grad_norm": 6.157007694244385,
       "learning_rate": 6.747747747747748e-06,
-      "loss": 1.1211,
+      "loss": 1.115,
       "step": 32490
     },
     {
       "epoch": 97.6,
-      "grad_norm": 5.159914970397949,
+      "grad_norm": 7.595696926116943,
       "learning_rate": 6.746746746746748e-06,
-      "loss": 1.2009,
+      "loss": 1.2011,
       "step": 32500
     },
     {
       "epoch": 97.63,
-      "grad_norm": 5.825040340423584,
+      "grad_norm": 12.609193801879883,
       "learning_rate": 6.745745745745746e-06,
-      "loss": 1.1114,
+      "loss": 1.1148,
       "step": 32510
     },
     {
       "epoch": 97.66,
-      "grad_norm": 5.6305670738220215,
+      "grad_norm": 6.163232803344727,
       "learning_rate": 6.744744744744746e-06,
-      "loss": 1.1029,
+      "loss": 1.1049,
       "step": 32520
     },
     {
       "epoch": 97.69,
-      "grad_norm": 5.1703691482543945,
+      "grad_norm": 6.423469066619873,
       "learning_rate": 6.743743743743744e-06,
-      "loss": 1.1525,
+      "loss": 1.1632,
       "step": 32530
     },
     {
       "epoch": 97.72,
-      "grad_norm": 7.31355619430542,
+      "grad_norm": 8.800883293151855,
       "learning_rate": 6.742742742742743e-06,
-      "loss": 1.1202,
+      "loss": 1.123,
       "step": 32540
     },
     {
       "epoch": 97.75,
-      "grad_norm": 6.144682884216309,
+      "grad_norm": 6.0406060218811035,
       "learning_rate": 6.7417417417417415e-06,
-      "loss": 1.0892,
+      "loss": 1.0955,
       "step": 32550
     },
     {
       "epoch": 97.78,
-      "grad_norm": 5.043186187744141,
+      "grad_norm": 5.231060981750488,
       "learning_rate": 6.740740740740741e-06,
-      "loss": 1.1755,
+      "loss": 1.1735,
       "step": 32560
     },
     {
       "epoch": 97.81,
-      "grad_norm": 7.292966842651367,
+      "grad_norm": 6.573897361755371,
       "learning_rate": 6.7397397397397404e-06,
-      "loss": 1.1595,
+      "loss": 1.1522,
       "step": 32570
     },
     {
       "epoch": 97.84,
-      "grad_norm": 8.75687026977539,
+      "grad_norm": 7.372313499450684,
       "learning_rate": 6.738738738738739e-06,
-      "loss": 1.067,
+      "loss": 1.0678,
       "step": 32580
     },
     {
       "epoch": 97.87,
-      "grad_norm": 6.014616012573242,
+      "grad_norm": 6.592780113220215,
       "learning_rate": 6.7377377377377385e-06,
-      "loss": 1.0817,
+      "loss": 1.0789,
       "step": 32590
     },
     {
       "epoch": 97.9,
-      "grad_norm": 9.430804252624512,
+      "grad_norm": 8.183270454406738,
       "learning_rate": 6.736736736736738e-06,
-      "loss": 1.1603,
+      "loss": 1.1533,
       "step": 32600
     },
     {
       "epoch": 97.93,
-      "grad_norm": 8.759592056274414,
+      "grad_norm": 6.089609622955322,
       "learning_rate": 6.7357357357357365e-06,
-      "loss": 1.2017,
+      "loss": 1.1901,
       "step": 32610
     },
     {
       "epoch": 97.96,
-      "grad_norm": 6.204437255859375,
+      "grad_norm": 8.173481941223145,
       "learning_rate": 6.734734734734736e-06,
-      "loss": 1.1738,
+      "loss": 1.1672,
       "step": 32620
     },
     {
       "epoch": 97.99,
-      "grad_norm": 7.609854698181152,
+      "grad_norm": 9.975370407104492,
       "learning_rate": 6.733733733733734e-06,
-      "loss": 1.198,
+      "loss": 1.2019,
       "step": 32630
     },
     {
       "epoch": 98.0,
-      "eval_accuracy": 0.8014,
-      "eval_loss": 0.649456262588501,
-      "eval_runtime": 5.5198,
-      "eval_samples_per_second": 1811.668,
-      "eval_steps_per_second": 7.247,
+      "eval_accuracy": 0.7988,
+      "eval_loss": 0.649478554725647,
+      "eval_runtime": 5.3527,
+      "eval_samples_per_second": 1868.202,
+      "eval_steps_per_second": 7.473,
       "step": 32634
     },
     {
       "epoch": 98.02,
-      "grad_norm": 11.618558883666992,
+      "grad_norm": 9.513919830322266,
       "learning_rate": 6.732732732732733e-06,
-      "loss": 1.2818,
+      "loss": 1.3054,
       "step": 32640
     },
     {
       "epoch": 98.05,
-      "grad_norm": 9.74620532989502,
+      "grad_norm": 8.098094940185547,
       "learning_rate": 6.731731731731732e-06,
-      "loss": 1.1912,
+      "loss": 1.1872,
       "step": 32650
     },
     {
       "epoch": 98.08,
-      "grad_norm": 8.102526664733887,
+      "grad_norm": 7.758856296539307,
       "learning_rate": 6.730730730730731e-06,
-      "loss": 1.1293,
+      "loss": 1.1174,
       "step": 32660
     },
     {
       "epoch": 98.11,
-      "grad_norm": 6.860073566436768,
+      "grad_norm": 7.286405563354492,
       "learning_rate": 6.729729729729731e-06,
-      "loss": 1.1139,
+      "loss": 1.1055,
       "step": 32670
     },
     {
       "epoch": 98.14,
-      "grad_norm": 7.7736406326293945,
+      "grad_norm": 9.895270347595215,
       "learning_rate": 6.728728728728729e-06,
-      "loss": 1.2189,
+      "loss": 1.2204,
       "step": 32680
     },
     {
       "epoch": 98.17,
-      "grad_norm": 6.615218162536621,
+      "grad_norm": 6.230807781219482,
       "learning_rate": 6.727727727727729e-06,
-      "loss": 1.2218,
+      "loss": 1.216,
       "step": 32690
     },
     {
       "epoch": 98.2,
-      "grad_norm": 6.261587142944336,
+      "grad_norm": 6.296480178833008,
       "learning_rate": 6.726726726726728e-06,
-      "loss": 1.1589,
+      "loss": 1.1559,
       "step": 32700
     },
     {
       "epoch": 98.23,
-      "grad_norm": 4.544332027435303,
+      "grad_norm": 7.422635555267334,
       "learning_rate": 6.725725725725726e-06,
-      "loss": 1.1344,
+      "loss": 1.1422,
       "step": 32710
     },
     {
       "epoch": 98.26,
-      "grad_norm": 8.705597877502441,
+      "grad_norm": 6.041656970977783,
       "learning_rate": 6.724724724724725e-06,
-      "loss": 1.1761,
+      "loss": 1.1672,
       "step": 32720
     },
     {
       "epoch": 98.29,
-      "grad_norm": 6.16464900970459,
+      "grad_norm": 7.040347576141357,
       "learning_rate": 6.723723723723724e-06,
-      "loss": 1.1161,
+      "loss": 1.1109,
       "step": 32730
     },
     {
       "epoch": 98.32,
-      "grad_norm": 6.940003871917725,
+      "grad_norm": 8.151779174804688,
       "learning_rate": 6.722722722722723e-06,
-      "loss": 1.176,
+      "loss": 1.1685,
       "step": 32740
     },
     {
       "epoch": 98.35,
-      "grad_norm": 7.379177093505859,
+      "grad_norm": 7.05086612701416,
       "learning_rate": 6.721721721721722e-06,
-      "loss": 1.1183,
+      "loss": 1.1184,
       "step": 32750
     },
     {
       "epoch": 98.38,
-      "grad_norm": 7.469712257385254,
+      "grad_norm": 10.170239448547363,
       "learning_rate": 6.720720720720721e-06,
-      "loss": 1.1621,
+      "loss": 1.1532,
       "step": 32760
     },
     {
       "epoch": 98.41,
-      "grad_norm": 5.6154093742370605,
+      "grad_norm": 7.045928001403809,
       "learning_rate": 6.719719719719721e-06,
-      "loss": 1.0895,
+      "loss": 1.0898,
       "step": 32770
     },
     {
       "epoch": 98.44,
-      "grad_norm": 7.352874279022217,
+      "grad_norm": 10.583979606628418,
       "learning_rate": 6.718718718718719e-06,
-      "loss": 1.0922,
+      "loss": 1.1028,
       "step": 32780
     },
     {
       "epoch": 98.47,
-      "grad_norm": 7.839776992797852,
+      "grad_norm": 10.50316047668457,
       "learning_rate": 6.717717717717718e-06,
-      "loss": 1.1583,
+      "loss": 1.1612,
       "step": 32790
     },
     {
       "epoch": 98.5,
-      "grad_norm": 5.158909797668457,
+      "grad_norm": 5.345668315887451,
       "learning_rate": 6.716716716716717e-06,
-      "loss": 1.1628,
+      "loss": 1.1674,
       "step": 32800
     },
     {
       "epoch": 98.53,
-      "grad_norm": 5.447737693786621,
+      "grad_norm": 6.920622825622559,
       "learning_rate": 6.715715715715716e-06,
-      "loss": 1.1652,
+      "loss": 1.1704,
       "step": 32810
     },
     {
       "epoch": 98.56,
-      "grad_norm": 7.571290016174316,
+      "grad_norm": 9.868077278137207,
       "learning_rate": 6.7147147147147155e-06,
-      "loss": 1.1126,
+      "loss": 1.1165,
       "step": 32820
     },
     {
       "epoch": 98.59,
-      "grad_norm": 8.398306846618652,
+      "grad_norm": 9.412837028503418,
       "learning_rate": 6.713713713713714e-06,
-      "loss": 1.1041,
+      "loss": 1.1076,
       "step": 32830
     },
     {
       "epoch": 98.62,
-      "grad_norm": 6.3214287757873535,
+      "grad_norm": 7.399446487426758,
       "learning_rate": 6.7127127127127135e-06,
-      "loss": 1.1518,
+      "loss": 1.1519,
       "step": 32840
     },
     {
       "epoch": 98.65,
-      "grad_norm": 7.169186115264893,
+      "grad_norm": 6.401692867279053,
       "learning_rate": 6.711711711711713e-06,
-      "loss": 1.0742,
+      "loss": 1.0746,
       "step": 32850
     },
     {
       "epoch": 98.68,
-      "grad_norm": 6.837580680847168,
+      "grad_norm": 7.928347110748291,
       "learning_rate": 6.7107107107107115e-06,
-      "loss": 1.2002,
+      "loss": 1.196,
       "step": 32860
     },
     {
       "epoch": 98.71,
-      "grad_norm": 7.794702053070068,
+      "grad_norm": 7.4038286209106445,
       "learning_rate": 6.709709709709711e-06,
-      "loss": 1.1499,
+      "loss": 1.1478,
       "step": 32870
     },
     {
       "epoch": 98.74,
-      "grad_norm": 8.49131965637207,
+      "grad_norm": 7.877062797546387,
       "learning_rate": 6.708708708708709e-06,
-      "loss": 1.1332,
+      "loss": 1.1262,
       "step": 32880
     },
     {
       "epoch": 98.77,
-      "grad_norm": 10.657488822937012,
+      "grad_norm": 7.267848491668701,
       "learning_rate": 6.707707707707708e-06,
-      "loss": 1.1849,
+      "loss": 1.1807,
       "step": 32890
     },
     {
       "epoch": 98.8,
-      "grad_norm": 9.375699996948242,
+      "grad_norm": 6.269059658050537,
       "learning_rate": 6.706706706706707e-06,
-      "loss": 1.1655,
+      "loss": 1.1645,
       "step": 32900
     },
     {
       "epoch": 98.83,
-      "grad_norm": 5.8153605461120605,
+      "grad_norm": 4.628307819366455,
       "learning_rate": 6.705705705705706e-06,
-      "loss": 1.1736,
+      "loss": 1.1784,
       "step": 32910
     },
     {
       "epoch": 98.86,
-      "grad_norm": 7.078489780426025,
+      "grad_norm": 7.606204509735107,
       "learning_rate": 6.704704704704706e-06,
-      "loss": 1.184,
+      "loss": 1.1803,
       "step": 32920
     },
     {
       "epoch": 98.89,
-      "grad_norm": 11.361342430114746,
+      "grad_norm": 7.606089115142822,
       "learning_rate": 6.703703703703704e-06,
-      "loss": 1.1411,
+      "loss": 1.1374,
       "step": 32930
     },
     {
       "epoch": 98.92,
-      "grad_norm": 8.924610137939453,
+      "grad_norm": 7.109785556793213,
       "learning_rate": 6.702702702702704e-06,
-      "loss": 1.1787,
+      "loss": 1.174,
       "step": 32940
     },
     {
       "epoch": 98.95,
-      "grad_norm": 6.6474690437316895,
+      "grad_norm": 6.729063034057617,
       "learning_rate": 6.701701701701703e-06,
-      "loss": 1.1061,
+      "loss": 1.1029,
       "step": 32950
     },
     {
       "epoch": 98.98,
-      "grad_norm": 6.122426509857178,
+      "grad_norm": 5.650670051574707,
       "learning_rate": 6.700700700700701e-06,
-      "loss": 1.0816,
+      "loss": 1.0851,
       "step": 32960
     },
     {
       "epoch": 99.0,
-      "eval_accuracy": 0.8008,
-      "eval_loss": 0.6501058340072632,
-      "eval_runtime": 5.1677,
-      "eval_samples_per_second": 1935.079,
-      "eval_steps_per_second": 7.74,
+      "eval_accuracy": 0.7985,
+      "eval_loss": 0.6518943309783936,
+      "eval_runtime": 5.123,
+      "eval_samples_per_second": 1951.999,
+      "eval_steps_per_second": 7.808,
       "step": 32967
     },
     {
       "epoch": 99.01,
-      "grad_norm": 5.883338928222656,
+      "grad_norm": 9.43350601196289,
       "learning_rate": 6.6996996996997e-06,
-      "loss": 1.3052,
+      "loss": 1.3143,
       "step": 32970
     },
     {
       "epoch": 99.04,
-      "grad_norm": 7.179941654205322,
+      "grad_norm": 8.9138822555542,
       "learning_rate": 6.698698698698699e-06,
-      "loss": 1.1833,
+      "loss": 1.1824,
       "step": 32980
     },
     {
       "epoch": 99.07,
-      "grad_norm": 7.171102523803711,
+      "grad_norm": 8.166499137878418,
       "learning_rate": 6.697697697697698e-06,
-      "loss": 1.0992,
+      "loss": 1.0889,
       "step": 32990
     },
     {
       "epoch": 99.1,
-      "grad_norm": 8.57737922668457,
+      "grad_norm": 7.1333465576171875,
       "learning_rate": 6.696696696696697e-06,
-      "loss": 1.0436,
+      "loss": 1.0491,
       "step": 33000
     },
     {
       "epoch": 99.13,
-      "grad_norm": 6.8690505027771,
+      "grad_norm": 7.249352931976318,
       "learning_rate": 6.695695695695696e-06,
-      "loss": 1.1664,
+      "loss": 1.1747,
       "step": 33010
     },
     {
       "epoch": 99.16,
-      "grad_norm": 5.285749912261963,
+      "grad_norm": 5.241756916046143,
       "learning_rate": 6.694694694694696e-06,
-      "loss": 1.1369,
+      "loss": 1.1293,
       "step": 33020
     },
     {
       "epoch": 99.19,
-      "grad_norm": 7.738404750823975,
+      "grad_norm": 6.801769256591797,
       "learning_rate": 6.693693693693694e-06,
-      "loss": 1.1417,
+      "loss": 1.1325,
       "step": 33030
     },
     {
       "epoch": 99.22,
-      "grad_norm": 5.8620476722717285,
+      "grad_norm": 7.038864612579346,
       "learning_rate": 6.692692692692693e-06,
-      "loss": 1.1396,
+      "loss": 1.1352,
       "step": 33040
     },
     {
       "epoch": 99.25,
-      "grad_norm": 5.988830089569092,
+      "grad_norm": 6.4121317863464355,
       "learning_rate": 6.691691691691692e-06,
-      "loss": 1.0907,
+      "loss": 1.0835,
       "step": 33050
     },
     {
       "epoch": 99.28,
-      "grad_norm": 8.482405662536621,
+      "grad_norm": 5.9150919914245605,
       "learning_rate": 6.690690690690691e-06,
-      "loss": 1.1545,
+      "loss": 1.1549,
       "step": 33060
     },
     {
       "epoch": 99.31,
-      "grad_norm": 8.114683151245117,
+      "grad_norm": 8.579070091247559,
       "learning_rate": 6.6896896896896905e-06,
-      "loss": 1.1569,
+      "loss": 1.1502,
       "step": 33070
     },
     {
       "epoch": 99.34,
-      "grad_norm": 7.364828109741211,
+      "grad_norm": 8.354649543762207,
       "learning_rate": 6.688688688688689e-06,
-      "loss": 1.1313,
+      "loss": 1.1323,
       "step": 33080
     },
     {
       "epoch": 99.37,
-      "grad_norm": 6.975665092468262,
+      "grad_norm": 6.824965953826904,
       "learning_rate": 6.6876876876876885e-06,
-      "loss": 1.2389,
+      "loss": 1.2358,
       "step": 33090
     },
     {
       "epoch": 99.4,
-      "grad_norm": 7.736310005187988,
+      "grad_norm": 7.0947675704956055,
       "learning_rate": 6.686686686686687e-06,
-      "loss": 1.1526,
+      "loss": 1.1565,
       "step": 33100
     },
     {
       "epoch": 99.43,
-      "grad_norm": 5.770545482635498,
+      "grad_norm": 8.744779586791992,
       "learning_rate": 6.6856856856856866e-06,
-      "loss": 1.044,
+      "loss": 1.0371,
       "step": 33110
     },
     {
       "epoch": 99.46,
-      "grad_norm": 6.389944076538086,
+      "grad_norm": 7.226615905761719,
       "learning_rate": 6.684684684684684e-06,
-      "loss": 1.1085,
+      "loss": 1.1074,
       "step": 33120
     },
     {
       "epoch": 99.49,
-      "grad_norm": 6.547369003295898,
+      "grad_norm": 5.27736234664917,
       "learning_rate": 6.683683683683684e-06,
-      "loss": 1.1031,
+      "loss": 1.1029,
       "step": 33130
     },
     {
       "epoch": 99.52,
-      "grad_norm": 6.149424076080322,
+      "grad_norm": 5.9466753005981445,
       "learning_rate": 6.682682682682683e-06,
-      "loss": 1.1282,
+      "loss": 1.1346,
       "step": 33140
     },
     {
       "epoch": 99.55,
-      "grad_norm": 6.861825466156006,
+      "grad_norm": 5.4361724853515625,
       "learning_rate": 6.681681681681682e-06,
-      "loss": 1.1479,
+      "loss": 1.1466,
       "step": 33150
     },
     {
       "epoch": 99.58,
-      "grad_norm": 4.737374782562256,
+      "grad_norm": 5.409894943237305,
       "learning_rate": 6.680680680680681e-06,
-      "loss": 1.155,
+      "loss": 1.1651,
       "step": 33160
     },
     {
       "epoch": 99.61,
-      "grad_norm": 10.938896179199219,
+      "grad_norm": 10.329731941223145,
       "learning_rate": 6.679679679679681e-06,
-      "loss": 1.1195,
+      "loss": 1.1215,
       "step": 33170
     },
     {
       "epoch": 99.64,
-      "grad_norm": 5.807580947875977,
+      "grad_norm": 5.243748188018799,
       "learning_rate": 6.678678678678679e-06,
-      "loss": 1.184,
+      "loss": 1.1758,
       "step": 33180
     },
     {
       "epoch": 99.67,
-      "grad_norm": 5.6478447914123535,
+      "grad_norm": 5.7614898681640625,
       "learning_rate": 6.677677677677679e-06,
-      "loss": 1.1097,
+      "loss": 1.1165,
       "step": 33190
     },
     {
       "epoch": 99.7,
-      "grad_norm": 6.450976371765137,
+      "grad_norm": 6.215134143829346,
       "learning_rate": 6.676676676676678e-06,
-      "loss": 1.123,
+      "loss": 1.1231,
       "step": 33200
     },
     {
       "epoch": 99.73,
-      "grad_norm": 7.994123458862305,
+      "grad_norm": 6.491041660308838,
       "learning_rate": 6.675675675675676e-06,
-      "loss": 1.1295,
+      "loss": 1.1174,
       "step": 33210
     },
     {
       "epoch": 99.76,
-      "grad_norm": 5.735809803009033,
+      "grad_norm": 7.006434440612793,
       "learning_rate": 6.6746746746746745e-06,
-      "loss": 1.0861,
+      "loss": 1.0804,
       "step": 33220
     },
     {
       "epoch": 99.79,
-      "grad_norm": 5.50828218460083,
+      "grad_norm": 4.622373104095459,
       "learning_rate": 6.673673673673674e-06,
-      "loss": 1.177,
+      "loss": 1.1763,
       "step": 33230
     },
     {
       "epoch": 99.82,
-      "grad_norm": 7.61718225479126,
+      "grad_norm": 12.852262496948242,
       "learning_rate": 6.672672672672673e-06,
-      "loss": 1.1369,
+      "loss": 1.1348,
       "step": 33240
     },
     {
       "epoch": 99.85,
-      "grad_norm": 5.226757049560547,
+      "grad_norm": 6.01716947555542,
       "learning_rate": 6.671671671671672e-06,
-      "loss": 1.1135,
+      "loss": 1.1022,
       "step": 33250
     },
     {
       "epoch": 99.88,
-      "grad_norm": 4.983853816986084,
+      "grad_norm": 5.247378826141357,
       "learning_rate": 6.670670670670671e-06,
-      "loss": 1.1545,
+      "loss": 1.1476,
       "step": 33260
     },
     {
       "epoch": 99.91,
-      "grad_norm": 8.712162971496582,
+      "grad_norm": 10.58740520477295,
       "learning_rate": 6.669669669669671e-06,
-      "loss": 1.1007,
+      "loss": 1.1017,
       "step": 33270
     },
     {
       "epoch": 99.94,
-      "grad_norm": 5.820200443267822,
+      "grad_norm": 6.939590930938721,
       "learning_rate": 6.6686686686686695e-06,
-      "loss": 1.1205,
+      "loss": 1.1208,
       "step": 33280
     },
     {
       "epoch": 99.97,
-      "grad_norm": 6.357318878173828,
+      "grad_norm": 7.243859767913818,
       "learning_rate": 6.667667667667668e-06,
-      "loss": 1.1306,
+      "loss": 1.1325,
       "step": 33290
     },
     {
       "epoch": 100.0,
-      "grad_norm": 34.81151580810547,
+      "grad_norm": 54.05570602416992,
       "learning_rate": 6.666666666666667e-06,
-      "loss": 1.1854,
+      "loss": 1.1985,
       "step": 33300
     },
     {
       "epoch": 100.0,
-      "eval_accuracy": 0.8007,
-      "eval_loss": 0.6525286436080933,
-      "eval_runtime": 5.7274,
-      "eval_samples_per_second": 1745.996,
-      "eval_steps_per_second": 6.984,
+      "eval_accuracy": 0.8002,
+      "eval_loss": 0.6521774530410767,
+      "eval_runtime": 5.6324,
+      "eval_samples_per_second": 1775.441,
+      "eval_steps_per_second": 7.102,
       "step": 33300
     },
     {
       "epoch": 100.03,
-      "grad_norm": 5.743560314178467,
+      "grad_norm": 6.577492713928223,
       "learning_rate": 6.665665665665666e-06,
-      "loss": 1.0939,
+      "loss": 1.0962,
       "step": 33310
     },
     {
       "epoch": 100.06,
-      "grad_norm": 6.298009872436523,
+      "grad_norm": 6.74983549118042,
       "learning_rate": 6.664664664664665e-06,
-      "loss": 1.1041,
+      "loss": 1.1076,
       "step": 33320
     },
     {
       "epoch": 100.09,
-      "grad_norm": 5.211252212524414,
+      "grad_norm": 7.423816204071045,
       "learning_rate": 6.663663663663664e-06,
-      "loss": 1.1006,
+      "loss": 1.1075,
       "step": 33330
     },
     {
       "epoch": 100.12,
-      "grad_norm": 5.535873889923096,
+      "grad_norm": 14.360748291015625,
       "learning_rate": 6.6626626626626636e-06,
-      "loss": 1.1287,
+      "loss": 1.1168,
       "step": 33340
     },
     {
       "epoch": 100.15,
-      "grad_norm": 6.115882873535156,
+      "grad_norm": 7.322205543518066,
       "learning_rate": 6.661661661661662e-06,
-      "loss": 1.0975,
+      "loss": 1.0915,
       "step": 33350
     },
     {
       "epoch": 100.18,
-      "grad_norm": 7.129404544830322,
+      "grad_norm": 6.235116958618164,
       "learning_rate": 6.660660660660662e-06,
-      "loss": 1.1553,
+      "loss": 1.1507,
       "step": 33360
     },
     {
       "epoch": 100.21,
-      "grad_norm": 6.124326229095459,
+      "grad_norm": 5.524021148681641,
       "learning_rate": 6.659659659659659e-06,
-      "loss": 1.1248,
+      "loss": 1.1275,
       "step": 33370
     },
     {
       "epoch": 100.24,
-      "grad_norm": 6.367180824279785,
+      "grad_norm": 6.342471599578857,
       "learning_rate": 6.658658658658659e-06,
-      "loss": 1.1428,
+      "loss": 1.1415,
       "step": 33380
     },
     {
       "epoch": 100.27,
-      "grad_norm": 5.114198207855225,
+      "grad_norm": 6.137682914733887,
       "learning_rate": 6.657657657657658e-06,
-      "loss": 1.0826,
+      "loss": 1.0901,
       "step": 33390
     },
     {
       "epoch": 100.3,
-      "grad_norm": 6.327525615692139,
+      "grad_norm": 6.088094711303711,
       "learning_rate": 6.656656656656657e-06,
-      "loss": 1.087,
+      "loss": 1.0813,
       "step": 33400
     },
     {
       "epoch": 100.33,
-      "grad_norm": 5.338654518127441,
+      "grad_norm": 4.858251571655273,
       "learning_rate": 6.655655655655656e-06,
-      "loss": 1.0959,
+      "loss": 1.092,
       "step": 33410
     },
     {
       "epoch": 100.36,
-      "grad_norm": 6.080212593078613,
+      "grad_norm": 8.151640892028809,
       "learning_rate": 6.654654654654656e-06,
-      "loss": 1.1812,
+      "loss": 1.1811,
       "step": 33420
     },
     {
       "epoch": 100.39,
-      "grad_norm": 6.28367805480957,
+      "grad_norm": 7.971110820770264,
       "learning_rate": 6.653653653653654e-06,
-      "loss": 1.1023,
+      "loss": 1.1027,
       "step": 33430
     },
     {
       "epoch": 100.42,
-      "grad_norm": 8.223723411560059,
+      "grad_norm": 7.210797309875488,
       "learning_rate": 6.652652652652654e-06,
-      "loss": 1.1559,
+      "loss": 1.1619,
       "step": 33440
     },
     {
       "epoch": 100.45,
-      "grad_norm": 6.616147041320801,
+      "grad_norm": 9.254097938537598,
       "learning_rate": 6.651651651651652e-06,
-      "loss": 1.0652,
+      "loss": 1.0507,
       "step": 33450
     },
     {
       "epoch": 100.48,
-      "grad_norm": 7.884021282196045,
+      "grad_norm": 7.306634426116943,
       "learning_rate": 6.650650650650651e-06,
-      "loss": 1.1823,
+      "loss": 1.1957,
       "step": 33460
     },
     {
       "epoch": 100.51,
-      "grad_norm": 8.82709789276123,
+      "grad_norm": 7.31254243850708,
       "learning_rate": 6.6496496496496495e-06,
-      "loss": 1.0501,
+      "loss": 1.0566,
       "step": 33470
     },
     {
       "epoch": 100.54,
-      "grad_norm": 5.583005428314209,
+      "grad_norm": 6.582447052001953,
       "learning_rate": 6.648648648648649e-06,
-      "loss": 1.1729,
+      "loss": 1.1706,
       "step": 33480
     },
     {
       "epoch": 100.57,
-      "grad_norm": 10.100821495056152,
+      "grad_norm": 7.73210334777832,
       "learning_rate": 6.647647647647648e-06,
-      "loss": 1.1903,
+      "loss": 1.1753,
       "step": 33490
     },
     {
       "epoch": 100.6,
-      "grad_norm": 5.963928699493408,
+      "grad_norm": 7.584439277648926,
       "learning_rate": 6.646646646646647e-06,
-      "loss": 1.163,
+      "loss": 1.1653,
       "step": 33500
     },
     {
       "epoch": 100.63,
-      "grad_norm": 7.735825061798096,
+      "grad_norm": 10.121824264526367,
       "learning_rate": 6.6456456456456465e-06,
-      "loss": 1.0889,
+      "loss": 1.0899,
       "step": 33510
     },
     {
       "epoch": 100.66,
-      "grad_norm": 5.290556907653809,
+      "grad_norm": 6.172900199890137,
       "learning_rate": 6.644644644644646e-06,
-      "loss": 1.1138,
+      "loss": 1.1085,
       "step": 33520
     },
     {
       "epoch": 100.69,
-      "grad_norm": 5.478888988494873,
+      "grad_norm": 6.511559963226318,
       "learning_rate": 6.6436436436436445e-06,
-      "loss": 1.1267,
+      "loss": 1.1245,
       "step": 33530
     },
     {
       "epoch": 100.72,
-      "grad_norm": 8.588815689086914,
+      "grad_norm": 7.061877727508545,
       "learning_rate": 6.642642642642643e-06,
-      "loss": 1.1193,
+      "loss": 1.1309,
       "step": 33540
     },
     {
       "epoch": 100.75,
-      "grad_norm": 5.83283805847168,
+      "grad_norm": 9.95302963256836,
       "learning_rate": 6.641641641641642e-06,
-      "loss": 1.0989,
+      "loss": 1.0916,
       "step": 33550
     },
     {
       "epoch": 100.78,
-      "grad_norm": 9.584370613098145,
+      "grad_norm": 15.828944206237793,
       "learning_rate": 6.640640640640641e-06,
-      "loss": 1.0418,
+      "loss": 1.0309,
       "step": 33560
     },
     {
       "epoch": 100.81,
-      "grad_norm": 8.48386287689209,
+      "grad_norm": 7.731821060180664,
       "learning_rate": 6.63963963963964e-06,
-      "loss": 1.0502,
+      "loss": 1.046,
       "step": 33570
     },
     {
       "epoch": 100.84,
-      "grad_norm": 7.05922269821167,
+      "grad_norm": 7.63286018371582,
       "learning_rate": 6.638638638638639e-06,
-      "loss": 1.0844,
+      "loss": 1.0799,
       "step": 33580
     },
     {
       "epoch": 100.87,
-      "grad_norm": 11.038089752197266,
+      "grad_norm": 8.039173126220703,
       "learning_rate": 6.637637637637639e-06,
-      "loss": 1.0395,
+      "loss": 1.0294,
       "step": 33590
     },
     {
       "epoch": 100.9,
-      "grad_norm": 6.256180286407471,
+      "grad_norm": 6.728932857513428,
       "learning_rate": 6.636636636636637e-06,
-      "loss": 1.1416,
+      "loss": 1.1381,
       "step": 33600
     },
     {
       "epoch": 100.93,
-      "grad_norm": 6.4898762702941895,
+      "grad_norm": 8.511120796203613,
       "learning_rate": 6.635635635635637e-06,
-      "loss": 1.1792,
+      "loss": 1.1806,
       "step": 33610
     },
     {
       "epoch": 100.96,
-      "grad_norm": 10.067558288574219,
+      "grad_norm": 7.556412220001221,
       "learning_rate": 6.634634634634634e-06,
-      "loss": 1.1161,
+      "loss": 1.1229,
       "step": 33620
     },
     {
       "epoch": 100.99,
-      "grad_norm": 7.577031135559082,
+      "grad_norm": 8.403305053710938,
       "learning_rate": 6.633633633633634e-06,
-      "loss": 1.1589,
+      "loss": 1.1456,
       "step": 33630
     },
     {
       "epoch": 101.0,
-      "eval_accuracy": 0.8004,
-      "eval_loss": 0.6483901143074036,
-      "eval_runtime": 5.2473,
-      "eval_samples_per_second": 1905.755,
-      "eval_steps_per_second": 7.623,
+      "eval_accuracy": 0.802,
+      "eval_loss": 0.6476743221282959,
+      "eval_runtime": 5.2909,
+      "eval_samples_per_second": 1890.041,
+      "eval_steps_per_second": 7.56,
       "step": 33633
     },
     {
       "epoch": 101.02,
-      "grad_norm": 6.027695178985596,
+      "grad_norm": 6.746572494506836,
       "learning_rate": 6.632632632632633e-06,
-      "loss": 1.3324,
+      "loss": 1.3127,
       "step": 33640
     },
     {
       "epoch": 101.05,
-      "grad_norm": 6.110278129577637,
+      "grad_norm": 8.586809158325195,
       "learning_rate": 6.631631631631632e-06,
-      "loss": 1.1841,
+      "loss": 1.1882,
       "step": 33650
     },
     {
       "epoch": 101.08,
-      "grad_norm": 5.9016242027282715,
+      "grad_norm": 5.67822265625,
       "learning_rate": 6.630630630630631e-06,
-      "loss": 1.15,
+      "loss": 1.1471,
       "step": 33660
     },
     {
       "epoch": 101.11,
-      "grad_norm": 5.808050155639648,
+      "grad_norm": 6.733979225158691,
       "learning_rate": 6.62962962962963e-06,
-      "loss": 1.1353,
+      "loss": 1.1282,
       "step": 33670
     },
     {
       "epoch": 101.14,
-      "grad_norm": 6.185662269592285,
+      "grad_norm": 6.855559349060059,
       "learning_rate": 6.628628628628629e-06,
-      "loss": 1.1693,
+      "loss": 1.177,
       "step": 33680
     },
     {
       "epoch": 101.17,
-      "grad_norm": 7.7011637687683105,
+      "grad_norm": 8.706315994262695,
       "learning_rate": 6.627627627627629e-06,
-      "loss": 1.1308,
+      "loss": 1.1388,
       "step": 33690
     },
     {
       "epoch": 101.2,
-      "grad_norm": 6.846614837646484,
+      "grad_norm": 5.8230414390563965,
       "learning_rate": 6.626626626626627e-06,
-      "loss": 1.1709,
+      "loss": 1.1579,
       "step": 33700
     },
     {
       "epoch": 101.23,
-      "grad_norm": 7.101274013519287,
+      "grad_norm": 7.249975204467773,
       "learning_rate": 6.625625625625626e-06,
-      "loss": 1.0434,
+      "loss": 1.0449,
       "step": 33710
     },
     {
       "epoch": 101.26,
-      "grad_norm": 4.891470909118652,
+      "grad_norm": 6.0238165855407715,
       "learning_rate": 6.6246246246246246e-06,
-      "loss": 1.193,
+      "loss": 1.1902,
       "step": 33720
     },
     {
       "epoch": 101.29,
-      "grad_norm": 4.658853530883789,
+      "grad_norm": 6.03178071975708,
       "learning_rate": 6.623623623623624e-06,
-      "loss": 1.1,
+      "loss": 1.0925,
       "step": 33730
     },
     {
       "epoch": 101.32,
-      "grad_norm": 6.063246726989746,
+      "grad_norm": 6.943318843841553,
       "learning_rate": 6.6226226226226235e-06,
-      "loss": 1.1119,
+      "loss": 1.1135,
       "step": 33740
     },
     {
       "epoch": 101.35,
-      "grad_norm": 6.6746110916137695,
+      "grad_norm": 9.507216453552246,
       "learning_rate": 6.621621621621622e-06,
-      "loss": 1.0806,
+      "loss": 1.0866,
       "step": 33750
     },
     {
       "epoch": 101.38,
-      "grad_norm": 7.611641883850098,
+      "grad_norm": 6.841081619262695,
       "learning_rate": 6.6206206206206215e-06,
-      "loss": 1.07,
+      "loss": 1.0652,
       "step": 33760
     },
     {
       "epoch": 101.41,
-      "grad_norm": 7.625645637512207,
+      "grad_norm": 6.043604373931885,
       "learning_rate": 6.619619619619621e-06,
-      "loss": 1.1295,
+      "loss": 1.1224,
       "step": 33770
     },
     {
       "epoch": 101.44,
-      "grad_norm": 7.649852275848389,
+      "grad_norm": 7.085200309753418,
       "learning_rate": 6.6186186186186195e-06,
-      "loss": 1.1717,
+      "loss": 1.1703,
       "step": 33780
     },
     {
       "epoch": 101.47,
-      "grad_norm": 7.775462627410889,
+      "grad_norm": 7.828639030456543,
       "learning_rate": 6.617617617617617e-06,
-      "loss": 1.1005,
+      "loss": 1.0896,
       "step": 33790
     },
     {
       "epoch": 101.5,
-      "grad_norm": 6.596286773681641,
+      "grad_norm": 6.964863300323486,
       "learning_rate": 6.616616616616617e-06,
-      "loss": 1.081,
+      "loss": 1.074,
       "step": 33800
     },
     {
       "epoch": 101.53,
-      "grad_norm": 8.754149436950684,
+      "grad_norm": 9.195528030395508,
       "learning_rate": 6.615615615615616e-06,
-      "loss": 1.1437,
+      "loss": 1.1438,
       "step": 33810
     },
     {
       "epoch": 101.56,
-      "grad_norm": 6.499390125274658,
+      "grad_norm": 8.080568313598633,
       "learning_rate": 6.614614614614615e-06,
-      "loss": 1.158,
+      "loss": 1.1453,
       "step": 33820
     },
     {
       "epoch": 101.59,
-      "grad_norm": 8.816536903381348,
+      "grad_norm": 7.66557502746582,
       "learning_rate": 6.613613613613614e-06,
-      "loss": 1.1248,
+      "loss": 1.1141,
       "step": 33830
     },
     {
       "epoch": 101.62,
-      "grad_norm": 8.939920425415039,
+      "grad_norm": 7.130758762359619,
       "learning_rate": 6.612612612612614e-06,
-      "loss": 1.1189,
+      "loss": 1.1242,
       "step": 33840
     },
     {
       "epoch": 101.65,
-      "grad_norm": 9.267573356628418,
+      "grad_norm": 9.105847358703613,
       "learning_rate": 6.611611611611612e-06,
-      "loss": 1.1846,
+      "loss": 1.1929,
       "step": 33850
     },
     {
       "epoch": 101.68,
-      "grad_norm": 6.727185249328613,
+      "grad_norm": 5.138393878936768,
       "learning_rate": 6.610610610610612e-06,
-      "loss": 1.098,
+      "loss": 1.1015,
       "step": 33860
     },
     {
       "epoch": 101.71,
-      "grad_norm": 8.451547622680664,
+      "grad_norm": 6.146358013153076,
       "learning_rate": 6.609609609609609e-06,
-      "loss": 1.1695,
+      "loss": 1.1634,
       "step": 33870
     },
     {
       "epoch": 101.74,
-      "grad_norm": 6.456782341003418,
+      "grad_norm": 5.744650840759277,
       "learning_rate": 6.608608608608609e-06,
-      "loss": 1.1178,
+      "loss": 1.1249,
       "step": 33880
     },
     {
       "epoch": 101.77,
-      "grad_norm": 6.1656880378723145,
+      "grad_norm": 7.579628944396973,
       "learning_rate": 6.607607607607608e-06,
-      "loss": 1.1359,
+      "loss": 1.1453,
       "step": 33890
     },
     {
       "epoch": 101.8,
-      "grad_norm": 8.428045272827148,
+      "grad_norm": 7.887132167816162,
       "learning_rate": 6.606606606606607e-06,
-      "loss": 1.1476,
+      "loss": 1.1434,
       "step": 33900
     },
     {
       "epoch": 101.83,
-      "grad_norm": 5.577824115753174,
+      "grad_norm": 6.029246807098389,
       "learning_rate": 6.605605605605606e-06,
-      "loss": 1.1132,
+      "loss": 1.1145,
       "step": 33910
     },
     {
       "epoch": 101.86,
-      "grad_norm": 5.276789665222168,
+      "grad_norm": 6.342925071716309,
       "learning_rate": 6.604604604604605e-06,
-      "loss": 1.0875,
+      "loss": 1.0885,
       "step": 33920
     },
     {
       "epoch": 101.89,
-      "grad_norm": 10.915855407714844,
+      "grad_norm": 10.828214645385742,
       "learning_rate": 6.603603603603604e-06,
-      "loss": 1.1035,
+      "loss": 1.0969,
       "step": 33930
     },
     {
       "epoch": 101.92,
-      "grad_norm": 7.605905055999756,
+      "grad_norm": 8.098462104797363,
       "learning_rate": 6.602602602602604e-06,
-      "loss": 1.1025,
+      "loss": 1.1051,
       "step": 33940
     },
     {
       "epoch": 101.95,
-      "grad_norm": 6.24487829208374,
+      "grad_norm": 8.480448722839355,
       "learning_rate": 6.601601601601602e-06,
-      "loss": 1.0569,
+      "loss": 1.0611,
       "step": 33950
     },
     {
       "epoch": 101.98,
-      "grad_norm": 10.42727279663086,
+      "grad_norm": 11.578932762145996,
       "learning_rate": 6.600600600600601e-06,
-      "loss": 1.1621,
+      "loss": 1.1648,
       "step": 33960
     },
     {
       "epoch": 102.0,
-      "eval_accuracy": 0.8028,
-      "eval_loss": 0.6455795168876648,
-      "eval_runtime": 5.5556,
-      "eval_samples_per_second": 1799.998,
-      "eval_steps_per_second": 7.2,
+      "eval_accuracy": 0.8013,
+      "eval_loss": 0.642520010471344,
+      "eval_runtime": 5.342,
+      "eval_samples_per_second": 1871.963,
+      "eval_steps_per_second": 7.488,
       "step": 33966
     },
     {
       "epoch": 102.01,
-      "grad_norm": 7.577363967895508,
+      "grad_norm": 8.125686645507812,
       "learning_rate": 6.5995995995996e-06,
-      "loss": 1.3517,
+      "loss": 1.3343,
       "step": 33970
     },
     {
       "epoch": 102.04,
-      "grad_norm": 8.412418365478516,
+      "grad_norm": 8.957856178283691,
       "learning_rate": 6.598598598598599e-06,
-      "loss": 1.1063,
+      "loss": 1.1033,
       "step": 33980
     },
     {
       "epoch": 102.07,
-      "grad_norm": 5.5821943283081055,
+      "grad_norm": 6.880398750305176,
       "learning_rate": 6.5975975975975985e-06,
-      "loss": 1.1068,
+      "loss": 1.1116,
       "step": 33990
     },
     {
       "epoch": 102.1,
-      "grad_norm": 5.765705585479736,
+      "grad_norm": 9.082393646240234,
       "learning_rate": 6.596596596596597e-06,
-      "loss": 1.1009,
+      "loss": 1.1022,
       "step": 34000
     },
     {
       "epoch": 102.13,
-      "grad_norm": 6.218010902404785,
+      "grad_norm": 8.626836776733398,
       "learning_rate": 6.5955955955955965e-06,
-      "loss": 1.0877,
+      "loss": 1.102,
       "step": 34010
     },
     {
       "epoch": 102.16,
-      "grad_norm": 6.556366920471191,
+      "grad_norm": 6.8483686447143555,
       "learning_rate": 6.594594594594595e-06,
-      "loss": 1.1191,
+      "loss": 1.1242,
       "step": 34020
     },
     {
       "epoch": 102.19,
-      "grad_norm": 5.7375168800354,
+      "grad_norm": 8.152750968933105,
       "learning_rate": 6.5935935935935946e-06,
-      "loss": 1.155,
+      "loss": 1.1488,
       "step": 34030
     },
     {
       "epoch": 102.22,
-      "grad_norm": 5.826480388641357,
+      "grad_norm": 7.038767337799072,
       "learning_rate": 6.592592592592592e-06,
-      "loss": 1.1125,
+      "loss": 1.1175,
       "step": 34040
     },
     {
       "epoch": 102.25,
-      "grad_norm": 9.080286979675293,
+      "grad_norm": 8.128741264343262,
       "learning_rate": 6.591591591591592e-06,
-      "loss": 1.1099,
+      "loss": 1.1163,
       "step": 34050
     },
     {
       "epoch": 102.28,
-      "grad_norm": 7.541192054748535,
+      "grad_norm": 8.101195335388184,
       "learning_rate": 6.590590590590591e-06,
-      "loss": 1.1434,
+      "loss": 1.1539,
       "step": 34060
     },
     {
       "epoch": 102.31,
-      "grad_norm": 8.04828929901123,
+      "grad_norm": 8.978123664855957,
       "learning_rate": 6.58958958958959e-06,
-      "loss": 1.0999,
+      "loss": 1.0985,
       "step": 34070
     },
     {
       "epoch": 102.34,
-      "grad_norm": 5.517087936401367,
+      "grad_norm": 6.0953850746154785,
       "learning_rate": 6.588588588588589e-06,
-      "loss": 1.0847,
+      "loss": 1.0852,
       "step": 34080
     },
     {
       "epoch": 102.37,
-      "grad_norm": 8.664015769958496,
+      "grad_norm": 7.036874771118164,
       "learning_rate": 6.587587587587589e-06,
-      "loss": 1.1316,
+      "loss": 1.1286,
       "step": 34090
     },
     {
       "epoch": 102.4,
-      "grad_norm": 8.77520751953125,
+      "grad_norm": 8.236102104187012,
       "learning_rate": 6.586586586586587e-06,
-      "loss": 1.1351,
+      "loss": 1.1253,
       "step": 34100
     },
     {
       "epoch": 102.43,
-      "grad_norm": 5.954639434814453,
+      "grad_norm": 7.535617351531982,
       "learning_rate": 6.585585585585587e-06,
-      "loss": 1.1606,
+      "loss": 1.1551,
       "step": 34110
     },
     {
       "epoch": 102.46,
-      "grad_norm": 8.057635307312012,
+      "grad_norm": 6.41841459274292,
       "learning_rate": 6.5845845845845845e-06,
-      "loss": 1.0996,
+      "loss": 1.0967,
       "step": 34120
     },
     {
       "epoch": 102.49,
-      "grad_norm": 8.137197494506836,
+      "grad_norm": 6.349040508270264,
       "learning_rate": 6.583583583583584e-06,
-      "loss": 1.1816,
+      "loss": 1.178,
       "step": 34130
     },
     {
       "epoch": 102.52,
-      "grad_norm": 6.502910614013672,
+      "grad_norm": 6.516385078430176,
       "learning_rate": 6.5825825825825825e-06,
-      "loss": 1.1176,
+      "loss": 1.111,
       "step": 34140
     },
     {
       "epoch": 102.55,
-      "grad_norm": 5.534854888916016,
+      "grad_norm": 8.206841468811035,
       "learning_rate": 6.581581581581582e-06,
-      "loss": 1.1193,
+      "loss": 1.1326,
       "step": 34150
     },
     {
       "epoch": 102.58,
-      "grad_norm": 6.770633220672607,
+      "grad_norm": 9.287317276000977,
       "learning_rate": 6.580580580580581e-06,
-      "loss": 1.199,
+      "loss": 1.2014,
       "step": 34160
     },
     {
       "epoch": 102.61,
-      "grad_norm": 6.960815906524658,
+      "grad_norm": 7.4901909828186035,
       "learning_rate": 6.57957957957958e-06,
-      "loss": 1.1671,
+      "loss": 1.1527,
       "step": 34170
     },
     {
       "epoch": 102.64,
-      "grad_norm": 7.238114833831787,
+      "grad_norm": 6.652816295623779,
       "learning_rate": 6.578578578578579e-06,
-      "loss": 1.1318,
+      "loss": 1.1313,
       "step": 34180
     },
     {
       "epoch": 102.67,
-      "grad_norm": 8.259832382202148,
+      "grad_norm": 12.266073226928711,
       "learning_rate": 6.577577577577579e-06,
-      "loss": 1.0841,
+      "loss": 1.0797,
       "step": 34190
     },
     {
       "epoch": 102.7,
-      "grad_norm": 5.898361682891846,
+      "grad_norm": 8.913169860839844,
       "learning_rate": 6.5765765765765775e-06,
-      "loss": 1.1093,
+      "loss": 1.1033,
       "step": 34200
     },
     {
       "epoch": 102.73,
-      "grad_norm": 7.628055572509766,
+      "grad_norm": 6.248297691345215,
       "learning_rate": 6.575575575575576e-06,
-      "loss": 1.0946,
+      "loss": 1.095,
       "step": 34210
     },
     {
       "epoch": 102.76,
-      "grad_norm": 8.015387535095215,
+      "grad_norm": 8.259164810180664,
       "learning_rate": 6.574574574574575e-06,
-      "loss": 1.0838,
+      "loss": 1.0831,
       "step": 34220
     },
     {
       "epoch": 102.79,
-      "grad_norm": 11.214966773986816,
+      "grad_norm": 8.552067756652832,
       "learning_rate": 6.573573573573574e-06,
-      "loss": 1.094,
+      "loss": 1.1019,
       "step": 34230
     },
     {
       "epoch": 102.82,
-      "grad_norm": 6.207174777984619,
+      "grad_norm": 7.350534439086914,
       "learning_rate": 6.572572572572573e-06,
       "loss": 1.095,
       "step": 34240
     },
     {
       "epoch": 102.85,
-      "grad_norm": 8.424463272094727,
+      "grad_norm": 8.951598167419434,
       "learning_rate": 6.571571571571572e-06,
-      "loss": 1.1474,
+      "loss": 1.1443,
       "step": 34250
     },
     {
       "epoch": 102.88,
-      "grad_norm": 7.197977542877197,
+      "grad_norm": 6.664317607879639,
       "learning_rate": 6.5705705705705716e-06,
-      "loss": 1.1133,
+      "loss": 1.1032,
       "step": 34260
     },
     {
       "epoch": 102.91,
-      "grad_norm": 6.939733982086182,
+      "grad_norm": 8.787091255187988,
       "learning_rate": 6.56956956956957e-06,
-      "loss": 1.1783,
+      "loss": 1.1609,
       "step": 34270
     },
     {
       "epoch": 102.94,
-      "grad_norm": 6.7816667556762695,
+      "grad_norm": 6.772167205810547,
       "learning_rate": 6.56856856856857e-06,
-      "loss": 1.1585,
+      "loss": 1.1616,
       "step": 34280
     },
     {
       "epoch": 102.97,
-      "grad_norm": 6.860868453979492,
+      "grad_norm": 6.265247344970703,
       "learning_rate": 6.567567567567567e-06,
-      "loss": 1.1066,
+      "loss": 1.1005,
       "step": 34290
     },
     {
       "epoch": 103.0,
-      "eval_accuracy": 0.8006,
-      "eval_loss": 0.6549428701400757,
-      "eval_runtime": 5.3441,
-      "eval_samples_per_second": 1871.235,
-      "eval_steps_per_second": 7.485,
+      "eval_accuracy": 0.8009,
+      "eval_loss": 0.6525696516036987,
+      "eval_runtime": 5.5759,
+      "eval_samples_per_second": 1793.446,
+      "eval_steps_per_second": 7.174,
       "step": 34299
     },
     {
       "epoch": 103.0,
-      "grad_norm": 6.734114646911621,
+      "grad_norm": 7.860905170440674,
       "learning_rate": 6.566566566566567e-06,
-      "loss": 1.2078,
+      "loss": 1.2183,
       "step": 34300
     },
     {
       "epoch": 103.03,
-      "grad_norm": 6.843709468841553,
+      "grad_norm": 9.787517547607422,
       "learning_rate": 6.565565565565566e-06,
-      "loss": 1.07,
+      "loss": 1.0742,
       "step": 34310
     },
     {
       "epoch": 103.06,
-      "grad_norm": 7.593029499053955,
+      "grad_norm": 7.319555759429932,
       "learning_rate": 6.564564564564565e-06,
-      "loss": 1.1276,
+      "loss": 1.1154,
       "step": 34320
     },
     {
       "epoch": 103.09,
-      "grad_norm": 5.523630619049072,
+      "grad_norm": 5.136215686798096,
       "learning_rate": 6.563563563563564e-06,
-      "loss": 1.1584,
+      "loss": 1.1534,
       "step": 34330
     },
     {
       "epoch": 103.12,
-      "grad_norm": 8.977666854858398,
+      "grad_norm": 10.25833511352539,
       "learning_rate": 6.562562562562564e-06,
-      "loss": 1.1216,
+      "loss": 1.1249,
       "step": 34340
     },
     {
       "epoch": 103.15,
-      "grad_norm": 6.877377510070801,
+      "grad_norm": 7.637105464935303,
       "learning_rate": 6.561561561561562e-06,
-      "loss": 1.1581,
+      "loss": 1.1587,
       "step": 34350
     },
     {
       "epoch": 103.18,
-      "grad_norm": 5.076304912567139,
+      "grad_norm": 5.662967681884766,
       "learning_rate": 6.560560560560562e-06,
-      "loss": 1.0886,
+      "loss": 1.0752,
       "step": 34360
     },
     {
       "epoch": 103.21,
-      "grad_norm": 8.886892318725586,
+      "grad_norm": 9.280594825744629,
       "learning_rate": 6.5595595595595595e-06,
-      "loss": 1.1367,
+      "loss": 1.1237,
       "step": 34370
     },
     {
       "epoch": 103.24,
-      "grad_norm": 6.370909214019775,
+      "grad_norm": 6.832572937011719,
       "learning_rate": 6.558558558558559e-06,
-      "loss": 1.0714,
+      "loss": 1.0696,
       "step": 34380
     },
     {
       "epoch": 103.27,
-      "grad_norm": 9.835305213928223,
+      "grad_norm": 10.555768013000488,
       "learning_rate": 6.5575575575575575e-06,
-      "loss": 1.0877,
+      "loss": 1.0764,
       "step": 34390
     },
     {
       "epoch": 103.3,
-      "grad_norm": 8.959345817565918,
+      "grad_norm": 7.1383209228515625,
       "learning_rate": 6.556556556556557e-06,
-      "loss": 1.1015,
+      "loss": 1.0988,
       "step": 34400
     },
     {
       "epoch": 103.33,
-      "grad_norm": 7.464613914489746,
+      "grad_norm": 8.769537925720215,
       "learning_rate": 6.555555555555556e-06,
-      "loss": 1.1467,
+      "loss": 1.1439,
       "step": 34410
     },
     {
       "epoch": 103.36,
-      "grad_norm": 8.288899421691895,
+      "grad_norm": 8.342437744140625,
       "learning_rate": 6.554554554554555e-06,
-      "loss": 1.1762,
+      "loss": 1.1662,
       "step": 34420
     },
     {
       "epoch": 103.39,
-      "grad_norm": 5.571897506713867,
+      "grad_norm": 5.852190971374512,
       "learning_rate": 6.5535535535535544e-06,
-      "loss": 1.1153,
+      "loss": 1.1125,
       "step": 34430
     },
     {
       "epoch": 103.42,
-      "grad_norm": 6.8669891357421875,
+      "grad_norm": 6.5291314125061035,
       "learning_rate": 6.552552552552554e-06,
-      "loss": 1.1178,
+      "loss": 1.0949,
       "step": 34440
     },
     {
       "epoch": 103.45,
-      "grad_norm": 8.482243537902832,
+      "grad_norm": 9.915285110473633,
       "learning_rate": 6.551551551551552e-06,
-      "loss": 1.0992,
+      "loss": 1.0967,
       "step": 34450
     },
     {
       "epoch": 103.48,
-      "grad_norm": 7.498608589172363,
+      "grad_norm": 8.365479469299316,
       "learning_rate": 6.550550550550551e-06,
-      "loss": 1.1477,
+      "loss": 1.1478,
       "step": 34460
     },
     {
       "epoch": 103.51,
-      "grad_norm": 5.398630619049072,
+      "grad_norm": 8.720650672912598,
       "learning_rate": 6.54954954954955e-06,
-      "loss": 1.1146,
+      "loss": 1.108,
       "step": 34470
     },
     {
       "epoch": 103.54,
-      "grad_norm": 5.720746994018555,
+      "grad_norm": 7.990910530090332,
       "learning_rate": 6.548548548548549e-06,
-      "loss": 1.1463,
+      "loss": 1.1477,
       "step": 34480
     },
     {
       "epoch": 103.57,
-      "grad_norm": 7.074902057647705,
+      "grad_norm": 7.141448497772217,
       "learning_rate": 6.547547547547548e-06,
-      "loss": 1.1696,
+      "loss": 1.1546,
       "step": 34490
     },
     {
       "epoch": 103.6,
-      "grad_norm": 10.958619117736816,
+      "grad_norm": 16.38246726989746,
       "learning_rate": 6.546546546546547e-06,
-      "loss": 1.1336,
+      "loss": 1.1273,
       "step": 34500
     },
     {
       "epoch": 103.63,
-      "grad_norm": 7.787295818328857,
+      "grad_norm": 10.88282585144043,
       "learning_rate": 6.545545545545547e-06,
-      "loss": 1.1459,
+      "loss": 1.1485,
       "step": 34510
     },
     {
       "epoch": 103.66,
-      "grad_norm": 6.550888538360596,
+      "grad_norm": 6.986879348754883,
       "learning_rate": 6.544544544544545e-06,
-      "loss": 1.2188,
+      "loss": 1.2218,
       "step": 34520
     },
     {
       "epoch": 103.69,
-      "grad_norm": 8.491141319274902,
+      "grad_norm": 8.470141410827637,
       "learning_rate": 6.543543543543545e-06,
-      "loss": 1.1376,
+      "loss": 1.1158,
       "step": 34530
     },
     {
       "epoch": 103.72,
-      "grad_norm": 4.541799068450928,
+      "grad_norm": 5.400335311889648,
       "learning_rate": 6.542542542542542e-06,
-      "loss": 1.0908,
+      "loss": 1.0892,
       "step": 34540
     },
     {
       "epoch": 103.75,
-      "grad_norm": 5.165287017822266,
+      "grad_norm": 6.44712495803833,
       "learning_rate": 6.541541541541542e-06,
-      "loss": 1.1075,
+      "loss": 1.1135,
       "step": 34550
     },
     {
       "epoch": 103.78,
-      "grad_norm": 6.739583969116211,
+      "grad_norm": 7.5720953941345215,
       "learning_rate": 6.540540540540541e-06,
-      "loss": 1.1092,
+      "loss": 1.1107,
       "step": 34560
     },
     {
       "epoch": 103.81,
-      "grad_norm": 7.453261852264404,
+      "grad_norm": 10.76776123046875,
       "learning_rate": 6.53953953953954e-06,
-      "loss": 1.2118,
+      "loss": 1.2107,
       "step": 34570
     },
     {
       "epoch": 103.84,
-      "grad_norm": 5.015534400939941,
+      "grad_norm": 4.192724704742432,
       "learning_rate": 6.538538538538539e-06,
-      "loss": 1.1111,
+      "loss": 1.115,
       "step": 34580
     },
     {
       "epoch": 103.87,
-      "grad_norm": 6.954635143280029,
+      "grad_norm": 5.184593200683594,
       "learning_rate": 6.537537537537538e-06,
-      "loss": 1.1306,
+      "loss": 1.1393,
       "step": 34590
     },
     {
       "epoch": 103.9,
-      "grad_norm": 4.959394454956055,
+      "grad_norm": 6.441568374633789,
       "learning_rate": 6.536536536536537e-06,
-      "loss": 1.1085,
+      "loss": 1.1015,
       "step": 34600
     },
     {
       "epoch": 103.93,
-      "grad_norm": 6.110310077667236,
+      "grad_norm": 6.579643249511719,
       "learning_rate": 6.535535535535537e-06,
-      "loss": 1.1355,
+      "loss": 1.129,
       "step": 34610
     },
     {
       "epoch": 103.96,
-      "grad_norm": 6.13616418838501,
+      "grad_norm": 8.185837745666504,
       "learning_rate": 6.5345345345345345e-06,
-      "loss": 1.1195,
+      "loss": 1.1176,
       "step": 34620
     },
     {
       "epoch": 103.99,
-      "grad_norm": 5.983326435089111,
+      "grad_norm": 9.785394668579102,
       "learning_rate": 6.533533533533534e-06,
-      "loss": 1.1108,
+      "loss": 1.104,
       "step": 34630
     },
     {
       "epoch": 104.0,
-      "eval_accuracy": 0.8016,
-      "eval_loss": 0.6474565267562866,
-      "eval_runtime": 5.6016,
-      "eval_samples_per_second": 1785.207,
-      "eval_steps_per_second": 7.141,
+      "eval_accuracy": 0.8007,
+      "eval_loss": 0.6469685435295105,
+      "eval_runtime": 5.5841,
+      "eval_samples_per_second": 1790.802,
+      "eval_steps_per_second": 7.163,
       "step": 34632
     },
     {
       "epoch": 104.02,
-      "grad_norm": 7.283252716064453,
+      "grad_norm": 6.31179141998291,
       "learning_rate": 6.5325325325325326e-06,
-      "loss": 1.3232,
+      "loss": 1.3286,
       "step": 34640
     },
     {
       "epoch": 104.05,
-      "grad_norm": 7.825419902801514,
+      "grad_norm": 8.898679733276367,
       "learning_rate": 6.531531531531532e-06,
-      "loss": 1.1151,
+      "loss": 1.1205,
       "step": 34650
     },
     {
       "epoch": 104.08,
-      "grad_norm": 6.120309829711914,
+      "grad_norm": 6.553800106048584,
       "learning_rate": 6.5305305305305314e-06,
-      "loss": 1.0777,
+      "loss": 1.0666,
       "step": 34660
     },
     {
       "epoch": 104.11,
-      "grad_norm": 5.09692907333374,
+      "grad_norm": 5.705324649810791,
       "learning_rate": 6.52952952952953e-06,
-      "loss": 1.1292,
+      "loss": 1.1228,
       "step": 34670
     },
     {
       "epoch": 104.14,
-      "grad_norm": 7.906500816345215,
+      "grad_norm": 8.621255874633789,
       "learning_rate": 6.5285285285285295e-06,
-      "loss": 1.1767,
+      "loss": 1.1768,
       "step": 34680
     },
     {
       "epoch": 104.17,
-      "grad_norm": 7.137513160705566,
+      "grad_norm": 6.650600433349609,
       "learning_rate": 6.527527527527528e-06,
-      "loss": 1.2141,
+      "loss": 1.212,
       "step": 34690
     },
     {
       "epoch": 104.2,
-      "grad_norm": 7.357808589935303,
+      "grad_norm": 7.043776035308838,
       "learning_rate": 6.526526526526527e-06,
-      "loss": 1.1154,
+      "loss": 1.1058,
       "step": 34700
     },
     {
       "epoch": 104.23,
-      "grad_norm": 6.93979549407959,
+      "grad_norm": 9.570466995239258,
       "learning_rate": 6.525525525525525e-06,
-      "loss": 1.1181,
+      "loss": 1.1182,
       "step": 34710
     },
     {
       "epoch": 104.26,
-      "grad_norm": 5.020732402801514,
+      "grad_norm": 5.7597336769104,
       "learning_rate": 6.524524524524525e-06,
-      "loss": 1.1072,
+      "loss": 1.0969,
       "step": 34720
     },
     {
       "epoch": 104.29,
-      "grad_norm": 6.141091346740723,
+      "grad_norm": 6.675995826721191,
       "learning_rate": 6.523523523523524e-06,
-      "loss": 1.1156,
+      "loss": 1.1116,
       "step": 34730
     },
     {
       "epoch": 104.32,
-      "grad_norm": 6.757984161376953,
+      "grad_norm": 7.0862860679626465,
       "learning_rate": 6.522522522522523e-06,
-      "loss": 1.0601,
+      "loss": 1.0591,
       "step": 34740
     },
     {
       "epoch": 104.35,
-      "grad_norm": 8.38156795501709,
+      "grad_norm": 8.663498878479004,
       "learning_rate": 6.521521521521522e-06,
-      "loss": 1.1374,
+      "loss": 1.1315,
       "step": 34750
     },
     {
       "epoch": 104.38,
-      "grad_norm": 9.337801933288574,
+      "grad_norm": 8.804123878479004,
       "learning_rate": 6.520520520520522e-06,
-      "loss": 1.1291,
+      "loss": 1.1307,
       "step": 34760
     },
     {
       "epoch": 104.41,
-      "grad_norm": 5.191306114196777,
+      "grad_norm": 6.035397529602051,
       "learning_rate": 6.51951951951952e-06,
-      "loss": 1.081,
+      "loss": 1.0719,
       "step": 34770
     },
     {
       "epoch": 104.44,
-      "grad_norm": 6.0454277992248535,
+      "grad_norm": 8.25857162475586,
       "learning_rate": 6.51851851851852e-06,
-      "loss": 1.1586,
+      "loss": 1.1628,
       "step": 34780
     },
     {
       "epoch": 104.47,
-      "grad_norm": 5.67290735244751,
+      "grad_norm": 5.478704929351807,
       "learning_rate": 6.517517517517517e-06,
-      "loss": 1.1134,
+      "loss": 1.1087,
       "step": 34790
     },
     {
       "epoch": 104.5,
-      "grad_norm": 7.948431968688965,
+      "grad_norm": 8.59283447265625,
       "learning_rate": 6.516516516516517e-06,
-      "loss": 1.1441,
+      "loss": 1.1421,
       "step": 34800
     },
     {
       "epoch": 104.53,
-      "grad_norm": 5.553799629211426,
+      "grad_norm": 6.095491886138916,
       "learning_rate": 6.5155155155155155e-06,
-      "loss": 1.0872,
+      "loss": 1.0963,
       "step": 34810
     },
     {
       "epoch": 104.56,
-      "grad_norm": 6.543584823608398,
+      "grad_norm": 10.41076374053955,
       "learning_rate": 6.514514514514515e-06,
-      "loss": 1.1677,
+      "loss": 1.1671,
       "step": 34820
     },
     {
       "epoch": 104.59,
-      "grad_norm": 6.198753833770752,
+      "grad_norm": 7.410000801086426,
       "learning_rate": 6.513513513513514e-06,
-      "loss": 1.1666,
+      "loss": 1.1565,
       "step": 34830
     },
     {
       "epoch": 104.62,
-      "grad_norm": 7.706489562988281,
+      "grad_norm": 9.313766479492188,
       "learning_rate": 6.512512512512513e-06,
-      "loss": 1.1044,
+      "loss": 1.1207,
       "step": 34840
     },
     {
       "epoch": 104.65,
-      "grad_norm": 5.600742816925049,
+      "grad_norm": 6.388516426086426,
       "learning_rate": 6.511511511511512e-06,
-      "loss": 1.1279,
+      "loss": 1.1224,
       "step": 34850
     },
     {
       "epoch": 104.68,
-      "grad_norm": 8.736764907836914,
+      "grad_norm": 9.2155122756958,
       "learning_rate": 6.510510510510512e-06,
-      "loss": 1.0741,
+      "loss": 1.0714,
       "step": 34860
     },
     {
       "epoch": 104.71,
-      "grad_norm": 6.025099277496338,
+      "grad_norm": 5.461924076080322,
       "learning_rate": 6.5095095095095096e-06,
-      "loss": 1.1338,
+      "loss": 1.1213,
       "step": 34870
     },
     {
       "epoch": 104.74,
-      "grad_norm": 7.686117172241211,
+      "grad_norm": 6.515641212463379,
       "learning_rate": 6.508508508508509e-06,
-      "loss": 1.0846,
+      "loss": 1.0894,
       "step": 34880
     },
     {
       "epoch": 104.77,
-      "grad_norm": 9.797392845153809,
+      "grad_norm": 9.53022289276123,
       "learning_rate": 6.507507507507508e-06,
-      "loss": 1.1396,
+      "loss": 1.1439,
       "step": 34890
     },
     {
       "epoch": 104.8,
-      "grad_norm": 7.93204927444458,
+      "grad_norm": 5.748436450958252,
       "learning_rate": 6.506506506506507e-06,
-      "loss": 1.0541,
+      "loss": 1.0553,
       "step": 34900
     },
     {
       "epoch": 104.83,
-      "grad_norm": 9.328483581542969,
+      "grad_norm": 8.395913124084473,
       "learning_rate": 6.5055055055055065e-06,
-      "loss": 1.0551,
+      "loss": 1.0589,
       "step": 34910
     },
     {
       "epoch": 104.86,
-      "grad_norm": 7.794367790222168,
+      "grad_norm": 7.813923358917236,
       "learning_rate": 6.504504504504505e-06,
-      "loss": 1.1108,
+      "loss": 1.1151,
       "step": 34920
     },
     {
       "epoch": 104.89,
-      "grad_norm": 5.83880090713501,
+      "grad_norm": 5.478094577789307,
       "learning_rate": 6.5035035035035045e-06,
-      "loss": 1.1687,
+      "loss": 1.1629,
       "step": 34930
     },
     {
       "epoch": 104.92,
-      "grad_norm": 6.966657638549805,
+      "grad_norm": 7.014945983886719,
       "learning_rate": 6.502502502502503e-06,
-      "loss": 1.0772,
+      "loss": 1.0789,
       "step": 34940
     },
     {
       "epoch": 104.95,
-      "grad_norm": 7.320685386657715,
+      "grad_norm": 8.269954681396484,
       "learning_rate": 6.501501501501502e-06,
-      "loss": 1.0995,
+      "loss": 1.0867,
       "step": 34950
     },
     {
       "epoch": 104.98,
-      "grad_norm": 5.927343845367432,
+      "grad_norm": 6.114131927490234,
       "learning_rate": 6.5005005005005e-06,
-      "loss": 1.1329,
+      "loss": 1.1291,
       "step": 34960
     },
     {
       "epoch": 105.0,
-      "eval_accuracy": 0.802,
-      "eval_loss": 0.6419815421104431,
-      "eval_runtime": 5.2788,
-      "eval_samples_per_second": 1894.359,
-      "eval_steps_per_second": 7.577,
+      "eval_accuracy": 0.8045,
+      "eval_loss": 0.640106737613678,
+      "eval_runtime": 5.3807,
+      "eval_samples_per_second": 1858.483,
+      "eval_steps_per_second": 7.434,
       "step": 34965
     },
     {
       "epoch": 105.02,
-      "grad_norm": 7.502290725708008,
+      "grad_norm": 6.432104587554932,
       "learning_rate": 6.4994994994995e-06,
-      "loss": 1.3414,
+      "loss": 1.337,
       "step": 34970
     },
     {
       "epoch": 105.05,
-      "grad_norm": 7.060651779174805,
+      "grad_norm": 6.965089797973633,
       "learning_rate": 6.498498498498499e-06,
-      "loss": 1.1504,
+      "loss": 1.143,
       "step": 34980
     },
     {
       "epoch": 105.08,
-      "grad_norm": 6.420904159545898,
+      "grad_norm": 6.071815490722656,
       "learning_rate": 6.497497497497498e-06,
-      "loss": 1.0997,
+      "loss": 1.0927,
       "step": 34990
     },
     {
       "epoch": 105.11,
-      "grad_norm": 7.491681098937988,
+      "grad_norm": 7.316691875457764,
       "learning_rate": 6.496496496496497e-06,
-      "loss": 1.0626,
+      "loss": 1.0597,
       "step": 35000
     },
     {
       "epoch": 105.14,
-      "grad_norm": 6.531076431274414,
+      "grad_norm": 7.456391334533691,
       "learning_rate": 6.495495495495497e-06,
-      "loss": 1.1488,
+      "loss": 1.1489,
       "step": 35010
     },
     {
       "epoch": 105.17,
-      "grad_norm": 6.497890949249268,
+      "grad_norm": 7.677006721496582,
       "learning_rate": 6.494494494494495e-06,
-      "loss": 1.0476,
+      "loss": 1.0453,
       "step": 35020
     },
     {
       "epoch": 105.2,
-      "grad_norm": 7.32771110534668,
+      "grad_norm": 9.488685607910156,
       "learning_rate": 6.493493493493495e-06,
-      "loss": 1.0389,
+      "loss": 1.0216,
       "step": 35030
     },
     {
       "epoch": 105.23,
-      "grad_norm": 10.244178771972656,
+      "grad_norm": 8.859675407409668,
       "learning_rate": 6.4924924924924924e-06,
-      "loss": 1.1483,
+      "loss": 1.138,
       "step": 35040
     },
     {
       "epoch": 105.26,
-      "grad_norm": 7.445093154907227,
+      "grad_norm": 8.25672721862793,
       "learning_rate": 6.491491491491492e-06,
-      "loss": 1.1118,
+      "loss": 1.1077,
       "step": 35050
     },
     {
       "epoch": 105.29,
-      "grad_norm": 4.334314346313477,
+      "grad_norm": 5.2663893699646,
       "learning_rate": 6.4904904904904905e-06,
-      "loss": 1.0889,
+      "loss": 1.0835,
       "step": 35060
     },
     {
       "epoch": 105.32,
-      "grad_norm": 9.797545433044434,
+      "grad_norm": 9.808889389038086,
       "learning_rate": 6.48948948948949e-06,
-      "loss": 1.1798,
+      "loss": 1.1819,
       "step": 35070
     },
     {
       "epoch": 105.35,
-      "grad_norm": 6.035510540008545,
+      "grad_norm": 6.533900737762451,
       "learning_rate": 6.488488488488489e-06,
-      "loss": 1.1147,
+      "loss": 1.113,
       "step": 35080
     },
     {
       "epoch": 105.38,
-      "grad_norm": 8.868050575256348,
+      "grad_norm": 8.76127815246582,
       "learning_rate": 6.487487487487488e-06,
-      "loss": 1.1006,
+      "loss": 1.1018,
       "step": 35090
     },
     {
       "epoch": 105.41,
-      "grad_norm": 6.962992191314697,
+      "grad_norm": 6.628868103027344,
       "learning_rate": 6.486486486486487e-06,
-      "loss": 1.0838,
+      "loss": 1.0768,
       "step": 35100
     },
     {
       "epoch": 105.44,
-      "grad_norm": 6.248010635375977,
+      "grad_norm": 5.472059726715088,
       "learning_rate": 6.485485485485487e-06,
-      "loss": 1.0801,
+      "loss": 1.076,
       "step": 35110
     },
     {
       "epoch": 105.47,
-      "grad_norm": 8.660426139831543,
+      "grad_norm": 7.668332576751709,
       "learning_rate": 6.484484484484485e-06,
-      "loss": 1.0972,
+      "loss": 1.0949,
       "step": 35120
     },
     {
       "epoch": 105.5,
-      "grad_norm": 5.058567047119141,
+      "grad_norm": 5.965418815612793,
       "learning_rate": 6.483483483483484e-06,
-      "loss": 1.0191,
+      "loss": 1.0177,
       "step": 35130
     },
     {
       "epoch": 105.53,
-      "grad_norm": 8.028244018554688,
+      "grad_norm": 8.072368621826172,
       "learning_rate": 6.482482482482483e-06,
-      "loss": 1.1128,
+      "loss": 1.1165,
       "step": 35140
     },
     {
       "epoch": 105.56,
-      "grad_norm": 9.233287811279297,
+      "grad_norm": 11.005446434020996,
       "learning_rate": 6.481481481481482e-06,
-      "loss": 1.1194,
+      "loss": 1.122,
       "step": 35150
     },
     {
       "epoch": 105.59,
-      "grad_norm": 6.70985746383667,
+      "grad_norm": 7.374541759490967,
       "learning_rate": 6.480480480480481e-06,
-      "loss": 1.1411,
+      "loss": 1.1306,
       "step": 35160
     },
     {
       "epoch": 105.62,
-      "grad_norm": 7.067668914794922,
+      "grad_norm": 10.29902458190918,
       "learning_rate": 6.47947947947948e-06,
-      "loss": 1.0591,
+      "loss": 1.0556,
       "step": 35170
     },
     {
       "epoch": 105.65,
-      "grad_norm": 8.52836799621582,
+      "grad_norm": 6.6759233474731445,
       "learning_rate": 6.4784784784784796e-06,
-      "loss": 1.1443,
+      "loss": 1.1332,
       "step": 35180
     },
     {
       "epoch": 105.68,
-      "grad_norm": 9.686338424682617,
+      "grad_norm": 6.889441967010498,
       "learning_rate": 6.477477477477478e-06,
-      "loss": 1.1421,
+      "loss": 1.1329,
       "step": 35190
     },
     {
       "epoch": 105.71,
-      "grad_norm": 6.132402420043945,
+      "grad_norm": 5.545436382293701,
       "learning_rate": 6.476476476476477e-06,
-      "loss": 1.1304,
+      "loss": 1.1324,
       "step": 35200
     },
     {
       "epoch": 105.74,
-      "grad_norm": 6.905637741088867,
+      "grad_norm": 8.173844337463379,
       "learning_rate": 6.475475475475475e-06,
-      "loss": 1.0791,
+      "loss": 1.083,
       "step": 35210
     },
     {
       "epoch": 105.77,
-      "grad_norm": 6.894477844238281,
+      "grad_norm": 6.120182037353516,
       "learning_rate": 6.474474474474475e-06,
-      "loss": 1.148,
+      "loss": 1.1548,
       "step": 35220
     },
     {
       "epoch": 105.8,
-      "grad_norm": 5.454221725463867,
+      "grad_norm": 8.270605087280273,
       "learning_rate": 6.473473473473474e-06,
-      "loss": 1.0679,
+      "loss": 1.0576,
       "step": 35230
     },
     {
       "epoch": 105.83,
-      "grad_norm": 6.763976573944092,
+      "grad_norm": 7.286891460418701,
       "learning_rate": 6.472472472472473e-06,
-      "loss": 1.1098,
+      "loss": 1.1172,
       "step": 35240
     },
     {
       "epoch": 105.86,
-      "grad_norm": 7.690670490264893,
+      "grad_norm": 9.909543991088867,
       "learning_rate": 6.471471471471472e-06,
-      "loss": 1.1264,
+      "loss": 1.1169,
       "step": 35250
     },
     {
       "epoch": 105.89,
-      "grad_norm": 7.582956790924072,
+      "grad_norm": 7.168170928955078,
       "learning_rate": 6.470470470470472e-06,
-      "loss": 1.1354,
+      "loss": 1.1367,
       "step": 35260
     },
     {
       "epoch": 105.92,
-      "grad_norm": 10.899483680725098,
+      "grad_norm": 14.442118644714355,
       "learning_rate": 6.46946946946947e-06,
-      "loss": 1.0622,
+      "loss": 1.0602,
       "step": 35270
     },
     {
       "epoch": 105.95,
-      "grad_norm": 6.843169689178467,
+      "grad_norm": 6.508365154266357,
       "learning_rate": 6.46846846846847e-06,
-      "loss": 1.0982,
+      "loss": 1.1077,
       "step": 35280
     },
     {
       "epoch": 105.98,
-      "grad_norm": 9.51691722869873,
+      "grad_norm": 11.880293846130371,
       "learning_rate": 6.4674674674674675e-06,
-      "loss": 1.084,
+      "loss": 1.0789,
       "step": 35290
     },
     {
       "epoch": 106.0,
-      "eval_accuracy": 0.8011,
-      "eval_loss": 0.6431878209114075,
-      "eval_runtime": 5.2233,
-      "eval_samples_per_second": 1914.486,
-      "eval_steps_per_second": 7.658,
+      "eval_accuracy": 0.8013,
+      "eval_loss": 0.6414195895195007,
+      "eval_runtime": 5.3381,
+      "eval_samples_per_second": 1873.334,
+      "eval_steps_per_second": 7.493,
       "step": 35298
     },
     {
       "epoch": 106.01,
-      "grad_norm": 6.83957052230835,
+      "grad_norm": 6.839313507080078,
       "learning_rate": 6.466466466466467e-06,
-      "loss": 1.445,
+      "loss": 1.3835,
       "step": 35300
     },
     {
       "epoch": 106.04,
-      "grad_norm": 5.634103298187256,
+      "grad_norm": 8.049918174743652,
       "learning_rate": 6.4654654654654655e-06,
-      "loss": 1.1098,
+      "loss": 1.1099,
       "step": 35310
     },
     {
       "epoch": 106.07,
-      "grad_norm": 9.201004981994629,
+      "grad_norm": 11.549914360046387,
       "learning_rate": 6.464464464464465e-06,
-      "loss": 1.1096,
+      "loss": 1.1155,
       "step": 35320
     },
     {
       "epoch": 106.1,
-      "grad_norm": 6.729426860809326,
+      "grad_norm": 6.223927974700928,
       "learning_rate": 6.463463463463464e-06,
-      "loss": 1.1631,
+      "loss": 1.1474,
       "step": 35330
     },
     {
       "epoch": 106.13,
-      "grad_norm": 9.532023429870605,
+      "grad_norm": 9.09906005859375,
       "learning_rate": 6.462462462462463e-06,
-      "loss": 1.1583,
+      "loss": 1.1367,
       "step": 35340
     },
     {
       "epoch": 106.16,
-      "grad_norm": 6.064061641693115,
+      "grad_norm": 8.34214973449707,
       "learning_rate": 6.4614614614614624e-06,
-      "loss": 1.1349,
+      "loss": 1.1277,
       "step": 35350
     },
     {
       "epoch": 106.19,
-      "grad_norm": 8.656634330749512,
+      "grad_norm": 8.753836631774902,
       "learning_rate": 6.460460460460462e-06,
-      "loss": 1.1256,
+      "loss": 1.1099,
       "step": 35360
     },
     {
       "epoch": 106.22,
-      "grad_norm": 7.311756610870361,
+      "grad_norm": 6.33612585067749,
       "learning_rate": 6.45945945945946e-06,
-      "loss": 1.1471,
+      "loss": 1.1602,
       "step": 35370
     },
     {
       "epoch": 106.25,
-      "grad_norm": 8.422447204589844,
+      "grad_norm": 12.477778434753418,
       "learning_rate": 6.458458458458458e-06,
-      "loss": 1.1138,
+      "loss": 1.1192,
       "step": 35380
     },
     {
       "epoch": 106.28,
-      "grad_norm": 8.125385284423828,
+      "grad_norm": 4.9372358322143555,
       "learning_rate": 6.457457457457458e-06,
-      "loss": 1.0673,
+      "loss": 1.0617,
       "step": 35390
     },
     {
       "epoch": 106.31,
-      "grad_norm": 6.735252857208252,
+      "grad_norm": 6.08016300201416,
       "learning_rate": 6.456456456456457e-06,
-      "loss": 1.1506,
+      "loss": 1.1464,
       "step": 35400
     },
     {
       "epoch": 106.34,
-      "grad_norm": 8.38672924041748,
+      "grad_norm": 9.437256813049316,
       "learning_rate": 6.455455455455456e-06,
-      "loss": 1.1339,
+      "loss": 1.1327,
       "step": 35410
     },
     {
       "epoch": 106.37,
-      "grad_norm": 6.55241060256958,
+      "grad_norm": 8.085865020751953,
       "learning_rate": 6.454454454454455e-06,
-      "loss": 1.149,
+      "loss": 1.1373,
       "step": 35420
     },
     {
       "epoch": 106.4,
-      "grad_norm": 7.824096202850342,
+      "grad_norm": 8.291666984558105,
       "learning_rate": 6.453453453453455e-06,
-      "loss": 1.1686,
+      "loss": 1.1785,
       "step": 35430
     },
     {
       "epoch": 106.43,
-      "grad_norm": 6.175295829772949,
+      "grad_norm": 6.747929573059082,
       "learning_rate": 6.452452452452453e-06,
-      "loss": 1.117,
+      "loss": 1.1122,
       "step": 35440
     },
     {
       "epoch": 106.46,
-      "grad_norm": 6.609597682952881,
+      "grad_norm": 6.550422668457031,
       "learning_rate": 6.451451451451452e-06,
-      "loss": 1.0168,
+      "loss": 1.0179,
       "step": 35450
     },
     {
       "epoch": 106.49,
-      "grad_norm": 7.415184020996094,
+      "grad_norm": 8.527907371520996,
       "learning_rate": 6.45045045045045e-06,
-      "loss": 1.1783,
+      "loss": 1.1808,
       "step": 35460
     },
     {
       "epoch": 106.52,
-      "grad_norm": 6.500551223754883,
+      "grad_norm": 6.658613681793213,
       "learning_rate": 6.44944944944945e-06,
-      "loss": 1.0966,
+      "loss": 1.089,
       "step": 35470
     },
     {
       "epoch": 106.55,
-      "grad_norm": 6.05951452255249,
+      "grad_norm": 7.716269016265869,
       "learning_rate": 6.448448448448449e-06,
-      "loss": 1.1147,
+      "loss": 1.1168,
       "step": 35480
     },
     {
       "epoch": 106.58,
-      "grad_norm": 5.803422927856445,
+      "grad_norm": 6.206985950469971,
       "learning_rate": 6.447447447447448e-06,
-      "loss": 1.1142,
+      "loss": 1.112,
       "step": 35490
     },
     {
       "epoch": 106.61,
-      "grad_norm": 7.877813339233398,
+      "grad_norm": 6.395627975463867,
       "learning_rate": 6.446446446446447e-06,
-      "loss": 1.0949,
+      "loss": 1.0888,
       "step": 35500
     },
     {
       "epoch": 106.64,
-      "grad_norm": 7.949633598327637,
+      "grad_norm": 9.67166519165039,
       "learning_rate": 6.445445445445446e-06,
-      "loss": 1.1181,
+      "loss": 1.1122,
       "step": 35510
     },
     {
       "epoch": 106.67,
-      "grad_norm": 6.661839008331299,
+      "grad_norm": 5.7236008644104,
       "learning_rate": 6.444444444444445e-06,
-      "loss": 1.0371,
+      "loss": 1.0501,
       "step": 35520
     },
     {
       "epoch": 106.7,
-      "grad_norm": 8.547270774841309,
+      "grad_norm": 11.815479278564453,
       "learning_rate": 6.443443443443445e-06,
-      "loss": 1.1004,
+      "loss": 1.1025,
       "step": 35530
     },
     {
       "epoch": 106.73,
-      "grad_norm": 7.438020706176758,
+      "grad_norm": 7.423037528991699,
       "learning_rate": 6.4424424424424425e-06,
-      "loss": 1.0767,
+      "loss": 1.0748,
       "step": 35540
     },
     {
       "epoch": 106.76,
-      "grad_norm": 7.018486499786377,
+      "grad_norm": 6.76237678527832,
       "learning_rate": 6.441441441441442e-06,
-      "loss": 1.1651,
+      "loss": 1.1665,
       "step": 35550
     },
     {
       "epoch": 106.79,
-      "grad_norm": 8.198371887207031,
+      "grad_norm": 8.438776969909668,
       "learning_rate": 6.4404404404404406e-06,
-      "loss": 1.0816,
+      "loss": 1.0732,
       "step": 35560
     },
     {
       "epoch": 106.82,
-      "grad_norm": 7.134220600128174,
+      "grad_norm": 7.487705707550049,
       "learning_rate": 6.43943943943944e-06,
-      "loss": 1.0873,
+      "loss": 1.0939,
       "step": 35570
     },
     {
       "epoch": 106.85,
-      "grad_norm": 6.889313220977783,
+      "grad_norm": 8.31131362915039,
       "learning_rate": 6.4384384384384394e-06,
-      "loss": 1.1301,
+      "loss": 1.1318,
       "step": 35580
     },
     {
       "epoch": 106.88,
-      "grad_norm": 6.56885290145874,
+      "grad_norm": 7.231139659881592,
       "learning_rate": 6.437437437437438e-06,
-      "loss": 1.1128,
+      "loss": 1.1106,
       "step": 35590
     },
     {
       "epoch": 106.91,
-      "grad_norm": 7.484628677368164,
+      "grad_norm": 8.20919418334961,
       "learning_rate": 6.4364364364364375e-06,
-      "loss": 1.0219,
+      "loss": 1.0109,
       "step": 35600
     },
     {
       "epoch": 106.94,
-      "grad_norm": 7.203248977661133,
+      "grad_norm": 9.13278865814209,
       "learning_rate": 6.435435435435436e-06,
-      "loss": 1.1,
+      "loss": 1.1041,
       "step": 35610
     },
     {
       "epoch": 106.97,
-      "grad_norm": 6.2799296379089355,
+      "grad_norm": 6.904733180999756,
       "learning_rate": 6.434434434434435e-06,
-      "loss": 1.0519,
+      "loss": 1.0522,
       "step": 35620
     },
     {
       "epoch": 107.0,
-      "grad_norm": 8.173545837402344,
+      "grad_norm": 8.574841499328613,
       "learning_rate": 6.433433433433433e-06,
-      "loss": 1.0535,
+      "loss": 1.0519,
       "step": 35630
     },
     {
       "epoch": 107.0,
-      "eval_accuracy": 0.8026,
-      "eval_loss": 0.6415277123451233,
-      "eval_runtime": 5.8351,
-      "eval_samples_per_second": 1713.759,
-      "eval_steps_per_second": 6.855,
+      "eval_accuracy": 0.8041,
+      "eval_loss": 0.6423982977867126,
+      "eval_runtime": 5.3614,
+      "eval_samples_per_second": 1865.2,
+      "eval_steps_per_second": 7.461,
       "step": 35631
     },
     {
       "epoch": 107.03,
-      "grad_norm": 4.842329978942871,
+      "grad_norm": 5.603639602661133,
       "learning_rate": 6.432432432432433e-06,
-      "loss": 1.2787,
+      "loss": 1.2613,
       "step": 35640
     },
     {
       "epoch": 107.06,
-      "grad_norm": 8.067398071289062,
+      "grad_norm": 7.765013694763184,
       "learning_rate": 6.431431431431432e-06,
-      "loss": 1.1504,
+      "loss": 1.1517,
       "step": 35650
     },
     {
       "epoch": 107.09,
-      "grad_norm": 6.958824634552002,
+      "grad_norm": 7.691516876220703,
       "learning_rate": 6.430430430430431e-06,
-      "loss": 1.1598,
+      "loss": 1.1572,
       "step": 35660
     },
     {
       "epoch": 107.12,
-      "grad_norm": 4.644047260284424,
+      "grad_norm": 6.271618366241455,
       "learning_rate": 6.42942942942943e-06,
-      "loss": 1.1373,
+      "loss": 1.13,
       "step": 35670
     },
     {
       "epoch": 107.15,
-      "grad_norm": 6.135988235473633,
+      "grad_norm": 6.347027778625488,
       "learning_rate": 6.42842842842843e-06,
-      "loss": 1.1317,
+      "loss": 1.126,
       "step": 35680
     },
     {
       "epoch": 107.18,
-      "grad_norm": 9.20081615447998,
+      "grad_norm": 11.334532737731934,
       "learning_rate": 6.427427427427428e-06,
-      "loss": 1.1227,
+      "loss": 1.1042,
       "step": 35690
     },
     {
       "epoch": 107.21,
-      "grad_norm": 5.689577102661133,
+      "grad_norm": 6.8818817138671875,
       "learning_rate": 6.426426426426427e-06,
-      "loss": 1.1357,
+      "loss": 1.1282,
       "step": 35700
     },
     {
       "epoch": 107.24,
-      "grad_norm": 14.354471206665039,
+      "grad_norm": 17.077898025512695,
       "learning_rate": 6.425425425425425e-06,
-      "loss": 1.1644,
+      "loss": 1.1669,
       "step": 35710
     },
     {
       "epoch": 107.27,
-      "grad_norm": 5.695476531982422,
+      "grad_norm": 6.594638824462891,
       "learning_rate": 6.424424424424425e-06,
-      "loss": 1.1131,
+      "loss": 1.1299,
       "step": 35720
     },
     {
       "epoch": 107.3,
-      "grad_norm": 7.414490699768066,
+      "grad_norm": 6.911393165588379,
       "learning_rate": 6.4234234234234234e-06,
-      "loss": 1.0802,
+      "loss": 1.0778,
       "step": 35730
     },
     {
       "epoch": 107.33,
-      "grad_norm": 8.130237579345703,
+      "grad_norm": 8.31602954864502,
       "learning_rate": 6.422422422422423e-06,
-      "loss": 1.0592,
+      "loss": 1.0628,
       "step": 35740
     },
     {
       "epoch": 107.36,
-      "grad_norm": 6.244762897491455,
+      "grad_norm": 7.348419189453125,
       "learning_rate": 6.421421421421422e-06,
-      "loss": 1.1552,
+      "loss": 1.1557,
       "step": 35750
     },
     {
       "epoch": 107.39,
-      "grad_norm": 8.520709037780762,
+      "grad_norm": 10.013178825378418,
       "learning_rate": 6.420420420420421e-06,
-      "loss": 1.0854,
+      "loss": 1.0721,
       "step": 35760
     },
     {
       "epoch": 107.42,
-      "grad_norm": 7.613659858703613,
+      "grad_norm": 7.506481647491455,
       "learning_rate": 6.41941941941942e-06,
-      "loss": 1.0626,
+      "loss": 1.0581,
       "step": 35770
     },
     {
       "epoch": 107.45,
-      "grad_norm": 5.382191181182861,
+      "grad_norm": 7.080164432525635,
       "learning_rate": 6.418418418418418e-06,
-      "loss": 1.0129,
+      "loss": 1.0254,
       "step": 35780
     },
     {
       "epoch": 107.48,
-      "grad_norm": 9.699453353881836,
+      "grad_norm": 9.005425453186035,
       "learning_rate": 6.4174174174174176e-06,
-      "loss": 1.0659,
+      "loss": 1.0656,
       "step": 35790
     },
     {
       "epoch": 107.51,
-      "grad_norm": 6.800751686096191,
+      "grad_norm": 7.859724521636963,
       "learning_rate": 6.416416416416417e-06,
-      "loss": 1.128,
+      "loss": 1.1397,
       "step": 35800
     },
     {
       "epoch": 107.54,
-      "grad_norm": 6.885721683502197,
+      "grad_norm": 6.384864330291748,
       "learning_rate": 6.415415415415416e-06,
-      "loss": 1.0737,
+      "loss": 1.0868,
       "step": 35810
     },
     {
       "epoch": 107.57,
-      "grad_norm": 7.165224075317383,
+      "grad_norm": 6.8480658531188965,
       "learning_rate": 6.414414414414415e-06,
-      "loss": 1.09,
+      "loss": 1.0955,
       "step": 35820
     },
     {
       "epoch": 107.6,
-      "grad_norm": 7.5895256996154785,
+      "grad_norm": 6.551271438598633,
       "learning_rate": 6.4134134134134145e-06,
-      "loss": 1.1598,
+      "loss": 1.1633,
       "step": 35830
     },
     {
       "epoch": 107.63,
-      "grad_norm": 5.34323263168335,
+      "grad_norm": 7.4410624504089355,
       "learning_rate": 6.412412412412413e-06,
-      "loss": 1.073,
+      "loss": 1.0758,
       "step": 35840
     },
     {
       "epoch": 107.66,
-      "grad_norm": 9.147034645080566,
+      "grad_norm": 6.792202949523926,
       "learning_rate": 6.4114114114114125e-06,
-      "loss": 1.0769,
+      "loss": 1.0673,
       "step": 35850
     },
     {
       "epoch": 107.69,
-      "grad_norm": 6.607875823974609,
+      "grad_norm": 8.488973617553711,
       "learning_rate": 6.410410410410411e-06,
-      "loss": 1.1789,
+      "loss": 1.1791,
       "step": 35860
     },
     {
       "epoch": 107.72,
-      "grad_norm": 7.623050689697266,
+      "grad_norm": 8.032159805297852,
       "learning_rate": 6.40940940940941e-06,
-      "loss": 1.044,
+      "loss": 1.0359,
       "step": 35870
     },
     {
       "epoch": 107.75,
-      "grad_norm": 5.531670570373535,
+      "grad_norm": 5.347622394561768,
       "learning_rate": 6.408408408408408e-06,
-      "loss": 1.1044,
+      "loss": 1.1028,
       "step": 35880
     },
     {
       "epoch": 107.78,
-      "grad_norm": 6.586122989654541,
+      "grad_norm": 7.600484371185303,
       "learning_rate": 6.407407407407408e-06,
-      "loss": 1.1212,
+      "loss": 1.1164,
       "step": 35890
     },
     {
       "epoch": 107.81,
-      "grad_norm": 6.149862289428711,
+      "grad_norm": 6.155834197998047,
       "learning_rate": 6.406406406406407e-06,
-      "loss": 1.1322,
+      "loss": 1.1319,
       "step": 35900
     },
     {
       "epoch": 107.84,
-      "grad_norm": 5.504378795623779,
+      "grad_norm": 6.506203651428223,
       "learning_rate": 6.405405405405406e-06,
-      "loss": 1.0875,
+      "loss": 1.0849,
       "step": 35910
     },
     {
       "epoch": 107.87,
-      "grad_norm": 5.606842994689941,
+      "grad_norm": 6.126054286956787,
       "learning_rate": 6.404404404404405e-06,
-      "loss": 1.0925,
+      "loss": 1.089,
       "step": 35920
     },
     {
       "epoch": 107.9,
-      "grad_norm": 5.223085880279541,
+      "grad_norm": 6.230494976043701,
       "learning_rate": 6.403403403403405e-06,
-      "loss": 1.0627,
+      "loss": 1.0683,
       "step": 35930
     },
     {
       "epoch": 107.93,
-      "grad_norm": 5.485172748565674,
+      "grad_norm": 5.9926252365112305,
       "learning_rate": 6.402402402402403e-06,
-      "loss": 1.1509,
+      "loss": 1.1487,
       "step": 35940
     },
     {
       "epoch": 107.96,
-      "grad_norm": 8.089557647705078,
+      "grad_norm": 7.572860240936279,
       "learning_rate": 6.401401401401402e-06,
-      "loss": 1.1317,
+      "loss": 1.1275,
       "step": 35950
     },
     {
       "epoch": 107.99,
-      "grad_norm": 6.666469573974609,
+      "grad_norm": 6.66692590713501,
       "learning_rate": 6.4004004004004004e-06,
-      "loss": 1.0708,
+      "loss": 1.0561,
       "step": 35960
     },
     {
       "epoch": 108.0,
-      "eval_accuracy": 0.8006,
-      "eval_loss": 0.6415180563926697,
-      "eval_runtime": 5.3999,
-      "eval_samples_per_second": 1851.881,
-      "eval_steps_per_second": 7.408,
+      "eval_accuracy": 0.8014,
+      "eval_loss": 0.6423989534378052,
+      "eval_runtime": 5.415,
+      "eval_samples_per_second": 1846.731,
+      "eval_steps_per_second": 7.387,
       "step": 35964
     },
     {
       "epoch": 108.02,
-      "grad_norm": 5.427980899810791,
+      "grad_norm": 7.998316764831543,
       "learning_rate": 6.3993993993994e-06,
-      "loss": 1.2935,
+      "loss": 1.3019,
       "step": 35970
     },
     {
       "epoch": 108.05,
-      "grad_norm": 6.176517486572266,
+      "grad_norm": 8.812718391418457,
       "learning_rate": 6.3983983983983985e-06,
-      "loss": 1.0708,
+      "loss": 1.0701,
       "step": 35980
     },
     {
       "epoch": 108.08,
-      "grad_norm": 5.631108283996582,
+      "grad_norm": 6.278778553009033,
       "learning_rate": 6.397397397397398e-06,
-      "loss": 1.1199,
+      "loss": 1.122,
       "step": 35990
     },
     {
       "epoch": 108.11,
-      "grad_norm": 8.144123077392578,
+      "grad_norm": 7.845185279846191,
       "learning_rate": 6.396396396396397e-06,
-      "loss": 1.0787,
+      "loss": 1.0795,
       "step": 36000
     },
     {
       "epoch": 108.14,
-      "grad_norm": 8.539996147155762,
+      "grad_norm": 7.439271926879883,
       "learning_rate": 6.395395395395396e-06,
-      "loss": 1.0979,
+      "loss": 1.0993,
       "step": 36010
     },
     {
       "epoch": 108.17,
-      "grad_norm": 7.117108345031738,
+      "grad_norm": 6.298420429229736,
       "learning_rate": 6.394394394394395e-06,
-      "loss": 1.1246,
+      "loss": 1.1172,
       "step": 36020
     },
     {
       "epoch": 108.2,
-      "grad_norm": 9.343645095825195,
+      "grad_norm": 10.313017845153809,
       "learning_rate": 6.393393393393393e-06,
-      "loss": 1.0596,
+      "loss": 1.0547,
       "step": 36030
     },
     {
       "epoch": 108.23,
-      "grad_norm": 8.216552734375,
+      "grad_norm": 7.427731037139893,
       "learning_rate": 6.392392392392393e-06,
-      "loss": 1.0887,
+      "loss": 1.0866,
       "step": 36040
     },
     {
       "epoch": 108.26,
-      "grad_norm": 8.308598518371582,
+      "grad_norm": 8.01746654510498,
       "learning_rate": 6.391391391391392e-06,
-      "loss": 1.0898,
+      "loss": 1.078,
       "step": 36050
     },
     {
       "epoch": 108.29,
-      "grad_norm": 9.685263633728027,
+      "grad_norm": 6.964483261108398,
       "learning_rate": 6.390390390390391e-06,
-      "loss": 1.1207,
+      "loss": 1.1266,
       "step": 36060
     },
     {
       "epoch": 108.32,
-      "grad_norm": 7.764200687408447,
+      "grad_norm": 7.580156326293945,
       "learning_rate": 6.38938938938939e-06,
-      "loss": 1.0856,
+      "loss": 1.0808,
       "step": 36070
     },
     {
       "epoch": 108.35,
-      "grad_norm": 7.031552791595459,
+      "grad_norm": 7.57322883605957,
       "learning_rate": 6.388388388388389e-06,
-      "loss": 1.1814,
+      "loss": 1.1751,
       "step": 36080
     },
     {
       "epoch": 108.38,
-      "grad_norm": 6.76999044418335,
+      "grad_norm": 6.028384685516357,
       "learning_rate": 6.387387387387388e-06,
-      "loss": 1.0776,
+      "loss": 1.088,
       "step": 36090
     },
     {
       "epoch": 108.41,
-      "grad_norm": 6.8597798347473145,
+      "grad_norm": 7.225150108337402,
       "learning_rate": 6.3863863863863875e-06,
-      "loss": 1.0649,
+      "loss": 1.0694,
       "step": 36100
     },
     {
       "epoch": 108.44,
-      "grad_norm": 6.974178314208984,
+      "grad_norm": 5.483529090881348,
       "learning_rate": 6.385385385385386e-06,
-      "loss": 1.1124,
+      "loss": 1.1038,
       "step": 36110
     },
     {
       "epoch": 108.47,
-      "grad_norm": 6.607979774475098,
+      "grad_norm": 5.5586838722229,
       "learning_rate": 6.384384384384385e-06,
-      "loss": 1.0419,
+      "loss": 1.043,
       "step": 36120
     },
     {
       "epoch": 108.5,
-      "grad_norm": 6.393781661987305,
+      "grad_norm": 5.790086269378662,
       "learning_rate": 6.383383383383383e-06,
-      "loss": 1.1096,
+      "loss": 1.0984,
       "step": 36130
     },
     {
       "epoch": 108.53,
-      "grad_norm": 5.433977127075195,
+      "grad_norm": 6.1773600578308105,
       "learning_rate": 6.382382382382383e-06,
-      "loss": 1.1886,
+      "loss": 1.1954,
       "step": 36140
     },
     {
       "epoch": 108.56,
-      "grad_norm": 6.569241046905518,
+      "grad_norm": 6.607522487640381,
       "learning_rate": 6.381381381381382e-06,
-      "loss": 1.1468,
+      "loss": 1.1491,
       "step": 36150
     },
     {
       "epoch": 108.59,
-      "grad_norm": 4.776759624481201,
+      "grad_norm": 5.100533485412598,
       "learning_rate": 6.380380380380381e-06,
-      "loss": 1.0796,
+      "loss": 1.0748,
       "step": 36160
     },
     {
       "epoch": 108.62,
-      "grad_norm": 5.166229724884033,
+      "grad_norm": 6.194584846496582,
       "learning_rate": 6.37937937937938e-06,
-      "loss": 1.1008,
+      "loss": 1.0988,
       "step": 36170
     },
     {
       "epoch": 108.65,
-      "grad_norm": 5.689291000366211,
+      "grad_norm": 6.483497142791748,
       "learning_rate": 6.378378378378379e-06,
-      "loss": 1.1692,
+      "loss": 1.1718,
       "step": 36180
     },
     {
       "epoch": 108.68,
-      "grad_norm": 5.751842975616455,
+      "grad_norm": 4.95058536529541,
       "learning_rate": 6.377377377377378e-06,
-      "loss": 1.1264,
+      "loss": 1.1284,
       "step": 36190
     },
     {
       "epoch": 108.71,
-      "grad_norm": 4.459030628204346,
+      "grad_norm": 5.91630744934082,
       "learning_rate": 6.376376376376376e-06,
-      "loss": 1.1121,
+      "loss": 1.1059,
       "step": 36200
     },
     {
       "epoch": 108.74,
-      "grad_norm": 7.6232147216796875,
+      "grad_norm": 6.076160430908203,
       "learning_rate": 6.3753753753753755e-06,
-      "loss": 1.096,
+      "loss": 1.0987,
       "step": 36210
     },
     {
       "epoch": 108.77,
-      "grad_norm": 7.053699493408203,
+      "grad_norm": 10.290274620056152,
       "learning_rate": 6.374374374374375e-06,
-      "loss": 1.1445,
+      "loss": 1.1427,
       "step": 36220
     },
     {
       "epoch": 108.8,
-      "grad_norm": 9.381733894348145,
+      "grad_norm": 8.345067024230957,
       "learning_rate": 6.3733733733733735e-06,
-      "loss": 1.0941,
+      "loss": 1.0915,
       "step": 36230
     },
     {
       "epoch": 108.83,
-      "grad_norm": 4.854085922241211,
+      "grad_norm": 4.56545352935791,
       "learning_rate": 6.372372372372373e-06,
-      "loss": 1.0934,
+      "loss": 1.1044,
       "step": 36240
     },
     {
       "epoch": 108.86,
-      "grad_norm": 7.329185962677002,
+      "grad_norm": 10.677928924560547,
       "learning_rate": 6.371371371371372e-06,
-      "loss": 1.0828,
+      "loss": 1.0805,
       "step": 36250
     },
     {
       "epoch": 108.89,
-      "grad_norm": 6.114192485809326,
+      "grad_norm": 5.04295015335083,
       "learning_rate": 6.370370370370371e-06,
-      "loss": 1.0942,
+      "loss": 1.0914,
       "step": 36260
     },
     {
       "epoch": 108.92,
-      "grad_norm": 4.39804744720459,
+      "grad_norm": 4.627781391143799,
       "learning_rate": 6.3693693693693704e-06,
-      "loss": 1.1062,
+      "loss": 1.1192,
       "step": 36270
     },
     {
       "epoch": 108.95,
-      "grad_norm": 5.842523097991943,
+      "grad_norm": 9.21884536743164,
       "learning_rate": 6.368368368368368e-06,
-      "loss": 1.1337,
+      "loss": 1.132,
       "step": 36280
     },
     {
       "epoch": 108.98,
-      "grad_norm": 5.752237796783447,
+      "grad_norm": 5.534067153930664,
       "learning_rate": 6.367367367367368e-06,
-      "loss": 1.0657,
+      "loss": 1.0709,
       "step": 36290
     },
     {
       "epoch": 109.0,
-      "eval_accuracy": 0.8033,
-      "eval_loss": 0.6397601366043091,
-      "eval_runtime": 5.3713,
-      "eval_samples_per_second": 1861.736,
-      "eval_steps_per_second": 7.447,
+      "eval_accuracy": 0.8035,
+      "eval_loss": 0.6389144659042358,
+      "eval_runtime": 5.2586,
+      "eval_samples_per_second": 1901.632,
+      "eval_steps_per_second": 7.607,
       "step": 36297
     },
     {
       "epoch": 109.01,
-      "grad_norm": 7.805610179901123,
+      "grad_norm": 9.162347793579102,
       "learning_rate": 6.366366366366366e-06,
-      "loss": 1.3708,
+      "loss": 1.3642,
       "step": 36300
     },
     {
       "epoch": 109.04,
-      "grad_norm": 8.266752243041992,
+      "grad_norm": 7.899369239807129,
       "learning_rate": 6.365365365365366e-06,
-      "loss": 1.0967,
+      "loss": 1.0953,
       "step": 36310
     },
     {
       "epoch": 109.07,
-      "grad_norm": 7.739986896514893,
+      "grad_norm": 10.528433799743652,
       "learning_rate": 6.364364364364365e-06,
-      "loss": 1.0588,
+      "loss": 1.058,
       "step": 36320
     },
     {
       "epoch": 109.1,
-      "grad_norm": 6.682123184204102,
+      "grad_norm": 6.7444539070129395,
       "learning_rate": 6.363363363363364e-06,
-      "loss": 1.1828,
+      "loss": 1.1783,
       "step": 36330
     },
     {
       "epoch": 109.13,
-      "grad_norm": 6.244485855102539,
+      "grad_norm": 6.283884048461914,
       "learning_rate": 6.362362362362363e-06,
-      "loss": 1.097,
+      "loss": 1.0992,
       "step": 36340
     },
     {
       "epoch": 109.16,
-      "grad_norm": 8.481616973876953,
+      "grad_norm": 7.256768703460693,
       "learning_rate": 6.361361361361363e-06,
-      "loss": 1.0802,
+      "loss": 1.07,
       "step": 36350
     },
     {
       "epoch": 109.19,
-      "grad_norm": 6.550126552581787,
+      "grad_norm": 6.9743475914001465,
       "learning_rate": 6.360360360360361e-06,
-      "loss": 1.0781,
+      "loss": 1.0752,
       "step": 36360
     },
     {
       "epoch": 109.22,
-      "grad_norm": 6.280527591705322,
+      "grad_norm": 7.33081579208374,
       "learning_rate": 6.35935935935936e-06,
-      "loss": 1.0706,
+      "loss": 1.0815,
       "step": 36370
     },
     {
       "epoch": 109.25,
-      "grad_norm": 8.174455642700195,
+      "grad_norm": 7.85202169418335,
       "learning_rate": 6.358358358358358e-06,
-      "loss": 1.0718,
+      "loss": 1.0879,
       "step": 36380
     },
     {
       "epoch": 109.28,
-      "grad_norm": 6.522738456726074,
+      "grad_norm": 6.5428876876831055,
       "learning_rate": 6.357357357357358e-06,
-      "loss": 1.1692,
+      "loss": 1.1715,
       "step": 36390
     },
     {
       "epoch": 109.31,
-      "grad_norm": 5.5773844718933105,
+      "grad_norm": 6.118473529815674,
       "learning_rate": 6.356356356356357e-06,
-      "loss": 1.0987,
+      "loss": 1.0861,
       "step": 36400
     },
     {
       "epoch": 109.34,
-      "grad_norm": 5.563918590545654,
+      "grad_norm": 5.79777717590332,
       "learning_rate": 6.355355355355356e-06,
-      "loss": 1.0824,
+      "loss": 1.07,
       "step": 36410
     },
     {
       "epoch": 109.37,
-      "grad_norm": 5.088877201080322,
+      "grad_norm": 6.082858085632324,
       "learning_rate": 6.354354354354355e-06,
-      "loss": 1.1296,
+      "loss": 1.1218,
       "step": 36420
     },
     {
       "epoch": 109.4,
-      "grad_norm": 5.7985758781433105,
+      "grad_norm": 4.763041973114014,
       "learning_rate": 6.353353353353354e-06,
-      "loss": 1.1281,
+      "loss": 1.1254,
       "step": 36430
     },
     {
       "epoch": 109.43,
-      "grad_norm": 9.031242370605469,
+      "grad_norm": 7.68549919128418,
       "learning_rate": 6.352352352352353e-06,
-      "loss": 1.0981,
+      "loss": 1.096,
       "step": 36440
     },
     {
       "epoch": 109.46,
-      "grad_norm": 6.58619499206543,
+      "grad_norm": 7.972354888916016,
       "learning_rate": 6.351351351351351e-06,
-      "loss": 1.102,
+      "loss": 1.1081,
       "step": 36450
     },
     {
       "epoch": 109.49,
-      "grad_norm": 8.549145698547363,
+      "grad_norm": 5.97420072555542,
       "learning_rate": 6.3503503503503505e-06,
-      "loss": 1.106,
+      "loss": 1.1156,
       "step": 36460
     },
     {
       "epoch": 109.52,
-      "grad_norm": 8.710298538208008,
+      "grad_norm": 8.250345230102539,
       "learning_rate": 6.34934934934935e-06,
-      "loss": 1.0844,
+      "loss": 1.0886,
       "step": 36470
     },
     {
       "epoch": 109.55,
-      "grad_norm": 6.683289051055908,
+      "grad_norm": 8.172926902770996,
       "learning_rate": 6.3483483483483485e-06,
-      "loss": 1.1687,
+      "loss": 1.1778,
       "step": 36480
     },
     {
       "epoch": 109.58,
-      "grad_norm": 6.353591442108154,
+      "grad_norm": 5.680255889892578,
       "learning_rate": 6.347347347347348e-06,
-      "loss": 1.0476,
+      "loss": 1.0452,
       "step": 36490
     },
     {
       "epoch": 109.61,
-      "grad_norm": 7.020476818084717,
+      "grad_norm": 6.061624526977539,
       "learning_rate": 6.3463463463463474e-06,
-      "loss": 1.0965,
+      "loss": 1.0929,
       "step": 36500
     },
     {
       "epoch": 109.64,
-      "grad_norm": 5.454484462738037,
+      "grad_norm": 6.673813343048096,
       "learning_rate": 6.345345345345346e-06,
-      "loss": 1.0649,
+      "loss": 1.0651,
       "step": 36510
     },
     {
       "epoch": 109.67,
-      "grad_norm": 6.515374183654785,
+      "grad_norm": 6.4644036293029785,
       "learning_rate": 6.3443443443443455e-06,
-      "loss": 1.0564,
+      "loss": 1.06,
       "step": 36520
     },
     {
       "epoch": 109.7,
-      "grad_norm": 6.403848648071289,
+      "grad_norm": 6.142782211303711,
       "learning_rate": 6.343343343343343e-06,
-      "loss": 1.0369,
+      "loss": 1.0223,
       "step": 36530
     },
     {
       "epoch": 109.73,
-      "grad_norm": 6.040570259094238,
+      "grad_norm": 6.409983158111572,
       "learning_rate": 6.342342342342343e-06,
-      "loss": 1.129,
+      "loss": 1.1312,
       "step": 36540
     },
     {
       "epoch": 109.76,
-      "grad_norm": 6.3075714111328125,
+      "grad_norm": 5.911462306976318,
       "learning_rate": 6.341341341341341e-06,
-      "loss": 1.0328,
+      "loss": 1.0213,
       "step": 36550
     },
     {
       "epoch": 109.79,
-      "grad_norm": 7.106643199920654,
+      "grad_norm": 5.667751789093018,
       "learning_rate": 6.340340340340341e-06,
-      "loss": 1.0796,
+      "loss": 1.0799,
       "step": 36560
     },
     {
       "epoch": 109.82,
-      "grad_norm": 6.258601188659668,
+      "grad_norm": 6.286153793334961,
       "learning_rate": 6.33933933933934e-06,
-      "loss": 1.0947,
+      "loss": 1.0896,
       "step": 36570
     },
     {
       "epoch": 109.85,
-      "grad_norm": 7.344061851501465,
+      "grad_norm": 7.923434257507324,
       "learning_rate": 6.338338338338339e-06,
-      "loss": 1.0808,
+      "loss": 1.0749,
       "step": 36580
     },
     {
       "epoch": 109.88,
-      "grad_norm": 4.420611381530762,
+      "grad_norm": 4.676229953765869,
       "learning_rate": 6.337337337337338e-06,
-      "loss": 1.173,
+      "loss": 1.1735,
       "step": 36590
     },
     {
       "epoch": 109.91,
-      "grad_norm": 6.692820072174072,
+      "grad_norm": 7.940841197967529,
       "learning_rate": 6.336336336336338e-06,
-      "loss": 1.1961,
+      "loss": 1.2063,
       "step": 36600
     },
     {
       "epoch": 109.94,
-      "grad_norm": 6.285885334014893,
+      "grad_norm": 6.169107437133789,
       "learning_rate": 6.335335335335336e-06,
-      "loss": 1.0732,
+      "loss": 1.0582,
       "step": 36610
     },
     {
       "epoch": 109.97,
-      "grad_norm": 7.030479431152344,
+      "grad_norm": 9.402519226074219,
       "learning_rate": 6.334334334334335e-06,
-      "loss": 1.1041,
+      "loss": 1.0965,
       "step": 36620
     },
     {
       "epoch": 110.0,
-      "grad_norm": 61.00059509277344,
+      "grad_norm": 49.66619873046875,
       "learning_rate": 6.333333333333333e-06,
-      "loss": 1.1575,
+      "loss": 1.1728,
       "step": 36630
     },
     {
       "epoch": 110.0,
-      "eval_accuracy": 0.802,
-      "eval_loss": 0.6462051868438721,
-      "eval_runtime": 5.2956,
-      "eval_samples_per_second": 1888.347,
-      "eval_steps_per_second": 7.553,
+      "eval_accuracy": 0.8029,
+      "eval_loss": 0.6464499235153198,
+      "eval_runtime": 5.1272,
+      "eval_samples_per_second": 1950.394,
+      "eval_steps_per_second": 7.802,
       "step": 36630
     },
     {
       "epoch": 110.03,
-      "grad_norm": 5.53280782699585,
+      "grad_norm": 7.25161075592041,
       "learning_rate": 6.332332332332333e-06,
-      "loss": 1.0454,
+      "loss": 1.0449,
       "step": 36640
     },
     {
       "epoch": 110.06,
-      "grad_norm": 6.343801498413086,
+      "grad_norm": 6.51585054397583,
       "learning_rate": 6.3313313313313314e-06,
-      "loss": 1.0945,
+      "loss": 1.0892,
       "step": 36650
     },
     {
       "epoch": 110.09,
-      "grad_norm": 4.960554122924805,
+      "grad_norm": 6.402839660644531,
       "learning_rate": 6.330330330330331e-06,
-      "loss": 1.0799,
+      "loss": 1.08,
       "step": 36660
     },
     {
       "epoch": 110.12,
-      "grad_norm": 8.975944519042969,
+      "grad_norm": 8.915799140930176,
       "learning_rate": 6.32932932932933e-06,
-      "loss": 1.2166,
+      "loss": 1.2218,
       "step": 36670
     },
     {
       "epoch": 110.15,
-      "grad_norm": 7.134684085845947,
+      "grad_norm": 6.735911846160889,
       "learning_rate": 6.328328328328329e-06,
-      "loss": 1.0754,
+      "loss": 1.0844,
       "step": 36680
     },
     {
       "epoch": 110.18,
-      "grad_norm": 5.809741973876953,
+      "grad_norm": 5.661557197570801,
       "learning_rate": 6.327327327327328e-06,
-      "loss": 1.1368,
+      "loss": 1.1356,
       "step": 36690
     },
     {
       "epoch": 110.21,
-      "grad_norm": 6.0756611824035645,
+      "grad_norm": 9.706396102905273,
       "learning_rate": 6.326326326326326e-06,
-      "loss": 1.204,
+      "loss": 1.1977,
       "step": 36700
     },
     {
       "epoch": 110.24,
-      "grad_norm": 9.7136869430542,
+      "grad_norm": 10.235123634338379,
       "learning_rate": 6.3253253253253255e-06,
-      "loss": 1.0674,
+      "loss": 1.0769,
       "step": 36710
     },
     {
       "epoch": 110.27,
-      "grad_norm": 5.2592949867248535,
+      "grad_norm": 7.408087730407715,
       "learning_rate": 6.324324324324325e-06,
-      "loss": 1.1288,
+      "loss": 1.1352,
       "step": 36720
     },
     {
       "epoch": 110.3,
-      "grad_norm": 6.711035251617432,
+      "grad_norm": 7.543276786804199,
       "learning_rate": 6.323323323323324e-06,
-      "loss": 1.1791,
+      "loss": 1.1802,
       "step": 36730
     },
     {
       "epoch": 110.33,
-      "grad_norm": 7.875670433044434,
+      "grad_norm": 7.065732955932617,
       "learning_rate": 6.322322322322323e-06,
-      "loss": 1.1311,
+      "loss": 1.1109,
       "step": 36740
     },
     {
       "epoch": 110.36,
-      "grad_norm": 6.852148056030273,
+      "grad_norm": 9.922021865844727,
       "learning_rate": 6.321321321321322e-06,
-      "loss": 1.1084,
+      "loss": 1.096,
       "step": 36750
     },
     {
       "epoch": 110.39,
-      "grad_norm": 7.046970367431641,
+      "grad_norm": 5.9354023933410645,
       "learning_rate": 6.320320320320321e-06,
-      "loss": 1.0594,
+      "loss": 1.0635,
       "step": 36760
     },
     {
       "epoch": 110.42,
-      "grad_norm": 6.30199670791626,
+      "grad_norm": 7.868113040924072,
       "learning_rate": 6.3193193193193205e-06,
-      "loss": 1.1653,
+      "loss": 1.1528,
       "step": 36770
     },
     {
       "epoch": 110.45,
-      "grad_norm": 5.556674480438232,
+      "grad_norm": 7.957797050476074,
       "learning_rate": 6.318318318318318e-06,
-      "loss": 1.0955,
+      "loss": 1.0993,
       "step": 36780
     },
     {
       "epoch": 110.48,
-      "grad_norm": 5.796895503997803,
+      "grad_norm": 6.614912986755371,
       "learning_rate": 6.317317317317318e-06,
-      "loss": 1.2169,
+      "loss": 1.209,
       "step": 36790
     },
     {
       "epoch": 110.51,
-      "grad_norm": 7.7267069816589355,
+      "grad_norm": 10.574417114257812,
       "learning_rate": 6.316316316316316e-06,
-      "loss": 1.124,
+      "loss": 1.1244,
       "step": 36800
     },
     {
       "epoch": 110.54,
-      "grad_norm": 6.742910385131836,
+      "grad_norm": 7.5989508628845215,
       "learning_rate": 6.315315315315316e-06,
-      "loss": 1.1123,
+      "loss": 1.111,
       "step": 36810
     },
     {
       "epoch": 110.57,
-      "grad_norm": 8.342729568481445,
+      "grad_norm": 7.102509498596191,
       "learning_rate": 6.314314314314315e-06,
-      "loss": 1.1028,
+      "loss": 1.1025,
       "step": 36820
     },
     {
       "epoch": 110.6,
-      "grad_norm": 5.269210338592529,
+      "grad_norm": 4.749288558959961,
       "learning_rate": 6.313313313313314e-06,
-      "loss": 1.1227,
+      "loss": 1.126,
       "step": 36830
     },
     {
       "epoch": 110.63,
-      "grad_norm": 8.1673583984375,
+      "grad_norm": 7.076495170593262,
       "learning_rate": 6.312312312312313e-06,
-      "loss": 1.059,
+      "loss": 1.0546,
       "step": 36840
     },
     {
       "epoch": 110.66,
-      "grad_norm": 6.926266193389893,
+      "grad_norm": 6.810534477233887,
       "learning_rate": 6.311311311311313e-06,
-      "loss": 1.0287,
+      "loss": 1.02,
       "step": 36850
     },
     {
       "epoch": 110.69,
-      "grad_norm": 6.527507781982422,
+      "grad_norm": 8.907082557678223,
       "learning_rate": 6.31031031031031e-06,
-      "loss": 1.0933,
+      "loss": 1.0756,
       "step": 36860
     },
     {
       "epoch": 110.72,
-      "grad_norm": 7.00398588180542,
+      "grad_norm": 6.7824249267578125,
       "learning_rate": 6.309309309309309e-06,
-      "loss": 1.0611,
+      "loss": 1.0569,
       "step": 36870
     },
     {
       "epoch": 110.75,
-      "grad_norm": 8.484260559082031,
+      "grad_norm": 9.075421333312988,
       "learning_rate": 6.3083083083083084e-06,
-      "loss": 1.1713,
+      "loss": 1.1576,
       "step": 36880
     },
     {
       "epoch": 110.78,
-      "grad_norm": 7.720708847045898,
+      "grad_norm": 6.882009506225586,
       "learning_rate": 6.307307307307308e-06,
-      "loss": 1.1019,
+      "loss": 1.0885,
       "step": 36890
     },
     {
       "epoch": 110.81,
-      "grad_norm": 6.870096683502197,
+      "grad_norm": 7.002157211303711,
       "learning_rate": 6.3063063063063065e-06,
-      "loss": 1.0861,
+      "loss": 1.0878,
       "step": 36900
     },
     {
       "epoch": 110.84,
-      "grad_norm": 7.794412136077881,
+      "grad_norm": 7.368155479431152,
       "learning_rate": 6.305305305305306e-06,
-      "loss": 1.06,
+      "loss": 1.0505,
       "step": 36910
     },
     {
       "epoch": 110.87,
-      "grad_norm": 6.999216079711914,
+      "grad_norm": 6.0107035636901855,
       "learning_rate": 6.304304304304305e-06,
-      "loss": 1.1077,
+      "loss": 1.1028,
       "step": 36920
     },
     {
       "epoch": 110.9,
-      "grad_norm": 6.908900260925293,
+      "grad_norm": 6.471554756164551,
       "learning_rate": 6.303303303303304e-06,
-      "loss": 1.1451,
+      "loss": 1.1458,
       "step": 36930
     },
     {
       "epoch": 110.93,
-      "grad_norm": 6.482051372528076,
+      "grad_norm": 7.1880693435668945,
       "learning_rate": 6.302302302302303e-06,
-      "loss": 1.1159,
+      "loss": 1.1136,
       "step": 36940
     },
     {
       "epoch": 110.96,
-      "grad_norm": 7.040231227874756,
+      "grad_norm": 7.7183942794799805,
       "learning_rate": 6.301301301301301e-06,
-      "loss": 1.094,
+      "loss": 1.1107,
       "step": 36950
     },
     {
       "epoch": 110.99,
-      "grad_norm": 5.166589736938477,
+      "grad_norm": 5.783481121063232,
       "learning_rate": 6.300300300300301e-06,
-      "loss": 1.0678,
+      "loss": 1.0629,
       "step": 36960
     },
     {
       "epoch": 111.0,
-      "eval_accuracy": 0.8028,
-      "eval_loss": 0.6389985084533691,
-      "eval_runtime": 5.6594,
-      "eval_samples_per_second": 1766.961,
-      "eval_steps_per_second": 7.068,
+      "eval_accuracy": 0.8051,
+      "eval_loss": 0.6364142298698425,
+      "eval_runtime": 5.3104,
+      "eval_samples_per_second": 1883.081,
+      "eval_steps_per_second": 7.532,
       "step": 36963
     },
     {
       "epoch": 111.02,
-      "grad_norm": 5.1207733154296875,
+      "grad_norm": 7.304844379425049,
       "learning_rate": 6.2992992992993e-06,
-      "loss": 1.2503,
+      "loss": 1.276,
       "step": 36970
     },
     {
       "epoch": 111.05,
-      "grad_norm": 7.386695384979248,
+      "grad_norm": 9.198808670043945,
       "learning_rate": 6.298298298298299e-06,
-      "loss": 0.9965,
+      "loss": 0.98,
       "step": 36980
     },
     {
       "epoch": 111.08,
-      "grad_norm": 6.669808864593506,
+      "grad_norm": 8.134875297546387,
       "learning_rate": 6.297297297297298e-06,
-      "loss": 1.0941,
+      "loss": 1.0834,
       "step": 36990
     },
     {
       "epoch": 111.11,
-      "grad_norm": 15.443010330200195,
+      "grad_norm": 9.584543228149414,
       "learning_rate": 6.296296296296297e-06,
-      "loss": 1.0511,
+      "loss": 1.062,
       "step": 37000
     },
     {
       "epoch": 111.14,
-      "grad_norm": 7.720875263214111,
+      "grad_norm": 8.382852554321289,
       "learning_rate": 6.295295295295296e-06,
-      "loss": 1.1033,
+      "loss": 1.116,
       "step": 37010
     },
     {
       "epoch": 111.17,
-      "grad_norm": 6.667705535888672,
+      "grad_norm": 6.604728698730469,
       "learning_rate": 6.2942942942942955e-06,
-      "loss": 1.0928,
+      "loss": 1.0906,
       "step": 37020
     },
     {
       "epoch": 111.2,
-      "grad_norm": 5.233231544494629,
+      "grad_norm": 5.214109420776367,
       "learning_rate": 6.293293293293293e-06,
-      "loss": 0.99,
+      "loss": 0.9853,
       "step": 37030
     },
     {
       "epoch": 111.23,
-      "grad_norm": 6.106170177459717,
+      "grad_norm": 6.34938907623291,
       "learning_rate": 6.292292292292293e-06,
-      "loss": 1.0151,
+      "loss": 1.0085,
       "step": 37040
     },
     {
       "epoch": 111.26,
-      "grad_norm": 7.28203010559082,
+      "grad_norm": 11.316672325134277,
       "learning_rate": 6.291291291291291e-06,
-      "loss": 1.1309,
+      "loss": 1.1341,
       "step": 37050
     },
     {
       "epoch": 111.29,
-      "grad_norm": 5.628142356872559,
+      "grad_norm": 5.532124042510986,
       "learning_rate": 6.290290290290291e-06,
-      "loss": 1.092,
+      "loss": 1.0909,
       "step": 37060
     },
     {
       "epoch": 111.32,
-      "grad_norm": 4.023622512817383,
+      "grad_norm": 4.862917900085449,
       "learning_rate": 6.28928928928929e-06,
-      "loss": 1.0247,
+      "loss": 1.0241,
       "step": 37070
     },
     {
       "epoch": 111.35,
-      "grad_norm": 7.968639850616455,
+      "grad_norm": 10.784178733825684,
       "learning_rate": 6.288288288288289e-06,
-      "loss": 1.0973,
+      "loss": 1.092,
       "step": 37080
     },
     {
       "epoch": 111.38,
-      "grad_norm": 6.851569175720215,
+      "grad_norm": 6.7134623527526855,
       "learning_rate": 6.287287287287288e-06,
-      "loss": 1.1283,
+      "loss": 1.1189,
       "step": 37090
     },
     {
       "epoch": 111.41,
-      "grad_norm": 8.786542892456055,
+      "grad_norm": 6.558589458465576,
       "learning_rate": 6.286286286286287e-06,
-      "loss": 1.1692,
+      "loss": 1.1572,
       "step": 37100
     },
     {
       "epoch": 111.44,
-      "grad_norm": 5.693499565124512,
+      "grad_norm": 5.171597957611084,
       "learning_rate": 6.2852852852852854e-06,
-      "loss": 1.1059,
+      "loss": 1.1122,
       "step": 37110
     },
     {
       "epoch": 111.47,
-      "grad_norm": 6.944270610809326,
+      "grad_norm": 5.89455509185791,
       "learning_rate": 6.284284284284284e-06,
-      "loss": 1.0858,
+      "loss": 1.0852,
       "step": 37120
     },
     {
       "epoch": 111.5,
-      "grad_norm": 6.055338382720947,
+      "grad_norm": 6.097740650177002,
       "learning_rate": 6.2832832832832835e-06,
-      "loss": 1.1202,
+      "loss": 1.1311,
       "step": 37130
     },
     {
       "epoch": 111.53,
-      "grad_norm": 5.085280418395996,
+      "grad_norm": 7.3804755210876465,
       "learning_rate": 6.282282282282283e-06,
-      "loss": 1.038,
+      "loss": 1.0396,
       "step": 37140
     },
     {
       "epoch": 111.56,
-      "grad_norm": 4.993185520172119,
+      "grad_norm": 5.681091785430908,
       "learning_rate": 6.2812812812812815e-06,
-      "loss": 1.1309,
+      "loss": 1.1281,
       "step": 37150
     },
     {
       "epoch": 111.59,
-      "grad_norm": 6.514731407165527,
+      "grad_norm": 7.229230880737305,
       "learning_rate": 6.280280280280281e-06,
-      "loss": 1.1443,
+      "loss": 1.1448,
       "step": 37160
     },
     {
       "epoch": 111.62,
-      "grad_norm": 6.2999773025512695,
+      "grad_norm": 6.980918884277344,
       "learning_rate": 6.27927927927928e-06,
-      "loss": 1.1293,
+      "loss": 1.1226,
       "step": 37170
     },
     {
       "epoch": 111.65,
-      "grad_norm": 8.613157272338867,
+      "grad_norm": 11.804841041564941,
       "learning_rate": 6.278278278278279e-06,
-      "loss": 1.0651,
+      "loss": 1.061,
       "step": 37180
     },
     {
       "epoch": 111.68,
-      "grad_norm": 7.147843360900879,
+      "grad_norm": 8.835192680358887,
       "learning_rate": 6.2772772772772784e-06,
-      "loss": 1.1417,
+      "loss": 1.1332,
       "step": 37190
     },
     {
       "epoch": 111.71,
-      "grad_norm": 5.999594211578369,
+      "grad_norm": 7.086983680725098,
       "learning_rate": 6.276276276276276e-06,
-      "loss": 1.1436,
+      "loss": 1.1417,
       "step": 37200
     },
     {
       "epoch": 111.74,
-      "grad_norm": 6.812475204467773,
+      "grad_norm": 7.575691223144531,
       "learning_rate": 6.275275275275276e-06,
-      "loss": 1.0917,
+      "loss": 1.083,
       "step": 37210
     },
     {
       "epoch": 111.77,
-      "grad_norm": 5.671546936035156,
+      "grad_norm": 4.596778869628906,
       "learning_rate": 6.274274274274274e-06,
-      "loss": 1.1131,
+      "loss": 1.1243,
       "step": 37220
     },
     {
       "epoch": 111.8,
-      "grad_norm": 4.7622857093811035,
+      "grad_norm": 5.485865592956543,
       "learning_rate": 6.273273273273274e-06,
-      "loss": 1.0968,
+      "loss": 1.0886,
       "step": 37230
     },
     {
       "epoch": 111.83,
-      "grad_norm": 5.108748435974121,
+      "grad_norm": 6.432788372039795,
       "learning_rate": 6.272272272272273e-06,
-      "loss": 1.0661,
+      "loss": 1.0505,
       "step": 37240
     },
     {
       "epoch": 111.86,
-      "grad_norm": 5.870114326477051,
+      "grad_norm": 8.057191848754883,
       "learning_rate": 6.271271271271272e-06,
-      "loss": 1.1224,
+      "loss": 1.1237,
       "step": 37250
     },
     {
       "epoch": 111.89,
-      "grad_norm": 7.506541728973389,
+      "grad_norm": 9.707804679870605,
       "learning_rate": 6.270270270270271e-06,
-      "loss": 1.071,
+      "loss": 1.0731,
       "step": 37260
     },
     {
       "epoch": 111.92,
-      "grad_norm": 8.233085632324219,
+      "grad_norm": 9.220712661743164,
       "learning_rate": 6.2692692692692706e-06,
-      "loss": 1.1321,
+      "loss": 1.1367,
       "step": 37270
     },
     {
       "epoch": 111.95,
-      "grad_norm": 6.2397613525390625,
+      "grad_norm": 10.294652938842773,
       "learning_rate": 6.268268268268268e-06,
-      "loss": 1.036,
+      "loss": 1.0345,
       "step": 37280
     },
     {
       "epoch": 111.98,
-      "grad_norm": 10.029104232788086,
+      "grad_norm": 7.708982944488525,
       "learning_rate": 6.267267267267268e-06,
       "loss": 1.1565,
       "step": 37290
     },
     {
       "epoch": 112.0,
-      "eval_accuracy": 0.8012,
-      "eval_loss": 0.6506243944168091,
-      "eval_runtime": 5.3621,
-      "eval_samples_per_second": 1864.93,
-      "eval_steps_per_second": 7.46,
+      "eval_accuracy": 0.8009,
+      "eval_loss": 0.6526205539703369,
+      "eval_runtime": 5.3867,
+      "eval_samples_per_second": 1856.431,
+      "eval_steps_per_second": 7.426,
       "step": 37296
     },
     {
       "epoch": 112.01,
-      "grad_norm": 9.613479614257812,
+      "grad_norm": 8.438828468322754,
       "learning_rate": 6.266266266266266e-06,
-      "loss": 1.1749,
+      "loss": 1.1791,
       "step": 37300
     },
     {
       "epoch": 112.04,
-      "grad_norm": 4.81672477722168,
+      "grad_norm": 6.319471836090088,
       "learning_rate": 6.265265265265266e-06,
-      "loss": 1.1267,
+      "loss": 1.1225,
       "step": 37310
     },
     {
       "epoch": 112.07,
-      "grad_norm": 10.312203407287598,
+      "grad_norm": 10.905848503112793,
       "learning_rate": 6.264264264264264e-06,
-      "loss": 1.0864,
+      "loss": 1.0819,
       "step": 37320
     },
     {
       "epoch": 112.1,
-      "grad_norm": 7.638431072235107,
+      "grad_norm": 8.869301795959473,
       "learning_rate": 6.263263263263264e-06,
-      "loss": 1.1376,
+      "loss": 1.143,
       "step": 37330
     },
     {
       "epoch": 112.13,
-      "grad_norm": 5.647699356079102,
+      "grad_norm": 7.3019280433654785,
       "learning_rate": 6.262262262262263e-06,
-      "loss": 1.1346,
+      "loss": 1.1292,
       "step": 37340
     },
     {
       "epoch": 112.16,
-      "grad_norm": 8.704755783081055,
+      "grad_norm": 7.236865043640137,
       "learning_rate": 6.261261261261262e-06,
-      "loss": 1.0264,
+      "loss": 1.0216,
       "step": 37350
     },
     {
       "epoch": 112.19,
-      "grad_norm": 7.388134956359863,
+      "grad_norm": 9.479453086853027,
       "learning_rate": 6.2602602602602605e-06,
-      "loss": 1.0628,
+      "loss": 1.0583,
       "step": 37360
     },
     {
       "epoch": 112.22,
-      "grad_norm": 5.7229323387146,
+      "grad_norm": 6.082587242126465,
       "learning_rate": 6.259259259259259e-06,
-      "loss": 1.0767,
+      "loss": 1.0789,
       "step": 37370
     },
     {
       "epoch": 112.25,
-      "grad_norm": 8.928657531738281,
+      "grad_norm": 8.293089866638184,
       "learning_rate": 6.2582582582582585e-06,
-      "loss": 1.0637,
+      "loss": 1.0576,
       "step": 37380
     },
     {
       "epoch": 112.28,
-      "grad_norm": 7.340973854064941,
+      "grad_norm": 8.833977699279785,
       "learning_rate": 6.257257257257258e-06,
-      "loss": 1.1696,
+      "loss": 1.1673,
       "step": 37390
     },
     {
       "epoch": 112.31,
-      "grad_norm": 5.499286651611328,
+      "grad_norm": 6.832913875579834,
       "learning_rate": 6.2562562562562565e-06,
-      "loss": 1.1071,
+      "loss": 1.1168,
       "step": 37400
     },
     {
       "epoch": 112.34,
-      "grad_norm": 7.621452808380127,
+      "grad_norm": 6.834465980529785,
       "learning_rate": 6.255255255255256e-06,
-      "loss": 1.0629,
+      "loss": 1.0628,
       "step": 37410
     },
     {
       "epoch": 112.37,
-      "grad_norm": 5.231354236602783,
+      "grad_norm": 5.660340309143066,
       "learning_rate": 6.2542542542542554e-06,
-      "loss": 1.0776,
+      "loss": 1.0635,
       "step": 37420
     },
     {
       "epoch": 112.4,
-      "grad_norm": 7.214157581329346,
+      "grad_norm": 6.073281288146973,
       "learning_rate": 6.253253253253254e-06,
-      "loss": 1.1502,
+      "loss": 1.1366,
       "step": 37430
     },
     {
       "epoch": 112.43,
-      "grad_norm": 6.018527030944824,
+      "grad_norm": 6.008208751678467,
       "learning_rate": 6.2522522522522535e-06,
-      "loss": 1.0117,
+      "loss": 1.013,
       "step": 37440
     },
     {
       "epoch": 112.46,
-      "grad_norm": 8.043930053710938,
+      "grad_norm": 7.49979305267334,
       "learning_rate": 6.251251251251251e-06,
-      "loss": 1.0704,
+      "loss": 1.0758,
       "step": 37450
     },
     {
       "epoch": 112.49,
-      "grad_norm": 6.183081150054932,
+      "grad_norm": 7.017287254333496,
       "learning_rate": 6.250250250250251e-06,
-      "loss": 1.0337,
+      "loss": 1.0173,
       "step": 37460
     },
     {
       "epoch": 112.52,
-      "grad_norm": 7.131575584411621,
+      "grad_norm": 6.007573127746582,
       "learning_rate": 6.249249249249249e-06,
-      "loss": 1.1225,
+      "loss": 1.1149,
       "step": 37470
     },
     {
       "epoch": 112.55,
-      "grad_norm": 9.371749877929688,
+      "grad_norm": 8.523208618164062,
       "learning_rate": 6.248248248248249e-06,
-      "loss": 1.1446,
+      "loss": 1.1494,
       "step": 37480
     },
     {
       "epoch": 112.58,
-      "grad_norm": 7.029998779296875,
+      "grad_norm": 8.921719551086426,
       "learning_rate": 6.247247247247248e-06,
-      "loss": 1.0238,
+      "loss": 1.0202,
       "step": 37490
     },
     {
       "epoch": 112.61,
-      "grad_norm": 5.996323585510254,
+      "grad_norm": 5.703799724578857,
       "learning_rate": 6.246246246246247e-06,
-      "loss": 1.0593,
+      "loss": 1.0512,
       "step": 37500
     },
     {
       "epoch": 112.64,
-      "grad_norm": 5.630310535430908,
+      "grad_norm": 5.035394191741943,
       "learning_rate": 6.245245245245246e-06,
-      "loss": 1.0475,
+      "loss": 1.0459,
       "step": 37510
     },
     {
       "epoch": 112.67,
-      "grad_norm": 10.96910572052002,
+      "grad_norm": 7.770929336547852,
       "learning_rate": 6.244244244244246e-06,
-      "loss": 1.0769,
+      "loss": 1.0652,
       "step": 37520
     },
     {
       "epoch": 112.7,
-      "grad_norm": 6.170958518981934,
+      "grad_norm": 5.866489410400391,
       "learning_rate": 6.243243243243243e-06,
-      "loss": 1.0813,
+      "loss": 1.0777,
       "step": 37530
     },
     {
       "epoch": 112.73,
-      "grad_norm": 8.187673568725586,
+      "grad_norm": 8.742467880249023,
       "learning_rate": 6.242242242242243e-06,
-      "loss": 1.0908,
+      "loss": 1.1024,
       "step": 37540
     },
     {
       "epoch": 112.76,
-      "grad_norm": 5.604761600494385,
+      "grad_norm": 5.667892932891846,
       "learning_rate": 6.241241241241241e-06,
-      "loss": 1.0284,
+      "loss": 1.0263,
       "step": 37550
     },
     {
       "epoch": 112.79,
-      "grad_norm": 6.874386310577393,
+      "grad_norm": 7.254565715789795,
       "learning_rate": 6.240240240240241e-06,
-      "loss": 1.0861,
+      "loss": 1.0894,
       "step": 37560
     },
     {
       "epoch": 112.82,
-      "grad_norm": 6.425081253051758,
+      "grad_norm": 6.6558403968811035,
       "learning_rate": 6.2392392392392394e-06,
-      "loss": 1.0798,
+      "loss": 1.082,
       "step": 37570
     },
     {
       "epoch": 112.85,
-      "grad_norm": 11.378520011901855,
+      "grad_norm": 11.500048637390137,
       "learning_rate": 6.238238238238239e-06,
-      "loss": 1.0794,
+      "loss": 1.0829,
       "step": 37580
     },
     {
       "epoch": 112.88,
-      "grad_norm": 5.561273097991943,
+      "grad_norm": 6.47641134262085,
       "learning_rate": 6.237237237237238e-06,
-      "loss": 1.0638,
+      "loss": 1.0664,
       "step": 37590
     },
     {
       "epoch": 112.91,
-      "grad_norm": 7.788337707519531,
+      "grad_norm": 13.944479942321777,
       "learning_rate": 6.236236236236237e-06,
-      "loss": 1.1608,
+      "loss": 1.1698,
       "step": 37600
     },
     {
       "epoch": 112.94,
-      "grad_norm": 7.264598369598389,
+      "grad_norm": 7.855220317840576,
       "learning_rate": 6.2352352352352355e-06,
-      "loss": 1.051,
+      "loss": 1.0498,
       "step": 37610
     },
     {
       "epoch": 112.97,
-      "grad_norm": 9.19202709197998,
+      "grad_norm": 6.999389171600342,
       "learning_rate": 6.234234234234234e-06,
-      "loss": 1.0379,
+      "loss": 1.0378,
       "step": 37620
     },
     {
       "epoch": 113.0,
-      "eval_accuracy": 0.8023,
-      "eval_loss": 0.6423593163490295,
-      "eval_runtime": 5.3849,
-      "eval_samples_per_second": 1857.04,
-      "eval_steps_per_second": 7.428,
+      "eval_accuracy": 0.8031,
+      "eval_loss": 0.6396318078041077,
+      "eval_runtime": 5.3174,
+      "eval_samples_per_second": 1880.609,
+      "eval_steps_per_second": 7.522,
       "step": 37629
     },
     {
       "epoch": 113.0,
-      "grad_norm": 6.541807651519775,
+      "grad_norm": 8.853808403015137,
       "learning_rate": 6.2332332332332335e-06,
-      "loss": 1.2158,
+      "loss": 1.2201,
       "step": 37630
     },
     {
       "epoch": 113.03,
-      "grad_norm": 8.213560104370117,
+      "grad_norm": 8.036919593811035,
       "learning_rate": 6.232232232232233e-06,
-      "loss": 1.0606,
+      "loss": 1.0399,
       "step": 37640
     },
     {
       "epoch": 113.06,
-      "grad_norm": 4.934196472167969,
+      "grad_norm": 6.757829666137695,
       "learning_rate": 6.2312312312312316e-06,
-      "loss": 1.076,
+      "loss": 1.0746,
       "step": 37650
     },
     {
       "epoch": 113.09,
-      "grad_norm": 8.244272232055664,
+      "grad_norm": 10.715744972229004,
       "learning_rate": 6.230230230230231e-06,
-      "loss": 1.1291,
+      "loss": 1.1247,
       "step": 37660
     },
     {
       "epoch": 113.12,
-      "grad_norm": 6.626892566680908,
+      "grad_norm": 8.031914710998535,
       "learning_rate": 6.22922922922923e-06,
-      "loss": 1.1392,
+      "loss": 1.1447,
       "step": 37670
     },
     {
       "epoch": 113.15,
-      "grad_norm": 7.3037848472595215,
+      "grad_norm": 8.682992935180664,
       "learning_rate": 6.228228228228229e-06,
-      "loss": 1.0821,
+      "loss": 1.0884,
       "step": 37680
     },
     {
       "epoch": 113.18,
-      "grad_norm": 10.491775512695312,
+      "grad_norm": 10.673492431640625,
       "learning_rate": 6.2272272272272285e-06,
-      "loss": 1.0844,
+      "loss": 1.0846,
       "step": 37690
     },
     {
       "epoch": 113.21,
-      "grad_norm": 7.037581920623779,
+      "grad_norm": 6.702084064483643,
       "learning_rate": 6.226226226226226e-06,
-      "loss": 1.1239,
+      "loss": 1.1257,
       "step": 37700
     },
     {
       "epoch": 113.24,
-      "grad_norm": 7.1789751052856445,
+      "grad_norm": 6.20263147354126,
       "learning_rate": 6.225225225225226e-06,
-      "loss": 1.1167,
+      "loss": 1.119,
       "step": 37710
     },
     {
       "epoch": 113.27,
-      "grad_norm": 5.513553142547607,
+      "grad_norm": 6.184822082519531,
       "learning_rate": 6.224224224224224e-06,
-      "loss": 1.0292,
+      "loss": 1.0269,
       "step": 37720
     },
     {
       "epoch": 113.3,
-      "grad_norm": 7.4515485763549805,
+      "grad_norm": 7.169656276702881,
       "learning_rate": 6.223223223223224e-06,
-      "loss": 1.0934,
+      "loss": 1.0872,
       "step": 37730
     },
     {
       "epoch": 113.33,
-      "grad_norm": 7.134352684020996,
+      "grad_norm": 7.631453037261963,
       "learning_rate": 6.222222222222223e-06,
-      "loss": 1.1498,
+      "loss": 1.1423,
       "step": 37740
     },
     {
       "epoch": 113.36,
-      "grad_norm": 5.884287357330322,
+      "grad_norm": 6.761605262756348,
       "learning_rate": 6.221221221221222e-06,
-      "loss": 1.1116,
+      "loss": 1.1126,
       "step": 37750
     },
     {
       "epoch": 113.39,
-      "grad_norm": 5.166376113891602,
+      "grad_norm": 7.547616481781006,
       "learning_rate": 6.220220220220221e-06,
-      "loss": 1.1546,
+      "loss": 1.1678,
       "step": 37760
     },
     {
       "epoch": 113.42,
-      "grad_norm": 5.81422233581543,
+      "grad_norm": 5.673600673675537,
       "learning_rate": 6.219219219219221e-06,
-      "loss": 1.0936,
+      "loss": 1.1011,
       "step": 37770
     },
     {
       "epoch": 113.45,
-      "grad_norm": 11.644503593444824,
+      "grad_norm": 8.955865859985352,
       "learning_rate": 6.218218218218218e-06,
-      "loss": 1.0521,
+      "loss": 1.0397,
       "step": 37780
     },
     {
       "epoch": 113.48,
-      "grad_norm": 8.044099807739258,
+      "grad_norm": 7.744366645812988,
       "learning_rate": 6.217217217217217e-06,
-      "loss": 1.1252,
+      "loss": 1.1195,
       "step": 37790
     },
     {
       "epoch": 113.51,
-      "grad_norm": 5.617173194885254,
+      "grad_norm": 6.849184036254883,
       "learning_rate": 6.2162162162162164e-06,
-      "loss": 1.1269,
+      "loss": 1.1238,
       "step": 37800
     },
     {
       "epoch": 113.54,
-      "grad_norm": 5.081953525543213,
+      "grad_norm": 6.648610591888428,
       "learning_rate": 6.215215215215216e-06,
-      "loss": 0.9948,
+      "loss": 0.9937,
       "step": 37810
     },
     {
       "epoch": 113.57,
-      "grad_norm": 5.246086597442627,
+      "grad_norm": 7.119449138641357,
       "learning_rate": 6.2142142142142145e-06,
-      "loss": 1.0982,
+      "loss": 1.0928,
       "step": 37820
     },
     {
       "epoch": 113.6,
-      "grad_norm": 5.773117542266846,
+      "grad_norm": 6.629194259643555,
       "learning_rate": 6.213213213213214e-06,
-      "loss": 1.0213,
+      "loss": 1.0218,
       "step": 37830
     },
     {
       "epoch": 113.63,
-      "grad_norm": 7.72771692276001,
+      "grad_norm": 8.632637023925781,
       "learning_rate": 6.212212212212213e-06,
-      "loss": 1.045,
+      "loss": 1.0433,
       "step": 37840
     },
     {
       "epoch": 113.66,
-      "grad_norm": 7.984820365905762,
+      "grad_norm": 7.981788158416748,
       "learning_rate": 6.211211211211212e-06,
-      "loss": 1.0625,
+      "loss": 1.0641,
       "step": 37850
     },
     {
       "epoch": 113.69,
-      "grad_norm": 6.090648651123047,
+      "grad_norm": 6.893254280090332,
       "learning_rate": 6.2102102102102105e-06,
-      "loss": 1.0794,
+      "loss": 1.073,
       "step": 37860
     },
     {
       "epoch": 113.72,
-      "grad_norm": 8.90555477142334,
+      "grad_norm": 9.154814720153809,
       "learning_rate": 6.209209209209209e-06,
       "loss": 1.0204,
       "step": 37870
     },
     {
       "epoch": 113.75,
-      "grad_norm": 6.59332799911499,
+      "grad_norm": 6.695834159851074,
       "learning_rate": 6.2082082082082086e-06,
-      "loss": 1.0992,
+      "loss": 1.0949,
       "step": 37880
     },
     {
       "epoch": 113.78,
-      "grad_norm": 5.626827239990234,
+      "grad_norm": 8.63681411743164,
       "learning_rate": 6.207207207207208e-06,
-      "loss": 1.1293,
+      "loss": 1.1227,
       "step": 37890
     },
     {
       "epoch": 113.81,
-      "grad_norm": 6.698348522186279,
+      "grad_norm": 6.8918890953063965,
       "learning_rate": 6.206206206206207e-06,
-      "loss": 1.1161,
+      "loss": 1.11,
       "step": 37900
     },
     {
       "epoch": 113.84,
-      "grad_norm": 5.040308952331543,
+      "grad_norm": 5.163510322570801,
       "learning_rate": 6.205205205205206e-06,
-      "loss": 1.0147,
+      "loss": 1.0138,
       "step": 37910
     },
     {
       "epoch": 113.87,
-      "grad_norm": 7.973019123077393,
+      "grad_norm": 11.211349487304688,
       "learning_rate": 6.204204204204205e-06,
-      "loss": 1.1258,
+      "loss": 1.1197,
       "step": 37920
     },
     {
       "epoch": 113.9,
-      "grad_norm": 4.211703777313232,
+      "grad_norm": 5.612077713012695,
       "learning_rate": 6.203203203203204e-06,
-      "loss": 1.1388,
+      "loss": 1.1369,
       "step": 37930
     },
     {
       "epoch": 113.93,
-      "grad_norm": 9.779862403869629,
+      "grad_norm": 11.407480239868164,
       "learning_rate": 6.2022022022022035e-06,
-      "loss": 1.1203,
+      "loss": 1.1206,
       "step": 37940
     },
     {
       "epoch": 113.96,
-      "grad_norm": 7.614193439483643,
+      "grad_norm": 6.612216472625732,
       "learning_rate": 6.201201201201201e-06,
-      "loss": 1.0936,
+      "loss": 1.0987,
       "step": 37950
     },
     {
       "epoch": 113.99,
-      "grad_norm": 6.6155571937561035,
+      "grad_norm": 10.073917388916016,
       "learning_rate": 6.200200200200201e-06,
-      "loss": 1.0942,
+      "loss": 1.0866,
       "step": 37960
     },
     {
       "epoch": 114.0,
-      "eval_accuracy": 0.8032,
-      "eval_loss": 0.6377995014190674,
-      "eval_runtime": 5.5001,
-      "eval_samples_per_second": 1818.143,
-      "eval_steps_per_second": 7.273,
+      "eval_accuracy": 0.8029,
+      "eval_loss": 0.6382209062576294,
+      "eval_runtime": 5.4371,
+      "eval_samples_per_second": 1839.224,
+      "eval_steps_per_second": 7.357,
       "step": 37962
     },
     {
       "epoch": 114.02,
-      "grad_norm": 6.352574825286865,
+      "grad_norm": 9.132283210754395,
       "learning_rate": 6.199199199199199e-06,
-      "loss": 1.1315,
+      "loss": 1.1358,
       "step": 37970
     },
     {
       "epoch": 114.05,
-      "grad_norm": 7.7886061668396,
+      "grad_norm": 7.5684919357299805,
       "learning_rate": 6.198198198198199e-06,
-      "loss": 1.1508,
+      "loss": 1.1317,
       "step": 37980
     },
     {
       "epoch": 114.08,
-      "grad_norm": 6.236972332000732,
+      "grad_norm": 6.280869007110596,
       "learning_rate": 6.197197197197198e-06,
-      "loss": 1.0988,
+      "loss": 1.0872,
       "step": 37990
     },
     {
       "epoch": 114.11,
-      "grad_norm": 5.0635666847229,
+      "grad_norm": 5.230590343475342,
       "learning_rate": 6.196196196196197e-06,
-      "loss": 1.0605,
+      "loss": 1.0622,
       "step": 38000
     },
     {
       "epoch": 114.14,
-      "grad_norm": 5.729677200317383,
+      "grad_norm": 5.879208564758301,
       "learning_rate": 6.195195195195196e-06,
-      "loss": 1.0466,
+      "loss": 1.0467,
       "step": 38010
     },
     {
       "epoch": 114.17,
-      "grad_norm": 6.128084659576416,
+      "grad_norm": 6.575455188751221,
       "learning_rate": 6.194194194194195e-06,
-      "loss": 1.0111,
+      "loss": 1.0037,
       "step": 38020
     },
     {
       "epoch": 114.2,
-      "grad_norm": 7.6433820724487305,
+      "grad_norm": 8.13090705871582,
       "learning_rate": 6.1931931931931934e-06,
-      "loss": 1.1696,
+      "loss": 1.1693,
       "step": 38030
     },
     {
       "epoch": 114.23,
-      "grad_norm": 7.8031744956970215,
+      "grad_norm": 11.15799331665039,
       "learning_rate": 6.192192192192192e-06,
-      "loss": 1.0897,
+      "loss": 1.0869,
       "step": 38040
     },
     {
       "epoch": 114.26,
-      "grad_norm": 6.142184257507324,
+      "grad_norm": 8.36663818359375,
       "learning_rate": 6.1911911911911915e-06,
-      "loss": 1.068,
+      "loss": 1.0645,
       "step": 38050
     },
     {
       "epoch": 114.29,
-      "grad_norm": 6.654850482940674,
+      "grad_norm": 6.791731834411621,
       "learning_rate": 6.190190190190191e-06,
-      "loss": 1.1461,
+      "loss": 1.1267,
       "step": 38060
     },
     {
       "epoch": 114.32,
-      "grad_norm": 8.182784080505371,
+      "grad_norm": 7.169312477111816,
       "learning_rate": 6.1891891891891895e-06,
-      "loss": 1.0426,
+      "loss": 1.0553,
       "step": 38070
     },
     {
       "epoch": 114.35,
-      "grad_norm": 7.438183307647705,
+      "grad_norm": 5.908121585845947,
       "learning_rate": 6.188188188188189e-06,
-      "loss": 1.18,
+      "loss": 1.1823,
       "step": 38080
     },
     {
       "epoch": 114.38,
-      "grad_norm": 7.077129364013672,
+      "grad_norm": 7.020669460296631,
       "learning_rate": 6.187187187187188e-06,
-      "loss": 1.1189,
+      "loss": 1.1183,
       "step": 38090
     },
     {
       "epoch": 114.41,
-      "grad_norm": 9.653081893920898,
+      "grad_norm": 9.45844554901123,
       "learning_rate": 6.186186186186187e-06,
-      "loss": 1.1091,
+      "loss": 1.0969,
       "step": 38100
     },
     {
       "epoch": 114.44,
-      "grad_norm": 6.013190269470215,
+      "grad_norm": 8.213235855102539,
       "learning_rate": 6.1851851851851856e-06,
-      "loss": 1.0802,
+      "loss": 1.0716,
       "step": 38110
     },
     {
       "epoch": 114.47,
-      "grad_norm": 5.412250518798828,
+      "grad_norm": 7.67694616317749,
       "learning_rate": 6.184184184184184e-06,
-      "loss": 1.1337,
+      "loss": 1.1297,
       "step": 38120
     },
     {
       "epoch": 114.5,
-      "grad_norm": 7.195882797241211,
+      "grad_norm": 7.194870471954346,
       "learning_rate": 6.183183183183184e-06,
-      "loss": 1.0945,
+      "loss": 1.092,
       "step": 38130
     },
     {
       "epoch": 114.53,
-      "grad_norm": 6.091780662536621,
+      "grad_norm": 7.556748867034912,
       "learning_rate": 6.182182182182182e-06,
-      "loss": 1.098,
+      "loss": 1.0917,
       "step": 38140
     },
     {
       "epoch": 114.56,
-      "grad_norm": 8.314773559570312,
+      "grad_norm": 6.2943339347839355,
       "learning_rate": 6.181181181181182e-06,
-      "loss": 1.0923,
+      "loss": 1.0789,
       "step": 38150
     },
     {
       "epoch": 114.59,
-      "grad_norm": 7.336097240447998,
+      "grad_norm": 8.307439804077148,
       "learning_rate": 6.180180180180181e-06,
-      "loss": 1.1188,
+      "loss": 1.1269,
       "step": 38160
     },
     {
       "epoch": 114.62,
-      "grad_norm": 6.4828033447265625,
+      "grad_norm": 8.274991035461426,
       "learning_rate": 6.17917917917918e-06,
-      "loss": 1.1546,
+      "loss": 1.1622,
       "step": 38170
     },
     {
       "epoch": 114.65,
-      "grad_norm": 7.746421813964844,
+      "grad_norm": 9.104543685913086,
       "learning_rate": 6.178178178178179e-06,
-      "loss": 1.0704,
+      "loss": 1.0576,
       "step": 38180
     },
     {
       "epoch": 114.68,
-      "grad_norm": 4.772459983825684,
+      "grad_norm": 6.266383171081543,
       "learning_rate": 6.177177177177177e-06,
-      "loss": 1.1328,
+      "loss": 1.1395,
       "step": 38190
     },
     {
       "epoch": 114.71,
-      "grad_norm": 9.287857055664062,
+      "grad_norm": 8.64734935760498,
       "learning_rate": 6.176176176176176e-06,
-      "loss": 1.0956,
+      "loss": 1.0816,
       "step": 38200
     },
     {
       "epoch": 114.74,
-      "grad_norm": 6.402691841125488,
+      "grad_norm": 6.0654730796813965,
       "learning_rate": 6.175175175175176e-06,
-      "loss": 1.1085,
+      "loss": 1.1015,
       "step": 38210
     },
     {
       "epoch": 114.77,
-      "grad_norm": 6.446484565734863,
+      "grad_norm": 6.369784832000732,
       "learning_rate": 6.174174174174174e-06,
-      "loss": 1.0954,
+      "loss": 1.0828,
       "step": 38220
     },
     {
       "epoch": 114.8,
-      "grad_norm": 6.492711067199707,
+      "grad_norm": 12.06479263305664,
       "learning_rate": 6.173173173173174e-06,
-      "loss": 1.0612,
+      "loss": 1.0587,
       "step": 38230
     },
     {
       "epoch": 114.83,
-      "grad_norm": 8.32776927947998,
+      "grad_norm": 13.29538345336914,
       "learning_rate": 6.172172172172172e-06,
-      "loss": 1.0273,
+      "loss": 1.0312,
       "step": 38240
     },
     {
       "epoch": 114.86,
-      "grad_norm": 5.714176177978516,
+      "grad_norm": 5.428351402282715,
       "learning_rate": 6.171171171171172e-06,
-      "loss": 1.0803,
+      "loss": 1.074,
       "step": 38250
     },
     {
       "epoch": 114.89,
-      "grad_norm": 7.545133113861084,
+      "grad_norm": 6.145267963409424,
       "learning_rate": 6.170170170170171e-06,
-      "loss": 1.1181,
+      "loss": 1.1215,
       "step": 38260
     },
     {
       "epoch": 114.92,
-      "grad_norm": 6.815221786499023,
+      "grad_norm": 6.752283573150635,
       "learning_rate": 6.16916916916917e-06,
-      "loss": 1.0112,
+      "loss": 1.0037,
       "step": 38270
     },
     {
       "epoch": 114.95,
-      "grad_norm": 6.180943489074707,
+      "grad_norm": 6.166947364807129,
       "learning_rate": 6.1681681681681685e-06,
-      "loss": 1.0211,
+      "loss": 1.0105,
       "step": 38280
     },
     {
       "epoch": 114.98,
-      "grad_norm": 10.584677696228027,
+      "grad_norm": 15.164694786071777,
       "learning_rate": 6.167167167167167e-06,
-      "loss": 1.0977,
+      "loss": 1.0864,
       "step": 38290
     },
     {
       "epoch": 115.0,
-      "eval_accuracy": 0.8069,
-      "eval_loss": 0.6247538328170776,
-      "eval_runtime": 5.6478,
-      "eval_samples_per_second": 1770.599,
-      "eval_steps_per_second": 7.082,
+      "eval_accuracy": 0.8067,
+      "eval_loss": 0.6238564848899841,
+      "eval_runtime": 5.4586,
+      "eval_samples_per_second": 1831.982,
+      "eval_steps_per_second": 7.328,
       "step": 38295
     },
     {
       "epoch": 115.02,
-      "grad_norm": 6.982071399688721,
+      "grad_norm": 6.741769313812256,
       "learning_rate": 6.1661661661661665e-06,
-      "loss": 1.0758,
+      "loss": 1.0798,
       "step": 38300
     },
     {
       "epoch": 115.05,
-      "grad_norm": 6.137686729431152,
+      "grad_norm": 7.004385471343994,
       "learning_rate": 6.165165165165166e-06,
-      "loss": 1.0841,
+      "loss": 1.0875,
       "step": 38310
     },
     {
       "epoch": 115.08,
-      "grad_norm": 5.589268207550049,
+      "grad_norm": 8.244741439819336,
       "learning_rate": 6.1641641641641645e-06,
-      "loss": 1.0824,
+      "loss": 1.0757,
       "step": 38320
     },
     {
       "epoch": 115.11,
-      "grad_norm": 7.055795192718506,
+      "grad_norm": 6.676910400390625,
       "learning_rate": 6.163163163163164e-06,
-      "loss": 1.1103,
+      "loss": 1.1165,
       "step": 38330
     },
     {
       "epoch": 115.14,
-      "grad_norm": 7.501062870025635,
+      "grad_norm": 7.7475810050964355,
       "learning_rate": 6.162162162162163e-06,
-      "loss": 1.0466,
+      "loss": 1.0365,
       "step": 38340
     },
     {
       "epoch": 115.17,
-      "grad_norm": 6.700809955596924,
+      "grad_norm": 6.7648115158081055,
       "learning_rate": 6.161161161161162e-06,
-      "loss": 1.1177,
+      "loss": 1.1138,
       "step": 38350
     },
     {
       "epoch": 115.2,
-      "grad_norm": 8.193719863891602,
+      "grad_norm": 6.794809818267822,
       "learning_rate": 6.16016016016016e-06,
-      "loss": 1.0766,
+      "loss": 1.0726,
       "step": 38360
     },
     {
       "epoch": 115.23,
-      "grad_norm": 6.224299430847168,
+      "grad_norm": 7.477279186248779,
       "learning_rate": 6.159159159159159e-06,
-      "loss": 1.0487,
+      "loss": 1.0532,
       "step": 38370
     },
     {
       "epoch": 115.26,
-      "grad_norm": 6.043673038482666,
+      "grad_norm": 7.371276378631592,
       "learning_rate": 6.158158158158159e-06,
-      "loss": 1.0761,
+      "loss": 1.0644,
       "step": 38380
     },
     {
       "epoch": 115.29,
-      "grad_norm": 8.22892951965332,
+      "grad_norm": 10.167928695678711,
       "learning_rate": 6.157157157157157e-06,
-      "loss": 1.0619,
+      "loss": 1.0568,
       "step": 38390
     },
     {
       "epoch": 115.32,
-      "grad_norm": 8.303683280944824,
+      "grad_norm": 9.575668334960938,
       "learning_rate": 6.156156156156157e-06,
-      "loss": 1.1619,
+      "loss": 1.1694,
       "step": 38400
     },
     {
       "epoch": 115.35,
-      "grad_norm": 6.939892768859863,
+      "grad_norm": 7.522953987121582,
       "learning_rate": 6.155155155155156e-06,
-      "loss": 1.1049,
+      "loss": 1.0928,
       "step": 38410
     },
     {
       "epoch": 115.38,
-      "grad_norm": 7.088039875030518,
+      "grad_norm": 8.044832229614258,
       "learning_rate": 6.154154154154155e-06,
-      "loss": 1.1087,
+      "loss": 1.1019,
       "step": 38420
     },
     {
       "epoch": 115.41,
-      "grad_norm": 6.501294136047363,
+      "grad_norm": 8.999271392822266,
       "learning_rate": 6.153153153153154e-06,
-      "loss": 1.0745,
+      "loss": 1.0761,
       "step": 38430
     },
     {
       "epoch": 115.44,
-      "grad_norm": 7.175703048706055,
+      "grad_norm": 6.0032148361206055,
       "learning_rate": 6.152152152152152e-06,
-      "loss": 1.0518,
+      "loss": 1.0484,
       "step": 38440
     },
     {
       "epoch": 115.47,
-      "grad_norm": 5.008584976196289,
+      "grad_norm": 5.252702713012695,
       "learning_rate": 6.151151151151151e-06,
-      "loss": 1.0388,
+      "loss": 1.0339,
       "step": 38450
     },
     {
       "epoch": 115.5,
-      "grad_norm": 4.891729831695557,
+      "grad_norm": 7.150794982910156,
       "learning_rate": 6.150150150150151e-06,
       "loss": 1.0689,
       "step": 38460
     },
     {
       "epoch": 115.53,
-      "grad_norm": 7.136166095733643,
+      "grad_norm": 5.979158401489258,
       "learning_rate": 6.149149149149149e-06,
-      "loss": 1.0653,
+      "loss": 1.0569,
       "step": 38470
     },
     {
       "epoch": 115.56,
-      "grad_norm": 8.944881439208984,
+      "grad_norm": 10.291590690612793,
       "learning_rate": 6.148148148148149e-06,
-      "loss": 1.0608,
+      "loss": 1.054,
       "step": 38480
     },
     {
       "epoch": 115.59,
-      "grad_norm": 5.104443073272705,
+      "grad_norm": 6.193352222442627,
       "learning_rate": 6.147147147147147e-06,
-      "loss": 1.1481,
+      "loss": 1.1524,
       "step": 38490
     },
     {
       "epoch": 115.62,
-      "grad_norm": 7.385315895080566,
+      "grad_norm": 11.042230606079102,
       "learning_rate": 6.146146146146147e-06,
-      "loss": 1.09,
+      "loss": 1.0906,
       "step": 38500
     },
     {
       "epoch": 115.65,
-      "grad_norm": 7.405564785003662,
+      "grad_norm": 8.939558982849121,
       "learning_rate": 6.145145145145146e-06,
-      "loss": 1.0298,
+      "loss": 1.0292,
       "step": 38510
     },
     {
       "epoch": 115.68,
-      "grad_norm": 7.355929374694824,
+      "grad_norm": 9.453753471374512,
       "learning_rate": 6.144144144144145e-06,
-      "loss": 1.1424,
+      "loss": 1.1345,
       "step": 38520
     },
     {
       "epoch": 115.71,
-      "grad_norm": 7.778319835662842,
+      "grad_norm": 6.927004814147949,
       "learning_rate": 6.1431431431431435e-06,
-      "loss": 1.0991,
+      "loss": 1.0809,
       "step": 38530
     },
     {
       "epoch": 115.74,
-      "grad_norm": 6.889627933502197,
+      "grad_norm": 7.291510105133057,
       "learning_rate": 6.142142142142142e-06,
-      "loss": 1.0806,
+      "loss": 1.0856,
       "step": 38540
     },
     {
       "epoch": 115.77,
-      "grad_norm": 6.277814865112305,
+      "grad_norm": 6.924350738525391,
       "learning_rate": 6.1411411411411415e-06,
-      "loss": 1.0811,
+      "loss": 1.0768,
       "step": 38550
     },
     {
       "epoch": 115.8,
-      "grad_norm": 7.529620170593262,
+      "grad_norm": 6.210633754730225,
       "learning_rate": 6.140140140140141e-06,
-      "loss": 1.1054,
+      "loss": 1.1025,
       "step": 38560
     },
     {
       "epoch": 115.83,
-      "grad_norm": 8.47703742980957,
+      "grad_norm": 7.285170078277588,
       "learning_rate": 6.1391391391391396e-06,
-      "loss": 1.1305,
+      "loss": 1.1345,
       "step": 38570
     },
     {
       "epoch": 115.86,
-      "grad_norm": 6.109755992889404,
+      "grad_norm": 6.2451581954956055,
       "learning_rate": 6.138138138138139e-06,
-      "loss": 1.1136,
+      "loss": 1.1056,
       "step": 38580
     },
     {
       "epoch": 115.89,
-      "grad_norm": 7.523285865783691,
+      "grad_norm": 9.358064651489258,
       "learning_rate": 6.137137137137138e-06,
-      "loss": 1.0563,
+      "loss": 1.0586,
       "step": 38590
     },
     {
       "epoch": 115.92,
-      "grad_norm": 10.578325271606445,
+      "grad_norm": 13.266717910766602,
       "learning_rate": 6.136136136136137e-06,
-      "loss": 1.2011,
+      "loss": 1.2066,
       "step": 38600
     },
     {
       "epoch": 115.95,
-      "grad_norm": 7.4092020988464355,
+      "grad_norm": 9.144289016723633,
       "learning_rate": 6.135135135135135e-06,
-      "loss": 1.155,
+      "loss": 1.1526,
       "step": 38610
     },
     {
       "epoch": 115.98,
-      "grad_norm": 8.544330596923828,
+      "grad_norm": 10.439029693603516,
       "learning_rate": 6.134134134134134e-06,
-      "loss": 1.1348,
+      "loss": 1.1246,
       "step": 38620
     },
     {
       "epoch": 116.0,
-      "eval_accuracy": 0.8082,
-      "eval_loss": 0.6264135837554932,
-      "eval_runtime": 5.2294,
-      "eval_samples_per_second": 1912.271,
-      "eval_steps_per_second": 7.649,
+      "eval_accuracy": 0.8085,
+      "eval_loss": 0.6257779002189636,
+      "eval_runtime": 5.3358,
+      "eval_samples_per_second": 1874.136,
+      "eval_steps_per_second": 7.497,
       "step": 38628
     },
     {
       "epoch": 116.01,
-      "grad_norm": 6.588642120361328,
+      "grad_norm": 9.019854545593262,
       "learning_rate": 6.133133133133134e-06,
-      "loss": 1.1569,
+      "loss": 1.1862,
       "step": 38630
     },
     {
       "epoch": 116.04,
-      "grad_norm": 9.143474578857422,
+      "grad_norm": 6.755329608917236,
       "learning_rate": 6.132132132132132e-06,
-      "loss": 1.1822,
+      "loss": 1.1776,
       "step": 38640
     },
     {
       "epoch": 116.07,
-      "grad_norm": 7.717463493347168,
+      "grad_norm": 7.433858394622803,
       "learning_rate": 6.131131131131132e-06,
-      "loss": 1.1209,
+      "loss": 1.118,
       "step": 38650
     },
     {
       "epoch": 116.1,
-      "grad_norm": 6.6299214363098145,
+      "grad_norm": 9.248769760131836,
       "learning_rate": 6.130130130130131e-06,
-      "loss": 1.0593,
+      "loss": 1.0603,
       "step": 38660
     },
     {
       "epoch": 116.13,
-      "grad_norm": 5.8547587394714355,
+      "grad_norm": 7.370952606201172,
       "learning_rate": 6.12912912912913e-06,
-      "loss": 1.0729,
+      "loss": 1.0735,
       "step": 38670
     },
     {
       "epoch": 116.16,
-      "grad_norm": 5.968803882598877,
+      "grad_norm": 7.740847110748291,
       "learning_rate": 6.128128128128129e-06,
-      "loss": 1.0409,
+      "loss": 1.0393,
       "step": 38680
     },
     {
       "epoch": 116.19,
-      "grad_norm": 6.659968852996826,
+      "grad_norm": 7.061502456665039,
       "learning_rate": 6.127127127127127e-06,
-      "loss": 1.0888,
+      "loss": 1.0903,
       "step": 38690
     },
     {
       "epoch": 116.22,
-      "grad_norm": 5.52916145324707,
+      "grad_norm": 6.872004985809326,
       "learning_rate": 6.126126126126126e-06,
-      "loss": 1.0868,
+      "loss": 1.0806,
       "step": 38700
     },
     {
       "epoch": 116.25,
-      "grad_norm": 6.599611282348633,
+      "grad_norm": 5.687037467956543,
       "learning_rate": 6.125125125125125e-06,
-      "loss": 1.0677,
+      "loss": 1.0648,
       "step": 38710
     },
     {
       "epoch": 116.28,
-      "grad_norm": 6.446048259735107,
+      "grad_norm": 8.513541221618652,
       "learning_rate": 6.124124124124124e-06,
-      "loss": 1.1122,
+      "loss": 1.1124,
       "step": 38720
     },
     {
       "epoch": 116.31,
-      "grad_norm": 7.391465187072754,
+      "grad_norm": 6.736770153045654,
       "learning_rate": 6.123123123123124e-06,
-      "loss": 1.1157,
+      "loss": 1.1181,
       "step": 38730
     },
     {
       "epoch": 116.34,
-      "grad_norm": 7.397523403167725,
+      "grad_norm": 7.32673454284668,
       "learning_rate": 6.1221221221221225e-06,
-      "loss": 1.1396,
+      "loss": 1.1341,
       "step": 38740
     },
     {
       "epoch": 116.37,
-      "grad_norm": 7.176833152770996,
+      "grad_norm": 8.949681282043457,
       "learning_rate": 6.121121121121122e-06,
-      "loss": 1.072,
+      "loss": 1.0607,
       "step": 38750
     },
     {
       "epoch": 116.4,
-      "grad_norm": 7.098006248474121,
+      "grad_norm": 6.170365333557129,
       "learning_rate": 6.120120120120121e-06,
-      "loss": 1.0492,
+      "loss": 1.054,
       "step": 38760
     },
     {
       "epoch": 116.43,
-      "grad_norm": 7.035533428192139,
+      "grad_norm": 7.212123870849609,
       "learning_rate": 6.11911911911912e-06,
-      "loss": 1.0623,
+      "loss": 1.063,
       "step": 38770
     },
     {
       "epoch": 116.46,
-      "grad_norm": 7.146546363830566,
+      "grad_norm": 7.851434230804443,
       "learning_rate": 6.1181181181181185e-06,
-      "loss": 1.0616,
+      "loss": 1.0684,
       "step": 38780
     },
     {
       "epoch": 116.49,
-      "grad_norm": 8.441097259521484,
+      "grad_norm": 8.679754257202148,
       "learning_rate": 6.117117117117117e-06,
-      "loss": 1.138,
+      "loss": 1.1376,
       "step": 38790
     },
     {
       "epoch": 116.52,
-      "grad_norm": 7.055720806121826,
+      "grad_norm": 5.119994640350342,
       "learning_rate": 6.1161161161161166e-06,
-      "loss": 1.0076,
+      "loss": 1.0116,
       "step": 38800
     },
     {
       "epoch": 116.55,
-      "grad_norm": 6.161187171936035,
+      "grad_norm": 8.5137357711792,
       "learning_rate": 6.115115115115115e-06,
-      "loss": 1.0715,
+      "loss": 1.0676,
       "step": 38810
     },
     {
       "epoch": 116.58,
-      "grad_norm": 7.9579997062683105,
+      "grad_norm": 7.9841814041137695,
       "learning_rate": 6.114114114114115e-06,
-      "loss": 1.1507,
+      "loss": 1.1276,
       "step": 38820
     },
     {
       "epoch": 116.61,
-      "grad_norm": 8.847417831420898,
+      "grad_norm": 8.651477813720703,
       "learning_rate": 6.113113113113114e-06,
       "loss": 1.1098,
       "step": 38830
     },
     {
       "epoch": 116.64,
-      "grad_norm": 5.773353099822998,
+      "grad_norm": 6.277093410491943,
       "learning_rate": 6.112112112112113e-06,
-      "loss": 1.0879,
+      "loss": 1.0805,
       "step": 38840
     },
     {
       "epoch": 116.67,
-      "grad_norm": 4.963371753692627,
+      "grad_norm": 5.623876571655273,
       "learning_rate": 6.111111111111112e-06,
-      "loss": 1.0519,
+      "loss": 1.0585,
       "step": 38850
     },
     {
       "epoch": 116.7,
-      "grad_norm": 11.065657615661621,
+      "grad_norm": 7.483618259429932,
       "learning_rate": 6.11011011011011e-06,
-      "loss": 1.0947,
+      "loss": 1.0859,
       "step": 38860
     },
     {
       "epoch": 116.73,
-      "grad_norm": 10.643662452697754,
+      "grad_norm": 8.838569641113281,
       "learning_rate": 6.109109109109109e-06,
-      "loss": 0.9468,
+      "loss": 0.9376,
       "step": 38870
     },
     {
       "epoch": 116.76,
-      "grad_norm": 7.243569374084473,
+      "grad_norm": 7.196118354797363,
       "learning_rate": 6.108108108108109e-06,
-      "loss": 1.0191,
+      "loss": 1.023,
       "step": 38880
     },
     {
       "epoch": 116.79,
-      "grad_norm": 5.9333014488220215,
+      "grad_norm": 6.297749042510986,
       "learning_rate": 6.107107107107107e-06,
-      "loss": 1.0843,
+      "loss": 1.0836,
       "step": 38890
     },
     {
       "epoch": 116.82,
-      "grad_norm": 5.177875995635986,
+      "grad_norm": 5.5909953117370605,
       "learning_rate": 6.106106106106107e-06,
-      "loss": 1.0612,
+      "loss": 1.0597,
       "step": 38900
     },
     {
       "epoch": 116.85,
-      "grad_norm": 6.676166534423828,
+      "grad_norm": 7.662276744842529,
       "learning_rate": 6.105105105105106e-06,
-      "loss": 1.083,
+      "loss": 1.096,
       "step": 38910
     },
     {
       "epoch": 116.88,
-      "grad_norm": 6.664681911468506,
+      "grad_norm": 6.791346549987793,
       "learning_rate": 6.104104104104105e-06,
-      "loss": 1.0335,
+      "loss": 1.0363,
       "step": 38920
     },
     {
       "epoch": 116.91,
-      "grad_norm": 6.053986072540283,
+      "grad_norm": 8.049251556396484,
       "learning_rate": 6.103103103103104e-06,
-      "loss": 1.1009,
+      "loss": 1.0985,
       "step": 38930
     },
     {
       "epoch": 116.94,
-      "grad_norm": 7.008050441741943,
+      "grad_norm": 6.136603832244873,
       "learning_rate": 6.102102102102102e-06,
-      "loss": 1.0942,
+      "loss": 1.0902,
       "step": 38940
     },
     {
       "epoch": 116.97,
-      "grad_norm": 6.382004261016846,
+      "grad_norm": 7.264875888824463,
       "learning_rate": 6.101101101101101e-06,
-      "loss": 1.071,
+      "loss": 1.0734,
       "step": 38950
     },
     {
       "epoch": 117.0,
-      "grad_norm": 7.557199478149414,
+      "grad_norm": 7.034822940826416,
       "learning_rate": 6.1001001001001e-06,
-      "loss": 1.0204,
+      "loss": 1.0202,
       "step": 38960
     },
     {
       "epoch": 117.0,
-      "eval_accuracy": 0.808,
-      "eval_loss": 0.625478982925415,
-      "eval_runtime": 5.4679,
-      "eval_samples_per_second": 1828.845,
-      "eval_steps_per_second": 7.315,
+      "eval_accuracy": 0.8082,
+      "eval_loss": 0.6250253319740295,
+      "eval_runtime": 5.2101,
+      "eval_samples_per_second": 1919.349,
+      "eval_steps_per_second": 7.677,
       "step": 38961
     },
     {
       "epoch": 117.03,
-      "grad_norm": 10.044252395629883,
+      "grad_norm": 9.245140075683594,
       "learning_rate": 6.0990990990990995e-06,
-      "loss": 1.151,
+      "loss": 1.1349,
       "step": 38970
     },
     {
       "epoch": 117.06,
-      "grad_norm": 6.766800403594971,
+      "grad_norm": 6.995547294616699,
       "learning_rate": 6.098098098098099e-06,
-      "loss": 1.113,
+      "loss": 1.111,
       "step": 38980
     },
     {
       "epoch": 117.09,
-      "grad_norm": 8.690776824951172,
+      "grad_norm": 7.991418361663818,
       "learning_rate": 6.0970970970970975e-06,
-      "loss": 1.0083,
+      "loss": 1.0097,
       "step": 38990
     },
     {
       "epoch": 117.12,
-      "grad_norm": 8.645395278930664,
+      "grad_norm": 7.584635257720947,
       "learning_rate": 6.096096096096097e-06,
-      "loss": 1.1687,
+      "loss": 1.1699,
       "step": 39000
     },
     {
       "epoch": 117.15,
-      "grad_norm": 6.34262752532959,
+      "grad_norm": 7.299264907836914,
       "learning_rate": 6.095095095095096e-06,
-      "loss": 1.0591,
+      "loss": 1.0443,
       "step": 39010
     },
     {
       "epoch": 117.18,
-      "grad_norm": 7.557326316833496,
+      "grad_norm": 7.191432476043701,
       "learning_rate": 6.094094094094095e-06,
-      "loss": 1.0334,
+      "loss": 1.0304,
       "step": 39020
     },
     {
       "epoch": 117.21,
-      "grad_norm": 8.071248054504395,
+      "grad_norm": 6.003443241119385,
       "learning_rate": 6.0930930930930936e-06,
-      "loss": 1.0536,
+      "loss": 1.0539,
       "step": 39030
     },
     {
       "epoch": 117.24,
-      "grad_norm": 6.410345554351807,
+      "grad_norm": 6.553544521331787,
       "learning_rate": 6.092092092092092e-06,
       "loss": 1.1148,
       "step": 39040
     },
     {
       "epoch": 117.27,
-      "grad_norm": 6.450709819793701,
+      "grad_norm": 5.394062042236328,
       "learning_rate": 6.091091091091092e-06,
-      "loss": 1.1189,
+      "loss": 1.127,
       "step": 39050
     },
     {
       "epoch": 117.3,
-      "grad_norm": 6.358497142791748,
+      "grad_norm": 10.245285987854004,
       "learning_rate": 6.09009009009009e-06,
-      "loss": 1.076,
+      "loss": 1.0883,
       "step": 39060
     },
     {
       "epoch": 117.33,
-      "grad_norm": 6.523952484130859,
+      "grad_norm": 10.047804832458496,
       "learning_rate": 6.08908908908909e-06,
-      "loss": 1.1253,
+      "loss": 1.1203,
       "step": 39070
     },
     {
       "epoch": 117.36,
-      "grad_norm": 7.297082424163818,
+      "grad_norm": 5.606284141540527,
       "learning_rate": 6.088088088088089e-06,
-      "loss": 1.0994,
+      "loss": 1.0963,
       "step": 39080
     },
     {
       "epoch": 117.39,
-      "grad_norm": 8.969975471496582,
+      "grad_norm": 7.284419059753418,
       "learning_rate": 6.087087087087088e-06,
-      "loss": 1.1374,
+      "loss": 1.1344,
       "step": 39090
     },
     {
       "epoch": 117.42,
-      "grad_norm": 4.738396167755127,
+      "grad_norm": 5.222155570983887,
       "learning_rate": 6.086086086086087e-06,
-      "loss": 1.0367,
+      "loss": 1.0424,
       "step": 39100
     },
     {
       "epoch": 117.45,
-      "grad_norm": 6.322540283203125,
+      "grad_norm": 5.377930164337158,
       "learning_rate": 6.085085085085085e-06,
-      "loss": 1.097,
+      "loss": 1.0996,
       "step": 39110
     },
     {
       "epoch": 117.48,
-      "grad_norm": 8.79407024383545,
+      "grad_norm": 10.453733444213867,
       "learning_rate": 6.084084084084084e-06,
-      "loss": 1.1356,
+      "loss": 1.1443,
       "step": 39120
     },
     {
       "epoch": 117.51,
-      "grad_norm": 7.099576950073242,
+      "grad_norm": 8.718679428100586,
       "learning_rate": 6.083083083083084e-06,
-      "loss": 1.0683,
+      "loss": 1.0616,
       "step": 39130
     },
     {
       "epoch": 117.54,
-      "grad_norm": 8.459528923034668,
+      "grad_norm": 7.620346546173096,
       "learning_rate": 6.082082082082082e-06,
-      "loss": 1.132,
+      "loss": 1.1205,
       "step": 39140
     },
     {
       "epoch": 117.57,
-      "grad_norm": 7.239151954650879,
+      "grad_norm": 8.342159271240234,
       "learning_rate": 6.081081081081082e-06,
-      "loss": 1.0712,
+      "loss": 1.076,
       "step": 39150
     },
     {
       "epoch": 117.6,
-      "grad_norm": 5.342985153198242,
+      "grad_norm": 7.158531188964844,
       "learning_rate": 6.08008008008008e-06,
-      "loss": 1.0892,
+      "loss": 1.0889,
       "step": 39160
     },
     {
       "epoch": 117.63,
-      "grad_norm": 4.411149024963379,
+      "grad_norm": 4.110936641693115,
       "learning_rate": 6.07907907907908e-06,
-      "loss": 1.039,
+      "loss": 1.0392,
       "step": 39170
     },
     {
       "epoch": 117.66,
-      "grad_norm": 6.665385723114014,
+      "grad_norm": 5.99495792388916,
       "learning_rate": 6.078078078078079e-06,
-      "loss": 1.0528,
+      "loss": 1.0406,
       "step": 39180
     },
     {
       "epoch": 117.69,
-      "grad_norm": 6.190800666809082,
+      "grad_norm": 8.643967628479004,
       "learning_rate": 6.077077077077077e-06,
-      "loss": 1.0515,
+      "loss": 1.0533,
       "step": 39190
     },
     {
       "epoch": 117.72,
-      "grad_norm": 10.21083927154541,
+      "grad_norm": 8.937430381774902,
       "learning_rate": 6.0760760760760765e-06,
-      "loss": 1.1105,
+      "loss": 1.0996,
       "step": 39200
     },
     {
       "epoch": 117.75,
-      "grad_norm": 6.745800495147705,
+      "grad_norm": 7.833702087402344,
       "learning_rate": 6.075075075075075e-06,
-      "loss": 1.1153,
+      "loss": 1.1156,
       "step": 39210
     },
     {
       "epoch": 117.78,
-      "grad_norm": 7.451457977294922,
+      "grad_norm": 7.135834693908691,
       "learning_rate": 6.0740740740740745e-06,
-      "loss": 1.0509,
+      "loss": 1.0453,
       "step": 39220
     },
     {
       "epoch": 117.81,
-      "grad_norm": 6.436119079589844,
+      "grad_norm": 7.214910984039307,
       "learning_rate": 6.073073073073074e-06,
-      "loss": 1.1063,
+      "loss": 1.1152,
       "step": 39230
     },
     {
       "epoch": 117.84,
-      "grad_norm": 8.729544639587402,
+      "grad_norm": 8.492382049560547,
       "learning_rate": 6.0720720720720725e-06,
-      "loss": 1.0604,
+      "loss": 1.0557,
       "step": 39240
     },
     {
       "epoch": 117.87,
-      "grad_norm": 8.518671035766602,
+      "grad_norm": 7.213019371032715,
       "learning_rate": 6.071071071071072e-06,
-      "loss": 1.0778,
+      "loss": 1.0807,
       "step": 39250
     },
     {
       "epoch": 117.9,
-      "grad_norm": 6.04960298538208,
+      "grad_norm": 8.937920570373535,
       "learning_rate": 6.0700700700700706e-06,
-      "loss": 1.0507,
+      "loss": 1.0384,
       "step": 39260
     },
     {
       "epoch": 117.93,
-      "grad_norm": 5.795947074890137,
+      "grad_norm": 8.569550514221191,
       "learning_rate": 6.06906906906907e-06,
-      "loss": 1.1411,
+      "loss": 1.1452,
       "step": 39270
     },
     {
       "epoch": 117.96,
-      "grad_norm": 11.41977310180664,
+      "grad_norm": 8.663778305053711,
       "learning_rate": 6.068068068068068e-06,
-      "loss": 1.0926,
+      "loss": 1.0946,
       "step": 39280
     },
     {
       "epoch": 117.99,
-      "grad_norm": 6.701806545257568,
+      "grad_norm": 9.38388729095459,
       "learning_rate": 6.067067067067067e-06,
-      "loss": 1.0201,
+      "loss": 1.0199,
       "step": 39290
     },
     {
       "epoch": 118.0,
-      "eval_accuracy": 0.8088,
-      "eval_loss": 0.6239674687385559,
-      "eval_runtime": 5.497,
-      "eval_samples_per_second": 1819.182,
-      "eval_steps_per_second": 7.277,
+      "eval_accuracy": 0.8073,
+      "eval_loss": 0.6232763528823853,
+      "eval_runtime": 5.1144,
+      "eval_samples_per_second": 1955.262,
+      "eval_steps_per_second": 7.821,
       "step": 39294
     },
     {
       "epoch": 118.02,
-      "grad_norm": 6.4342570304870605,
+      "grad_norm": 6.023465156555176,
       "learning_rate": 6.066066066066067e-06,
-      "loss": 1.1349,
+      "loss": 1.1271,
       "step": 39300
     },
     {
       "epoch": 118.05,
-      "grad_norm": 7.875779151916504,
+      "grad_norm": 7.856538772583008,
       "learning_rate": 6.065065065065065e-06,
-      "loss": 1.0447,
+      "loss": 1.0481,
       "step": 39310
     },
     {
       "epoch": 118.08,
-      "grad_norm": 7.834926605224609,
+      "grad_norm": 7.659352779388428,
       "learning_rate": 6.064064064064065e-06,
-      "loss": 1.0764,
+      "loss": 1.0715,
       "step": 39320
     },
     {
       "epoch": 118.11,
-      "grad_norm": 7.26174259185791,
+      "grad_norm": 7.624958038330078,
       "learning_rate": 6.063063063063064e-06,
-      "loss": 1.0928,
+      "loss": 1.0937,
       "step": 39330
     },
     {
       "epoch": 118.14,
-      "grad_norm": 7.114091396331787,
+      "grad_norm": 7.968799591064453,
       "learning_rate": 6.062062062062063e-06,
       "loss": 1.0205,
       "step": 39340
     },
     {
       "epoch": 118.17,
-      "grad_norm": 9.00168514251709,
+      "grad_norm": 8.86004638671875,
       "learning_rate": 6.061061061061062e-06,
-      "loss": 1.031,
+      "loss": 1.024,
       "step": 39350
     },
     {
       "epoch": 118.2,
-      "grad_norm": 7.262809753417969,
+      "grad_norm": 8.594239234924316,
       "learning_rate": 6.06006006006006e-06,
-      "loss": 1.1218,
+      "loss": 1.1098,
       "step": 39360
     },
     {
       "epoch": 118.23,
-      "grad_norm": 9.113847732543945,
+      "grad_norm": 10.566232681274414,
       "learning_rate": 6.059059059059059e-06,
-      "loss": 1.1098,
+      "loss": 1.1097,
       "step": 39370
     },
     {
       "epoch": 118.26,
-      "grad_norm": 4.905179977416992,
+      "grad_norm": 7.97852087020874,
       "learning_rate": 6.058058058058058e-06,
-      "loss": 1.0902,
+      "loss": 1.0764,
       "step": 39380
     },
     {
       "epoch": 118.29,
-      "grad_norm": 9.638078689575195,
+      "grad_norm": 7.437664031982422,
       "learning_rate": 6.057057057057057e-06,
-      "loss": 1.071,
+      "loss": 1.0739,
       "step": 39390
     },
     {
       "epoch": 118.32,
-      "grad_norm": 7.098802089691162,
+      "grad_norm": 9.770959854125977,
       "learning_rate": 6.056056056056057e-06,
-      "loss": 1.0691,
+      "loss": 1.071,
       "step": 39400
     },
     {
       "epoch": 118.35,
-      "grad_norm": 8.522515296936035,
+      "grad_norm": 6.6590657234191895,
       "learning_rate": 6.055055055055055e-06,
-      "loss": 1.1193,
+      "loss": 1.098,
       "step": 39410
     },
     {
       "epoch": 118.38,
-      "grad_norm": 7.333311080932617,
+      "grad_norm": 6.111998558044434,
       "learning_rate": 6.054054054054055e-06,
-      "loss": 1.0645,
+      "loss": 1.0531,
       "step": 39420
     },
     {
       "epoch": 118.41,
-      "grad_norm": 8.981118202209473,
+      "grad_norm": 13.720052719116211,
       "learning_rate": 6.053053053053054e-06,
-      "loss": 1.135,
+      "loss": 1.1316,
       "step": 39430
     },
     {
       "epoch": 118.44,
-      "grad_norm": 8.207225799560547,
+      "grad_norm": 6.890731334686279,
       "learning_rate": 6.052052052052052e-06,
-      "loss": 1.0526,
+      "loss": 1.0453,
       "step": 39440
     },
     {
       "epoch": 118.47,
-      "grad_norm": 6.698736667633057,
+      "grad_norm": 7.537546634674072,
       "learning_rate": 6.0510510510510515e-06,
-      "loss": 1.0784,
+      "loss": 1.0812,
       "step": 39450
     },
     {
       "epoch": 118.5,
-      "grad_norm": 6.494556427001953,
+      "grad_norm": 7.60024356842041,
       "learning_rate": 6.05005005005005e-06,
-      "loss": 1.1358,
+      "loss": 1.1445,
       "step": 39460
     },
     {
       "epoch": 118.53,
-      "grad_norm": 6.311203956604004,
+      "grad_norm": 7.614530086517334,
       "learning_rate": 6.0490490490490495e-06,
       "loss": 1.1305,
       "step": 39470
     },
     {
       "epoch": 118.56,
-      "grad_norm": 7.563522815704346,
+      "grad_norm": 8.09738540649414,
       "learning_rate": 6.048048048048049e-06,
-      "loss": 1.1113,
+      "loss": 1.1054,
       "step": 39480
     },
     {
       "epoch": 118.59,
-      "grad_norm": 6.640514850616455,
+      "grad_norm": 6.9457221031188965,
       "learning_rate": 6.0470470470470476e-06,
-      "loss": 1.0979,
+      "loss": 1.0763,
       "step": 39490
     },
     {
       "epoch": 118.62,
-      "grad_norm": 5.24219274520874,
+      "grad_norm": 6.452393054962158,
       "learning_rate": 6.046046046046047e-06,
-      "loss": 1.0984,
+      "loss": 1.0951,
       "step": 39500
     },
     {
       "epoch": 118.65,
-      "grad_norm": 5.727725505828857,
+      "grad_norm": 6.380261421203613,
       "learning_rate": 6.045045045045046e-06,
-      "loss": 1.0965,
+      "loss": 1.1051,
       "step": 39510
     },
     {
       "epoch": 118.68,
-      "grad_norm": 7.027763843536377,
+      "grad_norm": 7.8393168449401855,
       "learning_rate": 6.044044044044044e-06,
-      "loss": 1.0881,
+      "loss": 1.0948,
       "step": 39520
     },
     {
       "epoch": 118.71,
-      "grad_norm": 8.315985679626465,
+      "grad_norm": 6.072793483734131,
       "learning_rate": 6.043043043043043e-06,
-      "loss": 1.0683,
+      "loss": 1.053,
       "step": 39530
     },
     {
       "epoch": 118.74,
-      "grad_norm": 6.415372848510742,
+      "grad_norm": 7.581340312957764,
       "learning_rate": 6.042042042042042e-06,
-      "loss": 1.0601,
+      "loss": 1.0593,
       "step": 39540
     },
     {
       "epoch": 118.77,
-      "grad_norm": 5.143367767333984,
+      "grad_norm": 6.64141845703125,
       "learning_rate": 6.041041041041042e-06,
-      "loss": 0.9988,
+      "loss": 0.9983,
       "step": 39550
     },
     {
       "epoch": 118.8,
-      "grad_norm": 6.428032398223877,
+      "grad_norm": 5.777500629425049,
       "learning_rate": 6.04004004004004e-06,
-      "loss": 1.0257,
+      "loss": 1.0278,
       "step": 39560
     },
     {
       "epoch": 118.83,
-      "grad_norm": 8.352897644042969,
+      "grad_norm": 7.419256210327148,
       "learning_rate": 6.03903903903904e-06,
-      "loss": 1.0587,
+      "loss": 1.0677,
       "step": 39570
     },
     {
       "epoch": 118.86,
-      "grad_norm": 6.79310941696167,
+      "grad_norm": 7.181364059448242,
       "learning_rate": 6.038038038038039e-06,
-      "loss": 1.0616,
+      "loss": 1.0604,
       "step": 39580
     },
     {
       "epoch": 118.89,
-      "grad_norm": 5.323863983154297,
+      "grad_norm": 6.380545616149902,
       "learning_rate": 6.037037037037038e-06,
-      "loss": 1.0313,
+      "loss": 1.0311,
       "step": 39590
     },
     {
       "epoch": 118.92,
-      "grad_norm": 6.150847911834717,
+      "grad_norm": 6.605906009674072,
       "learning_rate": 6.036036036036037e-06,
-      "loss": 0.9946,
+      "loss": 1.0066,
       "step": 39600
     },
     {
       "epoch": 118.95,
-      "grad_norm": 6.2486090660095215,
+      "grad_norm": 9.63383674621582,
       "learning_rate": 6.035035035035035e-06,
-      "loss": 1.0714,
+      "loss": 1.07,
       "step": 39610
     },
     {
       "epoch": 118.98,
-      "grad_norm": 7.86260986328125,
+      "grad_norm": 9.457944869995117,
       "learning_rate": 6.034034034034034e-06,
-      "loss": 1.1539,
+      "loss": 1.1481,
       "step": 39620
     },
     {
       "epoch": 119.0,
-      "eval_accuracy": 0.8064,
-      "eval_loss": 0.625173032283783,
-      "eval_runtime": 5.4162,
-      "eval_samples_per_second": 1846.296,
-      "eval_steps_per_second": 7.385,
+      "eval_accuracy": 0.8086,
+      "eval_loss": 0.6227534413337708,
+      "eval_runtime": 5.278,
+      "eval_samples_per_second": 1894.647,
+      "eval_steps_per_second": 7.579,
       "step": 39627
     },
     {
       "epoch": 119.01,
-      "grad_norm": 11.121772766113281,
+      "grad_norm": 9.511966705322266,
       "learning_rate": 6.033033033033033e-06,
-      "loss": 1.1084,
+      "loss": 1.0872,
       "step": 39630
     },
     {
       "epoch": 119.04,
-      "grad_norm": 6.533758163452148,
+      "grad_norm": 9.70907211303711,
       "learning_rate": 6.032032032032032e-06,
-      "loss": 1.0921,
+      "loss": 1.0882,
       "step": 39640
     },
     {
       "epoch": 119.07,
-      "grad_norm": 8.334358215332031,
+      "grad_norm": 5.860357761383057,
       "learning_rate": 6.031031031031032e-06,
-      "loss": 1.1068,
+      "loss": 1.1015,
       "step": 39650
     },
     {
       "epoch": 119.1,
-      "grad_norm": 5.614645004272461,
+      "grad_norm": 5.563657283782959,
       "learning_rate": 6.0300300300300304e-06,
-      "loss": 1.0869,
+      "loss": 1.0837,
       "step": 39660
     },
     {
       "epoch": 119.13,
-      "grad_norm": 6.099480152130127,
+      "grad_norm": 6.161004066467285,
       "learning_rate": 6.02902902902903e-06,
-      "loss": 1.0685,
+      "loss": 1.0694,
       "step": 39670
     },
     {
       "epoch": 119.16,
-      "grad_norm": 7.068170547485352,
+      "grad_norm": 8.168277740478516,
       "learning_rate": 6.028028028028029e-06,
-      "loss": 1.1433,
+      "loss": 1.1431,
       "step": 39680
     },
     {
       "epoch": 119.19,
-      "grad_norm": 6.6027398109436035,
+      "grad_norm": 7.609015464782715,
       "learning_rate": 6.027027027027027e-06,
-      "loss": 1.064,
+      "loss": 1.065,
       "step": 39690
     },
     {
       "epoch": 119.22,
-      "grad_norm": 9.000941276550293,
+      "grad_norm": 10.84853744506836,
       "learning_rate": 6.0260260260260265e-06,
-      "loss": 1.1551,
+      "loss": 1.1416,
       "step": 39700
     },
     {
       "epoch": 119.25,
-      "grad_norm": 7.971297264099121,
+      "grad_norm": 6.894078254699707,
       "learning_rate": 6.025025025025025e-06,
-      "loss": 1.0901,
+      "loss": 1.0795,
       "step": 39710
     },
     {
       "epoch": 119.28,
-      "grad_norm": 6.541400909423828,
+      "grad_norm": 7.1685919761657715,
       "learning_rate": 6.0240240240240246e-06,
-      "loss": 1.0923,
+      "loss": 1.0846,
       "step": 39720
     },
     {
       "epoch": 119.31,
-      "grad_norm": 6.208627223968506,
+      "grad_norm": 7.1141862869262695,
       "learning_rate": 6.023023023023023e-06,
-      "loss": 1.0776,
+      "loss": 1.077,
       "step": 39730
     },
     {
       "epoch": 119.34,
-      "grad_norm": 8.132689476013184,
+      "grad_norm": 8.982442855834961,
       "learning_rate": 6.022022022022023e-06,
-      "loss": 1.1151,
+      "loss": 1.1212,
       "step": 39740
     },
     {
       "epoch": 119.37,
-      "grad_norm": 6.474180221557617,
+      "grad_norm": 6.61651611328125,
       "learning_rate": 6.021021021021022e-06,
-      "loss": 1.06,
+      "loss": 1.0436,
       "step": 39750
     },
     {
       "epoch": 119.4,
-      "grad_norm": 6.306558132171631,
+      "grad_norm": 9.964186668395996,
       "learning_rate": 6.020020020020021e-06,
-      "loss": 1.0216,
+      "loss": 1.0111,
       "step": 39760
     },
     {
       "epoch": 119.43,
-      "grad_norm": 6.275662422180176,
+      "grad_norm": 7.875288963317871,
       "learning_rate": 6.019019019019019e-06,
-      "loss": 1.0771,
+      "loss": 1.07,
       "step": 39770
     },
     {
       "epoch": 119.46,
-      "grad_norm": 6.72388219833374,
+      "grad_norm": 6.1498541831970215,
       "learning_rate": 6.018018018018018e-06,
-      "loss": 1.054,
+      "loss": 1.0449,
       "step": 39780
     },
     {
       "epoch": 119.49,
-      "grad_norm": 7.092596530914307,
+      "grad_norm": 5.558241844177246,
       "learning_rate": 6.017017017017017e-06,
-      "loss": 1.0638,
+      "loss": 1.0625,
       "step": 39790
     },
     {
       "epoch": 119.52,
-      "grad_norm": 8.308540344238281,
+      "grad_norm": 6.261958599090576,
       "learning_rate": 6.016016016016017e-06,
-      "loss": 1.232,
+      "loss": 1.2321,
       "step": 39800
     },
     {
       "epoch": 119.55,
-      "grad_norm": 8.732039451599121,
+      "grad_norm": 8.487439155578613,
       "learning_rate": 6.015015015015015e-06,
-      "loss": 1.1077,
+      "loss": 1.0962,
       "step": 39810
     },
     {
       "epoch": 119.58,
-      "grad_norm": 7.602521896362305,
+      "grad_norm": 7.086191177368164,
       "learning_rate": 6.014014014014015e-06,
-      "loss": 1.0633,
+      "loss": 1.0687,
       "step": 39820
     },
     {
       "epoch": 119.61,
-      "grad_norm": 6.90560245513916,
+      "grad_norm": 6.217952728271484,
       "learning_rate": 6.013013013013014e-06,
-      "loss": 1.0324,
+      "loss": 1.0236,
       "step": 39830
     },
     {
       "epoch": 119.64,
-      "grad_norm": 7.1514387130737305,
+      "grad_norm": 11.159260749816895,
       "learning_rate": 6.012012012012013e-06,
-      "loss": 1.1457,
+      "loss": 1.1465,
       "step": 39840
     },
     {
       "epoch": 119.67,
-      "grad_norm": 6.828918933868408,
+      "grad_norm": 7.659945964813232,
       "learning_rate": 6.011011011011012e-06,
-      "loss": 1.0286,
+      "loss": 1.0267,
       "step": 39850
     },
     {
       "epoch": 119.7,
-      "grad_norm": 7.941656589508057,
+      "grad_norm": 11.73331069946289,
       "learning_rate": 6.01001001001001e-06,
-      "loss": 1.0182,
+      "loss": 1.02,
       "step": 39860
     },
     {
       "epoch": 119.73,
-      "grad_norm": 9.062158584594727,
+      "grad_norm": 8.66811752319336,
       "learning_rate": 6.009009009009009e-06,
-      "loss": 1.1582,
+      "loss": 1.1388,
       "step": 39870
     },
     {
       "epoch": 119.76,
-      "grad_norm": 5.403141498565674,
+      "grad_norm": 5.475318908691406,
       "learning_rate": 6.008008008008008e-06,
-      "loss": 1.1024,
+      "loss": 1.0985,
       "step": 39880
     },
     {
       "epoch": 119.79,
-      "grad_norm": 7.4237236976623535,
+      "grad_norm": 8.370485305786133,
       "learning_rate": 6.0070070070070074e-06,
-      "loss": 1.041,
+      "loss": 1.0382,
       "step": 39890
     },
     {
       "epoch": 119.82,
-      "grad_norm": 9.719712257385254,
+      "grad_norm": 8.50804328918457,
       "learning_rate": 6.006006006006007e-06,
-      "loss": 1.0662,
+      "loss": 1.0643,
       "step": 39900
     },
     {
       "epoch": 119.85,
-      "grad_norm": 5.32233190536499,
+      "grad_norm": 6.808847904205322,
       "learning_rate": 6.0050050050050055e-06,
-      "loss": 1.0704,
+      "loss": 1.0632,
       "step": 39910
     },
     {
       "epoch": 119.88,
-      "grad_norm": 6.71970272064209,
+      "grad_norm": 6.422298908233643,
       "learning_rate": 6.004004004004005e-06,
-      "loss": 1.086,
+      "loss": 1.0849,
       "step": 39920
     },
     {
       "epoch": 119.91,
-      "grad_norm": 6.8390045166015625,
+      "grad_norm": 7.84655237197876,
       "learning_rate": 6.003003003003004e-06,
-      "loss": 1.0195,
+      "loss": 1.0176,
       "step": 39930
     },
     {
       "epoch": 119.94,
-      "grad_norm": 8.468725204467773,
+      "grad_norm": 12.13757610321045,
       "learning_rate": 6.002002002002002e-06,
-      "loss": 1.0263,
+      "loss": 1.0259,
       "step": 39940
     },
     {
       "epoch": 119.97,
-      "grad_norm": 6.629939556121826,
+      "grad_norm": 8.024450302124023,
       "learning_rate": 6.0010010010010016e-06,
-      "loss": 1.0741,
+      "loss": 1.0791,
       "step": 39950
     },
     {
       "epoch": 120.0,
-      "grad_norm": 45.852386474609375,
+      "grad_norm": 98.91258239746094,
       "learning_rate": 6e-06,
-      "loss": 1.3025,
+      "loss": 1.3129,
       "step": 39960
     },
     {
       "epoch": 120.0,
-      "eval_accuracy": 0.805,
-      "eval_loss": 0.6305273771286011,
-      "eval_runtime": 5.3369,
-      "eval_samples_per_second": 1873.762,
-      "eval_steps_per_second": 7.495,
+      "eval_accuracy": 0.807,
+      "eval_loss": 0.6295485496520996,
+      "eval_runtime": 5.2639,
+      "eval_samples_per_second": 1899.733,
+      "eval_steps_per_second": 7.599,
       "step": 39960
     },
     {
       "epoch": 120.03,
-      "grad_norm": 7.60516881942749,
+      "grad_norm": 5.482144832611084,
       "learning_rate": 5.998998998999e-06,
-      "loss": 1.0843,
+      "loss": 1.0766,
       "step": 39970
     },
     {
       "epoch": 120.06,
-      "grad_norm": 5.911736011505127,
+      "grad_norm": 8.766833305358887,
       "learning_rate": 5.997997997997998e-06,
-      "loss": 1.063,
+      "loss": 1.0671,
       "step": 39980
     },
     {
       "epoch": 120.09,
-      "grad_norm": 9.706130981445312,
+      "grad_norm": 11.36038875579834,
       "learning_rate": 5.996996996996998e-06,
-      "loss": 1.0807,
+      "loss": 1.0694,
       "step": 39990
     },
     {
       "epoch": 120.12,
-      "grad_norm": 5.048537254333496,
+      "grad_norm": 6.535454750061035,
       "learning_rate": 5.995995995995997e-06,
-      "loss": 1.038,
+      "loss": 1.0422,
       "step": 40000
     },
     {
       "epoch": 120.15,
-      "grad_norm": 5.515662670135498,
+      "grad_norm": 8.556401252746582,
       "learning_rate": 5.994994994994996e-06,
-      "loss": 1.0579,
+      "loss": 1.0646,
       "step": 40010
     },
     {
       "epoch": 120.18,
-      "grad_norm": 8.101012229919434,
+      "grad_norm": 7.081175327301025,
       "learning_rate": 5.993993993993994e-06,
-      "loss": 1.0785,
+      "loss": 1.0727,
       "step": 40020
     },
     {
       "epoch": 120.21,
-      "grad_norm": 5.762811183929443,
+      "grad_norm": 6.378342628479004,
       "learning_rate": 5.992992992992993e-06,
-      "loss": 1.0243,
+      "loss": 1.0194,
       "step": 40030
     },
     {
       "epoch": 120.24,
-      "grad_norm": 6.965129852294922,
+      "grad_norm": 8.753281593322754,
       "learning_rate": 5.991991991991992e-06,
-      "loss": 1.0637,
+      "loss": 1.0546,
       "step": 40040
     },
     {
       "epoch": 120.27,
-      "grad_norm": 4.904429912567139,
+      "grad_norm": 6.256892681121826,
       "learning_rate": 5.990990990990992e-06,
-      "loss": 1.047,
+      "loss": 1.0433,
       "step": 40050
     },
     {
       "epoch": 120.3,
-      "grad_norm": 7.7573723793029785,
+      "grad_norm": 7.375635147094727,
       "learning_rate": 5.98998998998999e-06,
-      "loss": 1.0135,
+      "loss": 1.0124,
       "step": 40060
     },
     {
       "epoch": 120.33,
-      "grad_norm": 8.883901596069336,
+      "grad_norm": 10.434895515441895,
       "learning_rate": 5.98898898898899e-06,
-      "loss": 1.04,
+      "loss": 1.0337,
       "step": 40070
     },
     {
       "epoch": 120.36,
-      "grad_norm": 5.610440254211426,
+      "grad_norm": 6.585990905761719,
       "learning_rate": 5.987987987987988e-06,
-      "loss": 1.0688,
+      "loss": 1.0662,
       "step": 40080
     },
     {
       "epoch": 120.39,
-      "grad_norm": 7.453373432159424,
+      "grad_norm": 8.310700416564941,
       "learning_rate": 5.986986986986988e-06,
-      "loss": 1.049,
+      "loss": 1.0401,
       "step": 40090
     },
     {
       "epoch": 120.42,
-      "grad_norm": 7.451935291290283,
+      "grad_norm": 7.074868679046631,
       "learning_rate": 5.985985985985987e-06,
-      "loss": 1.0207,
+      "loss": 1.0111,
       "step": 40100
     },
     {
       "epoch": 120.45,
-      "grad_norm": 8.98121452331543,
+      "grad_norm": 8.044998168945312,
       "learning_rate": 5.984984984984985e-06,
-      "loss": 1.0739,
+      "loss": 1.0612,
       "step": 40110
     },
     {
       "epoch": 120.48,
-      "grad_norm": 5.122777938842773,
+      "grad_norm": 6.736636161804199,
       "learning_rate": 5.9839839839839844e-06,
-      "loss": 1.0669,
+      "loss": 1.0575,
       "step": 40120
     },
     {
       "epoch": 120.51,
-      "grad_norm": 6.00630521774292,
+      "grad_norm": 8.246712684631348,
       "learning_rate": 5.982982982982983e-06,
-      "loss": 1.0179,
+      "loss": 1.0177,
       "step": 40130
     },
     {
       "epoch": 120.54,
-      "grad_norm": 7.797217845916748,
+      "grad_norm": 8.85960865020752,
       "learning_rate": 5.9819819819819825e-06,
-      "loss": 1.1583,
+      "loss": 1.1571,
       "step": 40140
     },
     {
       "epoch": 120.57,
-      "grad_norm": 6.868021011352539,
+      "grad_norm": 6.541440963745117,
       "learning_rate": 5.980980980980982e-06,
-      "loss": 1.0794,
+      "loss": 1.0789,
       "step": 40150
     },
     {
       "epoch": 120.6,
-      "grad_norm": 6.365246295928955,
+      "grad_norm": 7.554319858551025,
       "learning_rate": 5.9799799799799805e-06,
-      "loss": 1.0355,
+      "loss": 1.0301,
       "step": 40160
     },
     {
       "epoch": 120.63,
-      "grad_norm": 6.306880474090576,
+      "grad_norm": 6.77114725112915,
       "learning_rate": 5.97897897897898e-06,
-      "loss": 1.1718,
+      "loss": 1.1601,
       "step": 40170
     },
     {
       "epoch": 120.66,
-      "grad_norm": 6.286937236785889,
+      "grad_norm": 6.893423557281494,
       "learning_rate": 5.9779779779779786e-06,
-      "loss": 1.1156,
+      "loss": 1.1042,
       "step": 40180
     },
     {
       "epoch": 120.69,
-      "grad_norm": 5.8247971534729,
+      "grad_norm": 4.3482666015625,
       "learning_rate": 5.976976976976977e-06,
-      "loss": 1.0269,
+      "loss": 1.0237,
       "step": 40190
     },
     {
       "epoch": 120.72,
-      "grad_norm": 8.972111701965332,
+      "grad_norm": 9.065352439880371,
       "learning_rate": 5.975975975975976e-06,
-      "loss": 1.1576,
+      "loss": 1.1531,
       "step": 40200
     },
     {
       "epoch": 120.75,
-      "grad_norm": 9.059359550476074,
+      "grad_norm": 9.261029243469238,
       "learning_rate": 5.974974974974975e-06,
-      "loss": 1.0793,
+      "loss": 1.0844,
       "step": 40210
     },
     {
       "epoch": 120.78,
-      "grad_norm": 5.147841453552246,
+      "grad_norm": 5.972881317138672,
       "learning_rate": 5.973973973973975e-06,
-      "loss": 1.0942,
+      "loss": 1.0931,
       "step": 40220
     },
     {
       "epoch": 120.81,
-      "grad_norm": 7.460197925567627,
+      "grad_norm": 7.228552341461182,
       "learning_rate": 5.972972972972973e-06,
-      "loss": 1.1372,
+      "loss": 1.1269,
       "step": 40230
     },
     {
       "epoch": 120.84,
-      "grad_norm": 6.4169230461120605,
+      "grad_norm": 7.854869842529297,
       "learning_rate": 5.971971971971973e-06,
-      "loss": 1.1439,
+      "loss": 1.1478,
       "step": 40240
     },
     {
       "epoch": 120.87,
-      "grad_norm": 7.69524621963501,
+      "grad_norm": 7.62755823135376,
       "learning_rate": 5.970970970970972e-06,
-      "loss": 1.0756,
+      "loss": 1.0788,
       "step": 40250
     },
     {
       "epoch": 120.9,
-      "grad_norm": 6.57259464263916,
+      "grad_norm": 6.834786891937256,
       "learning_rate": 5.969969969969971e-06,
-      "loss": 1.0927,
+      "loss": 1.0969,
       "step": 40260
     },
     {
       "epoch": 120.93,
-      "grad_norm": 6.598838806152344,
+      "grad_norm": 5.635851860046387,
       "learning_rate": 5.968968968968969e-06,
-      "loss": 1.0351,
+      "loss": 1.0404,
       "step": 40270
     },
     {
       "epoch": 120.96,
-      "grad_norm": 7.719028472900391,
+      "grad_norm": 9.566873550415039,
       "learning_rate": 5.967967967967968e-06,
-      "loss": 1.0666,
+      "loss": 1.0725,
       "step": 40280
     },
     {
       "epoch": 120.99,
-      "grad_norm": 6.1904473304748535,
+      "grad_norm": 5.941823482513428,
       "learning_rate": 5.966966966966967e-06,
-      "loss": 1.0533,
+      "loss": 1.0525,
       "step": 40290
     },
     {
       "epoch": 121.0,
-      "eval_accuracy": 0.8065,
-      "eval_loss": 0.6283848285675049,
-      "eval_runtime": 5.4024,
-      "eval_samples_per_second": 1851.019,
-      "eval_steps_per_second": 7.404,
+      "eval_accuracy": 0.8088,
+      "eval_loss": 0.6273288726806641,
+      "eval_runtime": 5.4293,
+      "eval_samples_per_second": 1841.864,
+      "eval_steps_per_second": 7.367,
       "step": 40293
     },
     {
       "epoch": 121.02,
-      "grad_norm": 6.163480281829834,
+      "grad_norm": 9.127039909362793,
       "learning_rate": 5.965965965965966e-06,
-      "loss": 1.212,
+      "loss": 1.2103,
       "step": 40300
     },
     {
       "epoch": 121.05,
-      "grad_norm": 6.205991744995117,
+      "grad_norm": 6.968634128570557,
       "learning_rate": 5.964964964964965e-06,
-      "loss": 1.0621,
+      "loss": 1.0452,
       "step": 40310
     },
     {
       "epoch": 121.08,
-      "grad_norm": 6.016970634460449,
+      "grad_norm": 6.112205505371094,
       "learning_rate": 5.963963963963965e-06,
-      "loss": 1.0252,
+      "loss": 1.0147,
       "step": 40320
     },
     {
       "epoch": 121.11,
-      "grad_norm": 7.503122806549072,
+      "grad_norm": 7.897352695465088,
       "learning_rate": 5.962962962962963e-06,
-      "loss": 1.0992,
+      "loss": 1.0922,
       "step": 40330
     },
     {
       "epoch": 121.14,
-      "grad_norm": 7.458715438842773,
+      "grad_norm": 8.1462984085083,
       "learning_rate": 5.961961961961963e-06,
-      "loss": 1.0316,
+      "loss": 1.0303,
       "step": 40340
     },
     {
       "epoch": 121.17,
-      "grad_norm": 6.201898097991943,
+      "grad_norm": 7.887652397155762,
       "learning_rate": 5.960960960960962e-06,
-      "loss": 1.0592,
+      "loss": 1.0606,
       "step": 40350
     },
     {
       "epoch": 121.2,
-      "grad_norm": 7.301377773284912,
+      "grad_norm": 6.29365873336792,
       "learning_rate": 5.95995995995996e-06,
-      "loss": 1.0673,
+      "loss": 1.0784,
       "step": 40360
     },
     {
       "epoch": 121.23,
-      "grad_norm": 7.4832048416137695,
+      "grad_norm": 6.559881210327148,
       "learning_rate": 5.9589589589589595e-06,
-      "loss": 0.9986,
+      "loss": 1.004,
       "step": 40370
     },
     {
       "epoch": 121.26,
-      "grad_norm": 8.526287078857422,
+      "grad_norm": 11.348655700683594,
       "learning_rate": 5.957957957957958e-06,
-      "loss": 1.1271,
+      "loss": 1.1227,
       "step": 40380
     },
     {
       "epoch": 121.29,
-      "grad_norm": 6.2320637702941895,
+      "grad_norm": 5.875394821166992,
       "learning_rate": 5.9569569569569575e-06,
-      "loss": 0.9814,
+      "loss": 0.9715,
       "step": 40390
     },
     {
       "epoch": 121.32,
-      "grad_norm": 7.706111431121826,
+      "grad_norm": 7.30016565322876,
       "learning_rate": 5.955955955955957e-06,
-      "loss": 1.0112,
+      "loss": 0.9982,
       "step": 40400
     },
     {
       "epoch": 121.35,
-      "grad_norm": 5.891811370849609,
+      "grad_norm": 6.620288372039795,
       "learning_rate": 5.9549549549549556e-06,
-      "loss": 1.1293,
+      "loss": 1.1275,
       "step": 40410
     },
     {
       "epoch": 121.38,
-      "grad_norm": 7.885891437530518,
+      "grad_norm": 7.9077839851379395,
       "learning_rate": 5.953953953953955e-06,
-      "loss": 1.0379,
+      "loss": 1.0338,
       "step": 40420
     },
     {
       "epoch": 121.41,
-      "grad_norm": 7.298081398010254,
+      "grad_norm": 8.406597137451172,
       "learning_rate": 5.952952952952954e-06,
-      "loss": 1.0888,
+      "loss": 1.0936,
       "step": 40430
     },
     {
       "epoch": 121.44,
-      "grad_norm": 6.541845321655273,
+      "grad_norm": 6.524200439453125,
       "learning_rate": 5.951951951951952e-06,
-      "loss": 0.9889,
+      "loss": 1.0045,
       "step": 40440
     },
     {
       "epoch": 121.47,
-      "grad_norm": 7.653921127319336,
+      "grad_norm": 7.083986282348633,
       "learning_rate": 5.950950950950951e-06,
-      "loss": 1.089,
+      "loss": 1.0843,
       "step": 40450
     },
     {
       "epoch": 121.5,
-      "grad_norm": 7.710638523101807,
+      "grad_norm": 6.720353603363037,
       "learning_rate": 5.94994994994995e-06,
-      "loss": 1.0785,
+      "loss": 1.0701,
       "step": 40460
     },
     {
       "epoch": 121.53,
-      "grad_norm": 6.071182727813721,
+      "grad_norm": 6.197246074676514,
       "learning_rate": 5.94894894894895e-06,
-      "loss": 1.0814,
+      "loss": 1.0862,
       "step": 40470
     },
     {
       "epoch": 121.56,
-      "grad_norm": 7.421065330505371,
+      "grad_norm": 7.973300933837891,
       "learning_rate": 5.947947947947948e-06,
-      "loss": 1.1711,
+      "loss": 1.1717,
       "step": 40480
     },
     {
       "epoch": 121.59,
-      "grad_norm": 5.807744026184082,
+      "grad_norm": 6.652436256408691,
       "learning_rate": 5.946946946946948e-06,
-      "loss": 1.1416,
+      "loss": 1.1428,
       "step": 40490
     },
     {
       "epoch": 121.62,
-      "grad_norm": 7.822805404663086,
+      "grad_norm": 7.697293281555176,
       "learning_rate": 5.945945945945947e-06,
-      "loss": 1.101,
+      "loss": 1.0956,
       "step": 40500
     },
     {
       "epoch": 121.65,
-      "grad_norm": 4.989006519317627,
+      "grad_norm": 5.614060401916504,
       "learning_rate": 5.944944944944946e-06,
-      "loss": 1.0952,
+      "loss": 1.0973,
       "step": 40510
     },
     {
       "epoch": 121.68,
-      "grad_norm": 6.492006778717041,
+      "grad_norm": 8.017306327819824,
       "learning_rate": 5.943943943943944e-06,
-      "loss": 1.079,
+      "loss": 1.0776,
       "step": 40520
     },
     {
       "epoch": 121.71,
-      "grad_norm": 6.012423515319824,
+      "grad_norm": 5.49555778503418,
       "learning_rate": 5.942942942942943e-06,
-      "loss": 1.0681,
+      "loss": 1.0711,
       "step": 40530
     },
     {
       "epoch": 121.74,
-      "grad_norm": 6.595686435699463,
+      "grad_norm": 6.104950904846191,
       "learning_rate": 5.941941941941942e-06,
-      "loss": 1.0593,
+      "loss": 1.0668,
       "step": 40540
     },
     {
       "epoch": 121.77,
-      "grad_norm": 6.161447048187256,
+      "grad_norm": 7.696218013763428,
       "learning_rate": 5.940940940940941e-06,
-      "loss": 1.0461,
+      "loss": 1.0369,
       "step": 40550
     },
     {
       "epoch": 121.8,
-      "grad_norm": 5.184179782867432,
+      "grad_norm": 5.8477888107299805,
       "learning_rate": 5.93993993993994e-06,
-      "loss": 1.0528,
+      "loss": 1.0637,
       "step": 40560
     },
     {
       "epoch": 121.83,
-      "grad_norm": 5.936486721038818,
+      "grad_norm": 6.3194169998168945,
       "learning_rate": 5.93893893893894e-06,
-      "loss": 1.0506,
+      "loss": 1.054,
       "step": 40570
     },
     {
       "epoch": 121.86,
-      "grad_norm": 7.410540580749512,
+      "grad_norm": 11.24451732635498,
       "learning_rate": 5.9379379379379384e-06,
-      "loss": 1.091,
+      "loss": 1.0915,
       "step": 40580
     },
     {
       "epoch": 121.89,
-      "grad_norm": 6.471795082092285,
+      "grad_norm": 12.184638023376465,
       "learning_rate": 5.936936936936938e-06,
-      "loss": 1.0074,
+      "loss": 1.0109,
       "step": 40590
     },
     {
       "epoch": 121.92,
-      "grad_norm": 6.945420742034912,
+      "grad_norm": 5.720442295074463,
       "learning_rate": 5.935935935935936e-06,
-      "loss": 1.0749,
+      "loss": 1.0802,
       "step": 40600
     },
     {
       "epoch": 121.95,
-      "grad_norm": 10.045755386352539,
+      "grad_norm": 8.19951343536377,
       "learning_rate": 5.934934934934935e-06,
-      "loss": 1.0322,
+      "loss": 1.0243,
       "step": 40610
     },
     {
       "epoch": 121.98,
-      "grad_norm": 7.011258602142334,
+      "grad_norm": 7.172584056854248,
       "learning_rate": 5.9339339339339345e-06,
-      "loss": 0.9733,
+      "loss": 0.978,
       "step": 40620
     },
     {
       "epoch": 122.0,
-      "eval_accuracy": 0.8075,
-      "eval_loss": 0.6237444281578064,
-      "eval_runtime": 5.5069,
-      "eval_samples_per_second": 1815.908,
-      "eval_steps_per_second": 7.264,
+      "eval_accuracy": 0.8087,
+      "eval_loss": 0.623236894607544,
+      "eval_runtime": 5.3121,
+      "eval_samples_per_second": 1882.478,
+      "eval_steps_per_second": 7.53,
       "step": 40626
     },
     {
       "epoch": 122.01,
-      "grad_norm": 5.942174911499023,
+      "grad_norm": 7.356744289398193,
       "learning_rate": 5.932932932932933e-06,
-      "loss": 1.2162,
+      "loss": 1.2275,
       "step": 40630
     },
     {
       "epoch": 122.04,
-      "grad_norm": 6.4712629318237305,
+      "grad_norm": 7.018731117248535,
       "learning_rate": 5.9319319319319326e-06,
-      "loss": 1.1305,
+      "loss": 1.1198,
       "step": 40640
     },
     {
       "epoch": 122.07,
-      "grad_norm": 6.057232856750488,
+      "grad_norm": 8.396256446838379,
       "learning_rate": 5.930930930930931e-06,
-      "loss": 1.0666,
+      "loss": 1.0749,
       "step": 40650
     },
     {
       "epoch": 122.1,
-      "grad_norm": 6.953725337982178,
+      "grad_norm": 6.928677082061768,
       "learning_rate": 5.929929929929931e-06,
-      "loss": 1.1415,
+      "loss": 1.1318,
       "step": 40660
     },
     {
       "epoch": 122.13,
-      "grad_norm": 6.829768180847168,
+      "grad_norm": 7.627328872680664,
       "learning_rate": 5.92892892892893e-06,
-      "loss": 1.0201,
+      "loss": 1.0122,
       "step": 40670
     },
     {
       "epoch": 122.16,
-      "grad_norm": 8.106306076049805,
+      "grad_norm": 7.264162063598633,
       "learning_rate": 5.927927927927929e-06,
-      "loss": 1.0653,
+      "loss": 1.0443,
       "step": 40680
     },
     {
       "epoch": 122.19,
-      "grad_norm": 7.4329705238342285,
+      "grad_norm": 5.325633525848389,
       "learning_rate": 5.926926926926927e-06,
-      "loss": 1.1084,
+      "loss": 1.1033,
       "step": 40690
     },
     {
       "epoch": 122.22,
-      "grad_norm": 7.227511882781982,
+      "grad_norm": 8.091411590576172,
       "learning_rate": 5.925925925925926e-06,
-      "loss": 1.1325,
+      "loss": 1.1308,
       "step": 40700
     },
     {
       "epoch": 122.25,
-      "grad_norm": 5.128061771392822,
+      "grad_norm": 5.207223892211914,
       "learning_rate": 5.924924924924925e-06,
-      "loss": 1.0919,
+      "loss": 1.0979,
       "step": 40710
     },
     {
       "epoch": 122.28,
-      "grad_norm": 7.4826154708862305,
+      "grad_norm": 7.589554786682129,
       "learning_rate": 5.923923923923925e-06,
-      "loss": 1.1149,
+      "loss": 1.1029,
       "step": 40720
     },
     {
       "epoch": 122.31,
-      "grad_norm": 5.959225177764893,
+      "grad_norm": 5.261590957641602,
       "learning_rate": 5.922922922922923e-06,
-      "loss": 1.0863,
+      "loss": 1.0833,
       "step": 40730
     },
     {
       "epoch": 122.34,
-      "grad_norm": 7.300321578979492,
+      "grad_norm": 8.65865707397461,
       "learning_rate": 5.921921921921923e-06,
-      "loss": 1.1172,
+      "loss": 1.1169,
       "step": 40740
     },
     {
       "epoch": 122.37,
-      "grad_norm": 5.448154449462891,
+      "grad_norm": 7.725284099578857,
       "learning_rate": 5.920920920920921e-06,
-      "loss": 1.117,
+      "loss": 1.1367,
       "step": 40750
     },
     {
       "epoch": 122.4,
-      "grad_norm": 7.517981052398682,
+      "grad_norm": 6.5229034423828125,
       "learning_rate": 5.919919919919921e-06,
-      "loss": 1.021,
+      "loss": 1.0227,
       "step": 40760
     },
     {
       "epoch": 122.43,
-      "grad_norm": 6.373562335968018,
+      "grad_norm": 6.872967720031738,
       "learning_rate": 5.9189189189189185e-06,
-      "loss": 1.0619,
+      "loss": 1.059,
       "step": 40770
     },
     {
       "epoch": 122.46,
-      "grad_norm": 6.470538139343262,
+      "grad_norm": 5.828187942504883,
       "learning_rate": 5.917917917917918e-06,
-      "loss": 1.0735,
+      "loss": 1.0689,
       "step": 40780
     },
     {
       "epoch": 122.49,
-      "grad_norm": 5.32407808303833,
+      "grad_norm": 5.456592559814453,
       "learning_rate": 5.916916916916917e-06,
-      "loss": 1.0476,
+      "loss": 1.0428,
       "step": 40790
     },
     {
       "epoch": 122.52,
-      "grad_norm": 6.363615989685059,
+      "grad_norm": 6.7998785972595215,
       "learning_rate": 5.915915915915916e-06,
-      "loss": 1.1105,
+      "loss": 1.1063,
       "step": 40800
     },
     {
       "epoch": 122.55,
-      "grad_norm": 9.689352989196777,
+      "grad_norm": 9.292464256286621,
       "learning_rate": 5.9149149149149154e-06,
-      "loss": 1.1272,
+      "loss": 1.1113,
       "step": 40810
     },
     {
       "epoch": 122.58,
-      "grad_norm": 11.2840576171875,
+      "grad_norm": 5.984572410583496,
       "learning_rate": 5.913913913913915e-06,
-      "loss": 1.0047,
+      "loss": 1.008,
       "step": 40820
     },
     {
       "epoch": 122.61,
-      "grad_norm": 7.050837993621826,
+      "grad_norm": 6.656430244445801,
       "learning_rate": 5.9129129129129135e-06,
-      "loss": 1.0118,
+      "loss": 1.0245,
       "step": 40830
     },
     {
       "epoch": 122.64,
-      "grad_norm": 5.321043491363525,
+      "grad_norm": 7.254312992095947,
       "learning_rate": 5.911911911911913e-06,
-      "loss": 1.091,
+      "loss": 1.0912,
       "step": 40840
     },
     {
       "epoch": 122.67,
-      "grad_norm": 6.638062953948975,
+      "grad_norm": 7.263437271118164,
       "learning_rate": 5.910910910910911e-06,
-      "loss": 1.0467,
+      "loss": 1.0411,
       "step": 40850
     },
     {
       "epoch": 122.7,
-      "grad_norm": 4.170259952545166,
+      "grad_norm": 6.420060157775879,
       "learning_rate": 5.90990990990991e-06,
-      "loss": 1.0195,
+      "loss": 1.0164,
       "step": 40860
     },
     {
       "epoch": 122.73,
-      "grad_norm": 6.689126014709473,
+      "grad_norm": 9.275634765625,
       "learning_rate": 5.908908908908909e-06,
-      "loss": 1.0282,
+      "loss": 1.0353,
       "step": 40870
     },
     {
       "epoch": 122.76,
-      "grad_norm": 8.463713645935059,
+      "grad_norm": 8.931121826171875,
       "learning_rate": 5.907907907907908e-06,
-      "loss": 1.0296,
+      "loss": 1.0252,
       "step": 40880
     },
     {
       "epoch": 122.79,
-      "grad_norm": 7.434302806854248,
+      "grad_norm": 9.642162322998047,
       "learning_rate": 5.906906906906908e-06,
-      "loss": 1.0307,
+      "loss": 1.0098,
       "step": 40890
     },
     {
       "epoch": 122.82,
-      "grad_norm": 5.534347057342529,
+      "grad_norm": 5.173938751220703,
       "learning_rate": 5.905905905905906e-06,
-      "loss": 0.9988,
+      "loss": 1.0056,
       "step": 40900
     },
     {
       "epoch": 122.85,
-      "grad_norm": 9.356279373168945,
+      "grad_norm": 6.825785160064697,
       "learning_rate": 5.904904904904906e-06,
-      "loss": 1.1155,
+      "loss": 1.1054,
       "step": 40910
     },
     {
       "epoch": 122.88,
-      "grad_norm": 7.923691749572754,
+      "grad_norm": 6.632835865020752,
       "learning_rate": 5.903903903903905e-06,
-      "loss": 1.0244,
+      "loss": 1.0061,
       "step": 40920
     },
     {
       "epoch": 122.91,
-      "grad_norm": 7.455259799957275,
+      "grad_norm": 6.9716339111328125,
       "learning_rate": 5.902902902902904e-06,
-      "loss": 1.0276,
+      "loss": 1.0426,
       "step": 40930
     },
     {
       "epoch": 122.94,
-      "grad_norm": 5.345543384552002,
+      "grad_norm": 6.078738689422607,
       "learning_rate": 5.901901901901902e-06,
-      "loss": 1.0767,
+      "loss": 1.0739,
       "step": 40940
     },
     {
       "epoch": 122.97,
-      "grad_norm": 6.866259574890137,
+      "grad_norm": 5.6099138259887695,
       "learning_rate": 5.900900900900901e-06,
-      "loss": 1.0752,
+      "loss": 1.0759,
       "step": 40950
     },
     {
       "epoch": 123.0,
-      "eval_accuracy": 0.8098,
-      "eval_loss": 0.6218096613883972,
-      "eval_runtime": 5.7666,
-      "eval_samples_per_second": 1734.116,
-      "eval_steps_per_second": 6.936,
+      "eval_accuracy": 0.8097,
+      "eval_loss": 0.6218311786651611,
+      "eval_runtime": 5.3276,
+      "eval_samples_per_second": 1877.005,
+      "eval_steps_per_second": 7.508,
       "step": 40959
     },
     {
       "epoch": 123.0,
-      "grad_norm": 6.789834022521973,
+      "grad_norm": 6.56857442855835,
       "learning_rate": 5.8998998998999e-06,
-      "loss": 1.2813,
+      "loss": 1.2902,
       "step": 40960
     },
     {
       "epoch": 123.03,
-      "grad_norm": 7.741002559661865,
+      "grad_norm": 5.993311882019043,
       "learning_rate": 5.8988988988989e-06,
-      "loss": 1.0059,
+      "loss": 1.0036,
       "step": 40970
     },
     {
       "epoch": 123.06,
-      "grad_norm": 6.260262489318848,
+      "grad_norm": 8.41157341003418,
       "learning_rate": 5.897897897897898e-06,
-      "loss": 1.1238,
+      "loss": 1.1194,
       "step": 40980
     },
     {
       "epoch": 123.09,
-      "grad_norm": 9.12535285949707,
+      "grad_norm": 9.701403617858887,
       "learning_rate": 5.896896896896898e-06,
-      "loss": 1.0727,
+      "loss": 1.0637,
       "step": 40990
     },
     {
       "epoch": 123.12,
-      "grad_norm": 8.334635734558105,
+      "grad_norm": 6.822525978088379,
       "learning_rate": 5.895895895895896e-06,
-      "loss": 1.1049,
+      "loss": 1.0958,
       "step": 41000
     },
     {
       "epoch": 123.15,
-      "grad_norm": 6.178862571716309,
+      "grad_norm": 7.792285919189453,
       "learning_rate": 5.894894894894896e-06,
-      "loss": 1.0686,
+      "loss": 1.0539,
       "step": 41010
     },
     {
       "epoch": 123.18,
-      "grad_norm": 6.795135974884033,
+      "grad_norm": 6.874769687652588,
       "learning_rate": 5.8938938938938936e-06,
-      "loss": 1.0345,
+      "loss": 1.032,
       "step": 41020
     },
     {
       "epoch": 123.21,
-      "grad_norm": 10.760247230529785,
+      "grad_norm": 9.708645820617676,
       "learning_rate": 5.892892892892893e-06,
-      "loss": 1.1214,
+      "loss": 1.1142,
       "step": 41030
     },
     {
       "epoch": 123.24,
-      "grad_norm": 6.999403953552246,
+      "grad_norm": 7.8921003341674805,
       "learning_rate": 5.8918918918918924e-06,
-      "loss": 1.0728,
+      "loss": 1.0562,
       "step": 41040
     },
     {
       "epoch": 123.27,
-      "grad_norm": 5.922271251678467,
+      "grad_norm": 5.5052571296691895,
       "learning_rate": 5.890890890890891e-06,
-      "loss": 1.0481,
+      "loss": 1.0479,
       "step": 41050
     },
     {
       "epoch": 123.3,
-      "grad_norm": 8.0668306350708,
+      "grad_norm": 6.953878402709961,
       "learning_rate": 5.8898898898898905e-06,
-      "loss": 1.0732,
+      "loss": 1.0725,
       "step": 41060
     },
     {
       "epoch": 123.33,
-      "grad_norm": 6.252226829528809,
+      "grad_norm": 6.206379413604736,
       "learning_rate": 5.88888888888889e-06,
-      "loss": 1.1053,
+      "loss": 1.0957,
       "step": 41070
     },
     {
       "epoch": 123.36,
-      "grad_norm": 7.134433746337891,
+      "grad_norm": 11.365701675415039,
       "learning_rate": 5.8878878878878885e-06,
-      "loss": 1.0912,
+      "loss": 1.092,
       "step": 41080
     },
     {
       "epoch": 123.39,
-      "grad_norm": 7.458648681640625,
+      "grad_norm": 7.3408660888671875,
       "learning_rate": 5.886886886886888e-06,
-      "loss": 1.1186,
+      "loss": 1.1119,
       "step": 41090
     },
     {
       "epoch": 123.42,
-      "grad_norm": 9.221880912780762,
+      "grad_norm": 10.04033088684082,
       "learning_rate": 5.885885885885886e-06,
-      "loss": 0.9934,
+      "loss": 0.9915,
       "step": 41100
     },
     {
       "epoch": 123.45,
-      "grad_norm": 6.556828022003174,
+      "grad_norm": 7.788590431213379,
       "learning_rate": 5.884884884884885e-06,
-      "loss": 1.0198,
+      "loss": 1.0089,
       "step": 41110
     },
     {
       "epoch": 123.48,
-      "grad_norm": 4.97642707824707,
+      "grad_norm": 5.145765781402588,
       "learning_rate": 5.883883883883884e-06,
-      "loss": 1.0507,
+      "loss": 1.0433,
       "step": 41120
     },
     {
       "epoch": 123.51,
-      "grad_norm": 7.373807430267334,
+      "grad_norm": 10.013214111328125,
       "learning_rate": 5.882882882882883e-06,
-      "loss": 1.0756,
+      "loss": 1.0648,
       "step": 41130
     },
     {
       "epoch": 123.54,
-      "grad_norm": 4.804464817047119,
+      "grad_norm": 6.996456146240234,
       "learning_rate": 5.881881881881883e-06,
-      "loss": 1.1847,
+      "loss": 1.1944,
       "step": 41140
     },
     {
       "epoch": 123.57,
-      "grad_norm": 7.218160629272461,
+      "grad_norm": 6.546069145202637,
       "learning_rate": 5.880880880880881e-06,
-      "loss": 1.0986,
+      "loss": 1.0938,
       "step": 41150
     },
     {
       "epoch": 123.6,
-      "grad_norm": 6.0997514724731445,
+      "grad_norm": 6.6015095710754395,
       "learning_rate": 5.879879879879881e-06,
-      "loss": 1.0209,
+      "loss": 1.0402,
       "step": 41160
     },
     {
       "epoch": 123.63,
-      "grad_norm": 6.415432453155518,
+      "grad_norm": 6.199520587921143,
       "learning_rate": 5.87887887887888e-06,
-      "loss": 1.0383,
+      "loss": 1.0447,
       "step": 41170
     },
     {
       "epoch": 123.66,
-      "grad_norm": 5.5446271896362305,
+      "grad_norm": 6.390815258026123,
       "learning_rate": 5.877877877877879e-06,
-      "loss": 1.0348,
+      "loss": 1.0263,
       "step": 41180
     },
     {
       "epoch": 123.69,
-      "grad_norm": 5.008070945739746,
+      "grad_norm": 6.006152629852295,
       "learning_rate": 5.876876876876877e-06,
-      "loss": 1.0528,
+      "loss": 1.0556,
       "step": 41190
     },
     {
       "epoch": 123.72,
-      "grad_norm": 8.03830337524414,
+      "grad_norm": 6.2786455154418945,
       "learning_rate": 5.875875875875876e-06,
-      "loss": 1.1423,
+      "loss": 1.1426,
       "step": 41200
     },
     {
       "epoch": 123.75,
-      "grad_norm": 6.145468711853027,
+      "grad_norm": 6.074872016906738,
       "learning_rate": 5.874874874874875e-06,
-      "loss": 1.1538,
+      "loss": 1.1612,
       "step": 41210
     },
     {
       "epoch": 123.78,
-      "grad_norm": 5.045534133911133,
+      "grad_norm": 5.713789463043213,
       "learning_rate": 5.873873873873874e-06,
-      "loss": 1.0724,
+      "loss": 1.0714,
       "step": 41220
     },
     {
       "epoch": 123.81,
-      "grad_norm": 6.107951641082764,
+      "grad_norm": 6.6898932456970215,
       "learning_rate": 5.872872872872873e-06,
-      "loss": 1.0472,
+      "loss": 1.0359,
       "step": 41230
     },
     {
       "epoch": 123.84,
-      "grad_norm": 7.2071146965026855,
+      "grad_norm": 6.4471893310546875,
       "learning_rate": 5.871871871871873e-06,
-      "loss": 1.1184,
+      "loss": 1.1084,
       "step": 41240
     },
     {
       "epoch": 123.87,
-      "grad_norm": 5.681679725646973,
+      "grad_norm": 5.89757776260376,
       "learning_rate": 5.870870870870871e-06,
-      "loss": 1.0034,
+      "loss": 1.006,
       "step": 41250
     },
     {
       "epoch": 123.9,
-      "grad_norm": 6.2092671394348145,
+      "grad_norm": 6.460568428039551,
       "learning_rate": 5.869869869869871e-06,
-      "loss": 1.0679,
+      "loss": 1.0724,
       "step": 41260
     },
     {
       "epoch": 123.93,
-      "grad_norm": 5.777194023132324,
+      "grad_norm": 6.6421217918396,
       "learning_rate": 5.868868868868869e-06,
-      "loss": 1.1101,
+      "loss": 1.1136,
       "step": 41270
     },
     {
       "epoch": 123.96,
-      "grad_norm": 5.8160834312438965,
+      "grad_norm": 6.660964012145996,
       "learning_rate": 5.867867867867868e-06,
-      "loss": 1.0707,
+      "loss": 1.0679,
       "step": 41280
     },
     {
       "epoch": 123.99,
-      "grad_norm": 8.497125625610352,
+      "grad_norm": 6.548609256744385,
       "learning_rate": 5.8668668668668675e-06,
-      "loss": 1.1421,
+      "loss": 1.1361,
       "step": 41290
     },
     {
       "epoch": 124.0,
-      "eval_accuracy": 0.807,
-      "eval_loss": 0.6187058687210083,
-      "eval_runtime": 5.4622,
-      "eval_samples_per_second": 1830.759,
-      "eval_steps_per_second": 7.323,
+      "eval_accuracy": 0.8115,
+      "eval_loss": 0.6177389025688171,
+      "eval_runtime": 5.4742,
+      "eval_samples_per_second": 1826.766,
+      "eval_steps_per_second": 7.307,
       "step": 41292
     },
     {
       "epoch": 124.02,
-      "grad_norm": 8.052566528320312,
+      "grad_norm": 11.902168273925781,
       "learning_rate": 5.865865865865866e-06,
-      "loss": 1.2209,
+      "loss": 1.2047,
       "step": 41300
     },
     {
       "epoch": 124.05,
-      "grad_norm": 6.834409236907959,
+      "grad_norm": 6.930357933044434,
       "learning_rate": 5.8648648648648655e-06,
-      "loss": 1.103,
+      "loss": 1.1085,
       "step": 41310
     },
     {
       "epoch": 124.08,
-      "grad_norm": 5.471545696258545,
+      "grad_norm": 8.475255966186523,
       "learning_rate": 5.863863863863864e-06,
-      "loss": 1.1158,
+      "loss": 1.1078,
       "step": 41320
     },
     {
       "epoch": 124.11,
-      "grad_norm": 6.9391279220581055,
+      "grad_norm": 6.0257368087768555,
       "learning_rate": 5.8628628628628635e-06,
-      "loss": 1.0618,
+      "loss": 1.0729,
       "step": 41330
     },
     {
       "epoch": 124.14,
-      "grad_norm": 6.384866714477539,
+      "grad_norm": 6.1028032302856445,
       "learning_rate": 5.861861861861863e-06,
-      "loss": 1.0795,
+      "loss": 1.0719,
       "step": 41340
     },
     {
       "epoch": 124.17,
-      "grad_norm": 5.5030107498168945,
+      "grad_norm": 6.998908519744873,
       "learning_rate": 5.860860860860861e-06,
-      "loss": 1.0896,
+      "loss": 1.0857,
       "step": 41350
     },
     {
       "epoch": 124.2,
-      "grad_norm": 6.612639427185059,
+      "grad_norm": 8.876693725585938,
       "learning_rate": 5.85985985985986e-06,
-      "loss": 1.0368,
+      "loss": 1.0327,
       "step": 41360
     },
     {
       "epoch": 124.23,
-      "grad_norm": 5.419907569885254,
+      "grad_norm": 6.764688968658447,
       "learning_rate": 5.858858858858859e-06,
-      "loss": 1.1147,
+      "loss": 1.1078,
       "step": 41370
     },
     {
       "epoch": 124.26,
-      "grad_norm": 6.5069074630737305,
+      "grad_norm": 7.622865676879883,
       "learning_rate": 5.857857857857858e-06,
-      "loss": 1.0462,
+      "loss": 1.0439,
       "step": 41380
     },
     {
       "epoch": 124.29,
-      "grad_norm": 6.895297050476074,
+      "grad_norm": 6.256166934967041,
       "learning_rate": 5.856856856856858e-06,
-      "loss": 1.0214,
+      "loss": 1.0247,
       "step": 41390
     },
     {
       "epoch": 124.32,
-      "grad_norm": 4.6906304359436035,
+      "grad_norm": 5.390114784240723,
       "learning_rate": 5.855855855855856e-06,
-      "loss": 0.981,
+      "loss": 0.9786,
       "step": 41400
     },
     {
       "epoch": 124.35,
-      "grad_norm": 7.992466449737549,
+      "grad_norm": 7.9380927085876465,
       "learning_rate": 5.854854854854856e-06,
-      "loss": 1.0319,
+      "loss": 1.0261,
       "step": 41410
     },
     {
       "epoch": 124.38,
-      "grad_norm": 5.211793422698975,
+      "grad_norm": 5.986649513244629,
       "learning_rate": 5.853853853853855e-06,
-      "loss": 1.0487,
+      "loss": 1.052,
       "step": 41420
     },
     {
       "epoch": 124.41,
-      "grad_norm": 7.182945251464844,
+      "grad_norm": 9.327011108398438,
       "learning_rate": 5.852852852852854e-06,
-      "loss": 1.1195,
+      "loss": 1.12,
       "step": 41430
     },
     {
       "epoch": 124.44,
-      "grad_norm": 9.976508140563965,
+      "grad_norm": 9.332976341247559,
       "learning_rate": 5.8518518518518515e-06,
-      "loss": 1.103,
+      "loss": 1.101,
       "step": 41440
     },
     {
       "epoch": 124.47,
-      "grad_norm": 5.737894058227539,
+      "grad_norm": 5.53643274307251,
       "learning_rate": 5.850850850850851e-06,
-      "loss": 1.1051,
+      "loss": 1.1103,
       "step": 41450
     },
     {
       "epoch": 124.5,
-      "grad_norm": 6.834338665008545,
+      "grad_norm": 7.797732830047607,
       "learning_rate": 5.84984984984985e-06,
-      "loss": 1.0283,
+      "loss": 1.023,
       "step": 41460
     },
     {
       "epoch": 124.53,
-      "grad_norm": 8.924941062927246,
+      "grad_norm": 7.089838981628418,
       "learning_rate": 5.848848848848849e-06,
-      "loss": 1.0462,
+      "loss": 1.045,
       "step": 41470
     },
     {
       "epoch": 124.56,
-      "grad_norm": 6.465217590332031,
+      "grad_norm": 8.150374412536621,
       "learning_rate": 5.847847847847848e-06,
-      "loss": 1.0524,
+      "loss": 1.056,
       "step": 41480
     },
     {
       "epoch": 124.59,
-      "grad_norm": 6.964038372039795,
+      "grad_norm": 8.309186935424805,
       "learning_rate": 5.846846846846848e-06,
-      "loss": 1.0197,
+      "loss": 1.0179,
       "step": 41490
     },
     {
       "epoch": 124.62,
-      "grad_norm": 8.284086227416992,
+      "grad_norm": 6.576289176940918,
       "learning_rate": 5.8458458458458464e-06,
-      "loss": 1.099,
+      "loss": 1.0857,
       "step": 41500
     },
     {
       "epoch": 124.65,
-      "grad_norm": 5.69329833984375,
+      "grad_norm": 8.586288452148438,
       "learning_rate": 5.844844844844846e-06,
-      "loss": 1.082,
+      "loss": 1.0864,
       "step": 41510
     },
     {
       "epoch": 124.68,
-      "grad_norm": 5.958210468292236,
+      "grad_norm": 7.0304388999938965,
       "learning_rate": 5.843843843843844e-06,
-      "loss": 1.0525,
+      "loss": 1.0512,
       "step": 41520
     },
     {
       "epoch": 124.71,
-      "grad_norm": 5.838710784912109,
+      "grad_norm": 7.424017906188965,
       "learning_rate": 5.842842842842843e-06,
-      "loss": 1.0542,
+      "loss": 1.0404,
       "step": 41530
     },
     {
       "epoch": 124.74,
-      "grad_norm": 7.405842304229736,
+      "grad_norm": 6.272463798522949,
       "learning_rate": 5.8418418418418425e-06,
-      "loss": 1.0352,
+      "loss": 1.0256,
       "step": 41540
     },
     {
       "epoch": 124.77,
-      "grad_norm": 6.177097797393799,
+      "grad_norm": 4.929904937744141,
       "learning_rate": 5.840840840840841e-06,
-      "loss": 0.976,
+      "loss": 0.9867,
       "step": 41550
     },
     {
       "epoch": 124.8,
-      "grad_norm": 7.7285027503967285,
+      "grad_norm": 7.031424045562744,
       "learning_rate": 5.8398398398398405e-06,
-      "loss": 1.0533,
+      "loss": 1.0579,
       "step": 41560
     },
     {
       "epoch": 124.83,
-      "grad_norm": 5.428056240081787,
+      "grad_norm": 6.666329860687256,
       "learning_rate": 5.838838838838839e-06,
-      "loss": 1.0345,
+      "loss": 1.0425,
       "step": 41570
     },
     {
       "epoch": 124.86,
-      "grad_norm": 6.707502365112305,
+      "grad_norm": 6.5152788162231445,
       "learning_rate": 5.837837837837839e-06,
-      "loss": 1.1015,
+      "loss": 1.0922,
       "step": 41580
     },
     {
       "epoch": 124.89,
-      "grad_norm": 7.354709148406982,
+      "grad_norm": 10.001039505004883,
       "learning_rate": 5.836836836836838e-06,
-      "loss": 1.0914,
+      "loss": 1.0843,
       "step": 41590
     },
     {
       "epoch": 124.92,
-      "grad_norm": 7.513284683227539,
+      "grad_norm": 10.73922348022461,
       "learning_rate": 5.835835835835836e-06,
-      "loss": 1.0757,
+      "loss": 1.0789,
       "step": 41600
     },
     {
       "epoch": 124.95,
-      "grad_norm": 5.339949607849121,
+      "grad_norm": 7.9347944259643555,
       "learning_rate": 5.834834834834835e-06,
-      "loss": 1.0854,
+      "loss": 1.0864,
       "step": 41610
     },
     {
       "epoch": 124.98,
-      "grad_norm": 9.294020652770996,
+      "grad_norm": 9.064098358154297,
       "learning_rate": 5.833833833833834e-06,
-      "loss": 0.9842,
+      "loss": 1.0064,
       "step": 41620
     },
     {
       "epoch": 125.0,
-      "eval_accuracy": 0.8078,
-      "eval_loss": 0.6293672919273376,
-      "eval_runtime": 5.6479,
-      "eval_samples_per_second": 1770.556,
-      "eval_steps_per_second": 7.082,
+      "eval_accuracy": 0.8081,
+      "eval_loss": 0.629188597202301,
+      "eval_runtime": 5.4519,
+      "eval_samples_per_second": 1834.209,
+      "eval_steps_per_second": 7.337,
       "step": 41625
     },
     {
       "epoch": 125.02,
-      "grad_norm": 5.859194755554199,
+      "grad_norm": 7.538881778717041,
       "learning_rate": 5.832832832832833e-06,
-      "loss": 1.0957,
+      "loss": 1.1107,
       "step": 41630
     },
     {
       "epoch": 125.05,
-      "grad_norm": 5.751328468322754,
+      "grad_norm": 7.439499378204346,
       "learning_rate": 5.831831831831833e-06,
-      "loss": 1.0502,
+      "loss": 1.0475,
       "step": 41640
     },
     {
       "epoch": 125.08,
-      "grad_norm": 5.820566177368164,
+      "grad_norm": 7.4711079597473145,
       "learning_rate": 5.830830830830831e-06,
-      "loss": 1.0793,
+      "loss": 1.0788,
       "step": 41650
     },
     {
       "epoch": 125.11,
-      "grad_norm": 5.5453643798828125,
+      "grad_norm": 4.995654582977295,
       "learning_rate": 5.829829829829831e-06,
-      "loss": 1.0166,
+      "loss": 1.0016,
       "step": 41660
     },
     {
       "epoch": 125.14,
-      "grad_norm": 6.48692512512207,
+      "grad_norm": 6.768871307373047,
       "learning_rate": 5.828828828828829e-06,
-      "loss": 1.0601,
+      "loss": 1.0591,
       "step": 41670
     },
     {
       "epoch": 125.17,
-      "grad_norm": 8.325119972229004,
+      "grad_norm": 8.471821784973145,
       "learning_rate": 5.827827827827829e-06,
-      "loss": 1.0593,
+      "loss": 1.062,
       "step": 41680
     },
     {
       "epoch": 125.2,
-      "grad_norm": 5.747281074523926,
+      "grad_norm": 11.14253044128418,
       "learning_rate": 5.8268268268268265e-06,
-      "loss": 1.0524,
+      "loss": 1.0561,
       "step": 41690
     },
     {
       "epoch": 125.23,
-      "grad_norm": 10.23435115814209,
+      "grad_norm": 10.530852317810059,
       "learning_rate": 5.825825825825826e-06,
-      "loss": 1.141,
+      "loss": 1.1426,
       "step": 41700
     },
     {
       "epoch": 125.26,
-      "grad_norm": 8.103321075439453,
+      "grad_norm": 7.126166343688965,
       "learning_rate": 5.824824824824825e-06,
-      "loss": 1.0643,
+      "loss": 1.0562,
       "step": 41710
     },
     {
       "epoch": 125.29,
-      "grad_norm": 6.441063404083252,
+      "grad_norm": 7.816658973693848,
       "learning_rate": 5.823823823823824e-06,
-      "loss": 1.0862,
+      "loss": 1.0873,
       "step": 41720
     },
     {
       "epoch": 125.32,
-      "grad_norm": 6.83735990524292,
+      "grad_norm": 6.855495452880859,
       "learning_rate": 5.8228228228228234e-06,
-      "loss": 1.0617,
+      "loss": 1.048,
       "step": 41730
     },
     {
       "epoch": 125.35,
-      "grad_norm": 5.5967302322387695,
+      "grad_norm": 5.422829627990723,
       "learning_rate": 5.821821821821823e-06,
-      "loss": 1.0507,
+      "loss": 1.0576,
       "step": 41740
     },
     {
       "epoch": 125.38,
-      "grad_norm": 6.5646843910217285,
+      "grad_norm": 6.980844974517822,
       "learning_rate": 5.8208208208208215e-06,
-      "loss": 1.0025,
+      "loss": 1.005,
       "step": 41750
     },
     {
       "epoch": 125.41,
-      "grad_norm": 7.774322032928467,
+      "grad_norm": 7.752714157104492,
       "learning_rate": 5.819819819819821e-06,
-      "loss": 1.0144,
+      "loss": 1.006,
       "step": 41760
     },
     {
       "epoch": 125.44,
-      "grad_norm": 5.384167671203613,
+      "grad_norm": 6.650729656219482,
       "learning_rate": 5.818818818818819e-06,
-      "loss": 1.0729,
+      "loss": 1.0787,
       "step": 41770
     },
     {
       "epoch": 125.47,
-      "grad_norm": 5.986880779266357,
+      "grad_norm": 6.649991512298584,
       "learning_rate": 5.817817817817818e-06,
-      "loss": 1.0991,
+      "loss": 1.1085,
       "step": 41780
     },
     {
       "epoch": 125.5,
-      "grad_norm": 6.16122579574585,
+      "grad_norm": 9.008349418640137,
       "learning_rate": 5.816816816816817e-06,
-      "loss": 1.0459,
+      "loss": 1.0411,
       "step": 41790
     },
     {
       "epoch": 125.53,
-      "grad_norm": 7.329347610473633,
+      "grad_norm": 7.031470775604248,
       "learning_rate": 5.815815815815816e-06,
-      "loss": 1.0853,
+      "loss": 1.0876,
       "step": 41800
     },
     {
       "epoch": 125.56,
-      "grad_norm": 6.798847675323486,
+      "grad_norm": 7.017879962921143,
       "learning_rate": 5.814814814814816e-06,
-      "loss": 1.1478,
+      "loss": 1.1579,
       "step": 41810
     },
     {
       "epoch": 125.59,
-      "grad_norm": 7.537840366363525,
+      "grad_norm": 6.114543914794922,
       "learning_rate": 5.813813813813814e-06,
-      "loss": 0.9888,
+      "loss": 0.99,
       "step": 41820
     },
     {
       "epoch": 125.62,
-      "grad_norm": 8.395492553710938,
+      "grad_norm": 10.347532272338867,
       "learning_rate": 5.812812812812814e-06,
-      "loss": 1.0935,
+      "loss": 1.0875,
       "step": 41830
     },
     {
       "epoch": 125.65,
-      "grad_norm": 6.658613204956055,
+      "grad_norm": 6.544745922088623,
       "learning_rate": 5.811811811811813e-06,
-      "loss": 0.989,
+      "loss": 0.9933,
       "step": 41840
     },
     {
       "epoch": 125.68,
-      "grad_norm": 6.0951948165893555,
+      "grad_norm": 6.290530204772949,
       "learning_rate": 5.810810810810811e-06,
-      "loss": 1.0236,
+      "loss": 1.0207,
       "step": 41850
     },
     {
       "epoch": 125.71,
-      "grad_norm": 5.511558532714844,
+      "grad_norm": 4.669356346130371,
       "learning_rate": 5.80980980980981e-06,
-      "loss": 1.0156,
+      "loss": 1.0173,
       "step": 41860
     },
     {
       "epoch": 125.74,
-      "grad_norm": 8.030309677124023,
+      "grad_norm": 7.768892765045166,
       "learning_rate": 5.808808808808809e-06,
-      "loss": 1.1098,
+      "loss": 1.1122,
       "step": 41870
     },
     {
       "epoch": 125.77,
-      "grad_norm": 6.302718162536621,
+      "grad_norm": 6.510124206542969,
       "learning_rate": 5.807807807807808e-06,
-      "loss": 1.0427,
+      "loss": 1.0392,
       "step": 41880
     },
     {
       "epoch": 125.8,
-      "grad_norm": 5.969773292541504,
+      "grad_norm": 8.944371223449707,
       "learning_rate": 5.806806806806808e-06,
-      "loss": 1.0805,
+      "loss": 1.0802,
       "step": 41890
     },
     {
       "epoch": 125.83,
-      "grad_norm": 7.206167697906494,
+      "grad_norm": 6.971608638763428,
       "learning_rate": 5.805805805805806e-06,
-      "loss": 1.0561,
+      "loss": 1.0628,
       "step": 41900
     },
     {
       "epoch": 125.86,
-      "grad_norm": 6.8547587394714355,
+      "grad_norm": 8.726605415344238,
       "learning_rate": 5.804804804804806e-06,
-      "loss": 1.0135,
+      "loss": 1.015,
       "step": 41910
     },
     {
       "epoch": 125.89,
-      "grad_norm": 7.547189712524414,
+      "grad_norm": 6.342212677001953,
       "learning_rate": 5.803803803803804e-06,
-      "loss": 1.0605,
+      "loss": 1.0552,
       "step": 41920
     },
     {
       "epoch": 125.92,
-      "grad_norm": 8.997673988342285,
+      "grad_norm": 10.56101131439209,
       "learning_rate": 5.802802802802803e-06,
-      "loss": 1.0925,
+      "loss": 1.0867,
       "step": 41930
     },
     {
       "epoch": 125.95,
-      "grad_norm": 4.3642706871032715,
+      "grad_norm": 8.053936004638672,
       "learning_rate": 5.8018018018018015e-06,
-      "loss": 1.0681,
+      "loss": 1.0642,
       "step": 41940
     },
     {
       "epoch": 125.98,
-      "grad_norm": 7.6589837074279785,
+      "grad_norm": 6.002086639404297,
       "learning_rate": 5.800800800800801e-06,
-      "loss": 1.06,
+      "loss": 1.0619,
       "step": 41950
     },
     {
       "epoch": 126.0,
-      "eval_accuracy": 0.8094,
-      "eval_loss": 0.6174260973930359,
-      "eval_runtime": 5.3141,
-      "eval_samples_per_second": 1881.8,
-      "eval_steps_per_second": 7.527,
+      "eval_accuracy": 0.813,
+      "eval_loss": 0.6147308945655823,
+      "eval_runtime": 5.3235,
+      "eval_samples_per_second": 1878.446,
+      "eval_steps_per_second": 7.514,
       "step": 41958
     },
     {
       "epoch": 126.01,
-      "grad_norm": 7.5229339599609375,
+      "grad_norm": 7.608147144317627,
       "learning_rate": 5.7997997997998004e-06,
-      "loss": 1.1709,
+      "loss": 1.1699,
       "step": 41960
     },
     {
       "epoch": 126.04,
-      "grad_norm": 6.712756633758545,
+      "grad_norm": 7.489234924316406,
       "learning_rate": 5.798798798798799e-06,
-      "loss": 1.0511,
+      "loss": 1.0553,
       "step": 41970
     },
     {
       "epoch": 126.07,
-      "grad_norm": 4.666429042816162,
+      "grad_norm": 5.5367431640625,
       "learning_rate": 5.7977977977977985e-06,
-      "loss": 1.052,
+      "loss": 1.0472,
       "step": 41980
     },
     {
       "epoch": 126.1,
-      "grad_norm": 6.218411445617676,
+      "grad_norm": 6.798861980438232,
       "learning_rate": 5.796796796796798e-06,
-      "loss": 1.0379,
+      "loss": 1.0281,
       "step": 41990
     },
     {
       "epoch": 126.13,
-      "grad_norm": 6.931608200073242,
+      "grad_norm": 11.116792678833008,
       "learning_rate": 5.7957957957957965e-06,
-      "loss": 0.9826,
+      "loss": 0.9869,
       "step": 42000
     },
     {
       "epoch": 126.16,
-      "grad_norm": 9.20053482055664,
+      "grad_norm": 6.728570938110352,
       "learning_rate": 5.794794794794796e-06,
-      "loss": 1.0586,
+      "loss": 1.0497,
       "step": 42010
     },
     {
       "epoch": 126.19,
-      "grad_norm": 5.553876876831055,
+      "grad_norm": 6.7125043869018555,
       "learning_rate": 5.793793793793794e-06,
-      "loss": 0.9859,
+      "loss": 0.9825,
       "step": 42020
     },
     {
       "epoch": 126.22,
-      "grad_norm": 6.418639659881592,
+      "grad_norm": 5.98939847946167,
       "learning_rate": 5.792792792792793e-06,
-      "loss": 1.0461,
+      "loss": 1.0446,
       "step": 42030
     },
     {
       "epoch": 126.25,
-      "grad_norm": 8.067168235778809,
+      "grad_norm": 7.990087032318115,
       "learning_rate": 5.791791791791792e-06,
-      "loss": 1.103,
+      "loss": 1.1021,
       "step": 42040
     },
     {
       "epoch": 126.28,
-      "grad_norm": 5.962207794189453,
+      "grad_norm": 7.422767162322998,
       "learning_rate": 5.790790790790791e-06,
-      "loss": 1.0464,
+      "loss": 1.0372,
       "step": 42050
     },
     {
       "epoch": 126.31,
-      "grad_norm": 5.658208847045898,
+      "grad_norm": 10.487519264221191,
       "learning_rate": 5.789789789789791e-06,
-      "loss": 1.1425,
+      "loss": 1.151,
       "step": 42060
     },
     {
       "epoch": 126.34,
-      "grad_norm": 5.422247409820557,
+      "grad_norm": 6.4130072593688965,
       "learning_rate": 5.788788788788789e-06,
-      "loss": 1.0916,
+      "loss": 1.0921,
       "step": 42070
     },
     {
       "epoch": 126.37,
-      "grad_norm": 7.357682228088379,
+      "grad_norm": 8.828360557556152,
       "learning_rate": 5.787787787787789e-06,
-      "loss": 1.0325,
+      "loss": 1.0467,
       "step": 42080
     },
     {
       "epoch": 126.4,
-      "grad_norm": 7.297579765319824,
+      "grad_norm": 8.32227611541748,
       "learning_rate": 5.786786786786788e-06,
-      "loss": 1.1044,
+      "loss": 1.0858,
       "step": 42090
     },
     {
       "epoch": 126.43,
-      "grad_norm": 11.601985931396484,
+      "grad_norm": 10.063148498535156,
       "learning_rate": 5.785785785785786e-06,
-      "loss": 1.084,
+      "loss": 1.0897,
       "step": 42100
     },
     {
       "epoch": 126.46,
-      "grad_norm": 11.901975631713867,
+      "grad_norm": 9.878661155700684,
       "learning_rate": 5.784784784784785e-06,
-      "loss": 1.0736,
+      "loss": 1.0774,
       "step": 42110
     },
     {
       "epoch": 126.49,
-      "grad_norm": 7.362436294555664,
+      "grad_norm": 9.017292976379395,
       "learning_rate": 5.783783783783784e-06,
-      "loss": 1.1296,
+      "loss": 1.1344,
       "step": 42120
     },
     {
       "epoch": 126.52,
-      "grad_norm": 8.63421630859375,
+      "grad_norm": 8.736855506896973,
       "learning_rate": 5.782782782782783e-06,
-      "loss": 1.1218,
+      "loss": 1.1126,
       "step": 42130
     },
     {
       "epoch": 126.55,
-      "grad_norm": 6.050577163696289,
+      "grad_norm": 4.8707475662231445,
       "learning_rate": 5.781781781781782e-06,
-      "loss": 0.9945,
+      "loss": 0.9948,
       "step": 42140
     },
     {
       "epoch": 126.58,
-      "grad_norm": 8.316149711608887,
+      "grad_norm": 6.940192222595215,
       "learning_rate": 5.780780780780781e-06,
-      "loss": 1.072,
+      "loss": 1.059,
       "step": 42150
     },
     {
       "epoch": 126.61,
-      "grad_norm": 7.117032527923584,
+      "grad_norm": 7.267270088195801,
       "learning_rate": 5.779779779779781e-06,
-      "loss": 1.0625,
+      "loss": 1.052,
       "step": 42160
     },
     {
       "epoch": 126.64,
-      "grad_norm": 6.239290237426758,
+      "grad_norm": 11.108975410461426,
       "learning_rate": 5.778778778778779e-06,
-      "loss": 1.1204,
+      "loss": 1.1114,
       "step": 42170
     },
     {
       "epoch": 126.67,
-      "grad_norm": 7.495082378387451,
+      "grad_norm": 8.702690124511719,
       "learning_rate": 5.777777777777778e-06,
-      "loss": 1.0742,
+      "loss": 1.0841,
       "step": 42180
     },
     {
       "epoch": 126.7,
-      "grad_norm": 5.6787190437316895,
+      "grad_norm": 6.314316749572754,
       "learning_rate": 5.776776776776777e-06,
-      "loss": 1.0798,
+      "loss": 1.0731,
       "step": 42190
     },
     {
       "epoch": 126.73,
-      "grad_norm": 7.236333847045898,
+      "grad_norm": 8.07426929473877,
       "learning_rate": 5.775775775775776e-06,
-      "loss": 1.0879,
+      "loss": 1.0759,
       "step": 42200
     },
     {
       "epoch": 126.76,
-      "grad_norm": 6.262499809265137,
+      "grad_norm": 9.081854820251465,
       "learning_rate": 5.7747747747747755e-06,
-      "loss": 1.032,
+      "loss": 1.0317,
       "step": 42210
     },
     {
       "epoch": 126.79,
-      "grad_norm": 6.51117467880249,
+      "grad_norm": 6.923382759094238,
       "learning_rate": 5.773773773773774e-06,
-      "loss": 1.0804,
+      "loss": 1.0853,
       "step": 42220
     },
     {
       "epoch": 126.82,
-      "grad_norm": 6.922031402587891,
+      "grad_norm": 8.821578025817871,
       "learning_rate": 5.7727727727727735e-06,
-      "loss": 1.0276,
+      "loss": 1.0164,
       "step": 42230
     },
     {
       "epoch": 126.85,
-      "grad_norm": 6.9309306144714355,
+      "grad_norm": 7.56242561340332,
       "learning_rate": 5.771771771771772e-06,
-      "loss": 1.0396,
+      "loss": 1.0372,
       "step": 42240
     },
     {
       "epoch": 126.88,
-      "grad_norm": 9.395503044128418,
+      "grad_norm": 12.587566375732422,
       "learning_rate": 5.7707707707707715e-06,
-      "loss": 1.039,
+      "loss": 1.036,
       "step": 42250
     },
     {
       "epoch": 126.91,
-      "grad_norm": 8.188984870910645,
+      "grad_norm": 10.565802574157715,
       "learning_rate": 5.769769769769771e-06,
-      "loss": 1.0543,
+      "loss": 1.0523,
       "step": 42260
     },
     {
       "epoch": 126.94,
-      "grad_norm": 7.745103359222412,
+      "grad_norm": 6.690069198608398,
       "learning_rate": 5.768768768768769e-06,
-      "loss": 1.0806,
+      "loss": 1.0878,
       "step": 42270
     },
     {
       "epoch": 126.97,
-      "grad_norm": 6.0138258934021,
+      "grad_norm": 9.545936584472656,
       "learning_rate": 5.767767767767768e-06,
-      "loss": 1.01,
+      "loss": 1.0052,
       "step": 42280
     },
     {
       "epoch": 127.0,
-      "grad_norm": 7.018523216247559,
+      "grad_norm": 9.225862503051758,
       "learning_rate": 5.766766766766767e-06,
-      "loss": 1.1292,
+      "loss": 1.1252,
       "step": 42290
     },
     {
       "epoch": 127.0,
-      "eval_accuracy": 0.8084,
-      "eval_loss": 0.6206111311912537,
-      "eval_runtime": 5.695,
-      "eval_samples_per_second": 1755.929,
-      "eval_steps_per_second": 7.024,
+      "eval_accuracy": 0.8106,
+      "eval_loss": 0.619255781173706,
+      "eval_runtime": 5.4315,
+      "eval_samples_per_second": 1841.098,
+      "eval_steps_per_second": 7.364,
       "step": 42291
     },
     {
       "epoch": 127.03,
-      "grad_norm": 6.9051313400268555,
+      "grad_norm": 7.37378454208374,
       "learning_rate": 5.765765765765766e-06,
-      "loss": 1.1746,
+      "loss": 1.1481,
       "step": 42300
     },
     {
       "epoch": 127.06,
-      "grad_norm": 5.589310169219971,
+      "grad_norm": 6.679778575897217,
       "learning_rate": 5.764764764764766e-06,
-      "loss": 1.0071,
+      "loss": 0.9929,
       "step": 42310
     },
     {
       "epoch": 127.09,
-      "grad_norm": 6.393409252166748,
+      "grad_norm": 5.610591411590576,
       "learning_rate": 5.763763763763764e-06,
-      "loss": 1.0288,
+      "loss": 1.0244,
       "step": 42320
     },
     {
       "epoch": 127.12,
-      "grad_norm": 6.2181596755981445,
+      "grad_norm": 5.308989524841309,
       "learning_rate": 5.762762762762764e-06,
-      "loss": 1.1181,
+      "loss": 1.1152,
       "step": 42330
     },
     {
       "epoch": 127.15,
-      "grad_norm": 7.086292743682861,
+      "grad_norm": 7.5548505783081055,
       "learning_rate": 5.761761761761763e-06,
-      "loss": 1.1186,
+      "loss": 1.1199,
       "step": 42340
     },
     {
       "epoch": 127.18,
-      "grad_norm": 8.902729034423828,
+      "grad_norm": 5.595024108886719,
       "learning_rate": 5.760760760760761e-06,
-      "loss": 1.0015,
+      "loss": 0.9943,
       "step": 42350
     },
     {
       "epoch": 127.21,
-      "grad_norm": 7.664077281951904,
+      "grad_norm": 7.086429595947266,
       "learning_rate": 5.7597597597597595e-06,
-      "loss": 0.993,
+      "loss": 0.9873,
       "step": 42360
     },
     {
       "epoch": 127.24,
-      "grad_norm": 7.015576362609863,
+      "grad_norm": 5.840652942657471,
       "learning_rate": 5.758758758758759e-06,
-      "loss": 1.0837,
+      "loss": 1.0891,
       "step": 42370
     },
     {
       "epoch": 127.27,
-      "grad_norm": 6.188762187957764,
+      "grad_norm": 7.058176040649414,
       "learning_rate": 5.757757757757758e-06,
-      "loss": 1.0552,
+      "loss": 1.0598,
       "step": 42380
     },
     {
       "epoch": 127.3,
-      "grad_norm": 6.842144012451172,
+      "grad_norm": 8.84478759765625,
       "learning_rate": 5.756756756756757e-06,
-      "loss": 1.1169,
+      "loss": 1.1151,
       "step": 42390
     },
     {
       "epoch": 127.33,
-      "grad_norm": 7.338894367218018,
+      "grad_norm": 8.384414672851562,
       "learning_rate": 5.755755755755756e-06,
-      "loss": 1.0796,
+      "loss": 1.0835,
       "step": 42400
     },
     {
       "epoch": 127.36,
-      "grad_norm": 6.7559428215026855,
+      "grad_norm": 6.553741931915283,
       "learning_rate": 5.754754754754756e-06,
-      "loss": 1.0351,
+      "loss": 1.0232,
       "step": 42410
     },
     {
       "epoch": 127.39,
-      "grad_norm": 8.485050201416016,
+      "grad_norm": 7.641211986541748,
       "learning_rate": 5.7537537537537544e-06,
-      "loss": 1.0564,
+      "loss": 1.0511,
       "step": 42420
     },
     {
       "epoch": 127.42,
-      "grad_norm": 8.118146896362305,
+      "grad_norm": 10.408940315246582,
       "learning_rate": 5.752752752752753e-06,
-      "loss": 1.0809,
+      "loss": 1.0648,
       "step": 42430
     },
     {
       "epoch": 127.45,
-      "grad_norm": 6.247161865234375,
+      "grad_norm": 8.54533863067627,
       "learning_rate": 5.751751751751752e-06,
-      "loss": 1.0089,
+      "loss": 1.0032,
       "step": 42440
     },
     {
       "epoch": 127.48,
-      "grad_norm": 6.281066417694092,
+      "grad_norm": 6.294346332550049,
       "learning_rate": 5.750750750750751e-06,
-      "loss": 1.0479,
+      "loss": 1.0542,
       "step": 42450
     },
     {
       "epoch": 127.51,
-      "grad_norm": 7.427447319030762,
+      "grad_norm": 6.853056907653809,
       "learning_rate": 5.7497497497497505e-06,
-      "loss": 1.0436,
+      "loss": 1.0414,
       "step": 42460
     },
     {
       "epoch": 127.54,
-      "grad_norm": 7.854270935058594,
+      "grad_norm": 9.006463050842285,
       "learning_rate": 5.748748748748749e-06,
-      "loss": 1.0682,
+      "loss": 1.0742,
       "step": 42470
     },
     {
       "epoch": 127.57,
-      "grad_norm": 6.843516826629639,
+      "grad_norm": 5.883875846862793,
       "learning_rate": 5.7477477477477485e-06,
-      "loss": 1.0484,
+      "loss": 1.037,
       "step": 42480
     },
     {
       "epoch": 127.6,
-      "grad_norm": 8.650701522827148,
+      "grad_norm": 7.669888496398926,
       "learning_rate": 5.746746746746747e-06,
-      "loss": 1.0373,
+      "loss": 1.0452,
       "step": 42490
     },
     {
       "epoch": 127.63,
-      "grad_norm": 7.219196319580078,
+      "grad_norm": 6.298346042633057,
       "learning_rate": 5.7457457457457466e-06,
-      "loss": 1.0733,
+      "loss": 1.0792,
       "step": 42500
     },
     {
       "epoch": 127.66,
-      "grad_norm": 9.328327178955078,
+      "grad_norm": 8.346816062927246,
       "learning_rate": 5.744744744744746e-06,
-      "loss": 1.1047,
+      "loss": 1.1041,
       "step": 42510
     },
     {
       "epoch": 127.69,
-      "grad_norm": 6.193305015563965,
+      "grad_norm": 6.068596839904785,
       "learning_rate": 5.743743743743744e-06,
-      "loss": 1.0456,
+      "loss": 1.0454,
       "step": 42520
     },
     {
       "epoch": 127.72,
-      "grad_norm": 7.209783554077148,
+      "grad_norm": 8.481571197509766,
       "learning_rate": 5.742742742742743e-06,
-      "loss": 1.1025,
+      "loss": 1.1039,
       "step": 42530
     },
     {
       "epoch": 127.75,
-      "grad_norm": 7.719836711883545,
+      "grad_norm": 8.264073371887207,
       "learning_rate": 5.741741741741742e-06,
-      "loss": 1.0173,
+      "loss": 1.0127,
       "step": 42540
     },
     {
       "epoch": 127.78,
-      "grad_norm": 7.292041301727295,
+      "grad_norm": 9.011350631713867,
       "learning_rate": 5.740740740740741e-06,
-      "loss": 1.0895,
+      "loss": 1.0906,
       "step": 42550
     },
     {
       "epoch": 127.81,
-      "grad_norm": 5.080024719238281,
+      "grad_norm": 6.109772205352783,
       "learning_rate": 5.739739739739741e-06,
-      "loss": 1.0729,
+      "loss": 1.071,
       "step": 42560
     },
     {
       "epoch": 127.84,
-      "grad_norm": 7.099133014678955,
+      "grad_norm": 7.504342555999756,
       "learning_rate": 5.738738738738739e-06,
-      "loss": 1.0884,
+      "loss": 1.103,
       "step": 42570
     },
     {
       "epoch": 127.87,
-      "grad_norm": 6.13621187210083,
+      "grad_norm": 4.806710243225098,
       "learning_rate": 5.737737737737739e-06,
-      "loss": 1.0174,
+      "loss": 1.0089,
       "step": 42580
     },
     {
       "epoch": 127.9,
-      "grad_norm": 5.986080646514893,
+      "grad_norm": 7.289290904998779,
       "learning_rate": 5.736736736736737e-06,
-      "loss": 1.0125,
+      "loss": 1.0064,
       "step": 42590
     },
     {
       "epoch": 127.93,
-      "grad_norm": 6.402467250823975,
+      "grad_norm": 7.614872932434082,
       "learning_rate": 5.735735735735736e-06,
-      "loss": 1.0623,
+      "loss": 1.0531,
       "step": 42600
     },
     {
       "epoch": 127.96,
-      "grad_norm": 5.876680374145508,
+      "grad_norm": 7.900713920593262,
       "learning_rate": 5.7347347347347345e-06,
-      "loss": 1.0635,
+      "loss": 1.0565,
       "step": 42610
     },
     {
       "epoch": 127.99,
-      "grad_norm": 8.266478538513184,
+      "grad_norm": 10.576767921447754,
       "learning_rate": 5.733733733733734e-06,
-      "loss": 1.0878,
+      "loss": 1.0828,
       "step": 42620
     },
     {
       "epoch": 128.0,
-      "eval_accuracy": 0.8103,
-      "eval_loss": 0.6143720149993896,
-      "eval_runtime": 5.6007,
-      "eval_samples_per_second": 1785.498,
-      "eval_steps_per_second": 7.142,
+      "eval_accuracy": 0.8114,
+      "eval_loss": 0.6157733201980591,
+      "eval_runtime": 5.3796,
+      "eval_samples_per_second": 1858.866,
+      "eval_steps_per_second": 7.435,
       "step": 42624
     },
     {
       "epoch": 128.02,
-      "grad_norm": 5.120694637298584,
+      "grad_norm": 5.593405723571777,
       "learning_rate": 5.732732732732733e-06,
-      "loss": 1.1556,
+      "loss": 1.1553,
       "step": 42630
     },
     {
       "epoch": 128.05,
-      "grad_norm": 5.044877052307129,
+      "grad_norm": 5.22299861907959,
       "learning_rate": 5.731731731731732e-06,
-      "loss": 1.0239,
+      "loss": 1.0265,
       "step": 42640
     },
     {
       "epoch": 128.08,
-      "grad_norm": 6.738753795623779,
+      "grad_norm": 8.3890962600708,
       "learning_rate": 5.7307307307307314e-06,
-      "loss": 1.1036,
+      "loss": 1.0933,
       "step": 42650
     },
     {
       "epoch": 128.11,
-      "grad_norm": 6.5549445152282715,
+      "grad_norm": 8.715259552001953,
       "learning_rate": 5.729729729729731e-06,
-      "loss": 1.0273,
+      "loss": 1.0186,
       "step": 42660
     },
     {
       "epoch": 128.14,
-      "grad_norm": 4.753144264221191,
+      "grad_norm": 5.93709659576416,
       "learning_rate": 5.7287287287287295e-06,
-      "loss": 1.0045,
+      "loss": 1.0091,
       "step": 42670
     },
     {
       "epoch": 128.17,
-      "grad_norm": 6.567693710327148,
+      "grad_norm": 9.37485122680664,
       "learning_rate": 5.727727727727728e-06,
-      "loss": 1.0328,
+      "loss": 1.0453,
       "step": 42680
     },
     {
       "epoch": 128.2,
-      "grad_norm": 6.295573711395264,
+      "grad_norm": 7.640521049499512,
       "learning_rate": 5.726726726726727e-06,
-      "loss": 1.1002,
+      "loss": 1.09,
       "step": 42690
     },
     {
       "epoch": 128.23,
-      "grad_norm": 6.138082981109619,
+      "grad_norm": 7.239931583404541,
       "learning_rate": 5.725725725725726e-06,
-      "loss": 1.0323,
+      "loss": 1.0362,
       "step": 42700
     },
     {
       "epoch": 128.26,
-      "grad_norm": 6.794919967651367,
+      "grad_norm": 7.145660877227783,
       "learning_rate": 5.724724724724725e-06,
-      "loss": 1.0846,
+      "loss": 1.0735,
       "step": 42710
     },
     {
       "epoch": 128.29,
-      "grad_norm": 4.802174091339111,
+      "grad_norm": 7.262906551361084,
       "learning_rate": 5.723723723723724e-06,
-      "loss": 1.1095,
+      "loss": 1.1072,
       "step": 42720
     },
     {
       "epoch": 128.32,
-      "grad_norm": 4.720958232879639,
+      "grad_norm": 4.7792158126831055,
       "learning_rate": 5.7227227227227236e-06,
-      "loss": 1.0613,
+      "loss": 1.0534,
       "step": 42730
     },
     {
       "epoch": 128.35,
-      "grad_norm": 7.991726875305176,
+      "grad_norm": 8.959444999694824,
       "learning_rate": 5.721721721721722e-06,
-      "loss": 1.1369,
+      "loss": 1.1326,
       "step": 42740
     },
     {
       "epoch": 128.38,
-      "grad_norm": 8.015350341796875,
+      "grad_norm": 11.194101333618164,
       "learning_rate": 5.720720720720722e-06,
-      "loss": 1.1019,
+      "loss": 1.1063,
       "step": 42750
     },
     {
       "epoch": 128.41,
-      "grad_norm": 6.773718357086182,
+      "grad_norm": 6.243969917297363,
       "learning_rate": 5.719719719719721e-06,
-      "loss": 1.0706,
+      "loss": 1.0685,
       "step": 42760
     },
     {
       "epoch": 128.44,
-      "grad_norm": 7.129086494445801,
+      "grad_norm": 7.647986888885498,
       "learning_rate": 5.718718718718719e-06,
-      "loss": 1.0849,
+      "loss": 1.0895,
       "step": 42770
     },
     {
       "epoch": 128.47,
-      "grad_norm": 7.2590508460998535,
+      "grad_norm": 8.512574195861816,
       "learning_rate": 5.717717717717718e-06,
-      "loss": 1.0434,
+      "loss": 1.039,
       "step": 42780
     },
     {
       "epoch": 128.5,
-      "grad_norm": 5.63535213470459,
+      "grad_norm": 6.107377052307129,
       "learning_rate": 5.716716716716717e-06,
-      "loss": 1.0852,
+      "loss": 1.073,
       "step": 42790
     },
     {
       "epoch": 128.53,
-      "grad_norm": 10.273405075073242,
+      "grad_norm": 11.364848136901855,
       "learning_rate": 5.715715715715716e-06,
-      "loss": 1.1031,
+      "loss": 1.111,
       "step": 42800
     },
     {
       "epoch": 128.56,
-      "grad_norm": 8.19107437133789,
+      "grad_norm": 6.36797571182251,
       "learning_rate": 5.714714714714715e-06,
-      "loss": 1.0723,
+      "loss": 1.083,
       "step": 42810
     },
     {
       "epoch": 128.59,
-      "grad_norm": 8.94410514831543,
+      "grad_norm": 7.96816873550415,
       "learning_rate": 5.713713713713714e-06,
-      "loss": 1.0141,
+      "loss": 1.0164,
       "step": 42820
     },
     {
       "epoch": 128.62,
-      "grad_norm": 5.637784957885742,
+      "grad_norm": 6.8394598960876465,
       "learning_rate": 5.712712712712714e-06,
-      "loss": 1.0207,
+      "loss": 1.0228,
       "step": 42830
     },
     {
       "epoch": 128.65,
-      "grad_norm": 5.388303756713867,
+      "grad_norm": 6.308931350708008,
       "learning_rate": 5.711711711711712e-06,
-      "loss": 1.142,
+      "loss": 1.1367,
       "step": 42840
     },
     {
       "epoch": 128.68,
-      "grad_norm": 5.817361831665039,
+      "grad_norm": 6.880357265472412,
       "learning_rate": 5.710710710710711e-06,
-      "loss": 1.0909,
+      "loss": 1.0942,
       "step": 42850
     },
     {
       "epoch": 128.71,
-      "grad_norm": 4.132732391357422,
+      "grad_norm": 5.900829792022705,
       "learning_rate": 5.7097097097097095e-06,
-      "loss": 1.0625,
+      "loss": 1.079,
       "step": 42860
     },
     {
       "epoch": 128.74,
-      "grad_norm": 7.758181571960449,
+      "grad_norm": 8.388945579528809,
       "learning_rate": 5.708708708708709e-06,
-      "loss": 1.0554,
+      "loss": 1.0601,
       "step": 42870
     },
     {
       "epoch": 128.77,
-      "grad_norm": 4.966310501098633,
+      "grad_norm": 5.374595642089844,
       "learning_rate": 5.707707707707708e-06,
-      "loss": 0.9901,
+      "loss": 0.986,
       "step": 42880
     },
     {
       "epoch": 128.8,
-      "grad_norm": 6.602185249328613,
+      "grad_norm": 9.874635696411133,
       "learning_rate": 5.706706706706707e-06,
-      "loss": 1.001,
+      "loss": 0.9985,
       "step": 42890
     },
     {
       "epoch": 128.83,
-      "grad_norm": 7.587734699249268,
+      "grad_norm": 7.240536212921143,
       "learning_rate": 5.7057057057057065e-06,
-      "loss": 1.0778,
+      "loss": 1.0752,
       "step": 42900
     },
     {
       "epoch": 128.86,
-      "grad_norm": 8.935761451721191,
+      "grad_norm": 12.1953706741333,
       "learning_rate": 5.704704704704706e-06,
-      "loss": 1.049,
+      "loss": 1.0409,
       "step": 42910
     },
     {
       "epoch": 128.89,
-      "grad_norm": 5.434597492218018,
+      "grad_norm": 6.652745246887207,
       "learning_rate": 5.7037037037037045e-06,
-      "loss": 1.0308,
+      "loss": 1.0245,
       "step": 42920
     },
     {
       "epoch": 128.92,
-      "grad_norm": 10.215167999267578,
+      "grad_norm": 9.672639846801758,
       "learning_rate": 5.702702702702702e-06,
-      "loss": 1.0146,
+      "loss": 1.0167,
       "step": 42930
     },
     {
       "epoch": 128.95,
-      "grad_norm": 5.269582748413086,
+      "grad_norm": 7.310551166534424,
       "learning_rate": 5.701701701701702e-06,
-      "loss": 1.1073,
+      "loss": 1.1076,
       "step": 42940
     },
     {
       "epoch": 128.98,
-      "grad_norm": 5.209118366241455,
+      "grad_norm": 7.3925676345825195,
       "learning_rate": 5.700700700700701e-06,
-      "loss": 1.0766,
+      "loss": 1.0873,
       "step": 42950
     },
     {
       "epoch": 129.0,
-      "eval_accuracy": 0.8104,
-      "eval_loss": 0.6126055717468262,
-      "eval_runtime": 5.3853,
-      "eval_samples_per_second": 1856.912,
-      "eval_steps_per_second": 7.428,
+      "eval_accuracy": 0.8137,
+      "eval_loss": 0.6118718981742859,
+      "eval_runtime": 5.1655,
+      "eval_samples_per_second": 1935.933,
+      "eval_steps_per_second": 7.744,
       "step": 42957
     },
     {
       "epoch": 129.01,
-      "grad_norm": 6.9529876708984375,
+      "grad_norm": 9.38088321685791,
       "learning_rate": 5.6996996996997e-06,
-      "loss": 1.1852,
+      "loss": 1.2007,
       "step": 42960
     },
     {
       "epoch": 129.04,
-      "grad_norm": 5.7192463874816895,
+      "grad_norm": 7.545430660247803,
       "learning_rate": 5.698698698698699e-06,
-      "loss": 1.0758,
+      "loss": 1.0635,
       "step": 42970
     },
     {
       "epoch": 129.07,
-      "grad_norm": 10.307046890258789,
+      "grad_norm": 7.606111526489258,
       "learning_rate": 5.697697697697699e-06,
-      "loss": 1.0285,
+      "loss": 1.0289,
       "step": 42980
     },
     {
       "epoch": 129.1,
-      "grad_norm": 7.351574897766113,
+      "grad_norm": 9.640442848205566,
       "learning_rate": 5.696696696696697e-06,
-      "loss": 1.0891,
+      "loss": 1.1021,
       "step": 42990
     },
     {
       "epoch": 129.13,
-      "grad_norm": 5.717246055603027,
+      "grad_norm": 8.185342788696289,
       "learning_rate": 5.695695695695697e-06,
-      "loss": 1.0487,
+      "loss": 1.0589,
       "step": 43000
     },
     {
       "epoch": 129.16,
-      "grad_norm": 6.845102310180664,
+      "grad_norm": 5.259015083312988,
       "learning_rate": 5.694694694694696e-06,
-      "loss": 1.0162,
+      "loss": 1.0242,
       "step": 43010
     },
     {
       "epoch": 129.19,
-      "grad_norm": 5.503822326660156,
+      "grad_norm": 6.505908489227295,
       "learning_rate": 5.693693693693694e-06,
-      "loss": 1.0251,
+      "loss": 1.0161,
       "step": 43020
     },
     {
       "epoch": 129.22,
-      "grad_norm": 7.755375385284424,
+      "grad_norm": 8.584378242492676,
       "learning_rate": 5.692692692692693e-06,
-      "loss": 1.1031,
+      "loss": 1.096,
       "step": 43030
     },
     {
       "epoch": 129.25,
-      "grad_norm": 6.951054096221924,
+      "grad_norm": 9.774480819702148,
       "learning_rate": 5.691691691691692e-06,
-      "loss": 1.0485,
+      "loss": 1.0465,
       "step": 43040
     },
     {
       "epoch": 129.28,
-      "grad_norm": 6.725027084350586,
+      "grad_norm": 6.749785900115967,
       "learning_rate": 5.690690690690691e-06,
-      "loss": 0.9557,
+      "loss": 0.9609,
       "step": 43050
     },
     {
       "epoch": 129.31,
-      "grad_norm": 7.786320686340332,
+      "grad_norm": 7.507871150970459,
       "learning_rate": 5.68968968968969e-06,
-      "loss": 1.1176,
+      "loss": 1.1127,
       "step": 43060
     },
     {
       "epoch": 129.34,
-      "grad_norm": 5.400676727294922,
+      "grad_norm": 8.997273445129395,
       "learning_rate": 5.688688688688689e-06,
-      "loss": 0.9974,
+      "loss": 1.0037,
       "step": 43070
     },
     {
       "epoch": 129.37,
-      "grad_norm": 4.973848342895508,
+      "grad_norm": 6.664679050445557,
       "learning_rate": 5.687687687687689e-06,
-      "loss": 1.0697,
+      "loss": 1.0666,
       "step": 43080
     },
     {
       "epoch": 129.4,
-      "grad_norm": 6.567466735839844,
+      "grad_norm": 5.683163166046143,
       "learning_rate": 5.686686686686687e-06,
-      "loss": 1.0553,
+      "loss": 1.0639,
       "step": 43090
     },
     {
       "epoch": 129.43,
-      "grad_norm": 5.8075385093688965,
+      "grad_norm": 5.408211708068848,
       "learning_rate": 5.685685685685686e-06,
-      "loss": 1.097,
+      "loss": 1.0971,
       "step": 43100
     },
     {
       "epoch": 129.46,
-      "grad_norm": 7.981101036071777,
+      "grad_norm": 7.249408721923828,
       "learning_rate": 5.6846846846846846e-06,
-      "loss": 0.9945,
+      "loss": 0.9952,
       "step": 43110
     },
     {
       "epoch": 129.49,
-      "grad_norm": 5.4851226806640625,
+      "grad_norm": 6.00241756439209,
       "learning_rate": 5.683683683683684e-06,
-      "loss": 1.0647,
+      "loss": 1.0674,
       "step": 43120
     },
     {
       "epoch": 129.52,
-      "grad_norm": 6.681140422821045,
+      "grad_norm": 8.499560356140137,
       "learning_rate": 5.6826826826826835e-06,
-      "loss": 0.9736,
+      "loss": 0.9718,
       "step": 43130
     },
     {
       "epoch": 129.55,
-      "grad_norm": 5.457583427429199,
+      "grad_norm": 8.042102813720703,
       "learning_rate": 5.681681681681682e-06,
-      "loss": 1.1347,
+      "loss": 1.1294,
       "step": 43140
     },
     {
       "epoch": 129.58,
-      "grad_norm": 3.778116226196289,
+      "grad_norm": 5.853381633758545,
       "learning_rate": 5.6806806806806815e-06,
-      "loss": 1.0799,
+      "loss": 1.0946,
       "step": 43150
     },
     {
       "epoch": 129.61,
-      "grad_norm": 7.590839385986328,
+      "grad_norm": 8.178668975830078,
       "learning_rate": 5.67967967967968e-06,
-      "loss": 0.974,
+      "loss": 0.9723,
       "step": 43160
     },
     {
       "epoch": 129.64,
-      "grad_norm": 6.123884201049805,
+      "grad_norm": 6.389987468719482,
       "learning_rate": 5.6786786786786795e-06,
-      "loss": 1.0477,
+      "loss": 1.0519,
       "step": 43170
     },
     {
       "epoch": 129.67,
-      "grad_norm": 5.531554698944092,
+      "grad_norm": 6.241816997528076,
       "learning_rate": 5.677677677677677e-06,
-      "loss": 1.1508,
+      "loss": 1.1516,
       "step": 43180
     },
     {
       "epoch": 129.7,
-      "grad_norm": 6.759396076202393,
+      "grad_norm": 7.085709095001221,
       "learning_rate": 5.676676676676677e-06,
-      "loss": 1.056,
+      "loss": 1.0518,
       "step": 43190
     },
     {
       "epoch": 129.73,
-      "grad_norm": 10.044015884399414,
+      "grad_norm": 7.658003807067871,
       "learning_rate": 5.675675675675676e-06,
-      "loss": 0.9824,
+      "loss": 0.9764,
       "step": 43200
     },
     {
       "epoch": 129.76,
-      "grad_norm": 6.241951942443848,
+      "grad_norm": 6.17680549621582,
       "learning_rate": 5.674674674674675e-06,
-      "loss": 0.9955,
+      "loss": 0.9933,
       "step": 43210
     },
     {
       "epoch": 129.79,
-      "grad_norm": 7.370161056518555,
+      "grad_norm": 7.322697639465332,
       "learning_rate": 5.673673673673674e-06,
-      "loss": 1.0871,
+      "loss": 1.0913,
       "step": 43220
     },
     {
       "epoch": 129.82,
-      "grad_norm": 4.725358009338379,
+      "grad_norm": 8.218016624450684,
       "learning_rate": 5.672672672672674e-06,
-      "loss": 1.0381,
+      "loss": 1.0332,
       "step": 43230
     },
     {
       "epoch": 129.85,
-      "grad_norm": 6.76936149597168,
+      "grad_norm": 6.582704544067383,
       "learning_rate": 5.671671671671672e-06,
-      "loss": 1.1218,
+      "loss": 1.1265,
       "step": 43240
     },
     {
       "epoch": 129.88,
-      "grad_norm": 6.117926597595215,
+      "grad_norm": 6.096864700317383,
       "learning_rate": 5.670670670670672e-06,
-      "loss": 1.1193,
+      "loss": 1.1072,
       "step": 43250
     },
     {
       "epoch": 129.91,
-      "grad_norm": 4.78517484664917,
+      "grad_norm": 6.345552921295166,
       "learning_rate": 5.6696696696696694e-06,
-      "loss": 0.9945,
+      "loss": 0.9951,
       "step": 43260
     },
     {
       "epoch": 129.94,
-      "grad_norm": 5.7594709396362305,
+      "grad_norm": 5.06243896484375,
       "learning_rate": 5.668668668668669e-06,
-      "loss": 1.0561,
+      "loss": 1.0566,
       "step": 43270
     },
     {
       "epoch": 129.97,
-      "grad_norm": 5.686593532562256,
+      "grad_norm": 5.994672775268555,
       "learning_rate": 5.6676676676676675e-06,
-      "loss": 1.0351,
+      "loss": 1.0453,
       "step": 43280
     },
     {
       "epoch": 130.0,
-      "grad_norm": 26.71927261352539,
+      "grad_norm": 26.37158203125,
       "learning_rate": 5.666666666666667e-06,
-      "loss": 1.2749,
+      "loss": 1.2733,
       "step": 43290
     },
     {
       "epoch": 130.0,
-      "eval_accuracy": 0.8106,
-      "eval_loss": 0.612274706363678,
-      "eval_runtime": 5.377,
-      "eval_samples_per_second": 1859.758,
-      "eval_steps_per_second": 7.439,
+      "eval_accuracy": 0.8133,
+      "eval_loss": 0.6104779839515686,
+      "eval_runtime": 5.3188,
+      "eval_samples_per_second": 1880.136,
+      "eval_steps_per_second": 7.521,
       "step": 43290
     },
     {
       "epoch": 130.03,
-      "grad_norm": 8.28128719329834,
+      "grad_norm": 7.938552379608154,
       "learning_rate": 5.665665665665666e-06,
-      "loss": 1.0719,
+      "loss": 1.0823,
       "step": 43300
     },
     {
       "epoch": 130.06,
-      "grad_norm": 6.186869144439697,
+      "grad_norm": 8.45387077331543,
       "learning_rate": 5.664664664664665e-06,
-      "loss": 1.04,
+      "loss": 1.0564,
       "step": 43310
     },
     {
       "epoch": 130.09,
-      "grad_norm": 8.408331871032715,
+      "grad_norm": 5.452244281768799,
       "learning_rate": 5.663663663663664e-06,
-      "loss": 1.0213,
+      "loss": 1.0158,
       "step": 43320
     },
     {
       "epoch": 130.12,
-      "grad_norm": 5.894686698913574,
+      "grad_norm": 6.265316009521484,
       "learning_rate": 5.662662662662664e-06,
-      "loss": 1.0411,
+      "loss": 1.0447,
       "step": 43330
     },
     {
       "epoch": 130.15,
-      "grad_norm": 7.380275726318359,
+      "grad_norm": 8.513215065002441,
       "learning_rate": 5.661661661661662e-06,
-      "loss": 0.9749,
+      "loss": 0.9597,
       "step": 43340
     },
     {
       "epoch": 130.18,
-      "grad_norm": 6.275332450866699,
+      "grad_norm": 7.725722312927246,
       "learning_rate": 5.660660660660661e-06,
-      "loss": 1.0139,
+      "loss": 1.0117,
       "step": 43350
     },
     {
       "epoch": 130.21,
-      "grad_norm": 5.896085739135742,
+      "grad_norm": 4.893223762512207,
       "learning_rate": 5.65965965965966e-06,
-      "loss": 1.01,
+      "loss": 1.012,
       "step": 43360
     },
     {
       "epoch": 130.24,
-      "grad_norm": 6.679431438446045,
+      "grad_norm": 6.566874027252197,
       "learning_rate": 5.658658658658659e-06,
-      "loss": 1.0923,
+      "loss": 1.0888,
       "step": 43370
     },
     {
       "epoch": 130.27,
-      "grad_norm": 6.202209949493408,
+      "grad_norm": 8.723212242126465,
       "learning_rate": 5.657657657657658e-06,
-      "loss": 0.9882,
+      "loss": 0.992,
       "step": 43380
     },
     {
       "epoch": 130.3,
-      "grad_norm": 6.524528503417969,
+      "grad_norm": 6.635607719421387,
       "learning_rate": 5.656656656656657e-06,
-      "loss": 1.1193,
+      "loss": 1.1116,
       "step": 43390
     },
     {
       "epoch": 130.33,
-      "grad_norm": 6.4759440422058105,
+      "grad_norm": 9.40714168548584,
       "learning_rate": 5.6556556556556565e-06,
-      "loss": 1.1177,
+      "loss": 1.1099,
       "step": 43400
     },
     {
       "epoch": 130.36,
-      "grad_norm": 6.315952301025391,
+      "grad_norm": 6.9509196281433105,
       "learning_rate": 5.654654654654655e-06,
-      "loss": 1.1152,
+      "loss": 1.1172,
       "step": 43410
     },
     {
       "epoch": 130.39,
-      "grad_norm": 6.093774318695068,
+      "grad_norm": 9.829021453857422,
       "learning_rate": 5.6536536536536546e-06,
-      "loss": 0.9732,
+      "loss": 0.9664,
       "step": 43420
     },
     {
       "epoch": 130.42,
-      "grad_norm": 9.21229076385498,
+      "grad_norm": 9.202765464782715,
       "learning_rate": 5.652652652652652e-06,
-      "loss": 1.0758,
+      "loss": 1.0726,
       "step": 43430
     },
     {
       "epoch": 130.45,
-      "grad_norm": 8.042040824890137,
+      "grad_norm": 6.327095985412598,
       "learning_rate": 5.651651651651652e-06,
-      "loss": 1.0377,
+      "loss": 1.0369,
       "step": 43440
     },
     {
       "epoch": 130.48,
-      "grad_norm": 5.076534748077393,
+      "grad_norm": 7.771714687347412,
       "learning_rate": 5.650650650650651e-06,
-      "loss": 1.097,
+      "loss": 1.0873,
       "step": 43450
     },
     {
       "epoch": 130.51,
-      "grad_norm": 7.783759117126465,
+      "grad_norm": 4.371439456939697,
       "learning_rate": 5.64964964964965e-06,
-      "loss": 1.1172,
+      "loss": 1.1262,
       "step": 43460
     },
     {
       "epoch": 130.54,
-      "grad_norm": 8.248390197753906,
+      "grad_norm": 9.629596710205078,
       "learning_rate": 5.648648648648649e-06,
-      "loss": 1.0342,
+      "loss": 1.0314,
       "step": 43470
     },
     {
       "epoch": 130.57,
-      "grad_norm": 6.495917320251465,
+      "grad_norm": 7.339854717254639,
       "learning_rate": 5.647647647647649e-06,
-      "loss": 1.0522,
+      "loss": 1.0569,
       "step": 43480
     },
     {
       "epoch": 130.6,
-      "grad_norm": 7.410682201385498,
+      "grad_norm": 7.039999485015869,
       "learning_rate": 5.646646646646647e-06,
-      "loss": 1.0906,
+      "loss": 1.0833,
       "step": 43490
     },
     {
       "epoch": 130.63,
-      "grad_norm": 6.169040679931641,
+      "grad_norm": 9.453725814819336,
       "learning_rate": 5.645645645645647e-06,
-      "loss": 1.0353,
+      "loss": 1.0422,
       "step": 43500
     },
     {
       "epoch": 130.66,
-      "grad_norm": 4.490420341491699,
+      "grad_norm": 4.076911926269531,
       "learning_rate": 5.6446446446446445e-06,
-      "loss": 0.9585,
+      "loss": 0.9543,
       "step": 43510
     },
     {
       "epoch": 130.69,
-      "grad_norm": 6.923290729522705,
+      "grad_norm": 7.213587284088135,
       "learning_rate": 5.643643643643644e-06,
-      "loss": 1.0772,
+      "loss": 1.0791,
       "step": 43520
     },
     {
       "epoch": 130.72,
-      "grad_norm": 5.692728519439697,
+      "grad_norm": 6.168738842010498,
       "learning_rate": 5.6426426426426425e-06,
-      "loss": 1.0676,
+      "loss": 1.0693,
       "step": 43530
     },
     {
       "epoch": 130.75,
-      "grad_norm": 6.934525012969971,
+      "grad_norm": 7.725180625915527,
       "learning_rate": 5.641641641641642e-06,
-      "loss": 1.0275,
+      "loss": 1.0239,
       "step": 43540
     },
     {
       "epoch": 130.78,
-      "grad_norm": 7.507137775421143,
+      "grad_norm": 10.022038459777832,
       "learning_rate": 5.640640640640641e-06,
-      "loss": 1.0167,
+      "loss": 1.0175,
       "step": 43550
     },
     {
       "epoch": 130.81,
-      "grad_norm": 8.271552085876465,
+      "grad_norm": 8.696328163146973,
       "learning_rate": 5.63963963963964e-06,
-      "loss": 1.0124,
+      "loss": 1.0159,
       "step": 43560
     },
     {
       "epoch": 130.84,
-      "grad_norm": 7.302443981170654,
+      "grad_norm": 9.27116584777832,
       "learning_rate": 5.638638638638639e-06,
-      "loss": 1.0935,
+      "loss": 1.0889,
       "step": 43570
     },
     {
       "epoch": 130.87,
-      "grad_norm": 5.969295501708984,
+      "grad_norm": 6.3157572746276855,
       "learning_rate": 5.637637637637639e-06,
-      "loss": 1.055,
+      "loss": 1.0501,
       "step": 43580
     },
     {
       "epoch": 130.9,
-      "grad_norm": 5.20751953125,
+      "grad_norm": 8.258955955505371,
       "learning_rate": 5.6366366366366375e-06,
-      "loss": 1.0912,
+      "loss": 1.088,
       "step": 43590
     },
     {
       "epoch": 130.93,
-      "grad_norm": 5.690283298492432,
+      "grad_norm": 6.119899749755859,
       "learning_rate": 5.635635635635636e-06,
-      "loss": 1.0729,
+      "loss": 1.0662,
       "step": 43600
     },
     {
       "epoch": 130.96,
-      "grad_norm": 7.045783519744873,
+      "grad_norm": 5.478098392486572,
       "learning_rate": 5.634634634634635e-06,
-      "loss": 1.0457,
+      "loss": 1.0373,
       "step": 43610
     },
     {
       "epoch": 130.99,
-      "grad_norm": 4.315288066864014,
+      "grad_norm": 4.709186553955078,
       "learning_rate": 5.633633633633634e-06,
-      "loss": 1.1147,
+      "loss": 1.1138,
       "step": 43620
     },
     {
       "epoch": 131.0,
-      "eval_accuracy": 0.8105,
-      "eval_loss": 0.612774133682251,
-      "eval_runtime": 5.4823,
-      "eval_samples_per_second": 1824.05,
-      "eval_steps_per_second": 7.296,
+      "eval_accuracy": 0.8115,
+      "eval_loss": 0.6119075417518616,
+      "eval_runtime": 5.1592,
+      "eval_samples_per_second": 1938.276,
+      "eval_steps_per_second": 7.753,
       "step": 43623
     },
     {
       "epoch": 131.02,
-      "grad_norm": 6.413859844207764,
+      "grad_norm": 6.719168663024902,
       "learning_rate": 5.632632632632633e-06,
-      "loss": 1.2041,
+      "loss": 1.2221,
       "step": 43630
     },
     {
       "epoch": 131.05,
-      "grad_norm": 7.785806179046631,
+      "grad_norm": 6.464757442474365,
       "learning_rate": 5.631631631631632e-06,
-      "loss": 1.0567,
+      "loss": 1.0594,
       "step": 43640
     },
     {
       "epoch": 131.08,
-      "grad_norm": 6.754075050354004,
+      "grad_norm": 7.317476749420166,
       "learning_rate": 5.6306306306306316e-06,
-      "loss": 1.0421,
+      "loss": 1.0425,
       "step": 43650
     },
     {
       "epoch": 131.11,
-      "grad_norm": 5.4832563400268555,
+      "grad_norm": 7.182113170623779,
       "learning_rate": 5.62962962962963e-06,
-      "loss": 1.044,
+      "loss": 1.0418,
       "step": 43660
     },
     {
       "epoch": 131.14,
-      "grad_norm": 6.42626428604126,
+      "grad_norm": 9.223611831665039,
       "learning_rate": 5.62862862862863e-06,
-      "loss": 1.0667,
+      "loss": 1.0719,
       "step": 43670
     },
     {
       "epoch": 131.17,
-      "grad_norm": 7.451161861419678,
+      "grad_norm": 8.525914192199707,
       "learning_rate": 5.627627627627627e-06,
-      "loss": 1.0307,
+      "loss": 1.0202,
       "step": 43680
     },
     {
       "epoch": 131.2,
-      "grad_norm": 8.218131065368652,
+      "grad_norm": 9.475523948669434,
       "learning_rate": 5.626626626626627e-06,
-      "loss": 1.0671,
+      "loss": 1.0696,
       "step": 43690
     },
     {
       "epoch": 131.23,
-      "grad_norm": 8.037705421447754,
+      "grad_norm": 10.510824203491211,
       "learning_rate": 5.625625625625626e-06,
-      "loss": 1.0484,
+      "loss": 1.0514,
       "step": 43700
     },
     {
       "epoch": 131.26,
-      "grad_norm": 7.438817024230957,
+      "grad_norm": 10.046808242797852,
       "learning_rate": 5.624624624624625e-06,
-      "loss": 1.1009,
+      "loss": 1.1069,
       "step": 43710
     },
     {
       "epoch": 131.29,
-      "grad_norm": 7.395327091217041,
+      "grad_norm": 7.835263252258301,
       "learning_rate": 5.623623623623624e-06,
-      "loss": 1.0527,
+      "loss": 1.0536,
       "step": 43720
     },
     {
       "epoch": 131.32,
-      "grad_norm": 6.624264717102051,
+      "grad_norm": 7.2568864822387695,
       "learning_rate": 5.622622622622623e-06,
-      "loss": 1.0506,
+      "loss": 1.0572,
       "step": 43730
     },
     {
       "epoch": 131.35,
-      "grad_norm": 7.065186023712158,
+      "grad_norm": 6.617342948913574,
       "learning_rate": 5.621621621621622e-06,
-      "loss": 0.9746,
+      "loss": 0.9701,
       "step": 43740
     },
     {
       "epoch": 131.38,
-      "grad_norm": 6.068761348724365,
+      "grad_norm": 9.19428825378418,
       "learning_rate": 5.620620620620622e-06,
-      "loss": 1.1012,
+      "loss": 1.1108,
       "step": 43750
     },
     {
       "epoch": 131.41,
-      "grad_norm": 4.634932518005371,
+      "grad_norm": 6.478476047515869,
       "learning_rate": 5.6196196196196195e-06,
-      "loss": 0.9986,
+      "loss": 0.9939,
       "step": 43760
     },
     {
       "epoch": 131.44,
-      "grad_norm": 5.974303722381592,
+      "grad_norm": 6.378845691680908,
       "learning_rate": 5.618618618618619e-06,
-      "loss": 1.0245,
+      "loss": 1.0224,
       "step": 43770
     },
     {
       "epoch": 131.47,
-      "grad_norm": 5.71920108795166,
+      "grad_norm": 8.010004043579102,
       "learning_rate": 5.6176176176176175e-06,
-      "loss": 1.0124,
+      "loss": 1.0028,
       "step": 43780
     },
     {
       "epoch": 131.5,
-      "grad_norm": 7.914678573608398,
+      "grad_norm": 7.006198406219482,
       "learning_rate": 5.616616616616617e-06,
-      "loss": 1.0696,
+      "loss": 1.0695,
       "step": 43790
     },
     {
       "epoch": 131.53,
-      "grad_norm": 7.338247299194336,
+      "grad_norm": 9.294214248657227,
       "learning_rate": 5.615615615615616e-06,
-      "loss": 0.9479,
+      "loss": 0.9503,
       "step": 43800
     },
     {
       "epoch": 131.56,
-      "grad_norm": 5.676547050476074,
+      "grad_norm": 5.488953113555908,
       "learning_rate": 5.614614614614615e-06,
-      "loss": 1.0383,
+      "loss": 1.0314,
       "step": 43810
     },
     {
       "epoch": 131.59,
-      "grad_norm": 8.19979190826416,
+      "grad_norm": 7.289972305297852,
       "learning_rate": 5.6136136136136145e-06,
-      "loss": 1.0624,
+      "loss": 1.0559,
       "step": 43820
     },
     {
       "epoch": 131.62,
-      "grad_norm": 8.185334205627441,
+      "grad_norm": 9.740937232971191,
       "learning_rate": 5.612612612612614e-06,
-      "loss": 1.0404,
+      "loss": 1.0395,
       "step": 43830
     },
     {
       "epoch": 131.65,
-      "grad_norm": 5.522820949554443,
+      "grad_norm": 8.894689559936523,
       "learning_rate": 5.6116116116116125e-06,
-      "loss": 1.073,
+      "loss": 1.0722,
       "step": 43840
     },
     {
       "epoch": 131.68,
-      "grad_norm": 7.940682411193848,
+      "grad_norm": 8.574851989746094,
       "learning_rate": 5.61061061061061e-06,
-      "loss": 1.0582,
+      "loss": 1.0513,
       "step": 43850
     },
     {
       "epoch": 131.71,
-      "grad_norm": 6.620090007781982,
+      "grad_norm": 6.258894920349121,
       "learning_rate": 5.60960960960961e-06,
-      "loss": 1.0234,
+      "loss": 1.0177,
       "step": 43860
     },
     {
       "epoch": 131.74,
-      "grad_norm": 6.747226238250732,
+      "grad_norm": 8.458909034729004,
       "learning_rate": 5.608608608608609e-06,
-      "loss": 1.0543,
+      "loss": 1.0476,
       "step": 43870
     },
     {
       "epoch": 131.77,
-      "grad_norm": 5.977757453918457,
+      "grad_norm": 7.8876800537109375,
       "learning_rate": 5.607607607607608e-06,
-      "loss": 1.105,
+      "loss": 1.0993,
       "step": 43880
     },
     {
       "epoch": 131.8,
-      "grad_norm": 5.664333343505859,
+      "grad_norm": 5.728778839111328,
       "learning_rate": 5.606606606606607e-06,
-      "loss": 1.0895,
+      "loss": 1.091,
       "step": 43890
     },
     {
       "epoch": 131.83,
-      "grad_norm": 5.615564346313477,
+      "grad_norm": 5.491821765899658,
       "learning_rate": 5.605605605605607e-06,
-      "loss": 1.0441,
+      "loss": 1.0365,
       "step": 43900
     },
     {
       "epoch": 131.86,
-      "grad_norm": 5.15415096282959,
+      "grad_norm": 6.641414165496826,
       "learning_rate": 5.604604604604605e-06,
-      "loss": 1.058,
+      "loss": 1.0473,
       "step": 43910
     },
     {
       "epoch": 131.89,
-      "grad_norm": 5.026356220245361,
+      "grad_norm": 6.7104339599609375,
       "learning_rate": 5.603603603603605e-06,
-      "loss": 1.0764,
+      "loss": 1.0731,
       "step": 43920
     },
     {
       "epoch": 131.92,
-      "grad_norm": 6.412919998168945,
+      "grad_norm": 5.212334632873535,
       "learning_rate": 5.602602602602602e-06,
-      "loss": 1.0411,
+      "loss": 1.0314,
       "step": 43930
     },
     {
       "epoch": 131.95,
-      "grad_norm": 9.439248085021973,
+      "grad_norm": 8.561535835266113,
       "learning_rate": 5.601601601601602e-06,
-      "loss": 1.1599,
+      "loss": 1.1556,
       "step": 43940
     },
     {
       "epoch": 131.98,
-      "grad_norm": 7.656896114349365,
+      "grad_norm": 7.960371494293213,
       "learning_rate": 5.600600600600601e-06,
-      "loss": 1.0357,
+      "loss": 1.0391,
       "step": 43950
     },
     {
       "epoch": 132.0,
-      "eval_accuracy": 0.8138,
-      "eval_loss": 0.6059504151344299,
-      "eval_runtime": 5.4604,
-      "eval_samples_per_second": 1831.354,
-      "eval_steps_per_second": 7.325,
+      "eval_accuracy": 0.8118,
+      "eval_loss": 0.6064999103546143,
+      "eval_runtime": 5.2938,
+      "eval_samples_per_second": 1889.004,
+      "eval_steps_per_second": 7.556,
       "step": 43956
     },
     {
       "epoch": 132.01,
-      "grad_norm": 7.0264973640441895,
+      "grad_norm": 7.8103437423706055,
       "learning_rate": 5.5995995995996e-06,
-      "loss": 1.2065,
+      "loss": 1.1937,
       "step": 43960
     },
     {
       "epoch": 132.04,
-      "grad_norm": 8.200530052185059,
+      "grad_norm": 5.790878772735596,
       "learning_rate": 5.598598598598599e-06,
-      "loss": 1.0859,
+      "loss": 1.0798,
       "step": 43970
     },
     {
       "epoch": 132.07,
-      "grad_norm": 5.517313480377197,
+      "grad_norm": 6.4842209815979,
       "learning_rate": 5.597597597597598e-06,
-      "loss": 1.0312,
+      "loss": 1.0325,
       "step": 43980
     },
     {
       "epoch": 132.1,
-      "grad_norm": 6.03942346572876,
+      "grad_norm": 6.327904224395752,
       "learning_rate": 5.596596596596597e-06,
-      "loss": 1.061,
+      "loss": 1.0569,
       "step": 43990
     },
     {
       "epoch": 132.13,
-      "grad_norm": 5.7256178855896,
+      "grad_norm": 6.416882038116455,
       "learning_rate": 5.595595595595597e-06,
-      "loss": 1.0807,
+      "loss": 1.0882,
       "step": 44000
     },
     {
       "epoch": 132.16,
-      "grad_norm": 8.059256553649902,
+      "grad_norm": 7.830126762390137,
       "learning_rate": 5.5945945945945945e-06,
-      "loss": 1.048,
+      "loss": 1.0447,
       "step": 44010
     },
     {
       "epoch": 132.19,
-      "grad_norm": 6.639590263366699,
+      "grad_norm": 7.817397594451904,
       "learning_rate": 5.593593593593594e-06,
-      "loss": 1.0318,
+      "loss": 1.0287,
       "step": 44020
     },
     {
       "epoch": 132.22,
-      "grad_norm": 7.008256912231445,
+      "grad_norm": 6.9719672203063965,
       "learning_rate": 5.5925925925925926e-06,
-      "loss": 1.0638,
+      "loss": 1.0662,
       "step": 44030
     },
     {
       "epoch": 132.25,
-      "grad_norm": 8.61005687713623,
+      "grad_norm": 7.274653911590576,
       "learning_rate": 5.591591591591592e-06,
-      "loss": 1.0299,
+      "loss": 1.0234,
       "step": 44040
     },
     {
       "epoch": 132.28,
-      "grad_norm": 10.105414390563965,
+      "grad_norm": 11.008844375610352,
       "learning_rate": 5.5905905905905915e-06,
-      "loss": 1.0636,
+      "loss": 1.0617,
       "step": 44050
     },
     {
       "epoch": 132.31,
-      "grad_norm": 6.04059362411499,
+      "grad_norm": 6.187415599822998,
       "learning_rate": 5.58958958958959e-06,
-      "loss": 1.0364,
+      "loss": 1.0312,
       "step": 44060
     },
     {
       "epoch": 132.34,
-      "grad_norm": 6.740742206573486,
+      "grad_norm": 7.222479343414307,
       "learning_rate": 5.5885885885885895e-06,
-      "loss": 1.0504,
+      "loss": 1.0501,
       "step": 44070
     },
     {
       "epoch": 132.37,
-      "grad_norm": 8.183216094970703,
+      "grad_norm": 10.337240219116211,
       "learning_rate": 5.587587587587588e-06,
-      "loss": 1.0677,
+      "loss": 1.0675,
       "step": 44080
     },
     {
       "epoch": 132.4,
-      "grad_norm": 5.765298843383789,
+      "grad_norm": 7.715066909790039,
       "learning_rate": 5.5865865865865875e-06,
-      "loss": 0.9886,
+      "loss": 1.0001,
       "step": 44090
     },
     {
       "epoch": 132.43,
-      "grad_norm": 6.586012363433838,
+      "grad_norm": 8.008761405944824,
       "learning_rate": 5.585585585585585e-06,
-      "loss": 1.0473,
+      "loss": 1.0435,
       "step": 44100
     },
     {
       "epoch": 132.46,
-      "grad_norm": 5.947220802307129,
+      "grad_norm": 6.219756126403809,
       "learning_rate": 5.584584584584585e-06,
-      "loss": 1.1356,
+      "loss": 1.127,
       "step": 44110
     },
     {
       "epoch": 132.49,
-      "grad_norm": 9.120024681091309,
+      "grad_norm": 5.502279281616211,
       "learning_rate": 5.583583583583584e-06,
-      "loss": 1.0298,
+      "loss": 1.0257,
       "step": 44120
     },
     {
       "epoch": 132.52,
-      "grad_norm": 9.30467414855957,
+      "grad_norm": 8.742290496826172,
       "learning_rate": 5.582582582582583e-06,
-      "loss": 1.0163,
+      "loss": 1.0256,
       "step": 44130
     },
     {
       "epoch": 132.55,
-      "grad_norm": 5.702168941497803,
+      "grad_norm": 6.860732078552246,
       "learning_rate": 5.581581581581582e-06,
-      "loss": 1.0515,
+      "loss": 1.0569,
       "step": 44140
     },
     {
       "epoch": 132.58,
-      "grad_norm": 7.081487655639648,
+      "grad_norm": 7.1755571365356445,
       "learning_rate": 5.580580580580582e-06,
-      "loss": 1.0884,
+      "loss": 1.0896,
       "step": 44150
     },
     {
       "epoch": 132.61,
-      "grad_norm": 4.741261959075928,
+      "grad_norm": 7.872689723968506,
       "learning_rate": 5.57957957957958e-06,
-      "loss": 1.0044,
+      "loss": 1.0084,
       "step": 44160
     },
     {
       "epoch": 132.64,
-      "grad_norm": 6.852700710296631,
+      "grad_norm": 7.113485813140869,
       "learning_rate": 5.57857857857858e-06,
-      "loss": 1.0432,
+      "loss": 1.0369,
       "step": 44170
     },
     {
       "epoch": 132.67,
-      "grad_norm": 5.795170783996582,
+      "grad_norm": 7.755223274230957,
       "learning_rate": 5.577577577577577e-06,
-      "loss": 1.0959,
+      "loss": 1.0937,
       "step": 44180
     },
     {
       "epoch": 132.7,
-      "grad_norm": 7.352586269378662,
+      "grad_norm": 9.206199645996094,
       "learning_rate": 5.576576576576577e-06,
-      "loss": 1.0424,
+      "loss": 1.0365,
       "step": 44190
     },
     {
       "epoch": 132.73,
-      "grad_norm": 7.070411682128906,
+      "grad_norm": 8.265064239501953,
       "learning_rate": 5.5755755755755755e-06,
-      "loss": 1.0179,
+      "loss": 1.0059,
       "step": 44200
     },
     {
       "epoch": 132.76,
-      "grad_norm": 6.684155464172363,
+      "grad_norm": 7.873532295227051,
       "learning_rate": 5.574574574574575e-06,
-      "loss": 1.0705,
+      "loss": 1.0738,
       "step": 44210
     },
     {
       "epoch": 132.79,
-      "grad_norm": 4.720856666564941,
+      "grad_norm": 4.9005446434021,
       "learning_rate": 5.573573573573574e-06,
-      "loss": 1.0601,
+      "loss": 1.0549,
       "step": 44220
     },
     {
       "epoch": 132.82,
-      "grad_norm": 5.835788249969482,
+      "grad_norm": 6.7910380363464355,
       "learning_rate": 5.572572572572573e-06,
-      "loss": 1.0635,
+      "loss": 1.0478,
       "step": 44230
     },
     {
       "epoch": 132.85,
-      "grad_norm": 5.809361457824707,
+      "grad_norm": 7.070315361022949,
       "learning_rate": 5.571571571571572e-06,
-      "loss": 1.077,
+      "loss": 1.067,
       "step": 44240
     },
     {
       "epoch": 132.88,
-      "grad_norm": 7.94553279876709,
+      "grad_norm": 7.580498218536377,
       "learning_rate": 5.570570570570572e-06,
-      "loss": 1.0064,
+      "loss": 1.0138,
       "step": 44250
     },
     {
       "epoch": 132.91,
-      "grad_norm": 5.605452537536621,
+      "grad_norm": 6.0074663162231445,
       "learning_rate": 5.5695695695695696e-06,
-      "loss": 1.0123,
+      "loss": 1.0066,
       "step": 44260
     },
     {
       "epoch": 132.94,
-      "grad_norm": 6.241903305053711,
+      "grad_norm": 7.629077434539795,
       "learning_rate": 5.568568568568569e-06,
-      "loss": 1.1121,
+      "loss": 1.1141,
       "step": 44270
     },
     {
       "epoch": 132.97,
-      "grad_norm": 8.926802635192871,
+      "grad_norm": 7.1949462890625,
       "learning_rate": 5.567567567567568e-06,
-      "loss": 1.0424,
+      "loss": 1.0452,
       "step": 44280
     },
     {
       "epoch": 133.0,
-      "eval_accuracy": 0.8146,
-      "eval_loss": 0.606224536895752,
-      "eval_runtime": 5.3878,
-      "eval_samples_per_second": 1856.059,
-      "eval_steps_per_second": 7.424,
+      "eval_accuracy": 0.8157,
+      "eval_loss": 0.6054458022117615,
+      "eval_runtime": 5.2219,
+      "eval_samples_per_second": 1914.999,
+      "eval_steps_per_second": 7.66,
       "step": 44289
     },
     {
       "epoch": 133.0,
-      "grad_norm": 9.328606605529785,
+      "grad_norm": 14.592885971069336,
       "learning_rate": 5.566566566566567e-06,
-      "loss": 1.4171,
+      "loss": 1.4169,
       "step": 44290
     },
     {
       "epoch": 133.03,
-      "grad_norm": 7.5865888595581055,
+      "grad_norm": 8.658323287963867,
       "learning_rate": 5.565565565565566e-06,
-      "loss": 1.0388,
+      "loss": 1.0347,
       "step": 44300
     },
     {
       "epoch": 133.06,
-      "grad_norm": 6.25778865814209,
+      "grad_norm": 6.729433059692383,
       "learning_rate": 5.564564564564565e-06,
-      "loss": 1.0484,
+      "loss": 1.052,
       "step": 44310
     },
     {
       "epoch": 133.09,
-      "grad_norm": 7.7305908203125,
+      "grad_norm": 6.435615539550781,
       "learning_rate": 5.5635635635635645e-06,
-      "loss": 1.0097,
+      "loss": 1.0037,
       "step": 44320
     },
     {
       "epoch": 133.12,
-      "grad_norm": 7.864670276641846,
+      "grad_norm": 5.869286060333252,
       "learning_rate": 5.562562562562563e-06,
-      "loss": 1.0821,
+      "loss": 1.0973,
       "step": 44330
     },
     {
       "epoch": 133.15,
-      "grad_norm": 9.126596450805664,
+      "grad_norm": 7.691341400146484,
       "learning_rate": 5.561561561561562e-06,
-      "loss": 1.0666,
+      "loss": 1.0606,
       "step": 44340
     },
     {
       "epoch": 133.18,
-      "grad_norm": 6.5579705238342285,
+      "grad_norm": 7.507673740386963,
       "learning_rate": 5.56056056056056e-06,
-      "loss": 1.0018,
+      "loss": 1.0105,
       "step": 44350
     },
     {
       "epoch": 133.21,
-      "grad_norm": 6.001008987426758,
+      "grad_norm": 7.070250511169434,
       "learning_rate": 5.55955955955956e-06,
-      "loss": 1.0226,
+      "loss": 1.0279,
       "step": 44360
     },
     {
       "epoch": 133.24,
-      "grad_norm": 6.5694475173950195,
+      "grad_norm": 11.23974323272705,
       "learning_rate": 5.558558558558559e-06,
-      "loss": 1.0829,
+      "loss": 1.0774,
       "step": 44370
     },
     {
       "epoch": 133.27,
-      "grad_norm": 5.689926624298096,
+      "grad_norm": 5.527629375457764,
       "learning_rate": 5.557557557557558e-06,
-      "loss": 1.0469,
+      "loss": 1.0543,
       "step": 44380
     },
     {
       "epoch": 133.3,
-      "grad_norm": 7.157435894012451,
+      "grad_norm": 6.402199745178223,
       "learning_rate": 5.556556556556557e-06,
-      "loss": 1.0441,
+      "loss": 1.0429,
       "step": 44390
     },
     {
       "epoch": 133.33,
-      "grad_norm": 8.706755638122559,
+      "grad_norm": 8.349788665771484,
       "learning_rate": 5.555555555555557e-06,
-      "loss": 1.0719,
+      "loss": 1.0629,
       "step": 44400
     },
     {
       "epoch": 133.36,
-      "grad_norm": 5.572200775146484,
+      "grad_norm": 5.773845195770264,
       "learning_rate": 5.554554554554555e-06,
-      "loss": 1.0445,
+      "loss": 1.05,
       "step": 44410
     },
     {
       "epoch": 133.39,
-      "grad_norm": 5.756275653839111,
+      "grad_norm": 9.620352745056152,
       "learning_rate": 5.553553553553555e-06,
-      "loss": 1.0671,
+      "loss": 1.064,
       "step": 44420
     },
     {
       "epoch": 133.42,
-      "grad_norm": 5.571158409118652,
+      "grad_norm": 6.500404357910156,
       "learning_rate": 5.5525525525525525e-06,
-      "loss": 1.0317,
+      "loss": 1.0381,
       "step": 44430
     },
     {
       "epoch": 133.45,
-      "grad_norm": 7.440280914306641,
+      "grad_norm": 9.340810775756836,
       "learning_rate": 5.551551551551552e-06,
-      "loss": 1.0566,
+      "loss": 1.0497,
       "step": 44440
     },
     {
       "epoch": 133.48,
-      "grad_norm": 5.805731773376465,
+      "grad_norm": 7.259245872497559,
       "learning_rate": 5.5505505505505505e-06,
-      "loss": 0.9987,
+      "loss": 1.0005,
       "step": 44450
     },
     {
       "epoch": 133.51,
-      "grad_norm": 6.5628814697265625,
+      "grad_norm": 6.025730133056641,
       "learning_rate": 5.54954954954955e-06,
-      "loss": 1.1286,
+      "loss": 1.1275,
       "step": 44460
     },
     {
       "epoch": 133.54,
-      "grad_norm": 10.104125022888184,
+      "grad_norm": 9.196330070495605,
       "learning_rate": 5.548548548548549e-06,
-      "loss": 1.0776,
+      "loss": 1.0941,
       "step": 44470
     },
     {
       "epoch": 133.57,
-      "grad_norm": 7.604818820953369,
+      "grad_norm": 7.980248928070068,
       "learning_rate": 5.547547547547548e-06,
-      "loss": 1.0605,
+      "loss": 1.0629,
       "step": 44480
     },
     {
       "epoch": 133.6,
-      "grad_norm": 7.05104923248291,
+      "grad_norm": 6.423136234283447,
       "learning_rate": 5.546546546546547e-06,
-      "loss": 1.0152,
+      "loss": 1.0196,
       "step": 44490
     },
     {
       "epoch": 133.63,
-      "grad_norm": 7.211410045623779,
+      "grad_norm": 5.8467302322387695,
       "learning_rate": 5.545545545545547e-06,
-      "loss": 1.0293,
+      "loss": 1.0306,
       "step": 44500
     },
     {
       "epoch": 133.66,
-      "grad_norm": 6.2140421867370605,
+      "grad_norm": 12.092368125915527,
       "learning_rate": 5.544544544544545e-06,
-      "loss": 1.0215,
+      "loss": 1.0098,
       "step": 44510
     },
     {
       "epoch": 133.69,
-      "grad_norm": 5.109631061553955,
+      "grad_norm": 6.498576641082764,
       "learning_rate": 5.543543543543544e-06,
-      "loss": 0.9947,
+      "loss": 0.9949,
       "step": 44520
     },
     {
       "epoch": 133.72,
-      "grad_norm": 5.967331409454346,
+      "grad_norm": 7.485474109649658,
       "learning_rate": 5.542542542542543e-06,
-      "loss": 1.0416,
+      "loss": 1.037,
       "step": 44530
     },
     {
       "epoch": 133.75,
-      "grad_norm": 9.050482749938965,
+      "grad_norm": 7.196428298950195,
       "learning_rate": 5.541541541541542e-06,
-      "loss": 1.0139,
+      "loss": 1.0081,
       "step": 44540
     },
     {
       "epoch": 133.78,
-      "grad_norm": 5.472394943237305,
+      "grad_norm": 8.97492790222168,
       "learning_rate": 5.540540540540541e-06,
-      "loss": 1.0024,
+      "loss": 1.0043,
       "step": 44550
     },
     {
       "epoch": 133.81,
-      "grad_norm": 5.958432674407959,
+      "grad_norm": 5.397035598754883,
       "learning_rate": 5.53953953953954e-06,
-      "loss": 1.1607,
+      "loss": 1.1514,
       "step": 44560
     },
     {
       "epoch": 133.84,
-      "grad_norm": 5.930016040802002,
+      "grad_norm": 9.872901916503906,
       "learning_rate": 5.5385385385385396e-06,
-      "loss": 0.9489,
+      "loss": 0.9485,
       "step": 44570
     },
     {
       "epoch": 133.87,
-      "grad_norm": 7.6777544021606445,
+      "grad_norm": 5.9347944259643555,
       "learning_rate": 5.537537537537538e-06,
-      "loss": 1.073,
+      "loss": 1.0647,
       "step": 44580
     },
     {
       "epoch": 133.9,
-      "grad_norm": 6.499703407287598,
+      "grad_norm": 7.159059047698975,
       "learning_rate": 5.536536536536537e-06,
       "loss": 1.0353,
       "step": 44590
     },
     {
       "epoch": 133.93,
-      "grad_norm": 4.94793701171875,
+      "grad_norm": 5.291772365570068,
       "learning_rate": 5.535535535535535e-06,
-      "loss": 1.0752,
+      "loss": 1.0713,
       "step": 44600
     },
     {
       "epoch": 133.96,
-      "grad_norm": 7.174930095672607,
+      "grad_norm": 6.095190525054932,
       "learning_rate": 5.534534534534535e-06,
-      "loss": 1.029,
+      "loss": 1.0305,
       "step": 44610
     },
     {
       "epoch": 133.99,
-      "grad_norm": 7.512166976928711,
+      "grad_norm": 7.839895248413086,
       "learning_rate": 5.533533533533534e-06,
-      "loss": 1.0532,
+      "loss": 1.0543,
       "step": 44620
     },
     {
       "epoch": 134.0,
-      "eval_accuracy": 0.8121,
-      "eval_loss": 0.6124274730682373,
-      "eval_runtime": 5.4986,
-      "eval_samples_per_second": 1818.652,
-      "eval_steps_per_second": 7.275,
+      "eval_accuracy": 0.8123,
+      "eval_loss": 0.6118724346160889,
+      "eval_runtime": 5.5354,
+      "eval_samples_per_second": 1806.568,
+      "eval_steps_per_second": 7.226,
       "step": 44622
     },
     {
       "epoch": 134.02,
-      "grad_norm": 9.750910758972168,
+      "grad_norm": 9.228999137878418,
       "learning_rate": 5.532532532532533e-06,
-      "loss": 1.4189,
+      "loss": 1.4292,
       "step": 44630
     },
     {
       "epoch": 134.05,
-      "grad_norm": 8.481996536254883,
+      "grad_norm": 6.97711706161499,
       "learning_rate": 5.531531531531532e-06,
-      "loss": 1.0432,
+      "loss": 1.0403,
       "step": 44640
     },
     {
       "epoch": 134.08,
-      "grad_norm": 6.531315326690674,
+      "grad_norm": 8.336518287658691,
       "learning_rate": 5.530530530530531e-06,
-      "loss": 1.0527,
+      "loss": 1.0565,
       "step": 44650
     },
     {
       "epoch": 134.11,
-      "grad_norm": 6.288015365600586,
+      "grad_norm": 7.517739295959473,
       "learning_rate": 5.52952952952953e-06,
-      "loss": 0.9965,
+      "loss": 0.9897,
       "step": 44660
     },
     {
       "epoch": 134.14,
-      "grad_norm": 7.408775329589844,
+      "grad_norm": 7.145333766937256,
       "learning_rate": 5.52852852852853e-06,
-      "loss": 1.0777,
+      "loss": 1.0753,
       "step": 44670
     },
     {
       "epoch": 134.17,
-      "grad_norm": 6.120229244232178,
+      "grad_norm": 7.618991851806641,
       "learning_rate": 5.5275275275275275e-06,
-      "loss": 1.0716,
+      "loss": 1.0655,
       "step": 44680
     },
     {
       "epoch": 134.2,
-      "grad_norm": 6.15062952041626,
+      "grad_norm": 8.335567474365234,
       "learning_rate": 5.526526526526527e-06,
-      "loss": 1.0632,
+      "loss": 1.0559,
       "step": 44690
     },
     {
       "epoch": 134.23,
-      "grad_norm": 7.7783732414245605,
+      "grad_norm": 5.6085710525512695,
       "learning_rate": 5.5255255255255255e-06,
-      "loss": 1.0307,
+      "loss": 1.0319,
       "step": 44700
     },
     {
       "epoch": 134.26,
-      "grad_norm": 6.308168888092041,
+      "grad_norm": 8.948322296142578,
       "learning_rate": 5.524524524524525e-06,
-      "loss": 1.0926,
+      "loss": 1.0949,
       "step": 44710
     },
     {
       "epoch": 134.29,
-      "grad_norm": 6.908431053161621,
+      "grad_norm": 7.282476902008057,
       "learning_rate": 5.523523523523524e-06,
-      "loss": 1.058,
+      "loss": 1.0519,
       "step": 44720
     },
     {
       "epoch": 134.32,
-      "grad_norm": 5.881716728210449,
+      "grad_norm": 6.221504211425781,
       "learning_rate": 5.522522522522523e-06,
-      "loss": 1.0999,
+      "loss": 1.0946,
       "step": 44730
     },
     {
       "epoch": 134.35,
-      "grad_norm": 5.842805862426758,
+      "grad_norm": 5.354765892028809,
       "learning_rate": 5.5215215215215224e-06,
-      "loss": 1.0728,
+      "loss": 1.0747,
       "step": 44740
     },
     {
       "epoch": 134.38,
-      "grad_norm": 7.048940181732178,
+      "grad_norm": 7.383026599884033,
       "learning_rate": 5.520520520520521e-06,
-      "loss": 1.0516,
+      "loss": 1.0378,
       "step": 44750
     },
     {
       "epoch": 134.41,
-      "grad_norm": 5.222501754760742,
+      "grad_norm": 6.015492916107178,
       "learning_rate": 5.51951951951952e-06,
-      "loss": 0.989,
+      "loss": 0.9865,
       "step": 44760
     },
     {
       "epoch": 134.44,
-      "grad_norm": 8.270181655883789,
+      "grad_norm": 6.966856479644775,
       "learning_rate": 5.518518518518518e-06,
-      "loss": 1.0461,
+      "loss": 1.0381,
       "step": 44770
     },
     {
       "epoch": 134.47,
-      "grad_norm": 6.993601322174072,
+      "grad_norm": 6.4861159324646,
       "learning_rate": 5.517517517517518e-06,
-      "loss": 0.9925,
+      "loss": 0.9916,
       "step": 44780
     },
     {
       "epoch": 134.5,
-      "grad_norm": 6.914500713348389,
+      "grad_norm": 7.2377448081970215,
       "learning_rate": 5.516516516516517e-06,
-      "loss": 1.0897,
+      "loss": 1.0868,
       "step": 44790
     },
     {
       "epoch": 134.53,
-      "grad_norm": 6.980650901794434,
+      "grad_norm": 12.660313606262207,
       "learning_rate": 5.515515515515516e-06,
-      "loss": 1.1085,
+      "loss": 1.1177,
       "step": 44800
     },
     {
       "epoch": 134.56,
-      "grad_norm": 5.293242454528809,
+      "grad_norm": 5.399194717407227,
       "learning_rate": 5.514514514514515e-06,
-      "loss": 1.0316,
+      "loss": 1.024,
       "step": 44810
     },
     {
       "epoch": 134.59,
-      "grad_norm": 6.202395439147949,
+      "grad_norm": 7.218954086303711,
       "learning_rate": 5.513513513513515e-06,
-      "loss": 0.9982,
+      "loss": 1.0004,
       "step": 44820
     },
     {
       "epoch": 134.62,
-      "grad_norm": 8.26404094696045,
+      "grad_norm": 7.579561710357666,
       "learning_rate": 5.512512512512513e-06,
-      "loss": 1.0323,
+      "loss": 1.0357,
       "step": 44830
     },
     {
       "epoch": 134.65,
-      "grad_norm": 6.488289833068848,
+      "grad_norm": 8.804637908935547,
       "learning_rate": 5.511511511511512e-06,
-      "loss": 1.0775,
+      "loss": 1.0779,
       "step": 44840
     },
     {
       "epoch": 134.68,
-      "grad_norm": 6.1371870040893555,
+      "grad_norm": 6.660343170166016,
       "learning_rate": 5.51051051051051e-06,
-      "loss": 1.0248,
+      "loss": 1.0235,
       "step": 44850
     },
     {
       "epoch": 134.71,
-      "grad_norm": 5.551422595977783,
+      "grad_norm": 6.467170238494873,
       "learning_rate": 5.50950950950951e-06,
-      "loss": 0.9941,
+      "loss": 0.9855,
       "step": 44860
     },
     {
       "epoch": 134.74,
-      "grad_norm": 6.472884178161621,
+      "grad_norm": 6.5093674659729,
       "learning_rate": 5.508508508508508e-06,
-      "loss": 1.0639,
+      "loss": 1.0589,
       "step": 44870
     },
     {
       "epoch": 134.77,
-      "grad_norm": 6.869942665100098,
+      "grad_norm": 7.442628860473633,
       "learning_rate": 5.507507507507508e-06,
-      "loss": 1.0306,
+      "loss": 1.0242,
       "step": 44880
     },
     {
       "epoch": 134.8,
-      "grad_norm": 7.4356770515441895,
+      "grad_norm": 8.709906578063965,
       "learning_rate": 5.506506506506507e-06,
-      "loss": 1.0234,
+      "loss": 1.0249,
       "step": 44890
     },
     {
       "epoch": 134.83,
-      "grad_norm": 5.522848606109619,
+      "grad_norm": 6.170648097991943,
       "learning_rate": 5.505505505505506e-06,
-      "loss": 1.1289,
+      "loss": 1.1208,
       "step": 44900
     },
     {
       "epoch": 134.86,
-      "grad_norm": 6.083286762237549,
+      "grad_norm": 7.419956684112549,
       "learning_rate": 5.504504504504505e-06,
-      "loss": 1.0087,
+      "loss": 1.0038,
       "step": 44910
     },
     {
       "epoch": 134.89,
-      "grad_norm": 8.671223640441895,
+      "grad_norm": 7.882572174072266,
       "learning_rate": 5.503503503503505e-06,
-      "loss": 0.9911,
+      "loss": 0.9991,
       "step": 44920
     },
     {
       "epoch": 134.92,
-      "grad_norm": 8.060920715332031,
+      "grad_norm": 9.263543128967285,
       "learning_rate": 5.5025025025025025e-06,
-      "loss": 1.0487,
+      "loss": 1.056,
       "step": 44930
     },
     {
       "epoch": 134.95,
-      "grad_norm": 7.096294403076172,
+      "grad_norm": 11.132171630859375,
       "learning_rate": 5.501501501501502e-06,
-      "loss": 1.1222,
+      "loss": 1.1193,
       "step": 44940
     },
     {
       "epoch": 134.98,
-      "grad_norm": 5.774645805358887,
+      "grad_norm": 8.518119812011719,
       "learning_rate": 5.5005005005005006e-06,
-      "loss": 1.0163,
+      "loss": 1.0173,
       "step": 44950
     },
     {
       "epoch": 135.0,
-      "eval_accuracy": 0.8123,
-      "eval_loss": 0.6135919690132141,
-      "eval_runtime": 5.5156,
-      "eval_samples_per_second": 1813.028,
-      "eval_steps_per_second": 7.252,
+      "eval_accuracy": 0.8118,
+      "eval_loss": 0.6131147742271423,
+      "eval_runtime": 5.2345,
+      "eval_samples_per_second": 1910.387,
+      "eval_steps_per_second": 7.642,
       "step": 44955
     },
     {
       "epoch": 135.02,
-      "grad_norm": 5.397398471832275,
+      "grad_norm": 4.895552635192871,
       "learning_rate": 5.4994994994995e-06,
-      "loss": 1.132,
+      "loss": 1.1277,
       "step": 44960
     },
     {
       "epoch": 135.05,
-      "grad_norm": 5.9335737228393555,
+      "grad_norm": 5.410463809967041,
       "learning_rate": 5.4984984984984994e-06,
-      "loss": 1.0662,
+      "loss": 1.0576,
       "step": 44970
     },
     {
       "epoch": 135.08,
-      "grad_norm": 5.7754621505737305,
+      "grad_norm": 7.407782554626465,
       "learning_rate": 5.497497497497498e-06,
-      "loss": 1.041,
+      "loss": 1.0381,
       "step": 44980
     },
     {
       "epoch": 135.11,
-      "grad_norm": 5.607287883758545,
+      "grad_norm": 6.824173927307129,
       "learning_rate": 5.4964964964964975e-06,
-      "loss": 1.0447,
+      "loss": 1.0548,
       "step": 44990
     },
     {
       "epoch": 135.14,
-      "grad_norm": 6.7279372215271,
+      "grad_norm": 6.298985481262207,
       "learning_rate": 5.495495495495496e-06,
-      "loss": 1.0539,
+      "loss": 1.0626,
       "step": 45000
     },
     {
       "epoch": 135.17,
-      "grad_norm": 6.82113790512085,
+      "grad_norm": 6.916965961456299,
       "learning_rate": 5.494494494494495e-06,
-      "loss": 1.0661,
+      "loss": 1.0711,
       "step": 45010
     },
     {
       "epoch": 135.2,
-      "grad_norm": 7.874472618103027,
+      "grad_norm": 6.5690531730651855,
       "learning_rate": 5.493493493493493e-06,
-      "loss": 1.0748,
+      "loss": 1.0676,
       "step": 45020
     },
     {
       "epoch": 135.23,
-      "grad_norm": 6.4499945640563965,
+      "grad_norm": 6.567800045013428,
       "learning_rate": 5.492492492492493e-06,
-      "loss": 1.0274,
+      "loss": 1.0119,
       "step": 45030
     },
     {
       "epoch": 135.26,
-      "grad_norm": 6.416584491729736,
+      "grad_norm": 11.598808288574219,
       "learning_rate": 5.491491491491492e-06,
-      "loss": 1.1236,
+      "loss": 1.1295,
       "step": 45040
     },
     {
       "epoch": 135.29,
-      "grad_norm": 7.140875339508057,
+      "grad_norm": 6.194224834442139,
       "learning_rate": 5.490490490490491e-06,
-      "loss": 1.0742,
+      "loss": 1.0599,
       "step": 45050
     },
     {
       "epoch": 135.32,
-      "grad_norm": 5.652576923370361,
+      "grad_norm": 8.83958625793457,
       "learning_rate": 5.48948948948949e-06,
-      "loss": 1.0858,
+      "loss": 1.0841,
       "step": 45060
     },
     {
       "epoch": 135.35,
-      "grad_norm": 6.095656394958496,
+      "grad_norm": 7.79110050201416,
       "learning_rate": 5.48848848848849e-06,
-      "loss": 1.0107,
+      "loss": 1.0072,
       "step": 45070
     },
     {
       "epoch": 135.38,
-      "grad_norm": 6.217310905456543,
+      "grad_norm": 5.504159927368164,
       "learning_rate": 5.487487487487488e-06,
-      "loss": 0.9738,
+      "loss": 0.9655,
       "step": 45080
     },
     {
       "epoch": 135.41,
-      "grad_norm": 8.13300609588623,
+      "grad_norm": 6.208398342132568,
       "learning_rate": 5.486486486486487e-06,
-      "loss": 1.0061,
+      "loss": 1.0105,
       "step": 45090
     },
     {
       "epoch": 135.44,
-      "grad_norm": 6.74412202835083,
+      "grad_norm": 8.247613906860352,
       "learning_rate": 5.485485485485485e-06,
-      "loss": 1.0597,
+      "loss": 1.0471,
       "step": 45100
     },
     {
       "epoch": 135.47,
-      "grad_norm": 7.197597026824951,
+      "grad_norm": 6.370333194732666,
       "learning_rate": 5.484484484484485e-06,
-      "loss": 1.047,
+      "loss": 1.0559,
       "step": 45110
     },
     {
       "epoch": 135.5,
-      "grad_norm": 6.265678882598877,
+      "grad_norm": 5.392190456390381,
       "learning_rate": 5.4834834834834834e-06,
-      "loss": 0.9478,
+      "loss": 0.9443,
       "step": 45120
     },
     {
       "epoch": 135.53,
-      "grad_norm": 10.21109676361084,
+      "grad_norm": 6.285122394561768,
       "learning_rate": 5.482482482482483e-06,
-      "loss": 1.0634,
+      "loss": 1.0601,
       "step": 45130
     },
     {
       "epoch": 135.56,
-      "grad_norm": 7.9858527183532715,
+      "grad_norm": 9.601500511169434,
       "learning_rate": 5.481481481481482e-06,
-      "loss": 1.0616,
+      "loss": 1.0528,
       "step": 45140
     },
     {
       "epoch": 135.59,
-      "grad_norm": 5.83823299407959,
+      "grad_norm": 7.000296592712402,
       "learning_rate": 5.480480480480481e-06,
-      "loss": 0.9759,
+      "loss": 0.9663,
       "step": 45150
     },
     {
       "epoch": 135.62,
-      "grad_norm": 8.392951965332031,
+      "grad_norm": 8.02866268157959,
       "learning_rate": 5.47947947947948e-06,
-      "loss": 0.9778,
+      "loss": 0.9883,
       "step": 45160
     },
     {
       "epoch": 135.65,
-      "grad_norm": 5.648130416870117,
+      "grad_norm": 6.011983394622803,
       "learning_rate": 5.47847847847848e-06,
       "loss": 1.0169,
       "step": 45170
     },
     {
       "epoch": 135.68,
-      "grad_norm": 4.904910087585449,
+      "grad_norm": 5.964869976043701,
       "learning_rate": 5.4774774774774776e-06,
-      "loss": 1.0166,
+      "loss": 1.0188,
       "step": 45180
     },
     {
       "epoch": 135.71,
-      "grad_norm": 4.975961208343506,
+      "grad_norm": 6.078293800354004,
       "learning_rate": 5.476476476476477e-06,
-      "loss": 1.0341,
+      "loss": 1.0396,
       "step": 45190
     },
     {
       "epoch": 135.74,
-      "grad_norm": 5.25714111328125,
+      "grad_norm": 6.6797709465026855,
       "learning_rate": 5.475475475475476e-06,
-      "loss": 0.9841,
+      "loss": 0.994,
       "step": 45200
     },
     {
       "epoch": 135.77,
-      "grad_norm": 5.1715497970581055,
+      "grad_norm": 4.8800048828125,
       "learning_rate": 5.474474474474475e-06,
-      "loss": 1.0648,
+      "loss": 1.062,
       "step": 45210
     },
     {
       "epoch": 135.8,
-      "grad_norm": 7.5335283279418945,
+      "grad_norm": 9.307746887207031,
       "learning_rate": 5.473473473473474e-06,
-      "loss": 1.0918,
+      "loss": 1.0911,
       "step": 45220
     },
     {
       "epoch": 135.83,
-      "grad_norm": 6.073637008666992,
+      "grad_norm": 9.131691932678223,
       "learning_rate": 5.472472472472473e-06,
-      "loss": 1.0567,
+      "loss": 1.0537,
       "step": 45230
     },
     {
       "epoch": 135.86,
-      "grad_norm": 7.734210968017578,
+      "grad_norm": 9.97434139251709,
       "learning_rate": 5.4714714714714725e-06,
-      "loss": 0.9987,
+      "loss": 0.9966,
       "step": 45240
     },
     {
       "epoch": 135.89,
-      "grad_norm": 7.106470584869385,
+      "grad_norm": 6.528149127960205,
       "learning_rate": 5.470470470470471e-06,
-      "loss": 0.9985,
+      "loss": 1.0025,
       "step": 45250
     },
     {
       "epoch": 135.92,
-      "grad_norm": 7.990725994110107,
+      "grad_norm": 7.077942371368408,
       "learning_rate": 5.46946946946947e-06,
-      "loss": 1.1284,
+      "loss": 1.1266,
       "step": 45260
     },
     {
       "epoch": 135.95,
-      "grad_norm": 5.897083282470703,
+      "grad_norm": 6.076117992401123,
       "learning_rate": 5.468468468468468e-06,
-      "loss": 0.9762,
+      "loss": 0.9759,
       "step": 45270
     },
     {
       "epoch": 135.98,
-      "grad_norm": 6.013416767120361,
+      "grad_norm": 5.382653713226318,
       "learning_rate": 5.467467467467468e-06,
-      "loss": 1.0789,
+      "loss": 1.0912,
       "step": 45280
     },
     {
       "epoch": 136.0,
-      "eval_accuracy": 0.8122,
-      "eval_loss": 0.6143561601638794,
-      "eval_runtime": 6.1448,
-      "eval_samples_per_second": 1627.387,
-      "eval_steps_per_second": 6.51,
+      "eval_accuracy": 0.8126,
+      "eval_loss": 0.6154760718345642,
+      "eval_runtime": 5.4601,
+      "eval_samples_per_second": 1831.468,
+      "eval_steps_per_second": 7.326,
       "step": 45288
     },
     {
       "epoch": 136.01,
-      "grad_norm": 5.48843240737915,
+      "grad_norm": 7.657252311706543,
       "learning_rate": 5.466466466466467e-06,
-      "loss": 1.1116,
+      "loss": 1.1142,
       "step": 45290
     },
     {
       "epoch": 136.04,
-      "grad_norm": 6.248884677886963,
+      "grad_norm": 6.564085960388184,
       "learning_rate": 5.465465465465466e-06,
-      "loss": 1.0237,
+      "loss": 1.0172,
       "step": 45300
     },
     {
       "epoch": 136.07,
-      "grad_norm": 6.240085601806641,
+      "grad_norm": 7.352023601531982,
       "learning_rate": 5.464464464464465e-06,
-      "loss": 1.051,
+      "loss": 1.0518,
       "step": 45310
     },
     {
       "epoch": 136.1,
-      "grad_norm": 8.700919151306152,
+      "grad_norm": 5.713761329650879,
       "learning_rate": 5.463463463463464e-06,
-      "loss": 1.0163,
+      "loss": 1.0118,
       "step": 45320
     },
     {
       "epoch": 136.13,
-      "grad_norm": 4.644911766052246,
+      "grad_norm": 5.1132097244262695,
       "learning_rate": 5.462462462462463e-06,
-      "loss": 1.0418,
+      "loss": 1.0171,
       "step": 45330
     },
     {
       "epoch": 136.16,
-      "grad_norm": 6.026497840881348,
+      "grad_norm": 9.733572006225586,
       "learning_rate": 5.461461461461461e-06,
-      "loss": 1.0891,
+      "loss": 1.0855,
       "step": 45340
     },
     {
       "epoch": 136.19,
-      "grad_norm": 5.813206195831299,
+      "grad_norm": 5.927685260772705,
       "learning_rate": 5.4604604604604604e-06,
-      "loss": 1.053,
+      "loss": 1.0491,
       "step": 45350
     },
     {
       "epoch": 136.22,
-      "grad_norm": 6.0458455085754395,
+      "grad_norm": 6.215799331665039,
       "learning_rate": 5.45945945945946e-06,
-      "loss": 1.0421,
+      "loss": 1.0317,
       "step": 45360
     },
     {
       "epoch": 136.25,
-      "grad_norm": 7.615644454956055,
+      "grad_norm": 7.620630264282227,
       "learning_rate": 5.4584584584584585e-06,
-      "loss": 1.0634,
+      "loss": 1.0595,
       "step": 45370
     },
     {
       "epoch": 136.28,
-      "grad_norm": 6.42539644241333,
+      "grad_norm": 6.822344779968262,
       "learning_rate": 5.457457457457458e-06,
-      "loss": 1.0254,
+      "loss": 1.018,
       "step": 45380
     },
     {
       "epoch": 136.31,
-      "grad_norm": 6.430290699005127,
+      "grad_norm": 7.420136451721191,
       "learning_rate": 5.456456456456457e-06,
-      "loss": 0.9967,
+      "loss": 0.9872,
       "step": 45390
     },
     {
       "epoch": 136.34,
-      "grad_norm": 9.434653282165527,
+      "grad_norm": 11.498109817504883,
       "learning_rate": 5.455455455455456e-06,
-      "loss": 1.0553,
+      "loss": 1.0624,
       "step": 45400
     },
     {
       "epoch": 136.37,
-      "grad_norm": 6.9574809074401855,
+      "grad_norm": 6.123563766479492,
       "learning_rate": 5.454454454454455e-06,
-      "loss": 1.0578,
+      "loss": 1.0567,
       "step": 45410
     },
     {
       "epoch": 136.4,
-      "grad_norm": 6.14061164855957,
+      "grad_norm": 6.902637004852295,
       "learning_rate": 5.453453453453455e-06,
-      "loss": 1.0989,
+      "loss": 1.1104,
       "step": 45420
     },
     {
       "epoch": 136.43,
-      "grad_norm": 7.463507652282715,
+      "grad_norm": 8.046177864074707,
       "learning_rate": 5.452452452452453e-06,
-      "loss": 1.1182,
+      "loss": 1.1116,
       "step": 45430
     },
     {
       "epoch": 136.46,
-      "grad_norm": 6.009886264801025,
+      "grad_norm": 7.216541290283203,
       "learning_rate": 5.451451451451451e-06,
-      "loss": 1.0451,
+      "loss": 1.0498,
       "step": 45440
     },
     {
       "epoch": 136.49,
-      "grad_norm": 5.888139247894287,
+      "grad_norm": 7.717516899108887,
       "learning_rate": 5.450450450450451e-06,
-      "loss": 0.9944,
+      "loss": 0.9879,
       "step": 45450
     },
     {
       "epoch": 136.52,
-      "grad_norm": 5.4764628410339355,
+      "grad_norm": 7.879677772521973,
       "learning_rate": 5.44944944944945e-06,
-      "loss": 1.0316,
+      "loss": 1.0292,
       "step": 45460
     },
     {
       "epoch": 136.55,
-      "grad_norm": 6.060569763183594,
+      "grad_norm": 6.641725540161133,
       "learning_rate": 5.448448448448449e-06,
-      "loss": 1.0147,
+      "loss": 1.0168,
       "step": 45470
     },
     {
       "epoch": 136.58,
-      "grad_norm": 6.2233405113220215,
+      "grad_norm": 7.356192111968994,
       "learning_rate": 5.447447447447448e-06,
-      "loss": 0.9958,
+      "loss": 0.9866,
       "step": 45480
     },
     {
       "epoch": 136.61,
-      "grad_norm": 6.686922073364258,
+      "grad_norm": 7.970301628112793,
       "learning_rate": 5.4464464464464475e-06,
-      "loss": 1.077,
+      "loss": 1.0728,
       "step": 45490
     },
     {
       "epoch": 136.64,
-      "grad_norm": 6.027292251586914,
+      "grad_norm": 6.814348220825195,
       "learning_rate": 5.445445445445446e-06,
-      "loss": 1.0456,
+      "loss": 1.0208,
       "step": 45500
     },
     {
       "epoch": 136.67,
-      "grad_norm": 5.6394734382629395,
+      "grad_norm": 6.99000883102417,
       "learning_rate": 5.444444444444445e-06,
-      "loss": 1.0338,
+      "loss": 1.028,
       "step": 45510
     },
     {
       "epoch": 136.7,
-      "grad_norm": 8.169734954833984,
+      "grad_norm": 7.0701704025268555,
       "learning_rate": 5.443443443443443e-06,
-      "loss": 1.0639,
+      "loss": 1.0704,
       "step": 45520
     },
     {
       "epoch": 136.73,
-      "grad_norm": 6.107832431793213,
+      "grad_norm": 6.110533714294434,
       "learning_rate": 5.442442442442443e-06,
-      "loss": 0.9534,
+      "loss": 0.9627,
       "step": 45530
     },
     {
       "epoch": 136.76,
-      "grad_norm": 7.34189510345459,
+      "grad_norm": 10.164361953735352,
       "learning_rate": 5.441441441441442e-06,
-      "loss": 1.016,
+      "loss": 1.0061,
       "step": 45540
     },
     {
       "epoch": 136.79,
-      "grad_norm": 7.518131256103516,
+      "grad_norm": 6.80831241607666,
       "learning_rate": 5.440440440440441e-06,
-      "loss": 1.13,
+      "loss": 1.1213,
       "step": 45550
     },
     {
       "epoch": 136.82,
-      "grad_norm": 5.873836994171143,
+      "grad_norm": 6.304059028625488,
       "learning_rate": 5.43943943943944e-06,
-      "loss": 1.0446,
+      "loss": 1.0338,
       "step": 45560
     },
     {
       "epoch": 136.85,
-      "grad_norm": 5.250598907470703,
+      "grad_norm": 7.033069133758545,
       "learning_rate": 5.438438438438439e-06,
-      "loss": 1.0143,
+      "loss": 1.0221,
       "step": 45570
     },
     {
       "epoch": 136.88,
-      "grad_norm": 5.365545749664307,
+      "grad_norm": 7.176772594451904,
       "learning_rate": 5.437437437437438e-06,
-      "loss": 1.0915,
+      "loss": 1.079,
       "step": 45580
     },
     {
       "epoch": 136.91,
-      "grad_norm": 8.698531150817871,
+      "grad_norm": 7.062721252441406,
       "learning_rate": 5.436436436436436e-06,
-      "loss": 1.0458,
+      "loss": 1.0471,
       "step": 45590
     },
     {
       "epoch": 136.94,
-      "grad_norm": 8.217132568359375,
+      "grad_norm": 8.421354293823242,
       "learning_rate": 5.4354354354354355e-06,
-      "loss": 1.0656,
+      "loss": 1.0636,
       "step": 45600
     },
     {
       "epoch": 136.97,
-      "grad_norm": 5.501484394073486,
+      "grad_norm": 6.250651836395264,
       "learning_rate": 5.434434434434435e-06,
-      "loss": 1.0491,
+      "loss": 1.0591,
       "step": 45610
     },
     {
       "epoch": 137.0,
-      "grad_norm": 8.191163063049316,
+      "grad_norm": 14.710040092468262,
       "learning_rate": 5.4334334334334335e-06,
-      "loss": 0.9845,
+      "loss": 0.9771,
       "step": 45620
     },
     {
       "epoch": 137.0,
-      "eval_accuracy": 0.8114,
-      "eval_loss": 0.6118127703666687,
-      "eval_runtime": 5.7098,
-      "eval_samples_per_second": 1751.383,
-      "eval_steps_per_second": 7.006,
+      "eval_accuracy": 0.814,
+      "eval_loss": 0.6096869111061096,
+      "eval_runtime": 5.3178,
+      "eval_samples_per_second": 1880.466,
+      "eval_steps_per_second": 7.522,
       "step": 45621
     },
     {
       "epoch": 137.03,
-      "grad_norm": 7.515133380889893,
+      "grad_norm": 7.415700912475586,
       "learning_rate": 5.432432432432433e-06,
-      "loss": 1.1384,
+      "loss": 1.1108,
       "step": 45630
     },
     {
       "epoch": 137.06,
-      "grad_norm": 9.66165828704834,
+      "grad_norm": 5.472467422485352,
       "learning_rate": 5.431431431431432e-06,
-      "loss": 1.0455,
+      "loss": 1.0435,
       "step": 45640
     },
     {
       "epoch": 137.09,
-      "grad_norm": 7.029970169067383,
+      "grad_norm": 7.742816925048828,
       "learning_rate": 5.430430430430431e-06,
-      "loss": 1.0563,
+      "loss": 1.0503,
       "step": 45650
     },
     {
       "epoch": 137.12,
-      "grad_norm": 8.438450813293457,
+      "grad_norm": 8.076441764831543,
       "learning_rate": 5.4294294294294304e-06,
-      "loss": 1.0681,
+      "loss": 1.0595,
       "step": 45660
     },
     {
       "epoch": 137.15,
-      "grad_norm": 5.718472003936768,
+      "grad_norm": 5.881075382232666,
       "learning_rate": 5.428428428428428e-06,
-      "loss": 1.0405,
+      "loss": 1.0447,
       "step": 45670
     },
     {
       "epoch": 137.18,
-      "grad_norm": 7.022383689880371,
+      "grad_norm": 8.063398361206055,
       "learning_rate": 5.427427427427428e-06,
-      "loss": 1.0231,
+      "loss": 1.0263,
       "step": 45680
     },
     {
       "epoch": 137.21,
-      "grad_norm": 6.5866379737854,
+      "grad_norm": 8.14753246307373,
       "learning_rate": 5.426426426426426e-06,
-      "loss": 0.9855,
+      "loss": 0.9662,
       "step": 45690
     },
     {
       "epoch": 137.24,
-      "grad_norm": 7.87273645401001,
+      "grad_norm": 11.000223159790039,
       "learning_rate": 5.425425425425426e-06,
-      "loss": 1.0337,
+      "loss": 1.0369,
       "step": 45700
     },
     {
       "epoch": 137.27,
-      "grad_norm": 8.241292953491211,
+      "grad_norm": 13.694694519042969,
       "learning_rate": 5.424424424424425e-06,
-      "loss": 1.0296,
+      "loss": 1.0401,
       "step": 45710
     },
     {
       "epoch": 137.3,
-      "grad_norm": 6.533015251159668,
+      "grad_norm": 9.611083984375,
       "learning_rate": 5.423423423423424e-06,
-      "loss": 1.0645,
+      "loss": 1.0621,
       "step": 45720
     },
     {
       "epoch": 137.33,
-      "grad_norm": 5.183393955230713,
+      "grad_norm": 7.5067667961120605,
       "learning_rate": 5.422422422422423e-06,
-      "loss": 0.9479,
+      "loss": 0.9412,
       "step": 45730
     },
     {
       "epoch": 137.36,
-      "grad_norm": 6.907729625701904,
+      "grad_norm": 7.999253749847412,
       "learning_rate": 5.421421421421423e-06,
-      "loss": 1.0651,
+      "loss": 1.0508,
       "step": 45740
     },
     {
       "epoch": 137.39,
-      "grad_norm": 5.98319149017334,
+      "grad_norm": 5.4400715827941895,
       "learning_rate": 5.420420420420421e-06,
-      "loss": 1.0338,
+      "loss": 1.0185,
       "step": 45750
     },
     {
       "epoch": 137.42,
-      "grad_norm": 5.3992018699646,
+      "grad_norm": 4.913991451263428,
       "learning_rate": 5.41941941941942e-06,
-      "loss": 0.978,
+      "loss": 0.9762,
       "step": 45760
     },
     {
       "epoch": 137.45,
-      "grad_norm": 4.213364124298096,
+      "grad_norm": 5.470534324645996,
       "learning_rate": 5.418418418418418e-06,
-      "loss": 1.0389,
+      "loss": 1.0363,
       "step": 45770
     },
     {
       "epoch": 137.48,
-      "grad_norm": 7.779664039611816,
+      "grad_norm": 7.391083717346191,
       "learning_rate": 5.417417417417418e-06,
-      "loss": 1.0072,
+      "loss": 1.0057,
       "step": 45780
     },
     {
       "epoch": 137.51,
-      "grad_norm": 7.867021083831787,
+      "grad_norm": 6.832329750061035,
       "learning_rate": 5.416416416416416e-06,
-      "loss": 0.9971,
+      "loss": 0.9949,
       "step": 45790
     },
     {
       "epoch": 137.54,
-      "grad_norm": 7.2546186447143555,
+      "grad_norm": 6.791614055633545,
       "learning_rate": 5.415415415415416e-06,
-      "loss": 1.0763,
+      "loss": 1.0845,
       "step": 45800
     },
     {
       "epoch": 137.57,
-      "grad_norm": 6.693117141723633,
+      "grad_norm": 6.3377227783203125,
       "learning_rate": 5.414414414414415e-06,
-      "loss": 1.0458,
+      "loss": 1.0456,
       "step": 45810
     },
     {
       "epoch": 137.6,
-      "grad_norm": 7.490116596221924,
+      "grad_norm": 7.326231002807617,
       "learning_rate": 5.413413413413414e-06,
-      "loss": 1.022,
+      "loss": 1.0293,
       "step": 45820
     },
     {
       "epoch": 137.63,
-      "grad_norm": 7.541634559631348,
+      "grad_norm": 7.044851779937744,
       "learning_rate": 5.412412412412413e-06,
-      "loss": 1.0402,
+      "loss": 1.0485,
       "step": 45830
     },
     {
       "epoch": 137.66,
-      "grad_norm": 5.8742570877075195,
+      "grad_norm": 7.944100379943848,
       "learning_rate": 5.411411411411411e-06,
-      "loss": 1.0183,
+      "loss": 1.0103,
       "step": 45840
     },
     {
       "epoch": 137.69,
-      "grad_norm": 4.844747543334961,
+      "grad_norm": 6.653429985046387,
       "learning_rate": 5.4104104104104105e-06,
-      "loss": 0.9976,
+      "loss": 0.9988,
       "step": 45850
     },
     {
       "epoch": 137.72,
-      "grad_norm": 5.483329772949219,
+      "grad_norm": 5.542562484741211,
       "learning_rate": 5.40940940940941e-06,
-      "loss": 0.9596,
+      "loss": 0.9611,
       "step": 45860
     },
     {
       "epoch": 137.75,
-      "grad_norm": 5.9769439697265625,
+      "grad_norm": 7.229701995849609,
       "learning_rate": 5.4084084084084086e-06,
-      "loss": 1.0706,
+      "loss": 1.0805,
       "step": 45870
     },
     {
       "epoch": 137.78,
-      "grad_norm": 6.814098358154297,
+      "grad_norm": 7.4158854484558105,
       "learning_rate": 5.407407407407408e-06,
-      "loss": 1.0669,
+      "loss": 1.0668,
       "step": 45880
     },
     {
       "epoch": 137.81,
-      "grad_norm": 6.862179756164551,
+      "grad_norm": 7.905383586883545,
       "learning_rate": 5.4064064064064074e-06,
-      "loss": 1.0427,
+      "loss": 1.0382,
       "step": 45890
     },
     {
       "epoch": 137.84,
-      "grad_norm": 5.169375419616699,
+      "grad_norm": 6.2236223220825195,
       "learning_rate": 5.405405405405406e-06,
-      "loss": 1.0007,
+      "loss": 0.9971,
       "step": 45900
     },
     {
       "epoch": 137.87,
-      "grad_norm": 6.91594123840332,
+      "grad_norm": 8.042767524719238,
       "learning_rate": 5.4044044044044055e-06,
-      "loss": 1.0415,
+      "loss": 1.0441,
       "step": 45910
     },
     {
       "epoch": 137.9,
-      "grad_norm": 7.279690742492676,
+      "grad_norm": 8.51661491394043,
       "learning_rate": 5.403403403403403e-06,
-      "loss": 1.0231,
+      "loss": 1.027,
       "step": 45920
     },
     {
       "epoch": 137.93,
-      "grad_norm": 5.865286350250244,
+      "grad_norm": 6.366793632507324,
       "learning_rate": 5.402402402402403e-06,
-      "loss": 1.029,
+      "loss": 1.0331,
       "step": 45930
     },
     {
       "epoch": 137.96,
-      "grad_norm": 8.761519432067871,
+      "grad_norm": 9.938056945800781,
       "learning_rate": 5.401401401401401e-06,
-      "loss": 1.1202,
+      "loss": 1.1137,
       "step": 45940
     },
     {
       "epoch": 137.99,
-      "grad_norm": 5.382352352142334,
+      "grad_norm": 5.801309108734131,
       "learning_rate": 5.400400400400401e-06,
-      "loss": 1.0238,
+      "loss": 1.0431,
       "step": 45950
     },
     {
       "epoch": 138.0,
-      "eval_accuracy": 0.8123,
-      "eval_loss": 0.6073533892631531,
-      "eval_runtime": 5.5919,
-      "eval_samples_per_second": 1788.293,
-      "eval_steps_per_second": 7.153,
+      "eval_accuracy": 0.8159,
+      "eval_loss": 0.6076262593269348,
+      "eval_runtime": 5.3778,
+      "eval_samples_per_second": 1859.504,
+      "eval_steps_per_second": 7.438,
       "step": 45954
     },
     {
       "epoch": 138.02,
-      "grad_norm": 7.149085998535156,
+      "grad_norm": 6.1771063804626465,
       "learning_rate": 5.3993993993994e-06,
-      "loss": 1.1092,
+      "loss": 1.093,
       "step": 45960
     },
     {
       "epoch": 138.05,
-      "grad_norm": 7.283044338226318,
+      "grad_norm": 8.564148902893066,
       "learning_rate": 5.398398398398399e-06,
-      "loss": 1.1022,
+      "loss": 1.1015,
       "step": 45970
     },
     {
       "epoch": 138.08,
-      "grad_norm": 6.518016338348389,
+      "grad_norm": 7.4269185066223145,
       "learning_rate": 5.397397397397398e-06,
-      "loss": 1.0831,
+      "loss": 1.0838,
       "step": 45980
     },
     {
       "epoch": 138.11,
-      "grad_norm": 6.636559009552002,
+      "grad_norm": 6.934932708740234,
       "learning_rate": 5.396396396396398e-06,
-      "loss": 1.0622,
+      "loss": 1.0636,
       "step": 45990
     },
     {
       "epoch": 138.14,
-      "grad_norm": 5.118414878845215,
+      "grad_norm": 6.238371849060059,
       "learning_rate": 5.395395395395396e-06,
-      "loss": 1.0726,
+      "loss": 1.0834,
       "step": 46000
     },
     {
       "epoch": 138.17,
-      "grad_norm": 7.224179744720459,
+      "grad_norm": 9.000445365905762,
       "learning_rate": 5.394394394394394e-06,
-      "loss": 1.0063,
+      "loss": 1.0048,
       "step": 46010
     },
     {
       "epoch": 138.2,
-      "grad_norm": 8.51608657836914,
+      "grad_norm": 10.350955963134766,
       "learning_rate": 5.393393393393393e-06,
-      "loss": 1.0823,
+      "loss": 1.0716,
       "step": 46020
     },
     {
       "epoch": 138.23,
-      "grad_norm": 8.41434097290039,
+      "grad_norm": 9.764566421508789,
       "learning_rate": 5.392392392392393e-06,
-      "loss": 1.0117,
+      "loss": 1.0152,
       "step": 46030
     },
     {
       "epoch": 138.26,
-      "grad_norm": 6.474865913391113,
+      "grad_norm": 6.267009258270264,
       "learning_rate": 5.3913913913913914e-06,
-      "loss": 1.0746,
+      "loss": 1.0707,
       "step": 46040
     },
     {
       "epoch": 138.29,
-      "grad_norm": 6.239567756652832,
+      "grad_norm": 7.222299098968506,
       "learning_rate": 5.390390390390391e-06,
-      "loss": 1.0154,
+      "loss": 1.0159,
       "step": 46050
     },
     {
       "epoch": 138.32,
-      "grad_norm": 7.245998382568359,
+      "grad_norm": 6.544965744018555,
       "learning_rate": 5.38938938938939e-06,
-      "loss": 0.9955,
+      "loss": 0.9991,
       "step": 46060
     },
     {
       "epoch": 138.35,
-      "grad_norm": 6.7455596923828125,
+      "grad_norm": 6.403246879577637,
       "learning_rate": 5.388388388388389e-06,
-      "loss": 1.0116,
+      "loss": 1.0124,
       "step": 46070
     },
     {
       "epoch": 138.38,
-      "grad_norm": 6.188246250152588,
+      "grad_norm": 7.409077167510986,
       "learning_rate": 5.387387387387388e-06,
-      "loss": 1.0473,
+      "loss": 1.0367,
       "step": 46080
     },
     {
       "epoch": 138.41,
-      "grad_norm": 7.7095465660095215,
+      "grad_norm": 7.044198513031006,
       "learning_rate": 5.386386386386386e-06,
-      "loss": 1.0095,
+      "loss": 1.0198,
       "step": 46090
     },
     {
       "epoch": 138.44,
-      "grad_norm": 6.084375381469727,
+      "grad_norm": 6.302069187164307,
       "learning_rate": 5.3853853853853856e-06,
-      "loss": 1.0823,
+      "loss": 1.0937,
       "step": 46100
     },
     {
       "epoch": 138.47,
-      "grad_norm": 6.623564720153809,
+      "grad_norm": 11.892349243164062,
       "learning_rate": 5.384384384384385e-06,
-      "loss": 1.0266,
+      "loss": 1.0275,
       "step": 46110
     },
     {
       "epoch": 138.5,
-      "grad_norm": 6.939033508300781,
+      "grad_norm": 5.5102925300598145,
       "learning_rate": 5.383383383383384e-06,
-      "loss": 1.114,
+      "loss": 1.1159,
       "step": 46120
     },
     {
       "epoch": 138.53,
-      "grad_norm": 8.582728385925293,
+      "grad_norm": 6.4264936447143555,
       "learning_rate": 5.382382382382383e-06,
-      "loss": 1.0729,
+      "loss": 1.0757,
       "step": 46130
     },
     {
       "epoch": 138.56,
-      "grad_norm": 5.752279281616211,
+      "grad_norm": 7.969864368438721,
       "learning_rate": 5.381381381381382e-06,
-      "loss": 1.0812,
+      "loss": 1.0844,
       "step": 46140
     },
     {
       "epoch": 138.59,
-      "grad_norm": 6.958952903747559,
+      "grad_norm": 8.998806953430176,
       "learning_rate": 5.380380380380381e-06,
-      "loss": 1.0224,
+      "loss": 1.0068,
       "step": 46150
     },
     {
       "epoch": 138.62,
-      "grad_norm": 7.750551223754883,
+      "grad_norm": 8.953554153442383,
       "learning_rate": 5.3793793793793805e-06,
-      "loss": 0.9667,
+      "loss": 0.9651,
       "step": 46160
     },
     {
       "epoch": 138.65,
-      "grad_norm": 8.475152969360352,
+      "grad_norm": 8.57585334777832,
       "learning_rate": 5.378378378378378e-06,
-      "loss": 1.0314,
+      "loss": 1.0371,
       "step": 46170
     },
     {
       "epoch": 138.68,
-      "grad_norm": 7.299678802490234,
+      "grad_norm": 8.393003463745117,
       "learning_rate": 5.377377377377378e-06,
-      "loss": 1.0175,
+      "loss": 1.016,
       "step": 46180
     },
     {
       "epoch": 138.71,
-      "grad_norm": 5.292054653167725,
+      "grad_norm": 5.126714706420898,
       "learning_rate": 5.376376376376376e-06,
-      "loss": 1.0171,
+      "loss": 1.0172,
       "step": 46190
     },
     {
       "epoch": 138.74,
-      "grad_norm": 5.331376075744629,
+      "grad_norm": 6.141870498657227,
       "learning_rate": 5.375375375375376e-06,
-      "loss": 0.9512,
+      "loss": 0.956,
       "step": 46200
     },
     {
       "epoch": 138.77,
-      "grad_norm": 6.005594730377197,
+      "grad_norm": 9.65499210357666,
       "learning_rate": 5.374374374374375e-06,
-      "loss": 1.0328,
+      "loss": 1.0461,
       "step": 46210
     },
     {
       "epoch": 138.8,
-      "grad_norm": 5.533634185791016,
+      "grad_norm": 6.458467483520508,
       "learning_rate": 5.373373373373374e-06,
-      "loss": 0.9575,
+      "loss": 0.952,
       "step": 46220
     },
     {
       "epoch": 138.83,
-      "grad_norm": 5.983732223510742,
+      "grad_norm": 5.9537835121154785,
       "learning_rate": 5.372372372372373e-06,
-      "loss": 1.048,
+      "loss": 1.0471,
       "step": 46230
     },
     {
       "epoch": 138.86,
-      "grad_norm": 6.825761795043945,
+      "grad_norm": 5.2064032554626465,
       "learning_rate": 5.371371371371372e-06,
-      "loss": 1.0235,
+      "loss": 1.0192,
       "step": 46240
     },
     {
       "epoch": 138.89,
-      "grad_norm": 5.447849750518799,
+      "grad_norm": 5.718149185180664,
       "learning_rate": 5.370370370370371e-06,
-      "loss": 1.0325,
+      "loss": 1.0176,
       "step": 46250
     },
     {
       "epoch": 138.92,
-      "grad_norm": 5.814444065093994,
+      "grad_norm": 6.023651123046875,
       "learning_rate": 5.369369369369369e-06,
-      "loss": 1.1183,
+      "loss": 1.1115,
       "step": 46260
     },
     {
       "epoch": 138.95,
-      "grad_norm": 5.9441447257995605,
+      "grad_norm": 6.892574787139893,
       "learning_rate": 5.3683683683683684e-06,
-      "loss": 1.0509,
+      "loss": 1.0491,
       "step": 46270
     },
     {
       "epoch": 138.98,
-      "grad_norm": 10.669390678405762,
+      "grad_norm": 7.010969638824463,
       "learning_rate": 5.367367367367368e-06,
-      "loss": 1.0287,
+      "loss": 1.0191,
       "step": 46280
     },
     {
       "epoch": 139.0,
-      "eval_accuracy": 0.8135,
-      "eval_loss": 0.6098882555961609,
-      "eval_runtime": 5.5017,
-      "eval_samples_per_second": 1817.628,
-      "eval_steps_per_second": 7.271,
+      "eval_accuracy": 0.8144,
+      "eval_loss": 0.6111952066421509,
+      "eval_runtime": 5.4294,
+      "eval_samples_per_second": 1841.813,
+      "eval_steps_per_second": 7.367,
       "step": 46287
     },
     {
       "epoch": 139.01,
-      "grad_norm": 6.777736663818359,
+      "grad_norm": 6.1447672843933105,
       "learning_rate": 5.3663663663663665e-06,
-      "loss": 1.2663,
+      "loss": 1.2438,
       "step": 46290
     },
     {
       "epoch": 139.04,
-      "grad_norm": 7.840846538543701,
+      "grad_norm": 9.190237998962402,
       "learning_rate": 5.365365365365366e-06,
-      "loss": 1.0431,
+      "loss": 1.0466,
       "step": 46300
     },
     {
       "epoch": 139.07,
-      "grad_norm": 7.074498176574707,
+      "grad_norm": 5.854460716247559,
       "learning_rate": 5.364364364364365e-06,
-      "loss": 1.0655,
+      "loss": 1.0496,
       "step": 46310
     },
     {
       "epoch": 139.1,
-      "grad_norm": 7.154983043670654,
+      "grad_norm": 8.218518257141113,
       "learning_rate": 5.363363363363364e-06,
-      "loss": 1.0373,
+      "loss": 1.0308,
       "step": 46320
     },
     {
       "epoch": 139.13,
-      "grad_norm": 6.1080427169799805,
+      "grad_norm": 9.726752281188965,
       "learning_rate": 5.362362362362363e-06,
-      "loss": 0.9606,
+      "loss": 0.9593,
       "step": 46330
     },
     {
       "epoch": 139.16,
-      "grad_norm": 6.591211318969727,
+      "grad_norm": 7.5220184326171875,
       "learning_rate": 5.361361361361361e-06,
-      "loss": 1.0537,
+      "loss": 1.0522,
       "step": 46340
     },
     {
       "epoch": 139.19,
-      "grad_norm": 8.968505859375,
+      "grad_norm": 8.683764457702637,
       "learning_rate": 5.360360360360361e-06,
-      "loss": 1.0776,
+      "loss": 1.0719,
       "step": 46350
     },
     {
       "epoch": 139.22,
-      "grad_norm": 9.796473503112793,
+      "grad_norm": 10.520126342773438,
       "learning_rate": 5.359359359359359e-06,
-      "loss": 1.0582,
+      "loss": 1.058,
       "step": 46360
     },
     {
       "epoch": 139.25,
-      "grad_norm": 6.7947869300842285,
+      "grad_norm": 9.530558586120605,
       "learning_rate": 5.358358358358359e-06,
-      "loss": 1.0804,
+      "loss": 1.0751,
       "step": 46370
     },
     {
       "epoch": 139.28,
-      "grad_norm": 4.501155853271484,
+      "grad_norm": 6.416819095611572,
       "learning_rate": 5.357357357357358e-06,
-      "loss": 1.0235,
+      "loss": 1.0203,
       "step": 46380
     },
     {
       "epoch": 139.31,
-      "grad_norm": 8.488826751708984,
+      "grad_norm": 10.010324478149414,
       "learning_rate": 5.356356356356357e-06,
-      "loss": 1.0384,
+      "loss": 1.0405,
       "step": 46390
     },
     {
       "epoch": 139.34,
-      "grad_norm": 8.778350830078125,
+      "grad_norm": 7.013326168060303,
       "learning_rate": 5.355355355355356e-06,
-      "loss": 0.9725,
+      "loss": 0.9829,
       "step": 46400
     },
     {
       "epoch": 139.37,
-      "grad_norm": 6.47870397567749,
+      "grad_norm": 5.768508434295654,
       "learning_rate": 5.3543543543543555e-06,
-      "loss": 0.9909,
+      "loss": 0.9993,
       "step": 46410
     },
     {
       "epoch": 139.4,
-      "grad_norm": 8.823840141296387,
+      "grad_norm": 8.460158348083496,
       "learning_rate": 5.353353353353353e-06,
-      "loss": 1.0231,
+      "loss": 1.02,
       "step": 46420
     },
     {
       "epoch": 139.43,
-      "grad_norm": 5.878924369812012,
+      "grad_norm": 6.469432353973389,
       "learning_rate": 5.352352352352353e-06,
-      "loss": 1.0415,
+      "loss": 1.0399,
       "step": 46430
     },
     {
       "epoch": 139.46,
-      "grad_norm": 6.9654436111450195,
+      "grad_norm": 5.891149520874023,
       "learning_rate": 5.351351351351351e-06,
-      "loss": 1.0245,
+      "loss": 1.02,
       "step": 46440
     },
     {
       "epoch": 139.49,
-      "grad_norm": 6.403153896331787,
+      "grad_norm": 5.890532970428467,
       "learning_rate": 5.350350350350351e-06,
-      "loss": 1.0219,
+      "loss": 1.0181,
       "step": 46450
     },
     {
       "epoch": 139.52,
-      "grad_norm": 5.236534118652344,
+      "grad_norm": 5.484971523284912,
       "learning_rate": 5.34934934934935e-06,
-      "loss": 1.0488,
+      "loss": 1.0448,
       "step": 46460
     },
     {
       "epoch": 139.55,
-      "grad_norm": 6.172408103942871,
+      "grad_norm": 6.92724084854126,
       "learning_rate": 5.348348348348349e-06,
-      "loss": 1.0495,
+      "loss": 1.054,
       "step": 46470
     },
     {
       "epoch": 139.58,
-      "grad_norm": 7.020029067993164,
+      "grad_norm": 6.534867763519287,
       "learning_rate": 5.347347347347348e-06,
-      "loss": 1.0398,
+      "loss": 1.0368,
       "step": 46480
     },
     {
       "epoch": 139.61,
-      "grad_norm": 6.333571434020996,
+      "grad_norm": 6.544679164886475,
       "learning_rate": 5.346346346346347e-06,
-      "loss": 1.0263,
+      "loss": 1.0441,
       "step": 46490
     },
     {
       "epoch": 139.64,
-      "grad_norm": 5.062860488891602,
+      "grad_norm": 5.454566955566406,
       "learning_rate": 5.345345345345346e-06,
-      "loss": 1.0307,
+      "loss": 1.0436,
       "step": 46500
     },
     {
       "epoch": 139.67,
-      "grad_norm": 4.963271141052246,
+      "grad_norm": 6.1138014793396,
       "learning_rate": 5.344344344344344e-06,
-      "loss": 1.0626,
+      "loss": 1.0657,
       "step": 46510
     },
     {
       "epoch": 139.7,
-      "grad_norm": 7.217884540557861,
+      "grad_norm": 9.120600700378418,
       "learning_rate": 5.3433433433433435e-06,
-      "loss": 1.0248,
+      "loss": 1.0293,
       "step": 46520
     },
     {
       "epoch": 139.73,
-      "grad_norm": 6.480363845825195,
+      "grad_norm": 7.038805961608887,
       "learning_rate": 5.342342342342343e-06,
-      "loss": 1.0658,
+      "loss": 1.0583,
       "step": 46530
     },
     {
       "epoch": 139.76,
-      "grad_norm": 8.20897388458252,
+      "grad_norm": 11.326763153076172,
       "learning_rate": 5.3413413413413415e-06,
-      "loss": 1.0657,
+      "loss": 1.0606,
       "step": 46540
     },
     {
       "epoch": 139.79,
-      "grad_norm": 7.7469048500061035,
+      "grad_norm": 8.824652671813965,
       "learning_rate": 5.340340340340341e-06,
-      "loss": 1.0331,
+      "loss": 1.0265,
       "step": 46550
     },
     {
       "epoch": 139.82,
-      "grad_norm": 9.454681396484375,
+      "grad_norm": 8.985675811767578,
       "learning_rate": 5.33933933933934e-06,
-      "loss": 1.0688,
+      "loss": 1.0699,
       "step": 46560
     },
     {
       "epoch": 139.85,
-      "grad_norm": 6.313344955444336,
+      "grad_norm": 6.490138053894043,
       "learning_rate": 5.338338338338339e-06,
-      "loss": 1.0058,
+      "loss": 0.9985,
       "step": 46570
     },
     {
       "epoch": 139.88,
-      "grad_norm": 6.24948263168335,
+      "grad_norm": 6.1401753425598145,
       "learning_rate": 5.3373373373373384e-06,
-      "loss": 1.0831,
+      "loss": 1.0812,
       "step": 46580
     },
     {
       "epoch": 139.91,
-      "grad_norm": 5.325766563415527,
+      "grad_norm": 8.094685554504395,
       "learning_rate": 5.336336336336336e-06,
-      "loss": 0.9668,
+      "loss": 0.9689,
       "step": 46590
     },
     {
       "epoch": 139.94,
-      "grad_norm": 5.448654651641846,
+      "grad_norm": 5.560673236846924,
       "learning_rate": 5.335335335335336e-06,
-      "loss": 1.0227,
+      "loss": 1.0112,
       "step": 46600
     },
     {
       "epoch": 139.97,
-      "grad_norm": 8.100383758544922,
+      "grad_norm": 8.438315391540527,
       "learning_rate": 5.334334334334334e-06,
-      "loss": 1.0257,
+      "loss": 1.0203,
       "step": 46610
     },
     {
       "epoch": 140.0,
-      "grad_norm": 37.710044860839844,
+      "grad_norm": 63.77296829223633,
       "learning_rate": 5.333333333333334e-06,
-      "loss": 1.1634,
+      "loss": 1.1725,
       "step": 46620
     },
     {
       "epoch": 140.0,
-      "eval_accuracy": 0.8151,
-      "eval_loss": 0.6042969822883606,
-      "eval_runtime": 5.8568,
-      "eval_samples_per_second": 1707.423,
-      "eval_steps_per_second": 6.83,
+      "eval_accuracy": 0.8156,
+      "eval_loss": 0.6023972630500793,
+      "eval_runtime": 5.2059,
+      "eval_samples_per_second": 1920.911,
+      "eval_steps_per_second": 7.684,
       "step": 46620
     },
     {
       "epoch": 140.03,
-      "grad_norm": 6.27406120300293,
+      "grad_norm": 7.089919090270996,
       "learning_rate": 5.332332332332333e-06,
-      "loss": 0.9918,
+      "loss": 0.9878,
       "step": 46630
     },
     {
       "epoch": 140.06,
-      "grad_norm": 6.798130512237549,
+      "grad_norm": 7.026497840881348,
       "learning_rate": 5.331331331331332e-06,
-      "loss": 0.9993,
+      "loss": 1.0154,
       "step": 46640
     },
     {
       "epoch": 140.09,
-      "grad_norm": 6.592377185821533,
+      "grad_norm": 5.7509589195251465,
       "learning_rate": 5.330330330330331e-06,
-      "loss": 0.9803,
+      "loss": 0.9851,
       "step": 46650
     },
     {
       "epoch": 140.12,
-      "grad_norm": 5.323003768920898,
+      "grad_norm": 6.114601135253906,
       "learning_rate": 5.329329329329331e-06,
-      "loss": 1.0317,
+      "loss": 1.028,
       "step": 46660
     },
     {
       "epoch": 140.15,
-      "grad_norm": 6.05610990524292,
+      "grad_norm": 7.845508575439453,
       "learning_rate": 5.328328328328328e-06,
-      "loss": 0.9878,
+      "loss": 0.9912,
       "step": 46670
     },
     {
       "epoch": 140.18,
-      "grad_norm": 6.564503192901611,
+      "grad_norm": 6.991865158081055,
       "learning_rate": 5.327327327327328e-06,
-      "loss": 0.9941,
+      "loss": 1.0019,
       "step": 46680
     },
     {
       "epoch": 140.21,
-      "grad_norm": 7.168083667755127,
+      "grad_norm": 7.4719061851501465,
       "learning_rate": 5.326326326326326e-06,
-      "loss": 1.075,
+      "loss": 1.0675,
       "step": 46690
     },
     {
       "epoch": 140.24,
-      "grad_norm": 7.101056098937988,
+      "grad_norm": 9.661083221435547,
       "learning_rate": 5.325325325325326e-06,
-      "loss": 1.073,
+      "loss": 1.0808,
       "step": 46700
     },
     {
       "epoch": 140.27,
-      "grad_norm": 6.029839515686035,
+      "grad_norm": 7.504613399505615,
       "learning_rate": 5.324324324324324e-06,
-      "loss": 1.0159,
+      "loss": 1.0148,
       "step": 46710
     },
     {
       "epoch": 140.3,
-      "grad_norm": 7.69048547744751,
+      "grad_norm": 7.701639652252197,
       "learning_rate": 5.323323323323324e-06,
-      "loss": 1.016,
+      "loss": 1.0075,
       "step": 46720
     },
     {
       "epoch": 140.33,
-      "grad_norm": 6.468225002288818,
+      "grad_norm": 6.3550848960876465,
       "learning_rate": 5.322322322322323e-06,
-      "loss": 1.0633,
+      "loss": 1.0572,
       "step": 46730
     },
     {
       "epoch": 140.36,
-      "grad_norm": 6.246798038482666,
+      "grad_norm": 6.549104690551758,
       "learning_rate": 5.321321321321322e-06,
-      "loss": 1.0525,
+      "loss": 1.0494,
       "step": 46740
     },
     {
       "epoch": 140.39,
-      "grad_norm": 6.7331862449646,
+      "grad_norm": 7.651008605957031,
       "learning_rate": 5.320320320320321e-06,
-      "loss": 1.0216,
+      "loss": 1.0299,
       "step": 46750
     },
     {
       "epoch": 140.42,
-      "grad_norm": 7.686730861663818,
+      "grad_norm": 8.803339004516602,
       "learning_rate": 5.319319319319319e-06,
-      "loss": 0.9971,
+      "loss": 0.9859,
       "step": 46760
     },
     {
       "epoch": 140.45,
-      "grad_norm": 6.81760311126709,
+      "grad_norm": 6.1211442947387695,
       "learning_rate": 5.3183183183183185e-06,
-      "loss": 1.1096,
+      "loss": 1.1035,
       "step": 46770
     },
     {
       "epoch": 140.48,
-      "grad_norm": 6.483043670654297,
+      "grad_norm": 5.5641865730285645,
       "learning_rate": 5.317317317317318e-06,
-      "loss": 1.0252,
+      "loss": 1.021,
       "step": 46780
     },
     {
       "epoch": 140.51,
-      "grad_norm": 9.949965476989746,
+      "grad_norm": 6.85445499420166,
       "learning_rate": 5.3163163163163165e-06,
-      "loss": 1.0972,
+      "loss": 1.0891,
       "step": 46790
     },
     {
       "epoch": 140.54,
-      "grad_norm": 6.318627834320068,
+      "grad_norm": 6.437375545501709,
       "learning_rate": 5.315315315315316e-06,
-      "loss": 0.9577,
+      "loss": 0.9633,
       "step": 46800
     },
     {
       "epoch": 140.57,
-      "grad_norm": 6.11388635635376,
+      "grad_norm": 5.5897369384765625,
       "learning_rate": 5.314314314314315e-06,
-      "loss": 1.0221,
+      "loss": 1.024,
       "step": 46810
     },
     {
       "epoch": 140.6,
-      "grad_norm": 7.192134857177734,
+      "grad_norm": 8.960712432861328,
       "learning_rate": 5.313313313313314e-06,
-      "loss": 0.9811,
+      "loss": 0.9818,
       "step": 46820
     },
     {
       "epoch": 140.63,
-      "grad_norm": 5.867947578430176,
+      "grad_norm": 6.555196285247803,
       "learning_rate": 5.3123123123123135e-06,
-      "loss": 0.9796,
+      "loss": 0.9841,
       "step": 46830
     },
     {
       "epoch": 140.66,
-      "grad_norm": 7.666600704193115,
+      "grad_norm": 10.32123851776123,
       "learning_rate": 5.311311311311311e-06,
-      "loss": 0.9933,
+      "loss": 0.9842,
       "step": 46840
     },
     {
       "epoch": 140.69,
-      "grad_norm": 6.316259384155273,
+      "grad_norm": 7.085272789001465,
       "learning_rate": 5.310310310310311e-06,
-      "loss": 1.0975,
+      "loss": 1.1049,
       "step": 46850
     },
     {
       "epoch": 140.72,
-      "grad_norm": 8.398616790771484,
+      "grad_norm": 5.754984378814697,
       "learning_rate": 5.309309309309309e-06,
-      "loss": 1.0421,
+      "loss": 1.0314,
       "step": 46860
     },
     {
       "epoch": 140.75,
-      "grad_norm": 6.521101951599121,
+      "grad_norm": 7.8140549659729,
       "learning_rate": 5.308308308308309e-06,
-      "loss": 0.9974,
+      "loss": 0.9941,
       "step": 46870
     },
     {
       "epoch": 140.78,
-      "grad_norm": 10.452642440795898,
+      "grad_norm": 11.007996559143066,
       "learning_rate": 5.307307307307308e-06,
-      "loss": 1.0841,
+      "loss": 1.0687,
       "step": 46880
     },
     {
       "epoch": 140.81,
-      "grad_norm": 6.2719597816467285,
+      "grad_norm": 8.04029655456543,
       "learning_rate": 5.306306306306307e-06,
-      "loss": 1.0212,
+      "loss": 1.0084,
       "step": 46890
     },
     {
       "epoch": 140.84,
-      "grad_norm": 6.0748748779296875,
+      "grad_norm": 6.89626407623291,
       "learning_rate": 5.305305305305306e-06,
-      "loss": 1.0433,
+      "loss": 1.0435,
       "step": 46900
     },
     {
       "epoch": 140.87,
-      "grad_norm": 6.3779706954956055,
+      "grad_norm": 7.636281490325928,
       "learning_rate": 5.304304304304306e-06,
-      "loss": 1.0482,
+      "loss": 1.0448,
       "step": 46910
     },
     {
       "epoch": 140.9,
-      "grad_norm": 5.234277725219727,
+      "grad_norm": 6.380135536193848,
       "learning_rate": 5.303303303303303e-06,
-      "loss": 1.048,
+      "loss": 1.0443,
       "step": 46920
     },
     {
       "epoch": 140.93,
-      "grad_norm": 7.0167646408081055,
+      "grad_norm": 6.764390468597412,
       "learning_rate": 5.302302302302302e-06,
-      "loss": 1.0245,
+      "loss": 1.0233,
       "step": 46930
     },
     {
       "epoch": 140.96,
-      "grad_norm": 5.526442050933838,
+      "grad_norm": 9.548782348632812,
       "learning_rate": 5.301301301301301e-06,
-      "loss": 0.9438,
+      "loss": 0.94,
       "step": 46940
     },
     {
       "epoch": 140.99,
-      "grad_norm": 6.361134052276611,
+      "grad_norm": 6.580844879150391,
       "learning_rate": 5.300300300300301e-06,
-      "loss": 1.0906,
+      "loss": 1.086,
       "step": 46950
     },
     {
       "epoch": 141.0,
-      "eval_accuracy": 0.8134,
-      "eval_loss": 0.6070705652236938,
-      "eval_runtime": 5.572,
-      "eval_samples_per_second": 1794.693,
-      "eval_steps_per_second": 7.179,
+      "eval_accuracy": 0.8137,
+      "eval_loss": 0.6060789823532104,
+      "eval_runtime": 5.3615,
+      "eval_samples_per_second": 1865.156,
+      "eval_steps_per_second": 7.461,
       "step": 46953
     },
     {
       "epoch": 141.02,
-      "grad_norm": 9.43178653717041,
+      "grad_norm": 10.824030876159668,
       "learning_rate": 5.2992992992992994e-06,
-      "loss": 1.3226,
+      "loss": 1.3291,
       "step": 46960
     },
     {
       "epoch": 141.05,
-      "grad_norm": 5.211878299713135,
+      "grad_norm": 5.688042640686035,
       "learning_rate": 5.298298298298299e-06,
-      "loss": 1.0244,
+      "loss": 1.026,
       "step": 46970
     },
     {
       "epoch": 141.08,
-      "grad_norm": 4.6438307762146,
+      "grad_norm": 7.207863807678223,
       "learning_rate": 5.297297297297298e-06,
-      "loss": 1.0724,
+      "loss": 1.0794,
       "step": 46980
     },
     {
       "epoch": 141.11,
-      "grad_norm": 6.878786563873291,
+      "grad_norm": 9.05268383026123,
       "learning_rate": 5.296296296296297e-06,
-      "loss": 1.0386,
+      "loss": 1.029,
       "step": 46990
     },
     {
       "epoch": 141.14,
-      "grad_norm": 6.264168739318848,
+      "grad_norm": 7.707350254058838,
       "learning_rate": 5.2952952952952955e-06,
-      "loss": 1.0308,
+      "loss": 1.0298,
       "step": 47000
     },
     {
       "epoch": 141.17,
-      "grad_norm": 5.776508331298828,
+      "grad_norm": 5.3338541984558105,
       "learning_rate": 5.294294294294294e-06,
-      "loss": 0.9858,
+      "loss": 0.9862,
       "step": 47010
     },
     {
       "epoch": 141.2,
-      "grad_norm": 5.724595069885254,
+      "grad_norm": 6.678693771362305,
       "learning_rate": 5.2932932932932935e-06,
-      "loss": 1.0038,
+      "loss": 0.9957,
       "step": 47020
     },
     {
       "epoch": 141.23,
-      "grad_norm": 7.083037376403809,
+      "grad_norm": 7.350216865539551,
       "learning_rate": 5.292292292292293e-06,
-      "loss": 1.0273,
+      "loss": 1.0269,
       "step": 47030
     },
     {
       "epoch": 141.26,
-      "grad_norm": 7.12442684173584,
+      "grad_norm": 8.15230941772461,
       "learning_rate": 5.291291291291292e-06,
-      "loss": 1.0823,
+      "loss": 1.0663,
       "step": 47040
     },
     {
       "epoch": 141.29,
-      "grad_norm": 6.502033233642578,
+      "grad_norm": 8.659895896911621,
       "learning_rate": 5.290290290290291e-06,
-      "loss": 1.0549,
+      "loss": 1.058,
       "step": 47050
     },
     {
       "epoch": 141.32,
-      "grad_norm": 6.811048984527588,
+      "grad_norm": 7.386589050292969,
       "learning_rate": 5.28928928928929e-06,
-      "loss": 0.9806,
+      "loss": 0.9681,
       "step": 47060
     },
     {
       "epoch": 141.35,
-      "grad_norm": 6.6822285652160645,
+      "grad_norm": 6.353443145751953,
       "learning_rate": 5.288288288288289e-06,
-      "loss": 0.939,
+      "loss": 0.9444,
       "step": 47070
     },
     {
       "epoch": 141.38,
-      "grad_norm": 7.266794204711914,
+      "grad_norm": 7.697299003601074,
       "learning_rate": 5.2872872872872885e-06,
-      "loss": 1.0175,
+      "loss": 1.0071,
       "step": 47080
     },
     {
       "epoch": 141.41,
-      "grad_norm": 5.7770280838012695,
+      "grad_norm": 6.816279888153076,
       "learning_rate": 5.286286286286286e-06,
-      "loss": 1.0058,
+      "loss": 0.9996,
       "step": 47090
     },
     {
       "epoch": 141.44,
-      "grad_norm": 6.324966907501221,
+      "grad_norm": 6.532954216003418,
       "learning_rate": 5.285285285285286e-06,
-      "loss": 1.0911,
+      "loss": 1.0829,
       "step": 47100
     },
     {
       "epoch": 141.47,
-      "grad_norm": 6.5246992111206055,
+      "grad_norm": 6.952014446258545,
       "learning_rate": 5.284284284284284e-06,
-      "loss": 1.0014,
+      "loss": 0.9985,
       "step": 47110
     },
     {
       "epoch": 141.5,
-      "grad_norm": 5.969159126281738,
+      "grad_norm": 7.375873565673828,
       "learning_rate": 5.283283283283284e-06,
-      "loss": 1.0416,
+      "loss": 1.0403,
       "step": 47120
     },
     {
       "epoch": 141.53,
-      "grad_norm": 5.4994378089904785,
+      "grad_norm": 5.776439666748047,
       "learning_rate": 5.282282282282283e-06,
-      "loss": 0.9806,
+      "loss": 0.9905,
       "step": 47130
     },
     {
       "epoch": 141.56,
-      "grad_norm": 6.852823734283447,
+      "grad_norm": 6.373602390289307,
       "learning_rate": 5.281281281281282e-06,
-      "loss": 1.0319,
+      "loss": 1.0304,
       "step": 47140
     },
     {
       "epoch": 141.59,
-      "grad_norm": 6.511971950531006,
+      "grad_norm": 8.27077865600586,
       "learning_rate": 5.280280280280281e-06,
-      "loss": 1.0429,
+      "loss": 1.0371,
       "step": 47150
     },
     {
       "epoch": 141.62,
-      "grad_norm": 8.90776252746582,
+      "grad_norm": 9.617079734802246,
       "learning_rate": 5.27927927927928e-06,
-      "loss": 1.099,
+      "loss": 1.0881,
       "step": 47160
     },
     {
       "epoch": 141.65,
-      "grad_norm": 4.9805588722229,
+      "grad_norm": 6.0820159912109375,
       "learning_rate": 5.278278278278278e-06,
-      "loss": 1.0152,
+      "loss": 1.0205,
       "step": 47170
     },
     {
       "epoch": 141.68,
-      "grad_norm": 4.795029640197754,
+      "grad_norm": 5.149901390075684,
       "learning_rate": 5.277277277277277e-06,
-      "loss": 1.047,
+      "loss": 1.0396,
       "step": 47180
     },
     {
       "epoch": 141.71,
-      "grad_norm": 7.110479354858398,
+      "grad_norm": 9.076029777526855,
       "learning_rate": 5.2762762762762764e-06,
-      "loss": 1.0539,
+      "loss": 1.0495,
       "step": 47190
     },
     {
       "epoch": 141.74,
-      "grad_norm": 6.418974876403809,
+      "grad_norm": 6.079338550567627,
       "learning_rate": 5.275275275275276e-06,
-      "loss": 0.9952,
+      "loss": 0.9909,
       "step": 47200
     },
     {
       "epoch": 141.77,
-      "grad_norm": 7.617059230804443,
+      "grad_norm": 7.148215293884277,
       "learning_rate": 5.2742742742742745e-06,
-      "loss": 1.0543,
+      "loss": 1.0555,
       "step": 47210
     },
     {
       "epoch": 141.8,
-      "grad_norm": 8.041937828063965,
+      "grad_norm": 10.556778907775879,
       "learning_rate": 5.273273273273274e-06,
-      "loss": 1.017,
+      "loss": 1.0194,
       "step": 47220
     },
     {
       "epoch": 141.83,
-      "grad_norm": 6.690342426300049,
+      "grad_norm": 8.48851203918457,
       "learning_rate": 5.272272272272273e-06,
-      "loss": 1.0106,
+      "loss": 1.0143,
       "step": 47230
     },
     {
       "epoch": 141.86,
-      "grad_norm": 7.434762954711914,
+      "grad_norm": 6.761380672454834,
       "learning_rate": 5.271271271271272e-06,
-      "loss": 1.0528,
+      "loss": 1.0448,
       "step": 47240
     },
     {
       "epoch": 141.89,
-      "grad_norm": 6.899306774139404,
+      "grad_norm": 7.596887111663818,
       "learning_rate": 5.2702702702702705e-06,
-      "loss": 0.9657,
+      "loss": 0.9666,
       "step": 47250
     },
     {
       "epoch": 141.92,
-      "grad_norm": 7.003232955932617,
+      "grad_norm": 7.348509311676025,
       "learning_rate": 5.269269269269269e-06,
-      "loss": 1.0027,
+      "loss": 0.9976,
       "step": 47260
     },
     {
       "epoch": 141.95,
-      "grad_norm": 7.2447123527526855,
+      "grad_norm": 9.34985637664795,
       "learning_rate": 5.268268268268269e-06,
-      "loss": 1.0436,
+      "loss": 1.0424,
       "step": 47270
     },
     {
       "epoch": 141.98,
-      "grad_norm": 5.60079288482666,
+      "grad_norm": 10.9617338180542,
       "learning_rate": 5.267267267267267e-06,
-      "loss": 1.0672,
+      "loss": 1.0669,
       "step": 47280
     },
     {
       "epoch": 142.0,
-      "eval_accuracy": 0.8168,
-      "eval_loss": 0.6000540256500244,
-      "eval_runtime": 5.3414,
-      "eval_samples_per_second": 1872.155,
-      "eval_steps_per_second": 7.489,
+      "eval_accuracy": 0.8173,
+      "eval_loss": 0.6015468835830688,
+      "eval_runtime": 5.456,
+      "eval_samples_per_second": 1832.842,
+      "eval_steps_per_second": 7.331,
       "step": 47286
     },
     {
       "epoch": 142.01,
-      "grad_norm": 7.555996417999268,
+      "grad_norm": 8.864593505859375,
       "learning_rate": 5.266266266266267e-06,
-      "loss": 1.1276,
+      "loss": 1.1205,
       "step": 47290
     },
     {
       "epoch": 142.04,
-      "grad_norm": 5.104434013366699,
+      "grad_norm": 5.595366477966309,
       "learning_rate": 5.265265265265266e-06,
-      "loss": 0.9896,
+      "loss": 0.9875,
       "step": 47300
     },
     {
       "epoch": 142.07,
-      "grad_norm": 9.724295616149902,
+      "grad_norm": 10.559696197509766,
       "learning_rate": 5.264264264264265e-06,
-      "loss": 1.1305,
+      "loss": 1.1389,
       "step": 47310
     },
     {
       "epoch": 142.1,
-      "grad_norm": 5.2811784744262695,
+      "grad_norm": 7.180098056793213,
       "learning_rate": 5.263263263263264e-06,
-      "loss": 1.014,
+      "loss": 1.0201,
       "step": 47320
     },
     {
       "epoch": 142.13,
-      "grad_norm": 4.767124652862549,
+      "grad_norm": 5.541479110717773,
       "learning_rate": 5.2622622622622635e-06,
-      "loss": 1.0327,
+      "loss": 1.0239,
       "step": 47330
     },
     {
       "epoch": 142.16,
-      "grad_norm": 5.690001010894775,
+      "grad_norm": 6.282836437225342,
       "learning_rate": 5.261261261261261e-06,
-      "loss": 0.979,
+      "loss": 0.9767,
       "step": 47340
     },
     {
       "epoch": 142.19,
-      "grad_norm": 5.139133453369141,
+      "grad_norm": 5.28555965423584,
       "learning_rate": 5.260260260260261e-06,
-      "loss": 1.0555,
+      "loss": 1.0447,
       "step": 47350
     },
     {
       "epoch": 142.22,
-      "grad_norm": 6.579441070556641,
+      "grad_norm": 6.558642387390137,
       "learning_rate": 5.259259259259259e-06,
-      "loss": 1.0316,
+      "loss": 1.0246,
       "step": 47360
     },
     {
       "epoch": 142.25,
-      "grad_norm": 6.104929447174072,
+      "grad_norm": 9.13534164428711,
       "learning_rate": 5.258258258258259e-06,
-      "loss": 0.9808,
+      "loss": 0.9757,
       "step": 47370
     },
     {
       "epoch": 142.28,
-      "grad_norm": 5.4971699714660645,
+      "grad_norm": 6.319664478302002,
       "learning_rate": 5.257257257257257e-06,
-      "loss": 0.994,
+      "loss": 0.9929,
       "step": 47380
     },
     {
       "epoch": 142.31,
-      "grad_norm": 7.269001007080078,
+      "grad_norm": 8.788552284240723,
       "learning_rate": 5.256256256256257e-06,
-      "loss": 0.9971,
+      "loss": 1.0112,
       "step": 47390
     },
     {
       "epoch": 142.34,
-      "grad_norm": 6.593532562255859,
+      "grad_norm": 6.842126846313477,
       "learning_rate": 5.255255255255256e-06,
-      "loss": 1.0409,
+      "loss": 1.035,
       "step": 47400
     },
     {
       "epoch": 142.37,
-      "grad_norm": 6.457555770874023,
+      "grad_norm": 8.122811317443848,
       "learning_rate": 5.254254254254255e-06,
-      "loss": 1.0748,
+      "loss": 1.0724,
       "step": 47410
     },
     {
       "epoch": 142.4,
-      "grad_norm": 4.476595401763916,
+      "grad_norm": 5.575638294219971,
       "learning_rate": 5.2532532532532534e-06,
-      "loss": 0.9897,
+      "loss": 0.9938,
       "step": 47420
     },
     {
       "epoch": 142.43,
-      "grad_norm": 6.856764793395996,
+      "grad_norm": 8.269518852233887,
       "learning_rate": 5.252252252252252e-06,
-      "loss": 1.0102,
+      "loss": 1.0154,
       "step": 47430
     },
     {
       "epoch": 142.46,
-      "grad_norm": 6.486205101013184,
+      "grad_norm": 6.315513610839844,
       "learning_rate": 5.2512512512512515e-06,
-      "loss": 0.9867,
+      "loss": 0.9875,
       "step": 47440
     },
     {
       "epoch": 142.49,
-      "grad_norm": 5.109527111053467,
+      "grad_norm": 6.732626914978027,
       "learning_rate": 5.250250250250251e-06,
-      "loss": 1.074,
+      "loss": 1.0763,
       "step": 47450
     },
     {
       "epoch": 142.52,
-      "grad_norm": 5.367919445037842,
+      "grad_norm": 7.218102931976318,
       "learning_rate": 5.2492492492492495e-06,
-      "loss": 0.9735,
+      "loss": 0.9633,
       "step": 47460
     },
     {
       "epoch": 142.55,
-      "grad_norm": 4.924900054931641,
+      "grad_norm": 5.677657604217529,
       "learning_rate": 5.248248248248249e-06,
       "loss": 0.9951,
       "step": 47470
     },
     {
       "epoch": 142.58,
-      "grad_norm": 6.4463982582092285,
+      "grad_norm": 11.110333442687988,
       "learning_rate": 5.247247247247248e-06,
-      "loss": 1.0255,
+      "loss": 1.0394,
       "step": 47480
     },
     {
       "epoch": 142.61,
-      "grad_norm": 4.789453983306885,
+      "grad_norm": 6.7735137939453125,
       "learning_rate": 5.246246246246247e-06,
-      "loss": 1.0067,
+      "loss": 0.9935,
       "step": 47490
     },
     {
       "epoch": 142.64,
-      "grad_norm": 6.470994472503662,
+      "grad_norm": 9.984660148620605,
       "learning_rate": 5.245245245245245e-06,
-      "loss": 1.0411,
+      "loss": 1.0312,
       "step": 47500
     },
     {
       "epoch": 142.67,
-      "grad_norm": 7.14097261428833,
+      "grad_norm": 7.019810676574707,
       "learning_rate": 5.244244244244244e-06,
-      "loss": 0.9973,
+      "loss": 0.9927,
       "step": 47510
     },
     {
       "epoch": 142.7,
-      "grad_norm": 5.429264545440674,
+      "grad_norm": 6.480651378631592,
       "learning_rate": 5.243243243243244e-06,
-      "loss": 1.0325,
+      "loss": 1.0437,
       "step": 47520
     },
     {
       "epoch": 142.73,
-      "grad_norm": 5.163198471069336,
+      "grad_norm": 4.5865397453308105,
       "learning_rate": 5.242242242242242e-06,
-      "loss": 0.9636,
+      "loss": 0.9609,
       "step": 47530
     },
     {
       "epoch": 142.76,
-      "grad_norm": 5.9707932472229,
+      "grad_norm": 6.817288875579834,
       "learning_rate": 5.241241241241242e-06,
-      "loss": 0.9559,
+      "loss": 0.9508,
       "step": 47540
     },
     {
       "epoch": 142.79,
-      "grad_norm": 6.572181701660156,
+      "grad_norm": 7.758142948150635,
       "learning_rate": 5.240240240240241e-06,
-      "loss": 1.0332,
+      "loss": 1.0311,
       "step": 47550
     },
     {
       "epoch": 142.82,
-      "grad_norm": 5.375937461853027,
+      "grad_norm": 6.6133713722229,
       "learning_rate": 5.23923923923924e-06,
-      "loss": 1.0283,
+      "loss": 1.0337,
       "step": 47560
     },
     {
       "epoch": 142.85,
-      "grad_norm": 7.0314249992370605,
+      "grad_norm": 9.99130916595459,
       "learning_rate": 5.238238238238239e-06,
-      "loss": 1.0071,
+      "loss": 0.9952,
       "step": 47570
     },
     {
       "epoch": 142.88,
-      "grad_norm": 8.09426212310791,
+      "grad_norm": 9.343180656433105,
       "learning_rate": 5.2372372372372386e-06,
-      "loss": 1.0355,
+      "loss": 1.0326,
       "step": 47580
     },
     {
       "epoch": 142.91,
-      "grad_norm": 6.260875701904297,
+      "grad_norm": 6.988169193267822,
       "learning_rate": 5.236236236236236e-06,
-      "loss": 1.0378,
+      "loss": 1.0571,
       "step": 47590
     },
     {
       "epoch": 142.94,
-      "grad_norm": 6.354398727416992,
+      "grad_norm": 7.1053009033203125,
       "learning_rate": 5.235235235235236e-06,
-      "loss": 1.0268,
+      "loss": 1.0371,
       "step": 47600
     },
     {
       "epoch": 142.97,
-      "grad_norm": 7.164937973022461,
+      "grad_norm": 7.107811450958252,
       "learning_rate": 5.234234234234234e-06,
-      "loss": 1.0423,
+      "loss": 1.0308,
       "step": 47610
     },
     {
       "epoch": 143.0,
-      "eval_accuracy": 0.8144,
-      "eval_loss": 0.6077069640159607,
-      "eval_runtime": 5.659,
-      "eval_samples_per_second": 1767.1,
-      "eval_steps_per_second": 7.068,
+      "eval_accuracy": 0.814,
+      "eval_loss": 0.604745626449585,
+      "eval_runtime": 5.4143,
+      "eval_samples_per_second": 1846.952,
+      "eval_steps_per_second": 7.388,
       "step": 47619
     },
     {
       "epoch": 143.0,
-      "grad_norm": 8.83923053741455,
+      "grad_norm": 9.478767395019531,
       "learning_rate": 5.233233233233234e-06,
-      "loss": 1.2528,
+      "loss": 1.2357,
       "step": 47620
     },
     {
       "epoch": 143.03,
-      "grad_norm": 6.157909393310547,
+      "grad_norm": 8.707049369812012,
       "learning_rate": 5.232232232232232e-06,
-      "loss": 1.0178,
+      "loss": 1.0102,
       "step": 47630
     },
     {
       "epoch": 143.06,
-      "grad_norm": 5.669006824493408,
+      "grad_norm": 5.336564064025879,
       "learning_rate": 5.231231231231232e-06,
-      "loss": 1.0719,
+      "loss": 1.0631,
       "step": 47640
     },
     {
       "epoch": 143.09,
-      "grad_norm": 4.796868324279785,
+      "grad_norm": 8.793354034423828,
       "learning_rate": 5.230230230230231e-06,
-      "loss": 1.0388,
+      "loss": 1.0368,
       "step": 47650
     },
     {
       "epoch": 143.12,
-      "grad_norm": 8.472967147827148,
+      "grad_norm": 9.304779052734375,
       "learning_rate": 5.22922922922923e-06,
-      "loss": 0.9635,
+      "loss": 0.9504,
       "step": 47660
     },
     {
       "epoch": 143.15,
-      "grad_norm": 7.2833170890808105,
+      "grad_norm": 7.091740608215332,
       "learning_rate": 5.2282282282282285e-06,
-      "loss": 1.0447,
+      "loss": 1.0394,
       "step": 47670
     },
     {
       "epoch": 143.18,
-      "grad_norm": 8.10488224029541,
+      "grad_norm": 6.772148132324219,
       "learning_rate": 5.227227227227227e-06,
-      "loss": 1.0318,
+      "loss": 1.0326,
       "step": 47680
     },
     {
       "epoch": 143.21,
-      "grad_norm": 5.990333557128906,
+      "grad_norm": 8.011707305908203,
       "learning_rate": 5.2262262262262265e-06,
-      "loss": 1.0036,
+      "loss": 0.9956,
       "step": 47690
     },
     {
       "epoch": 143.24,
-      "grad_norm": 5.023184776306152,
+      "grad_norm": 5.304980754852295,
       "learning_rate": 5.225225225225226e-06,
-      "loss": 0.9736,
+      "loss": 0.9677,
       "step": 47700
     },
     {
       "epoch": 143.27,
-      "grad_norm": 6.4939680099487305,
+      "grad_norm": 7.066841125488281,
       "learning_rate": 5.2242242242242245e-06,
-      "loss": 0.9883,
+      "loss": 0.9847,
       "step": 47710
     },
     {
       "epoch": 143.3,
-      "grad_norm": 7.70677375793457,
+      "grad_norm": 8.946791648864746,
       "learning_rate": 5.223223223223224e-06,
-      "loss": 1.0391,
+      "loss": 1.034,
       "step": 47720
     },
     {
       "epoch": 143.33,
-      "grad_norm": 6.062836170196533,
+      "grad_norm": 6.5468220710754395,
       "learning_rate": 5.2222222222222226e-06,
-      "loss": 1.0556,
+      "loss": 1.055,
       "step": 47730
     },
     {
       "epoch": 143.36,
-      "grad_norm": 6.486807346343994,
+      "grad_norm": 5.640563488006592,
       "learning_rate": 5.221221221221222e-06,
-      "loss": 0.976,
+      "loss": 0.9698,
       "step": 47740
     },
     {
       "epoch": 143.39,
-      "grad_norm": 5.947448253631592,
+      "grad_norm": 5.173648357391357,
       "learning_rate": 5.22022022022022e-06,
-      "loss": 1.0441,
+      "loss": 1.0415,
       "step": 47750
     },
     {
       "epoch": 143.42,
-      "grad_norm": 5.600487232208252,
+      "grad_norm": 5.612419605255127,
       "learning_rate": 5.219219219219219e-06,
-      "loss": 0.9957,
+      "loss": 0.9929,
       "step": 47760
     },
     {
       "epoch": 143.45,
-      "grad_norm": 7.5771708488464355,
+      "grad_norm": 7.709536075592041,
       "learning_rate": 5.218218218218219e-06,
-      "loss": 0.9875,
+      "loss": 0.9897,
       "step": 47770
     },
     {
       "epoch": 143.48,
-      "grad_norm": 4.871819019317627,
+      "grad_norm": 6.236568927764893,
       "learning_rate": 5.217217217217217e-06,
-      "loss": 0.9937,
+      "loss": 0.9928,
       "step": 47780
     },
     {
       "epoch": 143.51,
-      "grad_norm": 5.509121417999268,
+      "grad_norm": 7.31873893737793,
       "learning_rate": 5.216216216216217e-06,
-      "loss": 0.9789,
+      "loss": 0.9707,
       "step": 47790
     },
     {
       "epoch": 143.54,
-      "grad_norm": 5.27621603012085,
+      "grad_norm": 8.332075119018555,
       "learning_rate": 5.215215215215216e-06,
-      "loss": 0.9481,
+      "loss": 0.9475,
       "step": 47800
     },
     {
       "epoch": 143.57,
-      "grad_norm": 8.907063484191895,
+      "grad_norm": 6.565253734588623,
       "learning_rate": 5.214214214214215e-06,
-      "loss": 1.0771,
+      "loss": 1.0829,
       "step": 47810
     },
     {
       "epoch": 143.6,
-      "grad_norm": 6.362974643707275,
+      "grad_norm": 6.101981163024902,
       "learning_rate": 5.213213213213214e-06,
-      "loss": 1.0277,
+      "loss": 1.0199,
       "step": 47820
     },
     {
       "epoch": 143.63,
-      "grad_norm": 7.8927321434021,
+      "grad_norm": 8.57576847076416,
       "learning_rate": 5.212212212212214e-06,
-      "loss": 0.9983,
+      "loss": 0.9997,
       "step": 47830
     },
     {
       "epoch": 143.66,
-      "grad_norm": 5.2228546142578125,
+      "grad_norm": 5.478487014770508,
       "learning_rate": 5.211211211211211e-06,
-      "loss": 1.0054,
+      "loss": 1.0019,
       "step": 47840
     },
     {
       "epoch": 143.69,
-      "grad_norm": 6.5567426681518555,
+      "grad_norm": 7.265838623046875,
       "learning_rate": 5.21021021021021e-06,
-      "loss": 0.9753,
+      "loss": 0.9704,
       "step": 47850
     },
     {
       "epoch": 143.72,
-      "grad_norm": 5.81366491317749,
+      "grad_norm": 6.250947952270508,
       "learning_rate": 5.209209209209209e-06,
-      "loss": 0.9699,
+      "loss": 0.9737,
       "step": 47860
     },
     {
       "epoch": 143.75,
-      "grad_norm": 7.932685375213623,
+      "grad_norm": 6.723698139190674,
       "learning_rate": 5.208208208208209e-06,
-      "loss": 1.03,
+      "loss": 1.0235,
       "step": 47870
     },
     {
       "epoch": 143.78,
-      "grad_norm": 6.8455095291137695,
+      "grad_norm": 8.378684997558594,
       "learning_rate": 5.2072072072072074e-06,
-      "loss": 0.9748,
+      "loss": 0.9855,
       "step": 47880
     },
     {
       "epoch": 143.81,
-      "grad_norm": 7.90122652053833,
+      "grad_norm": 6.063172817230225,
       "learning_rate": 5.206206206206207e-06,
-      "loss": 0.9731,
+      "loss": 0.9783,
       "step": 47890
     },
     {
       "epoch": 143.84,
-      "grad_norm": 6.4469404220581055,
+      "grad_norm": 7.104347229003906,
       "learning_rate": 5.205205205205206e-06,
-      "loss": 1.0602,
+      "loss": 1.0619,
       "step": 47900
     },
     {
       "epoch": 143.87,
-      "grad_norm": 8.116966247558594,
+      "grad_norm": 7.746407508850098,
       "learning_rate": 5.204204204204205e-06,
-      "loss": 0.9698,
+      "loss": 0.9632,
       "step": 47910
     },
     {
       "epoch": 143.9,
-      "grad_norm": 6.312359809875488,
+      "grad_norm": 7.495641231536865,
       "learning_rate": 5.2032032032032035e-06,
-      "loss": 1.0185,
+      "loss": 1.0097,
       "step": 47920
     },
     {
       "epoch": 143.93,
-      "grad_norm": 9.545312881469727,
+      "grad_norm": 7.342875003814697,
       "learning_rate": 5.202202202202202e-06,
-      "loss": 1.0007,
+      "loss": 1.0057,
       "step": 47930
     },
     {
       "epoch": 143.96,
-      "grad_norm": 5.9935126304626465,
+      "grad_norm": 5.970818042755127,
       "learning_rate": 5.2012012012012015e-06,
-      "loss": 1.0512,
+      "loss": 1.0474,
       "step": 47940
     },
     {
       "epoch": 143.99,
-      "grad_norm": 6.175382137298584,
+      "grad_norm": 8.777056694030762,
       "learning_rate": 5.200200200200201e-06,
-      "loss": 1.1038,
+      "loss": 1.0954,
       "step": 47950
     },
     {
       "epoch": 144.0,
-      "eval_accuracy": 0.8155,
-      "eval_loss": 0.6028499603271484,
-      "eval_runtime": 5.6749,
-      "eval_samples_per_second": 1762.141,
-      "eval_steps_per_second": 7.049,
+      "eval_accuracy": 0.8174,
+      "eval_loss": 0.6024371385574341,
+      "eval_runtime": 5.423,
+      "eval_samples_per_second": 1844.011,
+      "eval_steps_per_second": 7.376,
       "step": 47952
     },
     {
       "epoch": 144.02,
-      "grad_norm": 5.075616836547852,
+      "grad_norm": 5.105863094329834,
       "learning_rate": 5.1991991991991996e-06,
-      "loss": 1.135,
+      "loss": 1.1377,
       "step": 47960
     },
     {
       "epoch": 144.05,
-      "grad_norm": 7.548086643218994,
+      "grad_norm": 8.062420845031738,
       "learning_rate": 5.198198198198199e-06,
-      "loss": 1.0335,
+      "loss": 1.0445,
       "step": 47970
     },
     {
       "epoch": 144.08,
-      "grad_norm": 7.150139331817627,
+      "grad_norm": 7.460764408111572,
       "learning_rate": 5.197197197197198e-06,
-      "loss": 0.9615,
+      "loss": 0.9578,
       "step": 47980
     },
     {
       "epoch": 144.11,
-      "grad_norm": 7.535610675811768,
+      "grad_norm": 7.836721420288086,
       "learning_rate": 5.196196196196197e-06,
-      "loss": 1.0207,
+      "loss": 1.0107,
       "step": 47990
     },
     {
       "epoch": 144.14,
-      "grad_norm": 4.630771636962891,
+      "grad_norm": 4.940882205963135,
       "learning_rate": 5.195195195195195e-06,
-      "loss": 0.9888,
+      "loss": 0.9949,
       "step": 48000
     },
     {
       "epoch": 144.17,
-      "grad_norm": 5.447176456451416,
+      "grad_norm": 5.944962024688721,
       "learning_rate": 5.194194194194194e-06,
-      "loss": 1.0153,
+      "loss": 1.0221,
       "step": 48010
     },
     {
       "epoch": 144.2,
-      "grad_norm": 7.644443035125732,
+      "grad_norm": 9.727544784545898,
       "learning_rate": 5.193193193193194e-06,
-      "loss": 1.0133,
+      "loss": 1.0162,
       "step": 48020
     },
     {
       "epoch": 144.23,
-      "grad_norm": 6.131068706512451,
+      "grad_norm": 5.832951545715332,
       "learning_rate": 5.192192192192192e-06,
-      "loss": 1.0054,
+      "loss": 1.018,
       "step": 48030
     },
     {
       "epoch": 144.26,
-      "grad_norm": 6.426167964935303,
+      "grad_norm": 6.086207389831543,
       "learning_rate": 5.191191191191192e-06,
-      "loss": 1.1362,
+      "loss": 1.1337,
       "step": 48040
     },
     {
       "epoch": 144.29,
-      "grad_norm": 6.0597968101501465,
+      "grad_norm": 7.1738200187683105,
       "learning_rate": 5.190190190190191e-06,
-      "loss": 1.0003,
+      "loss": 1.0016,
       "step": 48050
     },
     {
       "epoch": 144.32,
-      "grad_norm": 5.537502765655518,
+      "grad_norm": 6.004421710968018,
       "learning_rate": 5.18918918918919e-06,
-      "loss": 0.994,
+      "loss": 0.9955,
       "step": 48060
     },
     {
       "epoch": 144.35,
-      "grad_norm": 8.956573486328125,
+      "grad_norm": 7.052433013916016,
       "learning_rate": 5.188188188188189e-06,
-      "loss": 1.044,
+      "loss": 1.0337,
       "step": 48070
     },
     {
       "epoch": 144.38,
-      "grad_norm": 7.066995620727539,
+      "grad_norm": 8.325255393981934,
       "learning_rate": 5.187187187187187e-06,
-      "loss": 1.0031,
+      "loss": 1.0043,
       "step": 48080
     },
     {
       "epoch": 144.41,
-      "grad_norm": 6.351221084594727,
+      "grad_norm": 5.3629069328308105,
       "learning_rate": 5.186186186186186e-06,
-      "loss": 0.9762,
+      "loss": 0.9629,
       "step": 48090
     },
     {
       "epoch": 144.44,
-      "grad_norm": 6.498474597930908,
+      "grad_norm": 8.071052551269531,
       "learning_rate": 5.185185185185185e-06,
-      "loss": 1.0896,
+      "loss": 1.0794,
       "step": 48100
     },
     {
       "epoch": 144.47,
-      "grad_norm": 7.171960353851318,
+      "grad_norm": 12.182059288024902,
       "learning_rate": 5.1841841841841844e-06,
-      "loss": 0.9915,
+      "loss": 0.999,
       "step": 48110
     },
     {
       "epoch": 144.5,
-      "grad_norm": 6.420947551727295,
+      "grad_norm": 6.960427284240723,
       "learning_rate": 5.183183183183184e-06,
-      "loss": 0.9548,
+      "loss": 0.9547,
       "step": 48120
     },
     {
       "epoch": 144.53,
-      "grad_norm": 7.904021739959717,
+      "grad_norm": 9.498191833496094,
       "learning_rate": 5.1821821821821825e-06,
-      "loss": 1.0131,
+      "loss": 1.0063,
       "step": 48130
     },
     {
       "epoch": 144.56,
-      "grad_norm": 6.076004981994629,
+      "grad_norm": 7.848505020141602,
       "learning_rate": 5.181181181181182e-06,
-      "loss": 1.1152,
+      "loss": 1.1091,
       "step": 48140
     },
     {
       "epoch": 144.59,
-      "grad_norm": 5.307633399963379,
+      "grad_norm": 7.179937839508057,
       "learning_rate": 5.180180180180181e-06,
-      "loss": 1.0653,
+      "loss": 1.0726,
       "step": 48150
     },
     {
       "epoch": 144.62,
-      "grad_norm": 7.961058616638184,
+      "grad_norm": 9.463897705078125,
       "learning_rate": 5.17917917917918e-06,
-      "loss": 1.0302,
+      "loss": 1.034,
       "step": 48160
     },
     {
       "epoch": 144.65,
-      "grad_norm": 8.538704872131348,
+      "grad_norm": 11.964587211608887,
       "learning_rate": 5.1781781781781785e-06,
-      "loss": 1.0138,
+      "loss": 1.0075,
       "step": 48170
     },
     {
       "epoch": 144.68,
-      "grad_norm": 5.876989364624023,
+      "grad_norm": 6.19171142578125,
       "learning_rate": 5.177177177177177e-06,
-      "loss": 1.0596,
+      "loss": 1.0585,
       "step": 48180
     },
     {
       "epoch": 144.71,
-      "grad_norm": 7.1260480880737305,
+      "grad_norm": 7.886561393737793,
       "learning_rate": 5.1761761761761766e-06,
-      "loss": 1.0777,
+      "loss": 1.081,
       "step": 48190
     },
     {
       "epoch": 144.74,
-      "grad_norm": 5.020293712615967,
+      "grad_norm": 7.116549015045166,
       "learning_rate": 5.175175175175175e-06,
-      "loss": 1.0515,
+      "loss": 1.0463,
       "step": 48200
     },
     {
       "epoch": 144.77,
-      "grad_norm": 12.212621688842773,
+      "grad_norm": 8.03917121887207,
       "learning_rate": 5.174174174174175e-06,
-      "loss": 1.0177,
+      "loss": 1.0165,
       "step": 48210
     },
     {
       "epoch": 144.8,
-      "grad_norm": 5.2442755699157715,
+      "grad_norm": 9.402873992919922,
       "learning_rate": 5.173173173173174e-06,
-      "loss": 1.0073,
+      "loss": 1.0139,
       "step": 48220
     },
     {
       "epoch": 144.83,
-      "grad_norm": 8.001505851745605,
+      "grad_norm": 8.933883666992188,
       "learning_rate": 5.172172172172173e-06,
-      "loss": 1.0211,
+      "loss": 1.0161,
       "step": 48230
     },
     {
       "epoch": 144.86,
-      "grad_norm": 6.308782577514648,
+      "grad_norm": 5.921163558959961,
       "learning_rate": 5.171171171171172e-06,
-      "loss": 1.0237,
+      "loss": 1.0327,
       "step": 48240
     },
     {
       "epoch": 144.89,
-      "grad_norm": 6.622931957244873,
+      "grad_norm": 6.735468864440918,
       "learning_rate": 5.17017017017017e-06,
-      "loss": 1.0819,
+      "loss": 1.0852,
       "step": 48250
     },
     {
       "epoch": 144.92,
-      "grad_norm": 4.671836853027344,
+      "grad_norm": 7.9569902420043945,
       "learning_rate": 5.169169169169169e-06,
-      "loss": 0.9952,
+      "loss": 0.9776,
       "step": 48260
     },
     {
       "epoch": 144.95,
-      "grad_norm": 4.604900360107422,
+      "grad_norm": 7.087533950805664,
       "learning_rate": 5.168168168168169e-06,
-      "loss": 1.0183,
+      "loss": 1.0157,
       "step": 48270
     },
     {
       "epoch": 144.98,
-      "grad_norm": 6.378788471221924,
+      "grad_norm": 6.288061141967773,
       "learning_rate": 5.167167167167167e-06,
-      "loss": 0.9353,
+      "loss": 0.9333,
       "step": 48280
     },
     {
       "epoch": 145.0,
-      "eval_accuracy": 0.8117,
-      "eval_loss": 0.6065496802330017,
-      "eval_runtime": 5.488,
-      "eval_samples_per_second": 1822.152,
-      "eval_steps_per_second": 7.289,
+      "eval_accuracy": 0.8172,
+      "eval_loss": 0.6064577102661133,
+      "eval_runtime": 5.418,
+      "eval_samples_per_second": 1845.69,
+      "eval_steps_per_second": 7.383,
       "step": 48285
     },
     {
       "epoch": 145.02,
-      "grad_norm": 7.043067455291748,
+      "grad_norm": 6.957758903503418,
       "learning_rate": 5.166166166166167e-06,
-      "loss": 1.2419,
+      "loss": 1.231,
       "step": 48290
     },
     {
       "epoch": 145.05,
-      "grad_norm": 7.470553398132324,
+      "grad_norm": 6.457559585571289,
       "learning_rate": 5.165165165165165e-06,
-      "loss": 1.0275,
+      "loss": 1.0241,
       "step": 48300
     },
     {
       "epoch": 145.08,
-      "grad_norm": 7.4756269454956055,
+      "grad_norm": 6.03047513961792,
       "learning_rate": 5.164164164164165e-06,
-      "loss": 1.0363,
+      "loss": 1.0313,
       "step": 48310
     },
     {
       "epoch": 145.11,
-      "grad_norm": 7.148989200592041,
+      "grad_norm": 7.905534267425537,
       "learning_rate": 5.163163163163164e-06,
-      "loss": 1.0663,
+      "loss": 1.0574,
       "step": 48320
     },
     {
       "epoch": 145.14,
-      "grad_norm": 5.574904918670654,
+      "grad_norm": 6.684157848358154,
       "learning_rate": 5.162162162162162e-06,
-      "loss": 1.0892,
+      "loss": 1.0824,
       "step": 48330
     },
     {
       "epoch": 145.17,
-      "grad_norm": 6.758440017700195,
+      "grad_norm": 8.36883544921875,
       "learning_rate": 5.161161161161161e-06,
-      "loss": 1.0712,
+      "loss": 1.0802,
       "step": 48340
     },
     {
       "epoch": 145.2,
-      "grad_norm": 6.785769462585449,
+      "grad_norm": 5.85011625289917,
       "learning_rate": 5.16016016016016e-06,
-      "loss": 0.9335,
+      "loss": 0.9301,
       "step": 48350
     },
     {
       "epoch": 145.23,
-      "grad_norm": 6.558292388916016,
+      "grad_norm": 6.696437835693359,
       "learning_rate": 5.1591591591591595e-06,
-      "loss": 1.0139,
+      "loss": 1.0167,
       "step": 48360
     },
     {
       "epoch": 145.26,
-      "grad_norm": 8.230819702148438,
+      "grad_norm": 10.681571006774902,
       "learning_rate": 5.158158158158159e-06,
-      "loss": 1.015,
+      "loss": 1.0231,
       "step": 48370
     },
     {
       "epoch": 145.29,
-      "grad_norm": 5.339791297912598,
+      "grad_norm": 4.787454605102539,
       "learning_rate": 5.1571571571571575e-06,
-      "loss": 1.0189,
+      "loss": 1.0261,
       "step": 48380
     },
     {
       "epoch": 145.32,
-      "grad_norm": 5.213869094848633,
+      "grad_norm": 7.414575099945068,
       "learning_rate": 5.156156156156157e-06,
-      "loss": 1.0027,
+      "loss": 0.9972,
       "step": 48390
     },
     {
       "epoch": 145.35,
-      "grad_norm": 10.181922912597656,
+      "grad_norm": 5.833670139312744,
       "learning_rate": 5.155155155155156e-06,
-      "loss": 1.0529,
+      "loss": 1.0564,
       "step": 48400
     },
     {
       "epoch": 145.38,
-      "grad_norm": 7.537327289581299,
+      "grad_norm": 7.160225868225098,
       "learning_rate": 5.154154154154155e-06,
-      "loss": 1.0492,
+      "loss": 1.0432,
       "step": 48410
     },
     {
       "epoch": 145.41,
-      "grad_norm": 7.463928699493408,
+      "grad_norm": 7.571688652038574,
       "learning_rate": 5.153153153153153e-06,
-      "loss": 0.9771,
+      "loss": 0.9732,
       "step": 48420
     },
     {
       "epoch": 145.44,
-      "grad_norm": 6.439528942108154,
+      "grad_norm": 6.970107555389404,
       "learning_rate": 5.152152152152152e-06,
-      "loss": 1.0045,
+      "loss": 0.9994,
       "step": 48430
     },
     {
       "epoch": 145.47,
-      "grad_norm": 6.276314735412598,
+      "grad_norm": 5.890781402587891,
       "learning_rate": 5.151151151151152e-06,
-      "loss": 1.0368,
+      "loss": 1.0339,
       "step": 48440
     },
     {
       "epoch": 145.5,
-      "grad_norm": 6.54655647277832,
+      "grad_norm": 8.98634147644043,
       "learning_rate": 5.15015015015015e-06,
-      "loss": 1.0311,
+      "loss": 1.027,
       "step": 48450
     },
     {
       "epoch": 145.53,
-      "grad_norm": 5.961723804473877,
+      "grad_norm": 6.865266799926758,
       "learning_rate": 5.14914914914915e-06,
-      "loss": 1.064,
+      "loss": 1.06,
       "step": 48460
     },
     {
       "epoch": 145.56,
-      "grad_norm": 6.154224395751953,
+      "grad_norm": 6.9112653732299805,
       "learning_rate": 5.148148148148149e-06,
-      "loss": 1.0252,
+      "loss": 1.0269,
       "step": 48470
     },
     {
       "epoch": 145.59,
-      "grad_norm": 7.961581707000732,
+      "grad_norm": 8.376065254211426,
       "learning_rate": 5.147147147147148e-06,
-      "loss": 1.0543,
+      "loss": 1.0487,
       "step": 48480
     },
     {
       "epoch": 145.62,
-      "grad_norm": 5.90988826751709,
+      "grad_norm": 6.129405498504639,
       "learning_rate": 5.146146146146147e-06,
-      "loss": 1.0159,
+      "loss": 1.0206,
       "step": 48490
     },
     {
       "epoch": 145.65,
-      "grad_norm": 7.133069038391113,
+      "grad_norm": 5.61231803894043,
       "learning_rate": 5.145145145145145e-06,
-      "loss": 0.9848,
+      "loss": 0.9878,
       "step": 48500
     },
     {
       "epoch": 145.68,
-      "grad_norm": 7.446450233459473,
+      "grad_norm": 8.358036041259766,
       "learning_rate": 5.144144144144144e-06,
-      "loss": 1.0535,
+      "loss": 1.0548,
       "step": 48510
     },
     {
       "epoch": 145.71,
-      "grad_norm": 7.89063835144043,
+      "grad_norm": 5.991701602935791,
       "learning_rate": 5.143143143143144e-06,
-      "loss": 1.0777,
+      "loss": 1.0808,
       "step": 48520
     },
     {
       "epoch": 145.74,
-      "grad_norm": 9.257742881774902,
+      "grad_norm": 7.173447608947754,
       "learning_rate": 5.142142142142142e-06,
-      "loss": 0.999,
+      "loss": 0.9918,
       "step": 48530
     },
     {
       "epoch": 145.77,
-      "grad_norm": 6.750062942504883,
+      "grad_norm": 7.334826946258545,
       "learning_rate": 5.141141141141142e-06,
-      "loss": 1.0108,
+      "loss": 1.0021,
       "step": 48540
     },
     {
       "epoch": 145.8,
-      "grad_norm": 5.453706741333008,
+      "grad_norm": 6.0422892570495605,
       "learning_rate": 5.14014014014014e-06,
-      "loss": 0.9841,
+      "loss": 0.9832,
       "step": 48550
     },
     {
       "epoch": 145.83,
-      "grad_norm": 6.44390344619751,
+      "grad_norm": 6.504695892333984,
       "learning_rate": 5.13913913913914e-06,
-      "loss": 1.0606,
+      "loss": 1.0723,
       "step": 48560
     },
     {
       "epoch": 145.86,
-      "grad_norm": 6.707788467407227,
+      "grad_norm": 6.187772274017334,
       "learning_rate": 5.138138138138139e-06,
-      "loss": 1.0404,
+      "loss": 1.0328,
       "step": 48570
     },
     {
       "epoch": 145.89,
-      "grad_norm": 8.374937057495117,
+      "grad_norm": 10.116083145141602,
       "learning_rate": 5.137137137137137e-06,
-      "loss": 1.0641,
+      "loss": 1.0755,
       "step": 48580
     },
     {
       "epoch": 145.92,
-      "grad_norm": 5.39868688583374,
+      "grad_norm": 4.875659942626953,
       "learning_rate": 5.1361361361361365e-06,
-      "loss": 1.0668,
+      "loss": 1.0601,
       "step": 48590
     },
     {
       "epoch": 145.95,
-      "grad_norm": 6.798660755157471,
+      "grad_norm": 5.5736870765686035,
       "learning_rate": 5.135135135135135e-06,
       "loss": 1.0126,
       "step": 48600
     },
     {
       "epoch": 145.98,
-      "grad_norm": 7.709324836730957,
+      "grad_norm": 7.046219825744629,
       "learning_rate": 5.1341341341341345e-06,
-      "loss": 1.0238,
+      "loss": 1.0231,
       "step": 48610
     },
     {
       "epoch": 146.0,
-      "eval_accuracy": 0.8151,
-      "eval_loss": 0.597857654094696,
-      "eval_runtime": 5.5507,
-      "eval_samples_per_second": 1801.577,
-      "eval_steps_per_second": 7.206,
+      "eval_accuracy": 0.8178,
+      "eval_loss": 0.5973260998725891,
+      "eval_runtime": 5.2537,
+      "eval_samples_per_second": 1903.432,
+      "eval_steps_per_second": 7.614,
       "step": 48618
     },
     {
       "epoch": 146.01,
-      "grad_norm": 7.68079948425293,
+      "grad_norm": 8.109574317932129,
       "learning_rate": 5.133133133133134e-06,
-      "loss": 0.9437,
+      "loss": 0.9294,
       "step": 48620
     },
     {
       "epoch": 146.04,
-      "grad_norm": 4.5637125968933105,
+      "grad_norm": 5.661199569702148,
       "learning_rate": 5.1321321321321325e-06,
-      "loss": 1.0268,
+      "loss": 1.0311,
       "step": 48630
     },
     {
       "epoch": 146.07,
-      "grad_norm": 7.258363246917725,
+      "grad_norm": 7.816080093383789,
       "learning_rate": 5.131131131131132e-06,
-      "loss": 1.0102,
+      "loss": 1.0141,
       "step": 48640
     },
     {
       "epoch": 146.1,
-      "grad_norm": 6.759359359741211,
+      "grad_norm": 7.997344493865967,
       "learning_rate": 5.1301301301301306e-06,
-      "loss": 1.0516,
+      "loss": 1.0455,
       "step": 48650
     },
     {
       "epoch": 146.13,
-      "grad_norm": 6.976968288421631,
+      "grad_norm": 7.645991325378418,
       "learning_rate": 5.12912912912913e-06,
-      "loss": 1.0057,
+      "loss": 1.0079,
       "step": 48660
     },
     {
       "epoch": 146.16,
-      "grad_norm": 7.572692394256592,
+      "grad_norm": 6.643763542175293,
       "learning_rate": 5.128128128128128e-06,
-      "loss": 1.028,
+      "loss": 1.0252,
       "step": 48670
     },
     {
       "epoch": 146.19,
-      "grad_norm": 5.423971652984619,
+      "grad_norm": 5.932886600494385,
       "learning_rate": 5.127127127127127e-06,
-      "loss": 1.0084,
+      "loss": 1.0076,
       "step": 48680
     },
     {
       "epoch": 146.22,
-      "grad_norm": 6.727510929107666,
+      "grad_norm": 6.616607189178467,
       "learning_rate": 5.126126126126127e-06,
-      "loss": 1.0502,
+      "loss": 1.0446,
       "step": 48690
     },
     {
       "epoch": 146.25,
-      "grad_norm": 6.7762908935546875,
+      "grad_norm": 7.262320041656494,
       "learning_rate": 5.125125125125125e-06,
-      "loss": 1.0003,
+      "loss": 1.0065,
       "step": 48700
     },
     {
       "epoch": 146.28,
-      "grad_norm": 5.333106994628906,
+      "grad_norm": 7.269801616668701,
       "learning_rate": 5.124124124124125e-06,
-      "loss": 0.9633,
+      "loss": 0.9581,
       "step": 48710
     },
     {
       "epoch": 146.31,
-      "grad_norm": 6.191470146179199,
+      "grad_norm": 6.859851837158203,
       "learning_rate": 5.123123123123124e-06,
-      "loss": 1.045,
+      "loss": 1.0551,
       "step": 48720
     },
     {
       "epoch": 146.34,
-      "grad_norm": 5.9976372718811035,
+      "grad_norm": 6.563263416290283,
       "learning_rate": 5.122122122122123e-06,
-      "loss": 1.0241,
+      "loss": 1.0092,
       "step": 48730
     },
     {
       "epoch": 146.37,
-      "grad_norm": 6.714161396026611,
+      "grad_norm": 6.835339069366455,
       "learning_rate": 5.121121121121122e-06,
-      "loss": 1.045,
+      "loss": 1.0556,
       "step": 48740
     },
     {
       "epoch": 146.4,
-      "grad_norm": 5.355071544647217,
+      "grad_norm": 5.95213508605957,
       "learning_rate": 5.12012012012012e-06,
-      "loss": 0.9868,
+      "loss": 0.9864,
       "step": 48750
     },
     {
       "epoch": 146.43,
-      "grad_norm": 5.219332695007324,
+      "grad_norm": 8.126996994018555,
       "learning_rate": 5.119119119119119e-06,
-      "loss": 0.956,
+      "loss": 0.9471,
       "step": 48760
     },
     {
       "epoch": 146.46,
-      "grad_norm": 6.305655002593994,
+      "grad_norm": 6.570484161376953,
       "learning_rate": 5.118118118118118e-06,
-      "loss": 1.0009,
+      "loss": 0.9979,
       "step": 48770
     },
     {
       "epoch": 146.49,
-      "grad_norm": 5.640052795410156,
+      "grad_norm": 5.935009956359863,
       "learning_rate": 5.117117117117117e-06,
-      "loss": 1.0067,
+      "loss": 1.0028,
       "step": 48780
     },
     {
       "epoch": 146.52,
-      "grad_norm": 6.892086982727051,
+      "grad_norm": 6.4679059982299805,
       "learning_rate": 5.116116116116117e-06,
-      "loss": 0.992,
+      "loss": 0.9894,
       "step": 48790
     },
     {
       "epoch": 146.55,
-      "grad_norm": 6.6011223793029785,
+      "grad_norm": 5.512022018432617,
       "learning_rate": 5.115115115115115e-06,
-      "loss": 1.0395,
+      "loss": 1.0544,
       "step": 48800
     },
     {
       "epoch": 146.58,
-      "grad_norm": 7.761758804321289,
+      "grad_norm": 8.656336784362793,
       "learning_rate": 5.114114114114115e-06,
-      "loss": 0.9822,
+      "loss": 0.993,
       "step": 48810
     },
     {
       "epoch": 146.61,
-      "grad_norm": 6.890510082244873,
+      "grad_norm": 7.6986541748046875,
       "learning_rate": 5.113113113113114e-06,
-      "loss": 1.1425,
+      "loss": 1.1404,
       "step": 48820
     },
     {
       "epoch": 146.64,
-      "grad_norm": 6.5892558097839355,
+      "grad_norm": 6.43289852142334,
       "learning_rate": 5.112112112112112e-06,
-      "loss": 1.0486,
+      "loss": 1.047,
       "step": 48830
     },
     {
       "epoch": 146.67,
-      "grad_norm": 6.174718856811523,
+      "grad_norm": 6.432590007781982,
       "learning_rate": 5.1111111111111115e-06,
-      "loss": 1.0216,
+      "loss": 1.0142,
       "step": 48840
     },
     {
       "epoch": 146.7,
-      "grad_norm": 5.251026630401611,
+      "grad_norm": 4.328456878662109,
       "learning_rate": 5.11011011011011e-06,
-      "loss": 0.9681,
+      "loss": 0.9644,
       "step": 48850
     },
     {
       "epoch": 146.73,
-      "grad_norm": 6.359548568725586,
+      "grad_norm": 7.803346157073975,
       "learning_rate": 5.1091091091091095e-06,
-      "loss": 1.0032,
+      "loss": 1.0057,
       "step": 48860
     },
     {
       "epoch": 146.76,
-      "grad_norm": 8.928146362304688,
+      "grad_norm": 8.96503734588623,
       "learning_rate": 5.108108108108108e-06,
-      "loss": 1.0288,
+      "loss": 1.0359,
       "step": 48870
     },
     {
       "epoch": 146.79,
-      "grad_norm": 7.210725784301758,
+      "grad_norm": 6.280088901519775,
       "learning_rate": 5.1071071071071076e-06,
-      "loss": 1.0997,
+      "loss": 1.0962,
       "step": 48880
     },
     {
       "epoch": 146.82,
-      "grad_norm": 5.523189067840576,
+      "grad_norm": 6.808392524719238,
       "learning_rate": 5.106106106106107e-06,
-      "loss": 1.0297,
+      "loss": 1.0282,
       "step": 48890
     },
     {
       "epoch": 146.85,
-      "grad_norm": 8.25831127166748,
+      "grad_norm": 10.077254295349121,
       "learning_rate": 5.105105105105106e-06,
-      "loss": 1.0375,
+      "loss": 1.0315,
       "step": 48900
     },
     {
       "epoch": 146.88,
-      "grad_norm": 6.340320110321045,
+      "grad_norm": 6.391401767730713,
       "learning_rate": 5.104104104104105e-06,
-      "loss": 1.0027,
+      "loss": 0.9969,
       "step": 48910
     },
     {
       "epoch": 146.91,
-      "grad_norm": 6.019315242767334,
+      "grad_norm": 8.048151969909668,
       "learning_rate": 5.103103103103103e-06,
-      "loss": 1.0269,
+      "loss": 1.032,
       "step": 48920
     },
     {
       "epoch": 146.94,
-      "grad_norm": 11.804777145385742,
+      "grad_norm": 6.808799743652344,
       "learning_rate": 5.102102102102102e-06,
-      "loss": 1.0704,
+      "loss": 1.0681,
       "step": 48930
     },
     {
       "epoch": 146.97,
-      "grad_norm": 7.368858337402344,
+      "grad_norm": 7.747023105621338,
       "learning_rate": 5.101101101101102e-06,
-      "loss": 0.9829,
+      "loss": 0.9881,
       "step": 48940
     },
     {
       "epoch": 147.0,
-      "grad_norm": 5.575592041015625,
+      "grad_norm": 6.36798095703125,
       "learning_rate": 5.1001001001001e-06,
-      "loss": 1.0313,
+      "loss": 1.0397,
       "step": 48950
     },
     {
       "epoch": 147.0,
-      "eval_accuracy": 0.8149,
-      "eval_loss": 0.6022176742553711,
-      "eval_runtime": 5.2792,
-      "eval_samples_per_second": 1894.235,
-      "eval_steps_per_second": 7.577,
+      "eval_accuracy": 0.8161,
+      "eval_loss": 0.6012245416641235,
+      "eval_runtime": 5.4308,
+      "eval_samples_per_second": 1841.363,
+      "eval_steps_per_second": 7.365,
       "step": 48951
     },
     {
       "epoch": 147.03,
-      "grad_norm": 7.250554084777832,
+      "grad_norm": 10.612808227539062,
       "learning_rate": 5.0990990990991e-06,
-      "loss": 1.3061,
+      "loss": 1.3009,
       "step": 48960
     },
     {
       "epoch": 147.06,
-      "grad_norm": 5.082711219787598,
+      "grad_norm": 5.699284076690674,
       "learning_rate": 5.098098098098099e-06,
-      "loss": 1.0342,
+      "loss": 1.0319,
       "step": 48970
     },
     {
       "epoch": 147.09,
-      "grad_norm": 5.4449639320373535,
+      "grad_norm": 4.948537826538086,
       "learning_rate": 5.097097097097098e-06,
-      "loss": 1.0262,
+      "loss": 1.0204,
       "step": 48980
     },
     {
       "epoch": 147.12,
-      "grad_norm": 6.517314910888672,
+      "grad_norm": 7.380295753479004,
       "learning_rate": 5.096096096096097e-06,
-      "loss": 0.9714,
+      "loss": 0.9708,
       "step": 48990
     },
     {
       "epoch": 147.15,
-      "grad_norm": 7.272191047668457,
+      "grad_norm": 8.036633491516113,
       "learning_rate": 5.095095095095095e-06,
-      "loss": 1.0063,
+      "loss": 0.9925,
       "step": 49000
     },
     {
       "epoch": 147.18,
-      "grad_norm": 6.898960590362549,
+      "grad_norm": 6.50740385055542,
       "learning_rate": 5.094094094094094e-06,
-      "loss": 1.0322,
+      "loss": 1.0253,
       "step": 49010
     },
     {
       "epoch": 147.21,
-      "grad_norm": 6.25504207611084,
+      "grad_norm": 7.783687114715576,
       "learning_rate": 5.093093093093093e-06,
-      "loss": 1.0772,
+      "loss": 1.0724,
       "step": 49020
     },
     {
       "epoch": 147.24,
-      "grad_norm": 7.269886016845703,
+      "grad_norm": 7.224920749664307,
       "learning_rate": 5.092092092092092e-06,
-      "loss": 0.9886,
+      "loss": 0.9948,
       "step": 49030
     },
     {
       "epoch": 147.27,
-      "grad_norm": 5.291088104248047,
+      "grad_norm": 5.018131732940674,
       "learning_rate": 5.091091091091092e-06,
-      "loss": 1.0374,
+      "loss": 1.0348,
       "step": 49040
     },
     {
       "epoch": 147.3,
-      "grad_norm": 7.0945281982421875,
+      "grad_norm": 6.251223087310791,
       "learning_rate": 5.0900900900900905e-06,
-      "loss": 1.044,
+      "loss": 1.0369,
       "step": 49050
     },
     {
       "epoch": 147.33,
-      "grad_norm": 10.194209098815918,
+      "grad_norm": 14.00920295715332,
       "learning_rate": 5.08908908908909e-06,
-      "loss": 0.9868,
+      "loss": 0.9757,
       "step": 49060
     },
     {
       "epoch": 147.36,
-      "grad_norm": 6.09758186340332,
+      "grad_norm": 6.785868167877197,
       "learning_rate": 5.088088088088089e-06,
-      "loss": 0.9664,
+      "loss": 0.9716,
       "step": 49070
     },
     {
       "epoch": 147.39,
-      "grad_norm": 6.63582706451416,
+      "grad_norm": 5.76088285446167,
       "learning_rate": 5.087087087087087e-06,
-      "loss": 1.0856,
+      "loss": 1.0775,
       "step": 49080
     },
     {
       "epoch": 147.42,
-      "grad_norm": 5.335341930389404,
+      "grad_norm": 7.826784610748291,
       "learning_rate": 5.0860860860860865e-06,
-      "loss": 1.0012,
+      "loss": 1.0002,
       "step": 49090
     },
     {
       "epoch": 147.45,
-      "grad_norm": 8.818526268005371,
+      "grad_norm": 9.421430587768555,
       "learning_rate": 5.085085085085085e-06,
-      "loss": 1.0042,
+      "loss": 1.01,
       "step": 49100
     },
     {
       "epoch": 147.48,
-      "grad_norm": 8.347199440002441,
+      "grad_norm": 9.863956451416016,
       "learning_rate": 5.0840840840840846e-06,
-      "loss": 1.0437,
+      "loss": 1.0331,
       "step": 49110
     },
     {
       "epoch": 147.51,
-      "grad_norm": 5.8957905769348145,
+      "grad_norm": 4.906596660614014,
       "learning_rate": 5.083083083083083e-06,
-      "loss": 0.9487,
+      "loss": 0.9608,
       "step": 49120
     },
     {
       "epoch": 147.54,
-      "grad_norm": 5.057541847229004,
+      "grad_norm": 6.488494396209717,
       "learning_rate": 5.082082082082083e-06,
-      "loss": 0.934,
+      "loss": 0.944,
       "step": 49130
     },
     {
       "epoch": 147.57,
-      "grad_norm": 6.406920909881592,
+      "grad_norm": 5.480376243591309,
       "learning_rate": 5.081081081081082e-06,
-      "loss": 1.0243,
+      "loss": 1.0288,
       "step": 49140
     },
     {
       "epoch": 147.6,
-      "grad_norm": 5.402125358581543,
+      "grad_norm": 7.154183387756348,
       "learning_rate": 5.080080080080081e-06,
-      "loss": 1.0448,
+      "loss": 1.034,
       "step": 49150
     },
     {
       "epoch": 147.63,
-      "grad_norm": 7.241135120391846,
+      "grad_norm": 5.800095081329346,
       "learning_rate": 5.07907907907908e-06,
-      "loss": 0.9941,
+      "loss": 0.9856,
       "step": 49160
     },
     {
       "epoch": 147.66,
-      "grad_norm": 6.8342976570129395,
+      "grad_norm": 8.676850318908691,
       "learning_rate": 5.078078078078078e-06,
-      "loss": 1.0353,
+      "loss": 1.0392,
       "step": 49170
     },
     {
       "epoch": 147.69,
-      "grad_norm": 6.737917423248291,
+      "grad_norm": 6.300413131713867,
       "learning_rate": 5.077077077077077e-06,
-      "loss": 1.0296,
+      "loss": 1.0359,
       "step": 49180
     },
     {
       "epoch": 147.72,
-      "grad_norm": 7.363367557525635,
+      "grad_norm": 6.525656223297119,
       "learning_rate": 5.076076076076077e-06,
-      "loss": 1.0609,
+      "loss": 1.0594,
       "step": 49190
     },
     {
       "epoch": 147.75,
-      "grad_norm": 6.454530715942383,
+      "grad_norm": 8.802899360656738,
       "learning_rate": 5.075075075075075e-06,
-      "loss": 1.0059,
+      "loss": 0.9998,
       "step": 49200
     },
     {
       "epoch": 147.78,
-      "grad_norm": 6.645541667938232,
+      "grad_norm": 7.941816329956055,
       "learning_rate": 5.074074074074075e-06,
-      "loss": 1.0479,
+      "loss": 1.0405,
       "step": 49210
     },
     {
       "epoch": 147.81,
-      "grad_norm": 7.529337406158447,
+      "grad_norm": 7.519719123840332,
       "learning_rate": 5.073073073073073e-06,
-      "loss": 1.0704,
+      "loss": 1.0564,
       "step": 49220
     },
     {
       "epoch": 147.84,
-      "grad_norm": 7.029292106628418,
+      "grad_norm": 5.251666069030762,
       "learning_rate": 5.072072072072073e-06,
-      "loss": 0.99,
+      "loss": 0.9846,
       "step": 49230
     },
     {
       "epoch": 147.87,
-      "grad_norm": 6.331614017486572,
+      "grad_norm": 5.957528114318848,
       "learning_rate": 5.071071071071072e-06,
-      "loss": 0.9691,
+      "loss": 0.9763,
       "step": 49240
     },
     {
       "epoch": 147.9,
-      "grad_norm": 7.972911357879639,
+      "grad_norm": 6.741492748260498,
       "learning_rate": 5.07007007007007e-06,
-      "loss": 0.942,
+      "loss": 0.9454,
       "step": 49250
     },
     {
       "epoch": 147.93,
-      "grad_norm": 5.582961559295654,
+      "grad_norm": 6.146005153656006,
       "learning_rate": 5.069069069069069e-06,
-      "loss": 1.0858,
+      "loss": 1.074,
       "step": 49260
     },
     {
       "epoch": 147.96,
-      "grad_norm": 5.50705099105835,
+      "grad_norm": 6.6803460121154785,
       "learning_rate": 5.068068068068068e-06,
-      "loss": 0.9771,
+      "loss": 0.9759,
       "step": 49270
     },
     {
       "epoch": 147.99,
-      "grad_norm": 8.457859992980957,
+      "grad_norm": 7.146859169006348,
       "learning_rate": 5.0670670670670675e-06,
-      "loss": 1.0897,
+      "loss": 1.0904,
       "step": 49280
     },
     {
       "epoch": 148.0,
       "eval_accuracy": 0.8179,
-      "eval_loss": 0.6007615923881531,
-      "eval_runtime": 5.4642,
-      "eval_samples_per_second": 1830.109,
-      "eval_steps_per_second": 7.32,
+      "eval_loss": 0.601088285446167,
+      "eval_runtime": 5.31,
+      "eval_samples_per_second": 1883.232,
+      "eval_steps_per_second": 7.533,
       "step": 49284
     },
     {
       "epoch": 148.02,
-      "grad_norm": 5.036593914031982,
+      "grad_norm": 7.465099334716797,
       "learning_rate": 5.066066066066067e-06,
-      "loss": 1.2314,
+      "loss": 1.2298,
       "step": 49290
     },
     {
       "epoch": 148.05,
-      "grad_norm": 7.305020332336426,
+      "grad_norm": 5.94172477722168,
       "learning_rate": 5.0650650650650655e-06,
-      "loss": 0.9801,
+      "loss": 0.9869,
       "step": 49300
     },
     {
       "epoch": 148.08,
-      "grad_norm": 7.2682695388793945,
+      "grad_norm": 6.7936015129089355,
       "learning_rate": 5.064064064064065e-06,
-      "loss": 0.9971,
+      "loss": 0.9999,
       "step": 49310
     },
     {
       "epoch": 148.11,
-      "grad_norm": 7.457651615142822,
+      "grad_norm": 7.728912830352783,
       "learning_rate": 5.0630630630630635e-06,
-      "loss": 1.0083,
+      "loss": 1.0007,
       "step": 49320
     },
     {
       "epoch": 148.14,
-      "grad_norm": 4.852605819702148,
+      "grad_norm": 7.956703186035156,
       "learning_rate": 5.062062062062062e-06,
-      "loss": 0.9655,
+      "loss": 0.9691,
       "step": 49330
     },
     {
       "epoch": 148.17,
-      "grad_norm": 6.206523418426514,
+      "grad_norm": 6.533988952636719,
       "learning_rate": 5.061061061061061e-06,
-      "loss": 1.0671,
+      "loss": 1.0692,
       "step": 49340
     },
     {
       "epoch": 148.2,
-      "grad_norm": 8.389702796936035,
+      "grad_norm": 6.272923469543457,
       "learning_rate": 5.06006006006006e-06,
-      "loss": 1.0504,
+      "loss": 1.0557,
       "step": 49350
     },
     {
       "epoch": 148.23,
-      "grad_norm": 5.724721431732178,
+      "grad_norm": 6.664149761199951,
       "learning_rate": 5.05905905905906e-06,
-      "loss": 0.9757,
+      "loss": 0.9776,
       "step": 49360
     },
     {
       "epoch": 148.26,
-      "grad_norm": 6.39667272567749,
+      "grad_norm": 8.07779598236084,
       "learning_rate": 5.058058058058058e-06,
-      "loss": 1.0491,
+      "loss": 1.0383,
       "step": 49370
     },
     {
       "epoch": 148.29,
-      "grad_norm": 8.481523513793945,
+      "grad_norm": 9.183378219604492,
       "learning_rate": 5.057057057057058e-06,
-      "loss": 1.0402,
+      "loss": 1.0317,
       "step": 49380
     },
     {
       "epoch": 148.32,
-      "grad_norm": 6.667923927307129,
+      "grad_norm": 6.908310413360596,
       "learning_rate": 5.056056056056057e-06,
-      "loss": 0.9674,
+      "loss": 0.9574,
       "step": 49390
     },
     {
       "epoch": 148.35,
-      "grad_norm": 7.46612548828125,
+      "grad_norm": 8.246929168701172,
       "learning_rate": 5.055055055055056e-06,
-      "loss": 1.0203,
+      "loss": 1.0129,
       "step": 49400
     },
     {
       "epoch": 148.38,
-      "grad_norm": 6.404944896697998,
+      "grad_norm": 7.418915271759033,
       "learning_rate": 5.054054054054054e-06,
-      "loss": 1.044,
+      "loss": 1.0515,
       "step": 49410
     },
     {
       "epoch": 148.41,
-      "grad_norm": 6.424440860748291,
+      "grad_norm": 5.934270858764648,
       "learning_rate": 5.053053053053053e-06,
-      "loss": 1.0019,
+      "loss": 1.0038,
       "step": 49420
     },
     {
       "epoch": 148.44,
-      "grad_norm": 4.754261493682861,
+      "grad_norm": 5.210793495178223,
       "learning_rate": 5.052052052052052e-06,
-      "loss": 1.0261,
+      "loss": 1.0338,
       "step": 49430
     },
     {
       "epoch": 148.47,
-      "grad_norm": 5.0437703132629395,
+      "grad_norm": 5.886385440826416,
       "learning_rate": 5.051051051051051e-06,
-      "loss": 1.0521,
+      "loss": 1.0411,
       "step": 49440
     },
     {
       "epoch": 148.5,
-      "grad_norm": 6.232791900634766,
+      "grad_norm": 6.174377918243408,
       "learning_rate": 5.05005005005005e-06,
-      "loss": 1.0779,
+      "loss": 1.0707,
       "step": 49450
     },
     {
       "epoch": 148.53,
-      "grad_norm": 6.752559185028076,
+      "grad_norm": 8.489816665649414,
       "learning_rate": 5.04904904904905e-06,
-      "loss": 0.9846,
+      "loss": 0.9763,
       "step": 49460
     },
     {
       "epoch": 148.56,
-      "grad_norm": 8.516922950744629,
+      "grad_norm": 12.720274925231934,
       "learning_rate": 5.048048048048048e-06,
-      "loss": 1.0392,
+      "loss": 1.0337,
       "step": 49470
     },
     {
       "epoch": 148.59,
-      "grad_norm": 7.973382472991943,
+      "grad_norm": 6.154221534729004,
       "learning_rate": 5.047047047047048e-06,
-      "loss": 0.9997,
+      "loss": 0.9833,
       "step": 49480
     },
     {
       "epoch": 148.62,
-      "grad_norm": 7.49485969543457,
+      "grad_norm": 7.989703178405762,
       "learning_rate": 5.046046046046047e-06,
-      "loss": 1.07,
+      "loss": 1.0687,
       "step": 49490
     },
     {
       "epoch": 148.65,
-      "grad_norm": 5.978147983551025,
+      "grad_norm": 6.362643241882324,
       "learning_rate": 5.045045045045045e-06,
-      "loss": 1.0572,
+      "loss": 1.0566,
       "step": 49500
     },
     {
       "epoch": 148.68,
-      "grad_norm": 5.847923278808594,
+      "grad_norm": 5.628790855407715,
       "learning_rate": 5.0440440440440444e-06,
-      "loss": 1.0307,
+      "loss": 1.0208,
       "step": 49510
     },
     {
       "epoch": 148.71,
-      "grad_norm": 6.187611103057861,
+      "grad_norm": 7.8349504470825195,
       "learning_rate": 5.043043043043043e-06,
-      "loss": 1.0161,
+      "loss": 1.004,
       "step": 49520
     },
     {
       "epoch": 148.74,
-      "grad_norm": 10.301688194274902,
+      "grad_norm": 8.48215103149414,
       "learning_rate": 5.0420420420420425e-06,
-      "loss": 1.0616,
+      "loss": 1.0542,
       "step": 49530
     },
     {
       "epoch": 148.77,
-      "grad_norm": 5.178468704223633,
+      "grad_norm": 5.765330791473389,
       "learning_rate": 5.041041041041042e-06,
-      "loss": 0.9989,
+      "loss": 1.0009,
       "step": 49540
     },
     {
       "epoch": 148.8,
-      "grad_norm": 8.381328582763672,
+      "grad_norm": 7.8314337730407715,
       "learning_rate": 5.0400400400400405e-06,
-      "loss": 1.0506,
+      "loss": 1.0532,
       "step": 49550
     },
     {
       "epoch": 148.83,
-      "grad_norm": 6.929701328277588,
+      "grad_norm": 6.052536964416504,
       "learning_rate": 5.03903903903904e-06,
-      "loss": 0.9741,
+      "loss": 0.9682,
       "step": 49560
     },
     {
       "epoch": 148.86,
-      "grad_norm": 6.30063009262085,
+      "grad_norm": 7.753597736358643,
       "learning_rate": 5.0380380380380386e-06,
-      "loss": 1.0212,
+      "loss": 1.0129,
       "step": 49570
     },
     {
       "epoch": 148.89,
-      "grad_norm": 8.280416488647461,
+      "grad_norm": 9.756672859191895,
       "learning_rate": 5.037037037037037e-06,
-      "loss": 1.0332,
+      "loss": 1.0383,
       "step": 49580
     },
     {
       "epoch": 148.92,
-      "grad_norm": 8.382351875305176,
+      "grad_norm": 9.741190910339355,
       "learning_rate": 5.036036036036036e-06,
-      "loss": 0.9713,
+      "loss": 0.9736,
       "step": 49590
     },
     {
       "epoch": 148.95,
-      "grad_norm": 5.618527889251709,
+      "grad_norm": 4.867576599121094,
       "learning_rate": 5.035035035035035e-06,
-      "loss": 0.9668,
+      "loss": 0.9737,
       "step": 49600
     },
     {
       "epoch": 148.98,
-      "grad_norm": 7.061892509460449,
+      "grad_norm": 8.11778736114502,
       "learning_rate": 5.034034034034035e-06,
-      "loss": 0.9711,
+      "loss": 0.9663,
       "step": 49610
     },
     {
       "epoch": 149.0,
       "eval_accuracy": 0.8148,
-      "eval_loss": 0.6039597392082214,
-      "eval_runtime": 5.6263,
-      "eval_samples_per_second": 1777.353,
-      "eval_steps_per_second": 7.109,
+      "eval_loss": 0.6043193936347961,
+      "eval_runtime": 5.2392,
+      "eval_samples_per_second": 1908.684,
+      "eval_steps_per_second": 7.635,
       "step": 49617
     },
     {
       "epoch": 149.01,
-      "grad_norm": 5.369565010070801,
+      "grad_norm": 7.061316967010498,
       "learning_rate": 5.033033033033033e-06,
-      "loss": 1.2299,
+      "loss": 1.2451,
       "step": 49620
     },
     {
       "epoch": 149.04,
-      "grad_norm": 6.766183376312256,
+      "grad_norm": 6.248096466064453,
       "learning_rate": 5.032032032032033e-06,
-      "loss": 1.078,
+      "loss": 1.0855,
       "step": 49630
     },
     {
       "epoch": 149.07,
-      "grad_norm": 4.377588272094727,
+      "grad_norm": 5.315594673156738,
       "learning_rate": 5.031031031031032e-06,
-      "loss": 1.0317,
+      "loss": 1.0308,
       "step": 49640
     },
     {
       "epoch": 149.1,
-      "grad_norm": 6.283676624298096,
+      "grad_norm": 7.882686614990234,
       "learning_rate": 5.030030030030031e-06,
-      "loss": 1.0362,
+      "loss": 1.0346,
       "step": 49650
     },
     {
       "epoch": 149.13,
-      "grad_norm": 7.964866638183594,
+      "grad_norm": 8.494651794433594,
       "learning_rate": 5.029029029029029e-06,
-      "loss": 0.9777,
+      "loss": 0.9688,
       "step": 49660
     },
     {
       "epoch": 149.16,
-      "grad_norm": 6.615139961242676,
+      "grad_norm": 8.238455772399902,
       "learning_rate": 5.028028028028028e-06,
-      "loss": 0.9902,
+      "loss": 0.9798,
       "step": 49670
     },
     {
       "epoch": 149.19,
-      "grad_norm": 5.706762790679932,
+      "grad_norm": 8.8111572265625,
       "learning_rate": 5.027027027027027e-06,
-      "loss": 0.9884,
+      "loss": 0.9925,
       "step": 49680
     },
     {
       "epoch": 149.22,
-      "grad_norm": 7.272326946258545,
+      "grad_norm": 6.84496545791626,
       "learning_rate": 5.026026026026026e-06,
-      "loss": 1.1129,
+      "loss": 1.1104,
       "step": 49690
     },
     {
       "epoch": 149.25,
-      "grad_norm": 6.612908840179443,
+      "grad_norm": 7.317546367645264,
       "learning_rate": 5.025025025025025e-06,
-      "loss": 1.0862,
+      "loss": 1.0876,
       "step": 49700
     },
     {
       "epoch": 149.28,
-      "grad_norm": 5.6386637687683105,
+      "grad_norm": 7.697225093841553,
       "learning_rate": 5.024024024024025e-06,
-      "loss": 1.0381,
+      "loss": 1.0351,
       "step": 49710
     },
     {
       "epoch": 149.31,
-      "grad_norm": 7.943356513977051,
+      "grad_norm": 8.607976913452148,
       "learning_rate": 5.023023023023023e-06,
-      "loss": 1.0204,
+      "loss": 1.028,
       "step": 49720
     },
     {
       "epoch": 149.34,
-      "grad_norm": 7.46380615234375,
+      "grad_norm": 5.613062858581543,
       "learning_rate": 5.022022022022023e-06,
-      "loss": 0.9606,
+      "loss": 0.9527,
       "step": 49730
     },
     {
       "epoch": 149.37,
-      "grad_norm": 5.869882583618164,
+      "grad_norm": 9.69284439086914,
       "learning_rate": 5.021021021021022e-06,
-      "loss": 1.0111,
+      "loss": 1.0068,
       "step": 49740
     },
     {
       "epoch": 149.4,
-      "grad_norm": 5.742044448852539,
+      "grad_norm": 4.368209362030029,
       "learning_rate": 5.02002002002002e-06,
-      "loss": 1.0614,
+      "loss": 1.0498,
       "step": 49750
     },
     {
       "epoch": 149.43,
-      "grad_norm": 5.3574090003967285,
+      "grad_norm": 5.427814483642578,
       "learning_rate": 5.0190190190190195e-06,
-      "loss": 0.957,
+      "loss": 0.9686,
       "step": 49760
     },
     {
       "epoch": 149.46,
-      "grad_norm": 4.564164161682129,
+      "grad_norm": 5.764521598815918,
       "learning_rate": 5.018018018018018e-06,
-      "loss": 1.0083,
+      "loss": 0.9946,
       "step": 49770
     },
     {
       "epoch": 149.49,
-      "grad_norm": 8.36137866973877,
+      "grad_norm": 10.702657699584961,
       "learning_rate": 5.0170170170170175e-06,
-      "loss": 0.9781,
+      "loss": 0.9724,
       "step": 49780
     },
     {
       "epoch": 149.52,
-      "grad_norm": 6.097189426422119,
+      "grad_norm": 6.182511806488037,
       "learning_rate": 5.016016016016016e-06,
-      "loss": 0.9805,
+      "loss": 0.9835,
       "step": 49790
     },
     {
       "epoch": 149.55,
-      "grad_norm": 7.823583602905273,
+      "grad_norm": 8.335213661193848,
       "learning_rate": 5.0150150150150156e-06,
-      "loss": 1.0113,
+      "loss": 1.0179,
       "step": 49800
     },
     {
       "epoch": 149.58,
-      "grad_norm": 7.214178085327148,
+      "grad_norm": 8.713489532470703,
       "learning_rate": 5.014014014014015e-06,
-      "loss": 1.0495,
+      "loss": 1.0548,
       "step": 49810
     },
     {
       "epoch": 149.61,
-      "grad_norm": 6.4203996658325195,
+      "grad_norm": 7.520198822021484,
       "learning_rate": 5.013013013013014e-06,
-      "loss": 1.0593,
+      "loss": 1.0526,
       "step": 49820
     },
     {
       "epoch": 149.64,
-      "grad_norm": 7.433349609375,
+      "grad_norm": 7.071113109588623,
       "learning_rate": 5.012012012012012e-06,
-      "loss": 1.0016,
+      "loss": 0.9955,
       "step": 49830
     },
     {
       "epoch": 149.67,
-      "grad_norm": 5.667852401733398,
+      "grad_norm": 7.068820953369141,
       "learning_rate": 5.011011011011011e-06,
-      "loss": 0.9954,
+      "loss": 1.0022,
       "step": 49840
     },
     {
       "epoch": 149.7,
-      "grad_norm": 7.209700584411621,
+      "grad_norm": 7.88516902923584,
       "learning_rate": 5.01001001001001e-06,
-      "loss": 1.1069,
+      "loss": 1.0932,
       "step": 49850
     },
     {
       "epoch": 149.73,
-      "grad_norm": 7.555225849151611,
+      "grad_norm": 6.902438640594482,
       "learning_rate": 5.00900900900901e-06,
-      "loss": 0.9891,
+      "loss": 0.9827,
       "step": 49860
     },
     {
       "epoch": 149.76,
-      "grad_norm": 10.19960880279541,
+      "grad_norm": 11.153833389282227,
       "learning_rate": 5.008008008008008e-06,
-      "loss": 0.9818,
+      "loss": 0.9856,
       "step": 49870
     },
     {
       "epoch": 149.79,
-      "grad_norm": 6.00209379196167,
+      "grad_norm": 5.54121732711792,
       "learning_rate": 5.007007007007008e-06,
-      "loss": 0.9872,
+      "loss": 0.978,
       "step": 49880
     },
     {
       "epoch": 149.82,
-      "grad_norm": 6.604346752166748,
+      "grad_norm": 7.746760845184326,
       "learning_rate": 5.006006006006007e-06,
-      "loss": 1.067,
+      "loss": 1.0613,
       "step": 49890
     },
     {
       "epoch": 149.85,
-      "grad_norm": 7.682033538818359,
+      "grad_norm": 7.194839954376221,
       "learning_rate": 5.005005005005006e-06,
-      "loss": 1.0196,
+      "loss": 1.0202,
       "step": 49900
     },
     {
       "epoch": 149.88,
-      "grad_norm": 7.127830982208252,
+      "grad_norm": 8.871628761291504,
       "learning_rate": 5.0040040040040035e-06,
-      "loss": 0.9926,
+      "loss": 0.9832,
       "step": 49910
     },
     {
       "epoch": 149.91,
-      "grad_norm": 6.9543561935424805,
+      "grad_norm": 7.765955448150635,
       "learning_rate": 5.003003003003003e-06,
-      "loss": 0.9899,
+      "loss": 0.9875,
       "step": 49920
     },
     {
       "epoch": 149.94,
-      "grad_norm": 6.896037578582764,
+      "grad_norm": 8.546931266784668,
       "learning_rate": 5.002002002002002e-06,
-      "loss": 1.0404,
+      "loss": 1.0373,
       "step": 49930
     },
     {
       "epoch": 149.97,
-      "grad_norm": 7.643980979919434,
+      "grad_norm": 6.715919017791748,
       "learning_rate": 5.001001001001001e-06,
-      "loss": 0.9772,
+      "loss": 0.9665,
       "step": 49940
     },
     {
       "epoch": 150.0,
-      "grad_norm": 28.41185760498047,
+      "grad_norm": 24.424030303955078,
       "learning_rate": 5e-06,
-      "loss": 1.2002,
+      "loss": 1.1889,
       "step": 49950
     },
     {
       "epoch": 150.0,
-      "eval_accuracy": 0.8162,
-      "eval_loss": 0.6017760038375854,
-      "eval_runtime": 5.7528,
-      "eval_samples_per_second": 1738.288,
-      "eval_steps_per_second": 6.953,
+      "eval_accuracy": 0.8178,
+      "eval_loss": 0.6016015410423279,
+      "eval_runtime": 5.3219,
+      "eval_samples_per_second": 1879.014,
+      "eval_steps_per_second": 7.516,
       "step": 49950
     },
     {
       "epoch": 150.03,
-      "grad_norm": 6.313272953033447,
+      "grad_norm": 5.753154754638672,
       "learning_rate": 4.998998998999e-06,
-      "loss": 1.0308,
+      "loss": 1.036,
       "step": 49960
     },
     {
       "epoch": 150.06,
-      "grad_norm": 10.549605369567871,
+      "grad_norm": 15.868500709533691,
       "learning_rate": 4.9979979979979984e-06,
-      "loss": 1.0514,
+      "loss": 1.0458,
       "step": 49970
     },
     {
       "epoch": 150.09,
-      "grad_norm": 6.077355861663818,
+      "grad_norm": 8.180094718933105,
       "learning_rate": 4.996996996996997e-06,
-      "loss": 1.0565,
+      "loss": 1.0637,
       "step": 49980
     },
     {
       "epoch": 150.12,
-      "grad_norm": 6.404356956481934,
+      "grad_norm": 8.474630355834961,
       "learning_rate": 4.9959959959959965e-06,
-      "loss": 1.0097,
+      "loss": 1.0025,
       "step": 49990
     },
     {
       "epoch": 150.15,
-      "grad_norm": 4.5395684242248535,
+      "grad_norm": 5.93121862411499,
       "learning_rate": 4.994994994994996e-06,
-      "loss": 0.9904,
+      "loss": 0.9892,
       "step": 50000
     },
     {
       "epoch": 150.18,
-      "grad_norm": 6.397698879241943,
+      "grad_norm": 7.828526973724365,
       "learning_rate": 4.9939939939939945e-06,
-      "loss": 1.0086,
+      "loss": 1.0099,
       "step": 50010
     },
     {
       "epoch": 150.21,
-      "grad_norm": 8.358365058898926,
+      "grad_norm": 6.3050761222839355,
       "learning_rate": 4.992992992992993e-06,
-      "loss": 1.0052,
+      "loss": 1.0189,
       "step": 50020
     },
     {
       "epoch": 150.24,
-      "grad_norm": 8.132730484008789,
+      "grad_norm": 7.706366539001465,
       "learning_rate": 4.9919919919919926e-06,
-      "loss": 1.0151,
+      "loss": 1.0071,
       "step": 50030
     },
     {
       "epoch": 150.27,
-      "grad_norm": 7.047395706176758,
+      "grad_norm": 8.104601860046387,
       "learning_rate": 4.990990990990991e-06,
-      "loss": 0.9739,
+      "loss": 0.9827,
       "step": 50040
     },
     {
       "epoch": 150.3,
-      "grad_norm": 6.100387096405029,
+      "grad_norm": 8.834508895874023,
       "learning_rate": 4.989989989989991e-06,
-      "loss": 0.9964,
+      "loss": 1.0041,
       "step": 50050
     },
     {
       "epoch": 150.33,
-      "grad_norm": 6.264103889465332,
+      "grad_norm": 10.258058547973633,
       "learning_rate": 4.988988988988989e-06,
-      "loss": 1.0051,
+      "loss": 0.9959,
       "step": 50060
     },
     {
       "epoch": 150.36,
-      "grad_norm": 5.288430213928223,
+      "grad_norm": 5.857507228851318,
       "learning_rate": 4.987987987987989e-06,
-      "loss": 0.9846,
+      "loss": 0.9953,
       "step": 50070
     },
     {
       "epoch": 150.39,
-      "grad_norm": 7.092759609222412,
+      "grad_norm": 9.147820472717285,
       "learning_rate": 4.986986986986987e-06,
-      "loss": 1.0074,
+      "loss": 0.9975,
       "step": 50080
     },
     {
       "epoch": 150.42,
-      "grad_norm": 6.5676774978637695,
+      "grad_norm": 8.001996994018555,
       "learning_rate": 4.985985985985987e-06,
-      "loss": 0.9492,
+      "loss": 0.9521,
       "step": 50090
     },
     {
       "epoch": 150.45,
-      "grad_norm": 6.799488544464111,
+      "grad_norm": 6.017551422119141,
       "learning_rate": 4.984984984984985e-06,
-      "loss": 1.0115,
+      "loss": 1.0119,
       "step": 50100
     },
     {
       "epoch": 150.48,
-      "grad_norm": 6.813508033752441,
+      "grad_norm": 6.997466087341309,
       "learning_rate": 4.983983983983985e-06,
-      "loss": 1.0033,
+      "loss": 0.992,
       "step": 50110
     },
     {
       "epoch": 150.51,
-      "grad_norm": 8.176109313964844,
+      "grad_norm": 6.256275177001953,
       "learning_rate": 4.982982982982983e-06,
-      "loss": 1.0214,
+      "loss": 1.0161,
       "step": 50120
     },
     {
       "epoch": 150.54,
-      "grad_norm": 6.554351329803467,
+      "grad_norm": 7.474444389343262,
       "learning_rate": 4.981981981981983e-06,
-      "loss": 1.0428,
+      "loss": 1.0461,
       "step": 50130
     },
     {
       "epoch": 150.57,
-      "grad_norm": 6.430649757385254,
+      "grad_norm": 7.203385353088379,
       "learning_rate": 4.980980980980981e-06,
-      "loss": 1.0845,
+      "loss": 1.0708,
       "step": 50140
     },
     {
       "epoch": 150.6,
-      "grad_norm": 6.034136772155762,
+      "grad_norm": 6.876626968383789,
       "learning_rate": 4.97997997997998e-06,
-      "loss": 1.0074,
+      "loss": 1.0016,
       "step": 50150
     },
     {
       "epoch": 150.63,
-      "grad_norm": 6.732151508331299,
+      "grad_norm": 9.291625022888184,
       "learning_rate": 4.978978978978979e-06,
-      "loss": 1.0287,
+      "loss": 1.022,
       "step": 50160
     },
     {
       "epoch": 150.66,
-      "grad_norm": 7.248101234436035,
+      "grad_norm": 8.658099174499512,
       "learning_rate": 4.977977977977979e-06,
-      "loss": 1.0271,
+      "loss": 1.0168,
       "step": 50170
     },
     {
       "epoch": 150.69,
-      "grad_norm": 4.586252689361572,
+      "grad_norm": 6.554454803466797,
       "learning_rate": 4.976976976976977e-06,
-      "loss": 0.9829,
+      "loss": 0.9922,
       "step": 50180
     },
     {
       "epoch": 150.72,
-      "grad_norm": 6.76125955581665,
+      "grad_norm": 6.980246067047119,
       "learning_rate": 4.975975975975976e-06,
-      "loss": 1.0546,
+      "loss": 1.0445,
       "step": 50190
     },
     {
       "epoch": 150.75,
-      "grad_norm": 7.200073719024658,
+      "grad_norm": 7.315950393676758,
       "learning_rate": 4.9749749749749754e-06,
-      "loss": 1.0083,
+      "loss": 1.0053,
       "step": 50200
     },
     {
       "epoch": 150.78,
-      "grad_norm": 6.879418849945068,
+      "grad_norm": 8.948052406311035,
       "learning_rate": 4.973973973973975e-06,
-      "loss": 1.0579,
+      "loss": 1.0681,
       "step": 50210
     },
     {
       "epoch": 150.81,
-      "grad_norm": 8.534775733947754,
+      "grad_norm": 8.698883056640625,
       "learning_rate": 4.9729729729729735e-06,
-      "loss": 1.0351,
+      "loss": 1.0369,
       "step": 50220
     },
     {
       "epoch": 150.84,
-      "grad_norm": 6.408534049987793,
+      "grad_norm": 6.561751365661621,
       "learning_rate": 4.971971971971972e-06,
-      "loss": 0.9826,
+      "loss": 0.9709,
       "step": 50230
     },
     {
       "epoch": 150.87,
-      "grad_norm": 5.353127956390381,
+      "grad_norm": 5.379207611083984,
       "learning_rate": 4.9709709709709715e-06,
-      "loss": 1.033,
+      "loss": 1.0366,
       "step": 50240
     },
     {
       "epoch": 150.9,
-      "grad_norm": 6.137934684753418,
+      "grad_norm": 5.595942974090576,
       "learning_rate": 4.96996996996997e-06,
-      "loss": 1.0612,
+      "loss": 1.0522,
       "step": 50250
     },
     {
       "epoch": 150.93,
-      "grad_norm": 7.479977130889893,
+      "grad_norm": 8.619698524475098,
       "learning_rate": 4.9689689689689696e-06,
-      "loss": 1.0525,
+      "loss": 1.0404,
       "step": 50260
     },
     {
       "epoch": 150.96,
-      "grad_norm": 5.374978542327881,
+      "grad_norm": 6.504605770111084,
       "learning_rate": 4.967967967967968e-06,
-      "loss": 0.9942,
+      "loss": 0.9947,
       "step": 50270
     },
     {
       "epoch": 150.99,
-      "grad_norm": 5.734890937805176,
+      "grad_norm": 8.1761474609375,
       "learning_rate": 4.966966966966968e-06,
-      "loss": 1.0154,
+      "loss": 1.027,
       "step": 50280
     },
     {
       "epoch": 151.0,
-      "eval_accuracy": 0.816,
-      "eval_loss": 0.6042366027832031,
-      "eval_runtime": 5.3965,
-      "eval_samples_per_second": 1853.042,
-      "eval_steps_per_second": 7.412,
+      "eval_accuracy": 0.8166,
+      "eval_loss": 0.6032113432884216,
+      "eval_runtime": 5.4128,
+      "eval_samples_per_second": 1847.467,
+      "eval_steps_per_second": 7.39,
       "step": 50283
     },
     {
       "epoch": 151.02,
-      "grad_norm": 6.569194316864014,
+      "grad_norm": 6.114285945892334,
       "learning_rate": 4.965965965965966e-06,
-      "loss": 1.2526,
+      "loss": 1.2639,
       "step": 50290
     },
     {
       "epoch": 151.05,
-      "grad_norm": 7.375256061553955,
+      "grad_norm": 7.620820045471191,
       "learning_rate": 4.964964964964966e-06,
-      "loss": 1.0095,
+      "loss": 1.0015,
       "step": 50300
     },
     {
       "epoch": 151.08,
-      "grad_norm": 8.47429370880127,
+      "grad_norm": 6.271129608154297,
       "learning_rate": 4.963963963963964e-06,
-      "loss": 1.0199,
+      "loss": 1.0269,
       "step": 50310
     },
     {
       "epoch": 151.11,
-      "grad_norm": 6.220510005950928,
+      "grad_norm": 7.3721489906311035,
       "learning_rate": 4.962962962962964e-06,
-      "loss": 1.0286,
+      "loss": 1.0194,
       "step": 50320
     },
     {
       "epoch": 151.14,
-      "grad_norm": 5.026158809661865,
+      "grad_norm": 7.46964168548584,
       "learning_rate": 4.961961961961962e-06,
-      "loss": 1.0319,
+      "loss": 1.0398,
       "step": 50330
     },
     {
       "epoch": 151.17,
-      "grad_norm": 6.18709135055542,
+      "grad_norm": 7.466350555419922,
       "learning_rate": 4.960960960960962e-06,
-      "loss": 1.0515,
+      "loss": 1.0383,
       "step": 50340
     },
     {
       "epoch": 151.2,
-      "grad_norm": 8.23216724395752,
+      "grad_norm": 8.42910385131836,
       "learning_rate": 4.95995995995996e-06,
-      "loss": 1.001,
+      "loss": 1.0092,
       "step": 50350
     },
     {
       "epoch": 151.23,
-      "grad_norm": 5.924607753753662,
+      "grad_norm": 6.221477031707764,
       "learning_rate": 4.958958958958959e-06,
-      "loss": 0.9416,
+      "loss": 0.9372,
       "step": 50360
     },
     {
       "epoch": 151.26,
-      "grad_norm": 5.823295593261719,
+      "grad_norm": 6.024626731872559,
       "learning_rate": 4.957957957957958e-06,
-      "loss": 1.0424,
+      "loss": 1.0383,
       "step": 50370
     },
     {
       "epoch": 151.29,
-      "grad_norm": 6.396824836730957,
+      "grad_norm": 6.763006687164307,
       "learning_rate": 4.956956956956958e-06,
-      "loss": 1.0031,
+      "loss": 0.9997,
       "step": 50380
     },
     {
       "epoch": 151.32,
-      "grad_norm": 4.121939659118652,
+      "grad_norm": 4.75070858001709,
       "learning_rate": 4.955955955955956e-06,
-      "loss": 0.9526,
+      "loss": 0.9583,
       "step": 50390
     },
     {
       "epoch": 151.35,
-      "grad_norm": 6.356105327606201,
+      "grad_norm": 9.7894287109375,
       "learning_rate": 4.954954954954955e-06,
-      "loss": 0.9583,
+      "loss": 0.9483,
       "step": 50400
     },
     {
       "epoch": 151.38,
-      "grad_norm": 6.585315227508545,
+      "grad_norm": 7.84089994430542,
       "learning_rate": 4.953953953953954e-06,
-      "loss": 1.0752,
+      "loss": 1.0617,
       "step": 50410
     },
     {
       "epoch": 151.41,
-      "grad_norm": 7.496552467346191,
+      "grad_norm": 8.234545707702637,
       "learning_rate": 4.952952952952954e-06,
-      "loss": 1.0307,
+      "loss": 1.0283,
       "step": 50420
     },
     {
       "epoch": 151.44,
-      "grad_norm": 8.961104393005371,
+      "grad_norm": 8.634352684020996,
       "learning_rate": 4.9519519519519524e-06,
-      "loss": 1.0663,
+      "loss": 1.0549,
       "step": 50430
     },
     {
       "epoch": 151.47,
-      "grad_norm": 5.62644100189209,
+      "grad_norm": 7.257065296173096,
       "learning_rate": 4.950950950950951e-06,
-      "loss": 0.9852,
+      "loss": 0.9849,
       "step": 50440
     },
     {
       "epoch": 151.5,
-      "grad_norm": 5.031961917877197,
+      "grad_norm": 5.941311359405518,
       "learning_rate": 4.9499499499499505e-06,
-      "loss": 0.9637,
+      "loss": 0.9675,
       "step": 50450
     },
     {
       "epoch": 151.53,
-      "grad_norm": 6.960273742675781,
+      "grad_norm": 7.050848007202148,
       "learning_rate": 4.94894894894895e-06,
-      "loss": 0.9597,
+      "loss": 0.9671,
       "step": 50460
     },
     {
       "epoch": 151.56,
-      "grad_norm": 5.802542209625244,
+      "grad_norm": 5.288051605224609,
       "learning_rate": 4.947947947947948e-06,
-      "loss": 1.0323,
+      "loss": 1.0236,
       "step": 50470
     },
     {
       "epoch": 151.59,
-      "grad_norm": 5.729872226715088,
+      "grad_norm": 6.931333065032959,
       "learning_rate": 4.946946946946947e-06,
-      "loss": 1.0373,
+      "loss": 1.0413,
       "step": 50480
     },
     {
       "epoch": 151.62,
-      "grad_norm": 6.67117977142334,
+      "grad_norm": 9.080257415771484,
       "learning_rate": 4.9459459459459466e-06,
-      "loss": 0.9753,
+      "loss": 0.9676,
       "step": 50490
     },
     {
       "epoch": 151.65,
-      "grad_norm": 7.2055888175964355,
+      "grad_norm": 6.547423839569092,
       "learning_rate": 4.944944944944945e-06,
-      "loss": 1.0076,
+      "loss": 1.0047,
       "step": 50500
     },
     {
       "epoch": 151.68,
-      "grad_norm": 10.156743049621582,
+      "grad_norm": 11.51767635345459,
       "learning_rate": 4.943943943943945e-06,
-      "loss": 1.0113,
+      "loss": 1.0166,
       "step": 50510
     },
     {
       "epoch": 151.71,
-      "grad_norm": 6.384617328643799,
+      "grad_norm": 6.665674209594727,
       "learning_rate": 4.942942942942943e-06,
-      "loss": 0.9456,
+      "loss": 0.9542,
       "step": 50520
     },
     {
       "epoch": 151.74,
-      "grad_norm": 9.276174545288086,
+      "grad_norm": 10.290655136108398,
       "learning_rate": 4.941941941941943e-06,
-      "loss": 1.0064,
+      "loss": 0.9999,
       "step": 50530
     },
     {
       "epoch": 151.77,
-      "grad_norm": 6.902745723724365,
+      "grad_norm": 6.949934005737305,
       "learning_rate": 4.940940940940941e-06,
-      "loss": 0.99,
+      "loss": 0.9984,
       "step": 50540
     },
     {
       "epoch": 151.8,
-      "grad_norm": 6.558669090270996,
+      "grad_norm": 5.1987996101379395,
       "learning_rate": 4.939939939939941e-06,
-      "loss": 1.0051,
+      "loss": 0.9943,
       "step": 50550
     },
     {
       "epoch": 151.83,
-      "grad_norm": 6.50363826751709,
+      "grad_norm": 6.0841898918151855,
       "learning_rate": 4.938938938938939e-06,
-      "loss": 1.0876,
+      "loss": 1.0878,
       "step": 50560
     },
     {
       "epoch": 151.86,
-      "grad_norm": 5.4437055587768555,
+      "grad_norm": 6.445475101470947,
       "learning_rate": 4.937937937937939e-06,
-      "loss": 0.9162,
+      "loss": 0.9163,
       "step": 50570
     },
     {
       "epoch": 151.89,
-      "grad_norm": 5.418394088745117,
+      "grad_norm": 6.019898891448975,
       "learning_rate": 4.936936936936937e-06,
-      "loss": 1.0353,
+      "loss": 1.0239,
       "step": 50580
     },
     {
       "epoch": 151.92,
-      "grad_norm": 8.711281776428223,
+      "grad_norm": 8.095691680908203,
       "learning_rate": 4.935935935935937e-06,
-      "loss": 1.0042,
+      "loss": 0.9992,
       "step": 50590
     },
     {
       "epoch": 151.95,
-      "grad_norm": 6.9909257888793945,
+      "grad_norm": 6.415546894073486,
       "learning_rate": 4.934934934934935e-06,
-      "loss": 0.9979,
+      "loss": 0.9881,
       "step": 50600
     },
     {
       "epoch": 151.98,
-      "grad_norm": 6.569545269012451,
+      "grad_norm": 5.60078239440918,
       "learning_rate": 4.933933933933934e-06,
-      "loss": 1.0561,
+      "loss": 1.0552,
       "step": 50610
     },
     {
       "epoch": 152.0,
-      "eval_accuracy": 0.8145,
-      "eval_loss": 0.604200005531311,
-      "eval_runtime": 5.3809,
-      "eval_samples_per_second": 1858.41,
-      "eval_steps_per_second": 7.434,
+      "eval_accuracy": 0.814,
+      "eval_loss": 0.6037417054176331,
+      "eval_runtime": 5.5859,
+      "eval_samples_per_second": 1790.225,
+      "eval_steps_per_second": 7.161,
       "step": 50616
     },
     {
       "epoch": 152.01,
-      "grad_norm": 5.63825798034668,
+      "grad_norm": 7.979721546173096,
       "learning_rate": 4.932932932932933e-06,
-      "loss": 1.1767,
+      "loss": 1.2056,
       "step": 50620
     },
     {
       "epoch": 152.04,
-      "grad_norm": 6.049804210662842,
+      "grad_norm": 7.742903232574463,
       "learning_rate": 4.931931931931933e-06,
-      "loss": 0.9959,
+      "loss": 0.9906,
       "step": 50630
     },
     {
       "epoch": 152.07,
-      "grad_norm": 4.825001239776611,
+      "grad_norm": 4.953359127044678,
       "learning_rate": 4.930930930930931e-06,
-      "loss": 1.014,
+      "loss": 1.0061,
       "step": 50640
     },
     {
       "epoch": 152.1,
-      "grad_norm": 6.174343585968018,
+      "grad_norm": 9.53330135345459,
       "learning_rate": 4.92992992992993e-06,
-      "loss": 0.9776,
+      "loss": 0.9887,
       "step": 50650
     },
     {
       "epoch": 152.13,
-      "grad_norm": 6.569102764129639,
+      "grad_norm": 6.8419718742370605,
       "learning_rate": 4.9289289289289294e-06,
-      "loss": 1.0252,
+      "loss": 1.0291,
       "step": 50660
     },
     {
       "epoch": 152.16,
-      "grad_norm": 7.379744529724121,
+      "grad_norm": 7.053258895874023,
       "learning_rate": 4.927927927927929e-06,
-      "loss": 0.987,
+      "loss": 0.9895,
       "step": 50670
     },
     {
       "epoch": 152.19,
-      "grad_norm": 5.6738409996032715,
+      "grad_norm": 7.096541404724121,
       "learning_rate": 4.9269269269269275e-06,
-      "loss": 0.9937,
+      "loss": 0.9946,
       "step": 50680
     },
     {
       "epoch": 152.22,
-      "grad_norm": 8.35814380645752,
+      "grad_norm": 7.7178449630737305,
       "learning_rate": 4.925925925925926e-06,
-      "loss": 0.9972,
+      "loss": 1.0042,
       "step": 50690
     },
     {
       "epoch": 152.25,
-      "grad_norm": 5.863307952880859,
+      "grad_norm": 6.677862167358398,
       "learning_rate": 4.9249249249249255e-06,
-      "loss": 1.0337,
+      "loss": 1.0358,
       "step": 50700
     },
     {
       "epoch": 152.28,
-      "grad_norm": 7.763482093811035,
+      "grad_norm": 10.774595260620117,
       "learning_rate": 4.923923923923924e-06,
-      "loss": 0.9503,
+      "loss": 0.9508,
       "step": 50710
     },
     {
       "epoch": 152.31,
-      "grad_norm": 5.8146514892578125,
+      "grad_norm": 7.783036231994629,
       "learning_rate": 4.922922922922923e-06,
-      "loss": 0.9724,
+      "loss": 0.9764,
       "step": 50720
     },
     {
       "epoch": 152.34,
-      "grad_norm": 6.161673545837402,
+      "grad_norm": 7.065697193145752,
       "learning_rate": 4.921921921921922e-06,
-      "loss": 1.0115,
+      "loss": 1.0111,
       "step": 50730
     },
     {
       "epoch": 152.37,
-      "grad_norm": 7.292535781860352,
+      "grad_norm": 8.572538375854492,
       "learning_rate": 4.920920920920922e-06,
-      "loss": 0.9236,
+      "loss": 0.9267,
       "step": 50740
     },
     {
       "epoch": 152.4,
-      "grad_norm": 8.74374008178711,
+      "grad_norm": 7.7647600173950195,
       "learning_rate": 4.91991991991992e-06,
-      "loss": 1.0357,
+      "loss": 1.0283,
       "step": 50750
     },
     {
       "epoch": 152.43,
-      "grad_norm": 5.612797260284424,
+      "grad_norm": 6.714043140411377,
       "learning_rate": 4.91891891891892e-06,
-      "loss": 1.0037,
+      "loss": 1.0147,
       "step": 50760
     },
     {
       "epoch": 152.46,
-      "grad_norm": 7.751625061035156,
+      "grad_norm": 8.270740509033203,
       "learning_rate": 4.917917917917918e-06,
-      "loss": 1.0008,
+      "loss": 1.0061,
       "step": 50770
     },
     {
       "epoch": 152.49,
-      "grad_norm": 5.007915496826172,
+      "grad_norm": 6.785730361938477,
       "learning_rate": 4.916916916916918e-06,
-      "loss": 0.9371,
+      "loss": 0.9328,
       "step": 50780
     },
     {
       "epoch": 152.52,
-      "grad_norm": 4.852046489715576,
+      "grad_norm": 5.627216339111328,
       "learning_rate": 4.915915915915916e-06,
-      "loss": 0.9738,
+      "loss": 0.9859,
       "step": 50790
     },
     {
       "epoch": 152.55,
-      "grad_norm": 7.54857063293457,
+      "grad_norm": 7.464947700500488,
       "learning_rate": 4.914914914914916e-06,
-      "loss": 0.994,
+      "loss": 0.9854,
       "step": 50800
     },
     {
       "epoch": 152.58,
-      "grad_norm": 8.757441520690918,
+      "grad_norm": 6.873262405395508,
       "learning_rate": 4.913913913913914e-06,
-      "loss": 0.9913,
+      "loss": 0.9867,
       "step": 50810
     },
     {
       "epoch": 152.61,
-      "grad_norm": 5.582734107971191,
+      "grad_norm": 7.448166847229004,
       "learning_rate": 4.912912912912913e-06,
-      "loss": 1.0006,
+      "loss": 0.9921,
       "step": 50820
     },
     {
       "epoch": 152.64,
-      "grad_norm": 5.342665672302246,
+      "grad_norm": 4.908132553100586,
       "learning_rate": 4.911911911911912e-06,
-      "loss": 1.0069,
+      "loss": 1.0047,
       "step": 50830
     },
     {
       "epoch": 152.67,
-      "grad_norm": 6.294168472290039,
+      "grad_norm": 7.57865047454834,
       "learning_rate": 4.910910910910912e-06,
-      "loss": 1.0094,
+      "loss": 1.0058,
       "step": 50840
     },
     {
       "epoch": 152.7,
-      "grad_norm": 7.759007453918457,
+      "grad_norm": 6.0419921875,
       "learning_rate": 4.90990990990991e-06,
-      "loss": 1.0124,
+      "loss": 1.003,
       "step": 50850
     },
     {
       "epoch": 152.73,
-      "grad_norm": 4.5836100578308105,
+      "grad_norm": 6.877859592437744,
       "learning_rate": 4.908908908908909e-06,
       "loss": 1.0135,
       "step": 50860
     },
     {
       "epoch": 152.76,
-      "grad_norm": 8.173589706420898,
+      "grad_norm": 7.80710506439209,
       "learning_rate": 4.907907907907908e-06,
-      "loss": 1.054,
+      "loss": 1.0489,
       "step": 50870
     },
     {
       "epoch": 152.79,
-      "grad_norm": 6.8938889503479,
+      "grad_norm": 8.753059387207031,
       "learning_rate": 4.906906906906908e-06,
-      "loss": 1.0783,
+      "loss": 1.0763,
       "step": 50880
     },
     {
       "epoch": 152.82,
-      "grad_norm": 5.351250648498535,
+      "grad_norm": 5.94732666015625,
       "learning_rate": 4.9059059059059064e-06,
-      "loss": 0.9685,
+      "loss": 0.9644,
       "step": 50890
     },
     {
       "epoch": 152.85,
-      "grad_norm": 5.2580084800720215,
+      "grad_norm": 6.15539026260376,
       "learning_rate": 4.904904904904905e-06,
-      "loss": 0.9815,
+      "loss": 0.9749,
       "step": 50900
     },
     {
       "epoch": 152.88,
-      "grad_norm": 8.68312931060791,
+      "grad_norm": 7.8478827476501465,
       "learning_rate": 4.9039039039039045e-06,
-      "loss": 1.0313,
+      "loss": 1.0364,
       "step": 50910
     },
     {
       "epoch": 152.91,
-      "grad_norm": 6.989731788635254,
+      "grad_norm": 6.619234561920166,
       "learning_rate": 4.902902902902904e-06,
-      "loss": 0.9282,
+      "loss": 0.922,
       "step": 50920
     },
     {
       "epoch": 152.94,
-      "grad_norm": 9.469709396362305,
+      "grad_norm": 7.882242679595947,
       "learning_rate": 4.901901901901902e-06,
-      "loss": 0.9958,
+      "loss": 0.9938,
       "step": 50930
     },
     {
       "epoch": 152.97,
-      "grad_norm": 5.6254425048828125,
+      "grad_norm": 5.993936061859131,
       "learning_rate": 4.900900900900901e-06,
-      "loss": 0.9962,
+      "loss": 1.0057,
       "step": 50940
     },
     {
       "epoch": 153.0,
-      "eval_accuracy": 0.8158,
-      "eval_loss": 0.6010604500770569,
-      "eval_runtime": 5.4701,
-      "eval_samples_per_second": 1828.119,
-      "eval_steps_per_second": 7.312,
+      "eval_accuracy": 0.8147,
+      "eval_loss": 0.5999048948287964,
+      "eval_runtime": 5.2218,
+      "eval_samples_per_second": 1915.033,
+      "eval_steps_per_second": 7.66,
       "step": 50949
     },
     {
       "epoch": 153.0,
-      "grad_norm": 6.5960822105407715,
+      "grad_norm": 6.856289386749268,
       "learning_rate": 4.8998998998999005e-06,
-      "loss": 1.1069,
+      "loss": 1.0788,
       "step": 50950
     },
     {
       "epoch": 153.03,
-      "grad_norm": 5.506674766540527,
+      "grad_norm": 6.139232635498047,
       "learning_rate": 4.898898898898899e-06,
-      "loss": 1.0073,
+      "loss": 1.0072,
       "step": 50960
     },
     {
       "epoch": 153.06,
-      "grad_norm": 4.498945236206055,
+      "grad_norm": 5.683094024658203,
       "learning_rate": 4.897897897897898e-06,
-      "loss": 0.9422,
+      "loss": 0.9416,
       "step": 50970
     },
     {
       "epoch": 153.09,
-      "grad_norm": 7.3863701820373535,
+      "grad_norm": 6.750290393829346,
       "learning_rate": 4.896896896896897e-06,
-      "loss": 0.9712,
+      "loss": 0.9591,
       "step": 50980
     },
     {
       "epoch": 153.12,
-      "grad_norm": 8.422163963317871,
+      "grad_norm": 6.275930881500244,
       "learning_rate": 4.895895895895897e-06,
-      "loss": 0.9944,
+      "loss": 0.9834,
       "step": 50990
     },
     {
       "epoch": 153.15,
-      "grad_norm": 6.241824626922607,
+      "grad_norm": 6.569389343261719,
       "learning_rate": 4.894894894894895e-06,
-      "loss": 1.0215,
+      "loss": 1.0198,
       "step": 51000
     },
     {
       "epoch": 153.18,
-      "grad_norm": 6.329225063323975,
+      "grad_norm": 7.830654144287109,
       "learning_rate": 4.893893893893894e-06,
-      "loss": 1.0327,
+      "loss": 1.0367,
       "step": 51010
     },
     {
       "epoch": 153.21,
-      "grad_norm": 7.1616530418396,
+      "grad_norm": 5.247340202331543,
       "learning_rate": 4.892892892892893e-06,
-      "loss": 0.9404,
+      "loss": 0.9379,
       "step": 51020
     },
     {
       "epoch": 153.24,
-      "grad_norm": 6.173557281494141,
+      "grad_norm": 9.432572364807129,
       "learning_rate": 4.891891891891893e-06,
-      "loss": 0.8953,
+      "loss": 0.8892,
       "step": 51030
     },
     {
       "epoch": 153.27,
-      "grad_norm": 6.686371803283691,
+      "grad_norm": 5.616205215454102,
       "learning_rate": 4.890890890890891e-06,
-      "loss": 0.9733,
+      "loss": 0.9745,
       "step": 51040
     },
     {
       "epoch": 153.3,
-      "grad_norm": 6.922730445861816,
+      "grad_norm": 5.669053554534912,
       "learning_rate": 4.889889889889891e-06,
-      "loss": 1.0058,
+      "loss": 0.9984,
       "step": 51050
     },
     {
       "epoch": 153.33,
-      "grad_norm": 8.147199630737305,
+      "grad_norm": 6.314888000488281,
       "learning_rate": 4.888888888888889e-06,
-      "loss": 0.9972,
+      "loss": 0.999,
       "step": 51060
     },
     {
       "epoch": 153.36,
-      "grad_norm": 7.199734210968018,
+      "grad_norm": 6.279026985168457,
       "learning_rate": 4.887887887887888e-06,
-      "loss": 0.9941,
+      "loss": 1.0078,
       "step": 51070
     },
     {
       "epoch": 153.39,
-      "grad_norm": 5.579361438751221,
+      "grad_norm": 5.640705585479736,
       "learning_rate": 4.886886886886887e-06,
-      "loss": 1.0465,
+      "loss": 1.0337,
       "step": 51080
     },
     {
       "epoch": 153.42,
-      "grad_norm": 7.710988998413086,
+      "grad_norm": 7.853842258453369,
       "learning_rate": 4.885885885885887e-06,
-      "loss": 0.9895,
+      "loss": 0.9792,
       "step": 51090
     },
     {
       "epoch": 153.45,
-      "grad_norm": 11.973464012145996,
+      "grad_norm": 11.211026191711426,
       "learning_rate": 4.884884884884885e-06,
-      "loss": 1.0385,
+      "loss": 1.0467,
       "step": 51100
     },
     {
       "epoch": 153.48,
-      "grad_norm": 7.851558685302734,
+      "grad_norm": 7.1522440910339355,
       "learning_rate": 4.883883883883884e-06,
-      "loss": 1.0164,
+      "loss": 1.0087,
       "step": 51110
     },
     {
       "epoch": 153.51,
-      "grad_norm": 6.300526142120361,
+      "grad_norm": 7.2842936515808105,
       "learning_rate": 4.8828828828828834e-06,
-      "loss": 1.0717,
+      "loss": 1.0643,
       "step": 51120
     },
     {
       "epoch": 153.54,
-      "grad_norm": 10.198640823364258,
+      "grad_norm": 10.566871643066406,
       "learning_rate": 4.881881881881883e-06,
-      "loss": 1.0277,
+      "loss": 1.0217,
       "step": 51130
     },
     {
       "epoch": 153.57,
-      "grad_norm": 6.672577857971191,
+      "grad_norm": 6.727398872375488,
       "learning_rate": 4.8808808808808815e-06,
-      "loss": 0.9579,
+      "loss": 0.958,
       "step": 51140
     },
     {
       "epoch": 153.6,
-      "grad_norm": 6.203878879547119,
+      "grad_norm": 6.5836710929870605,
       "learning_rate": 4.87987987987988e-06,
-      "loss": 0.9812,
+      "loss": 0.9682,
       "step": 51150
     },
     {
       "epoch": 153.63,
-      "grad_norm": 9.531362533569336,
+      "grad_norm": 9.577075004577637,
       "learning_rate": 4.8788788788788795e-06,
-      "loss": 1.0055,
+      "loss": 0.9981,
       "step": 51160
     },
     {
       "epoch": 153.66,
-      "grad_norm": 4.898408889770508,
+      "grad_norm": 4.939971923828125,
       "learning_rate": 4.877877877877878e-06,
-      "loss": 1.0219,
+      "loss": 1.0183,
       "step": 51170
     },
     {
       "epoch": 153.69,
-      "grad_norm": 7.827752590179443,
+      "grad_norm": 9.52396297454834,
       "learning_rate": 4.876876876876877e-06,
-      "loss": 1.0118,
+      "loss": 1.0113,
       "step": 51180
     },
     {
       "epoch": 153.72,
-      "grad_norm": 6.2607245445251465,
+      "grad_norm": 8.412007331848145,
       "learning_rate": 4.875875875875876e-06,
-      "loss": 1.0064,
+      "loss": 1.0137,
       "step": 51190
     },
     {
       "epoch": 153.75,
-      "grad_norm": 6.221005916595459,
+      "grad_norm": 6.532750606536865,
       "learning_rate": 4.874874874874876e-06,
-      "loss": 1.0058,
+      "loss": 0.9951,
       "step": 51200
     },
     {
       "epoch": 153.78,
-      "grad_norm": 6.909198760986328,
+      "grad_norm": 7.031577110290527,
       "learning_rate": 4.873873873873874e-06,
-      "loss": 1.0169,
+      "loss": 1.0118,
       "step": 51210
     },
     {
       "epoch": 153.81,
-      "grad_norm": 6.6310954093933105,
+      "grad_norm": 6.859035968780518,
       "learning_rate": 4.872872872872873e-06,
-      "loss": 0.919,
+      "loss": 0.9091,
       "step": 51220
     },
     {
       "epoch": 153.84,
-      "grad_norm": 4.579472541809082,
+      "grad_norm": 6.767331600189209,
       "learning_rate": 4.871871871871872e-06,
-      "loss": 0.9989,
+      "loss": 1.0018,
       "step": 51230
     },
     {
       "epoch": 153.87,
-      "grad_norm": 6.547811985015869,
+      "grad_norm": 8.018696784973145,
       "learning_rate": 4.870870870870872e-06,
-      "loss": 0.9967,
+      "loss": 1.0059,
       "step": 51240
     },
     {
       "epoch": 153.9,
-      "grad_norm": 5.590276718139648,
+      "grad_norm": 6.494315147399902,
       "learning_rate": 4.86986986986987e-06,
-      "loss": 0.9718,
+      "loss": 0.9648,
       "step": 51250
     },
     {
       "epoch": 153.93,
-      "grad_norm": 7.278170585632324,
+      "grad_norm": 8.255210876464844,
       "learning_rate": 4.868868868868869e-06,
-      "loss": 0.9661,
+      "loss": 0.9705,
       "step": 51260
     },
     {
       "epoch": 153.96,
-      "grad_norm": 6.62558126449585,
+      "grad_norm": 9.905619621276855,
       "learning_rate": 4.867867867867868e-06,
-      "loss": 0.8984,
+      "loss": 0.9055,
       "step": 51270
     },
     {
       "epoch": 153.99,
-      "grad_norm": 6.18053674697876,
+      "grad_norm": 7.4922261238098145,
       "learning_rate": 4.866866866866867e-06,
-      "loss": 1.0812,
+      "loss": 1.0762,
       "step": 51280
     },
     {
       "epoch": 154.0,
-      "eval_accuracy": 0.8165,
-      "eval_loss": 0.5961253643035889,
-      "eval_runtime": 5.5491,
-      "eval_samples_per_second": 1802.105,
-      "eval_steps_per_second": 7.208,
+      "eval_accuracy": 0.8169,
+      "eval_loss": 0.5963844656944275,
+      "eval_runtime": 5.548,
+      "eval_samples_per_second": 1802.458,
+      "eval_steps_per_second": 7.21,
       "step": 51282
     },
     {
       "epoch": 154.02,
-      "grad_norm": 6.16987419128418,
+      "grad_norm": 5.808726787567139,
       "learning_rate": 4.865865865865866e-06,
-      "loss": 1.1074,
+      "loss": 1.1051,
       "step": 51290
     },
     {
       "epoch": 154.05,
-      "grad_norm": 6.958493709564209,
+      "grad_norm": 6.995981216430664,
       "learning_rate": 4.864864864864866e-06,
-      "loss": 0.9965,
+      "loss": 0.9824,
       "step": 51300
     },
     {
       "epoch": 154.08,
-      "grad_norm": 7.045452117919922,
+      "grad_norm": 9.151211738586426,
       "learning_rate": 4.863863863863864e-06,
-      "loss": 0.9953,
+      "loss": 0.9937,
       "step": 51310
     },
     {
       "epoch": 154.11,
-      "grad_norm": 7.435359954833984,
+      "grad_norm": 11.425180435180664,
       "learning_rate": 4.862862862862863e-06,
-      "loss": 0.976,
+      "loss": 0.9736,
       "step": 51320
     },
     {
       "epoch": 154.14,
-      "grad_norm": 6.2447614669799805,
+      "grad_norm": 6.529719352722168,
       "learning_rate": 4.861861861861862e-06,
-      "loss": 0.973,
+      "loss": 0.972,
       "step": 51330
     },
     {
       "epoch": 154.17,
-      "grad_norm": 5.185611724853516,
+      "grad_norm": 7.025553226470947,
       "learning_rate": 4.860860860860862e-06,
-      "loss": 0.9922,
+      "loss": 0.9912,
       "step": 51340
     },
     {
       "epoch": 154.2,
-      "grad_norm": 6.182654857635498,
+      "grad_norm": 5.870179653167725,
       "learning_rate": 4.8598598598598604e-06,
-      "loss": 1.0761,
+      "loss": 1.0729,
       "step": 51350
     },
     {
       "epoch": 154.23,
-      "grad_norm": 5.654953956604004,
+      "grad_norm": 7.534576416015625,
       "learning_rate": 4.858858858858859e-06,
-      "loss": 0.9554,
+      "loss": 0.957,
       "step": 51360
     },
     {
       "epoch": 154.26,
-      "grad_norm": 9.840909957885742,
+      "grad_norm": 10.797396659851074,
       "learning_rate": 4.8578578578578585e-06,
-      "loss": 0.9903,
+      "loss": 0.9801,
       "step": 51370
     },
     {
       "epoch": 154.29,
-      "grad_norm": 7.616309642791748,
+      "grad_norm": 7.714526176452637,
       "learning_rate": 4.856856856856857e-06,
-      "loss": 0.9976,
+      "loss": 0.9945,
       "step": 51380
     },
     {
       "epoch": 154.32,
-      "grad_norm": 5.156238079071045,
+      "grad_norm": 8.071720123291016,
       "learning_rate": 4.855855855855856e-06,
-      "loss": 1.0055,
+      "loss": 1.0087,
       "step": 51390
     },
     {
       "epoch": 154.35,
-      "grad_norm": 8.555770874023438,
+      "grad_norm": 7.8826727867126465,
       "learning_rate": 4.854854854854855e-06,
-      "loss": 1.034,
+      "loss": 1.0201,
       "step": 51400
     },
     {
       "epoch": 154.38,
-      "grad_norm": 5.73854398727417,
+      "grad_norm": 6.537350654602051,
       "learning_rate": 4.8538538538538545e-06,
-      "loss": 1.0311,
+      "loss": 1.035,
       "step": 51410
     },
     {
       "epoch": 154.41,
-      "grad_norm": 5.652195453643799,
+      "grad_norm": 6.151535511016846,
       "learning_rate": 4.852852852852853e-06,
-      "loss": 1.0473,
+      "loss": 1.039,
       "step": 51420
     },
     {
       "epoch": 154.44,
-      "grad_norm": 5.826671123504639,
+      "grad_norm": 6.378010272979736,
       "learning_rate": 4.851851851851852e-06,
-      "loss": 1.0433,
+      "loss": 1.0452,
       "step": 51430
     },
     {
       "epoch": 154.47,
-      "grad_norm": 6.858035564422607,
+      "grad_norm": 8.243379592895508,
       "learning_rate": 4.850850850850851e-06,
-      "loss": 0.9877,
+      "loss": 0.9867,
       "step": 51440
     },
     {
       "epoch": 154.5,
-      "grad_norm": 6.211648464202881,
+      "grad_norm": 8.175056457519531,
       "learning_rate": 4.849849849849851e-06,
-      "loss": 0.9743,
+      "loss": 0.979,
       "step": 51450
     },
     {
       "epoch": 154.53,
-      "grad_norm": 11.158310890197754,
+      "grad_norm": 10.057835578918457,
       "learning_rate": 4.848848848848849e-06,
-      "loss": 0.9908,
+      "loss": 0.9755,
       "step": 51460
     },
     {
       "epoch": 154.56,
-      "grad_norm": 4.89386510848999,
+      "grad_norm": 6.073917388916016,
       "learning_rate": 4.847847847847848e-06,
-      "loss": 1.0531,
+      "loss": 1.0416,
       "step": 51470
     },
     {
       "epoch": 154.59,
-      "grad_norm": 6.144166946411133,
+      "grad_norm": 5.73057222366333,
       "learning_rate": 4.846846846846847e-06,
-      "loss": 0.9886,
+      "loss": 0.9869,
       "step": 51480
     },
     {
       "epoch": 154.62,
-      "grad_norm": 7.113414287567139,
+      "grad_norm": 5.986699104309082,
       "learning_rate": 4.845845845845847e-06,
-      "loss": 1.0346,
+      "loss": 1.0424,
       "step": 51490
     },
     {
       "epoch": 154.65,
-      "grad_norm": 6.389400959014893,
+      "grad_norm": 7.135775089263916,
       "learning_rate": 4.844844844844845e-06,
-      "loss": 0.9954,
+      "loss": 0.9984,
       "step": 51500
     },
     {
       "epoch": 154.68,
-      "grad_norm": 7.132532119750977,
+      "grad_norm": 7.872323513031006,
       "learning_rate": 4.843843843843844e-06,
-      "loss": 1.0073,
+      "loss": 1.0154,
       "step": 51510
     },
     {
       "epoch": 154.71,
-      "grad_norm": 6.453092098236084,
+      "grad_norm": 8.081791877746582,
       "learning_rate": 4.842842842842843e-06,
-      "loss": 1.0406,
+      "loss": 1.0253,
       "step": 51520
     },
     {
       "epoch": 154.74,
-      "grad_norm": 5.87326192855835,
+      "grad_norm": 5.587206840515137,
       "learning_rate": 4.841841841841842e-06,
-      "loss": 0.9971,
+      "loss": 0.9857,
       "step": 51530
     },
     {
       "epoch": 154.77,
-      "grad_norm": 6.4198479652404785,
+      "grad_norm": 7.478949546813965,
       "learning_rate": 4.840840840840841e-06,
-      "loss": 0.9854,
+      "loss": 0.9777,
       "step": 51540
     },
     {
       "epoch": 154.8,
-      "grad_norm": 5.421718120574951,
+      "grad_norm": 6.158984661102295,
       "learning_rate": 4.83983983983984e-06,
-      "loss": 0.9752,
+      "loss": 0.9845,
       "step": 51550
     },
     {
       "epoch": 154.83,
-      "grad_norm": 7.492892265319824,
+      "grad_norm": 8.400362014770508,
       "learning_rate": 4.838838838838839e-06,
-      "loss": 0.947,
+      "loss": 0.9456,
       "step": 51560
     },
     {
       "epoch": 154.86,
-      "grad_norm": 5.219449520111084,
+      "grad_norm": 5.41246223449707,
       "learning_rate": 4.837837837837838e-06,
-      "loss": 0.9669,
+      "loss": 0.9673,
       "step": 51570
     },
     {
       "epoch": 154.89,
-      "grad_norm": 5.294669151306152,
+      "grad_norm": 5.934901714324951,
       "learning_rate": 4.8368368368368374e-06,
-      "loss": 0.9993,
+      "loss": 0.9925,
       "step": 51580
     },
     {
       "epoch": 154.92,
-      "grad_norm": 5.123594760894775,
+      "grad_norm": 6.750782489776611,
       "learning_rate": 4.835835835835837e-06,
-      "loss": 0.9777,
+      "loss": 0.9775,
       "step": 51590
     },
     {
       "epoch": 154.95,
-      "grad_norm": 6.843083381652832,
+      "grad_norm": 6.525670528411865,
       "learning_rate": 4.8348348348348355e-06,
-      "loss": 1.0362,
+      "loss": 1.0389,
       "step": 51600
     },
     {
       "epoch": 154.98,
-      "grad_norm": 7.725347995758057,
+      "grad_norm": 8.068577766418457,
       "learning_rate": 4.833833833833834e-06,
-      "loss": 1.0307,
+      "loss": 1.034,
       "step": 51610
     },
     {
       "epoch": 155.0,
-      "eval_accuracy": 0.8152,
-      "eval_loss": 0.6053746342658997,
-      "eval_runtime": 5.6176,
-      "eval_samples_per_second": 1780.132,
-      "eval_steps_per_second": 7.121,
+      "eval_accuracy": 0.8157,
+      "eval_loss": 0.604640543460846,
+      "eval_runtime": 5.3681,
+      "eval_samples_per_second": 1862.853,
+      "eval_steps_per_second": 7.451,
       "step": 51615
     },
     {
       "epoch": 155.02,
-      "grad_norm": 4.521846294403076,
+      "grad_norm": 5.752431869506836,
       "learning_rate": 4.8328328328328335e-06,
-      "loss": 1.1685,
+      "loss": 1.1934,
       "step": 51620
     },
     {
       "epoch": 155.05,
-      "grad_norm": 5.575931549072266,
+      "grad_norm": 4.871572971343994,
       "learning_rate": 4.831831831831832e-06,
-      "loss": 1.0564,
+      "loss": 1.0556,
       "step": 51630
     },
     {
       "epoch": 155.08,
-      "grad_norm": 7.399127960205078,
+      "grad_norm": 8.1824951171875,
       "learning_rate": 4.830830830830831e-06,
-      "loss": 1.0027,
+      "loss": 1.0022,
       "step": 51640
     },
     {
       "epoch": 155.11,
-      "grad_norm": 5.823172569274902,
+      "grad_norm": 7.27178955078125,
       "learning_rate": 4.82982982982983e-06,
-      "loss": 1.0566,
+      "loss": 1.0526,
       "step": 51650
     },
     {
       "epoch": 155.14,
-      "grad_norm": 5.295358657836914,
+      "grad_norm": 9.185352325439453,
       "learning_rate": 4.82882882882883e-06,
-      "loss": 1.0187,
+      "loss": 1.0171,
       "step": 51660
     },
     {
       "epoch": 155.17,
-      "grad_norm": 8.864603042602539,
+      "grad_norm": 6.562739372253418,
       "learning_rate": 4.827827827827828e-06,
-      "loss": 0.9914,
+      "loss": 1.0014,
       "step": 51670
     },
     {
       "epoch": 155.2,
-      "grad_norm": 5.337499618530273,
+      "grad_norm": 5.874017238616943,
       "learning_rate": 4.826826826826827e-06,
-      "loss": 1.0533,
+      "loss": 1.0538,
       "step": 51680
     },
     {
       "epoch": 155.23,
-      "grad_norm": 6.414247512817383,
+      "grad_norm": 7.728109836578369,
       "learning_rate": 4.825825825825826e-06,
-      "loss": 0.9798,
+      "loss": 0.9727,
       "step": 51690
     },
     {
       "epoch": 155.26,
-      "grad_norm": 7.4430460929870605,
+      "grad_norm": 7.370913028717041,
       "learning_rate": 4.824824824824826e-06,
-      "loss": 1.0022,
+      "loss": 1.0135,
       "step": 51700
     },
     {
       "epoch": 155.29,
-      "grad_norm": 6.7195611000061035,
+      "grad_norm": 7.097131729125977,
       "learning_rate": 4.823823823823824e-06,
-      "loss": 1.0399,
+      "loss": 1.0296,
       "step": 51710
     },
     {
       "epoch": 155.32,
-      "grad_norm": 5.197562217712402,
+      "grad_norm": 6.29583740234375,
       "learning_rate": 4.822822822822823e-06,
-      "loss": 1.0284,
+      "loss": 1.0207,
       "step": 51720
     },
     {
       "epoch": 155.35,
-      "grad_norm": 7.139791965484619,
+      "grad_norm": 6.176142692565918,
       "learning_rate": 4.821821821821822e-06,
-      "loss": 0.9967,
+      "loss": 0.9966,
       "step": 51730
     },
     {
       "epoch": 155.38,
-      "grad_norm": 6.984912872314453,
+      "grad_norm": 8.931207656860352,
       "learning_rate": 4.820820820820821e-06,
-      "loss": 1.0399,
+      "loss": 1.0375,
       "step": 51740
     },
     {
       "epoch": 155.41,
-      "grad_norm": 5.453206539154053,
+      "grad_norm": 4.5027642250061035,
       "learning_rate": 4.81981981981982e-06,
-      "loss": 1.0318,
+      "loss": 1.0232,
       "step": 51750
     },
     {
       "epoch": 155.44,
-      "grad_norm": 6.336913585662842,
+      "grad_norm": 5.721016883850098,
       "learning_rate": 4.818818818818819e-06,
-      "loss": 1.0049,
+      "loss": 1.0094,
       "step": 51760
     },
     {
       "epoch": 155.47,
-      "grad_norm": 6.198503494262695,
+      "grad_norm": 5.486338138580322,
       "learning_rate": 4.817817817817818e-06,
-      "loss": 0.942,
+      "loss": 0.9368,
       "step": 51770
     },
     {
       "epoch": 155.5,
-      "grad_norm": 5.998616695404053,
+      "grad_norm": 8.093971252441406,
       "learning_rate": 4.816816816816817e-06,
-      "loss": 1.0051,
+      "loss": 1.0096,
       "step": 51780
     },
     {
       "epoch": 155.53,
-      "grad_norm": 5.325040340423584,
+      "grad_norm": 6.5227227210998535,
       "learning_rate": 4.815815815815816e-06,
-      "loss": 0.952,
+      "loss": 0.9539,
       "step": 51790
     },
     {
       "epoch": 155.56,
-      "grad_norm": 7.27346134185791,
+      "grad_norm": 7.953677654266357,
       "learning_rate": 4.814814814814815e-06,
-      "loss": 0.9894,
+      "loss": 0.9826,
       "step": 51800
     },
     {
       "epoch": 155.59,
-      "grad_norm": 6.727331161499023,
+      "grad_norm": 7.629272937774658,
       "learning_rate": 4.8138138138138144e-06,
-      "loss": 0.9414,
+      "loss": 0.9435,
       "step": 51810
     },
     {
       "epoch": 155.62,
-      "grad_norm": 5.022841453552246,
+      "grad_norm": 6.678920269012451,
       "learning_rate": 4.812812812812813e-06,
-      "loss": 0.9403,
+      "loss": 0.9363,
       "step": 51820
     },
     {
       "epoch": 155.65,
-      "grad_norm": 6.972978591918945,
+      "grad_norm": 6.086242198944092,
       "learning_rate": 4.8118118118118125e-06,
-      "loss": 0.9806,
+      "loss": 0.9749,
       "step": 51830
     },
     {
       "epoch": 155.68,
-      "grad_norm": 6.038856506347656,
+      "grad_norm": 6.85850191116333,
       "learning_rate": 4.810810810810811e-06,
-      "loss": 0.9303,
+      "loss": 0.9202,
       "step": 51840
     },
     {
       "epoch": 155.71,
-      "grad_norm": 6.96592903137207,
+      "grad_norm": 9.081768035888672,
       "learning_rate": 4.80980980980981e-06,
-      "loss": 0.9912,
+      "loss": 0.9903,
       "step": 51850
     },
     {
       "epoch": 155.74,
-      "grad_norm": 4.315644264221191,
+      "grad_norm": 5.6754584312438965,
       "learning_rate": 4.808808808808809e-06,
-      "loss": 0.988,
+      "loss": 0.9899,
       "step": 51860
     },
     {
       "epoch": 155.77,
-      "grad_norm": 4.274169921875,
+      "grad_norm": 7.143518447875977,
       "learning_rate": 4.8078078078078085e-06,
-      "loss": 0.9991,
+      "loss": 1.0039,
       "step": 51870
     },
     {
       "epoch": 155.8,
-      "grad_norm": 7.020811557769775,
+      "grad_norm": 6.227580547332764,
       "learning_rate": 4.806806806806807e-06,
-      "loss": 1.0282,
+      "loss": 1.0267,
       "step": 51880
     },
     {
       "epoch": 155.83,
-      "grad_norm": 7.978997707366943,
+      "grad_norm": 7.493308067321777,
       "learning_rate": 4.805805805805806e-06,
-      "loss": 1.0183,
+      "loss": 1.0027,
       "step": 51890
     },
     {
       "epoch": 155.86,
-      "grad_norm": 6.36984395980835,
+      "grad_norm": 6.015876293182373,
       "learning_rate": 4.804804804804805e-06,
-      "loss": 0.971,
+      "loss": 0.9657,
       "step": 51900
     },
     {
       "epoch": 155.89,
-      "grad_norm": 6.47339391708374,
+      "grad_norm": 5.510498046875,
       "learning_rate": 4.803803803803805e-06,
-      "loss": 1.0555,
+      "loss": 1.0438,
       "step": 51910
     },
     {
       "epoch": 155.92,
-      "grad_norm": 6.6458234786987305,
+      "grad_norm": 5.870835781097412,
       "learning_rate": 4.802802802802803e-06,
-      "loss": 1.0372,
+      "loss": 1.0284,
       "step": 51920
     },
     {
       "epoch": 155.95,
-      "grad_norm": 6.438012599945068,
+      "grad_norm": 5.911910533905029,
       "learning_rate": 4.801801801801802e-06,
-      "loss": 0.9595,
+      "loss": 0.9606,
       "step": 51930
     },
     {
       "epoch": 155.98,
-      "grad_norm": 7.4502339363098145,
+      "grad_norm": 6.464371681213379,
       "learning_rate": 4.800800800800801e-06,
-      "loss": 0.991,
+      "loss": 0.9891,
       "step": 51940
     },
     {
       "epoch": 156.0,
-      "eval_accuracy": 0.814,
-      "eval_loss": 0.6018582582473755,
-      "eval_runtime": 5.6697,
-      "eval_samples_per_second": 1763.758,
-      "eval_steps_per_second": 7.055,
+      "eval_accuracy": 0.8166,
+      "eval_loss": 0.6016373038291931,
+      "eval_runtime": 5.4515,
+      "eval_samples_per_second": 1834.351,
+      "eval_steps_per_second": 7.337,
       "step": 51948
     },
     {
       "epoch": 156.01,
-      "grad_norm": 6.3477630615234375,
+      "grad_norm": 6.430293083190918,
       "learning_rate": 4.7997997997998e-06,
-      "loss": 1.1036,
+      "loss": 1.1156,
       "step": 51950
     },
     {
       "epoch": 156.04,
-      "grad_norm": 4.333883762359619,
+      "grad_norm": 5.63566255569458,
       "learning_rate": 4.798798798798799e-06,
-      "loss": 0.9679,
+      "loss": 0.9688,
       "step": 51960
     },
     {
       "epoch": 156.07,
-      "grad_norm": 6.10819149017334,
+      "grad_norm": 6.683868885040283,
       "learning_rate": 4.797797797797798e-06,
-      "loss": 0.978,
+      "loss": 0.9679,
       "step": 51970
     },
     {
       "epoch": 156.1,
-      "grad_norm": 6.027104377746582,
+      "grad_norm": 6.855597496032715,
       "learning_rate": 4.796796796796797e-06,
-      "loss": 0.9774,
+      "loss": 0.9638,
       "step": 51980
     },
     {
       "epoch": 156.13,
-      "grad_norm": 5.568253040313721,
+      "grad_norm": 4.73078727722168,
       "learning_rate": 4.795795795795796e-06,
-      "loss": 1.0135,
+      "loss": 1.0119,
       "step": 51990
     },
     {
       "epoch": 156.16,
-      "grad_norm": 6.79562520980835,
+      "grad_norm": 7.463665962219238,
       "learning_rate": 4.794794794794795e-06,
-      "loss": 0.9746,
+      "loss": 0.9789,
       "step": 52000
     },
     {
       "epoch": 156.19,
-      "grad_norm": 7.702124118804932,
+      "grad_norm": 7.349396705627441,
       "learning_rate": 4.793793793793794e-06,
-      "loss": 0.9907,
+      "loss": 0.987,
       "step": 52010
     },
     {
       "epoch": 156.22,
-      "grad_norm": 7.873946666717529,
+      "grad_norm": 9.150490760803223,
       "learning_rate": 4.792792792792793e-06,
-      "loss": 1.001,
+      "loss": 1.0062,
       "step": 52020
     },
     {
       "epoch": 156.25,
-      "grad_norm": 10.730430603027344,
+      "grad_norm": 9.006162643432617,
       "learning_rate": 4.791791791791792e-06,
-      "loss": 0.9857,
+      "loss": 0.9752,
       "step": 52030
     },
     {
       "epoch": 156.28,
-      "grad_norm": 8.555842399597168,
+      "grad_norm": 9.94941520690918,
       "learning_rate": 4.7907907907907914e-06,
-      "loss": 1.0283,
+      "loss": 1.0237,
       "step": 52040
     },
     {
       "epoch": 156.31,
-      "grad_norm": 5.733790397644043,
+      "grad_norm": 6.390782833099365,
       "learning_rate": 4.78978978978979e-06,
-      "loss": 1.0299,
+      "loss": 1.0262,
       "step": 52050
     },
     {
       "epoch": 156.34,
-      "grad_norm": 7.186568737030029,
+      "grad_norm": 7.796817779541016,
       "learning_rate": 4.7887887887887895e-06,
-      "loss": 1.041,
+      "loss": 1.0442,
       "step": 52060
     },
     {
       "epoch": 156.37,
-      "grad_norm": 7.602048873901367,
+      "grad_norm": 10.010221481323242,
       "learning_rate": 4.787787787787788e-06,
-      "loss": 1.0074,
+      "loss": 1.0075,
       "step": 52070
     },
     {
       "epoch": 156.4,
-      "grad_norm": 6.5654215812683105,
+      "grad_norm": 6.2849626541137695,
       "learning_rate": 4.7867867867867875e-06,
-      "loss": 1.079,
+      "loss": 1.0738,
       "step": 52080
     },
     {
       "epoch": 156.43,
-      "grad_norm": 4.449295520782471,
+      "grad_norm": 8.467914581298828,
       "learning_rate": 4.785785785785786e-06,
-      "loss": 0.9509,
+      "loss": 0.9515,
       "step": 52090
     },
     {
       "epoch": 156.46,
-      "grad_norm": 8.583230018615723,
+      "grad_norm": 7.56186580657959,
       "learning_rate": 4.784784784784785e-06,
-      "loss": 1.0175,
+      "loss": 1.0198,
       "step": 52100
     },
     {
       "epoch": 156.49,
-      "grad_norm": 5.999931335449219,
+      "grad_norm": 6.1824846267700195,
       "learning_rate": 4.783783783783784e-06,
-      "loss": 0.96,
+      "loss": 0.9599,
       "step": 52110
     },
     {
       "epoch": 156.52,
-      "grad_norm": 5.816559314727783,
+      "grad_norm": 7.060883045196533,
       "learning_rate": 4.7827827827827836e-06,
-      "loss": 1.0166,
+      "loss": 1.015,
       "step": 52120
     },
     {
       "epoch": 156.55,
-      "grad_norm": 6.791285514831543,
+      "grad_norm": 9.580018043518066,
       "learning_rate": 4.781781781781782e-06,
-      "loss": 1.0285,
+      "loss": 1.0318,
       "step": 52130
     },
     {
       "epoch": 156.58,
-      "grad_norm": 5.310251235961914,
+      "grad_norm": 4.771050453186035,
       "learning_rate": 4.780780780780781e-06,
-      "loss": 1.0336,
+      "loss": 1.0276,
       "step": 52140
     },
     {
       "epoch": 156.61,
-      "grad_norm": 7.536881923675537,
+      "grad_norm": 9.617027282714844,
       "learning_rate": 4.77977977977978e-06,
-      "loss": 1.0016,
+      "loss": 1.0053,
       "step": 52150
     },
     {
       "epoch": 156.64,
-      "grad_norm": 5.33380126953125,
+      "grad_norm": 5.9789276123046875,
       "learning_rate": 4.77877877877878e-06,
-      "loss": 1.0551,
+      "loss": 1.053,
       "step": 52160
     },
     {
       "epoch": 156.67,
-      "grad_norm": 6.222975254058838,
+      "grad_norm": 7.3845534324646,
       "learning_rate": 4.777777777777778e-06,
-      "loss": 1.0182,
+      "loss": 1.014,
       "step": 52170
     },
     {
       "epoch": 156.7,
-      "grad_norm": 8.217728614807129,
+      "grad_norm": 6.557841777801514,
       "learning_rate": 4.776776776776777e-06,
-      "loss": 0.959,
+      "loss": 0.9631,
       "step": 52180
     },
     {
       "epoch": 156.73,
-      "grad_norm": 6.044974327087402,
+      "grad_norm": 7.5492939949035645,
       "learning_rate": 4.775775775775776e-06,
-      "loss": 1.0581,
+      "loss": 1.0554,
       "step": 52190
     },
     {
       "epoch": 156.76,
-      "grad_norm": 9.413265228271484,
+      "grad_norm": 9.741268157958984,
       "learning_rate": 4.774774774774775e-06,
-      "loss": 1.0451,
+      "loss": 1.0318,
       "step": 52200
     },
     {
       "epoch": 156.79,
-      "grad_norm": 7.392457008361816,
+      "grad_norm": 8.941689491271973,
       "learning_rate": 4.773773773773774e-06,
-      "loss": 1.0072,
+      "loss": 1.0066,
       "step": 52210
     },
     {
       "epoch": 156.82,
-      "grad_norm": 6.579846382141113,
+      "grad_norm": 6.818838119506836,
       "learning_rate": 4.772772772772773e-06,
-      "loss": 0.9342,
+      "loss": 0.9306,
       "step": 52220
     },
     {
       "epoch": 156.85,
-      "grad_norm": 6.996549129486084,
+      "grad_norm": 7.486064910888672,
       "learning_rate": 4.771771771771772e-06,
-      "loss": 1.009,
+      "loss": 1.0073,
       "step": 52230
     },
     {
       "epoch": 156.88,
-      "grad_norm": 7.164485454559326,
+      "grad_norm": 8.701003074645996,
       "learning_rate": 4.770770770770771e-06,
-      "loss": 1.0179,
+      "loss": 1.0093,
       "step": 52240
     },
     {
       "epoch": 156.91,
-      "grad_norm": 6.700099468231201,
+      "grad_norm": 7.77695894241333,
       "learning_rate": 4.76976976976977e-06,
-      "loss": 1.073,
+      "loss": 1.0663,
       "step": 52250
     },
     {
       "epoch": 156.94,
-      "grad_norm": 6.70352840423584,
+      "grad_norm": 9.841805458068848,
       "learning_rate": 4.768768768768769e-06,
-      "loss": 1.0678,
+      "loss": 1.0576,
       "step": 52260
     },
     {
       "epoch": 156.97,
-      "grad_norm": 7.225233554840088,
+      "grad_norm": 7.457268714904785,
       "learning_rate": 4.7677677677677684e-06,
-      "loss": 0.9312,
+      "loss": 0.9369,
       "step": 52270
     },
     {
       "epoch": 157.0,
-      "grad_norm": 6.481053352355957,
+      "grad_norm": 6.862839698791504,
       "learning_rate": 4.766766766766767e-06,
-      "loss": 1.0396,
+      "loss": 1.0351,
       "step": 52280
     },
     {
       "epoch": 157.0,
-      "eval_accuracy": 0.815,
-      "eval_loss": 0.6014113426208496,
-      "eval_runtime": 5.4358,
-      "eval_samples_per_second": 1839.649,
-      "eval_steps_per_second": 7.359,
+      "eval_accuracy": 0.8173,
+      "eval_loss": 0.5986879467964172,
+      "eval_runtime": 5.2299,
+      "eval_samples_per_second": 1912.081,
+      "eval_steps_per_second": 7.648,
       "step": 52281
     },
     {
       "epoch": 157.03,
-      "grad_norm": 7.957444190979004,
+      "grad_norm": 7.717632293701172,
       "learning_rate": 4.7657657657657665e-06,
-      "loss": 1.0592,
+      "loss": 1.0729,
       "step": 52290
     },
     {
       "epoch": 157.06,
-      "grad_norm": 6.013793468475342,
+      "grad_norm": 7.158042907714844,
       "learning_rate": 4.764764764764765e-06,
-      "loss": 0.9494,
+      "loss": 0.9555,
       "step": 52300
     },
     {
       "epoch": 157.09,
-      "grad_norm": 6.356362819671631,
+      "grad_norm": 7.8348588943481445,
       "learning_rate": 4.763763763763764e-06,
-      "loss": 0.9803,
+      "loss": 0.9671,
       "step": 52310
     },
     {
       "epoch": 157.12,
-      "grad_norm": 6.818270206451416,
+      "grad_norm": 10.802165985107422,
       "learning_rate": 4.762762762762763e-06,
-      "loss": 1.0253,
+      "loss": 1.0178,
       "step": 52320
     },
     {
       "epoch": 157.15,
-      "grad_norm": 7.023272514343262,
+      "grad_norm": 6.825409412384033,
       "learning_rate": 4.7617617617617625e-06,
-      "loss": 1.0224,
+      "loss": 1.0058,
       "step": 52330
     },
     {
       "epoch": 157.18,
-      "grad_norm": 7.4363813400268555,
+      "grad_norm": 5.123265743255615,
       "learning_rate": 4.760760760760761e-06,
-      "loss": 0.9432,
+      "loss": 0.9539,
       "step": 52340
     },
     {
       "epoch": 157.21,
-      "grad_norm": 7.9942946434021,
+      "grad_norm": 9.811251640319824,
       "learning_rate": 4.75975975975976e-06,
-      "loss": 0.9855,
+      "loss": 1.0035,
       "step": 52350
     },
     {
       "epoch": 157.24,
-      "grad_norm": 5.500067710876465,
+      "grad_norm": 4.670645236968994,
       "learning_rate": 4.758758758758759e-06,
-      "loss": 0.9789,
+      "loss": 0.9881,
       "step": 52360
     },
     {
       "epoch": 157.27,
-      "grad_norm": 5.856741428375244,
+      "grad_norm": 7.160909652709961,
       "learning_rate": 4.757757757757759e-06,
-      "loss": 1.007,
+      "loss": 1.0028,
       "step": 52370
     },
     {
       "epoch": 157.3,
-      "grad_norm": 6.369245529174805,
+      "grad_norm": 7.558296203613281,
       "learning_rate": 4.756756756756757e-06,
-      "loss": 1.0048,
+      "loss": 1.0064,
       "step": 52380
     },
     {
       "epoch": 157.33,
-      "grad_norm": 6.799499034881592,
+      "grad_norm": 7.529449939727783,
       "learning_rate": 4.755755755755756e-06,
-      "loss": 0.9685,
+      "loss": 0.9636,
       "step": 52390
     },
     {
       "epoch": 157.36,
-      "grad_norm": 7.893992900848389,
+      "grad_norm": 8.098579406738281,
       "learning_rate": 4.754754754754755e-06,
-      "loss": 1.0048,
+      "loss": 1.0026,
       "step": 52400
     },
     {
       "epoch": 157.39,
-      "grad_norm": 7.206137657165527,
+      "grad_norm": 6.022697925567627,
       "learning_rate": 4.753753753753754e-06,
-      "loss": 1.0422,
+      "loss": 1.0373,
       "step": 52410
     },
     {
       "epoch": 157.42,
-      "grad_norm": 6.815255641937256,
+      "grad_norm": 5.605231761932373,
       "learning_rate": 4.752752752752753e-06,
-      "loss": 1.035,
+      "loss": 1.0371,
       "step": 52420
     },
     {
       "epoch": 157.45,
-      "grad_norm": 6.9344868659973145,
+      "grad_norm": 10.150459289550781,
       "learning_rate": 4.751751751751752e-06,
-      "loss": 1.0101,
+      "loss": 1.0149,
       "step": 52430
     },
     {
       "epoch": 157.48,
-      "grad_norm": 6.703427314758301,
+      "grad_norm": 8.3174409866333,
       "learning_rate": 4.750750750750751e-06,
-      "loss": 1.0095,
+      "loss": 1.0108,
       "step": 52440
     },
     {
       "epoch": 157.51,
-      "grad_norm": 6.658684730529785,
+      "grad_norm": 7.6879096031188965,
       "learning_rate": 4.74974974974975e-06,
-      "loss": 1.004,
+      "loss": 1.0069,
       "step": 52450
     },
     {
       "epoch": 157.54,
-      "grad_norm": 6.681285858154297,
+      "grad_norm": 9.071887016296387,
       "learning_rate": 4.748748748748749e-06,
-      "loss": 0.9813,
+      "loss": 0.9803,
       "step": 52460
     },
     {
       "epoch": 157.57,
-      "grad_norm": 6.87605619430542,
+      "grad_norm": 7.254177570343018,
       "learning_rate": 4.747747747747748e-06,
-      "loss": 0.9451,
+      "loss": 0.9399,
       "step": 52470
     },
     {
       "epoch": 157.6,
-      "grad_norm": 6.351436614990234,
+      "grad_norm": 6.162106037139893,
       "learning_rate": 4.746746746746747e-06,
-      "loss": 1.0017,
+      "loss": 0.9949,
       "step": 52480
     },
     {
       "epoch": 157.63,
-      "grad_norm": 5.56697416305542,
+      "grad_norm": 5.8268609046936035,
       "learning_rate": 4.745745745745746e-06,
-      "loss": 1.0131,
+      "loss": 1.0028,
       "step": 52490
     },
     {
       "epoch": 157.66,
-      "grad_norm": 6.938794136047363,
+      "grad_norm": 7.904989719390869,
       "learning_rate": 4.7447447447447454e-06,
-      "loss": 1.04,
+      "loss": 1.0364,
       "step": 52500
     },
     {
       "epoch": 157.69,
-      "grad_norm": 8.070590019226074,
+      "grad_norm": 9.129644393920898,
       "learning_rate": 4.743743743743744e-06,
-      "loss": 0.9502,
+      "loss": 0.9677,
       "step": 52510
     },
     {
       "epoch": 157.72,
-      "grad_norm": 6.763585567474365,
+      "grad_norm": 5.62079381942749,
       "learning_rate": 4.7427427427427435e-06,
-      "loss": 1.0549,
+      "loss": 1.0611,
       "step": 52520
     },
     {
       "epoch": 157.75,
-      "grad_norm": 5.397427082061768,
+      "grad_norm": 7.764962196350098,
       "learning_rate": 4.741741741741742e-06,
-      "loss": 0.9952,
+      "loss": 0.9991,
       "step": 52530
     },
     {
       "epoch": 157.78,
-      "grad_norm": 7.9924468994140625,
+      "grad_norm": 8.008248329162598,
       "learning_rate": 4.7407407407407415e-06,
-      "loss": 1.0218,
+      "loss": 1.0445,
       "step": 52540
     },
     {
       "epoch": 157.81,
-      "grad_norm": 6.299176216125488,
+      "grad_norm": 9.261885643005371,
       "learning_rate": 4.73973973973974e-06,
-      "loss": 0.9775,
+      "loss": 0.9773,
       "step": 52550
     },
     {
       "epoch": 157.84,
-      "grad_norm": 6.5254950523376465,
+      "grad_norm": 6.0654377937316895,
       "learning_rate": 4.738738738738739e-06,
-      "loss": 0.9813,
+      "loss": 0.9807,
       "step": 52560
     },
     {
       "epoch": 157.87,
-      "grad_norm": 5.680192470550537,
+      "grad_norm": 6.339696407318115,
       "learning_rate": 4.737737737737738e-06,
-      "loss": 0.9582,
+      "loss": 0.9691,
       "step": 52570
     },
     {
       "epoch": 157.9,
-      "grad_norm": 6.4821319580078125,
+      "grad_norm": 10.95643138885498,
       "learning_rate": 4.7367367367367376e-06,
-      "loss": 1.0347,
+      "loss": 1.0305,
       "step": 52580
     },
     {
       "epoch": 157.93,
-      "grad_norm": 5.732822895050049,
+      "grad_norm": 5.843149185180664,
       "learning_rate": 4.735735735735736e-06,
-      "loss": 1.0189,
+      "loss": 1.0222,
       "step": 52590
     },
     {
       "epoch": 157.96,
-      "grad_norm": 5.769380569458008,
+      "grad_norm": 6.229001045227051,
       "learning_rate": 4.734734734734735e-06,
-      "loss": 0.9768,
+      "loss": 0.9659,
       "step": 52600
     },
     {
       "epoch": 157.99,
-      "grad_norm": 4.4922871589660645,
+      "grad_norm": 5.642692565917969,
       "learning_rate": 4.733733733733734e-06,
-      "loss": 1.0524,
+      "loss": 1.0586,
       "step": 52610
     },
     {
       "epoch": 158.0,
-      "eval_accuracy": 0.8164,
-      "eval_loss": 0.6015158891677856,
-      "eval_runtime": 5.354,
-      "eval_samples_per_second": 1867.77,
-      "eval_steps_per_second": 7.471,
+      "eval_accuracy": 0.8165,
+      "eval_loss": 0.5999394655227661,
+      "eval_runtime": 5.561,
+      "eval_samples_per_second": 1798.249,
+      "eval_steps_per_second": 7.193,
       "step": 52614
     },
     {
       "epoch": 158.02,
-      "grad_norm": 6.8282470703125,
+      "grad_norm": 5.881308555603027,
       "learning_rate": 4.732732732732734e-06,
-      "loss": 1.1156,
+      "loss": 1.1149,
       "step": 52620
     },
     {
       "epoch": 158.05,
-      "grad_norm": 6.593434810638428,
+      "grad_norm": 7.9621381759643555,
       "learning_rate": 4.731731731731732e-06,
-      "loss": 1.015,
+      "loss": 1.0096,
       "step": 52630
     },
     {
       "epoch": 158.08,
-      "grad_norm": 8.932826042175293,
+      "grad_norm": 6.250314712524414,
       "learning_rate": 4.730730730730731e-06,
-      "loss": 0.9919,
+      "loss": 0.9951,
       "step": 52640
     },
     {
       "epoch": 158.11,
-      "grad_norm": 5.881802082061768,
+      "grad_norm": 5.764659404754639,
       "learning_rate": 4.72972972972973e-06,
-      "loss": 1.0682,
+      "loss": 1.0585,
       "step": 52650
     },
     {
       "epoch": 158.14,
-      "grad_norm": 8.087949752807617,
+      "grad_norm": 6.837044715881348,
       "learning_rate": 4.728728728728729e-06,
-      "loss": 0.9333,
+      "loss": 0.9296,
       "step": 52660
     },
     {
       "epoch": 158.17,
-      "grad_norm": 5.095150470733643,
+      "grad_norm": 6.961920738220215,
       "learning_rate": 4.727727727727728e-06,
-      "loss": 0.9654,
+      "loss": 0.9604,
       "step": 52670
     },
     {
       "epoch": 158.2,
-      "grad_norm": 7.507325649261475,
+      "grad_norm": 7.531022071838379,
       "learning_rate": 4.726726726726727e-06,
-      "loss": 1.0276,
+      "loss": 1.0264,
       "step": 52680
     },
     {
       "epoch": 158.23,
-      "grad_norm": 6.987926006317139,
+      "grad_norm": 6.202571868896484,
       "learning_rate": 4.725725725725726e-06,
-      "loss": 0.9745,
+      "loss": 0.9661,
       "step": 52690
     },
     {
       "epoch": 158.26,
-      "grad_norm": 7.359043598175049,
+      "grad_norm": 8.115350723266602,
       "learning_rate": 4.724724724724725e-06,
-      "loss": 0.9945,
+      "loss": 0.9946,
       "step": 52700
     },
     {
       "epoch": 158.29,
-      "grad_norm": 7.119153022766113,
+      "grad_norm": 7.935606002807617,
       "learning_rate": 4.723723723723724e-06,
-      "loss": 1.0685,
+      "loss": 1.0541,
       "step": 52710
     },
     {
       "epoch": 158.32,
-      "grad_norm": 8.68436336517334,
+      "grad_norm": 8.413661003112793,
       "learning_rate": 4.722722722722723e-06,
-      "loss": 1.0153,
+      "loss": 1.0202,
       "step": 52720
     },
     {
       "epoch": 158.35,
-      "grad_norm": 6.446855545043945,
+      "grad_norm": 7.536243915557861,
       "learning_rate": 4.7217217217217224e-06,
-      "loss": 1.0042,
+      "loss": 1.0029,
       "step": 52730
     },
     {
       "epoch": 158.38,
-      "grad_norm": 5.727358341217041,
+      "grad_norm": 7.4769158363342285,
       "learning_rate": 4.720720720720721e-06,
-      "loss": 1.0164,
+      "loss": 1.0177,
       "step": 52740
     },
     {
       "epoch": 158.41,
-      "grad_norm": 6.667992115020752,
+      "grad_norm": 6.207727432250977,
       "learning_rate": 4.7197197197197205e-06,
-      "loss": 1.025,
+      "loss": 1.0235,
       "step": 52750
     },
     {
       "epoch": 158.44,
-      "grad_norm": 14.321820259094238,
+      "grad_norm": 8.511510848999023,
       "learning_rate": 4.718718718718719e-06,
-      "loss": 0.9718,
+      "loss": 0.9755,
       "step": 52760
     },
     {
       "epoch": 158.47,
-      "grad_norm": 7.147668361663818,
+      "grad_norm": 6.716568470001221,
       "learning_rate": 4.717717717717718e-06,
-      "loss": 1.0119,
+      "loss": 1.0066,
       "step": 52770
     },
     {
       "epoch": 158.5,
-      "grad_norm": 4.797133922576904,
+      "grad_norm": 7.192840576171875,
       "learning_rate": 4.716716716716717e-06,
-      "loss": 0.9961,
+      "loss": 1.0054,
       "step": 52780
     },
     {
       "epoch": 158.53,
-      "grad_norm": 4.942505836486816,
+      "grad_norm": 5.180251598358154,
       "learning_rate": 4.7157157157157165e-06,
-      "loss": 0.9279,
+      "loss": 0.9206,
       "step": 52790
     },
     {
       "epoch": 158.56,
-      "grad_norm": 7.190163612365723,
+      "grad_norm": 6.582204341888428,
       "learning_rate": 4.714714714714715e-06,
-      "loss": 0.9908,
+      "loss": 0.9889,
       "step": 52800
     },
     {
       "epoch": 158.59,
-      "grad_norm": 6.440318584442139,
+      "grad_norm": 6.786748886108398,
       "learning_rate": 4.713713713713714e-06,
-      "loss": 1.025,
+      "loss": 1.0312,
       "step": 52810
     },
     {
       "epoch": 158.62,
-      "grad_norm": 5.737502098083496,
+      "grad_norm": 6.267710208892822,
       "learning_rate": 4.712712712712713e-06,
-      "loss": 1.0056,
+      "loss": 1.0171,
       "step": 52820
     },
     {
       "epoch": 158.65,
-      "grad_norm": 5.317436695098877,
+      "grad_norm": 6.219812870025635,
       "learning_rate": 4.711711711711713e-06,
-      "loss": 0.9605,
+      "loss": 0.9533,
       "step": 52830
     },
     {
       "epoch": 158.68,
-      "grad_norm": 6.946715831756592,
+      "grad_norm": 8.951515197753906,
       "learning_rate": 4.710710710710711e-06,
-      "loss": 0.974,
+      "loss": 0.9833,
       "step": 52840
     },
     {
       "epoch": 158.71,
-      "grad_norm": 6.904016494750977,
+      "grad_norm": 6.173943519592285,
       "learning_rate": 4.70970970970971e-06,
-      "loss": 1.0628,
+      "loss": 1.0657,
       "step": 52850
     },
     {
       "epoch": 158.74,
-      "grad_norm": 5.448324680328369,
+      "grad_norm": 5.866313457489014,
       "learning_rate": 4.708708708708709e-06,
-      "loss": 0.9967,
+      "loss": 0.9949,
       "step": 52860
     },
     {
       "epoch": 158.77,
-      "grad_norm": 7.748785972595215,
+      "grad_norm": 7.081233978271484,
       "learning_rate": 4.707707707707708e-06,
-      "loss": 1.0374,
+      "loss": 1.0321,
       "step": 52870
     },
     {
       "epoch": 158.8,
-      "grad_norm": 5.972383975982666,
+      "grad_norm": 9.023186683654785,
       "learning_rate": 4.7067067067067064e-06,
-      "loss": 1.0303,
+      "loss": 1.0224,
       "step": 52880
     },
     {
       "epoch": 158.83,
-      "grad_norm": 7.968528747558594,
+      "grad_norm": 7.018273830413818,
       "learning_rate": 4.705705705705706e-06,
-      "loss": 1.0114,
+      "loss": 1.0104,
       "step": 52890
     },
     {
       "epoch": 158.86,
-      "grad_norm": 6.238701820373535,
+      "grad_norm": 6.943708419799805,
       "learning_rate": 4.704704704704705e-06,
-      "loss": 0.9715,
+      "loss": 0.9691,
       "step": 52900
     },
     {
       "epoch": 158.89,
-      "grad_norm": 5.37094783782959,
+      "grad_norm": 5.870466232299805,
       "learning_rate": 4.703703703703704e-06,
-      "loss": 0.9888,
+      "loss": 1.0002,
       "step": 52910
     },
     {
       "epoch": 158.92,
-      "grad_norm": 5.548099040985107,
+      "grad_norm": 8.166054725646973,
       "learning_rate": 4.702702702702703e-06,
-      "loss": 1.0008,
+      "loss": 0.9991,
       "step": 52920
     },
     {
       "epoch": 158.95,
-      "grad_norm": 7.410159587860107,
+      "grad_norm": 7.1879377365112305,
       "learning_rate": 4.701701701701702e-06,
-      "loss": 1.0164,
+      "loss": 1.0061,
       "step": 52930
     },
     {
       "epoch": 158.98,
-      "grad_norm": 10.482246398925781,
+      "grad_norm": 11.72937297821045,
       "learning_rate": 4.700700700700701e-06,
-      "loss": 0.9873,
+      "loss": 0.9828,
       "step": 52940
     },
     {
       "epoch": 159.0,
-      "eval_accuracy": 0.8152,
-      "eval_loss": 0.600149929523468,
-      "eval_runtime": 5.5736,
-      "eval_samples_per_second": 1794.163,
-      "eval_steps_per_second": 7.177,
+      "eval_accuracy": 0.8177,
+      "eval_loss": 0.5985073447227478,
+      "eval_runtime": 5.2737,
+      "eval_samples_per_second": 1896.212,
+      "eval_steps_per_second": 7.585,
       "step": 52947
     },
     {
       "epoch": 159.01,
-      "grad_norm": 5.987947940826416,
+      "grad_norm": 5.9571733474731445,
       "learning_rate": 4.6996996996997e-06,
-      "loss": 1.3453,
+      "loss": 1.3594,
       "step": 52950
     },
     {
       "epoch": 159.04,
-      "grad_norm": 7.2485527992248535,
+      "grad_norm": 5.845569133758545,
       "learning_rate": 4.698698698698699e-06,
-      "loss": 1.0091,
+      "loss": 1.0115,
       "step": 52960
     },
     {
       "epoch": 159.07,
-      "grad_norm": 8.97481632232666,
+      "grad_norm": 12.767858505249023,
       "learning_rate": 4.697697697697698e-06,
-      "loss": 1.054,
+      "loss": 1.0511,
       "step": 52970
     },
     {
       "epoch": 159.1,
-      "grad_norm": 7.409544467926025,
+      "grad_norm": 10.105382919311523,
       "learning_rate": 4.696696696696697e-06,
-      "loss": 1.0539,
+      "loss": 1.0385,
       "step": 52980
     },
     {
       "epoch": 159.13,
-      "grad_norm": 7.791576385498047,
+      "grad_norm": 8.059584617614746,
       "learning_rate": 4.695695695695696e-06,
-      "loss": 1.0236,
+      "loss": 1.0163,
       "step": 52990
     },
     {
       "epoch": 159.16,
-      "grad_norm": 7.2799973487854,
+      "grad_norm": 7.9343085289001465,
       "learning_rate": 4.6946946946946955e-06,
-      "loss": 0.9448,
+      "loss": 0.9505,
       "step": 53000
     },
     {
       "epoch": 159.19,
-      "grad_norm": 10.479708671569824,
+      "grad_norm": 7.4556684494018555,
       "learning_rate": 4.693693693693694e-06,
-      "loss": 0.976,
+      "loss": 0.9659,
       "step": 53010
     },
     {
       "epoch": 159.22,
-      "grad_norm": 5.711459636688232,
+      "grad_norm": 6.079098224639893,
       "learning_rate": 4.692692692692693e-06,
-      "loss": 1.0268,
+      "loss": 1.027,
       "step": 53020
     },
     {
       "epoch": 159.25,
-      "grad_norm": 7.230443000793457,
+      "grad_norm": 6.540330410003662,
       "learning_rate": 4.691691691691692e-06,
-      "loss": 0.9915,
+      "loss": 0.996,
       "step": 53030
     },
     {
       "epoch": 159.28,
-      "grad_norm": 6.772353172302246,
+      "grad_norm": 6.5042724609375,
       "learning_rate": 4.6906906906906916e-06,
-      "loss": 1.0021,
+      "loss": 0.996,
       "step": 53040
     },
     {
       "epoch": 159.31,
-      "grad_norm": 6.055197715759277,
+      "grad_norm": 9.370908737182617,
       "learning_rate": 4.68968968968969e-06,
-      "loss": 1.0085,
+      "loss": 0.9957,
       "step": 53050
     },
     {
       "epoch": 159.34,
-      "grad_norm": 4.689176082611084,
+      "grad_norm": 7.650594711303711,
       "learning_rate": 4.688688688688689e-06,
-      "loss": 0.9897,
+      "loss": 0.9859,
       "step": 53060
     },
     {
       "epoch": 159.37,
-      "grad_norm": 7.494970321655273,
+      "grad_norm": 6.754175662994385,
       "learning_rate": 4.687687687687688e-06,
-      "loss": 0.9963,
+      "loss": 0.9998,
       "step": 53070
     },
     {
       "epoch": 159.4,
-      "grad_norm": 4.865299224853516,
+      "grad_norm": 6.544069766998291,
       "learning_rate": 4.686686686686688e-06,
-      "loss": 1.0135,
+      "loss": 1.0088,
       "step": 53080
     },
     {
       "epoch": 159.43,
-      "grad_norm": 6.060498237609863,
+      "grad_norm": 7.627967834472656,
       "learning_rate": 4.685685685685686e-06,
-      "loss": 1.0309,
+      "loss": 1.0351,
       "step": 53090
     },
     {
       "epoch": 159.46,
-      "grad_norm": 6.602518081665039,
+      "grad_norm": 8.495121002197266,
       "learning_rate": 4.684684684684685e-06,
-      "loss": 0.9918,
+      "loss": 0.985,
       "step": 53100
     },
     {
       "epoch": 159.49,
-      "grad_norm": 5.257236003875732,
+      "grad_norm": 6.0775299072265625,
       "learning_rate": 4.683683683683684e-06,
-      "loss": 0.9779,
+      "loss": 0.9825,
       "step": 53110
     },
     {
       "epoch": 159.52,
-      "grad_norm": 7.730934143066406,
+      "grad_norm": 6.748131275177002,
       "learning_rate": 4.682682682682683e-06,
-      "loss": 0.9701,
+      "loss": 0.9586,
       "step": 53120
     },
     {
       "epoch": 159.55,
-      "grad_norm": 5.283814430236816,
+      "grad_norm": 6.445345878601074,
       "learning_rate": 4.6816816816816815e-06,
-      "loss": 1.0266,
+      "loss": 1.0195,
       "step": 53130
     },
     {
       "epoch": 159.58,
-      "grad_norm": 4.06208610534668,
+      "grad_norm": 7.386924743652344,
       "learning_rate": 4.680680680680681e-06,
-      "loss": 0.9279,
+      "loss": 0.9255,
       "step": 53140
     },
     {
       "epoch": 159.61,
-      "grad_norm": 5.505299091339111,
+      "grad_norm": 5.931994438171387,
       "learning_rate": 4.67967967967968e-06,
-      "loss": 0.9667,
+      "loss": 0.9789,
       "step": 53150
     },
     {
       "epoch": 159.64,
-      "grad_norm": 5.789005279541016,
+      "grad_norm": 6.027069568634033,
       "learning_rate": 4.678678678678679e-06,
-      "loss": 1.0534,
+      "loss": 1.0574,
       "step": 53160
     },
     {
       "epoch": 159.67,
-      "grad_norm": 5.323051929473877,
+      "grad_norm": 5.899692535400391,
       "learning_rate": 4.677677677677678e-06,
-      "loss": 1.0085,
+      "loss": 1.0023,
       "step": 53170
     },
     {
       "epoch": 159.7,
-      "grad_norm": 7.672285079956055,
+      "grad_norm": 7.133199691772461,
       "learning_rate": 4.676676676676677e-06,
-      "loss": 1.0092,
+      "loss": 1.0113,
       "step": 53180
     },
     {
       "epoch": 159.73,
-      "grad_norm": 7.38014554977417,
+      "grad_norm": 7.693531036376953,
       "learning_rate": 4.675675675675676e-06,
       "loss": 0.9429,
       "step": 53190
     },
     {
       "epoch": 159.76,
-      "grad_norm": 7.112912654876709,
+      "grad_norm": 5.379668235778809,
       "learning_rate": 4.674674674674675e-06,
-      "loss": 0.9538,
+      "loss": 0.9456,
       "step": 53200
     },
     {
       "epoch": 159.79,
-      "grad_norm": 7.153504848480225,
+      "grad_norm": 9.369929313659668,
       "learning_rate": 4.6736736736736745e-06,
-      "loss": 1.0577,
+      "loss": 1.0512,
       "step": 53210
     },
     {
       "epoch": 159.82,
-      "grad_norm": 6.183586597442627,
+      "grad_norm": 7.10391092300415,
       "learning_rate": 4.672672672672673e-06,
-      "loss": 0.971,
+      "loss": 0.9731,
       "step": 53220
     },
     {
       "epoch": 159.85,
-      "grad_norm": 9.325462341308594,
+      "grad_norm": 6.43406867980957,
       "learning_rate": 4.671671671671672e-06,
-      "loss": 0.9938,
+      "loss": 0.9799,
       "step": 53230
     },
     {
       "epoch": 159.88,
-      "grad_norm": 8.037854194641113,
+      "grad_norm": 6.784370422363281,
       "learning_rate": 4.670670670670671e-06,
-      "loss": 1.046,
+      "loss": 1.0408,
       "step": 53240
     },
     {
       "epoch": 159.91,
-      "grad_norm": 5.876431465148926,
+      "grad_norm": 5.402968883514404,
       "learning_rate": 4.6696696696696705e-06,
-      "loss": 1.0624,
+      "loss": 1.0602,
       "step": 53250
     },
     {
       "epoch": 159.94,
-      "grad_norm": 3.7963366508483887,
+      "grad_norm": 6.396175384521484,
       "learning_rate": 4.668668668668669e-06,
-      "loss": 0.9634,
+      "loss": 0.9718,
       "step": 53260
     },
     {
       "epoch": 159.97,
-      "grad_norm": 8.093559265136719,
+      "grad_norm": 7.287644863128662,
       "learning_rate": 4.667667667667668e-06,
-      "loss": 1.0632,
+      "loss": 1.0645,
       "step": 53270
     },
     {
       "epoch": 160.0,
-      "grad_norm": 41.49421310424805,
+      "grad_norm": 54.76237869262695,
       "learning_rate": 4.666666666666667e-06,
-      "loss": 1.0471,
+      "loss": 1.0671,
       "step": 53280
     },
     {
       "epoch": 160.0,
-      "eval_accuracy": 0.8165,
-      "eval_loss": 0.5988420844078064,
-      "eval_runtime": 5.5587,
-      "eval_samples_per_second": 1798.985,
-      "eval_steps_per_second": 7.196,
+      "eval_accuracy": 0.8172,
+      "eval_loss": 0.5979963541030884,
+      "eval_runtime": 5.2349,
+      "eval_samples_per_second": 1910.267,
+      "eval_steps_per_second": 7.641,
       "step": 53280
     },
     {
       "epoch": 160.03,
-      "grad_norm": 5.448941707611084,
+      "grad_norm": 6.68494987487793,
       "learning_rate": 4.665665665665667e-06,
-      "loss": 0.9371,
+      "loss": 0.9249,
       "step": 53290
     },
     {
       "epoch": 160.06,
-      "grad_norm": 6.419309139251709,
+      "grad_norm": 6.440035343170166,
       "learning_rate": 4.664664664664665e-06,
-      "loss": 1.0083,
+      "loss": 1.0059,
       "step": 53300
     },
     {
       "epoch": 160.09,
-      "grad_norm": 8.612344741821289,
+      "grad_norm": 6.0308380126953125,
       "learning_rate": 4.663663663663664e-06,
-      "loss": 0.9619,
+      "loss": 0.9621,
       "step": 53310
     },
     {
       "epoch": 160.12,
-      "grad_norm": 6.381759166717529,
+      "grad_norm": 7.647411823272705,
       "learning_rate": 4.662662662662663e-06,
-      "loss": 0.9534,
+      "loss": 0.9593,
       "step": 53320
     },
     {
       "epoch": 160.15,
-      "grad_norm": 9.411449432373047,
+      "grad_norm": 7.125226020812988,
       "learning_rate": 4.661661661661662e-06,
-      "loss": 1.0387,
+      "loss": 1.0398,
       "step": 53330
     },
     {
       "epoch": 160.18,
-      "grad_norm": 6.774388790130615,
+      "grad_norm": 9.116605758666992,
       "learning_rate": 4.6606606606606604e-06,
-      "loss": 1.004,
+      "loss": 1.0094,
       "step": 53340
     },
     {
       "epoch": 160.21,
-      "grad_norm": 6.455174922943115,
+      "grad_norm": 7.367770195007324,
       "learning_rate": 4.65965965965966e-06,
-      "loss": 1.0229,
+      "loss": 1.0203,
       "step": 53350
     },
     {
       "epoch": 160.24,
-      "grad_norm": 8.536046981811523,
+      "grad_norm": 7.664543628692627,
       "learning_rate": 4.658658658658659e-06,
-      "loss": 0.9605,
+      "loss": 0.9455,
       "step": 53360
     },
     {
       "epoch": 160.27,
-      "grad_norm": 6.88848352432251,
+      "grad_norm": 10.306459426879883,
       "learning_rate": 4.657657657657658e-06,
-      "loss": 1.0263,
+      "loss": 1.0227,
       "step": 53370
     },
     {
       "epoch": 160.3,
-      "grad_norm": 6.29094934463501,
+      "grad_norm": 7.001646041870117,
       "learning_rate": 4.6566566566566565e-06,
-      "loss": 1.0003,
+      "loss": 1.0088,
       "step": 53380
     },
     {
       "epoch": 160.33,
-      "grad_norm": 6.643807411193848,
+      "grad_norm": 8.340766906738281,
       "learning_rate": 4.655655655655656e-06,
-      "loss": 1.0255,
+      "loss": 1.0252,
       "step": 53390
     },
     {
       "epoch": 160.36,
-      "grad_norm": 6.661516189575195,
+      "grad_norm": 8.055726051330566,
       "learning_rate": 4.654654654654655e-06,
-      "loss": 0.9759,
+      "loss": 0.971,
       "step": 53400
     },
     {
       "epoch": 160.39,
-      "grad_norm": 6.610814094543457,
+      "grad_norm": 5.304447174072266,
       "learning_rate": 4.653653653653654e-06,
-      "loss": 0.938,
+      "loss": 0.9382,
       "step": 53410
     },
     {
       "epoch": 160.42,
-      "grad_norm": 5.175430774688721,
+      "grad_norm": 6.442556381225586,
       "learning_rate": 4.6526526526526526e-06,
-      "loss": 0.9748,
+      "loss": 0.9781,
       "step": 53420
     },
     {
       "epoch": 160.45,
-      "grad_norm": 5.687014579772949,
+      "grad_norm": 7.384301662445068,
       "learning_rate": 4.651651651651652e-06,
-      "loss": 0.9619,
+      "loss": 0.957,
       "step": 53430
     },
     {
       "epoch": 160.48,
-      "grad_norm": 5.980249404907227,
+      "grad_norm": 7.132927417755127,
       "learning_rate": 4.650650650650651e-06,
-      "loss": 0.9547,
+      "loss": 0.9582,
       "step": 53440
     },
     {
       "epoch": 160.51,
-      "grad_norm": 7.076175689697266,
+      "grad_norm": 8.032471656799316,
       "learning_rate": 4.64964964964965e-06,
-      "loss": 0.9973,
+      "loss": 0.9948,
       "step": 53450
     },
     {
       "epoch": 160.54,
-      "grad_norm": 5.5416579246521,
+      "grad_norm": 5.900144577026367,
       "learning_rate": 4.6486486486486495e-06,
-      "loss": 0.9585,
+      "loss": 0.9714,
       "step": 53460
     },
     {
       "epoch": 160.57,
-      "grad_norm": 4.956183433532715,
+      "grad_norm": 5.535617828369141,
       "learning_rate": 4.647647647647648e-06,
-      "loss": 0.9702,
+      "loss": 0.9754,
       "step": 53470
     },
     {
       "epoch": 160.6,
-      "grad_norm": 4.884345531463623,
+      "grad_norm": 5.367117404937744,
       "learning_rate": 4.646646646646647e-06,
-      "loss": 0.9739,
+      "loss": 0.9722,
       "step": 53480
     },
     {
       "epoch": 160.63,
-      "grad_norm": 5.782952308654785,
+      "grad_norm": 6.30300235748291,
       "learning_rate": 4.645645645645646e-06,
-      "loss": 1.0527,
+      "loss": 1.0477,
       "step": 53490
     },
     {
       "epoch": 160.66,
-      "grad_norm": 6.754055023193359,
+      "grad_norm": 8.655506134033203,
       "learning_rate": 4.6446446446446456e-06,
-      "loss": 1.0157,
+      "loss": 1.0164,
       "step": 53500
     },
     {
       "epoch": 160.69,
-      "grad_norm": 6.075368404388428,
+      "grad_norm": 5.040628910064697,
       "learning_rate": 4.643643643643644e-06,
-      "loss": 0.9611,
+      "loss": 0.964,
       "step": 53510
     },
     {
       "epoch": 160.72,
-      "grad_norm": 7.046881675720215,
+      "grad_norm": 5.2802934646606445,
       "learning_rate": 4.642642642642643e-06,
-      "loss": 1.0631,
+      "loss": 1.0583,
       "step": 53520
     },
     {
       "epoch": 160.75,
-      "grad_norm": 6.353909492492676,
+      "grad_norm": 6.852458477020264,
       "learning_rate": 4.641641641641642e-06,
-      "loss": 0.9899,
+      "loss": 0.9775,
       "step": 53530
     },
     {
       "epoch": 160.78,
-      "grad_norm": 9.166726112365723,
+      "grad_norm": 8.393898963928223,
       "learning_rate": 4.640640640640642e-06,
-      "loss": 0.9732,
+      "loss": 0.9807,
       "step": 53540
     },
     {
       "epoch": 160.81,
-      "grad_norm": 5.654964923858643,
+      "grad_norm": 4.450160503387451,
       "learning_rate": 4.63963963963964e-06,
-      "loss": 1.0563,
+      "loss": 1.0623,
       "step": 53550
     },
     {
       "epoch": 160.84,
-      "grad_norm": 6.209615707397461,
+      "grad_norm": 6.867171287536621,
       "learning_rate": 4.638638638638639e-06,
-      "loss": 1.0038,
+      "loss": 1.0089,
       "step": 53560
     },
     {
       "epoch": 160.87,
-      "grad_norm": 8.55988597869873,
+      "grad_norm": 6.757373332977295,
       "learning_rate": 4.637637637637638e-06,
-      "loss": 0.9684,
+      "loss": 0.961,
       "step": 53570
     },
     {
       "epoch": 160.9,
-      "grad_norm": 6.501730918884277,
+      "grad_norm": 6.699324607849121,
       "learning_rate": 4.636636636636637e-06,
-      "loss": 1.0852,
+      "loss": 1.0871,
       "step": 53580
     },
     {
       "epoch": 160.93,
-      "grad_norm": 4.852698802947998,
+      "grad_norm": 4.600536823272705,
       "learning_rate": 4.6356356356356355e-06,
-      "loss": 0.954,
+      "loss": 0.9461,
       "step": 53590
     },
     {
       "epoch": 160.96,
-      "grad_norm": 9.278258323669434,
+      "grad_norm": 9.135825157165527,
       "learning_rate": 4.634634634634635e-06,
-      "loss": 1.0414,
+      "loss": 1.0412,
       "step": 53600
     },
     {
       "epoch": 160.99,
-      "grad_norm": 6.9371490478515625,
+      "grad_norm": 7.819042682647705,
       "learning_rate": 4.633633633633634e-06,
-      "loss": 0.9178,
+      "loss": 0.9242,
       "step": 53610
     },
     {
       "epoch": 161.0,
-      "eval_accuracy": 0.8185,
-      "eval_loss": 0.5936145782470703,
-      "eval_runtime": 5.3137,
-      "eval_samples_per_second": 1881.915,
-      "eval_steps_per_second": 7.528,
+      "eval_accuracy": 0.8214,
+      "eval_loss": 0.5920769572257996,
+      "eval_runtime": 5.3531,
+      "eval_samples_per_second": 1868.076,
+      "eval_steps_per_second": 7.472,
       "step": 53613
     },
     {
       "epoch": 161.02,
-      "grad_norm": 6.387082099914551,
+      "grad_norm": 6.105080604553223,
       "learning_rate": 4.632632632632633e-06,
-      "loss": 1.0311,
+      "loss": 1.0191,
       "step": 53620
     },
     {
       "epoch": 161.05,
-      "grad_norm": 5.409492015838623,
+      "grad_norm": 5.985925197601318,
       "learning_rate": 4.6316316316316315e-06,
-      "loss": 0.9654,
+      "loss": 0.9646,
       "step": 53630
     },
     {
       "epoch": 161.08,
-      "grad_norm": 6.110879898071289,
+      "grad_norm": 6.7764177322387695,
       "learning_rate": 4.630630630630631e-06,
-      "loss": 1.0459,
+      "loss": 1.043,
       "step": 53640
     },
     {
       "epoch": 161.11,
-      "grad_norm": 7.8040289878845215,
+      "grad_norm": 10.360906600952148,
       "learning_rate": 4.62962962962963e-06,
-      "loss": 0.984,
+      "loss": 0.9674,
       "step": 53650
     },
     {
       "epoch": 161.14,
-      "grad_norm": 8.853523254394531,
+      "grad_norm": 11.331032752990723,
       "learning_rate": 4.628628628628629e-06,
-      "loss": 1.0572,
+      "loss": 1.0581,
       "step": 53660
     },
     {
       "epoch": 161.17,
-      "grad_norm": 4.436427593231201,
+      "grad_norm": 6.495529651641846,
       "learning_rate": 4.627627627627628e-06,
-      "loss": 0.9693,
+      "loss": 0.9689,
       "step": 53670
     },
     {
       "epoch": 161.2,
-      "grad_norm": 8.189730644226074,
+      "grad_norm": 6.866076946258545,
       "learning_rate": 4.626626626626627e-06,
-      "loss": 0.9902,
+      "loss": 0.9832,
       "step": 53680
     },
     {
       "epoch": 161.23,
-      "grad_norm": 7.248913764953613,
+      "grad_norm": 6.266598701477051,
       "learning_rate": 4.625625625625626e-06,
-      "loss": 1.0619,
+      "loss": 1.0622,
       "step": 53690
     },
     {
       "epoch": 161.26,
-      "grad_norm": 8.957290649414062,
+      "grad_norm": 6.659067153930664,
       "learning_rate": 4.624624624624625e-06,
-      "loss": 1.0676,
+      "loss": 1.0687,
       "step": 53700
     },
     {
       "epoch": 161.29,
-      "grad_norm": 4.299274921417236,
+      "grad_norm": 4.304501533508301,
       "learning_rate": 4.6236236236236245e-06,
-      "loss": 0.9927,
+      "loss": 0.9894,
       "step": 53710
     },
     {
       "epoch": 161.32,
-      "grad_norm": 8.12605094909668,
+      "grad_norm": 9.394484519958496,
       "learning_rate": 4.622622622622623e-06,
-      "loss": 1.0361,
+      "loss": 1.0262,
       "step": 53720
     },
     {
       "epoch": 161.35,
-      "grad_norm": 7.697816848754883,
+      "grad_norm": 7.9297332763671875,
       "learning_rate": 4.621621621621622e-06,
-      "loss": 0.9922,
+      "loss": 0.9946,
       "step": 53730
     },
     {
       "epoch": 161.38,
-      "grad_norm": 5.256916046142578,
+      "grad_norm": 7.792161464691162,
       "learning_rate": 4.620620620620621e-06,
-      "loss": 1.0157,
+      "loss": 1.0191,
       "step": 53740
     },
     {
       "epoch": 161.41,
-      "grad_norm": 10.358504295349121,
+      "grad_norm": 8.001723289489746,
       "learning_rate": 4.619619619619621e-06,
-      "loss": 0.9688,
+      "loss": 0.9604,
       "step": 53750
     },
     {
       "epoch": 161.44,
-      "grad_norm": 5.670978546142578,
+      "grad_norm": 6.047107696533203,
       "learning_rate": 4.618618618618619e-06,
-      "loss": 1.0112,
+      "loss": 1.0123,
       "step": 53760
     },
     {
       "epoch": 161.47,
-      "grad_norm": 7.69520378112793,
+      "grad_norm": 10.359033584594727,
       "learning_rate": 4.617617617617618e-06,
-      "loss": 1.0236,
+      "loss": 1.0187,
       "step": 53770
     },
     {
       "epoch": 161.5,
-      "grad_norm": 7.205333709716797,
+      "grad_norm": 5.734325885772705,
       "learning_rate": 4.616616616616617e-06,
-      "loss": 0.9759,
+      "loss": 0.9741,
       "step": 53780
     },
     {
       "epoch": 161.53,
-      "grad_norm": 6.849427700042725,
+      "grad_norm": 5.817820072174072,
       "learning_rate": 4.615615615615616e-06,
-      "loss": 0.9805,
+      "loss": 0.972,
       "step": 53790
     },
     {
       "epoch": 161.56,
-      "grad_norm": 7.2671637535095215,
+      "grad_norm": 9.420005798339844,
       "learning_rate": 4.614614614614614e-06,
-      "loss": 0.9588,
+      "loss": 0.9595,
       "step": 53800
     },
     {
       "epoch": 161.59,
-      "grad_norm": 4.452743053436279,
+      "grad_norm": 4.623761177062988,
       "learning_rate": 4.613613613613614e-06,
-      "loss": 1.0193,
+      "loss": 1.002,
       "step": 53810
     },
     {
       "epoch": 161.62,
-      "grad_norm": 6.5895514488220215,
+      "grad_norm": 7.1768646240234375,
       "learning_rate": 4.612612612612613e-06,
-      "loss": 0.9427,
+      "loss": 0.9378,
       "step": 53820
     },
     {
       "epoch": 161.65,
-      "grad_norm": 6.3752546310424805,
+      "grad_norm": 5.9058732986450195,
       "learning_rate": 4.611611611611612e-06,
-      "loss": 0.9456,
+      "loss": 0.9469,
       "step": 53830
     },
     {
       "epoch": 161.68,
-      "grad_norm": 6.892521381378174,
+      "grad_norm": 7.3201141357421875,
       "learning_rate": 4.6106106106106105e-06,
-      "loss": 0.9525,
+      "loss": 0.9535,
       "step": 53840
     },
     {
       "epoch": 161.71,
-      "grad_norm": 8.918933868408203,
+      "grad_norm": 12.06292724609375,
       "learning_rate": 4.60960960960961e-06,
-      "loss": 1.0286,
+      "loss": 1.0305,
       "step": 53850
     },
     {
       "epoch": 161.74,
-      "grad_norm": 10.49914836883545,
+      "grad_norm": 9.242270469665527,
       "learning_rate": 4.608608608608609e-06,
-      "loss": 0.9151,
+      "loss": 0.9119,
       "step": 53860
     },
     {
       "epoch": 161.77,
-      "grad_norm": 8.263923645019531,
+      "grad_norm": 7.711277008056641,
       "learning_rate": 4.607607607607608e-06,
-      "loss": 0.9889,
+      "loss": 0.9941,
       "step": 53870
     },
     {
       "epoch": 161.8,
-      "grad_norm": 5.3820343017578125,
+      "grad_norm": 6.423284530639648,
       "learning_rate": 4.6066066066066066e-06,
-      "loss": 1.052,
+      "loss": 1.0553,
       "step": 53880
     },
     {
       "epoch": 161.83,
-      "grad_norm": 7.2170586585998535,
+      "grad_norm": 5.670832633972168,
       "learning_rate": 4.605605605605606e-06,
-      "loss": 1.0172,
+      "loss": 1.019,
       "step": 53890
     },
     {
       "epoch": 161.86,
-      "grad_norm": 5.68372106552124,
+      "grad_norm": 5.948126316070557,
       "learning_rate": 4.604604604604605e-06,
-      "loss": 0.9602,
+      "loss": 0.9578,
       "step": 53900
     },
     {
       "epoch": 161.89,
-      "grad_norm": 6.137324333190918,
+      "grad_norm": 8.022018432617188,
       "learning_rate": 4.603603603603604e-06,
-      "loss": 0.9671,
+      "loss": 0.9658,
       "step": 53910
     },
     {
       "epoch": 161.92,
-      "grad_norm": 5.167444705963135,
+      "grad_norm": 4.222145080566406,
       "learning_rate": 4.602602602602603e-06,
-      "loss": 0.9745,
+      "loss": 0.9749,
       "step": 53920
     },
     {
       "epoch": 161.95,
-      "grad_norm": 6.009934902191162,
+      "grad_norm": 5.290371894836426,
       "learning_rate": 4.601601601601602e-06,
-      "loss": 1.012,
+      "loss": 1.014,
       "step": 53930
     },
     {
       "epoch": 161.98,
-      "grad_norm": 5.6088643074035645,
+      "grad_norm": 7.4633307456970215,
       "learning_rate": 4.600600600600601e-06,
-      "loss": 0.9738,
+      "loss": 0.9678,
       "step": 53940
     },
     {
       "epoch": 162.0,
-      "eval_accuracy": 0.8205,
-      "eval_loss": 0.5894484519958496,
-      "eval_runtime": 5.4687,
-      "eval_samples_per_second": 1828.604,
-      "eval_steps_per_second": 7.314,
+      "eval_accuracy": 0.8223,
+      "eval_loss": 0.5885720252990723,
+      "eval_runtime": 5.4106,
+      "eval_samples_per_second": 1848.219,
+      "eval_steps_per_second": 7.393,
       "step": 53946
     },
     {
       "epoch": 162.01,
-      "grad_norm": 4.641473770141602,
+      "grad_norm": 7.797116279602051,
       "learning_rate": 4.5995995995996e-06,
       "loss": 1.1247,
       "step": 53950
     },
     {
       "epoch": 162.04,
-      "grad_norm": 7.307536602020264,
+      "grad_norm": 5.93387508392334,
       "learning_rate": 4.5985985985985996e-06,
-      "loss": 0.9611,
+      "loss": 0.9494,
       "step": 53960
     },
     {
       "epoch": 162.07,
-      "grad_norm": 5.8001861572265625,
+      "grad_norm": 6.860488414764404,
       "learning_rate": 4.597597597597598e-06,
-      "loss": 1.0463,
+      "loss": 1.0467,
       "step": 53970
     },
     {
       "epoch": 162.1,
-      "grad_norm": 5.435837268829346,
+      "grad_norm": 6.020226955413818,
       "learning_rate": 4.596596596596597e-06,
-      "loss": 1.0394,
+      "loss": 1.0469,
       "step": 53980
     },
     {
       "epoch": 162.13,
-      "grad_norm": 7.4731950759887695,
+      "grad_norm": 7.200310707092285,
       "learning_rate": 4.595595595595596e-06,
-      "loss": 1.0348,
+      "loss": 1.0428,
       "step": 53990
     },
     {
       "epoch": 162.16,
-      "grad_norm": 6.203253746032715,
+      "grad_norm": 5.575154781341553,
       "learning_rate": 4.594594594594596e-06,
-      "loss": 1.0079,
+      "loss": 1.0101,
       "step": 54000
     },
     {
       "epoch": 162.19,
-      "grad_norm": 5.900053024291992,
+      "grad_norm": 6.571534156799316,
       "learning_rate": 4.593593593593593e-06,
-      "loss": 0.992,
+      "loss": 0.9797,
       "step": 54010
     },
     {
       "epoch": 162.22,
-      "grad_norm": 4.599884986877441,
+      "grad_norm": 4.125630855560303,
       "learning_rate": 4.592592592592593e-06,
-      "loss": 0.9215,
+      "loss": 0.9225,
       "step": 54020
     },
     {
       "epoch": 162.25,
-      "grad_norm": 7.358728885650635,
+      "grad_norm": 5.6402788162231445,
       "learning_rate": 4.591591591591592e-06,
-      "loss": 0.9542,
+      "loss": 0.9702,
       "step": 54030
     },
     {
       "epoch": 162.28,
-      "grad_norm": 6.5827860832214355,
+      "grad_norm": 8.896785736083984,
       "learning_rate": 4.590590590590591e-06,
-      "loss": 1.0281,
+      "loss": 1.0327,
       "step": 54040
     },
     {
       "epoch": 162.31,
-      "grad_norm": 6.895693302154541,
+      "grad_norm": 6.151606559753418,
       "learning_rate": 4.5895895895895895e-06,
-      "loss": 1.0359,
+      "loss": 1.0243,
       "step": 54050
     },
     {
       "epoch": 162.34,
-      "grad_norm": 6.505865097045898,
+      "grad_norm": 6.0738372802734375,
       "learning_rate": 4.588588588588589e-06,
-      "loss": 1.0496,
+      "loss": 1.0551,
       "step": 54060
     },
     {
       "epoch": 162.37,
-      "grad_norm": 6.381886005401611,
+      "grad_norm": 6.513808727264404,
       "learning_rate": 4.587587587587588e-06,
-      "loss": 0.9817,
+      "loss": 0.965,
       "step": 54070
     },
     {
       "epoch": 162.4,
-      "grad_norm": 6.373971462249756,
+      "grad_norm": 6.049448013305664,
       "learning_rate": 4.586586586586587e-06,
-      "loss": 1.0108,
+      "loss": 1.0049,
       "step": 54080
     },
     {
       "epoch": 162.43,
-      "grad_norm": 5.954143047332764,
+      "grad_norm": 6.2911763191223145,
       "learning_rate": 4.5855855855855855e-06,
-      "loss": 0.9472,
+      "loss": 0.9479,
       "step": 54090
     },
     {
       "epoch": 162.46,
-      "grad_norm": 6.259993553161621,
+      "grad_norm": 7.343038558959961,
       "learning_rate": 4.584584584584585e-06,
-      "loss": 1.1041,
+      "loss": 1.1009,
       "step": 54100
     },
     {
       "epoch": 162.49,
-      "grad_norm": 6.051600933074951,
+      "grad_norm": 6.051255226135254,
       "learning_rate": 4.583583583583584e-06,
-      "loss": 0.9834,
+      "loss": 0.976,
       "step": 54110
     },
     {
       "epoch": 162.52,
-      "grad_norm": 7.294398307800293,
+      "grad_norm": 9.735627174377441,
       "learning_rate": 4.582582582582583e-06,
-      "loss": 0.9713,
+      "loss": 0.9783,
       "step": 54120
     },
     {
       "epoch": 162.55,
-      "grad_norm": 6.568013668060303,
+      "grad_norm": 5.339836120605469,
       "learning_rate": 4.581581581581582e-06,
-      "loss": 0.9548,
+      "loss": 0.9528,
       "step": 54130
     },
     {
       "epoch": 162.58,
-      "grad_norm": 8.103934288024902,
+      "grad_norm": 7.637454986572266,
       "learning_rate": 4.580580580580581e-06,
-      "loss": 0.9769,
+      "loss": 0.976,
       "step": 54140
     },
     {
       "epoch": 162.61,
-      "grad_norm": 7.661509990692139,
+      "grad_norm": 10.314615249633789,
       "learning_rate": 4.57957957957958e-06,
-      "loss": 1.0024,
+      "loss": 1.0053,
       "step": 54150
     },
     {
       "epoch": 162.64,
-      "grad_norm": 6.689609050750732,
+      "grad_norm": 10.583759307861328,
       "learning_rate": 4.578578578578579e-06,
-      "loss": 0.9632,
+      "loss": 0.9682,
       "step": 54160
     },
     {
       "epoch": 162.67,
-      "grad_norm": 8.792977333068848,
+      "grad_norm": 9.799603462219238,
       "learning_rate": 4.577577577577578e-06,
-      "loss": 1.0155,
+      "loss": 1.0103,
       "step": 54170
     },
     {
       "epoch": 162.7,
-      "grad_norm": 6.2813720703125,
+      "grad_norm": 5.145797252655029,
       "learning_rate": 4.576576576576577e-06,
-      "loss": 1.0422,
+      "loss": 1.0351,
       "step": 54180
     },
     {
       "epoch": 162.73,
-      "grad_norm": 9.754347801208496,
+      "grad_norm": 8.119501113891602,
       "learning_rate": 4.575575575575576e-06,
-      "loss": 0.9872,
+      "loss": 0.9787,
       "step": 54190
     },
     {
       "epoch": 162.76,
-      "grad_norm": 7.649877548217773,
+      "grad_norm": 7.848708152770996,
       "learning_rate": 4.574574574574575e-06,
-      "loss": 1.0154,
+      "loss": 1.0087,
       "step": 54200
     },
     {
       "epoch": 162.79,
-      "grad_norm": 5.3908257484436035,
+      "grad_norm": 6.627180576324463,
       "learning_rate": 4.573573573573574e-06,
-      "loss": 1.0094,
+      "loss": 1.0101,
       "step": 54210
     },
     {
       "epoch": 162.82,
-      "grad_norm": 7.8652191162109375,
+      "grad_norm": 8.047491073608398,
       "learning_rate": 4.572572572572573e-06,
-      "loss": 0.9753,
+      "loss": 0.9637,
       "step": 54220
     },
     {
       "epoch": 162.85,
-      "grad_norm": 6.0022172927856445,
+      "grad_norm": 6.2589006423950195,
       "learning_rate": 4.571571571571572e-06,
-      "loss": 0.9592,
+      "loss": 0.9595,
       "step": 54230
     },
     {
       "epoch": 162.88,
-      "grad_norm": 6.252433776855469,
+      "grad_norm": 9.246963500976562,
       "learning_rate": 4.570570570570571e-06,
-      "loss": 0.9823,
+      "loss": 0.9785,
       "step": 54240
     },
     {
       "epoch": 162.91,
-      "grad_norm": 6.475395202636719,
+      "grad_norm": 6.684051513671875,
       "learning_rate": 4.56956956956957e-06,
-      "loss": 0.9889,
+      "loss": 0.9887,
       "step": 54250
     },
     {
       "epoch": 162.94,
-      "grad_norm": 6.912642955780029,
+      "grad_norm": 6.462885856628418,
       "learning_rate": 4.568568568568568e-06,
-      "loss": 1.0126,
+      "loss": 1.0079,
       "step": 54260
     },
     {
       "epoch": 162.97,
-      "grad_norm": 6.091406345367432,
+      "grad_norm": 8.650242805480957,
       "learning_rate": 4.567567567567568e-06,
-      "loss": 1.0487,
+      "loss": 1.0455,
       "step": 54270
     },
     {
       "epoch": 163.0,
-      "eval_accuracy": 0.8161,
-      "eval_loss": 0.596890926361084,
-      "eval_runtime": 5.8435,
-      "eval_samples_per_second": 1711.295,
-      "eval_steps_per_second": 6.845,
+      "eval_accuracy": 0.8169,
+      "eval_loss": 0.5959286093711853,
+      "eval_runtime": 5.3838,
+      "eval_samples_per_second": 1857.414,
+      "eval_steps_per_second": 7.43,
       "step": 54279
     },
     {
       "epoch": 163.0,
-      "grad_norm": 7.971333026885986,
+      "grad_norm": 8.218915939331055,
       "learning_rate": 4.566566566566567e-06,
-      "loss": 1.1063,
+      "loss": 1.1082,
       "step": 54280
     },
     {
       "epoch": 163.03,
-      "grad_norm": 5.669619083404541,
+      "grad_norm": 5.128223896026611,
       "learning_rate": 4.565565565565566e-06,
-      "loss": 0.9792,
+      "loss": 0.9703,
       "step": 54290
     },
     {
       "epoch": 163.06,
-      "grad_norm": 6.827602386474609,
+      "grad_norm": 6.045326232910156,
       "learning_rate": 4.5645645645645645e-06,
-      "loss": 1.0579,
+      "loss": 1.0382,
       "step": 54300
     },
     {
       "epoch": 163.09,
-      "grad_norm": 5.832051753997803,
+      "grad_norm": 8.270216941833496,
       "learning_rate": 4.563563563563564e-06,
-      "loss": 0.9887,
+      "loss": 0.9845,
       "step": 54310
     },
     {
       "epoch": 163.12,
-      "grad_norm": 6.2098236083984375,
+      "grad_norm": 7.184401035308838,
       "learning_rate": 4.562562562562563e-06,
-      "loss": 1.0104,
+      "loss": 1.0078,
       "step": 54320
     },
     {
       "epoch": 163.15,
-      "grad_norm": 11.937443733215332,
+      "grad_norm": 9.373382568359375,
       "learning_rate": 4.561561561561562e-06,
-      "loss": 1.1019,
+      "loss": 1.0888,
       "step": 54330
     },
     {
       "epoch": 163.18,
-      "grad_norm": 6.295870304107666,
+      "grad_norm": 5.43765115737915,
       "learning_rate": 4.5605605605605606e-06,
-      "loss": 0.9731,
+      "loss": 0.964,
       "step": 54340
     },
     {
       "epoch": 163.21,
-      "grad_norm": 7.467673301696777,
+      "grad_norm": 12.418783187866211,
       "learning_rate": 4.55955955955956e-06,
-      "loss": 1.0226,
+      "loss": 1.0279,
       "step": 54350
     },
     {
       "epoch": 163.24,
-      "grad_norm": 9.952975273132324,
+      "grad_norm": 7.67262601852417,
       "learning_rate": 4.558558558558559e-06,
-      "loss": 1.0156,
+      "loss": 1.022,
       "step": 54360
     },
     {
       "epoch": 163.27,
-      "grad_norm": 7.000222206115723,
+      "grad_norm": 6.421058654785156,
       "learning_rate": 4.557557557557558e-06,
-      "loss": 0.968,
+      "loss": 0.9622,
       "step": 54370
     },
     {
       "epoch": 163.3,
-      "grad_norm": 4.93173360824585,
+      "grad_norm": 5.418042182922363,
       "learning_rate": 4.556556556556557e-06,
-      "loss": 0.9607,
+      "loss": 0.9413,
       "step": 54380
     },
     {
       "epoch": 163.33,
-      "grad_norm": 7.319820404052734,
+      "grad_norm": 10.072169303894043,
       "learning_rate": 4.555555555555556e-06,
-      "loss": 0.9802,
+      "loss": 0.9697,
       "step": 54390
     },
     {
       "epoch": 163.36,
-      "grad_norm": 6.0701189041137695,
+      "grad_norm": 7.9035844802856445,
       "learning_rate": 4.554554554554555e-06,
-      "loss": 0.9976,
+      "loss": 0.9968,
       "step": 54400
     },
     {
       "epoch": 163.39,
-      "grad_norm": 10.454800605773926,
+      "grad_norm": 10.630640029907227,
       "learning_rate": 4.553553553553554e-06,
-      "loss": 0.9638,
+      "loss": 0.9748,
       "step": 54410
     },
     {
       "epoch": 163.42,
-      "grad_norm": 5.962608337402344,
+      "grad_norm": 7.971675872802734,
       "learning_rate": 4.552552552552553e-06,
-      "loss": 1.0382,
+      "loss": 1.028,
       "step": 54420
     },
     {
       "epoch": 163.45,
-      "grad_norm": 4.488060474395752,
+      "grad_norm": 4.570819854736328,
       "learning_rate": 4.551551551551552e-06,
-      "loss": 0.9665,
+      "loss": 0.9787,
       "step": 54430
     },
     {
       "epoch": 163.48,
-      "grad_norm": 8.17495346069336,
+      "grad_norm": 8.876788139343262,
       "learning_rate": 4.550550550550551e-06,
-      "loss": 0.9628,
+      "loss": 0.958,
       "step": 54440
     },
     {
       "epoch": 163.51,
-      "grad_norm": 5.727427959442139,
+      "grad_norm": 6.710515975952148,
       "learning_rate": 4.54954954954955e-06,
-      "loss": 1.0294,
+      "loss": 1.0258,
       "step": 54450
     },
     {
       "epoch": 163.54,
-      "grad_norm": 6.408252239227295,
+      "grad_norm": 7.038167476654053,
       "learning_rate": 4.548548548548549e-06,
-      "loss": 1.0552,
+      "loss": 1.052,
       "step": 54460
     },
     {
       "epoch": 163.57,
-      "grad_norm": 6.131401062011719,
+      "grad_norm": 7.905228137969971,
       "learning_rate": 4.547547547547547e-06,
-      "loss": 1.0419,
+      "loss": 1.0398,
       "step": 54470
     },
     {
       "epoch": 163.6,
-      "grad_norm": 6.062838554382324,
+      "grad_norm": 6.71134090423584,
       "learning_rate": 4.546546546546547e-06,
-      "loss": 1.0004,
+      "loss": 1.01,
       "step": 54480
     },
     {
       "epoch": 163.63,
-      "grad_norm": 5.807088851928711,
+      "grad_norm": 6.328392505645752,
       "learning_rate": 4.545545545545546e-06,
-      "loss": 0.9342,
+      "loss": 0.9263,
       "step": 54490
     },
     {
       "epoch": 163.66,
-      "grad_norm": 6.737570762634277,
+      "grad_norm": 8.0132417678833,
       "learning_rate": 4.544544544544545e-06,
-      "loss": 0.8922,
+      "loss": 0.8956,
       "step": 54500
     },
     {
       "epoch": 163.69,
-      "grad_norm": 9.672507286071777,
+      "grad_norm": 10.63634204864502,
       "learning_rate": 4.5435435435435435e-06,
-      "loss": 1.0096,
+      "loss": 0.9988,
       "step": 54510
     },
     {
       "epoch": 163.72,
-      "grad_norm": 7.476734161376953,
+      "grad_norm": 6.738472938537598,
       "learning_rate": 4.542542542542543e-06,
-      "loss": 0.9908,
+      "loss": 0.9943,
       "step": 54520
     },
     {
       "epoch": 163.75,
-      "grad_norm": 8.990476608276367,
+      "grad_norm": 6.861005783081055,
       "learning_rate": 4.541541541541542e-06,
-      "loss": 0.9462,
+      "loss": 0.9453,
       "step": 54530
     },
     {
       "epoch": 163.78,
-      "grad_norm": 5.771658897399902,
+      "grad_norm": 5.925947666168213,
       "learning_rate": 4.540540540540541e-06,
-      "loss": 0.9742,
+      "loss": 0.9874,
       "step": 54540
     },
     {
       "epoch": 163.81,
-      "grad_norm": 5.626770973205566,
+      "grad_norm": 4.521169662475586,
       "learning_rate": 4.5395395395395395e-06,
-      "loss": 0.9915,
+      "loss": 0.9953,
       "step": 54550
     },
     {
       "epoch": 163.84,
-      "grad_norm": 7.52364444732666,
+      "grad_norm": 8.441594123840332,
       "learning_rate": 4.538538538538539e-06,
-      "loss": 1.0382,
+      "loss": 1.0372,
       "step": 54560
     },
     {
       "epoch": 163.87,
-      "grad_norm": 6.490213394165039,
+      "grad_norm": 5.967195987701416,
       "learning_rate": 4.537537537537538e-06,
-      "loss": 0.912,
+      "loss": 0.9098,
       "step": 54570
     },
     {
       "epoch": 163.9,
-      "grad_norm": 7.625494003295898,
+      "grad_norm": 8.108692169189453,
       "learning_rate": 4.536536536536537e-06,
-      "loss": 1.0205,
+      "loss": 1.0152,
       "step": 54580
     },
     {
       "epoch": 163.93,
-      "grad_norm": 8.248053550720215,
+      "grad_norm": 8.278778076171875,
       "learning_rate": 4.535535535535536e-06,
-      "loss": 1.0749,
+      "loss": 1.0706,
       "step": 54590
     },
     {
       "epoch": 163.96,
-      "grad_norm": 6.347767353057861,
+      "grad_norm": 5.35346794128418,
       "learning_rate": 4.534534534534535e-06,
-      "loss": 1.0032,
+      "loss": 0.9833,
       "step": 54600
     },
     {
       "epoch": 163.99,
-      "grad_norm": 5.495870590209961,
+      "grad_norm": 8.511323928833008,
       "learning_rate": 4.533533533533534e-06,
-      "loss": 1.0434,
+      "loss": 1.0475,
       "step": 54610
     },
     {
       "epoch": 164.0,
-      "eval_accuracy": 0.8173,
-      "eval_loss": 0.5946216583251953,
-      "eval_runtime": 5.4002,
-      "eval_samples_per_second": 1851.787,
-      "eval_steps_per_second": 7.407,
+      "eval_accuracy": 0.821,
+      "eval_loss": 0.5922606587409973,
+      "eval_runtime": 5.4961,
+      "eval_samples_per_second": 1819.459,
+      "eval_steps_per_second": 7.278,
       "step": 54612
     },
     {
       "epoch": 164.02,
-      "grad_norm": 5.980034828186035,
+      "grad_norm": 5.854613304138184,
       "learning_rate": 4.532532532532533e-06,
-      "loss": 1.0636,
+      "loss": 1.0651,
       "step": 54620
     },
     {
       "epoch": 164.05,
-      "grad_norm": 6.761343479156494,
+      "grad_norm": 10.715947151184082,
       "learning_rate": 4.531531531531532e-06,
-      "loss": 1.0007,
+      "loss": 0.9992,
       "step": 54630
     },
     {
       "epoch": 164.08,
-      "grad_norm": 4.870825290679932,
+      "grad_norm": 5.568942546844482,
       "learning_rate": 4.530530530530531e-06,
-      "loss": 1.0157,
+      "loss": 1.0174,
       "step": 54640
     },
     {
       "epoch": 164.11,
-      "grad_norm": 6.831179141998291,
+      "grad_norm": 5.681856632232666,
       "learning_rate": 4.52952952952953e-06,
-      "loss": 0.9888,
+      "loss": 1.0005,
       "step": 54650
     },
     {
       "epoch": 164.14,
-      "grad_norm": 6.65612268447876,
+      "grad_norm": 6.430143356323242,
       "learning_rate": 4.528528528528529e-06,
-      "loss": 1.048,
+      "loss": 1.0398,
       "step": 54660
     },
     {
       "epoch": 164.17,
-      "grad_norm": 6.953721523284912,
+      "grad_norm": 6.374128818511963,
       "learning_rate": 4.527527527527528e-06,
-      "loss": 0.9537,
+      "loss": 0.9451,
       "step": 54670
     },
     {
       "epoch": 164.2,
-      "grad_norm": 5.729317665100098,
+      "grad_norm": 10.036372184753418,
       "learning_rate": 4.526526526526527e-06,
-      "loss": 1.0149,
+      "loss": 1.0254,
       "step": 54680
     },
     {
       "epoch": 164.23,
-      "grad_norm": 7.315549850463867,
+      "grad_norm": 7.193665981292725,
       "learning_rate": 4.525525525525526e-06,
-      "loss": 0.9534,
+      "loss": 0.9454,
       "step": 54690
     },
     {
       "epoch": 164.26,
-      "grad_norm": 5.5425190925598145,
+      "grad_norm": 7.364106178283691,
       "learning_rate": 4.524524524524525e-06,
-      "loss": 0.9832,
+      "loss": 0.9845,
       "step": 54700
     },
     {
       "epoch": 164.29,
-      "grad_norm": 4.742002964019775,
+      "grad_norm": 4.667991638183594,
       "learning_rate": 4.523523523523524e-06,
-      "loss": 0.9697,
+      "loss": 0.9649,
       "step": 54710
     },
     {
       "epoch": 164.32,
-      "grad_norm": 6.681662082672119,
+      "grad_norm": 7.346385955810547,
       "learning_rate": 4.522522522522522e-06,
-      "loss": 1.0103,
+      "loss": 1.0113,
       "step": 54720
     },
     {
       "epoch": 164.35,
-      "grad_norm": 6.070363521575928,
+      "grad_norm": 7.6182146072387695,
       "learning_rate": 4.521521521521522e-06,
-      "loss": 1.011,
+      "loss": 1.0115,
       "step": 54730
     },
     {
       "epoch": 164.38,
-      "grad_norm": 6.327245235443115,
+      "grad_norm": 6.9833269119262695,
       "learning_rate": 4.520520520520521e-06,
-      "loss": 1.0556,
+      "loss": 1.0541,
       "step": 54740
     },
     {
       "epoch": 164.41,
-      "grad_norm": 5.822882652282715,
+      "grad_norm": 6.4968061447143555,
       "learning_rate": 4.51951951951952e-06,
-      "loss": 1.0019,
+      "loss": 1.0033,
       "step": 54750
     },
     {
       "epoch": 164.44,
-      "grad_norm": 6.157587051391602,
+      "grad_norm": 7.213491439819336,
       "learning_rate": 4.5185185185185185e-06,
-      "loss": 1.0445,
+      "loss": 1.0471,
       "step": 54760
     },
     {
       "epoch": 164.47,
-      "grad_norm": 8.6544828414917,
+      "grad_norm": 10.853086471557617,
       "learning_rate": 4.517517517517518e-06,
-      "loss": 0.9741,
+      "loss": 0.9698,
       "step": 54770
     },
     {
       "epoch": 164.5,
-      "grad_norm": 5.210467338562012,
+      "grad_norm": 5.68464469909668,
       "learning_rate": 4.516516516516517e-06,
-      "loss": 0.9567,
+      "loss": 0.9657,
       "step": 54780
     },
     {
       "epoch": 164.53,
-      "grad_norm": 5.082479000091553,
+      "grad_norm": 5.849269866943359,
       "learning_rate": 4.515515515515516e-06,
-      "loss": 0.9702,
+      "loss": 0.9684,
       "step": 54790
     },
     {
       "epoch": 164.56,
-      "grad_norm": 8.50968074798584,
+      "grad_norm": 10.126883506774902,
       "learning_rate": 4.5145145145145146e-06,
-      "loss": 1.0041,
+      "loss": 1.0023,
       "step": 54800
     },
     {
       "epoch": 164.59,
-      "grad_norm": 6.1491780281066895,
+      "grad_norm": 11.052619934082031,
       "learning_rate": 4.513513513513514e-06,
-      "loss": 1.0462,
+      "loss": 1.0367,
       "step": 54810
     },
     {
       "epoch": 164.62,
-      "grad_norm": 8.555598258972168,
+      "grad_norm": 8.647212982177734,
       "learning_rate": 4.512512512512513e-06,
-      "loss": 0.9979,
+      "loss": 1.0103,
       "step": 54820
     },
     {
       "epoch": 164.65,
-      "grad_norm": 7.619609832763672,
+      "grad_norm": 9.249083518981934,
       "learning_rate": 4.511511511511512e-06,
-      "loss": 0.8769,
+      "loss": 0.8842,
       "step": 54830
     },
     {
       "epoch": 164.68,
-      "grad_norm": 6.0073137283325195,
+      "grad_norm": 5.8307976722717285,
       "learning_rate": 4.510510510510511e-06,
-      "loss": 0.9348,
+      "loss": 0.9337,
       "step": 54840
     },
     {
       "epoch": 164.71,
-      "grad_norm": 4.846462726593018,
+      "grad_norm": 5.322102069854736,
       "learning_rate": 4.50950950950951e-06,
-      "loss": 0.9897,
+      "loss": 0.9947,
       "step": 54850
     },
     {
       "epoch": 164.74,
-      "grad_norm": 8.565136909484863,
+      "grad_norm": 5.21377420425415,
       "learning_rate": 4.508508508508509e-06,
-      "loss": 0.9235,
+      "loss": 0.929,
       "step": 54860
     },
     {
       "epoch": 164.77,
-      "grad_norm": 7.590268611907959,
+      "grad_norm": 8.306624412536621,
       "learning_rate": 4.507507507507508e-06,
-      "loss": 0.9954,
+      "loss": 0.9978,
       "step": 54870
     },
     {
       "epoch": 164.8,
-      "grad_norm": 5.600142955780029,
+      "grad_norm": 7.268520832061768,
       "learning_rate": 4.506506506506507e-06,
-      "loss": 0.9763,
+      "loss": 0.9708,
       "step": 54880
     },
     {
       "epoch": 164.83,
-      "grad_norm": 7.954755783081055,
+      "grad_norm": 8.603504180908203,
       "learning_rate": 4.505505505505506e-06,
-      "loss": 1.0133,
+      "loss": 1.0236,
       "step": 54890
     },
     {
       "epoch": 164.86,
-      "grad_norm": 6.407777309417725,
+      "grad_norm": 7.5973711013793945,
       "learning_rate": 4.504504504504505e-06,
-      "loss": 1.0144,
+      "loss": 1.027,
       "step": 54900
     },
     {
       "epoch": 164.89,
-      "grad_norm": 6.071404457092285,
+      "grad_norm": 7.3321533203125,
       "learning_rate": 4.503503503503504e-06,
-      "loss": 0.9802,
+      "loss": 0.9791,
       "step": 54910
     },
     {
       "epoch": 164.92,
-      "grad_norm": 8.03786849975586,
+      "grad_norm": 8.218749046325684,
       "learning_rate": 4.502502502502503e-06,
-      "loss": 0.9796,
+      "loss": 0.985,
       "step": 54920
     },
     {
       "epoch": 164.95,
-      "grad_norm": 8.092543601989746,
+      "grad_norm": 7.738978385925293,
       "learning_rate": 4.501501501501501e-06,
-      "loss": 1.0709,
+      "loss": 1.0708,
       "step": 54930
     },
     {
       "epoch": 164.98,
-      "grad_norm": 5.243190765380859,
+      "grad_norm": 5.9854512214660645,
       "learning_rate": 4.500500500500501e-06,
-      "loss": 0.9916,
+      "loss": 0.9951,
       "step": 54940
     },
     {
       "epoch": 165.0,
-      "eval_accuracy": 0.8194,
-      "eval_loss": 0.5959897041320801,
-      "eval_runtime": 5.4037,
-      "eval_samples_per_second": 1850.594,
-      "eval_steps_per_second": 7.402,
+      "eval_accuracy": 0.8201,
+      "eval_loss": 0.5928794145584106,
+      "eval_runtime": 5.3601,
+      "eval_samples_per_second": 1865.622,
+      "eval_steps_per_second": 7.462,
       "step": 54945
     },
     {
       "epoch": 165.02,
-      "grad_norm": 7.208664417266846,
+      "grad_norm": 6.955272197723389,
       "learning_rate": 4.4994994994995e-06,
-      "loss": 1.2489,
+      "loss": 1.2465,
       "step": 54950
     },
     {
       "epoch": 165.05,
-      "grad_norm": 7.111008644104004,
+      "grad_norm": 6.774902820587158,
       "learning_rate": 4.498498498498499e-06,
-      "loss": 1.0287,
+      "loss": 1.0242,
       "step": 54960
     },
     {
       "epoch": 165.08,
-      "grad_norm": 5.975818634033203,
+      "grad_norm": 6.1631927490234375,
       "learning_rate": 4.4974974974974974e-06,
-      "loss": 1.029,
+      "loss": 1.0192,
       "step": 54970
     },
     {
       "epoch": 165.11,
-      "grad_norm": 7.545650482177734,
+      "grad_norm": 8.054332733154297,
       "learning_rate": 4.496496496496497e-06,
-      "loss": 1.0197,
+      "loss": 1.0174,
       "step": 54980
     },
     {
       "epoch": 165.14,
-      "grad_norm": 6.795822620391846,
+      "grad_norm": 6.782811641693115,
       "learning_rate": 4.495495495495496e-06,
-      "loss": 1.0826,
+      "loss": 1.0852,
       "step": 54990
     },
     {
       "epoch": 165.17,
-      "grad_norm": 4.729997634887695,
+      "grad_norm": 5.632568836212158,
       "learning_rate": 4.494494494494495e-06,
-      "loss": 0.9961,
+      "loss": 0.9882,
       "step": 55000
     },
     {
       "epoch": 165.2,
-      "grad_norm": 6.065668106079102,
+      "grad_norm": 6.7926435470581055,
       "learning_rate": 4.4934934934934935e-06,
-      "loss": 1.0341,
+      "loss": 1.0302,
       "step": 55010
     },
     {
       "epoch": 165.23,
-      "grad_norm": 7.720165729522705,
+      "grad_norm": 9.76214599609375,
       "learning_rate": 4.492492492492493e-06,
-      "loss": 1.06,
+      "loss": 1.0733,
       "step": 55020
     },
     {
       "epoch": 165.26,
-      "grad_norm": 8.411377906799316,
+      "grad_norm": 6.8757100105285645,
       "learning_rate": 4.491491491491492e-06,
-      "loss": 1.0395,
+      "loss": 1.0285,
       "step": 55030
     },
     {
       "epoch": 165.29,
-      "grad_norm": 6.325547218322754,
+      "grad_norm": 5.215798377990723,
       "learning_rate": 4.490490490490491e-06,
-      "loss": 1.0389,
+      "loss": 1.0396,
       "step": 55040
     },
     {
       "epoch": 165.32,
-      "grad_norm": 8.465347290039062,
+      "grad_norm": 8.041476249694824,
       "learning_rate": 4.48948948948949e-06,
-      "loss": 1.0275,
+      "loss": 1.0358,
       "step": 55050
     },
     {
       "epoch": 165.35,
-      "grad_norm": 7.782208442687988,
+      "grad_norm": 9.271126747131348,
       "learning_rate": 4.488488488488489e-06,
-      "loss": 1.0682,
+      "loss": 1.0597,
       "step": 55060
     },
     {
       "epoch": 165.38,
-      "grad_norm": 6.063454627990723,
+      "grad_norm": 5.961141109466553,
       "learning_rate": 4.487487487487488e-06,
-      "loss": 0.9375,
+      "loss": 0.9432,
       "step": 55070
     },
     {
       "epoch": 165.41,
-      "grad_norm": 4.847438812255859,
+      "grad_norm": 7.423973083496094,
       "learning_rate": 4.486486486486487e-06,
-      "loss": 1.0113,
+      "loss": 1.0142,
       "step": 55080
     },
     {
       "epoch": 165.44,
-      "grad_norm": 7.491527557373047,
+      "grad_norm": 11.59479808807373,
       "learning_rate": 4.485485485485486e-06,
-      "loss": 0.9696,
+      "loss": 0.9719,
       "step": 55090
     },
     {
       "epoch": 165.47,
-      "grad_norm": 7.092261791229248,
+      "grad_norm": 8.287652969360352,
       "learning_rate": 4.484484484484485e-06,
-      "loss": 1.0096,
+      "loss": 1.0141,
       "step": 55100
     },
     {
       "epoch": 165.5,
-      "grad_norm": 7.245370388031006,
+      "grad_norm": 6.82904052734375,
       "learning_rate": 4.483483483483484e-06,
-      "loss": 0.9472,
+      "loss": 0.9456,
       "step": 55110
     },
     {
       "epoch": 165.53,
-      "grad_norm": 6.250280380249023,
+      "grad_norm": 6.957817554473877,
       "learning_rate": 4.482482482482483e-06,
-      "loss": 0.9381,
+      "loss": 0.9375,
       "step": 55120
     },
     {
       "epoch": 165.56,
-      "grad_norm": 8.756875991821289,
+      "grad_norm": 6.699901103973389,
       "learning_rate": 4.481481481481482e-06,
-      "loss": 0.9805,
+      "loss": 0.977,
       "step": 55130
     },
     {
       "epoch": 165.59,
-      "grad_norm": 4.463512420654297,
+      "grad_norm": 4.095755100250244,
       "learning_rate": 4.480480480480481e-06,
-      "loss": 1.0476,
+      "loss": 1.0386,
       "step": 55140
     },
     {
       "epoch": 165.62,
-      "grad_norm": 6.420620441436768,
+      "grad_norm": 8.088288307189941,
       "learning_rate": 4.47947947947948e-06,
-      "loss": 0.9992,
+      "loss": 0.9889,
       "step": 55150
     },
     {
       "epoch": 165.65,
-      "grad_norm": 6.556023597717285,
+      "grad_norm": 6.104094505310059,
       "learning_rate": 4.478478478478479e-06,
-      "loss": 0.9824,
+      "loss": 0.9756,
       "step": 55160
     },
     {
       "epoch": 165.68,
-      "grad_norm": 4.8240532875061035,
+      "grad_norm": 5.763723373413086,
       "learning_rate": 4.477477477477478e-06,
-      "loss": 0.945,
+      "loss": 0.9428,
       "step": 55170
     },
     {
       "epoch": 165.71,
-      "grad_norm": 6.015388011932373,
+      "grad_norm": 7.43716287612915,
       "learning_rate": 4.476476476476476e-06,
-      "loss": 0.9898,
+      "loss": 0.9886,
       "step": 55180
     },
     {
       "epoch": 165.74,
-      "grad_norm": 6.17332124710083,
+      "grad_norm": 6.438056468963623,
       "learning_rate": 4.475475475475476e-06,
-      "loss": 0.8954,
+      "loss": 0.8921,
       "step": 55190
     },
     {
       "epoch": 165.77,
-      "grad_norm": 7.171682834625244,
+      "grad_norm": 5.822615623474121,
       "learning_rate": 4.474474474474475e-06,
-      "loss": 0.9377,
+      "loss": 0.9268,
       "step": 55200
     },
     {
       "epoch": 165.8,
-      "grad_norm": 6.105165004730225,
+      "grad_norm": 6.516354560852051,
       "learning_rate": 4.473473473473474e-06,
-      "loss": 0.9771,
+      "loss": 0.9855,
       "step": 55210
     },
     {
       "epoch": 165.83,
-      "grad_norm": 6.207922458648682,
+      "grad_norm": 8.400127410888672,
       "learning_rate": 4.4724724724724725e-06,
-      "loss": 0.8958,
+      "loss": 0.9004,
       "step": 55220
     },
     {
       "epoch": 165.86,
-      "grad_norm": 6.049530506134033,
+      "grad_norm": 5.810722351074219,
       "learning_rate": 4.471471471471472e-06,
-      "loss": 0.9963,
+      "loss": 1.0004,
       "step": 55230
     },
     {
       "epoch": 165.89,
-      "grad_norm": 8.149657249450684,
+      "grad_norm": 8.967561721801758,
       "learning_rate": 4.470470470470471e-06,
-      "loss": 0.9062,
+      "loss": 0.9134,
       "step": 55240
     },
     {
       "epoch": 165.92,
-      "grad_norm": 8.115234375,
+      "grad_norm": 6.568699836730957,
       "learning_rate": 4.46946946946947e-06,
-      "loss": 0.9974,
+      "loss": 0.9954,
       "step": 55250
     },
     {
       "epoch": 165.95,
-      "grad_norm": 9.334964752197266,
+      "grad_norm": 8.812065124511719,
       "learning_rate": 4.4684684684684686e-06,
-      "loss": 1.0331,
+      "loss": 1.0253,
       "step": 55260
     },
     {
       "epoch": 165.98,
-      "grad_norm": 5.135908126831055,
+      "grad_norm": 4.364644527435303,
       "learning_rate": 4.467467467467468e-06,
-      "loss": 0.9596,
+      "loss": 0.9673,
       "step": 55270
     },
     {
       "epoch": 166.0,
-      "eval_accuracy": 0.8194,
-      "eval_loss": 0.5889564752578735,
-      "eval_runtime": 5.5314,
-      "eval_samples_per_second": 1807.868,
-      "eval_steps_per_second": 7.231,
+      "eval_accuracy": 0.8218,
+      "eval_loss": 0.587411642074585,
+      "eval_runtime": 5.5615,
+      "eval_samples_per_second": 1798.061,
+      "eval_steps_per_second": 7.192,
       "step": 55278
     },
     {
       "epoch": 166.01,
-      "grad_norm": 6.010180473327637,
+      "grad_norm": 6.806262969970703,
       "learning_rate": 4.466466466466467e-06,
-      "loss": 0.9682,
+      "loss": 0.9898,
       "step": 55280
     },
     {
       "epoch": 166.04,
-      "grad_norm": 6.235501289367676,
+      "grad_norm": 5.86610221862793,
       "learning_rate": 4.465465465465465e-06,
-      "loss": 0.9853,
+      "loss": 0.9779,
       "step": 55290
     },
     {
       "epoch": 166.07,
-      "grad_norm": 6.489870548248291,
+      "grad_norm": 6.787936687469482,
       "learning_rate": 4.464464464464465e-06,
-      "loss": 1.0098,
+      "loss": 1.0003,
       "step": 55300
     },
     {
       "epoch": 166.1,
-      "grad_norm": 4.798956871032715,
+      "grad_norm": 5.119420528411865,
       "learning_rate": 4.463463463463464e-06,
-      "loss": 0.9764,
+      "loss": 0.9635,
       "step": 55310
     },
     {
       "epoch": 166.13,
-      "grad_norm": 5.8682732582092285,
+      "grad_norm": 9.043368339538574,
       "learning_rate": 4.462462462462463e-06,
-      "loss": 0.9884,
+      "loss": 0.9977,
       "step": 55320
     },
     {
       "epoch": 166.16,
-      "grad_norm": 8.350728988647461,
+      "grad_norm": 9.33957290649414,
       "learning_rate": 4.461461461461462e-06,
-      "loss": 1.0316,
+      "loss": 1.0364,
       "step": 55330
     },
     {
       "epoch": 166.19,
-      "grad_norm": 7.851647853851318,
+      "grad_norm": 5.668789386749268,
       "learning_rate": 4.460460460460461e-06,
-      "loss": 0.9667,
+      "loss": 0.9596,
       "step": 55340
     },
     {
       "epoch": 166.22,
-      "grad_norm": 6.996201992034912,
+      "grad_norm": 7.884900093078613,
       "learning_rate": 4.45945945945946e-06,
-      "loss": 0.9918,
+      "loss": 1.003,
       "step": 55350
     },
     {
       "epoch": 166.25,
-      "grad_norm": 4.51870584487915,
+      "grad_norm": 4.02178430557251,
       "learning_rate": 4.458458458458459e-06,
-      "loss": 0.949,
+      "loss": 0.9406,
       "step": 55360
     },
     {
       "epoch": 166.28,
-      "grad_norm": 7.017898082733154,
+      "grad_norm": 7.78130578994751,
       "learning_rate": 4.457457457457458e-06,
-      "loss": 0.9784,
+      "loss": 0.9761,
       "step": 55370
     },
     {
       "epoch": 166.31,
-      "grad_norm": 5.977457523345947,
+      "grad_norm": 6.919050693511963,
       "learning_rate": 4.456456456456457e-06,
-      "loss": 0.9438,
+      "loss": 0.9547,
       "step": 55380
     },
     {
       "epoch": 166.34,
-      "grad_norm": 6.226626396179199,
+      "grad_norm": 9.119017601013184,
       "learning_rate": 4.455455455455455e-06,
-      "loss": 1.0025,
+      "loss": 1.0109,
       "step": 55390
     },
     {
       "epoch": 166.37,
-      "grad_norm": 6.189627647399902,
+      "grad_norm": 6.62051248550415,
       "learning_rate": 4.454454454454455e-06,
-      "loss": 1.0416,
+      "loss": 1.0381,
       "step": 55400
     },
     {
       "epoch": 166.4,
-      "grad_norm": 8.30517864227295,
+      "grad_norm": 7.675502777099609,
       "learning_rate": 4.453453453453454e-06,
-      "loss": 1.0289,
+      "loss": 1.0259,
       "step": 55410
     },
     {
       "epoch": 166.43,
-      "grad_norm": 7.192183494567871,
+      "grad_norm": 7.143414497375488,
       "learning_rate": 4.452452452452453e-06,
       "loss": 0.9873,
       "step": 55420
     },
     {
       "epoch": 166.46,
-      "grad_norm": 6.198598861694336,
+      "grad_norm": 6.631393909454346,
       "learning_rate": 4.4514514514514514e-06,
-      "loss": 0.9811,
+      "loss": 0.9754,
       "step": 55430
     },
     {
       "epoch": 166.49,
-      "grad_norm": 5.076237678527832,
+      "grad_norm": 6.871507167816162,
       "learning_rate": 4.450450450450451e-06,
       "loss": 0.9739,
       "step": 55440
     },
     {
       "epoch": 166.52,
-      "grad_norm": 7.84921407699585,
+      "grad_norm": 10.210509300231934,
       "learning_rate": 4.44944944944945e-06,
-      "loss": 0.9525,
+      "loss": 0.955,
       "step": 55450
     },
     {
       "epoch": 166.55,
-      "grad_norm": 7.679599761962891,
+      "grad_norm": 7.137550354003906,
       "learning_rate": 4.448448448448449e-06,
-      "loss": 1.0005,
+      "loss": 0.9994,
       "step": 55460
     },
     {
       "epoch": 166.58,
-      "grad_norm": 5.570200443267822,
+      "grad_norm": 5.927725791931152,
       "learning_rate": 4.4474474474474475e-06,
-      "loss": 0.9451,
+      "loss": 0.9486,
       "step": 55470
     },
     {
       "epoch": 166.61,
-      "grad_norm": 7.050314426422119,
+      "grad_norm": 7.9632978439331055,
       "learning_rate": 4.446446446446447e-06,
-      "loss": 0.9565,
+      "loss": 0.9569,
       "step": 55480
     },
     {
       "epoch": 166.64,
-      "grad_norm": 8.683314323425293,
+      "grad_norm": 8.23519229888916,
       "learning_rate": 4.445445445445446e-06,
-      "loss": 0.9526,
+      "loss": 0.9611,
       "step": 55490
     },
     {
       "epoch": 166.67,
-      "grad_norm": 4.87830924987793,
+      "grad_norm": 5.404907703399658,
       "learning_rate": 4.444444444444444e-06,
-      "loss": 1.0146,
+      "loss": 1.0259,
       "step": 55500
     },
     {
       "epoch": 166.7,
-      "grad_norm": 7.148522853851318,
+      "grad_norm": 5.308506965637207,
       "learning_rate": 4.443443443443444e-06,
-      "loss": 0.9004,
+      "loss": 0.8969,
       "step": 55510
     },
     {
       "epoch": 166.73,
-      "grad_norm": 4.3923845291137695,
+      "grad_norm": 6.286486625671387,
       "learning_rate": 4.442442442442443e-06,
-      "loss": 0.9549,
+      "loss": 0.9543,
       "step": 55520
     },
     {
       "epoch": 166.76,
-      "grad_norm": 6.754518985748291,
+      "grad_norm": 7.1688618659973145,
       "learning_rate": 4.441441441441442e-06,
-      "loss": 0.988,
+      "loss": 0.9906,
       "step": 55530
     },
     {
       "epoch": 166.79,
-      "grad_norm": 6.797971248626709,
+      "grad_norm": 6.538938999176025,
       "learning_rate": 4.44044044044044e-06,
-      "loss": 0.986,
+      "loss": 0.9969,
       "step": 55540
     },
     {
       "epoch": 166.82,
-      "grad_norm": 5.765371799468994,
+      "grad_norm": 5.89948844909668,
       "learning_rate": 4.43943943943944e-06,
-      "loss": 0.9513,
+      "loss": 0.9417,
       "step": 55550
     },
     {
       "epoch": 166.85,
-      "grad_norm": 6.155567646026611,
+      "grad_norm": 7.0622758865356445,
       "learning_rate": 4.438438438438439e-06,
-      "loss": 1.0019,
+      "loss": 0.9919,
       "step": 55560
     },
     {
       "epoch": 166.88,
-      "grad_norm": 7.893054008483887,
+      "grad_norm": 6.6869988441467285,
       "learning_rate": 4.437437437437438e-06,
-      "loss": 1.0108,
+      "loss": 1.0116,
       "step": 55570
     },
     {
       "epoch": 166.91,
-      "grad_norm": 6.1578779220581055,
+      "grad_norm": 11.250052452087402,
       "learning_rate": 4.436436436436437e-06,
-      "loss": 1.0303,
+      "loss": 1.0266,
       "step": 55580
     },
     {
       "epoch": 166.94,
-      "grad_norm": 6.176051139831543,
+      "grad_norm": 8.425325393676758,
       "learning_rate": 4.435435435435436e-06,
-      "loss": 1.0847,
+      "loss": 1.0852,
       "step": 55590
     },
     {
       "epoch": 166.97,
-      "grad_norm": 6.799369812011719,
+      "grad_norm": 6.6908464431762695,
       "learning_rate": 4.434434434434435e-06,
-      "loss": 0.9601,
+      "loss": 0.9557,
       "step": 55600
     },
     {
       "epoch": 167.0,
-      "grad_norm": 6.191830635070801,
+      "grad_norm": 6.558169841766357,
       "learning_rate": 4.433433433433434e-06,
-      "loss": 1.0006,
+      "loss": 0.9937,
       "step": 55610
     },
     {
       "epoch": 167.0,
-      "eval_accuracy": 0.8176,
-      "eval_loss": 0.5910018086433411,
-      "eval_runtime": 5.3247,
-      "eval_samples_per_second": 1878.038,
-      "eval_steps_per_second": 7.512,
+      "eval_accuracy": 0.8214,
+      "eval_loss": 0.5880972743034363,
+      "eval_runtime": 5.6523,
+      "eval_samples_per_second": 1769.192,
+      "eval_steps_per_second": 7.077,
       "step": 55611
     },
     {
       "epoch": 167.03,
-      "grad_norm": 6.282098293304443,
+      "grad_norm": 7.877981662750244,
       "learning_rate": 4.432432432432433e-06,
-      "loss": 1.1348,
+      "loss": 1.1212,
       "step": 55620
     },
     {
       "epoch": 167.06,
-      "grad_norm": 6.893001079559326,
+      "grad_norm": 6.295651435852051,
       "learning_rate": 4.431431431431432e-06,
-      "loss": 1.0299,
+      "loss": 1.0371,
       "step": 55630
     },
     {
       "epoch": 167.09,
-      "grad_norm": 8.664212226867676,
+      "grad_norm": 5.9647016525268555,
       "learning_rate": 4.43043043043043e-06,
-      "loss": 0.9323,
+      "loss": 0.9316,
       "step": 55640
     },
     {
       "epoch": 167.12,
-      "grad_norm": 7.4459919929504395,
+      "grad_norm": 5.298347473144531,
       "learning_rate": 4.42942942942943e-06,
-      "loss": 0.97,
+      "loss": 0.9797,
       "step": 55650
     },
     {
       "epoch": 167.15,
-      "grad_norm": 10.176400184631348,
+      "grad_norm": 8.888267517089844,
       "learning_rate": 4.428428428428429e-06,
-      "loss": 0.8725,
+      "loss": 0.8775,
       "step": 55660
     },
     {
       "epoch": 167.18,
-      "grad_norm": 6.238811016082764,
+      "grad_norm": 4.8528666496276855,
       "learning_rate": 4.427427427427428e-06,
-      "loss": 1.0665,
+      "loss": 1.06,
       "step": 55670
     },
     {
       "epoch": 167.21,
-      "grad_norm": 6.449116230010986,
+      "grad_norm": 8.252571105957031,
       "learning_rate": 4.4264264264264265e-06,
-      "loss": 0.9767,
+      "loss": 0.9707,
       "step": 55680
     },
     {
       "epoch": 167.24,
-      "grad_norm": 6.373172283172607,
+      "grad_norm": 7.201282024383545,
       "learning_rate": 4.425425425425426e-06,
-      "loss": 0.9331,
+      "loss": 0.9318,
       "step": 55690
     },
     {
       "epoch": 167.27,
-      "grad_norm": 6.319322109222412,
+      "grad_norm": 7.416339874267578,
       "learning_rate": 4.424424424424425e-06,
-      "loss": 0.9916,
+      "loss": 0.9961,
       "step": 55700
     },
     {
       "epoch": 167.3,
-      "grad_norm": 7.1776909828186035,
+      "grad_norm": 5.524851322174072,
       "learning_rate": 4.423423423423424e-06,
-      "loss": 0.983,
+      "loss": 0.9766,
       "step": 55710
     },
     {
       "epoch": 167.33,
-      "grad_norm": 7.395866394042969,
+      "grad_norm": 9.862366676330566,
       "learning_rate": 4.4224224224224226e-06,
-      "loss": 0.9316,
+      "loss": 0.9408,
       "step": 55720
     },
     {
       "epoch": 167.36,
-      "grad_norm": 9.426618576049805,
+      "grad_norm": 7.398355960845947,
       "learning_rate": 4.421421421421422e-06,
-      "loss": 1.0112,
+      "loss": 1.0105,
       "step": 55730
     },
     {
       "epoch": 167.39,
-      "grad_norm": 5.29486608505249,
+      "grad_norm": 6.451447010040283,
       "learning_rate": 4.420420420420421e-06,
-      "loss": 0.9841,
+      "loss": 0.9856,
       "step": 55740
     },
     {
       "epoch": 167.42,
-      "grad_norm": 8.498709678649902,
+      "grad_norm": 9.375261306762695,
       "learning_rate": 4.419419419419419e-06,
-      "loss": 0.9426,
+      "loss": 0.9432,
       "step": 55750
     },
     {
       "epoch": 167.45,
-      "grad_norm": 5.2539167404174805,
+      "grad_norm": 4.9578776359558105,
       "learning_rate": 4.418418418418419e-06,
-      "loss": 0.9781,
+      "loss": 0.9792,
       "step": 55760
     },
     {
       "epoch": 167.48,
-      "grad_norm": 7.2603302001953125,
+      "grad_norm": 8.13465404510498,
       "learning_rate": 4.417417417417418e-06,
-      "loss": 0.9631,
+      "loss": 0.9618,
       "step": 55770
     },
     {
       "epoch": 167.51,
-      "grad_norm": 9.643048286437988,
+      "grad_norm": 11.21062183380127,
       "learning_rate": 4.416416416416417e-06,
-      "loss": 0.9476,
+      "loss": 0.9502,
       "step": 55780
     },
     {
       "epoch": 167.54,
-      "grad_norm": 5.283472537994385,
+      "grad_norm": 6.060259819030762,
       "learning_rate": 4.415415415415415e-06,
-      "loss": 0.9924,
+      "loss": 0.9894,
       "step": 55790
     },
     {
       "epoch": 167.57,
-      "grad_norm": 7.067764759063721,
+      "grad_norm": 9.918739318847656,
       "learning_rate": 4.414414414414415e-06,
-      "loss": 0.9852,
+      "loss": 0.9768,
       "step": 55800
     },
     {
       "epoch": 167.6,
-      "grad_norm": 6.940587520599365,
+      "grad_norm": 5.0517802238464355,
       "learning_rate": 4.413413413413414e-06,
-      "loss": 0.97,
+      "loss": 0.9635,
       "step": 55810
     },
     {
       "epoch": 167.63,
-      "grad_norm": 6.140127658843994,
+      "grad_norm": 6.0663557052612305,
       "learning_rate": 4.412412412412413e-06,
-      "loss": 1.0007,
+      "loss": 0.9986,
       "step": 55820
     },
     {
       "epoch": 167.66,
-      "grad_norm": 7.326564788818359,
+      "grad_norm": 6.754120349884033,
       "learning_rate": 4.411411411411412e-06,
-      "loss": 1.0452,
+      "loss": 1.04,
       "step": 55830
     },
     {
       "epoch": 167.69,
-      "grad_norm": 5.222813129425049,
+      "grad_norm": 6.458491325378418,
       "learning_rate": 4.410410410410411e-06,
-      "loss": 0.9503,
+      "loss": 0.9539,
       "step": 55840
     },
     {
       "epoch": 167.72,
-      "grad_norm": 6.863008499145508,
+      "grad_norm": 7.884034633636475,
       "learning_rate": 4.409409409409409e-06,
-      "loss": 0.9434,
+      "loss": 0.9443,
       "step": 55850
     },
     {
       "epoch": 167.75,
-      "grad_norm": 5.8695831298828125,
+      "grad_norm": 8.99539852142334,
       "learning_rate": 4.408408408408409e-06,
-      "loss": 1.0258,
+      "loss": 1.0198,
       "step": 55860
     },
     {
       "epoch": 167.78,
-      "grad_norm": 6.947471618652344,
+      "grad_norm": 8.149840354919434,
       "learning_rate": 4.407407407407408e-06,
-      "loss": 0.9897,
+      "loss": 0.9912,
       "step": 55870
     },
     {
       "epoch": 167.81,
-      "grad_norm": 7.393586158752441,
+      "grad_norm": 7.187238693237305,
       "learning_rate": 4.406406406406407e-06,
-      "loss": 1.0511,
+      "loss": 1.0529,
       "step": 55880
     },
     {
       "epoch": 167.84,
-      "grad_norm": 4.935520648956299,
+      "grad_norm": 5.753887176513672,
       "learning_rate": 4.4054054054054054e-06,
-      "loss": 1.0116,
+      "loss": 1.0095,
       "step": 55890
     },
     {
       "epoch": 167.87,
-      "grad_norm": 8.507771492004395,
+      "grad_norm": 7.635948181152344,
       "learning_rate": 4.404404404404405e-06,
-      "loss": 1.0337,
+      "loss": 1.0277,
       "step": 55900
     },
     {
       "epoch": 167.9,
-      "grad_norm": 8.076346397399902,
+      "grad_norm": 10.149164199829102,
       "learning_rate": 4.403403403403404e-06,
-      "loss": 1.0569,
+      "loss": 1.0575,
       "step": 55910
     },
     {
       "epoch": 167.93,
-      "grad_norm": 8.220253944396973,
+      "grad_norm": 5.737960338592529,
       "learning_rate": 4.402402402402403e-06,
-      "loss": 0.9655,
+      "loss": 0.9705,
       "step": 55920
     },
     {
       "epoch": 167.96,
-      "grad_norm": 7.549067974090576,
+      "grad_norm": 5.802128314971924,
       "learning_rate": 4.4014014014014015e-06,
-      "loss": 1.0429,
+      "loss": 1.0245,
       "step": 55930
     },
     {
       "epoch": 167.99,
-      "grad_norm": 5.633042812347412,
+      "grad_norm": 5.154459476470947,
       "learning_rate": 4.400400400400401e-06,
-      "loss": 0.99,
+      "loss": 0.9936,
       "step": 55940
     },
     {
       "epoch": 168.0,
-      "eval_accuracy": 0.8195,
-      "eval_loss": 0.5900973081588745,
-      "eval_runtime": 5.467,
-      "eval_samples_per_second": 1829.153,
-      "eval_steps_per_second": 7.317,
+      "eval_accuracy": 0.8226,
+      "eval_loss": 0.5904180407524109,
+      "eval_runtime": 5.34,
+      "eval_samples_per_second": 1872.655,
+      "eval_steps_per_second": 7.491,
       "step": 55944
     },
     {
       "epoch": 168.02,
-      "grad_norm": 6.7439446449279785,
+      "grad_norm": 6.0123372077941895,
       "learning_rate": 4.3993993993993996e-06,
-      "loss": 1.0821,
+      "loss": 1.0819,
       "step": 55950
     },
     {
       "epoch": 168.05,
-      "grad_norm": 5.206599712371826,
+      "grad_norm": 5.342199802398682,
       "learning_rate": 4.398398398398398e-06,
-      "loss": 1.0045,
+      "loss": 1.0102,
       "step": 55960
     },
     {
       "epoch": 168.08,
-      "grad_norm": 5.353061199188232,
+      "grad_norm": 6.0317816734313965,
       "learning_rate": 4.397397397397398e-06,
-      "loss": 0.9694,
+      "loss": 0.9791,
       "step": 55970
     },
     {
       "epoch": 168.11,
-      "grad_norm": 6.095679759979248,
+      "grad_norm": 9.784164428710938,
       "learning_rate": 4.396396396396397e-06,
-      "loss": 1.011,
+      "loss": 1.0113,
       "step": 55980
     },
     {
       "epoch": 168.14,
-      "grad_norm": 7.010627746582031,
+      "grad_norm": 6.445689678192139,
       "learning_rate": 4.395395395395396e-06,
-      "loss": 0.9452,
+      "loss": 0.9374,
       "step": 55990
     },
     {
       "epoch": 168.17,
-      "grad_norm": 7.1315131187438965,
+      "grad_norm": 6.322487831115723,
       "learning_rate": 4.394394394394394e-06,
-      "loss": 0.9632,
+      "loss": 0.9653,
       "step": 56000
     },
     {
       "epoch": 168.2,
-      "grad_norm": 10.419921875,
+      "grad_norm": 8.388384819030762,
       "learning_rate": 4.393393393393394e-06,
-      "loss": 1.0222,
+      "loss": 1.0143,
       "step": 56010
     },
     {
       "epoch": 168.23,
-      "grad_norm": 8.767400741577148,
+      "grad_norm": 7.552946090698242,
       "learning_rate": 4.392392392392393e-06,
-      "loss": 1.0622,
+      "loss": 1.0545,
       "step": 56020
     },
     {
       "epoch": 168.26,
-      "grad_norm": 7.010430812835693,
+      "grad_norm": 6.689023017883301,
       "learning_rate": 4.391391391391392e-06,
-      "loss": 0.9549,
+      "loss": 0.9538,
       "step": 56030
     },
     {
       "epoch": 168.29,
-      "grad_norm": 5.282059669494629,
+      "grad_norm": 9.028752326965332,
       "learning_rate": 4.39039039039039e-06,
-      "loss": 1.0307,
+      "loss": 1.0288,
       "step": 56040
     },
     {
       "epoch": 168.32,
-      "grad_norm": 6.484198570251465,
+      "grad_norm": 9.106201171875,
       "learning_rate": 4.38938938938939e-06,
-      "loss": 0.9423,
+      "loss": 0.9408,
       "step": 56050
     },
     {
       "epoch": 168.35,
-      "grad_norm": 5.851840019226074,
+      "grad_norm": 6.175382614135742,
       "learning_rate": 4.388388388388389e-06,
-      "loss": 0.9868,
+      "loss": 0.9807,
       "step": 56060
     },
     {
       "epoch": 168.38,
-      "grad_norm": 4.828789234161377,
+      "grad_norm": 4.853939056396484,
       "learning_rate": 4.387387387387388e-06,
-      "loss": 0.951,
+      "loss": 0.9608,
       "step": 56070
     },
     {
       "epoch": 168.41,
-      "grad_norm": 7.182310104370117,
+      "grad_norm": 7.503751277923584,
       "learning_rate": 4.386386386386386e-06,
-      "loss": 1.0155,
+      "loss": 1.0229,
       "step": 56080
     },
     {
       "epoch": 168.44,
-      "grad_norm": 5.618887901306152,
+      "grad_norm": 6.429169178009033,
       "learning_rate": 4.385385385385386e-06,
-      "loss": 0.9535,
+      "loss": 0.9474,
       "step": 56090
     },
     {
       "epoch": 168.47,
-      "grad_norm": 4.988004207611084,
+      "grad_norm": 5.289485931396484,
       "learning_rate": 4.384384384384384e-06,
-      "loss": 1.022,
+      "loss": 1.015,
       "step": 56100
     },
     {
       "epoch": 168.5,
-      "grad_norm": 7.986780166625977,
+      "grad_norm": 7.6761651039123535,
       "learning_rate": 4.383383383383384e-06,
-      "loss": 1.0483,
+      "loss": 1.0432,
       "step": 56110
     },
     {
       "epoch": 168.53,
-      "grad_norm": 7.08125638961792,
+      "grad_norm": 6.222444534301758,
       "learning_rate": 4.382382382382383e-06,
-      "loss": 0.9533,
+      "loss": 0.9481,
       "step": 56120
     },
     {
       "epoch": 168.56,
-      "grad_norm": 6.6041951179504395,
+      "grad_norm": 7.985228061676025,
       "learning_rate": 4.381381381381382e-06,
-      "loss": 0.9653,
+      "loss": 0.9765,
       "step": 56130
     },
     {
       "epoch": 168.59,
-      "grad_norm": 7.4777936935424805,
+      "grad_norm": 9.2854642868042,
       "learning_rate": 4.3803803803803805e-06,
-      "loss": 0.9923,
+      "loss": 0.989,
       "step": 56140
     },
     {
       "epoch": 168.62,
-      "grad_norm": 9.051115989685059,
+      "grad_norm": 7.151541233062744,
       "learning_rate": 4.37937937937938e-06,
-      "loss": 0.9994,
+      "loss": 1.0053,
       "step": 56150
     },
     {
       "epoch": 168.65,
-      "grad_norm": 8.427289009094238,
+      "grad_norm": 10.851372718811035,
       "learning_rate": 4.378378378378379e-06,
-      "loss": 1.0174,
+      "loss": 1.013,
       "step": 56160
     },
     {
       "epoch": 168.68,
-      "grad_norm": 5.239019870758057,
+      "grad_norm": 6.23702335357666,
       "learning_rate": 4.377377377377378e-06,
-      "loss": 1.0561,
+      "loss": 1.0586,
       "step": 56170
     },
     {
       "epoch": 168.71,
-      "grad_norm": 4.579071521759033,
+      "grad_norm": 5.569283485412598,
       "learning_rate": 4.3763763763763765e-06,
-      "loss": 0.867,
+      "loss": 0.864,
       "step": 56180
     },
     {
       "epoch": 168.74,
-      "grad_norm": 5.279940128326416,
+      "grad_norm": 7.153161525726318,
       "learning_rate": 4.375375375375376e-06,
-      "loss": 0.9147,
+      "loss": 0.9184,
       "step": 56190
     },
     {
       "epoch": 168.77,
-      "grad_norm": 6.257159233093262,
+      "grad_norm": 6.213150501251221,
       "learning_rate": 4.374374374374375e-06,
-      "loss": 0.9636,
+      "loss": 0.9659,
       "step": 56200
     },
     {
       "epoch": 168.8,
-      "grad_norm": 7.533535957336426,
+      "grad_norm": 7.825570106506348,
       "learning_rate": 4.373373373373373e-06,
-      "loss": 0.9281,
+      "loss": 0.9302,
       "step": 56210
     },
     {
       "epoch": 168.83,
-      "grad_norm": 5.896141052246094,
+      "grad_norm": 6.033956527709961,
       "learning_rate": 4.372372372372373e-06,
-      "loss": 0.9835,
+      "loss": 0.9803,
       "step": 56220
     },
     {
       "epoch": 168.86,
-      "grad_norm": 5.943737030029297,
+      "grad_norm": 6.121496200561523,
       "learning_rate": 4.371371371371372e-06,
-      "loss": 0.9997,
+      "loss": 1.0091,
       "step": 56230
     },
     {
       "epoch": 168.89,
-      "grad_norm": 6.835830211639404,
+      "grad_norm": 6.015477657318115,
       "learning_rate": 4.370370370370371e-06,
-      "loss": 0.9753,
+      "loss": 0.9693,
       "step": 56240
     },
     {
       "epoch": 168.92,
-      "grad_norm": 7.149149417877197,
+      "grad_norm": 8.931275367736816,
       "learning_rate": 4.369369369369369e-06,
-      "loss": 0.9836,
+      "loss": 0.9769,
       "step": 56250
     },
     {
       "epoch": 168.95,
-      "grad_norm": 6.570301532745361,
+      "grad_norm": 7.112145900726318,
       "learning_rate": 4.368368368368369e-06,
-      "loss": 0.9495,
+      "loss": 0.9512,
       "step": 56260
     },
     {
       "epoch": 168.98,
-      "grad_norm": 5.159536838531494,
+      "grad_norm": 8.257311820983887,
       "learning_rate": 4.367367367367368e-06,
-      "loss": 1.0125,
+      "loss": 1.0039,
       "step": 56270
     },
     {
       "epoch": 169.0,
-      "eval_accuracy": 0.8185,
-      "eval_loss": 0.594907820224762,
-      "eval_runtime": 5.5016,
-      "eval_samples_per_second": 1817.661,
-      "eval_steps_per_second": 7.271,
+      "eval_accuracy": 0.8163,
+      "eval_loss": 0.5948166251182556,
+      "eval_runtime": 5.2336,
+      "eval_samples_per_second": 1910.745,
+      "eval_steps_per_second": 7.643,
       "step": 56277
     },
     {
       "epoch": 169.01,
-      "grad_norm": 5.843102931976318,
+      "grad_norm": 6.924843788146973,
       "learning_rate": 4.366366366366367e-06,
-      "loss": 1.1611,
+      "loss": 1.1643,
       "step": 56280
     },
     {
       "epoch": 169.04,
-      "grad_norm": 4.481675148010254,
+      "grad_norm": 5.031331539154053,
       "learning_rate": 4.365365365365365e-06,
-      "loss": 0.9325,
+      "loss": 0.9324,
       "step": 56290
     },
     {
       "epoch": 169.07,
-      "grad_norm": 5.820013046264648,
+      "grad_norm": 5.429023742675781,
       "learning_rate": 4.364364364364365e-06,
-      "loss": 0.9714,
+      "loss": 0.9788,
       "step": 56300
     },
     {
       "epoch": 169.1,
-      "grad_norm": 8.211709976196289,
+      "grad_norm": 8.171993255615234,
       "learning_rate": 4.363363363363363e-06,
-      "loss": 1.1202,
+      "loss": 1.1109,
       "step": 56310
     },
     {
       "epoch": 169.13,
-      "grad_norm": 7.753154277801514,
+      "grad_norm": 5.29931640625,
       "learning_rate": 4.362362362362363e-06,
-      "loss": 1.016,
+      "loss": 1.0049,
       "step": 56320
     },
     {
       "epoch": 169.16,
-      "grad_norm": 5.5673828125,
+      "grad_norm": 6.695034027099609,
       "learning_rate": 4.361361361361361e-06,
-      "loss": 0.9815,
+      "loss": 0.9833,
       "step": 56330
     },
     {
       "epoch": 169.19,
-      "grad_norm": 5.563979625701904,
+      "grad_norm": 5.325242519378662,
       "learning_rate": 4.360360360360361e-06,
-      "loss": 0.9521,
+      "loss": 0.9435,
       "step": 56340
     },
     {
       "epoch": 169.22,
-      "grad_norm": 6.524745941162109,
+      "grad_norm": 6.231870174407959,
       "learning_rate": 4.3593593593593594e-06,
-      "loss": 0.9193,
+      "loss": 0.9126,
       "step": 56350
     },
     {
       "epoch": 169.25,
-      "grad_norm": 4.805704593658447,
+      "grad_norm": 5.314600944519043,
       "learning_rate": 4.358358358358359e-06,
-      "loss": 0.9436,
+      "loss": 0.9378,
       "step": 56360
     },
     {
       "epoch": 169.28,
-      "grad_norm": 5.709771633148193,
+      "grad_norm": 4.556664943695068,
       "learning_rate": 4.357357357357358e-06,
-      "loss": 0.9768,
+      "loss": 0.9674,
       "step": 56370
     },
     {
       "epoch": 169.31,
-      "grad_norm": 5.843330383300781,
+      "grad_norm": 7.124413013458252,
       "learning_rate": 4.356356356356357e-06,
-      "loss": 1.036,
+      "loss": 1.0236,
       "step": 56380
     },
     {
       "epoch": 169.34,
-      "grad_norm": 7.125544548034668,
+      "grad_norm": 7.851509094238281,
       "learning_rate": 4.3553553553553555e-06,
-      "loss": 0.9937,
+      "loss": 0.9927,
       "step": 56390
     },
     {
       "epoch": 169.37,
-      "grad_norm": 5.590447425842285,
+      "grad_norm": 7.322617053985596,
       "learning_rate": 4.354354354354355e-06,
-      "loss": 1.0082,
+      "loss": 1.0001,
       "step": 56400
     },
     {
       "epoch": 169.4,
-      "grad_norm": 6.876036643981934,
+      "grad_norm": 9.111237525939941,
       "learning_rate": 4.3533533533533535e-06,
-      "loss": 1.0034,
+      "loss": 1.0029,
       "step": 56410
     },
     {
       "epoch": 169.43,
-      "grad_norm": 6.731537818908691,
+      "grad_norm": 7.247379779815674,
       "learning_rate": 4.352352352352352e-06,
-      "loss": 0.9916,
+      "loss": 0.9954,
       "step": 56420
     },
     {
       "epoch": 169.46,
-      "grad_norm": 8.717347145080566,
+      "grad_norm": 6.96370267868042,
       "learning_rate": 4.351351351351352e-06,
-      "loss": 1.0203,
+      "loss": 1.0145,
       "step": 56430
     },
     {
       "epoch": 169.49,
-      "grad_norm": 10.916757583618164,
+      "grad_norm": 10.059701919555664,
       "learning_rate": 4.350350350350351e-06,
-      "loss": 0.999,
+      "loss": 0.9855,
       "step": 56440
     },
     {
       "epoch": 169.52,
-      "grad_norm": 6.431384086608887,
+      "grad_norm": 7.551264762878418,
       "learning_rate": 4.34934934934935e-06,
-      "loss": 0.9945,
+      "loss": 1.0073,
       "step": 56450
     },
     {
       "epoch": 169.55,
-      "grad_norm": 6.612633228302002,
+      "grad_norm": 7.821953773498535,
       "learning_rate": 4.348348348348348e-06,
-      "loss": 1.0262,
+      "loss": 1.028,
       "step": 56460
     },
     {
       "epoch": 169.58,
-      "grad_norm": 8.268803596496582,
+      "grad_norm": 9.027318000793457,
       "learning_rate": 4.347347347347348e-06,
-      "loss": 0.9945,
+      "loss": 0.9926,
       "step": 56470
     },
     {
       "epoch": 169.61,
-      "grad_norm": 7.941657543182373,
+      "grad_norm": 8.304580688476562,
       "learning_rate": 4.346346346346347e-06,
-      "loss": 1.094,
+      "loss": 1.098,
       "step": 56480
     },
     {
       "epoch": 169.64,
-      "grad_norm": 5.694316864013672,
+      "grad_norm": 5.552855491638184,
       "learning_rate": 4.345345345345346e-06,
-      "loss": 0.9219,
+      "loss": 0.9155,
       "step": 56490
     },
     {
       "epoch": 169.67,
-      "grad_norm": 6.701169013977051,
+      "grad_norm": 6.712451934814453,
       "learning_rate": 4.344344344344344e-06,
-      "loss": 1.0366,
+      "loss": 1.0325,
       "step": 56500
     },
     {
       "epoch": 169.7,
-      "grad_norm": 8.175214767456055,
+      "grad_norm": 7.1357903480529785,
       "learning_rate": 4.343343343343344e-06,
-      "loss": 0.9678,
+      "loss": 0.9668,
       "step": 56510
     },
     {
       "epoch": 169.73,
-      "grad_norm": 6.110684394836426,
+      "grad_norm": 6.286517143249512,
       "learning_rate": 4.342342342342343e-06,
-      "loss": 0.9919,
+      "loss": 0.9791,
       "step": 56520
     },
     {
       "epoch": 169.76,
-      "grad_norm": 9.001298904418945,
+      "grad_norm": 6.8825883865356445,
       "learning_rate": 4.341341341341342e-06,
-      "loss": 1.0362,
+      "loss": 1.0323,
       "step": 56530
     },
     {
       "epoch": 169.79,
-      "grad_norm": 6.749107360839844,
+      "grad_norm": 5.727536201477051,
       "learning_rate": 4.34034034034034e-06,
-      "loss": 0.9958,
+      "loss": 0.9847,
       "step": 56540
     },
     {
       "epoch": 169.82,
-      "grad_norm": 6.816410064697266,
+      "grad_norm": 6.622138977050781,
       "learning_rate": 4.33933933933934e-06,
-      "loss": 1.003,
+      "loss": 0.9915,
       "step": 56550
     },
     {
       "epoch": 169.85,
-      "grad_norm": 8.41502857208252,
+      "grad_norm": 5.993500709533691,
       "learning_rate": 4.338338338338338e-06,
-      "loss": 0.9871,
+      "loss": 0.9886,
       "step": 56560
     },
     {
       "epoch": 169.88,
-      "grad_norm": 6.962545394897461,
+      "grad_norm": 6.555738925933838,
       "learning_rate": 4.337337337337338e-06,
-      "loss": 0.9969,
+      "loss": 1.0043,
       "step": 56570
     },
     {
       "epoch": 169.91,
-      "grad_norm": 4.562180519104004,
+      "grad_norm": 4.420021057128906,
       "learning_rate": 4.3363363363363364e-06,
-      "loss": 0.8804,
+      "loss": 0.8777,
       "step": 56580
     },
     {
       "epoch": 169.94,
-      "grad_norm": 6.5855712890625,
+      "grad_norm": 7.449794292449951,
       "learning_rate": 4.335335335335336e-06,
-      "loss": 0.9752,
+      "loss": 0.976,
       "step": 56590
     },
     {
       "epoch": 169.97,
-      "grad_norm": 7.707968235015869,
+      "grad_norm": 7.161410808563232,
       "learning_rate": 4.3343343343343345e-06,
-      "loss": 1.0022,
+      "loss": 1.004,
       "step": 56600
     },
     {
       "epoch": 170.0,
-      "grad_norm": 30.620025634765625,
+      "grad_norm": 28.9823055267334,
       "learning_rate": 4.333333333333334e-06,
-      "loss": 1.0714,
+      "loss": 1.0677,
       "step": 56610
     },
     {
       "epoch": 170.0,
-      "eval_accuracy": 0.8193,
-      "eval_loss": 0.590685248374939,
-      "eval_runtime": 5.4824,
-      "eval_samples_per_second": 1824.008,
-      "eval_steps_per_second": 7.296,
+      "eval_accuracy": 0.8202,
+      "eval_loss": 0.5892015099525452,
+      "eval_runtime": 5.3297,
+      "eval_samples_per_second": 1876.295,
+      "eval_steps_per_second": 7.505,
       "step": 56610
     },
     {
       "epoch": 170.03,
-      "grad_norm": 5.490104675292969,
+      "grad_norm": 5.802674293518066,
       "learning_rate": 4.3323323323323325e-06,
-      "loss": 0.9089,
+      "loss": 0.91,
       "step": 56620
     },
     {
       "epoch": 170.06,
-      "grad_norm": 5.317603588104248,
+      "grad_norm": 6.986905574798584,
       "learning_rate": 4.331331331331332e-06,
-      "loss": 0.9725,
+      "loss": 0.9733,
       "step": 56630
     },
     {
       "epoch": 170.09,
-      "grad_norm": 7.429643154144287,
+      "grad_norm": 5.773680210113525,
       "learning_rate": 4.3303303303303305e-06,
-      "loss": 0.9803,
+      "loss": 0.9754,
       "step": 56640
     },
     {
       "epoch": 170.12,
-      "grad_norm": 9.472784996032715,
+      "grad_norm": 8.544479370117188,
       "learning_rate": 4.32932932932933e-06,
-      "loss": 0.9952,
+      "loss": 1.0008,
       "step": 56650
     },
     {
       "epoch": 170.15,
-      "grad_norm": 9.182112693786621,
+      "grad_norm": 9.040329933166504,
       "learning_rate": 4.328328328328329e-06,
-      "loss": 0.9107,
+      "loss": 0.9256,
       "step": 56660
     },
     {
       "epoch": 170.18,
-      "grad_norm": 6.440854549407959,
+      "grad_norm": 7.802104949951172,
       "learning_rate": 4.327327327327327e-06,
-      "loss": 0.9931,
+      "loss": 0.9935,
       "step": 56670
     },
     {
       "epoch": 170.21,
-      "grad_norm": 4.998106479644775,
+      "grad_norm": 4.919273376464844,
       "learning_rate": 4.326326326326327e-06,
-      "loss": 0.9376,
+      "loss": 0.9392,
       "step": 56680
     },
     {
       "epoch": 170.24,
-      "grad_norm": 7.0564680099487305,
+      "grad_norm": 9.068553924560547,
       "learning_rate": 4.325325325325326e-06,
-      "loss": 0.9587,
+      "loss": 0.9623,
       "step": 56690
     },
     {
       "epoch": 170.27,
-      "grad_norm": 6.391025543212891,
+      "grad_norm": 6.758793830871582,
       "learning_rate": 4.324324324324325e-06,
-      "loss": 0.9971,
+      "loss": 0.9904,
       "step": 56700
     },
     {
       "epoch": 170.3,
-      "grad_norm": 6.855101585388184,
+      "grad_norm": 8.718999862670898,
       "learning_rate": 4.323323323323323e-06,
-      "loss": 0.9523,
+      "loss": 0.9598,
       "step": 56710
     },
     {
       "epoch": 170.33,
-      "grad_norm": 6.619818687438965,
+      "grad_norm": 7.70663595199585,
       "learning_rate": 4.322322322322323e-06,
-      "loss": 0.9496,
+      "loss": 0.9455,
       "step": 56720
     },
     {
       "epoch": 170.36,
-      "grad_norm": 8.596623420715332,
+      "grad_norm": 6.499110698699951,
       "learning_rate": 4.321321321321322e-06,
-      "loss": 1.0262,
+      "loss": 1.0292,
       "step": 56730
     },
     {
       "epoch": 170.39,
-      "grad_norm": 4.825212478637695,
+      "grad_norm": 5.019879341125488,
       "learning_rate": 4.320320320320321e-06,
-      "loss": 1.0017,
+      "loss": 0.9945,
       "step": 56740
     },
     {
       "epoch": 170.42,
-      "grad_norm": 7.764327049255371,
+      "grad_norm": 8.448901176452637,
       "learning_rate": 4.319319319319319e-06,
-      "loss": 1.0597,
+      "loss": 1.0565,
       "step": 56750
     },
     {
       "epoch": 170.45,
-      "grad_norm": 5.882233142852783,
+      "grad_norm": 7.327672958374023,
       "learning_rate": 4.318318318318319e-06,
-      "loss": 0.9973,
+      "loss": 0.9979,
       "step": 56760
     },
     {
       "epoch": 170.48,
-      "grad_norm": 5.750434398651123,
+      "grad_norm": 6.711387634277344,
       "learning_rate": 4.317317317317317e-06,
-      "loss": 0.9648,
+      "loss": 0.9639,
       "step": 56770
     },
     {
       "epoch": 170.51,
-      "grad_norm": 6.567395210266113,
+      "grad_norm": 6.684991836547852,
       "learning_rate": 4.316316316316317e-06,
-      "loss": 0.9902,
+      "loss": 0.9914,
       "step": 56780
     },
     {
       "epoch": 170.54,
-      "grad_norm": 7.586670875549316,
+      "grad_norm": 11.05115795135498,
       "learning_rate": 4.315315315315315e-06,
-      "loss": 0.9227,
+      "loss": 0.9277,
       "step": 56790
     },
     {
       "epoch": 170.57,
-      "grad_norm": 6.738625526428223,
+      "grad_norm": 6.198968410491943,
       "learning_rate": 4.314314314314315e-06,
-      "loss": 1.0291,
+      "loss": 1.0278,
       "step": 56800
     },
     {
       "epoch": 170.6,
-      "grad_norm": 6.259493827819824,
+      "grad_norm": 8.037765502929688,
       "learning_rate": 4.3133133133133134e-06,
-      "loss": 0.9386,
+      "loss": 0.9344,
       "step": 56810
     },
     {
       "epoch": 170.63,
-      "grad_norm": 8.243454933166504,
+      "grad_norm": 11.837545394897461,
       "learning_rate": 4.312312312312313e-06,
-      "loss": 0.9991,
+      "loss": 0.999,
       "step": 56820
     },
     {
       "epoch": 170.66,
-      "grad_norm": 7.437496662139893,
+      "grad_norm": 6.0878071784973145,
       "learning_rate": 4.3113113113113115e-06,
-      "loss": 1.0382,
+      "loss": 1.0358,
       "step": 56830
     },
     {
       "epoch": 170.69,
-      "grad_norm": 6.3546624183654785,
+      "grad_norm": 9.397327423095703,
       "learning_rate": 4.310310310310311e-06,
-      "loss": 1.0135,
+      "loss": 1.013,
       "step": 56840
     },
     {
       "epoch": 170.72,
-      "grad_norm": 5.314104080200195,
+      "grad_norm": 6.422165870666504,
       "learning_rate": 4.3093093093093095e-06,
-      "loss": 0.9618,
+      "loss": 0.9687,
       "step": 56850
     },
     {
       "epoch": 170.75,
-      "grad_norm": 5.663581848144531,
+      "grad_norm": 7.128824710845947,
       "learning_rate": 4.308308308308309e-06,
-      "loss": 0.9407,
+      "loss": 0.9418,
       "step": 56860
     },
     {
       "epoch": 170.78,
-      "grad_norm": 5.130428791046143,
+      "grad_norm": 10.488344192504883,
       "learning_rate": 4.3073073073073075e-06,
-      "loss": 0.9835,
+      "loss": 0.9751,
       "step": 56870
     },
     {
       "epoch": 170.81,
-      "grad_norm": 8.1718168258667,
+      "grad_norm": 7.195184230804443,
       "learning_rate": 4.306306306306306e-06,
-      "loss": 0.9717,
+      "loss": 0.9839,
       "step": 56880
     },
     {
       "epoch": 170.84,
-      "grad_norm": 7.706439018249512,
+      "grad_norm": 8.487993240356445,
       "learning_rate": 4.305305305305306e-06,
-      "loss": 0.9728,
+      "loss": 0.9754,
       "step": 56890
     },
     {
       "epoch": 170.87,
-      "grad_norm": 6.178399562835693,
+      "grad_norm": 5.656654357910156,
       "learning_rate": 4.304304304304305e-06,
-      "loss": 1.0191,
+      "loss": 1.0077,
       "step": 56900
     },
     {
       "epoch": 170.9,
-      "grad_norm": 8.404486656188965,
+      "grad_norm": 8.010581016540527,
       "learning_rate": 4.303303303303304e-06,
-      "loss": 1.012,
+      "loss": 1.0239,
       "step": 56910
     },
     {
       "epoch": 170.93,
-      "grad_norm": 6.999275207519531,
+      "grad_norm": 6.412772178649902,
       "learning_rate": 4.302302302302302e-06,
-      "loss": 1.007,
+      "loss": 1.0155,
       "step": 56920
     },
     {
       "epoch": 170.96,
-      "grad_norm": 6.07869815826416,
+      "grad_norm": 8.956306457519531,
       "learning_rate": 4.301301301301302e-06,
-      "loss": 0.9935,
+      "loss": 0.9981,
       "step": 56930
     },
     {
       "epoch": 170.99,
-      "grad_norm": 6.49916934967041,
+      "grad_norm": 6.116293430328369,
       "learning_rate": 4.300300300300301e-06,
-      "loss": 1.011,
+      "loss": 1.0172,
       "step": 56940
     },
     {
       "epoch": 171.0,
-      "eval_accuracy": 0.8201,
-      "eval_loss": 0.5951588153839111,
-      "eval_runtime": 5.2887,
-      "eval_samples_per_second": 1890.817,
-      "eval_steps_per_second": 7.563,
+      "eval_accuracy": 0.8192,
+      "eval_loss": 0.5944841504096985,
+      "eval_runtime": 5.1848,
+      "eval_samples_per_second": 1928.723,
+      "eval_steps_per_second": 7.715,
       "step": 56943
     },
     {
       "epoch": 171.02,
-      "grad_norm": 5.040866374969482,
+      "grad_norm": 4.950351238250732,
       "learning_rate": 4.2992992992993e-06,
-      "loss": 1.0782,
+      "loss": 1.0651,
       "step": 56950
     },
     {
       "epoch": 171.05,
-      "grad_norm": 5.824924945831299,
+      "grad_norm": 7.10728645324707,
       "learning_rate": 4.298298298298298e-06,
-      "loss": 1.0116,
+      "loss": 1.0111,
       "step": 56960
     },
     {
       "epoch": 171.08,
-      "grad_norm": 8.43815803527832,
+      "grad_norm": 10.009684562683105,
       "learning_rate": 4.297297297297298e-06,
-      "loss": 0.9572,
+      "loss": 0.9617,
       "step": 56970
     },
     {
       "epoch": 171.11,
-      "grad_norm": 7.278158187866211,
+      "grad_norm": 7.082460403442383,
       "learning_rate": 4.296296296296296e-06,
-      "loss": 0.9323,
+      "loss": 0.9394,
       "step": 56980
     },
     {
       "epoch": 171.14,
-      "grad_norm": 7.7384934425354,
+      "grad_norm": 5.939715385437012,
       "learning_rate": 4.295295295295296e-06,
-      "loss": 0.9632,
+      "loss": 0.9638,
       "step": 56990
     },
     {
       "epoch": 171.17,
-      "grad_norm": 6.198485851287842,
+      "grad_norm": 7.281838893890381,
       "learning_rate": 4.294294294294294e-06,
-      "loss": 1.0298,
+      "loss": 1.0301,
       "step": 57000
     },
     {
       "epoch": 171.2,
-      "grad_norm": 8.445798873901367,
+      "grad_norm": 7.408473014831543,
       "learning_rate": 4.293293293293294e-06,
-      "loss": 0.9581,
+      "loss": 0.9573,
       "step": 57010
     },
     {
       "epoch": 171.23,
-      "grad_norm": 5.926905155181885,
+      "grad_norm": 6.613779067993164,
       "learning_rate": 4.292292292292292e-06,
-      "loss": 0.9952,
+      "loss": 1.0018,
       "step": 57020
     },
     {
       "epoch": 171.26,
-      "grad_norm": 8.326822280883789,
+      "grad_norm": 7.298211574554443,
       "learning_rate": 4.291291291291292e-06,
-      "loss": 1.0366,
+      "loss": 1.0208,
       "step": 57030
     },
     {
       "epoch": 171.29,
-      "grad_norm": 6.417525768280029,
+      "grad_norm": 6.488585948944092,
       "learning_rate": 4.2902902902902904e-06,
-      "loss": 0.981,
+      "loss": 0.9686,
       "step": 57040
     },
     {
       "epoch": 171.32,
-      "grad_norm": 7.017544269561768,
+      "grad_norm": 5.689070701599121,
       "learning_rate": 4.28928928928929e-06,
-      "loss": 0.9619,
+      "loss": 0.9698,
       "step": 57050
     },
     {
       "epoch": 171.35,
-      "grad_norm": 8.010387420654297,
+      "grad_norm": 7.418861389160156,
       "learning_rate": 4.2882882882882885e-06,
-      "loss": 1.0588,
+      "loss": 1.0534,
       "step": 57060
     },
     {
       "epoch": 171.38,
-      "grad_norm": 6.428537845611572,
+      "grad_norm": 7.157418251037598,
       "learning_rate": 4.287287287287288e-06,
-      "loss": 1.0201,
+      "loss": 1.0189,
       "step": 57070
     },
     {
       "epoch": 171.41,
-      "grad_norm": 5.345740795135498,
+      "grad_norm": 7.65828800201416,
       "learning_rate": 4.2862862862862865e-06,
-      "loss": 0.9432,
+      "loss": 0.9289,
       "step": 57080
     },
     {
       "epoch": 171.44,
-      "grad_norm": 6.588776588439941,
+      "grad_norm": 7.339501857757568,
       "learning_rate": 4.285285285285286e-06,
-      "loss": 0.9949,
+      "loss": 0.9947,
       "step": 57090
     },
     {
       "epoch": 171.47,
-      "grad_norm": 10.427704811096191,
+      "grad_norm": 8.524580955505371,
       "learning_rate": 4.2842842842842845e-06,
-      "loss": 0.8601,
+      "loss": 0.868,
       "step": 57100
     },
     {
       "epoch": 171.5,
-      "grad_norm": 8.383879661560059,
+      "grad_norm": 7.013965129852295,
       "learning_rate": 4.283283283283284e-06,
-      "loss": 1.0281,
+      "loss": 1.017,
       "step": 57110
     },
     {
       "epoch": 171.53,
-      "grad_norm": 6.146906852722168,
+      "grad_norm": 8.512283325195312,
       "learning_rate": 4.282282282282283e-06,
-      "loss": 0.9794,
+      "loss": 0.9883,
       "step": 57120
     },
     {
       "epoch": 171.56,
-      "grad_norm": 5.8249192237854,
+      "grad_norm": 7.077947616577148,
       "learning_rate": 4.281281281281281e-06,
-      "loss": 1.0448,
+      "loss": 1.0457,
       "step": 57130
     },
     {
       "epoch": 171.59,
-      "grad_norm": 8.164721488952637,
+      "grad_norm": 8.033219337463379,
       "learning_rate": 4.280280280280281e-06,
-      "loss": 1.012,
+      "loss": 0.9986,
       "step": 57140
     },
     {
       "epoch": 171.62,
-      "grad_norm": 8.785282135009766,
+      "grad_norm": 9.006940841674805,
       "learning_rate": 4.27927927927928e-06,
-      "loss": 0.9385,
+      "loss": 0.928,
       "step": 57150
     },
     {
       "epoch": 171.65,
-      "grad_norm": 6.605419158935547,
+      "grad_norm": 9.160282135009766,
       "learning_rate": 4.278278278278279e-06,
-      "loss": 0.9562,
+      "loss": 0.9738,
       "step": 57160
     },
     {
       "epoch": 171.68,
-      "grad_norm": 7.04688835144043,
+      "grad_norm": 6.151976585388184,
       "learning_rate": 4.277277277277277e-06,
-      "loss": 1.0467,
+      "loss": 1.0431,
       "step": 57170
     },
     {
       "epoch": 171.71,
-      "grad_norm": 7.077495574951172,
+      "grad_norm": 8.831791877746582,
       "learning_rate": 4.276276276276277e-06,
-      "loss": 1.0516,
+      "loss": 1.0456,
       "step": 57180
     },
     {
       "epoch": 171.74,
-      "grad_norm": 6.012141227722168,
+      "grad_norm": 8.4002685546875,
       "learning_rate": 4.275275275275276e-06,
-      "loss": 0.9306,
+      "loss": 0.922,
       "step": 57190
     },
     {
       "epoch": 171.77,
-      "grad_norm": 8.064560890197754,
+      "grad_norm": 8.04333782196045,
       "learning_rate": 4.274274274274275e-06,
-      "loss": 0.9586,
+      "loss": 0.9499,
       "step": 57200
     },
     {
       "epoch": 171.8,
-      "grad_norm": 5.532112121582031,
+      "grad_norm": 6.527125358581543,
       "learning_rate": 4.273273273273273e-06,
-      "loss": 0.9471,
+      "loss": 0.9521,
       "step": 57210
     },
     {
       "epoch": 171.83,
-      "grad_norm": 6.053155899047852,
+      "grad_norm": 8.13235855102539,
       "learning_rate": 4.272272272272273e-06,
-      "loss": 1.0248,
+      "loss": 1.0272,
       "step": 57220
     },
     {
       "epoch": 171.86,
-      "grad_norm": 5.451865196228027,
+      "grad_norm": 6.451412677764893,
       "learning_rate": 4.271271271271271e-06,
-      "loss": 0.9701,
+      "loss": 0.9644,
       "step": 57230
     },
     {
       "epoch": 171.89,
-      "grad_norm": 7.161723613739014,
+      "grad_norm": 6.254931449890137,
       "learning_rate": 4.270270270270271e-06,
-      "loss": 1.0376,
+      "loss": 1.0272,
       "step": 57240
     },
     {
       "epoch": 171.92,
-      "grad_norm": 6.154750823974609,
+      "grad_norm": 6.001988410949707,
       "learning_rate": 4.269269269269269e-06,
-      "loss": 0.9803,
+      "loss": 0.9899,
       "step": 57250
     },
     {
       "epoch": 171.95,
-      "grad_norm": 8.600580215454102,
+      "grad_norm": 10.811860084533691,
       "learning_rate": 4.268268268268269e-06,
-      "loss": 1.0099,
+      "loss": 1.003,
       "step": 57260
     },
     {
       "epoch": 171.98,
-      "grad_norm": 5.275885105133057,
+      "grad_norm": 5.569774150848389,
       "learning_rate": 4.2672672672672674e-06,
-      "loss": 0.9099,
+      "loss": 0.9155,
       "step": 57270
     },
     {
       "epoch": 172.0,
-      "eval_accuracy": 0.8169,
-      "eval_loss": 0.5905144810676575,
-      "eval_runtime": 5.5891,
-      "eval_samples_per_second": 1789.189,
-      "eval_steps_per_second": 7.157,
+      "eval_accuracy": 0.82,
+      "eval_loss": 0.5891578793525696,
+      "eval_runtime": 5.2767,
+      "eval_samples_per_second": 1895.113,
+      "eval_steps_per_second": 7.58,
       "step": 57276
     },
     {
       "epoch": 172.01,
-      "grad_norm": 7.995873928070068,
+      "grad_norm": 7.500364303588867,
       "learning_rate": 4.266266266266267e-06,
-      "loss": 1.2713,
+      "loss": 1.2704,
       "step": 57280
     },
     {
       "epoch": 172.04,
-      "grad_norm": 5.735530853271484,
+      "grad_norm": 5.647741317749023,
       "learning_rate": 4.2652652652652655e-06,
-      "loss": 0.953,
+      "loss": 0.9451,
       "step": 57290
     },
     {
       "epoch": 172.07,
-      "grad_norm": 6.427038192749023,
+      "grad_norm": 10.266873359680176,
       "learning_rate": 4.264264264264265e-06,
-      "loss": 0.964,
+      "loss": 0.962,
       "step": 57300
     },
     {
       "epoch": 172.1,
-      "grad_norm": 4.708528518676758,
+      "grad_norm": 5.55878210067749,
       "learning_rate": 4.2632632632632635e-06,
-      "loss": 0.9857,
+      "loss": 0.9811,
       "step": 57310
     },
     {
       "epoch": 172.13,
-      "grad_norm": 5.05018949508667,
+      "grad_norm": 5.194493770599365,
       "learning_rate": 4.262262262262263e-06,
-      "loss": 0.9993,
+      "loss": 0.9975,
       "step": 57320
     },
     {
       "epoch": 172.16,
-      "grad_norm": 4.961715221405029,
+      "grad_norm": 5.520266056060791,
       "learning_rate": 4.2612612612612615e-06,
-      "loss": 0.9627,
+      "loss": 0.9762,
       "step": 57330
     },
     {
       "epoch": 172.19,
-      "grad_norm": 6.519645690917969,
+      "grad_norm": 8.44970703125,
       "learning_rate": 4.26026026026026e-06,
-      "loss": 0.9719,
+      "loss": 0.9644,
       "step": 57340
     },
     {
       "epoch": 172.22,
-      "grad_norm": 4.330857753753662,
+      "grad_norm": 4.201999664306641,
       "learning_rate": 4.2592592592592596e-06,
-      "loss": 0.9204,
+      "loss": 0.9174,
       "step": 57350
     },
     {
       "epoch": 172.25,
-      "grad_norm": 10.768898010253906,
+      "grad_norm": 7.7078166007995605,
       "learning_rate": 4.258258258258259e-06,
-      "loss": 0.9336,
+      "loss": 0.9348,
       "step": 57360
     },
     {
       "epoch": 172.28,
-      "grad_norm": 6.093428611755371,
+      "grad_norm": 7.763128757476807,
       "learning_rate": 4.257257257257258e-06,
-      "loss": 0.9007,
+      "loss": 0.9068,
       "step": 57370
     },
     {
       "epoch": 172.31,
-      "grad_norm": 8.052370071411133,
+      "grad_norm": 6.794031620025635,
       "learning_rate": 4.256256256256256e-06,
-      "loss": 0.9751,
+      "loss": 0.9612,
       "step": 57380
     },
     {
       "epoch": 172.34,
-      "grad_norm": 7.486790180206299,
+      "grad_norm": 6.245731830596924,
       "learning_rate": 4.255255255255256e-06,
-      "loss": 0.8968,
+      "loss": 0.8944,
       "step": 57390
     },
     {
       "epoch": 172.37,
-      "grad_norm": 6.691470623016357,
+      "grad_norm": 7.237773418426514,
       "learning_rate": 4.254254254254255e-06,
-      "loss": 0.9341,
+      "loss": 0.9274,
       "step": 57400
     },
     {
       "epoch": 172.4,
-      "grad_norm": 5.719216823577881,
+      "grad_norm": 6.510280609130859,
       "learning_rate": 4.253253253253254e-06,
-      "loss": 1.0816,
+      "loss": 1.0719,
       "step": 57410
     },
     {
       "epoch": 172.43,
-      "grad_norm": 8.30722713470459,
+      "grad_norm": 7.967406272888184,
       "learning_rate": 4.252252252252252e-06,
-      "loss": 0.9786,
+      "loss": 0.9716,
       "step": 57420
     },
     {
       "epoch": 172.46,
-      "grad_norm": 6.678756237030029,
+      "grad_norm": 6.857039451599121,
       "learning_rate": 4.251251251251252e-06,
-      "loss": 0.9311,
+      "loss": 0.9329,
       "step": 57430
     },
     {
       "epoch": 172.49,
-      "grad_norm": 5.430222034454346,
+      "grad_norm": 5.27644681930542,
       "learning_rate": 4.25025025025025e-06,
-      "loss": 0.9203,
+      "loss": 0.9275,
       "step": 57440
     },
     {
       "epoch": 172.52,
-      "grad_norm": 6.501481533050537,
+      "grad_norm": 5.869178295135498,
       "learning_rate": 4.24924924924925e-06,
-      "loss": 0.9377,
+      "loss": 0.9259,
       "step": 57450
     },
     {
       "epoch": 172.55,
-      "grad_norm": 6.936547756195068,
+      "grad_norm": 10.140311241149902,
       "learning_rate": 4.248248248248248e-06,
-      "loss": 0.9974,
+      "loss": 1.0037,
       "step": 57460
     },
     {
       "epoch": 172.58,
-      "grad_norm": 6.074435234069824,
+      "grad_norm": 6.381155014038086,
       "learning_rate": 4.247247247247248e-06,
-      "loss": 0.9692,
+      "loss": 0.9609,
       "step": 57470
     },
     {
       "epoch": 172.61,
-      "grad_norm": 7.484917640686035,
+      "grad_norm": 6.242295265197754,
       "learning_rate": 4.246246246246246e-06,
-      "loss": 0.968,
+      "loss": 0.9598,
       "step": 57480
     },
     {
       "epoch": 172.64,
-      "grad_norm": 6.443945407867432,
+      "grad_norm": 6.6319427490234375,
       "learning_rate": 4.245245245245246e-06,
-      "loss": 1.0491,
+      "loss": 1.0474,
       "step": 57490
     },
     {
       "epoch": 172.67,
-      "grad_norm": 8.712279319763184,
+      "grad_norm": 6.078037261962891,
       "learning_rate": 4.2442442442442444e-06,
-      "loss": 1.0014,
+      "loss": 0.9933,
       "step": 57500
     },
     {
       "epoch": 172.7,
-      "grad_norm": 5.348664283752441,
+      "grad_norm": 4.209136009216309,
       "learning_rate": 4.243243243243244e-06,
       "loss": 1.0142,
       "step": 57510
     },
     {
       "epoch": 172.73,
-      "grad_norm": 5.946605682373047,
+      "grad_norm": 9.43163776397705,
       "learning_rate": 4.2422422422422425e-06,
-      "loss": 0.9121,
+      "loss": 0.9226,
       "step": 57520
     },
     {
       "epoch": 172.76,
-      "grad_norm": 6.6635236740112305,
+      "grad_norm": 6.871527671813965,
       "learning_rate": 4.241241241241242e-06,
-      "loss": 1.0351,
+      "loss": 1.0234,
       "step": 57530
     },
     {
       "epoch": 172.79,
-      "grad_norm": 5.259237766265869,
+      "grad_norm": 5.477137565612793,
       "learning_rate": 4.2402402402402405e-06,
-      "loss": 1.0304,
+      "loss": 1.032,
       "step": 57540
     },
     {
       "epoch": 172.82,
-      "grad_norm": 7.379566669464111,
+      "grad_norm": 7.341521263122559,
       "learning_rate": 4.23923923923924e-06,
-      "loss": 0.9966,
+      "loss": 0.9913,
       "step": 57550
     },
     {
       "epoch": 172.85,
-      "grad_norm": 6.366992950439453,
+      "grad_norm": 6.493458271026611,
       "learning_rate": 4.2382382382382385e-06,
-      "loss": 0.9299,
+      "loss": 0.9223,
       "step": 57560
     },
     {
       "epoch": 172.88,
-      "grad_norm": 7.3531389236450195,
+      "grad_norm": 7.29157018661499,
       "learning_rate": 4.237237237237238e-06,
-      "loss": 0.9633,
+      "loss": 0.9595,
       "step": 57570
     },
     {
       "epoch": 172.91,
-      "grad_norm": 4.167303562164307,
+      "grad_norm": 4.860920429229736,
       "learning_rate": 4.2362362362362366e-06,
-      "loss": 0.9239,
+      "loss": 0.9235,
       "step": 57580
     },
     {
       "epoch": 172.94,
-      "grad_norm": 5.63935661315918,
+      "grad_norm": 5.4882049560546875,
       "learning_rate": 4.235235235235235e-06,
-      "loss": 0.9953,
+      "loss": 1.0028,
       "step": 57590
     },
     {
       "epoch": 172.97,
-      "grad_norm": 7.719032287597656,
+      "grad_norm": 7.058857440948486,
       "learning_rate": 4.234234234234235e-06,
-      "loss": 0.9879,
+      "loss": 0.9894,
       "step": 57600
     },
     {
       "epoch": 173.0,
-      "eval_accuracy": 0.8201,
-      "eval_loss": 0.595493733882904,
-      "eval_runtime": 5.3325,
-      "eval_samples_per_second": 1875.307,
-      "eval_steps_per_second": 7.501,
+      "eval_accuracy": 0.8193,
+      "eval_loss": 0.5953618884086609,
+      "eval_runtime": 5.5654,
+      "eval_samples_per_second": 1796.802,
+      "eval_steps_per_second": 7.187,
       "step": 57609
     },
     {
       "epoch": 173.0,
-      "grad_norm": 6.578979015350342,
+      "grad_norm": 6.634015083312988,
       "learning_rate": 4.233233233233234e-06,
-      "loss": 1.1115,
+      "loss": 1.0981,
       "step": 57610
     },
     {
       "epoch": 173.03,
-      "grad_norm": 5.680975914001465,
+      "grad_norm": 6.90461540222168,
       "learning_rate": 4.232232232232233e-06,
-      "loss": 0.9667,
+      "loss": 0.9738,
       "step": 57620
     },
     {
       "epoch": 173.06,
-      "grad_norm": 6.637292385101318,
+      "grad_norm": 8.841856002807617,
       "learning_rate": 4.231231231231231e-06,
-      "loss": 0.987,
+      "loss": 0.9819,
       "step": 57630
     },
     {
       "epoch": 173.09,
-      "grad_norm": 5.649447917938232,
+      "grad_norm": 8.933527946472168,
       "learning_rate": 4.230230230230231e-06,
-      "loss": 0.9506,
+      "loss": 0.9551,
       "step": 57640
     },
     {
       "epoch": 173.12,
-      "grad_norm": 8.128000259399414,
+      "grad_norm": 8.824667930603027,
       "learning_rate": 4.22922922922923e-06,
-      "loss": 0.9776,
+      "loss": 0.9798,
       "step": 57650
     },
     {
       "epoch": 173.15,
-      "grad_norm": 6.3696699142456055,
+      "grad_norm": 6.375158786773682,
       "learning_rate": 4.228228228228229e-06,
-      "loss": 0.9476,
+      "loss": 0.9408,
       "step": 57660
     },
     {
       "epoch": 173.18,
-      "grad_norm": 7.179959774017334,
+      "grad_norm": 6.284273147583008,
       "learning_rate": 4.227227227227227e-06,
-      "loss": 0.9272,
+      "loss": 0.926,
       "step": 57670
     },
     {
       "epoch": 173.21,
-      "grad_norm": 5.680412292480469,
+      "grad_norm": 6.191782474517822,
       "learning_rate": 4.226226226226227e-06,
-      "loss": 1.0415,
+      "loss": 1.0425,
       "step": 57680
     },
     {
       "epoch": 173.24,
-      "grad_norm": 6.367232799530029,
+      "grad_norm": 6.010632514953613,
       "learning_rate": 4.225225225225225e-06,
-      "loss": 1.0015,
+      "loss": 1.004,
       "step": 57690
     },
     {
       "epoch": 173.27,
-      "grad_norm": 7.266384601593018,
+      "grad_norm": 6.733784198760986,
       "learning_rate": 4.224224224224225e-06,
-      "loss": 0.9537,
+      "loss": 0.9542,
       "step": 57700
     },
     {
       "epoch": 173.3,
-      "grad_norm": 5.722720146179199,
+      "grad_norm": 7.1573805809021,
       "learning_rate": 4.223223223223223e-06,
-      "loss": 1.0326,
+      "loss": 1.0309,
       "step": 57710
     },
     {
       "epoch": 173.33,
-      "grad_norm": 6.570093631744385,
+      "grad_norm": 5.748527526855469,
       "learning_rate": 4.222222222222223e-06,
-      "loss": 1.0131,
+      "loss": 1.0224,
       "step": 57720
     },
     {
       "epoch": 173.36,
-      "grad_norm": 5.449119567871094,
+      "grad_norm": 5.3648810386657715,
       "learning_rate": 4.2212212212212214e-06,
-      "loss": 0.9508,
+      "loss": 0.9501,
       "step": 57730
     },
     {
       "epoch": 173.39,
-      "grad_norm": 6.861564636230469,
+      "grad_norm": 6.141861438751221,
       "learning_rate": 4.220220220220221e-06,
-      "loss": 0.943,
+      "loss": 0.9423,
       "step": 57740
     },
     {
       "epoch": 173.42,
-      "grad_norm": 7.004024028778076,
+      "grad_norm": 6.866944313049316,
       "learning_rate": 4.2192192192192195e-06,
-      "loss": 1.0162,
+      "loss": 1.0073,
       "step": 57750
     },
     {
       "epoch": 173.45,
-      "grad_norm": 5.916383266448975,
+      "grad_norm": 7.327688694000244,
       "learning_rate": 4.218218218218219e-06,
-      "loss": 0.9621,
+      "loss": 0.9532,
       "step": 57760
     },
     {
       "epoch": 173.48,
-      "grad_norm": 4.602097034454346,
+      "grad_norm": 4.667279243469238,
       "learning_rate": 4.2172172172172175e-06,
-      "loss": 0.9804,
+      "loss": 0.9762,
       "step": 57770
     },
     {
       "epoch": 173.51,
-      "grad_norm": 6.606781959533691,
+      "grad_norm": 8.712727546691895,
       "learning_rate": 4.216216216216217e-06,
-      "loss": 0.9543,
+      "loss": 0.9559,
       "step": 57780
     },
     {
       "epoch": 173.54,
-      "grad_norm": 7.198867321014404,
+      "grad_norm": 8.319018363952637,
       "learning_rate": 4.2152152152152155e-06,
-      "loss": 0.9872,
+      "loss": 0.9724,
       "step": 57790
     },
     {
       "epoch": 173.57,
-      "grad_norm": 6.261878490447998,
+      "grad_norm": 7.39320182800293,
       "learning_rate": 4.214214214214214e-06,
-      "loss": 0.9817,
+      "loss": 0.9702,
       "step": 57800
     },
     {
       "epoch": 173.6,
-      "grad_norm": 6.593649864196777,
+      "grad_norm": 5.550201416015625,
       "learning_rate": 4.2132132132132136e-06,
-      "loss": 0.9274,
+      "loss": 0.9165,
       "step": 57810
     },
     {
       "epoch": 173.63,
-      "grad_norm": 6.105283737182617,
+      "grad_norm": 8.082877159118652,
       "learning_rate": 4.212212212212213e-06,
-      "loss": 0.9802,
+      "loss": 0.9783,
       "step": 57820
     },
     {
       "epoch": 173.66,
-      "grad_norm": 7.759555816650391,
+      "grad_norm": 9.308167457580566,
       "learning_rate": 4.211211211211212e-06,
-      "loss": 1.0265,
+      "loss": 1.026,
       "step": 57830
     },
     {
       "epoch": 173.69,
-      "grad_norm": 6.122910976409912,
+      "grad_norm": 10.583247184753418,
       "learning_rate": 4.21021021021021e-06,
-      "loss": 0.8643,
+      "loss": 0.8663,
       "step": 57840
     },
     {
       "epoch": 173.72,
-      "grad_norm": 5.64223051071167,
+      "grad_norm": 6.337741851806641,
       "learning_rate": 4.20920920920921e-06,
-      "loss": 0.9753,
+      "loss": 0.9744,
       "step": 57850
     },
     {
       "epoch": 173.75,
-      "grad_norm": 4.522459506988525,
+      "grad_norm": 4.486853122711182,
       "learning_rate": 4.208208208208209e-06,
-      "loss": 0.9651,
+      "loss": 0.9797,
       "step": 57860
     },
     {
       "epoch": 173.78,
-      "grad_norm": 6.292692184448242,
+      "grad_norm": 8.574864387512207,
       "learning_rate": 4.207207207207208e-06,
-      "loss": 1.0306,
+      "loss": 1.0284,
       "step": 57870
     },
     {
       "epoch": 173.81,
-      "grad_norm": 5.7735819816589355,
+      "grad_norm": 6.404019832611084,
       "learning_rate": 4.206206206206206e-06,
-      "loss": 0.979,
+      "loss": 0.9703,
       "step": 57880
     },
     {
       "epoch": 173.84,
-      "grad_norm": 7.4018168449401855,
+      "grad_norm": 7.932193756103516,
       "learning_rate": 4.205205205205206e-06,
-      "loss": 0.9848,
+      "loss": 0.9821,
       "step": 57890
     },
     {
       "epoch": 173.87,
-      "grad_norm": 7.323965072631836,
+      "grad_norm": 5.840383052825928,
       "learning_rate": 4.204204204204204e-06,
-      "loss": 0.9791,
+      "loss": 0.9757,
       "step": 57900
     },
     {
       "epoch": 173.9,
-      "grad_norm": 5.989774703979492,
+      "grad_norm": 7.002708435058594,
       "learning_rate": 4.203203203203203e-06,
-      "loss": 0.9385,
+      "loss": 0.9291,
       "step": 57910
     },
     {
       "epoch": 173.93,
-      "grad_norm": 5.1103901863098145,
+      "grad_norm": 3.832151412963867,
       "learning_rate": 4.202202202202202e-06,
-      "loss": 0.9318,
+      "loss": 0.9356,
       "step": 57920
     },
     {
       "epoch": 173.96,
-      "grad_norm": 9.376553535461426,
+      "grad_norm": 10.616680145263672,
       "learning_rate": 4.201201201201202e-06,
-      "loss": 1.0243,
+      "loss": 1.0171,
       "step": 57930
     },
     {
       "epoch": 173.99,
-      "grad_norm": 6.509884834289551,
+      "grad_norm": 5.112054824829102,
       "learning_rate": 4.2002002002002e-06,
-      "loss": 1.0559,
+      "loss": 1.0513,
       "step": 57940
     },
     {
       "epoch": 174.0,
-      "eval_accuracy": 0.8197,
-      "eval_loss": 0.5892412662506104,
-      "eval_runtime": 5.3429,
-      "eval_samples_per_second": 1871.651,
-      "eval_steps_per_second": 7.487,
+      "eval_accuracy": 0.8209,
+      "eval_loss": 0.5884857773780823,
+      "eval_runtime": 5.3483,
+      "eval_samples_per_second": 1869.758,
+      "eval_steps_per_second": 7.479,
       "step": 57942
     },
     {
       "epoch": 174.02,
-      "grad_norm": 5.853391170501709,
+      "grad_norm": 6.173883438110352,
       "learning_rate": 4.199199199199199e-06,
-      "loss": 1.0957,
+      "loss": 1.0961,
       "step": 57950
     },
     {
       "epoch": 174.05,
-      "grad_norm": 7.806312561035156,
+      "grad_norm": 8.601030349731445,
       "learning_rate": 4.1981981981981984e-06,
-      "loss": 1.0572,
+      "loss": 1.0617,
       "step": 57960
     },
     {
       "epoch": 174.08,
-      "grad_norm": 7.707867622375488,
+      "grad_norm": 8.365843772888184,
       "learning_rate": 4.197197197197198e-06,
-      "loss": 0.9288,
+      "loss": 0.9223,
       "step": 57970
     },
     {
       "epoch": 174.11,
-      "grad_norm": 6.284613132476807,
+      "grad_norm": 8.371768951416016,
       "learning_rate": 4.1961961961961965e-06,
-      "loss": 1.0123,
+      "loss": 1.014,
       "step": 57980
     },
     {
       "epoch": 174.14,
-      "grad_norm": 6.570204257965088,
+      "grad_norm": 7.29685640335083,
       "learning_rate": 4.195195195195196e-06,
-      "loss": 0.9707,
+      "loss": 0.9671,
       "step": 57990
     },
     {
       "epoch": 174.17,
-      "grad_norm": 5.463085174560547,
+      "grad_norm": 6.050613880157471,
       "learning_rate": 4.1941941941941945e-06,
-      "loss": 0.9696,
+      "loss": 0.9641,
       "step": 58000
     },
     {
       "epoch": 174.2,
-      "grad_norm": 7.718828201293945,
+      "grad_norm": 9.009360313415527,
       "learning_rate": 4.193193193193193e-06,
-      "loss": 1.0773,
+      "loss": 1.0776,
       "step": 58010
     },
     {
       "epoch": 174.23,
-      "grad_norm": 8.799976348876953,
+      "grad_norm": 7.811835289001465,
       "learning_rate": 4.1921921921921925e-06,
-      "loss": 0.9322,
+      "loss": 0.922,
       "step": 58020
     },
     {
       "epoch": 174.26,
-      "grad_norm": 8.720549583435059,
+      "grad_norm": 7.404264450073242,
       "learning_rate": 4.191191191191192e-06,
-      "loss": 0.9323,
+      "loss": 0.9334,
       "step": 58030
     },
     {
       "epoch": 174.29,
-      "grad_norm": 6.783285140991211,
+      "grad_norm": 8.908015251159668,
       "learning_rate": 4.1901901901901906e-06,
-      "loss": 1.0404,
+      "loss": 1.0292,
       "step": 58040
     },
     {
       "epoch": 174.32,
-      "grad_norm": 5.497209072113037,
+      "grad_norm": 6.204119682312012,
       "learning_rate": 4.189189189189189e-06,
-      "loss": 0.9406,
+      "loss": 0.9511,
       "step": 58050
     },
     {
       "epoch": 174.35,
-      "grad_norm": 6.747400760650635,
+      "grad_norm": 8.075756072998047,
       "learning_rate": 4.188188188188189e-06,
-      "loss": 1.0132,
+      "loss": 1.0004,
       "step": 58060
     },
     {
       "epoch": 174.38,
-      "grad_norm": 6.171955585479736,
+      "grad_norm": 8.040233612060547,
       "learning_rate": 4.187187187187188e-06,
-      "loss": 0.9758,
+      "loss": 0.9682,
       "step": 58070
     },
     {
       "epoch": 174.41,
-      "grad_norm": 4.95564079284668,
+      "grad_norm": 4.608746528625488,
       "learning_rate": 4.186186186186187e-06,
-      "loss": 0.957,
+      "loss": 0.9678,
       "step": 58080
     },
     {
       "epoch": 174.44,
-      "grad_norm": 4.54019021987915,
+      "grad_norm": 4.788285732269287,
       "learning_rate": 4.185185185185185e-06,
-      "loss": 0.9582,
+      "loss": 0.9568,
       "step": 58090
     },
     {
       "epoch": 174.47,
-      "grad_norm": 9.138566017150879,
+      "grad_norm": 6.005876064300537,
       "learning_rate": 4.184184184184185e-06,
-      "loss": 0.9636,
+      "loss": 0.9607,
       "step": 58100
     },
     {
       "epoch": 174.5,
-      "grad_norm": 6.8391523361206055,
+      "grad_norm": 9.562610626220703,
       "learning_rate": 4.183183183183184e-06,
-      "loss": 0.973,
+      "loss": 0.9797,
       "step": 58110
     },
     {
       "epoch": 174.53,
-      "grad_norm": 5.657789707183838,
+      "grad_norm": 8.039691925048828,
       "learning_rate": 4.182182182182183e-06,
-      "loss": 0.9469,
+      "loss": 0.9317,
       "step": 58120
     },
     {
       "epoch": 174.56,
-      "grad_norm": 7.3769917488098145,
+      "grad_norm": 7.46379280090332,
       "learning_rate": 4.181181181181181e-06,
-      "loss": 0.9868,
+      "loss": 0.9852,
       "step": 58130
     },
     {
       "epoch": 174.59,
-      "grad_norm": 7.1208815574646,
+      "grad_norm": 9.878857612609863,
       "learning_rate": 4.180180180180181e-06,
-      "loss": 1.0321,
+      "loss": 1.0288,
       "step": 58140
     },
     {
       "epoch": 174.62,
-      "grad_norm": 5.609315872192383,
+      "grad_norm": 5.694334983825684,
       "learning_rate": 4.179179179179179e-06,
-      "loss": 1.0079,
+      "loss": 0.9895,
       "step": 58150
     },
     {
       "epoch": 174.65,
-      "grad_norm": 7.822027683258057,
+      "grad_norm": 8.57294750213623,
       "learning_rate": 4.178178178178178e-06,
-      "loss": 1.0319,
+      "loss": 1.0361,
       "step": 58160
     },
     {
       "epoch": 174.68,
-      "grad_norm": 6.565327167510986,
+      "grad_norm": 7.998125076293945,
       "learning_rate": 4.177177177177177e-06,
-      "loss": 0.9908,
+      "loss": 0.9841,
       "step": 58170
     },
     {
       "epoch": 174.71,
-      "grad_norm": 5.376637935638428,
+      "grad_norm": 6.822649002075195,
       "learning_rate": 4.176176176176177e-06,
-      "loss": 0.9818,
+      "loss": 0.9701,
       "step": 58180
     },
     {
       "epoch": 174.74,
-      "grad_norm": 5.632751941680908,
+      "grad_norm": 5.1343488693237305,
       "learning_rate": 4.175175175175175e-06,
-      "loss": 0.9711,
+      "loss": 0.9709,
       "step": 58190
     },
     {
       "epoch": 174.77,
-      "grad_norm": 6.147202014923096,
+      "grad_norm": 6.383197784423828,
       "learning_rate": 4.174174174174174e-06,
-      "loss": 1.0039,
+      "loss": 1.0081,
       "step": 58200
     },
     {
       "epoch": 174.8,
-      "grad_norm": 6.187526702880859,
+      "grad_norm": 7.8227691650390625,
       "learning_rate": 4.1731731731731735e-06,
-      "loss": 0.9546,
+      "loss": 0.9545,
       "step": 58210
     },
     {
       "epoch": 174.83,
-      "grad_norm": 7.120652675628662,
+      "grad_norm": 8.257023811340332,
       "learning_rate": 4.172172172172173e-06,
-      "loss": 1.0204,
+      "loss": 1.015,
       "step": 58220
     },
     {
       "epoch": 174.86,
-      "grad_norm": 6.278927326202393,
+      "grad_norm": 5.72097110748291,
       "learning_rate": 4.1711711711711715e-06,
-      "loss": 0.9822,
+      "loss": 0.9739,
       "step": 58230
     },
     {
       "epoch": 174.89,
-      "grad_norm": 6.5379719734191895,
+      "grad_norm": 6.705107688903809,
       "learning_rate": 4.170170170170171e-06,
-      "loss": 0.9783,
+      "loss": 0.9753,
       "step": 58240
     },
     {
       "epoch": 174.92,
-      "grad_norm": 6.286907196044922,
+      "grad_norm": 5.1713666915893555,
       "learning_rate": 4.1691691691691695e-06,
-      "loss": 1.0041,
+      "loss": 0.9983,
       "step": 58250
     },
     {
       "epoch": 174.95,
-      "grad_norm": 6.343276500701904,
+      "grad_norm": 5.597589015960693,
       "learning_rate": 4.168168168168168e-06,
-      "loss": 1.005,
+      "loss": 1.0087,
       "step": 58260
     },
     {
       "epoch": 174.98,
-      "grad_norm": 6.695322513580322,
+      "grad_norm": 10.031847953796387,
       "learning_rate": 4.1671671671671676e-06,
-      "loss": 1.0002,
+      "loss": 0.9993,
       "step": 58270
     },
     {
       "epoch": 175.0,
-      "eval_accuracy": 0.8201,
-      "eval_loss": 0.5913780927658081,
-      "eval_runtime": 5.3941,
-      "eval_samples_per_second": 1853.882,
-      "eval_steps_per_second": 7.416,
+      "eval_accuracy": 0.8185,
+      "eval_loss": 0.5905084609985352,
+      "eval_runtime": 5.3729,
+      "eval_samples_per_second": 1861.205,
+      "eval_steps_per_second": 7.445,
       "step": 58275
     },
     {
       "epoch": 175.02,
-      "grad_norm": 7.172609806060791,
+      "grad_norm": 6.010506629943848,
       "learning_rate": 4.166166166166167e-06,
-      "loss": 1.1915,
+      "loss": 1.1632,
       "step": 58280
     },
     {
       "epoch": 175.05,
-      "grad_norm": 8.809659004211426,
+      "grad_norm": 10.169755935668945,
       "learning_rate": 4.165165165165166e-06,
-      "loss": 1.0057,
+      "loss": 1.0042,
       "step": 58290
     },
     {
       "epoch": 175.08,
-      "grad_norm": 7.96044921875,
+      "grad_norm": 7.647428512573242,
       "learning_rate": 4.164164164164164e-06,
-      "loss": 0.9941,
+      "loss": 0.984,
       "step": 58300
     },
     {
       "epoch": 175.11,
-      "grad_norm": 6.590322017669678,
+      "grad_norm": 8.654297828674316,
       "learning_rate": 4.163163163163164e-06,
-      "loss": 1.0192,
+      "loss": 1.024,
       "step": 58310
     },
     {
       "epoch": 175.14,
-      "grad_norm": 7.782774448394775,
+      "grad_norm": 7.196133613586426,
       "learning_rate": 4.162162162162163e-06,
-      "loss": 0.9382,
+      "loss": 0.9365,
       "step": 58320
     },
     {
       "epoch": 175.17,
-      "grad_norm": 7.630946159362793,
+      "grad_norm": 7.91795539855957,
       "learning_rate": 4.161161161161162e-06,
-      "loss": 0.9513,
+      "loss": 0.9447,
       "step": 58330
     },
     {
       "epoch": 175.2,
-      "grad_norm": 8.412841796875,
+      "grad_norm": 9.177855491638184,
       "learning_rate": 4.16016016016016e-06,
-      "loss": 0.9684,
+      "loss": 0.9675,
       "step": 58340
     },
     {
       "epoch": 175.23,
-      "grad_norm": 6.440208911895752,
+      "grad_norm": 5.512609004974365,
       "learning_rate": 4.15915915915916e-06,
-      "loss": 0.913,
+      "loss": 0.9109,
       "step": 58350
     },
     {
       "epoch": 175.26,
-      "grad_norm": 5.901694297790527,
+      "grad_norm": 5.867053031921387,
       "learning_rate": 4.158158158158158e-06,
-      "loss": 0.9802,
+      "loss": 0.9904,
       "step": 58360
     },
     {
       "epoch": 175.29,
-      "grad_norm": 4.9975762367248535,
+      "grad_norm": 5.592979431152344,
       "learning_rate": 4.157157157157157e-06,
-      "loss": 0.9838,
+      "loss": 0.9881,
       "step": 58370
     },
     {
       "epoch": 175.32,
-      "grad_norm": 5.109960079193115,
+      "grad_norm": 5.9607930183410645,
       "learning_rate": 4.156156156156156e-06,
-      "loss": 0.996,
+      "loss": 1.0081,
       "step": 58380
     },
     {
       "epoch": 175.35,
-      "grad_norm": 6.92929220199585,
+      "grad_norm": 6.80201530456543,
       "learning_rate": 4.155155155155156e-06,
-      "loss": 0.9878,
+      "loss": 0.9885,
       "step": 58390
     },
     {
       "epoch": 175.38,
-      "grad_norm": 6.672657012939453,
+      "grad_norm": 8.724517822265625,
       "learning_rate": 4.154154154154154e-06,
-      "loss": 0.9809,
+      "loss": 0.982,
       "step": 58400
     },
     {
       "epoch": 175.41,
-      "grad_norm": 5.0153093338012695,
+      "grad_norm": 4.616136074066162,
       "learning_rate": 4.153153153153153e-06,
-      "loss": 0.95,
+      "loss": 0.956,
       "step": 58410
     },
     {
       "epoch": 175.44,
-      "grad_norm": 8.40100383758545,
+      "grad_norm": 8.320347785949707,
       "learning_rate": 4.152152152152152e-06,
-      "loss": 0.9752,
+      "loss": 0.9647,
       "step": 58420
     },
     {
       "epoch": 175.47,
-      "grad_norm": 9.054948806762695,
+      "grad_norm": 8.598138809204102,
       "learning_rate": 4.151151151151152e-06,
-      "loss": 0.9555,
+      "loss": 0.9619,
       "step": 58430
     },
     {
       "epoch": 175.5,
-      "grad_norm": 8.451876640319824,
+      "grad_norm": 8.147773742675781,
       "learning_rate": 4.1501501501501505e-06,
-      "loss": 0.904,
+      "loss": 0.8926,
       "step": 58440
     },
     {
       "epoch": 175.53,
-      "grad_norm": 6.250154972076416,
+      "grad_norm": 6.211183547973633,
       "learning_rate": 4.149149149149149e-06,
-      "loss": 1.0268,
+      "loss": 1.0235,
       "step": 58450
     },
     {
       "epoch": 175.56,
-      "grad_norm": 5.5402703285217285,
+      "grad_norm": 6.525309085845947,
       "learning_rate": 4.1481481481481485e-06,
-      "loss": 0.9903,
+      "loss": 0.975,
       "step": 58460
     },
     {
       "epoch": 175.59,
-      "grad_norm": 5.348860740661621,
+      "grad_norm": 5.5716447830200195,
       "learning_rate": 4.147147147147147e-06,
-      "loss": 0.9079,
+      "loss": 0.9022,
       "step": 58470
     },
     {
       "epoch": 175.62,
-      "grad_norm": 5.3319268226623535,
+      "grad_norm": 5.316502094268799,
       "learning_rate": 4.1461461461461465e-06,
-      "loss": 1.0113,
+      "loss": 1.011,
       "step": 58480
     },
     {
       "epoch": 175.65,
-      "grad_norm": 7.987618923187256,
+      "grad_norm": 6.471047878265381,
       "learning_rate": 4.145145145145145e-06,
-      "loss": 0.9574,
+      "loss": 0.9573,
       "step": 58490
     },
     {
       "epoch": 175.68,
-      "grad_norm": 11.5653657913208,
+      "grad_norm": 8.515604972839355,
       "learning_rate": 4.1441441441441446e-06,
-      "loss": 1.0473,
+      "loss": 1.0496,
       "step": 58500
     },
     {
       "epoch": 175.71,
-      "grad_norm": 6.7287211418151855,
+      "grad_norm": 9.119002342224121,
       "learning_rate": 4.143143143143143e-06,
-      "loss": 1.0015,
+      "loss": 1.0001,
       "step": 58510
     },
     {
       "epoch": 175.74,
-      "grad_norm": 7.230432510375977,
+      "grad_norm": 8.893181800842285,
       "learning_rate": 4.142142142142143e-06,
-      "loss": 1.0024,
+      "loss": 1.0021,
       "step": 58520
     },
     {
       "epoch": 175.77,
-      "grad_norm": 7.97581148147583,
+      "grad_norm": 9.601815223693848,
       "learning_rate": 4.141141141141142e-06,
-      "loss": 0.9013,
+      "loss": 0.8987,
       "step": 58530
     },
     {
       "epoch": 175.8,
-      "grad_norm": 6.142941951751709,
+      "grad_norm": 8.190877914428711,
       "learning_rate": 4.140140140140141e-06,
-      "loss": 1.0274,
+      "loss": 1.0232,
       "step": 58540
     },
     {
       "epoch": 175.83,
-      "grad_norm": 11.120235443115234,
+      "grad_norm": 10.024890899658203,
       "learning_rate": 4.139139139139139e-06,
-      "loss": 0.9958,
+      "loss": 0.9822,
       "step": 58550
     },
     {
       "epoch": 175.86,
-      "grad_norm": 5.919925689697266,
+      "grad_norm": 6.275481700897217,
       "learning_rate": 4.138138138138139e-06,
-      "loss": 0.9668,
+      "loss": 0.9605,
       "step": 58560
     },
     {
       "epoch": 175.89,
-      "grad_norm": 5.732372760772705,
+      "grad_norm": 8.532727241516113,
       "learning_rate": 4.137137137137138e-06,
-      "loss": 1.0652,
+      "loss": 1.0686,
       "step": 58570
     },
     {
       "epoch": 175.92,
-      "grad_norm": 6.905848979949951,
+      "grad_norm": 6.871089458465576,
       "learning_rate": 4.136136136136137e-06,
-      "loss": 1.0533,
+      "loss": 1.0539,
       "step": 58580
     },
     {
       "epoch": 175.95,
-      "grad_norm": 9.146328926086426,
+      "grad_norm": 10.726669311523438,
       "learning_rate": 4.135135135135135e-06,
-      "loss": 1.0469,
+      "loss": 1.0354,
       "step": 58590
     },
     {
       "epoch": 175.98,
-      "grad_norm": 7.759104251861572,
+      "grad_norm": 7.439610481262207,
       "learning_rate": 4.134134134134135e-06,
-      "loss": 0.9461,
+      "loss": 0.9425,
       "step": 58600
     },
     {
       "epoch": 176.0,
       "eval_accuracy": 0.8214,
-      "eval_loss": 0.5866039991378784,
-      "eval_runtime": 5.5172,
-      "eval_samples_per_second": 1812.52,
-      "eval_steps_per_second": 7.25,
+      "eval_loss": 0.586967408657074,
+      "eval_runtime": 5.5647,
+      "eval_samples_per_second": 1797.058,
+      "eval_steps_per_second": 7.188,
       "step": 58608
     },
     {
       "epoch": 176.01,
-      "grad_norm": 6.91691255569458,
+      "grad_norm": 5.8647236824035645,
       "learning_rate": 4.133133133133133e-06,
-      "loss": 1.2031,
+      "loss": 1.211,
       "step": 58610
     },
     {
       "epoch": 176.04,
-      "grad_norm": 7.2989821434021,
+      "grad_norm": 6.754680156707764,
       "learning_rate": 4.132132132132132e-06,
-      "loss": 1.0071,
+      "loss": 0.9929,
       "step": 58620
     },
     {
       "epoch": 176.07,
-      "grad_norm": 7.9031758308410645,
+      "grad_norm": 5.933424472808838,
       "learning_rate": 4.131131131131131e-06,
-      "loss": 1.0337,
+      "loss": 1.0301,
       "step": 58630
     },
     {
       "epoch": 176.1,
-      "grad_norm": 6.815690994262695,
+      "grad_norm": 6.450525760650635,
       "learning_rate": 4.130130130130131e-06,
-      "loss": 0.9896,
+      "loss": 0.9848,
       "step": 58640
     },
     {
       "epoch": 176.13,
-      "grad_norm": 7.053646564483643,
+      "grad_norm": 6.530848979949951,
       "learning_rate": 4.129129129129129e-06,
-      "loss": 1.0168,
+      "loss": 1.0133,
       "step": 58650
     },
     {
       "epoch": 176.16,
-      "grad_norm": 5.050972938537598,
+      "grad_norm": 6.5562286376953125,
       "learning_rate": 4.128128128128128e-06,
-      "loss": 0.9015,
+      "loss": 0.9086,
       "step": 58660
     },
     {
       "epoch": 176.19,
-      "grad_norm": 6.892662048339844,
+      "grad_norm": 6.070527076721191,
       "learning_rate": 4.1271271271271275e-06,
-      "loss": 0.9701,
+      "loss": 0.9761,
       "step": 58670
     },
     {
       "epoch": 176.22,
-      "grad_norm": 6.430692672729492,
+      "grad_norm": 7.687796592712402,
       "learning_rate": 4.126126126126127e-06,
-      "loss": 0.9702,
+      "loss": 0.9613,
       "step": 58680
     },
     {
       "epoch": 176.25,
-      "grad_norm": 7.213876247406006,
+      "grad_norm": 7.27675199508667,
       "learning_rate": 4.1251251251251255e-06,
-      "loss": 1.0088,
+      "loss": 0.996,
       "step": 58690
     },
     {
       "epoch": 176.28,
-      "grad_norm": 5.2469162940979,
+      "grad_norm": 5.69980525970459,
       "learning_rate": 4.124124124124124e-06,
-      "loss": 0.9937,
+      "loss": 0.9845,
       "step": 58700
     },
     {
       "epoch": 176.31,
-      "grad_norm": 6.23927116394043,
+      "grad_norm": 6.803303241729736,
       "learning_rate": 4.1231231231231235e-06,
-      "loss": 1.0215,
+      "loss": 1.0149,
       "step": 58710
     },
     {
       "epoch": 176.34,
-      "grad_norm": 6.072787761688232,
+      "grad_norm": 7.583559989929199,
       "learning_rate": 4.122122122122122e-06,
-      "loss": 0.9464,
+      "loss": 0.9534,
       "step": 58720
     },
     {
       "epoch": 176.37,
-      "grad_norm": 7.561134338378906,
+      "grad_norm": 6.790592193603516,
       "learning_rate": 4.1211211211211216e-06,
-      "loss": 0.9453,
+      "loss": 0.9596,
       "step": 58730
     },
     {
       "epoch": 176.4,
-      "grad_norm": 4.724495887756348,
+      "grad_norm": 5.054145812988281,
       "learning_rate": 4.12012012012012e-06,
-      "loss": 0.8882,
+      "loss": 0.8928,
       "step": 58740
     },
     {
       "epoch": 176.43,
-      "grad_norm": 8.23934555053711,
+      "grad_norm": 9.477456092834473,
       "learning_rate": 4.11911911911912e-06,
-      "loss": 0.9282,
+      "loss": 0.9263,
       "step": 58750
     },
     {
       "epoch": 176.46,
-      "grad_norm": 8.265459060668945,
+      "grad_norm": 6.778630256652832,
       "learning_rate": 4.118118118118118e-06,
-      "loss": 0.9318,
+      "loss": 0.9249,
       "step": 58760
     },
     {
       "epoch": 176.49,
-      "grad_norm": 5.93311071395874,
+      "grad_norm": 5.631801128387451,
       "learning_rate": 4.117117117117118e-06,
-      "loss": 0.9399,
+      "loss": 0.9516,
       "step": 58770
     },
     {
       "epoch": 176.52,
-      "grad_norm": 7.962337970733643,
+      "grad_norm": 8.173842430114746,
       "learning_rate": 4.116116116116117e-06,
-      "loss": 0.9451,
+      "loss": 0.9379,
       "step": 58780
     },
     {
       "epoch": 176.55,
-      "grad_norm": 8.016249656677246,
+      "grad_norm": 11.543488502502441,
       "learning_rate": 4.115115115115116e-06,
-      "loss": 1.0109,
+      "loss": 1.0156,
       "step": 58790
     },
     {
       "epoch": 176.58,
-      "grad_norm": 6.1313323974609375,
+      "grad_norm": 6.78938102722168,
       "learning_rate": 4.114114114114114e-06,
-      "loss": 0.9019,
+      "loss": 0.8998,
       "step": 58800
     },
     {
       "epoch": 176.61,
-      "grad_norm": 8.012249946594238,
+      "grad_norm": 6.9497480392456055,
       "learning_rate": 4.113113113113114e-06,
-      "loss": 0.9315,
+      "loss": 0.9396,
       "step": 58810
     },
     {
       "epoch": 176.64,
-      "grad_norm": 5.715963840484619,
+      "grad_norm": 6.853225231170654,
       "learning_rate": 4.112112112112112e-06,
-      "loss": 0.8326,
+      "loss": 0.8256,
       "step": 58820
     },
     {
       "epoch": 176.67,
-      "grad_norm": 6.087641716003418,
+      "grad_norm": 8.79919147491455,
       "learning_rate": 4.111111111111111e-06,
-      "loss": 0.9626,
+      "loss": 0.9548,
       "step": 58830
     },
     {
       "epoch": 176.7,
-      "grad_norm": 6.494189262390137,
+      "grad_norm": 5.52346134185791,
       "learning_rate": 4.11011011011011e-06,
-      "loss": 0.9545,
+      "loss": 0.9541,
       "step": 58840
     },
     {
       "epoch": 176.73,
-      "grad_norm": 8.417920112609863,
+      "grad_norm": 7.920896053314209,
       "learning_rate": 4.10910910910911e-06,
-      "loss": 1.0042,
+      "loss": 1.0091,
       "step": 58850
     },
     {
       "epoch": 176.76,
-      "grad_norm": 8.029001235961914,
+      "grad_norm": 8.423033714294434,
       "learning_rate": 4.108108108108108e-06,
-      "loss": 0.8966,
+      "loss": 0.9026,
       "step": 58860
     },
     {
       "epoch": 176.79,
-      "grad_norm": 5.63749885559082,
+      "grad_norm": 7.461137771606445,
       "learning_rate": 4.107107107107107e-06,
-      "loss": 1.0173,
+      "loss": 1.0144,
       "step": 58870
     },
     {
       "epoch": 176.82,
-      "grad_norm": 5.569163799285889,
+      "grad_norm": 6.689416408538818,
       "learning_rate": 4.106106106106106e-06,
-      "loss": 0.9532,
+      "loss": 0.9505,
       "step": 58880
     },
     {
       "epoch": 176.85,
-      "grad_norm": 4.568454742431641,
+      "grad_norm": 6.058223247528076,
       "learning_rate": 4.105105105105106e-06,
-      "loss": 0.972,
+      "loss": 0.9687,
       "step": 58890
     },
     {
       "epoch": 176.88,
-      "grad_norm": 5.972090721130371,
+      "grad_norm": 5.8557353019714355,
       "learning_rate": 4.1041041041041045e-06,
-      "loss": 1.0056,
+      "loss": 0.9987,
       "step": 58900
     },
     {
       "epoch": 176.91,
-      "grad_norm": 6.339194297790527,
+      "grad_norm": 8.221125602722168,
       "learning_rate": 4.103103103103103e-06,
-      "loss": 0.9738,
+      "loss": 0.9819,
       "step": 58910
     },
     {
       "epoch": 176.94,
-      "grad_norm": 7.985220432281494,
+      "grad_norm": 12.524197578430176,
       "learning_rate": 4.1021021021021025e-06,
-      "loss": 1.0071,
+      "loss": 0.9944,
       "step": 58920
     },
     {
       "epoch": 176.97,
-      "grad_norm": 6.632358074188232,
+      "grad_norm": 6.800198078155518,
       "learning_rate": 4.101101101101101e-06,
-      "loss": 0.9517,
+      "loss": 0.9475,
       "step": 58930
     },
     {
       "epoch": 177.0,
-      "grad_norm": 5.184008598327637,
+      "grad_norm": 5.910663604736328,
       "learning_rate": 4.1001001001001005e-06,
-      "loss": 0.9624,
+      "loss": 0.9589,
       "step": 58940
     },
     {
       "epoch": 177.0,
-      "eval_accuracy": 0.8228,
-      "eval_loss": 0.586977481842041,
-      "eval_runtime": 5.4486,
-      "eval_samples_per_second": 1835.318,
-      "eval_steps_per_second": 7.341,
+      "eval_accuracy": 0.8237,
+      "eval_loss": 0.5849549770355225,
+      "eval_runtime": 5.4891,
+      "eval_samples_per_second": 1821.79,
+      "eval_steps_per_second": 7.287,
       "step": 58941
     },
     {
       "epoch": 177.03,
-      "grad_norm": 6.312966823577881,
+      "grad_norm": 7.296267509460449,
       "learning_rate": 4.099099099099099e-06,
-      "loss": 0.9912,
+      "loss": 1.0082,
       "step": 58950
     },
     {
       "epoch": 177.06,
-      "grad_norm": 6.7403364181518555,
+      "grad_norm": 7.410564422607422,
       "learning_rate": 4.0980980980980986e-06,
-      "loss": 0.9786,
+      "loss": 0.9766,
       "step": 58960
     },
     {
       "epoch": 177.09,
-      "grad_norm": 10.316843032836914,
+      "grad_norm": 8.078182220458984,
       "learning_rate": 4.097097097097097e-06,
-      "loss": 1.0198,
+      "loss": 1.0164,
       "step": 58970
     },
     {
       "epoch": 177.12,
-      "grad_norm": 5.928155899047852,
+      "grad_norm": 6.238293647766113,
       "learning_rate": 4.096096096096097e-06,
-      "loss": 0.9778,
+      "loss": 0.9637,
       "step": 58980
     },
     {
       "epoch": 177.15,
-      "grad_norm": 9.002729415893555,
+      "grad_norm": 8.370416641235352,
       "learning_rate": 4.095095095095095e-06,
-      "loss": 0.9122,
+      "loss": 0.9064,
       "step": 58990
     },
     {
       "epoch": 177.18,
-      "grad_norm": 5.512416839599609,
+      "grad_norm": 5.842286109924316,
       "learning_rate": 4.094094094094095e-06,
-      "loss": 0.8896,
+      "loss": 0.8905,
       "step": 59000
     },
     {
       "epoch": 177.21,
-      "grad_norm": 7.157411575317383,
+      "grad_norm": 5.184284687042236,
       "learning_rate": 4.093093093093093e-06,
-      "loss": 0.9462,
+      "loss": 0.9432,
       "step": 59010
     },
     {
       "epoch": 177.24,
-      "grad_norm": 5.240059852600098,
+      "grad_norm": 5.51246452331543,
       "learning_rate": 4.092092092092093e-06,
-      "loss": 0.9534,
+      "loss": 0.9441,
       "step": 59020
     },
     {
       "epoch": 177.27,
-      "grad_norm": 7.131496429443359,
+      "grad_norm": 7.640038967132568,
       "learning_rate": 4.091091091091091e-06,
-      "loss": 0.9794,
+      "loss": 0.9824,
       "step": 59030
     },
     {
       "epoch": 177.3,
-      "grad_norm": 8.588711738586426,
+      "grad_norm": 7.6236891746521,
       "learning_rate": 4.09009009009009e-06,
-      "loss": 0.9568,
+      "loss": 0.9584,
       "step": 59040
     },
     {
       "epoch": 177.33,
-      "grad_norm": 6.290826320648193,
+      "grad_norm": 7.353691101074219,
       "learning_rate": 4.089089089089089e-06,
-      "loss": 0.9411,
+      "loss": 0.9293,
       "step": 59050
     },
     {
       "epoch": 177.36,
-      "grad_norm": 4.969542026519775,
+      "grad_norm": 5.273831367492676,
       "learning_rate": 4.088088088088089e-06,
-      "loss": 0.9334,
+      "loss": 0.9232,
       "step": 59060
     },
     {
       "epoch": 177.39,
-      "grad_norm": 6.266404151916504,
+      "grad_norm": 6.454809665679932,
       "learning_rate": 4.087087087087087e-06,
-      "loss": 0.9591,
+      "loss": 0.9485,
       "step": 59070
     },
     {
       "epoch": 177.42,
-      "grad_norm": 7.29256010055542,
+      "grad_norm": 10.562555313110352,
       "learning_rate": 4.086086086086086e-06,
-      "loss": 1.0205,
+      "loss": 1.0129,
       "step": 59080
     },
     {
       "epoch": 177.45,
-      "grad_norm": 8.588647842407227,
+      "grad_norm": 9.029181480407715,
       "learning_rate": 4.085085085085085e-06,
-      "loss": 0.9912,
+      "loss": 0.9849,
       "step": 59090
     },
     {
       "epoch": 177.48,
-      "grad_norm": 9.692782402038574,
+      "grad_norm": 7.5036725997924805,
       "learning_rate": 4.084084084084085e-06,
-      "loss": 0.9669,
+      "loss": 0.9594,
       "step": 59100
     },
     {
       "epoch": 177.51,
-      "grad_norm": 5.764029502868652,
+      "grad_norm": 4.975048542022705,
       "learning_rate": 4.083083083083083e-06,
-      "loss": 0.9219,
+      "loss": 0.9174,
       "step": 59110
     },
     {
       "epoch": 177.54,
-      "grad_norm": 6.945075988769531,
+      "grad_norm": 6.3143630027771,
       "learning_rate": 4.082082082082082e-06,
-      "loss": 1.0031,
+      "loss": 1.0042,
       "step": 59120
     },
     {
       "epoch": 177.57,
-      "grad_norm": 8.184958457946777,
+      "grad_norm": 9.539887428283691,
       "learning_rate": 4.0810810810810815e-06,
-      "loss": 0.9302,
+      "loss": 0.9282,
       "step": 59130
     },
     {
       "epoch": 177.6,
-      "grad_norm": 5.418379306793213,
+      "grad_norm": 5.339883804321289,
       "learning_rate": 4.080080080080081e-06,
-      "loss": 0.9446,
+      "loss": 0.9444,
       "step": 59140
     },
     {
       "epoch": 177.63,
-      "grad_norm": 5.475845813751221,
+      "grad_norm": 4.2025017738342285,
       "learning_rate": 4.0790790790790795e-06,
-      "loss": 0.9298,
+      "loss": 0.928,
       "step": 59150
     },
     {
       "epoch": 177.66,
-      "grad_norm": 5.835297584533691,
+      "grad_norm": 7.884823322296143,
       "learning_rate": 4.078078078078078e-06,
-      "loss": 0.8967,
+      "loss": 0.9029,
       "step": 59160
     },
     {
       "epoch": 177.69,
-      "grad_norm": 6.831631660461426,
+      "grad_norm": 7.418975830078125,
       "learning_rate": 4.0770770770770775e-06,
-      "loss": 0.9901,
+      "loss": 0.9875,
       "step": 59170
     },
     {
       "epoch": 177.72,
-      "grad_norm": 5.634005069732666,
+      "grad_norm": 7.284444332122803,
       "learning_rate": 4.076076076076076e-06,
-      "loss": 1.0066,
+      "loss": 1.0108,
       "step": 59180
     },
     {
       "epoch": 177.75,
-      "grad_norm": 6.601406097412109,
+      "grad_norm": 5.639531135559082,
       "learning_rate": 4.0750750750750756e-06,
-      "loss": 1.0103,
+      "loss": 1.0131,
       "step": 59190
     },
     {
       "epoch": 177.78,
-      "grad_norm": 7.681553840637207,
+      "grad_norm": 6.029305458068848,
       "learning_rate": 4.074074074074074e-06,
-      "loss": 0.9086,
+      "loss": 0.9059,
       "step": 59200
     },
     {
       "epoch": 177.81,
-      "grad_norm": 5.995511054992676,
+      "grad_norm": 7.725765228271484,
       "learning_rate": 4.073073073073074e-06,
-      "loss": 0.9775,
+      "loss": 0.9817,
       "step": 59210
     },
     {
       "epoch": 177.84,
-      "grad_norm": 5.725948810577393,
+      "grad_norm": 7.0089874267578125,
       "learning_rate": 4.072072072072072e-06,
-      "loss": 1.0101,
+      "loss": 1.009,
       "step": 59220
     },
     {
       "epoch": 177.87,
-      "grad_norm": 7.413930416107178,
+      "grad_norm": 7.108767509460449,
       "learning_rate": 4.071071071071072e-06,
-      "loss": 0.9385,
+      "loss": 0.9341,
       "step": 59230
     },
     {
       "epoch": 177.9,
-      "grad_norm": 8.294300079345703,
+      "grad_norm": 9.356579780578613,
       "learning_rate": 4.07007007007007e-06,
-      "loss": 1.0444,
+      "loss": 1.0467,
       "step": 59240
     },
     {
       "epoch": 177.93,
-      "grad_norm": 5.722809314727783,
+      "grad_norm": 5.546184539794922,
       "learning_rate": 4.06906906906907e-06,
-      "loss": 0.9066,
+      "loss": 0.9028,
       "step": 59250
     },
     {
       "epoch": 177.96,
-      "grad_norm": 7.304844856262207,
+      "grad_norm": 8.075434684753418,
       "learning_rate": 4.068068068068068e-06,
-      "loss": 1.0373,
+      "loss": 1.0378,
       "step": 59260
     },
     {
       "epoch": 177.99,
-      "grad_norm": 6.885715961456299,
+      "grad_norm": 7.371589183807373,
       "learning_rate": 4.067067067067068e-06,
-      "loss": 0.9952,
+      "loss": 0.9953,
       "step": 59270
     },
     {
       "epoch": 178.0,
-      "eval_accuracy": 0.8199,
-      "eval_loss": 0.5919607281684875,
-      "eval_runtime": 5.9323,
-      "eval_samples_per_second": 1685.675,
-      "eval_steps_per_second": 6.743,
+      "eval_accuracy": 0.8195,
+      "eval_loss": 0.5912211537361145,
+      "eval_runtime": 5.7091,
+      "eval_samples_per_second": 1751.6,
+      "eval_steps_per_second": 7.006,
       "step": 59274
     },
     {
       "epoch": 178.02,
-      "grad_norm": 9.297898292541504,
+      "grad_norm": 6.883633613586426,
       "learning_rate": 4.066066066066066e-06,
-      "loss": 1.1356,
+      "loss": 1.1292,
       "step": 59280
     },
     {
       "epoch": 178.05,
-      "grad_norm": 5.214542865753174,
+      "grad_norm": 7.331965923309326,
       "learning_rate": 4.065065065065065e-06,
-      "loss": 1.0066,
+      "loss": 1.0172,
       "step": 59290
     },
     {
       "epoch": 178.08,
-      "grad_norm": 6.934868812561035,
+      "grad_norm": 9.249367713928223,
       "learning_rate": 4.064064064064064e-06,
-      "loss": 0.9281,
+      "loss": 0.932,
       "step": 59300
     },
     {
       "epoch": 178.11,
-      "grad_norm": 8.844340324401855,
+      "grad_norm": 10.441234588623047,
       "learning_rate": 4.063063063063064e-06,
-      "loss": 0.9649,
+      "loss": 0.9717,
       "step": 59310
     },
     {
       "epoch": 178.14,
-      "grad_norm": 6.534623622894287,
+      "grad_norm": 9.019403457641602,
       "learning_rate": 4.062062062062062e-06,
-      "loss": 0.9671,
+      "loss": 0.9582,
       "step": 59320
     },
     {
       "epoch": 178.17,
-      "grad_norm": 7.842656135559082,
+      "grad_norm": 9.614323616027832,
       "learning_rate": 4.061061061061061e-06,
-      "loss": 0.9476,
+      "loss": 0.9453,
       "step": 59330
     },
     {
       "epoch": 178.2,
-      "grad_norm": 4.758697986602783,
+      "grad_norm": 4.962985515594482,
       "learning_rate": 4.06006006006006e-06,
-      "loss": 0.948,
+      "loss": 0.9509,
       "step": 59340
     },
     {
       "epoch": 178.23,
-      "grad_norm": 8.111039161682129,
+      "grad_norm": 8.295848846435547,
       "learning_rate": 4.05905905905906e-06,
-      "loss": 0.9343,
+      "loss": 0.9386,
       "step": 59350
     },
     {
       "epoch": 178.26,
-      "grad_norm": 8.43136978149414,
+      "grad_norm": 8.011237144470215,
       "learning_rate": 4.0580580580580584e-06,
-      "loss": 0.98,
+      "loss": 0.9889,
       "step": 59360
     },
     {
       "epoch": 178.29,
-      "grad_norm": 4.225381851196289,
+      "grad_norm": 5.451390266418457,
       "learning_rate": 4.057057057057057e-06,
-      "loss": 0.9682,
+      "loss": 0.9582,
       "step": 59370
     },
     {
       "epoch": 178.32,
-      "grad_norm": 7.699958801269531,
+      "grad_norm": 8.401768684387207,
       "learning_rate": 4.0560560560560565e-06,
-      "loss": 1.0316,
+      "loss": 1.0303,
       "step": 59380
     },
     {
       "epoch": 178.35,
-      "grad_norm": 7.271623611450195,
+      "grad_norm": 8.551661491394043,
       "learning_rate": 4.055055055055055e-06,
-      "loss": 0.9061,
+      "loss": 0.9145,
       "step": 59390
     },
     {
       "epoch": 178.38,
-      "grad_norm": 5.242176532745361,
+      "grad_norm": 6.354477882385254,
       "learning_rate": 4.0540540540540545e-06,
-      "loss": 0.9277,
+      "loss": 0.9279,
       "step": 59400
     },
     {
       "epoch": 178.41,
-      "grad_norm": 6.657094478607178,
+      "grad_norm": 7.527266025543213,
       "learning_rate": 4.053053053053053e-06,
-      "loss": 0.9788,
+      "loss": 0.9847,
       "step": 59410
     },
     {
       "epoch": 178.44,
-      "grad_norm": 9.116490364074707,
+      "grad_norm": 7.9801344871521,
       "learning_rate": 4.0520520520520526e-06,
-      "loss": 0.9812,
+      "loss": 0.9765,
       "step": 59420
     },
     {
       "epoch": 178.47,
-      "grad_norm": 6.314152240753174,
+      "grad_norm": 6.574300289154053,
       "learning_rate": 4.051051051051051e-06,
-      "loss": 0.9543,
+      "loss": 0.9588,
       "step": 59430
     },
     {
       "epoch": 178.5,
-      "grad_norm": 8.576586723327637,
+      "grad_norm": 7.4991655349731445,
       "learning_rate": 4.050050050050051e-06,
-      "loss": 0.9797,
+      "loss": 0.9788,
       "step": 59440
     },
     {
       "epoch": 178.53,
-      "grad_norm": 7.649954795837402,
+      "grad_norm": 6.899722099304199,
       "learning_rate": 4.049049049049049e-06,
-      "loss": 1.016,
+      "loss": 1.0144,
       "step": 59450
     },
     {
       "epoch": 178.56,
-      "grad_norm": 6.692655563354492,
+      "grad_norm": 7.144195556640625,
       "learning_rate": 4.048048048048049e-06,
-      "loss": 0.9931,
+      "loss": 0.993,
       "step": 59460
     },
     {
       "epoch": 178.59,
-      "grad_norm": 7.0217366218566895,
+      "grad_norm": 7.605905532836914,
       "learning_rate": 4.047047047047047e-06,
-      "loss": 0.8842,
+      "loss": 0.8793,
       "step": 59470
     },
     {
       "epoch": 178.62,
-      "grad_norm": 5.817065238952637,
+      "grad_norm": 7.2137532234191895,
       "learning_rate": 4.046046046046047e-06,
-      "loss": 0.9808,
+      "loss": 0.989,
       "step": 59480
     },
     {
       "epoch": 178.65,
-      "grad_norm": 8.091401100158691,
+      "grad_norm": 8.046785354614258,
       "learning_rate": 4.045045045045045e-06,
-      "loss": 0.9446,
+      "loss": 0.943,
       "step": 59490
     },
     {
       "epoch": 178.68,
-      "grad_norm": 9.100872039794922,
+      "grad_norm": 9.41783332824707,
       "learning_rate": 4.044044044044044e-06,
-      "loss": 0.9816,
+      "loss": 0.9842,
       "step": 59500
     },
     {
       "epoch": 178.71,
-      "grad_norm": 7.53174352645874,
+      "grad_norm": 8.513449668884277,
       "learning_rate": 4.043043043043043e-06,
-      "loss": 1.0264,
+      "loss": 1.0244,
       "step": 59510
     },
     {
       "epoch": 178.74,
-      "grad_norm": 5.5866289138793945,
+      "grad_norm": 8.396339416503906,
       "learning_rate": 4.042042042042043e-06,
-      "loss": 0.9793,
+      "loss": 0.9663,
       "step": 59520
     },
     {
       "epoch": 178.77,
-      "grad_norm": 7.743398189544678,
+      "grad_norm": 6.937052249908447,
       "learning_rate": 4.041041041041041e-06,
-      "loss": 0.9729,
+      "loss": 0.9671,
       "step": 59530
     },
     {
       "epoch": 178.8,
-      "grad_norm": 4.225981712341309,
+      "grad_norm": 6.678079605102539,
       "learning_rate": 4.04004004004004e-06,
-      "loss": 0.9703,
+      "loss": 0.9699,
       "step": 59540
     },
     {
       "epoch": 178.83,
-      "grad_norm": 5.580868244171143,
+      "grad_norm": 5.194333553314209,
       "learning_rate": 4.039039039039039e-06,
-      "loss": 0.9616,
+      "loss": 0.9547,
       "step": 59550
     },
     {
       "epoch": 178.86,
-      "grad_norm": 7.6490888595581055,
+      "grad_norm": 9.401595115661621,
       "learning_rate": 4.038038038038039e-06,
-      "loss": 0.9304,
+      "loss": 0.9231,
       "step": 59560
     },
     {
       "epoch": 178.89,
-      "grad_norm": 10.122507095336914,
+      "grad_norm": 9.321494102478027,
       "learning_rate": 4.037037037037037e-06,
-      "loss": 1.037,
+      "loss": 1.0263,
       "step": 59570
     },
     {
       "epoch": 178.92,
-      "grad_norm": 6.580507278442383,
+      "grad_norm": 5.961899757385254,
       "learning_rate": 4.036036036036036e-06,
-      "loss": 1.0068,
+      "loss": 0.9989,
       "step": 59580
     },
     {
       "epoch": 178.95,
-      "grad_norm": 7.060535430908203,
+      "grad_norm": 6.056222438812256,
       "learning_rate": 4.0350350350350354e-06,
-      "loss": 0.9659,
+      "loss": 0.9588,
       "step": 59590
     },
     {
       "epoch": 178.98,
-      "grad_norm": 4.8992815017700195,
+      "grad_norm": 5.508442401885986,
       "learning_rate": 4.034034034034035e-06,
-      "loss": 1.0415,
+      "loss": 1.0399,
       "step": 59600
     },
     {
       "epoch": 179.0,
-      "eval_accuracy": 0.8193,
-      "eval_loss": 0.5925759673118591,
-      "eval_runtime": 5.6629,
-      "eval_samples_per_second": 1765.88,
-      "eval_steps_per_second": 7.064,
+      "eval_accuracy": 0.8211,
+      "eval_loss": 0.5911418795585632,
+      "eval_runtime": 5.1167,
+      "eval_samples_per_second": 1954.39,
+      "eval_steps_per_second": 7.818,
       "step": 59607
     },
     {
       "epoch": 179.01,
-      "grad_norm": 8.648131370544434,
+      "grad_norm": 7.499538898468018,
       "learning_rate": 4.0330330330330335e-06,
-      "loss": 1.1045,
+      "loss": 1.0867,
       "step": 59610
     },
     {
       "epoch": 179.04,
-      "grad_norm": 6.468850135803223,
+      "grad_norm": 7.302982330322266,
       "learning_rate": 4.032032032032032e-06,
-      "loss": 0.9595,
+      "loss": 0.9668,
       "step": 59620
     },
     {
       "epoch": 179.07,
-      "grad_norm": 5.842111110687256,
+      "grad_norm": 6.938148021697998,
       "learning_rate": 4.0310310310310315e-06,
-      "loss": 0.985,
+      "loss": 0.9799,
       "step": 59630
     },
     {
       "epoch": 179.1,
-      "grad_norm": 5.357604026794434,
+      "grad_norm": 7.483407497406006,
       "learning_rate": 4.03003003003003e-06,
-      "loss": 0.925,
+      "loss": 0.934,
       "step": 59640
     },
     {
       "epoch": 179.13,
-      "grad_norm": 8.031611442565918,
+      "grad_norm": 8.690408706665039,
       "learning_rate": 4.0290290290290296e-06,
-      "loss": 0.9656,
+      "loss": 0.9716,
       "step": 59650
     },
     {
       "epoch": 179.16,
-      "grad_norm": 5.517383098602295,
+      "grad_norm": 6.048762798309326,
       "learning_rate": 4.028028028028028e-06,
-      "loss": 0.9629,
+      "loss": 0.9653,
       "step": 59660
     },
     {
       "epoch": 179.19,
-      "grad_norm": 5.826840400695801,
+      "grad_norm": 7.0963921546936035,
       "learning_rate": 4.027027027027028e-06,
-      "loss": 0.9177,
+      "loss": 0.9101,
       "step": 59670
     },
     {
       "epoch": 179.22,
-      "grad_norm": 8.03215503692627,
+      "grad_norm": 8.076611518859863,
       "learning_rate": 4.026026026026026e-06,
-      "loss": 1.0092,
+      "loss": 0.9986,
       "step": 59680
     },
     {
       "epoch": 179.25,
-      "grad_norm": 5.746580123901367,
+      "grad_norm": 5.896687030792236,
       "learning_rate": 4.025025025025026e-06,
-      "loss": 0.9903,
+      "loss": 0.9885,
       "step": 59690
     },
     {
       "epoch": 179.28,
-      "grad_norm": 7.679192543029785,
+      "grad_norm": 7.363318920135498,
       "learning_rate": 4.024024024024024e-06,
-      "loss": 1.017,
+      "loss": 1.0135,
       "step": 59700
     },
     {
       "epoch": 179.31,
-      "grad_norm": 6.824161529541016,
+      "grad_norm": 6.913417339324951,
       "learning_rate": 4.023023023023024e-06,
-      "loss": 1.0102,
+      "loss": 1.0162,
       "step": 59710
     },
     {
       "epoch": 179.34,
-      "grad_norm": 8.1859712600708,
+      "grad_norm": 6.416074275970459,
       "learning_rate": 4.022022022022022e-06,
-      "loss": 0.9841,
+      "loss": 0.9898,
       "step": 59720
     },
     {
       "epoch": 179.37,
-      "grad_norm": 6.760232448577881,
+      "grad_norm": 5.941421031951904,
       "learning_rate": 4.021021021021022e-06,
-      "loss": 0.9674,
+      "loss": 0.9582,
       "step": 59730
     },
     {
       "epoch": 179.4,
-      "grad_norm": 9.86778450012207,
+      "grad_norm": 7.704451084136963,
       "learning_rate": 4.02002002002002e-06,
-      "loss": 0.9782,
+      "loss": 0.9779,
       "step": 59740
     },
     {
       "epoch": 179.43,
-      "grad_norm": 6.490161418914795,
+      "grad_norm": 7.161685943603516,
       "learning_rate": 4.019019019019019e-06,
-      "loss": 0.9609,
+      "loss": 0.9703,
       "step": 59750
     },
     {
       "epoch": 179.46,
-      "grad_norm": 6.072545051574707,
+      "grad_norm": 5.14897346496582,
       "learning_rate": 4.018018018018018e-06,
-      "loss": 0.9972,
+      "loss": 1.0028,
       "step": 59760
     },
     {
       "epoch": 179.49,
-      "grad_norm": 6.074062347412109,
+      "grad_norm": 6.851978778839111,
       "learning_rate": 4.017017017017018e-06,
-      "loss": 0.8927,
+      "loss": 0.9023,
       "step": 59770
     },
     {
       "epoch": 179.52,
-      "grad_norm": 7.905488014221191,
+      "grad_norm": 7.042934417724609,
       "learning_rate": 4.016016016016016e-06,
-      "loss": 0.9451,
+      "loss": 0.9423,
       "step": 59780
     },
     {
       "epoch": 179.55,
-      "grad_norm": 7.677745342254639,
+      "grad_norm": 6.448281288146973,
       "learning_rate": 4.015015015015015e-06,
-      "loss": 0.9242,
+      "loss": 0.9195,
       "step": 59790
     },
     {
       "epoch": 179.58,
-      "grad_norm": 5.667516708374023,
+      "grad_norm": 7.204006195068359,
       "learning_rate": 4.014014014014014e-06,
-      "loss": 1.0096,
+      "loss": 1.0056,
       "step": 59800
     },
     {
       "epoch": 179.61,
-      "grad_norm": 7.969326972961426,
+      "grad_norm": 10.442361831665039,
       "learning_rate": 4.013013013013014e-06,
-      "loss": 0.9757,
+      "loss": 0.9803,
       "step": 59810
     },
     {
       "epoch": 179.64,
-      "grad_norm": 6.181922912597656,
+      "grad_norm": 7.235001564025879,
       "learning_rate": 4.0120120120120124e-06,
-      "loss": 0.9199,
+      "loss": 0.922,
       "step": 59820
     },
     {
       "epoch": 179.67,
-      "grad_norm": 5.521868705749512,
+      "grad_norm": 7.128024578094482,
       "learning_rate": 4.011011011011011e-06,
-      "loss": 0.9795,
+      "loss": 0.9827,
       "step": 59830
     },
     {
       "epoch": 179.7,
-      "grad_norm": 7.393014907836914,
+      "grad_norm": 7.7471842765808105,
       "learning_rate": 4.0100100100100105e-06,
-      "loss": 0.9618,
+      "loss": 0.9612,
       "step": 59840
     },
     {
       "epoch": 179.73,
-      "grad_norm": 6.203984260559082,
+      "grad_norm": 6.5710577964782715,
       "learning_rate": 4.009009009009009e-06,
-      "loss": 0.9711,
+      "loss": 0.9753,
       "step": 59850
     },
     {
       "epoch": 179.76,
-      "grad_norm": 5.784074306488037,
+      "grad_norm": 5.701641082763672,
       "learning_rate": 4.0080080080080085e-06,
-      "loss": 0.9217,
+      "loss": 0.9235,
       "step": 59860
     },
     {
       "epoch": 179.79,
-      "grad_norm": 5.57011079788208,
+      "grad_norm": 8.069244384765625,
       "learning_rate": 4.007007007007007e-06,
-      "loss": 0.9786,
+      "loss": 0.9768,
       "step": 59870
     },
     {
       "epoch": 179.82,
-      "grad_norm": 8.136204719543457,
+      "grad_norm": 6.092705249786377,
       "learning_rate": 4.0060060060060066e-06,
-      "loss": 0.9246,
+      "loss": 0.9035,
       "step": 59880
     },
     {
       "epoch": 179.85,
-      "grad_norm": 5.158721446990967,
+      "grad_norm": 4.976334095001221,
       "learning_rate": 4.005005005005005e-06,
-      "loss": 0.9556,
+      "loss": 0.9499,
       "step": 59890
     },
     {
       "epoch": 179.88,
-      "grad_norm": 7.472280502319336,
+      "grad_norm": 6.6235432624816895,
       "learning_rate": 4.004004004004005e-06,
-      "loss": 0.9222,
+      "loss": 0.9263,
       "step": 59900
     },
     {
       "epoch": 179.91,
-      "grad_norm": 6.400526523590088,
+      "grad_norm": 7.031864166259766,
       "learning_rate": 4.003003003003003e-06,
-      "loss": 0.9628,
+      "loss": 0.9608,
       "step": 59910
     },
     {
       "epoch": 179.94,
-      "grad_norm": 6.690732002258301,
+      "grad_norm": 6.168126106262207,
       "learning_rate": 4.002002002002003e-06,
-      "loss": 1.0286,
+      "loss": 1.0169,
       "step": 59920
     },
     {
       "epoch": 179.97,
-      "grad_norm": 6.7889909744262695,
+      "grad_norm": 6.644896030426025,
       "learning_rate": 4.001001001001001e-06,
-      "loss": 1.0183,
+      "loss": 1.017,
       "step": 59930
     },
     {
       "epoch": 180.0,
-      "grad_norm": 48.54082107543945,
+      "grad_norm": 52.651275634765625,
       "learning_rate": 4.000000000000001e-06,
-      "loss": 1.0416,
+      "loss": 1.0385,
       "step": 59940
     },
     {
       "epoch": 180.0,
-      "eval_accuracy": 0.8206,
-      "eval_loss": 0.590079128742218,
-      "eval_runtime": 5.4866,
-      "eval_samples_per_second": 1822.609,
-      "eval_steps_per_second": 7.29,
+      "eval_accuracy": 0.8243,
+      "eval_loss": 0.5887439250946045,
+      "eval_runtime": 5.4467,
+      "eval_samples_per_second": 1835.964,
+      "eval_steps_per_second": 7.344,
       "step": 59940
     },
     {
       "epoch": 180.03,
-      "grad_norm": 9.334248542785645,
+      "grad_norm": 7.805271625518799,
       "learning_rate": 3.998998998998999e-06,
-      "loss": 0.983,
+      "loss": 0.9799,
       "step": 59950
     },
     {
       "epoch": 180.06,
-      "grad_norm": 6.895930290222168,
+      "grad_norm": 8.394794464111328,
       "learning_rate": 3.997997997997998e-06,
-      "loss": 0.9857,
+      "loss": 0.9735,
       "step": 59960
     },
     {
       "epoch": 180.09,
-      "grad_norm": 9.062329292297363,
+      "grad_norm": 7.367812633514404,
       "learning_rate": 3.996996996996997e-06,
-      "loss": 0.9488,
+      "loss": 0.9517,
       "step": 59970
     },
     {
       "epoch": 180.12,
-      "grad_norm": 7.126073837280273,
+      "grad_norm": 8.625434875488281,
       "learning_rate": 3.995995995995997e-06,
-      "loss": 0.9917,
+      "loss": 0.9928,
       "step": 59980
     },
     {
       "epoch": 180.15,
-      "grad_norm": 6.594622611999512,
+      "grad_norm": 6.119296073913574,
       "learning_rate": 3.994994994994995e-06,
-      "loss": 0.9752,
+      "loss": 0.9648,
       "step": 59990
     },
     {
       "epoch": 180.18,
-      "grad_norm": 8.826480865478516,
+      "grad_norm": 8.845227241516113,
       "learning_rate": 3.993993993993994e-06,
-      "loss": 0.9705,
+      "loss": 0.9688,
       "step": 60000
     },
     {
       "epoch": 180.21,
-      "grad_norm": 7.766104221343994,
+      "grad_norm": 7.248865127563477,
       "learning_rate": 3.992992992992993e-06,
-      "loss": 1.0128,
+      "loss": 1.0103,
       "step": 60010
     },
     {
       "epoch": 180.24,
-      "grad_norm": 7.213733673095703,
+      "grad_norm": 8.618568420410156,
       "learning_rate": 3.991991991991993e-06,
-      "loss": 0.9717,
+      "loss": 0.9672,
       "step": 60020
     },
     {
       "epoch": 180.27,
-      "grad_norm": 7.857553482055664,
+      "grad_norm": 5.711571216583252,
       "learning_rate": 3.990990990990991e-06,
-      "loss": 0.8823,
+      "loss": 0.8799,
       "step": 60030
     },
     {
       "epoch": 180.3,
-      "grad_norm": 5.872647285461426,
+      "grad_norm": 4.906576633453369,
       "learning_rate": 3.98998998998999e-06,
-      "loss": 1.013,
+      "loss": 1.0139,
       "step": 60040
     },
     {
       "epoch": 180.33,
-      "grad_norm": 6.716306686401367,
+      "grad_norm": 6.179689884185791,
       "learning_rate": 3.9889889889889894e-06,
-      "loss": 0.9195,
+      "loss": 0.9104,
       "step": 60050
     },
     {
       "epoch": 180.36,
-      "grad_norm": 4.424661636352539,
+      "grad_norm": 4.694790840148926,
       "learning_rate": 3.987987987987989e-06,
-      "loss": 0.998,
+      "loss": 0.9994,
       "step": 60060
     },
     {
       "epoch": 180.39,
-      "grad_norm": 8.070015907287598,
+      "grad_norm": 11.017017364501953,
       "learning_rate": 3.986986986986987e-06,
-      "loss": 0.9603,
+      "loss": 0.9648,
       "step": 60070
     },
     {
       "epoch": 180.42,
-      "grad_norm": 6.675721168518066,
+      "grad_norm": 6.914079666137695,
       "learning_rate": 3.985985985985986e-06,
-      "loss": 0.9657,
+      "loss": 0.9589,
       "step": 60080
     },
     {
       "epoch": 180.45,
-      "grad_norm": 6.9713616371154785,
+      "grad_norm": 6.145934581756592,
       "learning_rate": 3.9849849849849855e-06,
-      "loss": 0.9622,
+      "loss": 0.9535,
       "step": 60090
     },
     {
       "epoch": 180.48,
-      "grad_norm": 6.18906831741333,
+      "grad_norm": 6.88093900680542,
       "learning_rate": 3.983983983983984e-06,
-      "loss": 0.9315,
+      "loss": 0.9385,
       "step": 60100
     },
     {
       "epoch": 180.51,
-      "grad_norm": 6.998172283172607,
+      "grad_norm": 7.533775806427002,
       "learning_rate": 3.9829829829829836e-06,
-      "loss": 0.9752,
+      "loss": 0.9772,
       "step": 60110
     },
     {
       "epoch": 180.54,
-      "grad_norm": 7.993656635284424,
+      "grad_norm": 8.763158798217773,
       "learning_rate": 3.981981981981982e-06,
-      "loss": 0.9896,
+      "loss": 0.9997,
       "step": 60120
     },
     {
       "epoch": 180.57,
-      "grad_norm": 4.176937580108643,
+      "grad_norm": 4.6551737785339355,
       "learning_rate": 3.980980980980982e-06,
-      "loss": 0.8821,
+      "loss": 0.8897,
       "step": 60130
     },
     {
       "epoch": 180.6,
-      "grad_norm": 6.9626641273498535,
+      "grad_norm": 9.63400936126709,
       "learning_rate": 3.97997997997998e-06,
-      "loss": 1.0362,
+      "loss": 1.0312,
       "step": 60140
     },
     {
       "epoch": 180.63,
-      "grad_norm": 9.447319030761719,
+      "grad_norm": 6.736459732055664,
       "learning_rate": 3.97897897897898e-06,
-      "loss": 0.9823,
+      "loss": 0.9816,
       "step": 60150
     },
     {
       "epoch": 180.66,
-      "grad_norm": 8.223974227905273,
+      "grad_norm": 7.209841251373291,
       "learning_rate": 3.977977977977978e-06,
-      "loss": 0.9577,
+      "loss": 0.9601,
       "step": 60160
     },
     {
       "epoch": 180.69,
-      "grad_norm": 8.676678657531738,
+      "grad_norm": 5.898496150970459,
       "learning_rate": 3.976976976976978e-06,
-      "loss": 0.9876,
+      "loss": 0.9795,
       "step": 60170
     },
     {
       "epoch": 180.72,
-      "grad_norm": 7.482052326202393,
+      "grad_norm": 7.16316556930542,
       "learning_rate": 3.975975975975976e-06,
-      "loss": 0.9026,
+      "loss": 0.9007,
       "step": 60180
     },
     {
       "epoch": 180.75,
-      "grad_norm": 7.2139716148376465,
+      "grad_norm": 7.560372829437256,
       "learning_rate": 3.974974974974976e-06,
-      "loss": 0.9949,
+      "loss": 0.987,
       "step": 60190
     },
     {
       "epoch": 180.78,
-      "grad_norm": 5.192409038543701,
+      "grad_norm": 5.717531681060791,
       "learning_rate": 3.973973973973974e-06,
-      "loss": 1.039,
+      "loss": 1.0415,
       "step": 60200
     },
     {
       "epoch": 180.81,
-      "grad_norm": 5.190308094024658,
+      "grad_norm": 5.042459487915039,
       "learning_rate": 3.972972972972973e-06,
-      "loss": 0.901,
+      "loss": 0.8978,
       "step": 60210
     },
     {
       "epoch": 180.84,
-      "grad_norm": 6.153444766998291,
+      "grad_norm": 7.515032768249512,
       "learning_rate": 3.971971971971972e-06,
-      "loss": 0.9423,
+      "loss": 0.9311,
       "step": 60220
     },
     {
       "epoch": 180.87,
-      "grad_norm": 6.867712020874023,
+      "grad_norm": 6.8535871505737305,
       "learning_rate": 3.970970970970972e-06,
-      "loss": 0.9939,
+      "loss": 0.9829,
       "step": 60230
     },
     {
       "epoch": 180.9,
-      "grad_norm": 6.823915481567383,
+      "grad_norm": 6.3883795738220215,
       "learning_rate": 3.96996996996997e-06,
-      "loss": 1.0289,
+      "loss": 1.0291,
       "step": 60240
     },
     {
       "epoch": 180.93,
-      "grad_norm": 8.47498607635498,
+      "grad_norm": 9.42122745513916,
       "learning_rate": 3.968968968968969e-06,
-      "loss": 0.9367,
+      "loss": 0.9274,
       "step": 60250
     },
     {
       "epoch": 180.96,
-      "grad_norm": 6.381810188293457,
+      "grad_norm": 5.941969394683838,
       "learning_rate": 3.967967967967968e-06,
-      "loss": 0.9663,
+      "loss": 0.9765,
       "step": 60260
     },
     {
       "epoch": 180.99,
-      "grad_norm": 7.005844593048096,
+      "grad_norm": 6.9397873878479,
       "learning_rate": 3.966966966966968e-06,
-      "loss": 0.9467,
+      "loss": 0.945,
       "step": 60270
     },
     {
       "epoch": 181.0,
-      "eval_accuracy": 0.8216,
-      "eval_loss": 0.5910734534263611,
-      "eval_runtime": 5.526,
-      "eval_samples_per_second": 1809.621,
-      "eval_steps_per_second": 7.238,
+      "eval_accuracy": 0.8213,
+      "eval_loss": 0.5919082760810852,
+      "eval_runtime": 5.3433,
+      "eval_samples_per_second": 1871.493,
+      "eval_steps_per_second": 7.486,
       "step": 60273
     },
     {
       "epoch": 181.02,
-      "grad_norm": 6.47843599319458,
+      "grad_norm": 11.141839027404785,
       "learning_rate": 3.9659659659659664e-06,
-      "loss": 1.126,
+      "loss": 1.1006,
       "step": 60280
     },
     {
       "epoch": 181.05,
-      "grad_norm": 5.901058673858643,
+      "grad_norm": 8.140180587768555,
       "learning_rate": 3.964964964964965e-06,
-      "loss": 0.9237,
+      "loss": 0.9195,
       "step": 60290
     },
     {
       "epoch": 181.08,
-      "grad_norm": 6.291253566741943,
+      "grad_norm": 6.658304691314697,
       "learning_rate": 3.9639639639639645e-06,
-      "loss": 0.9498,
+      "loss": 0.9456,
       "step": 60300
     },
     {
       "epoch": 181.11,
-      "grad_norm": 6.645325660705566,
+      "grad_norm": 5.851245880126953,
       "learning_rate": 3.962962962962963e-06,
-      "loss": 0.9607,
+      "loss": 0.971,
       "step": 60310
     },
     {
       "epoch": 181.14,
-      "grad_norm": 6.676445007324219,
+      "grad_norm": 6.593517303466797,
       "learning_rate": 3.961961961961962e-06,
-      "loss": 0.9493,
+      "loss": 0.9482,
       "step": 60320
     },
     {
       "epoch": 181.17,
-      "grad_norm": 6.53177547454834,
+      "grad_norm": 7.580690383911133,
       "learning_rate": 3.960960960960961e-06,
-      "loss": 1.002,
+      "loss": 1.0061,
       "step": 60330
     },
     {
       "epoch": 181.2,
-      "grad_norm": 5.83335018157959,
+      "grad_norm": 7.298375606536865,
       "learning_rate": 3.9599599599599606e-06,
-      "loss": 0.9432,
+      "loss": 0.9407,
       "step": 60340
     },
     {
       "epoch": 181.23,
-      "grad_norm": 9.920125961303711,
+      "grad_norm": 11.751030921936035,
       "learning_rate": 3.958958958958959e-06,
-      "loss": 0.9871,
+      "loss": 0.9912,
       "step": 60350
     },
     {
       "epoch": 181.26,
-      "grad_norm": 6.5418701171875,
+      "grad_norm": 6.690117359161377,
       "learning_rate": 3.957957957957958e-06,
-      "loss": 1.0002,
+      "loss": 0.9952,
       "step": 60360
     },
     {
       "epoch": 181.29,
-      "grad_norm": 8.101462364196777,
+      "grad_norm": 7.410055637359619,
       "learning_rate": 3.956956956956957e-06,
-      "loss": 0.9336,
+      "loss": 0.9363,
       "step": 60370
     },
     {
       "epoch": 181.32,
-      "grad_norm": 5.949239730834961,
+      "grad_norm": 7.702282905578613,
       "learning_rate": 3.955955955955957e-06,
-      "loss": 1.0002,
+      "loss": 0.9943,
       "step": 60380
     },
     {
       "epoch": 181.35,
-      "grad_norm": 8.215999603271484,
+      "grad_norm": 7.784963607788086,
       "learning_rate": 3.954954954954955e-06,
-      "loss": 1.0464,
+      "loss": 1.0432,
       "step": 60390
     },
     {
       "epoch": 181.38,
-      "grad_norm": 5.646831512451172,
+      "grad_norm": 6.3151655197143555,
       "learning_rate": 3.953953953953955e-06,
-      "loss": 1.0036,
+      "loss": 1.0049,
       "step": 60400
     },
     {
       "epoch": 181.41,
-      "grad_norm": 5.518808364868164,
+      "grad_norm": 5.405436038970947,
       "learning_rate": 3.952952952952953e-06,
-      "loss": 0.9832,
+      "loss": 0.9872,
       "step": 60410
     },
     {
       "epoch": 181.44,
-      "grad_norm": 6.631608009338379,
+      "grad_norm": 5.960411071777344,
       "learning_rate": 3.951951951951952e-06,
-      "loss": 0.9514,
+      "loss": 0.9526,
       "step": 60420
     },
     {
       "epoch": 181.47,
-      "grad_norm": 6.08339786529541,
+      "grad_norm": 6.945624351501465,
       "learning_rate": 3.950950950950951e-06,
-      "loss": 0.941,
+      "loss": 0.9489,
       "step": 60430
     },
     {
       "epoch": 181.5,
-      "grad_norm": 6.093608379364014,
+      "grad_norm": 7.381246566772461,
       "learning_rate": 3.949949949949951e-06,
-      "loss": 0.9773,
+      "loss": 0.9652,
       "step": 60440
     },
     {
       "epoch": 181.53,
-      "grad_norm": 5.9904303550720215,
+      "grad_norm": 5.946290493011475,
       "learning_rate": 3.948948948948949e-06,
-      "loss": 0.9211,
+      "loss": 0.9237,
       "step": 60450
     },
     {
       "epoch": 181.56,
-      "grad_norm": 6.0152363777160645,
+      "grad_norm": 5.232936382293701,
       "learning_rate": 3.947947947947948e-06,
-      "loss": 0.9417,
+      "loss": 0.9461,
       "step": 60460
     },
     {
       "epoch": 181.59,
-      "grad_norm": 5.110505104064941,
+      "grad_norm": 6.582051753997803,
       "learning_rate": 3.946946946946947e-06,
-      "loss": 0.9734,
+      "loss": 0.9552,
       "step": 60470
     },
     {
       "epoch": 181.62,
-      "grad_norm": 5.980525970458984,
+      "grad_norm": 8.444500923156738,
       "learning_rate": 3.945945945945947e-06,
-      "loss": 0.9169,
+      "loss": 0.921,
       "step": 60480
     },
     {
       "epoch": 181.65,
-      "grad_norm": 6.474083423614502,
+      "grad_norm": 8.74199104309082,
       "learning_rate": 3.944944944944945e-06,
-      "loss": 0.9447,
+      "loss": 0.9484,
       "step": 60490
     },
     {
       "epoch": 181.68,
-      "grad_norm": 7.457697868347168,
+      "grad_norm": 7.496757507324219,
       "learning_rate": 3.943943943943944e-06,
-      "loss": 0.9619,
+      "loss": 0.9662,
       "step": 60500
     },
     {
       "epoch": 181.71,
-      "grad_norm": 7.364903450012207,
+      "grad_norm": 6.294915676116943,
       "learning_rate": 3.9429429429429434e-06,
-      "loss": 0.9125,
+      "loss": 0.9106,
       "step": 60510
     },
     {
       "epoch": 181.74,
-      "grad_norm": 6.149965286254883,
+      "grad_norm": 7.100608825683594,
       "learning_rate": 3.941941941941943e-06,
-      "loss": 0.9233,
+      "loss": 0.9175,
       "step": 60520
     },
     {
       "epoch": 181.77,
-      "grad_norm": 8.743706703186035,
+      "grad_norm": 7.9706010818481445,
       "learning_rate": 3.940940940940941e-06,
-      "loss": 0.9571,
+      "loss": 0.9476,
       "step": 60530
     },
     {
       "epoch": 181.8,
-      "grad_norm": 7.144441604614258,
+      "grad_norm": 8.865121841430664,
       "learning_rate": 3.93993993993994e-06,
-      "loss": 0.9255,
+      "loss": 0.9359,
       "step": 60540
     },
     {
       "epoch": 181.83,
-      "grad_norm": 4.92498254776001,
+      "grad_norm": 5.464580059051514,
       "learning_rate": 3.9389389389389395e-06,
-      "loss": 0.9208,
+      "loss": 0.9198,
       "step": 60550
     },
     {
       "epoch": 181.86,
-      "grad_norm": 7.704205513000488,
+      "grad_norm": 6.345850467681885,
       "learning_rate": 3.937937937937938e-06,
-      "loss": 0.9731,
+      "loss": 0.9594,
       "step": 60560
     },
     {
       "epoch": 181.89,
-      "grad_norm": 6.796319484710693,
+      "grad_norm": 5.988412857055664,
       "learning_rate": 3.936936936936937e-06,
-      "loss": 0.9704,
+      "loss": 0.9709,
       "step": 60570
     },
     {
       "epoch": 181.92,
-      "grad_norm": 7.594470500946045,
+      "grad_norm": 7.790589809417725,
       "learning_rate": 3.935935935935936e-06,
-      "loss": 0.995,
+      "loss": 0.9911,
       "step": 60580
     },
     {
       "epoch": 181.95,
-      "grad_norm": 4.445347785949707,
+      "grad_norm": 5.06175422668457,
       "learning_rate": 3.934934934934936e-06,
       "loss": 1.0071,
       "step": 60590
     },
     {
       "epoch": 181.98,
-      "grad_norm": 8.370203018188477,
+      "grad_norm": 10.781474113464355,
       "learning_rate": 3.933933933933934e-06,
-      "loss": 0.9783,
+      "loss": 0.9819,
       "step": 60600
     },
     {
       "epoch": 182.0,
-      "eval_accuracy": 0.8206,
-      "eval_loss": 0.583217203617096,
-      "eval_runtime": 5.3821,
-      "eval_samples_per_second": 1858.015,
-      "eval_steps_per_second": 7.432,
+      "eval_accuracy": 0.8218,
+      "eval_loss": 0.5839049220085144,
+      "eval_runtime": 5.5856,
+      "eval_samples_per_second": 1790.309,
+      "eval_steps_per_second": 7.161,
       "step": 60606
     },
     {
       "epoch": 182.01,
-      "grad_norm": 8.432976722717285,
+      "grad_norm": 7.188652515411377,
       "learning_rate": 3.932932932932933e-06,
-      "loss": 1.2639,
+      "loss": 1.2544,
       "step": 60610
     },
     {
       "epoch": 182.04,
-      "grad_norm": 5.845606327056885,
+      "grad_norm": 5.620754241943359,
       "learning_rate": 3.931931931931932e-06,
-      "loss": 1.038,
+      "loss": 1.0362,
       "step": 60620
     },
     {
       "epoch": 182.07,
-      "grad_norm": 7.878211975097656,
+      "grad_norm": 9.200063705444336,
       "learning_rate": 3.930930930930932e-06,
-      "loss": 0.9529,
+      "loss": 0.9565,
       "step": 60630
     },
     {
       "epoch": 182.1,
-      "grad_norm": 8.630556106567383,
+      "grad_norm": 5.82631254196167,
       "learning_rate": 3.92992992992993e-06,
-      "loss": 1.0117,
+      "loss": 1.0281,
       "step": 60640
     },
     {
       "epoch": 182.13,
-      "grad_norm": 6.0417962074279785,
+      "grad_norm": 6.977082252502441,
       "learning_rate": 3.92892892892893e-06,
-      "loss": 1.0318,
+      "loss": 1.0204,
       "step": 60650
     },
     {
       "epoch": 182.16,
-      "grad_norm": 6.140097141265869,
+      "grad_norm": 8.29778003692627,
       "learning_rate": 3.927927927927928e-06,
-      "loss": 0.9581,
+      "loss": 0.9565,
       "step": 60660
     },
     {
       "epoch": 182.19,
-      "grad_norm": 6.24277400970459,
+      "grad_norm": 5.993316173553467,
       "learning_rate": 3.926926926926927e-06,
-      "loss": 0.9124,
+      "loss": 0.9014,
       "step": 60670
     },
     {
       "epoch": 182.22,
-      "grad_norm": 7.934340000152588,
+      "grad_norm": 7.2459845542907715,
       "learning_rate": 3.925925925925926e-06,
-      "loss": 1.0289,
+      "loss": 1.0275,
       "step": 60680
     },
     {
       "epoch": 182.25,
-      "grad_norm": 6.270734786987305,
+      "grad_norm": 7.855305194854736,
       "learning_rate": 3.924924924924926e-06,
-      "loss": 0.919,
+      "loss": 0.9221,
       "step": 60690
     },
     {
       "epoch": 182.28,
-      "grad_norm": 8.420573234558105,
+      "grad_norm": 8.922479629516602,
       "learning_rate": 3.923923923923924e-06,
-      "loss": 0.9228,
+      "loss": 0.9345,
       "step": 60700
     },
     {
       "epoch": 182.31,
-      "grad_norm": 5.99972677230835,
+      "grad_norm": 6.362020969390869,
       "learning_rate": 3.922922922922923e-06,
-      "loss": 0.9466,
+      "loss": 0.9467,
       "step": 60710
     },
     {
       "epoch": 182.34,
-      "grad_norm": 6.027103424072266,
+      "grad_norm": 6.462217330932617,
       "learning_rate": 3.921921921921922e-06,
-      "loss": 0.9733,
+      "loss": 0.9816,
       "step": 60720
     },
     {
       "epoch": 182.37,
-      "grad_norm": 7.416637897491455,
+      "grad_norm": 5.6718854904174805,
       "learning_rate": 3.920920920920922e-06,
-      "loss": 0.8645,
+      "loss": 0.853,
       "step": 60730
     },
     {
       "epoch": 182.4,
-      "grad_norm": 4.842827796936035,
+      "grad_norm": 6.346751689910889,
       "learning_rate": 3.9199199199199204e-06,
-      "loss": 0.9379,
+      "loss": 0.9222,
       "step": 60740
     },
     {
       "epoch": 182.43,
-      "grad_norm": 10.11699390411377,
+      "grad_norm": 9.456245422363281,
       "learning_rate": 3.918918918918919e-06,
-      "loss": 1.0145,
+      "loss": 1.0118,
       "step": 60750
     },
     {
       "epoch": 182.46,
-      "grad_norm": 6.362307548522949,
+      "grad_norm": 8.04415225982666,
       "learning_rate": 3.9179179179179185e-06,
-      "loss": 1.0573,
+      "loss": 1.0479,
       "step": 60760
     },
     {
       "epoch": 182.49,
-      "grad_norm": 5.962704658508301,
+      "grad_norm": 6.516165256500244,
       "learning_rate": 3.916916916916917e-06,
-      "loss": 0.9169,
+      "loss": 0.9111,
       "step": 60770
     },
     {
       "epoch": 182.52,
-      "grad_norm": 4.629899978637695,
+      "grad_norm": 5.340964317321777,
       "learning_rate": 3.915915915915916e-06,
-      "loss": 0.8839,
+      "loss": 0.8931,
       "step": 60780
     },
     {
       "epoch": 182.55,
-      "grad_norm": 6.3740715980529785,
+      "grad_norm": 7.988334655761719,
       "learning_rate": 3.914914914914915e-06,
-      "loss": 0.982,
+      "loss": 0.988,
       "step": 60790
     },
     {
       "epoch": 182.58,
-      "grad_norm": 7.309112071990967,
+      "grad_norm": 8.554170608520508,
       "learning_rate": 3.9139139139139145e-06,
-      "loss": 0.9763,
+      "loss": 0.9744,
       "step": 60800
     },
     {
       "epoch": 182.61,
-      "grad_norm": 6.739979267120361,
+      "grad_norm": 8.665961265563965,
       "learning_rate": 3.912912912912913e-06,
-      "loss": 0.9676,
+      "loss": 0.9595,
       "step": 60810
     },
     {
       "epoch": 182.64,
-      "grad_norm": 6.464739799499512,
+      "grad_norm": 6.241861343383789,
       "learning_rate": 3.911911911911912e-06,
-      "loss": 0.9254,
+      "loss": 0.9189,
       "step": 60820
     },
     {
       "epoch": 182.67,
-      "grad_norm": 5.690426349639893,
+      "grad_norm": 8.330137252807617,
       "learning_rate": 3.910910910910911e-06,
-      "loss": 0.9939,
+      "loss": 0.9976,
       "step": 60830
     },
     {
       "epoch": 182.7,
-      "grad_norm": 6.886569976806641,
+      "grad_norm": 7.569728374481201,
       "learning_rate": 3.909909909909911e-06,
-      "loss": 1.0538,
+      "loss": 1.0304,
       "step": 60840
     },
     {
       "epoch": 182.73,
-      "grad_norm": 6.3680548667907715,
+      "grad_norm": 7.5609283447265625,
       "learning_rate": 3.908908908908909e-06,
-      "loss": 0.9212,
+      "loss": 0.9126,
       "step": 60850
     },
     {
       "epoch": 182.76,
-      "grad_norm": 8.910856246948242,
+      "grad_norm": 8.383936882019043,
       "learning_rate": 3.907907907907908e-06,
-      "loss": 0.9865,
+      "loss": 0.9756,
       "step": 60860
     },
     {
       "epoch": 182.79,
-      "grad_norm": 7.541000843048096,
+      "grad_norm": 6.28580904006958,
       "learning_rate": 3.906906906906907e-06,
-      "loss": 0.9526,
+      "loss": 0.947,
       "step": 60870
     },
     {
       "epoch": 182.82,
-      "grad_norm": 6.609584331512451,
+      "grad_norm": 11.383773803710938,
       "learning_rate": 3.905905905905906e-06,
-      "loss": 1.0514,
+      "loss": 1.0609,
       "step": 60880
     },
     {
       "epoch": 182.85,
-      "grad_norm": 4.92947244644165,
+      "grad_norm": 5.597952365875244,
       "learning_rate": 3.904904904904905e-06,
-      "loss": 0.9119,
+      "loss": 0.9101,
       "step": 60890
     },
     {
       "epoch": 182.88,
-      "grad_norm": 5.938350200653076,
+      "grad_norm": 6.4952521324157715,
       "learning_rate": 3.903903903903904e-06,
-      "loss": 0.9798,
+      "loss": 0.9829,
       "step": 60900
     },
     {
       "epoch": 182.91,
-      "grad_norm": 6.836793899536133,
+      "grad_norm": 5.805178165435791,
       "learning_rate": 3.902902902902903e-06,
-      "loss": 0.9906,
+      "loss": 0.9842,
       "step": 60910
     },
     {
       "epoch": 182.94,
-      "grad_norm": 7.324462890625,
+      "grad_norm": 6.830507755279541,
       "learning_rate": 3.901901901901902e-06,
-      "loss": 1.0143,
+      "loss": 1.0095,
       "step": 60920
     },
     {
       "epoch": 182.97,
-      "grad_norm": 6.263777732849121,
+      "grad_norm": 5.900455474853516,
       "learning_rate": 3.900900900900901e-06,
-      "loss": 0.9147,
+      "loss": 0.9231,
       "step": 60930
     },
     {
       "epoch": 183.0,
-      "eval_accuracy": 0.8223,
-      "eval_loss": 0.588080644607544,
-      "eval_runtime": 5.4994,
-      "eval_samples_per_second": 1818.379,
-      "eval_steps_per_second": 7.274,
+      "eval_accuracy": 0.8235,
+      "eval_loss": 0.5891535878181458,
+      "eval_runtime": 5.4251,
+      "eval_samples_per_second": 1843.272,
+      "eval_steps_per_second": 7.373,
       "step": 60939
     },
     {
       "epoch": 183.0,
-      "grad_norm": 7.4410271644592285,
+      "grad_norm": 6.5529632568359375,
       "learning_rate": 3.899899899899901e-06,
-      "loss": 1.1718,
+      "loss": 1.1269,
       "step": 60940
     },
     {
       "epoch": 183.03,
-      "grad_norm": 5.923596382141113,
+      "grad_norm": 6.328668594360352,
       "learning_rate": 3.898898898898899e-06,
-      "loss": 0.9005,
+      "loss": 0.8991,
       "step": 60950
     },
     {
       "epoch": 183.06,
-      "grad_norm": 6.131540775299072,
+      "grad_norm": 7.909534454345703,
       "learning_rate": 3.897897897897898e-06,
-      "loss": 0.9497,
+      "loss": 0.9452,
       "step": 60960
     },
     {
       "epoch": 183.09,
-      "grad_norm": 8.016525268554688,
+      "grad_norm": 7.323083400726318,
       "learning_rate": 3.8968968968968974e-06,
-      "loss": 0.9647,
+      "loss": 0.9669,
       "step": 60970
     },
     {
       "epoch": 183.12,
-      "grad_norm": 5.136233329772949,
+      "grad_norm": 7.100643634796143,
       "learning_rate": 3.895895895895896e-06,
-      "loss": 0.9075,
+      "loss": 0.9111,
       "step": 60980
     },
     {
       "epoch": 183.15,
-      "grad_norm": 6.226183891296387,
+      "grad_norm": 6.331968784332275,
       "learning_rate": 3.894894894894895e-06,
-      "loss": 0.9705,
+      "loss": 0.9679,
       "step": 60990
     },
     {
       "epoch": 183.18,
-      "grad_norm": 5.521854877471924,
+      "grad_norm": 8.621289253234863,
       "learning_rate": 3.893893893893894e-06,
-      "loss": 0.9821,
+      "loss": 0.9786,
       "step": 61000
     },
     {
       "epoch": 183.21,
-      "grad_norm": 7.994105815887451,
+      "grad_norm": 8.15859603881836,
       "learning_rate": 3.8928928928928935e-06,
-      "loss": 0.9818,
+      "loss": 0.9765,
       "step": 61010
     },
     {
       "epoch": 183.24,
-      "grad_norm": 7.634850978851318,
+      "grad_norm": 9.659664154052734,
       "learning_rate": 3.891891891891892e-06,
-      "loss": 0.9933,
+      "loss": 0.9926,
       "step": 61020
     },
     {
       "epoch": 183.27,
-      "grad_norm": 6.194941997528076,
+      "grad_norm": 6.52155065536499,
       "learning_rate": 3.890890890890891e-06,
-      "loss": 1.0308,
+      "loss": 1.013,
       "step": 61030
     },
     {
       "epoch": 183.3,
-      "grad_norm": 8.381722450256348,
+      "grad_norm": 8.440603256225586,
       "learning_rate": 3.88988988988989e-06,
-      "loss": 0.9244,
+      "loss": 0.9262,
       "step": 61040
     },
     {
       "epoch": 183.33,
-      "grad_norm": 6.031935691833496,
+      "grad_norm": 6.169778823852539,
       "learning_rate": 3.88888888888889e-06,
-      "loss": 1.0026,
+      "loss": 0.9925,
       "step": 61050
     },
     {
       "epoch": 183.36,
-      "grad_norm": 8.47339916229248,
+      "grad_norm": 6.70745849609375,
       "learning_rate": 3.887887887887888e-06,
-      "loss": 0.951,
+      "loss": 0.9566,
       "step": 61060
     },
     {
       "epoch": 183.39,
-      "grad_norm": 6.946739196777344,
+      "grad_norm": 9.573139190673828,
       "learning_rate": 3.886886886886887e-06,
-      "loss": 0.9275,
+      "loss": 0.9426,
       "step": 61070
     },
     {
       "epoch": 183.42,
-      "grad_norm": 6.622891902923584,
+      "grad_norm": 7.14495325088501,
       "learning_rate": 3.885885885885886e-06,
-      "loss": 0.9653,
+      "loss": 0.9577,
       "step": 61080
     },
     {
       "epoch": 183.45,
-      "grad_norm": 7.062820911407471,
+      "grad_norm": 7.680576801300049,
       "learning_rate": 3.884884884884886e-06,
-      "loss": 0.9265,
+      "loss": 0.9331,
       "step": 61090
     },
     {
       "epoch": 183.48,
-      "grad_norm": 5.957156658172607,
+      "grad_norm": 7.882317543029785,
       "learning_rate": 3.883883883883884e-06,
-      "loss": 1.0042,
+      "loss": 0.9958,
       "step": 61100
     },
     {
       "epoch": 183.51,
-      "grad_norm": 5.884033203125,
+      "grad_norm": 8.288481712341309,
       "learning_rate": 3.882882882882883e-06,
-      "loss": 0.9247,
+      "loss": 0.9265,
       "step": 61110
     },
     {
       "epoch": 183.54,
-      "grad_norm": 6.728677272796631,
+      "grad_norm": 8.068124771118164,
       "learning_rate": 3.881881881881882e-06,
-      "loss": 1.0507,
+      "loss": 1.041,
       "step": 61120
     },
     {
       "epoch": 183.57,
-      "grad_norm": 5.224053859710693,
+      "grad_norm": 5.95291805267334,
       "learning_rate": 3.880880880880881e-06,
-      "loss": 0.9428,
+      "loss": 0.9403,
       "step": 61130
     },
     {
       "epoch": 183.6,
-      "grad_norm": 5.250800609588623,
+      "grad_norm": 5.185598373413086,
       "learning_rate": 3.87987987987988e-06,
-      "loss": 0.8865,
+      "loss": 0.8862,
       "step": 61140
     },
     {
       "epoch": 183.63,
-      "grad_norm": 6.415823936462402,
+      "grad_norm": 7.002123832702637,
       "learning_rate": 3.878878878878879e-06,
-      "loss": 0.9529,
+      "loss": 0.9492,
       "step": 61150
     },
     {
       "epoch": 183.66,
-      "grad_norm": 6.064749240875244,
+      "grad_norm": 7.493641376495361,
       "learning_rate": 3.877877877877878e-06,
-      "loss": 0.9808,
+      "loss": 0.9865,
       "step": 61160
     },
     {
       "epoch": 183.69,
-      "grad_norm": 6.312156677246094,
+      "grad_norm": 8.015585899353027,
       "learning_rate": 3.876876876876877e-06,
-      "loss": 0.9003,
+      "loss": 0.8998,
       "step": 61170
     },
     {
       "epoch": 183.72,
-      "grad_norm": 7.960103988647461,
+      "grad_norm": 10.457655906677246,
       "learning_rate": 3.875875875875876e-06,
-      "loss": 0.9943,
+      "loss": 0.9913,
       "step": 61180
     },
     {
       "epoch": 183.75,
-      "grad_norm": 9.14918327331543,
+      "grad_norm": 9.002128601074219,
       "learning_rate": 3.874874874874876e-06,
-      "loss": 0.9502,
+      "loss": 0.9519,
       "step": 61190
     },
     {
       "epoch": 183.78,
-      "grad_norm": 5.505520820617676,
+      "grad_norm": 8.003687858581543,
       "learning_rate": 3.8738738738738744e-06,
-      "loss": 0.9985,
+      "loss": 0.9841,
       "step": 61200
     },
     {
       "epoch": 183.81,
-      "grad_norm": 7.873384475708008,
+      "grad_norm": 6.3261237144470215,
       "learning_rate": 3.872872872872873e-06,
-      "loss": 1.006,
+      "loss": 1.0103,
       "step": 61210
     },
     {
       "epoch": 183.84,
-      "grad_norm": 8.069258689880371,
+      "grad_norm": 10.753619194030762,
       "learning_rate": 3.8718718718718725e-06,
-      "loss": 0.9894,
+      "loss": 0.9891,
       "step": 61220
     },
     {
       "epoch": 183.87,
-      "grad_norm": 5.952986717224121,
+      "grad_norm": 7.977743625640869,
       "learning_rate": 3.870870870870871e-06,
-      "loss": 0.9647,
+      "loss": 0.963,
       "step": 61230
     },
     {
       "epoch": 183.9,
-      "grad_norm": 6.643319129943848,
+      "grad_norm": 4.642685890197754,
       "learning_rate": 3.86986986986987e-06,
-      "loss": 0.9173,
+      "loss": 0.9253,
       "step": 61240
     },
     {
       "epoch": 183.93,
-      "grad_norm": 6.923768043518066,
+      "grad_norm": 7.085824489593506,
       "learning_rate": 3.868868868868869e-06,
-      "loss": 0.9417,
+      "loss": 0.9364,
       "step": 61250
     },
     {
       "epoch": 183.96,
-      "grad_norm": 11.285629272460938,
+      "grad_norm": 12.570465087890625,
       "learning_rate": 3.8678678678678685e-06,
-      "loss": 1.0807,
+      "loss": 1.0744,
       "step": 61260
     },
     {
       "epoch": 183.99,
-      "grad_norm": 6.904598712921143,
+      "grad_norm": 7.777284622192383,
       "learning_rate": 3.866866866866867e-06,
-      "loss": 0.9848,
+      "loss": 0.9765,
       "step": 61270
     },
     {
       "epoch": 184.0,
-      "eval_accuracy": 0.8218,
-      "eval_loss": 0.5897899270057678,
-      "eval_runtime": 5.2898,
-      "eval_samples_per_second": 1890.418,
-      "eval_steps_per_second": 7.562,
+      "eval_accuracy": 0.8223,
+      "eval_loss": 0.5878111720085144,
+      "eval_runtime": 5.4971,
+      "eval_samples_per_second": 1819.136,
+      "eval_steps_per_second": 7.277,
       "step": 61272
     },
     {
       "epoch": 184.02,
-      "grad_norm": 4.989622116088867,
+      "grad_norm": 8.539281845092773,
       "learning_rate": 3.865865865865866e-06,
-      "loss": 1.2282,
+      "loss": 1.2006,
       "step": 61280
     },
     {
       "epoch": 184.05,
-      "grad_norm": 6.262873649597168,
+      "grad_norm": 8.802392959594727,
       "learning_rate": 3.864864864864865e-06,
-      "loss": 0.9908,
+      "loss": 1.0013,
       "step": 61290
     },
     {
       "epoch": 184.08,
-      "grad_norm": 5.8469438552856445,
+      "grad_norm": 5.434993743896484,
       "learning_rate": 3.863863863863865e-06,
-      "loss": 0.9035,
+      "loss": 0.9047,
       "step": 61300
     },
     {
       "epoch": 184.11,
-      "grad_norm": 8.519216537475586,
+      "grad_norm": 6.836068630218506,
       "learning_rate": 3.862862862862863e-06,
-      "loss": 0.9438,
+      "loss": 0.9462,
       "step": 61310
     },
     {
       "epoch": 184.14,
-      "grad_norm": 6.590601921081543,
+      "grad_norm": 7.65800142288208,
       "learning_rate": 3.861861861861862e-06,
-      "loss": 0.9693,
+      "loss": 0.9716,
       "step": 61320
     },
     {
       "epoch": 184.17,
-      "grad_norm": 4.944343090057373,
+      "grad_norm": 9.440774917602539,
       "learning_rate": 3.860860860860861e-06,
-      "loss": 0.9514,
+      "loss": 0.9375,
       "step": 61330
     },
     {
       "epoch": 184.2,
-      "grad_norm": 6.408355712890625,
+      "grad_norm": 6.719813823699951,
       "learning_rate": 3.85985985985986e-06,
-      "loss": 0.9667,
+      "loss": 0.9626,
       "step": 61340
     },
     {
       "epoch": 184.23,
-      "grad_norm": 5.170409202575684,
+      "grad_norm": 5.569274425506592,
       "learning_rate": 3.858858858858859e-06,
-      "loss": 1.0269,
+      "loss": 1.0341,
       "step": 61350
     },
     {
       "epoch": 184.26,
-      "grad_norm": 6.0051655769348145,
+      "grad_norm": 6.670773029327393,
       "learning_rate": 3.857857857857858e-06,
-      "loss": 0.9328,
+      "loss": 0.9345,
       "step": 61360
     },
     {
       "epoch": 184.29,
-      "grad_norm": 7.92279577255249,
+      "grad_norm": 8.643997192382812,
       "learning_rate": 3.856856856856857e-06,
-      "loss": 0.9015,
+      "loss": 0.8974,
       "step": 61370
     },
     {
       "epoch": 184.32,
-      "grad_norm": 5.878408908843994,
+      "grad_norm": 6.89208984375,
       "learning_rate": 3.855855855855856e-06,
-      "loss": 1.0301,
+      "loss": 1.0241,
       "step": 61380
     },
     {
       "epoch": 184.35,
-      "grad_norm": 6.435563564300537,
+      "grad_norm": 5.148653507232666,
       "learning_rate": 3.854854854854855e-06,
-      "loss": 0.9404,
+      "loss": 0.9317,
       "step": 61390
     },
     {
       "epoch": 184.38,
-      "grad_norm": 6.550014495849609,
+      "grad_norm": 4.985918045043945,
       "learning_rate": 3.853853853853854e-06,
-      "loss": 0.9057,
+      "loss": 0.9009,
       "step": 61400
     },
     {
       "epoch": 184.41,
-      "grad_norm": 8.188610076904297,
+      "grad_norm": 7.299058437347412,
       "learning_rate": 3.852852852852853e-06,
-      "loss": 1.0121,
+      "loss": 1.0055,
       "step": 61410
     },
     {
       "epoch": 184.44,
-      "grad_norm": 6.128997325897217,
+      "grad_norm": 6.074924468994141,
       "learning_rate": 3.851851851851852e-06,
-      "loss": 0.9271,
+      "loss": 0.9285,
       "step": 61420
     },
     {
       "epoch": 184.47,
-      "grad_norm": 5.929874897003174,
+      "grad_norm": 5.418395042419434,
       "learning_rate": 3.8508508508508514e-06,
-      "loss": 0.9375,
+      "loss": 0.9218,
       "step": 61430
     },
     {
       "epoch": 184.5,
-      "grad_norm": 5.920310974121094,
+      "grad_norm": 5.829029560089111,
       "learning_rate": 3.84984984984985e-06,
-      "loss": 0.9639,
+      "loss": 0.968,
       "step": 61440
     },
     {
       "epoch": 184.53,
-      "grad_norm": 6.0455002784729,
+      "grad_norm": 8.170210838317871,
       "learning_rate": 3.848848848848849e-06,
-      "loss": 1.0197,
+      "loss": 1.023,
       "step": 61450
     },
     {
       "epoch": 184.56,
-      "grad_norm": 7.094353199005127,
+      "grad_norm": 7.166289806365967,
       "learning_rate": 3.847847847847848e-06,
-      "loss": 0.9573,
+      "loss": 0.9493,
       "step": 61460
     },
     {
       "epoch": 184.59,
-      "grad_norm": 5.673529148101807,
+      "grad_norm": 8.246244430541992,
       "learning_rate": 3.8468468468468475e-06,
-      "loss": 0.9743,
+      "loss": 0.9689,
       "step": 61470
     },
     {
       "epoch": 184.62,
-      "grad_norm": 4.501720905303955,
+      "grad_norm": 5.54439115524292,
       "learning_rate": 3.845845845845846e-06,
-      "loss": 0.9204,
+      "loss": 0.9237,
       "step": 61480
     },
     {
       "epoch": 184.65,
-      "grad_norm": 5.358657360076904,
+      "grad_norm": 8.671210289001465,
       "learning_rate": 3.844844844844845e-06,
-      "loss": 0.9759,
+      "loss": 0.9793,
       "step": 61490
     },
     {
       "epoch": 184.68,
-      "grad_norm": 6.737173080444336,
+      "grad_norm": 7.122357368469238,
       "learning_rate": 3.843843843843844e-06,
-      "loss": 0.948,
+      "loss": 0.9461,
       "step": 61500
     },
     {
       "epoch": 184.71,
-      "grad_norm": 8.857156753540039,
+      "grad_norm": 7.470210552215576,
       "learning_rate": 3.842842842842844e-06,
-      "loss": 0.9569,
+      "loss": 0.9568,
       "step": 61510
     },
     {
       "epoch": 184.74,
-      "grad_norm": 8.097476959228516,
+      "grad_norm": 7.655945301055908,
       "learning_rate": 3.841841841841842e-06,
-      "loss": 1.0048,
+      "loss": 1.0025,
       "step": 61520
     },
     {
       "epoch": 184.77,
-      "grad_norm": 7.103294849395752,
+      "grad_norm": 9.131484985351562,
       "learning_rate": 3.840840840840841e-06,
-      "loss": 0.9675,
+      "loss": 0.9628,
       "step": 61530
     },
     {
       "epoch": 184.8,
-      "grad_norm": 8.827051162719727,
+      "grad_norm": 7.826813697814941,
       "learning_rate": 3.83983983983984e-06,
-      "loss": 0.8933,
+      "loss": 0.892,
       "step": 61540
     },
     {
       "epoch": 184.83,
-      "grad_norm": 4.671200752258301,
+      "grad_norm": 6.599262237548828,
       "learning_rate": 3.83883883883884e-06,
-      "loss": 0.9708,
+      "loss": 0.9674,
       "step": 61550
     },
     {
       "epoch": 184.86,
-      "grad_norm": 9.256632804870605,
+      "grad_norm": 6.696078300476074,
       "learning_rate": 3.837837837837838e-06,
-      "loss": 0.9717,
+      "loss": 0.9611,
       "step": 61560
     },
     {
       "epoch": 184.89,
-      "grad_norm": 7.6630144119262695,
+      "grad_norm": 6.566277503967285,
       "learning_rate": 3.836836836836837e-06,
-      "loss": 0.9308,
+      "loss": 0.9267,
       "step": 61570
     },
     {
       "epoch": 184.92,
-      "grad_norm": 6.010096549987793,
+      "grad_norm": 5.1474833488464355,
       "learning_rate": 3.835835835835836e-06,
-      "loss": 0.9566,
+      "loss": 0.9595,
       "step": 61580
     },
     {
       "epoch": 184.95,
-      "grad_norm": 5.938097953796387,
+      "grad_norm": 6.122076034545898,
       "learning_rate": 3.834834834834835e-06,
-      "loss": 0.9115,
+      "loss": 0.9145,
       "step": 61590
     },
     {
       "epoch": 184.98,
-      "grad_norm": 7.416814804077148,
+      "grad_norm": 8.287005424499512,
       "learning_rate": 3.833833833833834e-06,
-      "loss": 0.9454,
+      "loss": 0.9373,
       "step": 61600
     },
     {
       "epoch": 185.0,
-      "eval_accuracy": 0.8201,
-      "eval_loss": 0.5916330218315125,
-      "eval_runtime": 5.4905,
-      "eval_samples_per_second": 1821.322,
-      "eval_steps_per_second": 7.285,
+      "eval_accuracy": 0.8218,
+      "eval_loss": 0.5921128988265991,
+      "eval_runtime": 5.3591,
+      "eval_samples_per_second": 1865.987,
+      "eval_steps_per_second": 7.464,
       "step": 61605
     },
     {
       "epoch": 185.02,
-      "grad_norm": 5.840172290802002,
+      "grad_norm": 5.260549545288086,
       "learning_rate": 3.832832832832833e-06,
-      "loss": 1.2105,
+      "loss": 1.2279,
       "step": 61610
     },
     {
       "epoch": 185.05,
-      "grad_norm": 6.112273216247559,
+      "grad_norm": 6.280886650085449,
       "learning_rate": 3.831831831831832e-06,
-      "loss": 0.9712,
+      "loss": 0.965,
       "step": 61620
     },
     {
       "epoch": 185.08,
-      "grad_norm": 4.9560723304748535,
+      "grad_norm": 6.004842758178711,
       "learning_rate": 3.830830830830831e-06,
-      "loss": 0.998,
+      "loss": 0.9979,
       "step": 61630
     },
     {
       "epoch": 185.11,
-      "grad_norm": 7.557987213134766,
+      "grad_norm": 7.785391807556152,
       "learning_rate": 3.82982982982983e-06,
-      "loss": 0.9053,
+      "loss": 0.9132,
       "step": 61640
     },
     {
       "epoch": 185.14,
-      "grad_norm": 6.805177688598633,
+      "grad_norm": 7.029149532318115,
       "learning_rate": 3.828828828828829e-06,
-      "loss": 1.0017,
+      "loss": 0.9865,
       "step": 61650
     },
     {
       "epoch": 185.17,
-      "grad_norm": 9.28736686706543,
+      "grad_norm": 5.686846733093262,
       "learning_rate": 3.8278278278278284e-06,
-      "loss": 0.9296,
+      "loss": 0.9254,
       "step": 61660
     },
     {
       "epoch": 185.2,
-      "grad_norm": 6.5211358070373535,
+      "grad_norm": 5.298410892486572,
       "learning_rate": 3.826826826826827e-06,
-      "loss": 0.8687,
+      "loss": 0.8668,
       "step": 61670
     },
     {
       "epoch": 185.23,
-      "grad_norm": 7.9119720458984375,
+      "grad_norm": 8.215540885925293,
       "learning_rate": 3.8258258258258265e-06,
-      "loss": 0.9764,
+      "loss": 0.9652,
       "step": 61680
     },
     {
       "epoch": 185.26,
-      "grad_norm": 6.610804080963135,
+      "grad_norm": 6.389473915100098,
       "learning_rate": 3.824824824824825e-06,
-      "loss": 0.9683,
+      "loss": 0.9739,
       "step": 61690
     },
     {
       "epoch": 185.29,
-      "grad_norm": 6.886896133422852,
+      "grad_norm": 5.7342729568481445,
       "learning_rate": 3.823823823823824e-06,
-      "loss": 1.0138,
+      "loss": 1.0085,
       "step": 61700
     },
     {
       "epoch": 185.32,
-      "grad_norm": 8.163553237915039,
+      "grad_norm": 7.2613911628723145,
       "learning_rate": 3.822822822822823e-06,
-      "loss": 0.9243,
+      "loss": 0.9353,
       "step": 61710
     },
     {
       "epoch": 185.35,
-      "grad_norm": 7.83153772354126,
+      "grad_norm": 6.126305103302002,
       "learning_rate": 3.8218218218218225e-06,
-      "loss": 0.9659,
+      "loss": 0.9633,
       "step": 61720
     },
     {
       "epoch": 185.38,
-      "grad_norm": 6.777648448944092,
+      "grad_norm": 6.606804847717285,
       "learning_rate": 3.820820820820821e-06,
-      "loss": 0.9958,
+      "loss": 0.9894,
       "step": 61730
     },
     {
       "epoch": 185.41,
-      "grad_norm": 4.946415424346924,
+      "grad_norm": 7.502766132354736,
       "learning_rate": 3.81981981981982e-06,
-      "loss": 1.0112,
+      "loss": 1.0136,
       "step": 61740
     },
     {
       "epoch": 185.44,
-      "grad_norm": 6.207168102264404,
+      "grad_norm": 6.544745922088623,
       "learning_rate": 3.818818818818819e-06,
-      "loss": 0.9362,
+      "loss": 0.9367,
       "step": 61750
     },
     {
       "epoch": 185.47,
-      "grad_norm": 8.020191192626953,
+      "grad_norm": 6.3826704025268555,
       "learning_rate": 3.817817817817819e-06,
-      "loss": 0.9469,
+      "loss": 0.9489,
       "step": 61760
     },
     {
       "epoch": 185.5,
-      "grad_norm": 5.5348944664001465,
+      "grad_norm": 7.375263690948486,
       "learning_rate": 3.816816816816817e-06,
-      "loss": 0.9367,
+      "loss": 0.9316,
       "step": 61770
     },
     {
       "epoch": 185.53,
-      "grad_norm": 8.248254776000977,
+      "grad_norm": 7.362241744995117,
       "learning_rate": 3.815815815815816e-06,
       "loss": 0.9684,
       "step": 61780
     },
     {
       "epoch": 185.56,
-      "grad_norm": 7.02725076675415,
+      "grad_norm": 6.607955455780029,
       "learning_rate": 3.814814814814815e-06,
-      "loss": 0.9951,
+      "loss": 1.0094,
       "step": 61790
     },
     {
       "epoch": 185.59,
-      "grad_norm": 7.391062259674072,
+      "grad_norm": 8.526917457580566,
       "learning_rate": 3.8138138138138143e-06,
-      "loss": 0.9607,
+      "loss": 0.9512,
       "step": 61800
     },
     {
       "epoch": 185.62,
-      "grad_norm": 6.324107646942139,
+      "grad_norm": 5.790355682373047,
       "learning_rate": 3.8128128128128133e-06,
-      "loss": 0.8895,
+      "loss": 0.8761,
       "step": 61810
     },
     {
       "epoch": 185.65,
-      "grad_norm": 7.292401313781738,
+      "grad_norm": 8.17216968536377,
       "learning_rate": 3.811811811811812e-06,
-      "loss": 0.9787,
+      "loss": 0.9817,
       "step": 61820
     },
     {
       "epoch": 185.68,
-      "grad_norm": 7.630838394165039,
+      "grad_norm": 8.634063720703125,
       "learning_rate": 3.810810810810811e-06,
-      "loss": 0.8349,
+      "loss": 0.8304,
       "step": 61830
     },
     {
       "epoch": 185.71,
-      "grad_norm": 7.626615524291992,
+      "grad_norm": 7.3106465339660645,
       "learning_rate": 3.8098098098098103e-06,
-      "loss": 1.0398,
+      "loss": 1.0299,
       "step": 61840
     },
     {
       "epoch": 185.74,
-      "grad_norm": 5.808999061584473,
+      "grad_norm": 7.878754138946533,
       "learning_rate": 3.8088088088088094e-06,
-      "loss": 0.955,
+      "loss": 0.9584,
       "step": 61850
     },
     {
       "epoch": 185.77,
-      "grad_norm": 4.384845733642578,
+      "grad_norm": 5.131159782409668,
       "learning_rate": 3.807807807807808e-06,
-      "loss": 1.0179,
+      "loss": 1.0169,
       "step": 61860
     },
     {
       "epoch": 185.8,
-      "grad_norm": 5.047191143035889,
+      "grad_norm": 5.177952766418457,
       "learning_rate": 3.806806806806807e-06,
-      "loss": 0.9063,
+      "loss": 0.901,
       "step": 61870
     },
     {
       "epoch": 185.83,
-      "grad_norm": 6.130573272705078,
+      "grad_norm": 7.795198440551758,
       "learning_rate": 3.805805805805806e-06,
-      "loss": 0.9824,
+      "loss": 0.978,
       "step": 61880
     },
     {
       "epoch": 185.86,
-      "grad_norm": 4.851668357849121,
+      "grad_norm": 6.895273685455322,
       "learning_rate": 3.8048048048048054e-06,
-      "loss": 0.9652,
+      "loss": 0.9583,
       "step": 61890
     },
     {
       "epoch": 185.89,
-      "grad_norm": 7.233530521392822,
+      "grad_norm": 7.8442606925964355,
       "learning_rate": 3.803803803803804e-06,
-      "loss": 0.9976,
+      "loss": 1.0027,
       "step": 61900
     },
     {
       "epoch": 185.92,
-      "grad_norm": 4.8957037925720215,
+      "grad_norm": 8.109230995178223,
       "learning_rate": 3.802802802802803e-06,
-      "loss": 0.8965,
+      "loss": 0.9032,
       "step": 61910
     },
     {
       "epoch": 185.95,
-      "grad_norm": 5.63247537612915,
+      "grad_norm": 7.606938362121582,
       "learning_rate": 3.801801801801802e-06,
-      "loss": 0.9584,
+      "loss": 0.9614,
       "step": 61920
     },
     {
       "epoch": 185.98,
-      "grad_norm": 8.980535507202148,
+      "grad_norm": 15.001842498779297,
       "learning_rate": 3.800800800800801e-06,
-      "loss": 1.0287,
+      "loss": 1.0182,
       "step": 61930
     },
     {
       "epoch": 186.0,
-      "eval_accuracy": 0.8222,
-      "eval_loss": 0.5880201458930969,
-      "eval_runtime": 5.317,
-      "eval_samples_per_second": 1880.777,
-      "eval_steps_per_second": 7.523,
+      "eval_accuracy": 0.8228,
+      "eval_loss": 0.586543083190918,
+      "eval_runtime": 5.3281,
+      "eval_samples_per_second": 1876.857,
+      "eval_steps_per_second": 7.507,
       "step": 61938
     },
     {
       "epoch": 186.01,
-      "grad_norm": 5.78520393371582,
+      "grad_norm": 6.357463359832764,
       "learning_rate": 3.7997997997997997e-06,
-      "loss": 1.1191,
+      "loss": 1.0937,
       "step": 61940
     },
     {
       "epoch": 186.04,
-      "grad_norm": 5.568073749542236,
+      "grad_norm": 6.098045825958252,
       "learning_rate": 3.798798798798799e-06,
-      "loss": 0.9911,
+      "loss": 0.9928,
       "step": 61950
     },
     {
       "epoch": 186.07,
-      "grad_norm": 8.445097923278809,
+      "grad_norm": 7.982432842254639,
       "learning_rate": 3.797797797797798e-06,
-      "loss": 1.0214,
+      "loss": 1.0394,
       "step": 61960
     },
     {
       "epoch": 186.1,
-      "grad_norm": 6.563830375671387,
+      "grad_norm": 6.6531500816345215,
       "learning_rate": 3.796796796796797e-06,
-      "loss": 0.9828,
+      "loss": 0.9741,
       "step": 61970
     },
     {
       "epoch": 186.13,
-      "grad_norm": 7.2098822593688965,
+      "grad_norm": 7.222927093505859,
       "learning_rate": 3.795795795795796e-06,
-      "loss": 0.9352,
+      "loss": 0.937,
       "step": 61980
     },
     {
       "epoch": 186.16,
-      "grad_norm": 6.373877048492432,
+      "grad_norm": 6.259188175201416,
       "learning_rate": 3.7947947947947948e-06,
-      "loss": 0.9818,
+      "loss": 0.9762,
       "step": 61990
     },
     {
       "epoch": 186.19,
-      "grad_norm": 4.2350358963012695,
+      "grad_norm": 5.471319198608398,
       "learning_rate": 3.793793793793794e-06,
-      "loss": 0.9416,
+      "loss": 0.9425,
       "step": 62000
     },
     {
       "epoch": 186.22,
-      "grad_norm": 6.495857238769531,
+      "grad_norm": 5.929308891296387,
       "learning_rate": 3.7927927927927932e-06,
-      "loss": 0.9904,
+      "loss": 0.9835,
       "step": 62010
     },
     {
       "epoch": 186.25,
-      "grad_norm": 8.149953842163086,
+      "grad_norm": 8.413453102111816,
       "learning_rate": 3.7917917917917922e-06,
-      "loss": 0.9754,
+      "loss": 0.9834,
       "step": 62020
     },
     {
       "epoch": 186.28,
-      "grad_norm": 9.182705879211426,
+      "grad_norm": 6.640708923339844,
       "learning_rate": 3.790790790790791e-06,
-      "loss": 0.9901,
+      "loss": 0.9863,
       "step": 62030
     },
     {
       "epoch": 186.31,
-      "grad_norm": 6.538585186004639,
+      "grad_norm": 7.785292625427246,
       "learning_rate": 3.78978978978979e-06,
-      "loss": 1.0121,
+      "loss": 1.0151,
       "step": 62040
     },
     {
       "epoch": 186.34,
-      "grad_norm": 6.225589752197266,
+      "grad_norm": 10.576643943786621,
       "learning_rate": 3.7887887887887893e-06,
-      "loss": 0.9299,
+      "loss": 0.9326,
       "step": 62050
     },
     {
       "epoch": 186.37,
-      "grad_norm": 6.333828449249268,
+      "grad_norm": 6.046290397644043,
       "learning_rate": 3.7877877877877883e-06,
-      "loss": 0.9413,
+      "loss": 0.9438,
       "step": 62060
     },
     {
       "epoch": 186.4,
-      "grad_norm": 5.7685418128967285,
+      "grad_norm": 6.840560436248779,
       "learning_rate": 3.786786786786787e-06,
-      "loss": 0.9554,
+      "loss": 0.9501,
       "step": 62070
     },
     {
       "epoch": 186.43,
-      "grad_norm": 7.78047513961792,
+      "grad_norm": 4.883410453796387,
       "learning_rate": 3.785785785785786e-06,
-      "loss": 0.9142,
+      "loss": 0.9163,
       "step": 62080
     },
     {
       "epoch": 186.46,
-      "grad_norm": 9.091288566589355,
+      "grad_norm": 8.272229194641113,
       "learning_rate": 3.7847847847847854e-06,
-      "loss": 0.9954,
+      "loss": 1.0043,
       "step": 62090
     },
     {
       "epoch": 186.49,
-      "grad_norm": 6.681239128112793,
+      "grad_norm": 4.517034530639648,
       "learning_rate": 3.7837837837837844e-06,
-      "loss": 0.965,
+      "loss": 0.9699,
       "step": 62100
     },
     {
       "epoch": 186.52,
-      "grad_norm": 6.873343467712402,
+      "grad_norm": 6.151267051696777,
       "learning_rate": 3.782782782782783e-06,
-      "loss": 0.9575,
+      "loss": 0.9516,
       "step": 62110
     },
     {
       "epoch": 186.55,
-      "grad_norm": 6.787483215332031,
+      "grad_norm": 6.367547988891602,
       "learning_rate": 3.781781781781782e-06,
-      "loss": 0.9333,
+      "loss": 0.9301,
       "step": 62120
     },
     {
       "epoch": 186.58,
-      "grad_norm": 8.72840404510498,
+      "grad_norm": 11.630304336547852,
       "learning_rate": 3.780780780780781e-06,
-      "loss": 0.9562,
+      "loss": 0.9497,
       "step": 62130
     },
     {
       "epoch": 186.61,
-      "grad_norm": 5.855907917022705,
+      "grad_norm": 7.148751735687256,
       "learning_rate": 3.7797797797797805e-06,
-      "loss": 0.9632,
+      "loss": 0.9603,
       "step": 62140
     },
     {
       "epoch": 186.64,
-      "grad_norm": 6.44469690322876,
+      "grad_norm": 6.12123441696167,
       "learning_rate": 3.778778778778779e-06,
-      "loss": 0.9461,
+      "loss": 0.9517,
       "step": 62150
     },
     {
       "epoch": 186.67,
-      "grad_norm": 8.560931205749512,
+      "grad_norm": 10.263585090637207,
       "learning_rate": 3.777777777777778e-06,
-      "loss": 0.993,
+      "loss": 0.9944,
       "step": 62160
     },
     {
       "epoch": 186.7,
-      "grad_norm": 5.043516635894775,
+      "grad_norm": 6.450360298156738,
       "learning_rate": 3.776776776776777e-06,
-      "loss": 0.9051,
+      "loss": 0.8965,
       "step": 62170
     },
     {
       "epoch": 186.73,
-      "grad_norm": 9.655406951904297,
+      "grad_norm": 8.5963716506958,
       "learning_rate": 3.775775775775776e-06,
-      "loss": 1.0066,
+      "loss": 1.0118,
       "step": 62180
     },
     {
       "epoch": 186.76,
-      "grad_norm": 7.613271236419678,
+      "grad_norm": 8.921076774597168,
       "learning_rate": 3.7747747747747747e-06,
-      "loss": 0.9174,
+      "loss": 0.9197,
       "step": 62190
     },
     {
       "epoch": 186.79,
-      "grad_norm": 10.9385347366333,
+      "grad_norm": 9.003583908081055,
       "learning_rate": 3.773773773773774e-06,
-      "loss": 1.0366,
+      "loss": 1.0389,
       "step": 62200
     },
     {
       "epoch": 186.82,
-      "grad_norm": 7.465693473815918,
+      "grad_norm": 8.064310073852539,
       "learning_rate": 3.772772772772773e-06,
-      "loss": 0.9388,
+      "loss": 0.935,
       "step": 62210
     },
     {
       "epoch": 186.85,
-      "grad_norm": 6.574164390563965,
+      "grad_norm": 6.640136241912842,
       "learning_rate": 3.771771771771772e-06,
-      "loss": 1.0068,
+      "loss": 0.9963,
       "step": 62220
     },
     {
       "epoch": 186.88,
-      "grad_norm": 6.579254627227783,
+      "grad_norm": 8.309174537658691,
       "learning_rate": 3.7707707707707708e-06,
-      "loss": 0.9806,
+      "loss": 0.9814,
       "step": 62230
     },
     {
       "epoch": 186.91,
-      "grad_norm": 6.6318678855896,
+      "grad_norm": 6.587475299835205,
       "learning_rate": 3.76976976976977e-06,
-      "loss": 0.9723,
+      "loss": 0.9786,
       "step": 62240
     },
     {
       "epoch": 186.94,
-      "grad_norm": 7.745762825012207,
+      "grad_norm": 8.527280807495117,
       "learning_rate": 3.7687687687687692e-06,
-      "loss": 0.9351,
+      "loss": 0.9306,
       "step": 62250
     },
     {
       "epoch": 186.97,
-      "grad_norm": 5.357268333435059,
+      "grad_norm": 6.676831245422363,
       "learning_rate": 3.7677677677677683e-06,
-      "loss": 1.0265,
+      "loss": 1.0174,
       "step": 62260
     },
     {
       "epoch": 187.0,
-      "grad_norm": 6.128688335418701,
+      "grad_norm": 5.943664073944092,
       "learning_rate": 3.7667667667667673e-06,
-      "loss": 0.9336,
+      "loss": 0.9342,
       "step": 62270
     },
     {
       "epoch": 187.0,
-      "eval_accuracy": 0.8221,
-      "eval_loss": 0.5856371521949768,
-      "eval_runtime": 5.4917,
-      "eval_samples_per_second": 1820.937,
-      "eval_steps_per_second": 7.284,
+      "eval_accuracy": 0.8233,
+      "eval_loss": 0.5839989185333252,
+      "eval_runtime": 5.6792,
+      "eval_samples_per_second": 1760.823,
+      "eval_steps_per_second": 7.043,
       "step": 62271
     },
     {
       "epoch": 187.03,
-      "grad_norm": 6.878393650054932,
+      "grad_norm": 6.722506046295166,
       "learning_rate": 3.765765765765766e-06,
-      "loss": 1.0323,
+      "loss": 1.0298,
       "step": 62280
     },
     {
       "epoch": 187.06,
-      "grad_norm": 6.738706588745117,
+      "grad_norm": 7.302229881286621,
       "learning_rate": 3.764764764764765e-06,
-      "loss": 0.95,
+      "loss": 0.9465,
       "step": 62290
     },
     {
       "epoch": 187.09,
-      "grad_norm": 7.025619983673096,
+      "grad_norm": 6.042829990386963,
       "learning_rate": 3.7637637637637643e-06,
-      "loss": 0.9525,
+      "loss": 0.9559,
       "step": 62300
     },
     {
       "epoch": 187.12,
-      "grad_norm": 7.269512176513672,
+      "grad_norm": 5.968397617340088,
       "learning_rate": 3.7627627627627634e-06,
-      "loss": 1.056,
+      "loss": 1.0632,
       "step": 62310
     },
     {
       "epoch": 187.15,
-      "grad_norm": 6.492223739624023,
+      "grad_norm": 6.8339996337890625,
       "learning_rate": 3.761761761761762e-06,
-      "loss": 0.9726,
+      "loss": 0.9738,
       "step": 62320
     },
     {
       "epoch": 187.18,
-      "grad_norm": 8.125943183898926,
+      "grad_norm": 10.180669784545898,
       "learning_rate": 3.760760760760761e-06,
-      "loss": 1.0219,
+      "loss": 1.0257,
       "step": 62330
     },
     {
       "epoch": 187.21,
-      "grad_norm": 6.579796314239502,
+      "grad_norm": 6.387184143066406,
       "learning_rate": 3.75975975975976e-06,
-      "loss": 0.9275,
+      "loss": 0.9315,
       "step": 62340
     },
     {
       "epoch": 187.24,
-      "grad_norm": 9.965291976928711,
+      "grad_norm": 13.002514839172363,
       "learning_rate": 3.7587587587587594e-06,
-      "loss": 1.0006,
+      "loss": 0.9995,
       "step": 62350
     },
     {
       "epoch": 187.27,
-      "grad_norm": 6.565539836883545,
+      "grad_norm": 6.1140899658203125,
       "learning_rate": 3.757757757757758e-06,
-      "loss": 0.8993,
+      "loss": 0.9029,
       "step": 62360
     },
     {
       "epoch": 187.3,
-      "grad_norm": 6.934856414794922,
+      "grad_norm": 8.800145149230957,
       "learning_rate": 3.756756756756757e-06,
-      "loss": 0.9061,
+      "loss": 0.8988,
       "step": 62370
     },
     {
       "epoch": 187.33,
-      "grad_norm": 7.616572856903076,
+      "grad_norm": 7.158848762512207,
       "learning_rate": 3.755755755755756e-06,
-      "loss": 0.9188,
+      "loss": 0.9128,
       "step": 62380
     },
     {
       "epoch": 187.36,
-      "grad_norm": 5.3663835525512695,
+      "grad_norm": 4.938182830810547,
       "learning_rate": 3.754754754754755e-06,
-      "loss": 0.9226,
+      "loss": 0.921,
       "step": 62390
     },
     {
       "epoch": 187.39,
-      "grad_norm": 6.518790245056152,
+      "grad_norm": 5.896596431732178,
       "learning_rate": 3.7537537537537537e-06,
-      "loss": 0.9582,
+      "loss": 0.9592,
       "step": 62400
     },
     {
       "epoch": 187.42,
-      "grad_norm": 7.394542217254639,
+      "grad_norm": 6.820781230926514,
       "learning_rate": 3.752752752752753e-06,
-      "loss": 0.9491,
+      "loss": 0.9523,
       "step": 62410
     },
     {
       "epoch": 187.45,
-      "grad_norm": 5.404160499572754,
+      "grad_norm": 6.836843490600586,
       "learning_rate": 3.751751751751752e-06,
-      "loss": 0.9736,
+      "loss": 0.9566,
       "step": 62420
     },
     {
       "epoch": 187.48,
-      "grad_norm": 6.551852226257324,
+      "grad_norm": 6.505169868469238,
       "learning_rate": 3.750750750750751e-06,
-      "loss": 0.9303,
+      "loss": 0.9383,
       "step": 62430
     },
     {
       "epoch": 187.51,
-      "grad_norm": 7.22265625,
+      "grad_norm": 6.646194934844971,
       "learning_rate": 3.7497497497497497e-06,
-      "loss": 0.9528,
+      "loss": 0.952,
       "step": 62440
     },
     {
       "epoch": 187.54,
-      "grad_norm": 5.082779884338379,
+      "grad_norm": 8.990429878234863,
       "learning_rate": 3.7487487487487488e-06,
-      "loss": 0.9615,
+      "loss": 0.9608,
       "step": 62450
     },
     {
       "epoch": 187.57,
-      "grad_norm": 4.986639499664307,
+      "grad_norm": 7.04560661315918,
       "learning_rate": 3.747747747747748e-06,
-      "loss": 1.0128,
+      "loss": 1.0193,
       "step": 62460
     },
     {
       "epoch": 187.6,
-      "grad_norm": 8.13492202758789,
+      "grad_norm": 9.049245834350586,
       "learning_rate": 3.7467467467467472e-06,
-      "loss": 0.9671,
+      "loss": 0.9609,
       "step": 62470
     },
     {
       "epoch": 187.63,
-      "grad_norm": 6.180700778961182,
+      "grad_norm": 6.906842231750488,
       "learning_rate": 3.745745745745746e-06,
-      "loss": 0.9428,
+      "loss": 0.951,
       "step": 62480
     },
     {
       "epoch": 187.66,
-      "grad_norm": 7.034291744232178,
+      "grad_norm": 6.981168270111084,
       "learning_rate": 3.744744744744745e-06,
-      "loss": 0.9104,
+      "loss": 0.9126,
       "step": 62490
     },
     {
       "epoch": 187.69,
-      "grad_norm": 6.762683868408203,
+      "grad_norm": 8.220126152038574,
       "learning_rate": 3.743743743743744e-06,
-      "loss": 0.9139,
+      "loss": 0.9111,
       "step": 62500
     },
     {
       "epoch": 187.72,
-      "grad_norm": 6.9419474601745605,
+      "grad_norm": 6.090997695922852,
       "learning_rate": 3.7427427427427433e-06,
-      "loss": 1.0048,
+      "loss": 1.0076,
       "step": 62510
     },
     {
       "epoch": 187.75,
-      "grad_norm": 6.989166259765625,
+      "grad_norm": 6.63918924331665,
       "learning_rate": 3.7417417417417423e-06,
-      "loss": 0.9663,
+      "loss": 0.9713,
       "step": 62520
     },
     {
       "epoch": 187.78,
-      "grad_norm": 5.558735370635986,
+      "grad_norm": 6.93048095703125,
       "learning_rate": 3.740740740740741e-06,
-      "loss": 0.9769,
+      "loss": 0.9768,
       "step": 62530
     },
     {
       "epoch": 187.81,
-      "grad_norm": 6.041651248931885,
+      "grad_norm": 5.978641986846924,
       "learning_rate": 3.73973973973974e-06,
-      "loss": 0.9428,
+      "loss": 0.9371,
       "step": 62540
     },
     {
       "epoch": 187.84,
-      "grad_norm": 5.985527515411377,
+      "grad_norm": 6.457122325897217,
       "learning_rate": 3.7387387387387394e-06,
-      "loss": 0.9882,
+      "loss": 0.9763,
       "step": 62550
     },
     {
       "epoch": 187.87,
-      "grad_norm": 6.206282615661621,
+      "grad_norm": 11.797399520874023,
       "learning_rate": 3.7377377377377384e-06,
-      "loss": 1.0038,
+      "loss": 0.999,
       "step": 62560
     },
     {
       "epoch": 187.9,
-      "grad_norm": 8.8430814743042,
+      "grad_norm": 8.350495338439941,
       "learning_rate": 3.736736736736737e-06,
-      "loss": 1.0108,
+      "loss": 1.0053,
       "step": 62570
     },
     {
       "epoch": 187.93,
-      "grad_norm": 6.695288181304932,
+      "grad_norm": 6.762351989746094,
       "learning_rate": 3.735735735735736e-06,
-      "loss": 0.9161,
+      "loss": 0.9147,
       "step": 62580
     },
     {
       "epoch": 187.96,
-      "grad_norm": 5.887287139892578,
+      "grad_norm": 7.402275085449219,
       "learning_rate": 3.734734734734735e-06,
-      "loss": 1.0147,
+      "loss": 1.0108,
       "step": 62590
     },
     {
       "epoch": 187.99,
-      "grad_norm": 6.939835548400879,
+      "grad_norm": 8.291165351867676,
       "learning_rate": 3.7337337337337345e-06,
-      "loss": 1.0148,
+      "loss": 1.0178,
       "step": 62600
     },
     {
       "epoch": 188.0,
-      "eval_accuracy": 0.8205,
-      "eval_loss": 0.5903107523918152,
-      "eval_runtime": 5.4918,
-      "eval_samples_per_second": 1820.91,
-      "eval_steps_per_second": 7.284,
+      "eval_accuracy": 0.8198,
+      "eval_loss": 0.5892007946968079,
+      "eval_runtime": 5.4301,
+      "eval_samples_per_second": 1841.598,
+      "eval_steps_per_second": 7.366,
       "step": 62604
     },
     {
       "epoch": 188.02,
-      "grad_norm": 6.065162658691406,
+      "grad_norm": 6.099264144897461,
       "learning_rate": 3.732732732732733e-06,
-      "loss": 1.0973,
+      "loss": 1.106,
       "step": 62610
     },
     {
       "epoch": 188.05,
-      "grad_norm": 9.761616706848145,
+      "grad_norm": 7.167239665985107,
       "learning_rate": 3.731731731731732e-06,
-      "loss": 0.9835,
+      "loss": 0.9792,
       "step": 62620
     },
     {
       "epoch": 188.08,
-      "grad_norm": 8.3375825881958,
+      "grad_norm": 6.353151798248291,
       "learning_rate": 3.730730730730731e-06,
-      "loss": 0.9707,
+      "loss": 0.9574,
       "step": 62630
     },
     {
       "epoch": 188.11,
-      "grad_norm": 6.363694190979004,
+      "grad_norm": 6.0030412673950195,
       "learning_rate": 3.72972972972973e-06,
-      "loss": 0.9338,
+      "loss": 0.9182,
       "step": 62640
     },
     {
       "epoch": 188.14,
-      "grad_norm": 6.968137741088867,
+      "grad_norm": 6.719593524932861,
       "learning_rate": 3.7287287287287287e-06,
-      "loss": 0.9346,
+      "loss": 0.9347,
       "step": 62650
     },
     {
       "epoch": 188.17,
-      "grad_norm": 5.660830497741699,
+      "grad_norm": 5.719414234161377,
       "learning_rate": 3.727727727727728e-06,
-      "loss": 0.9619,
+      "loss": 0.9564,
       "step": 62660
     },
     {
       "epoch": 188.2,
-      "grad_norm": 9.215130805969238,
+      "grad_norm": 7.855037212371826,
       "learning_rate": 3.726726726726727e-06,
-      "loss": 0.9666,
+      "loss": 0.9512,
       "step": 62670
     },
     {
       "epoch": 188.23,
-      "grad_norm": 6.2149128913879395,
+      "grad_norm": 5.926156520843506,
       "learning_rate": 3.725725725725726e-06,
-      "loss": 0.9184,
+      "loss": 0.9236,
       "step": 62680
     },
     {
       "epoch": 188.26,
-      "grad_norm": 6.809688091278076,
+      "grad_norm": 6.995885372161865,
       "learning_rate": 3.7247247247247248e-06,
-      "loss": 0.9277,
+      "loss": 0.9145,
       "step": 62690
     },
     {
       "epoch": 188.29,
-      "grad_norm": 9.579018592834473,
+      "grad_norm": 6.909655570983887,
       "learning_rate": 3.723723723723724e-06,
-      "loss": 0.995,
+      "loss": 0.99,
       "step": 62700
     },
     {
       "epoch": 188.32,
-      "grad_norm": 7.55750846862793,
+      "grad_norm": 8.0038480758667,
       "learning_rate": 3.7227227227227232e-06,
-      "loss": 0.9074,
+      "loss": 0.9112,
       "step": 62710
     },
     {
       "epoch": 188.35,
-      "grad_norm": 6.561522006988525,
+      "grad_norm": 8.21249771118164,
       "learning_rate": 3.7217217217217223e-06,
-      "loss": 1.0319,
+      "loss": 1.0434,
       "step": 62720
     },
     {
       "epoch": 188.38,
-      "grad_norm": 5.518377780914307,
+      "grad_norm": 7.038833141326904,
       "learning_rate": 3.720720720720721e-06,
-      "loss": 0.9442,
+      "loss": 0.9495,
       "step": 62730
     },
     {
       "epoch": 188.41,
-      "grad_norm": 6.635398864746094,
+      "grad_norm": 10.275400161743164,
       "learning_rate": 3.71971971971972e-06,
-      "loss": 1.0001,
+      "loss": 0.9894,
       "step": 62740
     },
     {
       "epoch": 188.44,
-      "grad_norm": 8.890018463134766,
+      "grad_norm": 8.484926223754883,
       "learning_rate": 3.718718718718719e-06,
-      "loss": 0.9036,
+      "loss": 0.9088,
       "step": 62750
     },
     {
       "epoch": 188.47,
-      "grad_norm": 5.859053611755371,
+      "grad_norm": 6.415878772735596,
       "learning_rate": 3.7177177177177183e-06,
-      "loss": 0.9337,
+      "loss": 0.936,
       "step": 62760
     },
     {
       "epoch": 188.5,
-      "grad_norm": 5.842217445373535,
+      "grad_norm": 7.406108379364014,
       "learning_rate": 3.716716716716717e-06,
-      "loss": 0.9804,
+      "loss": 0.9856,
       "step": 62770
     },
     {
       "epoch": 188.53,
-      "grad_norm": 6.1079511642456055,
+      "grad_norm": 6.303323745727539,
       "learning_rate": 3.715715715715716e-06,
-      "loss": 0.8968,
+      "loss": 0.8896,
       "step": 62780
     },
     {
       "epoch": 188.56,
-      "grad_norm": 7.58093786239624,
+      "grad_norm": 7.867791175842285,
       "learning_rate": 3.714714714714715e-06,
-      "loss": 0.9761,
+      "loss": 0.9905,
       "step": 62790
     },
     {
       "epoch": 188.59,
-      "grad_norm": 9.046996116638184,
+      "grad_norm": 10.589790344238281,
       "learning_rate": 3.713713713713714e-06,
-      "loss": 0.98,
+      "loss": 0.9752,
       "step": 62800
     },
     {
       "epoch": 188.62,
-      "grad_norm": 7.121416091918945,
+      "grad_norm": 6.339098930358887,
       "learning_rate": 3.7127127127127134e-06,
-      "loss": 0.9732,
+      "loss": 0.9821,
       "step": 62810
     },
     {
       "epoch": 188.65,
-      "grad_norm": 5.834244728088379,
+      "grad_norm": 6.25957727432251,
       "learning_rate": 3.711711711711712e-06,
-      "loss": 0.9204,
+      "loss": 0.9156,
       "step": 62820
     },
     {
       "epoch": 188.68,
-      "grad_norm": 7.644344806671143,
+      "grad_norm": 9.11428451538086,
       "learning_rate": 3.710710710710711e-06,
-      "loss": 1.0455,
+      "loss": 1.0403,
       "step": 62830
     },
     {
       "epoch": 188.71,
-      "grad_norm": 6.899954795837402,
+      "grad_norm": 6.76245641708374,
       "learning_rate": 3.70970970970971e-06,
-      "loss": 0.9445,
+      "loss": 0.9478,
       "step": 62840
     },
     {
       "epoch": 188.74,
-      "grad_norm": 6.48959493637085,
+      "grad_norm": 7.978260040283203,
       "learning_rate": 3.708708708708709e-06,
-      "loss": 0.9952,
+      "loss": 1.0007,
       "step": 62850
     },
     {
       "epoch": 188.77,
-      "grad_norm": 7.36521053314209,
+      "grad_norm": 6.637894153594971,
       "learning_rate": 3.7077077077077077e-06,
-      "loss": 1.0006,
+      "loss": 0.9867,
       "step": 62860
     },
     {
       "epoch": 188.8,
-      "grad_norm": 6.158368110656738,
+      "grad_norm": 7.951972484588623,
       "learning_rate": 3.706706706706707e-06,
-      "loss": 0.9675,
+      "loss": 0.9684,
       "step": 62870
     },
     {
       "epoch": 188.83,
-      "grad_norm": 6.001492500305176,
+      "grad_norm": 6.273138523101807,
       "learning_rate": 3.705705705705706e-06,
-      "loss": 0.9184,
+      "loss": 0.9123,
       "step": 62880
     },
     {
       "epoch": 188.86,
-      "grad_norm": 5.497878074645996,
+      "grad_norm": 5.316123008728027,
       "learning_rate": 3.704704704704705e-06,
-      "loss": 0.9164,
+      "loss": 0.9243,
       "step": 62890
     },
     {
       "epoch": 188.89,
-      "grad_norm": 7.078121185302734,
+      "grad_norm": 6.548666477203369,
       "learning_rate": 3.7037037037037037e-06,
-      "loss": 0.9378,
+      "loss": 0.9426,
       "step": 62900
     },
     {
       "epoch": 188.92,
-      "grad_norm": 8.084782600402832,
+      "grad_norm": 7.285181522369385,
       "learning_rate": 3.7027027027027028e-06,
-      "loss": 0.9624,
+      "loss": 0.9593,
       "step": 62910
     },
     {
       "epoch": 188.95,
-      "grad_norm": 6.55314302444458,
+      "grad_norm": 7.125308036804199,
       "learning_rate": 3.701701701701702e-06,
-      "loss": 0.9026,
+      "loss": 0.8987,
       "step": 62920
     },
     {
       "epoch": 188.98,
-      "grad_norm": 5.136220455169678,
+      "grad_norm": 5.211755275726318,
       "learning_rate": 3.7007007007007012e-06,
-      "loss": 0.9184,
+      "loss": 0.9279,
       "step": 62930
     },
     {
       "epoch": 189.0,
-      "eval_accuracy": 0.8217,
-      "eval_loss": 0.5811169743537903,
-      "eval_runtime": 5.3795,
-      "eval_samples_per_second": 1858.892,
-      "eval_steps_per_second": 7.436,
+      "eval_accuracy": 0.8229,
+      "eval_loss": 0.5813032984733582,
+      "eval_runtime": 5.2945,
+      "eval_samples_per_second": 1888.768,
+      "eval_steps_per_second": 7.555,
       "step": 62937
     },
     {
       "epoch": 189.01,
-      "grad_norm": 10.037946701049805,
+      "grad_norm": 10.271612167358398,
       "learning_rate": 3.6996996996997e-06,
-      "loss": 1.1574,
+      "loss": 1.15,
       "step": 62940
     },
     {
       "epoch": 189.04,
-      "grad_norm": 6.004737377166748,
+      "grad_norm": 6.405943393707275,
       "learning_rate": 3.698698698698699e-06,
-      "loss": 0.992,
+      "loss": 0.9981,
       "step": 62950
     },
     {
       "epoch": 189.07,
-      "grad_norm": 6.575816631317139,
+      "grad_norm": 6.473845958709717,
       "learning_rate": 3.697697697697698e-06,
-      "loss": 0.9937,
+      "loss": 0.9876,
       "step": 62960
     },
     {
       "epoch": 189.1,
-      "grad_norm": 5.862962245941162,
+      "grad_norm": 6.152550220489502,
       "learning_rate": 3.6966966966966973e-06,
-      "loss": 0.9268,
+      "loss": 0.9269,
       "step": 62970
     },
     {
       "epoch": 189.13,
-      "grad_norm": 8.257712364196777,
+      "grad_norm": 7.680664539337158,
       "learning_rate": 3.695695695695696e-06,
-      "loss": 0.9508,
+      "loss": 0.9456,
       "step": 62980
     },
     {
       "epoch": 189.16,
-      "grad_norm": 7.007780075073242,
+      "grad_norm": 8.852792739868164,
       "learning_rate": 3.694694694694695e-06,
-      "loss": 0.9896,
+      "loss": 0.9838,
       "step": 62990
     },
     {
       "epoch": 189.19,
-      "grad_norm": 6.013000965118408,
+      "grad_norm": 7.093292713165283,
       "learning_rate": 3.693693693693694e-06,
-      "loss": 0.9473,
+      "loss": 0.9395,
       "step": 63000
     },
     {
       "epoch": 189.22,
-      "grad_norm": 4.382778644561768,
+      "grad_norm": 5.466906547546387,
       "learning_rate": 3.692692692692693e-06,
-      "loss": 0.918,
+      "loss": 0.9154,
       "step": 63010
     },
     {
       "epoch": 189.25,
-      "grad_norm": 5.340714931488037,
+      "grad_norm": 6.610934257507324,
       "learning_rate": 3.6916916916916915e-06,
-      "loss": 0.9228,
+      "loss": 0.9229,
       "step": 63020
     },
     {
       "epoch": 189.28,
-      "grad_norm": 8.455182075500488,
+      "grad_norm": 7.839634895324707,
       "learning_rate": 3.690690690690691e-06,
-      "loss": 0.9826,
+      "loss": 0.9796,
       "step": 63030
     },
     {
       "epoch": 189.31,
-      "grad_norm": 6.952402114868164,
+      "grad_norm": 7.217624664306641,
       "learning_rate": 3.68968968968969e-06,
-      "loss": 0.8849,
+      "loss": 0.8967,
       "step": 63040
     },
     {
       "epoch": 189.34,
-      "grad_norm": 8.294417381286621,
+      "grad_norm": 9.072685241699219,
       "learning_rate": 3.688688688688689e-06,
-      "loss": 0.954,
+      "loss": 0.9577,
       "step": 63050
     },
     {
       "epoch": 189.37,
-      "grad_norm": 7.007102012634277,
+      "grad_norm": 6.005403518676758,
       "learning_rate": 3.6876876876876885e-06,
-      "loss": 0.9953,
+      "loss": 0.9945,
       "step": 63060
     },
     {
       "epoch": 189.4,
-      "grad_norm": 4.395355224609375,
+      "grad_norm": 5.712246417999268,
       "learning_rate": 3.6866866866866866e-06,
-      "loss": 0.8476,
+      "loss": 0.8519,
       "step": 63070
     },
     {
       "epoch": 189.43,
-      "grad_norm": 8.113494873046875,
+      "grad_norm": 7.3385009765625,
       "learning_rate": 3.685685685685686e-06,
-      "loss": 0.9876,
+      "loss": 0.9863,
       "step": 63080
     },
     {
       "epoch": 189.46,
-      "grad_norm": 4.8790812492370605,
+      "grad_norm": 6.829197406768799,
       "learning_rate": 3.684684684684685e-06,
-      "loss": 0.9713,
+      "loss": 0.9732,
       "step": 63090
     },
     {
       "epoch": 189.49,
-      "grad_norm": 10.283567428588867,
+      "grad_norm": 9.821922302246094,
       "learning_rate": 3.683683683683684e-06,
-      "loss": 0.8844,
+      "loss": 0.8864,
       "step": 63100
     },
     {
       "epoch": 189.52,
-      "grad_norm": 7.061463832855225,
+      "grad_norm": 8.10657787322998,
       "learning_rate": 3.6826826826826827e-06,
-      "loss": 0.8655,
+      "loss": 0.8549,
       "step": 63110
     },
     {
       "epoch": 189.55,
-      "grad_norm": 5.684554576873779,
+      "grad_norm": 7.262762546539307,
       "learning_rate": 3.681681681681682e-06,
-      "loss": 0.9222,
+      "loss": 0.9293,
       "step": 63120
     },
     {
       "epoch": 189.58,
-      "grad_norm": 7.896756172180176,
+      "grad_norm": 10.920295715332031,
       "learning_rate": 3.680680680680681e-06,
-      "loss": 1.0166,
+      "loss": 1.0109,
       "step": 63130
     },
     {
       "epoch": 189.61,
-      "grad_norm": 8.464810371398926,
+      "grad_norm": 5.683691501617432,
       "learning_rate": 3.67967967967968e-06,
-      "loss": 0.9843,
+      "loss": 0.9805,
       "step": 63140
     },
     {
       "epoch": 189.64,
-      "grad_norm": 7.465626239776611,
+      "grad_norm": 7.644509792327881,
       "learning_rate": 3.6786786786786788e-06,
-      "loss": 0.9854,
+      "loss": 0.9865,
       "step": 63150
     },
     {
       "epoch": 189.67,
-      "grad_norm": 6.505306243896484,
+      "grad_norm": 6.412083625793457,
       "learning_rate": 3.677677677677678e-06,
-      "loss": 0.9501,
+      "loss": 0.9505,
       "step": 63160
     },
     {
       "epoch": 189.7,
-      "grad_norm": 6.190248012542725,
+      "grad_norm": 6.560059070587158,
       "learning_rate": 3.6766766766766772e-06,
-      "loss": 0.8963,
+      "loss": 0.8913,
       "step": 63170
     },
     {
       "epoch": 189.73,
-      "grad_norm": 5.682215213775635,
+      "grad_norm": 6.547494411468506,
       "learning_rate": 3.6756756756756763e-06,
-      "loss": 0.9346,
+      "loss": 0.9317,
       "step": 63180
     },
     {
       "epoch": 189.76,
-      "grad_norm": 6.6893534660339355,
+      "grad_norm": 6.199770450592041,
       "learning_rate": 3.674674674674675e-06,
-      "loss": 0.9284,
+      "loss": 0.9146,
       "step": 63190
     },
     {
       "epoch": 189.79,
-      "grad_norm": 7.734560489654541,
+      "grad_norm": 8.025856971740723,
       "learning_rate": 3.673673673673674e-06,
-      "loss": 0.9978,
+      "loss": 0.9824,
       "step": 63200
     },
     {
       "epoch": 189.82,
-      "grad_norm": 6.906240463256836,
+      "grad_norm": 6.617509365081787,
       "learning_rate": 3.672672672672673e-06,
-      "loss": 0.9935,
+      "loss": 0.9916,
       "step": 63210
     },
     {
       "epoch": 189.85,
-      "grad_norm": 5.948399543762207,
+      "grad_norm": 6.55916690826416,
       "learning_rate": 3.6716716716716723e-06,
-      "loss": 0.9253,
+      "loss": 0.9271,
       "step": 63220
     },
     {
       "epoch": 189.88,
-      "grad_norm": 4.275208950042725,
+      "grad_norm": 4.67455530166626,
       "learning_rate": 3.670670670670671e-06,
-      "loss": 1.095,
+      "loss": 1.0888,
       "step": 63230
     },
     {
       "epoch": 189.91,
-      "grad_norm": 8.194496154785156,
+      "grad_norm": 10.118561744689941,
       "learning_rate": 3.66966966966967e-06,
-      "loss": 0.9606,
+      "loss": 0.9608,
       "step": 63240
     },
     {
       "epoch": 189.94,
-      "grad_norm": 7.992068767547607,
+      "grad_norm": 8.073339462280273,
       "learning_rate": 3.668668668668669e-06,
-      "loss": 0.9963,
+      "loss": 0.9995,
       "step": 63250
     },
     {
       "epoch": 189.97,
-      "grad_norm": 6.340768337249756,
+      "grad_norm": 6.628989219665527,
       "learning_rate": 3.667667667667668e-06,
-      "loss": 0.9667,
+      "loss": 0.9659,
       "step": 63260
     },
     {
       "epoch": 190.0,
-      "grad_norm": 51.95118713378906,
+      "grad_norm": 56.592918395996094,
       "learning_rate": 3.6666666666666666e-06,
-      "loss": 1.2194,
+      "loss": 1.193,
       "step": 63270
     },
     {
       "epoch": 190.0,
-      "eval_accuracy": 0.8214,
-      "eval_loss": 0.585174024105072,
-      "eval_runtime": 5.3881,
-      "eval_samples_per_second": 1855.936,
-      "eval_steps_per_second": 7.424,
+      "eval_accuracy": 0.8233,
+      "eval_loss": 0.5849727988243103,
+      "eval_runtime": 5.3058,
+      "eval_samples_per_second": 1884.712,
+      "eval_steps_per_second": 7.539,
       "step": 63270
     },
     {
       "epoch": 190.03,
-      "grad_norm": 7.025871753692627,
+      "grad_norm": 7.175314426422119,
       "learning_rate": 3.665665665665666e-06,
-      "loss": 0.9573,
+      "loss": 0.9531,
       "step": 63280
     },
     {
       "epoch": 190.06,
-      "grad_norm": 5.7278337478637695,
+      "grad_norm": 6.746385097503662,
       "learning_rate": 3.664664664664665e-06,
-      "loss": 0.9792,
+      "loss": 0.9775,
       "step": 63290
     },
     {
       "epoch": 190.09,
-      "grad_norm": 5.966313362121582,
+      "grad_norm": 8.508235931396484,
       "learning_rate": 3.663663663663664e-06,
-      "loss": 0.9119,
+      "loss": 0.9123,
       "step": 63300
     },
     {
       "epoch": 190.12,
-      "grad_norm": 6.412699222564697,
+      "grad_norm": 6.913599967956543,
       "learning_rate": 3.662662662662663e-06,
-      "loss": 0.9279,
+      "loss": 0.9191,
       "step": 63310
     },
     {
       "epoch": 190.15,
-      "grad_norm": 5.91480016708374,
+      "grad_norm": 5.249114990234375,
       "learning_rate": 3.6616616616616617e-06,
-      "loss": 0.9055,
+      "loss": 0.9059,
       "step": 63320
     },
     {
       "epoch": 190.18,
-      "grad_norm": 6.9377055168151855,
+      "grad_norm": 7.83038330078125,
       "learning_rate": 3.660660660660661e-06,
-      "loss": 0.9688,
+      "loss": 0.9698,
       "step": 63330
     },
     {
       "epoch": 190.21,
-      "grad_norm": 6.6478962898254395,
+      "grad_norm": 8.05334758758545,
       "learning_rate": 3.65965965965966e-06,
-      "loss": 0.9707,
+      "loss": 0.9719,
       "step": 63340
     },
     {
       "epoch": 190.24,
-      "grad_norm": 7.822774410247803,
+      "grad_norm": 9.463396072387695,
       "learning_rate": 3.658658658658659e-06,
-      "loss": 1.0102,
+      "loss": 1.0117,
       "step": 63350
     },
     {
       "epoch": 190.27,
-      "grad_norm": 9.106644630432129,
+      "grad_norm": 13.753082275390625,
       "learning_rate": 3.6576576576576577e-06,
-      "loss": 1.041,
+      "loss": 1.0295,
       "step": 63360
     },
     {
       "epoch": 190.3,
-      "grad_norm": 5.878053665161133,
+      "grad_norm": 6.9020843505859375,
       "learning_rate": 3.6566566566566568e-06,
-      "loss": 0.9748,
+      "loss": 0.9627,
       "step": 63370
     },
     {
       "epoch": 190.33,
-      "grad_norm": 6.316137313842773,
+      "grad_norm": 7.892165660858154,
       "learning_rate": 3.655655655655656e-06,
-      "loss": 0.9414,
+      "loss": 0.9426,
       "step": 63380
     },
     {
       "epoch": 190.36,
-      "grad_norm": 6.322136878967285,
+      "grad_norm": 6.114143371582031,
       "learning_rate": 3.6546546546546552e-06,
-      "loss": 0.9176,
+      "loss": 0.9141,
       "step": 63390
     },
     {
       "epoch": 190.39,
-      "grad_norm": 6.134067058563232,
+      "grad_norm": 7.500816822052002,
       "learning_rate": 3.653653653653654e-06,
-      "loss": 0.9513,
+      "loss": 0.951,
       "step": 63400
     },
     {
       "epoch": 190.42,
-      "grad_norm": 6.681760311126709,
+      "grad_norm": 7.4892354011535645,
       "learning_rate": 3.652652652652653e-06,
-      "loss": 0.911,
+      "loss": 0.9117,
       "step": 63410
     },
     {
       "epoch": 190.45,
-      "grad_norm": 11.396188735961914,
+      "grad_norm": 8.427125930786133,
       "learning_rate": 3.651651651651652e-06,
-      "loss": 0.9231,
+      "loss": 0.9254,
       "step": 63420
     },
     {
       "epoch": 190.48,
-      "grad_norm": 5.833983898162842,
+      "grad_norm": 6.1100897789001465,
       "learning_rate": 3.6506506506506513e-06,
-      "loss": 0.9012,
+      "loss": 0.8901,
       "step": 63430
     },
     {
       "epoch": 190.51,
-      "grad_norm": 9.734024047851562,
+      "grad_norm": 10.542137145996094,
       "learning_rate": 3.64964964964965e-06,
-      "loss": 0.9768,
+      "loss": 0.9786,
       "step": 63440
     },
     {
       "epoch": 190.54,
-      "grad_norm": 6.9378533363342285,
+      "grad_norm": 6.3958234786987305,
       "learning_rate": 3.648648648648649e-06,
-      "loss": 0.9293,
+      "loss": 0.9213,
       "step": 63450
     },
     {
       "epoch": 190.57,
-      "grad_norm": 6.744111061096191,
+      "grad_norm": 7.317327976226807,
       "learning_rate": 3.647647647647648e-06,
-      "loss": 0.9275,
+      "loss": 0.9313,
       "step": 63460
     },
     {
       "epoch": 190.6,
-      "grad_norm": 5.714186668395996,
+      "grad_norm": 9.0894136428833,
       "learning_rate": 3.646646646646647e-06,
-      "loss": 0.9071,
+      "loss": 0.9162,
       "step": 63470
     },
     {
       "epoch": 190.63,
-      "grad_norm": 6.694404125213623,
+      "grad_norm": 7.439560413360596,
       "learning_rate": 3.6456456456456455e-06,
-      "loss": 0.9948,
+      "loss": 1.0025,
       "step": 63480
     },
     {
       "epoch": 190.66,
-      "grad_norm": 6.8883819580078125,
+      "grad_norm": 6.302369594573975,
       "learning_rate": 3.644644644644645e-06,
-      "loss": 0.9103,
+      "loss": 0.9202,
       "step": 63490
     },
     {
       "epoch": 190.69,
-      "grad_norm": 7.185611724853516,
+      "grad_norm": 9.118399620056152,
       "learning_rate": 3.643643643643644e-06,
-      "loss": 0.924,
+      "loss": 0.9222,
       "step": 63500
     },
     {
       "epoch": 190.72,
-      "grad_norm": 7.9870500564575195,
+      "grad_norm": 9.612536430358887,
       "learning_rate": 3.642642642642643e-06,
-      "loss": 0.9804,
+      "loss": 0.9823,
       "step": 63510
     },
     {
       "epoch": 190.75,
-      "grad_norm": 5.76318359375,
+      "grad_norm": 4.850242614746094,
       "learning_rate": 3.6416416416416416e-06,
-      "loss": 0.8977,
+      "loss": 0.8967,
       "step": 63520
     },
     {
       "epoch": 190.78,
-      "grad_norm": 7.563267707824707,
+      "grad_norm": 6.17231559753418,
       "learning_rate": 3.6406406406406406e-06,
-      "loss": 1.0148,
+      "loss": 1.0118,
       "step": 63530
     },
     {
       "epoch": 190.81,
-      "grad_norm": 6.1282453536987305,
+      "grad_norm": 6.803900718688965,
       "learning_rate": 3.63963963963964e-06,
-      "loss": 0.9641,
+      "loss": 0.9614,
       "step": 63540
     },
     {
       "epoch": 190.84,
-      "grad_norm": 8.130715370178223,
+      "grad_norm": 9.25068187713623,
       "learning_rate": 3.638638638638639e-06,
-      "loss": 1.0097,
+      "loss": 1.012,
       "step": 63550
     },
     {
       "epoch": 190.87,
-      "grad_norm": 6.481472492218018,
+      "grad_norm": 5.833984851837158,
       "learning_rate": 3.6376376376376377e-06,
-      "loss": 0.9881,
+      "loss": 0.9885,
       "step": 63560
     },
     {
       "epoch": 190.9,
-      "grad_norm": 6.937613487243652,
+      "grad_norm": 6.072689533233643,
       "learning_rate": 3.6366366366366367e-06,
-      "loss": 0.9234,
+      "loss": 0.9228,
       "step": 63570
     },
     {
       "epoch": 190.93,
-      "grad_norm": 4.87177848815918,
+      "grad_norm": 6.5354719161987305,
       "learning_rate": 3.635635635635636e-06,
-      "loss": 0.9728,
+      "loss": 0.972,
       "step": 63580
     },
     {
       "epoch": 190.96,
-      "grad_norm": 6.538061618804932,
+      "grad_norm": 8.679620742797852,
       "learning_rate": 3.634634634634635e-06,
-      "loss": 1.0257,
+      "loss": 1.0192,
       "step": 63590
     },
     {
       "epoch": 190.99,
-      "grad_norm": 8.001168251037598,
+      "grad_norm": 7.873843669891357,
       "learning_rate": 3.633633633633634e-06,
-      "loss": 0.9717,
+      "loss": 0.9696,
       "step": 63600
     },
     {
       "epoch": 191.0,
-      "eval_accuracy": 0.8204,
-      "eval_loss": 0.5872608423233032,
-      "eval_runtime": 5.4167,
-      "eval_samples_per_second": 1846.13,
-      "eval_steps_per_second": 7.385,
+      "eval_accuracy": 0.8223,
+      "eval_loss": 0.5850110054016113,
+      "eval_runtime": 5.3042,
+      "eval_samples_per_second": 1885.312,
+      "eval_steps_per_second": 7.541,
       "step": 63603
     },
     {
       "epoch": 191.02,
-      "grad_norm": 6.5709099769592285,
+      "grad_norm": 4.704038619995117,
       "learning_rate": 3.6326326326326328e-06,
-      "loss": 1.1472,
+      "loss": 1.1852,
       "step": 63610
     },
     {
       "epoch": 191.05,
-      "grad_norm": 7.422120094299316,
+      "grad_norm": 6.52982234954834,
       "learning_rate": 3.631631631631632e-06,
-      "loss": 0.9335,
+      "loss": 0.9323,
       "step": 63620
     },
     {
       "epoch": 191.08,
-      "grad_norm": 4.825226306915283,
+      "grad_norm": 6.599318027496338,
       "learning_rate": 3.6306306306306312e-06,
-      "loss": 0.9749,
+      "loss": 0.9727,
       "step": 63630
     },
     {
       "epoch": 191.11,
-      "grad_norm": 5.661325931549072,
+      "grad_norm": 6.828115940093994,
       "learning_rate": 3.6296296296296302e-06,
-      "loss": 0.8894,
+      "loss": 0.893,
       "step": 63640
     },
     {
       "epoch": 191.14,
-      "grad_norm": 7.692465305328369,
+      "grad_norm": 6.436712265014648,
       "learning_rate": 3.628628628628629e-06,
-      "loss": 0.9738,
+      "loss": 0.9666,
       "step": 63650
     },
     {
       "epoch": 191.17,
-      "grad_norm": 6.855896472930908,
+      "grad_norm": 6.113524436950684,
       "learning_rate": 3.627627627627628e-06,
-      "loss": 0.9009,
+      "loss": 0.9055,
       "step": 63660
     },
     {
       "epoch": 191.2,
-      "grad_norm": 6.079876899719238,
+      "grad_norm": 6.123745441436768,
       "learning_rate": 3.626626626626627e-06,
-      "loss": 0.9975,
+      "loss": 1.0012,
       "step": 63670
     },
     {
       "epoch": 191.23,
-      "grad_norm": 8.46035099029541,
+      "grad_norm": 7.173342227935791,
       "learning_rate": 3.6256256256256263e-06,
-      "loss": 0.9876,
+      "loss": 0.9813,
       "step": 63680
     },
     {
       "epoch": 191.26,
-      "grad_norm": 7.213501930236816,
+      "grad_norm": 8.691679000854492,
       "learning_rate": 3.624624624624625e-06,
-      "loss": 0.9339,
+      "loss": 0.9385,
       "step": 63690
     },
     {
       "epoch": 191.29,
-      "grad_norm": 7.970086574554443,
+      "grad_norm": 7.73565149307251,
       "learning_rate": 3.623623623623624e-06,
-      "loss": 1.0247,
+      "loss": 1.0165,
       "step": 63700
     },
     {
       "epoch": 191.32,
-      "grad_norm": 5.500356197357178,
+      "grad_norm": 6.133037567138672,
       "learning_rate": 3.622622622622623e-06,
-      "loss": 0.955,
+      "loss": 0.9453,
       "step": 63710
     },
     {
       "epoch": 191.35,
-      "grad_norm": 10.283978462219238,
+      "grad_norm": 10.475796699523926,
       "learning_rate": 3.621621621621622e-06,
-      "loss": 0.9763,
+      "loss": 0.979,
       "step": 63720
     },
     {
       "epoch": 191.38,
-      "grad_norm": 7.11979866027832,
+      "grad_norm": 7.434528350830078,
       "learning_rate": 3.6206206206206206e-06,
-      "loss": 0.9361,
+      "loss": 0.9355,
       "step": 63730
     },
     {
       "epoch": 191.41,
-      "grad_norm": 5.696298599243164,
+      "grad_norm": 6.021180152893066,
       "learning_rate": 3.61961961961962e-06,
-      "loss": 1.001,
+      "loss": 1.0,
       "step": 63740
     },
     {
       "epoch": 191.44,
-      "grad_norm": 9.378618240356445,
+      "grad_norm": 8.502116203308105,
       "learning_rate": 3.618618618618619e-06,
-      "loss": 0.9494,
+      "loss": 0.9526,
       "step": 63750
     },
     {
       "epoch": 191.47,
-      "grad_norm": 7.119906902313232,
+      "grad_norm": 6.760183334350586,
       "learning_rate": 3.617617617617618e-06,
-      "loss": 0.9357,
+      "loss": 0.92,
       "step": 63760
     },
     {
       "epoch": 191.5,
-      "grad_norm": 4.775322437286377,
+      "grad_norm": 6.342414855957031,
       "learning_rate": 3.6166166166166166e-06,
-      "loss": 0.9761,
+      "loss": 0.9816,
       "step": 63770
     },
     {
       "epoch": 191.53,
-      "grad_norm": 6.899207592010498,
+      "grad_norm": 5.839179515838623,
       "learning_rate": 3.6156156156156157e-06,
-      "loss": 0.9836,
+      "loss": 0.973,
       "step": 63780
     },
     {
       "epoch": 191.56,
-      "grad_norm": 5.959933757781982,
+      "grad_norm": 6.631348133087158,
       "learning_rate": 3.614614614614615e-06,
-      "loss": 0.948,
+      "loss": 0.9447,
       "step": 63790
     },
     {
       "epoch": 191.59,
-      "grad_norm": 5.903765678405762,
+      "grad_norm": 5.268239974975586,
       "learning_rate": 3.613613613613614e-06,
-      "loss": 0.9923,
+      "loss": 0.9806,
       "step": 63800
     },
     {
       "epoch": 191.62,
-      "grad_norm": 6.836181163787842,
+      "grad_norm": 7.909390449523926,
       "learning_rate": 3.6126126126126127e-06,
-      "loss": 0.9796,
+      "loss": 0.9718,
       "step": 63810
     },
     {
       "epoch": 191.65,
-      "grad_norm": 8.549881935119629,
+      "grad_norm": 12.633269309997559,
       "learning_rate": 3.6116116116116117e-06,
-      "loss": 0.9399,
+      "loss": 0.9222,
       "step": 63820
     },
     {
       "epoch": 191.68,
-      "grad_norm": 5.982388973236084,
+      "grad_norm": 7.342348098754883,
       "learning_rate": 3.6106106106106108e-06,
-      "loss": 0.9251,
+      "loss": 0.9176,
       "step": 63830
     },
     {
       "epoch": 191.71,
-      "grad_norm": 9.006813049316406,
+      "grad_norm": 6.348480224609375,
       "learning_rate": 3.60960960960961e-06,
-      "loss": 0.9818,
+      "loss": 0.9729,
       "step": 63840
     },
     {
       "epoch": 191.74,
-      "grad_norm": 7.994421482086182,
+      "grad_norm": 8.672937393188477,
       "learning_rate": 3.608608608608609e-06,
-      "loss": 1.0184,
+      "loss": 1.0181,
       "step": 63850
     },
     {
       "epoch": 191.77,
-      "grad_norm": 5.823777675628662,
+      "grad_norm": 6.38429069519043,
       "learning_rate": 3.607607607607608e-06,
-      "loss": 0.9173,
+      "loss": 0.9186,
       "step": 63860
     },
     {
       "epoch": 191.8,
-      "grad_norm": 11.306929588317871,
+      "grad_norm": 13.935694694519043,
       "learning_rate": 3.606606606606607e-06,
-      "loss": 0.9623,
+      "loss": 0.9534,
       "step": 63870
     },
     {
       "epoch": 191.83,
-      "grad_norm": 6.158831596374512,
+      "grad_norm": 6.1728901863098145,
       "learning_rate": 3.605605605605606e-06,
-      "loss": 0.9959,
+      "loss": 0.9979,
       "step": 63880
     },
     {
       "epoch": 191.86,
-      "grad_norm": 6.156647682189941,
+      "grad_norm": 7.185746669769287,
       "learning_rate": 3.6046046046046053e-06,
-      "loss": 0.9156,
+      "loss": 0.9179,
       "step": 63890
     },
     {
       "epoch": 191.89,
-      "grad_norm": 7.369698524475098,
+      "grad_norm": 7.4732441902160645,
       "learning_rate": 3.603603603603604e-06,
-      "loss": 0.9018,
+      "loss": 0.9048,
       "step": 63900
     },
     {
       "epoch": 191.92,
-      "grad_norm": 6.733109474182129,
+      "grad_norm": 6.415408611297607,
       "learning_rate": 3.602602602602603e-06,
-      "loss": 0.9044,
+      "loss": 0.9066,
       "step": 63910
     },
     {
       "epoch": 191.95,
-      "grad_norm": 12.482758522033691,
+      "grad_norm": 8.548994064331055,
       "learning_rate": 3.601601601601602e-06,
-      "loss": 0.9842,
+      "loss": 0.9927,
       "step": 63920
     },
     {
       "epoch": 191.98,
-      "grad_norm": 6.514369010925293,
+      "grad_norm": 6.67390775680542,
       "learning_rate": 3.600600600600601e-06,
-      "loss": 1.0003,
+      "loss": 0.9883,
       "step": 63930
     },
     {
       "epoch": 192.0,
-      "eval_accuracy": 0.8239,
-      "eval_loss": 0.58356112241745,
-      "eval_runtime": 5.4145,
-      "eval_samples_per_second": 1846.891,
-      "eval_steps_per_second": 7.388,
+      "eval_accuracy": 0.8245,
+      "eval_loss": 0.5813835859298706,
+      "eval_runtime": 5.1573,
+      "eval_samples_per_second": 1938.997,
+      "eval_steps_per_second": 7.756,
       "step": 63936
     },
     {
       "epoch": 192.01,
-      "grad_norm": 7.237217903137207,
+      "grad_norm": 7.9695634841918945,
       "learning_rate": 3.5995995995995995e-06,
-      "loss": 1.0986,
+      "loss": 1.0777,
       "step": 63940
     },
     {
       "epoch": 192.04,
-      "grad_norm": 8.581924438476562,
+      "grad_norm": 7.010540962219238,
       "learning_rate": 3.598598598598599e-06,
-      "loss": 0.9578,
+      "loss": 0.9431,
       "step": 63950
     },
     {
       "epoch": 192.07,
-      "grad_norm": 6.727468490600586,
+      "grad_norm": 6.577169895172119,
       "learning_rate": 3.597597597597598e-06,
-      "loss": 0.9732,
+      "loss": 0.9713,
       "step": 63960
     },
     {
       "epoch": 192.1,
-      "grad_norm": 9.34692096710205,
+      "grad_norm": 14.088850021362305,
       "learning_rate": 3.596596596596597e-06,
-      "loss": 0.9969,
+      "loss": 0.9866,
       "step": 63970
     },
     {
       "epoch": 192.13,
-      "grad_norm": 5.411579608917236,
+      "grad_norm": 6.432487487792969,
       "learning_rate": 3.5955955955955956e-06,
-      "loss": 0.9859,
+      "loss": 0.993,
       "step": 63980
     },
     {
       "epoch": 192.16,
-      "grad_norm": 7.742833137512207,
+      "grad_norm": 11.357644081115723,
       "learning_rate": 3.5945945945945946e-06,
-      "loss": 0.9723,
+      "loss": 0.9781,
       "step": 63990
     },
     {
       "epoch": 192.19,
-      "grad_norm": 5.286495685577393,
+      "grad_norm": 7.509390354156494,
       "learning_rate": 3.593593593593594e-06,
-      "loss": 0.874,
+      "loss": 0.8694,
       "step": 64000
     },
     {
       "epoch": 192.22,
-      "grad_norm": 5.311521053314209,
+      "grad_norm": 5.71651554107666,
       "learning_rate": 3.592592592592593e-06,
-      "loss": 0.912,
+      "loss": 0.9116,
       "step": 64010
     },
     {
       "epoch": 192.25,
-      "grad_norm": 5.601346492767334,
+      "grad_norm": 6.1304144859313965,
       "learning_rate": 3.5915915915915917e-06,
-      "loss": 0.9031,
+      "loss": 0.8961,
       "step": 64020
     },
     {
       "epoch": 192.28,
-      "grad_norm": 4.555202007293701,
+      "grad_norm": 4.834120273590088,
       "learning_rate": 3.5905905905905907e-06,
-      "loss": 0.8792,
+      "loss": 0.8811,
       "step": 64030
     },
     {
       "epoch": 192.31,
-      "grad_norm": 6.810363292694092,
+      "grad_norm": 8.362278938293457,
       "learning_rate": 3.5895895895895897e-06,
-      "loss": 0.9551,
+      "loss": 0.9539,
       "step": 64040
     },
     {
       "epoch": 192.34,
-      "grad_norm": 6.135502338409424,
+      "grad_norm": 6.719781875610352,
       "learning_rate": 3.588588588588589e-06,
-      "loss": 0.9785,
+      "loss": 0.9813,
       "step": 64050
     },
     {
       "epoch": 192.37,
-      "grad_norm": 4.976435661315918,
+      "grad_norm": 6.332519054412842,
       "learning_rate": 3.5875875875875877e-06,
-      "loss": 0.9521,
+      "loss": 0.955,
       "step": 64060
     },
     {
       "epoch": 192.4,
-      "grad_norm": 6.949187278747559,
+      "grad_norm": 8.932093620300293,
       "learning_rate": 3.5865865865865868e-06,
-      "loss": 0.9774,
+      "loss": 0.9714,
       "step": 64070
     },
     {
       "epoch": 192.43,
-      "grad_norm": 3.9578161239624023,
+      "grad_norm": 5.232165813446045,
       "learning_rate": 3.5855855855855858e-06,
-      "loss": 0.953,
+      "loss": 0.9466,
       "step": 64080
     },
     {
       "epoch": 192.46,
-      "grad_norm": 7.038139343261719,
+      "grad_norm": 6.681798458099365,
       "learning_rate": 3.5845845845845852e-06,
-      "loss": 0.9241,
+      "loss": 0.9219,
       "step": 64090
     },
     {
       "epoch": 192.49,
-      "grad_norm": 5.282907009124756,
+      "grad_norm": 5.890224933624268,
       "learning_rate": 3.5835835835835834e-06,
-      "loss": 0.9323,
+      "loss": 0.9261,
       "step": 64100
     },
     {
       "epoch": 192.52,
-      "grad_norm": 6.002531051635742,
+      "grad_norm": 6.923965930938721,
       "learning_rate": 3.582582582582583e-06,
-      "loss": 0.9417,
+      "loss": 0.9473,
       "step": 64110
     },
     {
       "epoch": 192.55,
-      "grad_norm": 7.5364203453063965,
+      "grad_norm": 6.338657855987549,
       "learning_rate": 3.581581581581582e-06,
-      "loss": 0.9006,
+      "loss": 0.893,
       "step": 64120
     },
     {
       "epoch": 192.58,
-      "grad_norm": 7.1572065353393555,
+      "grad_norm": 7.2794976234436035,
       "learning_rate": 3.580580580580581e-06,
-      "loss": 0.943,
+      "loss": 0.9414,
       "step": 64130
     },
     {
       "epoch": 192.61,
-      "grad_norm": 7.995449542999268,
+      "grad_norm": 5.682925701141357,
       "learning_rate": 3.5795795795795803e-06,
-      "loss": 0.9779,
+      "loss": 0.9812,
       "step": 64140
     },
     {
       "epoch": 192.64,
-      "grad_norm": 5.896825790405273,
+      "grad_norm": 5.071952819824219,
       "learning_rate": 3.578578578578579e-06,
-      "loss": 0.9255,
+      "loss": 0.9249,
       "step": 64150
     },
     {
       "epoch": 192.67,
-      "grad_norm": 6.813307762145996,
+      "grad_norm": 8.5750732421875,
       "learning_rate": 3.577577577577578e-06,
-      "loss": 0.9609,
+      "loss": 0.9702,
       "step": 64160
     },
     {
       "epoch": 192.7,
-      "grad_norm": 6.186652660369873,
+      "grad_norm": 7.8035359382629395,
       "learning_rate": 3.576576576576577e-06,
-      "loss": 0.8732,
+      "loss": 0.8799,
       "step": 64170
     },
     {
       "epoch": 192.73,
-      "grad_norm": 5.366335391998291,
+      "grad_norm": 6.587234973907471,
       "learning_rate": 3.575575575575576e-06,
-      "loss": 0.9557,
+      "loss": 0.9591,
       "step": 64180
     },
     {
       "epoch": 192.76,
-      "grad_norm": 5.5560784339904785,
+      "grad_norm": 10.343474388122559,
       "learning_rate": 3.5745745745745746e-06,
-      "loss": 0.8957,
+      "loss": 0.8912,
       "step": 64190
     },
     {
       "epoch": 192.79,
-      "grad_norm": 5.447986602783203,
+      "grad_norm": 5.517196178436279,
       "learning_rate": 3.573573573573574e-06,
-      "loss": 0.9253,
+      "loss": 0.9175,
       "step": 64200
     },
     {
       "epoch": 192.82,
-      "grad_norm": 5.428342819213867,
+      "grad_norm": 6.708267688751221,
       "learning_rate": 3.572572572572573e-06,
-      "loss": 0.9957,
+      "loss": 0.9901,
       "step": 64210
     },
     {
       "epoch": 192.85,
-      "grad_norm": 6.779325485229492,
+      "grad_norm": 8.799589157104492,
       "learning_rate": 3.571571571571572e-06,
-      "loss": 0.9228,
+      "loss": 0.9271,
       "step": 64220
     },
     {
       "epoch": 192.88,
-      "grad_norm": 5.3189778327941895,
+      "grad_norm": 5.718320846557617,
       "learning_rate": 3.5705705705705706e-06,
-      "loss": 1.0098,
+      "loss": 1.0053,
       "step": 64230
     },
     {
       "epoch": 192.91,
-      "grad_norm": 10.001474380493164,
+      "grad_norm": 10.05222225189209,
       "learning_rate": 3.5695695695695697e-06,
-      "loss": 0.9518,
+      "loss": 0.9514,
       "step": 64240
     },
     {
       "epoch": 192.94,
-      "grad_norm": 6.578002452850342,
+      "grad_norm": 7.258294582366943,
       "learning_rate": 3.568568568568569e-06,
-      "loss": 1.005,
+      "loss": 1.001,
       "step": 64250
     },
     {
       "epoch": 192.97,
-      "grad_norm": 12.511248588562012,
+      "grad_norm": 7.690788745880127,
       "learning_rate": 3.567567567567568e-06,
-      "loss": 0.9657,
+      "loss": 0.9665,
       "step": 64260
     },
     {
       "epoch": 193.0,
-      "eval_accuracy": 0.8243,
-      "eval_loss": 0.5805965065956116,
-      "eval_runtime": 5.4403,
-      "eval_samples_per_second": 1838.14,
-      "eval_steps_per_second": 7.353,
+      "eval_accuracy": 0.8246,
+      "eval_loss": 0.5783727765083313,
+      "eval_runtime": 5.3783,
+      "eval_samples_per_second": 1859.321,
+      "eval_steps_per_second": 7.437,
       "step": 64269
     },
     {
       "epoch": 193.0,
-      "grad_norm": 6.194021224975586,
+      "grad_norm": 6.5794219970703125,
       "learning_rate": 3.5665665665665667e-06,
-      "loss": 1.0685,
+      "loss": 1.0739,
       "step": 64270
     },
     {
       "epoch": 193.03,
-      "grad_norm": 5.340394020080566,
+      "grad_norm": 5.895750999450684,
       "learning_rate": 3.5655655655655657e-06,
-      "loss": 0.9617,
+      "loss": 0.9631,
       "step": 64280
     },
     {
       "epoch": 193.06,
-      "grad_norm": 7.0331926345825195,
+      "grad_norm": 6.9944748878479,
       "learning_rate": 3.5645645645645647e-06,
-      "loss": 0.9226,
+      "loss": 0.9311,
       "step": 64290
     },
     {
       "epoch": 193.09,
-      "grad_norm": 5.713209629058838,
+      "grad_norm": 5.432283878326416,
       "learning_rate": 3.563563563563564e-06,
-      "loss": 0.9846,
+      "loss": 0.9804,
       "step": 64300
     },
     {
       "epoch": 193.12,
-      "grad_norm": 4.6061248779296875,
+      "grad_norm": 5.111244201660156,
       "learning_rate": 3.5625625625625628e-06,
-      "loss": 1.0323,
+      "loss": 1.0319,
       "step": 64310
     },
     {
       "epoch": 193.15,
-      "grad_norm": 6.705684185028076,
+      "grad_norm": 6.428986549377441,
       "learning_rate": 3.561561561561562e-06,
-      "loss": 0.9481,
+      "loss": 0.948,
       "step": 64320
     },
     {
       "epoch": 193.18,
-      "grad_norm": 14.938161849975586,
+      "grad_norm": 12.419626235961914,
       "learning_rate": 3.560560560560561e-06,
-      "loss": 0.9338,
+      "loss": 0.9292,
       "step": 64330
     },
     {
       "epoch": 193.21,
-      "grad_norm": 6.611726760864258,
+      "grad_norm": 8.314090728759766,
       "learning_rate": 3.55955955955956e-06,
-      "loss": 0.9627,
+      "loss": 0.9649,
       "step": 64340
     },
     {
       "epoch": 193.24,
-      "grad_norm": 6.737940311431885,
+      "grad_norm": 9.138510704040527,
       "learning_rate": 3.5585585585585584e-06,
-      "loss": 0.9978,
+      "loss": 0.9933,
       "step": 64350
     },
     {
       "epoch": 193.27,
-      "grad_norm": 7.016870498657227,
+      "grad_norm": 7.428427696228027,
       "learning_rate": 3.557557557557558e-06,
-      "loss": 0.8937,
+      "loss": 0.9025,
       "step": 64360
     },
     {
       "epoch": 193.3,
-      "grad_norm": 8.270951271057129,
+      "grad_norm": 8.646157264709473,
       "learning_rate": 3.556556556556557e-06,
-      "loss": 0.9405,
+      "loss": 0.9411,
       "step": 64370
     },
     {
       "epoch": 193.33,
-      "grad_norm": 5.626846790313721,
+      "grad_norm": 4.9220404624938965,
       "learning_rate": 3.555555555555556e-06,
-      "loss": 0.9282,
+      "loss": 0.926,
       "step": 64380
     },
     {
       "epoch": 193.36,
-      "grad_norm": 6.768457889556885,
+      "grad_norm": 7.204169750213623,
       "learning_rate": 3.554554554554555e-06,
-      "loss": 0.8706,
+      "loss": 0.8696,
       "step": 64390
     },
     {
       "epoch": 193.39,
-      "grad_norm": 5.760232448577881,
+      "grad_norm": 5.590950012207031,
       "learning_rate": 3.5535535535535535e-06,
-      "loss": 0.9186,
+      "loss": 0.9118,
       "step": 64400
     },
     {
       "epoch": 193.42,
-      "grad_norm": 6.616115570068359,
+      "grad_norm": 6.444925308227539,
       "learning_rate": 3.552552552552553e-06,
-      "loss": 0.9247,
+      "loss": 0.9284,
       "step": 64410
     },
     {
       "epoch": 193.45,
-      "grad_norm": 6.447544097900391,
+      "grad_norm": 7.865789890289307,
       "learning_rate": 3.551551551551552e-06,
-      "loss": 0.9872,
+      "loss": 0.9883,
       "step": 64420
     },
     {
       "epoch": 193.48,
-      "grad_norm": 6.897136688232422,
+      "grad_norm": 7.061901092529297,
       "learning_rate": 3.550550550550551e-06,
-      "loss": 0.9542,
+      "loss": 0.9589,
       "step": 64430
     },
     {
       "epoch": 193.51,
-      "grad_norm": 8.468799591064453,
+      "grad_norm": 7.189478397369385,
       "learning_rate": 3.5495495495495496e-06,
-      "loss": 0.9179,
+      "loss": 0.9199,
       "step": 64440
     },
     {
       "epoch": 193.54,
-      "grad_norm": 9.021018028259277,
+      "grad_norm": 10.328020095825195,
       "learning_rate": 3.5485485485485486e-06,
-      "loss": 0.8913,
+      "loss": 0.8767,
       "step": 64450
     },
     {
       "epoch": 193.57,
-      "grad_norm": 7.557258605957031,
+      "grad_norm": 5.339158058166504,
       "learning_rate": 3.547547547547548e-06,
-      "loss": 0.908,
+      "loss": 0.9087,
       "step": 64460
     },
     {
       "epoch": 193.6,
-      "grad_norm": 7.520648956298828,
+      "grad_norm": 7.2761969566345215,
       "learning_rate": 3.546546546546547e-06,
-      "loss": 0.8966,
+      "loss": 0.8889,
       "step": 64470
     },
     {
       "epoch": 193.63,
-      "grad_norm": 7.482933521270752,
+      "grad_norm": 9.365469932556152,
       "learning_rate": 3.5455455455455457e-06,
-      "loss": 0.9092,
+      "loss": 0.9155,
       "step": 64480
     },
     {
       "epoch": 193.66,
-      "grad_norm": 7.440866947174072,
+      "grad_norm": 8.635586738586426,
       "learning_rate": 3.5445445445445447e-06,
-      "loss": 0.9727,
+      "loss": 0.9722,
       "step": 64490
     },
     {
       "epoch": 193.69,
-      "grad_norm": 5.267864227294922,
+      "grad_norm": 8.014188766479492,
       "learning_rate": 3.5435435435435437e-06,
-      "loss": 0.9785,
+      "loss": 0.975,
       "step": 64500
     },
     {
       "epoch": 193.72,
-      "grad_norm": 6.3476433753967285,
+      "grad_norm": 8.6399507522583,
       "learning_rate": 3.542542542542543e-06,
-      "loss": 0.9551,
+      "loss": 0.9443,
       "step": 64510
     },
     {
       "epoch": 193.75,
-      "grad_norm": 8.087379455566406,
+      "grad_norm": 8.435717582702637,
       "learning_rate": 3.5415415415415417e-06,
-      "loss": 0.9378,
+      "loss": 0.9324,
       "step": 64520
     },
     {
       "epoch": 193.78,
-      "grad_norm": 7.711241722106934,
+      "grad_norm": 7.26228141784668,
       "learning_rate": 3.5405405405405408e-06,
-      "loss": 0.9507,
+      "loss": 0.9484,
       "step": 64530
     },
     {
       "epoch": 193.81,
-      "grad_norm": 7.453726768493652,
+      "grad_norm": 7.951531410217285,
       "learning_rate": 3.5395395395395398e-06,
-      "loss": 0.9315,
+      "loss": 0.9416,
       "step": 64540
     },
     {
       "epoch": 193.84,
-      "grad_norm": 7.166509628295898,
+      "grad_norm": 6.004051208496094,
       "learning_rate": 3.5385385385385392e-06,
-      "loss": 0.8906,
+      "loss": 0.9008,
       "step": 64550
     },
     {
       "epoch": 193.87,
-      "grad_norm": 6.427428722381592,
+      "grad_norm": 6.407548904418945,
       "learning_rate": 3.5375375375375374e-06,
-      "loss": 0.9807,
+      "loss": 0.9747,
       "step": 64560
     },
     {
       "epoch": 193.9,
-      "grad_norm": 5.590991973876953,
+      "grad_norm": 7.067380905151367,
       "learning_rate": 3.536536536536537e-06,
-      "loss": 0.9851,
+      "loss": 0.9812,
       "step": 64570
     },
     {
       "epoch": 193.93,
-      "grad_norm": 7.2621684074401855,
+      "grad_norm": 6.942722797393799,
       "learning_rate": 3.535535535535536e-06,
-      "loss": 1.0268,
+      "loss": 1.0298,
       "step": 64580
     },
     {
       "epoch": 193.96,
-      "grad_norm": 7.558947563171387,
+      "grad_norm": 9.51594066619873,
       "learning_rate": 3.534534534534535e-06,
-      "loss": 0.9583,
+      "loss": 0.9435,
       "step": 64590
     },
     {
       "epoch": 193.99,
-      "grad_norm": 7.2830491065979,
+      "grad_norm": 8.088098526000977,
       "learning_rate": 3.5335335335335335e-06,
-      "loss": 0.9865,
+      "loss": 0.9842,
       "step": 64600
     },
     {
       "epoch": 194.0,
-      "eval_accuracy": 0.8225,
-      "eval_loss": 0.5838760733604431,
-      "eval_runtime": 5.4562,
-      "eval_samples_per_second": 1832.768,
-      "eval_steps_per_second": 7.331,
+      "eval_accuracy": 0.8241,
+      "eval_loss": 0.5830530524253845,
+      "eval_runtime": 5.4126,
+      "eval_samples_per_second": 1847.531,
+      "eval_steps_per_second": 7.39,
       "step": 64602
     },
     {
       "epoch": 194.02,
-      "grad_norm": 6.215984344482422,
+      "grad_norm": 5.951772212982178,
       "learning_rate": 3.532532532532533e-06,
-      "loss": 1.0317,
+      "loss": 1.0227,
       "step": 64610
     },
     {
       "epoch": 194.05,
-      "grad_norm": 6.518527507781982,
+      "grad_norm": 6.51089334487915,
       "learning_rate": 3.531531531531532e-06,
-      "loss": 0.9302,
+      "loss": 0.9258,
       "step": 64620
     },
     {
       "epoch": 194.08,
-      "grad_norm": 5.257604598999023,
+      "grad_norm": 5.301957607269287,
       "learning_rate": 3.530530530530531e-06,
-      "loss": 0.9183,
+      "loss": 0.9146,
       "step": 64630
     },
     {
       "epoch": 194.11,
-      "grad_norm": 9.397860527038574,
+      "grad_norm": 11.603909492492676,
       "learning_rate": 3.5295295295295295e-06,
-      "loss": 0.9553,
+      "loss": 0.9443,
       "step": 64640
     },
     {
       "epoch": 194.14,
-      "grad_norm": 6.506763458251953,
+      "grad_norm": 7.710113048553467,
       "learning_rate": 3.5285285285285286e-06,
-      "loss": 0.9173,
+      "loss": 0.9163,
       "step": 64650
     },
     {
       "epoch": 194.17,
-      "grad_norm": 6.8301825523376465,
+      "grad_norm": 7.3544464111328125,
       "learning_rate": 3.527527527527528e-06,
-      "loss": 0.9573,
+      "loss": 0.9657,
       "step": 64660
     },
     {
       "epoch": 194.2,
-      "grad_norm": 5.90817403793335,
+      "grad_norm": 6.559677600860596,
       "learning_rate": 3.526526526526527e-06,
-      "loss": 0.8963,
+      "loss": 0.8973,
       "step": 64670
     },
     {
       "epoch": 194.23,
-      "grad_norm": 7.149304389953613,
+      "grad_norm": 4.850852012634277,
       "learning_rate": 3.525525525525526e-06,
-      "loss": 0.8986,
+      "loss": 0.8928,
       "step": 64680
     },
     {
       "epoch": 194.26,
-      "grad_norm": 7.136353969573975,
+      "grad_norm": 4.759112358093262,
       "learning_rate": 3.5245245245245246e-06,
-      "loss": 0.8864,
+      "loss": 0.8841,
       "step": 64690
     },
     {
       "epoch": 194.29,
-      "grad_norm": 5.898740291595459,
+      "grad_norm": 6.694949626922607,
       "learning_rate": 3.5235235235235237e-06,
-      "loss": 0.9696,
+      "loss": 0.9663,
       "step": 64700
     },
     {
       "epoch": 194.32,
-      "grad_norm": 5.678613185882568,
+      "grad_norm": 7.855728626251221,
       "learning_rate": 3.522522522522523e-06,
-      "loss": 0.946,
+      "loss": 0.9391,
       "step": 64710
     },
     {
       "epoch": 194.35,
-      "grad_norm": 5.471475601196289,
+      "grad_norm": 6.160037517547607,
       "learning_rate": 3.521521521521522e-06,
-      "loss": 0.9063,
+      "loss": 0.9105,
       "step": 64720
     },
     {
       "epoch": 194.38,
-      "grad_norm": 7.718780040740967,
+      "grad_norm": 6.395085334777832,
       "learning_rate": 3.5205205205205207e-06,
-      "loss": 0.9132,
+      "loss": 0.9062,
       "step": 64730
     },
     {
       "epoch": 194.41,
-      "grad_norm": 5.420345306396484,
+      "grad_norm": 6.914565086364746,
       "learning_rate": 3.5195195195195197e-06,
-      "loss": 0.9601,
+      "loss": 0.9685,
       "step": 64740
     },
     {
       "epoch": 194.44,
-      "grad_norm": 6.1703572273254395,
+      "grad_norm": 7.157896518707275,
       "learning_rate": 3.5185185185185187e-06,
-      "loss": 0.9826,
+      "loss": 0.9865,
       "step": 64750
     },
     {
       "epoch": 194.47,
-      "grad_norm": 5.648353099822998,
+      "grad_norm": 8.20499038696289,
       "learning_rate": 3.517517517517518e-06,
-      "loss": 0.9794,
+      "loss": 0.9744,
       "step": 64760
     },
     {
       "epoch": 194.5,
-      "grad_norm": 9.022456169128418,
+      "grad_norm": 10.187125205993652,
       "learning_rate": 3.5165165165165168e-06,
-      "loss": 0.9822,
+      "loss": 0.9805,
       "step": 64770
     },
     {
       "epoch": 194.53,
-      "grad_norm": 5.214541435241699,
+      "grad_norm": 6.3292622566223145,
       "learning_rate": 3.515515515515516e-06,
-      "loss": 0.9582,
+      "loss": 0.9398,
       "step": 64780
     },
     {
       "epoch": 194.56,
-      "grad_norm": 5.824596881866455,
+      "grad_norm": 7.087769985198975,
       "learning_rate": 3.514514514514515e-06,
-      "loss": 0.9909,
+      "loss": 0.9968,
       "step": 64790
     },
     {
       "epoch": 194.59,
-      "grad_norm": 6.3842926025390625,
+      "grad_norm": 8.639760971069336,
       "learning_rate": 3.513513513513514e-06,
-      "loss": 0.9845,
+      "loss": 0.9792,
       "step": 64800
     },
     {
       "epoch": 194.62,
-      "grad_norm": 4.825592041015625,
+      "grad_norm": 6.038487434387207,
       "learning_rate": 3.5125125125125124e-06,
-      "loss": 0.9064,
+      "loss": 0.9088,
       "step": 64810
     },
     {
       "epoch": 194.65,
-      "grad_norm": 4.923450946807861,
+      "grad_norm": 6.510197639465332,
       "learning_rate": 3.511511511511512e-06,
-      "loss": 0.951,
+      "loss": 0.9472,
       "step": 64820
     },
     {
       "epoch": 194.68,
-      "grad_norm": 5.440247535705566,
+      "grad_norm": 7.498666763305664,
       "learning_rate": 3.510510510510511e-06,
-      "loss": 0.915,
+      "loss": 0.9069,
       "step": 64830
     },
     {
       "epoch": 194.71,
-      "grad_norm": 7.426842212677002,
+      "grad_norm": 8.058786392211914,
       "learning_rate": 3.50950950950951e-06,
-      "loss": 0.9462,
+      "loss": 0.9385,
       "step": 64840
     },
     {
       "epoch": 194.74,
-      "grad_norm": 6.606391906738281,
+      "grad_norm": 9.28315544128418,
       "learning_rate": 3.5085085085085085e-06,
-      "loss": 1.0147,
+      "loss": 0.9931,
       "step": 64850
     },
     {
       "epoch": 194.77,
-      "grad_norm": 7.025200366973877,
+      "grad_norm": 6.2826619148254395,
       "learning_rate": 3.5075075075075075e-06,
-      "loss": 0.925,
+      "loss": 0.9166,
       "step": 64860
     },
     {
       "epoch": 194.8,
-      "grad_norm": 5.132568359375,
+      "grad_norm": 7.395730495452881,
       "learning_rate": 3.506506506506507e-06,
-      "loss": 0.9501,
+      "loss": 0.9456,
       "step": 64870
     },
     {
       "epoch": 194.83,
-      "grad_norm": 5.869955062866211,
+      "grad_norm": 5.464653968811035,
       "learning_rate": 3.505505505505506e-06,
-      "loss": 0.9583,
+      "loss": 0.965,
       "step": 64880
     },
     {
       "epoch": 194.86,
-      "grad_norm": 8.706028938293457,
+      "grad_norm": 12.247793197631836,
       "learning_rate": 3.5045045045045046e-06,
-      "loss": 1.036,
+      "loss": 1.0337,
       "step": 64890
     },
     {
       "epoch": 194.89,
-      "grad_norm": 6.917304992675781,
+      "grad_norm": 7.048732280731201,
       "learning_rate": 3.5035035035035036e-06,
-      "loss": 0.9902,
+      "loss": 0.9735,
       "step": 64900
     },
     {
       "epoch": 194.92,
-      "grad_norm": 5.493220329284668,
+      "grad_norm": 6.336861610412598,
       "learning_rate": 3.5025025025025026e-06,
-      "loss": 0.9592,
+      "loss": 0.9538,
       "step": 64910
     },
     {
       "epoch": 194.95,
-      "grad_norm": 6.807432174682617,
+      "grad_norm": 7.697917461395264,
       "learning_rate": 3.501501501501502e-06,
-      "loss": 0.9329,
+      "loss": 0.9204,
       "step": 64920
     },
     {
       "epoch": 194.98,
-      "grad_norm": 6.404989719390869,
+      "grad_norm": 7.685586452484131,
       "learning_rate": 3.500500500500501e-06,
-      "loss": 0.9642,
+      "loss": 0.9672,
       "step": 64930
     },
     {
       "epoch": 195.0,
-      "eval_accuracy": 0.8219,
-      "eval_loss": 0.5850241184234619,
-      "eval_runtime": 5.3443,
-      "eval_samples_per_second": 1871.16,
-      "eval_steps_per_second": 7.485,
+      "eval_accuracy": 0.825,
+      "eval_loss": 0.5826992392539978,
+      "eval_runtime": 5.3343,
+      "eval_samples_per_second": 1874.676,
+      "eval_steps_per_second": 7.499,
       "step": 64935
     },
     {
       "epoch": 195.02,
-      "grad_norm": 6.296675682067871,
+      "grad_norm": 7.683665752410889,
       "learning_rate": 3.4994994994994997e-06,
-      "loss": 1.0064,
+      "loss": 0.9741,
       "step": 64940
     },
     {
       "epoch": 195.05,
-      "grad_norm": 6.159897327423096,
+      "grad_norm": 5.855304718017578,
       "learning_rate": 3.4984984984984987e-06,
-      "loss": 0.8627,
+      "loss": 0.8583,
       "step": 64950
     },
     {
       "epoch": 195.08,
-      "grad_norm": 6.329672336578369,
+      "grad_norm": 6.137430191040039,
       "learning_rate": 3.4974974974974977e-06,
-      "loss": 0.9843,
+      "loss": 0.9813,
       "step": 64960
     },
     {
       "epoch": 195.11,
-      "grad_norm": 6.9494452476501465,
+      "grad_norm": 7.776120662689209,
       "learning_rate": 3.496496496496497e-06,
-      "loss": 0.9953,
+      "loss": 0.9951,
       "step": 64970
     },
     {
       "epoch": 195.14,
-      "grad_norm": 5.167768955230713,
+      "grad_norm": 4.700627326965332,
       "learning_rate": 3.4954954954954957e-06,
-      "loss": 0.8822,
+      "loss": 0.8789,
       "step": 64980
     },
     {
       "epoch": 195.17,
-      "grad_norm": 5.712154865264893,
+      "grad_norm": 8.73577880859375,
       "learning_rate": 3.4944944944944948e-06,
-      "loss": 0.9813,
+      "loss": 0.9799,
       "step": 64990
     },
     {
       "epoch": 195.2,
-      "grad_norm": 7.48964786529541,
+      "grad_norm": 6.254727363586426,
       "learning_rate": 3.4934934934934938e-06,
-      "loss": 0.8953,
+      "loss": 0.8987,
       "step": 65000
     },
     {
       "epoch": 195.23,
-      "grad_norm": 6.811525821685791,
+      "grad_norm": 7.0531325340271,
       "learning_rate": 3.492492492492493e-06,
-      "loss": 0.9699,
+      "loss": 0.9805,
       "step": 65010
     },
     {
       "epoch": 195.26,
-      "grad_norm": 7.303476810455322,
+      "grad_norm": 7.177323341369629,
       "learning_rate": 3.4914914914914914e-06,
-      "loss": 0.9289,
+      "loss": 0.922,
       "step": 65020
     },
     {
       "epoch": 195.29,
-      "grad_norm": 5.510158061981201,
+      "grad_norm": 5.856313705444336,
       "learning_rate": 3.490490490490491e-06,
-      "loss": 0.992,
+      "loss": 0.9887,
       "step": 65030
     },
     {
       "epoch": 195.32,
-      "grad_norm": 8.767537117004395,
+      "grad_norm": 12.345643997192383,
       "learning_rate": 3.48948948948949e-06,
-      "loss": 0.8892,
+      "loss": 0.8847,
       "step": 65040
     },
     {
       "epoch": 195.35,
-      "grad_norm": 8.583027839660645,
+      "grad_norm": 7.534376621246338,
       "learning_rate": 3.488488488488489e-06,
-      "loss": 0.9499,
+      "loss": 0.9491,
       "step": 65050
     },
     {
       "epoch": 195.38,
-      "grad_norm": 5.04366397857666,
+      "grad_norm": 5.6023101806640625,
       "learning_rate": 3.4874874874874875e-06,
-      "loss": 0.897,
+      "loss": 0.8919,
       "step": 65060
     },
     {
       "epoch": 195.41,
-      "grad_norm": 7.101098537445068,
+      "grad_norm": 7.501373767852783,
       "learning_rate": 3.4864864864864865e-06,
-      "loss": 0.997,
+      "loss": 1.0024,
       "step": 65070
     },
     {
       "epoch": 195.44,
-      "grad_norm": 10.924664497375488,
+      "grad_norm": 10.297907829284668,
       "learning_rate": 3.485485485485486e-06,
-      "loss": 0.973,
+      "loss": 0.9646,
       "step": 65080
     },
     {
       "epoch": 195.47,
-      "grad_norm": 9.738886833190918,
+      "grad_norm": 8.763750076293945,
       "learning_rate": 3.484484484484485e-06,
-      "loss": 0.9574,
+      "loss": 0.9543,
       "step": 65090
     },
     {
       "epoch": 195.5,
-      "grad_norm": 7.146059513092041,
+      "grad_norm": 5.39033842086792,
       "learning_rate": 3.4834834834834835e-06,
-      "loss": 1.0089,
+      "loss": 1.0014,
       "step": 65100
     },
     {
       "epoch": 195.53,
-      "grad_norm": 7.540902614593506,
+      "grad_norm": 5.567760467529297,
       "learning_rate": 3.4824824824824826e-06,
-      "loss": 1.0117,
+      "loss": 1.0022,
       "step": 65110
     },
     {
       "epoch": 195.56,
-      "grad_norm": 9.438485145568848,
+      "grad_norm": 8.712343215942383,
       "learning_rate": 3.481481481481482e-06,
-      "loss": 0.8915,
+      "loss": 0.891,
       "step": 65120
     },
     {
       "epoch": 195.59,
-      "grad_norm": 5.575406551361084,
+      "grad_norm": 5.04828405380249,
       "learning_rate": 3.480480480480481e-06,
-      "loss": 0.9052,
+      "loss": 0.9095,
       "step": 65130
     },
     {
       "epoch": 195.62,
-      "grad_norm": 7.179344654083252,
+      "grad_norm": 7.784473896026611,
       "learning_rate": 3.4794794794794796e-06,
-      "loss": 0.9779,
+      "loss": 0.9838,
       "step": 65140
     },
     {
       "epoch": 195.65,
-      "grad_norm": 10.02873420715332,
+      "grad_norm": 7.50624418258667,
       "learning_rate": 3.4784784784784786e-06,
-      "loss": 0.8761,
+      "loss": 0.88,
       "step": 65150
     },
     {
       "epoch": 195.68,
-      "grad_norm": 6.868290424346924,
+      "grad_norm": 8.999715805053711,
       "learning_rate": 3.4774774774774776e-06,
-      "loss": 0.9456,
+      "loss": 0.9408,
       "step": 65160
     },
     {
       "epoch": 195.71,
-      "grad_norm": 7.79679012298584,
+      "grad_norm": 6.4573140144348145,
       "learning_rate": 3.476476476476477e-06,
-      "loss": 1.0556,
+      "loss": 1.0562,
       "step": 65170
     },
     {
       "epoch": 195.74,
-      "grad_norm": 6.2746686935424805,
+      "grad_norm": 10.534112930297852,
       "learning_rate": 3.475475475475476e-06,
-      "loss": 0.9623,
+      "loss": 0.9566,
       "step": 65180
     },
     {
       "epoch": 195.77,
-      "grad_norm": 7.8665690422058105,
+      "grad_norm": 8.076118469238281,
       "learning_rate": 3.4744744744744747e-06,
-      "loss": 0.8946,
+      "loss": 0.8925,
       "step": 65190
     },
     {
       "epoch": 195.8,
-      "grad_norm": 8.22524642944336,
+      "grad_norm": 8.375502586364746,
       "learning_rate": 3.4734734734734737e-06,
-      "loss": 1.0149,
+      "loss": 1.0094,
       "step": 65200
     },
     {
       "epoch": 195.83,
-      "grad_norm": 5.470314979553223,
+      "grad_norm": 4.891530513763428,
       "learning_rate": 3.4724724724724727e-06,
-      "loss": 0.9386,
+      "loss": 0.9451,
       "step": 65210
     },
     {
       "epoch": 195.86,
-      "grad_norm": 7.7399773597717285,
+      "grad_norm": 8.53266716003418,
       "learning_rate": 3.471471471471472e-06,
-      "loss": 0.9665,
+      "loss": 0.9483,
       "step": 65220
     },
     {
       "epoch": 195.89,
-      "grad_norm": 9.8233003616333,
+      "grad_norm": 10.301980018615723,
       "learning_rate": 3.4704704704704708e-06,
-      "loss": 0.9195,
+      "loss": 0.9269,
       "step": 65230
     },
     {
       "epoch": 195.92,
-      "grad_norm": 6.0244460105896,
+      "grad_norm": 5.741480827331543,
       "learning_rate": 3.46946946946947e-06,
-      "loss": 0.9431,
+      "loss": 0.9438,
       "step": 65240
     },
     {
       "epoch": 195.95,
-      "grad_norm": 6.786837100982666,
+      "grad_norm": 5.6558403968811035,
       "learning_rate": 3.468468468468469e-06,
-      "loss": 0.9108,
+      "loss": 0.9213,
       "step": 65250
     },
     {
       "epoch": 195.98,
-      "grad_norm": 7.457739353179932,
+      "grad_norm": 8.549616813659668,
       "learning_rate": 3.467467467467468e-06,
-      "loss": 0.9839,
+      "loss": 0.9824,
       "step": 65260
     },
     {
       "epoch": 196.0,
-      "eval_accuracy": 0.8246,
-      "eval_loss": 0.5814548134803772,
-      "eval_runtime": 5.4965,
-      "eval_samples_per_second": 1819.329,
-      "eval_steps_per_second": 7.277,
+      "eval_accuracy": 0.8244,
+      "eval_loss": 0.5777956247329712,
+      "eval_runtime": 5.5013,
+      "eval_samples_per_second": 1817.745,
+      "eval_steps_per_second": 7.271,
       "step": 65268
     },
     {
       "epoch": 196.01,
-      "grad_norm": 8.353056907653809,
+      "grad_norm": 6.348392486572266,
       "learning_rate": 3.4664664664664664e-06,
-      "loss": 1.0061,
+      "loss": 1.0123,
       "step": 65270
     },
     {
       "epoch": 196.04,
-      "grad_norm": 5.094268798828125,
+      "grad_norm": 7.4910197257995605,
       "learning_rate": 3.465465465465466e-06,
-      "loss": 0.9761,
+      "loss": 0.9714,
       "step": 65280
     },
     {
       "epoch": 196.07,
-      "grad_norm": 7.1268463134765625,
+      "grad_norm": 10.852710723876953,
       "learning_rate": 3.464464464464465e-06,
-      "loss": 1.0024,
+      "loss": 0.9862,
       "step": 65290
     },
     {
       "epoch": 196.1,
-      "grad_norm": 6.37305212020874,
+      "grad_norm": 6.7594170570373535,
       "learning_rate": 3.463463463463464e-06,
-      "loss": 1.0329,
+      "loss": 1.0314,
       "step": 65300
     },
     {
       "epoch": 196.13,
-      "grad_norm": 6.686522960662842,
+      "grad_norm": 5.433690547943115,
       "learning_rate": 3.4624624624624625e-06,
-      "loss": 0.9833,
+      "loss": 0.9803,
       "step": 65310
     },
     {
       "epoch": 196.16,
-      "grad_norm": 5.428120136260986,
+      "grad_norm": 6.628762245178223,
       "learning_rate": 3.4614614614614615e-06,
-      "loss": 1.0312,
+      "loss": 1.0331,
       "step": 65320
     },
     {
       "epoch": 196.19,
-      "grad_norm": 6.451683521270752,
+      "grad_norm": 6.386612892150879,
       "learning_rate": 3.460460460460461e-06,
-      "loss": 1.0103,
+      "loss": 1.0153,
       "step": 65330
     },
     {
       "epoch": 196.22,
-      "grad_norm": 7.041833400726318,
+      "grad_norm": 8.768787384033203,
       "learning_rate": 3.45945945945946e-06,
-      "loss": 1.0212,
+      "loss": 1.0249,
       "step": 65340
     },
     {
       "epoch": 196.25,
-      "grad_norm": 6.129427433013916,
+      "grad_norm": 7.1480607986450195,
       "learning_rate": 3.4584584584584586e-06,
-      "loss": 0.913,
+      "loss": 0.9155,
       "step": 65350
     },
     {
       "epoch": 196.28,
-      "grad_norm": 8.157512664794922,
+      "grad_norm": 6.683876991271973,
       "learning_rate": 3.4574574574574576e-06,
-      "loss": 0.9049,
+      "loss": 0.8993,
       "step": 65360
     },
     {
       "epoch": 196.31,
-      "grad_norm": 8.397013664245605,
+      "grad_norm": 10.570980072021484,
       "learning_rate": 3.4564564564564566e-06,
-      "loss": 0.9338,
+      "loss": 0.9314,
       "step": 65370
     },
     {
       "epoch": 196.34,
-      "grad_norm": 6.023127555847168,
+      "grad_norm": 8.718035697937012,
       "learning_rate": 3.455455455455456e-06,
-      "loss": 0.9004,
+      "loss": 0.9041,
       "step": 65380
     },
     {
       "epoch": 196.37,
-      "grad_norm": 6.677743434906006,
+      "grad_norm": 6.019381046295166,
       "learning_rate": 3.4544544544544546e-06,
-      "loss": 0.8952,
+      "loss": 0.8827,
       "step": 65390
     },
     {
       "epoch": 196.4,
-      "grad_norm": 5.246835708618164,
+      "grad_norm": 6.695023536682129,
       "learning_rate": 3.4534534534534537e-06,
-      "loss": 0.9616,
+      "loss": 0.9618,
       "step": 65400
     },
     {
       "epoch": 196.43,
-      "grad_norm": 5.344621181488037,
+      "grad_norm": 5.560020446777344,
       "learning_rate": 3.4524524524524527e-06,
-      "loss": 0.9553,
+      "loss": 0.9479,
       "step": 65410
     },
     {
       "epoch": 196.46,
-      "grad_norm": 5.173955917358398,
+      "grad_norm": 5.386765480041504,
       "learning_rate": 3.4514514514514517e-06,
-      "loss": 0.9226,
+      "loss": 0.921,
       "step": 65420
     },
     {
       "epoch": 196.49,
-      "grad_norm": 6.328470230102539,
+      "grad_norm": 8.096517562866211,
       "learning_rate": 3.4504504504504503e-06,
-      "loss": 0.9526,
+      "loss": 0.9485,
       "step": 65430
     },
     {
       "epoch": 196.52,
-      "grad_norm": 6.167508125305176,
+      "grad_norm": 7.633049011230469,
       "learning_rate": 3.4494494494494497e-06,
-      "loss": 0.9946,
+      "loss": 0.9885,
       "step": 65440
     },
     {
       "epoch": 196.55,
-      "grad_norm": 6.668299198150635,
+      "grad_norm": 7.070896148681641,
       "learning_rate": 3.4484484484484488e-06,
-      "loss": 0.9924,
+      "loss": 0.9842,
       "step": 65450
     },
     {
       "epoch": 196.58,
-      "grad_norm": 7.8197021484375,
+      "grad_norm": 8.481072425842285,
       "learning_rate": 3.4474474474474478e-06,
-      "loss": 0.9016,
+      "loss": 0.9001,
       "step": 65460
     },
     {
       "epoch": 196.61,
-      "grad_norm": 6.635308265686035,
+      "grad_norm": 6.502707481384277,
       "learning_rate": 3.446446446446447e-06,
-      "loss": 0.9292,
+      "loss": 0.9362,
       "step": 65470
     },
     {
       "epoch": 196.64,
-      "grad_norm": 6.540359020233154,
+      "grad_norm": 7.72848653793335,
       "learning_rate": 3.4454454454454454e-06,
-      "loss": 0.9235,
+      "loss": 0.9343,
       "step": 65480
     },
     {
       "epoch": 196.67,
-      "grad_norm": 7.316467761993408,
+      "grad_norm": 5.801799774169922,
       "learning_rate": 3.444444444444445e-06,
-      "loss": 0.9069,
+      "loss": 0.9101,
       "step": 65490
     },
     {
       "epoch": 196.7,
-      "grad_norm": 9.187167167663574,
+      "grad_norm": 7.184073448181152,
       "learning_rate": 3.443443443443444e-06,
-      "loss": 0.9387,
+      "loss": 0.9404,
       "step": 65500
     },
     {
       "epoch": 196.73,
-      "grad_norm": 6.829257011413574,
+      "grad_norm": 6.281429290771484,
       "learning_rate": 3.442442442442443e-06,
-      "loss": 0.9723,
+      "loss": 0.9592,
       "step": 65510
     },
     {
       "epoch": 196.76,
-      "grad_norm": 4.397656440734863,
+      "grad_norm": 6.056782245635986,
       "learning_rate": 3.4414414414414415e-06,
-      "loss": 0.9511,
+      "loss": 0.9505,
       "step": 65520
     },
     {
       "epoch": 196.79,
-      "grad_norm": 11.452910423278809,
+      "grad_norm": 8.721957206726074,
       "learning_rate": 3.4404404404404405e-06,
-      "loss": 0.9518,
+      "loss": 0.9493,
       "step": 65530
     },
     {
       "epoch": 196.82,
-      "grad_norm": 5.808438301086426,
+      "grad_norm": 5.0055060386657715,
       "learning_rate": 3.43943943943944e-06,
-      "loss": 0.9626,
+      "loss": 0.9547,
       "step": 65540
     },
     {
       "epoch": 196.85,
-      "grad_norm": 7.496082305908203,
+      "grad_norm": 7.639198303222656,
       "learning_rate": 3.438438438438439e-06,
-      "loss": 0.9912,
+      "loss": 0.9869,
       "step": 65550
     },
     {
       "epoch": 196.88,
-      "grad_norm": 5.83450984954834,
+      "grad_norm": 6.849318981170654,
       "learning_rate": 3.4374374374374375e-06,
-      "loss": 1.0209,
+      "loss": 1.0296,
       "step": 65560
     },
     {
       "epoch": 196.91,
-      "grad_norm": 5.763981819152832,
+      "grad_norm": 5.248751163482666,
       "learning_rate": 3.4364364364364366e-06,
-      "loss": 0.9872,
+      "loss": 0.9892,
       "step": 65570
     },
     {
       "epoch": 196.94,
-      "grad_norm": 5.58937931060791,
+      "grad_norm": 6.361146450042725,
       "learning_rate": 3.435435435435436e-06,
-      "loss": 0.9369,
+      "loss": 0.9442,
       "step": 65580
     },
     {
       "epoch": 196.97,
-      "grad_norm": 9.824167251586914,
+      "grad_norm": 10.414185523986816,
       "learning_rate": 3.434434434434435e-06,
-      "loss": 0.9429,
+      "loss": 0.9362,
       "step": 65590
     },
     {
       "epoch": 197.0,
-      "grad_norm": 7.091148853302002,
+      "grad_norm": 8.057053565979004,
       "learning_rate": 3.4334334334334336e-06,
-      "loss": 0.999,
+      "loss": 0.9906,
       "step": 65600
     },
     {
       "epoch": 197.0,
-      "eval_accuracy": 0.8252,
-      "eval_loss": 0.5786702632904053,
-      "eval_runtime": 5.3045,
-      "eval_samples_per_second": 1885.194,
-      "eval_steps_per_second": 7.541,
+      "eval_accuracy": 0.8247,
+      "eval_loss": 0.5745610594749451,
+      "eval_runtime": 5.3012,
+      "eval_samples_per_second": 1886.35,
+      "eval_steps_per_second": 7.545,
       "step": 65601
     },
     {
       "epoch": 197.03,
-      "grad_norm": 6.921757221221924,
+      "grad_norm": 6.794538497924805,
       "learning_rate": 3.4324324324324326e-06,
-      "loss": 1.0718,
+      "loss": 1.0828,
       "step": 65610
     },
     {
       "epoch": 197.06,
-      "grad_norm": 5.95070219039917,
+      "grad_norm": 6.730975151062012,
       "learning_rate": 3.4314314314314316e-06,
-      "loss": 0.9165,
+      "loss": 0.9118,
       "step": 65620
     },
     {
       "epoch": 197.09,
-      "grad_norm": 8.587211608886719,
+      "grad_norm": 5.446534156799316,
       "learning_rate": 3.430430430430431e-06,
-      "loss": 0.9944,
+      "loss": 0.9818,
       "step": 65630
     },
     {
       "epoch": 197.12,
-      "grad_norm": 7.157227516174316,
+      "grad_norm": 8.303618431091309,
       "learning_rate": 3.4294294294294293e-06,
-      "loss": 0.9529,
+      "loss": 0.9542,
       "step": 65640
     },
     {
       "epoch": 197.15,
-      "grad_norm": 8.171802520751953,
+      "grad_norm": 6.884123802185059,
       "learning_rate": 3.4284284284284287e-06,
-      "loss": 0.9802,
+      "loss": 0.979,
       "step": 65650
     },
     {
       "epoch": 197.18,
-      "grad_norm": 6.770232200622559,
+      "grad_norm": 7.20119047164917,
       "learning_rate": 3.4274274274274277e-06,
-      "loss": 0.9769,
+      "loss": 0.9741,
       "step": 65660
     },
     {
       "epoch": 197.21,
-      "grad_norm": 6.770742893218994,
+      "grad_norm": 6.117996692657471,
       "learning_rate": 3.4264264264264267e-06,
-      "loss": 0.8906,
+      "loss": 0.8924,
       "step": 65670
     },
     {
       "epoch": 197.24,
-      "grad_norm": 5.558035373687744,
+      "grad_norm": 6.321722984313965,
       "learning_rate": 3.4254254254254253e-06,
-      "loss": 0.896,
+      "loss": 0.8962,
       "step": 65680
     },
     {
       "epoch": 197.27,
-      "grad_norm": 6.368505001068115,
+      "grad_norm": 7.171252727508545,
       "learning_rate": 3.4244244244244248e-06,
-      "loss": 0.9936,
+      "loss": 0.9847,
       "step": 65690
     },
     {
       "epoch": 197.3,
-      "grad_norm": 6.816822052001953,
+      "grad_norm": 8.965282440185547,
       "learning_rate": 3.423423423423424e-06,
-      "loss": 0.9363,
+      "loss": 0.9514,
       "step": 65700
     },
     {
       "epoch": 197.33,
-      "grad_norm": 5.7965545654296875,
+      "grad_norm": 6.060356616973877,
       "learning_rate": 3.422422422422423e-06,
-      "loss": 0.9959,
+      "loss": 0.9949,
       "step": 65710
     },
     {
       "epoch": 197.36,
-      "grad_norm": 8.035578727722168,
+      "grad_norm": 6.540942192077637,
       "learning_rate": 3.421421421421422e-06,
-      "loss": 1.0823,
+      "loss": 1.0828,
       "step": 65720
     },
     {
       "epoch": 197.39,
-      "grad_norm": 7.299635887145996,
+      "grad_norm": 6.766290664672852,
       "learning_rate": 3.4204204204204204e-06,
-      "loss": 1.0053,
+      "loss": 1.0024,
       "step": 65730
     },
     {
       "epoch": 197.42,
-      "grad_norm": 5.903440952301025,
+      "grad_norm": 4.941584587097168,
       "learning_rate": 3.41941941941942e-06,
-      "loss": 0.8749,
+      "loss": 0.8923,
       "step": 65740
     },
     {
       "epoch": 197.45,
-      "grad_norm": 6.828120231628418,
+      "grad_norm": 5.0399489402771,
       "learning_rate": 3.418418418418419e-06,
-      "loss": 0.8991,
+      "loss": 0.9003,
       "step": 65750
     },
     {
       "epoch": 197.48,
-      "grad_norm": 7.050987243652344,
+      "grad_norm": 6.9538373947143555,
       "learning_rate": 3.417417417417418e-06,
-      "loss": 0.9295,
+      "loss": 0.9307,
       "step": 65760
     },
     {
       "epoch": 197.51,
-      "grad_norm": 6.105101585388184,
+      "grad_norm": 9.241317749023438,
       "learning_rate": 3.4164164164164165e-06,
-      "loss": 0.9678,
+      "loss": 0.9565,
       "step": 65770
     },
     {
       "epoch": 197.54,
-      "grad_norm": 8.227157592773438,
+      "grad_norm": 6.206630706787109,
       "learning_rate": 3.4154154154154155e-06,
-      "loss": 0.9408,
+      "loss": 0.9366,
       "step": 65780
     },
     {
       "epoch": 197.57,
-      "grad_norm": 7.356484889984131,
+      "grad_norm": 6.787926197052002,
       "learning_rate": 3.414414414414415e-06,
-      "loss": 0.9937,
+      "loss": 0.9925,
       "step": 65790
     },
     {
       "epoch": 197.6,
-      "grad_norm": 7.90440034866333,
+      "grad_norm": 10.909749031066895,
       "learning_rate": 3.413413413413414e-06,
-      "loss": 1.012,
+      "loss": 1.0089,
       "step": 65800
     },
     {
       "epoch": 197.63,
-      "grad_norm": 5.480818748474121,
+      "grad_norm": 6.408101558685303,
       "learning_rate": 3.4124124124124126e-06,
-      "loss": 0.9089,
+      "loss": 0.8967,
       "step": 65810
     },
     {
       "epoch": 197.66,
-      "grad_norm": 6.822171688079834,
+      "grad_norm": 6.415679931640625,
       "learning_rate": 3.4114114114114116e-06,
-      "loss": 0.985,
+      "loss": 0.9882,
       "step": 65820
     },
     {
       "epoch": 197.69,
-      "grad_norm": 7.260140895843506,
+      "grad_norm": 6.119547367095947,
       "learning_rate": 3.4104104104104106e-06,
-      "loss": 0.9409,
+      "loss": 0.947,
       "step": 65830
     },
     {
       "epoch": 197.72,
-      "grad_norm": 7.406771183013916,
+      "grad_norm": 5.825259685516357,
       "learning_rate": 3.40940940940941e-06,
-      "loss": 1.0065,
+      "loss": 1.0008,
       "step": 65840
     },
     {
       "epoch": 197.75,
-      "grad_norm": 6.198680400848389,
+      "grad_norm": 7.113783836364746,
       "learning_rate": 3.4084084084084086e-06,
-      "loss": 0.9583,
+      "loss": 0.9445,
       "step": 65850
     },
     {
       "epoch": 197.78,
-      "grad_norm": 10.97604751586914,
+      "grad_norm": 11.929317474365234,
       "learning_rate": 3.4074074074074077e-06,
-      "loss": 0.9748,
+      "loss": 0.9679,
       "step": 65860
     },
     {
       "epoch": 197.81,
-      "grad_norm": 9.156510353088379,
+      "grad_norm": 8.631925582885742,
       "learning_rate": 3.4064064064064067e-06,
-      "loss": 0.9486,
+      "loss": 0.9427,
       "step": 65870
     },
     {
       "epoch": 197.84,
-      "grad_norm": 9.117813110351562,
+      "grad_norm": 7.500669956207275,
       "learning_rate": 3.4054054054054057e-06,
-      "loss": 0.9449,
+      "loss": 0.9374,
       "step": 65880
     },
     {
       "epoch": 197.87,
-      "grad_norm": 5.266153812408447,
+      "grad_norm": 6.059218406677246,
       "learning_rate": 3.4044044044044043e-06,
-      "loss": 0.971,
+      "loss": 0.9735,
       "step": 65890
     },
     {
       "epoch": 197.9,
-      "grad_norm": 7.2110795974731445,
+      "grad_norm": 8.3154296875,
       "learning_rate": 3.4034034034034037e-06,
-      "loss": 1.0074,
+      "loss": 1.0172,
       "step": 65900
     },
     {
       "epoch": 197.93,
-      "grad_norm": 6.699283599853516,
+      "grad_norm": 6.685973167419434,
       "learning_rate": 3.4024024024024028e-06,
-      "loss": 0.9262,
+      "loss": 0.922,
       "step": 65910
     },
     {
       "epoch": 197.96,
-      "grad_norm": 6.652071475982666,
+      "grad_norm": 6.542595863342285,
       "learning_rate": 3.4014014014014018e-06,
-      "loss": 0.9195,
+      "loss": 0.9128,
       "step": 65920
     },
     {
       "epoch": 197.99,
-      "grad_norm": 8.043095588684082,
+      "grad_norm": 8.896978378295898,
       "learning_rate": 3.4004004004004004e-06,
-      "loss": 0.9957,
+      "loss": 1.0019,
       "step": 65930
     },
     {
       "epoch": 198.0,
-      "eval_accuracy": 0.821,
-      "eval_loss": 0.5853809714317322,
-      "eval_runtime": 5.3287,
-      "eval_samples_per_second": 1876.628,
-      "eval_steps_per_second": 7.507,
+      "eval_accuracy": 0.8245,
+      "eval_loss": 0.5823343992233276,
+      "eval_runtime": 5.3782,
+      "eval_samples_per_second": 1859.351,
+      "eval_steps_per_second": 7.437,
       "step": 65934
     },
     {
       "epoch": 198.02,
-      "grad_norm": 8.130030632019043,
+      "grad_norm": 6.730227470397949,
       "learning_rate": 3.3993993993993994e-06,
-      "loss": 1.1264,
+      "loss": 1.0965,
       "step": 65940
     },
     {
       "epoch": 198.05,
-      "grad_norm": 6.899663925170898,
+      "grad_norm": 7.415763854980469,
       "learning_rate": 3.398398398398399e-06,
-      "loss": 0.8851,
+      "loss": 0.8887,
       "step": 65950
     },
     {
       "epoch": 198.08,
-      "grad_norm": 10.987419128417969,
+      "grad_norm": 12.590709686279297,
       "learning_rate": 3.397397397397398e-06,
-      "loss": 0.9318,
+      "loss": 0.9345,
       "step": 65960
     },
     {
       "epoch": 198.11,
-      "grad_norm": 6.419900417327881,
+      "grad_norm": 7.4090752601623535,
       "learning_rate": 3.3963963963963964e-06,
-      "loss": 0.9994,
+      "loss": 1.0147,
       "step": 65970
     },
     {
       "epoch": 198.14,
-      "grad_norm": 7.243531227111816,
+      "grad_norm": 7.293807506561279,
       "learning_rate": 3.3953953953953955e-06,
-      "loss": 1.0082,
+      "loss": 1.0002,
       "step": 65980
     },
     {
       "epoch": 198.17,
-      "grad_norm": 9.354218482971191,
+      "grad_norm": 8.2396821975708,
       "learning_rate": 3.3943943943943945e-06,
-      "loss": 0.9644,
+      "loss": 0.9639,
       "step": 65990
     },
     {
       "epoch": 198.2,
-      "grad_norm": 5.794597148895264,
+      "grad_norm": 6.236245155334473,
       "learning_rate": 3.393393393393394e-06,
-      "loss": 0.9522,
+      "loss": 0.9426,
       "step": 66000
     },
     {
       "epoch": 198.23,
-      "grad_norm": 9.385222434997559,
+      "grad_norm": 7.990300178527832,
       "learning_rate": 3.392392392392393e-06,
-      "loss": 0.9624,
+      "loss": 0.967,
       "step": 66010
     },
     {
       "epoch": 198.26,
-      "grad_norm": 8.587743759155273,
+      "grad_norm": 8.895379066467285,
       "learning_rate": 3.3913913913913915e-06,
-      "loss": 0.8486,
+      "loss": 0.8529,
       "step": 66020
     },
     {
       "epoch": 198.29,
-      "grad_norm": 8.295408248901367,
+      "grad_norm": 8.728877067565918,
       "learning_rate": 3.3903903903903905e-06,
-      "loss": 0.9593,
+      "loss": 0.9712,
       "step": 66030
     },
     {
       "epoch": 198.32,
-      "grad_norm": 8.083918571472168,
+      "grad_norm": 8.412097930908203,
       "learning_rate": 3.3893893893893896e-06,
-      "loss": 0.9388,
+      "loss": 0.937,
       "step": 66040
     },
     {
       "epoch": 198.35,
-      "grad_norm": 6.82302188873291,
+      "grad_norm": 9.1718168258667,
       "learning_rate": 3.388388388388389e-06,
-      "loss": 0.9284,
+      "loss": 0.9131,
       "step": 66050
     },
     {
       "epoch": 198.38,
-      "grad_norm": 5.062270164489746,
+      "grad_norm": 6.204099178314209,
       "learning_rate": 3.3873873873873876e-06,
-      "loss": 0.9358,
+      "loss": 0.9208,
       "step": 66060
     },
     {
       "epoch": 198.41,
-      "grad_norm": 9.601099967956543,
+      "grad_norm": 7.915976047515869,
       "learning_rate": 3.3863863863863866e-06,
-      "loss": 0.8921,
+      "loss": 0.8947,
       "step": 66070
     },
     {
       "epoch": 198.44,
-      "grad_norm": 5.022820949554443,
+      "grad_norm": 8.971877098083496,
       "learning_rate": 3.3853853853853856e-06,
-      "loss": 1.0244,
+      "loss": 1.0185,
       "step": 66080
     },
     {
       "epoch": 198.47,
-      "grad_norm": 5.641826152801514,
+      "grad_norm": 7.478358745574951,
       "learning_rate": 3.384384384384385e-06,
-      "loss": 0.88,
+      "loss": 0.8761,
       "step": 66090
     },
     {
       "epoch": 198.5,
-      "grad_norm": 3.9019558429718018,
+      "grad_norm": 5.654519557952881,
       "learning_rate": 3.3833833833833833e-06,
-      "loss": 0.9248,
+      "loss": 0.9131,
       "step": 66100
     },
     {
       "epoch": 198.53,
-      "grad_norm": 5.882186412811279,
+      "grad_norm": 6.685442924499512,
       "learning_rate": 3.3823823823823827e-06,
-      "loss": 0.9808,
+      "loss": 0.9745,
       "step": 66110
     },
     {
       "epoch": 198.56,
-      "grad_norm": 7.165634632110596,
+      "grad_norm": 6.528355121612549,
       "learning_rate": 3.3813813813813817e-06,
-      "loss": 1.0176,
+      "loss": 1.0066,
       "step": 66120
     },
     {
       "epoch": 198.59,
-      "grad_norm": 4.820448398590088,
+      "grad_norm": 5.438409328460693,
       "learning_rate": 3.3803803803803807e-06,
-      "loss": 0.9036,
+      "loss": 0.9143,
       "step": 66130
     },
     {
       "epoch": 198.62,
-      "grad_norm": 5.916661739349365,
+      "grad_norm": 5.51942253112793,
       "learning_rate": 3.3793793793793793e-06,
-      "loss": 0.9412,
+      "loss": 0.9388,
       "step": 66140
     },
     {
       "epoch": 198.65,
-      "grad_norm": 5.14828634262085,
+      "grad_norm": 4.697368621826172,
       "learning_rate": 3.3783783783783788e-06,
-      "loss": 0.8824,
+      "loss": 0.8847,
       "step": 66150
     },
     {
       "epoch": 198.68,
-      "grad_norm": 7.3419270515441895,
+      "grad_norm": 6.331083297729492,
       "learning_rate": 3.3773773773773778e-06,
-      "loss": 0.9827,
+      "loss": 0.972,
       "step": 66160
     },
     {
       "epoch": 198.71,
-      "grad_norm": 5.1212005615234375,
+      "grad_norm": 6.9242939949035645,
       "learning_rate": 3.376376376376377e-06,
-      "loss": 0.9446,
+      "loss": 0.9412,
       "step": 66170
     },
     {
       "epoch": 198.74,
-      "grad_norm": 6.283556938171387,
+      "grad_norm": 5.955799579620361,
       "learning_rate": 3.3753753753753754e-06,
-      "loss": 0.9589,
+      "loss": 0.9563,
       "step": 66180
     },
     {
       "epoch": 198.77,
-      "grad_norm": 6.823297023773193,
+      "grad_norm": 8.137931823730469,
       "learning_rate": 3.3743743743743744e-06,
-      "loss": 0.9127,
+      "loss": 0.9158,
       "step": 66190
     },
     {
       "epoch": 198.8,
-      "grad_norm": 6.024405002593994,
+      "grad_norm": 6.51961088180542,
       "learning_rate": 3.373373373373374e-06,
-      "loss": 0.9976,
+      "loss": 0.9839,
       "step": 66200
     },
     {
       "epoch": 198.83,
-      "grad_norm": 4.724828720092773,
+      "grad_norm": 5.53079080581665,
       "learning_rate": 3.372372372372373e-06,
-      "loss": 0.9916,
+      "loss": 0.9933,
       "step": 66210
     },
     {
       "epoch": 198.86,
-      "grad_norm": 6.886871337890625,
+      "grad_norm": 7.0902814865112305,
       "learning_rate": 3.3713713713713715e-06,
-      "loss": 0.9157,
+      "loss": 0.9145,
       "step": 66220
     },
     {
       "epoch": 198.89,
-      "grad_norm": 5.276222229003906,
+      "grad_norm": 5.720056056976318,
       "learning_rate": 3.3703703703703705e-06,
-      "loss": 0.9108,
+      "loss": 0.9043,
       "step": 66230
     },
     {
       "epoch": 198.92,
-      "grad_norm": 4.507023334503174,
+      "grad_norm": 6.4624176025390625,
       "learning_rate": 3.3693693693693695e-06,
-      "loss": 0.8956,
+      "loss": 0.8888,
       "step": 66240
     },
     {
       "epoch": 198.95,
-      "grad_norm": 7.211368560791016,
+      "grad_norm": 7.01402473449707,
       "learning_rate": 3.368368368368369e-06,
-      "loss": 1.0259,
+      "loss": 1.0266,
       "step": 66250
     },
     {
       "epoch": 198.98,
-      "grad_norm": 5.345754146575928,
+      "grad_norm": 5.9599690437316895,
       "learning_rate": 3.367367367367368e-06,
-      "loss": 0.9442,
+      "loss": 0.9386,
       "step": 66260
     },
     {
       "epoch": 199.0,
-      "eval_accuracy": 0.8189,
-      "eval_loss": 0.5894225835800171,
-      "eval_runtime": 5.5203,
-      "eval_samples_per_second": 1811.491,
-      "eval_steps_per_second": 7.246,
+      "eval_accuracy": 0.8218,
+      "eval_loss": 0.5878944993019104,
+      "eval_runtime": 5.4757,
+      "eval_samples_per_second": 1826.257,
+      "eval_steps_per_second": 7.305,
       "step": 66267
     },
     {
       "epoch": 199.01,
-      "grad_norm": 6.280285358428955,
+      "grad_norm": 6.940164089202881,
       "learning_rate": 3.3663663663663666e-06,
-      "loss": 1.0997,
+      "loss": 1.0993,
       "step": 66270
     },
     {
       "epoch": 199.04,
-      "grad_norm": 7.020377159118652,
+      "grad_norm": 5.9485182762146,
       "learning_rate": 3.3653653653653656e-06,
-      "loss": 0.9341,
+      "loss": 0.942,
       "step": 66280
     },
     {
       "epoch": 199.07,
-      "grad_norm": 6.972303867340088,
+      "grad_norm": 6.486828327178955,
       "learning_rate": 3.3643643643643646e-06,
-      "loss": 0.9557,
+      "loss": 0.9477,
       "step": 66290
     },
     {
       "epoch": 199.1,
-      "grad_norm": 7.399078845977783,
+      "grad_norm": 9.753954887390137,
       "learning_rate": 3.363363363363364e-06,
-      "loss": 0.9441,
+      "loss": 0.9446,
       "step": 66300
     },
     {
       "epoch": 199.13,
-      "grad_norm": 7.427445888519287,
+      "grad_norm": 6.179585933685303,
       "learning_rate": 3.3623623623623626e-06,
-      "loss": 0.8913,
+      "loss": 0.8869,
       "step": 66310
     },
     {
       "epoch": 199.16,
-      "grad_norm": 7.291522979736328,
+      "grad_norm": 8.520916938781738,
       "learning_rate": 3.3613613613613617e-06,
-      "loss": 0.9573,
+      "loss": 0.9548,
       "step": 66320
     },
     {
       "epoch": 199.19,
-      "grad_norm": 6.827638149261475,
+      "grad_norm": 6.762482166290283,
       "learning_rate": 3.3603603603603607e-06,
-      "loss": 0.9286,
+      "loss": 0.9303,
       "step": 66330
     },
     {
       "epoch": 199.22,
-      "grad_norm": 7.338196277618408,
+      "grad_norm": 7.758122444152832,
       "learning_rate": 3.3593593593593597e-06,
-      "loss": 0.9249,
+      "loss": 0.9399,
       "step": 66340
     },
     {
       "epoch": 199.25,
-      "grad_norm": 4.654018402099609,
+      "grad_norm": 5.909079551696777,
       "learning_rate": 3.3583583583583583e-06,
-      "loss": 0.962,
+      "loss": 0.9632,
       "step": 66350
     },
     {
       "epoch": 199.28,
-      "grad_norm": 9.104846000671387,
+      "grad_norm": 6.539394378662109,
       "learning_rate": 3.3573573573573577e-06,
-      "loss": 0.9598,
+      "loss": 0.9548,
       "step": 66360
     },
     {
       "epoch": 199.31,
-      "grad_norm": 6.937545299530029,
+      "grad_norm": 5.028235912322998,
       "learning_rate": 3.3563563563563567e-06,
-      "loss": 0.8732,
+      "loss": 0.8668,
       "step": 66370
     },
     {
       "epoch": 199.34,
-      "grad_norm": 4.934935092926025,
+      "grad_norm": 4.7256059646606445,
       "learning_rate": 3.3553553553553558e-06,
-      "loss": 0.9522,
+      "loss": 0.9461,
       "step": 66380
     },
     {
       "epoch": 199.37,
-      "grad_norm": 7.56602668762207,
+      "grad_norm": 5.830316066741943,
       "learning_rate": 3.3543543543543544e-06,
-      "loss": 0.9283,
+      "loss": 0.9199,
       "step": 66390
     },
     {
       "epoch": 199.4,
-      "grad_norm": 6.213950157165527,
+      "grad_norm": 6.398956775665283,
       "learning_rate": 3.3533533533533534e-06,
-      "loss": 0.9087,
+      "loss": 0.9042,
       "step": 66400
     },
     {
       "epoch": 199.43,
-      "grad_norm": 5.72544002532959,
+      "grad_norm": 6.102389812469482,
       "learning_rate": 3.352352352352353e-06,
-      "loss": 0.981,
+      "loss": 0.9654,
       "step": 66410
     },
     {
       "epoch": 199.46,
-      "grad_norm": 7.1792683601379395,
+      "grad_norm": 8.569842338562012,
       "learning_rate": 3.351351351351352e-06,
-      "loss": 0.9694,
+      "loss": 0.9629,
       "step": 66420
     },
     {
       "epoch": 199.49,
-      "grad_norm": 5.201842308044434,
+      "grad_norm": 4.938758373260498,
       "learning_rate": 3.3503503503503504e-06,
-      "loss": 0.9375,
+      "loss": 0.9419,
       "step": 66430
     },
     {
       "epoch": 199.52,
-      "grad_norm": 6.981503009796143,
+      "grad_norm": 8.12291145324707,
       "learning_rate": 3.3493493493493495e-06,
-      "loss": 0.9897,
+      "loss": 0.9893,
       "step": 66440
     },
     {
       "epoch": 199.55,
-      "grad_norm": 11.444555282592773,
+      "grad_norm": 8.169855117797852,
       "learning_rate": 3.3483483483483485e-06,
-      "loss": 0.9496,
+      "loss": 0.956,
       "step": 66450
     },
     {
       "epoch": 199.58,
-      "grad_norm": 6.21297025680542,
+      "grad_norm": 7.276932716369629,
       "learning_rate": 3.347347347347348e-06,
-      "loss": 0.9112,
+      "loss": 0.92,
       "step": 66460
     },
     {
       "epoch": 199.61,
-      "grad_norm": 7.578511714935303,
+      "grad_norm": 7.817767143249512,
       "learning_rate": 3.3463463463463465e-06,
-      "loss": 0.9652,
+      "loss": 0.9514,
       "step": 66470
     },
     {
       "epoch": 199.64,
-      "grad_norm": 6.570349216461182,
+      "grad_norm": 5.566411018371582,
       "learning_rate": 3.3453453453453455e-06,
-      "loss": 0.9411,
+      "loss": 0.937,
       "step": 66480
     },
     {
       "epoch": 199.67,
-      "grad_norm": 6.140203952789307,
+      "grad_norm": 7.885871887207031,
       "learning_rate": 3.3443443443443445e-06,
-      "loss": 0.9263,
+      "loss": 0.9323,
       "step": 66490
     },
     {
       "epoch": 199.7,
-      "grad_norm": 6.535429000854492,
+      "grad_norm": 10.266615867614746,
       "learning_rate": 3.3433433433433436e-06,
-      "loss": 0.91,
+      "loss": 0.9048,
       "step": 66500
     },
     {
       "epoch": 199.73,
-      "grad_norm": 7.497563362121582,
+      "grad_norm": 7.423398971557617,
       "learning_rate": 3.342342342342342e-06,
-      "loss": 0.9139,
+      "loss": 0.9076,
       "step": 66510
     },
     {
       "epoch": 199.76,
-      "grad_norm": 6.346972465515137,
+      "grad_norm": 7.979969024658203,
       "learning_rate": 3.3413413413413416e-06,
-      "loss": 0.9222,
+      "loss": 0.9206,
       "step": 66520
     },
     {
       "epoch": 199.79,
-      "grad_norm": 7.473598480224609,
+      "grad_norm": 5.264063835144043,
       "learning_rate": 3.3403403403403406e-06,
-      "loss": 0.9384,
+      "loss": 0.9465,
       "step": 66530
     },
     {
       "epoch": 199.82,
-      "grad_norm": 6.052313327789307,
+      "grad_norm": 8.796684265136719,
       "learning_rate": 3.3393393393393396e-06,
-      "loss": 0.993,
+      "loss": 0.9951,
       "step": 66540
     },
     {
       "epoch": 199.85,
-      "grad_norm": 5.627513408660889,
+      "grad_norm": 4.973764419555664,
       "learning_rate": 3.338338338338339e-06,
-      "loss": 0.9136,
+      "loss": 0.9169,
       "step": 66550
     },
     {
       "epoch": 199.88,
-      "grad_norm": 6.213540554046631,
+      "grad_norm": 8.824650764465332,
       "learning_rate": 3.3373373373373373e-06,
-      "loss": 0.9438,
+      "loss": 0.9445,
       "step": 66560
     },
     {
       "epoch": 199.91,
-      "grad_norm": 5.668242454528809,
+      "grad_norm": 7.359220027923584,
       "learning_rate": 3.3363363363363367e-06,
-      "loss": 0.9303,
+      "loss": 0.9336,
       "step": 66570
     },
     {
       "epoch": 199.94,
-      "grad_norm": 6.327518463134766,
+      "grad_norm": 6.48329496383667,
       "learning_rate": 3.3353353353353357e-06,
-      "loss": 0.9781,
+      "loss": 0.984,
       "step": 66580
     },
     {
       "epoch": 199.97,
-      "grad_norm": 8.253247261047363,
+      "grad_norm": 9.570948600769043,
       "learning_rate": 3.3343343343343347e-06,
-      "loss": 0.9961,
+      "loss": 0.9836,
       "step": 66590
     },
     {
       "epoch": 200.0,
-      "grad_norm": 50.72835159301758,
+      "grad_norm": 51.752952575683594,
       "learning_rate": 3.3333333333333333e-06,
-      "loss": 1.1311,
+      "loss": 1.1359,
       "step": 66600
     },
     {
       "epoch": 200.0,
-      "eval_accuracy": 0.8235,
-      "eval_loss": 0.5785021185874939,
-      "eval_runtime": 5.6403,
-      "eval_samples_per_second": 1772.967,
-      "eval_steps_per_second": 7.092,
+      "eval_accuracy": 0.8244,
+      "eval_loss": 0.5789966583251953,
+      "eval_runtime": 5.264,
+      "eval_samples_per_second": 1899.694,
+      "eval_steps_per_second": 7.599,
       "step": 66600
     },
     {
       "epoch": 200.03,
-      "grad_norm": 6.414413928985596,
+      "grad_norm": 7.498602867126465,
       "learning_rate": 3.3323323323323323e-06,
-      "loss": 0.9214,
+      "loss": 0.9181,
       "step": 66610
     },
     {
       "epoch": 200.06,
-      "grad_norm": 5.565194129943848,
+      "grad_norm": 6.602406978607178,
       "learning_rate": 3.3313313313313318e-06,
-      "loss": 0.9762,
+      "loss": 0.9705,
       "step": 66620
     },
     {
       "epoch": 200.09,
-      "grad_norm": 7.326407432556152,
+      "grad_norm": 6.083890438079834,
       "learning_rate": 3.330330330330331e-06,
-      "loss": 0.9453,
+      "loss": 0.9529,
       "step": 66630
     },
     {
       "epoch": 200.12,
-      "grad_norm": 8.94873332977295,
+      "grad_norm": 9.357783317565918,
       "learning_rate": 3.3293293293293294e-06,
-      "loss": 0.9881,
+      "loss": 0.9917,
       "step": 66640
     },
     {
       "epoch": 200.15,
-      "grad_norm": 7.277824401855469,
+      "grad_norm": 6.065515995025635,
       "learning_rate": 3.3283283283283284e-06,
-      "loss": 0.9565,
+      "loss": 0.952,
       "step": 66650
     },
     {
       "epoch": 200.18,
-      "grad_norm": 5.567216396331787,
+      "grad_norm": 7.869673252105713,
       "learning_rate": 3.327327327327328e-06,
-      "loss": 0.9878,
+      "loss": 0.984,
       "step": 66660
     },
     {
       "epoch": 200.21,
-      "grad_norm": 5.655640125274658,
+      "grad_norm": 8.30904483795166,
       "learning_rate": 3.326326326326327e-06,
-      "loss": 0.9156,
+      "loss": 0.9175,
       "step": 66670
     },
     {
       "epoch": 200.24,
-      "grad_norm": 6.285792827606201,
+      "grad_norm": 5.929512977600098,
       "learning_rate": 3.3253253253253255e-06,
-      "loss": 0.9423,
+      "loss": 0.9436,
       "step": 66680
     },
     {
       "epoch": 200.27,
-      "grad_norm": 8.436494827270508,
+      "grad_norm": 8.449394226074219,
       "learning_rate": 3.3243243243243245e-06,
-      "loss": 0.8973,
+      "loss": 0.8939,
       "step": 66690
     },
     {
       "epoch": 200.3,
-      "grad_norm": 6.768445014953613,
+      "grad_norm": 8.028549194335938,
       "learning_rate": 3.3233233233233235e-06,
-      "loss": 0.9409,
+      "loss": 0.9411,
       "step": 66700
     },
     {
       "epoch": 200.33,
-      "grad_norm": 6.154605388641357,
+      "grad_norm": 6.900210380554199,
       "learning_rate": 3.322322322322323e-06,
-      "loss": 0.9294,
+      "loss": 0.9203,
       "step": 66710
     },
     {
       "epoch": 200.36,
-      "grad_norm": 6.589270114898682,
+      "grad_norm": 8.515894889831543,
       "learning_rate": 3.3213213213213215e-06,
-      "loss": 0.9286,
+      "loss": 0.9301,
       "step": 66720
     },
     {
       "epoch": 200.39,
-      "grad_norm": 9.308235168457031,
+      "grad_norm": 7.755313396453857,
       "learning_rate": 3.3203203203203206e-06,
-      "loss": 0.9145,
+      "loss": 0.9275,
       "step": 66730
     },
     {
       "epoch": 200.42,
-      "grad_norm": 5.7957329750061035,
+      "grad_norm": 6.648796558380127,
       "learning_rate": 3.3193193193193196e-06,
-      "loss": 0.8869,
+      "loss": 0.883,
       "step": 66740
     },
     {
       "epoch": 200.45,
-      "grad_norm": 5.222961902618408,
+      "grad_norm": 4.540383338928223,
       "learning_rate": 3.3183183183183186e-06,
-      "loss": 0.9266,
+      "loss": 0.9288,
       "step": 66750
     },
     {
       "epoch": 200.48,
-      "grad_norm": 5.903299808502197,
+      "grad_norm": 6.823963165283203,
       "learning_rate": 3.317317317317317e-06,
-      "loss": 0.9373,
+      "loss": 0.9406,
       "step": 66760
     },
     {
       "epoch": 200.51,
-      "grad_norm": 6.023310661315918,
+      "grad_norm": 5.9887614250183105,
       "learning_rate": 3.3163163163163166e-06,
-      "loss": 0.898,
+      "loss": 0.8956,
       "step": 66770
     },
     {
       "epoch": 200.54,
-      "grad_norm": 6.360662937164307,
+      "grad_norm": 7.033992767333984,
       "learning_rate": 3.3153153153153157e-06,
-      "loss": 0.956,
+      "loss": 0.9459,
       "step": 66780
     },
     {
       "epoch": 200.57,
-      "grad_norm": 4.855058193206787,
+      "grad_norm": 7.155275821685791,
       "learning_rate": 3.3143143143143147e-06,
-      "loss": 0.9221,
+      "loss": 0.9245,
       "step": 66790
     },
     {
       "epoch": 200.6,
-      "grad_norm": 7.3659491539001465,
+      "grad_norm": 12.11591911315918,
       "learning_rate": 3.3133133133133137e-06,
-      "loss": 0.9844,
+      "loss": 0.9826,
       "step": 66800
     },
     {
       "epoch": 200.63,
-      "grad_norm": 5.604797840118408,
+      "grad_norm": 5.708730220794678,
       "learning_rate": 3.3123123123123123e-06,
-      "loss": 0.9445,
+      "loss": 0.9345,
       "step": 66810
     },
     {
       "epoch": 200.66,
-      "grad_norm": 7.2179999351501465,
+      "grad_norm": 7.173712253570557,
       "learning_rate": 3.3113113113113117e-06,
-      "loss": 0.938,
+      "loss": 0.9431,
       "step": 66820
     },
     {
       "epoch": 200.69,
-      "grad_norm": 6.4805731773376465,
+      "grad_norm": 8.676033020019531,
       "learning_rate": 3.3103103103103107e-06,
-      "loss": 1.0484,
+      "loss": 1.0378,
       "step": 66830
     },
     {
       "epoch": 200.72,
-      "grad_norm": 6.618391036987305,
+      "grad_norm": 5.87326717376709,
       "learning_rate": 3.3093093093093098e-06,
-      "loss": 0.9471,
+      "loss": 0.9432,
       "step": 66840
     },
     {
       "epoch": 200.75,
-      "grad_norm": 9.652891159057617,
+      "grad_norm": 8.620593070983887,
       "learning_rate": 3.3083083083083084e-06,
-      "loss": 0.9518,
+      "loss": 0.9539,
       "step": 66850
     },
     {
       "epoch": 200.78,
-      "grad_norm": 4.934891223907471,
+      "grad_norm": 7.407347202301025,
       "learning_rate": 3.3073073073073074e-06,
-      "loss": 0.98,
+      "loss": 0.9808,
       "step": 66860
     },
     {
       "epoch": 200.81,
-      "grad_norm": 13.773032188415527,
+      "grad_norm": 10.70297908782959,
       "learning_rate": 3.306306306306307e-06,
-      "loss": 0.9746,
+      "loss": 0.964,
       "step": 66870
     },
     {
       "epoch": 200.84,
-      "grad_norm": 7.526971340179443,
+      "grad_norm": 11.389545440673828,
       "learning_rate": 3.305305305305306e-06,
-      "loss": 0.9359,
+      "loss": 0.9288,
       "step": 66880
     },
     {
       "epoch": 200.87,
-      "grad_norm": 4.523584842681885,
+      "grad_norm": 4.883289337158203,
       "learning_rate": 3.3043043043043044e-06,
-      "loss": 0.9045,
+      "loss": 0.894,
       "step": 66890
     },
     {
       "epoch": 200.9,
-      "grad_norm": 9.509800910949707,
+      "grad_norm": 9.475319862365723,
       "learning_rate": 3.3033033033033035e-06,
-      "loss": 0.9625,
+      "loss": 0.9702,
       "step": 66900
     },
     {
       "epoch": 200.93,
-      "grad_norm": 5.9504570960998535,
+      "grad_norm": 7.3884429931640625,
       "learning_rate": 3.3023023023023025e-06,
-      "loss": 0.9557,
+      "loss": 0.9506,
       "step": 66910
     },
     {
       "epoch": 200.96,
-      "grad_norm": 6.2885260581970215,
+      "grad_norm": 9.535300254821777,
       "learning_rate": 3.301301301301302e-06,
-      "loss": 0.9127,
+      "loss": 0.9041,
       "step": 66920
     },
     {
       "epoch": 200.99,
-      "grad_norm": 7.944681644439697,
+      "grad_norm": 6.421842098236084,
       "learning_rate": 3.3003003003003005e-06,
-      "loss": 0.9542,
+      "loss": 0.9553,
       "step": 66930
     },
     {
       "epoch": 201.0,
-      "eval_accuracy": 0.824,
-      "eval_loss": 0.5783321857452393,
-      "eval_runtime": 5.553,
-      "eval_samples_per_second": 1800.821,
+      "eval_accuracy": 0.827,
+      "eval_loss": 0.5749549865722656,
+      "eval_runtime": 5.5534,
+      "eval_samples_per_second": 1800.69,
       "eval_steps_per_second": 7.203,
       "step": 66933
     },
     {
       "epoch": 201.02,
-      "grad_norm": 5.328568935394287,
+      "grad_norm": 5.452633857727051,
       "learning_rate": 3.2992992992992995e-06,
-      "loss": 1.0367,
+      "loss": 1.0313,
       "step": 66940
     },
     {
       "epoch": 201.05,
-      "grad_norm": 9.882721900939941,
+      "grad_norm": 10.672093391418457,
       "learning_rate": 3.2982982982982985e-06,
-      "loss": 0.9614,
+      "loss": 0.9576,
       "step": 66950
     },
     {
       "epoch": 201.08,
-      "grad_norm": 8.346414566040039,
+      "grad_norm": 6.576988697052002,
       "learning_rate": 3.2972972972972976e-06,
-      "loss": 0.8854,
+      "loss": 0.874,
       "step": 66960
     },
     {
       "epoch": 201.11,
-      "grad_norm": 5.791387557983398,
+      "grad_norm": 6.535892963409424,
       "learning_rate": 3.296296296296296e-06,
-      "loss": 0.9636,
+      "loss": 0.9572,
       "step": 66970
     },
     {
       "epoch": 201.14,
-      "grad_norm": 6.264197826385498,
+      "grad_norm": 13.647068977355957,
       "learning_rate": 3.2952952952952956e-06,
-      "loss": 0.9535,
+      "loss": 0.9689,
       "step": 66980
     },
     {
       "epoch": 201.17,
-      "grad_norm": 5.326868057250977,
+      "grad_norm": 6.934617042541504,
       "learning_rate": 3.2942942942942946e-06,
-      "loss": 0.9198,
+      "loss": 0.9282,
       "step": 66990
     },
     {
       "epoch": 201.2,
-      "grad_norm": 7.013060092926025,
+      "grad_norm": 8.945302963256836,
       "learning_rate": 3.2932932932932936e-06,
-      "loss": 0.9428,
+      "loss": 0.9457,
       "step": 67000
     },
     {
       "epoch": 201.23,
-      "grad_norm": 6.806612014770508,
+      "grad_norm": 9.374344825744629,
       "learning_rate": 3.2922922922922922e-06,
-      "loss": 0.9522,
+      "loss": 0.9613,
       "step": 67010
     },
     {
       "epoch": 201.26,
-      "grad_norm": 4.742289066314697,
+      "grad_norm": 5.941964626312256,
       "learning_rate": 3.2912912912912912e-06,
-      "loss": 0.934,
+      "loss": 0.9268,
       "step": 67020
     },
     {
       "epoch": 201.29,
-      "grad_norm": 5.596538066864014,
+      "grad_norm": 7.23925256729126,
       "learning_rate": 3.2902902902902907e-06,
-      "loss": 0.9443,
+      "loss": 0.9416,
       "step": 67030
     },
     {
       "epoch": 201.32,
-      "grad_norm": 8.723917007446289,
+      "grad_norm": 6.381480693817139,
       "learning_rate": 3.2892892892892897e-06,
-      "loss": 0.9645,
+      "loss": 0.9632,
       "step": 67040
     },
     {
       "epoch": 201.35,
-      "grad_norm": 6.655569076538086,
+      "grad_norm": 7.479490280151367,
       "learning_rate": 3.2882882882882887e-06,
-      "loss": 0.9021,
+      "loss": 0.9062,
       "step": 67050
     },
     {
       "epoch": 201.38,
-      "grad_norm": 4.973470687866211,
+      "grad_norm": 7.174620151519775,
       "learning_rate": 3.2872872872872873e-06,
-      "loss": 0.8595,
+      "loss": 0.8631,
       "step": 67060
     },
     {
       "epoch": 201.41,
-      "grad_norm": 7.195031642913818,
+      "grad_norm": 6.2910966873168945,
       "learning_rate": 3.2862862862862863e-06,
-      "loss": 0.8891,
+      "loss": 0.8857,
       "step": 67070
     },
     {
       "epoch": 201.44,
-      "grad_norm": 4.912550449371338,
+      "grad_norm": 6.85276460647583,
       "learning_rate": 3.2852852852852858e-06,
-      "loss": 0.8929,
+      "loss": 0.8885,
       "step": 67080
     },
     {
       "epoch": 201.47,
-      "grad_norm": 6.530505180358887,
+      "grad_norm": 4.810078144073486,
       "learning_rate": 3.284284284284285e-06,
-      "loss": 0.9491,
+      "loss": 0.9453,
       "step": 67090
     },
     {
       "epoch": 201.5,
-      "grad_norm": 6.08884334564209,
+      "grad_norm": 7.792052268981934,
       "learning_rate": 3.2832832832832834e-06,
-      "loss": 0.8895,
+      "loss": 0.8822,
       "step": 67100
     },
     {
       "epoch": 201.53,
-      "grad_norm": 5.935667037963867,
+      "grad_norm": 7.061225891113281,
       "learning_rate": 3.2822822822822824e-06,
-      "loss": 0.9181,
+      "loss": 0.9093,
       "step": 67110
     },
     {
       "epoch": 201.56,
-      "grad_norm": 6.6623945236206055,
+      "grad_norm": 10.039162635803223,
       "learning_rate": 3.281281281281282e-06,
-      "loss": 0.9317,
+      "loss": 0.9263,
       "step": 67120
     },
     {
       "epoch": 201.59,
-      "grad_norm": 5.999873161315918,
+      "grad_norm": 6.8974504470825195,
       "learning_rate": 3.280280280280281e-06,
-      "loss": 0.9599,
+      "loss": 0.9585,
       "step": 67130
     },
     {
       "epoch": 201.62,
-      "grad_norm": 6.787379264831543,
+      "grad_norm": 7.667761325836182,
       "learning_rate": 3.2792792792792795e-06,
-      "loss": 0.9119,
+      "loss": 0.9133,
       "step": 67140
     },
     {
       "epoch": 201.65,
-      "grad_norm": 6.233237266540527,
+      "grad_norm": 9.80319595336914,
       "learning_rate": 3.2782782782782785e-06,
-      "loss": 0.9682,
+      "loss": 0.9769,
       "step": 67150
     },
     {
       "epoch": 201.68,
-      "grad_norm": 6.387402534484863,
+      "grad_norm": 5.90784215927124,
       "learning_rate": 3.2772772772772775e-06,
-      "loss": 0.933,
+      "loss": 0.9259,
       "step": 67160
     },
     {
       "epoch": 201.71,
-      "grad_norm": 6.098155975341797,
+      "grad_norm": 6.535787105560303,
       "learning_rate": 3.276276276276277e-06,
-      "loss": 0.963,
+      "loss": 0.9563,
       "step": 67170
     },
     {
       "epoch": 201.74,
-      "grad_norm": 7.249209403991699,
+      "grad_norm": 6.800347328186035,
       "learning_rate": 3.2752752752752755e-06,
-      "loss": 0.9735,
+      "loss": 0.9625,
       "step": 67180
     },
     {
       "epoch": 201.77,
-      "grad_norm": 7.028077602386475,
+      "grad_norm": 8.789681434631348,
       "learning_rate": 3.2742742742742746e-06,
-      "loss": 0.9776,
+      "loss": 0.9696,
       "step": 67190
     },
     {
       "epoch": 201.8,
-      "grad_norm": 7.760516166687012,
+      "grad_norm": 5.45648193359375,
       "learning_rate": 3.2732732732732736e-06,
-      "loss": 0.9699,
+      "loss": 0.9643,
       "step": 67200
     },
     {
       "epoch": 201.83,
-      "grad_norm": 8.67701530456543,
+      "grad_norm": 8.765012741088867,
       "learning_rate": 3.2722722722722726e-06,
-      "loss": 0.9315,
+      "loss": 0.9378,
       "step": 67210
     },
     {
       "epoch": 201.86,
-      "grad_norm": 5.846559047698975,
+      "grad_norm": 7.846569538116455,
       "learning_rate": 3.271271271271271e-06,
-      "loss": 0.9587,
+      "loss": 0.9667,
       "step": 67220
     },
     {
       "epoch": 201.89,
-      "grad_norm": 6.428820610046387,
+      "grad_norm": 5.575026035308838,
       "learning_rate": 3.2702702702702706e-06,
-      "loss": 0.9187,
+      "loss": 0.9143,
       "step": 67230
     },
     {
       "epoch": 201.92,
-      "grad_norm": 6.0345072746276855,
+      "grad_norm": 5.685698986053467,
       "learning_rate": 3.2692692692692696e-06,
-      "loss": 1.0002,
+      "loss": 1.0055,
       "step": 67240
     },
     {
       "epoch": 201.95,
-      "grad_norm": 6.895460605621338,
+      "grad_norm": 6.427411079406738,
       "learning_rate": 3.2682682682682687e-06,
-      "loss": 1.0008,
+      "loss": 1.001,
       "step": 67250
     },
     {
       "epoch": 201.98,
-      "grad_norm": 6.701959609985352,
+      "grad_norm": 6.6697163581848145,
       "learning_rate": 3.2672672672672673e-06,
-      "loss": 0.9352,
+      "loss": 0.9381,
       "step": 67260
     },
     {
       "epoch": 202.0,
-      "eval_accuracy": 0.8231,
-      "eval_loss": 0.5810949206352234,
-      "eval_runtime": 5.6056,
-      "eval_samples_per_second": 1783.922,
-      "eval_steps_per_second": 7.136,
+      "eval_accuracy": 0.8252,
+      "eval_loss": 0.5817432403564453,
+      "eval_runtime": 5.3699,
+      "eval_samples_per_second": 1862.248,
+      "eval_steps_per_second": 7.449,
       "step": 67266
     },
     {
       "epoch": 202.01,
-      "grad_norm": 7.5295915603637695,
+      "grad_norm": 7.041280746459961,
       "learning_rate": 3.2662662662662663e-06,
-      "loss": 1.235,
+      "loss": 1.207,
       "step": 67270
     },
     {
       "epoch": 202.04,
-      "grad_norm": 4.79722785949707,
+      "grad_norm": 6.204293727874756,
       "learning_rate": 3.2652652652652657e-06,
-      "loss": 0.9047,
+      "loss": 0.9016,
       "step": 67280
     },
     {
       "epoch": 202.07,
-      "grad_norm": 5.550017356872559,
+      "grad_norm": 6.436374664306641,
       "learning_rate": 3.2642642642642647e-06,
-      "loss": 0.8758,
+      "loss": 0.8807,
       "step": 67290
     },
     {
       "epoch": 202.1,
-      "grad_norm": 8.57181453704834,
+      "grad_norm": 8.346945762634277,
       "learning_rate": 3.2632632632632633e-06,
-      "loss": 0.9817,
+      "loss": 0.9705,
       "step": 67300
     },
     {
       "epoch": 202.13,
-      "grad_norm": 7.079439163208008,
+      "grad_norm": 9.456299781799316,
       "learning_rate": 3.2622622622622624e-06,
-      "loss": 0.9169,
+      "loss": 0.9163,
       "step": 67310
     },
     {
       "epoch": 202.16,
-      "grad_norm": 12.149038314819336,
+      "grad_norm": 8.836966514587402,
       "learning_rate": 3.2612612612612614e-06,
-      "loss": 0.9688,
+      "loss": 0.967,
       "step": 67320
     },
     {
       "epoch": 202.19,
-      "grad_norm": 8.978081703186035,
+      "grad_norm": 6.306703090667725,
       "learning_rate": 3.260260260260261e-06,
-      "loss": 1.0592,
+      "loss": 1.0533,
       "step": 67330
     },
     {
       "epoch": 202.22,
-      "grad_norm": 3.8721277713775635,
+      "grad_norm": 5.268733978271484,
       "learning_rate": 3.25925925925926e-06,
-      "loss": 0.9213,
+      "loss": 0.9264,
       "step": 67340
     },
     {
       "epoch": 202.25,
-      "grad_norm": 10.476312637329102,
+      "grad_norm": 12.476459503173828,
       "learning_rate": 3.2582582582582584e-06,
-      "loss": 1.0019,
+      "loss": 0.9993,
       "step": 67350
     },
     {
       "epoch": 202.28,
-      "grad_norm": 7.821645259857178,
+      "grad_norm": 9.245137214660645,
       "learning_rate": 3.2572572572572574e-06,
-      "loss": 0.9377,
+      "loss": 0.9343,
       "step": 67360
     },
     {
       "epoch": 202.31,
-      "grad_norm": 5.746581554412842,
+      "grad_norm": 5.714376926422119,
       "learning_rate": 3.2562562562562565e-06,
-      "loss": 0.9534,
+      "loss": 0.9395,
       "step": 67370
     },
     {
       "epoch": 202.34,
-      "grad_norm": 6.952554702758789,
+      "grad_norm": 5.136667728424072,
       "learning_rate": 3.255255255255256e-06,
-      "loss": 0.9988,
+      "loss": 0.988,
       "step": 67380
     },
     {
       "epoch": 202.37,
-      "grad_norm": 6.806423187255859,
+      "grad_norm": 6.784618854522705,
       "learning_rate": 3.2542542542542545e-06,
-      "loss": 0.9558,
+      "loss": 0.9573,
       "step": 67390
     },
     {
       "epoch": 202.4,
-      "grad_norm": 6.82014799118042,
+      "grad_norm": 8.036700248718262,
       "learning_rate": 3.2532532532532535e-06,
-      "loss": 0.9506,
+      "loss": 0.9428,
       "step": 67400
     },
     {
       "epoch": 202.43,
-      "grad_norm": 6.056450366973877,
+      "grad_norm": 7.138790130615234,
       "learning_rate": 3.2522522522522525e-06,
-      "loss": 0.9136,
+      "loss": 0.9196,
       "step": 67410
     },
     {
       "epoch": 202.46,
-      "grad_norm": 6.208089351654053,
+      "grad_norm": 5.177746772766113,
       "learning_rate": 3.2512512512512516e-06,
-      "loss": 0.9361,
+      "loss": 0.9384,
       "step": 67420
     },
     {
       "epoch": 202.49,
-      "grad_norm": 5.554291725158691,
+      "grad_norm": 6.652961730957031,
       "learning_rate": 3.25025025025025e-06,
-      "loss": 0.9212,
+      "loss": 0.91,
       "step": 67430
     },
     {
       "epoch": 202.52,
-      "grad_norm": 6.87625789642334,
+      "grad_norm": 9.526432037353516,
       "learning_rate": 3.2492492492492496e-06,
-      "loss": 0.9008,
+      "loss": 0.9044,
       "step": 67440
     },
     {
       "epoch": 202.55,
-      "grad_norm": 6.865622520446777,
+      "grad_norm": 7.328756332397461,
       "learning_rate": 3.2482482482482486e-06,
-      "loss": 0.8688,
+      "loss": 0.8674,
       "step": 67450
     },
     {
       "epoch": 202.58,
-      "grad_norm": 9.866365432739258,
+      "grad_norm": 10.152843475341797,
       "learning_rate": 3.2472472472472476e-06,
-      "loss": 0.9132,
+      "loss": 0.9161,
       "step": 67460
     },
     {
       "epoch": 202.61,
-      "grad_norm": 5.7068586349487305,
+      "grad_norm": 9.194047927856445,
       "learning_rate": 3.2462462462462462e-06,
-      "loss": 0.9947,
+      "loss": 0.984,
       "step": 67470
     },
     {
       "epoch": 202.64,
-      "grad_norm": 4.158270835876465,
+      "grad_norm": 4.581419944763184,
       "learning_rate": 3.2452452452452452e-06,
-      "loss": 0.9226,
+      "loss": 0.925,
       "step": 67480
     },
     {
       "epoch": 202.67,
-      "grad_norm": 8.994667053222656,
+      "grad_norm": 8.49370288848877,
       "learning_rate": 3.2442442442442447e-06,
-      "loss": 0.937,
+      "loss": 0.9217,
       "step": 67490
     },
     {
       "epoch": 202.7,
-      "grad_norm": 7.816835403442383,
+      "grad_norm": 7.271054267883301,
       "learning_rate": 3.2432432432432437e-06,
-      "loss": 0.921,
+      "loss": 0.9262,
       "step": 67500
     },
     {
       "epoch": 202.73,
-      "grad_norm": 7.112462520599365,
+      "grad_norm": 6.067651271820068,
       "learning_rate": 3.2422422422422423e-06,
-      "loss": 0.8805,
+      "loss": 0.8754,
       "step": 67510
     },
     {
       "epoch": 202.76,
-      "grad_norm": 6.501965045928955,
+      "grad_norm": 6.013833522796631,
       "learning_rate": 3.2412412412412413e-06,
-      "loss": 1.0501,
+      "loss": 1.0258,
       "step": 67520
     },
     {
       "epoch": 202.79,
-      "grad_norm": 6.808049201965332,
+      "grad_norm": 7.282985687255859,
       "learning_rate": 3.2402402402402403e-06,
-      "loss": 1.0392,
+      "loss": 1.0413,
       "step": 67530
     },
     {
       "epoch": 202.82,
-      "grad_norm": 6.27401876449585,
+      "grad_norm": 6.192715644836426,
       "learning_rate": 3.2392392392392398e-06,
-      "loss": 1.027,
+      "loss": 1.0152,
       "step": 67540
     },
     {
       "epoch": 202.85,
-      "grad_norm": 5.948026180267334,
+      "grad_norm": 5.446735858917236,
       "learning_rate": 3.2382382382382384e-06,
-      "loss": 0.959,
+      "loss": 0.9549,
       "step": 67550
     },
     {
       "epoch": 202.88,
-      "grad_norm": 6.476720333099365,
+      "grad_norm": 6.065450191497803,
       "learning_rate": 3.2372372372372374e-06,
-      "loss": 0.946,
+      "loss": 0.94,
       "step": 67560
     },
     {
       "epoch": 202.91,
-      "grad_norm": 7.47920036315918,
+      "grad_norm": 8.553975105285645,
       "learning_rate": 3.2362362362362364e-06,
-      "loss": 0.9995,
+      "loss": 0.998,
       "step": 67570
     },
     {
       "epoch": 202.94,
-      "grad_norm": 5.066440105438232,
+      "grad_norm": 8.22885799407959,
       "learning_rate": 3.235235235235236e-06,
-      "loss": 0.9345,
+      "loss": 0.9438,
       "step": 67580
     },
     {
       "epoch": 202.97,
-      "grad_norm": 6.421878814697266,
+      "grad_norm": 7.264782428741455,
       "learning_rate": 3.234234234234235e-06,
-      "loss": 0.9764,
+      "loss": 0.973,
       "step": 67590
     },
     {
       "epoch": 203.0,
-      "eval_accuracy": 0.8198,
-      "eval_loss": 0.5897536277770996,
-      "eval_runtime": 5.5104,
-      "eval_samples_per_second": 1814.755,
-      "eval_steps_per_second": 7.259,
+      "eval_accuracy": 0.8229,
+      "eval_loss": 0.5869828462600708,
+      "eval_runtime": 5.0916,
+      "eval_samples_per_second": 1964.011,
+      "eval_steps_per_second": 7.856,
       "step": 67599
     },
     {
       "epoch": 203.0,
-      "grad_norm": 7.527539253234863,
+      "grad_norm": 8.036152839660645,
       "learning_rate": 3.2332332332332335e-06,
-      "loss": 1.0859,
+      "loss": 1.1008,
       "step": 67600
     },
     {
       "epoch": 203.03,
-      "grad_norm": 6.032872200012207,
+      "grad_norm": 10.559889793395996,
       "learning_rate": 3.2322322322322325e-06,
-      "loss": 1.0121,
+      "loss": 1.0159,
       "step": 67610
     },
     {
       "epoch": 203.06,
-      "grad_norm": 9.38195514678955,
+      "grad_norm": 6.371625900268555,
       "learning_rate": 3.2312312312312315e-06,
-      "loss": 0.942,
+      "loss": 0.9342,
       "step": 67620
     },
     {
       "epoch": 203.09,
-      "grad_norm": 5.8065619468688965,
+      "grad_norm": 5.753761291503906,
       "learning_rate": 3.230230230230231e-06,
-      "loss": 0.9209,
+      "loss": 0.9156,
       "step": 67630
     },
     {
       "epoch": 203.12,
-      "grad_norm": 8.643946647644043,
+      "grad_norm": 13.578285217285156,
       "learning_rate": 3.229229229229229e-06,
-      "loss": 0.912,
+      "loss": 0.9123,
       "step": 67640
     },
     {
       "epoch": 203.15,
-      "grad_norm": 6.42996072769165,
+      "grad_norm": 7.86279821395874,
       "learning_rate": 3.2282282282282286e-06,
-      "loss": 0.9376,
+      "loss": 0.9347,
       "step": 67650
     },
     {
       "epoch": 203.18,
-      "grad_norm": 7.1373610496521,
+      "grad_norm": 7.334069728851318,
       "learning_rate": 3.2272272272272276e-06,
-      "loss": 0.8658,
+      "loss": 0.8594,
       "step": 67660
     },
     {
       "epoch": 203.21,
-      "grad_norm": 5.617201805114746,
+      "grad_norm": 6.128842830657959,
       "learning_rate": 3.2262262262262266e-06,
-      "loss": 0.9782,
+      "loss": 0.9848,
       "step": 67670
     },
     {
       "epoch": 203.24,
-      "grad_norm": 6.766232490539551,
+      "grad_norm": 8.411584854125977,
       "learning_rate": 3.225225225225225e-06,
-      "loss": 0.9234,
+      "loss": 0.9232,
       "step": 67680
     },
     {
       "epoch": 203.27,
-      "grad_norm": 5.231124401092529,
+      "grad_norm": 4.849358558654785,
       "learning_rate": 3.2242242242242246e-06,
-      "loss": 0.953,
+      "loss": 0.9603,
       "step": 67690
     },
     {
       "epoch": 203.3,
-      "grad_norm": 9.157536506652832,
+      "grad_norm": 7.968724250793457,
       "learning_rate": 3.2232232232232236e-06,
-      "loss": 0.958,
+      "loss": 0.9578,
       "step": 67700
     },
     {
       "epoch": 203.33,
-      "grad_norm": 5.8346381187438965,
+      "grad_norm": 7.024089813232422,
       "learning_rate": 3.2222222222222227e-06,
-      "loss": 0.9715,
+      "loss": 0.9737,
       "step": 67710
     },
     {
       "epoch": 203.36,
-      "grad_norm": 8.016871452331543,
+      "grad_norm": 7.102033615112305,
       "learning_rate": 3.2212212212212213e-06,
-      "loss": 0.9675,
+      "loss": 0.9641,
       "step": 67720
     },
     {
       "epoch": 203.39,
-      "grad_norm": 5.6558380126953125,
+      "grad_norm": 6.555839538574219,
       "learning_rate": 3.2202202202202203e-06,
-      "loss": 0.9544,
+      "loss": 0.951,
       "step": 67730
     },
     {
       "epoch": 203.42,
-      "grad_norm": 7.688144683837891,
+      "grad_norm": 7.428567886352539,
       "learning_rate": 3.2192192192192197e-06,
-      "loss": 0.9177,
+      "loss": 0.9149,
       "step": 67740
     },
     {
       "epoch": 203.45,
-      "grad_norm": 5.196394443511963,
+      "grad_norm": 6.4125847816467285,
       "learning_rate": 3.2182182182182187e-06,
-      "loss": 1.0244,
+      "loss": 1.0088,
       "step": 67750
     },
     {
       "epoch": 203.48,
-      "grad_norm": 4.609150409698486,
+      "grad_norm": 4.990539073944092,
       "learning_rate": 3.2172172172172173e-06,
-      "loss": 0.9253,
+      "loss": 0.9236,
       "step": 67760
     },
     {
       "epoch": 203.51,
-      "grad_norm": 6.804192543029785,
+      "grad_norm": 8.210829734802246,
       "learning_rate": 3.2162162162162164e-06,
-      "loss": 0.9772,
+      "loss": 0.9678,
       "step": 67770
     },
     {
       "epoch": 203.54,
-      "grad_norm": 7.834531784057617,
+      "grad_norm": 9.421709060668945,
       "learning_rate": 3.2152152152152154e-06,
-      "loss": 0.9603,
+      "loss": 0.9718,
       "step": 67780
     },
     {
       "epoch": 203.57,
-      "grad_norm": 6.848029136657715,
+      "grad_norm": 5.505189895629883,
       "learning_rate": 3.214214214214215e-06,
-      "loss": 0.9502,
+      "loss": 0.9574,
       "step": 67790
     },
     {
       "epoch": 203.6,
-      "grad_norm": 6.728769779205322,
+      "grad_norm": 6.2151198387146,
       "learning_rate": 3.2132132132132134e-06,
-      "loss": 0.9709,
+      "loss": 0.9707,
       "step": 67800
     },
     {
       "epoch": 203.63,
-      "grad_norm": 7.900119781494141,
+      "grad_norm": 6.697146892547607,
       "learning_rate": 3.2122122122122124e-06,
-      "loss": 0.9538,
+      "loss": 0.9429,
       "step": 67810
     },
     {
       "epoch": 203.66,
-      "grad_norm": 6.626853942871094,
+      "grad_norm": 9.296228408813477,
       "learning_rate": 3.2112112112112114e-06,
-      "loss": 0.9737,
+      "loss": 0.9798,
       "step": 67820
     },
     {
       "epoch": 203.69,
-      "grad_norm": 5.588141918182373,
+      "grad_norm": 7.712438106536865,
       "learning_rate": 3.2102102102102105e-06,
-      "loss": 0.9471,
+      "loss": 0.9393,
       "step": 67830
     },
     {
       "epoch": 203.72,
-      "grad_norm": 7.4419403076171875,
+      "grad_norm": 9.1465425491333,
       "learning_rate": 3.209209209209209e-06,
-      "loss": 0.9818,
+      "loss": 0.9827,
       "step": 67840
     },
     {
       "epoch": 203.75,
-      "grad_norm": 4.939833641052246,
+      "grad_norm": 6.424969673156738,
       "learning_rate": 3.2082082082082085e-06,
-      "loss": 0.9721,
+      "loss": 0.9686,
       "step": 67850
     },
     {
       "epoch": 203.78,
-      "grad_norm": 6.454975128173828,
+      "grad_norm": 7.150141716003418,
       "learning_rate": 3.2072072072072075e-06,
-      "loss": 0.9077,
+      "loss": 0.9021,
       "step": 67860
     },
     {
       "epoch": 203.81,
-      "grad_norm": 6.659292221069336,
+      "grad_norm": 6.952312469482422,
       "learning_rate": 3.2062062062062065e-06,
-      "loss": 0.9246,
+      "loss": 0.9234,
       "step": 67870
     },
     {
       "epoch": 203.84,
-      "grad_norm": 9.287436485290527,
+      "grad_norm": 7.514066696166992,
       "learning_rate": 3.2052052052052056e-06,
-      "loss": 0.9093,
+      "loss": 0.913,
       "step": 67880
     },
     {
       "epoch": 203.87,
-      "grad_norm": 7.42474889755249,
+      "grad_norm": 9.55164623260498,
       "learning_rate": 3.204204204204204e-06,
-      "loss": 0.9169,
+      "loss": 0.9091,
       "step": 67890
     },
     {
       "epoch": 203.9,
-      "grad_norm": 6.301826477050781,
+      "grad_norm": 6.409942150115967,
       "learning_rate": 3.2032032032032036e-06,
-      "loss": 0.8849,
+      "loss": 0.8788,
       "step": 67900
     },
     {
       "epoch": 203.93,
-      "grad_norm": 8.865294456481934,
+      "grad_norm": 9.292878150939941,
       "learning_rate": 3.2022022022022026e-06,
-      "loss": 0.9408,
+      "loss": 0.9387,
       "step": 67910
     },
     {
       "epoch": 203.96,
-      "grad_norm": 5.812084674835205,
+      "grad_norm": 7.036854267120361,
       "learning_rate": 3.2012012012012016e-06,
-      "loss": 0.9563,
+      "loss": 0.956,
       "step": 67920
     },
     {
       "epoch": 203.99,
-      "grad_norm": 8.843096733093262,
+      "grad_norm": 8.8355131149292,
       "learning_rate": 3.2002002002002002e-06,
-      "loss": 0.9557,
+      "loss": 0.9493,
       "step": 67930
     },
     {
       "epoch": 204.0,
-      "eval_accuracy": 0.8239,
-      "eval_loss": 0.5757169723510742,
-      "eval_runtime": 5.7041,
-      "eval_samples_per_second": 1753.13,
-      "eval_steps_per_second": 7.013,
+      "eval_accuracy": 0.8262,
+      "eval_loss": 0.5743384957313538,
+      "eval_runtime": 5.3584,
+      "eval_samples_per_second": 1866.244,
+      "eval_steps_per_second": 7.465,
       "step": 67932
     },
     {
       "epoch": 204.02,
-      "grad_norm": 6.566263198852539,
+      "grad_norm": 7.106286525726318,
       "learning_rate": 3.1991991991991992e-06,
-      "loss": 1.0041,
+      "loss": 0.9941,
       "step": 67940
     },
     {
       "epoch": 204.05,
-      "grad_norm": 7.150760173797607,
+      "grad_norm": 7.594667911529541,
       "learning_rate": 3.1981981981981987e-06,
-      "loss": 1.0251,
+      "loss": 1.0311,
       "step": 67950
     },
     {
       "epoch": 204.08,
-      "grad_norm": 7.530023574829102,
+      "grad_norm": 9.017526626586914,
       "learning_rate": 3.1971971971971977e-06,
-      "loss": 0.9375,
+      "loss": 0.9428,
       "step": 67960
     },
     {
       "epoch": 204.11,
-      "grad_norm": 5.700253486633301,
+      "grad_norm": 6.103317737579346,
       "learning_rate": 3.1961961961961963e-06,
-      "loss": 0.9551,
+      "loss": 0.9494,
       "step": 67970
     },
     {
       "epoch": 204.14,
-      "grad_norm": 6.6335320472717285,
+      "grad_norm": 8.236000061035156,
       "learning_rate": 3.1951951951951953e-06,
-      "loss": 0.9315,
+      "loss": 0.9249,
       "step": 67980
     },
     {
       "epoch": 204.17,
-      "grad_norm": 5.662507057189941,
+      "grad_norm": 6.892899990081787,
       "learning_rate": 3.1941941941941943e-06,
-      "loss": 0.9924,
+      "loss": 0.9921,
       "step": 67990
     },
     {
       "epoch": 204.2,
-      "grad_norm": 7.243688583374023,
+      "grad_norm": 6.057879447937012,
       "learning_rate": 3.1931931931931938e-06,
-      "loss": 0.9901,
+      "loss": 0.9772,
       "step": 68000
     },
     {
       "epoch": 204.23,
-      "grad_norm": 7.238772392272949,
+      "grad_norm": 7.644697189331055,
       "learning_rate": 3.1921921921921924e-06,
-      "loss": 0.9683,
+      "loss": 0.9695,
       "step": 68010
     },
     {
       "epoch": 204.26,
-      "grad_norm": 4.808843612670898,
+      "grad_norm": 4.550437927246094,
       "learning_rate": 3.1911911911911914e-06,
-      "loss": 0.9301,
+      "loss": 0.9187,
       "step": 68020
     },
     {
       "epoch": 204.29,
-      "grad_norm": 5.289605140686035,
+      "grad_norm": 6.460263729095459,
       "learning_rate": 3.1901901901901904e-06,
-      "loss": 0.9525,
+      "loss": 0.9376,
       "step": 68030
     },
     {
       "epoch": 204.32,
-      "grad_norm": 7.600884437561035,
+      "grad_norm": 6.3038482666015625,
       "learning_rate": 3.1891891891891894e-06,
-      "loss": 0.9362,
+      "loss": 0.9246,
       "step": 68040
     },
     {
       "epoch": 204.35,
-      "grad_norm": 6.060369968414307,
+      "grad_norm": 6.359384059906006,
       "learning_rate": 3.188188188188188e-06,
-      "loss": 0.9281,
+      "loss": 0.9326,
       "step": 68050
     },
     {
       "epoch": 204.38,
-      "grad_norm": 5.259487152099609,
+      "grad_norm": 5.778562545776367,
       "learning_rate": 3.1871871871871875e-06,
-      "loss": 0.9124,
+      "loss": 0.9085,
       "step": 68060
     },
     {
       "epoch": 204.41,
-      "grad_norm": 8.999384880065918,
+      "grad_norm": 16.243289947509766,
       "learning_rate": 3.1861861861861865e-06,
-      "loss": 0.9381,
+      "loss": 0.9351,
       "step": 68070
     },
     {
       "epoch": 204.44,
-      "grad_norm": 6.603750228881836,
+      "grad_norm": 10.085220336914062,
       "learning_rate": 3.1851851851851855e-06,
-      "loss": 0.9307,
+      "loss": 0.9198,
       "step": 68080
     },
     {
       "epoch": 204.47,
-      "grad_norm": 7.279811382293701,
+      "grad_norm": 6.564107894897461,
       "learning_rate": 3.184184184184184e-06,
-      "loss": 0.9601,
+      "loss": 0.958,
       "step": 68090
     },
     {
       "epoch": 204.5,
-      "grad_norm": 4.609686374664307,
+      "grad_norm": 5.299106597900391,
       "learning_rate": 3.183183183183183e-06,
-      "loss": 0.9183,
+      "loss": 0.9103,
       "step": 68100
     },
     {
       "epoch": 204.53,
-      "grad_norm": 7.645833969116211,
+      "grad_norm": 8.523272514343262,
       "learning_rate": 3.1821821821821826e-06,
-      "loss": 0.9138,
+      "loss": 0.9141,
       "step": 68110
     },
     {
       "epoch": 204.56,
-      "grad_norm": 7.847474098205566,
+      "grad_norm": 7.6792802810668945,
       "learning_rate": 3.1811811811811816e-06,
-      "loss": 0.9347,
+      "loss": 0.9276,
       "step": 68120
     },
     {
       "epoch": 204.59,
-      "grad_norm": 7.754746437072754,
+      "grad_norm": 7.77717924118042,
       "learning_rate": 3.1801801801801806e-06,
-      "loss": 0.9287,
+      "loss": 0.9248,
       "step": 68130
     },
     {
       "epoch": 204.62,
-      "grad_norm": 7.413646221160889,
+      "grad_norm": 6.1110005378723145,
       "learning_rate": 3.179179179179179e-06,
-      "loss": 0.9434,
+      "loss": 0.9335,
       "step": 68140
     },
     {
       "epoch": 204.65,
-      "grad_norm": 6.200901031494141,
+      "grad_norm": 6.05645227432251,
       "learning_rate": 3.1781781781781786e-06,
-      "loss": 0.9686,
+      "loss": 0.9753,
       "step": 68150
     },
     {
       "epoch": 204.68,
-      "grad_norm": 5.4597296714782715,
+      "grad_norm": 5.939512252807617,
       "learning_rate": 3.1771771771771776e-06,
-      "loss": 0.9417,
+      "loss": 0.9342,
       "step": 68160
     },
     {
       "epoch": 204.71,
-      "grad_norm": 4.928735733032227,
+      "grad_norm": 6.206231594085693,
       "learning_rate": 3.1761761761761767e-06,
-      "loss": 0.9986,
+      "loss": 0.9943,
       "step": 68170
     },
     {
       "epoch": 204.74,
-      "grad_norm": 4.541771411895752,
+      "grad_norm": 4.68715238571167,
       "learning_rate": 3.1751751751751753e-06,
-      "loss": 0.9189,
+      "loss": 0.9204,
       "step": 68180
     },
     {
       "epoch": 204.77,
-      "grad_norm": 7.736173629760742,
+      "grad_norm": 7.897682189941406,
       "learning_rate": 3.1741741741741743e-06,
-      "loss": 0.9743,
+      "loss": 0.9739,
       "step": 68190
     },
     {
       "epoch": 204.8,
-      "grad_norm": 8.660161018371582,
+      "grad_norm": 7.960850715637207,
       "learning_rate": 3.1731731731731737e-06,
-      "loss": 0.9283,
+      "loss": 0.9193,
       "step": 68200
     },
     {
       "epoch": 204.83,
-      "grad_norm": 11.180505752563477,
+      "grad_norm": 11.830942153930664,
       "learning_rate": 3.1721721721721727e-06,
-      "loss": 0.9692,
+      "loss": 0.974,
       "step": 68210
     },
     {
       "epoch": 204.86,
-      "grad_norm": 8.169069290161133,
+      "grad_norm": 6.383305072784424,
       "learning_rate": 3.1711711711711713e-06,
-      "loss": 0.9642,
+      "loss": 0.9648,
       "step": 68220
     },
     {
       "epoch": 204.89,
-      "grad_norm": 5.823331356048584,
+      "grad_norm": 5.159214496612549,
       "learning_rate": 3.1701701701701703e-06,
-      "loss": 0.9498,
+      "loss": 0.9527,
       "step": 68230
     },
     {
       "epoch": 204.92,
-      "grad_norm": 6.235563278198242,
+      "grad_norm": 6.517946720123291,
       "learning_rate": 3.1691691691691694e-06,
-      "loss": 0.8741,
+      "loss": 0.8801,
       "step": 68240
     },
     {
       "epoch": 204.95,
-      "grad_norm": 4.604632377624512,
+      "grad_norm": 6.824657440185547,
       "learning_rate": 3.168168168168169e-06,
       "loss": 0.8933,
       "step": 68250
     },
     {
       "epoch": 204.98,
-      "grad_norm": 6.302891254425049,
+      "grad_norm": 6.205295085906982,
       "learning_rate": 3.1671671671671674e-06,
-      "loss": 0.9073,
+      "loss": 0.9091,
       "step": 68260
     },
     {
       "epoch": 205.0,
-      "eval_accuracy": 0.8227,
-      "eval_loss": 0.5837671756744385,
-      "eval_runtime": 5.5443,
-      "eval_samples_per_second": 1803.661,
-      "eval_steps_per_second": 7.215,
+      "eval_accuracy": 0.8257,
+      "eval_loss": 0.5819355845451355,
+      "eval_runtime": 5.2038,
+      "eval_samples_per_second": 1921.655,
+      "eval_steps_per_second": 7.687,
       "step": 68265
     },
     {
       "epoch": 205.02,
-      "grad_norm": 5.892078876495361,
+      "grad_norm": 6.964521884918213,
       "learning_rate": 3.1661661661661664e-06,
-      "loss": 1.0903,
+      "loss": 1.1213,
       "step": 68270
     },
     {
       "epoch": 205.05,
-      "grad_norm": 6.590725421905518,
+      "grad_norm": 5.843968868255615,
       "learning_rate": 3.1651651651651654e-06,
-      "loss": 0.9167,
+      "loss": 0.9177,
       "step": 68280
     },
     {
       "epoch": 205.08,
-      "grad_norm": 5.641725540161133,
+      "grad_norm": 7.861984729766846,
       "learning_rate": 3.1641641641641645e-06,
-      "loss": 0.9787,
+      "loss": 0.9798,
       "step": 68290
     },
     {
       "epoch": 205.11,
-      "grad_norm": 7.212670803070068,
+      "grad_norm": 10.339973449707031,
       "learning_rate": 3.163163163163163e-06,
-      "loss": 0.9404,
+      "loss": 0.9428,
       "step": 68300
     },
     {
       "epoch": 205.14,
-      "grad_norm": 6.689985752105713,
+      "grad_norm": 6.861701965332031,
       "learning_rate": 3.1621621621621625e-06,
-      "loss": 0.9748,
+      "loss": 0.9654,
       "step": 68310
     },
     {
       "epoch": 205.17,
-      "grad_norm": 6.627857208251953,
+      "grad_norm": 6.862594127655029,
       "learning_rate": 3.1611611611611615e-06,
-      "loss": 0.9775,
+      "loss": 0.969,
       "step": 68320
     },
     {
       "epoch": 205.2,
-      "grad_norm": 8.196226119995117,
+      "grad_norm": 4.314332485198975,
       "learning_rate": 3.1601601601601605e-06,
-      "loss": 0.8655,
+      "loss": 0.8598,
       "step": 68330
     },
     {
       "epoch": 205.23,
-      "grad_norm": 5.730470180511475,
+      "grad_norm": 8.309332847595215,
       "learning_rate": 3.159159159159159e-06,
-      "loss": 1.0111,
+      "loss": 1.0056,
       "step": 68340
     },
     {
       "epoch": 205.26,
-      "grad_norm": 5.938358783721924,
+      "grad_norm": 8.652680397033691,
       "learning_rate": 3.158158158158158e-06,
-      "loss": 0.9843,
+      "loss": 0.9854,
       "step": 68350
     },
     {
       "epoch": 205.29,
-      "grad_norm": 7.605236053466797,
+      "grad_norm": 6.995643138885498,
       "learning_rate": 3.1571571571571576e-06,
-      "loss": 0.92,
+      "loss": 0.916,
       "step": 68360
     },
     {
       "epoch": 205.32,
-      "grad_norm": 8.771563529968262,
+      "grad_norm": 6.593210697174072,
       "learning_rate": 3.1561561561561566e-06,
-      "loss": 0.9416,
+      "loss": 0.9435,
       "step": 68370
     },
     {
       "epoch": 205.35,
-      "grad_norm": 7.579165935516357,
+      "grad_norm": 7.685965061187744,
       "learning_rate": 3.155155155155155e-06,
-      "loss": 0.8937,
+      "loss": 0.8848,
       "step": 68380
     },
     {
       "epoch": 205.38,
-      "grad_norm": 6.398184299468994,
+      "grad_norm": 6.005645275115967,
       "learning_rate": 3.1541541541541542e-06,
-      "loss": 0.9244,
+      "loss": 0.9212,
       "step": 68390
     },
     {
       "epoch": 205.41,
-      "grad_norm": 6.781321048736572,
+      "grad_norm": 8.439756393432617,
       "learning_rate": 3.1531531531531532e-06,
-      "loss": 1.0075,
+      "loss": 1.0103,
       "step": 68400
     },
     {
       "epoch": 205.44,
-      "grad_norm": 8.330238342285156,
+      "grad_norm": 10.699536323547363,
       "learning_rate": 3.1521521521521527e-06,
-      "loss": 0.9728,
+      "loss": 0.9635,
       "step": 68410
     },
     {
       "epoch": 205.47,
-      "grad_norm": 4.9651265144348145,
+      "grad_norm": 6.50493860244751,
       "learning_rate": 3.1511511511511517e-06,
-      "loss": 0.904,
+      "loss": 0.9125,
       "step": 68420
     },
     {
       "epoch": 205.5,
-      "grad_norm": 7.146533012390137,
+      "grad_norm": 8.068642616271973,
       "learning_rate": 3.1501501501501503e-06,
-      "loss": 0.8729,
+      "loss": 0.8646,
       "step": 68430
     },
     {
       "epoch": 205.53,
-      "grad_norm": 5.380227565765381,
+      "grad_norm": 4.1941704750061035,
       "learning_rate": 3.1491491491491493e-06,
-      "loss": 0.9631,
+      "loss": 0.9594,
       "step": 68440
     },
     {
       "epoch": 205.56,
-      "grad_norm": 5.55513334274292,
+      "grad_norm": 5.530524253845215,
       "learning_rate": 3.1481481481481483e-06,
-      "loss": 0.8989,
+      "loss": 0.8955,
       "step": 68450
     },
     {
       "epoch": 205.59,
-      "grad_norm": 8.86141586303711,
+      "grad_norm": 9.452862739562988,
       "learning_rate": 3.1471471471471478e-06,
-      "loss": 0.9982,
+      "loss": 0.9923,
       "step": 68460
     },
     {
       "epoch": 205.62,
-      "grad_norm": 5.462921619415283,
+      "grad_norm": 7.339123249053955,
       "learning_rate": 3.1461461461461464e-06,
-      "loss": 0.9311,
+      "loss": 0.9337,
       "step": 68470
     },
     {
       "epoch": 205.65,
-      "grad_norm": 9.78079891204834,
+      "grad_norm": 8.131109237670898,
       "learning_rate": 3.1451451451451454e-06,
-      "loss": 0.9577,
+      "loss": 0.9567,
       "step": 68480
     },
     {
       "epoch": 205.68,
-      "grad_norm": 5.89876651763916,
+      "grad_norm": 7.1874680519104,
       "learning_rate": 3.1441441441441444e-06,
-      "loss": 0.9526,
+      "loss": 0.9556,
       "step": 68490
     },
     {
       "epoch": 205.71,
-      "grad_norm": 6.958099365234375,
+      "grad_norm": 10.430352210998535,
       "learning_rate": 3.1431431431431434e-06,
-      "loss": 0.9577,
+      "loss": 0.9595,
       "step": 68500
     },
     {
       "epoch": 205.74,
-      "grad_norm": 7.872494220733643,
+      "grad_norm": 8.885042190551758,
       "learning_rate": 3.142142142142142e-06,
-      "loss": 0.9097,
+      "loss": 0.9056,
       "step": 68510
     },
     {
       "epoch": 205.77,
-      "grad_norm": 7.514162063598633,
+      "grad_norm": 6.47009801864624,
       "learning_rate": 3.1411411411411415e-06,
-      "loss": 0.9865,
+      "loss": 0.9823,
       "step": 68520
     },
     {
       "epoch": 205.8,
-      "grad_norm": 8.55041217803955,
+      "grad_norm": 7.0363969802856445,
       "learning_rate": 3.1401401401401405e-06,
-      "loss": 1.0212,
+      "loss": 1.007,
       "step": 68530
     },
     {
       "epoch": 205.83,
-      "grad_norm": 7.162789821624756,
+      "grad_norm": 8.508967399597168,
       "learning_rate": 3.1391391391391395e-06,
-      "loss": 0.9401,
+      "loss": 0.927,
       "step": 68540
     },
     {
       "epoch": 205.86,
-      "grad_norm": 8.742140769958496,
+      "grad_norm": 6.990058422088623,
       "learning_rate": 3.138138138138138e-06,
-      "loss": 0.9336,
+      "loss": 0.9359,
       "step": 68550
     },
     {
       "epoch": 205.89,
-      "grad_norm": 8.137758255004883,
+      "grad_norm": 7.285504341125488,
       "learning_rate": 3.137137137137137e-06,
-      "loss": 0.93,
+      "loss": 0.9238,
       "step": 68560
     },
     {
       "epoch": 205.92,
-      "grad_norm": 6.103230953216553,
+      "grad_norm": 6.657644748687744,
       "learning_rate": 3.1361361361361365e-06,
-      "loss": 0.9025,
+      "loss": 0.8991,
       "step": 68570
     },
     {
       "epoch": 205.95,
-      "grad_norm": 6.550155162811279,
+      "grad_norm": 6.319234848022461,
       "learning_rate": 3.1351351351351356e-06,
-      "loss": 0.951,
+      "loss": 0.9437,
       "step": 68580
     },
     {
       "epoch": 205.98,
-      "grad_norm": 8.28841781616211,
+      "grad_norm": 7.582201957702637,
       "learning_rate": 3.134134134134134e-06,
-      "loss": 0.9087,
+      "loss": 0.9128,
       "step": 68590
     },
     {
       "epoch": 206.0,
-      "eval_accuracy": 0.8234,
-      "eval_loss": 0.5783612132072449,
-      "eval_runtime": 5.3417,
-      "eval_samples_per_second": 1872.052,
-      "eval_steps_per_second": 7.488,
+      "eval_accuracy": 0.8273,
+      "eval_loss": 0.5779500007629395,
+      "eval_runtime": 5.3051,
+      "eval_samples_per_second": 1884.985,
+      "eval_steps_per_second": 7.54,
       "step": 68598
     },
     {
       "epoch": 206.01,
-      "grad_norm": 6.081384658813477,
+      "grad_norm": 6.856604099273682,
       "learning_rate": 3.133133133133133e-06,
-      "loss": 1.1467,
+      "loss": 1.1443,
       "step": 68600
     },
     {
       "epoch": 206.04,
-      "grad_norm": 6.3800835609436035,
+      "grad_norm": 7.3200459480285645,
       "learning_rate": 3.132132132132132e-06,
-      "loss": 0.869,
+      "loss": 0.876,
       "step": 68610
     },
     {
       "epoch": 206.07,
-      "grad_norm": 5.0156450271606445,
+      "grad_norm": 6.438960552215576,
       "learning_rate": 3.1311311311311316e-06,
-      "loss": 0.9391,
+      "loss": 0.9254,
       "step": 68620
     },
     {
       "epoch": 206.1,
-      "grad_norm": 6.770200729370117,
+      "grad_norm": 8.327792167663574,
       "learning_rate": 3.1301301301301302e-06,
-      "loss": 0.9196,
+      "loss": 0.9,
       "step": 68630
     },
     {
       "epoch": 206.13,
-      "grad_norm": 6.11611270904541,
+      "grad_norm": 7.780411720275879,
       "learning_rate": 3.1291291291291293e-06,
-      "loss": 0.9349,
+      "loss": 0.9224,
       "step": 68640
     },
     {
       "epoch": 206.16,
-      "grad_norm": 7.84370231628418,
+      "grad_norm": 9.579241752624512,
       "learning_rate": 3.1281281281281283e-06,
-      "loss": 0.9774,
+      "loss": 0.9705,
       "step": 68650
     },
     {
       "epoch": 206.19,
-      "grad_norm": 5.795220375061035,
+      "grad_norm": 7.311451435089111,
       "learning_rate": 3.1271271271271277e-06,
-      "loss": 0.959,
+      "loss": 0.9458,
       "step": 68660
     },
     {
       "epoch": 206.22,
-      "grad_norm": 5.275188446044922,
+      "grad_norm": 5.750716209411621,
       "learning_rate": 3.1261261261261267e-06,
-      "loss": 0.8631,
+      "loss": 0.8633,
       "step": 68670
     },
     {
       "epoch": 206.25,
-      "grad_norm": 7.370100975036621,
+      "grad_norm": 8.434530258178711,
       "learning_rate": 3.1251251251251253e-06,
-      "loss": 0.9526,
+      "loss": 0.9638,
       "step": 68680
     },
     {
       "epoch": 206.28,
-      "grad_norm": 6.9948320388793945,
+      "grad_norm": 11.412034034729004,
       "learning_rate": 3.1241241241241243e-06,
-      "loss": 0.9492,
+      "loss": 0.9591,
       "step": 68690
     },
     {
       "epoch": 206.31,
-      "grad_norm": 8.285123825073242,
+      "grad_norm": 7.294414520263672,
       "learning_rate": 3.1231231231231234e-06,
-      "loss": 0.9034,
+      "loss": 0.895,
       "step": 68700
     },
     {
       "epoch": 206.34,
-      "grad_norm": 7.2914204597473145,
+      "grad_norm": 7.273905277252197,
       "learning_rate": 3.122122122122123e-06,
-      "loss": 0.9732,
+      "loss": 0.9663,
       "step": 68710
     },
     {
       "epoch": 206.37,
-      "grad_norm": 7.158438205718994,
+      "grad_norm": 7.668645858764648,
       "learning_rate": 3.1211211211211214e-06,
-      "loss": 0.8916,
+      "loss": 0.887,
       "step": 68720
     },
     {
       "epoch": 206.4,
-      "grad_norm": 4.975956916809082,
+      "grad_norm": 4.504548072814941,
       "learning_rate": 3.1201201201201204e-06,
-      "loss": 0.9989,
+      "loss": 1.0088,
       "step": 68730
     },
     {
       "epoch": 206.43,
-      "grad_norm": 5.942568302154541,
+      "grad_norm": 7.672431468963623,
       "learning_rate": 3.1191191191191194e-06,
-      "loss": 0.9623,
+      "loss": 0.9675,
       "step": 68740
     },
     {
       "epoch": 206.46,
-      "grad_norm": 8.515399932861328,
+      "grad_norm": 10.689260482788086,
       "learning_rate": 3.1181181181181185e-06,
-      "loss": 0.9904,
+      "loss": 0.9931,
       "step": 68750
     },
     {
       "epoch": 206.49,
-      "grad_norm": 6.663768291473389,
+      "grad_norm": 6.123605251312256,
       "learning_rate": 3.117117117117117e-06,
-      "loss": 0.9058,
+      "loss": 0.9208,
       "step": 68760
     },
     {
       "epoch": 206.52,
-      "grad_norm": 8.621114730834961,
+      "grad_norm": 6.440463542938232,
       "learning_rate": 3.1161161161161165e-06,
-      "loss": 0.9912,
+      "loss": 0.9952,
       "step": 68770
     },
     {
       "epoch": 206.55,
-      "grad_norm": 8.944323539733887,
+      "grad_norm": 7.64757776260376,
       "learning_rate": 3.1151151151151155e-06,
-      "loss": 0.9656,
+      "loss": 0.9744,
       "step": 68780
     },
     {
       "epoch": 206.58,
-      "grad_norm": 8.782063484191895,
+      "grad_norm": 6.355199337005615,
       "learning_rate": 3.1141141141141145e-06,
-      "loss": 0.9238,
+      "loss": 0.9128,
       "step": 68790
     },
     {
       "epoch": 206.61,
-      "grad_norm": 5.38347864151001,
+      "grad_norm": 5.575047016143799,
       "learning_rate": 3.113113113113113e-06,
-      "loss": 0.9259,
+      "loss": 0.9206,
       "step": 68800
     },
     {
       "epoch": 206.64,
-      "grad_norm": 7.262223243713379,
+      "grad_norm": 6.807714462280273,
       "learning_rate": 3.112112112112112e-06,
-      "loss": 0.9905,
+      "loss": 0.9883,
       "step": 68810
     },
     {
       "epoch": 206.67,
-      "grad_norm": 8.194812774658203,
+      "grad_norm": 6.64863395690918,
       "learning_rate": 3.1111111111111116e-06,
-      "loss": 0.8826,
+      "loss": 0.8794,
       "step": 68820
     },
     {
       "epoch": 206.7,
-      "grad_norm": 5.4751973152160645,
+      "grad_norm": 7.729517936706543,
       "learning_rate": 3.1101101101101106e-06,
-      "loss": 0.9492,
+      "loss": 0.9452,
       "step": 68830
     },
     {
       "epoch": 206.73,
-      "grad_norm": 7.842098236083984,
+      "grad_norm": 6.163824081420898,
       "learning_rate": 3.109109109109109e-06,
-      "loss": 1.0403,
+      "loss": 1.0275,
       "step": 68840
     },
     {
       "epoch": 206.76,
-      "grad_norm": 6.306462287902832,
+      "grad_norm": 5.813175678253174,
       "learning_rate": 3.1081081081081082e-06,
-      "loss": 1.0233,
+      "loss": 1.0248,
       "step": 68850
     },
     {
       "epoch": 206.79,
-      "grad_norm": 9.100236892700195,
+      "grad_norm": 8.333999633789062,
       "learning_rate": 3.1071071071071072e-06,
-      "loss": 0.9924,
+      "loss": 0.9922,
       "step": 68860
     },
     {
       "epoch": 206.82,
-      "grad_norm": 6.449151515960693,
+      "grad_norm": 5.891481399536133,
       "learning_rate": 3.1061061061061067e-06,
-      "loss": 0.8867,
+      "loss": 0.8903,
       "step": 68870
     },
     {
       "epoch": 206.85,
-      "grad_norm": 8.103368759155273,
+      "grad_norm": 8.018962860107422,
       "learning_rate": 3.1051051051051053e-06,
-      "loss": 0.9753,
+      "loss": 0.9666,
       "step": 68880
     },
     {
       "epoch": 206.88,
-      "grad_norm": 7.204677104949951,
+      "grad_norm": 7.908066749572754,
       "learning_rate": 3.1041041041041043e-06,
-      "loss": 0.938,
+      "loss": 0.9367,
       "step": 68890
     },
     {
       "epoch": 206.91,
-      "grad_norm": 5.239597320556641,
+      "grad_norm": 6.368007183074951,
       "learning_rate": 3.1031031031031033e-06,
-      "loss": 0.9234,
+      "loss": 0.9152,
       "step": 68900
     },
     {
       "epoch": 206.94,
-      "grad_norm": 6.58310604095459,
+      "grad_norm": 6.617549896240234,
       "learning_rate": 3.1021021021021023e-06,
-      "loss": 0.8865,
+      "loss": 0.8863,
       "step": 68910
     },
     {
       "epoch": 206.97,
-      "grad_norm": 4.605504989624023,
+      "grad_norm": 5.381767272949219,
       "learning_rate": 3.1011011011011018e-06,
-      "loss": 0.9994,
+      "loss": 1.0005,
       "step": 68920
     },
     {
       "epoch": 207.0,
-      "grad_norm": 6.946932792663574,
+      "grad_norm": 5.824578762054443,
       "learning_rate": 3.1001001001001004e-06,
-      "loss": 0.9194,
+      "loss": 0.9142,
       "step": 68930
     },
     {
       "epoch": 207.0,
-      "eval_accuracy": 0.8212,
-      "eval_loss": 0.5789391398429871,
-      "eval_runtime": 5.3883,
-      "eval_samples_per_second": 1855.888,
-      "eval_steps_per_second": 7.424,
+      "eval_accuracy": 0.8248,
+      "eval_loss": 0.5760163068771362,
+      "eval_runtime": 5.4386,
+      "eval_samples_per_second": 1838.702,
+      "eval_steps_per_second": 7.355,
       "step": 68931
     },
     {
       "epoch": 207.03,
-      "grad_norm": 5.8618645668029785,
+      "grad_norm": 6.858852863311768,
       "learning_rate": 3.0990990990990994e-06,
-      "loss": 1.1196,
+      "loss": 1.0992,
       "step": 68940
     },
     {
       "epoch": 207.06,
-      "grad_norm": 6.267481803894043,
+      "grad_norm": 7.078822135925293,
       "learning_rate": 3.0980980980980984e-06,
-      "loss": 0.9803,
+      "loss": 0.9838,
       "step": 68950
     },
     {
       "epoch": 207.09,
-      "grad_norm": 7.160161018371582,
+      "grad_norm": 7.014674663543701,
       "learning_rate": 3.0970970970970974e-06,
-      "loss": 0.9569,
+      "loss": 0.9579,
       "step": 68960
     },
     {
       "epoch": 207.12,
-      "grad_norm": 7.122020244598389,
+      "grad_norm": 6.940105438232422,
       "learning_rate": 3.096096096096096e-06,
-      "loss": 0.9888,
+      "loss": 0.9806,
       "step": 68970
     },
     {
       "epoch": 207.15,
-      "grad_norm": 8.340481758117676,
+      "grad_norm": 7.829718112945557,
       "learning_rate": 3.0950950950950955e-06,
-      "loss": 0.9704,
+      "loss": 0.9677,
       "step": 68980
     },
     {
       "epoch": 207.18,
-      "grad_norm": 12.144529342651367,
+      "grad_norm": 12.351561546325684,
       "learning_rate": 3.0940940940940945e-06,
-      "loss": 0.9452,
+      "loss": 0.9458,
       "step": 68990
     },
     {
       "epoch": 207.21,
-      "grad_norm": 5.8381547927856445,
+      "grad_norm": 6.104592800140381,
       "learning_rate": 3.0930930930930935e-06,
-      "loss": 0.8984,
+      "loss": 0.8978,
       "step": 69000
     },
     {
       "epoch": 207.24,
-      "grad_norm": 7.483870983123779,
+      "grad_norm": 6.2907490730285645,
       "learning_rate": 3.092092092092092e-06,
-      "loss": 0.9417,
+      "loss": 0.9462,
       "step": 69010
     },
     {
       "epoch": 207.27,
-      "grad_norm": 8.575736999511719,
+      "grad_norm": 7.346668243408203,
       "learning_rate": 3.091091091091091e-06,
-      "loss": 0.9967,
+      "loss": 0.9852,
       "step": 69020
     },
     {
       "epoch": 207.3,
-      "grad_norm": 5.6348676681518555,
+      "grad_norm": 4.530812740325928,
       "learning_rate": 3.0900900900900905e-06,
-      "loss": 0.9779,
+      "loss": 0.9741,
       "step": 69030
     },
     {
       "epoch": 207.33,
-      "grad_norm": 6.080043315887451,
+      "grad_norm": 5.434144496917725,
       "learning_rate": 3.0890890890890896e-06,
-      "loss": 0.918,
+      "loss": 0.9212,
       "step": 69040
     },
     {
       "epoch": 207.36,
-      "grad_norm": 3.895646810531616,
+      "grad_norm": 7.149373531341553,
       "learning_rate": 3.088088088088088e-06,
-      "loss": 0.9256,
+      "loss": 0.9321,
       "step": 69050
     },
     {
       "epoch": 207.39,
-      "grad_norm": 8.321868896484375,
+      "grad_norm": 7.312071323394775,
       "learning_rate": 3.087087087087087e-06,
-      "loss": 0.9984,
+      "loss": 0.9999,
       "step": 69060
     },
     {
       "epoch": 207.42,
-      "grad_norm": 8.827521324157715,
+      "grad_norm": 10.896991729736328,
       "learning_rate": 3.086086086086086e-06,
-      "loss": 0.8939,
+      "loss": 0.9041,
       "step": 69070
     },
     {
       "epoch": 207.45,
-      "grad_norm": 5.814652442932129,
+      "grad_norm": 6.403542518615723,
       "learning_rate": 3.0850850850850856e-06,
-      "loss": 0.9531,
+      "loss": 0.9465,
       "step": 69080
     },
     {
       "epoch": 207.48,
-      "grad_norm": 5.153205871582031,
+      "grad_norm": 5.313580513000488,
       "learning_rate": 3.0840840840840842e-06,
-      "loss": 0.9513,
+      "loss": 0.9517,
       "step": 69090
     },
     {
       "epoch": 207.51,
-      "grad_norm": 7.836008548736572,
+      "grad_norm": 11.527223587036133,
       "learning_rate": 3.0830830830830832e-06,
-      "loss": 0.9578,
+      "loss": 0.9597,
       "step": 69100
     },
     {
       "epoch": 207.54,
-      "grad_norm": 6.202759742736816,
+      "grad_norm": 6.886634826660156,
       "learning_rate": 3.0820820820820823e-06,
-      "loss": 0.8852,
+      "loss": 0.8878,
       "step": 69110
     },
     {
       "epoch": 207.57,
-      "grad_norm": 10.179849624633789,
+      "grad_norm": 11.409738540649414,
       "learning_rate": 3.0810810810810817e-06,
-      "loss": 0.9139,
+      "loss": 0.9026,
       "step": 69120
     },
     {
       "epoch": 207.6,
-      "grad_norm": 7.230825424194336,
+      "grad_norm": 6.361260890960693,
       "learning_rate": 3.08008008008008e-06,
-      "loss": 0.9121,
+      "loss": 0.9079,
       "step": 69130
     },
     {
       "epoch": 207.63,
-      "grad_norm": 6.715890884399414,
+      "grad_norm": 7.64415168762207,
       "learning_rate": 3.0790790790790793e-06,
-      "loss": 0.8802,
+      "loss": 0.8834,
       "step": 69140
     },
     {
       "epoch": 207.66,
-      "grad_norm": 7.28183126449585,
+      "grad_norm": 8.979188919067383,
       "learning_rate": 3.0780780780780783e-06,
-      "loss": 1.0094,
+      "loss": 1.0071,
       "step": 69150
     },
     {
       "epoch": 207.69,
-      "grad_norm": 6.783226013183594,
+      "grad_norm": 6.8188862800598145,
       "learning_rate": 3.0770770770770774e-06,
-      "loss": 0.9948,
+      "loss": 1.0009,
       "step": 69160
     },
     {
       "epoch": 207.72,
-      "grad_norm": 8.366267204284668,
+      "grad_norm": 6.603209972381592,
       "learning_rate": 3.076076076076076e-06,
-      "loss": 0.9234,
+      "loss": 0.9188,
       "step": 69170
     },
     {
       "epoch": 207.75,
-      "grad_norm": 8.021089553833008,
+      "grad_norm": 9.942479133605957,
       "learning_rate": 3.0750750750750754e-06,
-      "loss": 0.9469,
+      "loss": 0.9539,
       "step": 69180
     },
     {
       "epoch": 207.78,
-      "grad_norm": 8.012323379516602,
+      "grad_norm": 8.171456336975098,
       "learning_rate": 3.0740740740740744e-06,
-      "loss": 0.9523,
+      "loss": 0.9502,
       "step": 69190
     },
     {
       "epoch": 207.81,
-      "grad_norm": 6.206038475036621,
+      "grad_norm": 6.55024528503418,
       "learning_rate": 3.0730730730730734e-06,
-      "loss": 0.9339,
+      "loss": 0.9168,
       "step": 69200
     },
     {
       "epoch": 207.84,
-      "grad_norm": 6.6083149909973145,
+      "grad_norm": 7.19061279296875,
       "learning_rate": 3.0720720720720725e-06,
-      "loss": 0.9486,
+      "loss": 0.9414,
       "step": 69210
     },
     {
       "epoch": 207.87,
-      "grad_norm": 6.7516560554504395,
+      "grad_norm": 8.153666496276855,
       "learning_rate": 3.071071071071071e-06,
-      "loss": 1.0019,
+      "loss": 1.0009,
       "step": 69220
     },
     {
       "epoch": 207.9,
-      "grad_norm": 6.140419006347656,
+      "grad_norm": 4.484971046447754,
       "learning_rate": 3.0700700700700705e-06,
-      "loss": 0.9706,
+      "loss": 0.972,
       "step": 69230
     },
     {
       "epoch": 207.93,
-      "grad_norm": 6.15202522277832,
+      "grad_norm": 7.0130181312561035,
       "learning_rate": 3.0690690690690695e-06,
-      "loss": 0.9597,
+      "loss": 0.9577,
       "step": 69240
     },
     {
       "epoch": 207.96,
-      "grad_norm": 8.512612342834473,
+      "grad_norm": 7.629635334014893,
       "learning_rate": 3.0680680680680685e-06,
-      "loss": 0.9632,
+      "loss": 0.9587,
       "step": 69250
     },
     {
       "epoch": 207.99,
-      "grad_norm": 6.691050052642822,
+      "grad_norm": 8.440251350402832,
       "learning_rate": 3.067067067067067e-06,
-      "loss": 0.9406,
+      "loss": 0.9364,
       "step": 69260
     },
     {
       "epoch": 208.0,
-      "eval_accuracy": 0.826,
-      "eval_loss": 0.5724438428878784,
-      "eval_runtime": 5.4323,
-      "eval_samples_per_second": 1840.831,
-      "eval_steps_per_second": 7.363,
+      "eval_accuracy": 0.829,
+      "eval_loss": 0.5724502205848694,
+      "eval_runtime": 5.2864,
+      "eval_samples_per_second": 1891.641,
+      "eval_steps_per_second": 7.567,
       "step": 69264
     },
     {
       "epoch": 208.02,
-      "grad_norm": 7.280801296234131,
+      "grad_norm": 9.543071746826172,
       "learning_rate": 3.066066066066066e-06,
-      "loss": 1.0978,
+      "loss": 1.0575,
       "step": 69270
     },
     {
       "epoch": 208.05,
-      "grad_norm": 5.032446384429932,
+      "grad_norm": 4.491264343261719,
       "learning_rate": 3.0650650650650656e-06,
-      "loss": 0.8977,
+      "loss": 0.9002,
       "step": 69280
     },
     {
       "epoch": 208.08,
-      "grad_norm": 5.405389308929443,
+      "grad_norm": 7.6449971199035645,
       "learning_rate": 3.0640640640640646e-06,
-      "loss": 0.948,
+      "loss": 0.9373,
       "step": 69290
     },
     {
       "epoch": 208.11,
-      "grad_norm": 5.380246162414551,
+      "grad_norm": 6.8451056480407715,
       "learning_rate": 3.063063063063063e-06,
-      "loss": 0.9128,
+      "loss": 0.9099,
       "step": 69300
     },
     {
       "epoch": 208.14,
-      "grad_norm": 5.868793964385986,
+      "grad_norm": 6.3333821296691895,
       "learning_rate": 3.062062062062062e-06,
-      "loss": 0.9543,
+      "loss": 0.9595,
       "step": 69310
     },
     {
       "epoch": 208.17,
-      "grad_norm": 7.311581134796143,
+      "grad_norm": 5.5611982345581055,
       "learning_rate": 3.0610610610610612e-06,
-      "loss": 0.8917,
+      "loss": 0.9031,
       "step": 69320
     },
     {
       "epoch": 208.2,
-      "grad_norm": 7.77628231048584,
+      "grad_norm": 8.328622817993164,
       "learning_rate": 3.0600600600600607e-06,
-      "loss": 0.9484,
+      "loss": 0.9354,
       "step": 69330
     },
     {
       "epoch": 208.23,
-      "grad_norm": 5.3106231689453125,
+      "grad_norm": 5.3200812339782715,
       "learning_rate": 3.0590590590590593e-06,
-      "loss": 0.902,
+      "loss": 0.895,
       "step": 69340
     },
     {
       "epoch": 208.26,
-      "grad_norm": 5.7591423988342285,
+      "grad_norm": 6.076287269592285,
       "learning_rate": 3.0580580580580583e-06,
-      "loss": 0.8846,
+      "loss": 0.8772,
       "step": 69350
     },
     {
       "epoch": 208.29,
-      "grad_norm": 6.30707311630249,
+      "grad_norm": 7.742225646972656,
       "learning_rate": 3.0570570570570573e-06,
-      "loss": 0.9498,
+      "loss": 0.9443,
       "step": 69360
     },
     {
       "epoch": 208.32,
-      "grad_norm": 5.3920769691467285,
+      "grad_norm": 6.62296199798584,
       "learning_rate": 3.0560560560560563e-06,
-      "loss": 0.9661,
+      "loss": 0.9577,
       "step": 69370
     },
     {
       "epoch": 208.35,
-      "grad_norm": 6.277547836303711,
+      "grad_norm": 6.002671718597412,
       "learning_rate": 3.055055055055055e-06,
-      "loss": 0.9745,
+      "loss": 0.9749,
       "step": 69380
     },
     {
       "epoch": 208.38,
-      "grad_norm": 8.638790130615234,
+      "grad_norm": 6.228697299957275,
       "learning_rate": 3.0540540540540544e-06,
-      "loss": 0.9975,
+      "loss": 0.9847,
       "step": 69390
     },
     {
       "epoch": 208.41,
-      "grad_norm": 6.728004455566406,
+      "grad_norm": 6.330023765563965,
       "learning_rate": 3.0530530530530534e-06,
-      "loss": 0.912,
+      "loss": 0.9022,
       "step": 69400
     },
     {
       "epoch": 208.44,
-      "grad_norm": 5.2022786140441895,
+      "grad_norm": 4.905588626861572,
       "learning_rate": 3.0520520520520524e-06,
-      "loss": 0.8873,
+      "loss": 0.8762,
       "step": 69410
     },
     {
       "epoch": 208.47,
-      "grad_norm": 7.829473495483398,
+      "grad_norm": 7.412399768829346,
       "learning_rate": 3.051051051051051e-06,
-      "loss": 0.9406,
+      "loss": 0.9333,
       "step": 69420
     },
     {
       "epoch": 208.5,
-      "grad_norm": 7.810801982879639,
+      "grad_norm": 7.913746356964111,
       "learning_rate": 3.05005005005005e-06,
-      "loss": 0.918,
+      "loss": 0.9126,
       "step": 69430
     },
     {
       "epoch": 208.53,
-      "grad_norm": 5.713109016418457,
+      "grad_norm": 6.043558120727539,
       "learning_rate": 3.0490490490490494e-06,
       "loss": 0.9112,
       "step": 69440
     },
     {
       "epoch": 208.56,
-      "grad_norm": 6.016617774963379,
+      "grad_norm": 8.265094757080078,
       "learning_rate": 3.0480480480480485e-06,
-      "loss": 0.9517,
+      "loss": 0.9652,
       "step": 69450
     },
     {
       "epoch": 208.59,
-      "grad_norm": 5.792275428771973,
+      "grad_norm": 7.097805976867676,
       "learning_rate": 3.0470470470470475e-06,
-      "loss": 0.8927,
+      "loss": 0.8935,
       "step": 69460
     },
     {
       "epoch": 208.62,
-      "grad_norm": 6.913943767547607,
+      "grad_norm": 8.384971618652344,
       "learning_rate": 3.046046046046046e-06,
-      "loss": 0.9589,
+      "loss": 0.9626,
       "step": 69470
     },
     {
       "epoch": 208.65,
-      "grad_norm": 5.964015483856201,
+      "grad_norm": 5.547875881195068,
       "learning_rate": 3.045045045045045e-06,
-      "loss": 0.9624,
+      "loss": 0.9587,
       "step": 69480
     },
     {
       "epoch": 208.68,
-      "grad_norm": 7.550143241882324,
+      "grad_norm": 9.43061351776123,
       "learning_rate": 3.0440440440440445e-06,
-      "loss": 0.9324,
+      "loss": 0.9309,
       "step": 69490
     },
     {
       "epoch": 208.71,
-      "grad_norm": 5.055290222167969,
+      "grad_norm": 8.326706886291504,
       "learning_rate": 3.0430430430430436e-06,
-      "loss": 0.9508,
+      "loss": 0.9538,
       "step": 69500
     },
     {
       "epoch": 208.74,
-      "grad_norm": 6.017611980438232,
+      "grad_norm": 5.538358211517334,
       "learning_rate": 3.042042042042042e-06,
-      "loss": 0.9933,
+      "loss": 0.9862,
       "step": 69510
     },
     {
       "epoch": 208.77,
-      "grad_norm": 8.473211288452148,
+      "grad_norm": 6.871679782867432,
       "learning_rate": 3.041041041041041e-06,
-      "loss": 0.9519,
+      "loss": 0.9486,
       "step": 69520
     },
     {
       "epoch": 208.8,
-      "grad_norm": 7.34412956237793,
+      "grad_norm": 6.86476993560791,
       "learning_rate": 3.04004004004004e-06,
-      "loss": 0.9599,
+      "loss": 0.9586,
       "step": 69530
     },
     {
       "epoch": 208.83,
-      "grad_norm": 5.824643611907959,
+      "grad_norm": 6.114565849304199,
       "learning_rate": 3.0390390390390396e-06,
-      "loss": 0.8888,
+      "loss": 0.8832,
       "step": 69540
     },
     {
       "epoch": 208.86,
-      "grad_norm": 8.349218368530273,
+      "grad_norm": 7.698303699493408,
       "learning_rate": 3.0380380380380382e-06,
-      "loss": 0.9287,
+      "loss": 0.918,
       "step": 69550
     },
     {
       "epoch": 208.89,
-      "grad_norm": 5.8702874183654785,
+      "grad_norm": 6.540242671966553,
       "learning_rate": 3.0370370370370372e-06,
-      "loss": 0.9733,
+      "loss": 0.9782,
       "step": 69560
     },
     {
       "epoch": 208.92,
-      "grad_norm": 8.380675315856934,
+      "grad_norm": 5.804974555969238,
       "learning_rate": 3.0360360360360363e-06,
-      "loss": 1.0007,
+      "loss": 1.0068,
       "step": 69570
     },
     {
       "epoch": 208.95,
-      "grad_norm": 5.374027729034424,
+      "grad_norm": 4.88348913192749,
       "learning_rate": 3.0350350350350353e-06,
-      "loss": 0.9409,
+      "loss": 0.9388,
       "step": 69580
     },
     {
       "epoch": 208.98,
-      "grad_norm": 5.058649063110352,
+      "grad_norm": 5.9299726486206055,
       "learning_rate": 3.034034034034034e-06,
-      "loss": 0.8866,
+      "loss": 0.8886,
       "step": 69590
     },
     {
       "epoch": 209.0,
-      "eval_accuracy": 0.8247,
-      "eval_loss": 0.5773116946220398,
-      "eval_runtime": 5.37,
-      "eval_samples_per_second": 1862.212,
-      "eval_steps_per_second": 7.449,
+      "eval_accuracy": 0.8255,
+      "eval_loss": 0.5769652128219604,
+      "eval_runtime": 5.2655,
+      "eval_samples_per_second": 1899.17,
+      "eval_steps_per_second": 7.597,
       "step": 69597
     },
     {
       "epoch": 209.01,
-      "grad_norm": 6.219779968261719,
+      "grad_norm": 6.49110221862793,
       "learning_rate": 3.0330330330330333e-06,
-      "loss": 0.9452,
+      "loss": 0.9628,
       "step": 69600
     },
     {
       "epoch": 209.04,
-      "grad_norm": 4.546178817749023,
+      "grad_norm": 4.639133930206299,
       "learning_rate": 3.0320320320320323e-06,
-      "loss": 0.8449,
+      "loss": 0.8567,
       "step": 69610
     },
     {
       "epoch": 209.07,
-      "grad_norm": 5.37351655960083,
+      "grad_norm": 5.942988395690918,
       "learning_rate": 3.0310310310310314e-06,
-      "loss": 0.9157,
+      "loss": 0.9118,
       "step": 69620
     },
     {
       "epoch": 209.1,
-      "grad_norm": 7.0489959716796875,
+      "grad_norm": 7.658497333526611,
       "learning_rate": 3.03003003003003e-06,
-      "loss": 0.9308,
+      "loss": 0.9491,
       "step": 69630
     },
     {
       "epoch": 209.13,
-      "grad_norm": 5.2226033210754395,
+      "grad_norm": 7.194228172302246,
       "learning_rate": 3.029029029029029e-06,
-      "loss": 0.8928,
+      "loss": 0.8849,
       "step": 69640
     },
     {
       "epoch": 209.16,
-      "grad_norm": 7.199307441711426,
+      "grad_norm": 6.845555782318115,
       "learning_rate": 3.0280280280280284e-06,
-      "loss": 0.8989,
+      "loss": 0.8956,
       "step": 69650
     },
     {
       "epoch": 209.19,
-      "grad_norm": 6.1649370193481445,
+      "grad_norm": 6.954123020172119,
       "learning_rate": 3.0270270270270274e-06,
-      "loss": 1.0293,
+      "loss": 1.0208,
       "step": 69660
     },
     {
       "epoch": 209.22,
-      "grad_norm": 7.130064964294434,
+      "grad_norm": 9.886425018310547,
       "learning_rate": 3.026026026026026e-06,
-      "loss": 0.9378,
+      "loss": 0.9444,
       "step": 69670
     },
     {
       "epoch": 209.25,
-      "grad_norm": 6.312561511993408,
+      "grad_norm": 6.986354351043701,
       "learning_rate": 3.025025025025025e-06,
-      "loss": 0.8656,
+      "loss": 0.8639,
       "step": 69680
     },
     {
       "epoch": 209.28,
-      "grad_norm": 6.228979587554932,
+      "grad_norm": 7.6822662353515625,
       "learning_rate": 3.0240240240240245e-06,
-      "loss": 0.9434,
+      "loss": 0.9369,
       "step": 69690
     },
     {
       "epoch": 209.31,
-      "grad_norm": 5.9282355308532715,
+      "grad_norm": 6.8445539474487305,
       "learning_rate": 3.0230230230230235e-06,
-      "loss": 0.9364,
+      "loss": 0.9278,
       "step": 69700
     },
     {
       "epoch": 209.34,
-      "grad_norm": 8.763745307922363,
+      "grad_norm": 7.7928056716918945,
       "learning_rate": 3.022022022022022e-06,
-      "loss": 0.9244,
+      "loss": 0.9294,
       "step": 69710
     },
     {
       "epoch": 209.37,
-      "grad_norm": 4.523809432983398,
+      "grad_norm": 5.086730480194092,
       "learning_rate": 3.021021021021021e-06,
-      "loss": 1.0203,
+      "loss": 1.0211,
       "step": 69720
     },
     {
       "epoch": 209.4,
-      "grad_norm": 5.472871780395508,
+      "grad_norm": 4.35189151763916,
       "learning_rate": 3.02002002002002e-06,
-      "loss": 0.9558,
+      "loss": 0.9618,
       "step": 69730
     },
     {
       "epoch": 209.43,
-      "grad_norm": 5.88560676574707,
+      "grad_norm": 5.86196231842041,
       "learning_rate": 3.0190190190190196e-06,
-      "loss": 0.9386,
+      "loss": 0.9274,
       "step": 69740
     },
     {
       "epoch": 209.46,
-      "grad_norm": 6.5412163734436035,
+      "grad_norm": 5.844173908233643,
       "learning_rate": 3.0180180180180186e-06,
-      "loss": 0.964,
+      "loss": 0.9654,
       "step": 69750
     },
     {
       "epoch": 209.49,
-      "grad_norm": 8.490387916564941,
+      "grad_norm": 7.318161487579346,
       "learning_rate": 3.017017017017017e-06,
-      "loss": 0.9696,
+      "loss": 0.9611,
       "step": 69760
     },
     {
       "epoch": 209.52,
-      "grad_norm": 6.777596950531006,
+      "grad_norm": 6.912865161895752,
       "learning_rate": 3.016016016016016e-06,
-      "loss": 0.9234,
+      "loss": 0.9192,
       "step": 69770
     },
     {
       "epoch": 209.55,
-      "grad_norm": 4.821336269378662,
+      "grad_norm": 7.610154151916504,
       "learning_rate": 3.0150150150150152e-06,
-      "loss": 0.9461,
+      "loss": 0.9404,
       "step": 69780
     },
     {
       "epoch": 209.58,
-      "grad_norm": 7.548585891723633,
+      "grad_norm": 7.194675445556641,
       "learning_rate": 3.0140140140140147e-06,
-      "loss": 0.9492,
+      "loss": 0.9554,
       "step": 69790
     },
     {
       "epoch": 209.61,
-      "grad_norm": 8.158026695251465,
+      "grad_norm": 9.756567001342773,
       "learning_rate": 3.0130130130130133e-06,
-      "loss": 0.9712,
+      "loss": 0.9662,
       "step": 69800
     },
     {
       "epoch": 209.64,
-      "grad_norm": 5.607148170471191,
+      "grad_norm": 7.491135120391846,
       "learning_rate": 3.0120120120120123e-06,
-      "loss": 0.8704,
+      "loss": 0.8542,
       "step": 69810
     },
     {
       "epoch": 209.67,
-      "grad_norm": 7.326943397521973,
+      "grad_norm": 8.11842155456543,
       "learning_rate": 3.0110110110110113e-06,
-      "loss": 0.9879,
+      "loss": 0.9768,
       "step": 69820
     },
     {
       "epoch": 209.7,
-      "grad_norm": 6.06195068359375,
+      "grad_norm": 5.503032207489014,
       "learning_rate": 3.0100100100100103e-06,
-      "loss": 0.9215,
+      "loss": 0.9184,
       "step": 69830
     },
     {
       "epoch": 209.73,
-      "grad_norm": 7.151528835296631,
+      "grad_norm": 6.354783058166504,
       "learning_rate": 3.009009009009009e-06,
-      "loss": 0.9285,
+      "loss": 0.9238,
       "step": 69840
     },
     {
       "epoch": 209.76,
-      "grad_norm": 7.403812408447266,
+      "grad_norm": 6.4209208488464355,
       "learning_rate": 3.0080080080080084e-06,
-      "loss": 0.9418,
+      "loss": 0.944,
       "step": 69850
     },
     {
       "epoch": 209.79,
-      "grad_norm": 12.310784339904785,
+      "grad_norm": 12.36369514465332,
       "learning_rate": 3.0070070070070074e-06,
-      "loss": 0.9033,
+      "loss": 0.902,
       "step": 69860
     },
     {
       "epoch": 209.82,
-      "grad_norm": 4.789976119995117,
+      "grad_norm": 5.929891586303711,
       "learning_rate": 3.0060060060060064e-06,
-      "loss": 0.9681,
+      "loss": 0.965,
       "step": 69870
     },
     {
       "epoch": 209.85,
-      "grad_norm": 6.4490580558776855,
+      "grad_norm": 5.074809551239014,
       "learning_rate": 3.005005005005005e-06,
-      "loss": 0.906,
+      "loss": 0.913,
       "step": 69880
     },
     {
       "epoch": 209.88,
-      "grad_norm": 6.364231586456299,
+      "grad_norm": 5.746668815612793,
       "learning_rate": 3.004004004004004e-06,
-      "loss": 0.8628,
+      "loss": 0.8688,
       "step": 69890
     },
     {
       "epoch": 209.91,
-      "grad_norm": 15.704768180847168,
+      "grad_norm": 14.746356964111328,
       "learning_rate": 3.0030030030030034e-06,
-      "loss": 0.9246,
+      "loss": 0.9197,
       "step": 69900
     },
     {
       "epoch": 209.94,
-      "grad_norm": 12.165377616882324,
+      "grad_norm": 8.448393821716309,
       "learning_rate": 3.0020020020020025e-06,
-      "loss": 0.9593,
+      "loss": 0.9594,
       "step": 69910
     },
     {
       "epoch": 209.97,
-      "grad_norm": 6.081918239593506,
+      "grad_norm": 5.830969333648682,
       "learning_rate": 3.001001001001001e-06,
-      "loss": 0.9816,
+      "loss": 0.9892,
       "step": 69920
     },
     {
       "epoch": 210.0,
-      "grad_norm": 56.03330612182617,
+      "grad_norm": 38.92162322998047,
       "learning_rate": 3e-06,
       "loss": 1.0926,
       "step": 69930
     },
     {
       "epoch": 210.0,
-      "eval_accuracy": 0.8232,
-      "eval_loss": 0.5829669237136841,
-      "eval_runtime": 5.3745,
-      "eval_samples_per_second": 1860.633,
-      "eval_steps_per_second": 7.443,
+      "eval_accuracy": 0.8245,
+      "eval_loss": 0.5825836658477783,
+      "eval_runtime": 5.1259,
+      "eval_samples_per_second": 1950.893,
+      "eval_steps_per_second": 7.804,
       "step": 69930
     },
     {
       "epoch": 210.03,
-      "grad_norm": 7.15715217590332,
+      "grad_norm": 9.997931480407715,
       "learning_rate": 2.998998998998999e-06,
-      "loss": 0.9253,
+      "loss": 0.9386,
       "step": 69940
     },
     {
       "epoch": 210.06,
-      "grad_norm": 5.30460262298584,
+      "grad_norm": 6.606369972229004,
       "learning_rate": 2.9979979979979985e-06,
-      "loss": 0.9145,
+      "loss": 0.9163,
       "step": 69950
     },
     {
       "epoch": 210.09,
-      "grad_norm": 4.930269718170166,
+      "grad_norm": 5.981235027313232,
       "learning_rate": 2.996996996996997e-06,
-      "loss": 0.9486,
+      "loss": 0.9526,
       "step": 69960
     },
     {
       "epoch": 210.12,
-      "grad_norm": 7.956613063812256,
+      "grad_norm": 9.356122970581055,
       "learning_rate": 2.995995995995996e-06,
-      "loss": 1.0303,
+      "loss": 1.028,
       "step": 69970
     },
     {
       "epoch": 210.15,
-      "grad_norm": 7.427796840667725,
+      "grad_norm": 8.7911376953125,
       "learning_rate": 2.994994994994995e-06,
-      "loss": 0.9602,
+      "loss": 0.9713,
       "step": 69980
     },
     {
       "epoch": 210.18,
-      "grad_norm": 5.843893527984619,
+      "grad_norm": 5.167124271392822,
       "learning_rate": 2.993993993993994e-06,
-      "loss": 0.9782,
+      "loss": 0.9668,
       "step": 69990
     },
     {
       "epoch": 210.21,
-      "grad_norm": 7.108913898468018,
+      "grad_norm": 6.594429969787598,
       "learning_rate": 2.9929929929929936e-06,
-      "loss": 0.9677,
+      "loss": 0.9551,
       "step": 70000
     },
     {
       "epoch": 210.24,
-      "grad_norm": 6.257572174072266,
+      "grad_norm": 8.82521915435791,
       "learning_rate": 2.9919919919919922e-06,
-      "loss": 0.9493,
+      "loss": 0.9596,
       "step": 70010
     },
     {
       "epoch": 210.27,
-      "grad_norm": 7.959559917449951,
+      "grad_norm": 9.124839782714844,
       "learning_rate": 2.9909909909909912e-06,
-      "loss": 0.8825,
+      "loss": 0.875,
       "step": 70020
     },
     {
       "epoch": 210.3,
-      "grad_norm": 8.460164070129395,
+      "grad_norm": 6.634910583496094,
       "learning_rate": 2.9899899899899903e-06,
-      "loss": 0.9214,
+      "loss": 0.92,
       "step": 70030
     },
     {
       "epoch": 210.33,
-      "grad_norm": 6.806526184082031,
+      "grad_norm": 5.642166614532471,
       "learning_rate": 2.9889889889889893e-06,
-      "loss": 0.9877,
+      "loss": 0.9888,
       "step": 70040
     },
     {
       "epoch": 210.36,
-      "grad_norm": 8.267486572265625,
+      "grad_norm": 9.527459144592285,
       "learning_rate": 2.987987987987988e-06,
-      "loss": 0.9398,
+      "loss": 0.9363,
       "step": 70050
     },
     {
       "epoch": 210.39,
-      "grad_norm": 8.185225486755371,
+      "grad_norm": 8.675488471984863,
       "learning_rate": 2.9869869869869873e-06,
-      "loss": 0.9322,
+      "loss": 0.9354,
       "step": 70060
     },
     {
       "epoch": 210.42,
-      "grad_norm": 8.960143089294434,
+      "grad_norm": 11.113280296325684,
       "learning_rate": 2.9859859859859863e-06,
-      "loss": 0.8817,
+      "loss": 0.8873,
       "step": 70070
     },
     {
       "epoch": 210.45,
-      "grad_norm": 4.878829002380371,
+      "grad_norm": 5.079079627990723,
       "learning_rate": 2.9849849849849854e-06,
-      "loss": 0.9338,
+      "loss": 0.9356,
       "step": 70080
     },
     {
       "epoch": 210.48,
-      "grad_norm": 13.309867858886719,
+      "grad_norm": 9.241683959960938,
       "learning_rate": 2.983983983983984e-06,
-      "loss": 0.9517,
+      "loss": 0.9594,
       "step": 70090
     },
     {
       "epoch": 210.51,
-      "grad_norm": 4.566751003265381,
+      "grad_norm": 4.997034072875977,
       "learning_rate": 2.982982982982983e-06,
-      "loss": 1.0005,
+      "loss": 0.9983,
       "step": 70100
     },
     {
       "epoch": 210.54,
-      "grad_norm": 5.855860710144043,
+      "grad_norm": 5.677606582641602,
       "learning_rate": 2.9819819819819824e-06,
-      "loss": 0.9174,
+      "loss": 0.9071,
       "step": 70110
     },
     {
       "epoch": 210.57,
-      "grad_norm": 6.159372806549072,
+      "grad_norm": 8.671745300292969,
       "learning_rate": 2.9809809809809814e-06,
-      "loss": 0.9443,
+      "loss": 0.96,
       "step": 70120
     },
     {
       "epoch": 210.6,
-      "grad_norm": 7.412740707397461,
+      "grad_norm": 7.380494117736816,
       "learning_rate": 2.97997997997998e-06,
-      "loss": 0.9612,
+      "loss": 0.9538,
       "step": 70130
     },
     {
       "epoch": 210.63,
-      "grad_norm": 7.842422962188721,
+      "grad_norm": 7.461397171020508,
       "learning_rate": 2.978978978978979e-06,
-      "loss": 0.9398,
+      "loss": 0.9371,
       "step": 70140
     },
     {
       "epoch": 210.66,
-      "grad_norm": 7.116514205932617,
+      "grad_norm": 10.31419563293457,
       "learning_rate": 2.9779779779779785e-06,
-      "loss": 1.0717,
+      "loss": 1.0687,
       "step": 70150
     },
     {
       "epoch": 210.69,
-      "grad_norm": 8.88432502746582,
+      "grad_norm": 7.26830530166626,
       "learning_rate": 2.9769769769769775e-06,
-      "loss": 0.971,
+      "loss": 0.9785,
       "step": 70160
     },
     {
       "epoch": 210.72,
-      "grad_norm": 5.881826877593994,
+      "grad_norm": 7.58211612701416,
       "learning_rate": 2.975975975975976e-06,
-      "loss": 0.8979,
+      "loss": 0.9082,
       "step": 70170
     },
     {
       "epoch": 210.75,
-      "grad_norm": 6.739663124084473,
+      "grad_norm": 7.844005584716797,
       "learning_rate": 2.974974974974975e-06,
-      "loss": 1.0062,
+      "loss": 0.9985,
       "step": 70180
     },
     {
       "epoch": 210.78,
-      "grad_norm": 7.539654731750488,
+      "grad_norm": 7.471923828125,
       "learning_rate": 2.973973973973974e-06,
-      "loss": 0.9146,
+      "loss": 0.9132,
       "step": 70190
     },
     {
       "epoch": 210.81,
-      "grad_norm": 10.696638107299805,
+      "grad_norm": 7.532835960388184,
       "learning_rate": 2.9729729729729736e-06,
-      "loss": 0.8772,
+      "loss": 0.8721,
       "step": 70200
     },
     {
       "epoch": 210.84,
-      "grad_norm": 6.0039825439453125,
+      "grad_norm": 6.859649658203125,
       "learning_rate": 2.971971971971972e-06,
-      "loss": 0.9091,
+      "loss": 0.9053,
       "step": 70210
     },
     {
       "epoch": 210.87,
-      "grad_norm": 8.066340446472168,
+      "grad_norm": 7.344182014465332,
       "learning_rate": 2.970970970970971e-06,
-      "loss": 0.9716,
+      "loss": 0.9743,
       "step": 70220
     },
     {
       "epoch": 210.9,
-      "grad_norm": 10.904821395874023,
+      "grad_norm": 6.098062038421631,
       "learning_rate": 2.96996996996997e-06,
-      "loss": 0.8959,
+      "loss": 0.9,
       "step": 70230
     },
     {
       "epoch": 210.93,
-      "grad_norm": 7.852315902709961,
+      "grad_norm": 6.560287952423096,
       "learning_rate": 2.9689689689689692e-06,
-      "loss": 1.0693,
+      "loss": 1.0732,
       "step": 70240
     },
     {
       "epoch": 210.96,
-      "grad_norm": 8.371537208557129,
+      "grad_norm": 10.057839393615723,
       "learning_rate": 2.967967967967968e-06,
-      "loss": 0.8826,
+      "loss": 0.8928,
       "step": 70250
     },
     {
       "epoch": 210.99,
-      "grad_norm": 6.473704814910889,
+      "grad_norm": 6.830259799957275,
       "learning_rate": 2.9669669669669673e-06,
-      "loss": 0.9185,
+      "loss": 0.9182,
       "step": 70260
     },
     {
       "epoch": 211.0,
-      "eval_accuracy": 0.8246,
-      "eval_loss": 0.5779848694801331,
-      "eval_runtime": 5.2734,
-      "eval_samples_per_second": 1896.324,
-      "eval_steps_per_second": 7.585,
+      "eval_accuracy": 0.8254,
+      "eval_loss": 0.5784494876861572,
+      "eval_runtime": 5.2413,
+      "eval_samples_per_second": 1907.918,
+      "eval_steps_per_second": 7.632,
       "step": 70263
     },
     {
       "epoch": 211.02,
-      "grad_norm": 5.748218059539795,
+      "grad_norm": 6.286260604858398,
       "learning_rate": 2.9659659659659663e-06,
-      "loss": 1.1098,
+      "loss": 1.0928,
       "step": 70270
     },
     {
       "epoch": 211.05,
-      "grad_norm": 5.554972171783447,
+      "grad_norm": 5.855459690093994,
       "learning_rate": 2.9649649649649653e-06,
-      "loss": 0.875,
+      "loss": 0.8771,
       "step": 70280
     },
     {
       "epoch": 211.08,
-      "grad_norm": 5.947975158691406,
+      "grad_norm": 5.325944900512695,
       "learning_rate": 2.9639639639639643e-06,
-      "loss": 0.9704,
+      "loss": 0.9705,
       "step": 70290
     },
     {
       "epoch": 211.11,
-      "grad_norm": 6.501935005187988,
+      "grad_norm": 9.516717910766602,
       "learning_rate": 2.962962962962963e-06,
-      "loss": 0.9943,
+      "loss": 0.9881,
       "step": 70300
     },
     {
       "epoch": 211.14,
-      "grad_norm": 5.784550666809082,
+      "grad_norm": 5.677460670471191,
       "learning_rate": 2.9619619619619623e-06,
-      "loss": 0.9927,
+      "loss": 0.9921,
       "step": 70310
     },
     {
       "epoch": 211.17,
-      "grad_norm": 7.672462463378906,
+      "grad_norm": 9.1759614944458,
       "learning_rate": 2.9609609609609614e-06,
-      "loss": 0.9302,
+      "loss": 0.9318,
       "step": 70320
     },
     {
       "epoch": 211.2,
-      "grad_norm": 4.940243721008301,
+      "grad_norm": 5.037707805633545,
       "learning_rate": 2.9599599599599604e-06,
-      "loss": 0.9631,
+      "loss": 0.9617,
       "step": 70330
     },
     {
       "epoch": 211.23,
-      "grad_norm": 9.952421188354492,
+      "grad_norm": 13.978858947753906,
       "learning_rate": 2.958958958958959e-06,
-      "loss": 0.9528,
+      "loss": 0.9515,
       "step": 70340
     },
     {
       "epoch": 211.26,
-      "grad_norm": 6.976062297821045,
+      "grad_norm": 9.758131980895996,
       "learning_rate": 2.957957957957958e-06,
-      "loss": 0.9589,
+      "loss": 0.9692,
       "step": 70350
     },
     {
       "epoch": 211.29,
-      "grad_norm": 8.35143756866455,
+      "grad_norm": 5.857941150665283,
       "learning_rate": 2.9569569569569574e-06,
-      "loss": 0.9825,
+      "loss": 0.9837,
       "step": 70360
     },
     {
       "epoch": 211.32,
-      "grad_norm": 6.988082408905029,
+      "grad_norm": 6.024428367614746,
       "learning_rate": 2.9559559559559565e-06,
-      "loss": 0.9069,
+      "loss": 0.905,
       "step": 70370
     },
     {
       "epoch": 211.35,
-      "grad_norm": 7.831850051879883,
+      "grad_norm": 10.744120597839355,
       "learning_rate": 2.954954954954955e-06,
-      "loss": 0.8506,
+      "loss": 0.8392,
       "step": 70380
     },
     {
       "epoch": 211.38,
-      "grad_norm": 7.290890693664551,
+      "grad_norm": 5.325667858123779,
       "learning_rate": 2.953953953953954e-06,
-      "loss": 0.9254,
+      "loss": 0.9223,
       "step": 70390
     },
     {
       "epoch": 211.41,
-      "grad_norm": 5.880447864532471,
+      "grad_norm": 5.704004287719727,
       "learning_rate": 2.952952952952953e-06,
-      "loss": 0.9206,
+      "loss": 0.9218,
       "step": 70400
     },
     {
       "epoch": 211.44,
-      "grad_norm": 5.3326005935668945,
+      "grad_norm": 6.537552833557129,
       "learning_rate": 2.9519519519519525e-06,
-      "loss": 0.9029,
+      "loss": 0.9122,
       "step": 70410
     },
     {
       "epoch": 211.47,
-      "grad_norm": 6.693124771118164,
+      "grad_norm": 8.3465576171875,
       "learning_rate": 2.950950950950951e-06,
-      "loss": 1.0025,
+      "loss": 0.991,
       "step": 70420
     },
     {
       "epoch": 211.5,
-      "grad_norm": 5.837429046630859,
+      "grad_norm": 7.336887836456299,
       "learning_rate": 2.94994994994995e-06,
-      "loss": 0.8814,
+      "loss": 0.8878,
       "step": 70430
     },
     {
       "epoch": 211.53,
-      "grad_norm": 7.947801113128662,
+      "grad_norm": 6.692720890045166,
       "learning_rate": 2.948948948948949e-06,
-      "loss": 0.9081,
+      "loss": 0.9151,
       "step": 70440
     },
     {
       "epoch": 211.56,
-      "grad_norm": 7.30239725112915,
+      "grad_norm": 6.526366233825684,
       "learning_rate": 2.947947947947948e-06,
-      "loss": 0.9039,
+      "loss": 0.9109,
       "step": 70450
     },
     {
       "epoch": 211.59,
-      "grad_norm": 10.763333320617676,
+      "grad_norm": 10.397397994995117,
       "learning_rate": 2.9469469469469468e-06,
-      "loss": 0.9667,
+      "loss": 0.9764,
       "step": 70460
     },
     {
       "epoch": 211.62,
-      "grad_norm": 5.45380163192749,
+      "grad_norm": 6.4652934074401855,
       "learning_rate": 2.9459459459459462e-06,
-      "loss": 0.9711,
+      "loss": 0.9668,
       "step": 70470
     },
     {
       "epoch": 211.65,
-      "grad_norm": 9.560152053833008,
+      "grad_norm": 12.990030288696289,
       "learning_rate": 2.9449449449449452e-06,
-      "loss": 0.9149,
+      "loss": 0.9097,
       "step": 70480
     },
     {
       "epoch": 211.68,
-      "grad_norm": 9.80037784576416,
+      "grad_norm": 8.813663482666016,
       "learning_rate": 2.9439439439439443e-06,
-      "loss": 0.9258,
+      "loss": 0.9304,
       "step": 70490
     },
     {
       "epoch": 211.71,
-      "grad_norm": 6.5132856369018555,
+      "grad_norm": 8.055133819580078,
       "learning_rate": 2.942942942942943e-06,
-      "loss": 0.897,
+      "loss": 0.901,
       "step": 70500
     },
     {
       "epoch": 211.74,
-      "grad_norm": 6.947678089141846,
+      "grad_norm": 7.556239128112793,
       "learning_rate": 2.941941941941942e-06,
-      "loss": 0.8559,
+      "loss": 0.8535,
       "step": 70510
     },
     {
       "epoch": 211.77,
-      "grad_norm": 6.894594192504883,
+      "grad_norm": 12.702672004699707,
       "learning_rate": 2.9409409409409413e-06,
-      "loss": 0.979,
+      "loss": 0.9802,
       "step": 70520
     },
     {
       "epoch": 211.8,
-      "grad_norm": 5.020425319671631,
+      "grad_norm": 8.066522598266602,
       "learning_rate": 2.9399399399399403e-06,
-      "loss": 0.9084,
+      "loss": 0.8934,
       "step": 70530
     },
     {
       "epoch": 211.83,
-      "grad_norm": 7.251352310180664,
+      "grad_norm": 8.69083023071289,
       "learning_rate": 2.9389389389389393e-06,
-      "loss": 0.9307,
+      "loss": 0.9268,
       "step": 70540
     },
     {
       "epoch": 211.86,
-      "grad_norm": 6.632795333862305,
+      "grad_norm": 8.99774169921875,
       "learning_rate": 2.937937937937938e-06,
-      "loss": 0.8712,
+      "loss": 0.8695,
       "step": 70550
     },
     {
       "epoch": 211.89,
-      "grad_norm": 6.250014305114746,
+      "grad_norm": 5.640748500823975,
       "learning_rate": 2.936936936936937e-06,
-      "loss": 0.9827,
+      "loss": 0.9786,
       "step": 70560
     },
     {
       "epoch": 211.92,
-      "grad_norm": 4.7609758377075195,
+      "grad_norm": 5.212128639221191,
       "learning_rate": 2.9359359359359364e-06,
-      "loss": 0.9439,
+      "loss": 0.9418,
       "step": 70570
     },
     {
       "epoch": 211.95,
-      "grad_norm": 8.88786792755127,
+      "grad_norm": 7.803885459899902,
       "learning_rate": 2.9349349349349354e-06,
-      "loss": 0.9262,
+      "loss": 0.9261,
       "step": 70580
     },
     {
       "epoch": 211.98,
-      "grad_norm": 5.682873725891113,
+      "grad_norm": 6.214936256408691,
       "learning_rate": 2.933933933933934e-06,
-      "loss": 0.9636,
+      "loss": 0.9493,
       "step": 70590
     },
     {
       "epoch": 212.0,
-      "eval_accuracy": 0.8252,
-      "eval_loss": 0.5779120326042175,
-      "eval_runtime": 5.3177,
-      "eval_samples_per_second": 1880.496,
-      "eval_steps_per_second": 7.522,
+      "eval_accuracy": 0.8271,
+      "eval_loss": 0.5778210163116455,
+      "eval_runtime": 5.4486,
+      "eval_samples_per_second": 1835.329,
+      "eval_steps_per_second": 7.341,
       "step": 70596
     },
     {
       "epoch": 212.01,
-      "grad_norm": 8.0556001663208,
+      "grad_norm": 6.7266035079956055,
       "learning_rate": 2.932932932932933e-06,
-      "loss": 1.1592,
+      "loss": 1.1256,
       "step": 70600
     },
     {
       "epoch": 212.04,
-      "grad_norm": 7.342432498931885,
+      "grad_norm": 9.317503929138184,
       "learning_rate": 2.931931931931932e-06,
-      "loss": 0.9003,
+      "loss": 0.8949,
       "step": 70610
     },
     {
       "epoch": 212.07,
-      "grad_norm": 7.710801124572754,
+      "grad_norm": 7.37905216217041,
       "learning_rate": 2.9309309309309315e-06,
-      "loss": 0.9407,
+      "loss": 0.943,
       "step": 70620
     },
     {
       "epoch": 212.1,
-      "grad_norm": 4.820013523101807,
+      "grad_norm": 5.489761829376221,
       "learning_rate": 2.92992992992993e-06,
-      "loss": 0.9671,
+      "loss": 0.9659,
       "step": 70630
     },
     {
       "epoch": 212.13,
-      "grad_norm": 8.11202335357666,
+      "grad_norm": 8.509293556213379,
       "learning_rate": 2.928928928928929e-06,
-      "loss": 1.0073,
+      "loss": 1.0012,
       "step": 70640
     },
     {
       "epoch": 212.16,
-      "grad_norm": 7.120957851409912,
+      "grad_norm": 5.341659069061279,
       "learning_rate": 2.927927927927928e-06,
-      "loss": 0.9994,
+      "loss": 0.9882,
       "step": 70650
     },
     {
       "epoch": 212.19,
-      "grad_norm": 6.852572441101074,
+      "grad_norm": 7.081976413726807,
       "learning_rate": 2.9269269269269276e-06,
-      "loss": 0.9296,
+      "loss": 0.9244,
       "step": 70660
     },
     {
       "epoch": 212.22,
-      "grad_norm": 6.536323070526123,
+      "grad_norm": 8.697579383850098,
       "learning_rate": 2.9259259259259257e-06,
-      "loss": 0.9494,
+      "loss": 0.9548,
       "step": 70670
     },
     {
       "epoch": 212.25,
-      "grad_norm": 8.854889869689941,
+      "grad_norm": 5.859123229980469,
       "learning_rate": 2.924924924924925e-06,
-      "loss": 0.9693,
+      "loss": 0.9728,
       "step": 70680
     },
     {
       "epoch": 212.28,
-      "grad_norm": 6.4071784019470215,
+      "grad_norm": 6.2694268226623535,
       "learning_rate": 2.923923923923924e-06,
-      "loss": 0.9142,
+      "loss": 0.9091,
       "step": 70690
     },
     {
       "epoch": 212.31,
-      "grad_norm": 7.703823566436768,
+      "grad_norm": 6.330312728881836,
       "learning_rate": 2.9229229229229232e-06,
-      "loss": 0.9195,
+      "loss": 0.9213,
       "step": 70700
     },
     {
       "epoch": 212.34,
-      "grad_norm": 6.637999057769775,
+      "grad_norm": 6.9649977684021,
       "learning_rate": 2.921921921921922e-06,
-      "loss": 0.9541,
+      "loss": 0.9586,
       "step": 70710
     },
     {
       "epoch": 212.37,
-      "grad_norm": 7.6909074783325195,
+      "grad_norm": 9.405986785888672,
       "learning_rate": 2.9209209209209213e-06,
-      "loss": 0.9477,
+      "loss": 0.9488,
       "step": 70720
     },
     {
       "epoch": 212.4,
-      "grad_norm": 6.581507205963135,
+      "grad_norm": 8.637341499328613,
       "learning_rate": 2.9199199199199203e-06,
-      "loss": 0.9137,
+      "loss": 0.917,
       "step": 70730
     },
     {
       "epoch": 212.43,
-      "grad_norm": 4.401835918426514,
+      "grad_norm": 6.896022319793701,
       "learning_rate": 2.9189189189189193e-06,
-      "loss": 0.9437,
+      "loss": 0.9376,
       "step": 70740
     },
     {
       "epoch": 212.46,
-      "grad_norm": 7.005627632141113,
+      "grad_norm": 9.948395729064941,
       "learning_rate": 2.917917917917918e-06,
-      "loss": 0.8996,
+      "loss": 0.8887,
       "step": 70750
     },
     {
       "epoch": 212.49,
-      "grad_norm": 7.5717902183532715,
+      "grad_norm": 9.184518814086914,
       "learning_rate": 2.916916916916917e-06,
-      "loss": 0.9782,
+      "loss": 0.9787,
       "step": 70760
     },
     {
       "epoch": 212.52,
-      "grad_norm": 6.626132011413574,
+      "grad_norm": 5.221960544586182,
       "learning_rate": 2.9159159159159163e-06,
-      "loss": 0.8938,
+      "loss": 0.8903,
       "step": 70770
     },
     {
       "epoch": 212.55,
-      "grad_norm": 9.742942810058594,
+      "grad_norm": 8.032197952270508,
       "learning_rate": 2.9149149149149154e-06,
-      "loss": 0.971,
+      "loss": 0.9749,
       "step": 70780
     },
     {
       "epoch": 212.58,
-      "grad_norm": 8.823344230651855,
+      "grad_norm": 8.884482383728027,
       "learning_rate": 2.9139139139139144e-06,
-      "loss": 0.9676,
+      "loss": 0.9681,
       "step": 70790
     },
     {
       "epoch": 212.61,
-      "grad_norm": 5.8972601890563965,
+      "grad_norm": 7.063982009887695,
       "learning_rate": 2.912912912912913e-06,
-      "loss": 0.9619,
+      "loss": 0.967,
       "step": 70800
     },
     {
       "epoch": 212.64,
-      "grad_norm": 9.301518440246582,
+      "grad_norm": 10.451930046081543,
       "learning_rate": 2.911911911911912e-06,
-      "loss": 0.9539,
+      "loss": 0.9523,
       "step": 70810
     },
     {
       "epoch": 212.67,
-      "grad_norm": 5.4780120849609375,
+      "grad_norm": 6.929025173187256,
       "learning_rate": 2.9109109109109114e-06,
-      "loss": 0.8893,
+      "loss": 0.9002,
       "step": 70820
     },
     {
       "epoch": 212.7,
-      "grad_norm": 5.719918727874756,
+      "grad_norm": 7.621103286743164,
       "learning_rate": 2.9099099099099105e-06,
-      "loss": 0.9463,
+      "loss": 0.949,
       "step": 70830
     },
     {
       "epoch": 212.73,
-      "grad_norm": 8.476021766662598,
+      "grad_norm": 9.302020072937012,
       "learning_rate": 2.908908908908909e-06,
-      "loss": 0.9214,
+      "loss": 0.9315,
       "step": 70840
     },
     {
       "epoch": 212.76,
-      "grad_norm": 17.804977416992188,
+      "grad_norm": 12.663199424743652,
       "learning_rate": 2.907907907907908e-06,
-      "loss": 0.9917,
+      "loss": 0.9842,
       "step": 70850
     },
     {
       "epoch": 212.79,
-      "grad_norm": 6.449276924133301,
+      "grad_norm": 5.0961198806762695,
       "learning_rate": 2.906906906906907e-06,
-      "loss": 0.9108,
+      "loss": 0.9215,
       "step": 70860
     },
     {
       "epoch": 212.82,
-      "grad_norm": 4.923229217529297,
+      "grad_norm": 6.161968231201172,
       "learning_rate": 2.9059059059059065e-06,
-      "loss": 0.9407,
+      "loss": 0.9498,
       "step": 70870
     },
     {
       "epoch": 212.85,
-      "grad_norm": 5.6815505027771,
+      "grad_norm": 5.63728141784668,
       "learning_rate": 2.904904904904905e-06,
-      "loss": 0.9534,
+      "loss": 0.9549,
       "step": 70880
     },
     {
       "epoch": 212.88,
-      "grad_norm": 5.510833263397217,
+      "grad_norm": 6.0853471755981445,
       "learning_rate": 2.903903903903904e-06,
-      "loss": 0.9398,
+      "loss": 0.9348,
       "step": 70890
     },
     {
       "epoch": 212.91,
-      "grad_norm": 5.4648847579956055,
+      "grad_norm": 8.08102798461914,
       "learning_rate": 2.902902902902903e-06,
-      "loss": 0.8299,
+      "loss": 0.8431,
       "step": 70900
     },
     {
       "epoch": 212.94,
-      "grad_norm": 5.488482475280762,
+      "grad_norm": 5.6815056800842285,
       "learning_rate": 2.901901901901902e-06,
-      "loss": 0.9073,
+      "loss": 0.9096,
       "step": 70910
     },
     {
       "epoch": 212.97,
-      "grad_norm": 6.466689109802246,
+      "grad_norm": 6.148865222930908,
       "learning_rate": 2.9009009009009008e-06,
-      "loss": 0.9503,
+      "loss": 0.9581,
       "step": 70920
     },
     {
       "epoch": 213.0,
-      "eval_accuracy": 0.8242,
-      "eval_loss": 0.5780848860740662,
-      "eval_runtime": 5.2891,
-      "eval_samples_per_second": 1890.679,
-      "eval_steps_per_second": 7.563,
+      "eval_accuracy": 0.8261,
+      "eval_loss": 0.5769987106323242,
+      "eval_runtime": 5.4461,
+      "eval_samples_per_second": 1836.179,
+      "eval_steps_per_second": 7.345,
       "step": 70929
     },
     {
       "epoch": 213.0,
-      "grad_norm": 7.25825309753418,
+      "grad_norm": 6.347087860107422,
       "learning_rate": 2.8998998998999002e-06,
-      "loss": 1.1768,
+      "loss": 1.1448,
       "step": 70930
     },
     {
       "epoch": 213.03,
-      "grad_norm": 8.583785057067871,
+      "grad_norm": 9.027023315429688,
       "learning_rate": 2.8988988988988992e-06,
-      "loss": 0.9226,
+      "loss": 0.9212,
       "step": 70940
     },
     {
       "epoch": 213.06,
-      "grad_norm": 7.139830112457275,
+      "grad_norm": 8.004554748535156,
       "learning_rate": 2.8978978978978983e-06,
-      "loss": 0.9738,
+      "loss": 0.9681,
       "step": 70950
     },
     {
       "epoch": 213.09,
-      "grad_norm": 8.233696937561035,
+      "grad_norm": 6.571313858032227,
       "learning_rate": 2.896896896896897e-06,
       "loss": 0.9487,
       "step": 70960
     },
     {
       "epoch": 213.12,
-      "grad_norm": 6.032533645629883,
+      "grad_norm": 5.649678707122803,
       "learning_rate": 2.895895895895896e-06,
-      "loss": 0.885,
+      "loss": 0.886,
       "step": 70970
     },
     {
       "epoch": 213.15,
-      "grad_norm": 5.894317626953125,
+      "grad_norm": 6.538634777069092,
       "learning_rate": 2.8948948948948953e-06,
-      "loss": 0.9661,
+      "loss": 0.9688,
       "step": 70980
     },
     {
       "epoch": 213.18,
-      "grad_norm": 6.487376689910889,
+      "grad_norm": 5.6969451904296875,
       "learning_rate": 2.8938938938938943e-06,
-      "loss": 0.9257,
+      "loss": 0.9199,
       "step": 70990
     },
     {
       "epoch": 213.21,
-      "grad_norm": 6.366428852081299,
+      "grad_norm": 6.162879943847656,
       "learning_rate": 2.892892892892893e-06,
-      "loss": 0.9217,
+      "loss": 0.9221,
       "step": 71000
     },
     {
       "epoch": 213.24,
-      "grad_norm": 5.76392126083374,
+      "grad_norm": 6.16183614730835,
       "learning_rate": 2.891891891891892e-06,
-      "loss": 0.8504,
+      "loss": 0.8581,
       "step": 71010
     },
     {
       "epoch": 213.27,
-      "grad_norm": 6.625319480895996,
+      "grad_norm": 5.9928669929504395,
       "learning_rate": 2.890890890890891e-06,
-      "loss": 0.9681,
+      "loss": 0.962,
       "step": 71020
     },
     {
       "epoch": 213.3,
-      "grad_norm": 5.63862419128418,
+      "grad_norm": 5.460396766662598,
       "learning_rate": 2.8898898898898904e-06,
-      "loss": 0.9591,
+      "loss": 0.9541,
       "step": 71030
     },
     {
       "epoch": 213.33,
-      "grad_norm": 5.995763778686523,
+      "grad_norm": 6.779233455657959,
       "learning_rate": 2.888888888888889e-06,
-      "loss": 0.9441,
+      "loss": 0.9428,
       "step": 71040
     },
     {
       "epoch": 213.36,
-      "grad_norm": 6.586748123168945,
+      "grad_norm": 5.508744239807129,
       "learning_rate": 2.887887887887888e-06,
-      "loss": 0.9278,
+      "loss": 0.9279,
       "step": 71050
     },
     {
       "epoch": 213.39,
-      "grad_norm": 7.901427745819092,
+      "grad_norm": 8.168306350708008,
       "learning_rate": 2.886886886886887e-06,
-      "loss": 0.9248,
+      "loss": 0.9198,
       "step": 71060
     },
     {
       "epoch": 213.42,
-      "grad_norm": 6.587465286254883,
+      "grad_norm": 6.493582248687744,
       "learning_rate": 2.885885885885886e-06,
-      "loss": 1.0495,
+      "loss": 1.0457,
       "step": 71070
     },
     {
       "epoch": 213.45,
-      "grad_norm": 6.89039945602417,
+      "grad_norm": 6.731003284454346,
       "learning_rate": 2.8848848848848855e-06,
-      "loss": 0.9763,
+      "loss": 0.978,
       "step": 71080
     },
     {
       "epoch": 213.48,
-      "grad_norm": 5.3392462730407715,
+      "grad_norm": 7.448592662811279,
       "learning_rate": 2.883883883883884e-06,
-      "loss": 0.857,
+      "loss": 0.8651,
       "step": 71090
     },
     {
       "epoch": 213.51,
-      "grad_norm": 4.616884708404541,
+      "grad_norm": 4.392862319946289,
       "learning_rate": 2.882882882882883e-06,
-      "loss": 0.9131,
+      "loss": 0.9081,
       "step": 71100
     },
     {
       "epoch": 213.54,
-      "grad_norm": 6.315646648406982,
+      "grad_norm": 6.118627071380615,
       "learning_rate": 2.881881881881882e-06,
-      "loss": 0.9304,
+      "loss": 0.9346,
       "step": 71110
     },
     {
       "epoch": 213.57,
-      "grad_norm": 5.7306060791015625,
+      "grad_norm": 5.452793598175049,
       "learning_rate": 2.8808808808808816e-06,
-      "loss": 0.9594,
+      "loss": 0.9527,
       "step": 71120
     },
     {
       "epoch": 213.6,
-      "grad_norm": 7.248140811920166,
+      "grad_norm": 8.551169395446777,
       "learning_rate": 2.8798798798798797e-06,
-      "loss": 0.965,
+      "loss": 0.9676,
       "step": 71130
     },
     {
       "epoch": 213.63,
-      "grad_norm": 10.00749683380127,
+      "grad_norm": 9.346454620361328,
       "learning_rate": 2.878878878878879e-06,
-      "loss": 0.9017,
+      "loss": 0.9137,
       "step": 71140
     },
     {
       "epoch": 213.66,
-      "grad_norm": 13.83051872253418,
+      "grad_norm": 9.049593925476074,
       "learning_rate": 2.877877877877878e-06,
-      "loss": 0.9546,
+      "loss": 0.9523,
       "step": 71150
     },
     {
       "epoch": 213.69,
-      "grad_norm": 5.985548973083496,
+      "grad_norm": 6.251873016357422,
       "learning_rate": 2.8768768768768772e-06,
-      "loss": 0.9399,
+      "loss": 0.9407,
       "step": 71160
     },
     {
       "epoch": 213.72,
-      "grad_norm": 8.166257858276367,
+      "grad_norm": 7.572188377380371,
       "learning_rate": 2.875875875875876e-06,
-      "loss": 0.9055,
+      "loss": 0.9092,
       "step": 71170
     },
     {
       "epoch": 213.75,
-      "grad_norm": 10.742035865783691,
+      "grad_norm": 11.273395538330078,
       "learning_rate": 2.8748748748748753e-06,
-      "loss": 0.9654,
+      "loss": 0.9639,
       "step": 71180
     },
     {
       "epoch": 213.78,
-      "grad_norm": 6.093210220336914,
+      "grad_norm": 7.599462985992432,
       "learning_rate": 2.8738738738738743e-06,
-      "loss": 0.8983,
+      "loss": 0.8959,
       "step": 71190
     },
     {
       "epoch": 213.81,
-      "grad_norm": 6.56252384185791,
+      "grad_norm": 5.909582614898682,
       "learning_rate": 2.8728728728728733e-06,
-      "loss": 0.9682,
+      "loss": 0.9581,
       "step": 71200
     },
     {
       "epoch": 213.84,
-      "grad_norm": 5.076033592224121,
+      "grad_norm": 5.22104549407959,
       "learning_rate": 2.871871871871872e-06,
-      "loss": 0.9464,
+      "loss": 0.9534,
       "step": 71210
     },
     {
       "epoch": 213.87,
-      "grad_norm": 7.053407669067383,
+      "grad_norm": 6.969611167907715,
       "learning_rate": 2.870870870870871e-06,
       "loss": 0.9545,
       "step": 71220
     },
     {
       "epoch": 213.9,
-      "grad_norm": 9.800572395324707,
+      "grad_norm": 9.0591402053833,
       "learning_rate": 2.8698698698698703e-06,
-      "loss": 0.9346,
+      "loss": 0.926,
       "step": 71230
     },
     {
       "epoch": 213.93,
-      "grad_norm": 6.475651741027832,
+      "grad_norm": 5.738166809082031,
       "learning_rate": 2.8688688688688694e-06,
-      "loss": 0.946,
+      "loss": 0.9442,
       "step": 71240
     },
     {
       "epoch": 213.96,
-      "grad_norm": 7.078562259674072,
+      "grad_norm": 6.363611698150635,
       "learning_rate": 2.867867867867868e-06,
-      "loss": 0.9342,
+      "loss": 0.9313,
       "step": 71250
     },
     {
       "epoch": 213.99,
-      "grad_norm": 6.849552154541016,
+      "grad_norm": 6.039224624633789,
       "learning_rate": 2.866866866866867e-06,
-      "loss": 0.9006,
+      "loss": 0.9033,
       "step": 71260
     },
     {
       "epoch": 214.0,
-      "eval_accuracy": 0.8237,
-      "eval_loss": 0.5856425762176514,
-      "eval_runtime": 5.3686,
-      "eval_samples_per_second": 1862.681,
-      "eval_steps_per_second": 7.451,
+      "eval_accuracy": 0.823,
+      "eval_loss": 0.5862734317779541,
+      "eval_runtime": 5.1687,
+      "eval_samples_per_second": 1934.715,
+      "eval_steps_per_second": 7.739,
       "step": 71262
     },
     {
       "epoch": 214.02,
-      "grad_norm": 10.199888229370117,
+      "grad_norm": 10.285968780517578,
       "learning_rate": 2.865865865865866e-06,
-      "loss": 1.2477,
+      "loss": 1.2463,
       "step": 71270
     },
     {
       "epoch": 214.05,
-      "grad_norm": 6.641946792602539,
+      "grad_norm": 9.792351722717285,
       "learning_rate": 2.8648648648648654e-06,
-      "loss": 0.9586,
+      "loss": 0.9599,
       "step": 71280
     },
     {
       "epoch": 214.08,
-      "grad_norm": 8.621333122253418,
+      "grad_norm": 9.673145294189453,
       "learning_rate": 2.863863863863864e-06,
-      "loss": 0.9178,
+      "loss": 0.9311,
       "step": 71290
     },
     {
       "epoch": 214.11,
-      "grad_norm": 7.953291416168213,
+      "grad_norm": 9.567935943603516,
       "learning_rate": 2.862862862862863e-06,
-      "loss": 0.9662,
+      "loss": 0.9634,
       "step": 71300
     },
     {
       "epoch": 214.14,
-      "grad_norm": 5.455259799957275,
+      "grad_norm": 6.069672107696533,
       "learning_rate": 2.861861861861862e-06,
-      "loss": 0.9635,
+      "loss": 0.958,
       "step": 71310
     },
     {
       "epoch": 214.17,
-      "grad_norm": 6.778347492218018,
+      "grad_norm": 6.033369064331055,
       "learning_rate": 2.860860860860861e-06,
-      "loss": 0.9455,
+      "loss": 0.9527,
       "step": 71320
     },
     {
       "epoch": 214.2,
-      "grad_norm": 8.389677047729492,
+      "grad_norm": 7.833511829376221,
       "learning_rate": 2.8598598598598605e-06,
-      "loss": 0.9296,
+      "loss": 0.919,
       "step": 71330
     },
     {
       "epoch": 214.23,
-      "grad_norm": 7.155240535736084,
+      "grad_norm": 7.207285404205322,
       "learning_rate": 2.858858858858859e-06,
-      "loss": 0.9587,
+      "loss": 0.9637,
       "step": 71340
     },
     {
       "epoch": 214.26,
-      "grad_norm": 10.492780685424805,
+      "grad_norm": 9.01717472076416,
       "learning_rate": 2.857857857857858e-06,
-      "loss": 0.9414,
+      "loss": 0.9393,
       "step": 71350
     },
     {
       "epoch": 214.29,
-      "grad_norm": 5.13774299621582,
+      "grad_norm": 4.7567901611328125,
       "learning_rate": 2.856856856856857e-06,
-      "loss": 0.9476,
+      "loss": 0.941,
       "step": 71360
     },
     {
       "epoch": 214.32,
-      "grad_norm": 8.81657600402832,
+      "grad_norm": 7.509118556976318,
       "learning_rate": 2.855855855855856e-06,
-      "loss": 0.9245,
+      "loss": 0.9152,
       "step": 71370
     },
     {
       "epoch": 214.35,
-      "grad_norm": 7.8307342529296875,
+      "grad_norm": 8.470617294311523,
       "learning_rate": 2.8548548548548548e-06,
-      "loss": 0.9537,
+      "loss": 0.945,
       "step": 71380
     },
     {
       "epoch": 214.38,
-      "grad_norm": 5.653038024902344,
+      "grad_norm": 7.019521236419678,
       "learning_rate": 2.853853853853854e-06,
-      "loss": 0.9906,
+      "loss": 0.9836,
       "step": 71390
     },
     {
       "epoch": 214.41,
-      "grad_norm": 6.601412773132324,
+      "grad_norm": 7.862439155578613,
       "learning_rate": 2.8528528528528532e-06,
-      "loss": 0.8804,
+      "loss": 0.8764,
       "step": 71400
     },
     {
       "epoch": 214.44,
-      "grad_norm": 7.095278739929199,
+      "grad_norm": 5.8733086585998535,
       "learning_rate": 2.8518518518518522e-06,
-      "loss": 0.9467,
+      "loss": 0.9564,
       "step": 71410
     },
     {
       "epoch": 214.47,
-      "grad_norm": 7.323110103607178,
+      "grad_norm": 11.333822250366211,
       "learning_rate": 2.850850850850851e-06,
-      "loss": 0.9612,
+      "loss": 0.9734,
       "step": 71420
     },
     {
       "epoch": 214.5,
-      "grad_norm": 5.868104934692383,
+      "grad_norm": 5.957848072052002,
       "learning_rate": 2.84984984984985e-06,
-      "loss": 0.8894,
+      "loss": 0.8901,
       "step": 71430
     },
     {
       "epoch": 214.53,
-      "grad_norm": 7.155386447906494,
+      "grad_norm": 6.33123254776001,
       "learning_rate": 2.8488488488488493e-06,
-      "loss": 0.9776,
+      "loss": 0.9628,
       "step": 71440
     },
     {
       "epoch": 214.56,
-      "grad_norm": 7.350404739379883,
+      "grad_norm": 8.410058975219727,
       "learning_rate": 2.8478478478478483e-06,
-      "loss": 0.9257,
+      "loss": 0.928,
       "step": 71450
     },
     {
       "epoch": 214.59,
-      "grad_norm": 5.679640769958496,
+      "grad_norm": 10.536490440368652,
       "learning_rate": 2.846846846846847e-06,
-      "loss": 0.9371,
+      "loss": 0.9465,
       "step": 71460
     },
     {
       "epoch": 214.62,
-      "grad_norm": 5.321216106414795,
+      "grad_norm": 6.301913261413574,
       "learning_rate": 2.845845845845846e-06,
-      "loss": 0.9415,
+      "loss": 0.9366,
       "step": 71470
     },
     {
       "epoch": 214.65,
-      "grad_norm": 5.418609619140625,
+      "grad_norm": 7.224236011505127,
       "learning_rate": 2.844844844844845e-06,
-      "loss": 0.9062,
+      "loss": 0.9104,
       "step": 71480
     },
     {
       "epoch": 214.68,
-      "grad_norm": 7.571637153625488,
+      "grad_norm": 6.474426746368408,
       "learning_rate": 2.8438438438438444e-06,
-      "loss": 0.9358,
+      "loss": 0.9385,
       "step": 71490
     },
     {
       "epoch": 214.71,
-      "grad_norm": 5.990438461303711,
+      "grad_norm": 8.294471740722656,
       "learning_rate": 2.842842842842843e-06,
-      "loss": 0.8707,
+      "loss": 0.8799,
       "step": 71500
     },
     {
       "epoch": 214.74,
-      "grad_norm": 7.12021541595459,
+      "grad_norm": 8.47632122039795,
       "learning_rate": 2.841841841841842e-06,
-      "loss": 0.9147,
+      "loss": 0.9195,
       "step": 71510
     },
     {
       "epoch": 214.77,
-      "grad_norm": 7.193082809448242,
+      "grad_norm": 6.9369401931762695,
       "learning_rate": 2.840840840840841e-06,
-      "loss": 0.9657,
+      "loss": 0.9738,
       "step": 71520
     },
     {
       "epoch": 214.8,
-      "grad_norm": 5.8618292808532715,
+      "grad_norm": 5.548161029815674,
       "learning_rate": 2.83983983983984e-06,
-      "loss": 0.9281,
+      "loss": 0.9287,
       "step": 71530
     },
     {
       "epoch": 214.83,
-      "grad_norm": 11.49142074584961,
+      "grad_norm": 8.2123441696167,
       "learning_rate": 2.8388388388388386e-06,
-      "loss": 0.9307,
+      "loss": 0.9341,
       "step": 71540
     },
     {
       "epoch": 214.86,
-      "grad_norm": 5.793619632720947,
+      "grad_norm": 8.470561027526855,
       "learning_rate": 2.837837837837838e-06,
-      "loss": 0.9127,
+      "loss": 0.9067,
       "step": 71550
     },
     {
       "epoch": 214.89,
-      "grad_norm": 8.082571029663086,
+      "grad_norm": 6.721054553985596,
       "learning_rate": 2.836836836836837e-06,
-      "loss": 0.9027,
+      "loss": 0.9112,
       "step": 71560
     },
     {
       "epoch": 214.92,
-      "grad_norm": 8.523369789123535,
+      "grad_norm": 8.80095100402832,
       "learning_rate": 2.835835835835836e-06,
-      "loss": 0.9288,
+      "loss": 0.9159,
       "step": 71570
     },
     {
       "epoch": 214.95,
-      "grad_norm": 5.276920795440674,
+      "grad_norm": 7.342271327972412,
       "learning_rate": 2.8348348348348347e-06,
-      "loss": 0.9547,
+      "loss": 0.9484,
       "step": 71580
     },
     {
       "epoch": 214.98,
-      "grad_norm": 5.977039337158203,
+      "grad_norm": 6.531291961669922,
       "learning_rate": 2.8338338338338337e-06,
-      "loss": 0.9294,
+      "loss": 0.9275,
       "step": 71590
     },
     {
       "epoch": 215.0,
-      "eval_accuracy": 0.8244,
-      "eval_loss": 0.5737177729606628,
-      "eval_runtime": 5.4272,
-      "eval_samples_per_second": 1842.561,
-      "eval_steps_per_second": 7.37,
+      "eval_accuracy": 0.8282,
+      "eval_loss": 0.5729073882102966,
+      "eval_runtime": 5.3117,
+      "eval_samples_per_second": 1882.642,
+      "eval_steps_per_second": 7.531,
       "step": 71595
     },
     {
       "epoch": 215.02,
-      "grad_norm": 7.1488213539123535,
+      "grad_norm": 7.103471279144287,
       "learning_rate": 2.832832832832833e-06,
-      "loss": 1.1331,
+      "loss": 1.1403,
       "step": 71600
     },
     {
       "epoch": 215.05,
-      "grad_norm": 6.437438011169434,
+      "grad_norm": 5.614165306091309,
       "learning_rate": 2.831831831831832e-06,
-      "loss": 0.9917,
+      "loss": 0.9904,
       "step": 71610
     },
     {
       "epoch": 215.08,
-      "grad_norm": 7.898987293243408,
+      "grad_norm": 8.738224029541016,
       "learning_rate": 2.830830830830831e-06,
-      "loss": 0.9309,
+      "loss": 0.9289,
       "step": 71620
     },
     {
       "epoch": 215.11,
-      "grad_norm": 7.408633708953857,
+      "grad_norm": 7.683968544006348,
       "learning_rate": 2.82982982982983e-06,
-      "loss": 0.932,
+      "loss": 0.9307,
       "step": 71630
     },
     {
       "epoch": 215.14,
-      "grad_norm": 6.186914443969727,
+      "grad_norm": 7.885513782501221,
       "learning_rate": 2.828828828828829e-06,
-      "loss": 0.9193,
+      "loss": 0.9171,
       "step": 71640
     },
     {
       "epoch": 215.17,
-      "grad_norm": 4.9870171546936035,
+      "grad_norm": 3.7662012577056885,
       "learning_rate": 2.8278278278278283e-06,
-      "loss": 0.9372,
+      "loss": 0.9434,
       "step": 71650
     },
     {
       "epoch": 215.2,
-      "grad_norm": 5.690165042877197,
+      "grad_norm": 6.839529991149902,
       "learning_rate": 2.8268268268268273e-06,
-      "loss": 0.9138,
+      "loss": 0.9175,
       "step": 71660
     },
     {
       "epoch": 215.23,
-      "grad_norm": 6.741197109222412,
+      "grad_norm": 7.896917819976807,
       "learning_rate": 2.825825825825826e-06,
-      "loss": 0.9297,
+      "loss": 0.9277,
       "step": 71670
     },
     {
       "epoch": 215.26,
-      "grad_norm": 7.724676132202148,
+      "grad_norm": 7.337800979614258,
       "learning_rate": 2.824824824824825e-06,
-      "loss": 0.9664,
+      "loss": 0.9559,
       "step": 71680
     },
     {
       "epoch": 215.29,
-      "grad_norm": 6.874542713165283,
+      "grad_norm": 7.427122592926025,
       "learning_rate": 2.8238238238238243e-06,
-      "loss": 0.8681,
+      "loss": 0.8578,
       "step": 71690
     },
     {
       "epoch": 215.32,
-      "grad_norm": 7.584020137786865,
+      "grad_norm": 9.199481964111328,
       "learning_rate": 2.8228228228228234e-06,
-      "loss": 0.9614,
+      "loss": 0.9724,
       "step": 71700
     },
     {
       "epoch": 215.35,
-      "grad_norm": 10.87896728515625,
+      "grad_norm": 7.897259712219238,
       "learning_rate": 2.821821821821822e-06,
-      "loss": 0.9168,
+      "loss": 0.9181,
       "step": 71710
     },
     {
       "epoch": 215.38,
-      "grad_norm": 8.645072937011719,
+      "grad_norm": 11.318656921386719,
       "learning_rate": 2.820820820820821e-06,
-      "loss": 0.9965,
+      "loss": 0.9837,
       "step": 71720
     },
     {
       "epoch": 215.41,
-      "grad_norm": 4.78159761428833,
+      "grad_norm": 7.407173156738281,
       "learning_rate": 2.81981981981982e-06,
-      "loss": 0.9013,
+      "loss": 0.9085,
       "step": 71730
     },
     {
       "epoch": 215.44,
-      "grad_norm": 8.060081481933594,
+      "grad_norm": 8.378826141357422,
       "learning_rate": 2.8188188188188194e-06,
-      "loss": 0.8717,
+      "loss": 0.8795,
       "step": 71740
     },
     {
       "epoch": 215.47,
-      "grad_norm": 6.799717426300049,
+      "grad_norm": 6.768989562988281,
       "learning_rate": 2.817817817817818e-06,
-      "loss": 0.9146,
+      "loss": 0.9175,
       "step": 71750
     },
     {
       "epoch": 215.5,
-      "grad_norm": 7.413451671600342,
+      "grad_norm": 10.125837326049805,
       "learning_rate": 2.816816816816817e-06,
-      "loss": 0.9203,
+      "loss": 0.9253,
       "step": 71760
     },
     {
       "epoch": 215.53,
-      "grad_norm": 8.454544067382812,
+      "grad_norm": 7.671896457672119,
       "learning_rate": 2.815815815815816e-06,
-      "loss": 0.9412,
+      "loss": 0.9583,
       "step": 71770
     },
     {
       "epoch": 215.56,
-      "grad_norm": 4.262670516967773,
+      "grad_norm": 5.403573513031006,
       "learning_rate": 2.814814814814815e-06,
-      "loss": 0.8938,
+      "loss": 0.8815,
       "step": 71780
     },
     {
       "epoch": 215.59,
-      "grad_norm": 10.155774116516113,
+      "grad_norm": 6.568564414978027,
       "learning_rate": 2.8138138138138137e-06,
-      "loss": 0.962,
+      "loss": 0.9656,
       "step": 71790
     },
     {
       "epoch": 215.62,
-      "grad_norm": 5.936293125152588,
+      "grad_norm": 5.83669376373291,
       "learning_rate": 2.812812812812813e-06,
-      "loss": 0.9876,
+      "loss": 0.9941,
       "step": 71800
     },
     {
       "epoch": 215.65,
-      "grad_norm": 8.385005950927734,
+      "grad_norm": 4.776449680328369,
       "learning_rate": 2.811811811811812e-06,
-      "loss": 0.8976,
+      "loss": 0.8886,
       "step": 71810
     },
     {
       "epoch": 215.68,
-      "grad_norm": 4.433788776397705,
+      "grad_norm": 5.50703239440918,
       "learning_rate": 2.810810810810811e-06,
-      "loss": 0.9025,
+      "loss": 0.9057,
       "step": 71820
     },
     {
       "epoch": 215.71,
-      "grad_norm": 7.585935592651367,
+      "grad_norm": 7.830026626586914,
       "learning_rate": 2.8098098098098097e-06,
-      "loss": 0.9996,
+      "loss": 1.0021,
       "step": 71830
     },
     {
       "epoch": 215.74,
-      "grad_norm": 6.769144058227539,
+      "grad_norm": 6.7971601486206055,
       "learning_rate": 2.8088088088088088e-06,
-      "loss": 0.9849,
+      "loss": 0.9836,
       "step": 71840
     },
     {
       "epoch": 215.77,
-      "grad_norm": 4.552196979522705,
+      "grad_norm": 5.079774856567383,
       "learning_rate": 2.807807807807808e-06,
-      "loss": 0.8717,
+      "loss": 0.8745,
       "step": 71850
     },
     {
       "epoch": 215.8,
-      "grad_norm": 8.262500762939453,
+      "grad_norm": 10.160796165466309,
       "learning_rate": 2.8068068068068072e-06,
-      "loss": 0.869,
+      "loss": 0.8767,
       "step": 71860
     },
     {
       "epoch": 215.83,
-      "grad_norm": 7.819615364074707,
+      "grad_norm": 8.818652153015137,
       "learning_rate": 2.8058058058058062e-06,
-      "loss": 1.0242,
+      "loss": 1.0379,
       "step": 71870
     },
     {
       "epoch": 215.86,
-      "grad_norm": 6.979524612426758,
+      "grad_norm": 6.547286510467529,
       "learning_rate": 2.804804804804805e-06,
-      "loss": 0.9136,
+      "loss": 0.9175,
       "step": 71880
     },
     {
       "epoch": 215.89,
-      "grad_norm": 6.760934829711914,
+      "grad_norm": 8.231706619262695,
       "learning_rate": 2.803803803803804e-06,
-      "loss": 0.9066,
+      "loss": 0.9018,
       "step": 71890
     },
     {
       "epoch": 215.92,
-      "grad_norm": 5.486483573913574,
+      "grad_norm": 5.867310523986816,
       "learning_rate": 2.8028028028028033e-06,
-      "loss": 0.9147,
+      "loss": 0.9,
       "step": 71900
     },
     {
       "epoch": 215.95,
-      "grad_norm": 6.163059711456299,
+      "grad_norm": 7.086264133453369,
       "learning_rate": 2.8018018018018023e-06,
-      "loss": 0.9303,
+      "loss": 0.9255,
       "step": 71910
     },
     {
       "epoch": 215.98,
-      "grad_norm": 6.3927531242370605,
+      "grad_norm": 6.288363933563232,
       "learning_rate": 2.800800800800801e-06,
-      "loss": 1.0017,
+      "loss": 1.0075,
       "step": 71920
     },
     {
       "epoch": 216.0,
-      "eval_accuracy": 0.8245,
-      "eval_loss": 0.5802210569381714,
-      "eval_runtime": 5.3379,
-      "eval_samples_per_second": 1873.401,
-      "eval_steps_per_second": 7.494,
+      "eval_accuracy": 0.826,
+      "eval_loss": 0.578909158706665,
+      "eval_runtime": 5.4939,
+      "eval_samples_per_second": 1820.215,
+      "eval_steps_per_second": 7.281,
       "step": 71928
     },
     {
       "epoch": 216.01,
-      "grad_norm": 6.3035736083984375,
+      "grad_norm": 5.770232677459717,
       "learning_rate": 2.7997997997998e-06,
-      "loss": 1.1504,
+      "loss": 1.1642,
       "step": 71930
     },
     {
       "epoch": 216.04,
-      "grad_norm": 6.735114097595215,
+      "grad_norm": 9.336710929870605,
       "learning_rate": 2.798798798798799e-06,
-      "loss": 0.906,
+      "loss": 0.9082,
       "step": 71940
     },
     {
       "epoch": 216.07,
-      "grad_norm": 7.5865797996521,
+      "grad_norm": 8.075202941894531,
       "learning_rate": 2.7977977977977984e-06,
-      "loss": 0.9061,
+      "loss": 0.8993,
       "step": 71950
     },
     {
       "epoch": 216.1,
-      "grad_norm": 9.86451530456543,
+      "grad_norm": 9.572845458984375,
       "learning_rate": 2.796796796796797e-06,
-      "loss": 0.9176,
+      "loss": 0.9182,
       "step": 71960
     },
     {
       "epoch": 216.13,
-      "grad_norm": 7.518916606903076,
+      "grad_norm": 8.95561408996582,
       "learning_rate": 2.795795795795796e-06,
-      "loss": 1.0454,
+      "loss": 1.053,
       "step": 71970
     },
     {
       "epoch": 216.16,
-      "grad_norm": 8.271027565002441,
+      "grad_norm": 5.668359279632568,
       "learning_rate": 2.794794794794795e-06,
-      "loss": 0.9798,
+      "loss": 0.9728,
       "step": 71980
     },
     {
       "epoch": 216.19,
-      "grad_norm": 6.727694034576416,
+      "grad_norm": 7.966418743133545,
       "learning_rate": 2.793793793793794e-06,
-      "loss": 0.8922,
+      "loss": 0.882,
       "step": 71990
     },
     {
       "epoch": 216.22,
-      "grad_norm": 8.208390235900879,
+      "grad_norm": 15.188437461853027,
       "learning_rate": 2.7927927927927926e-06,
-      "loss": 0.8847,
+      "loss": 0.892,
       "step": 72000
     },
     {
       "epoch": 216.25,
-      "grad_norm": 6.122936725616455,
+      "grad_norm": 6.069741725921631,
       "learning_rate": 2.791791791791792e-06,
-      "loss": 0.9062,
+      "loss": 0.9018,
       "step": 72010
     },
     {
       "epoch": 216.28,
-      "grad_norm": 6.76113224029541,
+      "grad_norm": 7.851790428161621,
       "learning_rate": 2.790790790790791e-06,
-      "loss": 0.9709,
+      "loss": 0.9669,
       "step": 72020
     },
     {
       "epoch": 216.31,
-      "grad_norm": 5.83543586730957,
+      "grad_norm": 7.86965799331665,
       "learning_rate": 2.78978978978979e-06,
-      "loss": 0.9214,
+      "loss": 0.9288,
       "step": 72030
     },
     {
       "epoch": 216.34,
-      "grad_norm": 9.777892112731934,
+      "grad_norm": 7.272867202758789,
       "learning_rate": 2.7887887887887887e-06,
-      "loss": 0.8677,
+      "loss": 0.8664,
       "step": 72040
     },
     {
       "epoch": 216.37,
-      "grad_norm": 7.432103633880615,
+      "grad_norm": 7.464724063873291,
       "learning_rate": 2.7877877877877877e-06,
-      "loss": 0.9554,
+      "loss": 0.9629,
       "step": 72050
     },
     {
       "epoch": 216.4,
-      "grad_norm": 6.633342742919922,
+      "grad_norm": 5.833007335662842,
       "learning_rate": 2.786786786786787e-06,
-      "loss": 0.9381,
+      "loss": 0.9315,
       "step": 72060
     },
     {
       "epoch": 216.43,
-      "grad_norm": 6.028522491455078,
+      "grad_norm": 7.9702582359313965,
       "learning_rate": 2.785785785785786e-06,
-      "loss": 1.003,
+      "loss": 1.0007,
       "step": 72070
     },
     {
       "epoch": 216.46,
-      "grad_norm": 8.175390243530273,
+      "grad_norm": 7.5997161865234375,
       "learning_rate": 2.7847847847847848e-06,
-      "loss": 0.924,
+      "loss": 0.9217,
       "step": 72080
     },
     {
       "epoch": 216.49,
-      "grad_norm": 5.614952087402344,
+      "grad_norm": 6.810493469238281,
       "learning_rate": 2.783783783783784e-06,
-      "loss": 0.9166,
+      "loss": 0.9132,
       "step": 72090
     },
     {
       "epoch": 216.52,
-      "grad_norm": 7.307552814483643,
+      "grad_norm": 7.272316932678223,
       "learning_rate": 2.782782782782783e-06,
-      "loss": 0.9596,
+      "loss": 0.9502,
       "step": 72100
     },
     {
       "epoch": 216.55,
-      "grad_norm": 8.23143196105957,
+      "grad_norm": 9.458951950073242,
       "learning_rate": 2.7817817817817823e-06,
-      "loss": 0.9227,
+      "loss": 0.9108,
       "step": 72110
     },
     {
       "epoch": 216.58,
-      "grad_norm": 6.382566452026367,
+      "grad_norm": 5.700687885284424,
       "learning_rate": 2.780780780780781e-06,
-      "loss": 0.9035,
+      "loss": 0.9134,
       "step": 72120
     },
     {
       "epoch": 216.61,
-      "grad_norm": 6.257585048675537,
+      "grad_norm": 5.460498809814453,
       "learning_rate": 2.77977977977978e-06,
-      "loss": 0.9349,
+      "loss": 0.9408,
       "step": 72130
     },
     {
       "epoch": 216.64,
-      "grad_norm": 6.503777027130127,
+      "grad_norm": 6.828941345214844,
       "learning_rate": 2.778778778778779e-06,
-      "loss": 0.9343,
+      "loss": 0.9249,
       "step": 72140
     },
     {
       "epoch": 216.67,
-      "grad_norm": 4.63131046295166,
+      "grad_norm": 8.284524917602539,
       "learning_rate": 2.7777777777777783e-06,
-      "loss": 0.9714,
+      "loss": 0.9747,
       "step": 72150
     },
     {
       "epoch": 216.7,
-      "grad_norm": 7.052418231964111,
+      "grad_norm": 9.699789047241211,
       "learning_rate": 2.7767767767767774e-06,
-      "loss": 0.9588,
+      "loss": 0.9521,
       "step": 72160
     },
     {
       "epoch": 216.73,
-      "grad_norm": 6.856894016265869,
+      "grad_norm": 9.315030097961426,
       "learning_rate": 2.775775775775776e-06,
-      "loss": 0.9341,
+      "loss": 0.9193,
       "step": 72170
     },
     {
       "epoch": 216.76,
-      "grad_norm": 7.109679698944092,
+      "grad_norm": 5.938282012939453,
       "learning_rate": 2.774774774774775e-06,
-      "loss": 0.9159,
+      "loss": 0.9134,
       "step": 72180
     },
     {
       "epoch": 216.79,
-      "grad_norm": 7.910999298095703,
+      "grad_norm": 8.954614639282227,
       "learning_rate": 2.773773773773774e-06,
-      "loss": 0.9139,
+      "loss": 0.9263,
       "step": 72190
     },
     {
       "epoch": 216.82,
-      "grad_norm": 7.7920379638671875,
+      "grad_norm": 7.239497661590576,
       "learning_rate": 2.7727727727727734e-06,
-      "loss": 0.9822,
+      "loss": 0.9793,
       "step": 72200
     },
     {
       "epoch": 216.85,
-      "grad_norm": 7.967574119567871,
+      "grad_norm": 7.271762371063232,
       "learning_rate": 2.771771771771772e-06,
-      "loss": 1.0077,
+      "loss": 1.002,
       "step": 72210
     },
     {
       "epoch": 216.88,
-      "grad_norm": 7.080910682678223,
+      "grad_norm": 7.269804000854492,
       "learning_rate": 2.770770770770771e-06,
-      "loss": 0.916,
+      "loss": 0.9118,
       "step": 72220
     },
     {
       "epoch": 216.91,
-      "grad_norm": 4.8986124992370605,
+      "grad_norm": 9.05311393737793,
       "learning_rate": 2.76976976976977e-06,
-      "loss": 0.9,
+      "loss": 0.9029,
       "step": 72230
     },
     {
       "epoch": 216.94,
-      "grad_norm": 5.205753803253174,
+      "grad_norm": 5.632934093475342,
       "learning_rate": 2.768768768768769e-06,
-      "loss": 0.9761,
+      "loss": 0.9807,
       "step": 72240
     },
     {
       "epoch": 216.97,
-      "grad_norm": 6.262618064880371,
+      "grad_norm": 6.387950897216797,
       "learning_rate": 2.7677677677677677e-06,
-      "loss": 0.9452,
+      "loss": 0.9488,
       "step": 72250
     },
     {
       "epoch": 217.0,
-      "grad_norm": 7.140390396118164,
+      "grad_norm": 6.874744415283203,
       "learning_rate": 2.766766766766767e-06,
-      "loss": 0.9228,
+      "loss": 0.9201,
       "step": 72260
     },
     {
       "epoch": 217.0,
-      "eval_accuracy": 0.8243,
-      "eval_loss": 0.5795694589614868,
-      "eval_runtime": 5.4316,
-      "eval_samples_per_second": 1841.066,
-      "eval_steps_per_second": 7.364,
+      "eval_accuracy": 0.8259,
+      "eval_loss": 0.5779998302459717,
+      "eval_runtime": 5.4416,
+      "eval_samples_per_second": 1837.697,
+      "eval_steps_per_second": 7.351,
       "step": 72261
     },
     {
       "epoch": 217.03,
-      "grad_norm": 12.041172981262207,
+      "grad_norm": 13.919999122619629,
       "learning_rate": 2.765765765765766e-06,
-      "loss": 1.0399,
+      "loss": 1.0394,
       "step": 72270
     },
     {
       "epoch": 217.06,
-      "grad_norm": 6.357243537902832,
+      "grad_norm": 7.055396556854248,
       "learning_rate": 2.764764764764765e-06,
-      "loss": 0.8902,
+      "loss": 0.8852,
       "step": 72280
     },
     {
       "epoch": 217.09,
-      "grad_norm": 6.120850086212158,
+      "grad_norm": 6.79082727432251,
       "learning_rate": 2.7637637637637637e-06,
-      "loss": 0.9189,
+      "loss": 0.9121,
       "step": 72290
     },
     {
       "epoch": 217.12,
-      "grad_norm": 7.561592102050781,
+      "grad_norm": 7.088555335998535,
       "learning_rate": 2.7627627627627628e-06,
-      "loss": 0.9333,
+      "loss": 0.9372,
       "step": 72300
     },
     {
       "epoch": 217.15,
-      "grad_norm": 5.7195281982421875,
+      "grad_norm": 5.989465236663818,
       "learning_rate": 2.761761761761762e-06,
-      "loss": 0.9457,
+      "loss": 0.937,
       "step": 72310
     },
     {
       "epoch": 217.18,
-      "grad_norm": 6.931924819946289,
+      "grad_norm": 7.637596607208252,
       "learning_rate": 2.7607607607607612e-06,
-      "loss": 0.9689,
+      "loss": 0.9615,
       "step": 72320
     },
     {
       "epoch": 217.21,
-      "grad_norm": 5.469174385070801,
+      "grad_norm": 8.6825590133667,
       "learning_rate": 2.75975975975976e-06,
-      "loss": 1.013,
+      "loss": 1.0235,
       "step": 72330
     },
     {
       "epoch": 217.24,
-      "grad_norm": 10.930634498596191,
+      "grad_norm": 8.57563304901123,
       "learning_rate": 2.758758758758759e-06,
-      "loss": 0.9512,
+      "loss": 0.9595,
       "step": 72340
     },
     {
       "epoch": 217.27,
-      "grad_norm": 5.366798400878906,
+      "grad_norm": 5.881012916564941,
       "learning_rate": 2.757757757757758e-06,
-      "loss": 0.9753,
+      "loss": 0.9735,
       "step": 72350
     },
     {
       "epoch": 217.3,
-      "grad_norm": 7.373318672180176,
+      "grad_norm": 7.8500847816467285,
       "learning_rate": 2.7567567567567573e-06,
-      "loss": 0.9708,
+      "loss": 0.9675,
       "step": 72360
     },
     {
       "epoch": 217.33,
-      "grad_norm": 5.937170028686523,
+      "grad_norm": 7.601897716522217,
       "learning_rate": 2.755755755755756e-06,
-      "loss": 0.9009,
+      "loss": 0.8948,
       "step": 72370
     },
     {
       "epoch": 217.36,
-      "grad_norm": 5.760615825653076,
+      "grad_norm": 8.111116409301758,
       "learning_rate": 2.754754754754755e-06,
-      "loss": 0.925,
+      "loss": 0.9314,
       "step": 72380
     },
     {
       "epoch": 217.39,
-      "grad_norm": 6.727514743804932,
+      "grad_norm": 7.26259183883667,
       "learning_rate": 2.753753753753754e-06,
-      "loss": 0.8981,
+      "loss": 0.9042,
       "step": 72390
     },
     {
       "epoch": 217.42,
-      "grad_norm": 4.983912944793701,
+      "grad_norm": 5.461738109588623,
       "learning_rate": 2.752752752752753e-06,
-      "loss": 0.9056,
+      "loss": 0.9059,
       "step": 72400
     },
     {
       "epoch": 217.45,
-      "grad_norm": 6.961135387420654,
+      "grad_norm": 6.983610153198242,
       "learning_rate": 2.7517517517517524e-06,
-      "loss": 0.9865,
+      "loss": 0.9821,
       "step": 72410
     },
     {
       "epoch": 217.48,
-      "grad_norm": 6.95792818069458,
+      "grad_norm": 5.32417106628418,
       "learning_rate": 2.750750750750751e-06,
-      "loss": 0.9344,
+      "loss": 0.9286,
       "step": 72420
     },
     {
       "epoch": 217.51,
-      "grad_norm": 7.964688301086426,
+      "grad_norm": 6.897685527801514,
       "learning_rate": 2.74974974974975e-06,
-      "loss": 0.9534,
+      "loss": 0.9496,
       "step": 72430
     },
     {
       "epoch": 217.54,
-      "grad_norm": 7.777303695678711,
+      "grad_norm": 8.244394302368164,
       "learning_rate": 2.748748748748749e-06,
-      "loss": 0.8632,
+      "loss": 0.8572,
       "step": 72440
     },
     {
       "epoch": 217.57,
-      "grad_norm": 6.262092590332031,
+      "grad_norm": 6.767581939697266,
       "learning_rate": 2.747747747747748e-06,
-      "loss": 0.9424,
+      "loss": 0.9387,
       "step": 72450
     },
     {
       "epoch": 217.6,
-      "grad_norm": 8.705977439880371,
+      "grad_norm": 10.288528442382812,
       "learning_rate": 2.7467467467467466e-06,
-      "loss": 0.9279,
+      "loss": 0.9256,
       "step": 72460
     },
     {
       "epoch": 217.63,
-      "grad_norm": 5.958741664886475,
+      "grad_norm": 9.592744827270508,
       "learning_rate": 2.745745745745746e-06,
-      "loss": 0.9449,
+      "loss": 0.9356,
       "step": 72470
     },
     {
       "epoch": 217.66,
-      "grad_norm": 8.844047546386719,
+      "grad_norm": 6.3012847900390625,
       "learning_rate": 2.744744744744745e-06,
-      "loss": 0.8889,
+      "loss": 0.8888,
       "step": 72480
     },
     {
       "epoch": 217.69,
-      "grad_norm": 5.768106460571289,
+      "grad_norm": 6.30490779876709,
       "learning_rate": 2.743743743743744e-06,
-      "loss": 0.9049,
+      "loss": 0.9063,
       "step": 72490
     },
     {
       "epoch": 217.72,
-      "grad_norm": 9.38657283782959,
+      "grad_norm": 7.621927261352539,
       "learning_rate": 2.7427427427427427e-06,
-      "loss": 0.9425,
+      "loss": 0.9349,
       "step": 72500
     },
     {
       "epoch": 217.75,
-      "grad_norm": 5.362709045410156,
+      "grad_norm": 7.683348655700684,
       "learning_rate": 2.7417417417417417e-06,
-      "loss": 0.9254,
+      "loss": 0.9183,
       "step": 72510
     },
     {
       "epoch": 217.78,
-      "grad_norm": 4.530629634857178,
+      "grad_norm": 6.049909591674805,
       "learning_rate": 2.740740740740741e-06,
-      "loss": 0.9568,
+      "loss": 0.9585,
       "step": 72520
     },
     {
       "epoch": 217.81,
-      "grad_norm": 6.522764205932617,
+      "grad_norm": 6.608031749725342,
       "learning_rate": 2.73973973973974e-06,
-      "loss": 0.925,
+      "loss": 0.9232,
       "step": 72530
     },
     {
       "epoch": 217.84,
-      "grad_norm": 7.488985538482666,
+      "grad_norm": 8.487922668457031,
       "learning_rate": 2.7387387387387388e-06,
-      "loss": 0.9553,
+      "loss": 0.9664,
       "step": 72540
     },
     {
       "epoch": 217.87,
-      "grad_norm": 5.385812282562256,
+      "grad_norm": 5.805594444274902,
       "learning_rate": 2.737737737737738e-06,
-      "loss": 0.9289,
+      "loss": 0.9225,
       "step": 72550
     },
     {
       "epoch": 217.9,
-      "grad_norm": 7.321783065795898,
+      "grad_norm": 8.490604400634766,
       "learning_rate": 2.736736736736737e-06,
-      "loss": 0.9326,
+      "loss": 0.9401,
       "step": 72560
     },
     {
       "epoch": 217.93,
-      "grad_norm": 5.685489177703857,
+      "grad_norm": 6.131755828857422,
       "learning_rate": 2.7357357357357363e-06,
-      "loss": 0.9379,
+      "loss": 0.9355,
       "step": 72570
     },
     {
       "epoch": 217.96,
-      "grad_norm": 5.9327802658081055,
+      "grad_norm": 6.23167085647583,
       "learning_rate": 2.734734734734735e-06,
-      "loss": 0.9612,
+      "loss": 0.9582,
       "step": 72580
     },
     {
       "epoch": 217.99,
-      "grad_norm": 10.532814025878906,
+      "grad_norm": 10.67435073852539,
       "learning_rate": 2.733733733733734e-06,
-      "loss": 0.9644,
+      "loss": 0.9624,
       "step": 72590
     },
     {
       "epoch": 218.0,
-      "eval_accuracy": 0.8219,
-      "eval_loss": 0.5859174728393555,
-      "eval_runtime": 5.6389,
-      "eval_samples_per_second": 1773.393,
-      "eval_steps_per_second": 7.094,
+      "eval_accuracy": 0.8236,
+      "eval_loss": 0.5827905535697937,
+      "eval_runtime": 5.2601,
+      "eval_samples_per_second": 1901.116,
+      "eval_steps_per_second": 7.604,
       "step": 72594
     },
     {
       "epoch": 218.02,
-      "grad_norm": 9.306097984313965,
+      "grad_norm": 7.097799301147461,
       "learning_rate": 2.732732732732733e-06,
-      "loss": 1.236,
+      "loss": 1.2379,
       "step": 72600
     },
     {
       "epoch": 218.05,
-      "grad_norm": 9.399117469787598,
+      "grad_norm": 7.958342552185059,
       "learning_rate": 2.731731731731732e-06,
-      "loss": 0.8765,
+      "loss": 0.8768,
       "step": 72610
     },
     {
       "epoch": 218.08,
-      "grad_norm": 5.533222675323486,
+      "grad_norm": 4.53885555267334,
       "learning_rate": 2.7307307307307305e-06,
-      "loss": 0.8995,
+      "loss": 0.8933,
       "step": 72620
     },
     {
       "epoch": 218.11,
-      "grad_norm": 7.908480644226074,
+      "grad_norm": 7.708099365234375,
       "learning_rate": 2.72972972972973e-06,
-      "loss": 0.9702,
+      "loss": 0.9752,
       "step": 72630
     },
     {
       "epoch": 218.14,
-      "grad_norm": 5.921170711517334,
+      "grad_norm": 6.683178424835205,
       "learning_rate": 2.728728728728729e-06,
-      "loss": 0.9031,
+      "loss": 0.9057,
       "step": 72640
     },
     {
       "epoch": 218.17,
-      "grad_norm": 5.913274765014648,
+      "grad_norm": 6.364109992980957,
       "learning_rate": 2.727727727727728e-06,
-      "loss": 0.945,
+      "loss": 0.9473,
       "step": 72650
     },
     {
       "epoch": 218.2,
-      "grad_norm": 7.699877738952637,
+      "grad_norm": 6.330079078674316,
       "learning_rate": 2.7267267267267274e-06,
-      "loss": 0.8865,
+      "loss": 0.8939,
       "step": 72660
     },
     {
       "epoch": 218.23,
-      "grad_norm": 5.8549933433532715,
+      "grad_norm": 7.843347072601318,
       "learning_rate": 2.7257257257257256e-06,
-      "loss": 0.9596,
+      "loss": 0.9578,
       "step": 72670
     },
     {
       "epoch": 218.26,
-      "grad_norm": 5.486217975616455,
+      "grad_norm": 5.755901336669922,
       "learning_rate": 2.724724724724725e-06,
-      "loss": 0.8721,
+      "loss": 0.8809,
       "step": 72680
     },
     {
       "epoch": 218.29,
-      "grad_norm": 6.3092360496521,
+      "grad_norm": 7.000547885894775,
       "learning_rate": 2.723723723723724e-06,
-      "loss": 0.8992,
+      "loss": 0.8927,
       "step": 72690
     },
     {
       "epoch": 218.32,
-      "grad_norm": 5.997005462646484,
+      "grad_norm": 9.1029634475708,
       "learning_rate": 2.722722722722723e-06,
-      "loss": 0.9111,
+      "loss": 0.9094,
       "step": 72700
     },
     {
       "epoch": 218.35,
-      "grad_norm": 6.3835768699646,
+      "grad_norm": 6.088730335235596,
       "learning_rate": 2.7217217217217217e-06,
-      "loss": 0.9329,
+      "loss": 0.937,
       "step": 72710
     },
     {
       "epoch": 218.38,
-      "grad_norm": 9.882909774780273,
+      "grad_norm": 6.871915817260742,
       "learning_rate": 2.720720720720721e-06,
-      "loss": 0.9578,
+      "loss": 0.9597,
       "step": 72720
     },
     {
       "epoch": 218.41,
-      "grad_norm": 6.438920021057129,
+      "grad_norm": 7.44312858581543,
       "learning_rate": 2.71971971971972e-06,
-      "loss": 0.9636,
+      "loss": 0.9606,
       "step": 72730
     },
     {
       "epoch": 218.44,
-      "grad_norm": 10.05977725982666,
+      "grad_norm": 10.182854652404785,
       "learning_rate": 2.718718718718719e-06,
-      "loss": 0.916,
+      "loss": 0.9087,
       "step": 72740
     },
     {
       "epoch": 218.47,
-      "grad_norm": 8.613099098205566,
+      "grad_norm": 7.92831563949585,
       "learning_rate": 2.7177177177177177e-06,
-      "loss": 0.9381,
+      "loss": 0.9417,
       "step": 72750
     },
     {
       "epoch": 218.5,
-      "grad_norm": 5.401729583740234,
+      "grad_norm": 6.851491451263428,
       "learning_rate": 2.7167167167167168e-06,
-      "loss": 0.8773,
+      "loss": 0.8754,
       "step": 72760
     },
     {
       "epoch": 218.53,
-      "grad_norm": 8.367191314697266,
+      "grad_norm": 9.066606521606445,
       "learning_rate": 2.715715715715716e-06,
-      "loss": 0.8891,
+      "loss": 0.8779,
       "step": 72770
     },
     {
       "epoch": 218.56,
-      "grad_norm": 6.743752479553223,
+      "grad_norm": 5.933196544647217,
       "learning_rate": 2.7147147147147152e-06,
-      "loss": 0.8985,
+      "loss": 0.893,
       "step": 72780
     },
     {
       "epoch": 218.59,
-      "grad_norm": 5.6828484535217285,
+      "grad_norm": 5.562318801879883,
       "learning_rate": 2.713713713713714e-06,
-      "loss": 0.9788,
+      "loss": 0.9744,
       "step": 72790
     },
     {
       "epoch": 218.62,
-      "grad_norm": 5.55059814453125,
+      "grad_norm": 5.71994686126709,
       "learning_rate": 2.712712712712713e-06,
-      "loss": 0.897,
+      "loss": 0.895,
       "step": 72800
     },
     {
       "epoch": 218.65,
-      "grad_norm": 9.172086715698242,
+      "grad_norm": 12.900361061096191,
       "learning_rate": 2.711711711711712e-06,
-      "loss": 0.8746,
+      "loss": 0.869,
       "step": 72810
     },
     {
       "epoch": 218.68,
-      "grad_norm": 7.2971954345703125,
+      "grad_norm": 8.585466384887695,
       "learning_rate": 2.7107107107107113e-06,
-      "loss": 0.946,
+      "loss": 0.949,
       "step": 72820
     },
     {
       "epoch": 218.71,
-      "grad_norm": 5.520009517669678,
+      "grad_norm": 7.784217834472656,
       "learning_rate": 2.70970970970971e-06,
-      "loss": 0.9955,
+      "loss": 0.9976,
       "step": 72830
     },
     {
       "epoch": 218.74,
-      "grad_norm": 6.278941631317139,
+      "grad_norm": 7.183433532714844,
       "learning_rate": 2.708708708708709e-06,
       "loss": 0.9361,
       "step": 72840
     },
     {
       "epoch": 218.77,
-      "grad_norm": 6.358395576477051,
+      "grad_norm": 6.958471775054932,
       "learning_rate": 2.707707707707708e-06,
-      "loss": 0.8909,
+      "loss": 0.8782,
       "step": 72850
     },
     {
       "epoch": 218.8,
-      "grad_norm": 5.277022838592529,
+      "grad_norm": 5.698826313018799,
       "learning_rate": 2.706706706706707e-06,
-      "loss": 0.8703,
+      "loss": 0.8657,
       "step": 72860
     },
     {
       "epoch": 218.83,
-      "grad_norm": 5.604706287384033,
+      "grad_norm": 7.760838031768799,
       "learning_rate": 2.7057057057057055e-06,
-      "loss": 0.926,
+      "loss": 0.9244,
       "step": 72870
     },
     {
       "epoch": 218.86,
-      "grad_norm": 6.780740737915039,
+      "grad_norm": 6.333276748657227,
       "learning_rate": 2.704704704704705e-06,
-      "loss": 0.9624,
+      "loss": 0.9693,
       "step": 72880
     },
     {
       "epoch": 218.89,
-      "grad_norm": 5.856489658355713,
+      "grad_norm": 5.705721855163574,
       "learning_rate": 2.703703703703704e-06,
-      "loss": 0.927,
+      "loss": 0.9213,
       "step": 72890
     },
     {
       "epoch": 218.92,
-      "grad_norm": 7.708815574645996,
+      "grad_norm": 8.012775421142578,
       "learning_rate": 2.702702702702703e-06,
-      "loss": 0.993,
+      "loss": 0.9941,
       "step": 72900
     },
     {
       "epoch": 218.95,
-      "grad_norm": 7.581908702850342,
+      "grad_norm": 7.459386825561523,
       "learning_rate": 2.7017017017017016e-06,
-      "loss": 0.9129,
+      "loss": 0.9106,
       "step": 72910
     },
     {
       "epoch": 218.98,
-      "grad_norm": 8.937755584716797,
+      "grad_norm": 9.367036819458008,
       "learning_rate": 2.7007007007007006e-06,
-      "loss": 0.8991,
+      "loss": 0.8968,
       "step": 72920
     },
     {
       "epoch": 219.0,
-      "eval_accuracy": 0.8235,
-      "eval_loss": 0.5794757008552551,
-      "eval_runtime": 5.4233,
-      "eval_samples_per_second": 1843.905,
-      "eval_steps_per_second": 7.376,
+      "eval_accuracy": 0.8276,
+      "eval_loss": 0.577774703502655,
+      "eval_runtime": 5.6281,
+      "eval_samples_per_second": 1776.796,
+      "eval_steps_per_second": 7.107,
       "step": 72927
     },
     {
       "epoch": 219.01,
-      "grad_norm": 6.018161296844482,
+      "grad_norm": 6.799353122711182,
       "learning_rate": 2.6996996996997e-06,
-      "loss": 0.9826,
+      "loss": 0.9776,
       "step": 72930
     },
     {
       "epoch": 219.04,
-      "grad_norm": 6.745750904083252,
+      "grad_norm": 7.892888069152832,
       "learning_rate": 2.698698698698699e-06,
-      "loss": 0.8766,
+      "loss": 0.8805,
       "step": 72940
     },
     {
       "epoch": 219.07,
-      "grad_norm": 9.698201179504395,
+      "grad_norm": 7.911108016967773,
       "learning_rate": 2.697697697697698e-06,
       "loss": 0.979,
       "step": 72950
     },
     {
       "epoch": 219.1,
-      "grad_norm": 7.5127787590026855,
+      "grad_norm": 6.964560508728027,
       "learning_rate": 2.6966966966966967e-06,
-      "loss": 0.9302,
+      "loss": 0.9327,
       "step": 72960
     },
     {
       "epoch": 219.13,
-      "grad_norm": 7.344473361968994,
+      "grad_norm": 7.591625213623047,
       "learning_rate": 2.6956956956956957e-06,
-      "loss": 0.9203,
+      "loss": 0.9245,
       "step": 72970
     },
     {
       "epoch": 219.16,
-      "grad_norm": 9.137619018554688,
+      "grad_norm": 8.3609037399292,
       "learning_rate": 2.694694694694695e-06,
-      "loss": 1.0137,
+      "loss": 1.0184,
       "step": 72980
     },
     {
       "epoch": 219.19,
-      "grad_norm": 9.447741508483887,
+      "grad_norm": 10.235336303710938,
       "learning_rate": 2.693693693693694e-06,
-      "loss": 0.9522,
+      "loss": 0.9592,
       "step": 72990
     },
     {
       "epoch": 219.22,
-      "grad_norm": 5.883161544799805,
+      "grad_norm": 6.992069244384766,
       "learning_rate": 2.6926926926926928e-06,
-      "loss": 0.9183,
+      "loss": 0.9082,
       "step": 73000
     },
     {
       "epoch": 219.25,
-      "grad_norm": 4.839803695678711,
+      "grad_norm": 4.844293594360352,
       "learning_rate": 2.691691691691692e-06,
-      "loss": 0.8611,
+      "loss": 0.8601,
       "step": 73010
     },
     {
       "epoch": 219.28,
-      "grad_norm": 6.2444915771484375,
+      "grad_norm": 6.1902031898498535,
       "learning_rate": 2.690690690690691e-06,
-      "loss": 0.9652,
+      "loss": 0.9673,
       "step": 73020
     },
     {
       "epoch": 219.31,
-      "grad_norm": 5.773776054382324,
+      "grad_norm": 6.05573034286499,
       "learning_rate": 2.6896896896896903e-06,
-      "loss": 0.9175,
+      "loss": 0.9174,
       "step": 73030
     },
     {
       "epoch": 219.34,
-      "grad_norm": 5.484992980957031,
+      "grad_norm": 5.244838714599609,
       "learning_rate": 2.688688688688689e-06,
-      "loss": 0.9473,
+      "loss": 0.9327,
       "step": 73040
     },
     {
       "epoch": 219.37,
-      "grad_norm": 6.773443698883057,
+      "grad_norm": 6.842052459716797,
       "learning_rate": 2.687687687687688e-06,
-      "loss": 0.9437,
+      "loss": 0.9417,
       "step": 73050
     },
     {
       "epoch": 219.4,
-      "grad_norm": 7.300702095031738,
+      "grad_norm": 6.528975486755371,
       "learning_rate": 2.686686686686687e-06,
-      "loss": 0.9392,
+      "loss": 0.9402,
       "step": 73060
     },
     {
       "epoch": 219.43,
-      "grad_norm": 7.428591728210449,
+      "grad_norm": 6.301028728485107,
       "learning_rate": 2.685685685685686e-06,
-      "loss": 0.9312,
+      "loss": 0.9294,
       "step": 73070
     },
     {
       "epoch": 219.46,
-      "grad_norm": 7.491966724395752,
+      "grad_norm": 10.67099380493164,
       "learning_rate": 2.6846846846846845e-06,
-      "loss": 0.9717,
+      "loss": 0.9611,
       "step": 73080
     },
     {
       "epoch": 219.49,
-      "grad_norm": 8.162881851196289,
+      "grad_norm": 8.409181594848633,
       "learning_rate": 2.683683683683684e-06,
-      "loss": 0.8986,
+      "loss": 0.9074,
       "step": 73090
     },
     {
       "epoch": 219.52,
-      "grad_norm": 8.521729469299316,
+      "grad_norm": 6.286962032318115,
       "learning_rate": 2.682682682682683e-06,
-      "loss": 0.9522,
+      "loss": 0.9532,
       "step": 73100
     },
     {
       "epoch": 219.55,
-      "grad_norm": 6.315744876861572,
+      "grad_norm": 6.169939041137695,
       "learning_rate": 2.681681681681682e-06,
-      "loss": 0.8868,
+      "loss": 0.8924,
       "step": 73110
     },
     {
       "epoch": 219.58,
-      "grad_norm": 6.352571487426758,
+      "grad_norm": 6.383563041687012,
       "learning_rate": 2.6806806806806806e-06,
-      "loss": 0.9252,
+      "loss": 0.9176,
       "step": 73120
     },
     {
       "epoch": 219.61,
-      "grad_norm": 4.693980693817139,
+      "grad_norm": 6.170619964599609,
       "learning_rate": 2.6796796796796796e-06,
-      "loss": 0.9769,
+      "loss": 0.971,
       "step": 73130
     },
     {
       "epoch": 219.64,
-      "grad_norm": 7.85822057723999,
+      "grad_norm": 7.945230007171631,
       "learning_rate": 2.678678678678679e-06,
-      "loss": 0.8854,
+      "loss": 0.8795,
       "step": 73140
     },
     {
       "epoch": 219.67,
-      "grad_norm": 6.742613792419434,
+      "grad_norm": 6.799906253814697,
       "learning_rate": 2.677677677677678e-06,
-      "loss": 0.9636,
+      "loss": 0.9529,
       "step": 73150
     },
     {
       "epoch": 219.7,
-      "grad_norm": 5.795612812042236,
+      "grad_norm": 5.66868257522583,
       "learning_rate": 2.6766766766766766e-06,
-      "loss": 0.9524,
+      "loss": 0.9476,
       "step": 73160
     },
     {
       "epoch": 219.73,
-      "grad_norm": 4.989810943603516,
+      "grad_norm": 6.038729190826416,
       "learning_rate": 2.6756756756756757e-06,
-      "loss": 0.9053,
+      "loss": 0.9177,
       "step": 73170
     },
     {
       "epoch": 219.76,
-      "grad_norm": 7.986024856567383,
+      "grad_norm": 10.488251686096191,
       "learning_rate": 2.674674674674675e-06,
-      "loss": 0.9687,
+      "loss": 0.9799,
       "step": 73180
     },
     {
       "epoch": 219.79,
-      "grad_norm": 8.241772651672363,
+      "grad_norm": 11.756155967712402,
       "learning_rate": 2.673673673673674e-06,
-      "loss": 0.9515,
+      "loss": 0.9564,
       "step": 73190
     },
     {
       "epoch": 219.82,
-      "grad_norm": 6.7512526512146,
+      "grad_norm": 7.065455913543701,
       "learning_rate": 2.672672672672673e-06,
-      "loss": 0.929,
+      "loss": 0.9375,
       "step": 73200
     },
     {
       "epoch": 219.85,
-      "grad_norm": 5.826577663421631,
+      "grad_norm": 5.751431941986084,
       "learning_rate": 2.6716716716716717e-06,
-      "loss": 0.9012,
+      "loss": 0.9002,
       "step": 73210
     },
     {
       "epoch": 219.88,
-      "grad_norm": 11.410309791564941,
+      "grad_norm": 10.696011543273926,
       "learning_rate": 2.6706706706706708e-06,
-      "loss": 1.0161,
+      "loss": 1.0206,
       "step": 73220
     },
     {
       "epoch": 219.91,
-      "grad_norm": 5.597487449645996,
+      "grad_norm": 6.757604122161865,
       "learning_rate": 2.66966966966967e-06,
-      "loss": 0.8205,
+      "loss": 0.8236,
       "step": 73230
     },
     {
       "epoch": 219.94,
-      "grad_norm": 7.889488220214844,
+      "grad_norm": 8.578100204467773,
       "learning_rate": 2.6686686686686692e-06,
-      "loss": 0.9755,
+      "loss": 0.9762,
       "step": 73240
     },
     {
       "epoch": 219.97,
-      "grad_norm": 5.090219497680664,
+      "grad_norm": 6.6927266120910645,
       "learning_rate": 2.667667667667668e-06,
-      "loss": 0.9812,
+      "loss": 0.9937,
       "step": 73250
     },
     {
       "epoch": 220.0,
-      "grad_norm": 37.34755325317383,
+      "grad_norm": 38.61555099487305,
       "learning_rate": 2.666666666666667e-06,
-      "loss": 1.1149,
+      "loss": 1.1072,
       "step": 73260
     },
     {
       "epoch": 220.0,
-      "eval_accuracy": 0.8253,
-      "eval_loss": 0.5778186321258545,
-      "eval_runtime": 5.3564,
-      "eval_samples_per_second": 1866.923,
-      "eval_steps_per_second": 7.468,
+      "eval_accuracy": 0.829,
+      "eval_loss": 0.57671058177948,
+      "eval_runtime": 5.4324,
+      "eval_samples_per_second": 1840.8,
+      "eval_steps_per_second": 7.363,
       "step": 73260
     },
     {
       "epoch": 220.03,
-      "grad_norm": 5.351058006286621,
+      "grad_norm": 5.673396587371826,
       "learning_rate": 2.665665665665666e-06,
-      "loss": 0.9177,
+      "loss": 0.9065,
       "step": 73270
     },
     {
       "epoch": 220.06,
-      "grad_norm": 7.195152282714844,
+      "grad_norm": 7.475549221038818,
       "learning_rate": 2.6646646646646653e-06,
-      "loss": 0.8396,
+      "loss": 0.821,
       "step": 73280
     },
     {
       "epoch": 220.09,
-      "grad_norm": 6.274990558624268,
+      "grad_norm": 6.855286121368408,
       "learning_rate": 2.663663663663664e-06,
-      "loss": 0.9349,
+      "loss": 0.9264,
       "step": 73290
     },
     {
       "epoch": 220.12,
-      "grad_norm": 5.832813739776611,
+      "grad_norm": 6.398316383361816,
       "learning_rate": 2.662662662662663e-06,
-      "loss": 0.8794,
+      "loss": 0.8715,
       "step": 73300
     },
     {
       "epoch": 220.15,
-      "grad_norm": 5.781298637390137,
+      "grad_norm": 8.712723731994629,
       "learning_rate": 2.661661661661662e-06,
-      "loss": 0.9022,
+      "loss": 0.9008,
       "step": 73310
     },
     {
       "epoch": 220.18,
-      "grad_norm": 5.781087875366211,
+      "grad_norm": 5.670687675476074,
       "learning_rate": 2.660660660660661e-06,
-      "loss": 0.9914,
+      "loss": 0.9878,
       "step": 73320
     },
     {
       "epoch": 220.21,
-      "grad_norm": 7.396402359008789,
+      "grad_norm": 6.515443325042725,
       "learning_rate": 2.6596596596596595e-06,
-      "loss": 0.9423,
+      "loss": 0.9382,
       "step": 73330
     },
     {
       "epoch": 220.24,
-      "grad_norm": 5.230351448059082,
+      "grad_norm": 6.8233537673950195,
       "learning_rate": 2.658658658658659e-06,
-      "loss": 0.9204,
+      "loss": 0.9255,
       "step": 73340
     },
     {
       "epoch": 220.27,
-      "grad_norm": 6.178385257720947,
+      "grad_norm": 6.105470180511475,
       "learning_rate": 2.657657657657658e-06,
-      "loss": 0.9376,
+      "loss": 0.9315,
       "step": 73350
     },
     {
       "epoch": 220.3,
-      "grad_norm": 6.422088623046875,
+      "grad_norm": 7.173359394073486,
       "learning_rate": 2.656656656656657e-06,
-      "loss": 0.8736,
+      "loss": 0.8756,
       "step": 73360
     },
     {
       "epoch": 220.33,
-      "grad_norm": 6.458597660064697,
+      "grad_norm": 5.832638740539551,
       "learning_rate": 2.6556556556556556e-06,
-      "loss": 0.9472,
+      "loss": 0.9526,
       "step": 73370
     },
     {
       "epoch": 220.36,
-      "grad_norm": 6.792240142822266,
+      "grad_norm": 9.97787094116211,
       "learning_rate": 2.6546546546546546e-06,
-      "loss": 0.9027,
+      "loss": 0.8903,
       "step": 73380
     },
     {
       "epoch": 220.39,
-      "grad_norm": 5.539630889892578,
+      "grad_norm": 7.70621395111084,
       "learning_rate": 2.653653653653654e-06,
-      "loss": 0.8883,
+      "loss": 0.8854,
       "step": 73390
     },
     {
       "epoch": 220.42,
-      "grad_norm": 6.841437339782715,
+      "grad_norm": 6.39276647567749,
       "learning_rate": 2.652652652652653e-06,
-      "loss": 0.9729,
+      "loss": 0.9696,
       "step": 73400
     },
     {
       "epoch": 220.45,
-      "grad_norm": 8.384631156921387,
+      "grad_norm": 8.049924850463867,
       "learning_rate": 2.6516516516516517e-06,
-      "loss": 0.9585,
+      "loss": 0.9694,
       "step": 73410
     },
     {
       "epoch": 220.48,
-      "grad_norm": 7.719588756561279,
+      "grad_norm": 6.877997875213623,
       "learning_rate": 2.6506506506506507e-06,
-      "loss": 0.9036,
+      "loss": 0.8973,
       "step": 73420
     },
     {
       "epoch": 220.51,
-      "grad_norm": 7.742959976196289,
+      "grad_norm": 9.818087577819824,
       "learning_rate": 2.6496496496496497e-06,
-      "loss": 0.9516,
+      "loss": 0.9535,
       "step": 73430
     },
     {
       "epoch": 220.54,
-      "grad_norm": 12.321094512939453,
+      "grad_norm": 6.212538242340088,
       "learning_rate": 2.648648648648649e-06,
-      "loss": 0.8859,
+      "loss": 0.8884,
       "step": 73440
     },
     {
       "epoch": 220.57,
-      "grad_norm": 6.094039440155029,
+      "grad_norm": 8.376958847045898,
       "learning_rate": 2.6476476476476478e-06,
-      "loss": 0.9644,
+      "loss": 0.958,
       "step": 73450
     },
     {
       "epoch": 220.6,
-      "grad_norm": 5.046603679656982,
+      "grad_norm": 5.057199001312256,
       "learning_rate": 2.6466466466466468e-06,
-      "loss": 0.8937,
+      "loss": 0.8945,
       "step": 73460
     },
     {
       "epoch": 220.63,
-      "grad_norm": 8.348424911499023,
+      "grad_norm": 8.375308990478516,
       "learning_rate": 2.645645645645646e-06,
-      "loss": 0.9624,
+      "loss": 0.9647,
       "step": 73470
     },
     {
       "epoch": 220.66,
-      "grad_norm": 8.484251022338867,
+      "grad_norm": 11.146285057067871,
       "learning_rate": 2.644644644644645e-06,
-      "loss": 0.8817,
+      "loss": 0.8763,
       "step": 73480
     },
     {
       "epoch": 220.69,
-      "grad_norm": 7.793914794921875,
+      "grad_norm": 6.138522148132324,
       "learning_rate": 2.6436436436436443e-06,
-      "loss": 0.9387,
+      "loss": 0.9361,
       "step": 73490
     },
     {
       "epoch": 220.72,
-      "grad_norm": 6.482189655303955,
+      "grad_norm": 7.711483955383301,
       "learning_rate": 2.642642642642643e-06,
-      "loss": 0.9365,
+      "loss": 0.9354,
       "step": 73500
     },
     {
       "epoch": 220.75,
-      "grad_norm": 5.552371978759766,
+      "grad_norm": 6.0607171058654785,
       "learning_rate": 2.641641641641642e-06,
-      "loss": 0.9739,
+      "loss": 0.9756,
       "step": 73510
     },
     {
       "epoch": 220.78,
-      "grad_norm": 4.960691452026367,
+      "grad_norm": 7.09754753112793,
       "learning_rate": 2.640640640640641e-06,
-      "loss": 0.9744,
+      "loss": 0.9788,
       "step": 73520
     },
     {
       "epoch": 220.81,
-      "grad_norm": 7.206474304199219,
+      "grad_norm": 9.048027038574219,
       "learning_rate": 2.63963963963964e-06,
-      "loss": 0.9458,
+      "loss": 0.9475,
       "step": 73530
     },
     {
       "epoch": 220.84,
-      "grad_norm": 6.860537528991699,
+      "grad_norm": 6.837305545806885,
       "learning_rate": 2.6386386386386385e-06,
-      "loss": 1.0026,
+      "loss": 0.9913,
       "step": 73540
     },
     {
       "epoch": 220.87,
-      "grad_norm": 7.079867839813232,
+      "grad_norm": 7.541650295257568,
       "learning_rate": 2.637637637637638e-06,
-      "loss": 0.9286,
+      "loss": 0.9358,
       "step": 73550
     },
     {
       "epoch": 220.9,
-      "grad_norm": 6.690513610839844,
+      "grad_norm": 7.188553333282471,
       "learning_rate": 2.636636636636637e-06,
-      "loss": 0.9478,
+      "loss": 0.9515,
       "step": 73560
     },
     {
       "epoch": 220.93,
-      "grad_norm": 8.442537307739258,
+      "grad_norm": 9.016393661499023,
       "learning_rate": 2.635635635635636e-06,
-      "loss": 0.9124,
+      "loss": 0.9098,
       "step": 73570
     },
     {
       "epoch": 220.96,
-      "grad_norm": 5.903593063354492,
+      "grad_norm": 6.521063804626465,
       "learning_rate": 2.6346346346346346e-06,
-      "loss": 0.9685,
+      "loss": 0.9509,
       "step": 73580
     },
     {
       "epoch": 220.99,
-      "grad_norm": 6.76936674118042,
+      "grad_norm": 12.757721900939941,
       "learning_rate": 2.6336336336336336e-06,
-      "loss": 0.9295,
+      "loss": 0.923,
       "step": 73590
     },
     {
       "epoch": 221.0,
-      "eval_accuracy": 0.8251,
-      "eval_loss": 0.5784720778465271,
-      "eval_runtime": 5.4244,
-      "eval_samples_per_second": 1843.521,
-      "eval_steps_per_second": 7.374,
+      "eval_accuracy": 0.827,
+      "eval_loss": 0.5777686238288879,
+      "eval_runtime": 5.2447,
+      "eval_samples_per_second": 1906.68,
+      "eval_steps_per_second": 7.627,
       "step": 73593
     },
     {
       "epoch": 221.02,
-      "grad_norm": 7.177062034606934,
+      "grad_norm": 9.986597061157227,
       "learning_rate": 2.632632632632633e-06,
-      "loss": 1.0413,
+      "loss": 1.0432,
       "step": 73600
     },
     {
       "epoch": 221.05,
-      "grad_norm": 7.1002984046936035,
+      "grad_norm": 7.157760143280029,
       "learning_rate": 2.631631631631632e-06,
-      "loss": 0.9471,
+      "loss": 0.9564,
       "step": 73610
     },
     {
       "epoch": 221.08,
-      "grad_norm": 6.397023677825928,
+      "grad_norm": 6.388676166534424,
       "learning_rate": 2.6306306306306306e-06,
-      "loss": 0.9575,
+      "loss": 0.9521,
       "step": 73620
     },
     {
       "epoch": 221.11,
-      "grad_norm": 6.601405620574951,
+      "grad_norm": 7.7597270011901855,
       "learning_rate": 2.6296296296296297e-06,
-      "loss": 0.938,
+      "loss": 0.9458,
       "step": 73630
     },
     {
       "epoch": 221.14,
-      "grad_norm": 9.297335624694824,
+      "grad_norm": 9.724603652954102,
       "learning_rate": 2.6286286286286287e-06,
-      "loss": 0.8921,
+      "loss": 0.8898,
       "step": 73640
     },
     {
       "epoch": 221.17,
-      "grad_norm": 5.492344856262207,
+      "grad_norm": 8.29684066772461,
       "learning_rate": 2.627627627627628e-06,
-      "loss": 0.9319,
+      "loss": 0.9321,
       "step": 73650
     },
     {
       "epoch": 221.2,
-      "grad_norm": 6.488237380981445,
+      "grad_norm": 5.970361709594727,
       "learning_rate": 2.6266266266266267e-06,
-      "loss": 0.9,
+      "loss": 0.8993,
       "step": 73660
     },
     {
       "epoch": 221.23,
-      "grad_norm": 8.145695686340332,
+      "grad_norm": 8.008609771728516,
       "learning_rate": 2.6256256256256257e-06,
-      "loss": 0.8945,
+      "loss": 0.8864,
       "step": 73670
     },
     {
       "epoch": 221.26,
-      "grad_norm": 6.548057556152344,
+      "grad_norm": 5.625511169433594,
       "learning_rate": 2.6246246246246248e-06,
-      "loss": 0.8549,
+      "loss": 0.8617,
       "step": 73680
     },
     {
       "epoch": 221.29,
-      "grad_norm": 9.02611255645752,
+      "grad_norm": 10.187233924865723,
       "learning_rate": 2.623623623623624e-06,
-      "loss": 0.9144,
+      "loss": 0.9091,
       "step": 73690
     },
     {
       "epoch": 221.32,
-      "grad_norm": 11.721210479736328,
+      "grad_norm": 10.51606559753418,
       "learning_rate": 2.6226226226226224e-06,
-      "loss": 0.9638,
+      "loss": 0.9519,
       "step": 73700
     },
     {
       "epoch": 221.35,
-      "grad_norm": 6.370357990264893,
+      "grad_norm": 6.200037479400635,
       "learning_rate": 2.621621621621622e-06,
-      "loss": 0.8973,
+      "loss": 0.9046,
       "step": 73710
     },
     {
       "epoch": 221.38,
-      "grad_norm": 5.62892484664917,
+      "grad_norm": 6.057072162628174,
       "learning_rate": 2.620620620620621e-06,
-      "loss": 0.9033,
+      "loss": 0.9019,
       "step": 73720
     },
     {
       "epoch": 221.41,
-      "grad_norm": 7.292811870574951,
+      "grad_norm": 8.356407165527344,
       "learning_rate": 2.61961961961962e-06,
-      "loss": 0.949,
+      "loss": 0.9414,
       "step": 73730
     },
     {
       "epoch": 221.44,
-      "grad_norm": 5.365231513977051,
+      "grad_norm": 4.5120849609375,
       "learning_rate": 2.6186186186186193e-06,
-      "loss": 0.8889,
+      "loss": 0.8865,
       "step": 73740
     },
     {
       "epoch": 221.47,
-      "grad_norm": 5.843244552612305,
+      "grad_norm": 6.609121799468994,
       "learning_rate": 2.617617617617618e-06,
-      "loss": 0.9545,
+      "loss": 0.9507,
       "step": 73750
     },
     {
       "epoch": 221.5,
-      "grad_norm": 6.870596885681152,
+      "grad_norm": 6.759649753570557,
       "learning_rate": 2.616616616616617e-06,
-      "loss": 0.9833,
+      "loss": 0.9841,
       "step": 73760
     },
     {
       "epoch": 221.53,
-      "grad_norm": 6.247148513793945,
+      "grad_norm": 9.197484970092773,
       "learning_rate": 2.615615615615616e-06,
-      "loss": 0.8725,
+      "loss": 0.8673,
       "step": 73770
     },
     {
       "epoch": 221.56,
-      "grad_norm": 5.311408996582031,
+      "grad_norm": 5.661225318908691,
       "learning_rate": 2.614614614614615e-06,
-      "loss": 0.8836,
+      "loss": 0.8862,
       "step": 73780
     },
     {
       "epoch": 221.59,
-      "grad_norm": 7.437593936920166,
+      "grad_norm": 7.871275901794434,
       "learning_rate": 2.6136136136136135e-06,
-      "loss": 1.0136,
+      "loss": 1.0094,
       "step": 73790
     },
     {
       "epoch": 221.62,
-      "grad_norm": 7.131797790527344,
+      "grad_norm": 6.304747581481934,
       "learning_rate": 2.612612612612613e-06,
-      "loss": 0.9533,
+      "loss": 0.9603,
       "step": 73800
     },
     {
       "epoch": 221.65,
-      "grad_norm": 5.440120220184326,
+      "grad_norm": 5.130614280700684,
       "learning_rate": 2.611611611611612e-06,
-      "loss": 0.9372,
+      "loss": 0.9324,
       "step": 73810
     },
     {
       "epoch": 221.68,
-      "grad_norm": 5.777447700500488,
+      "grad_norm": 6.1323137283325195,
       "learning_rate": 2.610610610610611e-06,
-      "loss": 0.9081,
+      "loss": 0.9084,
       "step": 73820
     },
     {
       "epoch": 221.71,
-      "grad_norm": 6.064887046813965,
+      "grad_norm": 6.476076126098633,
       "learning_rate": 2.6096096096096096e-06,
-      "loss": 0.947,
+      "loss": 0.95,
       "step": 73830
     },
     {
       "epoch": 221.74,
-      "grad_norm": 5.820742130279541,
+      "grad_norm": 5.702173233032227,
       "learning_rate": 2.6086086086086086e-06,
-      "loss": 0.9964,
+      "loss": 0.9814,
       "step": 73840
     },
     {
       "epoch": 221.77,
-      "grad_norm": 6.874671936035156,
+      "grad_norm": 7.079371929168701,
       "learning_rate": 2.607607607607608e-06,
-      "loss": 0.9333,
+      "loss": 0.9239,
       "step": 73850
     },
     {
       "epoch": 221.8,
-      "grad_norm": 4.872808933258057,
+      "grad_norm": 5.0653486251831055,
       "learning_rate": 2.606606606606607e-06,
-      "loss": 0.9163,
+      "loss": 0.9183,
       "step": 73860
     },
     {
       "epoch": 221.83,
-      "grad_norm": 5.172772407531738,
+      "grad_norm": 7.492395401000977,
       "learning_rate": 2.6056056056056057e-06,
-      "loss": 0.9422,
+      "loss": 0.9348,
       "step": 73870
     },
     {
       "epoch": 221.86,
-      "grad_norm": 14.91858959197998,
+      "grad_norm": 16.446496963500977,
       "learning_rate": 2.6046046046046047e-06,
-      "loss": 0.9382,
+      "loss": 0.93,
       "step": 73880
     },
     {
       "epoch": 221.89,
-      "grad_norm": 9.39197063446045,
+      "grad_norm": 6.306950569152832,
       "learning_rate": 2.6036036036036037e-06,
-      "loss": 0.9809,
+      "loss": 0.9817,
       "step": 73890
     },
     {
       "epoch": 221.92,
-      "grad_norm": 6.634497165679932,
+      "grad_norm": 8.093420028686523,
       "learning_rate": 2.602602602602603e-06,
       "loss": 0.9457,
       "step": 73900
     },
     {
       "epoch": 221.95,
-      "grad_norm": 6.501752853393555,
+      "grad_norm": 9.873311042785645,
       "learning_rate": 2.6016016016016017e-06,
-      "loss": 0.9779,
+      "loss": 0.9631,
       "step": 73910
     },
     {
       "epoch": 221.98,
-      "grad_norm": 5.442018032073975,
+      "grad_norm": 4.565486907958984,
       "learning_rate": 2.6006006006006008e-06,
-      "loss": 0.9376,
+      "loss": 0.9438,
       "step": 73920
     },
     {
       "epoch": 222.0,
-      "eval_accuracy": 0.8255,
-      "eval_loss": 0.5769674777984619,
-      "eval_runtime": 5.662,
-      "eval_samples_per_second": 1766.145,
-      "eval_steps_per_second": 7.065,
+      "eval_accuracy": 0.8278,
+      "eval_loss": 0.5759829878807068,
+      "eval_runtime": 5.0951,
+      "eval_samples_per_second": 1962.676,
+      "eval_steps_per_second": 7.851,
       "step": 73926
     },
     {
       "epoch": 222.01,
-      "grad_norm": 5.190420627593994,
+      "grad_norm": 5.919711589813232,
       "learning_rate": 2.5995995995995998e-06,
-      "loss": 1.037,
+      "loss": 1.018,
       "step": 73930
     },
     {
       "epoch": 222.04,
-      "grad_norm": 7.738117694854736,
+      "grad_norm": 7.365616798400879,
       "learning_rate": 2.598598598598599e-06,
-      "loss": 0.8956,
+      "loss": 0.8964,
       "step": 73940
     },
     {
       "epoch": 222.07,
-      "grad_norm": 6.058897972106934,
+      "grad_norm": 5.585597991943359,
       "learning_rate": 2.5975975975975974e-06,
-      "loss": 0.8936,
+      "loss": 0.8931,
       "step": 73950
     },
     {
       "epoch": 222.1,
-      "grad_norm": 6.553525924682617,
+      "grad_norm": 7.502383232116699,
       "learning_rate": 2.596596596596597e-06,
-      "loss": 0.9741,
+      "loss": 0.9551,
       "step": 73960
     },
     {
       "epoch": 222.13,
-      "grad_norm": 7.308353424072266,
+      "grad_norm": 6.419597625732422,
       "learning_rate": 2.595595595595596e-06,
-      "loss": 0.8925,
+      "loss": 0.8873,
       "step": 73970
     },
     {
       "epoch": 222.16,
-      "grad_norm": 7.797952651977539,
+      "grad_norm": 8.016457557678223,
       "learning_rate": 2.594594594594595e-06,
-      "loss": 0.9332,
+      "loss": 0.9319,
       "step": 73980
     },
     {
       "epoch": 222.19,
-      "grad_norm": 7.358076572418213,
+      "grad_norm": 7.2769575119018555,
       "learning_rate": 2.5935935935935935e-06,
-      "loss": 0.8799,
+      "loss": 0.8842,
       "step": 73990
     },
     {
       "epoch": 222.22,
-      "grad_norm": 6.540506362915039,
+      "grad_norm": 6.478492736816406,
       "learning_rate": 2.5925925925925925e-06,
-      "loss": 0.9172,
+      "loss": 0.9243,
       "step": 74000
     },
     {
       "epoch": 222.25,
-      "grad_norm": 8.85946273803711,
+      "grad_norm": 8.65617847442627,
       "learning_rate": 2.591591591591592e-06,
-      "loss": 0.9372,
+      "loss": 0.9374,
       "step": 74010
     },
     {
       "epoch": 222.28,
-      "grad_norm": 6.7123613357543945,
+      "grad_norm": 6.994907855987549,
       "learning_rate": 2.590590590590591e-06,
-      "loss": 1.0147,
+      "loss": 1.0162,
       "step": 74020
     },
     {
       "epoch": 222.31,
-      "grad_norm": 8.543917655944824,
+      "grad_norm": 9.297606468200684,
       "learning_rate": 2.58958958958959e-06,
-      "loss": 0.9943,
+      "loss": 0.9852,
       "step": 74030
     },
     {
       "epoch": 222.34,
-      "grad_norm": 7.667257308959961,
+      "grad_norm": 8.333304405212402,
       "learning_rate": 2.5885885885885886e-06,
-      "loss": 0.9434,
+      "loss": 0.9348,
       "step": 74040
     },
     {
       "epoch": 222.37,
-      "grad_norm": 6.042251110076904,
+      "grad_norm": 8.932394981384277,
       "learning_rate": 2.5875875875875876e-06,
-      "loss": 0.8881,
+      "loss": 0.8884,
       "step": 74050
     },
     {
       "epoch": 222.4,
-      "grad_norm": 6.216823577880859,
+      "grad_norm": 9.478158950805664,
       "learning_rate": 2.586586586586587e-06,
-      "loss": 0.8845,
+      "loss": 0.8817,
       "step": 74060
     },
     {
       "epoch": 222.43,
-      "grad_norm": 6.23570442199707,
+      "grad_norm": 8.103350639343262,
       "learning_rate": 2.585585585585586e-06,
-      "loss": 1.0012,
+      "loss": 0.9927,
       "step": 74070
     },
     {
       "epoch": 222.46,
-      "grad_norm": 5.778531551361084,
+      "grad_norm": 6.401103496551514,
       "learning_rate": 2.5845845845845846e-06,
-      "loss": 0.9752,
+      "loss": 0.9715,
       "step": 74080
     },
     {
       "epoch": 222.49,
-      "grad_norm": 6.917848587036133,
+      "grad_norm": 10.067179679870605,
       "learning_rate": 2.5835835835835837e-06,
-      "loss": 0.8699,
+      "loss": 0.8563,
       "step": 74090
     },
     {
       "epoch": 222.52,
-      "grad_norm": 5.54561185836792,
+      "grad_norm": 7.626528263092041,
       "learning_rate": 2.5825825825825827e-06,
-      "loss": 0.9799,
+      "loss": 0.9849,
       "step": 74100
     },
     {
       "epoch": 222.55,
-      "grad_norm": 6.321262836456299,
+      "grad_norm": 6.745280742645264,
       "learning_rate": 2.581581581581582e-06,
-      "loss": 0.884,
+      "loss": 0.8934,
       "step": 74110
     },
     {
       "epoch": 222.58,
-      "grad_norm": 7.205796241760254,
+      "grad_norm": 10.384072303771973,
       "learning_rate": 2.5805805805805807e-06,
-      "loss": 0.9331,
+      "loss": 0.9323,
       "step": 74120
     },
     {
       "epoch": 222.61,
-      "grad_norm": 4.964723587036133,
+      "grad_norm": 5.38451623916626,
       "learning_rate": 2.5795795795795797e-06,
-      "loss": 0.9691,
+      "loss": 0.9724,
       "step": 74130
     },
     {
       "epoch": 222.64,
-      "grad_norm": 9.014298439025879,
+      "grad_norm": 8.957328796386719,
       "learning_rate": 2.5785785785785787e-06,
-      "loss": 0.9664,
+      "loss": 0.9542,
       "step": 74140
     },
     {
       "epoch": 222.67,
-      "grad_norm": 7.847906112670898,
+      "grad_norm": 9.463728904724121,
       "learning_rate": 2.577577577577578e-06,
-      "loss": 0.9455,
+      "loss": 0.942,
       "step": 74150
     },
     {
       "epoch": 222.7,
-      "grad_norm": 5.840590953826904,
+      "grad_norm": 5.849276542663574,
       "learning_rate": 2.5765765765765764e-06,
-      "loss": 0.9635,
+      "loss": 0.9587,
       "step": 74160
     },
     {
       "epoch": 222.73,
-      "grad_norm": 7.6421918869018555,
+      "grad_norm": 8.888472557067871,
       "learning_rate": 2.575575575575576e-06,
-      "loss": 0.9501,
+      "loss": 0.9567,
       "step": 74170
     },
     {
       "epoch": 222.76,
-      "grad_norm": 6.255091667175293,
+      "grad_norm": 4.3170247077941895,
       "learning_rate": 2.574574574574575e-06,
-      "loss": 0.8562,
+      "loss": 0.8435,
       "step": 74180
     },
     {
       "epoch": 222.79,
-      "grad_norm": 6.597495079040527,
+      "grad_norm": 7.045745849609375,
       "learning_rate": 2.573573573573574e-06,
-      "loss": 0.9212,
+      "loss": 0.9096,
       "step": 74190
     },
     {
       "epoch": 222.82,
-      "grad_norm": 6.09224796295166,
+      "grad_norm": 6.111021041870117,
       "learning_rate": 2.5725725725725724e-06,
-      "loss": 0.9179,
+      "loss": 0.9121,
       "step": 74200
     },
     {
       "epoch": 222.85,
-      "grad_norm": 6.229904651641846,
+      "grad_norm": 6.307273864746094,
       "learning_rate": 2.571571571571572e-06,
-      "loss": 0.9689,
+      "loss": 0.968,
       "step": 74210
     },
     {
       "epoch": 222.88,
-      "grad_norm": 5.8077778816223145,
+      "grad_norm": 6.707244396209717,
       "learning_rate": 2.570570570570571e-06,
-      "loss": 0.9147,
+      "loss": 0.9138,
       "step": 74220
     },
     {
       "epoch": 222.91,
-      "grad_norm": 7.209314823150635,
+      "grad_norm": 6.896584987640381,
       "learning_rate": 2.56956956956957e-06,
-      "loss": 0.931,
+      "loss": 0.9401,
       "step": 74230
     },
     {
       "epoch": 222.94,
-      "grad_norm": 6.754297733306885,
+      "grad_norm": 7.570929050445557,
       "learning_rate": 2.5685685685685685e-06,
-      "loss": 1.0016,
+      "loss": 0.9968,
       "step": 74240
     },
     {
       "epoch": 222.97,
-      "grad_norm": 6.429629325866699,
+      "grad_norm": 7.327308177947998,
       "learning_rate": 2.5675675675675675e-06,
-      "loss": 0.8995,
+      "loss": 0.8914,
       "step": 74250
     },
     {
       "epoch": 223.0,
-      "eval_accuracy": 0.8251,
-      "eval_loss": 0.5791096687316895,
-      "eval_runtime": 5.3935,
-      "eval_samples_per_second": 1854.081,
-      "eval_steps_per_second": 7.416,
+      "eval_accuracy": 0.8265,
+      "eval_loss": 0.5770121812820435,
+      "eval_runtime": 5.1983,
+      "eval_samples_per_second": 1923.709,
+      "eval_steps_per_second": 7.695,
       "step": 74259
     },
     {
       "epoch": 223.0,
-      "grad_norm": 6.019561767578125,
+      "grad_norm": 7.917863368988037,
       "learning_rate": 2.566566566566567e-06,
-      "loss": 1.039,
+      "loss": 1.0448,
       "step": 74260
     },
     {
       "epoch": 223.03,
-      "grad_norm": 5.3815507888793945,
+      "grad_norm": 6.662524700164795,
       "learning_rate": 2.565565565565566e-06,
-      "loss": 0.9008,
+      "loss": 0.9029,
       "step": 74270
     },
     {
       "epoch": 223.06,
-      "grad_norm": 4.733813762664795,
+      "grad_norm": 4.936094760894775,
       "learning_rate": 2.564564564564565e-06,
-      "loss": 0.8977,
+      "loss": 0.8975,
       "step": 74280
     },
     {
       "epoch": 223.09,
-      "grad_norm": 5.605366230010986,
+      "grad_norm": 6.748191833496094,
       "learning_rate": 2.5635635635635636e-06,
-      "loss": 0.9034,
+      "loss": 0.9,
       "step": 74290
     },
     {
       "epoch": 223.12,
-      "grad_norm": 6.222340106964111,
+      "grad_norm": 8.181519508361816,
       "learning_rate": 2.5625625625625626e-06,
-      "loss": 0.9273,
+      "loss": 0.9302,
       "step": 74300
     },
     {
       "epoch": 223.15,
-      "grad_norm": 9.072317123413086,
+      "grad_norm": 7.332512378692627,
       "learning_rate": 2.561561561561562e-06,
-      "loss": 0.9255,
+      "loss": 0.9328,
       "step": 74310
     },
     {
       "epoch": 223.18,
-      "grad_norm": 4.786988735198975,
+      "grad_norm": 6.509355545043945,
       "learning_rate": 2.560560560560561e-06,
-      "loss": 0.8813,
+      "loss": 0.8876,
       "step": 74320
     },
     {
       "epoch": 223.21,
-      "grad_norm": 6.8199849128723145,
+      "grad_norm": 6.553347587585449,
       "learning_rate": 2.5595595595595597e-06,
-      "loss": 0.9953,
+      "loss": 0.9913,
       "step": 74330
     },
     {
       "epoch": 223.24,
-      "grad_norm": 6.405803203582764,
+      "grad_norm": 5.9009552001953125,
       "learning_rate": 2.5585585585585587e-06,
       "loss": 0.9057,
       "step": 74340
     },
     {
       "epoch": 223.27,
-      "grad_norm": 7.105982780456543,
+      "grad_norm": 6.058664321899414,
       "learning_rate": 2.5575575575575577e-06,
-      "loss": 0.9307,
+      "loss": 0.9303,
       "step": 74350
     },
     {
       "epoch": 223.3,
-      "grad_norm": 6.282003879547119,
+      "grad_norm": 6.366199493408203,
       "learning_rate": 2.556556556556557e-06,
-      "loss": 0.8703,
+      "loss": 0.8537,
       "step": 74360
     },
     {
       "epoch": 223.33,
-      "grad_norm": 6.726934432983398,
+      "grad_norm": 10.70263385772705,
       "learning_rate": 2.5555555555555557e-06,
-      "loss": 0.933,
+      "loss": 0.9224,
       "step": 74370
     },
     {
       "epoch": 223.36,
-      "grad_norm": 6.302299976348877,
+      "grad_norm": 8.522193908691406,
       "learning_rate": 2.5545545545545548e-06,
-      "loss": 0.94,
+      "loss": 0.9541,
       "step": 74380
     },
     {
       "epoch": 223.39,
-      "grad_norm": 8.592718124389648,
+      "grad_norm": 6.065980911254883,
       "learning_rate": 2.5535535535535538e-06,
-      "loss": 0.9143,
+      "loss": 0.9132,
       "step": 74390
     },
     {
       "epoch": 223.42,
-      "grad_norm": 7.27379035949707,
+      "grad_norm": 7.252096176147461,
       "learning_rate": 2.552552552552553e-06,
-      "loss": 0.9177,
+      "loss": 0.9035,
       "step": 74400
     },
     {
       "epoch": 223.45,
-      "grad_norm": 11.154006958007812,
+      "grad_norm": 9.853524208068848,
       "learning_rate": 2.5515515515515514e-06,
-      "loss": 0.968,
+      "loss": 0.9623,
       "step": 74410
     },
     {
       "epoch": 223.48,
-      "grad_norm": 6.255345344543457,
+      "grad_norm": 8.721223831176758,
       "learning_rate": 2.550550550550551e-06,
-      "loss": 0.8785,
+      "loss": 0.8697,
       "step": 74420
     },
     {
       "epoch": 223.51,
-      "grad_norm": 6.750766277313232,
+      "grad_norm": 8.319167137145996,
       "learning_rate": 2.54954954954955e-06,
-      "loss": 0.9618,
+      "loss": 0.9546,
       "step": 74430
     },
     {
       "epoch": 223.54,
-      "grad_norm": 6.698027610778809,
+      "grad_norm": 6.748810291290283,
       "learning_rate": 2.548548548548549e-06,
-      "loss": 0.9574,
+      "loss": 0.9672,
       "step": 74440
     },
     {
       "epoch": 223.57,
-      "grad_norm": 7.346471786499023,
+      "grad_norm": 7.520939826965332,
       "learning_rate": 2.5475475475475475e-06,
-      "loss": 0.9369,
+      "loss": 0.9438,
       "step": 74450
     },
     {
       "epoch": 223.6,
-      "grad_norm": 7.015319347381592,
+      "grad_norm": 9.136316299438477,
       "learning_rate": 2.5465465465465465e-06,
-      "loss": 0.9167,
+      "loss": 0.9121,
       "step": 74460
     },
     {
       "epoch": 223.63,
-      "grad_norm": 7.76850700378418,
+      "grad_norm": 6.85337495803833,
       "learning_rate": 2.545545545545546e-06,
-      "loss": 0.9695,
+      "loss": 0.9758,
       "step": 74470
     },
     {
       "epoch": 223.66,
-      "grad_norm": 9.976987838745117,
+      "grad_norm": 8.062381744384766,
       "learning_rate": 2.544544544544545e-06,
-      "loss": 0.9266,
+      "loss": 0.9298,
       "step": 74480
     },
     {
       "epoch": 223.69,
-      "grad_norm": 8.472260475158691,
+      "grad_norm": 10.43685531616211,
       "learning_rate": 2.5435435435435435e-06,
-      "loss": 0.9593,
+      "loss": 0.9674,
       "step": 74490
     },
     {
       "epoch": 223.72,
-      "grad_norm": 8.222759246826172,
+      "grad_norm": 6.837255477905273,
       "learning_rate": 2.5425425425425426e-06,
-      "loss": 0.9446,
+      "loss": 0.941,
       "step": 74500
     },
     {
       "epoch": 223.75,
-      "grad_norm": 8.98372745513916,
+      "grad_norm": 7.767437934875488,
       "learning_rate": 2.5415415415415416e-06,
-      "loss": 0.9224,
+      "loss": 0.9266,
       "step": 74510
     },
     {
       "epoch": 223.78,
-      "grad_norm": 6.270655632019043,
+      "grad_norm": 6.015834808349609,
       "learning_rate": 2.540540540540541e-06,
-      "loss": 0.8953,
+      "loss": 0.9104,
       "step": 74520
     },
     {
       "epoch": 223.81,
-      "grad_norm": 9.513577461242676,
+      "grad_norm": 7.071451187133789,
       "learning_rate": 2.53953953953954e-06,
-      "loss": 0.8782,
+      "loss": 0.867,
       "step": 74530
     },
     {
       "epoch": 223.84,
-      "grad_norm": 7.895255088806152,
+      "grad_norm": 10.016141891479492,
       "learning_rate": 2.5385385385385386e-06,
-      "loss": 0.928,
+      "loss": 0.9278,
       "step": 74540
     },
     {
       "epoch": 223.87,
-      "grad_norm": 6.233753204345703,
+      "grad_norm": 7.1616530418396,
       "learning_rate": 2.5375375375375377e-06,
-      "loss": 0.9217,
+      "loss": 0.9091,
       "step": 74550
     },
     {
       "epoch": 223.9,
-      "grad_norm": 5.042590618133545,
+      "grad_norm": 6.690983295440674,
       "learning_rate": 2.5365365365365367e-06,
-      "loss": 0.9249,
+      "loss": 0.9192,
       "step": 74560
     },
     {
       "epoch": 223.93,
-      "grad_norm": 8.883967399597168,
+      "grad_norm": 7.306665897369385,
       "learning_rate": 2.535535535535536e-06,
-      "loss": 0.9399,
+      "loss": 0.9496,
       "step": 74570
     },
     {
       "epoch": 223.96,
-      "grad_norm": 9.629941940307617,
+      "grad_norm": 7.320693492889404,
       "learning_rate": 2.5345345345345347e-06,
-      "loss": 0.9101,
+      "loss": 0.9141,
       "step": 74580
     },
     {
       "epoch": 223.99,
-      "grad_norm": 7.151265621185303,
+      "grad_norm": 10.578472137451172,
       "learning_rate": 2.5335335335335337e-06,
-      "loss": 0.8994,
+      "loss": 0.8957,
       "step": 74590
     },
     {
       "epoch": 224.0,
-      "eval_accuracy": 0.8266,
-      "eval_loss": 0.571617603302002,
-      "eval_runtime": 5.5163,
-      "eval_samples_per_second": 1812.8,
-      "eval_steps_per_second": 7.251,
+      "eval_accuracy": 0.8286,
+      "eval_loss": 0.57008957862854,
+      "eval_runtime": 5.2871,
+      "eval_samples_per_second": 1891.392,
+      "eval_steps_per_second": 7.566,
       "step": 74592
     },
     {
       "epoch": 224.02,
-      "grad_norm": 6.524591445922852,
+      "grad_norm": 6.604224681854248,
       "learning_rate": 2.5325325325325327e-06,
-      "loss": 1.0998,
+      "loss": 1.1182,
       "step": 74600
     },
     {
       "epoch": 224.05,
-      "grad_norm": 8.450315475463867,
+      "grad_norm": 10.129018783569336,
       "learning_rate": 2.5315315315315318e-06,
-      "loss": 0.9053,
+      "loss": 0.9078,
       "step": 74610
     },
     {
       "epoch": 224.08,
-      "grad_norm": 8.242596626281738,
+      "grad_norm": 8.119063377380371,
       "learning_rate": 2.5305305305305304e-06,
-      "loss": 0.9084,
+      "loss": 0.8968,
       "step": 74620
     },
     {
       "epoch": 224.11,
-      "grad_norm": 4.3206048011779785,
+      "grad_norm": 5.15307092666626,
       "learning_rate": 2.52952952952953e-06,
-      "loss": 0.8953,
+      "loss": 0.8957,
       "step": 74630
     },
     {
       "epoch": 224.14,
-      "grad_norm": 11.02884292602539,
+      "grad_norm": 10.398945808410645,
       "learning_rate": 2.528528528528529e-06,
-      "loss": 0.8983,
+      "loss": 0.9042,
       "step": 74640
     },
     {
       "epoch": 224.17,
-      "grad_norm": 8.10582160949707,
+      "grad_norm": 10.193520545959473,
       "learning_rate": 2.527527527527528e-06,
-      "loss": 0.9363,
+      "loss": 0.9428,
       "step": 74650
     },
     {
       "epoch": 224.2,
-      "grad_norm": 5.4516777992248535,
+      "grad_norm": 7.356095790863037,
       "learning_rate": 2.5265265265265264e-06,
-      "loss": 0.9108,
+      "loss": 0.9078,
       "step": 74660
     },
     {
       "epoch": 224.23,
-      "grad_norm": 13.315930366516113,
+      "grad_norm": 10.015776634216309,
       "learning_rate": 2.5255255255255254e-06,
-      "loss": 0.9283,
+      "loss": 0.9235,
       "step": 74670
     },
     {
       "epoch": 224.26,
-      "grad_norm": 5.0367279052734375,
+      "grad_norm": 5.614912033081055,
       "learning_rate": 2.524524524524525e-06,
-      "loss": 0.8641,
+      "loss": 0.8555,
       "step": 74680
     },
     {
       "epoch": 224.29,
-      "grad_norm": 6.676198959350586,
+      "grad_norm": 5.952676296234131,
       "learning_rate": 2.523523523523524e-06,
-      "loss": 0.9148,
+      "loss": 0.9177,
       "step": 74690
     },
     {
       "epoch": 224.32,
-      "grad_norm": 5.47479248046875,
+      "grad_norm": 5.654432773590088,
       "learning_rate": 2.5225225225225225e-06,
-      "loss": 0.8726,
+      "loss": 0.871,
       "step": 74700
     },
     {
       "epoch": 224.35,
-      "grad_norm": 7.639857769012451,
+      "grad_norm": 6.294461727142334,
       "learning_rate": 2.5215215215215215e-06,
-      "loss": 0.9075,
+      "loss": 0.9042,
       "step": 74710
     },
     {
       "epoch": 224.38,
-      "grad_norm": 9.766960144042969,
+      "grad_norm": 10.133631706237793,
       "learning_rate": 2.520520520520521e-06,
-      "loss": 0.8929,
+      "loss": 0.8892,
       "step": 74720
     },
     {
       "epoch": 224.41,
-      "grad_norm": 6.96771764755249,
+      "grad_norm": 4.900798797607422,
       "learning_rate": 2.51951951951952e-06,
-      "loss": 0.9912,
+      "loss": 0.9962,
       "step": 74730
     },
     {
       "epoch": 224.44,
-      "grad_norm": 5.707102298736572,
+      "grad_norm": 6.446041107177734,
       "learning_rate": 2.5185185185185186e-06,
-      "loss": 0.8385,
+      "loss": 0.838,
       "step": 74740
     },
     {
       "epoch": 224.47,
-      "grad_norm": 5.323373317718506,
+      "grad_norm": 5.252581596374512,
       "learning_rate": 2.5175175175175176e-06,
-      "loss": 0.888,
+      "loss": 0.8852,
       "step": 74750
     },
     {
       "epoch": 224.5,
-      "grad_norm": 5.204634666442871,
+      "grad_norm": 5.980249404907227,
       "learning_rate": 2.5165165165165166e-06,
-      "loss": 0.934,
+      "loss": 0.9365,
       "step": 74760
     },
     {
       "epoch": 224.53,
-      "grad_norm": 6.643959045410156,
+      "grad_norm": 5.990653038024902,
       "learning_rate": 2.515515515515516e-06,
-      "loss": 0.9379,
+      "loss": 0.9439,
       "step": 74770
     },
     {
       "epoch": 224.56,
-      "grad_norm": 7.302264213562012,
+      "grad_norm": 6.5474162101745605,
       "learning_rate": 2.5145145145145147e-06,
-      "loss": 0.8808,
+      "loss": 0.8791,
       "step": 74780
     },
     {
       "epoch": 224.59,
-      "grad_norm": 6.088219165802002,
+      "grad_norm": 6.155712604522705,
       "learning_rate": 2.5135135135135137e-06,
-      "loss": 0.9671,
+      "loss": 0.9717,
       "step": 74790
     },
     {
       "epoch": 224.62,
-      "grad_norm": 4.886549472808838,
+      "grad_norm": 5.877361297607422,
       "learning_rate": 2.5125125125125127e-06,
-      "loss": 0.889,
+      "loss": 0.8856,
       "step": 74800
     },
     {
       "epoch": 224.65,
-      "grad_norm": 7.195277690887451,
+      "grad_norm": 5.9478440284729,
       "learning_rate": 2.5115115115115117e-06,
-      "loss": 0.8727,
+      "loss": 0.8636,
       "step": 74810
     },
     {
       "epoch": 224.68,
-      "grad_norm": 8.750214576721191,
+      "grad_norm": 8.356313705444336,
       "learning_rate": 2.510510510510511e-06,
-      "loss": 0.9458,
+      "loss": 0.9455,
       "step": 74820
     },
     {
       "epoch": 224.71,
-      "grad_norm": 8.439972877502441,
+      "grad_norm": 8.916015625,
       "learning_rate": 2.5095095095095097e-06,
-      "loss": 0.8538,
+      "loss": 0.859,
       "step": 74830
     },
     {
       "epoch": 224.74,
-      "grad_norm": 4.960003852844238,
+      "grad_norm": 5.350255489349365,
       "learning_rate": 2.5085085085085088e-06,
-      "loss": 1.0254,
+      "loss": 1.0209,
       "step": 74840
     },
     {
       "epoch": 224.77,
-      "grad_norm": 6.330846309661865,
+      "grad_norm": 6.9167985916137695,
       "learning_rate": 2.5075075075075078e-06,
-      "loss": 0.9489,
+      "loss": 0.9499,
       "step": 74850
     },
     {
       "epoch": 224.8,
-      "grad_norm": 7.6932806968688965,
+      "grad_norm": 7.823912143707275,
       "learning_rate": 2.506506506506507e-06,
-      "loss": 0.922,
+      "loss": 0.918,
       "step": 74860
     },
     {
       "epoch": 224.83,
-      "grad_norm": 4.542179584503174,
+      "grad_norm": 6.848102569580078,
       "learning_rate": 2.5055055055055054e-06,
-      "loss": 0.8723,
+      "loss": 0.861,
       "step": 74870
     },
     {
       "epoch": 224.86,
-      "grad_norm": 9.03976058959961,
+      "grad_norm": 6.438268661499023,
       "learning_rate": 2.504504504504505e-06,
-      "loss": 0.9011,
+      "loss": 0.9054,
       "step": 74880
     },
     {
       "epoch": 224.89,
-      "grad_norm": 6.287628650665283,
+      "grad_norm": 7.034922122955322,
       "learning_rate": 2.503503503503504e-06,
-      "loss": 0.9413,
+      "loss": 0.9303,
       "step": 74890
     },
     {
       "epoch": 224.92,
-      "grad_norm": 11.093245506286621,
+      "grad_norm": 8.802770614624023,
       "learning_rate": 2.502502502502503e-06,
-      "loss": 0.9495,
+      "loss": 0.9536,
       "step": 74900
     },
     {
       "epoch": 224.95,
-      "grad_norm": 6.584711074829102,
+      "grad_norm": 5.882460594177246,
       "learning_rate": 2.5015015015015015e-06,
-      "loss": 0.9677,
+      "loss": 0.9557,
       "step": 74910
     },
     {
       "epoch": 224.98,
-      "grad_norm": 8.790388107299805,
+      "grad_norm": 9.70560073852539,
       "learning_rate": 2.5005005005005005e-06,
-      "loss": 0.908,
+      "loss": 0.9086,
       "step": 74920
     },
     {
       "epoch": 225.0,
-      "eval_accuracy": 0.825,
-      "eval_loss": 0.5742149949073792,
-      "eval_runtime": 5.5041,
-      "eval_samples_per_second": 1816.811,
-      "eval_steps_per_second": 7.267,
+      "eval_accuracy": 0.8272,
+      "eval_loss": 0.5724507570266724,
+      "eval_runtime": 5.3607,
+      "eval_samples_per_second": 1865.429,
+      "eval_steps_per_second": 7.462,
       "step": 74925
     },
     {
       "epoch": 225.02,
-      "grad_norm": 6.253717422485352,
+      "grad_norm": 6.325930595397949,
       "learning_rate": 2.4994994994995e-06,
-      "loss": 1.2304,
+      "loss": 1.223,
       "step": 74930
     },
     {
       "epoch": 225.05,
-      "grad_norm": 7.210975170135498,
+      "grad_norm": 7.203902721405029,
       "learning_rate": 2.4984984984984985e-06,
-      "loss": 0.9374,
+      "loss": 0.9324,
       "step": 74940
     },
     {
       "epoch": 225.08,
-      "grad_norm": 8.135015487670898,
+      "grad_norm": 6.170255661010742,
       "learning_rate": 2.497497497497498e-06,
-      "loss": 0.9551,
+      "loss": 0.9579,
       "step": 74950
     },
     {
       "epoch": 225.11,
-      "grad_norm": 5.425216197967529,
+      "grad_norm": 5.973384857177734,
       "learning_rate": 2.4964964964964966e-06,
-      "loss": 0.9695,
+      "loss": 0.9588,
       "step": 74960
     },
     {
       "epoch": 225.14,
-      "grad_norm": 6.576850414276123,
+      "grad_norm": 6.144732475280762,
       "learning_rate": 2.4954954954954956e-06,
-      "loss": 0.9132,
+      "loss": 0.9154,
       "step": 74970
     },
     {
       "epoch": 225.17,
-      "grad_norm": 10.272638320922852,
+      "grad_norm": 7.2395501136779785,
       "learning_rate": 2.4944944944944946e-06,
-      "loss": 0.879,
+      "loss": 0.88,
       "step": 74980
     },
     {
       "epoch": 225.2,
-      "grad_norm": 7.420457363128662,
+      "grad_norm": 7.746460914611816,
       "learning_rate": 2.4934934934934936e-06,
-      "loss": 0.8665,
+      "loss": 0.8663,
       "step": 74990
     },
     {
       "epoch": 225.23,
-      "grad_norm": 6.233279228210449,
+      "grad_norm": 6.93830680847168,
       "learning_rate": 2.4924924924924926e-06,
-      "loss": 0.9789,
+      "loss": 0.98,
       "step": 75000
     },
     {
       "epoch": 225.26,
-      "grad_norm": 7.205628395080566,
+      "grad_norm": 5.9508538246154785,
       "learning_rate": 2.4914914914914916e-06,
-      "loss": 0.935,
+      "loss": 0.9472,
       "step": 75010
     },
     {
       "epoch": 225.29,
-      "grad_norm": 5.983125686645508,
+      "grad_norm": 5.7842278480529785,
       "learning_rate": 2.4904904904904907e-06,
-      "loss": 0.8311,
+      "loss": 0.8349,
       "step": 75020
     },
     {
       "epoch": 225.32,
-      "grad_norm": 9.39879322052002,
+      "grad_norm": 8.813572883605957,
       "learning_rate": 2.4894894894894897e-06,
-      "loss": 0.9246,
+      "loss": 0.9287,
       "step": 75030
     },
     {
       "epoch": 225.35,
-      "grad_norm": 6.07961368560791,
+      "grad_norm": 7.523661136627197,
       "learning_rate": 2.4884884884884887e-06,
-      "loss": 0.9381,
+      "loss": 0.9406,
       "step": 75040
     },
     {
       "epoch": 225.38,
-      "grad_norm": 5.919928550720215,
+      "grad_norm": 7.392022132873535,
       "learning_rate": 2.4874874874874877e-06,
-      "loss": 0.8387,
+      "loss": 0.8353,
       "step": 75050
     },
     {
       "epoch": 225.41,
-      "grad_norm": 7.9262375831604,
+      "grad_norm": 7.740636348724365,
       "learning_rate": 2.4864864864864867e-06,
-      "loss": 0.9341,
+      "loss": 0.9218,
       "step": 75060
     },
     {
       "epoch": 225.44,
-      "grad_norm": 8.35022258758545,
+      "grad_norm": 8.379388809204102,
       "learning_rate": 2.4854854854854858e-06,
-      "loss": 0.8987,
+      "loss": 0.8874,
       "step": 75070
     },
     {
       "epoch": 225.47,
-      "grad_norm": 6.792163848876953,
+      "grad_norm": 7.774751663208008,
       "learning_rate": 2.4844844844844848e-06,
-      "loss": 0.9908,
+      "loss": 0.9781,
       "step": 75080
     },
     {
       "epoch": 225.5,
-      "grad_norm": 9.036714553833008,
+      "grad_norm": 7.804744720458984,
       "learning_rate": 2.483483483483484e-06,
-      "loss": 0.8806,
+      "loss": 0.8848,
       "step": 75090
     },
     {
       "epoch": 225.53,
-      "grad_norm": 9.351632118225098,
+      "grad_norm": 7.57387638092041,
       "learning_rate": 2.482482482482483e-06,
-      "loss": 0.9061,
+      "loss": 0.907,
       "step": 75100
     },
     {
       "epoch": 225.56,
-      "grad_norm": 4.703951358795166,
+      "grad_norm": 5.042533874511719,
       "learning_rate": 2.481481481481482e-06,
-      "loss": 0.9651,
+      "loss": 0.9541,
       "step": 75110
     },
     {
       "epoch": 225.59,
-      "grad_norm": 8.401101112365723,
+      "grad_norm": 10.726346969604492,
       "learning_rate": 2.480480480480481e-06,
-      "loss": 0.9304,
+      "loss": 0.9326,
       "step": 75120
     },
     {
       "epoch": 225.62,
-      "grad_norm": 9.737022399902344,
+      "grad_norm": 8.381173133850098,
       "learning_rate": 2.4794794794794794e-06,
-      "loss": 0.919,
+      "loss": 0.9183,
       "step": 75130
     },
     {
       "epoch": 225.65,
-      "grad_norm": 6.211675643920898,
+      "grad_norm": 7.164147853851318,
       "learning_rate": 2.478478478478479e-06,
-      "loss": 0.9112,
+      "loss": 0.9117,
       "step": 75140
     },
     {
       "epoch": 225.68,
-      "grad_norm": 5.487493991851807,
+      "grad_norm": 7.8083367347717285,
       "learning_rate": 2.4774774774774775e-06,
-      "loss": 0.9007,
+      "loss": 0.8996,
       "step": 75150
     },
     {
       "epoch": 225.71,
-      "grad_norm": 9.082365989685059,
+      "grad_norm": 8.88070011138916,
       "learning_rate": 2.476476476476477e-06,
-      "loss": 0.9556,
+      "loss": 0.9516,
       "step": 75160
     },
     {
       "epoch": 225.74,
-      "grad_norm": 5.31629753112793,
+      "grad_norm": 6.303347110748291,
       "learning_rate": 2.4754754754754755e-06,
-      "loss": 0.8828,
+      "loss": 0.8787,
       "step": 75170
     },
     {
       "epoch": 225.77,
-      "grad_norm": 5.094243049621582,
+      "grad_norm": 5.7390923500061035,
       "learning_rate": 2.474474474474475e-06,
-      "loss": 0.9603,
+      "loss": 0.9523,
       "step": 75180
     },
     {
       "epoch": 225.8,
-      "grad_norm": 5.534727096557617,
+      "grad_norm": 6.279656887054443,
       "learning_rate": 2.4734734734734736e-06,
-      "loss": 0.9512,
+      "loss": 0.9445,
       "step": 75190
     },
     {
       "epoch": 225.83,
-      "grad_norm": 5.695630073547363,
+      "grad_norm": 5.056314468383789,
       "learning_rate": 2.4724724724724726e-06,
-      "loss": 0.8959,
+      "loss": 0.8881,
       "step": 75200
     },
     {
       "epoch": 225.86,
-      "grad_norm": 5.958025932312012,
+      "grad_norm": 6.7159833908081055,
       "learning_rate": 2.4714714714714716e-06,
-      "loss": 0.9639,
+      "loss": 0.9633,
       "step": 75210
     },
     {
       "epoch": 225.89,
-      "grad_norm": 4.745790481567383,
+      "grad_norm": 3.9337379932403564,
       "learning_rate": 2.4704704704704706e-06,
-      "loss": 0.9178,
+      "loss": 0.9093,
       "step": 75220
     },
     {
       "epoch": 225.92,
-      "grad_norm": 5.098117351531982,
+      "grad_norm": 7.433837890625,
       "learning_rate": 2.4694694694694696e-06,
-      "loss": 0.8654,
+      "loss": 0.8669,
       "step": 75230
     },
     {
       "epoch": 225.95,
-      "grad_norm": 7.408809185028076,
+      "grad_norm": 7.520740985870361,
       "learning_rate": 2.4684684684684686e-06,
-      "loss": 0.9247,
+      "loss": 0.9177,
       "step": 75240
     },
     {
       "epoch": 225.98,
-      "grad_norm": 7.872175216674805,
+      "grad_norm": 10.443479537963867,
       "learning_rate": 2.4674674674674677e-06,
-      "loss": 0.9579,
+      "loss": 0.9508,
       "step": 75250
     },
     {
       "epoch": 226.0,
-      "eval_accuracy": 0.8234,
-      "eval_loss": 0.5761713981628418,
-      "eval_runtime": 5.4716,
-      "eval_samples_per_second": 1827.605,
-      "eval_steps_per_second": 7.31,
+      "eval_accuracy": 0.8276,
+      "eval_loss": 0.5761318802833557,
+      "eval_runtime": 5.4031,
+      "eval_samples_per_second": 1850.773,
+      "eval_steps_per_second": 7.403,
       "step": 75258
     },
     {
       "epoch": 226.01,
-      "grad_norm": 5.972994327545166,
+      "grad_norm": 6.557273864746094,
       "learning_rate": 2.4664664664664667e-06,
-      "loss": 1.1579,
+      "loss": 1.181,
       "step": 75260
     },
     {
       "epoch": 226.04,
-      "grad_norm": 8.373980522155762,
+      "grad_norm": 10.304239273071289,
       "learning_rate": 2.4654654654654657e-06,
-      "loss": 0.8998,
+      "loss": 0.903,
       "step": 75270
     },
     {
       "epoch": 226.07,
-      "grad_norm": 5.554232120513916,
+      "grad_norm": 6.290005683898926,
       "learning_rate": 2.4644644644644647e-06,
-      "loss": 0.9513,
+      "loss": 0.9478,
       "step": 75280
     },
     {
       "epoch": 226.1,
-      "grad_norm": 6.122977256774902,
+      "grad_norm": 5.872477054595947,
       "learning_rate": 2.4634634634634637e-06,
-      "loss": 0.9314,
+      "loss": 0.9284,
       "step": 75290
     },
     {
       "epoch": 226.13,
-      "grad_norm": 5.175888538360596,
+      "grad_norm": 6.2103705406188965,
       "learning_rate": 2.4624624624624628e-06,
-      "loss": 0.9379,
+      "loss": 0.9376,
       "step": 75300
     },
     {
       "epoch": 226.16,
-      "grad_norm": 11.852643013000488,
+      "grad_norm": 9.100605964660645,
       "learning_rate": 2.4614614614614614e-06,
-      "loss": 0.93,
+      "loss": 0.925,
       "step": 75310
     },
     {
       "epoch": 226.19,
-      "grad_norm": 6.011014938354492,
+      "grad_norm": 5.862051010131836,
       "learning_rate": 2.460460460460461e-06,
-      "loss": 0.9642,
+      "loss": 0.9607,
       "step": 75320
     },
     {
       "epoch": 226.22,
-      "grad_norm": 6.178789138793945,
+      "grad_norm": 6.718263626098633,
       "learning_rate": 2.45945945945946e-06,
-      "loss": 1.0075,
+      "loss": 1.0084,
       "step": 75330
     },
     {
       "epoch": 226.25,
-      "grad_norm": 8.072211265563965,
+      "grad_norm": 8.47504711151123,
       "learning_rate": 2.458458458458459e-06,
-      "loss": 0.9573,
+      "loss": 0.9646,
       "step": 75340
     },
     {
       "epoch": 226.28,
-      "grad_norm": 6.576203346252441,
+      "grad_norm": 8.079773902893066,
       "learning_rate": 2.457457457457458e-06,
-      "loss": 0.9117,
+      "loss": 0.9087,
       "step": 75350
     },
     {
       "epoch": 226.31,
-      "grad_norm": 9.733781814575195,
+      "grad_norm": 6.5371246337890625,
       "learning_rate": 2.4564564564564564e-06,
       "loss": 0.8751,
       "step": 75360
     },
     {
       "epoch": 226.34,
-      "grad_norm": 7.18788480758667,
+      "grad_norm": 6.3766703605651855,
       "learning_rate": 2.455455455455456e-06,
-      "loss": 0.9733,
+      "loss": 0.9622,
       "step": 75370
     },
     {
       "epoch": 226.37,
-      "grad_norm": 6.536566257476807,
+      "grad_norm": 5.011795997619629,
       "learning_rate": 2.4544544544544545e-06,
-      "loss": 0.87,
+      "loss": 0.8793,
       "step": 75380
     },
     {
       "epoch": 226.4,
-      "grad_norm": 4.961673736572266,
+      "grad_norm": 5.63006067276001,
       "learning_rate": 2.453453453453454e-06,
-      "loss": 0.8977,
+      "loss": 0.9014,
       "step": 75390
     },
     {
       "epoch": 226.43,
-      "grad_norm": 6.693857669830322,
+      "grad_norm": 5.938263893127441,
       "learning_rate": 2.4524524524524525e-06,
-      "loss": 0.8814,
+      "loss": 0.8829,
       "step": 75400
     },
     {
       "epoch": 226.46,
-      "grad_norm": 8.87157917022705,
+      "grad_norm": 9.705885887145996,
       "learning_rate": 2.451451451451452e-06,
-      "loss": 0.8711,
+      "loss": 0.8688,
       "step": 75410
     },
     {
       "epoch": 226.49,
-      "grad_norm": 5.287412166595459,
+      "grad_norm": 6.37297248840332,
       "learning_rate": 2.4504504504504506e-06,
-      "loss": 0.9343,
+      "loss": 0.9379,
       "step": 75420
     },
     {
       "epoch": 226.52,
-      "grad_norm": 7.799742221832275,
+      "grad_norm": 8.543729782104492,
       "learning_rate": 2.4494494494494496e-06,
-      "loss": 0.9304,
+      "loss": 0.9211,
       "step": 75430
     },
     {
       "epoch": 226.55,
-      "grad_norm": 8.808121681213379,
+      "grad_norm": 6.9567999839782715,
       "learning_rate": 2.4484484484484486e-06,
-      "loss": 0.9859,
+      "loss": 0.9837,
       "step": 75440
     },
     {
       "epoch": 226.58,
-      "grad_norm": 7.620312690734863,
+      "grad_norm": 7.018962383270264,
       "learning_rate": 2.4474474474474476e-06,
-      "loss": 0.9144,
+      "loss": 0.9266,
       "step": 75450
     },
     {
       "epoch": 226.61,
-      "grad_norm": 7.5413384437561035,
+      "grad_norm": 7.44931697845459,
       "learning_rate": 2.4464464464464466e-06,
-      "loss": 0.8983,
+      "loss": 0.8994,
       "step": 75460
     },
     {
       "epoch": 226.64,
-      "grad_norm": 6.789848804473877,
+      "grad_norm": 7.560495853424072,
       "learning_rate": 2.4454454454454456e-06,
-      "loss": 0.8952,
+      "loss": 0.9006,
       "step": 75470
     },
     {
       "epoch": 226.67,
-      "grad_norm": 9.872308731079102,
+      "grad_norm": 9.642467498779297,
       "learning_rate": 2.4444444444444447e-06,
-      "loss": 0.9817,
+      "loss": 0.9838,
       "step": 75480
     },
     {
       "epoch": 226.7,
-      "grad_norm": 6.7767815589904785,
+      "grad_norm": 7.648903846740723,
       "learning_rate": 2.4434434434434437e-06,
-      "loss": 0.9143,
+      "loss": 0.9081,
       "step": 75490
     },
     {
       "epoch": 226.73,
-      "grad_norm": 6.272486686706543,
+      "grad_norm": 8.756503105163574,
       "learning_rate": 2.4424424424424427e-06,
-      "loss": 0.8862,
+      "loss": 0.8751,
       "step": 75500
     },
     {
       "epoch": 226.76,
-      "grad_norm": 9.17968463897705,
+      "grad_norm": 8.645489692687988,
       "learning_rate": 2.4414414414414417e-06,
-      "loss": 0.922,
+      "loss": 0.9075,
       "step": 75510
     },
     {
       "epoch": 226.79,
-      "grad_norm": 4.510869026184082,
+      "grad_norm": 6.604963302612305,
       "learning_rate": 2.4404404404404407e-06,
-      "loss": 0.9933,
+      "loss": 0.9836,
       "step": 75520
     },
     {
       "epoch": 226.82,
-      "grad_norm": 7.652945518493652,
+      "grad_norm": 9.790445327758789,
       "learning_rate": 2.4394394394394398e-06,
-      "loss": 0.8713,
+      "loss": 0.8737,
       "step": 75530
     },
     {
       "epoch": 226.85,
-      "grad_norm": 6.472902774810791,
+      "grad_norm": 6.197530746459961,
       "learning_rate": 2.4384384384384383e-06,
-      "loss": 0.8979,
+      "loss": 0.8901,
       "step": 75540
     },
     {
       "epoch": 226.88,
-      "grad_norm": 7.25363826751709,
+      "grad_norm": 9.509366035461426,
       "learning_rate": 2.437437437437438e-06,
-      "loss": 0.9778,
+      "loss": 0.9729,
       "step": 75550
     },
     {
       "epoch": 226.91,
-      "grad_norm": 5.586451530456543,
+      "grad_norm": 9.229921340942383,
       "learning_rate": 2.4364364364364364e-06,
-      "loss": 0.9226,
+      "loss": 0.9233,
       "step": 75560
     },
     {
       "epoch": 226.94,
-      "grad_norm": 7.84766149520874,
+      "grad_norm": 7.970900535583496,
       "learning_rate": 2.435435435435436e-06,
-      "loss": 0.8997,
+      "loss": 0.9019,
       "step": 75570
     },
     {
       "epoch": 226.97,
-      "grad_norm": 4.675437927246094,
+      "grad_norm": 4.388794422149658,
       "learning_rate": 2.4344344344344344e-06,
-      "loss": 0.9253,
+      "loss": 0.9174,
       "step": 75580
     },
     {
       "epoch": 227.0,
-      "grad_norm": 9.89514446258545,
+      "grad_norm": 8.043439865112305,
       "learning_rate": 2.4334334334334334e-06,
-      "loss": 0.9263,
+      "loss": 0.9358,
       "step": 75590
     },
     {
       "epoch": 227.0,
-      "eval_accuracy": 0.8247,
-      "eval_loss": 0.5745235085487366,
-      "eval_runtime": 5.4168,
-      "eval_samples_per_second": 1846.093,
-      "eval_steps_per_second": 7.384,
+      "eval_accuracy": 0.8277,
+      "eval_loss": 0.5756272673606873,
+      "eval_runtime": 5.0339,
+      "eval_samples_per_second": 1986.545,
+      "eval_steps_per_second": 7.946,
       "step": 75591
     },
     {
       "epoch": 227.03,
-      "grad_norm": 5.761136054992676,
+      "grad_norm": 6.296458721160889,
       "learning_rate": 2.432432432432433e-06,
-      "loss": 1.0736,
+      "loss": 1.0569,
       "step": 75600
     },
     {
       "epoch": 227.06,
-      "grad_norm": 10.246672630310059,
+      "grad_norm": 8.608980178833008,
       "learning_rate": 2.4314314314314315e-06,
-      "loss": 0.8784,
+      "loss": 0.8844,
       "step": 75610
     },
     {
       "epoch": 227.09,
-      "grad_norm": 10.661499977111816,
+      "grad_norm": 9.069744110107422,
       "learning_rate": 2.430430430430431e-06,
-      "loss": 0.9651,
+      "loss": 0.9619,
       "step": 75620
     },
     {
       "epoch": 227.12,
-      "grad_norm": 4.933978080749512,
+      "grad_norm": 5.394620895385742,
       "learning_rate": 2.4294294294294295e-06,
-      "loss": 0.8989,
+      "loss": 0.8964,
       "step": 75630
     },
     {
       "epoch": 227.15,
-      "grad_norm": 6.667109489440918,
+      "grad_norm": 6.643839359283447,
       "learning_rate": 2.4284284284284285e-06,
-      "loss": 0.9027,
+      "loss": 0.9033,
       "step": 75640
     },
     {
       "epoch": 227.18,
-      "grad_norm": 9.535804748535156,
+      "grad_norm": 7.849951267242432,
       "learning_rate": 2.4274274274274276e-06,
-      "loss": 0.8899,
+      "loss": 0.8802,
       "step": 75650
     },
     {
       "epoch": 227.21,
-      "grad_norm": 8.364187240600586,
+      "grad_norm": 11.094429016113281,
       "learning_rate": 2.4264264264264266e-06,
-      "loss": 0.9111,
+      "loss": 0.9164,
       "step": 75660
     },
     {
       "epoch": 227.24,
-      "grad_norm": 10.704109191894531,
+      "grad_norm": 9.78442096710205,
       "learning_rate": 2.4254254254254256e-06,
-      "loss": 0.9289,
+      "loss": 0.9245,
       "step": 75670
     },
     {
       "epoch": 227.27,
-      "grad_norm": 5.118391990661621,
+      "grad_norm": 4.533906936645508,
       "learning_rate": 2.4244244244244246e-06,
-      "loss": 0.938,
+      "loss": 0.9326,
       "step": 75680
     },
     {
       "epoch": 227.3,
-      "grad_norm": 5.776532173156738,
+      "grad_norm": 6.335336685180664,
       "learning_rate": 2.4234234234234236e-06,
-      "loss": 0.9146,
+      "loss": 0.9121,
       "step": 75690
     },
     {
       "epoch": 227.33,
-      "grad_norm": 6.7468581199646,
+      "grad_norm": 6.8262224197387695,
       "learning_rate": 2.4224224224224226e-06,
-      "loss": 0.948,
+      "loss": 0.9422,
       "step": 75700
     },
     {
       "epoch": 227.36,
-      "grad_norm": 7.2680487632751465,
+      "grad_norm": 8.174407958984375,
       "learning_rate": 2.4214214214214217e-06,
-      "loss": 0.9563,
+      "loss": 0.9519,
       "step": 75710
     },
     {
       "epoch": 227.39,
-      "grad_norm": 9.05387020111084,
+      "grad_norm": 9.448970794677734,
       "learning_rate": 2.4204204204204207e-06,
-      "loss": 0.9786,
+      "loss": 0.9835,
       "step": 75720
     },
     {
       "epoch": 227.42,
-      "grad_norm": 6.191096782684326,
+      "grad_norm": 4.5160298347473145,
       "learning_rate": 2.4194194194194197e-06,
-      "loss": 0.9057,
+      "loss": 0.9069,
       "step": 75730
     },
     {
       "epoch": 227.45,
-      "grad_norm": 6.623605728149414,
+      "grad_norm": 7.615314960479736,
       "learning_rate": 2.4184184184184187e-06,
-      "loss": 0.9162,
+      "loss": 0.9097,
       "step": 75740
     },
     {
       "epoch": 227.48,
-      "grad_norm": 6.182514667510986,
+      "grad_norm": 10.369890213012695,
       "learning_rate": 2.4174174174174177e-06,
       "loss": 0.9503,
       "step": 75750
     },
     {
       "epoch": 227.51,
-      "grad_norm": 4.829371452331543,
+      "grad_norm": 7.514602184295654,
       "learning_rate": 2.4164164164164168e-06,
-      "loss": 0.8974,
+      "loss": 0.905,
       "step": 75760
     },
     {
       "epoch": 227.54,
-      "grad_norm": 4.592681407928467,
+      "grad_norm": 6.313732147216797,
       "learning_rate": 2.4154154154154153e-06,
-      "loss": 0.9817,
+      "loss": 0.9811,
       "step": 75770
     },
     {
       "epoch": 227.57,
-      "grad_norm": 7.007162094116211,
+      "grad_norm": 7.262298583984375,
       "learning_rate": 2.414414414414415e-06,
-      "loss": 0.9313,
+      "loss": 0.926,
       "step": 75780
     },
     {
       "epoch": 227.6,
-      "grad_norm": 8.652573585510254,
+      "grad_norm": 14.395648956298828,
       "learning_rate": 2.4134134134134134e-06,
-      "loss": 0.9161,
+      "loss": 0.9135,
       "step": 75790
     },
     {
       "epoch": 227.63,
-      "grad_norm": 6.652675628662109,
+      "grad_norm": 6.319643020629883,
       "learning_rate": 2.412412412412413e-06,
-      "loss": 0.9453,
+      "loss": 0.9448,
       "step": 75800
     },
     {
       "epoch": 227.66,
-      "grad_norm": 6.402853012084961,
+      "grad_norm": 8.00029182434082,
       "learning_rate": 2.4114114114114114e-06,
-      "loss": 0.9137,
+      "loss": 0.916,
       "step": 75810
     },
     {
       "epoch": 227.69,
-      "grad_norm": 6.473983287811279,
+      "grad_norm": 5.8164753913879395,
       "learning_rate": 2.4104104104104104e-06,
-      "loss": 0.9102,
+      "loss": 0.9104,
       "step": 75820
     },
     {
       "epoch": 227.72,
-      "grad_norm": 7.095130443572998,
+      "grad_norm": 6.314164638519287,
       "learning_rate": 2.4094094094094095e-06,
-      "loss": 0.9298,
+      "loss": 0.9257,
       "step": 75830
     },
     {
       "epoch": 227.75,
-      "grad_norm": 7.281118392944336,
+      "grad_norm": 9.453041076660156,
       "learning_rate": 2.4084084084084085e-06,
-      "loss": 0.9751,
+      "loss": 0.978,
       "step": 75840
     },
     {
       "epoch": 227.78,
-      "grad_norm": 4.101138114929199,
+      "grad_norm": 5.6125030517578125,
       "learning_rate": 2.4074074074074075e-06,
-      "loss": 0.9075,
+      "loss": 0.9028,
       "step": 75850
     },
     {
       "epoch": 227.81,
-      "grad_norm": 6.056380271911621,
+      "grad_norm": 7.5742316246032715,
       "learning_rate": 2.4064064064064065e-06,
-      "loss": 0.9165,
+      "loss": 0.9244,
       "step": 75860
     },
     {
       "epoch": 227.84,
-      "grad_norm": 6.82618522644043,
+      "grad_norm": 7.120845317840576,
       "learning_rate": 2.4054054054054055e-06,
-      "loss": 0.9391,
+      "loss": 0.9297,
       "step": 75870
     },
     {
       "epoch": 227.87,
-      "grad_norm": 5.371558666229248,
+      "grad_norm": 7.114173889160156,
       "learning_rate": 2.4044044044044045e-06,
-      "loss": 0.8318,
+      "loss": 0.8418,
       "step": 75880
     },
     {
       "epoch": 227.9,
-      "grad_norm": 6.211153507232666,
+      "grad_norm": 6.941005706787109,
       "learning_rate": 2.4034034034034036e-06,
-      "loss": 0.9519,
+      "loss": 0.949,
       "step": 75890
     },
     {
       "epoch": 227.93,
-      "grad_norm": 6.795780658721924,
+      "grad_norm": 7.249866962432861,
       "learning_rate": 2.4024024024024026e-06,
-      "loss": 0.8999,
+      "loss": 0.9002,
       "step": 75900
     },
     {
       "epoch": 227.96,
-      "grad_norm": 7.140031814575195,
+      "grad_norm": 8.431568145751953,
       "learning_rate": 2.4014014014014016e-06,
-      "loss": 0.8987,
+      "loss": 0.8937,
       "step": 75910
     },
     {
       "epoch": 227.99,
-      "grad_norm": 7.032166004180908,
+      "grad_norm": 6.918605804443359,
       "learning_rate": 2.4004004004004006e-06,
-      "loss": 0.9343,
+      "loss": 0.9334,
       "step": 75920
     },
     {
       "epoch": 228.0,
-      "eval_accuracy": 0.8273,
-      "eval_loss": 0.5735989212989807,
-      "eval_runtime": 5.2288,
-      "eval_samples_per_second": 1912.5,
-      "eval_steps_per_second": 7.65,
+      "eval_accuracy": 0.8269,
+      "eval_loss": 0.5752361416816711,
+      "eval_runtime": 5.6137,
+      "eval_samples_per_second": 1781.345,
+      "eval_steps_per_second": 7.125,
       "step": 75924
     },
     {
       "epoch": 228.02,
-      "grad_norm": 7.771353244781494,
+      "grad_norm": 7.724119663238525,
       "learning_rate": 2.3993993993993996e-06,
-      "loss": 1.1001,
+      "loss": 1.0773,
       "step": 75930
     },
     {
       "epoch": 228.05,
-      "grad_norm": 10.544669151306152,
+      "grad_norm": 9.2847261428833,
       "learning_rate": 2.3983983983983987e-06,
-      "loss": 1.0408,
+      "loss": 1.0585,
       "step": 75940
     },
     {
       "epoch": 228.08,
-      "grad_norm": 8.5191068649292,
+      "grad_norm": 11.56100082397461,
       "learning_rate": 2.3973973973973977e-06,
-      "loss": 0.9398,
+      "loss": 0.9427,
       "step": 75950
     },
     {
       "epoch": 228.11,
-      "grad_norm": 5.122549057006836,
+      "grad_norm": 5.055180072784424,
       "learning_rate": 2.3963963963963967e-06,
-      "loss": 0.9135,
+      "loss": 0.9073,
       "step": 75960
     },
     {
       "epoch": 228.14,
-      "grad_norm": 6.551916122436523,
+      "grad_norm": 6.8139495849609375,
       "learning_rate": 2.3953953953953957e-06,
-      "loss": 0.9421,
+      "loss": 0.9454,
       "step": 75970
     },
     {
       "epoch": 228.17,
-      "grad_norm": 7.992487907409668,
+      "grad_norm": 7.747501850128174,
       "learning_rate": 2.3943943943943947e-06,
-      "loss": 0.9847,
+      "loss": 0.9785,
       "step": 75980
     },
     {
       "epoch": 228.2,
-      "grad_norm": 5.00302267074585,
+      "grad_norm": 5.05192232131958,
       "learning_rate": 2.3933933933933938e-06,
-      "loss": 0.8786,
+      "loss": 0.8832,
       "step": 75990
     },
     {
       "epoch": 228.23,
-      "grad_norm": 11.692514419555664,
+      "grad_norm": 9.720171928405762,
       "learning_rate": 2.3923923923923923e-06,
-      "loss": 0.9238,
+      "loss": 0.9284,
       "step": 76000
     },
     {
       "epoch": 228.26,
-      "grad_norm": 6.821290969848633,
+      "grad_norm": 6.140380382537842,
       "learning_rate": 2.3913913913913918e-06,
-      "loss": 0.9543,
+      "loss": 0.9451,
       "step": 76010
     },
     {
       "epoch": 228.29,
-      "grad_norm": 6.957498073577881,
+      "grad_norm": 7.154791355133057,
       "learning_rate": 2.3903903903903904e-06,
-      "loss": 0.8686,
+      "loss": 0.8734,
       "step": 76020
     },
     {
       "epoch": 228.32,
-      "grad_norm": 6.518402576446533,
+      "grad_norm": 5.93328857421875,
       "learning_rate": 2.38938938938939e-06,
-      "loss": 0.9097,
+      "loss": 0.9058,
       "step": 76030
     },
     {
       "epoch": 228.35,
-      "grad_norm": 6.979775905609131,
+      "grad_norm": 6.772213459014893,
       "learning_rate": 2.3883883883883884e-06,
-      "loss": 0.9509,
+      "loss": 0.9482,
       "step": 76040
     },
     {
       "epoch": 228.38,
-      "grad_norm": 5.707918167114258,
+      "grad_norm": 6.074962615966797,
       "learning_rate": 2.3873873873873874e-06,
-      "loss": 0.8695,
+      "loss": 0.8618,
       "step": 76050
     },
     {
       "epoch": 228.41,
-      "grad_norm": 7.362729072570801,
+      "grad_norm": 7.345481872558594,
       "learning_rate": 2.3863863863863865e-06,
-      "loss": 0.9366,
+      "loss": 0.9482,
       "step": 76060
     },
     {
       "epoch": 228.44,
-      "grad_norm": 7.80725622177124,
+      "grad_norm": 7.142408847808838,
       "learning_rate": 2.3853853853853855e-06,
-      "loss": 0.9626,
+      "loss": 0.9569,
       "step": 76070
     },
     {
       "epoch": 228.47,
-      "grad_norm": 6.519937038421631,
+      "grad_norm": 9.798394203186035,
       "learning_rate": 2.3843843843843845e-06,
-      "loss": 0.9213,
+      "loss": 0.9193,
       "step": 76080
     },
     {
       "epoch": 228.5,
-      "grad_norm": 6.769364833831787,
+      "grad_norm": 12.395088195800781,
       "learning_rate": 2.3833833833833835e-06,
-      "loss": 0.9532,
+      "loss": 0.9569,
       "step": 76090
     },
     {
       "epoch": 228.53,
-      "grad_norm": 7.014290809631348,
+      "grad_norm": 7.017335891723633,
       "learning_rate": 2.3823823823823825e-06,
-      "loss": 0.884,
+      "loss": 0.8781,
       "step": 76100
     },
     {
       "epoch": 228.56,
-      "grad_norm": 5.3063435554504395,
+      "grad_norm": 7.550616264343262,
       "learning_rate": 2.3813813813813815e-06,
-      "loss": 0.9195,
+      "loss": 0.9083,
       "step": 76110
     },
     {
       "epoch": 228.59,
-      "grad_norm": 8.027153968811035,
+      "grad_norm": 7.592893123626709,
       "learning_rate": 2.3803803803803806e-06,
-      "loss": 0.988,
+      "loss": 0.9794,
       "step": 76120
     },
     {
       "epoch": 228.62,
-      "grad_norm": 6.46621561050415,
+      "grad_norm": 5.648293495178223,
       "learning_rate": 2.3793793793793796e-06,
-      "loss": 0.888,
+      "loss": 0.8887,
       "step": 76130
     },
     {
       "epoch": 228.65,
-      "grad_norm": 8.328484535217285,
+      "grad_norm": 8.970434188842773,
       "learning_rate": 2.3783783783783786e-06,
-      "loss": 0.9122,
+      "loss": 0.9001,
       "step": 76140
     },
     {
       "epoch": 228.68,
-      "grad_norm": 6.724803447723389,
+      "grad_norm": 6.495261192321777,
       "learning_rate": 2.3773773773773776e-06,
-      "loss": 0.9111,
+      "loss": 0.9053,
       "step": 76150
     },
     {
       "epoch": 228.71,
-      "grad_norm": 8.000367164611816,
+      "grad_norm": 7.2382917404174805,
       "learning_rate": 2.3763763763763766e-06,
-      "loss": 0.8928,
+      "loss": 0.8912,
       "step": 76160
     },
     {
       "epoch": 228.74,
-      "grad_norm": 7.3445940017700195,
+      "grad_norm": 9.314358711242676,
       "learning_rate": 2.3753753753753757e-06,
-      "loss": 0.916,
+      "loss": 0.9122,
       "step": 76170
     },
     {
       "epoch": 228.77,
-      "grad_norm": 8.973849296569824,
+      "grad_norm": 7.6770806312561035,
       "learning_rate": 2.3743743743743747e-06,
-      "loss": 0.9204,
+      "loss": 0.9186,
       "step": 76180
     },
     {
       "epoch": 228.8,
-      "grad_norm": 5.814183235168457,
+      "grad_norm": 6.0336408615112305,
       "learning_rate": 2.3733733733733737e-06,
-      "loss": 0.9153,
+      "loss": 0.8985,
       "step": 76190
     },
     {
       "epoch": 228.83,
-      "grad_norm": 7.004820346832275,
+      "grad_norm": 8.082853317260742,
       "learning_rate": 2.3723723723723727e-06,
-      "loss": 0.9994,
+      "loss": 1.0056,
       "step": 76200
     },
     {
       "epoch": 228.86,
-      "grad_norm": 7.29182767868042,
+      "grad_norm": 7.381521701812744,
       "learning_rate": 2.3713713713713717e-06,
-      "loss": 0.9596,
+      "loss": 0.9562,
       "step": 76210
     },
     {
       "epoch": 228.89,
-      "grad_norm": 5.670565128326416,
+      "grad_norm": 8.469666481018066,
       "learning_rate": 2.3703703703703707e-06,
-      "loss": 0.9403,
+      "loss": 0.9295,
       "step": 76220
     },
     {
       "epoch": 228.92,
-      "grad_norm": 7.476768970489502,
+      "grad_norm": 6.109925746917725,
       "learning_rate": 2.3693693693693693e-06,
-      "loss": 0.9347,
+      "loss": 0.9245,
       "step": 76230
     },
     {
       "epoch": 228.95,
-      "grad_norm": 4.61652135848999,
+      "grad_norm": 5.529149532318115,
       "learning_rate": 2.3683683683683688e-06,
-      "loss": 0.8314,
+      "loss": 0.8317,
       "step": 76240
     },
     {
       "epoch": 228.98,
-      "grad_norm": 5.36049222946167,
+      "grad_norm": 6.73313045501709,
       "learning_rate": 2.3673673673673674e-06,
-      "loss": 0.8955,
+      "loss": 0.9001,
       "step": 76250
     },
     {
       "epoch": 229.0,
-      "eval_accuracy": 0.8248,
-      "eval_loss": 0.5760202407836914,
-      "eval_runtime": 5.4146,
-      "eval_samples_per_second": 1846.855,
-      "eval_steps_per_second": 7.387,
+      "eval_accuracy": 0.829,
+      "eval_loss": 0.577608048915863,
+      "eval_runtime": 5.3976,
+      "eval_samples_per_second": 1852.659,
+      "eval_steps_per_second": 7.411,
       "step": 76257
     },
     {
       "epoch": 229.01,
-      "grad_norm": 6.600915431976318,
+      "grad_norm": 6.474968433380127,
       "learning_rate": 2.366366366366367e-06,
-      "loss": 1.1412,
+      "loss": 1.1297,
       "step": 76260
     },
     {
       "epoch": 229.04,
-      "grad_norm": 9.119160652160645,
+      "grad_norm": 7.808255672454834,
       "learning_rate": 2.3653653653653654e-06,
-      "loss": 0.8835,
+      "loss": 0.8857,
       "step": 76270
     },
     {
       "epoch": 229.07,
-      "grad_norm": 9.377240180969238,
+      "grad_norm": 8.710619926452637,
       "learning_rate": 2.3643643643643644e-06,
-      "loss": 0.916,
+      "loss": 0.9199,
       "step": 76280
     },
     {
       "epoch": 229.1,
-      "grad_norm": 6.251333713531494,
+      "grad_norm": 6.52146577835083,
       "learning_rate": 2.3633633633633635e-06,
-      "loss": 0.8858,
+      "loss": 0.8749,
       "step": 76290
     },
     {
       "epoch": 229.13,
-      "grad_norm": 5.690656661987305,
+      "grad_norm": 5.04657506942749,
       "learning_rate": 2.3623623623623625e-06,
-      "loss": 0.9224,
+      "loss": 0.9204,
       "step": 76300
     },
     {
       "epoch": 229.16,
-      "grad_norm": 5.893414497375488,
+      "grad_norm": 6.898827075958252,
       "learning_rate": 2.3613613613613615e-06,
-      "loss": 0.8312,
+      "loss": 0.8323,
       "step": 76310
     },
     {
       "epoch": 229.19,
-      "grad_norm": 8.203968048095703,
+      "grad_norm": 8.50917911529541,
       "learning_rate": 2.3603603603603605e-06,
-      "loss": 0.9183,
+      "loss": 0.9121,
       "step": 76320
     },
     {
       "epoch": 229.22,
-      "grad_norm": 7.264726161956787,
+      "grad_norm": 7.442723274230957,
       "learning_rate": 2.3593593593593595e-06,
-      "loss": 0.9378,
+      "loss": 0.9413,
       "step": 76330
     },
     {
       "epoch": 229.25,
-      "grad_norm": 4.894717216491699,
+      "grad_norm": 7.395277500152588,
       "learning_rate": 2.3583583583583585e-06,
-      "loss": 0.9376,
+      "loss": 0.944,
       "step": 76340
     },
     {
       "epoch": 229.28,
-      "grad_norm": 6.476467132568359,
+      "grad_norm": 7.738025665283203,
       "learning_rate": 2.3573573573573576e-06,
-      "loss": 0.9361,
+      "loss": 0.9265,
       "step": 76350
     },
     {
       "epoch": 229.31,
-      "grad_norm": 6.739269256591797,
+      "grad_norm": 6.630418300628662,
       "learning_rate": 2.3563563563563566e-06,
-      "loss": 0.8702,
+      "loss": 0.8752,
       "step": 76360
     },
     {
       "epoch": 229.34,
-      "grad_norm": 5.532363414764404,
+      "grad_norm": 6.684784412384033,
       "learning_rate": 2.3553553553553556e-06,
-      "loss": 0.9379,
+      "loss": 0.9366,
       "step": 76370
     },
     {
       "epoch": 229.37,
-      "grad_norm": 5.163920879364014,
+      "grad_norm": 7.033756256103516,
       "learning_rate": 2.3543543543543546e-06,
-      "loss": 0.9611,
+      "loss": 0.9579,
       "step": 76380
     },
     {
       "epoch": 229.4,
-      "grad_norm": 6.418565273284912,
+      "grad_norm": 6.494988918304443,
       "learning_rate": 2.3533533533533532e-06,
-      "loss": 0.8852,
+      "loss": 0.8727,
       "step": 76390
     },
     {
       "epoch": 229.43,
-      "grad_norm": 6.363382339477539,
+      "grad_norm": 7.0168914794921875,
       "learning_rate": 2.3523523523523527e-06,
-      "loss": 0.9325,
+      "loss": 0.9276,
       "step": 76400
     },
     {
       "epoch": 229.46,
-      "grad_norm": 7.332528114318848,
+      "grad_norm": 6.139852046966553,
       "learning_rate": 2.3513513513513517e-06,
-      "loss": 0.8902,
+      "loss": 0.8924,
       "step": 76410
     },
     {
       "epoch": 229.49,
-      "grad_norm": 10.936832427978516,
+      "grad_norm": 9.174834251403809,
       "learning_rate": 2.3503503503503507e-06,
-      "loss": 0.8866,
+      "loss": 0.8804,
       "step": 76420
     },
     {
       "epoch": 229.52,
-      "grad_norm": 6.084210395812988,
+      "grad_norm": 5.707581520080566,
       "learning_rate": 2.3493493493493497e-06,
-      "loss": 0.8811,
+      "loss": 0.8711,
       "step": 76430
     },
     {
       "epoch": 229.55,
-      "grad_norm": 5.861665725708008,
+      "grad_norm": 6.332797050476074,
       "learning_rate": 2.3483483483483483e-06,
-      "loss": 0.916,
+      "loss": 0.9172,
       "step": 76440
     },
     {
       "epoch": 229.58,
-      "grad_norm": 5.921432018280029,
+      "grad_norm": 7.120185375213623,
       "learning_rate": 2.3473473473473477e-06,
-      "loss": 0.9626,
+      "loss": 0.9575,
       "step": 76450
     },
     {
       "epoch": 229.61,
-      "grad_norm": 6.361980438232422,
+      "grad_norm": 6.4954118728637695,
       "learning_rate": 2.3463463463463463e-06,
-      "loss": 0.9067,
+      "loss": 0.9195,
       "step": 76460
     },
     {
       "epoch": 229.64,
-      "grad_norm": 7.009181499481201,
+      "grad_norm": 8.000288963317871,
       "learning_rate": 2.3453453453453458e-06,
-      "loss": 0.9132,
+      "loss": 0.9195,
       "step": 76470
     },
     {
       "epoch": 229.67,
-      "grad_norm": 5.66736364364624,
+      "grad_norm": 6.153746604919434,
       "learning_rate": 2.3443443443443444e-06,
-      "loss": 0.9173,
+      "loss": 0.9207,
       "step": 76480
     },
     {
       "epoch": 229.7,
-      "grad_norm": 7.689339637756348,
+      "grad_norm": 5.382575035095215,
       "learning_rate": 2.343343343343344e-06,
-      "loss": 0.9855,
+      "loss": 0.9917,
       "step": 76490
     },
     {
       "epoch": 229.73,
-      "grad_norm": 6.6281418800354,
+      "grad_norm": 7.345523357391357,
       "learning_rate": 2.3423423423423424e-06,
-      "loss": 0.9076,
+      "loss": 0.9115,
       "step": 76500
     },
     {
       "epoch": 229.76,
-      "grad_norm": 5.96934175491333,
+      "grad_norm": 8.954519271850586,
       "learning_rate": 2.3413413413413414e-06,
-      "loss": 0.8912,
+      "loss": 0.8953,
       "step": 76510
     },
     {
       "epoch": 229.79,
-      "grad_norm": 5.885605812072754,
+      "grad_norm": 6.34539794921875,
       "learning_rate": 2.3403403403403405e-06,
-      "loss": 0.9038,
+      "loss": 0.9044,
       "step": 76520
     },
     {
       "epoch": 229.82,
-      "grad_norm": 7.083038806915283,
+      "grad_norm": 6.804876327514648,
       "learning_rate": 2.3393393393393395e-06,
-      "loss": 0.8939,
+      "loss": 0.8847,
       "step": 76530
     },
     {
       "epoch": 229.85,
-      "grad_norm": 4.7936506271362305,
+      "grad_norm": 5.981973171234131,
       "learning_rate": 2.3383383383383385e-06,
-      "loss": 0.9365,
+      "loss": 0.9265,
       "step": 76540
     },
     {
       "epoch": 229.88,
-      "grad_norm": 13.187565803527832,
+      "grad_norm": 10.301629066467285,
       "learning_rate": 2.3373373373373375e-06,
-      "loss": 0.9178,
+      "loss": 0.9112,
       "step": 76550
     },
     {
       "epoch": 229.91,
-      "grad_norm": 6.157358646392822,
+      "grad_norm": 5.630910396575928,
       "learning_rate": 2.3363363363363365e-06,
-      "loss": 0.8551,
+      "loss": 0.8566,
       "step": 76560
     },
     {
       "epoch": 229.94,
-      "grad_norm": 8.309229850769043,
+      "grad_norm": 8.451387405395508,
       "learning_rate": 2.3353353353353355e-06,
-      "loss": 0.9434,
+      "loss": 0.9498,
       "step": 76570
     },
     {
       "epoch": 229.97,
-      "grad_norm": 4.460539817810059,
+      "grad_norm": 6.459830284118652,
       "learning_rate": 2.3343343343343346e-06,
-      "loss": 0.9731,
+      "loss": 0.9657,
       "step": 76580
     },
     {
       "epoch": 230.0,
-      "grad_norm": 46.52259063720703,
+      "grad_norm": 32.80706787109375,
       "learning_rate": 2.3333333333333336e-06,
-      "loss": 1.063,
+      "loss": 1.0413,
       "step": 76590
     },
     {
       "epoch": 230.0,
-      "eval_accuracy": 0.8259,
-      "eval_loss": 0.5766270160675049,
-      "eval_runtime": 5.3249,
-      "eval_samples_per_second": 1877.962,
-      "eval_steps_per_second": 7.512,
+      "eval_accuracy": 0.827,
+      "eval_loss": 0.5761530995368958,
+      "eval_runtime": 5.4779,
+      "eval_samples_per_second": 1825.514,
+      "eval_steps_per_second": 7.302,
       "step": 76590
     },
     {
       "epoch": 230.03,
-      "grad_norm": 5.230226039886475,
+      "grad_norm": 5.456832408905029,
       "learning_rate": 2.3323323323323326e-06,
-      "loss": 0.922,
+      "loss": 0.9314,
       "step": 76600
     },
     {
       "epoch": 230.06,
-      "grad_norm": 6.307478427886963,
+      "grad_norm": 4.610037326812744,
       "learning_rate": 2.3313313313313316e-06,
-      "loss": 0.8436,
+      "loss": 0.8515,
       "step": 76610
     },
     {
       "epoch": 230.09,
-      "grad_norm": 8.62086009979248,
+      "grad_norm": 6.529040813446045,
       "learning_rate": 2.3303303303303302e-06,
-      "loss": 0.8907,
+      "loss": 0.8858,
       "step": 76620
     },
     {
       "epoch": 230.12,
-      "grad_norm": 8.375288009643555,
+      "grad_norm": 6.795773029327393,
       "learning_rate": 2.3293293293293297e-06,
-      "loss": 0.9752,
+      "loss": 0.9805,
       "step": 76630
     },
     {
       "epoch": 230.15,
-      "grad_norm": 6.865284442901611,
+      "grad_norm": 8.444944381713867,
       "learning_rate": 2.3283283283283282e-06,
-      "loss": 0.9222,
+      "loss": 0.9146,
       "step": 76640
     },
     {
       "epoch": 230.18,
-      "grad_norm": 6.6450934410095215,
+      "grad_norm": 5.8272175788879395,
       "learning_rate": 2.3273273273273277e-06,
-      "loss": 0.8933,
+      "loss": 0.8906,
       "step": 76650
     },
     {
       "epoch": 230.21,
-      "grad_norm": 5.517233848571777,
+      "grad_norm": 5.921377182006836,
       "learning_rate": 2.3263263263263263e-06,
-      "loss": 0.9472,
+      "loss": 0.9492,
       "step": 76660
     },
     {
       "epoch": 230.24,
-      "grad_norm": 6.831721305847168,
+      "grad_norm": 7.806619167327881,
       "learning_rate": 2.3253253253253253e-06,
-      "loss": 0.9066,
+      "loss": 0.9006,
       "step": 76670
     },
     {
       "epoch": 230.27,
-      "grad_norm": 6.299792766571045,
+      "grad_norm": 6.682567119598389,
       "learning_rate": 2.3243243243243247e-06,
-      "loss": 0.8512,
+      "loss": 0.8435,
       "step": 76680
     },
     {
       "epoch": 230.3,
-      "grad_norm": 6.988855838775635,
+      "grad_norm": 7.086954593658447,
       "learning_rate": 2.3233233233233233e-06,
-      "loss": 0.8779,
+      "loss": 0.8633,
       "step": 76690
     },
     {
       "epoch": 230.33,
-      "grad_norm": 5.716799736022949,
+      "grad_norm": 6.526058197021484,
       "learning_rate": 2.3223223223223228e-06,
-      "loss": 0.9276,
+      "loss": 0.9307,
       "step": 76700
     },
     {
       "epoch": 230.36,
-      "grad_norm": 8.07087516784668,
+      "grad_norm": 8.466130256652832,
       "learning_rate": 2.3213213213213214e-06,
-      "loss": 0.9182,
+      "loss": 0.9105,
       "step": 76710
     },
     {
       "epoch": 230.39,
-      "grad_norm": 6.648508548736572,
+      "grad_norm": 6.896261692047119,
       "learning_rate": 2.320320320320321e-06,
-      "loss": 0.9341,
+      "loss": 0.9283,
       "step": 76720
     },
     {
       "epoch": 230.42,
-      "grad_norm": 4.999793529510498,
+      "grad_norm": 4.812035083770752,
       "learning_rate": 2.3193193193193194e-06,
-      "loss": 0.9461,
+      "loss": 0.9327,
       "step": 76730
     },
     {
       "epoch": 230.45,
-      "grad_norm": 7.556336879730225,
+      "grad_norm": 5.997191905975342,
       "learning_rate": 2.3183183183183184e-06,
-      "loss": 0.9202,
+      "loss": 0.9155,
       "step": 76740
     },
     {
       "epoch": 230.48,
-      "grad_norm": 9.299432754516602,
+      "grad_norm": 8.33647632598877,
       "learning_rate": 2.3173173173173175e-06,
-      "loss": 0.8707,
+      "loss": 0.8655,
       "step": 76750
     },
     {
       "epoch": 230.51,
-      "grad_norm": 8.261966705322266,
+      "grad_norm": 5.728392124176025,
       "learning_rate": 2.3163163163163165e-06,
-      "loss": 0.8786,
+      "loss": 0.8788,
       "step": 76760
     },
     {
       "epoch": 230.54,
-      "grad_norm": 6.134219169616699,
+      "grad_norm": 6.32240629196167,
       "learning_rate": 2.3153153153153155e-06,
-      "loss": 0.9095,
+      "loss": 0.8992,
       "step": 76770
     },
     {
       "epoch": 230.57,
-      "grad_norm": 6.2185139656066895,
+      "grad_norm": 7.251842021942139,
       "learning_rate": 2.3143143143143145e-06,
-      "loss": 0.9276,
+      "loss": 0.9083,
       "step": 76780
     },
     {
       "epoch": 230.6,
-      "grad_norm": 8.914837837219238,
+      "grad_norm": 9.217578887939453,
       "learning_rate": 2.3133133133133135e-06,
-      "loss": 0.9279,
+      "loss": 0.9153,
       "step": 76790
     },
     {
       "epoch": 230.63,
-      "grad_norm": 8.568059921264648,
+      "grad_norm": 9.557445526123047,
       "learning_rate": 2.3123123123123125e-06,
-      "loss": 0.9024,
+      "loss": 0.9053,
       "step": 76800
     },
     {
       "epoch": 230.66,
-      "grad_norm": 6.224318981170654,
+      "grad_norm": 5.354879856109619,
       "learning_rate": 2.3113113113113116e-06,
-      "loss": 0.933,
+      "loss": 0.9304,
       "step": 76810
     },
     {
       "epoch": 230.69,
-      "grad_norm": 7.373044967651367,
+      "grad_norm": 8.576372146606445,
       "learning_rate": 2.3103103103103106e-06,
-      "loss": 0.9644,
+      "loss": 0.9555,
       "step": 76820
     },
     {
       "epoch": 230.72,
-      "grad_norm": 6.902327537536621,
+      "grad_norm": 6.624667644500732,
       "learning_rate": 2.3093093093093096e-06,
-      "loss": 0.9573,
+      "loss": 0.9584,
       "step": 76830
     },
     {
       "epoch": 230.75,
-      "grad_norm": 6.070363521575928,
+      "grad_norm": 6.799314975738525,
       "learning_rate": 2.3083083083083086e-06,
-      "loss": 1.0095,
+      "loss": 0.995,
       "step": 76840
     },
     {
       "epoch": 230.78,
-      "grad_norm": 5.999297618865967,
+      "grad_norm": 5.044459819793701,
       "learning_rate": 2.307307307307307e-06,
-      "loss": 0.9306,
+      "loss": 0.9269,
       "step": 76850
     },
     {
       "epoch": 230.81,
-      "grad_norm": 6.289979457855225,
+      "grad_norm": 7.232490062713623,
       "learning_rate": 2.3063063063063067e-06,
-      "loss": 0.9919,
+      "loss": 0.9844,
       "step": 76860
     },
     {
       "epoch": 230.84,
-      "grad_norm": 6.4955363273620605,
+      "grad_norm": 8.536572456359863,
       "learning_rate": 2.3053053053053052e-06,
-      "loss": 0.9572,
+      "loss": 0.9545,
       "step": 76870
     },
     {
       "epoch": 230.87,
-      "grad_norm": 8.098984718322754,
+      "grad_norm": 10.488628387451172,
       "learning_rate": 2.3043043043043047e-06,
-      "loss": 0.95,
+      "loss": 0.9599,
       "step": 76880
     },
     {
       "epoch": 230.9,
-      "grad_norm": 5.4253621101379395,
+      "grad_norm": 10.695523262023926,
       "learning_rate": 2.3033033033033033e-06,
-      "loss": 0.9443,
+      "loss": 0.9528,
       "step": 76890
     },
     {
       "epoch": 230.93,
-      "grad_norm": 8.445652961730957,
+      "grad_norm": 6.93255615234375,
       "learning_rate": 2.3023023023023023e-06,
-      "loss": 0.9738,
+      "loss": 0.9712,
       "step": 76900
     },
     {
       "epoch": 230.96,
-      "grad_norm": 6.29515266418457,
+      "grad_norm": 5.909913063049316,
       "learning_rate": 2.3013013013013013e-06,
-      "loss": 0.8696,
+      "loss": 0.8828,
       "step": 76910
     },
     {
       "epoch": 230.99,
-      "grad_norm": 6.898852348327637,
+      "grad_norm": 5.260955810546875,
       "learning_rate": 2.3003003003003003e-06,
-      "loss": 0.9331,
+      "loss": 0.9277,
       "step": 76920
     },
     {
       "epoch": 231.0,
-      "eval_accuracy": 0.826,
-      "eval_loss": 0.5766454339027405,
-      "eval_runtime": 5.679,
-      "eval_samples_per_second": 1760.866,
-      "eval_steps_per_second": 7.043,
+      "eval_accuracy": 0.8292,
+      "eval_loss": 0.5746156573295593,
+      "eval_runtime": 5.3984,
+      "eval_samples_per_second": 1852.403,
+      "eval_steps_per_second": 7.41,
       "step": 76923
     },
     {
       "epoch": 231.02,
-      "grad_norm": 6.621943950653076,
+      "grad_norm": 9.60738754272461,
       "learning_rate": 2.2992992992992998e-06,
-      "loss": 1.0055,
+      "loss": 1.014,
       "step": 76930
     },
     {
       "epoch": 231.05,
-      "grad_norm": 7.153619289398193,
+      "grad_norm": 7.381382942199707,
       "learning_rate": 2.2982982982982984e-06,
-      "loss": 0.8866,
+      "loss": 0.8856,
       "step": 76940
     },
     {
       "epoch": 231.08,
-      "grad_norm": 4.913899898529053,
+      "grad_norm": 6.167375087738037,
       "learning_rate": 2.297297297297298e-06,
-      "loss": 0.9217,
+      "loss": 0.9157,
       "step": 76950
     },
     {
       "epoch": 231.11,
-      "grad_norm": 8.148483276367188,
+      "grad_norm": 6.905209541320801,
       "learning_rate": 2.2962962962962964e-06,
-      "loss": 0.8913,
+      "loss": 0.8842,
       "step": 76960
     },
     {
       "epoch": 231.14,
-      "grad_norm": 5.847254276275635,
+      "grad_norm": 7.431135654449463,
       "learning_rate": 2.2952952952952954e-06,
-      "loss": 0.8797,
+      "loss": 0.8799,
       "step": 76970
     },
     {
       "epoch": 231.17,
-      "grad_norm": 6.169511795043945,
+      "grad_norm": 8.127669334411621,
       "learning_rate": 2.2942942942942944e-06,
-      "loss": 0.9149,
+      "loss": 0.8968,
       "step": 76980
     },
     {
       "epoch": 231.2,
-      "grad_norm": 6.041130065917969,
+      "grad_norm": 6.199949741363525,
       "learning_rate": 2.2932932932932935e-06,
-      "loss": 0.9084,
+      "loss": 0.9086,
       "step": 76990
     },
     {
       "epoch": 231.23,
-      "grad_norm": 7.670273303985596,
+      "grad_norm": 6.7686004638671875,
       "learning_rate": 2.2922922922922925e-06,
-      "loss": 0.9604,
+      "loss": 0.9598,
       "step": 77000
     },
     {
       "epoch": 231.26,
-      "grad_norm": 7.385842800140381,
+      "grad_norm": 8.024151802062988,
       "learning_rate": 2.2912912912912915e-06,
-      "loss": 0.9313,
+      "loss": 0.9328,
       "step": 77010
     },
     {
       "epoch": 231.29,
-      "grad_norm": 4.794745445251465,
+      "grad_norm": 5.160857677459717,
       "learning_rate": 2.2902902902902905e-06,
-      "loss": 0.902,
+      "loss": 0.9011,
       "step": 77020
     },
     {
       "epoch": 231.32,
-      "grad_norm": 7.012855529785156,
+      "grad_norm": 6.236949443817139,
       "learning_rate": 2.2892892892892895e-06,
-      "loss": 0.9048,
+      "loss": 0.9091,
       "step": 77030
     },
     {
       "epoch": 231.35,
-      "grad_norm": 6.710078239440918,
+      "grad_norm": 5.944578647613525,
       "learning_rate": 2.2882882882882886e-06,
-      "loss": 0.9266,
+      "loss": 0.9177,
       "step": 77040
     },
     {
       "epoch": 231.38,
-      "grad_norm": 8.88840103149414,
+      "grad_norm": 12.785058975219727,
       "learning_rate": 2.2872872872872876e-06,
-      "loss": 0.9398,
+      "loss": 0.9329,
       "step": 77050
     },
     {
       "epoch": 231.41,
-      "grad_norm": 6.448197841644287,
+      "grad_norm": 7.243312835693359,
       "learning_rate": 2.2862862862862866e-06,
-      "loss": 0.8683,
+      "loss": 0.8692,
       "step": 77060
     },
     {
       "epoch": 231.44,
-      "grad_norm": 5.382036209106445,
+      "grad_norm": 5.983648777008057,
       "learning_rate": 2.2852852852852856e-06,
-      "loss": 0.9264,
+      "loss": 0.9189,
       "step": 77070
     },
     {
       "epoch": 231.47,
-      "grad_norm": 9.522404670715332,
+      "grad_norm": 7.615947723388672,
       "learning_rate": 2.284284284284284e-06,
-      "loss": 0.8216,
+      "loss": 0.8327,
       "step": 77080
     },
     {
       "epoch": 231.5,
-      "grad_norm": 6.701741695404053,
+      "grad_norm": 5.978594779968262,
       "learning_rate": 2.2832832832832837e-06,
-      "loss": 0.8474,
+      "loss": 0.8457,
       "step": 77090
     },
     {
       "epoch": 231.53,
-      "grad_norm": 7.147876739501953,
+      "grad_norm": 8.174367904663086,
       "learning_rate": 2.2822822822822822e-06,
-      "loss": 0.96,
+      "loss": 0.962,
       "step": 77100
     },
     {
       "epoch": 231.56,
-      "grad_norm": 6.132070064544678,
+      "grad_norm": 6.135376930236816,
       "learning_rate": 2.2812812812812817e-06,
-      "loss": 0.9784,
+      "loss": 0.9694,
       "step": 77110
     },
     {
       "epoch": 231.59,
-      "grad_norm": 4.843452453613281,
+      "grad_norm": 6.718967437744141,
       "learning_rate": 2.2802802802802803e-06,
-      "loss": 0.9137,
+      "loss": 0.903,
       "step": 77120
     },
     {
       "epoch": 231.62,
-      "grad_norm": 6.572041034698486,
+      "grad_norm": 6.273033142089844,
       "learning_rate": 2.2792792792792793e-06,
-      "loss": 0.9023,
+      "loss": 0.9073,
       "step": 77130
     },
     {
       "epoch": 231.65,
-      "grad_norm": 6.773430824279785,
+      "grad_norm": 8.302186965942383,
       "learning_rate": 2.2782782782782783e-06,
-      "loss": 0.8526,
+      "loss": 0.8395,
       "step": 77140
     },
     {
       "epoch": 231.68,
-      "grad_norm": 6.416208267211914,
+      "grad_norm": 6.822917938232422,
       "learning_rate": 2.2772772772772773e-06,
-      "loss": 0.8621,
+      "loss": 0.8704,
       "step": 77150
     },
     {
       "epoch": 231.71,
-      "grad_norm": 7.583805084228516,
+      "grad_norm": 7.169437885284424,
       "learning_rate": 2.2762762762762764e-06,
-      "loss": 0.8934,
+      "loss": 0.8897,
       "step": 77160
     },
     {
       "epoch": 231.74,
-      "grad_norm": 7.194533348083496,
+      "grad_norm": 6.1197404861450195,
       "learning_rate": 2.2752752752752754e-06,
-      "loss": 0.9512,
+      "loss": 0.9584,
       "step": 77170
     },
     {
       "epoch": 231.77,
-      "grad_norm": 8.224298477172852,
+      "grad_norm": 7.396502494812012,
       "learning_rate": 2.2742742742742744e-06,
-      "loss": 0.9793,
+      "loss": 0.9784,
       "step": 77180
     },
     {
       "epoch": 231.8,
-      "grad_norm": 9.804091453552246,
+      "grad_norm": 10.200528144836426,
       "learning_rate": 2.2732732732732734e-06,
-      "loss": 0.9722,
+      "loss": 0.9701,
       "step": 77190
     },
     {
       "epoch": 231.83,
-      "grad_norm": 5.952477931976318,
+      "grad_norm": 6.55221700668335,
       "learning_rate": 2.2722722722722724e-06,
-      "loss": 0.8778,
+      "loss": 0.8676,
       "step": 77200
     },
     {
       "epoch": 231.86,
-      "grad_norm": 4.96974515914917,
+      "grad_norm": 5.17174768447876,
       "learning_rate": 2.2712712712712714e-06,
-      "loss": 0.9375,
+      "loss": 0.9371,
       "step": 77210
     },
     {
       "epoch": 231.89,
-      "grad_norm": 5.971379280090332,
+      "grad_norm": 6.141185283660889,
       "learning_rate": 2.2702702702702705e-06,
-      "loss": 0.8975,
+      "loss": 0.9005,
       "step": 77220
     },
     {
       "epoch": 231.92,
-      "grad_norm": 8.156781196594238,
+      "grad_norm": 10.473756790161133,
       "learning_rate": 2.2692692692692695e-06,
-      "loss": 0.9956,
+      "loss": 0.9948,
       "step": 77230
     },
     {
       "epoch": 231.95,
-      "grad_norm": 7.251339435577393,
+      "grad_norm": 7.121708393096924,
       "learning_rate": 2.2682682682682685e-06,
-      "loss": 0.9232,
+      "loss": 0.9212,
       "step": 77240
     },
     {
       "epoch": 231.98,
-      "grad_norm": 8.241951942443848,
+      "grad_norm": 6.960716247558594,
       "learning_rate": 2.2672672672672675e-06,
-      "loss": 0.9409,
+      "loss": 0.938,
       "step": 77250
     },
     {
       "epoch": 232.0,
-      "eval_accuracy": 0.8242,
-      "eval_loss": 0.582554042339325,
-      "eval_runtime": 5.3152,
-      "eval_samples_per_second": 1881.404,
-      "eval_steps_per_second": 7.526,
+      "eval_accuracy": 0.8263,
+      "eval_loss": 0.5817124247550964,
+      "eval_runtime": 5.1503,
+      "eval_samples_per_second": 1941.636,
+      "eval_steps_per_second": 7.767,
       "step": 77256
     },
     {
       "epoch": 232.01,
-      "grad_norm": 6.776363372802734,
+      "grad_norm": 8.135272979736328,
       "learning_rate": 2.2662662662662665e-06,
-      "loss": 1.2593,
+      "loss": 1.2494,
       "step": 77260
     },
     {
       "epoch": 232.04,
-      "grad_norm": 8.710539817810059,
+      "grad_norm": 7.235963344573975,
       "learning_rate": 2.2652652652652656e-06,
-      "loss": 0.9395,
+      "loss": 0.9496,
       "step": 77270
     },
     {
       "epoch": 232.07,
-      "grad_norm": 8.385071754455566,
+      "grad_norm": 8.216541290283203,
       "learning_rate": 2.2642642642642646e-06,
-      "loss": 0.9955,
+      "loss": 1.0035,
       "step": 77280
     },
     {
       "epoch": 232.1,
-      "grad_norm": 5.6491217613220215,
+      "grad_norm": 7.051952362060547,
       "learning_rate": 2.2632632632632636e-06,
-      "loss": 0.8869,
+      "loss": 0.8936,
       "step": 77290
     },
     {
       "epoch": 232.13,
-      "grad_norm": 7.829229831695557,
+      "grad_norm": 9.10755443572998,
       "learning_rate": 2.2622622622622626e-06,
-      "loss": 0.9688,
+      "loss": 0.9721,
       "step": 77300
     },
     {
       "epoch": 232.16,
-      "grad_norm": 7.885899066925049,
+      "grad_norm": 6.149399757385254,
       "learning_rate": 2.261261261261261e-06,
-      "loss": 0.9044,
+      "loss": 0.9085,
       "step": 77310
     },
     {
       "epoch": 232.19,
-      "grad_norm": 5.953463554382324,
+      "grad_norm": 5.494118690490723,
       "learning_rate": 2.2602602602602606e-06,
-      "loss": 0.891,
+      "loss": 0.8814,
       "step": 77320
     },
     {
       "epoch": 232.22,
-      "grad_norm": 8.589982032775879,
+      "grad_norm": 10.153294563293457,
       "learning_rate": 2.2592592592592592e-06,
-      "loss": 0.9343,
+      "loss": 0.9322,
       "step": 77330
     },
     {
       "epoch": 232.25,
-      "grad_norm": 6.253246307373047,
+      "grad_norm": 7.586155414581299,
       "learning_rate": 2.2582582582582587e-06,
-      "loss": 0.901,
+      "loss": 0.8915,
       "step": 77340
     },
     {
       "epoch": 232.28,
-      "grad_norm": 7.078223705291748,
+      "grad_norm": 6.284738063812256,
       "learning_rate": 2.2572572572572573e-06,
-      "loss": 0.9306,
+      "loss": 0.9275,
       "step": 77350
     },
     {
       "epoch": 232.31,
-      "grad_norm": 6.360637664794922,
+      "grad_norm": 7.63737678527832,
       "learning_rate": 2.2562562562562563e-06,
-      "loss": 1.0108,
+      "loss": 0.9978,
       "step": 77360
     },
     {
       "epoch": 232.34,
-      "grad_norm": 5.329776287078857,
+      "grad_norm": 7.5266523361206055,
       "learning_rate": 2.2552552552552553e-06,
-      "loss": 0.9127,
+      "loss": 0.906,
       "step": 77370
     },
     {
       "epoch": 232.37,
-      "grad_norm": 5.968944549560547,
+      "grad_norm": 8.597719192504883,
       "learning_rate": 2.2542542542542543e-06,
-      "loss": 0.9577,
+      "loss": 0.9507,
       "step": 77380
     },
     {
       "epoch": 232.4,
-      "grad_norm": 6.155536651611328,
+      "grad_norm": 6.987178802490234,
       "learning_rate": 2.2532532532532534e-06,
-      "loss": 0.8758,
+      "loss": 0.8714,
       "step": 77390
     },
     {
       "epoch": 232.43,
-      "grad_norm": 6.4144768714904785,
+      "grad_norm": 7.51460075378418,
       "learning_rate": 2.2522522522522524e-06,
-      "loss": 0.8921,
+      "loss": 0.8979,
       "step": 77400
     },
     {
       "epoch": 232.46,
-      "grad_norm": 7.929412364959717,
+      "grad_norm": 8.164068222045898,
       "learning_rate": 2.2512512512512514e-06,
-      "loss": 0.9664,
+      "loss": 0.9592,
       "step": 77410
     },
     {
       "epoch": 232.49,
-      "grad_norm": 8.094645500183105,
+      "grad_norm": 10.190463066101074,
       "learning_rate": 2.2502502502502504e-06,
-      "loss": 0.8805,
+      "loss": 0.8809,
       "step": 77420
     },
     {
       "epoch": 232.52,
-      "grad_norm": 6.735079765319824,
+      "grad_norm": 6.834677219390869,
       "learning_rate": 2.2492492492492494e-06,
-      "loss": 0.9739,
+      "loss": 0.9748,
       "step": 77430
     },
     {
       "epoch": 232.55,
-      "grad_norm": 5.4663496017456055,
+      "grad_norm": 4.9428887367248535,
       "learning_rate": 2.2482482482482484e-06,
-      "loss": 0.899,
+      "loss": 0.8998,
       "step": 77440
     },
     {
       "epoch": 232.58,
-      "grad_norm": 8.838212966918945,
+      "grad_norm": 6.180479526519775,
       "learning_rate": 2.2472472472472475e-06,
-      "loss": 0.8653,
+      "loss": 0.8771,
       "step": 77450
     },
     {
       "epoch": 232.61,
-      "grad_norm": 7.071592330932617,
+      "grad_norm": 5.603001594543457,
       "learning_rate": 2.2462462462462465e-06,
-      "loss": 0.8951,
+      "loss": 0.8933,
       "step": 77460
     },
     {
       "epoch": 232.64,
-      "grad_norm": 5.172804832458496,
+      "grad_norm": 6.335277080535889,
       "learning_rate": 2.2452452452452455e-06,
-      "loss": 0.8846,
+      "loss": 0.8809,
       "step": 77470
     },
     {
       "epoch": 232.67,
-      "grad_norm": 7.767462253570557,
+      "grad_norm": 11.84900951385498,
       "learning_rate": 2.2442442442442445e-06,
-      "loss": 0.9488,
+      "loss": 0.9612,
       "step": 77480
     },
     {
       "epoch": 232.7,
-      "grad_norm": 10.067631721496582,
+      "grad_norm": 8.026116371154785,
       "learning_rate": 2.2432432432432435e-06,
-      "loss": 0.9005,
+      "loss": 0.8934,
       "step": 77490
     },
     {
       "epoch": 232.73,
-      "grad_norm": 7.372007846832275,
+      "grad_norm": 5.990533828735352,
       "learning_rate": 2.2422422422422426e-06,
-      "loss": 0.8809,
+      "loss": 0.877,
       "step": 77500
     },
     {
       "epoch": 232.76,
-      "grad_norm": 12.325270652770996,
+      "grad_norm": 9.878954887390137,
       "learning_rate": 2.2412412412412416e-06,
-      "loss": 0.9279,
+      "loss": 0.9189,
       "step": 77510
     },
     {
       "epoch": 232.79,
-      "grad_norm": 6.726020336151123,
+      "grad_norm": 6.607303619384766,
       "learning_rate": 2.2402402402402406e-06,
-      "loss": 0.999,
+      "loss": 0.9908,
       "step": 77520
     },
     {
       "epoch": 232.82,
-      "grad_norm": 5.781802654266357,
+      "grad_norm": 6.197053909301758,
       "learning_rate": 2.2392392392392396e-06,
-      "loss": 0.9048,
+      "loss": 0.9017,
       "step": 77530
     },
     {
       "epoch": 232.85,
-      "grad_norm": 5.680134296417236,
+      "grad_norm": 6.451783180236816,
       "learning_rate": 2.238238238238238e-06,
-      "loss": 0.9778,
+      "loss": 0.9755,
       "step": 77540
     },
     {
       "epoch": 232.88,
-      "grad_norm": 8.939736366271973,
+      "grad_norm": 6.729145526885986,
       "learning_rate": 2.2372372372372376e-06,
-      "loss": 0.9571,
+      "loss": 0.9561,
       "step": 77550
     },
     {
       "epoch": 232.91,
-      "grad_norm": 5.613337993621826,
+      "grad_norm": 5.456052780151367,
       "learning_rate": 2.2362362362362362e-06,
-      "loss": 0.8438,
+      "loss": 0.8504,
       "step": 77560
     },
     {
       "epoch": 232.94,
-      "grad_norm": 8.531615257263184,
+      "grad_norm": 6.871230125427246,
       "learning_rate": 2.2352352352352357e-06,
-      "loss": 0.931,
+      "loss": 0.9383,
       "step": 77570
     },
     {
       "epoch": 232.97,
-      "grad_norm": 11.178622245788574,
+      "grad_norm": 10.713069915771484,
       "learning_rate": 2.2342342342342343e-06,
-      "loss": 0.9361,
+      "loss": 0.933,
       "step": 77580
     },
     {
       "epoch": 233.0,
-      "eval_accuracy": 0.8279,
-      "eval_loss": 0.571701169013977,
-      "eval_runtime": 5.1364,
-      "eval_samples_per_second": 1946.874,
-      "eval_steps_per_second": 7.787,
+      "eval_accuracy": 0.8285,
+      "eval_loss": 0.5703023672103882,
+      "eval_runtime": 5.2637,
+      "eval_samples_per_second": 1899.815,
+      "eval_steps_per_second": 7.599,
       "step": 77589
     },
     {
       "epoch": 233.0,
-      "grad_norm": 6.860579490661621,
+      "grad_norm": 6.603271484375,
       "learning_rate": 2.2332332332332333e-06,
-      "loss": 0.997,
+      "loss": 0.9967,
       "step": 77590
     },
     {
       "epoch": 233.03,
-      "grad_norm": 6.592562675476074,
+      "grad_norm": 8.83248233795166,
       "learning_rate": 2.2322322322322323e-06,
-      "loss": 0.9019,
+      "loss": 0.9091,
       "step": 77600
     },
     {
       "epoch": 233.06,
-      "grad_norm": 5.211353302001953,
+      "grad_norm": 5.11921501159668,
       "learning_rate": 2.2312312312312313e-06,
-      "loss": 0.9701,
+      "loss": 0.9777,
       "step": 77610
     },
     {
       "epoch": 233.09,
-      "grad_norm": 6.265829563140869,
+      "grad_norm": 7.035044193267822,
       "learning_rate": 2.2302302302302304e-06,
-      "loss": 0.8999,
+      "loss": 0.8977,
       "step": 77620
     },
     {
       "epoch": 233.12,
-      "grad_norm": 7.270330429077148,
+      "grad_norm": 7.487051010131836,
       "learning_rate": 2.2292292292292294e-06,
-      "loss": 0.8916,
+      "loss": 0.8924,
       "step": 77630
     },
     {
       "epoch": 233.15,
-      "grad_norm": 6.6357855796813965,
+      "grad_norm": 5.831080436706543,
       "learning_rate": 2.2282282282282284e-06,
-      "loss": 0.9364,
+      "loss": 0.9259,
       "step": 77640
     },
     {
       "epoch": 233.18,
-      "grad_norm": 9.041322708129883,
+      "grad_norm": 8.768872261047363,
       "learning_rate": 2.2272272272272274e-06,
-      "loss": 0.9813,
+      "loss": 0.9841,
       "step": 77650
     },
     {
       "epoch": 233.21,
-      "grad_norm": 6.436062335968018,
+      "grad_norm": 6.437468528747559,
       "learning_rate": 2.2262262262262264e-06,
-      "loss": 0.9099,
+      "loss": 0.9146,
       "step": 77660
     },
     {
       "epoch": 233.24,
-      "grad_norm": 6.235227108001709,
+      "grad_norm": 7.283465385437012,
       "learning_rate": 2.2252252252252254e-06,
-      "loss": 0.932,
+      "loss": 0.9344,
       "step": 77670
     },
     {
       "epoch": 233.27,
-      "grad_norm": 8.540156364440918,
+      "grad_norm": 7.467930793762207,
       "learning_rate": 2.2242242242242245e-06,
-      "loss": 0.9085,
+      "loss": 0.9011,
       "step": 77680
     },
     {
       "epoch": 233.3,
-      "grad_norm": 6.413301944732666,
+      "grad_norm": 8.33309555053711,
       "learning_rate": 2.2232232232232235e-06,
-      "loss": 0.8933,
+      "loss": 0.8707,
       "step": 77690
     },
     {
       "epoch": 233.33,
-      "grad_norm": 5.195923328399658,
+      "grad_norm": 4.946023464202881,
       "learning_rate": 2.222222222222222e-06,
-      "loss": 0.9258,
+      "loss": 0.9174,
       "step": 77700
     },
     {
       "epoch": 233.36,
-      "grad_norm": 5.433948993682861,
+      "grad_norm": 6.002856254577637,
       "learning_rate": 2.2212212212212215e-06,
-      "loss": 0.9376,
+      "loss": 0.9295,
       "step": 77710
     },
     {
       "epoch": 233.39,
-      "grad_norm": 10.12315559387207,
+      "grad_norm": 9.267861366271973,
       "learning_rate": 2.22022022022022e-06,
-      "loss": 1.0253,
+      "loss": 1.016,
       "step": 77720
     },
     {
       "epoch": 233.42,
-      "grad_norm": 5.145108222961426,
+      "grad_norm": 5.980457782745361,
       "learning_rate": 2.2192192192192196e-06,
-      "loss": 0.81,
+      "loss": 0.8049,
       "step": 77730
     },
     {
       "epoch": 233.45,
-      "grad_norm": 8.95115852355957,
+      "grad_norm": 6.880836486816406,
       "learning_rate": 2.2182182182182186e-06,
-      "loss": 0.942,
+      "loss": 0.9381,
       "step": 77740
     },
     {
       "epoch": 233.48,
-      "grad_norm": 5.335904121398926,
+      "grad_norm": 7.069659233093262,
       "learning_rate": 2.2172172172172176e-06,
-      "loss": 0.8855,
+      "loss": 0.8787,
       "step": 77750
     },
     {
       "epoch": 233.51,
-      "grad_norm": 7.803169250488281,
+      "grad_norm": 8.721061706542969,
       "learning_rate": 2.2162162162162166e-06,
-      "loss": 0.9568,
+      "loss": 0.9566,
       "step": 77760
     },
     {
       "epoch": 233.54,
-      "grad_norm": 7.157209873199463,
+      "grad_norm": 10.311488151550293,
       "learning_rate": 2.215215215215215e-06,
-      "loss": 0.9656,
+      "loss": 0.9537,
       "step": 77770
     },
     {
       "epoch": 233.57,
-      "grad_norm": 6.619572639465332,
+      "grad_norm": 9.001455307006836,
       "learning_rate": 2.2142142142142146e-06,
-      "loss": 0.9176,
+      "loss": 0.9306,
       "step": 77780
     },
     {
       "epoch": 233.6,
-      "grad_norm": 7.046349048614502,
+      "grad_norm": 9.750514030456543,
       "learning_rate": 2.2132132132132132e-06,
-      "loss": 0.8154,
+      "loss": 0.8095,
       "step": 77790
     },
     {
       "epoch": 233.63,
-      "grad_norm": 5.6041083335876465,
+      "grad_norm": 7.6474504470825195,
       "learning_rate": 2.2122122122122127e-06,
-      "loss": 0.8958,
+      "loss": 0.8901,
       "step": 77800
     },
     {
       "epoch": 233.66,
-      "grad_norm": 5.470070838928223,
+      "grad_norm": 8.789084434509277,
       "learning_rate": 2.2112112112112113e-06,
-      "loss": 0.9597,
+      "loss": 0.962,
       "step": 77810
     },
     {
       "epoch": 233.69,
-      "grad_norm": 7.785733222961426,
+      "grad_norm": 7.050671577453613,
       "learning_rate": 2.2102102102102103e-06,
-      "loss": 0.8862,
+      "loss": 0.8874,
       "step": 77820
     },
     {
       "epoch": 233.72,
-      "grad_norm": 5.676257133483887,
+      "grad_norm": 6.830833911895752,
       "learning_rate": 2.2092092092092093e-06,
-      "loss": 0.9167,
+      "loss": 0.9126,
       "step": 77830
     },
     {
       "epoch": 233.75,
-      "grad_norm": 7.109170436859131,
+      "grad_norm": 6.947665691375732,
       "learning_rate": 2.2082082082082083e-06,
-      "loss": 0.948,
+      "loss": 0.9468,
       "step": 77840
     },
     {
       "epoch": 233.78,
-      "grad_norm": 6.427718162536621,
+      "grad_norm": 6.534623622894287,
       "learning_rate": 2.2072072072072073e-06,
-      "loss": 0.9652,
+      "loss": 0.965,
       "step": 77850
     },
     {
       "epoch": 233.81,
-      "grad_norm": 10.473333358764648,
+      "grad_norm": 7.61991548538208,
       "learning_rate": 2.2062062062062064e-06,
-      "loss": 0.8142,
+      "loss": 0.8233,
       "step": 77860
     },
     {
       "epoch": 233.84,
-      "grad_norm": 8.3261079788208,
+      "grad_norm": 6.48091983795166,
       "learning_rate": 2.2052052052052054e-06,
-      "loss": 0.901,
+      "loss": 0.8961,
       "step": 77870
     },
     {
       "epoch": 233.87,
-      "grad_norm": 5.686007976531982,
+      "grad_norm": 5.485215663909912,
       "learning_rate": 2.2042042042042044e-06,
-      "loss": 0.9269,
+      "loss": 0.918,
       "step": 77880
     },
     {
       "epoch": 233.9,
-      "grad_norm": 6.1266632080078125,
+      "grad_norm": 6.72568416595459,
       "learning_rate": 2.2032032032032034e-06,
-      "loss": 0.9551,
+      "loss": 0.9533,
       "step": 77890
     },
     {
       "epoch": 233.93,
-      "grad_norm": 8.794051170349121,
+      "grad_norm": 7.380227088928223,
       "learning_rate": 2.2022022022022024e-06,
-      "loss": 0.8566,
+      "loss": 0.8516,
       "step": 77900
     },
     {
       "epoch": 233.96,
-      "grad_norm": 6.45546293258667,
+      "grad_norm": 6.073263168334961,
       "learning_rate": 2.2012012012012015e-06,
-      "loss": 0.902,
+      "loss": 0.9062,
       "step": 77910
     },
     {
       "epoch": 233.99,
-      "grad_norm": 9.693825721740723,
+      "grad_norm": 7.228025913238525,
       "learning_rate": 2.2002002002002005e-06,
-      "loss": 0.922,
+      "loss": 0.9192,
       "step": 77920
     },
     {
       "epoch": 234.0,
-      "eval_accuracy": 0.8262,
-      "eval_loss": 0.5721768736839294,
-      "eval_runtime": 5.5494,
-      "eval_samples_per_second": 1802.001,
-      "eval_steps_per_second": 7.208,
+      "eval_accuracy": 0.8291,
+      "eval_loss": 0.571128249168396,
+      "eval_runtime": 5.4556,
+      "eval_samples_per_second": 1832.991,
+      "eval_steps_per_second": 7.332,
       "step": 77922
     },
     {
       "epoch": 234.02,
-      "grad_norm": 6.2369914054870605,
+      "grad_norm": 6.323684215545654,
       "learning_rate": 2.199199199199199e-06,
-      "loss": 1.0523,
+      "loss": 1.0537,
       "step": 77930
     },
     {
       "epoch": 234.05,
-      "grad_norm": 7.014670372009277,
+      "grad_norm": 7.293963432312012,
       "learning_rate": 2.1981981981981985e-06,
-      "loss": 0.9075,
+      "loss": 0.9033,
       "step": 77940
     },
     {
       "epoch": 234.08,
-      "grad_norm": 7.476398468017578,
+      "grad_norm": 8.843942642211914,
       "learning_rate": 2.197197197197197e-06,
-      "loss": 0.9548,
+      "loss": 0.9581,
       "step": 77950
     },
     {
       "epoch": 234.11,
-      "grad_norm": 5.924478054046631,
+      "grad_norm": 5.707630634307861,
       "learning_rate": 2.1961961961961966e-06,
-      "loss": 0.8971,
+      "loss": 0.8774,
       "step": 77960
     },
     {
       "epoch": 234.14,
-      "grad_norm": 7.236154556274414,
+      "grad_norm": 9.725639343261719,
       "learning_rate": 2.195195195195195e-06,
-      "loss": 0.9302,
+      "loss": 0.9268,
       "step": 77970
     },
     {
       "epoch": 234.17,
-      "grad_norm": 6.874781608581543,
+      "grad_norm": 7.502881050109863,
       "learning_rate": 2.1941941941941946e-06,
-      "loss": 0.8917,
+      "loss": 0.895,
       "step": 77980
     },
     {
       "epoch": 234.2,
-      "grad_norm": 6.92641544342041,
+      "grad_norm": 6.301641941070557,
       "learning_rate": 2.193193193193193e-06,
-      "loss": 0.9175,
+      "loss": 0.9047,
       "step": 77990
     },
     {
       "epoch": 234.23,
-      "grad_norm": 6.900253772735596,
+      "grad_norm": 6.493235111236572,
       "learning_rate": 2.192192192192192e-06,
-      "loss": 0.9153,
+      "loss": 0.9123,
       "step": 78000
     },
     {
       "epoch": 234.26,
-      "grad_norm": 5.783236980438232,
+      "grad_norm": 7.103471755981445,
       "learning_rate": 2.1911911911911916e-06,
-      "loss": 0.9725,
+      "loss": 0.9757,
       "step": 78010
     },
     {
       "epoch": 234.29,
-      "grad_norm": 7.4990739822387695,
+      "grad_norm": 8.74559211730957,
       "learning_rate": 2.1901901901901902e-06,
-      "loss": 0.9075,
+      "loss": 0.9035,
       "step": 78020
     },
     {
       "epoch": 234.32,
-      "grad_norm": 4.335666656494141,
+      "grad_norm": 3.719987630844116,
       "learning_rate": 2.1891891891891897e-06,
-      "loss": 0.7952,
+      "loss": 0.7898,
       "step": 78030
     },
     {
       "epoch": 234.35,
-      "grad_norm": 6.512816905975342,
+      "grad_norm": 6.183246612548828,
       "learning_rate": 2.1881881881881883e-06,
-      "loss": 0.8324,
+      "loss": 0.8262,
       "step": 78040
     },
     {
       "epoch": 234.38,
-      "grad_norm": 6.214112281799316,
+      "grad_norm": 6.7937469482421875,
       "learning_rate": 2.1871871871871873e-06,
-      "loss": 0.9166,
+      "loss": 0.9161,
       "step": 78050
     },
     {
       "epoch": 234.41,
-      "grad_norm": 8.091532707214355,
+      "grad_norm": 9.973197937011719,
       "learning_rate": 2.1861861861861863e-06,
-      "loss": 0.8893,
+      "loss": 0.8894,
       "step": 78060
     },
     {
       "epoch": 234.44,
-      "grad_norm": 7.947592258453369,
+      "grad_norm": 7.951627254486084,
       "learning_rate": 2.1851851851851853e-06,
-      "loss": 0.93,
+      "loss": 0.9352,
       "step": 78070
     },
     {
       "epoch": 234.47,
-      "grad_norm": 8.433773040771484,
+      "grad_norm": 7.380625247955322,
       "learning_rate": 2.1841841841841843e-06,
-      "loss": 0.8887,
+      "loss": 0.8796,
       "step": 78080
     },
     {
       "epoch": 234.5,
-      "grad_norm": 5.490725994110107,
+      "grad_norm": 6.483063697814941,
       "learning_rate": 2.1831831831831834e-06,
-      "loss": 0.9128,
+      "loss": 0.9122,
       "step": 78090
     },
     {
       "epoch": 234.53,
-      "grad_norm": 8.573627471923828,
+      "grad_norm": 6.8720879554748535,
       "learning_rate": 2.1821821821821824e-06,
-      "loss": 0.8556,
+      "loss": 0.8524,
       "step": 78100
     },
     {
       "epoch": 234.56,
-      "grad_norm": 6.472256183624268,
+      "grad_norm": 6.4283061027526855,
       "learning_rate": 2.1811811811811814e-06,
-      "loss": 0.8423,
+      "loss": 0.8418,
       "step": 78110
     },
     {
       "epoch": 234.59,
-      "grad_norm": 5.301424980163574,
+      "grad_norm": 8.63646125793457,
       "learning_rate": 2.1801801801801804e-06,
-      "loss": 0.9237,
+      "loss": 0.9204,
       "step": 78120
     },
     {
       "epoch": 234.62,
-      "grad_norm": 8.395922660827637,
+      "grad_norm": 7.458379745483398,
       "learning_rate": 2.1791791791791794e-06,
-      "loss": 0.9948,
+      "loss": 0.9874,
       "step": 78130
     },
     {
       "epoch": 234.65,
-      "grad_norm": 9.153078079223633,
+      "grad_norm": 7.430457592010498,
       "learning_rate": 2.1781781781781785e-06,
-      "loss": 0.9468,
+      "loss": 0.9513,
       "step": 78140
     },
     {
       "epoch": 234.68,
-      "grad_norm": 6.9626784324646,
+      "grad_norm": 6.111026287078857,
       "learning_rate": 2.1771771771771775e-06,
-      "loss": 0.9029,
+      "loss": 0.9078,
       "step": 78150
     },
     {
       "epoch": 234.71,
-      "grad_norm": 6.766234397888184,
+      "grad_norm": 5.448149681091309,
       "learning_rate": 2.176176176176176e-06,
-      "loss": 0.8911,
+      "loss": 0.8987,
       "step": 78160
     },
     {
       "epoch": 234.74,
-      "grad_norm": 7.940955638885498,
+      "grad_norm": 5.862161636352539,
       "learning_rate": 2.1751751751751755e-06,
-      "loss": 0.8965,
+      "loss": 0.8952,
       "step": 78170
     },
     {
       "epoch": 234.77,
-      "grad_norm": 6.049698352813721,
+      "grad_norm": 5.449333190917969,
       "learning_rate": 2.174174174174174e-06,
-      "loss": 0.9006,
+      "loss": 0.8965,
       "step": 78180
     },
     {
       "epoch": 234.8,
-      "grad_norm": 8.069253921508789,
+      "grad_norm": 7.082630157470703,
       "learning_rate": 2.1731731731731735e-06,
-      "loss": 0.974,
+      "loss": 0.9764,
       "step": 78190
     },
     {
       "epoch": 234.83,
-      "grad_norm": 8.171812057495117,
+      "grad_norm": 9.406994819641113,
       "learning_rate": 2.172172172172172e-06,
-      "loss": 0.8871,
+      "loss": 0.8795,
       "step": 78200
     },
     {
       "epoch": 234.86,
-      "grad_norm": 7.061241149902344,
+      "grad_norm": 6.693833827972412,
       "learning_rate": 2.1711711711711716e-06,
-      "loss": 0.9218,
+      "loss": 0.925,
       "step": 78210
     },
     {
       "epoch": 234.89,
-      "grad_norm": 6.313744068145752,
+      "grad_norm": 8.332348823547363,
       "learning_rate": 2.17017017017017e-06,
-      "loss": 0.9889,
+      "loss": 0.9905,
       "step": 78220
     },
     {
       "epoch": 234.92,
-      "grad_norm": 6.111332893371582,
+      "grad_norm": 9.066394805908203,
       "learning_rate": 2.169169169169169e-06,
-      "loss": 0.9397,
+      "loss": 0.9265,
       "step": 78230
     },
     {
       "epoch": 234.95,
-      "grad_norm": 9.318737983703613,
+      "grad_norm": 8.215083122253418,
       "learning_rate": 2.1681681681681682e-06,
-      "loss": 0.9063,
+      "loss": 0.912,
       "step": 78240
     },
     {
       "epoch": 234.98,
-      "grad_norm": 7.514743328094482,
+      "grad_norm": 7.085821151733398,
       "learning_rate": 2.1671671671671672e-06,
-      "loss": 0.9189,
+      "loss": 0.9118,
       "step": 78250
     },
     {
       "epoch": 235.0,
-      "eval_accuracy": 0.8278,
-      "eval_loss": 0.5670110583305359,
-      "eval_runtime": 5.4333,
-      "eval_samples_per_second": 1840.51,
-      "eval_steps_per_second": 7.362,
+      "eval_accuracy": 0.8309,
+      "eval_loss": 0.5672502517700195,
+      "eval_runtime": 5.3288,
+      "eval_samples_per_second": 1876.609,
+      "eval_steps_per_second": 7.506,
       "step": 78255
     },
     {
       "epoch": 235.02,
-      "grad_norm": 8.011584281921387,
+      "grad_norm": 6.796942234039307,
       "learning_rate": 2.1661661661661663e-06,
-      "loss": 1.1048,
+      "loss": 1.1074,
       "step": 78260
     },
     {
       "epoch": 235.05,
-      "grad_norm": 9.050116539001465,
+      "grad_norm": 12.941593170166016,
       "learning_rate": 2.1651651651651653e-06,
-      "loss": 0.881,
+      "loss": 0.8865,
       "step": 78270
     },
     {
       "epoch": 235.08,
-      "grad_norm": 6.336880683898926,
+      "grad_norm": 5.861849308013916,
       "learning_rate": 2.1641641641641643e-06,
-      "loss": 0.8979,
+      "loss": 0.8988,
       "step": 78280
     },
     {
       "epoch": 235.11,
-      "grad_norm": 5.591732025146484,
+      "grad_norm": 6.914661884307861,
       "learning_rate": 2.1631631631631633e-06,
-      "loss": 0.9212,
+      "loss": 0.9113,
       "step": 78290
     },
     {
       "epoch": 235.14,
-      "grad_norm": 5.6512770652771,
+      "grad_norm": 5.719606876373291,
       "learning_rate": 2.1621621621621623e-06,
-      "loss": 0.9545,
+      "loss": 0.9611,
       "step": 78300
     },
     {
       "epoch": 235.17,
-      "grad_norm": 7.562773704528809,
+      "grad_norm": 9.372536659240723,
       "learning_rate": 2.1611611611611613e-06,
-      "loss": 0.9074,
+      "loss": 0.9119,
       "step": 78310
     },
     {
       "epoch": 235.2,
-      "grad_norm": 4.46877384185791,
+      "grad_norm": 4.830929279327393,
       "learning_rate": 2.1601601601601604e-06,
-      "loss": 0.9318,
+      "loss": 0.9209,
       "step": 78320
     },
     {
       "epoch": 235.23,
-      "grad_norm": 6.481410026550293,
+      "grad_norm": 7.6742353439331055,
       "learning_rate": 2.1591591591591594e-06,
-      "loss": 0.938,
+      "loss": 0.9344,
       "step": 78330
     },
     {
       "epoch": 235.26,
-      "grad_norm": 7.206664085388184,
+      "grad_norm": 7.8953070640563965,
       "learning_rate": 2.1581581581581584e-06,
-      "loss": 0.8465,
+      "loss": 0.8396,
       "step": 78340
     },
     {
       "epoch": 235.29,
-      "grad_norm": 7.663712024688721,
+      "grad_norm": 7.843598365783691,
       "learning_rate": 2.1571571571571574e-06,
-      "loss": 0.9387,
+      "loss": 0.9437,
       "step": 78350
     },
     {
       "epoch": 235.32,
-      "grad_norm": 6.413125038146973,
+      "grad_norm": 7.542398929595947,
       "learning_rate": 2.1561561561561564e-06,
-      "loss": 0.925,
+      "loss": 0.9263,
       "step": 78360
     },
     {
       "epoch": 235.35,
-      "grad_norm": 7.542961597442627,
+      "grad_norm": 5.3237810134887695,
       "learning_rate": 2.1551551551551555e-06,
-      "loss": 0.869,
+      "loss": 0.8694,
       "step": 78370
     },
     {
       "epoch": 235.38,
-      "grad_norm": 5.2424445152282715,
+      "grad_norm": 4.538902759552002,
       "learning_rate": 2.1541541541541545e-06,
-      "loss": 0.8255,
+      "loss": 0.8265,
       "step": 78380
     },
     {
       "epoch": 235.41,
-      "grad_norm": 6.881526947021484,
+      "grad_norm": 6.967612266540527,
       "learning_rate": 2.153153153153153e-06,
-      "loss": 0.9583,
+      "loss": 0.9435,
       "step": 78390
     },
     {
       "epoch": 235.44,
-      "grad_norm": 8.569782257080078,
+      "grad_norm": 6.96912956237793,
       "learning_rate": 2.1521521521521525e-06,
-      "loss": 0.8917,
+      "loss": 0.9019,
       "step": 78400
     },
     {
       "epoch": 235.47,
-      "grad_norm": 7.4873151779174805,
+      "grad_norm": 7.1526875495910645,
       "learning_rate": 2.151151151151151e-06,
-      "loss": 0.8989,
+      "loss": 0.9059,
       "step": 78410
     },
     {
       "epoch": 235.5,
-      "grad_norm": 6.340014457702637,
+      "grad_norm": 7.677491664886475,
       "learning_rate": 2.1501501501501505e-06,
-      "loss": 0.9128,
+      "loss": 0.9052,
       "step": 78420
     },
     {
       "epoch": 235.53,
-      "grad_norm": 6.527122497558594,
+      "grad_norm": 5.699955463409424,
       "learning_rate": 2.149149149149149e-06,
-      "loss": 0.924,
+      "loss": 0.9249,
       "step": 78430
     },
     {
       "epoch": 235.56,
-      "grad_norm": 6.671878337860107,
+      "grad_norm": 4.994957447052002,
       "learning_rate": 2.148148148148148e-06,
       "loss": 0.9408,
       "step": 78440
     },
     {
       "epoch": 235.59,
-      "grad_norm": 5.462967395782471,
+      "grad_norm": 6.39487886428833,
       "learning_rate": 2.147147147147147e-06,
-      "loss": 0.9619,
+      "loss": 0.9644,
       "step": 78450
     },
     {
       "epoch": 235.62,
-      "grad_norm": 13.27970027923584,
+      "grad_norm": 12.136369705200195,
       "learning_rate": 2.146146146146146e-06,
-      "loss": 0.8855,
+      "loss": 0.8782,
       "step": 78460
     },
     {
       "epoch": 235.65,
-      "grad_norm": 9.291534423828125,
+      "grad_norm": 8.658941268920898,
       "learning_rate": 2.1451451451451452e-06,
-      "loss": 0.9652,
+      "loss": 0.9669,
       "step": 78470
     },
     {
       "epoch": 235.68,
-      "grad_norm": 9.636818885803223,
+      "grad_norm": 8.58685302734375,
       "learning_rate": 2.1441441441441442e-06,
-      "loss": 0.9413,
+      "loss": 0.934,
       "step": 78480
     },
     {
       "epoch": 235.71,
-      "grad_norm": 6.927985668182373,
+      "grad_norm": 7.841699600219727,
       "learning_rate": 2.1431431431431433e-06,
-      "loss": 0.8547,
+      "loss": 0.844,
       "step": 78490
     },
     {
       "epoch": 235.74,
-      "grad_norm": 5.829562664031982,
+      "grad_norm": 5.748601913452148,
       "learning_rate": 2.1421421421421423e-06,
-      "loss": 0.933,
+      "loss": 0.9259,
       "step": 78500
     },
     {
       "epoch": 235.77,
-      "grad_norm": 11.930814743041992,
+      "grad_norm": 9.534053802490234,
       "learning_rate": 2.1411411411411413e-06,
-      "loss": 0.9491,
+      "loss": 0.9553,
       "step": 78510
     },
     {
       "epoch": 235.8,
-      "grad_norm": 6.18719482421875,
+      "grad_norm": 6.647080898284912,
       "learning_rate": 2.1401401401401403e-06,
-      "loss": 0.9539,
+      "loss": 0.9523,
       "step": 78520
     },
     {
       "epoch": 235.83,
-      "grad_norm": 7.901001930236816,
+      "grad_norm": 5.316436767578125,
       "learning_rate": 2.1391391391391393e-06,
-      "loss": 0.9602,
+      "loss": 0.968,
       "step": 78530
     },
     {
       "epoch": 235.86,
-      "grad_norm": 6.1194047927856445,
+      "grad_norm": 7.4282331466674805,
       "learning_rate": 2.1381381381381383e-06,
-      "loss": 0.9,
+      "loss": 0.9054,
       "step": 78540
     },
     {
       "epoch": 235.89,
-      "grad_norm": 6.192215919494629,
+      "grad_norm": 6.049473285675049,
       "learning_rate": 2.1371371371371374e-06,
-      "loss": 0.9493,
+      "loss": 0.9425,
       "step": 78550
     },
     {
       "epoch": 235.92,
-      "grad_norm": 6.430853843688965,
+      "grad_norm": 6.990262508392334,
       "learning_rate": 2.1361361361361364e-06,
-      "loss": 0.9089,
+      "loss": 0.9123,
       "step": 78560
     },
     {
       "epoch": 235.95,
-      "grad_norm": 7.563118934631348,
+      "grad_norm": 8.213315963745117,
       "learning_rate": 2.1351351351351354e-06,
-      "loss": 0.8864,
+      "loss": 0.8865,
       "step": 78570
     },
     {
       "epoch": 235.98,
-      "grad_norm": 7.34456205368042,
+      "grad_norm": 8.82752799987793,
       "learning_rate": 2.1341341341341344e-06,
-      "loss": 0.835,
+      "loss": 0.8398,
       "step": 78580
     },
     {
       "epoch": 236.0,
-      "eval_accuracy": 0.8274,
-      "eval_loss": 0.5674439072608948,
-      "eval_runtime": 5.4937,
-      "eval_samples_per_second": 1820.254,
-      "eval_steps_per_second": 7.281,
+      "eval_accuracy": 0.8311,
+      "eval_loss": 0.5661165714263916,
+      "eval_runtime": 5.2878,
+      "eval_samples_per_second": 1891.145,
+      "eval_steps_per_second": 7.565,
       "step": 78588
     },
     {
       "epoch": 236.01,
-      "grad_norm": 5.868900299072266,
+      "grad_norm": 8.320673942565918,
       "learning_rate": 2.1331331331331334e-06,
-      "loss": 1.1247,
+      "loss": 1.1216,
       "step": 78590
     },
     {
       "epoch": 236.04,
-      "grad_norm": 10.059922218322754,
+      "grad_norm": 7.024194240570068,
       "learning_rate": 2.1321321321321325e-06,
-      "loss": 0.9136,
+      "loss": 0.8986,
       "step": 78600
     },
     {
       "epoch": 236.07,
-      "grad_norm": 6.116087913513184,
+      "grad_norm": 6.341711044311523,
       "learning_rate": 2.1311311311311315e-06,
-      "loss": 0.8925,
+      "loss": 0.896,
       "step": 78610
     },
     {
       "epoch": 236.1,
-      "grad_norm": 10.305264472961426,
+      "grad_norm": 12.292296409606934,
       "learning_rate": 2.13013013013013e-06,
-      "loss": 0.9016,
+      "loss": 0.893,
       "step": 78620
     },
     {
       "epoch": 236.13,
-      "grad_norm": 8.764731407165527,
+      "grad_norm": 8.637133598327637,
       "learning_rate": 2.1291291291291295e-06,
-      "loss": 0.9477,
+      "loss": 0.9483,
       "step": 78630
     },
     {
       "epoch": 236.16,
-      "grad_norm": 6.114943027496338,
+      "grad_norm": 5.261669635772705,
       "learning_rate": 2.128128128128128e-06,
-      "loss": 0.8898,
+      "loss": 0.9008,
       "step": 78640
     },
     {
       "epoch": 236.19,
-      "grad_norm": 6.336632251739502,
+      "grad_norm": 6.096682071685791,
       "learning_rate": 2.1271271271271275e-06,
-      "loss": 0.9397,
+      "loss": 0.9325,
       "step": 78650
     },
     {
       "epoch": 236.22,
-      "grad_norm": 5.824873924255371,
+      "grad_norm": 5.881407260894775,
       "learning_rate": 2.126126126126126e-06,
-      "loss": 0.9114,
+      "loss": 0.9097,
       "step": 78660
     },
     {
       "epoch": 236.25,
-      "grad_norm": 6.878175735473633,
+      "grad_norm": 6.401546478271484,
       "learning_rate": 2.125125125125125e-06,
-      "loss": 0.8804,
+      "loss": 0.8679,
       "step": 78670
     },
     {
       "epoch": 236.28,
-      "grad_norm": 7.377870559692383,
+      "grad_norm": 5.920120716094971,
       "learning_rate": 2.124124124124124e-06,
-      "loss": 0.9592,
+      "loss": 0.9488,
       "step": 78680
     },
     {
       "epoch": 236.31,
-      "grad_norm": 6.068465709686279,
+      "grad_norm": 6.2438740730285645,
       "learning_rate": 2.123123123123123e-06,
-      "loss": 0.8826,
+      "loss": 0.879,
       "step": 78690
     },
     {
       "epoch": 236.34,
-      "grad_norm": 5.4476094245910645,
+      "grad_norm": 7.217743396759033,
       "learning_rate": 2.1221221221221222e-06,
-      "loss": 0.9063,
+      "loss": 0.918,
       "step": 78700
     },
     {
       "epoch": 236.37,
-      "grad_norm": 9.126185417175293,
+      "grad_norm": 7.64132022857666,
       "learning_rate": 2.1211211211211212e-06,
-      "loss": 0.9136,
+      "loss": 0.9125,
       "step": 78710
     },
     {
       "epoch": 236.4,
-      "grad_norm": 6.952035427093506,
+      "grad_norm": 5.952597141265869,
       "learning_rate": 2.1201201201201203e-06,
-      "loss": 0.8742,
+      "loss": 0.8672,
       "step": 78720
     },
     {
       "epoch": 236.43,
-      "grad_norm": 7.853507041931152,
+      "grad_norm": 8.917959213256836,
       "learning_rate": 2.1191191191191193e-06,
-      "loss": 0.9334,
+      "loss": 0.9291,
       "step": 78730
     },
     {
       "epoch": 236.46,
-      "grad_norm": 8.721124649047852,
+      "grad_norm": 8.804521560668945,
       "learning_rate": 2.1181181181181183e-06,
-      "loss": 0.9209,
+      "loss": 0.9181,
       "step": 78740
     },
     {
       "epoch": 236.49,
-      "grad_norm": 5.982433795928955,
+      "grad_norm": 9.2962064743042,
       "learning_rate": 2.1171171171171173e-06,
-      "loss": 0.9165,
+      "loss": 0.9166,
       "step": 78750
     },
     {
       "epoch": 236.52,
-      "grad_norm": 6.723097801208496,
+      "grad_norm": 5.861688137054443,
       "learning_rate": 2.1161161161161163e-06,
-      "loss": 0.9451,
+      "loss": 0.9344,
       "step": 78760
     },
     {
       "epoch": 236.55,
-      "grad_norm": 7.299095153808594,
+      "grad_norm": 6.535707950592041,
       "learning_rate": 2.1151151151151153e-06,
-      "loss": 0.9453,
+      "loss": 0.954,
       "step": 78770
     },
     {
       "epoch": 236.58,
-      "grad_norm": 5.886063098907471,
+      "grad_norm": 6.2168049812316895,
       "learning_rate": 2.1141141141141144e-06,
-      "loss": 0.8725,
+      "loss": 0.8758,
       "step": 78780
     },
     {
       "epoch": 236.61,
-      "grad_norm": 6.25809907913208,
+      "grad_norm": 7.136752605438232,
       "learning_rate": 2.1131131131131134e-06,
-      "loss": 0.919,
+      "loss": 0.9178,
       "step": 78790
     },
     {
       "epoch": 236.64,
-      "grad_norm": 6.663574695587158,
+      "grad_norm": 5.694781303405762,
       "learning_rate": 2.1121121121121124e-06,
-      "loss": 0.8303,
+      "loss": 0.8336,
       "step": 78800
     },
     {
       "epoch": 236.67,
-      "grad_norm": 5.870668411254883,
+      "grad_norm": 6.563995838165283,
       "learning_rate": 2.1111111111111114e-06,
-      "loss": 0.9532,
+      "loss": 0.9534,
       "step": 78810
     },
     {
       "epoch": 236.7,
-      "grad_norm": 7.352170467376709,
+      "grad_norm": 7.172924995422363,
       "learning_rate": 2.1101101101101104e-06,
-      "loss": 0.9746,
+      "loss": 0.971,
       "step": 78820
     },
     {
       "epoch": 236.73,
-      "grad_norm": 8.324058532714844,
+      "grad_norm": 10.03869915008545,
       "learning_rate": 2.1091091091091095e-06,
-      "loss": 0.9246,
+      "loss": 0.9252,
       "step": 78830
     },
     {
       "epoch": 236.76,
-      "grad_norm": 8.368306159973145,
+      "grad_norm": 5.567151069641113,
       "learning_rate": 2.1081081081081085e-06,
-      "loss": 0.8754,
+      "loss": 0.8806,
       "step": 78840
     },
     {
       "epoch": 236.79,
-      "grad_norm": 11.57325553894043,
+      "grad_norm": 6.556605339050293,
       "learning_rate": 2.107107107107107e-06,
-      "loss": 0.9689,
+      "loss": 0.9552,
       "step": 78850
     },
     {
       "epoch": 236.82,
-      "grad_norm": 8.177701950073242,
+      "grad_norm": 8.613974571228027,
       "learning_rate": 2.1061061061061065e-06,
-      "loss": 0.9205,
+      "loss": 0.9137,
       "step": 78860
     },
     {
       "epoch": 236.85,
-      "grad_norm": 6.8446125984191895,
+      "grad_norm": 5.9837751388549805,
       "learning_rate": 2.105105105105105e-06,
-      "loss": 0.9273,
+      "loss": 0.927,
       "step": 78870
     },
     {
       "epoch": 236.88,
-      "grad_norm": 5.831998825073242,
+      "grad_norm": 8.157293319702148,
       "learning_rate": 2.1041041041041045e-06,
-      "loss": 0.8769,
+      "loss": 0.8867,
       "step": 78880
     },
     {
       "epoch": 236.91,
-      "grad_norm": 9.079029083251953,
+      "grad_norm": 10.386295318603516,
       "learning_rate": 2.103103103103103e-06,
-      "loss": 0.925,
+      "loss": 0.922,
       "step": 78890
     },
     {
       "epoch": 236.94,
-      "grad_norm": 7.762720584869385,
+      "grad_norm": 7.774256229400635,
       "learning_rate": 2.102102102102102e-06,
-      "loss": 0.9221,
+      "loss": 0.9207,
       "step": 78900
     },
     {
       "epoch": 236.97,
-      "grad_norm": 5.851589202880859,
+      "grad_norm": 8.27913761138916,
       "learning_rate": 2.101101101101101e-06,
-      "loss": 0.8663,
+      "loss": 0.8684,
       "step": 78910
     },
     {
       "epoch": 237.0,
-      "grad_norm": 7.907855987548828,
+      "grad_norm": 9.497757911682129,
       "learning_rate": 2.1001001001001e-06,
-      "loss": 1.0082,
+      "loss": 1.0123,
       "step": 78920
     },
     {
       "epoch": 237.0,
-      "eval_accuracy": 0.8256,
-      "eval_loss": 0.5738283395767212,
-      "eval_runtime": 5.6252,
-      "eval_samples_per_second": 1777.723,
-      "eval_steps_per_second": 7.111,
+      "eval_accuracy": 0.8274,
+      "eval_loss": 0.5746479630470276,
+      "eval_runtime": 5.2622,
+      "eval_samples_per_second": 1900.34,
+      "eval_steps_per_second": 7.601,
       "step": 78921
     },
     {
       "epoch": 237.03,
-      "grad_norm": 10.682631492614746,
+      "grad_norm": 7.048360347747803,
       "learning_rate": 2.0990990990990992e-06,
-      "loss": 1.0616,
+      "loss": 1.081,
       "step": 78930
     },
     {
       "epoch": 237.06,
-      "grad_norm": 7.331655025482178,
+      "grad_norm": 7.391964435577393,
       "learning_rate": 2.0980980980980982e-06,
-      "loss": 0.9202,
+      "loss": 0.9237,
       "step": 78940
     },
     {
       "epoch": 237.09,
-      "grad_norm": 5.253720283508301,
+      "grad_norm": 4.9896697998046875,
       "learning_rate": 2.0970970970970972e-06,
-      "loss": 0.9054,
+      "loss": 0.9062,
       "step": 78950
     },
     {
       "epoch": 237.12,
-      "grad_norm": 5.370076656341553,
+      "grad_norm": 5.192037582397461,
       "learning_rate": 2.0960960960960963e-06,
-      "loss": 0.8996,
+      "loss": 0.8874,
       "step": 78960
     },
     {
       "epoch": 237.15,
-      "grad_norm": 6.0759596824646,
+      "grad_norm": 8.238003730773926,
       "learning_rate": 2.0950950950950953e-06,
-      "loss": 0.9302,
+      "loss": 0.934,
       "step": 78970
     },
     {
       "epoch": 237.18,
-      "grad_norm": 7.79900598526001,
+      "grad_norm": 6.310318946838379,
       "learning_rate": 2.0940940940940943e-06,
-      "loss": 0.9582,
+      "loss": 0.9546,
       "step": 78980
     },
     {
       "epoch": 237.21,
-      "grad_norm": 9.454049110412598,
+      "grad_norm": 9.431255340576172,
       "learning_rate": 2.0930930930930933e-06,
-      "loss": 0.8908,
+      "loss": 0.8791,
       "step": 78990
     },
     {
       "epoch": 237.24,
-      "grad_norm": 6.041835308074951,
+      "grad_norm": 7.567631244659424,
       "learning_rate": 2.0920920920920923e-06,
-      "loss": 0.9238,
+      "loss": 0.9292,
       "step": 79000
     },
     {
       "epoch": 237.27,
-      "grad_norm": 9.205573081970215,
+      "grad_norm": 6.934300899505615,
       "learning_rate": 2.0910910910910914e-06,
-      "loss": 0.9506,
+      "loss": 0.9491,
       "step": 79010
     },
     {
       "epoch": 237.3,
-      "grad_norm": 4.1607184410095215,
+      "grad_norm": 4.751235008239746,
       "learning_rate": 2.0900900900900904e-06,
-      "loss": 0.9072,
+      "loss": 0.895,
       "step": 79020
     },
     {
       "epoch": 237.33,
-      "grad_norm": 8.39903450012207,
+      "grad_norm": 7.728883266448975,
       "learning_rate": 2.089089089089089e-06,
-      "loss": 0.9308,
+      "loss": 0.9289,
       "step": 79030
     },
     {
       "epoch": 237.36,
-      "grad_norm": 9.024653434753418,
+      "grad_norm": 9.052355766296387,
       "learning_rate": 2.0880880880880884e-06,
-      "loss": 0.9323,
+      "loss": 0.9302,
       "step": 79040
     },
     {
       "epoch": 237.39,
-      "grad_norm": 8.282200813293457,
+      "grad_norm": 6.41889762878418,
       "learning_rate": 2.087087087087087e-06,
-      "loss": 0.9043,
+      "loss": 0.8965,
       "step": 79050
     },
     {
       "epoch": 237.42,
-      "grad_norm": 6.648709297180176,
+      "grad_norm": 7.041486740112305,
       "learning_rate": 2.0860860860860865e-06,
-      "loss": 0.9164,
+      "loss": 0.921,
       "step": 79060
     },
     {
       "epoch": 237.45,
-      "grad_norm": 5.790912628173828,
+      "grad_norm": 6.814572811126709,
       "learning_rate": 2.0850850850850855e-06,
-      "loss": 0.9368,
+      "loss": 0.9339,
       "step": 79070
     },
     {
       "epoch": 237.48,
-      "grad_norm": 6.185238361358643,
+      "grad_norm": 7.775021553039551,
       "learning_rate": 2.084084084084084e-06,
-      "loss": 0.8862,
+      "loss": 0.8858,
       "step": 79080
     },
     {
       "epoch": 237.51,
-      "grad_norm": 6.041407585144043,
+      "grad_norm": 7.4046244621276855,
       "learning_rate": 2.0830830830830835e-06,
-      "loss": 0.8982,
+      "loss": 0.8983,
       "step": 79090
     },
     {
       "epoch": 237.54,
-      "grad_norm": 4.037703514099121,
+      "grad_norm": 4.402141571044922,
       "learning_rate": 2.082082082082082e-06,
-      "loss": 0.9069,
+      "loss": 0.9111,
       "step": 79100
     },
     {
       "epoch": 237.57,
-      "grad_norm": 8.084012985229492,
+      "grad_norm": 7.402466297149658,
       "learning_rate": 2.0810810810810815e-06,
-      "loss": 0.9806,
+      "loss": 0.9852,
       "step": 79110
     },
     {
       "epoch": 237.6,
-      "grad_norm": 4.542537212371826,
+      "grad_norm": 6.740317344665527,
       "learning_rate": 2.08008008008008e-06,
-      "loss": 0.9511,
+      "loss": 0.9515,
       "step": 79120
     },
     {
       "epoch": 237.63,
-      "grad_norm": 7.9864020347595215,
+      "grad_norm": 6.193668365478516,
       "learning_rate": 2.079079079079079e-06,
-      "loss": 0.8979,
+      "loss": 0.9,
       "step": 79130
     },
     {
       "epoch": 237.66,
-      "grad_norm": 9.053156852722168,
+      "grad_norm": 7.695112228393555,
       "learning_rate": 2.078078078078078e-06,
-      "loss": 0.9221,
+      "loss": 0.9263,
       "step": 79140
     },
     {
       "epoch": 237.69,
-      "grad_norm": 6.959280014038086,
+      "grad_norm": 6.416289806365967,
       "learning_rate": 2.077077077077077e-06,
-      "loss": 1.0026,
+      "loss": 0.9914,
       "step": 79150
     },
     {
       "epoch": 237.72,
-      "grad_norm": 7.732060432434082,
+      "grad_norm": 6.3660383224487305,
       "learning_rate": 2.076076076076076e-06,
-      "loss": 1.0338,
+      "loss": 1.0403,
       "step": 79160
     },
     {
       "epoch": 237.75,
-      "grad_norm": 6.1724677085876465,
+      "grad_norm": 6.039412975311279,
       "learning_rate": 2.0750750750750752e-06,
-      "loss": 0.8594,
+      "loss": 0.8612,
       "step": 79170
     },
     {
       "epoch": 237.78,
-      "grad_norm": 6.478306770324707,
+      "grad_norm": 6.154743671417236,
       "learning_rate": 2.0740740740740742e-06,
-      "loss": 0.916,
+      "loss": 0.9242,
       "step": 79180
     },
     {
       "epoch": 237.81,
-      "grad_norm": 8.3814115524292,
+      "grad_norm": 6.880722999572754,
       "learning_rate": 2.0730730730730733e-06,
-      "loss": 0.8891,
+      "loss": 0.887,
       "step": 79190
     },
     {
       "epoch": 237.84,
-      "grad_norm": 9.270408630371094,
+      "grad_norm": 6.726841449737549,
       "learning_rate": 2.0720720720720723e-06,
-      "loss": 0.8958,
+      "loss": 0.8983,
       "step": 79200
     },
     {
       "epoch": 237.87,
-      "grad_norm": 5.378468036651611,
+      "grad_norm": 5.810263156890869,
       "learning_rate": 2.0710710710710713e-06,
-      "loss": 0.9024,
+      "loss": 0.8994,
       "step": 79210
     },
     {
       "epoch": 237.9,
-      "grad_norm": 6.967156410217285,
+      "grad_norm": 8.133099555969238,
       "learning_rate": 2.0700700700700703e-06,
-      "loss": 0.9386,
+      "loss": 0.942,
       "step": 79220
     },
     {
       "epoch": 237.93,
-      "grad_norm": 4.955174446105957,
+      "grad_norm": 5.589068412780762,
       "learning_rate": 2.0690690690690693e-06,
-      "loss": 0.9446,
+      "loss": 0.9416,
       "step": 79230
     },
     {
       "epoch": 237.96,
-      "grad_norm": 6.470575332641602,
+      "grad_norm": 6.69540548324585,
       "learning_rate": 2.0680680680680684e-06,
-      "loss": 0.9042,
+      "loss": 0.8962,
       "step": 79240
     },
     {
       "epoch": 237.99,
-      "grad_norm": 8.558087348937988,
+      "grad_norm": 10.802363395690918,
       "learning_rate": 2.0670670670670674e-06,
-      "loss": 0.9356,
+      "loss": 0.9227,
       "step": 79250
     },
     {
       "epoch": 238.0,
-      "eval_accuracy": 0.8277,
-      "eval_loss": 0.5700724124908447,
-      "eval_runtime": 5.5305,
-      "eval_samples_per_second": 1808.154,
-      "eval_steps_per_second": 7.233,
+      "eval_accuracy": 0.8305,
+      "eval_loss": 0.5690770745277405,
+      "eval_runtime": 5.3162,
+      "eval_samples_per_second": 1881.03,
+      "eval_steps_per_second": 7.524,
       "step": 79254
     },
     {
       "epoch": 238.02,
-      "grad_norm": 4.595211505889893,
+      "grad_norm": 4.621455669403076,
       "learning_rate": 2.066066066066066e-06,
-      "loss": 1.0174,
+      "loss": 1.0187,
       "step": 79260
     },
     {
       "epoch": 238.05,
-      "grad_norm": 4.422239303588867,
+      "grad_norm": 4.206650733947754,
       "learning_rate": 2.0650650650650654e-06,
-      "loss": 0.9137,
+      "loss": 0.9185,
       "step": 79270
     },
     {
       "epoch": 238.08,
-      "grad_norm": 7.209720134735107,
+      "grad_norm": 8.36380672454834,
       "learning_rate": 2.064064064064064e-06,
-      "loss": 0.9006,
+      "loss": 0.8955,
       "step": 79280
     },
     {
       "epoch": 238.11,
-      "grad_norm": 5.85824728012085,
+      "grad_norm": 6.201006889343262,
       "learning_rate": 2.0630630630630634e-06,
-      "loss": 0.899,
+      "loss": 0.8914,
       "step": 79290
     },
     {
       "epoch": 238.14,
-      "grad_norm": 7.951224327087402,
+      "grad_norm": 6.982199192047119,
       "learning_rate": 2.062062062062062e-06,
-      "loss": 0.9495,
+      "loss": 0.9492,
       "step": 79300
     },
     {
       "epoch": 238.17,
-      "grad_norm": 5.390174388885498,
+      "grad_norm": 7.765468120574951,
       "learning_rate": 2.061061061061061e-06,
-      "loss": 0.9192,
+      "loss": 0.9123,
       "step": 79310
     },
     {
       "epoch": 238.2,
-      "grad_norm": 8.032416343688965,
+      "grad_norm": 8.517627716064453,
       "learning_rate": 2.06006006006006e-06,
-      "loss": 0.9552,
+      "loss": 0.9564,
       "step": 79320
     },
     {
       "epoch": 238.23,
-      "grad_norm": 4.978128433227539,
+      "grad_norm": 6.989130973815918,
       "learning_rate": 2.059059059059059e-06,
-      "loss": 0.8852,
+      "loss": 0.8811,
       "step": 79330
     },
     {
       "epoch": 238.26,
-      "grad_norm": 7.137717247009277,
+      "grad_norm": 5.454930782318115,
       "learning_rate": 2.0580580580580585e-06,
-      "loss": 0.9355,
+      "loss": 0.9302,
       "step": 79340
     },
     {
       "epoch": 238.29,
-      "grad_norm": 5.4523820877075195,
+      "grad_norm": 7.488287925720215,
       "learning_rate": 2.057057057057057e-06,
-      "loss": 0.9122,
+      "loss": 0.9155,
       "step": 79350
     },
     {
       "epoch": 238.32,
-      "grad_norm": 6.521915912628174,
+      "grad_norm": 7.18209171295166,
       "learning_rate": 2.056056056056056e-06,
-      "loss": 0.9674,
+      "loss": 0.9661,
       "step": 79360
     },
     {
       "epoch": 238.35,
-      "grad_norm": 9.1487398147583,
+      "grad_norm": 6.5546393394470215,
       "learning_rate": 2.055055055055055e-06,
-      "loss": 0.933,
+      "loss": 0.9185,
       "step": 79370
     },
     {
       "epoch": 238.38,
-      "grad_norm": 6.724252700805664,
+      "grad_norm": 5.951749801635742,
       "learning_rate": 2.054054054054054e-06,
-      "loss": 0.9243,
+      "loss": 0.9248,
       "step": 79380
     },
     {
       "epoch": 238.41,
-      "grad_norm": 8.253846168518066,
+      "grad_norm": 9.474506378173828,
       "learning_rate": 2.053053053053053e-06,
-      "loss": 0.8749,
+      "loss": 0.8767,
       "step": 79390
     },
     {
       "epoch": 238.44,
-      "grad_norm": 7.003234386444092,
+      "grad_norm": 7.854935646057129,
       "learning_rate": 2.0520520520520522e-06,
-      "loss": 0.9661,
+      "loss": 0.9616,
       "step": 79400
     },
     {
       "epoch": 238.47,
-      "grad_norm": 7.469712734222412,
+      "grad_norm": 8.45244026184082,
       "learning_rate": 2.0510510510510512e-06,
-      "loss": 0.9559,
+      "loss": 0.9482,
       "step": 79410
     },
     {
       "epoch": 238.5,
-      "grad_norm": 6.683964252471924,
+      "grad_norm": 8.774053573608398,
       "learning_rate": 2.0500500500500503e-06,
-      "loss": 0.9437,
+      "loss": 0.9374,
       "step": 79420
     },
     {
       "epoch": 238.53,
-      "grad_norm": 5.920119762420654,
+      "grad_norm": 8.333842277526855,
       "learning_rate": 2.0490490490490493e-06,
-      "loss": 0.9268,
+      "loss": 0.9277,
       "step": 79430
     },
     {
       "epoch": 238.56,
-      "grad_norm": 5.904017925262451,
+      "grad_norm": 8.038762092590332,
       "learning_rate": 2.0480480480480483e-06,
-      "loss": 0.9554,
+      "loss": 0.9527,
       "step": 79440
     },
     {
       "epoch": 238.59,
-      "grad_norm": 8.074451446533203,
+      "grad_norm": 6.325090408325195,
       "learning_rate": 2.0470470470470473e-06,
-      "loss": 0.9379,
+      "loss": 0.9382,
       "step": 79450
     },
     {
       "epoch": 238.62,
-      "grad_norm": 9.723038673400879,
+      "grad_norm": 9.116259574890137,
       "learning_rate": 2.0460460460460463e-06,
-      "loss": 0.9402,
+      "loss": 0.9514,
       "step": 79460
     },
     {
       "epoch": 238.65,
-      "grad_norm": 8.67032527923584,
+      "grad_norm": 11.616948127746582,
       "learning_rate": 2.045045045045045e-06,
-      "loss": 0.8812,
+      "loss": 0.8844,
       "step": 79470
     },
     {
       "epoch": 238.68,
-      "grad_norm": 3.9169485569000244,
+      "grad_norm": 4.38642692565918,
       "learning_rate": 2.0440440440440444e-06,
-      "loss": 0.8275,
+      "loss": 0.8291,
       "step": 79480
     },
     {
       "epoch": 238.71,
-      "grad_norm": 8.231185913085938,
+      "grad_norm": 7.843504905700684,
       "learning_rate": 2.043043043043043e-06,
-      "loss": 0.9441,
+      "loss": 0.9519,
       "step": 79490
     },
     {
       "epoch": 238.74,
-      "grad_norm": 8.569451332092285,
+      "grad_norm": 10.395471572875977,
       "learning_rate": 2.0420420420420424e-06,
-      "loss": 0.9716,
+      "loss": 0.9646,
       "step": 79500
     },
     {
       "epoch": 238.77,
-      "grad_norm": 7.799859046936035,
+      "grad_norm": 9.08304500579834,
       "learning_rate": 2.041041041041041e-06,
-      "loss": 0.894,
+      "loss": 0.9065,
       "step": 79510
     },
     {
       "epoch": 238.8,
-      "grad_norm": 8.295920372009277,
+      "grad_norm": 8.05537223815918,
       "learning_rate": 2.0400400400400404e-06,
-      "loss": 0.9607,
+      "loss": 0.9527,
       "step": 79520
     },
     {
       "epoch": 238.83,
-      "grad_norm": 6.2205729484558105,
+      "grad_norm": 5.676578521728516,
       "learning_rate": 2.039039039039039e-06,
-      "loss": 0.8632,
+      "loss": 0.87,
       "step": 79530
     },
     {
       "epoch": 238.86,
-      "grad_norm": 5.702792167663574,
+      "grad_norm": 10.652159690856934,
       "learning_rate": 2.038038038038038e-06,
-      "loss": 0.9582,
+      "loss": 0.9576,
       "step": 79540
     },
     {
       "epoch": 238.89,
-      "grad_norm": 5.220892906188965,
+      "grad_norm": 5.329954624176025,
       "learning_rate": 2.037037037037037e-06,
-      "loss": 0.9144,
+      "loss": 0.9226,
       "step": 79550
     },
     {
       "epoch": 238.92,
-      "grad_norm": 4.117013931274414,
+      "grad_norm": 4.639743328094482,
       "learning_rate": 2.036036036036036e-06,
-      "loss": 0.8583,
+      "loss": 0.8588,
       "step": 79560
     },
     {
       "epoch": 238.95,
-      "grad_norm": 6.001654148101807,
+      "grad_norm": 6.423077583312988,
       "learning_rate": 2.035035035035035e-06,
-      "loss": 0.9276,
+      "loss": 0.9176,
       "step": 79570
     },
     {
       "epoch": 238.98,
-      "grad_norm": 5.467064380645752,
+      "grad_norm": 5.387325763702393,
       "learning_rate": 2.034034034034034e-06,
-      "loss": 0.888,
+      "loss": 0.889,
       "step": 79580
     },
     {
       "epoch": 239.0,
-      "eval_accuracy": 0.8267,
-      "eval_loss": 0.5701087117195129,
-      "eval_runtime": 5.3318,
-      "eval_samples_per_second": 1875.534,
-      "eval_steps_per_second": 7.502,
+      "eval_accuracy": 0.8299,
+      "eval_loss": 0.570182204246521,
+      "eval_runtime": 5.4722,
+      "eval_samples_per_second": 1827.422,
+      "eval_steps_per_second": 7.31,
       "step": 79587
     },
     {
       "epoch": 239.01,
-      "grad_norm": 8.053563117980957,
+      "grad_norm": 7.797935485839844,
       "learning_rate": 2.033033033033033e-06,
-      "loss": 0.9997,
+      "loss": 1.0114,
       "step": 79590
     },
     {
       "epoch": 239.04,
-      "grad_norm": 6.82689905166626,
+      "grad_norm": 7.086243152618408,
       "learning_rate": 2.032032032032032e-06,
-      "loss": 0.9292,
+      "loss": 0.9285,
       "step": 79600
     },
     {
       "epoch": 239.07,
-      "grad_norm": 7.067287445068359,
+      "grad_norm": 6.769633769989014,
       "learning_rate": 2.031031031031031e-06,
-      "loss": 0.9627,
+      "loss": 0.9634,
       "step": 79610
     },
     {
       "epoch": 239.1,
-      "grad_norm": 5.714476585388184,
+      "grad_norm": 6.112381935119629,
       "learning_rate": 2.03003003003003e-06,
-      "loss": 0.8581,
+      "loss": 0.8565,
       "step": 79620
     },
     {
       "epoch": 239.13,
-      "grad_norm": 6.998996257781982,
+      "grad_norm": 6.669654846191406,
       "learning_rate": 2.0290290290290292e-06,
-      "loss": 0.9326,
+      "loss": 0.9187,
       "step": 79630
     },
     {
       "epoch": 239.16,
-      "grad_norm": 7.050302982330322,
+      "grad_norm": 6.42284631729126,
       "learning_rate": 2.0280280280280282e-06,
-      "loss": 0.9092,
+      "loss": 0.9063,
       "step": 79640
     },
     {
       "epoch": 239.19,
-      "grad_norm": 7.030488014221191,
+      "grad_norm": 8.7627534866333,
       "learning_rate": 2.0270270270270273e-06,
-      "loss": 0.9603,
+      "loss": 0.957,
       "step": 79650
     },
     {
       "epoch": 239.22,
-      "grad_norm": 7.096002578735352,
+      "grad_norm": 6.162563800811768,
       "learning_rate": 2.0260260260260263e-06,
-      "loss": 0.8881,
+      "loss": 0.8844,
       "step": 79660
     },
     {
       "epoch": 239.25,
-      "grad_norm": 5.6830153465271,
+      "grad_norm": 7.5658159255981445,
       "learning_rate": 2.0250250250250253e-06,
-      "loss": 0.9335,
+      "loss": 0.9292,
       "step": 79670
     },
     {
       "epoch": 239.28,
-      "grad_norm": 5.237267017364502,
+      "grad_norm": 7.350477695465088,
       "learning_rate": 2.0240240240240243e-06,
-      "loss": 0.9485,
+      "loss": 0.9529,
       "step": 79680
     },
     {
       "epoch": 239.31,
-      "grad_norm": 8.349817276000977,
+      "grad_norm": 11.210776329040527,
       "learning_rate": 2.0230230230230233e-06,
-      "loss": 0.8868,
+      "loss": 0.8799,
       "step": 79690
     },
     {
       "epoch": 239.34,
-      "grad_norm": 4.338740825653076,
+      "grad_norm": 4.373857021331787,
       "learning_rate": 2.022022022022022e-06,
-      "loss": 0.9292,
+      "loss": 0.9085,
       "step": 79700
     },
     {
       "epoch": 239.37,
-      "grad_norm": 6.839813709259033,
+      "grad_norm": 7.250993728637695,
       "learning_rate": 2.0210210210210214e-06,
-      "loss": 0.9484,
+      "loss": 0.9495,
       "step": 79710
     },
     {
       "epoch": 239.4,
-      "grad_norm": 9.339956283569336,
+      "grad_norm": 8.075458526611328,
       "learning_rate": 2.02002002002002e-06,
-      "loss": 0.9123,
+      "loss": 0.9128,
       "step": 79720
     },
     {
       "epoch": 239.43,
-      "grad_norm": 7.723249912261963,
+      "grad_norm": 6.225409984588623,
       "learning_rate": 2.0190190190190194e-06,
-      "loss": 0.9276,
+      "loss": 0.9309,
       "step": 79730
     },
     {
       "epoch": 239.46,
-      "grad_norm": 8.889991760253906,
+      "grad_norm": 10.801323890686035,
       "learning_rate": 2.018018018018018e-06,
-      "loss": 0.9312,
+      "loss": 0.9288,
       "step": 79740
     },
     {
       "epoch": 239.49,
-      "grad_norm": 7.889145374298096,
+      "grad_norm": 8.917524337768555,
       "learning_rate": 2.0170170170170174e-06,
-      "loss": 0.9144,
+      "loss": 0.9085,
       "step": 79750
     },
     {
       "epoch": 239.52,
-      "grad_norm": 5.735894203186035,
+      "grad_norm": 6.392947196960449,
       "learning_rate": 2.016016016016016e-06,
-      "loss": 0.8548,
+      "loss": 0.8529,
       "step": 79760
     },
     {
       "epoch": 239.55,
-      "grad_norm": 9.55145263671875,
+      "grad_norm": 6.783271789550781,
       "learning_rate": 2.015015015015015e-06,
-      "loss": 0.9338,
+      "loss": 0.9449,
       "step": 79770
     },
     {
       "epoch": 239.58,
-      "grad_norm": 13.139598846435547,
+      "grad_norm": 11.912308692932129,
       "learning_rate": 2.014014014014014e-06,
-      "loss": 0.9127,
+      "loss": 0.9147,
       "step": 79780
     },
     {
       "epoch": 239.61,
-      "grad_norm": 8.005192756652832,
+      "grad_norm": 6.044051170349121,
       "learning_rate": 2.013013013013013e-06,
-      "loss": 0.8817,
+      "loss": 0.8846,
       "step": 79790
     },
     {
       "epoch": 239.64,
-      "grad_norm": 5.7790207862854,
+      "grad_norm": 6.022842884063721,
       "learning_rate": 2.012012012012012e-06,
-      "loss": 0.9292,
+      "loss": 0.9345,
       "step": 79800
     },
     {
       "epoch": 239.67,
-      "grad_norm": 5.90079927444458,
+      "grad_norm": 5.972153186798096,
       "learning_rate": 2.011011011011011e-06,
-      "loss": 0.871,
+      "loss": 0.8746,
       "step": 79810
     },
     {
       "epoch": 239.7,
-      "grad_norm": 6.05889368057251,
+      "grad_norm": 7.071434020996094,
       "learning_rate": 2.01001001001001e-06,
-      "loss": 0.9493,
+      "loss": 0.9422,
       "step": 79820
     },
     {
       "epoch": 239.73,
-      "grad_norm": 4.567934036254883,
+      "grad_norm": 5.203873634338379,
       "learning_rate": 2.009009009009009e-06,
-      "loss": 0.8916,
+      "loss": 0.8924,
       "step": 79830
     },
     {
       "epoch": 239.76,
-      "grad_norm": 10.93428897857666,
+      "grad_norm": 7.433931350708008,
       "learning_rate": 2.008008008008008e-06,
-      "loss": 0.9611,
+      "loss": 0.9632,
       "step": 79840
     },
     {
       "epoch": 239.79,
-      "grad_norm": 6.052842140197754,
+      "grad_norm": 6.0935282707214355,
       "learning_rate": 2.007007007007007e-06,
-      "loss": 0.9002,
+      "loss": 0.9011,
       "step": 79850
     },
     {
       "epoch": 239.82,
-      "grad_norm": 4.563061714172363,
+      "grad_norm": 5.713609218597412,
       "learning_rate": 2.0060060060060062e-06,
-      "loss": 0.858,
+      "loss": 0.8675,
       "step": 79860
     },
     {
       "epoch": 239.85,
-      "grad_norm": 7.202702045440674,
+      "grad_norm": 6.371826648712158,
       "learning_rate": 2.0050050050050052e-06,
-      "loss": 0.9803,
+      "loss": 0.9794,
       "step": 79870
     },
     {
       "epoch": 239.88,
-      "grad_norm": 6.810670852661133,
+      "grad_norm": 8.568815231323242,
       "learning_rate": 2.0040040040040043e-06,
-      "loss": 0.9227,
+      "loss": 0.9225,
       "step": 79880
     },
     {
       "epoch": 239.91,
-      "grad_norm": 7.678959369659424,
+      "grad_norm": 10.045238494873047,
       "learning_rate": 2.0030030030030033e-06,
-      "loss": 0.9423,
+      "loss": 0.9342,
       "step": 79890
     },
     {
       "epoch": 239.94,
-      "grad_norm": 6.030205249786377,
+      "grad_norm": 6.546368598937988,
       "learning_rate": 2.0020020020020023e-06,
-      "loss": 0.8758,
+      "loss": 0.8602,
       "step": 79900
     },
     {
       "epoch": 239.97,
-      "grad_norm": 7.461560249328613,
+      "grad_norm": 11.242127418518066,
       "learning_rate": 2.0010010010010013e-06,
-      "loss": 0.993,
+      "loss": 0.9909,
       "step": 79910
     },
     {
       "epoch": 240.0,
-      "grad_norm": 58.39124298095703,
+      "grad_norm": 69.17842102050781,
       "learning_rate": 2.0000000000000003e-06,
-      "loss": 1.2103,
+      "loss": 1.1921,
       "step": 79920
     },
     {
       "epoch": 240.0,
-      "eval_accuracy": 0.8237,
-      "eval_loss": 0.5769767761230469,
-      "eval_runtime": 5.5635,
-      "eval_samples_per_second": 1797.418,
-      "eval_steps_per_second": 7.19,
+      "eval_accuracy": 0.8281,
+      "eval_loss": 0.5765551328659058,
+      "eval_runtime": 5.4844,
+      "eval_samples_per_second": 1823.351,
+      "eval_steps_per_second": 7.293,
       "step": 79920
     },
     {
       "epoch": 240.03,
-      "grad_norm": 6.164283275604248,
+      "grad_norm": 6.979743957519531,
       "learning_rate": 1.998998998998999e-06,
-      "loss": 0.8476,
+      "loss": 0.8439,
       "step": 79930
     },
     {
       "epoch": 240.06,
-      "grad_norm": 6.0582475662231445,
+      "grad_norm": 6.525388717651367,
       "learning_rate": 1.9979979979979984e-06,
-      "loss": 0.9074,
+      "loss": 0.9191,
       "step": 79940
     },
     {
       "epoch": 240.09,
-      "grad_norm": 6.848342418670654,
+      "grad_norm": 6.544798851013184,
       "learning_rate": 1.996996996996997e-06,
-      "loss": 0.9449,
+      "loss": 0.9405,
       "step": 79950
     },
     {
       "epoch": 240.12,
-      "grad_norm": 9.105920791625977,
+      "grad_norm": 8.849898338317871,
       "learning_rate": 1.9959959959959964e-06,
-      "loss": 0.9809,
+      "loss": 0.9827,
       "step": 79960
     },
     {
       "epoch": 240.15,
-      "grad_norm": 5.711158275604248,
+      "grad_norm": 5.036855697631836,
       "learning_rate": 1.994994994994995e-06,
-      "loss": 0.9807,
+      "loss": 0.9851,
       "step": 79970
     },
     {
       "epoch": 240.18,
-      "grad_norm": 7.027157306671143,
+      "grad_norm": 6.855947494506836,
       "learning_rate": 1.9939939939939944e-06,
-      "loss": 0.9658,
+      "loss": 0.9607,
       "step": 79980
     },
     {
       "epoch": 240.21,
-      "grad_norm": 7.2197136878967285,
+      "grad_norm": 8.473443984985352,
       "learning_rate": 1.992992992992993e-06,
-      "loss": 0.894,
+      "loss": 0.8955,
       "step": 79990
     },
     {
       "epoch": 240.24,
-      "grad_norm": 7.476736068725586,
+      "grad_norm": 7.210253715515137,
       "learning_rate": 1.991991991991992e-06,
-      "loss": 0.9114,
+      "loss": 0.9053,
       "step": 80000
     },
     {
       "epoch": 240.27,
-      "grad_norm": 6.049589157104492,
+      "grad_norm": 8.556487083435059,
       "learning_rate": 1.990990990990991e-06,
-      "loss": 0.8935,
+      "loss": 0.8831,
       "step": 80010
     },
     {
       "epoch": 240.3,
-      "grad_norm": 8.871532440185547,
+      "grad_norm": 12.8659086227417,
       "learning_rate": 1.98998998998999e-06,
-      "loss": 0.8872,
+      "loss": 0.8854,
       "step": 80020
     },
     {
       "epoch": 240.33,
-      "grad_norm": 4.981647491455078,
+      "grad_norm": 5.98153018951416,
       "learning_rate": 1.988988988988989e-06,
-      "loss": 0.9496,
+      "loss": 0.9511,
       "step": 80030
     },
     {
       "epoch": 240.36,
-      "grad_norm": 6.580799579620361,
+      "grad_norm": 6.0896430015563965,
       "learning_rate": 1.987987987987988e-06,
-      "loss": 0.8675,
+      "loss": 0.8534,
       "step": 80040
     },
     {
       "epoch": 240.39,
-      "grad_norm": 5.923852443695068,
+      "grad_norm": 6.669402599334717,
       "learning_rate": 1.986986986986987e-06,
-      "loss": 0.8722,
+      "loss": 0.8631,
       "step": 80050
     },
     {
       "epoch": 240.42,
-      "grad_norm": 7.994172096252441,
+      "grad_norm": 6.423517227172852,
       "learning_rate": 1.985985985985986e-06,
-      "loss": 0.8825,
+      "loss": 0.8821,
       "step": 80060
     },
     {
       "epoch": 240.45,
-      "grad_norm": 5.332455158233643,
+      "grad_norm": 5.597962856292725,
       "learning_rate": 1.984984984984985e-06,
-      "loss": 0.9091,
+      "loss": 0.907,
       "step": 80070
     },
     {
       "epoch": 240.48,
-      "grad_norm": 8.139275550842285,
+      "grad_norm": 8.394994735717773,
       "learning_rate": 1.983983983983984e-06,
-      "loss": 0.9109,
+      "loss": 0.9119,
       "step": 80080
     },
     {
       "epoch": 240.51,
-      "grad_norm": 6.4767231941223145,
+      "grad_norm": 6.764835357666016,
       "learning_rate": 1.9829829829829832e-06,
-      "loss": 0.9272,
+      "loss": 0.9287,
       "step": 80090
     },
     {
       "epoch": 240.54,
-      "grad_norm": 5.963770389556885,
+      "grad_norm": 7.044877529144287,
       "learning_rate": 1.9819819819819822e-06,
-      "loss": 0.909,
+      "loss": 0.9149,
       "step": 80100
     },
     {
       "epoch": 240.57,
-      "grad_norm": 7.653842926025391,
+      "grad_norm": 8.942560195922852,
       "learning_rate": 1.980980980980981e-06,
-      "loss": 0.9334,
+      "loss": 0.9436,
       "step": 80110
     },
     {
       "epoch": 240.6,
-      "grad_norm": 6.3791632652282715,
+      "grad_norm": 8.54296875,
       "learning_rate": 1.9799799799799803e-06,
-      "loss": 0.8778,
+      "loss": 0.8772,
       "step": 80120
     },
     {
       "epoch": 240.63,
-      "grad_norm": 5.845083236694336,
+      "grad_norm": 5.9875807762146,
       "learning_rate": 1.978978978978979e-06,
-      "loss": 0.8492,
+      "loss": 0.8513,
       "step": 80130
     },
     {
       "epoch": 240.66,
-      "grad_norm": 5.227138996124268,
+      "grad_norm": 5.50302791595459,
       "learning_rate": 1.9779779779779783e-06,
-      "loss": 0.887,
+      "loss": 0.8727,
       "step": 80140
     },
     {
       "epoch": 240.69,
-      "grad_norm": 5.825594902038574,
+      "grad_norm": 7.572701454162598,
       "learning_rate": 1.9769769769769773e-06,
-      "loss": 0.9441,
+      "loss": 0.9448,
       "step": 80150
     },
     {
       "epoch": 240.72,
-      "grad_norm": 7.962092399597168,
+      "grad_norm": 12.2985258102417,
       "learning_rate": 1.975975975975976e-06,
-      "loss": 1.0068,
+      "loss": 1.0056,
       "step": 80160
     },
     {
       "epoch": 240.75,
-      "grad_norm": 7.863005638122559,
+      "grad_norm": 6.754595756530762,
       "learning_rate": 1.9749749749749754e-06,
-      "loss": 0.8849,
+      "loss": 0.8818,
       "step": 80170
     },
     {
       "epoch": 240.78,
-      "grad_norm": 5.959181308746338,
+      "grad_norm": 6.293399810791016,
       "learning_rate": 1.973973973973974e-06,
-      "loss": 0.8568,
+      "loss": 0.8522,
       "step": 80180
     },
     {
       "epoch": 240.81,
-      "grad_norm": 6.385360240936279,
+      "grad_norm": 6.135404109954834,
       "learning_rate": 1.9729729729729734e-06,
-      "loss": 0.8819,
+      "loss": 0.8797,
       "step": 80190
     },
     {
       "epoch": 240.84,
-      "grad_norm": 6.85277795791626,
+      "grad_norm": 9.690993309020996,
       "learning_rate": 1.971971971971972e-06,
-      "loss": 0.9863,
+      "loss": 0.9673,
       "step": 80200
     },
     {
       "epoch": 240.87,
-      "grad_norm": 5.956753730773926,
+      "grad_norm": 7.737615585327148,
       "learning_rate": 1.9709709709709714e-06,
-      "loss": 0.8941,
+      "loss": 0.9089,
       "step": 80210
     },
     {
       "epoch": 240.9,
-      "grad_norm": 6.1411871910095215,
+      "grad_norm": 6.634458541870117,
       "learning_rate": 1.96996996996997e-06,
-      "loss": 0.9221,
+      "loss": 0.9217,
       "step": 80220
     },
     {
       "epoch": 240.93,
-      "grad_norm": 7.25338077545166,
+      "grad_norm": 7.342255592346191,
       "learning_rate": 1.968968968968969e-06,
-      "loss": 0.9406,
+      "loss": 0.9355,
       "step": 80230
     },
     {
       "epoch": 240.96,
-      "grad_norm": 6.4250807762146,
+      "grad_norm": 7.3768310546875,
       "learning_rate": 1.967967967967968e-06,
-      "loss": 0.8358,
+      "loss": 0.8373,
       "step": 80240
     },
     {
       "epoch": 240.99,
-      "grad_norm": 7.490467548370361,
+      "grad_norm": 9.435381889343262,
       "learning_rate": 1.966966966966967e-06,
-      "loss": 0.9076,
+      "loss": 0.8856,
       "step": 80250
     },
     {
       "epoch": 241.0,
-      "eval_accuracy": 0.8223,
-      "eval_loss": 0.5892135500907898,
-      "eval_runtime": 5.5065,
-      "eval_samples_per_second": 1816.027,
-      "eval_steps_per_second": 7.264,
+      "eval_accuracy": 0.8245,
+      "eval_loss": 0.5886385440826416,
+      "eval_runtime": 5.3896,
+      "eval_samples_per_second": 1855.429,
+      "eval_steps_per_second": 7.422,
       "step": 80253
     },
     {
       "epoch": 241.02,
-      "grad_norm": 6.66855525970459,
+      "grad_norm": 8.041311264038086,
       "learning_rate": 1.965965965965966e-06,
-      "loss": 0.9745,
+      "loss": 0.9987,
       "step": 80260
     },
     {
       "epoch": 241.05,
-      "grad_norm": 8.966917037963867,
+      "grad_norm": 11.787075996398926,
       "learning_rate": 1.964964964964965e-06,
-      "loss": 0.9637,
+      "loss": 0.9614,
       "step": 80270
     },
     {
       "epoch": 241.08,
-      "grad_norm": 6.302736282348633,
+      "grad_norm": 15.151451110839844,
       "learning_rate": 1.963963963963964e-06,
-      "loss": 0.8987,
+      "loss": 0.8934,
       "step": 80280
     },
     {
       "epoch": 241.11,
-      "grad_norm": 8.333646774291992,
+      "grad_norm": 8.792181968688965,
       "learning_rate": 1.962962962962963e-06,
-      "loss": 0.8979,
+      "loss": 0.8876,
       "step": 80290
     },
     {
       "epoch": 241.14,
-      "grad_norm": 5.3812384605407715,
+      "grad_norm": 5.28138542175293,
       "learning_rate": 1.961961961961962e-06,
-      "loss": 0.9039,
+      "loss": 0.8951,
       "step": 80300
     },
     {
       "epoch": 241.17,
-      "grad_norm": 7.015513896942139,
+      "grad_norm": 6.877613544464111,
       "learning_rate": 1.960960960960961e-06,
-      "loss": 0.9343,
+      "loss": 0.9344,
       "step": 80310
     },
     {
       "epoch": 241.2,
-      "grad_norm": 4.664414882659912,
+      "grad_norm": 7.688151836395264,
       "learning_rate": 1.9599599599599602e-06,
-      "loss": 0.907,
+      "loss": 0.9035,
       "step": 80320
     },
     {
       "epoch": 241.23,
-      "grad_norm": 9.074508666992188,
+      "grad_norm": 11.282307624816895,
       "learning_rate": 1.9589589589589592e-06,
-      "loss": 0.9029,
+      "loss": 0.9075,
       "step": 80330
     },
     {
       "epoch": 241.26,
-      "grad_norm": 6.875227451324463,
+      "grad_norm": 8.2286376953125,
       "learning_rate": 1.957957957957958e-06,
-      "loss": 0.9227,
+      "loss": 0.912,
       "step": 80340
     },
     {
       "epoch": 241.29,
-      "grad_norm": 7.568996429443359,
+      "grad_norm": 6.711334228515625,
       "learning_rate": 1.9569569569569573e-06,
-      "loss": 0.9222,
+      "loss": 0.9201,
       "step": 80350
     },
     {
       "epoch": 241.32,
-      "grad_norm": 9.566666603088379,
+      "grad_norm": 8.006731986999512,
       "learning_rate": 1.955955955955956e-06,
-      "loss": 0.8714,
+      "loss": 0.8746,
       "step": 80360
     },
     {
       "epoch": 241.35,
-      "grad_norm": 7.9757513999938965,
+      "grad_norm": 7.443068504333496,
       "learning_rate": 1.9549549549549553e-06,
-      "loss": 0.9049,
+      "loss": 0.9074,
       "step": 80370
     },
     {
       "epoch": 241.38,
-      "grad_norm": 9.784894943237305,
+      "grad_norm": 10.597254753112793,
       "learning_rate": 1.953953953953954e-06,
-      "loss": 0.9918,
+      "loss": 0.9782,
       "step": 80380
     },
     {
       "epoch": 241.41,
-      "grad_norm": 6.164865016937256,
+      "grad_norm": 7.066493511199951,
       "learning_rate": 1.952952952952953e-06,
-      "loss": 0.9011,
+      "loss": 0.9081,
       "step": 80390
     },
     {
       "epoch": 241.44,
-      "grad_norm": 5.300997734069824,
+      "grad_norm": 5.303035736083984,
       "learning_rate": 1.951951951951952e-06,
-      "loss": 0.9048,
+      "loss": 0.89,
       "step": 80400
     },
     {
       "epoch": 241.47,
-      "grad_norm": 4.9040727615356445,
+      "grad_norm": 7.186089992523193,
       "learning_rate": 1.950950950950951e-06,
-      "loss": 0.9253,
+      "loss": 0.9232,
       "step": 80410
     },
     {
       "epoch": 241.5,
-      "grad_norm": 5.924160003662109,
+      "grad_norm": 6.2192864418029785,
       "learning_rate": 1.9499499499499504e-06,
-      "loss": 0.9046,
+      "loss": 0.9192,
       "step": 80420
     },
     {
       "epoch": 241.53,
-      "grad_norm": 6.614039897918701,
+      "grad_norm": 6.97580099105835,
       "learning_rate": 1.948948948948949e-06,
-      "loss": 0.914,
+      "loss": 0.9159,
       "step": 80430
     },
     {
       "epoch": 241.56,
-      "grad_norm": 6.282723903656006,
+      "grad_norm": 7.01101541519165,
       "learning_rate": 1.947947947947948e-06,
-      "loss": 0.8846,
+      "loss": 0.8878,
       "step": 80440
     },
     {
       "epoch": 241.59,
-      "grad_norm": 6.587308406829834,
+      "grad_norm": 8.588425636291504,
       "learning_rate": 1.946946946946947e-06,
-      "loss": 0.9223,
+      "loss": 0.9209,
       "step": 80450
     },
     {
       "epoch": 241.62,
-      "grad_norm": 9.335346221923828,
+      "grad_norm": 6.8470354080200195,
       "learning_rate": 1.945945945945946e-06,
-      "loss": 0.8762,
+      "loss": 0.872,
       "step": 80460
     },
     {
       "epoch": 241.65,
-      "grad_norm": 8.374478340148926,
+      "grad_norm": 6.474335670471191,
       "learning_rate": 1.944944944944945e-06,
-      "loss": 0.8954,
+      "loss": 0.9028,
       "step": 80470
     },
     {
       "epoch": 241.68,
-      "grad_norm": 6.552872657775879,
+      "grad_norm": 6.742740154266357,
       "learning_rate": 1.943943943943944e-06,
-      "loss": 0.9318,
+      "loss": 0.9363,
       "step": 80480
     },
     {
       "epoch": 241.71,
-      "grad_norm": 8.56727123260498,
+      "grad_norm": 12.094325065612793,
       "learning_rate": 1.942942942942943e-06,
-      "loss": 0.9089,
+      "loss": 0.9219,
       "step": 80490
     },
     {
       "epoch": 241.74,
-      "grad_norm": 7.4070611000061035,
+      "grad_norm": 7.015802383422852,
       "learning_rate": 1.941941941941942e-06,
-      "loss": 0.9652,
+      "loss": 0.9573,
       "step": 80500
     },
     {
       "epoch": 241.77,
-      "grad_norm": 6.457737922668457,
+      "grad_norm": 7.739763259887695,
       "learning_rate": 1.940940940940941e-06,
-      "loss": 0.9581,
+      "loss": 0.9553,
       "step": 80510
     },
     {
       "epoch": 241.8,
-      "grad_norm": 6.278329849243164,
+      "grad_norm": 6.3842453956604,
       "learning_rate": 1.93993993993994e-06,
-      "loss": 0.9674,
+      "loss": 0.9702,
       "step": 80520
     },
     {
       "epoch": 241.83,
-      "grad_norm": 8.636343955993652,
+      "grad_norm": 9.312710762023926,
       "learning_rate": 1.938938938938939e-06,
-      "loss": 0.8819,
+      "loss": 0.8766,
       "step": 80530
     },
     {
       "epoch": 241.86,
-      "grad_norm": 7.093334197998047,
+      "grad_norm": 8.664167404174805,
       "learning_rate": 1.937937937937938e-06,
-      "loss": 0.9393,
+      "loss": 0.9451,
       "step": 80540
     },
     {
       "epoch": 241.89,
-      "grad_norm": 9.108506202697754,
+      "grad_norm": 8.704667091369629,
       "learning_rate": 1.9369369369369372e-06,
-      "loss": 0.8752,
+      "loss": 0.8766,
       "step": 80550
     },
     {
       "epoch": 241.92,
-      "grad_norm": 6.697867393493652,
+      "grad_norm": 6.075462341308594,
       "learning_rate": 1.9359359359359362e-06,
-      "loss": 0.8808,
+      "loss": 0.8791,
       "step": 80560
     },
     {
       "epoch": 241.95,
-      "grad_norm": 6.440984725952148,
+      "grad_norm": 7.756100654602051,
       "learning_rate": 1.934934934934935e-06,
-      "loss": 0.9103,
+      "loss": 0.9022,
       "step": 80570
     },
     {
       "epoch": 241.98,
-      "grad_norm": 6.692206382751465,
+      "grad_norm": 5.190834045410156,
       "learning_rate": 1.9339339339339343e-06,
-      "loss": 0.8956,
+      "loss": 0.8914,
       "step": 80580
     },
     {
       "epoch": 242.0,
-      "eval_accuracy": 0.8264,
-      "eval_loss": 0.5717228055000305,
-      "eval_runtime": 5.2828,
-      "eval_samples_per_second": 1892.933,
-      "eval_steps_per_second": 7.572,
+      "eval_accuracy": 0.8296,
+      "eval_loss": 0.5715660452842712,
+      "eval_runtime": 5.3682,
+      "eval_samples_per_second": 1862.822,
+      "eval_steps_per_second": 7.451,
       "step": 80586
     },
     {
       "epoch": 242.01,
-      "grad_norm": 11.158843040466309,
+      "grad_norm": 8.474370956420898,
       "learning_rate": 1.932932932932933e-06,
-      "loss": 1.0486,
+      "loss": 1.0406,
       "step": 80590
     },
     {
       "epoch": 242.04,
-      "grad_norm": 6.238728046417236,
+      "grad_norm": 7.52123498916626,
       "learning_rate": 1.9319319319319323e-06,
-      "loss": 0.8884,
+      "loss": 0.89,
       "step": 80600
     },
     {
       "epoch": 242.07,
-      "grad_norm": 6.116739273071289,
+      "grad_norm": 6.286978244781494,
       "learning_rate": 1.930930930930931e-06,
-      "loss": 0.8849,
+      "loss": 0.8762,
       "step": 80610
     },
     {
       "epoch": 242.1,
-      "grad_norm": 6.454004287719727,
+      "grad_norm": 7.721216201782227,
       "learning_rate": 1.92992992992993e-06,
-      "loss": 0.8365,
+      "loss": 0.8409,
       "step": 80620
     },
     {
       "epoch": 242.13,
-      "grad_norm": 4.085811614990234,
+      "grad_norm": 4.464795112609863,
       "learning_rate": 1.928928928928929e-06,
-      "loss": 0.9067,
+      "loss": 0.9112,
       "step": 80630
     },
     {
       "epoch": 242.16,
-      "grad_norm": 4.708978176116943,
+      "grad_norm": 5.348705768585205,
       "learning_rate": 1.927927927927928e-06,
-      "loss": 0.9604,
+      "loss": 0.9545,
       "step": 80640
     },
     {
       "epoch": 242.19,
-      "grad_norm": 5.686951160430908,
+      "grad_norm": 6.752450942993164,
       "learning_rate": 1.926926926926927e-06,
-      "loss": 0.9242,
+      "loss": 0.9214,
       "step": 80650
     },
     {
       "epoch": 242.22,
-      "grad_norm": 8.256346702575684,
+      "grad_norm": 10.673188209533691,
       "learning_rate": 1.925925925925926e-06,
-      "loss": 0.9684,
+      "loss": 0.9641,
       "step": 80660
     },
     {
       "epoch": 242.25,
-      "grad_norm": 5.142011642456055,
+      "grad_norm": 6.469410419464111,
       "learning_rate": 1.924924924924925e-06,
-      "loss": 0.8762,
+      "loss": 0.8741,
       "step": 80670
     },
     {
       "epoch": 242.28,
-      "grad_norm": 7.740803241729736,
+      "grad_norm": 8.783241271972656,
       "learning_rate": 1.923923923923924e-06,
-      "loss": 0.9187,
+      "loss": 0.9174,
       "step": 80680
     },
     {
       "epoch": 242.31,
-      "grad_norm": 7.330544471740723,
+      "grad_norm": 7.996604919433594,
       "learning_rate": 1.922922922922923e-06,
-      "loss": 0.903,
+      "loss": 0.8996,
       "step": 80690
     },
     {
       "epoch": 242.34,
-      "grad_norm": 5.593299865722656,
+      "grad_norm": 7.1548895835876465,
       "learning_rate": 1.921921921921922e-06,
-      "loss": 0.8794,
+      "loss": 0.8715,
       "step": 80700
     },
     {
       "epoch": 242.37,
-      "grad_norm": 5.004030704498291,
+      "grad_norm": 5.77497673034668,
       "learning_rate": 1.920920920920921e-06,
-      "loss": 0.9448,
+      "loss": 0.9371,
       "step": 80710
     },
     {
       "epoch": 242.4,
-      "grad_norm": 6.923647880554199,
+      "grad_norm": 6.683837890625,
       "learning_rate": 1.91991991991992e-06,
-      "loss": 0.9301,
+      "loss": 0.9323,
       "step": 80720
     },
     {
       "epoch": 242.43,
-      "grad_norm": 9.46280574798584,
+      "grad_norm": 9.809531211853027,
       "learning_rate": 1.918918918918919e-06,
-      "loss": 0.9517,
+      "loss": 0.9475,
       "step": 80730
     },
     {
       "epoch": 242.46,
-      "grad_norm": 5.602623462677002,
+      "grad_norm": 6.940591812133789,
       "learning_rate": 1.917917917917918e-06,
-      "loss": 0.9187,
+      "loss": 0.9142,
       "step": 80740
     },
     {
       "epoch": 242.49,
-      "grad_norm": 6.234764575958252,
+      "grad_norm": 7.204885482788086,
       "learning_rate": 1.916916916916917e-06,
-      "loss": 0.9169,
+      "loss": 0.9174,
       "step": 80750
     },
     {
       "epoch": 242.52,
-      "grad_norm": 5.823460578918457,
+      "grad_norm": 6.8960795402526855,
       "learning_rate": 1.915915915915916e-06,
-      "loss": 0.912,
+      "loss": 0.9114,
       "step": 80760
     },
     {
       "epoch": 242.55,
-      "grad_norm": 7.1544294357299805,
+      "grad_norm": 7.126483917236328,
       "learning_rate": 1.914914914914915e-06,
-      "loss": 0.9227,
+      "loss": 0.9189,
       "step": 80770
     },
     {
       "epoch": 242.58,
-      "grad_norm": 5.034264087677002,
+      "grad_norm": 6.288402080535889,
       "learning_rate": 1.9139139139139142e-06,
-      "loss": 0.985,
+      "loss": 0.9885,
       "step": 80780
     },
     {
       "epoch": 242.61,
-      "grad_norm": 7.190291881561279,
+      "grad_norm": 11.23485279083252,
       "learning_rate": 1.9129129129129132e-06,
-      "loss": 0.9091,
+      "loss": 0.9159,
       "step": 80790
     },
     {
       "epoch": 242.64,
-      "grad_norm": 5.796281814575195,
+      "grad_norm": 7.938747882843018,
       "learning_rate": 1.911911911911912e-06,
-      "loss": 1.008,
+      "loss": 1.006,
       "step": 80800
     },
     {
       "epoch": 242.67,
-      "grad_norm": 5.501676082611084,
+      "grad_norm": 5.600336074829102,
       "learning_rate": 1.9109109109109113e-06,
-      "loss": 0.8918,
+      "loss": 0.8874,
       "step": 80810
     },
     {
       "epoch": 242.7,
-      "grad_norm": 6.532558917999268,
+      "grad_norm": 5.663583278656006,
       "learning_rate": 1.90990990990991e-06,
-      "loss": 0.8549,
+      "loss": 0.8548,
       "step": 80820
     },
     {
       "epoch": 242.73,
-      "grad_norm": 8.113168716430664,
+      "grad_norm": 6.393566608428955,
       "learning_rate": 1.9089089089089093e-06,
-      "loss": 0.9218,
+      "loss": 0.9122,
       "step": 80830
     },
     {
       "epoch": 242.76,
-      "grad_norm": 7.722615718841553,
+      "grad_norm": 8.965408325195312,
       "learning_rate": 1.907907907907908e-06,
-      "loss": 0.8778,
+      "loss": 0.8735,
       "step": 80840
     },
     {
       "epoch": 242.79,
-      "grad_norm": 6.12741756439209,
+      "grad_norm": 6.89486837387085,
       "learning_rate": 1.9069069069069071e-06,
-      "loss": 0.9329,
+      "loss": 0.9255,
       "step": 80850
     },
     {
       "epoch": 242.82,
-      "grad_norm": 8.127429008483887,
+      "grad_norm": 6.120883941650391,
       "learning_rate": 1.905905905905906e-06,
-      "loss": 0.9129,
+      "loss": 0.9142,
       "step": 80860
     },
     {
       "epoch": 242.85,
-      "grad_norm": 7.210639953613281,
+      "grad_norm": 6.9525957107543945,
       "learning_rate": 1.9049049049049052e-06,
-      "loss": 0.9592,
+      "loss": 0.9577,
       "step": 80870
     },
     {
       "epoch": 242.88,
-      "grad_norm": 6.179612636566162,
+      "grad_norm": 5.355838775634766,
       "learning_rate": 1.903903903903904e-06,
-      "loss": 0.8488,
+      "loss": 0.8336,
       "step": 80880
     },
     {
       "epoch": 242.91,
-      "grad_norm": 5.938989162445068,
+      "grad_norm": 6.443424224853516,
       "learning_rate": 1.902902902902903e-06,
-      "loss": 0.8908,
+      "loss": 0.8972,
       "step": 80890
     },
     {
       "epoch": 242.94,
-      "grad_norm": 6.968057155609131,
+      "grad_norm": 9.214920043945312,
       "learning_rate": 1.901901901901902e-06,
-      "loss": 1.0053,
+      "loss": 1.0052,
       "step": 80900
     },
     {
       "epoch": 242.97,
-      "grad_norm": 6.436769962310791,
+      "grad_norm": 6.439895153045654,
       "learning_rate": 1.900900900900901e-06,
-      "loss": 0.8968,
+      "loss": 0.8951,
       "step": 80910
     },
     {
       "epoch": 243.0,
-      "eval_accuracy": 0.8238,
-      "eval_loss": 0.5746627449989319,
-      "eval_runtime": 5.3895,
-      "eval_samples_per_second": 1855.452,
-      "eval_steps_per_second": 7.422,
+      "eval_accuracy": 0.8254,
+      "eval_loss": 0.5734505653381348,
+      "eval_runtime": 5.4531,
+      "eval_samples_per_second": 1833.814,
+      "eval_steps_per_second": 7.335,
       "step": 80919
     },
     {
       "epoch": 243.0,
-      "grad_norm": 7.5595927238464355,
+      "grad_norm": 5.259466648101807,
       "learning_rate": 1.8998998998998998e-06,
-      "loss": 0.9602,
+      "loss": 0.9483,
       "step": 80920
     },
     {
       "epoch": 243.03,
-      "grad_norm": 8.287642478942871,
+      "grad_norm": 7.675260066986084,
       "learning_rate": 1.898898898898899e-06,
-      "loss": 0.9406,
+      "loss": 0.934,
       "step": 80930
     },
     {
       "epoch": 243.06,
-      "grad_norm": 6.956298828125,
+      "grad_norm": 6.203985691070557,
       "learning_rate": 1.897897897897898e-06,
-      "loss": 0.9446,
+      "loss": 0.9452,
       "step": 80940
     },
     {
       "epoch": 243.09,
-      "grad_norm": 6.1405839920043945,
+      "grad_norm": 6.996640682220459,
       "learning_rate": 1.896896896896897e-06,
-      "loss": 0.924,
+      "loss": 0.9246,
       "step": 80950
     },
     {
       "epoch": 243.12,
-      "grad_norm": 14.44858169555664,
+      "grad_norm": 8.629411697387695,
       "learning_rate": 1.8958958958958961e-06,
-      "loss": 0.9337,
+      "loss": 0.9328,
       "step": 80960
     },
     {
       "epoch": 243.15,
-      "grad_norm": 5.964333534240723,
+      "grad_norm": 7.127389907836914,
       "learning_rate": 1.894894894894895e-06,
-      "loss": 0.9786,
+      "loss": 0.9769,
       "step": 80970
     },
     {
       "epoch": 243.18,
-      "grad_norm": 11.090689659118652,
+      "grad_norm": 13.431004524230957,
       "learning_rate": 1.8938938938938942e-06,
-      "loss": 0.8791,
+      "loss": 0.8863,
       "step": 80980
     },
     {
       "epoch": 243.21,
-      "grad_norm": 7.755176067352295,
+      "grad_norm": 4.982303142547607,
       "learning_rate": 1.892892892892893e-06,
-      "loss": 0.8647,
+      "loss": 0.8764,
       "step": 80990
     },
     {
       "epoch": 243.24,
-      "grad_norm": 6.860938549041748,
+      "grad_norm": 5.5991740226745605,
       "learning_rate": 1.8918918918918922e-06,
-      "loss": 0.9399,
+      "loss": 0.9428,
       "step": 81000
     },
     {
       "epoch": 243.27,
-      "grad_norm": 6.2509942054748535,
+      "grad_norm": 6.934983730316162,
       "learning_rate": 1.890890890890891e-06,
-      "loss": 0.9556,
+      "loss": 0.9665,
       "step": 81010
     },
     {
       "epoch": 243.3,
-      "grad_norm": 5.545879364013672,
+      "grad_norm": 5.1035542488098145,
       "learning_rate": 1.8898898898898902e-06,
-      "loss": 0.8622,
+      "loss": 0.8616,
       "step": 81020
     },
     {
       "epoch": 243.33,
-      "grad_norm": 9.162583351135254,
+      "grad_norm": 6.914124011993408,
       "learning_rate": 1.888888888888889e-06,
-      "loss": 0.9044,
+      "loss": 0.9041,
       "step": 81030
     },
     {
       "epoch": 243.36,
-      "grad_norm": 7.526141166687012,
+      "grad_norm": 9.972723960876465,
       "learning_rate": 1.887887887887888e-06,
-      "loss": 0.917,
+      "loss": 0.9142,
       "step": 81040
     },
     {
       "epoch": 243.39,
-      "grad_norm": 5.112117290496826,
+      "grad_norm": 6.8643412590026855,
       "learning_rate": 1.886886886886887e-06,
-      "loss": 0.9557,
+      "loss": 0.9466,
       "step": 81050
     },
     {
       "epoch": 243.42,
-      "grad_norm": 5.738952159881592,
+      "grad_norm": 5.010369777679443,
       "learning_rate": 1.885885885885886e-06,
-      "loss": 0.9012,
+      "loss": 0.9014,
       "step": 81060
     },
     {
       "epoch": 243.45,
-      "grad_norm": 4.738990306854248,
+      "grad_norm": 5.320896148681641,
       "learning_rate": 1.884884884884885e-06,
-      "loss": 0.8944,
+      "loss": 0.8992,
       "step": 81070
     },
     {
       "epoch": 243.48,
-      "grad_norm": 5.660656452178955,
+      "grad_norm": 5.73524284362793,
       "learning_rate": 1.8838838838838841e-06,
-      "loss": 0.9588,
+      "loss": 0.9615,
       "step": 81080
     },
     {
       "epoch": 243.51,
-      "grad_norm": 6.136982440948486,
+      "grad_norm": 7.029303073883057,
       "learning_rate": 1.882882882882883e-06,
-      "loss": 0.9297,
+      "loss": 0.9337,
       "step": 81090
     },
     {
       "epoch": 243.54,
-      "grad_norm": 7.463130474090576,
+      "grad_norm": 8.153009414672852,
       "learning_rate": 1.8818818818818822e-06,
-      "loss": 0.9305,
+      "loss": 0.9382,
       "step": 81100
     },
     {
       "epoch": 243.57,
-      "grad_norm": 5.955556392669678,
+      "grad_norm": 7.937097072601318,
       "learning_rate": 1.880880880880881e-06,
-      "loss": 0.9359,
+      "loss": 0.9357,
       "step": 81110
     },
     {
       "epoch": 243.6,
-      "grad_norm": 10.524014472961426,
+      "grad_norm": 10.129352569580078,
       "learning_rate": 1.87987987987988e-06,
-      "loss": 0.8293,
+      "loss": 0.8335,
       "step": 81120
     },
     {
       "epoch": 243.63,
-      "grad_norm": 7.580707550048828,
+      "grad_norm": 10.574562072753906,
       "learning_rate": 1.878878878878879e-06,
-      "loss": 0.9793,
+      "loss": 0.987,
       "step": 81130
     },
     {
       "epoch": 243.66,
-      "grad_norm": 5.241048812866211,
+      "grad_norm": 5.055078983306885,
       "learning_rate": 1.877877877877878e-06,
-      "loss": 0.8601,
+      "loss": 0.8605,
       "step": 81140
     },
     {
       "epoch": 243.69,
-      "grad_norm": 4.5927252769470215,
+      "grad_norm": 4.984067916870117,
       "learning_rate": 1.8768768768768768e-06,
-      "loss": 0.8635,
+      "loss": 0.868,
       "step": 81150
     },
     {
       "epoch": 243.72,
-      "grad_norm": 8.852167129516602,
+      "grad_norm": 9.038091659545898,
       "learning_rate": 1.875875875875876e-06,
-      "loss": 0.9648,
+      "loss": 0.9563,
       "step": 81160
     },
     {
       "epoch": 243.75,
-      "grad_norm": 12.651496887207031,
+      "grad_norm": 11.686128616333008,
       "learning_rate": 1.8748748748748749e-06,
-      "loss": 0.9865,
+      "loss": 0.9874,
       "step": 81170
     },
     {
       "epoch": 243.78,
-      "grad_norm": 7.7990803718566895,
+      "grad_norm": 8.563899993896484,
       "learning_rate": 1.873873873873874e-06,
-      "loss": 0.9241,
+      "loss": 0.9168,
       "step": 81180
     },
     {
       "epoch": 243.81,
-      "grad_norm": 10.322568893432617,
+      "grad_norm": 13.635272026062012,
       "learning_rate": 1.872872872872873e-06,
-      "loss": 0.9081,
+      "loss": 0.9115,
       "step": 81190
     },
     {
       "epoch": 243.84,
-      "grad_norm": 10.13730239868164,
+      "grad_norm": 7.072484493255615,
       "learning_rate": 1.871871871871872e-06,
-      "loss": 0.8784,
+      "loss": 0.8742,
       "step": 81200
     },
     {
       "epoch": 243.87,
-      "grad_norm": 8.143447875976562,
+      "grad_norm": 7.543489456176758,
       "learning_rate": 1.8708708708708712e-06,
-      "loss": 0.9207,
+      "loss": 0.9184,
       "step": 81210
     },
     {
       "epoch": 243.9,
-      "grad_norm": 8.41710376739502,
+      "grad_norm": 7.284512519836426,
       "learning_rate": 1.86986986986987e-06,
-      "loss": 0.9364,
+      "loss": 0.9408,
       "step": 81220
     },
     {
       "epoch": 243.93,
-      "grad_norm": 7.141921520233154,
+      "grad_norm": 8.365551948547363,
       "learning_rate": 1.8688688688688692e-06,
-      "loss": 0.9809,
+      "loss": 0.9812,
       "step": 81230
     },
     {
       "epoch": 243.96,
-      "grad_norm": 5.555495262145996,
+      "grad_norm": 6.900781631469727,
       "learning_rate": 1.867867867867868e-06,
-      "loss": 0.9353,
+      "loss": 0.9319,
       "step": 81240
     },
     {
       "epoch": 243.99,
-      "grad_norm": 6.869070053100586,
+      "grad_norm": 7.516140937805176,
       "learning_rate": 1.8668668668668672e-06,
-      "loss": 0.9055,
+      "loss": 0.9073,
       "step": 81250
     },
     {
       "epoch": 244.0,
-      "eval_accuracy": 0.8246,
-      "eval_loss": 0.574576735496521,
-      "eval_runtime": 5.3896,
-      "eval_samples_per_second": 1855.429,
-      "eval_steps_per_second": 7.422,
+      "eval_accuracy": 0.8281,
+      "eval_loss": 0.5737076997756958,
+      "eval_runtime": 5.3375,
+      "eval_samples_per_second": 1873.53,
+      "eval_steps_per_second": 7.494,
       "step": 81252
     },
     {
       "epoch": 244.02,
-      "grad_norm": 6.233569622039795,
+      "grad_norm": 6.166440486907959,
       "learning_rate": 1.865865865865866e-06,
-      "loss": 1.0357,
+      "loss": 1.0305,
       "step": 81260
     },
     {
       "epoch": 244.05,
-      "grad_norm": 7.431661605834961,
+      "grad_norm": 9.739580154418945,
       "learning_rate": 1.864864864864865e-06,
-      "loss": 0.9271,
+      "loss": 0.9279,
       "step": 81270
     },
     {
       "epoch": 244.08,
-      "grad_norm": 7.358098983764648,
+      "grad_norm": 8.650444984436035,
       "learning_rate": 1.863863863863864e-06,
-      "loss": 0.9142,
+      "loss": 0.9094,
       "step": 81280
     },
     {
       "epoch": 244.11,
-      "grad_norm": 7.3902587890625,
+      "grad_norm": 6.224049091339111,
       "learning_rate": 1.862862862862863e-06,
-      "loss": 0.9107,
+      "loss": 0.9016,
       "step": 81290
     },
     {
       "epoch": 244.14,
-      "grad_norm": 7.644387722015381,
+      "grad_norm": 9.317683219909668,
       "learning_rate": 1.861861861861862e-06,
-      "loss": 0.9009,
+      "loss": 0.8844,
       "step": 81300
     },
     {
       "epoch": 244.17,
-      "grad_norm": 9.713181495666504,
+      "grad_norm": 8.04351806640625,
       "learning_rate": 1.8608608608608611e-06,
-      "loss": 0.9277,
+      "loss": 0.9201,
       "step": 81310
     },
     {
       "epoch": 244.2,
-      "grad_norm": 6.214000225067139,
+      "grad_norm": 5.035552024841309,
       "learning_rate": 1.85985985985986e-06,
-      "loss": 0.8782,
+      "loss": 0.8775,
       "step": 81320
     },
     {
       "epoch": 244.23,
-      "grad_norm": 6.11535120010376,
+      "grad_norm": 7.82926082611084,
       "learning_rate": 1.8588588588588592e-06,
-      "loss": 0.9011,
+      "loss": 0.8923,
       "step": 81330
     },
     {
       "epoch": 244.26,
-      "grad_norm": 5.244883060455322,
+      "grad_norm": 7.26019287109375,
       "learning_rate": 1.857857857857858e-06,
-      "loss": 0.9262,
+      "loss": 0.9182,
       "step": 81340
     },
     {
       "epoch": 244.29,
-      "grad_norm": 9.498286247253418,
+      "grad_norm": 8.793850898742676,
       "learning_rate": 1.856856856856857e-06,
-      "loss": 0.8709,
+      "loss": 0.8783,
       "step": 81350
     },
     {
       "epoch": 244.32,
-      "grad_norm": 9.456997871398926,
+      "grad_norm": 9.169475555419922,
       "learning_rate": 1.855855855855856e-06,
-      "loss": 0.8475,
+      "loss": 0.839,
       "step": 81360
     },
     {
       "epoch": 244.35,
-      "grad_norm": 6.024025917053223,
+      "grad_norm": 6.969535827636719,
       "learning_rate": 1.854854854854855e-06,
-      "loss": 0.949,
+      "loss": 0.9348,
       "step": 81370
     },
     {
       "epoch": 244.38,
-      "grad_norm": 8.432114601135254,
+      "grad_norm": 10.292616844177246,
       "learning_rate": 1.8538538538538538e-06,
-      "loss": 0.893,
+      "loss": 0.8908,
       "step": 81380
     },
     {
       "epoch": 244.41,
-      "grad_norm": 5.089940071105957,
+      "grad_norm": 6.183272361755371,
       "learning_rate": 1.852852852852853e-06,
-      "loss": 0.9902,
+      "loss": 0.9903,
       "step": 81390
     },
     {
       "epoch": 244.44,
-      "grad_norm": 5.170373916625977,
+      "grad_norm": 5.12890100479126,
       "learning_rate": 1.8518518518518519e-06,
-      "loss": 0.9489,
+      "loss": 0.9561,
       "step": 81400
     },
     {
       "epoch": 244.47,
-      "grad_norm": 6.586550712585449,
+      "grad_norm": 5.322021961212158,
       "learning_rate": 1.850850850850851e-06,
-      "loss": 0.8567,
+      "loss": 0.861,
       "step": 81410
     },
     {
       "epoch": 244.5,
-      "grad_norm": 11.989113807678223,
+      "grad_norm": 5.046477317810059,
       "learning_rate": 1.84984984984985e-06,
-      "loss": 0.9275,
+      "loss": 0.9171,
       "step": 81420
     },
     {
       "epoch": 244.53,
-      "grad_norm": 7.6232194900512695,
+      "grad_norm": 8.3214111328125,
       "learning_rate": 1.848848848848849e-06,
-      "loss": 0.8582,
+      "loss": 0.8629,
       "step": 81430
     },
     {
       "epoch": 244.56,
-      "grad_norm": 5.409687519073486,
+      "grad_norm": 6.316959857940674,
       "learning_rate": 1.847847847847848e-06,
-      "loss": 0.8723,
+      "loss": 0.8685,
       "step": 81440
     },
     {
       "epoch": 244.59,
-      "grad_norm": 7.97031307220459,
+      "grad_norm": 5.867912769317627,
       "learning_rate": 1.846846846846847e-06,
-      "loss": 0.9726,
+      "loss": 0.9649,
       "step": 81450
     },
     {
       "epoch": 244.62,
-      "grad_norm": 5.558863639831543,
+      "grad_norm": 4.997313499450684,
       "learning_rate": 1.8458458458458458e-06,
-      "loss": 0.9052,
+      "loss": 0.8953,
       "step": 81460
     },
     {
       "epoch": 244.65,
-      "grad_norm": 7.330705165863037,
+      "grad_norm": 6.551809787750244,
       "learning_rate": 1.844844844844845e-06,
-      "loss": 0.8915,
+      "loss": 0.8966,
       "step": 81470
     },
     {
       "epoch": 244.68,
-      "grad_norm": 5.633613586425781,
+      "grad_norm": 5.351194858551025,
       "learning_rate": 1.8438438438438442e-06,
-      "loss": 0.9607,
+      "loss": 0.9594,
       "step": 81480
     },
     {
       "epoch": 244.71,
-      "grad_norm": 5.915164947509766,
+      "grad_norm": 5.820681571960449,
       "learning_rate": 1.842842842842843e-06,
-      "loss": 0.8904,
+      "loss": 0.8928,
       "step": 81490
     },
     {
       "epoch": 244.74,
-      "grad_norm": 6.335549354553223,
+      "grad_norm": 6.153966903686523,
       "learning_rate": 1.841841841841842e-06,
-      "loss": 0.981,
+      "loss": 0.9862,
       "step": 81500
     },
     {
       "epoch": 244.77,
-      "grad_norm": 7.6401214599609375,
+      "grad_norm": 6.743399620056152,
       "learning_rate": 1.840840840840841e-06,
-      "loss": 0.9387,
+      "loss": 0.95,
       "step": 81510
     },
     {
       "epoch": 244.8,
-      "grad_norm": 5.262809753417969,
+      "grad_norm": 5.588888645172119,
       "learning_rate": 1.83983983983984e-06,
-      "loss": 0.8671,
+      "loss": 0.8676,
       "step": 81520
     },
     {
       "epoch": 244.83,
-      "grad_norm": 5.030666351318359,
+      "grad_norm": 5.696310520172119,
       "learning_rate": 1.838838838838839e-06,
-      "loss": 0.8876,
+      "loss": 0.8988,
       "step": 81530
     },
     {
       "epoch": 244.86,
-      "grad_norm": 7.382626056671143,
+      "grad_norm": 6.475600242614746,
       "learning_rate": 1.8378378378378381e-06,
-      "loss": 0.897,
+      "loss": 0.8915,
       "step": 81540
     },
     {
       "epoch": 244.89,
-      "grad_norm": 7.485737323760986,
+      "grad_norm": 7.570889472961426,
       "learning_rate": 1.836836836836837e-06,
-      "loss": 0.9185,
+      "loss": 0.9172,
       "step": 81550
     },
     {
       "epoch": 244.92,
-      "grad_norm": 6.6439900398254395,
+      "grad_norm": 7.745683193206787,
       "learning_rate": 1.8358358358358362e-06,
-      "loss": 0.9296,
+      "loss": 0.9283,
       "step": 81560
     },
     {
       "epoch": 244.95,
-      "grad_norm": 6.76557731628418,
+      "grad_norm": 6.401723861694336,
       "learning_rate": 1.834834834834835e-06,
-      "loss": 0.9158,
+      "loss": 0.9183,
       "step": 81570
     },
     {
       "epoch": 244.98,
-      "grad_norm": 7.5168914794921875,
+      "grad_norm": 6.619222640991211,
       "learning_rate": 1.833833833833834e-06,
-      "loss": 0.8601,
+      "loss": 0.8557,
       "step": 81580
     },
     {
       "epoch": 245.0,
-      "eval_accuracy": 0.8269,
-      "eval_loss": 0.5728638172149658,
-      "eval_runtime": 5.5803,
-      "eval_samples_per_second": 1792.03,
-      "eval_steps_per_second": 7.168,
+      "eval_accuracy": 0.8292,
+      "eval_loss": 0.5713900327682495,
+      "eval_runtime": 5.2962,
+      "eval_samples_per_second": 1888.154,
+      "eval_steps_per_second": 7.553,
       "step": 81585
     },
     {
       "epoch": 245.02,
-      "grad_norm": 4.977352619171143,
+      "grad_norm": 4.986618518829346,
       "learning_rate": 1.832832832832833e-06,
-      "loss": 0.9714,
+      "loss": 0.9816,
       "step": 81590
     },
     {
       "epoch": 245.05,
-      "grad_norm": 5.108889579772949,
+      "grad_norm": 4.598616600036621,
       "learning_rate": 1.831831831831832e-06,
-      "loss": 0.9181,
+      "loss": 0.9106,
       "step": 81600
     },
     {
       "epoch": 245.08,
-      "grad_norm": 7.4922685623168945,
+      "grad_norm": 10.091522216796875,
       "learning_rate": 1.8308308308308308e-06,
-      "loss": 0.88,
+      "loss": 0.8867,
       "step": 81610
     },
     {
       "epoch": 245.11,
-      "grad_norm": 6.1169843673706055,
+      "grad_norm": 7.8551025390625,
       "learning_rate": 1.82982982982983e-06,
-      "loss": 0.9598,
+      "loss": 0.9485,
       "step": 81620
     },
     {
       "epoch": 245.14,
-      "grad_norm": 5.840083599090576,
+      "grad_norm": 7.1884260177612305,
       "learning_rate": 1.8288288288288289e-06,
-      "loss": 0.9225,
+      "loss": 0.9062,
       "step": 81630
     },
     {
       "epoch": 245.17,
-      "grad_norm": 6.631268501281738,
+      "grad_norm": 6.956891059875488,
       "learning_rate": 1.827827827827828e-06,
-      "loss": 0.8953,
+      "loss": 0.8901,
       "step": 81640
     },
     {
       "epoch": 245.2,
-      "grad_norm": 6.04162073135376,
+      "grad_norm": 4.856355667114258,
       "learning_rate": 1.826826826826827e-06,
-      "loss": 0.9148,
+      "loss": 0.9071,
       "step": 81650
     },
     {
       "epoch": 245.23,
-      "grad_norm": 6.986083507537842,
+      "grad_norm": 8.51058578491211,
       "learning_rate": 1.825825825825826e-06,
-      "loss": 0.93,
+      "loss": 0.9208,
       "step": 81660
     },
     {
       "epoch": 245.26,
-      "grad_norm": 9.670221328735352,
+      "grad_norm": 9.184491157531738,
       "learning_rate": 1.824824824824825e-06,
-      "loss": 0.8809,
+      "loss": 0.8835,
       "step": 81670
     },
     {
       "epoch": 245.29,
-      "grad_norm": 6.576682090759277,
+      "grad_norm": 8.759415626525879,
       "learning_rate": 1.823823823823824e-06,
-      "loss": 0.8347,
+      "loss": 0.8333,
       "step": 81680
     },
     {
       "epoch": 245.32,
-      "grad_norm": 7.9672532081604,
+      "grad_norm": 7.941183090209961,
       "learning_rate": 1.8228228228228228e-06,
-      "loss": 0.8966,
+      "loss": 0.8915,
       "step": 81690
     },
     {
       "epoch": 245.35,
-      "grad_norm": 8.479757308959961,
+      "grad_norm": 6.025416374206543,
       "learning_rate": 1.821821821821822e-06,
-      "loss": 0.9228,
+      "loss": 0.9169,
       "step": 81700
     },
     {
       "epoch": 245.38,
-      "grad_norm": 7.419148921966553,
+      "grad_norm": 7.747071743011475,
       "learning_rate": 1.8208208208208208e-06,
-      "loss": 0.9761,
+      "loss": 0.9767,
       "step": 81710
     },
     {
       "epoch": 245.41,
-      "grad_norm": 6.7250823974609375,
+      "grad_norm": 5.571617603302002,
       "learning_rate": 1.81981981981982e-06,
-      "loss": 0.9352,
+      "loss": 0.9323,
       "step": 81720
     },
     {
       "epoch": 245.44,
-      "grad_norm": 5.623191833496094,
+      "grad_norm": 6.943173885345459,
       "learning_rate": 1.8188188188188188e-06,
-      "loss": 0.9283,
+      "loss": 0.9205,
       "step": 81730
     },
     {
       "epoch": 245.47,
-      "grad_norm": 7.374078750610352,
+      "grad_norm": 9.473219871520996,
       "learning_rate": 1.817817817817818e-06,
-      "loss": 0.9584,
+      "loss": 0.9575,
       "step": 81740
     },
     {
       "epoch": 245.5,
-      "grad_norm": 6.124556064605713,
+      "grad_norm": 7.292981147766113,
       "learning_rate": 1.816816816816817e-06,
-      "loss": 0.8259,
+      "loss": 0.8105,
       "step": 81750
     },
     {
       "epoch": 245.53,
-      "grad_norm": 6.721776485443115,
+      "grad_norm": 8.685430526733398,
       "learning_rate": 1.815815815815816e-06,
-      "loss": 0.924,
+      "loss": 0.9233,
       "step": 81760
     },
     {
       "epoch": 245.56,
-      "grad_norm": 6.847118377685547,
+      "grad_norm": 8.33391284942627,
       "learning_rate": 1.8148148148148151e-06,
-      "loss": 0.9274,
+      "loss": 0.9296,
       "step": 81770
     },
     {
       "epoch": 245.59,
-      "grad_norm": 5.625936985015869,
+      "grad_norm": 6.309103012084961,
       "learning_rate": 1.813813813813814e-06,
-      "loss": 0.8801,
+      "loss": 0.8784,
       "step": 81780
     },
     {
       "epoch": 245.62,
-      "grad_norm": 6.148820400238037,
+      "grad_norm": 7.111942768096924,
       "learning_rate": 1.8128128128128132e-06,
       "loss": 0.8846,
       "step": 81790
     },
     {
       "epoch": 245.65,
-      "grad_norm": 11.810250282287598,
+      "grad_norm": 10.395215034484863,
       "learning_rate": 1.811811811811812e-06,
-      "loss": 0.8892,
+      "loss": 0.8867,
       "step": 81800
     },
     {
       "epoch": 245.68,
-      "grad_norm": 6.717804431915283,
+      "grad_norm": 6.841365814208984,
       "learning_rate": 1.810810810810811e-06,
-      "loss": 0.8862,
+      "loss": 0.8719,
       "step": 81810
     },
     {
       "epoch": 245.71,
-      "grad_norm": 6.806272506713867,
+      "grad_norm": 6.710789203643799,
       "learning_rate": 1.80980980980981e-06,
-      "loss": 0.8936,
+      "loss": 0.886,
       "step": 81820
     },
     {
       "epoch": 245.74,
-      "grad_norm": 5.222176551818848,
+      "grad_norm": 6.181480884552002,
       "learning_rate": 1.808808808808809e-06,
-      "loss": 0.9238,
+      "loss": 0.9261,
       "step": 81830
     },
     {
       "epoch": 245.77,
-      "grad_norm": 5.613964557647705,
+      "grad_norm": 5.773025035858154,
       "learning_rate": 1.8078078078078078e-06,
-      "loss": 0.903,
+      "loss": 0.8937,
       "step": 81840
     },
     {
       "epoch": 245.8,
-      "grad_norm": 8.941377639770508,
+      "grad_norm": 12.707913398742676,
       "learning_rate": 1.806806806806807e-06,
-      "loss": 0.9332,
+      "loss": 0.9379,
       "step": 81850
     },
     {
       "epoch": 245.83,
-      "grad_norm": 5.66418981552124,
+      "grad_norm": 6.953979969024658,
       "learning_rate": 1.8058058058058059e-06,
-      "loss": 0.9044,
+      "loss": 0.9101,
       "step": 81860
     },
     {
       "epoch": 245.86,
-      "grad_norm": 6.670485019683838,
+      "grad_norm": 8.256784439086914,
       "learning_rate": 1.804804804804805e-06,
-      "loss": 0.9678,
+      "loss": 0.958,
       "step": 81870
     },
     {
       "epoch": 245.89,
-      "grad_norm": 6.023720741271973,
+      "grad_norm": 4.674495697021484,
       "learning_rate": 1.803803803803804e-06,
-      "loss": 0.8778,
+      "loss": 0.8843,
       "step": 81880
     },
     {
       "epoch": 245.92,
-      "grad_norm": 7.530257701873779,
+      "grad_norm": 5.008884906768799,
       "learning_rate": 1.802802802802803e-06,
-      "loss": 0.9359,
+      "loss": 0.9334,
       "step": 81890
     },
     {
       "epoch": 245.95,
-      "grad_norm": 6.860295295715332,
+      "grad_norm": 5.1742143630981445,
       "learning_rate": 1.801801801801802e-06,
-      "loss": 0.9349,
+      "loss": 0.9235,
       "step": 81900
     },
     {
       "epoch": 245.98,
-      "grad_norm": 6.969515323638916,
+      "grad_norm": 10.466066360473633,
       "learning_rate": 1.800800800800801e-06,
-      "loss": 0.9811,
+      "loss": 0.9765,
       "step": 81910
     },
     {
       "epoch": 246.0,
-      "eval_accuracy": 0.8257,
-      "eval_loss": 0.5690745115280151,
-      "eval_runtime": 5.4949,
-      "eval_samples_per_second": 1819.858,
-      "eval_steps_per_second": 7.279,
+      "eval_accuracy": 0.831,
+      "eval_loss": 0.5691348314285278,
+      "eval_runtime": 5.393,
+      "eval_samples_per_second": 1854.257,
+      "eval_steps_per_second": 7.417,
       "step": 81918
     },
     {
       "epoch": 246.01,
-      "grad_norm": 5.666224479675293,
+      "grad_norm": 6.944030284881592,
       "learning_rate": 1.7997997997997998e-06,
-      "loss": 0.9811,
+      "loss": 0.95,
       "step": 81920
     },
     {
       "epoch": 246.04,
-      "grad_norm": 5.821582317352295,
+      "grad_norm": 6.3374152183532715,
       "learning_rate": 1.798798798798799e-06,
-      "loss": 0.828,
+      "loss": 0.8212,
       "step": 81930
     },
     {
       "epoch": 246.07,
-      "grad_norm": 6.008464336395264,
+      "grad_norm": 6.181510925292969,
       "learning_rate": 1.7977977977977978e-06,
-      "loss": 0.9272,
+      "loss": 0.9184,
       "step": 81940
     },
     {
       "epoch": 246.1,
-      "grad_norm": 5.2241716384887695,
+      "grad_norm": 5.732426643371582,
       "learning_rate": 1.796796796796797e-06,
-      "loss": 0.9242,
+      "loss": 0.9269,
       "step": 81950
     },
     {
       "epoch": 246.13,
-      "grad_norm": 4.555519104003906,
+      "grad_norm": 4.870442867279053,
       "learning_rate": 1.7957957957957958e-06,
-      "loss": 0.9257,
+      "loss": 0.9225,
       "step": 81960
     },
     {
       "epoch": 246.16,
-      "grad_norm": 8.916729927062988,
+      "grad_norm": 9.381394386291504,
       "learning_rate": 1.7947947947947949e-06,
-      "loss": 0.9946,
+      "loss": 0.997,
       "step": 81970
     },
     {
       "epoch": 246.19,
-      "grad_norm": 3.9670238494873047,
+      "grad_norm": 6.156876564025879,
       "learning_rate": 1.7937937937937939e-06,
-      "loss": 0.9194,
+      "loss": 0.9212,
       "step": 81980
     },
     {
       "epoch": 246.22,
-      "grad_norm": 9.57213306427002,
+      "grad_norm": 6.722533226013184,
       "learning_rate": 1.7927927927927929e-06,
-      "loss": 0.8922,
+      "loss": 0.8901,
       "step": 81990
     },
     {
       "epoch": 246.25,
-      "grad_norm": 7.916652202606201,
+      "grad_norm": 6.7347412109375,
       "learning_rate": 1.7917917917917917e-06,
-      "loss": 0.883,
+      "loss": 0.9,
       "step": 82000
     },
     {
       "epoch": 246.28,
-      "grad_norm": 6.4782867431640625,
+      "grad_norm": 6.532482147216797,
       "learning_rate": 1.790790790790791e-06,
-      "loss": 0.9191,
+      "loss": 0.9078,
       "step": 82010
     },
     {
       "epoch": 246.31,
-      "grad_norm": 5.739466190338135,
+      "grad_norm": 5.884617328643799,
       "learning_rate": 1.7897897897897902e-06,
-      "loss": 0.8773,
+      "loss": 0.8801,
       "step": 82020
     },
     {
       "epoch": 246.34,
-      "grad_norm": 8.349599838256836,
+      "grad_norm": 7.60919713973999,
       "learning_rate": 1.788788788788789e-06,
-      "loss": 0.8934,
+      "loss": 0.8959,
       "step": 82030
     },
     {
       "epoch": 246.37,
-      "grad_norm": 5.951127529144287,
+      "grad_norm": 4.925541400909424,
       "learning_rate": 1.787787787787788e-06,
-      "loss": 0.8254,
+      "loss": 0.831,
       "step": 82040
     },
     {
       "epoch": 246.4,
-      "grad_norm": 11.313461303710938,
+      "grad_norm": 16.841739654541016,
       "learning_rate": 1.786786786786787e-06,
-      "loss": 0.9789,
+      "loss": 0.978,
       "step": 82050
     },
     {
       "epoch": 246.43,
-      "grad_norm": 5.706259727478027,
+      "grad_norm": 6.336519718170166,
       "learning_rate": 1.785785785785786e-06,
-      "loss": 0.9012,
+      "loss": 0.9187,
       "step": 82060
     },
     {
       "epoch": 246.46,
-      "grad_norm": 6.833477020263672,
+      "grad_norm": 8.307379722595215,
       "learning_rate": 1.7847847847847848e-06,
-      "loss": 0.9452,
+      "loss": 0.9313,
       "step": 82070
     },
     {
       "epoch": 246.49,
-      "grad_norm": 7.663867473602295,
+      "grad_norm": 6.954379558563232,
       "learning_rate": 1.783783783783784e-06,
-      "loss": 0.9027,
+      "loss": 0.8953,
       "step": 82080
     },
     {
       "epoch": 246.52,
-      "grad_norm": 7.09483003616333,
+      "grad_norm": 6.404439449310303,
       "learning_rate": 1.7827827827827829e-06,
-      "loss": 0.9442,
+      "loss": 0.9363,
       "step": 82090
     },
     {
       "epoch": 246.55,
-      "grad_norm": 7.88062047958374,
+      "grad_norm": 9.062873840332031,
       "learning_rate": 1.781781781781782e-06,
-      "loss": 0.8716,
+      "loss": 0.8757,
       "step": 82100
     },
     {
       "epoch": 246.58,
-      "grad_norm": 4.889739990234375,
+      "grad_norm": 5.088563442230225,
       "learning_rate": 1.780780780780781e-06,
-      "loss": 0.8675,
+      "loss": 0.8668,
       "step": 82110
     },
     {
       "epoch": 246.61,
-      "grad_norm": 7.493253231048584,
+      "grad_norm": 9.592589378356934,
       "learning_rate": 1.77977977977978e-06,
-      "loss": 0.858,
+      "loss": 0.8515,
       "step": 82120
     },
     {
       "epoch": 246.64,
-      "grad_norm": 6.183606147766113,
+      "grad_norm": 5.716060638427734,
       "learning_rate": 1.778778778778779e-06,
       "loss": 0.8925,
       "step": 82130
     },
     {
       "epoch": 246.67,
-      "grad_norm": 6.415461540222168,
+      "grad_norm": 6.699799060821533,
       "learning_rate": 1.777777777777778e-06,
-      "loss": 0.9685,
+      "loss": 0.9606,
       "step": 82140
     },
     {
       "epoch": 246.7,
-      "grad_norm": 7.306652069091797,
+      "grad_norm": 7.110442638397217,
       "learning_rate": 1.7767767767767768e-06,
-      "loss": 0.89,
+      "loss": 0.8915,
       "step": 82150
     },
     {
       "epoch": 246.73,
-      "grad_norm": 5.418728351593018,
+      "grad_norm": 9.359130859375,
       "learning_rate": 1.775775775775776e-06,
-      "loss": 0.9157,
+      "loss": 0.9151,
       "step": 82160
     },
     {
       "epoch": 246.76,
-      "grad_norm": 3.9428951740264893,
+      "grad_norm": 5.174387454986572,
       "learning_rate": 1.7747747747747748e-06,
-      "loss": 0.8982,
+      "loss": 0.9052,
       "step": 82170
     },
     {
       "epoch": 246.79,
-      "grad_norm": 4.725827217102051,
+      "grad_norm": 6.065400123596191,
       "learning_rate": 1.773773773773774e-06,
-      "loss": 0.9501,
+      "loss": 0.9479,
       "step": 82180
     },
     {
       "epoch": 246.82,
-      "grad_norm": 6.865681171417236,
+      "grad_norm": 7.66332483291626,
       "learning_rate": 1.7727727727727728e-06,
-      "loss": 0.9789,
+      "loss": 0.9767,
       "step": 82190
     },
     {
       "epoch": 246.85,
-      "grad_norm": 6.1781415939331055,
+      "grad_norm": 7.667489528656006,
       "learning_rate": 1.7717717717717719e-06,
-      "loss": 0.911,
+      "loss": 0.9172,
       "step": 82200
     },
     {
       "epoch": 246.88,
-      "grad_norm": 5.72120475769043,
+      "grad_norm": 6.346189975738525,
       "learning_rate": 1.7707707707707709e-06,
-      "loss": 0.908,
+      "loss": 0.9111,
       "step": 82210
     },
     {
       "epoch": 246.91,
-      "grad_norm": 9.425262451171875,
+      "grad_norm": 6.00930118560791,
       "learning_rate": 1.7697697697697699e-06,
-      "loss": 0.9492,
+      "loss": 0.9323,
       "step": 82220
     },
     {
       "epoch": 246.94,
-      "grad_norm": 4.271500587463379,
+      "grad_norm": 5.512545585632324,
       "learning_rate": 1.7687687687687687e-06,
-      "loss": 0.8671,
+      "loss": 0.8638,
       "step": 82230
     },
     {
       "epoch": 246.97,
-      "grad_norm": 5.96638298034668,
+      "grad_norm": 7.8675055503845215,
       "learning_rate": 1.767767767767768e-06,
-      "loss": 0.9343,
+      "loss": 0.9491,
       "step": 82240
     },
     {
       "epoch": 247.0,
-      "grad_norm": 5.803853988647461,
+      "grad_norm": 5.705637454986572,
       "learning_rate": 1.7667667667667667e-06,
-      "loss": 0.9001,
+      "loss": 0.9095,
       "step": 82250
     },
     {
       "epoch": 247.0,
-      "eval_accuracy": 0.8247,
-      "eval_loss": 0.5832415223121643,
-      "eval_runtime": 5.2574,
-      "eval_samples_per_second": 1902.073,
-      "eval_steps_per_second": 7.608,
+      "eval_accuracy": 0.8262,
+      "eval_loss": 0.5814629197120667,
+      "eval_runtime": 5.4428,
+      "eval_samples_per_second": 1837.279,
+      "eval_steps_per_second": 7.349,
       "step": 82251
     },
     {
       "epoch": 247.03,
-      "grad_norm": 6.76224946975708,
+      "grad_norm": 9.380415916442871,
       "learning_rate": 1.765765765765766e-06,
-      "loss": 1.2861,
+      "loss": 1.2735,
       "step": 82260
     },
     {
       "epoch": 247.06,
-      "grad_norm": 5.596285343170166,
+      "grad_norm": 4.867734909057617,
       "learning_rate": 1.7647647647647648e-06,
-      "loss": 0.9345,
+      "loss": 0.9503,
       "step": 82270
     },
     {
       "epoch": 247.09,
-      "grad_norm": 6.583803176879883,
+      "grad_norm": 7.036834716796875,
       "learning_rate": 1.763763763763764e-06,
-      "loss": 0.974,
+      "loss": 0.968,
       "step": 82280
     },
     {
       "epoch": 247.12,
-      "grad_norm": 8.825424194335938,
+      "grad_norm": 11.117753028869629,
       "learning_rate": 1.762762762762763e-06,
-      "loss": 0.8972,
+      "loss": 0.8897,
       "step": 82290
     },
     {
       "epoch": 247.15,
-      "grad_norm": 7.51292085647583,
+      "grad_norm": 8.069042205810547,
       "learning_rate": 1.7617617617617618e-06,
-      "loss": 0.854,
+      "loss": 0.861,
       "step": 82300
     },
     {
       "epoch": 247.18,
-      "grad_norm": 7.345648765563965,
+      "grad_norm": 7.856075286865234,
       "learning_rate": 1.760760760760761e-06,
-      "loss": 0.9198,
+      "loss": 0.9237,
       "step": 82310
     },
     {
       "epoch": 247.21,
-      "grad_norm": 8.687010765075684,
+      "grad_norm": 12.393721580505371,
       "learning_rate": 1.7597597597597599e-06,
-      "loss": 0.8906,
+      "loss": 0.8905,
       "step": 82320
     },
     {
       "epoch": 247.24,
-      "grad_norm": 5.173910140991211,
+      "grad_norm": 6.789896488189697,
       "learning_rate": 1.758758758758759e-06,
-      "loss": 0.8668,
+      "loss": 0.8687,
       "step": 82330
     },
     {
       "epoch": 247.27,
-      "grad_norm": 7.518939018249512,
+      "grad_norm": 5.790414333343506,
       "learning_rate": 1.757757757757758e-06,
-      "loss": 0.958,
+      "loss": 0.9519,
       "step": 82340
     },
     {
       "epoch": 247.3,
-      "grad_norm": 7.800258636474609,
+      "grad_norm": 7.116460800170898,
       "learning_rate": 1.756756756756757e-06,
-      "loss": 0.8671,
+      "loss": 0.8643,
       "step": 82350
     },
     {
       "epoch": 247.33,
-      "grad_norm": 7.329277992248535,
+      "grad_norm": 7.095160961151123,
       "learning_rate": 1.755755755755756e-06,
       "loss": 0.9455,
       "step": 82360
     },
     {
       "epoch": 247.36,
-      "grad_norm": 7.124632835388184,
+      "grad_norm": 7.386099815368652,
       "learning_rate": 1.754754754754755e-06,
-      "loss": 0.9381,
+      "loss": 0.924,
       "step": 82370
     },
     {
       "epoch": 247.39,
-      "grad_norm": 7.561051368713379,
+      "grad_norm": 9.624646186828613,
       "learning_rate": 1.7537537537537538e-06,
-      "loss": 0.9141,
+      "loss": 0.9087,
       "step": 82380
     },
     {
       "epoch": 247.42,
-      "grad_norm": 4.540841579437256,
+      "grad_norm": 4.059447288513184,
       "learning_rate": 1.752752752752753e-06,
-      "loss": 0.8963,
+      "loss": 0.8958,
       "step": 82390
     },
     {
       "epoch": 247.45,
-      "grad_norm": 7.419632434844971,
+      "grad_norm": 10.095588684082031,
       "learning_rate": 1.7517517517517518e-06,
-      "loss": 0.9447,
+      "loss": 0.9429,
       "step": 82400
     },
     {
       "epoch": 247.48,
-      "grad_norm": 9.32021713256836,
+      "grad_norm": 7.32695198059082,
       "learning_rate": 1.750750750750751e-06,
-      "loss": 0.9374,
+      "loss": 0.9352,
       "step": 82410
     },
     {
       "epoch": 247.51,
-      "grad_norm": 7.50142765045166,
+      "grad_norm": 7.2545857429504395,
       "learning_rate": 1.7497497497497498e-06,
-      "loss": 0.9078,
+      "loss": 0.8957,
       "step": 82420
     },
     {
       "epoch": 247.54,
-      "grad_norm": 6.288571357727051,
+      "grad_norm": 5.995945930480957,
       "learning_rate": 1.7487487487487489e-06,
-      "loss": 0.9543,
+      "loss": 0.956,
       "step": 82430
     },
     {
       "epoch": 247.57,
-      "grad_norm": 6.8484930992126465,
+      "grad_norm": 7.532522678375244,
       "learning_rate": 1.7477477477477479e-06,
-      "loss": 0.9304,
+      "loss": 0.9196,
       "step": 82440
     },
     {
       "epoch": 247.6,
-      "grad_norm": 11.541560173034668,
+      "grad_norm": 14.090354919433594,
       "learning_rate": 1.7467467467467469e-06,
-      "loss": 0.9435,
+      "loss": 0.945,
       "step": 82450
     },
     {
       "epoch": 247.63,
-      "grad_norm": 7.897765636444092,
+      "grad_norm": 9.011261940002441,
       "learning_rate": 1.7457457457457457e-06,
-      "loss": 0.8787,
+      "loss": 0.885,
       "step": 82460
     },
     {
       "epoch": 247.66,
-      "grad_norm": 11.051445007324219,
+      "grad_norm": 8.654801368713379,
       "learning_rate": 1.744744744744745e-06,
-      "loss": 0.8759,
+      "loss": 0.8685,
       "step": 82470
     },
     {
       "epoch": 247.69,
-      "grad_norm": 7.070724010467529,
+      "grad_norm": 7.028858661651611,
       "learning_rate": 1.7437437437437437e-06,
-      "loss": 0.8773,
+      "loss": 0.8753,
       "step": 82480
     },
     {
       "epoch": 247.72,
-      "grad_norm": 5.3209919929504395,
+      "grad_norm": 6.362143039703369,
       "learning_rate": 1.742742742742743e-06,
-      "loss": 0.9057,
+      "loss": 0.906,
       "step": 82490
     },
     {
       "epoch": 247.75,
-      "grad_norm": 8.453730583190918,
+      "grad_norm": 4.849142074584961,
       "learning_rate": 1.7417417417417418e-06,
-      "loss": 0.9288,
+      "loss": 0.9242,
       "step": 82500
     },
     {
       "epoch": 247.78,
-      "grad_norm": 5.965101718902588,
+      "grad_norm": 7.036520957946777,
       "learning_rate": 1.740740740740741e-06,
-      "loss": 0.976,
+      "loss": 0.9782,
       "step": 82510
     },
     {
       "epoch": 247.81,
-      "grad_norm": 6.9444804191589355,
+      "grad_norm": 7.328978538513184,
       "learning_rate": 1.7397397397397398e-06,
-      "loss": 0.868,
+      "loss": 0.8729,
       "step": 82520
     },
     {
       "epoch": 247.84,
-      "grad_norm": 5.126283645629883,
+      "grad_norm": 5.373159885406494,
       "learning_rate": 1.7387387387387388e-06,
-      "loss": 0.956,
+      "loss": 0.9404,
       "step": 82530
     },
     {
       "epoch": 247.87,
-      "grad_norm": 6.86461067199707,
+      "grad_norm": 10.82804012298584,
       "learning_rate": 1.737737737737738e-06,
-      "loss": 0.9146,
+      "loss": 0.9143,
       "step": 82540
     },
     {
       "epoch": 247.9,
-      "grad_norm": 4.6471171379089355,
+      "grad_norm": 5.6615891456604,
       "learning_rate": 1.7367367367367369e-06,
-      "loss": 0.8754,
+      "loss": 0.8834,
       "step": 82550
     },
     {
       "epoch": 247.93,
-      "grad_norm": 5.713619232177734,
+      "grad_norm": 6.626134395599365,
       "learning_rate": 1.735735735735736e-06,
-      "loss": 0.8929,
+      "loss": 0.8894,
       "step": 82560
     },
     {
       "epoch": 247.96,
-      "grad_norm": 7.651586532592773,
+      "grad_norm": 6.756115436553955,
       "learning_rate": 1.734734734734735e-06,
-      "loss": 0.825,
+      "loss": 0.8292,
       "step": 82570
     },
     {
       "epoch": 247.99,
-      "grad_norm": 7.766658306121826,
+      "grad_norm": 8.033681869506836,
       "learning_rate": 1.733733733733734e-06,
-      "loss": 0.9668,
+      "loss": 0.971,
       "step": 82580
     },
     {
       "epoch": 248.0,
-      "eval_accuracy": 0.8268,
-      "eval_loss": 0.576914370059967,
-      "eval_runtime": 5.6585,
-      "eval_samples_per_second": 1767.267,
-      "eval_steps_per_second": 7.069,
+      "eval_accuracy": 0.8275,
+      "eval_loss": 0.5754186511039734,
+      "eval_runtime": 5.3305,
+      "eval_samples_per_second": 1876.004,
+      "eval_steps_per_second": 7.504,
       "step": 82584
     },
     {
       "epoch": 248.02,
-      "grad_norm": 6.43449068069458,
+      "grad_norm": 8.700984001159668,
       "learning_rate": 1.732732732732733e-06,
-      "loss": 1.0521,
+      "loss": 1.0453,
       "step": 82590
     },
     {
       "epoch": 248.05,
-      "grad_norm": 5.2402424812316895,
+      "grad_norm": 7.008136749267578,
       "learning_rate": 1.731731731731732e-06,
-      "loss": 0.9807,
+      "loss": 0.9781,
       "step": 82600
     },
     {
       "epoch": 248.08,
-      "grad_norm": 5.047797679901123,
+      "grad_norm": 7.803192615509033,
       "learning_rate": 1.7307307307307308e-06,
-      "loss": 0.9459,
+      "loss": 0.9454,
       "step": 82610
     },
     {
       "epoch": 248.11,
-      "grad_norm": 6.811305999755859,
+      "grad_norm": 6.70537805557251,
       "learning_rate": 1.72972972972973e-06,
-      "loss": 0.8722,
+      "loss": 0.8748,
       "step": 82620
     },
     {
       "epoch": 248.14,
-      "grad_norm": 5.195559501647949,
+      "grad_norm": 5.9423418045043945,
       "learning_rate": 1.7287287287287288e-06,
-      "loss": 0.9284,
+      "loss": 0.9264,
       "step": 82630
     },
     {
       "epoch": 248.17,
-      "grad_norm": 5.938811779022217,
+      "grad_norm": 5.94176721572876,
       "learning_rate": 1.727727727727728e-06,
-      "loss": 0.9229,
+      "loss": 0.9277,
       "step": 82640
     },
     {
       "epoch": 248.2,
-      "grad_norm": 8.494793891906738,
+      "grad_norm": 9.21778392791748,
       "learning_rate": 1.7267267267267268e-06,
-      "loss": 0.916,
+      "loss": 0.9086,
       "step": 82650
     },
     {
       "epoch": 248.23,
-      "grad_norm": 5.945128440856934,
+      "grad_norm": 8.682890892028809,
       "learning_rate": 1.7257257257257259e-06,
-      "loss": 0.9166,
+      "loss": 0.9147,
       "step": 82660
     },
     {
       "epoch": 248.26,
-      "grad_norm": 8.22391128540039,
+      "grad_norm": 10.712349891662598,
       "learning_rate": 1.7247247247247249e-06,
-      "loss": 0.9171,
+      "loss": 0.9232,
       "step": 82670
     },
     {
       "epoch": 248.29,
-      "grad_norm": 7.077451229095459,
+      "grad_norm": 8.972993850708008,
       "learning_rate": 1.7237237237237239e-06,
-      "loss": 0.8923,
+      "loss": 0.906,
       "step": 82680
     },
     {
       "epoch": 248.32,
-      "grad_norm": 7.915292739868164,
+      "grad_norm": 9.748563766479492,
       "learning_rate": 1.7227227227227227e-06,
-      "loss": 0.9347,
+      "loss": 0.9274,
       "step": 82690
     },
     {
       "epoch": 248.35,
-      "grad_norm": 7.362069129943848,
+      "grad_norm": 6.854910373687744,
       "learning_rate": 1.721721721721722e-06,
-      "loss": 0.8451,
+      "loss": 0.8609,
       "step": 82700
     },
     {
       "epoch": 248.38,
-      "grad_norm": 5.822973251342773,
+      "grad_norm": 7.018612384796143,
       "learning_rate": 1.7207207207207207e-06,
-      "loss": 1.0117,
+      "loss": 1.0172,
       "step": 82710
     },
     {
       "epoch": 248.41,
-      "grad_norm": 6.919643402099609,
+      "grad_norm": 5.99204683303833,
       "learning_rate": 1.71971971971972e-06,
-      "loss": 0.8778,
+      "loss": 0.8801,
       "step": 82720
     },
     {
       "epoch": 248.44,
-      "grad_norm": 5.544931888580322,
+      "grad_norm": 6.342594146728516,
       "learning_rate": 1.7187187187187188e-06,
-      "loss": 0.8277,
+      "loss": 0.8293,
       "step": 82730
     },
     {
       "epoch": 248.47,
-      "grad_norm": 5.363962650299072,
+      "grad_norm": 4.721466064453125,
       "learning_rate": 1.717717717717718e-06,
-      "loss": 0.8539,
+      "loss": 0.8579,
       "step": 82740
     },
     {
       "epoch": 248.5,
-      "grad_norm": 6.420175075531006,
+      "grad_norm": 6.593843460083008,
       "learning_rate": 1.7167167167167168e-06,
-      "loss": 0.9549,
+      "loss": 0.9474,
       "step": 82750
     },
     {
       "epoch": 248.53,
-      "grad_norm": 6.431835651397705,
+      "grad_norm": 6.902481555938721,
       "learning_rate": 1.7157157157157158e-06,
-      "loss": 0.8998,
+      "loss": 0.8899,
       "step": 82760
     },
     {
       "epoch": 248.56,
-      "grad_norm": 10.827522277832031,
+      "grad_norm": 6.491876602172852,
       "learning_rate": 1.7147147147147146e-06,
-      "loss": 0.8773,
+      "loss": 0.8762,
       "step": 82770
     },
     {
       "epoch": 248.59,
-      "grad_norm": 5.675964832305908,
+      "grad_norm": 5.887409687042236,
       "learning_rate": 1.7137137137137139e-06,
-      "loss": 0.9093,
+      "loss": 0.9159,
       "step": 82780
     },
     {
       "epoch": 248.62,
-      "grad_norm": 4.970299243927002,
+      "grad_norm": 5.092787265777588,
       "learning_rate": 1.7127127127127127e-06,
-      "loss": 0.9333,
+      "loss": 0.9287,
       "step": 82790
     },
     {
       "epoch": 248.65,
-      "grad_norm": 7.117290496826172,
+      "grad_norm": 10.287413597106934,
       "learning_rate": 1.711711711711712e-06,
-      "loss": 0.9306,
+      "loss": 0.9371,
       "step": 82800
     },
     {
       "epoch": 248.68,
-      "grad_norm": 4.252988338470459,
+      "grad_norm": 6.475136756896973,
       "learning_rate": 1.710710710710711e-06,
-      "loss": 0.8461,
+      "loss": 0.8545,
       "step": 82810
     },
     {
       "epoch": 248.71,
-      "grad_norm": 6.483274936676025,
+      "grad_norm": 8.924600601196289,
       "learning_rate": 1.70970970970971e-06,
-      "loss": 0.9069,
+      "loss": 0.9013,
       "step": 82820
     },
     {
       "epoch": 248.74,
-      "grad_norm": 9.22056770324707,
+      "grad_norm": 8.572921752929688,
       "learning_rate": 1.708708708708709e-06,
-      "loss": 1.0018,
+      "loss": 0.9939,
       "step": 82830
     },
     {
       "epoch": 248.77,
-      "grad_norm": 4.65700101852417,
+      "grad_norm": 4.681609630584717,
       "learning_rate": 1.7077077077077078e-06,
-      "loss": 0.9342,
+      "loss": 0.9397,
       "step": 82840
     },
     {
       "epoch": 248.8,
-      "grad_norm": 7.3195366859436035,
+      "grad_norm": 6.9191131591796875,
       "learning_rate": 1.706706706706707e-06,
-      "loss": 0.887,
+      "loss": 0.8969,
       "step": 82850
     },
     {
       "epoch": 248.83,
-      "grad_norm": 8.078057289123535,
+      "grad_norm": 8.40163803100586,
       "learning_rate": 1.7057057057057058e-06,
-      "loss": 0.8683,
+      "loss": 0.8568,
       "step": 82860
     },
     {
       "epoch": 248.86,
-      "grad_norm": 8.099111557006836,
+      "grad_norm": 7.19192361831665,
       "learning_rate": 1.704704704704705e-06,
-      "loss": 0.9277,
+      "loss": 0.922,
       "step": 82870
     },
     {
       "epoch": 248.89,
-      "grad_norm": 6.895734786987305,
+      "grad_norm": 6.479654788970947,
       "learning_rate": 1.7037037037037038e-06,
-      "loss": 0.9519,
+      "loss": 0.9371,
       "step": 82880
     },
     {
       "epoch": 248.92,
-      "grad_norm": 8.272894859313965,
+      "grad_norm": 8.959220886230469,
       "learning_rate": 1.7027027027027028e-06,
-      "loss": 0.988,
+      "loss": 0.9922,
       "step": 82890
     },
     {
       "epoch": 248.95,
-      "grad_norm": 7.910833835601807,
+      "grad_norm": 7.994198322296143,
       "learning_rate": 1.7017017017017019e-06,
-      "loss": 0.8949,
+      "loss": 0.9023,
       "step": 82900
     },
     {
       "epoch": 248.98,
-      "grad_norm": 7.495825290679932,
+      "grad_norm": 6.666601181030273,
       "learning_rate": 1.7007007007007009e-06,
-      "loss": 0.9281,
+      "loss": 0.9333,
       "step": 82910
     },
     {
       "epoch": 249.0,
-      "eval_accuracy": 0.8257,
-      "eval_loss": 0.5739601850509644,
-      "eval_runtime": 5.4666,
-      "eval_samples_per_second": 1829.284,
-      "eval_steps_per_second": 7.317,
+      "eval_accuracy": 0.8277,
+      "eval_loss": 0.5732537508010864,
+      "eval_runtime": 5.2904,
+      "eval_samples_per_second": 1890.219,
+      "eval_steps_per_second": 7.561,
       "step": 82917
     },
     {
       "epoch": 249.01,
-      "grad_norm": 6.445218563079834,
+      "grad_norm": 7.9849419593811035,
       "learning_rate": 1.6996996996996997e-06,
-      "loss": 0.9536,
+      "loss": 0.9455,
       "step": 82920
     },
     {
       "epoch": 249.04,
-      "grad_norm": 4.955441951751709,
+      "grad_norm": 6.265883922576904,
       "learning_rate": 1.698698698698699e-06,
-      "loss": 0.9234,
+      "loss": 0.9315,
       "step": 82930
     },
     {
       "epoch": 249.07,
-      "grad_norm": 5.141169548034668,
+      "grad_norm": 5.499331951141357,
       "learning_rate": 1.6976976976976977e-06,
-      "loss": 0.8275,
+      "loss": 0.8152,
       "step": 82940
     },
     {
       "epoch": 249.1,
-      "grad_norm": 6.113393306732178,
+      "grad_norm": 5.583036422729492,
       "learning_rate": 1.696696696696697e-06,
-      "loss": 0.8683,
+      "loss": 0.8651,
       "step": 82950
     },
     {
       "epoch": 249.13,
-      "grad_norm": 7.700004577636719,
+      "grad_norm": 6.8923845291137695,
       "learning_rate": 1.6956956956956958e-06,
-      "loss": 0.9904,
+      "loss": 0.9958,
       "step": 82960
     },
     {
       "epoch": 249.16,
-      "grad_norm": 7.225183963775635,
+      "grad_norm": 6.415700912475586,
       "learning_rate": 1.6946946946946948e-06,
-      "loss": 0.9101,
+      "loss": 0.906,
       "step": 82970
     },
     {
       "epoch": 249.19,
-      "grad_norm": 6.4946675300598145,
+      "grad_norm": 7.246401786804199,
       "learning_rate": 1.6936936936936938e-06,
-      "loss": 0.9065,
+      "loss": 0.9056,
       "step": 82980
     },
     {
       "epoch": 249.22,
-      "grad_norm": 7.876481056213379,
+      "grad_norm": 7.760812759399414,
       "learning_rate": 1.6926926926926928e-06,
       "loss": 0.8904,
       "step": 82990
     },
     {
       "epoch": 249.25,
-      "grad_norm": 8.055803298950195,
+      "grad_norm": 10.696921348571777,
       "learning_rate": 1.6916916916916916e-06,
-      "loss": 0.8912,
+      "loss": 0.8959,
       "step": 83000
     },
     {
       "epoch": 249.28,
-      "grad_norm": 8.39159107208252,
+      "grad_norm": 7.663586139678955,
       "learning_rate": 1.6906906906906909e-06,
-      "loss": 0.8822,
+      "loss": 0.882,
       "step": 83010
     },
     {
       "epoch": 249.31,
-      "grad_norm": 6.3304033279418945,
+      "grad_norm": 6.935215950012207,
       "learning_rate": 1.6896896896896897e-06,
-      "loss": 0.854,
+      "loss": 0.8557,
       "step": 83020
     },
     {
       "epoch": 249.34,
-      "grad_norm": 7.300128936767578,
+      "grad_norm": 6.826028823852539,
       "learning_rate": 1.6886886886886889e-06,
-      "loss": 1.0082,
+      "loss": 0.9934,
       "step": 83030
     },
     {
       "epoch": 249.37,
-      "grad_norm": 7.37366247177124,
+      "grad_norm": 7.159852504730225,
       "learning_rate": 1.6876876876876877e-06,
-      "loss": 0.9218,
+      "loss": 0.9168,
       "step": 83040
     },
     {
       "epoch": 249.4,
-      "grad_norm": 9.102727890014648,
+      "grad_norm": 7.122452735900879,
       "learning_rate": 1.686686686686687e-06,
-      "loss": 0.9506,
+      "loss": 0.9527,
       "step": 83050
     },
     {
       "epoch": 249.43,
-      "grad_norm": 7.2056565284729,
+      "grad_norm": 6.752143383026123,
       "learning_rate": 1.6856856856856857e-06,
-      "loss": 0.9349,
+      "loss": 0.9347,
       "step": 83060
     },
     {
       "epoch": 249.46,
-      "grad_norm": 10.729330062866211,
+      "grad_norm": 10.086822509765625,
       "learning_rate": 1.6846846846846848e-06,
-      "loss": 0.9198,
+      "loss": 0.9348,
       "step": 83070
     },
     {
       "epoch": 249.49,
-      "grad_norm": 7.335421562194824,
+      "grad_norm": 5.5517659187316895,
       "learning_rate": 1.683683683683684e-06,
-      "loss": 0.9125,
+      "loss": 0.9067,
       "step": 83080
     },
     {
       "epoch": 249.52,
-      "grad_norm": 7.729903697967529,
+      "grad_norm": 9.466414451599121,
       "learning_rate": 1.6826826826826828e-06,
-      "loss": 0.8993,
+      "loss": 0.8877,
       "step": 83090
     },
     {
       "epoch": 249.55,
-      "grad_norm": 6.726460933685303,
+      "grad_norm": 5.922889709472656,
       "learning_rate": 1.681681681681682e-06,
-      "loss": 0.9306,
+      "loss": 0.925,
       "step": 83100
     },
     {
       "epoch": 249.58,
-      "grad_norm": 4.764982223510742,
+      "grad_norm": 6.295448303222656,
       "learning_rate": 1.6806806806806808e-06,
-      "loss": 0.8594,
+      "loss": 0.86,
       "step": 83110
     },
     {
       "epoch": 249.61,
-      "grad_norm": 7.874682426452637,
+      "grad_norm": 7.656684875488281,
       "learning_rate": 1.6796796796796798e-06,
-      "loss": 0.9114,
+      "loss": 0.908,
       "step": 83120
     },
     {
       "epoch": 249.64,
-      "grad_norm": 5.025953769683838,
+      "grad_norm": 7.848729133605957,
       "learning_rate": 1.6786786786786789e-06,
-      "loss": 0.8792,
+      "loss": 0.883,
       "step": 83130
     },
     {
       "epoch": 249.67,
-      "grad_norm": 7.88132905960083,
+      "grad_norm": 7.571971893310547,
       "learning_rate": 1.6776776776776779e-06,
-      "loss": 0.9399,
+      "loss": 0.9452,
       "step": 83140
     },
     {
       "epoch": 249.7,
-      "grad_norm": 7.510242938995361,
+      "grad_norm": 8.218785285949707,
       "learning_rate": 1.6766766766766767e-06,
-      "loss": 0.9252,
+      "loss": 0.9104,
       "step": 83150
     },
     {
       "epoch": 249.73,
-      "grad_norm": 6.578803062438965,
+      "grad_norm": 9.922873497009277,
       "learning_rate": 1.675675675675676e-06,
-      "loss": 0.9649,
+      "loss": 0.9664,
       "step": 83160
     },
     {
       "epoch": 249.76,
-      "grad_norm": 7.819990158081055,
+      "grad_norm": 7.865416526794434,
       "learning_rate": 1.6746746746746747e-06,
-      "loss": 0.887,
+      "loss": 0.8819,
       "step": 83170
     },
     {
       "epoch": 249.79,
-      "grad_norm": 8.782036781311035,
+      "grad_norm": 7.2745232582092285,
       "learning_rate": 1.673673673673674e-06,
-      "loss": 0.892,
+      "loss": 0.8917,
       "step": 83180
     },
     {
       "epoch": 249.82,
-      "grad_norm": 6.500641822814941,
+      "grad_norm": 6.60286283493042,
       "learning_rate": 1.6726726726726728e-06,
-      "loss": 0.8892,
+      "loss": 0.8931,
       "step": 83190
     },
     {
       "epoch": 249.85,
-      "grad_norm": 7.9875946044921875,
+      "grad_norm": 6.110004901885986,
       "learning_rate": 1.6716716716716718e-06,
-      "loss": 0.9846,
+      "loss": 0.9906,
       "step": 83200
     },
     {
       "epoch": 249.88,
-      "grad_norm": 6.2925214767456055,
+      "grad_norm": 6.846839904785156,
       "learning_rate": 1.6706706706706708e-06,
-      "loss": 0.9434,
+      "loss": 0.9511,
       "step": 83210
     },
     {
       "epoch": 249.91,
-      "grad_norm": 5.769713878631592,
+      "grad_norm": 5.721170425415039,
       "learning_rate": 1.6696696696696698e-06,
-      "loss": 0.8448,
+      "loss": 0.8452,
       "step": 83220
     },
     {
       "epoch": 249.94,
-      "grad_norm": 6.825571537017822,
+      "grad_norm": 6.994389057159424,
       "learning_rate": 1.6686686686686686e-06,
-      "loss": 0.9208,
+      "loss": 0.9193,
       "step": 83230
     },
     {
       "epoch": 249.97,
-      "grad_norm": 11.461763381958008,
+      "grad_norm": 14.947723388671875,
       "learning_rate": 1.6676676676676679e-06,
-      "loss": 0.9165,
+      "loss": 0.9281,
       "step": 83240
     },
     {
       "epoch": 250.0,
-      "grad_norm": 77.48682403564453,
+      "grad_norm": 119.1614761352539,
       "learning_rate": 1.6666666666666667e-06,
-      "loss": 0.9167,
+      "loss": 0.9193,
       "step": 83250
     },
     {
       "epoch": 250.0,
-      "eval_accuracy": 0.8265,
-      "eval_loss": 0.5752727389335632,
-      "eval_runtime": 5.3166,
-      "eval_samples_per_second": 1880.907,
-      "eval_steps_per_second": 7.524,
+      "eval_accuracy": 0.8264,
+      "eval_loss": 0.5749737620353699,
+      "eval_runtime": 5.2172,
+      "eval_samples_per_second": 1916.732,
+      "eval_steps_per_second": 7.667,
       "step": 83250
     },
     {
       "epoch": 250.03,
-      "grad_norm": 6.112013816833496,
+      "grad_norm": 4.566414833068848,
       "learning_rate": 1.6656656656656659e-06,
-      "loss": 0.9032,
+      "loss": 0.9058,
       "step": 83260
     },
     {
       "epoch": 250.06,
-      "grad_norm": 5.813735008239746,
+      "grad_norm": 6.86842155456543,
       "learning_rate": 1.6646646646646647e-06,
-      "loss": 0.8424,
+      "loss": 0.851,
       "step": 83270
     },
     {
       "epoch": 250.09,
-      "grad_norm": 6.282175064086914,
+      "grad_norm": 5.2747626304626465,
       "learning_rate": 1.663663663663664e-06,
-      "loss": 0.8951,
+      "loss": 0.8974,
       "step": 83280
     },
     {
       "epoch": 250.12,
-      "grad_norm": 10.458329200744629,
+      "grad_norm": 8.732887268066406,
       "learning_rate": 1.6626626626626627e-06,
-      "loss": 0.9027,
+      "loss": 0.9015,
       "step": 83290
     },
     {
       "epoch": 250.15,
-      "grad_norm": 7.8553147315979,
+      "grad_norm": 7.330347061157227,
       "learning_rate": 1.6616616616616618e-06,
-      "loss": 0.9054,
+      "loss": 0.9041,
       "step": 83300
     },
     {
       "epoch": 250.18,
-      "grad_norm": 9.047595977783203,
+      "grad_norm": 8.366328239440918,
       "learning_rate": 1.6606606606606608e-06,
-      "loss": 0.918,
+      "loss": 0.9139,
       "step": 83310
     },
     {
       "epoch": 250.21,
-      "grad_norm": 7.7049880027771,
+      "grad_norm": 11.186152458190918,
       "learning_rate": 1.6596596596596598e-06,
-      "loss": 0.9557,
+      "loss": 0.9518,
       "step": 83320
     },
     {
       "epoch": 250.24,
-      "grad_norm": 7.63035249710083,
+      "grad_norm": 9.633931159973145,
       "learning_rate": 1.6586586586586586e-06,
-      "loss": 0.917,
+      "loss": 0.9102,
       "step": 83330
     },
     {
       "epoch": 250.27,
-      "grad_norm": 6.788579940795898,
+      "grad_norm": 10.226130485534668,
       "learning_rate": 1.6576576576576578e-06,
-      "loss": 0.9274,
+      "loss": 0.9215,
       "step": 83340
     },
     {
       "epoch": 250.3,
-      "grad_norm": 7.128540992736816,
+      "grad_norm": 8.327975273132324,
       "learning_rate": 1.6566566566566568e-06,
-      "loss": 0.899,
+      "loss": 0.8929,
       "step": 83350
     },
     {
       "epoch": 250.33,
-      "grad_norm": 5.202794075012207,
+      "grad_norm": 5.617682456970215,
       "learning_rate": 1.6556556556556559e-06,
-      "loss": 0.8568,
+      "loss": 0.8454,
       "step": 83360
     },
     {
       "epoch": 250.36,
-      "grad_norm": 7.241932392120361,
+      "grad_norm": 7.56367301940918,
       "learning_rate": 1.6546546546546549e-06,
-      "loss": 0.93,
+      "loss": 0.9107,
       "step": 83370
     },
     {
       "epoch": 250.39,
-      "grad_norm": 7.217708110809326,
+      "grad_norm": 6.8665852546691895,
       "learning_rate": 1.6536536536536537e-06,
-      "loss": 0.962,
+      "loss": 0.9634,
       "step": 83380
     },
     {
       "epoch": 250.42,
-      "grad_norm": 9.447193145751953,
+      "grad_norm": 8.491464614868164,
       "learning_rate": 1.652652652652653e-06,
-      "loss": 0.9395,
+      "loss": 0.9308,
       "step": 83390
     },
     {
       "epoch": 250.45,
-      "grad_norm": 7.843297004699707,
+      "grad_norm": 9.55777359008789,
       "learning_rate": 1.6516516516516517e-06,
-      "loss": 0.9195,
+      "loss": 0.9284,
       "step": 83400
     },
     {
       "epoch": 250.48,
-      "grad_norm": 5.9093194007873535,
+      "grad_norm": 7.750838756561279,
       "learning_rate": 1.650650650650651e-06,
-      "loss": 0.8933,
+      "loss": 0.8912,
       "step": 83410
     },
     {
       "epoch": 250.51,
-      "grad_norm": 8.474127769470215,
+      "grad_norm": 9.576678276062012,
       "learning_rate": 1.6496496496496498e-06,
-      "loss": 0.9544,
+      "loss": 0.9574,
       "step": 83420
     },
     {
       "epoch": 250.54,
-      "grad_norm": 4.360866069793701,
+      "grad_norm": 4.163198947906494,
       "learning_rate": 1.6486486486486488e-06,
-      "loss": 0.8925,
+      "loss": 0.8951,
       "step": 83430
     },
     {
       "epoch": 250.57,
-      "grad_norm": 9.579529762268066,
+      "grad_norm": 9.968963623046875,
       "learning_rate": 1.6476476476476478e-06,
-      "loss": 0.8991,
+      "loss": 0.8945,
       "step": 83440
     },
     {
       "epoch": 250.6,
-      "grad_norm": 6.169926166534424,
+      "grad_norm": 4.950615882873535,
       "learning_rate": 1.6466466466466468e-06,
-      "loss": 0.9196,
+      "loss": 0.9267,
       "step": 83450
     },
     {
       "epoch": 250.63,
-      "grad_norm": 6.487170696258545,
+      "grad_norm": 6.0487260818481445,
       "learning_rate": 1.6456456456456456e-06,
-      "loss": 0.9022,
+      "loss": 0.8999,
       "step": 83460
     },
     {
       "epoch": 250.66,
-      "grad_norm": 6.266751289367676,
+      "grad_norm": 6.319337368011475,
       "learning_rate": 1.6446446446446449e-06,
-      "loss": 0.8917,
+      "loss": 0.8924,
       "step": 83470
     },
     {
       "epoch": 250.69,
-      "grad_norm": 7.276398181915283,
+      "grad_norm": 7.645631790161133,
       "learning_rate": 1.6436436436436437e-06,
-      "loss": 0.9315,
+      "loss": 0.9215,
       "step": 83480
     },
     {
       "epoch": 250.72,
-      "grad_norm": 7.580574989318848,
+      "grad_norm": 6.6868133544921875,
       "learning_rate": 1.6426426426426429e-06,
-      "loss": 0.9184,
+      "loss": 0.9164,
       "step": 83490
     },
     {
       "epoch": 250.75,
-      "grad_norm": 7.208494186401367,
+      "grad_norm": 7.256258964538574,
       "learning_rate": 1.6416416416416417e-06,
-      "loss": 0.8552,
+      "loss": 0.8522,
       "step": 83500
     },
     {
       "epoch": 250.78,
-      "grad_norm": 4.3448710441589355,
+      "grad_norm": 5.6176323890686035,
       "learning_rate": 1.640640640640641e-06,
-      "loss": 0.9085,
+      "loss": 0.9108,
       "step": 83510
     },
     {
       "epoch": 250.81,
-      "grad_norm": 7.319076061248779,
+      "grad_norm": 6.856009006500244,
       "learning_rate": 1.6396396396396397e-06,
-      "loss": 0.9067,
+      "loss": 0.9028,
       "step": 83520
     },
     {
       "epoch": 250.84,
-      "grad_norm": 7.862907886505127,
+      "grad_norm": 5.952432155609131,
       "learning_rate": 1.6386386386386388e-06,
-      "loss": 0.8911,
+      "loss": 0.8785,
       "step": 83530
     },
     {
       "epoch": 250.87,
-      "grad_norm": 6.301018238067627,
+      "grad_norm": 6.937488555908203,
       "learning_rate": 1.6376376376376378e-06,
-      "loss": 0.9302,
+      "loss": 0.9223,
       "step": 83540
     },
     {
       "epoch": 250.9,
-      "grad_norm": 4.577832221984863,
+      "grad_norm": 6.148772716522217,
       "learning_rate": 1.6366366366366368e-06,
-      "loss": 0.8656,
+      "loss": 0.8569,
       "step": 83550
     },
     {
       "epoch": 250.93,
-      "grad_norm": 6.821409702301025,
+      "grad_norm": 8.44312858581543,
       "learning_rate": 1.6356356356356356e-06,
-      "loss": 1.0341,
+      "loss": 1.0374,
       "step": 83560
     },
     {
       "epoch": 250.96,
-      "grad_norm": 5.342735290527344,
+      "grad_norm": 6.892063617706299,
       "learning_rate": 1.6346346346346348e-06,
-      "loss": 0.9666,
+      "loss": 0.9695,
       "step": 83570
     },
     {
       "epoch": 250.99,
-      "grad_norm": 5.48955774307251,
+      "grad_norm": 4.9565253257751465,
       "learning_rate": 1.6336336336336336e-06,
-      "loss": 1.0039,
+      "loss": 0.9961,
       "step": 83580
     },
     {
       "epoch": 251.0,
-      "eval_accuracy": 0.8273,
-      "eval_loss": 0.5730262398719788,
-      "eval_runtime": 5.5793,
-      "eval_samples_per_second": 1792.343,
-      "eval_steps_per_second": 7.169,
+      "eval_accuracy": 0.8281,
+      "eval_loss": 0.5739744901657104,
+      "eval_runtime": 5.4604,
+      "eval_samples_per_second": 1831.359,
+      "eval_steps_per_second": 7.325,
       "step": 83583
     },
     {
       "epoch": 251.02,
-      "grad_norm": 6.2092061042785645,
+      "grad_norm": 6.0686869621276855,
       "learning_rate": 1.6326326326326329e-06,
-      "loss": 0.9917,
+      "loss": 0.9766,
       "step": 83590
     },
     {
       "epoch": 251.05,
-      "grad_norm": 8.475850105285645,
+      "grad_norm": 7.604572772979736,
       "learning_rate": 1.6316316316316317e-06,
-      "loss": 0.8984,
+      "loss": 0.8836,
       "step": 83600
     },
     {
       "epoch": 251.08,
-      "grad_norm": 7.951423168182373,
+      "grad_norm": 8.043323516845703,
       "learning_rate": 1.6306306306306307e-06,
-      "loss": 0.8905,
+      "loss": 0.8888,
       "step": 83610
     },
     {
       "epoch": 251.11,
-      "grad_norm": 7.133674621582031,
+      "grad_norm": 7.179011821746826,
       "learning_rate": 1.62962962962963e-06,
-      "loss": 0.9629,
+      "loss": 0.9445,
       "step": 83620
     },
     {
       "epoch": 251.14,
-      "grad_norm": 7.1227006912231445,
+      "grad_norm": 5.349400997161865,
       "learning_rate": 1.6286286286286287e-06,
-      "loss": 0.9004,
+      "loss": 0.9073,
       "step": 83630
     },
     {
       "epoch": 251.17,
-      "grad_norm": 9.219672203063965,
+      "grad_norm": 6.292776107788086,
       "learning_rate": 1.627627627627628e-06,
-      "loss": 0.9062,
+      "loss": 0.914,
       "step": 83640
     },
     {
       "epoch": 251.2,
-      "grad_norm": 6.088767051696777,
+      "grad_norm": 8.4429292678833,
       "learning_rate": 1.6266266266266268e-06,
-      "loss": 0.9244,
+      "loss": 0.9085,
       "step": 83650
     },
     {
       "epoch": 251.23,
-      "grad_norm": 6.76792573928833,
+      "grad_norm": 6.3061299324035645,
       "learning_rate": 1.6256256256256258e-06,
-      "loss": 0.8882,
+      "loss": 0.8844,
       "step": 83660
     },
     {
       "epoch": 251.26,
-      "grad_norm": 6.610317707061768,
+      "grad_norm": 5.282219409942627,
       "learning_rate": 1.6246246246246248e-06,
-      "loss": 0.9659,
+      "loss": 0.9528,
       "step": 83670
     },
     {
       "epoch": 251.29,
-      "grad_norm": 7.895793914794922,
+      "grad_norm": 8.871384620666504,
       "learning_rate": 1.6236236236236238e-06,
-      "loss": 0.8847,
+      "loss": 0.8853,
       "step": 83680
     },
     {
       "epoch": 251.32,
-      "grad_norm": 7.285521030426025,
+      "grad_norm": 6.416481018066406,
       "learning_rate": 1.6226226226226226e-06,
-      "loss": 0.873,
+      "loss": 0.8633,
       "step": 83690
     },
     {
       "epoch": 251.35,
-      "grad_norm": 6.931979179382324,
+      "grad_norm": 6.2389631271362305,
       "learning_rate": 1.6216216216216219e-06,
-      "loss": 0.8474,
+      "loss": 0.8465,
       "step": 83700
     },
     {
       "epoch": 251.38,
-      "grad_norm": 10.453474044799805,
+      "grad_norm": 9.526310920715332,
       "learning_rate": 1.6206206206206207e-06,
-      "loss": 0.926,
+      "loss": 0.9171,
       "step": 83710
     },
     {
       "epoch": 251.41,
-      "grad_norm": 5.674417495727539,
+      "grad_norm": 6.043628692626953,
       "learning_rate": 1.6196196196196199e-06,
-      "loss": 0.8985,
+      "loss": 0.8955,
       "step": 83720
     },
     {
       "epoch": 251.44,
-      "grad_norm": 5.438640117645264,
+      "grad_norm": 7.6861467361450195,
       "learning_rate": 1.6186186186186187e-06,
-      "loss": 0.9114,
+      "loss": 0.9093,
       "step": 83730
     },
     {
       "epoch": 251.47,
-      "grad_norm": 5.256451606750488,
+      "grad_norm": 6.190312385559082,
       "learning_rate": 1.617617617617618e-06,
-      "loss": 0.8433,
+      "loss": 0.8571,
       "step": 83740
     },
     {
       "epoch": 251.5,
-      "grad_norm": 5.784649848937988,
+      "grad_norm": 6.1911749839782715,
       "learning_rate": 1.6166166166166167e-06,
-      "loss": 0.8502,
+      "loss": 0.8479,
       "step": 83750
     },
     {
       "epoch": 251.53,
-      "grad_norm": 10.780577659606934,
+      "grad_norm": 9.150038719177246,
       "learning_rate": 1.6156156156156157e-06,
-      "loss": 0.9201,
+      "loss": 0.91,
       "step": 83760
     },
     {
       "epoch": 251.56,
-      "grad_norm": 7.6048383712768555,
+      "grad_norm": 5.542619228363037,
       "learning_rate": 1.6146146146146146e-06,
-      "loss": 0.8722,
+      "loss": 0.8752,
       "step": 83770
     },
     {
       "epoch": 251.59,
-      "grad_norm": 4.413800239562988,
+      "grad_norm": 5.081509113311768,
       "learning_rate": 1.6136136136136138e-06,
-      "loss": 0.9475,
+      "loss": 0.9537,
       "step": 83780
     },
     {
       "epoch": 251.62,
-      "grad_norm": 8.409502983093262,
+      "grad_norm": 7.473005771636963,
       "learning_rate": 1.6126126126126126e-06,
-      "loss": 0.944,
+      "loss": 0.9401,
       "step": 83790
     },
     {
       "epoch": 251.65,
-      "grad_norm": 6.121049404144287,
+      "grad_norm": 7.086778163909912,
       "learning_rate": 1.6116116116116118e-06,
-      "loss": 0.8897,
+      "loss": 0.8853,
       "step": 83800
     },
     {
       "epoch": 251.68,
-      "grad_norm": 6.6293511390686035,
+      "grad_norm": 5.098796844482422,
       "learning_rate": 1.6106106106106106e-06,
-      "loss": 0.9267,
+      "loss": 0.93,
       "step": 83810
     },
     {
       "epoch": 251.71,
-      "grad_norm": 6.5103068351745605,
+      "grad_norm": 8.065882682800293,
       "learning_rate": 1.6096096096096099e-06,
-      "loss": 0.9211,
+      "loss": 0.9228,
       "step": 83820
     },
     {
       "epoch": 251.74,
-      "grad_norm": 7.922293663024902,
+      "grad_norm": 5.946536064147949,
       "learning_rate": 1.6086086086086087e-06,
-      "loss": 0.8219,
+      "loss": 0.8212,
       "step": 83830
     },
     {
       "epoch": 251.77,
-      "grad_norm": 5.448889255523682,
+      "grad_norm": 5.500353813171387,
       "learning_rate": 1.6076076076076077e-06,
-      "loss": 0.9235,
+      "loss": 0.9323,
       "step": 83840
     },
     {
       "epoch": 251.8,
-      "grad_norm": 7.14418363571167,
+      "grad_norm": 9.561652183532715,
       "learning_rate": 1.6066066066066067e-06,
-      "loss": 0.9712,
+      "loss": 0.9641,
       "step": 83850
     },
     {
       "epoch": 251.83,
-      "grad_norm": 7.872790336608887,
+      "grad_norm": 9.883591651916504,
       "learning_rate": 1.6056056056056057e-06,
-      "loss": 0.9481,
+      "loss": 0.9281,
       "step": 83860
     },
     {
       "epoch": 251.86,
-      "grad_norm": 10.167057991027832,
+      "grad_norm": 8.534424781799316,
       "learning_rate": 1.6046046046046045e-06,
-      "loss": 0.8905,
+      "loss": 0.899,
       "step": 83870
     },
     {
       "epoch": 251.89,
-      "grad_norm": 9.889140129089355,
+      "grad_norm": 13.179317474365234,
       "learning_rate": 1.6036036036036038e-06,
-      "loss": 0.91,
+      "loss": 0.9153,
       "step": 83880
     },
     {
       "epoch": 251.92,
-      "grad_norm": 5.923210144042969,
+      "grad_norm": 6.683413505554199,
       "learning_rate": 1.6026026026026028e-06,
-      "loss": 0.9605,
+      "loss": 0.9581,
       "step": 83890
     },
     {
       "epoch": 251.95,
-      "grad_norm": 6.136784076690674,
+      "grad_norm": 5.428544044494629,
       "learning_rate": 1.6016016016016018e-06,
-      "loss": 0.9724,
+      "loss": 0.9808,
       "step": 83900
     },
     {
       "epoch": 251.98,
-      "grad_norm": 5.906559944152832,
+      "grad_norm": 6.225046157836914,
       "learning_rate": 1.6006006006006008e-06,
-      "loss": 0.9624,
+      "loss": 0.9489,
       "step": 83910
     },
     {
       "epoch": 252.0,
-      "eval_accuracy": 0.8291,
-      "eval_loss": 0.5666501522064209,
-      "eval_runtime": 5.4366,
-      "eval_samples_per_second": 1839.401,
-      "eval_steps_per_second": 7.358,
+      "eval_accuracy": 0.8304,
+      "eval_loss": 0.5659887194633484,
+      "eval_runtime": 5.317,
+      "eval_samples_per_second": 1880.757,
+      "eval_steps_per_second": 7.523,
       "step": 83916
     },
     {
       "epoch": 252.01,
-      "grad_norm": 9.002867698669434,
+      "grad_norm": 9.139132499694824,
       "learning_rate": 1.5995995995995996e-06,
-      "loss": 0.9724,
+      "loss": 0.9624,
       "step": 83920
     },
     {
       "epoch": 252.04,
-      "grad_norm": 8.485791206359863,
+      "grad_norm": 8.10392951965332,
       "learning_rate": 1.5985985985985988e-06,
-      "loss": 0.9365,
+      "loss": 0.9322,
       "step": 83930
     },
     {
       "epoch": 252.07,
-      "grad_norm": 6.162415981292725,
+      "grad_norm": 7.210274696350098,
       "learning_rate": 1.5975975975975977e-06,
-      "loss": 0.9402,
+      "loss": 0.9348,
       "step": 83940
     },
     {
       "epoch": 252.1,
-      "grad_norm": 7.492887496948242,
+      "grad_norm": 6.971937656402588,
       "learning_rate": 1.5965965965965969e-06,
-      "loss": 0.8785,
+      "loss": 0.8872,
       "step": 83950
     },
     {
       "epoch": 252.13,
-      "grad_norm": 5.458399772644043,
+      "grad_norm": 5.56447696685791,
       "learning_rate": 1.5955955955955957e-06,
-      "loss": 0.887,
+      "loss": 0.8854,
       "step": 83960
     },
     {
       "epoch": 252.16,
-      "grad_norm": 5.125554084777832,
+      "grad_norm": 7.299841403961182,
       "learning_rate": 1.5945945945945947e-06,
-      "loss": 0.943,
+      "loss": 0.9383,
       "step": 83970
     },
     {
       "epoch": 252.19,
-      "grad_norm": 5.221033573150635,
+      "grad_norm": 6.045350551605225,
       "learning_rate": 1.5935935935935937e-06,
-      "loss": 0.8852,
+      "loss": 0.8895,
       "step": 83980
     },
     {
       "epoch": 252.22,
-      "grad_norm": 6.0900092124938965,
+      "grad_norm": 7.927063941955566,
       "learning_rate": 1.5925925925925927e-06,
-      "loss": 0.955,
+      "loss": 0.9338,
       "step": 83990
     },
     {
       "epoch": 252.25,
-      "grad_norm": 6.747593879699707,
+      "grad_norm": 7.3414626121521,
       "learning_rate": 1.5915915915915916e-06,
-      "loss": 0.8801,
+      "loss": 0.8799,
       "step": 84000
     },
     {
       "epoch": 252.28,
-      "grad_norm": 7.616765975952148,
+      "grad_norm": 5.828044891357422,
       "learning_rate": 1.5905905905905908e-06,
-      "loss": 0.9429,
+      "loss": 0.9483,
       "step": 84010
     },
     {
       "epoch": 252.31,
-      "grad_norm": 5.828289031982422,
+      "grad_norm": 6.1322102546691895,
       "learning_rate": 1.5895895895895896e-06,
-      "loss": 0.904,
+      "loss": 0.9031,
       "step": 84020
     },
     {
       "epoch": 252.34,
-      "grad_norm": 5.580169200897217,
+      "grad_norm": 5.269176483154297,
       "learning_rate": 1.5885885885885888e-06,
-      "loss": 0.9432,
+      "loss": 0.9425,
       "step": 84030
     },
     {
       "epoch": 252.37,
-      "grad_norm": 7.581920623779297,
+      "grad_norm": 8.632560729980469,
       "learning_rate": 1.5875875875875876e-06,
-      "loss": 0.9589,
+      "loss": 0.9604,
       "step": 84040
     },
     {
       "epoch": 252.4,
-      "grad_norm": 7.55603551864624,
+      "grad_norm": 5.040329456329346,
       "learning_rate": 1.5865865865865869e-06,
-      "loss": 0.9588,
+      "loss": 0.9525,
       "step": 84050
     },
     {
       "epoch": 252.43,
-      "grad_norm": 8.501259803771973,
+      "grad_norm": 5.760193824768066,
       "learning_rate": 1.5855855855855857e-06,
-      "loss": 0.9366,
+      "loss": 0.9299,
       "step": 84060
     },
     {
       "epoch": 252.46,
-      "grad_norm": 8.04964542388916,
+      "grad_norm": 8.602758407592773,
       "learning_rate": 1.5845845845845847e-06,
-      "loss": 0.8532,
+      "loss": 0.8437,
       "step": 84070
     },
     {
       "epoch": 252.49,
-      "grad_norm": 8.624190330505371,
+      "grad_norm": 9.327435493469238,
       "learning_rate": 1.5835835835835837e-06,
-      "loss": 0.8502,
+      "loss": 0.839,
       "step": 84080
     },
     {
       "epoch": 252.52,
-      "grad_norm": 5.55385160446167,
+      "grad_norm": 9.785548210144043,
       "learning_rate": 1.5825825825825827e-06,
-      "loss": 0.886,
+      "loss": 0.8819,
       "step": 84090
     },
     {
       "epoch": 252.55,
-      "grad_norm": 6.408483028411865,
+      "grad_norm": 5.303720951080322,
       "learning_rate": 1.5815815815815815e-06,
-      "loss": 0.9497,
+      "loss": 0.9533,
       "step": 84100
     },
     {
       "epoch": 252.58,
-      "grad_norm": 5.80289363861084,
+      "grad_norm": 6.644670009613037,
       "learning_rate": 1.5805805805805808e-06,
-      "loss": 0.9504,
+      "loss": 0.9373,
       "step": 84110
     },
     {
       "epoch": 252.61,
-      "grad_norm": 6.146097183227539,
+      "grad_norm": 8.175010681152344,
       "learning_rate": 1.5795795795795796e-06,
-      "loss": 0.8842,
+      "loss": 0.8943,
       "step": 84120
     },
     {
       "epoch": 252.64,
-      "grad_norm": 6.590002536773682,
+      "grad_norm": 8.059051513671875,
       "learning_rate": 1.5785785785785788e-06,
-      "loss": 0.9672,
+      "loss": 0.9652,
       "step": 84130
     },
     {
       "epoch": 252.67,
-      "grad_norm": 5.740342617034912,
+      "grad_norm": 7.64532995223999,
       "learning_rate": 1.5775775775775776e-06,
-      "loss": 0.9258,
+      "loss": 0.929,
       "step": 84140
     },
     {
       "epoch": 252.7,
-      "grad_norm": 6.339529514312744,
+      "grad_norm": 7.477354526519775,
       "learning_rate": 1.5765765765765766e-06,
-      "loss": 0.9433,
+      "loss": 0.945,
       "step": 84150
     },
     {
       "epoch": 252.73,
-      "grad_norm": 7.57461404800415,
+      "grad_norm": 6.062965393066406,
       "learning_rate": 1.5755755755755758e-06,
-      "loss": 0.8255,
+      "loss": 0.8185,
       "step": 84160
     },
     {
       "epoch": 252.76,
-      "grad_norm": 8.819400787353516,
+      "grad_norm": 6.593989372253418,
       "learning_rate": 1.5745745745745747e-06,
-      "loss": 0.9179,
+      "loss": 0.9202,
       "step": 84170
     },
     {
       "epoch": 252.79,
-      "grad_norm": 5.376256942749023,
+      "grad_norm": 6.367208957672119,
       "learning_rate": 1.5735735735735739e-06,
-      "loss": 0.9132,
+      "loss": 0.9043,
       "step": 84180
     },
     {
       "epoch": 252.82,
-      "grad_norm": 7.442547798156738,
+      "grad_norm": 7.422072410583496,
       "learning_rate": 1.5725725725725727e-06,
-      "loss": 0.8875,
+      "loss": 0.8796,
       "step": 84190
     },
     {
       "epoch": 252.85,
-      "grad_norm": 6.877533435821533,
+      "grad_norm": 6.984378337860107,
       "learning_rate": 1.5715715715715717e-06,
-      "loss": 0.9034,
+      "loss": 0.8876,
       "step": 84200
     },
     {
       "epoch": 252.88,
-      "grad_norm": 7.744998931884766,
+      "grad_norm": 7.514007568359375,
       "learning_rate": 1.5705705705705707e-06,
-      "loss": 0.8526,
+      "loss": 0.8524,
       "step": 84210
     },
     {
       "epoch": 252.91,
-      "grad_norm": 6.293265342712402,
+      "grad_norm": 5.589169502258301,
       "learning_rate": 1.5695695695695697e-06,
-      "loss": 0.8585,
+      "loss": 0.8504,
       "step": 84220
     },
     {
       "epoch": 252.94,
-      "grad_norm": 9.381341934204102,
+      "grad_norm": 6.5556721687316895,
       "learning_rate": 1.5685685685685686e-06,
-      "loss": 0.8616,
+      "loss": 0.8631,
       "step": 84230
     },
     {
       "epoch": 252.97,
-      "grad_norm": 10.043846130371094,
+      "grad_norm": 10.296335220336914,
       "learning_rate": 1.5675675675675678e-06,
-      "loss": 0.8988,
+      "loss": 0.8926,
       "step": 84240
     },
     {
       "epoch": 253.0,
-      "eval_accuracy": 0.8255,
-      "eval_loss": 0.5751023292541504,
-      "eval_runtime": 5.4466,
-      "eval_samples_per_second": 1836.018,
-      "eval_steps_per_second": 7.344,
+      "eval_accuracy": 0.8281,
+      "eval_loss": 0.5729944109916687,
+      "eval_runtime": 5.2927,
+      "eval_samples_per_second": 1889.384,
+      "eval_steps_per_second": 7.558,
       "step": 84249
     },
     {
       "epoch": 253.0,
-      "grad_norm": 6.572518825531006,
+      "grad_norm": 7.051578998565674,
       "learning_rate": 1.5665665665665666e-06,
-      "loss": 0.9386,
+      "loss": 0.9504,
       "step": 84250
     },
     {
       "epoch": 253.03,
-      "grad_norm": 5.213061809539795,
+      "grad_norm": 6.364383220672607,
       "learning_rate": 1.5655655655655658e-06,
-      "loss": 0.8726,
+      "loss": 0.8784,
       "step": 84260
     },
     {
       "epoch": 253.06,
-      "grad_norm": 4.920195579528809,
+      "grad_norm": 5.9996137619018555,
       "learning_rate": 1.5645645645645646e-06,
-      "loss": 0.8969,
+      "loss": 0.8984,
       "step": 84270
     },
     {
       "epoch": 253.09,
-      "grad_norm": 7.96111536026001,
+      "grad_norm": 8.31523609161377,
       "learning_rate": 1.5635635635635639e-06,
-      "loss": 0.862,
+      "loss": 0.8701,
       "step": 84280
     },
     {
       "epoch": 253.12,
-      "grad_norm": 7.261661529541016,
+      "grad_norm": 10.158432006835938,
       "learning_rate": 1.5625625625625627e-06,
-      "loss": 0.9672,
+      "loss": 0.9691,
       "step": 84290
     },
     {
       "epoch": 253.15,
-      "grad_norm": 5.605440616607666,
+      "grad_norm": 5.565643787384033,
       "learning_rate": 1.5615615615615617e-06,
-      "loss": 0.9192,
+      "loss": 0.9069,
       "step": 84300
     },
     {
       "epoch": 253.18,
-      "grad_norm": 5.887853622436523,
+      "grad_norm": 4.922473907470703,
       "learning_rate": 1.5605605605605607e-06,
-      "loss": 0.9009,
+      "loss": 0.8948,
       "step": 84310
     },
     {
       "epoch": 253.21,
-      "grad_norm": 7.247108459472656,
+      "grad_norm": 6.283976078033447,
       "learning_rate": 1.5595595595595597e-06,
-      "loss": 0.9287,
+      "loss": 0.9186,
       "step": 84320
     },
     {
       "epoch": 253.24,
-      "grad_norm": 6.522788047790527,
+      "grad_norm": 9.048806190490723,
       "learning_rate": 1.5585585585585585e-06,
-      "loss": 0.9618,
+      "loss": 0.9606,
       "step": 84330
     },
     {
       "epoch": 253.27,
-      "grad_norm": 7.813406467437744,
+      "grad_norm": 7.864142894744873,
       "learning_rate": 1.5575575575575578e-06,
-      "loss": 0.9712,
+      "loss": 0.981,
       "step": 84340
     },
     {
       "epoch": 253.3,
-      "grad_norm": 6.64035177230835,
+      "grad_norm": 6.8521342277526855,
       "learning_rate": 1.5565565565565566e-06,
-      "loss": 0.9534,
+      "loss": 0.9482,
       "step": 84350
     },
     {
       "epoch": 253.33,
-      "grad_norm": 6.611476898193359,
+      "grad_norm": 5.921032905578613,
       "learning_rate": 1.5555555555555558e-06,
-      "loss": 0.992,
+      "loss": 0.9837,
       "step": 84360
     },
     {
       "epoch": 253.36,
-      "grad_norm": 6.942587375640869,
+      "grad_norm": 6.361745834350586,
       "learning_rate": 1.5545545545545546e-06,
-      "loss": 0.9802,
+      "loss": 0.9742,
       "step": 84370
     },
     {
       "epoch": 253.39,
-      "grad_norm": 4.943953037261963,
+      "grad_norm": 4.909067153930664,
       "learning_rate": 1.5535535535535536e-06,
-      "loss": 0.8699,
+      "loss": 0.857,
       "step": 84380
     },
     {
       "epoch": 253.42,
-      "grad_norm": 10.061647415161133,
+      "grad_norm": 8.537304878234863,
       "learning_rate": 1.5525525525525526e-06,
-      "loss": 0.9791,
+      "loss": 0.9832,
       "step": 84390
     },
     {
       "epoch": 253.45,
-      "grad_norm": 5.06413459777832,
+      "grad_norm": 6.945516586303711,
       "learning_rate": 1.5515515515515517e-06,
-      "loss": 0.9155,
+      "loss": 0.9123,
       "step": 84400
     },
     {
       "epoch": 253.48,
-      "grad_norm": 7.4124579429626465,
+      "grad_norm": 8.369254112243652,
       "learning_rate": 1.5505505505505509e-06,
-      "loss": 0.8929,
+      "loss": 0.8861,
       "step": 84410
     },
     {
       "epoch": 253.51,
-      "grad_norm": 4.882434844970703,
+      "grad_norm": 6.73359489440918,
       "learning_rate": 1.5495495495495497e-06,
-      "loss": 0.9077,
+      "loss": 0.8991,
       "step": 84420
     },
     {
       "epoch": 253.54,
-      "grad_norm": 6.653915882110596,
+      "grad_norm": 6.246030330657959,
       "learning_rate": 1.5485485485485487e-06,
-      "loss": 0.9718,
+      "loss": 0.9626,
       "step": 84430
     },
     {
       "epoch": 253.57,
-      "grad_norm": 8.23105239868164,
+      "grad_norm": 7.062889575958252,
       "learning_rate": 1.5475475475475477e-06,
-      "loss": 1.0029,
+      "loss": 0.9969,
       "step": 84440
     },
     {
       "epoch": 253.6,
-      "grad_norm": 5.51410436630249,
+      "grad_norm": 5.995069980621338,
       "learning_rate": 1.5465465465465467e-06,
-      "loss": 0.9665,
+      "loss": 0.9566,
       "step": 84450
     },
     {
       "epoch": 253.63,
-      "grad_norm": 6.008918762207031,
+      "grad_norm": 6.265349864959717,
       "learning_rate": 1.5455455455455456e-06,
-      "loss": 0.9209,
+      "loss": 0.91,
       "step": 84460
     },
     {
       "epoch": 253.66,
-      "grad_norm": 6.537206172943115,
+      "grad_norm": 9.935664176940918,
       "learning_rate": 1.5445445445445448e-06,
-      "loss": 0.8895,
+      "loss": 0.8911,
       "step": 84470
     },
     {
       "epoch": 253.69,
-      "grad_norm": 8.06179428100586,
+      "grad_norm": 11.493086814880371,
       "learning_rate": 1.5435435435435436e-06,
-      "loss": 0.8913,
+      "loss": 0.8922,
       "step": 84480
     },
     {
       "epoch": 253.72,
-      "grad_norm": 8.43299388885498,
+      "grad_norm": 7.621009826660156,
       "learning_rate": 1.5425425425425428e-06,
-      "loss": 0.8762,
+      "loss": 0.8861,
       "step": 84490
     },
     {
       "epoch": 253.75,
-      "grad_norm": 8.541954040527344,
+      "grad_norm": 6.810401916503906,
       "learning_rate": 1.5415415415415416e-06,
-      "loss": 0.8812,
+      "loss": 0.8819,
       "step": 84500
     },
     {
       "epoch": 253.78,
-      "grad_norm": 9.586172103881836,
+      "grad_norm": 12.53210163116455,
       "learning_rate": 1.5405405405405409e-06,
-      "loss": 0.9221,
+      "loss": 0.9257,
       "step": 84510
     },
     {
       "epoch": 253.81,
-      "grad_norm": 6.257017135620117,
+      "grad_norm": 7.261014461517334,
       "learning_rate": 1.5395395395395397e-06,
-      "loss": 0.8386,
+      "loss": 0.8412,
       "step": 84520
     },
     {
       "epoch": 253.84,
-      "grad_norm": 6.340400695800781,
+      "grad_norm": 6.711306571960449,
       "learning_rate": 1.5385385385385387e-06,
-      "loss": 0.9768,
+      "loss": 0.9756,
       "step": 84530
     },
     {
       "epoch": 253.87,
-      "grad_norm": 6.842611789703369,
+      "grad_norm": 6.646900653839111,
       "learning_rate": 1.5375375375375377e-06,
-      "loss": 0.8766,
+      "loss": 0.876,
       "step": 84540
     },
     {
       "epoch": 253.9,
-      "grad_norm": 6.2786545753479,
+      "grad_norm": 6.9448466300964355,
       "learning_rate": 1.5365365365365367e-06,
-      "loss": 1.0157,
+      "loss": 1.0155,
       "step": 84550
     },
     {
       "epoch": 253.93,
-      "grad_norm": 6.93209981918335,
+      "grad_norm": 7.026188850402832,
       "learning_rate": 1.5355355355355355e-06,
-      "loss": 0.883,
+      "loss": 0.8921,
       "step": 84560
     },
     {
       "epoch": 253.96,
-      "grad_norm": 6.133819580078125,
+      "grad_norm": 6.291578769683838,
       "learning_rate": 1.5345345345345348e-06,
-      "loss": 0.9102,
+      "loss": 0.9039,
       "step": 84570
     },
     {
       "epoch": 253.99,
-      "grad_norm": 5.14694356918335,
+      "grad_norm": 5.335271835327148,
       "learning_rate": 1.5335335335335336e-06,
-      "loss": 1.0041,
+      "loss": 0.9938,
       "step": 84580
     },
     {
       "epoch": 254.0,
-      "eval_accuracy": 0.8267,
-      "eval_loss": 0.571767270565033,
-      "eval_runtime": 5.4918,
-      "eval_samples_per_second": 1820.882,
-      "eval_steps_per_second": 7.284,
+      "eval_accuracy": 0.8303,
+      "eval_loss": 0.5709617137908936,
+      "eval_runtime": 5.3347,
+      "eval_samples_per_second": 1874.527,
+      "eval_steps_per_second": 7.498,
       "step": 84582
     },
     {
       "epoch": 254.02,
-      "grad_norm": 8.18069076538086,
+      "grad_norm": 5.762665271759033,
       "learning_rate": 1.5325325325325328e-06,
-      "loss": 1.0205,
+      "loss": 1.0127,
       "step": 84590
     },
     {
       "epoch": 254.05,
-      "grad_norm": 6.9300150871276855,
+      "grad_norm": 5.840947151184082,
       "learning_rate": 1.5315315315315316e-06,
-      "loss": 0.9795,
+      "loss": 0.9744,
       "step": 84600
     },
     {
       "epoch": 254.08,
-      "grad_norm": 8.79443073272705,
+      "grad_norm": 7.614108562469482,
       "learning_rate": 1.5305305305305306e-06,
-      "loss": 0.8417,
+      "loss": 0.8474,
       "step": 84610
     },
     {
       "epoch": 254.11,
-      "grad_norm": 6.620852470397949,
+      "grad_norm": 9.71542739868164,
       "learning_rate": 1.5295295295295296e-06,
-      "loss": 0.8867,
+      "loss": 0.8964,
       "step": 84620
     },
     {
       "epoch": 254.14,
-      "grad_norm": 5.753002643585205,
+      "grad_norm": 10.431770324707031,
       "learning_rate": 1.5285285285285287e-06,
-      "loss": 0.926,
+      "loss": 0.9276,
       "step": 84630
     },
     {
       "epoch": 254.17,
-      "grad_norm": 6.533583164215088,
+      "grad_norm": 6.640063762664795,
       "learning_rate": 1.5275275275275275e-06,
-      "loss": 0.8354,
+      "loss": 0.8292,
       "step": 84640
     },
     {
       "epoch": 254.2,
-      "grad_norm": 7.511693000793457,
+      "grad_norm": 8.465803146362305,
       "learning_rate": 1.5265265265265267e-06,
-      "loss": 0.8941,
+      "loss": 0.8972,
       "step": 84650
     },
     {
       "epoch": 254.23,
-      "grad_norm": 6.715672016143799,
+      "grad_norm": 6.547680854797363,
       "learning_rate": 1.5255255255255255e-06,
-      "loss": 0.887,
+      "loss": 0.8884,
       "step": 84660
     },
     {
       "epoch": 254.26,
-      "grad_norm": 7.71122932434082,
+      "grad_norm": 10.179974555969238,
       "learning_rate": 1.5245245245245247e-06,
-      "loss": 0.9198,
+      "loss": 0.9242,
       "step": 84670
     },
     {
       "epoch": 254.29,
-      "grad_norm": 4.6507344245910645,
+      "grad_norm": 6.530936241149902,
       "learning_rate": 1.5235235235235237e-06,
-      "loss": 0.8805,
+      "loss": 0.88,
       "step": 84680
     },
     {
       "epoch": 254.32,
-      "grad_norm": 6.547906398773193,
+      "grad_norm": 6.34619140625,
       "learning_rate": 1.5225225225225225e-06,
-      "loss": 0.9718,
+      "loss": 0.9812,
       "step": 84690
     },
     {
       "epoch": 254.35,
-      "grad_norm": 11.470182418823242,
+      "grad_norm": 9.991228103637695,
       "learning_rate": 1.5215215215215218e-06,
-      "loss": 0.913,
+      "loss": 0.911,
       "step": 84700
     },
     {
       "epoch": 254.38,
-      "grad_norm": 6.340887546539307,
+      "grad_norm": 6.576843738555908,
       "learning_rate": 1.5205205205205206e-06,
-      "loss": 0.9159,
+      "loss": 0.9188,
       "step": 84710
     },
     {
       "epoch": 254.41,
-      "grad_norm": 8.461610794067383,
+      "grad_norm": 9.158276557922363,
       "learning_rate": 1.5195195195195198e-06,
-      "loss": 0.9018,
+      "loss": 0.9033,
       "step": 84720
     },
     {
       "epoch": 254.44,
-      "grad_norm": 7.558836460113525,
+      "grad_norm": 6.991607189178467,
       "learning_rate": 1.5185185185185186e-06,
-      "loss": 0.8886,
+      "loss": 0.8968,
       "step": 84730
     },
     {
       "epoch": 254.47,
-      "grad_norm": 6.140642166137695,
+      "grad_norm": 6.535127639770508,
       "learning_rate": 1.5175175175175176e-06,
-      "loss": 0.874,
+      "loss": 0.8656,
       "step": 84740
     },
     {
       "epoch": 254.5,
-      "grad_norm": 7.548120975494385,
+      "grad_norm": 11.475275993347168,
       "learning_rate": 1.5165165165165167e-06,
-      "loss": 0.8502,
+      "loss": 0.8567,
       "step": 84750
     },
     {
       "epoch": 254.53,
-      "grad_norm": 7.601109981536865,
+      "grad_norm": 6.579020023345947,
       "learning_rate": 1.5155155155155157e-06,
-      "loss": 0.9428,
+      "loss": 0.9431,
       "step": 84760
     },
     {
       "epoch": 254.56,
-      "grad_norm": 7.604586601257324,
+      "grad_norm": 6.739378929138184,
       "learning_rate": 1.5145145145145145e-06,
-      "loss": 0.8521,
+      "loss": 0.8519,
       "step": 84770
     },
     {
       "epoch": 254.59,
-      "grad_norm": 6.815893173217773,
+      "grad_norm": 7.493529319763184,
       "learning_rate": 1.5135135135135137e-06,
-      "loss": 0.9103,
+      "loss": 0.9231,
       "step": 84780
     },
     {
       "epoch": 254.62,
-      "grad_norm": 5.949065208435059,
+      "grad_norm": 8.494226455688477,
       "learning_rate": 1.5125125125125125e-06,
-      "loss": 0.8538,
+      "loss": 0.8474,
       "step": 84790
     },
     {
       "epoch": 254.65,
-      "grad_norm": 5.29038667678833,
+      "grad_norm": 7.613640308380127,
       "learning_rate": 1.5115115115115118e-06,
-      "loss": 0.8908,
+      "loss": 0.8998,
       "step": 84800
     },
     {
       "epoch": 254.68,
-      "grad_norm": 5.984964370727539,
+      "grad_norm": 6.958147048950195,
       "learning_rate": 1.5105105105105106e-06,
-      "loss": 0.9028,
+      "loss": 0.8935,
       "step": 84810
     },
     {
       "epoch": 254.71,
-      "grad_norm": 4.760894775390625,
+      "grad_norm": 6.053483963012695,
       "learning_rate": 1.5095095095095098e-06,
-      "loss": 0.9204,
+      "loss": 0.921,
       "step": 84820
     },
     {
       "epoch": 254.74,
-      "grad_norm": 7.966403961181641,
+      "grad_norm": 7.520346164703369,
       "learning_rate": 1.5085085085085086e-06,
-      "loss": 0.9674,
+      "loss": 0.9601,
       "step": 84830
     },
     {
       "epoch": 254.77,
-      "grad_norm": 7.730035305023193,
+      "grad_norm": 8.379653930664062,
       "learning_rate": 1.5075075075075076e-06,
-      "loss": 0.9074,
+      "loss": 0.9097,
       "step": 84840
     },
     {
       "epoch": 254.8,
-      "grad_norm": 7.353894233703613,
+      "grad_norm": 8.262823104858398,
       "learning_rate": 1.5065065065065066e-06,
-      "loss": 0.9242,
+      "loss": 0.9209,
       "step": 84850
     },
     {
       "epoch": 254.83,
-      "grad_norm": 5.562304973602295,
+      "grad_norm": 5.410043239593506,
       "learning_rate": 1.5055055055055056e-06,
-      "loss": 0.8324,
+      "loss": 0.8232,
       "step": 84860
     },
     {
       "epoch": 254.86,
-      "grad_norm": 5.34063196182251,
+      "grad_norm": 6.3639984130859375,
       "learning_rate": 1.5045045045045045e-06,
-      "loss": 0.9333,
+      "loss": 0.9314,
       "step": 84870
     },
     {
       "epoch": 254.89,
-      "grad_norm": 5.588039398193359,
+      "grad_norm": 5.869692325592041,
       "learning_rate": 1.5035035035035037e-06,
-      "loss": 0.9046,
+      "loss": 0.9142,
       "step": 84880
     },
     {
       "epoch": 254.92,
-      "grad_norm": 6.6342973709106445,
+      "grad_norm": 4.796920299530029,
       "learning_rate": 1.5025025025025025e-06,
-      "loss": 0.9052,
+      "loss": 0.9086,
       "step": 84890
     },
     {
       "epoch": 254.95,
-      "grad_norm": 5.139523029327393,
+      "grad_norm": 6.648270606994629,
       "learning_rate": 1.5015015015015017e-06,
-      "loss": 0.8448,
+      "loss": 0.8434,
       "step": 84900
     },
     {
       "epoch": 254.98,
-      "grad_norm": 7.282650470733643,
+      "grad_norm": 6.771457195281982,
       "learning_rate": 1.5005005005005005e-06,
-      "loss": 0.8924,
+      "loss": 0.8881,
       "step": 84910
     },
     {
       "epoch": 255.0,
-      "eval_accuracy": 0.8253,
-      "eval_loss": 0.5740662813186646,
-      "eval_runtime": 5.3555,
-      "eval_samples_per_second": 1867.229,
-      "eval_steps_per_second": 7.469,
+      "eval_accuracy": 0.8276,
+      "eval_loss": 0.5730119347572327,
+      "eval_runtime": 5.1741,
+      "eval_samples_per_second": 1932.69,
+      "eval_steps_per_second": 7.731,
       "step": 84915
     },
     {
       "epoch": 255.02,
-      "grad_norm": 6.377302169799805,
+      "grad_norm": 6.897431373596191,
       "learning_rate": 1.4994994994994995e-06,
-      "loss": 1.0956,
+      "loss": 1.1023,
       "step": 84920
     },
     {
       "epoch": 255.05,
-      "grad_norm": 6.386617660522461,
+      "grad_norm": 6.157604217529297,
       "learning_rate": 1.4984984984984986e-06,
-      "loss": 0.9247,
+      "loss": 0.926,
       "step": 84930
     },
     {
       "epoch": 255.08,
-      "grad_norm": 7.8961310386657715,
+      "grad_norm": 5.17223596572876,
       "learning_rate": 1.4974974974974976e-06,
-      "loss": 0.8468,
+      "loss": 0.8582,
       "step": 84940
     },
     {
       "epoch": 255.11,
-      "grad_norm": 6.612208366394043,
+      "grad_norm": 5.91739559173584,
       "learning_rate": 1.4964964964964968e-06,
-      "loss": 0.898,
+      "loss": 0.9092,
       "step": 84950
     },
     {
       "epoch": 255.14,
-      "grad_norm": 6.582457065582275,
+      "grad_norm": 7.560092926025391,
       "learning_rate": 1.4954954954954956e-06,
-      "loss": 0.9673,
+      "loss": 0.9699,
       "step": 84960
     },
     {
       "epoch": 255.17,
-      "grad_norm": 7.979331970214844,
+      "grad_norm": 6.971930980682373,
       "learning_rate": 1.4944944944944946e-06,
-      "loss": 0.9818,
+      "loss": 0.9935,
       "step": 84970
     },
     {
       "epoch": 255.2,
-      "grad_norm": 6.643911361694336,
+      "grad_norm": 5.194787979125977,
       "learning_rate": 1.4934934934934937e-06,
-      "loss": 0.9027,
+      "loss": 0.8918,
       "step": 84980
     },
     {
       "epoch": 255.23,
-      "grad_norm": 4.691190242767334,
+      "grad_norm": 5.568471908569336,
       "learning_rate": 1.4924924924924927e-06,
-      "loss": 0.9189,
+      "loss": 0.9186,
       "step": 84990
     },
     {
       "epoch": 255.26,
-      "grad_norm": 5.48145055770874,
+      "grad_norm": 10.326565742492676,
       "learning_rate": 1.4914914914914915e-06,
-      "loss": 0.924,
+      "loss": 0.9251,
       "step": 85000
     },
     {
       "epoch": 255.29,
-      "grad_norm": 5.364309310913086,
+      "grad_norm": 7.4031782150268555,
       "learning_rate": 1.4904904904904907e-06,
-      "loss": 0.9197,
+      "loss": 0.9274,
       "step": 85010
     },
     {
       "epoch": 255.32,
-      "grad_norm": 6.54665470123291,
+      "grad_norm": 7.208024501800537,
       "learning_rate": 1.4894894894894895e-06,
-      "loss": 0.9069,
+      "loss": 0.9125,
       "step": 85020
     },
     {
       "epoch": 255.35,
-      "grad_norm": 8.25593090057373,
+      "grad_norm": 7.005957126617432,
       "learning_rate": 1.4884884884884887e-06,
-      "loss": 0.9119,
+      "loss": 0.8914,
       "step": 85030
     },
     {
       "epoch": 255.38,
-      "grad_norm": 6.676808834075928,
+      "grad_norm": 8.226602554321289,
       "learning_rate": 1.4874874874874876e-06,
-      "loss": 0.879,
+      "loss": 0.8861,
       "step": 85040
     },
     {
       "epoch": 255.41,
-      "grad_norm": 6.44677734375,
+      "grad_norm": 6.628051280975342,
       "learning_rate": 1.4864864864864868e-06,
-      "loss": 0.9121,
+      "loss": 0.9142,
       "step": 85050
     },
     {
       "epoch": 255.44,
-      "grad_norm": 6.552823543548584,
+      "grad_norm": 7.625163555145264,
       "learning_rate": 1.4854854854854856e-06,
-      "loss": 0.9075,
+      "loss": 0.9057,
       "step": 85060
     },
     {
       "epoch": 255.47,
-      "grad_norm": 5.190671443939209,
+      "grad_norm": 5.873938083648682,
       "learning_rate": 1.4844844844844846e-06,
-      "loss": 0.9253,
+      "loss": 0.9286,
       "step": 85070
     },
     {
       "epoch": 255.5,
-      "grad_norm": 5.751422882080078,
+      "grad_norm": 6.8152923583984375,
       "learning_rate": 1.4834834834834836e-06,
-      "loss": 0.8783,
+      "loss": 0.8697,
       "step": 85080
     },
     {
       "epoch": 255.53,
-      "grad_norm": 6.975903034210205,
+      "grad_norm": 6.5579657554626465,
       "learning_rate": 1.4824824824824826e-06,
-      "loss": 0.9245,
+      "loss": 0.9214,
       "step": 85090
     },
     {
       "epoch": 255.56,
-      "grad_norm": 5.245737552642822,
+      "grad_norm": 5.878832817077637,
       "learning_rate": 1.4814814814814815e-06,
-      "loss": 0.9223,
+      "loss": 0.9205,
       "step": 85100
     },
     {
       "epoch": 255.59,
-      "grad_norm": 6.387447834014893,
+      "grad_norm": 7.900144100189209,
       "learning_rate": 1.4804804804804807e-06,
-      "loss": 0.9249,
+      "loss": 0.9342,
       "step": 85110
     },
     {
       "epoch": 255.62,
-      "grad_norm": 6.8927483558654785,
+      "grad_norm": 6.681351661682129,
       "learning_rate": 1.4794794794794795e-06,
-      "loss": 0.9144,
+      "loss": 0.8998,
       "step": 85120
     },
     {
       "epoch": 255.65,
-      "grad_norm": 8.244065284729004,
+      "grad_norm": 8.937679290771484,
       "learning_rate": 1.4784784784784787e-06,
-      "loss": 0.9153,
+      "loss": 0.9139,
       "step": 85130
     },
     {
       "epoch": 255.68,
-      "grad_norm": 6.090306282043457,
+      "grad_norm": 6.100274562835693,
       "learning_rate": 1.4774774774774775e-06,
-      "loss": 0.9216,
+      "loss": 0.915,
       "step": 85140
     },
     {
       "epoch": 255.71,
-      "grad_norm": 7.000442981719971,
+      "grad_norm": 5.968214511871338,
       "learning_rate": 1.4764764764764765e-06,
-      "loss": 0.8914,
+      "loss": 0.8836,
       "step": 85150
     },
     {
       "epoch": 255.74,
-      "grad_norm": 9.285639762878418,
+      "grad_norm": 9.58726978302002,
       "learning_rate": 1.4754754754754756e-06,
-      "loss": 0.8966,
+      "loss": 0.8933,
       "step": 85160
     },
     {
       "epoch": 255.77,
-      "grad_norm": 7.161376953125,
+      "grad_norm": 6.380039215087891,
       "learning_rate": 1.4744744744744746e-06,
-      "loss": 0.8538,
+      "loss": 0.8509,
       "step": 85170
     },
     {
       "epoch": 255.8,
-      "grad_norm": 7.989756107330322,
+      "grad_norm": 6.3738484382629395,
       "learning_rate": 1.4734734734734734e-06,
-      "loss": 0.9505,
+      "loss": 0.9592,
       "step": 85180
     },
     {
       "epoch": 255.83,
-      "grad_norm": 6.42799711227417,
+      "grad_norm": 9.240232467651367,
       "learning_rate": 1.4724724724724726e-06,
-      "loss": 0.866,
+      "loss": 0.8533,
       "step": 85190
     },
     {
       "epoch": 255.86,
-      "grad_norm": 9.54860782623291,
+      "grad_norm": 10.576532363891602,
       "learning_rate": 1.4714714714714714e-06,
-      "loss": 0.9951,
+      "loss": 0.9843,
       "step": 85200
     },
     {
       "epoch": 255.89,
-      "grad_norm": 5.981546878814697,
+      "grad_norm": 8.72454833984375,
       "learning_rate": 1.4704704704704707e-06,
-      "loss": 0.9296,
+      "loss": 0.9307,
       "step": 85210
     },
     {
       "epoch": 255.92,
-      "grad_norm": 5.3776535987854,
+      "grad_norm": 7.345282077789307,
       "learning_rate": 1.4694694694694697e-06,
-      "loss": 0.8846,
+      "loss": 0.8705,
       "step": 85220
     },
     {
       "epoch": 255.95,
-      "grad_norm": 6.826462745666504,
+      "grad_norm": 5.641618728637695,
       "learning_rate": 1.4684684684684685e-06,
-      "loss": 0.8905,
+      "loss": 0.8922,
       "step": 85230
     },
     {
       "epoch": 255.98,
-      "grad_norm": 7.969820499420166,
+      "grad_norm": 9.193278312683105,
       "learning_rate": 1.4674674674674677e-06,
-      "loss": 0.9587,
+      "loss": 0.9577,
       "step": 85240
     },
     {
       "epoch": 256.0,
-      "eval_accuracy": 0.8277,
-      "eval_loss": 0.5665184855461121,
-      "eval_runtime": 5.1744,
-      "eval_samples_per_second": 1932.586,
-      "eval_steps_per_second": 7.73,
+      "eval_accuracy": 0.8298,
+      "eval_loss": 0.5663203597068787,
+      "eval_runtime": 5.1375,
+      "eval_samples_per_second": 1946.465,
+      "eval_steps_per_second": 7.786,
       "step": 85248
     },
     {
       "epoch": 256.01,
-      "grad_norm": 6.416565418243408,
+      "grad_norm": 6.231107234954834,
       "learning_rate": 1.4664664664664665e-06,
-      "loss": 1.021,
+      "loss": 1.0094,
       "step": 85250
     },
     {
       "epoch": 256.04,
-      "grad_norm": 6.562569618225098,
+      "grad_norm": 5.685012340545654,
       "learning_rate": 1.4654654654654657e-06,
-      "loss": 0.8885,
+      "loss": 0.8716,
       "step": 85260
     },
     {
       "epoch": 256.07,
-      "grad_norm": 10.416216850280762,
+      "grad_norm": 10.583894729614258,
       "learning_rate": 1.4644644644644646e-06,
-      "loss": 0.9291,
+      "loss": 0.9344,
       "step": 85270
     },
     {
       "epoch": 256.1,
-      "grad_norm": 6.177861213684082,
+      "grad_norm": 6.383655071258545,
       "learning_rate": 1.4634634634634638e-06,
-      "loss": 0.8988,
+      "loss": 0.8954,
       "step": 85280
     },
     {
       "epoch": 256.13,
-      "grad_norm": 8.727339744567871,
+      "grad_norm": 6.301596164703369,
       "learning_rate": 1.4624624624624626e-06,
-      "loss": 0.9249,
+      "loss": 0.9283,
       "step": 85290
     },
     {
       "epoch": 256.16,
-      "grad_norm": 6.53867769241333,
+      "grad_norm": 6.4253644943237305,
       "learning_rate": 1.4614614614614616e-06,
-      "loss": 0.9425,
+      "loss": 0.9412,
       "step": 85300
     },
     {
       "epoch": 256.19,
-      "grad_norm": 7.678091526031494,
+      "grad_norm": 7.3071088790893555,
       "learning_rate": 1.4604604604604606e-06,
-      "loss": 0.9104,
+      "loss": 0.9126,
       "step": 85310
     },
     {
       "epoch": 256.22,
-      "grad_norm": 5.742837905883789,
+      "grad_norm": 6.868616580963135,
       "learning_rate": 1.4594594594594596e-06,
-      "loss": 0.9279,
+      "loss": 0.929,
       "step": 85320
     },
     {
       "epoch": 256.25,
-      "grad_norm": 7.092319965362549,
+      "grad_norm": 9.040030479431152,
       "learning_rate": 1.4584584584584585e-06,
-      "loss": 0.9343,
+      "loss": 0.9358,
       "step": 85330
     },
     {
       "epoch": 256.28,
-      "grad_norm": 6.13466739654541,
+      "grad_norm": 8.17119312286377,
       "learning_rate": 1.4574574574574577e-06,
-      "loss": 0.9013,
+      "loss": 0.8975,
       "step": 85340
     },
     {
       "epoch": 256.31,
-      "grad_norm": 4.773816108703613,
+      "grad_norm": 5.887320518493652,
       "learning_rate": 1.4564564564564565e-06,
-      "loss": 0.8768,
+      "loss": 0.8792,
       "step": 85350
     },
     {
       "epoch": 256.34,
-      "grad_norm": 6.24235200881958,
+      "grad_norm": 6.025355339050293,
       "learning_rate": 1.4554554554554557e-06,
-      "loss": 0.8735,
+      "loss": 0.8661,
       "step": 85360
     },
     {
       "epoch": 256.37,
-      "grad_norm": 5.496309280395508,
+      "grad_norm": 5.986965179443359,
       "learning_rate": 1.4544544544544545e-06,
-      "loss": 0.8896,
+      "loss": 0.8922,
       "step": 85370
     },
     {
       "epoch": 256.4,
-      "grad_norm": 8.140124320983887,
+      "grad_norm": 7.646963596343994,
       "learning_rate": 1.4534534534534535e-06,
-      "loss": 0.8731,
+      "loss": 0.8766,
       "step": 85380
     },
     {
       "epoch": 256.43,
-      "grad_norm": 6.465211391448975,
+      "grad_norm": 6.640980243682861,
       "learning_rate": 1.4524524524524526e-06,
-      "loss": 0.8854,
+      "loss": 0.8837,
       "step": 85390
     },
     {
       "epoch": 256.46,
-      "grad_norm": 7.59878396987915,
+      "grad_norm": 6.493606090545654,
       "learning_rate": 1.4514514514514516e-06,
-      "loss": 0.9106,
+      "loss": 0.9096,
       "step": 85400
     },
     {
       "epoch": 256.49,
-      "grad_norm": 6.414132595062256,
+      "grad_norm": 7.748528480529785,
       "learning_rate": 1.4504504504504504e-06,
-      "loss": 0.9338,
+      "loss": 0.9351,
       "step": 85410
     },
     {
       "epoch": 256.52,
-      "grad_norm": 6.7392754554748535,
+      "grad_norm": 7.655106067657471,
       "learning_rate": 1.4494494494494496e-06,
-      "loss": 0.92,
+      "loss": 0.9225,
       "step": 85420
     },
     {
       "epoch": 256.55,
-      "grad_norm": 5.166225433349609,
+      "grad_norm": 5.413015842437744,
       "learning_rate": 1.4484484484484484e-06,
-      "loss": 0.9754,
+      "loss": 0.9731,
       "step": 85430
     },
     {
       "epoch": 256.58,
-      "grad_norm": 4.429004192352295,
+      "grad_norm": 5.31254768371582,
       "learning_rate": 1.4474474474474477e-06,
-      "loss": 0.9583,
+      "loss": 0.9601,
       "step": 85440
     },
     {
       "epoch": 256.61,
-      "grad_norm": 5.319735050201416,
+      "grad_norm": 5.8816819190979,
       "learning_rate": 1.4464464464464465e-06,
-      "loss": 0.8855,
+      "loss": 0.8871,
       "step": 85450
     },
     {
       "epoch": 256.64,
-      "grad_norm": 6.062988758087158,
+      "grad_norm": 7.158030033111572,
       "learning_rate": 1.4454454454454455e-06,
-      "loss": 0.8958,
+      "loss": 0.8932,
       "step": 85460
     },
     {
       "epoch": 256.67,
-      "grad_norm": 6.250340461730957,
+      "grad_norm": 6.021441459655762,
       "learning_rate": 1.4444444444444445e-06,
-      "loss": 0.9407,
+      "loss": 0.9436,
       "step": 85470
     },
     {
       "epoch": 256.7,
-      "grad_norm": 5.062887191772461,
+      "grad_norm": 6.086477279663086,
       "learning_rate": 1.4434434434434435e-06,
-      "loss": 0.9366,
+      "loss": 0.9389,
       "step": 85480
     },
     {
       "epoch": 256.73,
-      "grad_norm": 6.425777435302734,
+      "grad_norm": 6.531158924102783,
       "learning_rate": 1.4424424424424427e-06,
-      "loss": 0.9218,
+      "loss": 0.92,
       "step": 85490
     },
     {
       "epoch": 256.76,
-      "grad_norm": 6.75443696975708,
+      "grad_norm": 7.323708534240723,
       "learning_rate": 1.4414414414414416e-06,
-      "loss": 0.9173,
+      "loss": 0.9191,
       "step": 85500
     },
     {
       "epoch": 256.79,
-      "grad_norm": 6.575555801391602,
+      "grad_norm": 8.371753692626953,
       "learning_rate": 1.4404404404404408e-06,
-      "loss": 0.8795,
+      "loss": 0.8831,
       "step": 85510
     },
     {
       "epoch": 256.82,
-      "grad_norm": 5.830583572387695,
+      "grad_norm": 7.382374286651611,
       "learning_rate": 1.4394394394394396e-06,
-      "loss": 0.8826,
+      "loss": 0.8903,
       "step": 85520
     },
     {
       "epoch": 256.85,
-      "grad_norm": 7.606679439544678,
+      "grad_norm": 6.328585147857666,
       "learning_rate": 1.4384384384384386e-06,
-      "loss": 0.9218,
+      "loss": 0.9176,
       "step": 85530
     },
     {
       "epoch": 256.88,
-      "grad_norm": 5.915466785430908,
+      "grad_norm": 6.915104866027832,
       "learning_rate": 1.4374374374374376e-06,
-      "loss": 0.9583,
+      "loss": 0.9575,
       "step": 85540
     },
     {
       "epoch": 256.91,
-      "grad_norm": 6.911581039428711,
+      "grad_norm": 7.377707004547119,
       "learning_rate": 1.4364364364364366e-06,
-      "loss": 0.9768,
+      "loss": 0.9671,
       "step": 85550
     },
     {
       "epoch": 256.94,
-      "grad_norm": 6.235986709594727,
+      "grad_norm": 5.675912857055664,
       "learning_rate": 1.4354354354354354e-06,
-      "loss": 0.8738,
+      "loss": 0.8822,
       "step": 85560
     },
     {
       "epoch": 256.97,
-      "grad_norm": 7.2933268547058105,
+      "grad_norm": 5.294546604156494,
       "learning_rate": 1.4344344344344347e-06,
-      "loss": 0.9429,
+      "loss": 0.9389,
       "step": 85570
     },
     {
       "epoch": 257.0,
-      "grad_norm": 9.095274925231934,
+      "grad_norm": 13.37707805633545,
       "learning_rate": 1.4334334334334335e-06,
-      "loss": 0.959,
+      "loss": 0.9663,
       "step": 85580
     },
     {
       "epoch": 257.0,
-      "eval_accuracy": 0.8292,
-      "eval_loss": 0.5679036378860474,
-      "eval_runtime": 5.6079,
-      "eval_samples_per_second": 1783.186,
-      "eval_steps_per_second": 7.133,
+      "eval_accuracy": 0.8305,
+      "eval_loss": 0.5667198896408081,
+      "eval_runtime": 5.2021,
+      "eval_samples_per_second": 1922.297,
+      "eval_steps_per_second": 7.689,
       "step": 85581
     },
     {
       "epoch": 257.03,
-      "grad_norm": 7.560002326965332,
+      "grad_norm": 8.435630798339844,
       "learning_rate": 1.4324324324324327e-06,
-      "loss": 1.0271,
+      "loss": 1.0033,
       "step": 85590
     },
     {
       "epoch": 257.06,
-      "grad_norm": 4.8626790046691895,
+      "grad_norm": 5.226444244384766,
       "learning_rate": 1.4314314314314315e-06,
-      "loss": 0.8298,
+      "loss": 0.8313,
       "step": 85600
     },
     {
       "epoch": 257.09,
-      "grad_norm": 5.925969123840332,
+      "grad_norm": 7.9320902824401855,
       "learning_rate": 1.4304304304304305e-06,
-      "loss": 0.9039,
+      "loss": 0.9078,
       "step": 85610
     },
     {
       "epoch": 257.12,
-      "grad_norm": 6.273458480834961,
+      "grad_norm": 7.658085346221924,
       "learning_rate": 1.4294294294294296e-06,
-      "loss": 0.9537,
+      "loss": 0.9626,
       "step": 85620
     },
     {
       "epoch": 257.15,
-      "grad_norm": 8.40398120880127,
+      "grad_norm": 10.230605125427246,
       "learning_rate": 1.4284284284284286e-06,
-      "loss": 1.0328,
+      "loss": 1.0211,
       "step": 85630
     },
     {
       "epoch": 257.18,
-      "grad_norm": 6.493854522705078,
+      "grad_norm": 6.88485860824585,
       "learning_rate": 1.4274274274274274e-06,
-      "loss": 0.8857,
+      "loss": 0.8724,
       "step": 85640
     },
     {
       "epoch": 257.21,
-      "grad_norm": 6.513256549835205,
+      "grad_norm": 7.060079097747803,
       "learning_rate": 1.4264264264264266e-06,
-      "loss": 0.8726,
+      "loss": 0.8622,
       "step": 85650
     },
     {
       "epoch": 257.24,
-      "grad_norm": 5.532011032104492,
+      "grad_norm": 6.997701644897461,
       "learning_rate": 1.4254254254254254e-06,
-      "loss": 0.8861,
+      "loss": 0.8768,
       "step": 85660
     },
     {
       "epoch": 257.27,
-      "grad_norm": 5.654436111450195,
+      "grad_norm": 6.983120441436768,
       "learning_rate": 1.4244244244244247e-06,
-      "loss": 0.8673,
+      "loss": 0.862,
       "step": 85670
     },
     {
       "epoch": 257.3,
-      "grad_norm": 6.979704856872559,
+      "grad_norm": 5.945994853973389,
       "learning_rate": 1.4234234234234235e-06,
-      "loss": 0.9237,
+      "loss": 0.9184,
       "step": 85680
     },
     {
       "epoch": 257.33,
-      "grad_norm": 7.596399307250977,
+      "grad_norm": 7.625288486480713,
       "learning_rate": 1.4224224224224225e-06,
-      "loss": 0.9165,
+      "loss": 0.9098,
       "step": 85690
     },
     {
       "epoch": 257.36,
-      "grad_norm": 5.338690280914307,
+      "grad_norm": 5.199169158935547,
       "learning_rate": 1.4214214214214215e-06,
-      "loss": 0.9149,
+      "loss": 0.9097,
       "step": 85700
     },
     {
       "epoch": 257.39,
-      "grad_norm": 5.071985721588135,
+      "grad_norm": 5.5570244789123535,
       "learning_rate": 1.4204204204204205e-06,
-      "loss": 0.9341,
+      "loss": 0.932,
       "step": 85710
     },
     {
       "epoch": 257.42,
-      "grad_norm": 6.517214298248291,
+      "grad_norm": 5.978963851928711,
       "learning_rate": 1.4194194194194193e-06,
-      "loss": 0.9477,
+      "loss": 0.9608,
       "step": 85720
     },
     {
       "epoch": 257.45,
-      "grad_norm": 7.4336724281311035,
+      "grad_norm": 7.819144248962402,
       "learning_rate": 1.4184184184184185e-06,
-      "loss": 0.9312,
+      "loss": 0.9264,
       "step": 85730
     },
     {
       "epoch": 257.48,
-      "grad_norm": 6.355810642242432,
+      "grad_norm": 9.145334243774414,
       "learning_rate": 1.4174174174174174e-06,
-      "loss": 0.8587,
+      "loss": 0.8528,
       "step": 85740
     },
     {
       "epoch": 257.51,
-      "grad_norm": 6.924180507659912,
+      "grad_norm": 6.380894184112549,
       "learning_rate": 1.4164164164164166e-06,
-      "loss": 0.9074,
+      "loss": 0.8999,
       "step": 85750
     },
     {
       "epoch": 257.54,
-      "grad_norm": 6.34348201751709,
+      "grad_norm": 6.709489345550537,
       "learning_rate": 1.4154154154154156e-06,
-      "loss": 0.8599,
+      "loss": 0.8591,
       "step": 85760
     },
     {
       "epoch": 257.57,
-      "grad_norm": 8.203125953674316,
+      "grad_norm": 10.214401245117188,
       "learning_rate": 1.4144144144144144e-06,
-      "loss": 0.8534,
+      "loss": 0.8544,
       "step": 85770
     },
     {
       "epoch": 257.6,
-      "grad_norm": 5.833655834197998,
+      "grad_norm": 6.540119647979736,
       "learning_rate": 1.4134134134134136e-06,
-      "loss": 0.9459,
+      "loss": 0.9478,
       "step": 85780
     },
     {
       "epoch": 257.63,
-      "grad_norm": 7.578148365020752,
+      "grad_norm": 9.267709732055664,
       "learning_rate": 1.4124124124124124e-06,
-      "loss": 0.903,
+      "loss": 0.9112,
       "step": 85790
     },
     {
       "epoch": 257.66,
-      "grad_norm": 5.982785701751709,
+      "grad_norm": 5.38242244720459,
       "learning_rate": 1.4114114114114117e-06,
-      "loss": 0.9555,
+      "loss": 0.946,
       "step": 85800
     },
     {
       "epoch": 257.69,
-      "grad_norm": 5.215327262878418,
+      "grad_norm": 5.761265754699707,
       "learning_rate": 1.4104104104104105e-06,
-      "loss": 0.9385,
+      "loss": 0.936,
       "step": 85810
     },
     {
       "epoch": 257.72,
-      "grad_norm": 6.107519626617432,
+      "grad_norm": 6.391301155090332,
       "learning_rate": 1.4094094094094097e-06,
-      "loss": 0.9832,
+      "loss": 0.972,
       "step": 85820
     },
     {
       "epoch": 257.75,
-      "grad_norm": 6.24550199508667,
+      "grad_norm": 7.31610107421875,
       "learning_rate": 1.4084084084084085e-06,
-      "loss": 0.9363,
+      "loss": 0.9423,
       "step": 85830
     },
     {
       "epoch": 257.78,
-      "grad_norm": 6.2621612548828125,
+      "grad_norm": 5.740571022033691,
       "learning_rate": 1.4074074074074075e-06,
-      "loss": 0.8788,
+      "loss": 0.8735,
       "step": 85840
     },
     {
       "epoch": 257.81,
-      "grad_norm": 5.503775596618652,
+      "grad_norm": 6.26387357711792,
       "learning_rate": 1.4064064064064066e-06,
-      "loss": 0.9321,
+      "loss": 0.9336,
       "step": 85850
     },
     {
       "epoch": 257.84,
-      "grad_norm": 6.621520519256592,
+      "grad_norm": 5.592531681060791,
       "learning_rate": 1.4054054054054056e-06,
-      "loss": 0.9679,
+      "loss": 0.97,
       "step": 85860
     },
     {
       "epoch": 257.87,
-      "grad_norm": 8.40252685546875,
+      "grad_norm": 6.066006183624268,
       "learning_rate": 1.4044044044044044e-06,
-      "loss": 0.8678,
+      "loss": 0.8651,
       "step": 85870
     },
     {
       "epoch": 257.9,
-      "grad_norm": 8.074043273925781,
+      "grad_norm": 7.120665073394775,
       "learning_rate": 1.4034034034034036e-06,
-      "loss": 0.9202,
+      "loss": 0.9182,
       "step": 85880
     },
     {
       "epoch": 257.93,
-      "grad_norm": 5.703863620758057,
+      "grad_norm": 5.573312759399414,
       "learning_rate": 1.4024024024024024e-06,
-      "loss": 0.8634,
+      "loss": 0.8747,
       "step": 85890
     },
     {
       "epoch": 257.96,
-      "grad_norm": 5.213403224945068,
+      "grad_norm": 4.299238204956055,
       "learning_rate": 1.4014014014014016e-06,
-      "loss": 0.8708,
+      "loss": 0.8678,
       "step": 85900
     },
     {
       "epoch": 257.99,
-      "grad_norm": 3.839101552963257,
+      "grad_norm": 5.813861846923828,
       "learning_rate": 1.4004004004004005e-06,
-      "loss": 0.8092,
+      "loss": 0.8089,
       "step": 85910
     },
     {
       "epoch": 258.0,
-      "eval_accuracy": 0.8281,
-      "eval_loss": 0.5719311833381653,
-      "eval_runtime": 5.3692,
-      "eval_samples_per_second": 1862.488,
-      "eval_steps_per_second": 7.45,
+      "eval_accuracy": 0.8294,
+      "eval_loss": 0.5705278515815735,
+      "eval_runtime": 5.3874,
+      "eval_samples_per_second": 1856.193,
+      "eval_steps_per_second": 7.425,
       "step": 85914
     },
     {
       "epoch": 258.02,
-      "grad_norm": 7.993595600128174,
+      "grad_norm": 7.27652645111084,
       "learning_rate": 1.3993993993993995e-06,
-      "loss": 0.8977,
+      "loss": 0.9064,
       "step": 85920
     },
     {
       "epoch": 258.05,
-      "grad_norm": 7.16609525680542,
+      "grad_norm": 7.364129543304443,
       "learning_rate": 1.3983983983983985e-06,
-      "loss": 0.9855,
+      "loss": 0.9862,
       "step": 85930
     },
     {
       "epoch": 258.08,
-      "grad_norm": 7.5865864753723145,
+      "grad_norm": 5.972012996673584,
       "learning_rate": 1.3973973973973975e-06,
-      "loss": 0.9201,
+      "loss": 0.915,
       "step": 85940
     },
     {
       "epoch": 258.11,
-      "grad_norm": 5.222933292388916,
+      "grad_norm": 6.010472774505615,
       "learning_rate": 1.3963963963963963e-06,
-      "loss": 0.881,
+      "loss": 0.884,
       "step": 85950
     },
     {
       "epoch": 258.14,
-      "grad_norm": 5.3418869972229,
+      "grad_norm": 5.959405422210693,
       "learning_rate": 1.3953953953953955e-06,
-      "loss": 0.915,
+      "loss": 0.9208,
       "step": 85960
     },
     {
       "epoch": 258.17,
-      "grad_norm": 7.204357624053955,
+      "grad_norm": 8.553051948547363,
       "learning_rate": 1.3943943943943944e-06,
-      "loss": 1.0101,
+      "loss": 1.0028,
       "step": 85970
     },
     {
       "epoch": 258.2,
-      "grad_norm": 9.302352905273438,
+      "grad_norm": 5.86819314956665,
       "learning_rate": 1.3933933933933936e-06,
-      "loss": 0.9369,
+      "loss": 0.9261,
       "step": 85980
     },
     {
       "epoch": 258.23,
-      "grad_norm": 5.549462795257568,
+      "grad_norm": 6.586169242858887,
       "learning_rate": 1.3923923923923924e-06,
-      "loss": 0.8987,
+      "loss": 0.8979,
       "step": 85990
     },
     {
       "epoch": 258.26,
-      "grad_norm": 10.541217803955078,
+      "grad_norm": 9.918671607971191,
       "learning_rate": 1.3913913913913914e-06,
-      "loss": 0.9293,
+      "loss": 0.9245,
       "step": 86000
     },
     {
       "epoch": 258.29,
-      "grad_norm": 12.466519355773926,
+      "grad_norm": 8.987183570861816,
       "learning_rate": 1.3903903903903904e-06,
-      "loss": 0.9239,
+      "loss": 0.9301,
       "step": 86010
     },
     {
       "epoch": 258.32,
-      "grad_norm": 12.290989875793457,
+      "grad_norm": 7.627177715301514,
       "learning_rate": 1.3893893893893894e-06,
-      "loss": 0.878,
+      "loss": 0.8748,
       "step": 86020
     },
     {
       "epoch": 258.35,
-      "grad_norm": 7.323269844055176,
+      "grad_norm": 6.568332672119141,
       "learning_rate": 1.3883883883883887e-06,
-      "loss": 0.9765,
+      "loss": 0.9747,
       "step": 86030
     },
     {
       "epoch": 258.38,
-      "grad_norm": 9.810335159301758,
+      "grad_norm": 9.306078910827637,
       "learning_rate": 1.3873873873873875e-06,
-      "loss": 0.9036,
+      "loss": 0.8908,
       "step": 86040
     },
     {
       "epoch": 258.41,
-      "grad_norm": 8.178544044494629,
+      "grad_norm": 9.087903022766113,
       "learning_rate": 1.3863863863863867e-06,
-      "loss": 0.9779,
+      "loss": 0.9816,
       "step": 86050
     },
     {
       "epoch": 258.44,
-      "grad_norm": 7.68113899230957,
+      "grad_norm": 7.4390459060668945,
       "learning_rate": 1.3853853853853855e-06,
-      "loss": 0.8933,
+      "loss": 0.8997,
       "step": 86060
     },
     {
       "epoch": 258.47,
-      "grad_norm": 6.6607346534729,
+      "grad_norm": 6.390323162078857,
       "learning_rate": 1.3843843843843845e-06,
-      "loss": 1.0118,
+      "loss": 1.006,
       "step": 86070
     },
     {
       "epoch": 258.5,
-      "grad_norm": 5.672527313232422,
+      "grad_norm": 6.921362400054932,
       "learning_rate": 1.3833833833833836e-06,
-      "loss": 0.9035,
+      "loss": 0.9069,
       "step": 86080
     },
     {
       "epoch": 258.53,
-      "grad_norm": 6.6333770751953125,
+      "grad_norm": 6.306578159332275,
       "learning_rate": 1.3823823823823826e-06,
-      "loss": 0.8342,
+      "loss": 0.8313,
       "step": 86090
     },
     {
       "epoch": 258.56,
-      "grad_norm": 7.209149360656738,
+      "grad_norm": 8.526222229003906,
       "learning_rate": 1.3813813813813814e-06,
-      "loss": 0.9346,
+      "loss": 0.9343,
       "step": 86100
     },
     {
       "epoch": 258.59,
-      "grad_norm": 10.04890251159668,
+      "grad_norm": 7.494597911834717,
       "learning_rate": 1.3803803803803806e-06,
-      "loss": 0.9025,
+      "loss": 0.9093,
       "step": 86110
     },
     {
       "epoch": 258.62,
-      "grad_norm": 5.727445602416992,
+      "grad_norm": 5.888759613037109,
       "learning_rate": 1.3793793793793794e-06,
-      "loss": 0.853,
+      "loss": 0.8465,
       "step": 86120
     },
     {
       "epoch": 258.65,
-      "grad_norm": 6.18079137802124,
+      "grad_norm": 5.288328170776367,
       "learning_rate": 1.3783783783783786e-06,
-      "loss": 0.9467,
+      "loss": 0.9468,
       "step": 86130
     },
     {
       "epoch": 258.68,
-      "grad_norm": 7.103862762451172,
+      "grad_norm": 8.650757789611816,
       "learning_rate": 1.3773773773773775e-06,
-      "loss": 0.9471,
+      "loss": 0.9541,
       "step": 86140
     },
     {
       "epoch": 258.71,
-      "grad_norm": 7.028067111968994,
+      "grad_norm": 6.509782791137695,
       "learning_rate": 1.3763763763763765e-06,
-      "loss": 0.8694,
+      "loss": 0.8644,
       "step": 86150
     },
     {
       "epoch": 258.74,
-      "grad_norm": 6.866818428039551,
+      "grad_norm": 5.376341342926025,
       "learning_rate": 1.3753753753753755e-06,
-      "loss": 0.897,
+      "loss": 0.9003,
       "step": 86160
     },
     {
       "epoch": 258.77,
-      "grad_norm": 6.288191795349121,
+      "grad_norm": 8.226888656616211,
       "learning_rate": 1.3743743743743745e-06,
       "loss": 0.9247,
       "step": 86170
     },
     {
       "epoch": 258.8,
-      "grad_norm": 6.5060930252075195,
+      "grad_norm": 6.810420989990234,
       "learning_rate": 1.3733733733733733e-06,
-      "loss": 0.9074,
+      "loss": 0.9081,
       "step": 86180
     },
     {
       "epoch": 258.83,
-      "grad_norm": 5.800911903381348,
+      "grad_norm": 6.304223537445068,
       "learning_rate": 1.3723723723723725e-06,
-      "loss": 0.8857,
+      "loss": 0.8804,
       "step": 86190
     },
     {
       "epoch": 258.86,
-      "grad_norm": 6.884032726287842,
+      "grad_norm": 7.2686662673950195,
       "learning_rate": 1.3713713713713714e-06,
-      "loss": 0.9074,
+      "loss": 0.9115,
       "step": 86200
     },
     {
       "epoch": 258.89,
-      "grad_norm": 6.697781562805176,
+      "grad_norm": 10.214600563049316,
       "learning_rate": 1.3703703703703706e-06,
-      "loss": 0.8776,
+      "loss": 0.8724,
       "step": 86210
     },
     {
       "epoch": 258.92,
-      "grad_norm": 4.5814208984375,
+      "grad_norm": 5.896456718444824,
       "learning_rate": 1.3693693693693694e-06,
-      "loss": 0.9435,
+      "loss": 0.9489,
       "step": 86220
     },
     {
       "epoch": 258.95,
-      "grad_norm": 6.848178863525391,
+      "grad_norm": 7.668984889984131,
       "learning_rate": 1.3683683683683684e-06,
-      "loss": 0.9506,
+      "loss": 0.9443,
       "step": 86230
     },
     {
       "epoch": 258.98,
-      "grad_norm": 6.793423175811768,
+      "grad_norm": 5.166197776794434,
       "learning_rate": 1.3673673673673674e-06,
-      "loss": 0.9023,
+      "loss": 0.9041,
       "step": 86240
     },
     {
       "epoch": 259.0,
-      "eval_accuracy": 0.8282,
-      "eval_loss": 0.5691922903060913,
-      "eval_runtime": 5.3837,
-      "eval_samples_per_second": 1857.444,
-      "eval_steps_per_second": 7.43,
+      "eval_accuracy": 0.8303,
+      "eval_loss": 0.5670434236526489,
+      "eval_runtime": 5.1569,
+      "eval_samples_per_second": 1939.148,
+      "eval_steps_per_second": 7.757,
       "step": 86247
     },
     {
       "epoch": 259.01,
-      "grad_norm": 5.283283233642578,
+      "grad_norm": 5.558002471923828,
       "learning_rate": 1.3663663663663664e-06,
-      "loss": 1.0489,
+      "loss": 1.0284,
       "step": 86250
     },
     {
       "epoch": 259.04,
-      "grad_norm": 6.597677230834961,
+      "grad_norm": 10.320869445800781,
       "learning_rate": 1.3653653653653653e-06,
-      "loss": 0.8993,
+      "loss": 0.8971,
       "step": 86260
     },
     {
       "epoch": 259.07,
-      "grad_norm": 5.55370569229126,
+      "grad_norm": 7.285730838775635,
       "learning_rate": 1.3643643643643645e-06,
-      "loss": 0.91,
+      "loss": 0.9087,
       "step": 86270
     },
     {
       "epoch": 259.1,
-      "grad_norm": 5.213360786437988,
+      "grad_norm": 6.198106288909912,
       "learning_rate": 1.3633633633633637e-06,
-      "loss": 0.9811,
+      "loss": 0.9933,
       "step": 86280
     },
     {
       "epoch": 259.13,
-      "grad_norm": 6.573848247528076,
+      "grad_norm": 4.844527721405029,
       "learning_rate": 1.3623623623623625e-06,
-      "loss": 0.9042,
+      "loss": 0.9075,
       "step": 86290
     },
     {
       "epoch": 259.16,
-      "grad_norm": 8.76867961883545,
+      "grad_norm": 7.784147262573242,
       "learning_rate": 1.3613613613613615e-06,
-      "loss": 0.9839,
+      "loss": 0.9735,
       "step": 86300
     },
     {
       "epoch": 259.19,
-      "grad_norm": 9.399032592773438,
+      "grad_norm": 6.043144702911377,
       "learning_rate": 1.3603603603603606e-06,
-      "loss": 0.8999,
+      "loss": 0.8988,
       "step": 86310
     },
     {
       "epoch": 259.22,
-      "grad_norm": 5.980886936187744,
+      "grad_norm": 6.261370658874512,
       "learning_rate": 1.3593593593593596e-06,
-      "loss": 0.8482,
+      "loss": 0.8465,
       "step": 86320
     },
     {
       "epoch": 259.25,
-      "grad_norm": 8.258718490600586,
+      "grad_norm": 7.007743835449219,
       "learning_rate": 1.3583583583583584e-06,
-      "loss": 0.8959,
+      "loss": 0.8972,
       "step": 86330
     },
     {
       "epoch": 259.28,
-      "grad_norm": 4.949344635009766,
+      "grad_norm": 4.6346940994262695,
       "learning_rate": 1.3573573573573576e-06,
-      "loss": 0.8709,
+      "loss": 0.8721,
       "step": 86340
     },
     {
       "epoch": 259.31,
-      "grad_norm": 6.651227951049805,
+      "grad_norm": 4.961434841156006,
       "learning_rate": 1.3563563563563564e-06,
-      "loss": 0.8969,
+      "loss": 0.8842,
       "step": 86350
     },
     {
       "epoch": 259.34,
-      "grad_norm": 5.526999473571777,
+      "grad_norm": 4.941037654876709,
       "learning_rate": 1.3553553553553556e-06,
-      "loss": 0.9097,
+      "loss": 0.8975,
       "step": 86360
     },
     {
       "epoch": 259.37,
-      "grad_norm": 6.172590732574463,
+      "grad_norm": 5.940287113189697,
       "learning_rate": 1.3543543543543545e-06,
-      "loss": 0.8666,
+      "loss": 0.8789,
       "step": 86370
     },
     {
       "epoch": 259.4,
-      "grad_norm": 8.476754188537598,
+      "grad_norm": 6.828104496002197,
       "learning_rate": 1.3533533533533535e-06,
-      "loss": 0.8811,
+      "loss": 0.8844,
       "step": 86380
     },
     {
       "epoch": 259.43,
-      "grad_norm": 9.282114028930664,
+      "grad_norm": 7.769406795501709,
       "learning_rate": 1.3523523523523525e-06,
-      "loss": 0.9853,
+      "loss": 0.9844,
       "step": 86390
     },
     {
       "epoch": 259.46,
-      "grad_norm": 7.619330883026123,
+      "grad_norm": 6.785083293914795,
       "learning_rate": 1.3513513513513515e-06,
-      "loss": 0.8562,
+      "loss": 0.8548,
       "step": 86400
     },
     {
       "epoch": 259.49,
-      "grad_norm": 7.478638648986816,
+      "grad_norm": 6.574766159057617,
       "learning_rate": 1.3503503503503503e-06,
-      "loss": 0.8829,
+      "loss": 0.878,
       "step": 86410
     },
     {
       "epoch": 259.52,
-      "grad_norm": 5.562349796295166,
+      "grad_norm": 12.015536308288574,
       "learning_rate": 1.3493493493493495e-06,
-      "loss": 0.8503,
+      "loss": 0.8451,
       "step": 86420
     },
     {
       "epoch": 259.55,
-      "grad_norm": 6.01262092590332,
+      "grad_norm": 7.125096797943115,
       "learning_rate": 1.3483483483483484e-06,
-      "loss": 0.9503,
+      "loss": 0.9475,
       "step": 86430
     },
     {
       "epoch": 259.58,
-      "grad_norm": 5.7615647315979,
+      "grad_norm": 6.37600564956665,
       "learning_rate": 1.3473473473473476e-06,
-      "loss": 0.9133,
+      "loss": 0.9058,
       "step": 86440
     },
     {
       "epoch": 259.61,
-      "grad_norm": 5.122636795043945,
+      "grad_norm": 5.298223495483398,
       "learning_rate": 1.3463463463463464e-06,
-      "loss": 0.9096,
+      "loss": 0.9041,
       "step": 86450
     },
     {
       "epoch": 259.64,
-      "grad_norm": 6.1981658935546875,
+      "grad_norm": 5.185858726501465,
       "learning_rate": 1.3453453453453454e-06,
-      "loss": 0.9377,
+      "loss": 0.948,
       "step": 86460
     },
     {
       "epoch": 259.67,
-      "grad_norm": 5.193418502807617,
+      "grad_norm": 8.881721496582031,
       "learning_rate": 1.3443443443443444e-06,
-      "loss": 0.8964,
+      "loss": 0.8909,
       "step": 86470
     },
     {
       "epoch": 259.7,
-      "grad_norm": 6.6203742027282715,
+      "grad_norm": 7.374054431915283,
       "learning_rate": 1.3433433433433434e-06,
-      "loss": 0.9347,
+      "loss": 0.9427,
       "step": 86480
     },
     {
       "epoch": 259.73,
-      "grad_norm": 6.327735900878906,
+      "grad_norm": 11.05753231048584,
       "learning_rate": 1.3423423423423422e-06,
-      "loss": 0.9129,
+      "loss": 0.9063,
       "step": 86490
     },
     {
       "epoch": 259.76,
-      "grad_norm": 9.18451976776123,
+      "grad_norm": 7.883667469024658,
       "learning_rate": 1.3413413413413415e-06,
-      "loss": 0.8811,
+      "loss": 0.8772,
       "step": 86500
     },
     {
       "epoch": 259.79,
-      "grad_norm": 7.156782627105713,
+      "grad_norm": 7.524628639221191,
       "learning_rate": 1.3403403403403403e-06,
-      "loss": 0.9062,
+      "loss": 0.8984,
       "step": 86510
     },
     {
       "epoch": 259.82,
-      "grad_norm": 6.27545166015625,
+      "grad_norm": 5.502562522888184,
       "learning_rate": 1.3393393393393395e-06,
-      "loss": 0.8599,
+      "loss": 0.8589,
       "step": 86520
     },
     {
       "epoch": 259.85,
-      "grad_norm": 7.924337863922119,
+      "grad_norm": 6.045924663543701,
       "learning_rate": 1.3383383383383383e-06,
-      "loss": 0.8477,
+      "loss": 0.8481,
       "step": 86530
     },
     {
       "epoch": 259.88,
-      "grad_norm": 7.261173248291016,
+      "grad_norm": 7.349027633666992,
       "learning_rate": 1.3373373373373376e-06,
-      "loss": 0.8938,
+      "loss": 0.89,
       "step": 86540
     },
     {
       "epoch": 259.91,
-      "grad_norm": 7.8857035636901855,
+      "grad_norm": 6.031946659088135,
       "learning_rate": 1.3363363363363366e-06,
-      "loss": 0.9163,
+      "loss": 0.9118,
       "step": 86550
     },
     {
       "epoch": 259.94,
-      "grad_norm": 5.236282825469971,
+      "grad_norm": 6.904636383056641,
       "learning_rate": 1.3353353353353354e-06,
-      "loss": 0.8973,
+      "loss": 0.9027,
       "step": 86560
     },
     {
       "epoch": 259.97,
-      "grad_norm": 8.263518333435059,
+      "grad_norm": 9.517247200012207,
       "learning_rate": 1.3343343343343346e-06,
-      "loss": 0.9773,
+      "loss": 0.9762,
       "step": 86570
     },
     {
       "epoch": 260.0,
-      "grad_norm": 26.910358428955078,
+      "grad_norm": 22.714916229248047,
       "learning_rate": 1.3333333333333334e-06,
-      "loss": 1.0531,
+      "loss": 1.0149,
       "step": 86580
     },
     {
       "epoch": 260.0,
-      "eval_accuracy": 0.8271,
-      "eval_loss": 0.5706588625907898,
-      "eval_runtime": 5.5218,
-      "eval_samples_per_second": 1811.013,
-      "eval_steps_per_second": 7.244,
+      "eval_accuracy": 0.8312,
+      "eval_loss": 0.5689736604690552,
+      "eval_runtime": 5.3773,
+      "eval_samples_per_second": 1859.659,
+      "eval_steps_per_second": 7.439,
       "step": 86580
     },
     {
       "epoch": 260.03,
-      "grad_norm": 5.52992057800293,
+      "grad_norm": 7.6355299949646,
       "learning_rate": 1.3323323323323326e-06,
-      "loss": 0.8907,
+      "loss": 0.892,
       "step": 86590
     },
     {
       "epoch": 260.06,
-      "grad_norm": 7.555476665496826,
+      "grad_norm": 7.395337104797363,
       "learning_rate": 1.3313313313313315e-06,
-      "loss": 0.8125,
+      "loss": 0.8045,
       "step": 86600
     },
     {
       "epoch": 260.09,
-      "grad_norm": 7.0363874435424805,
+      "grad_norm": 7.348328113555908,
       "learning_rate": 1.3303303303303305e-06,
-      "loss": 0.8829,
+      "loss": 0.8785,
       "step": 86610
     },
     {
       "epoch": 260.12,
-      "grad_norm": 5.506227016448975,
+      "grad_norm": 5.165286540985107,
       "learning_rate": 1.3293293293293295e-06,
-      "loss": 0.9355,
+      "loss": 0.9359,
       "step": 86620
     },
     {
       "epoch": 260.15,
-      "grad_norm": 7.359404563903809,
+      "grad_norm": 6.971673965454102,
       "learning_rate": 1.3283283283283285e-06,
-      "loss": 0.8931,
+      "loss": 0.8902,
       "step": 86630
     },
     {
       "epoch": 260.18,
-      "grad_norm": 8.759977340698242,
+      "grad_norm": 9.537163734436035,
       "learning_rate": 1.3273273273273273e-06,
-      "loss": 0.9366,
+      "loss": 0.9342,
       "step": 86640
     },
     {
       "epoch": 260.21,
-      "grad_norm": 4.071224212646484,
+      "grad_norm": 5.692164421081543,
       "learning_rate": 1.3263263263263265e-06,
-      "loss": 0.9405,
+      "loss": 0.9348,
       "step": 86650
     },
     {
       "epoch": 260.24,
-      "grad_norm": 7.529251575469971,
+      "grad_norm": 10.931696891784668,
       "learning_rate": 1.3253253253253253e-06,
-      "loss": 0.9285,
+      "loss": 0.9183,
       "step": 86660
     },
     {
       "epoch": 260.27,
-      "grad_norm": 7.178025245666504,
+      "grad_norm": 7.074573516845703,
       "learning_rate": 1.3243243243243246e-06,
-      "loss": 0.8781,
+      "loss": 0.8802,
       "step": 86670
     },
     {
       "epoch": 260.3,
-      "grad_norm": 6.090941429138184,
+      "grad_norm": 6.566608905792236,
       "learning_rate": 1.3233233233233234e-06,
-      "loss": 0.9571,
+      "loss": 0.9518,
       "step": 86680
     },
     {
       "epoch": 260.33,
-      "grad_norm": 4.834505558013916,
+      "grad_norm": 6.944140911102295,
       "learning_rate": 1.3223223223223224e-06,
-      "loss": 0.8849,
+      "loss": 0.8816,
       "step": 86690
     },
     {
       "epoch": 260.36,
-      "grad_norm": 7.7118730545043945,
+      "grad_norm": 6.5585737228393555,
       "learning_rate": 1.3213213213213214e-06,
-      "loss": 0.9404,
+      "loss": 0.9319,
       "step": 86700
     },
     {
       "epoch": 260.39,
-      "grad_norm": 6.921819686889648,
+      "grad_norm": 7.547276496887207,
       "learning_rate": 1.3203203203203204e-06,
-      "loss": 0.9608,
+      "loss": 0.9534,
       "step": 86710
     },
     {
       "epoch": 260.42,
-      "grad_norm": 7.20330286026001,
+      "grad_norm": 5.9131951332092285,
       "learning_rate": 1.3193193193193192e-06,
-      "loss": 0.9404,
+      "loss": 0.9482,
       "step": 86720
     },
     {
       "epoch": 260.45,
-      "grad_norm": 6.529550552368164,
+      "grad_norm": 6.089829921722412,
       "learning_rate": 1.3183183183183185e-06,
-      "loss": 0.8825,
+      "loss": 0.8778,
       "step": 86730
     },
     {
       "epoch": 260.48,
-      "grad_norm": 4.403013229370117,
+      "grad_norm": 7.368325233459473,
       "learning_rate": 1.3173173173173173e-06,
-      "loss": 0.9401,
+      "loss": 0.931,
       "step": 86740
     },
     {
       "epoch": 260.51,
-      "grad_norm": 6.5081787109375,
+      "grad_norm": 6.127671718597412,
       "learning_rate": 1.3163163163163165e-06,
-      "loss": 0.9253,
+      "loss": 0.9265,
       "step": 86750
     },
     {
       "epoch": 260.54,
-      "grad_norm": 7.118206977844238,
+      "grad_norm": 6.0797810554504395,
       "learning_rate": 1.3153153153153153e-06,
-      "loss": 0.8895,
+      "loss": 0.8876,
       "step": 86760
     },
     {
       "epoch": 260.57,
-      "grad_norm": 6.505639553070068,
+      "grad_norm": 5.921451568603516,
       "learning_rate": 1.3143143143143143e-06,
-      "loss": 0.9494,
+      "loss": 0.9496,
       "step": 86770
     },
     {
       "epoch": 260.6,
-      "grad_norm": 6.113704204559326,
+      "grad_norm": 6.890771865844727,
       "learning_rate": 1.3133133133133134e-06,
-      "loss": 0.9342,
+      "loss": 0.936,
       "step": 86780
     },
     {
       "epoch": 260.63,
-      "grad_norm": 5.758663177490234,
+      "grad_norm": 6.53848123550415,
       "learning_rate": 1.3123123123123124e-06,
-      "loss": 0.93,
+      "loss": 0.9283,
       "step": 86790
     },
     {
       "epoch": 260.66,
-      "grad_norm": 6.249421119689941,
+      "grad_norm": 6.359808444976807,
       "learning_rate": 1.3113113113113112e-06,
-      "loss": 0.9582,
+      "loss": 0.9517,
       "step": 86800
     },
     {
       "epoch": 260.69,
-      "grad_norm": 7.765279769897461,
+      "grad_norm": 7.926748752593994,
       "learning_rate": 1.3103103103103104e-06,
-      "loss": 0.8448,
+      "loss": 0.8418,
       "step": 86810
     },
     {
       "epoch": 260.72,
-      "grad_norm": 5.720190525054932,
+      "grad_norm": 7.748381614685059,
       "learning_rate": 1.3093093093093096e-06,
-      "loss": 0.9322,
+      "loss": 0.936,
       "step": 86820
     },
     {
       "epoch": 260.75,
-      "grad_norm": 4.800948143005371,
+      "grad_norm": 5.312899589538574,
       "learning_rate": 1.3083083083083084e-06,
-      "loss": 0.8831,
+      "loss": 0.8848,
       "step": 86830
     },
     {
       "epoch": 260.78,
-      "grad_norm": 7.956442356109619,
+      "grad_norm": 7.573738098144531,
       "learning_rate": 1.3073073073073075e-06,
-      "loss": 0.873,
+      "loss": 0.8739,
       "step": 86840
     },
     {
       "epoch": 260.81,
-      "grad_norm": 6.124589920043945,
+      "grad_norm": 6.224379062652588,
       "learning_rate": 1.3063063063063065e-06,
-      "loss": 0.8505,
+      "loss": 0.8462,
       "step": 86850
     },
     {
       "epoch": 260.84,
-      "grad_norm": 6.320127964019775,
+      "grad_norm": 7.068265438079834,
       "learning_rate": 1.3053053053053055e-06,
-      "loss": 0.9081,
+      "loss": 0.9151,
       "step": 86860
     },
     {
       "epoch": 260.87,
-      "grad_norm": 8.148602485656738,
+      "grad_norm": 8.8317289352417,
       "learning_rate": 1.3043043043043043e-06,
-      "loss": 0.9185,
+      "loss": 0.9242,
       "step": 86870
     },
     {
       "epoch": 260.9,
-      "grad_norm": 6.909006595611572,
+      "grad_norm": 6.942406177520752,
       "learning_rate": 1.3033033033033035e-06,
-      "loss": 0.8771,
+      "loss": 0.8866,
       "step": 86880
     },
     {
       "epoch": 260.93,
-      "grad_norm": 6.89491605758667,
+      "grad_norm": 6.587258338928223,
       "learning_rate": 1.3023023023023023e-06,
-      "loss": 0.8731,
+      "loss": 0.8713,
       "step": 86890
     },
     {
       "epoch": 260.96,
-      "grad_norm": 8.913177490234375,
+      "grad_norm": 6.555168628692627,
       "learning_rate": 1.3013013013013016e-06,
-      "loss": 0.9444,
+      "loss": 0.9493,
       "step": 86900
     },
     {
       "epoch": 260.99,
-      "grad_norm": 6.368200778961182,
+      "grad_norm": 6.683752059936523,
       "learning_rate": 1.3003003003003004e-06,
-      "loss": 0.9112,
+      "loss": 0.899,
       "step": 86910
     },
     {
       "epoch": 261.0,
-      "eval_accuracy": 0.8259,
-      "eval_loss": 0.5703931450843811,
-      "eval_runtime": 5.6106,
-      "eval_samples_per_second": 1782.341,
-      "eval_steps_per_second": 7.129,
+      "eval_accuracy": 0.8302,
+      "eval_loss": 0.5688945055007935,
+      "eval_runtime": 5.3819,
+      "eval_samples_per_second": 1858.075,
+      "eval_steps_per_second": 7.432,
       "step": 86913
     },
     {
       "epoch": 261.02,
-      "grad_norm": 14.427462577819824,
+      "grad_norm": 16.68572998046875,
       "learning_rate": 1.2992992992992994e-06,
-      "loss": 1.0394,
+      "loss": 1.0323,
       "step": 86920
     },
     {
       "epoch": 261.05,
-      "grad_norm": 11.779111862182617,
+      "grad_norm": 6.549874305725098,
       "learning_rate": 1.2982982982982984e-06,
-      "loss": 0.8743,
+      "loss": 0.8761,
       "step": 86930
     },
     {
       "epoch": 261.08,
-      "grad_norm": 6.568065166473389,
+      "grad_norm": 6.917584419250488,
       "learning_rate": 1.2972972972972974e-06,
-      "loss": 0.7905,
+      "loss": 0.7818,
       "step": 86940
     },
     {
       "epoch": 261.11,
-      "grad_norm": 7.426055431365967,
+      "grad_norm": 8.458977699279785,
       "learning_rate": 1.2962962962962962e-06,
-      "loss": 0.89,
+      "loss": 0.8841,
       "step": 86950
     },
     {
       "epoch": 261.14,
-      "grad_norm": 6.790721893310547,
+      "grad_norm": 7.06601095199585,
       "learning_rate": 1.2952952952952955e-06,
-      "loss": 0.9258,
+      "loss": 0.9178,
       "step": 86960
     },
     {
       "epoch": 261.17,
-      "grad_norm": 6.415890216827393,
+      "grad_norm": 6.013164043426514,
       "learning_rate": 1.2942942942942943e-06,
-      "loss": 1.0072,
+      "loss": 0.9945,
       "step": 86970
     },
     {
       "epoch": 261.2,
-      "grad_norm": 6.819501876831055,
+      "grad_norm": 6.7138872146606445,
       "learning_rate": 1.2932932932932935e-06,
-      "loss": 0.9142,
+      "loss": 0.9204,
       "step": 86980
     },
     {
       "epoch": 261.23,
-      "grad_norm": 8.195550918579102,
+      "grad_norm": 8.562260627746582,
       "learning_rate": 1.2922922922922923e-06,
-      "loss": 0.9729,
+      "loss": 0.9712,
       "step": 86990
     },
     {
       "epoch": 261.26,
-      "grad_norm": 11.096035957336426,
+      "grad_norm": 5.977176666259766,
       "learning_rate": 1.2912912912912913e-06,
-      "loss": 0.9041,
+      "loss": 0.9092,
       "step": 87000
     },
     {
       "epoch": 261.29,
-      "grad_norm": 6.927455902099609,
+      "grad_norm": 6.141510486602783,
       "learning_rate": 1.2902902902902904e-06,
-      "loss": 0.906,
+      "loss": 0.9093,
       "step": 87010
     },
     {
       "epoch": 261.32,
-      "grad_norm": 6.050289630889893,
+      "grad_norm": 9.19703197479248,
       "learning_rate": 1.2892892892892894e-06,
-      "loss": 0.8628,
+      "loss": 0.8732,
       "step": 87020
     },
     {
       "epoch": 261.35,
-      "grad_norm": 12.181004524230957,
+      "grad_norm": 12.694092750549316,
       "learning_rate": 1.2882882882882882e-06,
-      "loss": 0.9197,
+      "loss": 0.9199,
       "step": 87030
     },
     {
       "epoch": 261.38,
-      "grad_norm": 4.973325729370117,
+      "grad_norm": 6.575229644775391,
       "learning_rate": 1.2872872872872874e-06,
-      "loss": 0.9658,
+      "loss": 0.9686,
       "step": 87040
     },
     {
       "epoch": 261.41,
-      "grad_norm": 5.448077201843262,
+      "grad_norm": 6.649130344390869,
       "learning_rate": 1.2862862862862862e-06,
-      "loss": 0.9385,
+      "loss": 0.9344,
       "step": 87050
     },
     {
       "epoch": 261.44,
-      "grad_norm": 7.303735733032227,
+      "grad_norm": 8.420234680175781,
       "learning_rate": 1.2852852852852854e-06,
-      "loss": 0.9517,
+      "loss": 0.9549,
       "step": 87060
     },
     {
       "epoch": 261.47,
-      "grad_norm": 7.768982410430908,
+      "grad_norm": 9.728636741638184,
       "learning_rate": 1.2842842842842843e-06,
-      "loss": 0.9385,
+      "loss": 0.9473,
       "step": 87070
     },
     {
       "epoch": 261.5,
-      "grad_norm": 6.073237419128418,
+      "grad_norm": 5.942495346069336,
       "learning_rate": 1.2832832832832835e-06,
-      "loss": 0.8853,
+      "loss": 0.8929,
       "step": 87080
     },
     {
       "epoch": 261.53,
-      "grad_norm": 7.272352695465088,
+      "grad_norm": 6.815118789672852,
       "learning_rate": 1.2822822822822825e-06,
-      "loss": 0.8684,
+      "loss": 0.8616,
       "step": 87090
     },
     {
       "epoch": 261.56,
-      "grad_norm": 7.584596633911133,
+      "grad_norm": 7.806725978851318,
       "learning_rate": 1.2812812812812813e-06,
-      "loss": 0.9124,
+      "loss": 0.9021,
       "step": 87100
     },
     {
       "epoch": 261.59,
-      "grad_norm": 6.96652889251709,
+      "grad_norm": 5.658957004547119,
       "learning_rate": 1.2802802802802805e-06,
-      "loss": 0.9643,
+      "loss": 0.9599,
       "step": 87110
     },
     {
       "epoch": 261.62,
-      "grad_norm": 8.796903610229492,
+      "grad_norm": 7.5366387367248535,
       "learning_rate": 1.2792792792792793e-06,
-      "loss": 0.9073,
+      "loss": 0.8996,
       "step": 87120
     },
     {
       "epoch": 261.65,
-      "grad_norm": 8.066240310668945,
+      "grad_norm": 9.00859546661377,
       "learning_rate": 1.2782782782782786e-06,
-      "loss": 0.9155,
+      "loss": 0.9229,
       "step": 87130
     },
     {
       "epoch": 261.68,
-      "grad_norm": 5.078109264373779,
+      "grad_norm": 5.025231838226318,
       "learning_rate": 1.2772772772772774e-06,
-      "loss": 0.8886,
+      "loss": 0.8781,
       "step": 87140
     },
     {
       "epoch": 261.71,
-      "grad_norm": 8.630660057067871,
+      "grad_norm": 7.192399501800537,
       "learning_rate": 1.2762762762762764e-06,
-      "loss": 0.923,
+      "loss": 0.9309,
       "step": 87150
     },
     {
       "epoch": 261.74,
-      "grad_norm": 6.1905341148376465,
+      "grad_norm": 6.0245771408081055,
       "learning_rate": 1.2752752752752754e-06,
-      "loss": 0.959,
+      "loss": 0.954,
       "step": 87160
     },
     {
       "epoch": 261.77,
-      "grad_norm": 5.371632099151611,
+      "grad_norm": 7.14896821975708,
       "learning_rate": 1.2742742742742744e-06,
-      "loss": 0.8802,
+      "loss": 0.8824,
       "step": 87170
     },
     {
       "epoch": 261.8,
-      "grad_norm": 5.949338436126709,
+      "grad_norm": 6.8357696533203125,
       "learning_rate": 1.2732732732732732e-06,
-      "loss": 0.9164,
+      "loss": 0.9272,
       "step": 87180
     },
     {
       "epoch": 261.83,
-      "grad_norm": 5.651325225830078,
+      "grad_norm": 5.894567966461182,
       "learning_rate": 1.2722722722722725e-06,
-      "loss": 0.9093,
+      "loss": 0.9022,
       "step": 87190
     },
     {
       "epoch": 261.86,
-      "grad_norm": 7.709237575531006,
+      "grad_norm": 5.239195823669434,
       "learning_rate": 1.2712712712712713e-06,
       "loss": 0.9269,
       "step": 87200
     },
     {
       "epoch": 261.89,
-      "grad_norm": 4.976951599121094,
+      "grad_norm": 5.7491984367370605,
       "learning_rate": 1.2702702702702705e-06,
-      "loss": 0.9159,
+      "loss": 0.9079,
       "step": 87210
     },
     {
       "epoch": 261.92,
-      "grad_norm": 7.994369029998779,
+      "grad_norm": 8.17073917388916,
       "learning_rate": 1.2692692692692693e-06,
-      "loss": 0.9211,
+      "loss": 0.9156,
       "step": 87220
     },
     {
       "epoch": 261.95,
-      "grad_norm": 8.073983192443848,
+      "grad_norm": 6.386580944061279,
       "learning_rate": 1.2682682682682683e-06,
-      "loss": 0.9163,
+      "loss": 0.9088,
       "step": 87230
     },
     {
       "epoch": 261.98,
-      "grad_norm": 8.068414688110352,
+      "grad_norm": 6.744256019592285,
       "learning_rate": 1.2672672672672674e-06,
-      "loss": 0.8781,
+      "loss": 0.8802,
       "step": 87240
     },
     {
       "epoch": 262.0,
-      "eval_accuracy": 0.826,
-      "eval_loss": 0.574149489402771,
-      "eval_runtime": 5.3409,
-      "eval_samples_per_second": 1872.345,
-      "eval_steps_per_second": 7.489,
+      "eval_accuracy": 0.8287,
+      "eval_loss": 0.5728639960289001,
+      "eval_runtime": 5.2985,
+      "eval_samples_per_second": 1887.336,
+      "eval_steps_per_second": 7.549,
       "step": 87246
     },
     {
       "epoch": 262.01,
-      "grad_norm": 9.071558952331543,
+      "grad_norm": 6.763991832733154,
       "learning_rate": 1.2662662662662664e-06,
-      "loss": 0.9792,
+      "loss": 0.9468,
       "step": 87250
     },
     {
       "epoch": 262.04,
-      "grad_norm": 5.922547340393066,
+      "grad_norm": 6.852902412414551,
       "learning_rate": 1.2652652652652652e-06,
-      "loss": 0.8883,
+      "loss": 0.8823,
       "step": 87260
     },
     {
       "epoch": 262.07,
-      "grad_norm": 8.857194900512695,
+      "grad_norm": 7.845478534698486,
       "learning_rate": 1.2642642642642644e-06,
-      "loss": 0.9457,
+      "loss": 0.9488,
       "step": 87270
     },
     {
       "epoch": 262.1,
-      "grad_norm": 5.685113906860352,
+      "grad_norm": 6.888769626617432,
       "learning_rate": 1.2632632632632632e-06,
-      "loss": 0.8958,
+      "loss": 0.9023,
       "step": 87280
     },
     {
       "epoch": 262.13,
-      "grad_norm": 6.662797927856445,
+      "grad_norm": 7.449899673461914,
       "learning_rate": 1.2622622622622624e-06,
-      "loss": 0.9175,
+      "loss": 0.9036,
       "step": 87290
     },
     {
       "epoch": 262.16,
-      "grad_norm": 6.776182651519775,
+      "grad_norm": 7.400008678436279,
       "learning_rate": 1.2612612612612613e-06,
-      "loss": 0.9422,
+      "loss": 0.9396,
       "step": 87300
     },
     {
       "epoch": 262.19,
-      "grad_norm": 5.575989723205566,
+      "grad_norm": 8.305876731872559,
       "learning_rate": 1.2602602602602605e-06,
-      "loss": 0.9023,
+      "loss": 0.8905,
       "step": 87310
     },
     {
       "epoch": 262.22,
-      "grad_norm": 5.5593461990356445,
+      "grad_norm": 6.349542140960693,
       "learning_rate": 1.2592592592592593e-06,
-      "loss": 0.8779,
+      "loss": 0.876,
       "step": 87320
     },
     {
       "epoch": 262.25,
-      "grad_norm": 5.869960784912109,
+      "grad_norm": 7.812323093414307,
       "learning_rate": 1.2582582582582583e-06,
-      "loss": 0.918,
+      "loss": 0.9142,
       "step": 87330
     },
     {
       "epoch": 262.28,
-      "grad_norm": 4.651187419891357,
+      "grad_norm": 6.62490701675415,
       "learning_rate": 1.2572572572572573e-06,
-      "loss": 0.8336,
+      "loss": 0.8257,
       "step": 87340
     },
     {
       "epoch": 262.31,
-      "grad_norm": 6.8043413162231445,
+      "grad_norm": 7.077232837677002,
       "learning_rate": 1.2562562562562563e-06,
-      "loss": 0.898,
+      "loss": 0.8876,
       "step": 87350
     },
     {
       "epoch": 262.34,
-      "grad_norm": 13.763994216918945,
+      "grad_norm": 11.54252815246582,
       "learning_rate": 1.2552552552552556e-06,
-      "loss": 0.9612,
+      "loss": 0.9714,
       "step": 87360
     },
     {
       "epoch": 262.37,
-      "grad_norm": 5.985256671905518,
+      "grad_norm": 5.938597202301025,
       "learning_rate": 1.2542542542542544e-06,
-      "loss": 0.9325,
+      "loss": 0.9303,
       "step": 87370
     },
     {
       "epoch": 262.4,
-      "grad_norm": 8.907166481018066,
+      "grad_norm": 8.549055099487305,
       "learning_rate": 1.2532532532532534e-06,
-      "loss": 0.9858,
+      "loss": 0.9925,
       "step": 87380
     },
     {
       "epoch": 262.43,
-      "grad_norm": 10.039406776428223,
+      "grad_norm": 7.273135185241699,
       "learning_rate": 1.2522522522522524e-06,
-      "loss": 0.8486,
+      "loss": 0.8466,
       "step": 87390
     },
     {
       "epoch": 262.46,
-      "grad_norm": 6.426618576049805,
+      "grad_norm": 7.521265029907227,
       "learning_rate": 1.2512512512512514e-06,
-      "loss": 0.9086,
+      "loss": 0.9148,
       "step": 87400
     },
     {
       "epoch": 262.49,
-      "grad_norm": 5.629103183746338,
+      "grad_norm": 5.775476455688477,
       "learning_rate": 1.2502502502502502e-06,
-      "loss": 0.9384,
+      "loss": 0.944,
       "step": 87410
     },
     {
       "epoch": 262.52,
-      "grad_norm": 6.647545337677002,
+      "grad_norm": 8.672942161560059,
       "learning_rate": 1.2492492492492493e-06,
-      "loss": 0.8648,
+      "loss": 0.8601,
       "step": 87420
     },
     {
       "epoch": 262.55,
-      "grad_norm": 6.3198347091674805,
+      "grad_norm": 6.227295875549316,
       "learning_rate": 1.2482482482482483e-06,
-      "loss": 0.8344,
+      "loss": 0.8331,
       "step": 87430
     },
     {
       "epoch": 262.58,
-      "grad_norm": 5.943576812744141,
+      "grad_norm": 5.86444616317749,
       "learning_rate": 1.2472472472472473e-06,
-      "loss": 0.9146,
+      "loss": 0.9141,
       "step": 87440
     },
     {
       "epoch": 262.61,
-      "grad_norm": 5.665236473083496,
+      "grad_norm": 8.557504653930664,
       "learning_rate": 1.2462462462462463e-06,
-      "loss": 0.8686,
+      "loss": 0.8651,
       "step": 87450
     },
     {
       "epoch": 262.64,
-      "grad_norm": 6.646903991699219,
+      "grad_norm": 6.972194671630859,
       "learning_rate": 1.2452452452452453e-06,
-      "loss": 0.9068,
+      "loss": 0.9023,
       "step": 87460
     },
     {
       "epoch": 262.67,
-      "grad_norm": 6.962013244628906,
+      "grad_norm": 9.120635986328125,
       "learning_rate": 1.2442442442442444e-06,
-      "loss": 0.909,
+      "loss": 0.9068,
       "step": 87470
     },
     {
       "epoch": 262.7,
-      "grad_norm": 8.721837043762207,
+      "grad_norm": 8.994202613830566,
       "learning_rate": 1.2432432432432434e-06,
-      "loss": 0.9363,
+      "loss": 0.9291,
       "step": 87480
     },
     {
       "epoch": 262.73,
-      "grad_norm": 7.30488920211792,
+      "grad_norm": 7.635751724243164,
       "learning_rate": 1.2422422422422424e-06,
-      "loss": 0.9526,
+      "loss": 0.9457,
       "step": 87490
     },
     {
       "epoch": 262.76,
-      "grad_norm": 6.92581033706665,
+      "grad_norm": 6.53345251083374,
       "learning_rate": 1.2412412412412414e-06,
-      "loss": 0.9289,
+      "loss": 0.9136,
       "step": 87500
     },
     {
       "epoch": 262.79,
-      "grad_norm": 6.847958564758301,
+      "grad_norm": 6.925932884216309,
       "learning_rate": 1.2402402402402404e-06,
-      "loss": 0.8946,
+      "loss": 0.8908,
       "step": 87510
     },
     {
       "epoch": 262.82,
-      "grad_norm": 5.257356643676758,
+      "grad_norm": 6.017780303955078,
       "learning_rate": 1.2392392392392394e-06,
-      "loss": 0.9138,
+      "loss": 0.9005,
       "step": 87520
     },
     {
       "epoch": 262.85,
-      "grad_norm": 6.45369291305542,
+      "grad_norm": 7.109196186065674,
       "learning_rate": 1.2382382382382385e-06,
-      "loss": 0.901,
+      "loss": 0.9039,
       "step": 87530
     },
     {
       "epoch": 262.88,
-      "grad_norm": 7.0991926193237305,
+      "grad_norm": 9.883187294006348,
       "learning_rate": 1.2372372372372375e-06,
-      "loss": 0.8314,
+      "loss": 0.8254,
       "step": 87540
     },
     {
       "epoch": 262.91,
-      "grad_norm": 7.709216117858887,
+      "grad_norm": 7.598671913146973,
       "learning_rate": 1.2362362362362363e-06,
-      "loss": 0.8985,
+      "loss": 0.8973,
       "step": 87550
     },
     {
       "epoch": 262.94,
-      "grad_norm": 7.470425605773926,
+      "grad_norm": 6.277622222900391,
       "learning_rate": 1.2352352352352353e-06,
-      "loss": 0.935,
+      "loss": 0.9336,
       "step": 87560
     },
     {
       "epoch": 262.97,
-      "grad_norm": 5.9771246910095215,
+      "grad_norm": 12.377532958984375,
       "learning_rate": 1.2342342342342343e-06,
-      "loss": 0.8708,
+      "loss": 0.8718,
       "step": 87570
     },
     {
       "epoch": 263.0,
-      "eval_accuracy": 0.829,
-      "eval_loss": 0.5654478669166565,
-      "eval_runtime": 5.4088,
-      "eval_samples_per_second": 1848.823,
+      "eval_accuracy": 0.8295,
+      "eval_loss": 0.565258800983429,
+      "eval_runtime": 5.4089,
+      "eval_samples_per_second": 1848.799,
       "eval_steps_per_second": 7.395,
       "step": 87579
     },
     {
       "epoch": 263.0,
-      "grad_norm": 6.324307441711426,
+      "grad_norm": 7.199764728546143,
       "learning_rate": 1.2332332332332333e-06,
-      "loss": 1.0627,
+      "loss": 1.0727,
       "step": 87580
     },
     {
       "epoch": 263.03,
-      "grad_norm": 7.564157485961914,
+      "grad_norm": 5.709984302520752,
       "learning_rate": 1.2322322322322324e-06,
-      "loss": 0.8648,
+      "loss": 0.8657,
       "step": 87590
     },
     {
       "epoch": 263.06,
-      "grad_norm": 5.410341262817383,
+      "grad_norm": 8.29051685333252,
       "learning_rate": 1.2312312312312314e-06,
-      "loss": 0.8765,
+      "loss": 0.8708,
       "step": 87600
     },
     {
       "epoch": 263.09,
-      "grad_norm": 6.600475311279297,
+      "grad_norm": 8.093719482421875,
       "learning_rate": 1.2302302302302304e-06,
-      "loss": 0.8962,
+      "loss": 0.8985,
       "step": 87610
     },
     {
       "epoch": 263.12,
-      "grad_norm": 9.137456893920898,
+      "grad_norm": 7.92423677444458,
       "learning_rate": 1.2292292292292294e-06,
-      "loss": 0.9011,
+      "loss": 0.9035,
       "step": 87620
     },
     {
       "epoch": 263.15,
-      "grad_norm": 7.641957759857178,
+      "grad_norm": 8.291467666625977,
       "learning_rate": 1.2282282282282282e-06,
-      "loss": 0.949,
+      "loss": 0.9447,
       "step": 87630
     },
     {
       "epoch": 263.18,
-      "grad_norm": 7.7544379234313965,
+      "grad_norm": 6.61473274230957,
       "learning_rate": 1.2272272272272272e-06,
-      "loss": 0.9072,
+      "loss": 0.8976,
       "step": 87640
     },
     {
       "epoch": 263.21,
-      "grad_norm": 7.129770278930664,
+      "grad_norm": 8.45995807647705,
       "learning_rate": 1.2262262262262263e-06,
-      "loss": 0.9248,
+      "loss": 0.9128,
       "step": 87650
     },
     {
       "epoch": 263.24,
-      "grad_norm": 8.871810913085938,
+      "grad_norm": 5.920749664306641,
       "learning_rate": 1.2252252252252253e-06,
-      "loss": 0.9173,
+      "loss": 0.9136,
       "step": 87660
     },
     {
       "epoch": 263.27,
-      "grad_norm": 5.552388668060303,
+      "grad_norm": 5.558622360229492,
       "learning_rate": 1.2242242242242243e-06,
-      "loss": 0.8384,
+      "loss": 0.8324,
       "step": 87670
     },
     {
       "epoch": 263.3,
-      "grad_norm": 8.915651321411133,
+      "grad_norm": 12.318538665771484,
       "learning_rate": 1.2232232232232233e-06,
-      "loss": 0.9339,
+      "loss": 0.9378,
       "step": 87680
     },
     {
       "epoch": 263.33,
-      "grad_norm": 7.9364705085754395,
+      "grad_norm": 7.442850589752197,
       "learning_rate": 1.2222222222222223e-06,
-      "loss": 0.9502,
+      "loss": 0.9416,
       "step": 87690
     },
     {
       "epoch": 263.36,
-      "grad_norm": 9.596973419189453,
+      "grad_norm": 10.359882354736328,
       "learning_rate": 1.2212212212212213e-06,
-      "loss": 0.916,
+      "loss": 0.9053,
       "step": 87700
     },
     {
       "epoch": 263.39,
-      "grad_norm": 5.7425689697265625,
+      "grad_norm": 7.672226428985596,
       "learning_rate": 1.2202202202202204e-06,
-      "loss": 0.9213,
+      "loss": 0.922,
       "step": 87710
     },
     {
       "epoch": 263.42,
-      "grad_norm": 7.803315162658691,
+      "grad_norm": 6.871875762939453,
       "learning_rate": 1.2192192192192192e-06,
-      "loss": 0.8964,
+      "loss": 0.8939,
       "step": 87720
     },
     {
       "epoch": 263.45,
-      "grad_norm": 5.370334148406982,
+      "grad_norm": 6.1575517654418945,
       "learning_rate": 1.2182182182182182e-06,
-      "loss": 0.8983,
+      "loss": 0.9047,
       "step": 87730
     },
     {
       "epoch": 263.48,
-      "grad_norm": 5.131213188171387,
+      "grad_norm": 5.368655681610107,
       "learning_rate": 1.2172172172172172e-06,
-      "loss": 0.9395,
+      "loss": 0.9378,
       "step": 87740
     },
     {
       "epoch": 263.51,
-      "grad_norm": 4.734607219696045,
+      "grad_norm": 5.642495632171631,
       "learning_rate": 1.2162162162162164e-06,
-      "loss": 0.8588,
+      "loss": 0.8504,
       "step": 87750
     },
     {
       "epoch": 263.54,
-      "grad_norm": 8.464799880981445,
+      "grad_norm": 10.359393119812012,
       "learning_rate": 1.2152152152152155e-06,
-      "loss": 0.858,
+      "loss": 0.8629,
       "step": 87760
     },
     {
       "epoch": 263.57,
-      "grad_norm": 8.509507179260254,
+      "grad_norm": 5.328103542327881,
       "learning_rate": 1.2142142142142143e-06,
-      "loss": 0.863,
+      "loss": 0.8556,
       "step": 87770
     },
     {
       "epoch": 263.6,
-      "grad_norm": 4.580979824066162,
+      "grad_norm": 5.505081653594971,
       "learning_rate": 1.2132132132132133e-06,
-      "loss": 0.8341,
+      "loss": 0.8296,
       "step": 87780
     },
     {
       "epoch": 263.63,
-      "grad_norm": 7.472888469696045,
+      "grad_norm": 6.999597549438477,
       "learning_rate": 1.2122122122122123e-06,
-      "loss": 0.882,
+      "loss": 0.8709,
       "step": 87790
     },
     {
       "epoch": 263.66,
-      "grad_norm": 5.9962239265441895,
+      "grad_norm": 7.268321514129639,
       "learning_rate": 1.2112112112112113e-06,
-      "loss": 0.8308,
+      "loss": 0.8328,
       "step": 87800
     },
     {
       "epoch": 263.69,
-      "grad_norm": 5.395132541656494,
+      "grad_norm": 6.276205539703369,
       "learning_rate": 1.2102102102102103e-06,
-      "loss": 0.927,
+      "loss": 0.9285,
       "step": 87810
     },
     {
       "epoch": 263.72,
-      "grad_norm": 6.564889430999756,
+      "grad_norm": 5.92985725402832,
       "learning_rate": 1.2092092092092094e-06,
-      "loss": 0.9224,
+      "loss": 0.9164,
       "step": 87820
     },
     {
       "epoch": 263.75,
-      "grad_norm": 6.465169429779053,
+      "grad_norm": 7.771515369415283,
       "learning_rate": 1.2082082082082084e-06,
-      "loss": 0.8999,
+      "loss": 0.9003,
       "step": 87830
     },
     {
       "epoch": 263.78,
-      "grad_norm": 6.011582374572754,
+      "grad_norm": 6.411550521850586,
       "learning_rate": 1.2072072072072074e-06,
-      "loss": 0.9991,
+      "loss": 0.991,
       "step": 87840
     },
     {
       "epoch": 263.81,
-      "grad_norm": 5.672259330749512,
+      "grad_norm": 6.253134727478027,
       "learning_rate": 1.2062062062062064e-06,
-      "loss": 0.8735,
+      "loss": 0.8543,
       "step": 87850
     },
     {
       "epoch": 263.84,
-      "grad_norm": 8.442880630493164,
+      "grad_norm": 7.218719005584717,
       "learning_rate": 1.2052052052052052e-06,
-      "loss": 0.9625,
+      "loss": 0.9582,
       "step": 87860
     },
     {
       "epoch": 263.87,
-      "grad_norm": 6.293641090393066,
+      "grad_norm": 7.035247802734375,
       "learning_rate": 1.2042042042042042e-06,
-      "loss": 0.9028,
+      "loss": 0.9054,
       "step": 87870
     },
     {
       "epoch": 263.9,
-      "grad_norm": 8.529809951782227,
+      "grad_norm": 8.805032730102539,
       "learning_rate": 1.2032032032032033e-06,
-      "loss": 0.879,
+      "loss": 0.876,
       "step": 87880
     },
     {
       "epoch": 263.93,
-      "grad_norm": 6.5636887550354,
+      "grad_norm": 7.34840726852417,
       "learning_rate": 1.2022022022022023e-06,
-      "loss": 0.9603,
+      "loss": 0.946,
       "step": 87890
     },
     {
       "epoch": 263.96,
-      "grad_norm": 6.489180564880371,
+      "grad_norm": 7.009896278381348,
       "learning_rate": 1.2012012012012013e-06,
-      "loss": 0.9488,
+      "loss": 0.9438,
       "step": 87900
     },
     {
       "epoch": 263.99,
-      "grad_norm": 5.44530725479126,
+      "grad_norm": 5.906706809997559,
       "learning_rate": 1.2002002002002003e-06,
-      "loss": 0.8706,
+      "loss": 0.8753,
       "step": 87910
     },
     {
       "epoch": 264.0,
-      "eval_accuracy": 0.8259,
-      "eval_loss": 0.5742642879486084,
-      "eval_runtime": 5.6573,
-      "eval_samples_per_second": 1767.616,
-      "eval_steps_per_second": 7.07,
+      "eval_accuracy": 0.8275,
+      "eval_loss": 0.5729371309280396,
+      "eval_runtime": 5.4595,
+      "eval_samples_per_second": 1831.659,
+      "eval_steps_per_second": 7.327,
       "step": 87912
     },
     {
       "epoch": 264.02,
-      "grad_norm": 6.877518177032471,
+      "grad_norm": 4.9998345375061035,
       "learning_rate": 1.1991991991991993e-06,
-      "loss": 1.0448,
+      "loss": 1.0405,
       "step": 87920
     },
     {
       "epoch": 264.05,
-      "grad_norm": 6.318953514099121,
+      "grad_norm": 6.385214805603027,
       "learning_rate": 1.1981981981981983e-06,
-      "loss": 0.8778,
+      "loss": 0.8735,
       "step": 87930
     },
     {
       "epoch": 264.08,
-      "grad_norm": 8.707093238830566,
+      "grad_norm": 7.340390682220459,
       "learning_rate": 1.1971971971971974e-06,
-      "loss": 0.9214,
+      "loss": 0.9215,
       "step": 87940
     },
     {
       "epoch": 264.11,
-      "grad_norm": 6.478142738342285,
+      "grad_norm": 7.970912456512451,
       "learning_rate": 1.1961961961961962e-06,
-      "loss": 0.9134,
+      "loss": 0.9067,
       "step": 87950
     },
     {
       "epoch": 264.14,
-      "grad_norm": 6.317121982574463,
+      "grad_norm": 6.320915222167969,
       "learning_rate": 1.1951951951951952e-06,
-      "loss": 0.8719,
+      "loss": 0.8605,
       "step": 87960
     },
     {
       "epoch": 264.17,
-      "grad_norm": 5.597269535064697,
+      "grad_norm": 5.922503471374512,
       "learning_rate": 1.1941941941941942e-06,
-      "loss": 0.8462,
+      "loss": 0.8489,
       "step": 87970
     },
     {
       "epoch": 264.2,
-      "grad_norm": 7.67670202255249,
+      "grad_norm": 5.83127498626709,
       "learning_rate": 1.1931931931931932e-06,
-      "loss": 0.9153,
+      "loss": 0.919,
       "step": 87980
     },
     {
       "epoch": 264.23,
-      "grad_norm": 5.583221435546875,
+      "grad_norm": 7.03609561920166,
       "learning_rate": 1.1921921921921922e-06,
-      "loss": 0.9064,
+      "loss": 0.8978,
       "step": 87990
     },
     {
       "epoch": 264.26,
-      "grad_norm": 8.533260345458984,
+      "grad_norm": 7.5621843338012695,
       "learning_rate": 1.1911911911911913e-06,
-      "loss": 0.9152,
+      "loss": 0.918,
       "step": 88000
     },
     {
       "epoch": 264.29,
-      "grad_norm": 5.3090362548828125,
+      "grad_norm": 6.936607360839844,
       "learning_rate": 1.1901901901901903e-06,
-      "loss": 0.8897,
+      "loss": 0.8774,
       "step": 88010
     },
     {
       "epoch": 264.32,
-      "grad_norm": 6.097050189971924,
+      "grad_norm": 5.63986349105835,
       "learning_rate": 1.1891891891891893e-06,
-      "loss": 0.8418,
+      "loss": 0.845,
       "step": 88020
     },
     {
       "epoch": 264.35,
-      "grad_norm": 8.677290916442871,
+      "grad_norm": 8.005338668823242,
       "learning_rate": 1.1881881881881883e-06,
-      "loss": 0.906,
+      "loss": 0.923,
       "step": 88030
     },
     {
       "epoch": 264.38,
-      "grad_norm": 7.6109819412231445,
+      "grad_norm": 9.56567096710205,
       "learning_rate": 1.1871871871871873e-06,
-      "loss": 0.9185,
+      "loss": 0.911,
       "step": 88040
     },
     {
       "epoch": 264.41,
-      "grad_norm": 6.24796199798584,
+      "grad_norm": 6.268744945526123,
       "learning_rate": 1.1861861861861864e-06,
-      "loss": 0.9215,
+      "loss": 0.9101,
       "step": 88050
     },
     {
       "epoch": 264.44,
-      "grad_norm": 7.585369110107422,
+      "grad_norm": 5.373239040374756,
       "learning_rate": 1.1851851851851854e-06,
-      "loss": 0.9061,
+      "loss": 0.8945,
       "step": 88060
     },
     {
       "epoch": 264.47,
-      "grad_norm": 6.473397254943848,
+      "grad_norm": 5.668124198913574,
       "learning_rate": 1.1841841841841844e-06,
-      "loss": 0.9584,
+      "loss": 0.951,
       "step": 88070
     },
     {
       "epoch": 264.5,
-      "grad_norm": 5.835780143737793,
+      "grad_norm": 6.980535507202148,
       "learning_rate": 1.1831831831831834e-06,
-      "loss": 0.8746,
+      "loss": 0.8735,
       "step": 88080
     },
     {
       "epoch": 264.53,
-      "grad_norm": 8.741701126098633,
+      "grad_norm": 6.928752422332764,
       "learning_rate": 1.1821821821821822e-06,
-      "loss": 0.9536,
+      "loss": 0.95,
       "step": 88090
     },
     {
       "epoch": 264.56,
-      "grad_norm": 6.178422451019287,
+      "grad_norm": 6.347029209136963,
       "learning_rate": 1.1811811811811812e-06,
-      "loss": 0.9019,
+      "loss": 0.9049,
       "step": 88100
     },
     {
       "epoch": 264.59,
-      "grad_norm": 6.566783428192139,
+      "grad_norm": 8.823602676391602,
       "learning_rate": 1.1801801801801803e-06,
-      "loss": 0.89,
+      "loss": 0.9044,
       "step": 88110
     },
     {
       "epoch": 264.62,
-      "grad_norm": 6.828951835632324,
+      "grad_norm": 7.084773063659668,
       "learning_rate": 1.1791791791791793e-06,
-      "loss": 0.9172,
+      "loss": 0.9153,
       "step": 88120
     },
     {
       "epoch": 264.65,
-      "grad_norm": 5.583493709564209,
+      "grad_norm": 6.1874680519104,
       "learning_rate": 1.1781781781781783e-06,
-      "loss": 0.9273,
+      "loss": 0.9238,
       "step": 88130
     },
     {
       "epoch": 264.68,
-      "grad_norm": 7.464349269866943,
+      "grad_norm": 9.041900634765625,
       "learning_rate": 1.1771771771771773e-06,
-      "loss": 0.9175,
+      "loss": 0.9254,
       "step": 88140
     },
     {
       "epoch": 264.71,
-      "grad_norm": 7.118016719818115,
+      "grad_norm": 8.208795547485352,
       "learning_rate": 1.1761761761761763e-06,
-      "loss": 1.0097,
+      "loss": 1.0033,
       "step": 88150
     },
     {
       "epoch": 264.74,
-      "grad_norm": 7.106741905212402,
+      "grad_norm": 6.117269992828369,
       "learning_rate": 1.1751751751751753e-06,
-      "loss": 0.8853,
+      "loss": 0.881,
       "step": 88160
     },
     {
       "epoch": 264.77,
-      "grad_norm": 6.765805244445801,
+      "grad_norm": 8.450068473815918,
       "learning_rate": 1.1741741741741742e-06,
-      "loss": 0.8952,
+      "loss": 0.8845,
       "step": 88170
     },
     {
       "epoch": 264.8,
-      "grad_norm": 10.41069221496582,
+      "grad_norm": 10.999812126159668,
       "learning_rate": 1.1731731731731732e-06,
-      "loss": 0.9153,
+      "loss": 0.9184,
       "step": 88180
     },
     {
       "epoch": 264.83,
-      "grad_norm": 4.561258792877197,
+      "grad_norm": 5.786318778991699,
       "learning_rate": 1.1721721721721722e-06,
-      "loss": 0.8992,
+      "loss": 0.8941,
       "step": 88190
     },
     {
       "epoch": 264.86,
-      "grad_norm": 7.265446662902832,
+      "grad_norm": 10.902735710144043,
       "learning_rate": 1.1711711711711712e-06,
-      "loss": 0.8862,
+      "loss": 0.8959,
       "step": 88200
     },
     {
       "epoch": 264.89,
-      "grad_norm": 4.920050144195557,
+      "grad_norm": 5.138391494750977,
       "learning_rate": 1.1701701701701702e-06,
-      "loss": 0.8828,
+      "loss": 0.8987,
       "step": 88210
     },
     {
       "epoch": 264.92,
-      "grad_norm": 5.4820356369018555,
+      "grad_norm": 7.0100507736206055,
       "learning_rate": 1.1691691691691692e-06,
-      "loss": 0.8529,
+      "loss": 0.8497,
       "step": 88220
     },
     {
       "epoch": 264.95,
-      "grad_norm": 15.351093292236328,
+      "grad_norm": 7.150150775909424,
       "learning_rate": 1.1681681681681683e-06,
-      "loss": 0.9016,
+      "loss": 0.8948,
       "step": 88230
     },
     {
       "epoch": 264.98,
-      "grad_norm": 6.284565448760986,
+      "grad_norm": 5.747974395751953,
       "learning_rate": 1.1671671671671673e-06,
-      "loss": 0.8743,
+      "loss": 0.8684,
       "step": 88240
     },
     {
       "epoch": 265.0,
-      "eval_accuracy": 0.8283,
-      "eval_loss": 0.567086398601532,
-      "eval_runtime": 5.4385,
-      "eval_samples_per_second": 1838.736,
-      "eval_steps_per_second": 7.355,
+      "eval_accuracy": 0.8295,
+      "eval_loss": 0.5672117471694946,
+      "eval_runtime": 5.432,
+      "eval_samples_per_second": 1840.943,
+      "eval_steps_per_second": 7.364,
       "step": 88245
     },
     {
       "epoch": 265.02,
-      "grad_norm": 8.35142993927002,
+      "grad_norm": 9.814544677734375,
       "learning_rate": 1.1661661661661663e-06,
-      "loss": 1.088,
+      "loss": 1.0923,
       "step": 88250
     },
     {
       "epoch": 265.05,
-      "grad_norm": 6.1624369621276855,
+      "grad_norm": 8.2359037399292,
       "learning_rate": 1.1651651651651651e-06,
-      "loss": 0.9234,
+      "loss": 0.9203,
       "step": 88260
     },
     {
       "epoch": 265.08,
-      "grad_norm": 4.8497209548950195,
+      "grad_norm": 5.801060199737549,
       "learning_rate": 1.1641641641641641e-06,
-      "loss": 0.9343,
+      "loss": 0.9328,
       "step": 88270
     },
     {
       "epoch": 265.11,
-      "grad_norm": 5.210460186004639,
+      "grad_norm": 6.843503475189209,
       "learning_rate": 1.1631631631631631e-06,
-      "loss": 0.8375,
+      "loss": 0.8322,
       "step": 88280
     },
     {
       "epoch": 265.14,
-      "grad_norm": 6.695747375488281,
+      "grad_norm": 7.886844635009766,
       "learning_rate": 1.1621621621621624e-06,
-      "loss": 0.9556,
+      "loss": 0.965,
       "step": 88290
     },
     {
       "epoch": 265.17,
-      "grad_norm": 9.557063102722168,
+      "grad_norm": 8.32577896118164,
       "learning_rate": 1.1611611611611614e-06,
-      "loss": 0.9687,
+      "loss": 0.9802,
       "step": 88300
     },
     {
       "epoch": 265.2,
-      "grad_norm": 9.358255386352539,
+      "grad_norm": 9.409056663513184,
       "learning_rate": 1.1601601601601604e-06,
-      "loss": 0.9443,
+      "loss": 0.9367,
       "step": 88310
     },
     {
       "epoch": 265.23,
-      "grad_norm": 7.384705543518066,
+      "grad_norm": 6.2976393699646,
       "learning_rate": 1.1591591591591592e-06,
-      "loss": 0.8792,
+      "loss": 0.8716,
       "step": 88320
     },
     {
       "epoch": 265.26,
-      "grad_norm": 5.784841060638428,
+      "grad_norm": 6.833142280578613,
       "learning_rate": 1.1581581581581582e-06,
-      "loss": 0.931,
+      "loss": 0.9305,
       "step": 88330
     },
     {
       "epoch": 265.29,
-      "grad_norm": 6.078474044799805,
+      "grad_norm": 4.92150354385376,
       "learning_rate": 1.1571571571571573e-06,
-      "loss": 0.8648,
+      "loss": 0.8638,
       "step": 88340
     },
     {
       "epoch": 265.32,
-      "grad_norm": 5.7668890953063965,
+      "grad_norm": 5.494810581207275,
       "learning_rate": 1.1561561561561563e-06,
-      "loss": 0.83,
+      "loss": 0.8259,
       "step": 88350
     },
     {
       "epoch": 265.35,
-      "grad_norm": 7.599725723266602,
+      "grad_norm": 10.259489059448242,
       "learning_rate": 1.1551551551551553e-06,
-      "loss": 0.9394,
+      "loss": 0.9318,
       "step": 88360
     },
     {
       "epoch": 265.38,
-      "grad_norm": 11.528825759887695,
+      "grad_norm": 7.908612251281738,
       "learning_rate": 1.1541541541541543e-06,
-      "loss": 0.888,
+      "loss": 0.8865,
       "step": 88370
     },
     {
       "epoch": 265.41,
-      "grad_norm": 8.611010551452637,
+      "grad_norm": 8.873766899108887,
       "learning_rate": 1.1531531531531533e-06,
-      "loss": 0.8935,
+      "loss": 0.9072,
       "step": 88380
     },
     {
       "epoch": 265.44,
-      "grad_norm": 6.458399295806885,
+      "grad_norm": 5.179502010345459,
       "learning_rate": 1.1521521521521523e-06,
-      "loss": 0.9528,
+      "loss": 0.9474,
       "step": 88390
     },
     {
       "epoch": 265.47,
-      "grad_norm": 8.270788192749023,
+      "grad_norm": 9.555233001708984,
       "learning_rate": 1.1511511511511512e-06,
-      "loss": 0.9648,
+      "loss": 0.9571,
       "step": 88400
     },
     {
       "epoch": 265.5,
-      "grad_norm": 8.720144271850586,
+      "grad_norm": 9.231612205505371,
       "learning_rate": 1.1501501501501502e-06,
-      "loss": 0.8909,
+      "loss": 0.8788,
       "step": 88410
     },
     {
       "epoch": 265.53,
-      "grad_norm": 6.961448669433594,
+      "grad_norm": 12.042704582214355,
       "learning_rate": 1.1491491491491492e-06,
-      "loss": 0.8827,
+      "loss": 0.8763,
       "step": 88420
     },
     {
       "epoch": 265.56,
-      "grad_norm": 5.6030778884887695,
+      "grad_norm": 4.5163397789001465,
       "learning_rate": 1.1481481481481482e-06,
-      "loss": 0.8664,
+      "loss": 0.8639,
       "step": 88430
     },
     {
       "epoch": 265.59,
-      "grad_norm": 8.065391540527344,
+      "grad_norm": 8.0321626663208,
       "learning_rate": 1.1471471471471472e-06,
-      "loss": 0.9344,
+      "loss": 0.9426,
       "step": 88440
     },
     {
       "epoch": 265.62,
-      "grad_norm": 4.6076979637146,
+      "grad_norm": 5.074545383453369,
       "learning_rate": 1.1461461461461462e-06,
-      "loss": 0.8886,
+      "loss": 0.8875,
       "step": 88450
     },
     {
       "epoch": 265.65,
-      "grad_norm": 5.827554225921631,
+      "grad_norm": 5.658087730407715,
       "learning_rate": 1.1451451451451453e-06,
       "loss": 0.8701,
       "step": 88460
     },
     {
       "epoch": 265.68,
-      "grad_norm": 6.625877857208252,
+      "grad_norm": 6.733203887939453,
       "learning_rate": 1.1441441441441443e-06,
-      "loss": 0.9363,
+      "loss": 0.9373,
       "step": 88470
     },
     {
       "epoch": 265.71,
-      "grad_norm": 6.282569408416748,
+      "grad_norm": 10.077024459838867,
       "learning_rate": 1.1431431431431433e-06,
-      "loss": 0.8797,
+      "loss": 0.8867,
       "step": 88480
     },
     {
       "epoch": 265.74,
-      "grad_norm": 6.511885643005371,
+      "grad_norm": 6.923529148101807,
       "learning_rate": 1.142142142142142e-06,
-      "loss": 0.8956,
+      "loss": 0.8935,
       "step": 88490
     },
     {
       "epoch": 265.77,
-      "grad_norm": 6.379395484924316,
+      "grad_norm": 4.834393501281738,
       "learning_rate": 1.1411411411411411e-06,
-      "loss": 0.9167,
+      "loss": 0.9087,
       "step": 88500
     },
     {
       "epoch": 265.8,
-      "grad_norm": 6.362008571624756,
+      "grad_norm": 5.579935073852539,
       "learning_rate": 1.1401401401401401e-06,
-      "loss": 0.9172,
+      "loss": 0.9125,
       "step": 88510
     },
     {
       "epoch": 265.83,
-      "grad_norm": 6.049336910247803,
+      "grad_norm": 6.63053035736084,
       "learning_rate": 1.1391391391391392e-06,
-      "loss": 0.9655,
+      "loss": 0.9726,
       "step": 88520
     },
     {
       "epoch": 265.86,
-      "grad_norm": 6.466190338134766,
+      "grad_norm": 8.258739471435547,
       "learning_rate": 1.1381381381381382e-06,
-      "loss": 0.9459,
+      "loss": 0.9394,
       "step": 88530
     },
     {
       "epoch": 265.89,
-      "grad_norm": 8.505845069885254,
+      "grad_norm": 7.646618366241455,
       "learning_rate": 1.1371371371371372e-06,
-      "loss": 0.9225,
+      "loss": 0.9134,
       "step": 88540
     },
     {
       "epoch": 265.92,
-      "grad_norm": 7.271946907043457,
+      "grad_norm": 8.297128677368164,
       "learning_rate": 1.1361361361361362e-06,
-      "loss": 0.9396,
+      "loss": 0.9312,
       "step": 88550
     },
     {
       "epoch": 265.95,
-      "grad_norm": 5.789492607116699,
+      "grad_norm": 7.674524307250977,
       "learning_rate": 1.1351351351351352e-06,
-      "loss": 0.8386,
+      "loss": 0.8376,
       "step": 88560
     },
     {
       "epoch": 265.98,
-      "grad_norm": 6.425172328948975,
+      "grad_norm": 8.582612991333008,
       "learning_rate": 1.1341341341341343e-06,
-      "loss": 0.9297,
+      "loss": 0.9311,
       "step": 88570
     },
     {
       "epoch": 266.0,
-      "eval_accuracy": 0.8294,
-      "eval_loss": 0.5726470351219177,
-      "eval_runtime": 5.4551,
-      "eval_samples_per_second": 1833.151,
-      "eval_steps_per_second": 7.333,
+      "eval_accuracy": 0.829,
+      "eval_loss": 0.5724761486053467,
+      "eval_runtime": 5.3542,
+      "eval_samples_per_second": 1867.678,
+      "eval_steps_per_second": 7.471,
       "step": 88578
     },
     {
       "epoch": 266.01,
-      "grad_norm": 5.523800373077393,
+      "grad_norm": 4.501252174377441,
       "learning_rate": 1.1331331331331333e-06,
-      "loss": 1.2119,
+      "loss": 1.2398,
       "step": 88580
     },
     {
       "epoch": 266.04,
-      "grad_norm": 7.587114334106445,
+      "grad_norm": 7.873360633850098,
       "learning_rate": 1.1321321321321323e-06,
-      "loss": 0.9016,
+      "loss": 0.9036,
       "step": 88590
     },
     {
       "epoch": 266.07,
-      "grad_norm": 7.816417694091797,
+      "grad_norm": 6.3114705085754395,
       "learning_rate": 1.1311311311311313e-06,
-      "loss": 0.8671,
+      "loss": 0.8652,
       "step": 88600
     },
     {
       "epoch": 266.1,
-      "grad_norm": 5.829999923706055,
+      "grad_norm": 7.154572486877441,
       "learning_rate": 1.1301301301301303e-06,
-      "loss": 0.9349,
+      "loss": 0.9298,
       "step": 88610
     },
     {
       "epoch": 266.13,
-      "grad_norm": 4.622116565704346,
+      "grad_norm": 5.629049777984619,
       "learning_rate": 1.1291291291291293e-06,
-      "loss": 0.9278,
+      "loss": 0.9306,
       "step": 88620
     },
     {
       "epoch": 266.16,
-      "grad_norm": 8.934247970581055,
+      "grad_norm": 8.22643756866455,
       "learning_rate": 1.1281281281281281e-06,
-      "loss": 0.8943,
+      "loss": 0.8873,
       "step": 88630
     },
     {
       "epoch": 266.19,
-      "grad_norm": 6.3735198974609375,
+      "grad_norm": 6.9525675773620605,
       "learning_rate": 1.1271271271271272e-06,
-      "loss": 0.9532,
+      "loss": 0.9526,
       "step": 88640
     },
     {
       "epoch": 266.22,
-      "grad_norm": 6.956852436065674,
+      "grad_norm": 7.547227382659912,
       "learning_rate": 1.1261261261261262e-06,
-      "loss": 0.932,
+      "loss": 0.9228,
       "step": 88650
     },
     {
       "epoch": 266.25,
-      "grad_norm": 6.5581889152526855,
+      "grad_norm": 10.586564064025879,
       "learning_rate": 1.1251251251251252e-06,
-      "loss": 0.9161,
+      "loss": 0.9175,
       "step": 88660
     },
     {
       "epoch": 266.28,
-      "grad_norm": 7.28289794921875,
+      "grad_norm": 7.498548984527588,
       "learning_rate": 1.1241241241241242e-06,
-      "loss": 0.9226,
+      "loss": 0.922,
       "step": 88670
     },
     {
       "epoch": 266.31,
-      "grad_norm": 5.707620620727539,
+      "grad_norm": 7.442488670349121,
       "learning_rate": 1.1231231231231232e-06,
-      "loss": 0.8811,
+      "loss": 0.8768,
       "step": 88680
     },
     {
       "epoch": 266.34,
-      "grad_norm": 6.801765441894531,
+      "grad_norm": 7.483349323272705,
       "learning_rate": 1.1221221221221223e-06,
-      "loss": 0.9222,
+      "loss": 0.9128,
       "step": 88690
     },
     {
       "epoch": 266.37,
-      "grad_norm": 6.183238506317139,
+      "grad_norm": 8.681123733520508,
       "learning_rate": 1.1211211211211213e-06,
-      "loss": 0.9248,
+      "loss": 0.915,
       "step": 88700
     },
     {
       "epoch": 266.4,
-      "grad_norm": 5.8214592933654785,
+      "grad_norm": 8.117988586425781,
       "learning_rate": 1.1201201201201203e-06,
-      "loss": 0.895,
+      "loss": 0.9011,
       "step": 88710
     },
     {
       "epoch": 266.43,
-      "grad_norm": 7.899796485900879,
+      "grad_norm": 6.873058795928955,
       "learning_rate": 1.119119119119119e-06,
-      "loss": 0.9494,
+      "loss": 0.9546,
       "step": 88720
     },
     {
       "epoch": 266.46,
-      "grad_norm": 4.912955284118652,
+      "grad_norm": 7.456656455993652,
       "learning_rate": 1.1181181181181181e-06,
-      "loss": 0.9156,
+      "loss": 0.9169,
       "step": 88730
     },
     {
       "epoch": 266.49,
-      "grad_norm": 6.340951442718506,
+      "grad_norm": 6.10158109664917,
       "learning_rate": 1.1171171171171171e-06,
-      "loss": 0.847,
+      "loss": 0.8337,
       "step": 88740
     },
     {
       "epoch": 266.52,
-      "grad_norm": 8.702261924743652,
+      "grad_norm": 8.099601745605469,
       "learning_rate": 1.1161161161161162e-06,
-      "loss": 0.8746,
+      "loss": 0.8715,
       "step": 88750
     },
     {
       "epoch": 266.55,
-      "grad_norm": 6.0486955642700195,
+      "grad_norm": 6.417669773101807,
       "learning_rate": 1.1151151151151152e-06,
-      "loss": 0.8627,
+      "loss": 0.8585,
       "step": 88760
     },
     {
       "epoch": 266.58,
-      "grad_norm": 5.74891996383667,
+      "grad_norm": 7.2861714363098145,
       "learning_rate": 1.1141141141141142e-06,
-      "loss": 0.9602,
+      "loss": 0.9609,
       "step": 88770
     },
     {
       "epoch": 266.61,
-      "grad_norm": 5.995080471038818,
+      "grad_norm": 6.7808003425598145,
       "learning_rate": 1.1131131131131132e-06,
-      "loss": 0.8933,
+      "loss": 0.8785,
       "step": 88780
     },
     {
       "epoch": 266.64,
-      "grad_norm": 6.412467002868652,
+      "grad_norm": 6.755853176116943,
       "learning_rate": 1.1121121121121122e-06,
-      "loss": 0.8858,
+      "loss": 0.8797,
       "step": 88790
     },
     {
       "epoch": 266.67,
-      "grad_norm": 6.3512282371521,
+      "grad_norm": 6.106185436248779,
       "learning_rate": 1.111111111111111e-06,
-      "loss": 0.913,
+      "loss": 0.9098,
       "step": 88800
     },
     {
       "epoch": 266.7,
-      "grad_norm": 7.2207112312316895,
+      "grad_norm": 9.069862365722656,
       "learning_rate": 1.11011011011011e-06,
-      "loss": 0.9323,
+      "loss": 0.9244,
       "step": 88810
     },
     {
       "epoch": 266.73,
-      "grad_norm": 6.750175952911377,
+      "grad_norm": 5.275566101074219,
       "learning_rate": 1.1091091091091093e-06,
-      "loss": 0.9091,
+      "loss": 0.9145,
       "step": 88820
     },
     {
       "epoch": 266.76,
-      "grad_norm": 5.406076908111572,
+      "grad_norm": 6.903679847717285,
       "learning_rate": 1.1081081081081083e-06,
-      "loss": 0.8944,
+      "loss": 0.8839,
       "step": 88830
     },
     {
       "epoch": 266.79,
-      "grad_norm": 5.688664436340332,
+      "grad_norm": 6.221364974975586,
       "learning_rate": 1.1071071071071073e-06,
-      "loss": 0.9237,
+      "loss": 0.918,
       "step": 88840
     },
     {
       "epoch": 266.82,
-      "grad_norm": 6.372854232788086,
+      "grad_norm": 10.268579483032227,
       "learning_rate": 1.1061061061061063e-06,
-      "loss": 0.8802,
+      "loss": 0.8774,
       "step": 88850
     },
     {
       "epoch": 266.85,
-      "grad_norm": 6.8278374671936035,
+      "grad_norm": 4.931076526641846,
       "learning_rate": 1.1051051051051051e-06,
-      "loss": 0.9073,
+      "loss": 0.8932,
       "step": 88860
     },
     {
       "epoch": 266.88,
-      "grad_norm": 5.077925205230713,
+      "grad_norm": 5.550628185272217,
       "learning_rate": 1.1041041041041042e-06,
-      "loss": 0.9103,
+      "loss": 0.9153,
       "step": 88870
     },
     {
       "epoch": 266.91,
-      "grad_norm": 6.200553894042969,
+      "grad_norm": 6.437033176422119,
       "learning_rate": 1.1031031031031032e-06,
-      "loss": 0.8926,
+      "loss": 0.8988,
       "step": 88880
     },
     {
       "epoch": 266.94,
-      "grad_norm": 4.459239482879639,
+      "grad_norm": 6.604865074157715,
       "learning_rate": 1.1021021021021022e-06,
-      "loss": 0.9092,
+      "loss": 0.9104,
       "step": 88890
     },
     {
       "epoch": 266.97,
-      "grad_norm": 6.766184329986572,
+      "grad_norm": 8.114964485168457,
       "learning_rate": 1.1011011011011012e-06,
-      "loss": 0.897,
+      "loss": 0.8974,
       "step": 88900
     },
     {
       "epoch": 267.0,
-      "grad_norm": 5.523940563201904,
+      "grad_norm": 5.92495059967041,
       "learning_rate": 1.1001001001001002e-06,
-      "loss": 0.9017,
+      "loss": 0.8922,
       "step": 88910
     },
     {
       "epoch": 267.0,
-      "eval_accuracy": 0.8282,
-      "eval_loss": 0.5751814246177673,
-      "eval_runtime": 5.3728,
-      "eval_samples_per_second": 1861.227,
-      "eval_steps_per_second": 7.445,
+      "eval_accuracy": 0.8295,
+      "eval_loss": 0.5742242932319641,
+      "eval_runtime": 5.3959,
+      "eval_samples_per_second": 1853.247,
+      "eval_steps_per_second": 7.413,
       "step": 88911
     },
     {
       "epoch": 267.03,
-      "grad_norm": 4.817924976348877,
+      "grad_norm": 6.472391605377197,
       "learning_rate": 1.0990990990990993e-06,
-      "loss": 1.0525,
+      "loss": 1.0606,
       "step": 88920
     },
     {
       "epoch": 267.06,
-      "grad_norm": 5.040170192718506,
+      "grad_norm": 5.887213230133057,
       "learning_rate": 1.0980980980980983e-06,
-      "loss": 0.9132,
+      "loss": 0.91,
       "step": 88930
     },
     {
       "epoch": 267.09,
-      "grad_norm": 5.7593793869018555,
+      "grad_norm": 7.266271114349365,
       "learning_rate": 1.0970970970970973e-06,
-      "loss": 0.9313,
+      "loss": 0.9347,
       "step": 88940
     },
     {
       "epoch": 267.12,
-      "grad_norm": 8.789324760437012,
+      "grad_norm": 7.452495098114014,
       "learning_rate": 1.096096096096096e-06,
-      "loss": 0.8999,
+      "loss": 0.9028,
       "step": 88950
     },
     {
       "epoch": 267.15,
-      "grad_norm": 4.94945764541626,
+      "grad_norm": 4.768435001373291,
       "learning_rate": 1.0950950950950951e-06,
-      "loss": 0.8617,
+      "loss": 0.8566,
       "step": 88960
     },
     {
       "epoch": 267.18,
-      "grad_norm": 6.082598686218262,
+      "grad_norm": 9.150209426879883,
       "learning_rate": 1.0940940940940941e-06,
-      "loss": 0.8976,
+      "loss": 0.8931,
       "step": 88970
     },
     {
       "epoch": 267.21,
-      "grad_norm": 5.144423007965088,
+      "grad_norm": 6.373785495758057,
       "learning_rate": 1.0930930930930932e-06,
-      "loss": 0.9651,
+      "loss": 0.955,
       "step": 88980
     },
     {
       "epoch": 267.24,
-      "grad_norm": 6.405306339263916,
+      "grad_norm": 7.749217510223389,
       "learning_rate": 1.0920920920920922e-06,
       "loss": 0.8783,
       "step": 88990
     },
     {
       "epoch": 267.27,
-      "grad_norm": 5.530896186828613,
+      "grad_norm": 5.468283653259277,
       "learning_rate": 1.0910910910910912e-06,
-      "loss": 0.891,
+      "loss": 0.8948,
       "step": 89000
     },
     {
       "epoch": 267.3,
-      "grad_norm": 4.865549087524414,
+      "grad_norm": 4.801423072814941,
       "learning_rate": 1.0900900900900902e-06,
-      "loss": 0.9312,
+      "loss": 0.9392,
       "step": 89010
     },
     {
       "epoch": 267.33,
-      "grad_norm": 7.323440074920654,
+      "grad_norm": 7.921946048736572,
       "learning_rate": 1.0890890890890892e-06,
-      "loss": 0.9398,
+      "loss": 0.9359,
       "step": 89020
     },
     {
       "epoch": 267.36,
-      "grad_norm": 5.587676048278809,
+      "grad_norm": 5.856663227081299,
       "learning_rate": 1.088088088088088e-06,
-      "loss": 0.927,
+      "loss": 0.9242,
       "step": 89030
     },
     {
       "epoch": 267.39,
-      "grad_norm": 5.821826934814453,
+      "grad_norm": 5.745721817016602,
       "learning_rate": 1.087087087087087e-06,
-      "loss": 0.87,
+      "loss": 0.8606,
       "step": 89040
     },
     {
       "epoch": 267.42,
-      "grad_norm": 5.46101713180542,
+      "grad_norm": 6.14887809753418,
       "learning_rate": 1.086086086086086e-06,
-      "loss": 0.8762,
+      "loss": 0.8656,
       "step": 89050
     },
     {
       "epoch": 267.45,
-      "grad_norm": 6.145132541656494,
+      "grad_norm": 6.032020092010498,
       "learning_rate": 1.085085085085085e-06,
-      "loss": 0.8869,
+      "loss": 0.883,
       "step": 89060
     },
     {
       "epoch": 267.48,
-      "grad_norm": 6.78792667388916,
+      "grad_norm": 5.851413726806641,
       "learning_rate": 1.0840840840840841e-06,
-      "loss": 0.9131,
+      "loss": 0.9036,
       "step": 89070
     },
     {
       "epoch": 267.51,
-      "grad_norm": 8.100372314453125,
+      "grad_norm": 9.498384475708008,
       "learning_rate": 1.0830830830830831e-06,
-      "loss": 0.979,
+      "loss": 0.9846,
       "step": 89080
     },
     {
       "epoch": 267.54,
-      "grad_norm": 7.921938896179199,
+      "grad_norm": 8.68620777130127,
       "learning_rate": 1.0820820820820821e-06,
-      "loss": 0.8818,
+      "loss": 0.8696,
       "step": 89090
     },
     {
       "epoch": 267.57,
-      "grad_norm": 6.779901027679443,
+      "grad_norm": 7.498527526855469,
       "learning_rate": 1.0810810810810812e-06,
-      "loss": 0.9328,
+      "loss": 0.9389,
       "step": 89100
     },
     {
       "epoch": 267.6,
-      "grad_norm": 6.827539920806885,
+      "grad_norm": 6.1228227615356445,
       "learning_rate": 1.0800800800800802e-06,
-      "loss": 0.9443,
+      "loss": 0.9415,
       "step": 89110
     },
     {
       "epoch": 267.63,
-      "grad_norm": 6.285449504852295,
+      "grad_norm": 7.35258674621582,
       "learning_rate": 1.0790790790790792e-06,
-      "loss": 0.876,
+      "loss": 0.8805,
       "step": 89120
     },
     {
       "epoch": 267.66,
-      "grad_norm": 8.2321138381958,
+      "grad_norm": 6.913578510284424,
       "learning_rate": 1.0780780780780782e-06,
-      "loss": 0.9513,
+      "loss": 0.9459,
       "step": 89130
     },
     {
       "epoch": 267.69,
-      "grad_norm": 7.305550575256348,
+      "grad_norm": 8.741549491882324,
       "learning_rate": 1.0770770770770772e-06,
-      "loss": 0.8727,
+      "loss": 0.8728,
       "step": 89140
     },
     {
       "epoch": 267.72,
-      "grad_norm": 5.967371940612793,
+      "grad_norm": 6.8460516929626465,
       "learning_rate": 1.0760760760760763e-06,
-      "loss": 0.8882,
+      "loss": 0.874,
       "step": 89150
     },
     {
       "epoch": 267.75,
-      "grad_norm": 8.321992874145508,
+      "grad_norm": 11.910537719726562,
       "learning_rate": 1.0750750750750753e-06,
-      "loss": 0.9054,
+      "loss": 0.9077,
       "step": 89160
     },
     {
       "epoch": 267.78,
-      "grad_norm": 6.265888214111328,
+      "grad_norm": 5.896631717681885,
       "learning_rate": 1.074074074074074e-06,
-      "loss": 0.9207,
+      "loss": 0.9281,
       "step": 89170
     },
     {
       "epoch": 267.81,
-      "grad_norm": 5.6585822105407715,
+      "grad_norm": 4.758500576019287,
       "learning_rate": 1.073073073073073e-06,
-      "loss": 0.9019,
+      "loss": 0.8981,
       "step": 89180
     },
     {
       "epoch": 267.84,
-      "grad_norm": 5.8806610107421875,
+      "grad_norm": 7.970211029052734,
       "learning_rate": 1.0720720720720721e-06,
-      "loss": 0.9189,
+      "loss": 0.9268,
       "step": 89190
     },
     {
       "epoch": 267.87,
-      "grad_norm": 6.458502292633057,
+      "grad_norm": 6.590489387512207,
       "learning_rate": 1.0710710710710711e-06,
-      "loss": 0.9265,
+      "loss": 0.9228,
       "step": 89200
     },
     {
       "epoch": 267.9,
-      "grad_norm": 7.724511623382568,
+      "grad_norm": 6.646440029144287,
       "learning_rate": 1.0700700700700702e-06,
-      "loss": 0.8741,
+      "loss": 0.8849,
       "step": 89210
     },
     {
       "epoch": 267.93,
-      "grad_norm": 6.562798023223877,
+      "grad_norm": 7.889343738555908,
       "learning_rate": 1.0690690690690692e-06,
-      "loss": 0.9357,
+      "loss": 0.9337,
       "step": 89220
     },
     {
       "epoch": 267.96,
-      "grad_norm": 7.238651275634766,
+      "grad_norm": 6.382678985595703,
       "learning_rate": 1.0680680680680682e-06,
-      "loss": 0.9276,
+      "loss": 0.9217,
       "step": 89230
     },
     {
       "epoch": 267.99,
-      "grad_norm": 5.598898410797119,
+      "grad_norm": 5.86279296875,
       "learning_rate": 1.0670670670670672e-06,
-      "loss": 0.9106,
+      "loss": 0.9141,
       "step": 89240
     },
     {
       "epoch": 268.0,
-      "eval_accuracy": 0.8268,
-      "eval_loss": 0.5731987357139587,
-      "eval_runtime": 5.6658,
-      "eval_samples_per_second": 1764.991,
-      "eval_steps_per_second": 7.06,
+      "eval_accuracy": 0.8304,
+      "eval_loss": 0.5731929540634155,
+      "eval_runtime": 5.2339,
+      "eval_samples_per_second": 1910.63,
+      "eval_steps_per_second": 7.643,
       "step": 89244
     },
     {
       "epoch": 268.02,
-      "grad_norm": 6.596947193145752,
+      "grad_norm": 11.170550346374512,
       "learning_rate": 1.0660660660660662e-06,
-      "loss": 1.1077,
+      "loss": 1.0621,
       "step": 89250
     },
     {
       "epoch": 268.05,
-      "grad_norm": 8.028807640075684,
+      "grad_norm": 7.036696910858154,
       "learning_rate": 1.065065065065065e-06,
-      "loss": 0.9315,
+      "loss": 0.9367,
       "step": 89260
     },
     {
       "epoch": 268.08,
-      "grad_norm": 6.0277099609375,
+      "grad_norm": 5.66010856628418,
       "learning_rate": 1.064064064064064e-06,
-      "loss": 0.8668,
+      "loss": 0.8674,
       "step": 89270
     },
     {
       "epoch": 268.11,
-      "grad_norm": 5.408481121063232,
+      "grad_norm": 5.423344612121582,
       "learning_rate": 1.063063063063063e-06,
-      "loss": 0.9362,
+      "loss": 0.9365,
       "step": 89280
     },
     {
       "epoch": 268.14,
-      "grad_norm": 6.680815696716309,
+      "grad_norm": 8.000452041625977,
       "learning_rate": 1.062062062062062e-06,
-      "loss": 0.9725,
+      "loss": 0.9759,
       "step": 89290
     },
     {
       "epoch": 268.17,
-      "grad_norm": 7.501251220703125,
+      "grad_norm": 7.9209184646606445,
       "learning_rate": 1.0610610610610611e-06,
-      "loss": 0.9161,
+      "loss": 0.9075,
       "step": 89300
     },
     {
       "epoch": 268.2,
-      "grad_norm": 10.448281288146973,
+      "grad_norm": 10.221400260925293,
       "learning_rate": 1.0600600600600601e-06,
-      "loss": 0.9579,
+      "loss": 0.9509,
       "step": 89310
     },
     {
       "epoch": 268.23,
-      "grad_norm": 4.210631847381592,
+      "grad_norm": 4.472344875335693,
       "learning_rate": 1.0590590590590591e-06,
-      "loss": 0.9197,
+      "loss": 0.9237,
       "step": 89320
     },
     {
       "epoch": 268.26,
-      "grad_norm": 5.769217014312744,
+      "grad_norm": 5.757817268371582,
       "learning_rate": 1.0580580580580582e-06,
-      "loss": 0.9342,
+      "loss": 0.9286,
       "step": 89330
     },
     {
       "epoch": 268.29,
-      "grad_norm": 5.318361282348633,
+      "grad_norm": 7.0810394287109375,
       "learning_rate": 1.0570570570570572e-06,
-      "loss": 0.9014,
+      "loss": 0.8997,
       "step": 89340
     },
     {
       "epoch": 268.32,
-      "grad_norm": 6.711041450500488,
+      "grad_norm": 6.851109981536865,
       "learning_rate": 1.0560560560560562e-06,
-      "loss": 0.905,
+      "loss": 0.9056,
       "step": 89350
     },
     {
       "epoch": 268.35,
-      "grad_norm": 6.427831649780273,
+      "grad_norm": 7.510689735412598,
       "learning_rate": 1.0550550550550552e-06,
-      "loss": 0.8903,
+      "loss": 0.8781,
       "step": 89360
     },
     {
       "epoch": 268.38,
-      "grad_norm": 3.842618703842163,
+      "grad_norm": 3.9822592735290527,
       "learning_rate": 1.0540540540540542e-06,
-      "loss": 0.9289,
+      "loss": 0.917,
       "step": 89370
     },
     {
       "epoch": 268.41,
-      "grad_norm": 8.297942161560059,
+      "grad_norm": 9.11953067779541,
       "learning_rate": 1.0530530530530533e-06,
-      "loss": 0.9264,
+      "loss": 0.9253,
       "step": 89380
     },
     {
       "epoch": 268.44,
-      "grad_norm": 6.650976181030273,
+      "grad_norm": 7.0758233070373535,
       "learning_rate": 1.0520520520520523e-06,
-      "loss": 0.9609,
+      "loss": 0.9596,
       "step": 89390
     },
     {
       "epoch": 268.47,
-      "grad_norm": 6.079313278198242,
+      "grad_norm": 6.141099452972412,
       "learning_rate": 1.051051051051051e-06,
-      "loss": 0.8398,
+      "loss": 0.8344,
       "step": 89400
     },
     {
       "epoch": 268.5,
-      "grad_norm": 7.978201389312744,
+      "grad_norm": 8.381059646606445,
       "learning_rate": 1.05005005005005e-06,
-      "loss": 0.8659,
+      "loss": 0.8704,
       "step": 89410
     },
     {
       "epoch": 268.53,
-      "grad_norm": 6.7075724601745605,
+      "grad_norm": 6.950074672698975,
       "learning_rate": 1.0490490490490491e-06,
-      "loss": 0.9451,
+      "loss": 0.9495,
       "step": 89420
     },
     {
       "epoch": 268.56,
-      "grad_norm": 5.3412251472473145,
+      "grad_norm": 10.150196075439453,
       "learning_rate": 1.0480480480480481e-06,
-      "loss": 0.8952,
+      "loss": 0.8767,
       "step": 89430
     },
     {
       "epoch": 268.59,
-      "grad_norm": 4.642005920410156,
+      "grad_norm": 4.718700408935547,
       "learning_rate": 1.0470470470470472e-06,
-      "loss": 0.918,
+      "loss": 0.8995,
       "step": 89440
     },
     {
       "epoch": 268.62,
-      "grad_norm": 7.051477432250977,
+      "grad_norm": 8.827250480651855,
       "learning_rate": 1.0460460460460462e-06,
-      "loss": 0.9633,
+      "loss": 0.9684,
       "step": 89450
     },
     {
       "epoch": 268.65,
-      "grad_norm": 5.847287178039551,
+      "grad_norm": 7.035213470458984,
       "learning_rate": 1.0450450450450452e-06,
-      "loss": 0.962,
+      "loss": 0.9599,
       "step": 89460
     },
     {
       "epoch": 268.68,
-      "grad_norm": 7.819766998291016,
+      "grad_norm": 6.1768927574157715,
       "learning_rate": 1.0440440440440442e-06,
-      "loss": 0.9262,
+      "loss": 0.9268,
       "step": 89470
     },
     {
       "epoch": 268.71,
-      "grad_norm": 6.215864181518555,
+      "grad_norm": 5.230839729309082,
       "learning_rate": 1.0430430430430432e-06,
-      "loss": 0.9201,
+      "loss": 0.9122,
       "step": 89480
     },
     {
       "epoch": 268.74,
-      "grad_norm": 7.638839244842529,
+      "grad_norm": 7.411840915679932,
       "learning_rate": 1.042042042042042e-06,
-      "loss": 0.8891,
+      "loss": 0.8761,
       "step": 89490
     },
     {
       "epoch": 268.77,
-      "grad_norm": 7.270205497741699,
+      "grad_norm": 6.465211868286133,
       "learning_rate": 1.041041041041041e-06,
-      "loss": 0.8906,
+      "loss": 0.8913,
       "step": 89500
     },
     {
       "epoch": 268.8,
-      "grad_norm": 10.421573638916016,
+      "grad_norm": 10.050432205200195,
       "learning_rate": 1.04004004004004e-06,
-      "loss": 0.8801,
+      "loss": 0.8837,
       "step": 89510
     },
     {
       "epoch": 268.83,
-      "grad_norm": 5.567302703857422,
+      "grad_norm": 7.91253137588501,
       "learning_rate": 1.039039039039039e-06,
-      "loss": 0.8996,
+      "loss": 0.8929,
       "step": 89520
     },
     {
       "epoch": 268.86,
-      "grad_norm": 5.994752407073975,
+      "grad_norm": 6.489618301391602,
       "learning_rate": 1.038038038038038e-06,
-      "loss": 0.9428,
+      "loss": 0.9322,
       "step": 89530
     },
     {
       "epoch": 268.89,
-      "grad_norm": 9.31883430480957,
+      "grad_norm": 9.616990089416504,
       "learning_rate": 1.0370370370370371e-06,
-      "loss": 0.8674,
+      "loss": 0.8619,
       "step": 89540
     },
     {
       "epoch": 268.92,
-      "grad_norm": 6.696113586425781,
+      "grad_norm": 6.915709018707275,
       "learning_rate": 1.0360360360360361e-06,
-      "loss": 0.875,
+      "loss": 0.8811,
       "step": 89550
     },
     {
       "epoch": 268.95,
-      "grad_norm": 6.253473281860352,
+      "grad_norm": 5.892404079437256,
       "learning_rate": 1.0350350350350352e-06,
-      "loss": 0.8901,
+      "loss": 0.8936,
       "step": 89560
     },
     {
       "epoch": 268.98,
-      "grad_norm": 6.782052993774414,
+      "grad_norm": 8.174165725708008,
       "learning_rate": 1.0340340340340342e-06,
-      "loss": 0.8829,
+      "loss": 0.8886,
       "step": 89570
     },
     {
       "epoch": 269.0,
-      "eval_accuracy": 0.827,
-      "eval_loss": 0.5749877095222473,
-      "eval_runtime": 5.4699,
-      "eval_samples_per_second": 1828.199,
-      "eval_steps_per_second": 7.313,
+      "eval_accuracy": 0.8281,
+      "eval_loss": 0.5734755396842957,
+      "eval_runtime": 5.265,
+      "eval_samples_per_second": 1899.329,
+      "eval_steps_per_second": 7.597,
       "step": 89577
     },
     {
       "epoch": 269.01,
-      "grad_norm": 6.708944797515869,
+      "grad_norm": 6.279465675354004,
       "learning_rate": 1.033033033033033e-06,
-      "loss": 1.0802,
+      "loss": 1.114,
       "step": 89580
     },
     {
       "epoch": 269.04,
-      "grad_norm": 9.407207489013672,
+      "grad_norm": 10.253714561462402,
       "learning_rate": 1.032032032032032e-06,
-      "loss": 0.9289,
+      "loss": 0.935,
       "step": 89590
     },
     {
       "epoch": 269.07,
-      "grad_norm": 11.667288780212402,
+      "grad_norm": 14.890660285949707,
       "learning_rate": 1.031031031031031e-06,
-      "loss": 0.9386,
+      "loss": 0.9395,
       "step": 89600
     },
     {
       "epoch": 269.1,
-      "grad_norm": 5.944511890411377,
+      "grad_norm": 8.68957805633545,
       "learning_rate": 1.03003003003003e-06,
-      "loss": 0.9624,
+      "loss": 0.9528,
       "step": 89610
     },
     {
       "epoch": 269.13,
-      "grad_norm": 5.485231399536133,
+      "grad_norm": 8.499131202697754,
       "learning_rate": 1.0290290290290293e-06,
-      "loss": 0.9105,
+      "loss": 0.9072,
       "step": 89620
     },
     {
       "epoch": 269.16,
-      "grad_norm": 6.472165584564209,
+      "grad_norm": 6.443289279937744,
       "learning_rate": 1.028028028028028e-06,
-      "loss": 0.8986,
+      "loss": 0.896,
       "step": 89630
     },
     {
       "epoch": 269.19,
-      "grad_norm": 7.321653842926025,
+      "grad_norm": 10.259716033935547,
       "learning_rate": 1.027027027027027e-06,
-      "loss": 0.9117,
+      "loss": 0.9143,
       "step": 89640
     },
     {
       "epoch": 269.22,
-      "grad_norm": 5.4554266929626465,
+      "grad_norm": 4.984058856964111,
       "learning_rate": 1.0260260260260261e-06,
-      "loss": 0.9239,
+      "loss": 0.9231,
       "step": 89650
     },
     {
       "epoch": 269.25,
-      "grad_norm": 6.436707019805908,
+      "grad_norm": 8.491416931152344,
       "learning_rate": 1.0250250250250251e-06,
-      "loss": 0.9271,
+      "loss": 0.9422,
       "step": 89660
     },
     {
       "epoch": 269.28,
-      "grad_norm": 6.371859550476074,
+      "grad_norm": 8.194557189941406,
       "learning_rate": 1.0240240240240242e-06,
-      "loss": 0.9223,
+      "loss": 0.9221,
       "step": 89670
     },
     {
       "epoch": 269.31,
-      "grad_norm": 6.713272571563721,
+      "grad_norm": 6.978273868560791,
       "learning_rate": 1.0230230230230232e-06,
-      "loss": 0.8882,
+      "loss": 0.8872,
       "step": 89680
     },
     {
       "epoch": 269.34,
-      "grad_norm": 4.803586483001709,
+      "grad_norm": 5.597606182098389,
       "learning_rate": 1.0220220220220222e-06,
-      "loss": 0.9402,
+      "loss": 0.943,
       "step": 89690
     },
     {
       "epoch": 269.37,
-      "grad_norm": 5.225164413452148,
+      "grad_norm": 7.434798717498779,
       "learning_rate": 1.0210210210210212e-06,
-      "loss": 0.9547,
+      "loss": 0.9527,
       "step": 89700
     },
     {
       "epoch": 269.4,
-      "grad_norm": 5.968810558319092,
+      "grad_norm": 5.431873321533203,
       "learning_rate": 1.0200200200200202e-06,
-      "loss": 0.7973,
+      "loss": 0.7994,
       "step": 89710
     },
     {
       "epoch": 269.43,
-      "grad_norm": 6.963784217834473,
+      "grad_norm": 6.9171576499938965,
       "learning_rate": 1.019019019019019e-06,
-      "loss": 0.9563,
+      "loss": 0.9531,
       "step": 89720
     },
     {
       "epoch": 269.46,
-      "grad_norm": 5.316221714019775,
+      "grad_norm": 7.277230739593506,
       "learning_rate": 1.018018018018018e-06,
-      "loss": 0.8887,
+      "loss": 0.8879,
       "step": 89730
     },
     {
       "epoch": 269.49,
-      "grad_norm": 7.06106424331665,
+      "grad_norm": 6.709704875946045,
       "learning_rate": 1.017017017017017e-06,
-      "loss": 0.9309,
+      "loss": 0.9208,
       "step": 89740
     },
     {
       "epoch": 269.52,
-      "grad_norm": 7.265474796295166,
+      "grad_norm": 6.93162727355957,
       "learning_rate": 1.016016016016016e-06,
-      "loss": 0.8998,
+      "loss": 0.8969,
       "step": 89750
     },
     {
       "epoch": 269.55,
-      "grad_norm": 6.428163528442383,
+      "grad_norm": 7.006763458251953,
       "learning_rate": 1.015015015015015e-06,
-      "loss": 0.956,
+      "loss": 0.957,
       "step": 89760
     },
     {
       "epoch": 269.58,
-      "grad_norm": 4.637615203857422,
+      "grad_norm": 5.0267205238342285,
       "learning_rate": 1.0140140140140141e-06,
-      "loss": 0.8121,
+      "loss": 0.8113,
       "step": 89770
     },
     {
       "epoch": 269.61,
-      "grad_norm": 6.649044990539551,
+      "grad_norm": 5.431989669799805,
       "learning_rate": 1.0130130130130131e-06,
-      "loss": 0.8344,
+      "loss": 0.8474,
       "step": 89780
     },
     {
       "epoch": 269.64,
-      "grad_norm": 5.7938232421875,
+      "grad_norm": 5.823074817657471,
       "learning_rate": 1.0120120120120122e-06,
-      "loss": 0.9169,
+      "loss": 0.9117,
       "step": 89790
     },
     {
       "epoch": 269.67,
-      "grad_norm": 12.082904815673828,
+      "grad_norm": 10.780046463012695,
       "learning_rate": 1.011011011011011e-06,
-      "loss": 0.9359,
+      "loss": 0.933,
       "step": 89800
     },
     {
       "epoch": 269.7,
-      "grad_norm": 8.75696086883545,
+      "grad_norm": 9.852265357971191,
       "learning_rate": 1.01001001001001e-06,
-      "loss": 0.8656,
+      "loss": 0.8506,
       "step": 89810
     },
     {
       "epoch": 269.73,
-      "grad_norm": 6.793710231781006,
+      "grad_norm": 10.426013946533203,
       "learning_rate": 1.009009009009009e-06,
-      "loss": 0.933,
+      "loss": 0.9398,
       "step": 89820
     },
     {
       "epoch": 269.76,
-      "grad_norm": 7.5803070068359375,
+      "grad_norm": 7.8563337326049805,
       "learning_rate": 1.008008008008008e-06,
-      "loss": 0.861,
+      "loss": 0.8576,
       "step": 89830
     },
     {
       "epoch": 269.79,
-      "grad_norm": 6.326283931732178,
+      "grad_norm": 7.682973861694336,
       "learning_rate": 1.007007007007007e-06,
-      "loss": 0.9022,
+      "loss": 0.918,
       "step": 89840
     },
     {
       "epoch": 269.82,
-      "grad_norm": 6.912482261657715,
+      "grad_norm": 9.109933853149414,
       "learning_rate": 1.006006006006006e-06,
-      "loss": 0.9135,
+      "loss": 0.9146,
       "step": 89850
     },
     {
       "epoch": 269.85,
-      "grad_norm": 4.616462707519531,
+      "grad_norm": 5.560709476470947,
       "learning_rate": 1.005005005005005e-06,
-      "loss": 0.8685,
+      "loss": 0.8684,
       "step": 89860
     },
     {
       "epoch": 269.88,
-      "grad_norm": 7.78359842300415,
+      "grad_norm": 5.380096912384033,
       "learning_rate": 1.004004004004004e-06,
-      "loss": 0.9904,
+      "loss": 0.9805,
       "step": 89870
     },
     {
       "epoch": 269.91,
-      "grad_norm": 8.085410118103027,
+      "grad_norm": 9.553788185119629,
       "learning_rate": 1.0030030030030031e-06,
-      "loss": 0.9582,
+      "loss": 0.9559,
       "step": 89880
     },
     {
       "epoch": 269.94,
-      "grad_norm": 7.957455635070801,
+      "grad_norm": 10.975713729858398,
       "learning_rate": 1.0020020020020021e-06,
-      "loss": 0.9388,
+      "loss": 0.9451,
       "step": 89890
     },
     {
       "epoch": 269.97,
-      "grad_norm": 5.201528072357178,
+      "grad_norm": 6.334826946258545,
       "learning_rate": 1.0010010010010011e-06,
-      "loss": 0.9111,
+      "loss": 0.9161,
       "step": 89900
     },
     {
       "epoch": 270.0,
-      "grad_norm": 43.30949401855469,
+      "grad_norm": 70.54277038574219,
       "learning_rate": 1.0000000000000002e-06,
-      "loss": 1.1097,
+      "loss": 1.1046,
       "step": 89910
     },
     {
       "epoch": 270.0,
-      "eval_accuracy": 0.8277,
-      "eval_loss": 0.5733190178871155,
-      "eval_runtime": 5.2407,
-      "eval_samples_per_second": 1908.129,
-      "eval_steps_per_second": 7.633,
+      "eval_accuracy": 0.8299,
+      "eval_loss": 0.5709655284881592,
+      "eval_runtime": 5.2087,
+      "eval_samples_per_second": 1919.864,
+      "eval_steps_per_second": 7.679,
       "step": 89910
     },
     {
       "epoch": 270.03,
-      "grad_norm": 7.134408473968506,
+      "grad_norm": 6.000466823577881,
       "learning_rate": 9.989989989989992e-07,
-      "loss": 0.8682,
+      "loss": 0.861,
       "step": 89920
     },
     {
       "epoch": 270.06,
-      "grad_norm": 8.553323745727539,
+      "grad_norm": 9.260699272155762,
       "learning_rate": 9.979979979979982e-07,
-      "loss": 0.8599,
+      "loss": 0.8639,
       "step": 89930
     },
     {
       "epoch": 270.09,
-      "grad_norm": 6.073080062866211,
+      "grad_norm": 8.557622909545898,
       "learning_rate": 9.969969969969972e-07,
-      "loss": 0.9436,
+      "loss": 0.9473,
       "step": 89940
     },
     {
       "epoch": 270.12,
-      "grad_norm": 6.151409149169922,
+      "grad_norm": 7.800375461578369,
       "learning_rate": 9.95995995995996e-07,
-      "loss": 0.9068,
+      "loss": 0.914,
       "step": 89950
     },
     {
       "epoch": 270.15,
-      "grad_norm": 5.631007671356201,
+      "grad_norm": 5.28271484375,
       "learning_rate": 9.94994994994995e-07,
-      "loss": 0.9158,
+      "loss": 0.906,
       "step": 89960
     },
     {
       "epoch": 270.18,
-      "grad_norm": 12.0625,
+      "grad_norm": 9.937504768371582,
       "learning_rate": 9.93993993993994e-07,
-      "loss": 0.9192,
+      "loss": 0.9203,
       "step": 89970
     },
     {
       "epoch": 270.21,
-      "grad_norm": 5.82733154296875,
+      "grad_norm": 6.369692325592041,
       "learning_rate": 9.92992992992993e-07,
-      "loss": 0.9421,
+      "loss": 0.9341,
       "step": 89980
     },
     {
       "epoch": 270.24,
-      "grad_norm": 6.189974308013916,
+      "grad_norm": 6.01729154586792,
       "learning_rate": 9.91991991991992e-07,
-      "loss": 0.8792,
+      "loss": 0.8771,
       "step": 89990
     },
     {
       "epoch": 270.27,
-      "grad_norm": 5.896833896636963,
+      "grad_norm": 5.822775363922119,
       "learning_rate": 9.909909909909911e-07,
-      "loss": 0.913,
+      "loss": 0.9145,
       "step": 90000
     },
     {
       "epoch": 270.3,
-      "grad_norm": 8.48974609375,
+      "grad_norm": 7.539471626281738,
       "learning_rate": 9.899899899899901e-07,
-      "loss": 0.8988,
+      "loss": 0.8915,
       "step": 90010
     },
     {
       "epoch": 270.33,
-      "grad_norm": 5.549361228942871,
+      "grad_norm": 6.818711757659912,
       "learning_rate": 9.889889889889892e-07,
-      "loss": 0.9228,
+      "loss": 0.9304,
       "step": 90020
     },
     {
       "epoch": 270.36,
-      "grad_norm": 8.595791816711426,
+      "grad_norm": 8.611358642578125,
       "learning_rate": 9.87987987987988e-07,
-      "loss": 0.9026,
+      "loss": 0.8957,
       "step": 90030
     },
     {
       "epoch": 270.39,
-      "grad_norm": 9.409880638122559,
+      "grad_norm": 7.352656364440918,
       "learning_rate": 9.86986986986987e-07,
-      "loss": 0.8778,
+      "loss": 0.8688,
       "step": 90040
     },
     {
       "epoch": 270.42,
-      "grad_norm": 9.45514965057373,
+      "grad_norm": 7.651889801025391,
       "learning_rate": 9.85985985985986e-07,
-      "loss": 0.9696,
+      "loss": 0.9578,
       "step": 90050
     },
     {
       "epoch": 270.45,
-      "grad_norm": 5.824161529541016,
+      "grad_norm": 5.271396636962891,
       "learning_rate": 9.84984984984985e-07,
-      "loss": 0.929,
+      "loss": 0.9189,
       "step": 90060
     },
     {
       "epoch": 270.48,
-      "grad_norm": 7.313093185424805,
+      "grad_norm": 7.694922924041748,
       "learning_rate": 9.83983983983984e-07,
-      "loss": 0.8932,
+      "loss": 0.8945,
       "step": 90070
     },
     {
       "epoch": 270.51,
-      "grad_norm": 7.234867095947266,
+      "grad_norm": 9.339532852172852,
       "learning_rate": 9.82982982982983e-07,
-      "loss": 0.8882,
+      "loss": 0.8974,
       "step": 90080
     },
     {
       "epoch": 270.54,
-      "grad_norm": 6.309026718139648,
+      "grad_norm": 6.770332336425781,
       "learning_rate": 9.81981981981982e-07,
-      "loss": 0.9139,
+      "loss": 0.9072,
       "step": 90090
     },
     {
       "epoch": 270.57,
-      "grad_norm": 5.696401596069336,
+      "grad_norm": 6.766361713409424,
       "learning_rate": 9.80980980980981e-07,
-      "loss": 0.9206,
+      "loss": 0.9103,
       "step": 90100
     },
     {
       "epoch": 270.6,
-      "grad_norm": 6.764082908630371,
+      "grad_norm": 9.416958808898926,
       "learning_rate": 9.799799799799801e-07,
-      "loss": 0.9567,
+      "loss": 0.9609,
       "step": 90110
     },
     {
       "epoch": 270.63,
-      "grad_norm": 7.966604232788086,
+      "grad_norm": 10.677770614624023,
       "learning_rate": 9.78978978978979e-07,
-      "loss": 0.9343,
+      "loss": 0.9322,
       "step": 90120
     },
     {
       "epoch": 270.66,
-      "grad_norm": 5.2955098152160645,
+      "grad_norm": 4.8728227615356445,
       "learning_rate": 9.77977977977978e-07,
-      "loss": 0.9037,
+      "loss": 0.8969,
       "step": 90130
     },
     {
       "epoch": 270.69,
-      "grad_norm": 5.6447272300720215,
+      "grad_norm": 7.692988872528076,
       "learning_rate": 9.76976976976977e-07,
-      "loss": 0.8629,
+      "loss": 0.8659,
       "step": 90140
     },
     {
       "epoch": 270.72,
-      "grad_norm": 5.487011432647705,
+      "grad_norm": 5.804048538208008,
       "learning_rate": 9.75975975975976e-07,
-      "loss": 0.9619,
+      "loss": 0.9641,
       "step": 90150
     },
     {
       "epoch": 270.75,
-      "grad_norm": 7.016127586364746,
+      "grad_norm": 7.070906162261963,
       "learning_rate": 9.749749749749752e-07,
-      "loss": 0.9039,
+      "loss": 0.9032,
       "step": 90160
     },
     {
       "epoch": 270.78,
-      "grad_norm": 6.304392337799072,
+      "grad_norm": 8.771299362182617,
       "learning_rate": 9.73973973973974e-07,
-      "loss": 0.8695,
+      "loss": 0.8703,
       "step": 90170
     },
     {
       "epoch": 270.81,
-      "grad_norm": 8.180070877075195,
+      "grad_norm": 9.133467674255371,
       "learning_rate": 9.72972972972973e-07,
-      "loss": 0.882,
+      "loss": 0.8956,
       "step": 90180
     },
     {
       "epoch": 270.84,
-      "grad_norm": 6.075888633728027,
+      "grad_norm": 6.785087585449219,
       "learning_rate": 9.71971971971972e-07,
-      "loss": 0.8507,
+      "loss": 0.8508,
       "step": 90190
     },
     {
       "epoch": 270.87,
-      "grad_norm": 7.298079490661621,
+      "grad_norm": 9.17315673828125,
       "learning_rate": 9.70970970970971e-07,
-      "loss": 0.9133,
+      "loss": 0.9045,
       "step": 90200
     },
     {
       "epoch": 270.9,
-      "grad_norm": 5.171963691711426,
+      "grad_norm": 5.3233642578125,
       "learning_rate": 9.6996996996997e-07,
-      "loss": 0.8669,
+      "loss": 0.8689,
       "step": 90210
     },
     {
       "epoch": 270.93,
-      "grad_norm": 6.367233753204346,
+      "grad_norm": 5.530008792877197,
       "learning_rate": 9.68968968968969e-07,
-      "loss": 0.9473,
+      "loss": 0.941,
       "step": 90220
     },
     {
       "epoch": 270.96,
-      "grad_norm": 5.635465145111084,
+      "grad_norm": 6.818763732910156,
       "learning_rate": 9.679679679679681e-07,
-      "loss": 0.8923,
+      "loss": 0.8875,
       "step": 90230
     },
     {
       "epoch": 270.99,
-      "grad_norm": 4.777095794677734,
+      "grad_norm": 6.519830703735352,
       "learning_rate": 9.669669669669671e-07,
-      "loss": 0.9004,
+      "loss": 0.8992,
       "step": 90240
     },
     {
       "epoch": 271.0,
-      "eval_accuracy": 0.8291,
-      "eval_loss": 0.5687561631202698,
-      "eval_runtime": 5.3644,
-      "eval_samples_per_second": 1864.133,
-      "eval_steps_per_second": 7.457,
+      "eval_accuracy": 0.8293,
+      "eval_loss": 0.5697559118270874,
+      "eval_runtime": 5.4985,
+      "eval_samples_per_second": 1818.69,
+      "eval_steps_per_second": 7.275,
       "step": 90243
     },
     {
       "epoch": 271.02,
-      "grad_norm": 6.397618770599365,
+      "grad_norm": 7.1113600730896,
       "learning_rate": 9.659659659659662e-07,
-      "loss": 0.8774,
+      "loss": 0.8706,
       "step": 90250
     },
     {
       "epoch": 271.05,
-      "grad_norm": 6.749684810638428,
+      "grad_norm": 7.7295918464660645,
       "learning_rate": 9.64964964964965e-07,
-      "loss": 0.852,
+      "loss": 0.8398,
       "step": 90260
     },
     {
       "epoch": 271.08,
-      "grad_norm": 10.856700897216797,
+      "grad_norm": 13.083982467651367,
       "learning_rate": 9.63963963963964e-07,
-      "loss": 0.9476,
+      "loss": 0.9357,
       "step": 90270
     },
     {
       "epoch": 271.11,
-      "grad_norm": 5.873683929443359,
+      "grad_norm": 7.057460308074951,
       "learning_rate": 9.62962962962963e-07,
-      "loss": 0.891,
+      "loss": 0.8891,
       "step": 90280
     },
     {
       "epoch": 271.14,
-      "grad_norm": 7.817655086517334,
+      "grad_norm": 10.642827987670898,
       "learning_rate": 9.61961961961962e-07,
-      "loss": 0.8569,
+      "loss": 0.8496,
       "step": 90290
     },
     {
       "epoch": 271.17,
-      "grad_norm": 6.965671062469482,
+      "grad_norm": 8.316075325012207,
       "learning_rate": 9.60960960960961e-07,
-      "loss": 0.9298,
+      "loss": 0.9377,
       "step": 90300
     },
     {
       "epoch": 271.2,
-      "grad_norm": 5.643367290496826,
+      "grad_norm": 5.754316806793213,
       "learning_rate": 9.5995995995996e-07,
-      "loss": 0.8793,
+      "loss": 0.8843,
       "step": 90310
     },
     {
       "epoch": 271.23,
-      "grad_norm": 8.553305625915527,
+      "grad_norm": 9.658960342407227,
       "learning_rate": 9.58958958958959e-07,
-      "loss": 0.9304,
+      "loss": 0.9384,
       "step": 90320
     },
     {
       "epoch": 271.26,
-      "grad_norm": 5.34323787689209,
+      "grad_norm": 5.357593059539795,
       "learning_rate": 9.57957957957958e-07,
-      "loss": 0.9376,
+      "loss": 0.9431,
       "step": 90330
     },
     {
       "epoch": 271.29,
-      "grad_norm": 9.47058391571045,
+      "grad_norm": 10.89166259765625,
       "learning_rate": 9.569569569569571e-07,
-      "loss": 0.9638,
+      "loss": 0.9662,
       "step": 90340
     },
     {
       "epoch": 271.32,
-      "grad_norm": 7.233519554138184,
+      "grad_norm": 7.147401809692383,
       "learning_rate": 9.55955955955956e-07,
-      "loss": 0.9653,
+      "loss": 0.958,
       "step": 90350
     },
     {
       "epoch": 271.35,
-      "grad_norm": 7.283676624298096,
+      "grad_norm": 6.037397861480713,
       "learning_rate": 9.54954954954955e-07,
-      "loss": 0.933,
+      "loss": 0.9271,
       "step": 90360
     },
     {
       "epoch": 271.38,
-      "grad_norm": 6.529357433319092,
+      "grad_norm": 5.506972789764404,
       "learning_rate": 9.53953953953954e-07,
-      "loss": 0.8501,
+      "loss": 0.8512,
       "step": 90370
     },
     {
       "epoch": 271.41,
-      "grad_norm": 8.030519485473633,
+      "grad_norm": 8.16527271270752,
       "learning_rate": 9.52952952952953e-07,
-      "loss": 0.9954,
+      "loss": 0.9937,
       "step": 90380
     },
     {
       "epoch": 271.44,
-      "grad_norm": 4.6728997230529785,
+      "grad_norm": 5.618782043457031,
       "learning_rate": 9.51951951951952e-07,
-      "loss": 0.853,
+      "loss": 0.8513,
       "step": 90390
     },
     {
       "epoch": 271.47,
-      "grad_norm": 5.664414405822754,
+      "grad_norm": 6.2026801109313965,
       "learning_rate": 9.50950950950951e-07,
-      "loss": 0.8493,
+      "loss": 0.8446,
       "step": 90400
     },
     {
       "epoch": 271.5,
-      "grad_norm": 7.173210620880127,
+      "grad_norm": 7.680527210235596,
       "learning_rate": 9.499499499499499e-07,
-      "loss": 0.8411,
+      "loss": 0.8565,
       "step": 90410
     },
     {
       "epoch": 271.53,
-      "grad_norm": 7.784193992614746,
+      "grad_norm": 11.231497764587402,
       "learning_rate": 9.48948948948949e-07,
       "loss": 0.8524,
       "step": 90420
     },
     {
       "epoch": 271.56,
-      "grad_norm": 7.888918876647949,
+      "grad_norm": 6.6059064865112305,
       "learning_rate": 9.479479479479481e-07,
-      "loss": 0.9637,
+      "loss": 0.9668,
       "step": 90430
     },
     {
       "epoch": 271.59,
-      "grad_norm": 7.200760364532471,
+      "grad_norm": 7.656988143920898,
       "learning_rate": 9.469469469469471e-07,
-      "loss": 0.8423,
+      "loss": 0.8311,
       "step": 90440
     },
     {
       "epoch": 271.62,
-      "grad_norm": 8.929557800292969,
+      "grad_norm": 9.582619667053223,
       "learning_rate": 9.459459459459461e-07,
-      "loss": 0.9228,
+      "loss": 0.9254,
       "step": 90450
     },
     {
       "epoch": 271.65,
-      "grad_norm": 6.957674980163574,
+      "grad_norm": 6.7285661697387695,
       "learning_rate": 9.449449449449451e-07,
-      "loss": 0.9418,
+      "loss": 0.9421,
       "step": 90460
     },
     {
       "epoch": 271.68,
-      "grad_norm": 6.361674785614014,
+      "grad_norm": 7.078763008117676,
       "learning_rate": 9.43943943943944e-07,
-      "loss": 0.8884,
+      "loss": 0.8845,
       "step": 90470
     },
     {
       "epoch": 271.71,
-      "grad_norm": 4.1985602378845215,
+      "grad_norm": 5.1893486976623535,
       "learning_rate": 9.42942942942943e-07,
-      "loss": 0.7982,
+      "loss": 0.7991,
       "step": 90480
     },
     {
       "epoch": 271.74,
-      "grad_norm": 8.889386177062988,
+      "grad_norm": 9.769329071044922,
       "learning_rate": 9.419419419419421e-07,
-      "loss": 0.9553,
+      "loss": 0.9581,
       "step": 90490
     },
     {
       "epoch": 271.77,
-      "grad_norm": 6.517195224761963,
+      "grad_norm": 7.795509338378906,
       "learning_rate": 9.409409409409411e-07,
-      "loss": 0.8369,
+      "loss": 0.8378,
       "step": 90500
     },
     {
       "epoch": 271.8,
-      "grad_norm": 6.003081798553467,
+      "grad_norm": 6.634261131286621,
       "learning_rate": 9.3993993993994e-07,
-      "loss": 0.8855,
+      "loss": 0.8795,
       "step": 90510
     },
     {
       "epoch": 271.83,
-      "grad_norm": 7.35237455368042,
+      "grad_norm": 6.991076469421387,
       "learning_rate": 9.38938938938939e-07,
-      "loss": 0.9142,
+      "loss": 0.9128,
       "step": 90520
     },
     {
       "epoch": 271.86,
-      "grad_norm": 5.997629642486572,
+      "grad_norm": 7.934773921966553,
       "learning_rate": 9.37937937937938e-07,
-      "loss": 0.8862,
+      "loss": 0.8853,
       "step": 90530
     },
     {
       "epoch": 271.89,
-      "grad_norm": 5.267559051513672,
+      "grad_norm": 7.453751087188721,
       "learning_rate": 9.36936936936937e-07,
-      "loss": 0.9754,
+      "loss": 0.9819,
       "step": 90540
     },
     {
       "epoch": 271.92,
-      "grad_norm": 4.837818622589111,
+      "grad_norm": 5.817561626434326,
       "learning_rate": 9.35935935935936e-07,
-      "loss": 0.8984,
+      "loss": 0.9119,
       "step": 90550
     },
     {
       "epoch": 271.95,
-      "grad_norm": 6.265627384185791,
+      "grad_norm": 7.748689651489258,
       "learning_rate": 9.34934934934935e-07,
-      "loss": 0.877,
+      "loss": 0.8757,
       "step": 90560
     },
     {
       "epoch": 271.98,
-      "grad_norm": 6.418170928955078,
+      "grad_norm": 5.566091060638428,
       "learning_rate": 9.33933933933934e-07,
-      "loss": 0.8734,
+      "loss": 0.8777,
       "step": 90570
     },
     {
       "epoch": 272.0,
-      "eval_accuracy": 0.8275,
-      "eval_loss": 0.5707234144210815,
-      "eval_runtime": 5.3206,
-      "eval_samples_per_second": 1879.495,
-      "eval_steps_per_second": 7.518,
+      "eval_accuracy": 0.8321,
+      "eval_loss": 0.568728506565094,
+      "eval_runtime": 5.5708,
+      "eval_samples_per_second": 1795.061,
+      "eval_steps_per_second": 7.18,
       "step": 90576
     },
     {
       "epoch": 272.01,
-      "grad_norm": 4.629018306732178,
+      "grad_norm": 5.422853946685791,
       "learning_rate": 9.32932932932933e-07,
-      "loss": 0.961,
+      "loss": 0.9673,
       "step": 90580
     },
     {
       "epoch": 272.04,
-      "grad_norm": 7.307394027709961,
+      "grad_norm": 10.39578914642334,
       "learning_rate": 9.31931931931932e-07,
-      "loss": 0.8911,
+      "loss": 0.8878,
       "step": 90590
     },
     {
       "epoch": 272.07,
-      "grad_norm": 6.956962585449219,
+      "grad_norm": 8.140064239501953,
       "learning_rate": 9.30930930930931e-07,
-      "loss": 0.8959,
+      "loss": 0.896,
       "step": 90600
     },
     {
       "epoch": 272.1,
-      "grad_norm": 6.853639125823975,
+      "grad_norm": 9.493454933166504,
       "learning_rate": 9.2992992992993e-07,
-      "loss": 0.9417,
+      "loss": 0.9444,
       "step": 90610
     },
     {
       "epoch": 272.13,
-      "grad_norm": 5.958841800689697,
+      "grad_norm": 5.732733726501465,
       "learning_rate": 9.28928928928929e-07,
-      "loss": 0.9427,
+      "loss": 0.918,
       "step": 90620
     },
     {
       "epoch": 272.16,
-      "grad_norm": 6.708520889282227,
+      "grad_norm": 5.783646583557129,
       "learning_rate": 9.27927927927928e-07,
-      "loss": 0.8276,
+      "loss": 0.8244,
       "step": 90630
     },
     {
       "epoch": 272.19,
-      "grad_norm": 6.1837897300720215,
+      "grad_norm": 5.117488384246826,
       "learning_rate": 9.269269269269269e-07,
-      "loss": 0.9018,
+      "loss": 0.899,
       "step": 90640
     },
     {
       "epoch": 272.22,
-      "grad_norm": 7.452227592468262,
+      "grad_norm": 5.668084144592285,
       "learning_rate": 9.259259259259259e-07,
-      "loss": 0.8969,
+      "loss": 0.8947,
       "step": 90650
     },
     {
       "epoch": 272.25,
-      "grad_norm": 7.662498474121094,
+      "grad_norm": 10.369906425476074,
       "learning_rate": 9.24924924924925e-07,
-      "loss": 0.9011,
+      "loss": 0.9108,
       "step": 90660
     },
     {
       "epoch": 272.28,
-      "grad_norm": 6.97830867767334,
+      "grad_norm": 6.959672927856445,
       "learning_rate": 9.23923923923924e-07,
-      "loss": 0.8436,
+      "loss": 0.8481,
       "step": 90670
     },
     {
       "epoch": 272.31,
-      "grad_norm": 6.8944597244262695,
+      "grad_norm": 7.178253173828125,
       "learning_rate": 9.229229229229229e-07,
-      "loss": 0.9022,
+      "loss": 0.9052,
       "step": 90680
     },
     {
       "epoch": 272.34,
-      "grad_norm": 5.813831806182861,
+      "grad_norm": 5.471269607543945,
       "learning_rate": 9.219219219219221e-07,
-      "loss": 0.9158,
+      "loss": 0.9213,
       "step": 90690
     },
     {
       "epoch": 272.37,
-      "grad_norm": 6.449962139129639,
+      "grad_norm": 6.361780166625977,
       "learning_rate": 9.20920920920921e-07,
-      "loss": 0.8398,
+      "loss": 0.8409,
       "step": 90700
     },
     {
       "epoch": 272.4,
-      "grad_norm": 9.756831169128418,
+      "grad_norm": 7.302761077880859,
       "learning_rate": 9.1991991991992e-07,
-      "loss": 0.9627,
+      "loss": 0.9527,
       "step": 90710
     },
     {
       "epoch": 272.43,
-      "grad_norm": 4.89442777633667,
+      "grad_norm": 7.235621929168701,
       "learning_rate": 9.189189189189191e-07,
-      "loss": 0.9364,
+      "loss": 0.9316,
       "step": 90720
     },
     {
       "epoch": 272.46,
-      "grad_norm": 5.669458389282227,
+      "grad_norm": 5.479763031005859,
       "learning_rate": 9.179179179179181e-07,
-      "loss": 0.9739,
+      "loss": 0.9801,
       "step": 90730
     },
     {
       "epoch": 272.49,
-      "grad_norm": 7.076022624969482,
+      "grad_norm": 10.261174201965332,
       "learning_rate": 9.16916916916917e-07,
-      "loss": 0.9147,
+      "loss": 0.9093,
       "step": 90740
     },
     {
       "epoch": 272.52,
-      "grad_norm": 5.0733208656311035,
+      "grad_norm": 6.660607814788818,
       "learning_rate": 9.15915915915916e-07,
-      "loss": 0.8439,
+      "loss": 0.843,
       "step": 90750
     },
     {
       "epoch": 272.55,
-      "grad_norm": 6.091789245605469,
+      "grad_norm": 8.173662185668945,
       "learning_rate": 9.14914914914915e-07,
-      "loss": 0.8901,
+      "loss": 0.8878,
       "step": 90760
     },
     {
       "epoch": 272.58,
-      "grad_norm": 10.564282417297363,
+      "grad_norm": 11.688820838928223,
       "learning_rate": 9.13913913913914e-07,
-      "loss": 0.9154,
+      "loss": 0.92,
       "step": 90770
     },
     {
       "epoch": 272.61,
-      "grad_norm": 8.715259552001953,
+      "grad_norm": 7.818812847137451,
       "learning_rate": 9.12912912912913e-07,
-      "loss": 0.8971,
+      "loss": 0.8945,
       "step": 90780
     },
     {
       "epoch": 272.64,
-      "grad_norm": 5.601443290710449,
+      "grad_norm": 7.689454555511475,
       "learning_rate": 9.11911911911912e-07,
-      "loss": 1.051,
+      "loss": 1.053,
       "step": 90790
     },
     {
       "epoch": 272.67,
-      "grad_norm": 8.623851776123047,
+      "grad_norm": 5.910655975341797,
       "learning_rate": 9.10910910910911e-07,
-      "loss": 0.8974,
+      "loss": 0.897,
       "step": 90800
     },
     {
       "epoch": 272.7,
-      "grad_norm": 5.754073143005371,
+      "grad_norm": 5.026261329650879,
       "learning_rate": 9.0990990990991e-07,
-      "loss": 0.9011,
+      "loss": 0.9047,
       "step": 90810
     },
     {
       "epoch": 272.73,
-      "grad_norm": 10.301849365234375,
+      "grad_norm": 6.420774459838867,
       "learning_rate": 9.08908908908909e-07,
-      "loss": 0.8883,
+      "loss": 0.8842,
       "step": 90820
     },
     {
       "epoch": 272.76,
-      "grad_norm": 9.982394218444824,
+      "grad_norm": 10.997797966003418,
       "learning_rate": 9.07907907907908e-07,
-      "loss": 0.9703,
+      "loss": 0.9625,
       "step": 90830
     },
     {
       "epoch": 272.79,
-      "grad_norm": 5.63289213180542,
+      "grad_norm": 5.273614406585693,
       "learning_rate": 9.06906906906907e-07,
-      "loss": 0.9956,
+      "loss": 1.0019,
       "step": 90840
     },
     {
       "epoch": 272.82,
-      "grad_norm": 7.7048139572143555,
+      "grad_norm": 8.257291793823242,
       "learning_rate": 9.05905905905906e-07,
-      "loss": 0.9626,
+      "loss": 0.9687,
       "step": 90850
     },
     {
       "epoch": 272.85,
-      "grad_norm": 5.313968181610107,
+      "grad_norm": 6.585517883300781,
       "learning_rate": 9.04904904904905e-07,
-      "loss": 0.8614,
+      "loss": 0.8502,
       "step": 90860
     },
     {
       "epoch": 272.88,
-      "grad_norm": 3.9588465690612793,
+      "grad_norm": 4.211805820465088,
       "learning_rate": 9.039039039039039e-07,
-      "loss": 0.9418,
+      "loss": 0.9442,
       "step": 90870
     },
     {
       "epoch": 272.91,
-      "grad_norm": 7.048387050628662,
+      "grad_norm": 8.180560111999512,
       "learning_rate": 9.029029029029029e-07,
-      "loss": 0.9465,
+      "loss": 0.954,
       "step": 90880
     },
     {
       "epoch": 272.94,
-      "grad_norm": 5.813858985900879,
+      "grad_norm": 7.278139114379883,
       "learning_rate": 9.01901901901902e-07,
-      "loss": 0.8865,
+      "loss": 0.8988,
       "step": 90890
     },
     {
       "epoch": 272.97,
-      "grad_norm": 10.472726821899414,
+      "grad_norm": 7.917295455932617,
       "learning_rate": 9.00900900900901e-07,
-      "loss": 0.8877,
+      "loss": 0.8804,
       "step": 90900
     },
     {
       "epoch": 273.0,
-      "eval_accuracy": 0.8265,
-      "eval_loss": 0.5711143612861633,
-      "eval_runtime": 5.4353,
-      "eval_samples_per_second": 1839.808,
-      "eval_steps_per_second": 7.359,
+      "eval_accuracy": 0.8279,
+      "eval_loss": 0.5697124004364014,
+      "eval_runtime": 5.4596,
+      "eval_samples_per_second": 1831.644,
+      "eval_steps_per_second": 7.327,
       "step": 90909
     },
     {
       "epoch": 273.0,
-      "grad_norm": 6.543977737426758,
+      "grad_norm": 7.207139015197754,
       "learning_rate": 8.998998998998999e-07,
-      "loss": 1.14,
+      "loss": 1.1457,
       "step": 90910
     },
     {
       "epoch": 273.03,
-      "grad_norm": 5.768892765045166,
+      "grad_norm": 7.36659574508667,
       "learning_rate": 8.988988988988989e-07,
-      "loss": 0.8635,
+      "loss": 0.8611,
       "step": 90920
     },
     {
       "epoch": 273.06,
-      "grad_norm": 7.047219753265381,
+      "grad_norm": 7.086922645568848,
       "learning_rate": 8.978978978978979e-07,
-      "loss": 0.9403,
+      "loss": 0.9413,
       "step": 90930
     },
     {
       "epoch": 273.09,
-      "grad_norm": 7.619346618652344,
+      "grad_norm": 6.817042350769043,
       "learning_rate": 8.968968968968969e-07,
-      "loss": 0.8939,
+      "loss": 0.8871,
       "step": 90940
     },
     {
       "epoch": 273.12,
-      "grad_norm": 7.570402145385742,
+      "grad_norm": 7.129281520843506,
       "learning_rate": 8.958958958958958e-07,
-      "loss": 0.8857,
+      "loss": 0.8823,
       "step": 90950
     },
     {
       "epoch": 273.15,
-      "grad_norm": 9.08796215057373,
+      "grad_norm": 9.916000366210938,
       "learning_rate": 8.948948948948951e-07,
-      "loss": 0.8602,
+      "loss": 0.858,
       "step": 90960
     },
     {
       "epoch": 273.18,
-      "grad_norm": 5.634189128875732,
+      "grad_norm": 8.161724090576172,
       "learning_rate": 8.93893893893894e-07,
-      "loss": 0.9507,
+      "loss": 0.946,
       "step": 90970
     },
     {
       "epoch": 273.21,
-      "grad_norm": 6.623793601989746,
+      "grad_norm": 4.857535362243652,
       "learning_rate": 8.92892892892893e-07,
-      "loss": 0.931,
+      "loss": 0.9378,
       "step": 90980
     },
     {
       "epoch": 273.24,
-      "grad_norm": 5.9908366203308105,
+      "grad_norm": 7.017720699310303,
       "learning_rate": 8.91891891891892e-07,
-      "loss": 0.9763,
+      "loss": 0.9813,
       "step": 90990
     },
     {
       "epoch": 273.27,
-      "grad_norm": 7.130666255950928,
+      "grad_norm": 6.766077995300293,
       "learning_rate": 8.90890890890891e-07,
-      "loss": 0.9372,
+      "loss": 0.9294,
       "step": 91000
     },
     {
       "epoch": 273.3,
-      "grad_norm": 8.352115631103516,
+      "grad_norm": 10.71668529510498,
       "learning_rate": 8.8988988988989e-07,
-      "loss": 0.9224,
+      "loss": 0.935,
       "step": 91010
     },
     {
       "epoch": 273.33,
-      "grad_norm": 4.272334575653076,
+      "grad_norm": 6.140296459197998,
       "learning_rate": 8.88888888888889e-07,
-      "loss": 0.9205,
+      "loss": 0.9309,
       "step": 91020
     },
     {
       "epoch": 273.36,
-      "grad_norm": 7.07313346862793,
+      "grad_norm": 6.504068851470947,
       "learning_rate": 8.87887887887888e-07,
-      "loss": 0.921,
+      "loss": 0.9189,
       "step": 91030
     },
     {
       "epoch": 273.39,
-      "grad_norm": 5.933520317077637,
+      "grad_norm": 8.089334487915039,
       "learning_rate": 8.86886886886887e-07,
-      "loss": 0.956,
+      "loss": 0.9528,
       "step": 91040
     },
     {
       "epoch": 273.42,
-      "grad_norm": 5.730103492736816,
+      "grad_norm": 5.883544445037842,
       "learning_rate": 8.858858858858859e-07,
-      "loss": 0.9018,
+      "loss": 0.8989,
       "step": 91050
     },
     {
       "epoch": 273.45,
-      "grad_norm": 6.431992530822754,
+      "grad_norm": 7.354516506195068,
       "learning_rate": 8.848848848848849e-07,
-      "loss": 0.855,
+      "loss": 0.8492,
       "step": 91060
     },
     {
       "epoch": 273.48,
-      "grad_norm": 8.511468887329102,
+      "grad_norm": 8.367471694946289,
       "learning_rate": 8.83883883883884e-07,
-      "loss": 0.8766,
+      "loss": 0.8855,
       "step": 91070
     },
     {
       "epoch": 273.51,
-      "grad_norm": 7.47554874420166,
+      "grad_norm": 7.655096054077148,
       "learning_rate": 8.82882882882883e-07,
-      "loss": 0.9483,
+      "loss": 0.9412,
       "step": 91080
     },
     {
       "epoch": 273.54,
-      "grad_norm": 6.445580959320068,
+      "grad_norm": 6.652777194976807,
       "learning_rate": 8.81881881881882e-07,
-      "loss": 0.9012,
+      "loss": 0.8902,
       "step": 91090
     },
     {
       "epoch": 273.57,
-      "grad_norm": 6.047555923461914,
+      "grad_norm": 5.786723613739014,
       "learning_rate": 8.808808808808809e-07,
-      "loss": 0.9106,
+      "loss": 0.9125,
       "step": 91100
     },
     {
       "epoch": 273.6,
-      "grad_norm": 6.577455997467041,
+      "grad_norm": 6.053332328796387,
       "learning_rate": 8.798798798798799e-07,
-      "loss": 0.8979,
+      "loss": 0.9056,
       "step": 91110
     },
     {
       "epoch": 273.63,
-      "grad_norm": 6.818726539611816,
+      "grad_norm": 8.400137901306152,
       "learning_rate": 8.78878878878879e-07,
-      "loss": 0.9347,
+      "loss": 0.939,
       "step": 91120
     },
     {
       "epoch": 273.66,
-      "grad_norm": 5.5112433433532715,
+      "grad_norm": 5.530249118804932,
       "learning_rate": 8.77877877877878e-07,
-      "loss": 0.8831,
+      "loss": 0.8826,
       "step": 91130
     },
     {
       "epoch": 273.69,
-      "grad_norm": 7.192471504211426,
+      "grad_norm": 7.967604160308838,
       "learning_rate": 8.768768768768769e-07,
-      "loss": 0.9228,
+      "loss": 0.924,
       "step": 91140
     },
     {
       "epoch": 273.72,
-      "grad_norm": 5.49111270904541,
+      "grad_norm": 7.1502766609191895,
       "learning_rate": 8.758758758758759e-07,
-      "loss": 0.9305,
+      "loss": 0.9342,
       "step": 91150
     },
     {
       "epoch": 273.75,
-      "grad_norm": 8.27436637878418,
+      "grad_norm": 6.937205791473389,
       "learning_rate": 8.748748748748749e-07,
-      "loss": 0.9353,
+      "loss": 0.9365,
       "step": 91160
     },
     {
       "epoch": 273.78,
-      "grad_norm": 4.823544025421143,
+      "grad_norm": 5.584310531616211,
       "learning_rate": 8.738738738738739e-07,
-      "loss": 0.9591,
+      "loss": 0.9679,
       "step": 91170
     },
     {
       "epoch": 273.81,
-      "grad_norm": 5.841024398803711,
+      "grad_norm": 9.109628677368164,
       "learning_rate": 8.728728728728728e-07,
-      "loss": 0.9352,
+      "loss": 0.9366,
       "step": 91180
     },
     {
       "epoch": 273.84,
-      "grad_norm": 6.897751331329346,
+      "grad_norm": 8.714279174804688,
       "learning_rate": 8.718718718718719e-07,
-      "loss": 0.8411,
+      "loss": 0.8443,
       "step": 91190
     },
     {
       "epoch": 273.87,
-      "grad_norm": 11.117254257202148,
+      "grad_norm": 8.407544136047363,
       "learning_rate": 8.708708708708709e-07,
-      "loss": 0.9368,
+      "loss": 0.9335,
       "step": 91200
     },
     {
       "epoch": 273.9,
-      "grad_norm": 6.401161193847656,
+      "grad_norm": 6.611410617828369,
       "learning_rate": 8.698698698698699e-07,
-      "loss": 0.9473,
+      "loss": 0.9399,
       "step": 91210
     },
     {
       "epoch": 273.93,
-      "grad_norm": 6.311308860778809,
+      "grad_norm": 8.46375846862793,
       "learning_rate": 8.68868868868869e-07,
-      "loss": 0.8665,
+      "loss": 0.8562,
       "step": 91220
     },
     {
       "epoch": 273.96,
-      "grad_norm": 6.343881130218506,
+      "grad_norm": 6.457398891448975,
       "learning_rate": 8.67867867867868e-07,
-      "loss": 0.85,
+      "loss": 0.8476,
       "step": 91230
     },
     {
       "epoch": 273.99,
-      "grad_norm": 4.212661266326904,
+      "grad_norm": 4.285398483276367,
       "learning_rate": 8.66866866866867e-07,
-      "loss": 0.9416,
+      "loss": 0.946,
       "step": 91240
     },
     {
       "epoch": 274.0,
-      "eval_accuracy": 0.8273,
-      "eval_loss": 0.5671635866165161,
-      "eval_runtime": 5.6554,
-      "eval_samples_per_second": 1768.229,
-      "eval_steps_per_second": 7.073,
+      "eval_accuracy": 0.829,
+      "eval_loss": 0.5672373175621033,
+      "eval_runtime": 5.4004,
+      "eval_samples_per_second": 1851.721,
+      "eval_steps_per_second": 7.407,
       "step": 91242
     },
     {
       "epoch": 274.02,
-      "grad_norm": 8.39102840423584,
+      "grad_norm": 11.973392486572266,
       "learning_rate": 8.65865865865866e-07,
-      "loss": 1.0702,
+      "loss": 1.0638,
       "step": 91250
     },
     {
       "epoch": 274.05,
-      "grad_norm": 8.362641334533691,
+      "grad_norm": 10.172039985656738,
       "learning_rate": 8.64864864864865e-07,
-      "loss": 0.8899,
+      "loss": 0.8886,
       "step": 91260
     },
     {
       "epoch": 274.08,
-      "grad_norm": 6.498374938964844,
+      "grad_norm": 6.325463771820068,
       "learning_rate": 8.63863863863864e-07,
-      "loss": 0.9003,
+      "loss": 0.9031,
       "step": 91270
     },
     {
       "epoch": 274.11,
-      "grad_norm": 5.487128257751465,
+      "grad_norm": 5.429495334625244,
       "learning_rate": 8.628628628628629e-07,
-      "loss": 0.8788,
+      "loss": 0.878,
       "step": 91280
     },
     {
       "epoch": 274.14,
-      "grad_norm": 10.083869934082031,
+      "grad_norm": 6.957854270935059,
       "learning_rate": 8.618618618618619e-07,
-      "loss": 0.9672,
+      "loss": 0.9545,
       "step": 91290
     },
     {
       "epoch": 274.17,
-      "grad_norm": 11.850189208984375,
+      "grad_norm": 10.935525894165039,
       "learning_rate": 8.60860860860861e-07,
-      "loss": 0.9683,
+      "loss": 0.9708,
       "step": 91300
     },
     {
       "epoch": 274.2,
-      "grad_norm": 6.648820400238037,
+      "grad_norm": 7.747117042541504,
       "learning_rate": 8.5985985985986e-07,
-      "loss": 0.8909,
+      "loss": 0.8928,
       "step": 91310
     },
     {
       "epoch": 274.23,
-      "grad_norm": 7.2796244621276855,
+      "grad_norm": 6.923698425292969,
       "learning_rate": 8.58858858858859e-07,
-      "loss": 0.9169,
+      "loss": 0.9068,
       "step": 91320
     },
     {
       "epoch": 274.26,
-      "grad_norm": 7.504072189331055,
+      "grad_norm": 9.331808090209961,
       "learning_rate": 8.578578578578579e-07,
-      "loss": 0.869,
+      "loss": 0.8643,
       "step": 91330
     },
     {
       "epoch": 274.29,
-      "grad_norm": 4.677946090698242,
+      "grad_norm": 4.587507247924805,
       "learning_rate": 8.568568568568569e-07,
-      "loss": 0.8102,
+      "loss": 0.8135,
       "step": 91340
     },
     {
       "epoch": 274.32,
-      "grad_norm": 7.264098644256592,
+      "grad_norm": 7.100182056427002,
       "learning_rate": 8.55855855855856e-07,
-      "loss": 0.9612,
+      "loss": 0.9597,
       "step": 91350
     },
     {
       "epoch": 274.35,
-      "grad_norm": 5.637818336486816,
+      "grad_norm": 9.513097763061523,
       "learning_rate": 8.54854854854855e-07,
-      "loss": 0.9486,
+      "loss": 0.9467,
       "step": 91360
     },
     {
       "epoch": 274.38,
-      "grad_norm": 6.605020523071289,
+      "grad_norm": 6.019412517547607,
       "learning_rate": 8.538538538538539e-07,
-      "loss": 0.9038,
+      "loss": 0.8946,
       "step": 91370
     },
     {
       "epoch": 274.41,
-      "grad_norm": 6.6918864250183105,
+      "grad_norm": 7.797547340393066,
       "learning_rate": 8.528528528528529e-07,
-      "loss": 0.8783,
+      "loss": 0.875,
       "step": 91380
     },
     {
       "epoch": 274.44,
-      "grad_norm": 6.465898513793945,
+      "grad_norm": 7.650733470916748,
       "learning_rate": 8.518518518518519e-07,
-      "loss": 0.905,
+      "loss": 0.8982,
       "step": 91390
     },
     {
       "epoch": 274.47,
-      "grad_norm": 7.486813068389893,
+      "grad_norm": 9.175434112548828,
       "learning_rate": 8.508508508508509e-07,
-      "loss": 0.856,
+      "loss": 0.8557,
       "step": 91400
     },
     {
       "epoch": 274.5,
-      "grad_norm": 5.5920090675354,
+      "grad_norm": 6.517074108123779,
       "learning_rate": 8.498498498498498e-07,
-      "loss": 0.974,
+      "loss": 0.9694,
       "step": 91410
     },
     {
       "epoch": 274.53,
-      "grad_norm": 5.872392654418945,
+      "grad_norm": 6.921928405761719,
       "learning_rate": 8.488488488488489e-07,
-      "loss": 0.9206,
+      "loss": 0.9078,
       "step": 91420
     },
     {
       "epoch": 274.56,
-      "grad_norm": 5.966666221618652,
+      "grad_norm": 6.4471025466918945,
       "learning_rate": 8.478478478478479e-07,
-      "loss": 0.8162,
+      "loss": 0.8152,
       "step": 91430
     },
     {
       "epoch": 274.59,
-      "grad_norm": 6.155245780944824,
+      "grad_norm": 6.917952060699463,
       "learning_rate": 8.468468468468469e-07,
-      "loss": 0.9439,
+      "loss": 0.9308,
       "step": 91440
     },
     {
       "epoch": 274.62,
-      "grad_norm": 7.487297534942627,
+      "grad_norm": 10.134401321411133,
       "learning_rate": 8.458458458458458e-07,
-      "loss": 0.9036,
+      "loss": 0.9038,
       "step": 91450
     },
     {
       "epoch": 274.65,
-      "grad_norm": 12.434550285339355,
+      "grad_norm": 12.526215553283691,
       "learning_rate": 8.448448448448448e-07,
-      "loss": 0.9722,
+      "loss": 0.9608,
       "step": 91460
     },
     {
       "epoch": 274.68,
-      "grad_norm": 6.742058277130127,
+      "grad_norm": 7.47263765335083,
       "learning_rate": 8.438438438438439e-07,
-      "loss": 0.8935,
+      "loss": 0.8932,
       "step": 91470
     },
     {
       "epoch": 274.71,
-      "grad_norm": 7.276824474334717,
+      "grad_norm": 9.700422286987305,
       "learning_rate": 8.428428428428429e-07,
-      "loss": 0.9898,
+      "loss": 0.988,
       "step": 91480
     },
     {
       "epoch": 274.74,
-      "grad_norm": 8.426352500915527,
+      "grad_norm": 9.270098686218262,
       "learning_rate": 8.41841841841842e-07,
-      "loss": 0.9398,
+      "loss": 0.9389,
       "step": 91490
     },
     {
       "epoch": 274.77,
-      "grad_norm": 6.419271945953369,
+      "grad_norm": 8.54174518585205,
       "learning_rate": 8.40840840840841e-07,
-      "loss": 0.8559,
+      "loss": 0.8634,
       "step": 91500
     },
     {
       "epoch": 274.8,
-      "grad_norm": 7.106082439422607,
+      "grad_norm": 6.10016393661499,
       "learning_rate": 8.398398398398399e-07,
-      "loss": 0.8867,
+      "loss": 0.883,
       "step": 91510
     },
     {
       "epoch": 274.83,
-      "grad_norm": 5.838918209075928,
+      "grad_norm": 7.088839054107666,
       "learning_rate": 8.388388388388389e-07,
-      "loss": 0.8748,
+      "loss": 0.8813,
       "step": 91520
     },
     {
       "epoch": 274.86,
-      "grad_norm": 5.159956455230713,
+      "grad_norm": 5.576127052307129,
       "learning_rate": 8.37837837837838e-07,
-      "loss": 0.9275,
+      "loss": 0.9189,
       "step": 91530
     },
     {
       "epoch": 274.89,
-      "grad_norm": 6.641724586486816,
+      "grad_norm": 8.696996688842773,
       "learning_rate": 8.36836836836837e-07,
-      "loss": 0.9,
+      "loss": 0.8894,
       "step": 91540
     },
     {
       "epoch": 274.92,
-      "grad_norm": 10.16651725769043,
+      "grad_norm": 7.44417142868042,
       "learning_rate": 8.358358358358359e-07,
-      "loss": 0.9881,
+      "loss": 0.9863,
       "step": 91550
     },
     {
       "epoch": 274.95,
-      "grad_norm": 9.833551406860352,
+      "grad_norm": 8.91939926147461,
       "learning_rate": 8.348348348348349e-07,
-      "loss": 0.929,
+      "loss": 0.9307,
       "step": 91560
     },
     {
       "epoch": 274.98,
-      "grad_norm": 7.015872001647949,
+      "grad_norm": 8.704158782958984,
       "learning_rate": 8.338338338338339e-07,
-      "loss": 0.9037,
+      "loss": 0.9012,
       "step": 91570
     },
     {
       "epoch": 275.0,
       "eval_accuracy": 0.8284,
-      "eval_loss": 0.5688610672950745,
-      "eval_runtime": 5.1583,
-      "eval_samples_per_second": 1938.612,
-      "eval_steps_per_second": 7.754,
+      "eval_loss": 0.5673578381538391,
+      "eval_runtime": 5.3884,
+      "eval_samples_per_second": 1855.838,
+      "eval_steps_per_second": 7.423,
       "step": 91575
     },
     {
       "epoch": 275.02,
-      "grad_norm": 4.761903285980225,
+      "grad_norm": 6.822493553161621,
       "learning_rate": 8.328328328328329e-07,
-      "loss": 0.9909,
+      "loss": 1.0079,
       "step": 91580
     },
     {
       "epoch": 275.05,
-      "grad_norm": 5.5306396484375,
+      "grad_norm": 4.445763111114502,
       "learning_rate": 8.31831831831832e-07,
-      "loss": 0.897,
+      "loss": 0.9021,
       "step": 91590
     },
     {
       "epoch": 275.08,
-      "grad_norm": 7.136261940002441,
+      "grad_norm": 8.470526695251465,
       "learning_rate": 8.308308308308309e-07,
-      "loss": 0.936,
+      "loss": 0.9341,
       "step": 91600
     },
     {
       "epoch": 275.11,
-      "grad_norm": 8.094276428222656,
+      "grad_norm": 6.379286766052246,
       "learning_rate": 8.298298298298299e-07,
-      "loss": 0.8895,
+      "loss": 0.8805,
       "step": 91610
     },
     {
       "epoch": 275.14,
-      "grad_norm": 7.3319830894470215,
+      "grad_norm": 6.376497745513916,
       "learning_rate": 8.288288288288289e-07,
-      "loss": 0.9382,
+      "loss": 0.9355,
       "step": 91620
     },
     {
       "epoch": 275.17,
-      "grad_norm": 6.432563304901123,
+      "grad_norm": 6.155558109283447,
       "learning_rate": 8.278278278278279e-07,
-      "loss": 0.9016,
+      "loss": 0.8986,
       "step": 91630
     },
     {
       "epoch": 275.2,
-      "grad_norm": 8.505309104919434,
+      "grad_norm": 10.209965705871582,
       "learning_rate": 8.268268268268268e-07,
-      "loss": 0.9051,
+      "loss": 0.8988,
       "step": 91640
     },
     {
       "epoch": 275.23,
-      "grad_norm": 6.027726650238037,
+      "grad_norm": 8.561792373657227,
       "learning_rate": 8.258258258258259e-07,
-      "loss": 0.9091,
+      "loss": 0.9009,
       "step": 91650
     },
     {
       "epoch": 275.26,
-      "grad_norm": 7.022714138031006,
+      "grad_norm": 6.824260711669922,
       "learning_rate": 8.248248248248249e-07,
-      "loss": 0.8798,
+      "loss": 0.8769,
       "step": 91660
     },
     {
       "epoch": 275.29,
-      "grad_norm": 7.435145378112793,
+      "grad_norm": 7.7584075927734375,
       "learning_rate": 8.238238238238239e-07,
-      "loss": 0.8964,
+      "loss": 0.893,
       "step": 91670
     },
     {
       "epoch": 275.32,
-      "grad_norm": 6.519006729125977,
+      "grad_norm": 6.5052170753479,
       "learning_rate": 8.228228228228228e-07,
-      "loss": 0.9817,
+      "loss": 0.9812,
       "step": 91680
     },
     {
       "epoch": 275.35,
-      "grad_norm": 4.476500511169434,
+      "grad_norm": 6.5220465660095215,
       "learning_rate": 8.218218218218218e-07,
-      "loss": 0.8853,
+      "loss": 0.8895,
       "step": 91690
     },
     {
       "epoch": 275.38,
-      "grad_norm": 9.16887092590332,
+      "grad_norm": 10.052521705627441,
       "learning_rate": 8.208208208208208e-07,
-      "loss": 0.924,
+      "loss": 0.9207,
       "step": 91700
     },
     {
       "epoch": 275.41,
-      "grad_norm": 5.600899696350098,
+      "grad_norm": 5.553134918212891,
       "learning_rate": 8.198198198198199e-07,
-      "loss": 0.941,
+      "loss": 0.928,
       "step": 91710
     },
     {
       "epoch": 275.44,
-      "grad_norm": 7.847200870513916,
+      "grad_norm": 9.035257339477539,
       "learning_rate": 8.188188188188189e-07,
-      "loss": 0.8897,
+      "loss": 0.884,
       "step": 91720
     },
     {
       "epoch": 275.47,
-      "grad_norm": 7.427462577819824,
+      "grad_norm": 9.062533378601074,
       "learning_rate": 8.178178178178178e-07,
-      "loss": 1.0082,
+      "loss": 1.0019,
       "step": 91730
     },
     {
       "epoch": 275.5,
-      "grad_norm": 8.345793724060059,
+      "grad_norm": 10.715699195861816,
       "learning_rate": 8.168168168168168e-07,
-      "loss": 0.8336,
+      "loss": 0.837,
       "step": 91740
     },
     {
       "epoch": 275.53,
-      "grad_norm": 8.046209335327148,
+      "grad_norm": 6.369086265563965,
       "learning_rate": 8.158158158158158e-07,
-      "loss": 0.9486,
+      "loss": 0.934,
       "step": 91750
     },
     {
       "epoch": 275.56,
-      "grad_norm": 5.332213878631592,
+      "grad_norm": 6.432453155517578,
       "learning_rate": 8.14814814814815e-07,
-      "loss": 0.94,
+      "loss": 0.9344,
       "step": 91760
     },
     {
       "epoch": 275.59,
-      "grad_norm": 5.431699275970459,
+      "grad_norm": 10.288597106933594,
       "learning_rate": 8.13813813813814e-07,
-      "loss": 0.8122,
+      "loss": 0.8211,
       "step": 91770
     },
     {
       "epoch": 275.62,
-      "grad_norm": 6.515787124633789,
+      "grad_norm": 6.503774166107178,
       "learning_rate": 8.128128128128129e-07,
-      "loss": 0.9347,
+      "loss": 0.936,
       "step": 91780
     },
     {
       "epoch": 275.65,
-      "grad_norm": 6.027890682220459,
+      "grad_norm": 5.800983905792236,
       "learning_rate": 8.118118118118119e-07,
-      "loss": 0.9328,
+      "loss": 0.9265,
       "step": 91790
     },
     {
       "epoch": 275.68,
-      "grad_norm": 4.883716583251953,
+      "grad_norm": 7.688254356384277,
       "learning_rate": 8.108108108108109e-07,
-      "loss": 0.8784,
+      "loss": 0.8752,
       "step": 91800
     },
     {
       "epoch": 275.71,
-      "grad_norm": 5.75476598739624,
+      "grad_norm": 5.7397847175598145,
       "learning_rate": 8.098098098098099e-07,
-      "loss": 0.9328,
+      "loss": 0.9116,
       "step": 91810
     },
     {
       "epoch": 275.74,
-      "grad_norm": 4.574604511260986,
+      "grad_norm": 5.247100353240967,
       "learning_rate": 8.08808808808809e-07,
-      "loss": 0.9101,
+      "loss": 0.9079,
       "step": 91820
     },
     {
       "epoch": 275.77,
-      "grad_norm": 8.07053279876709,
+      "grad_norm": 6.510951995849609,
       "learning_rate": 8.078078078078079e-07,
-      "loss": 0.8857,
+      "loss": 0.8901,
       "step": 91830
     },
     {
       "epoch": 275.8,
-      "grad_norm": 5.851452827453613,
+      "grad_norm": 6.431622505187988,
       "learning_rate": 8.068068068068069e-07,
-      "loss": 0.8724,
+      "loss": 0.8809,
       "step": 91840
     },
     {
       "epoch": 275.83,
-      "grad_norm": 7.251561164855957,
+      "grad_norm": 10.179264068603516,
       "learning_rate": 8.058058058058059e-07,
-      "loss": 0.874,
+      "loss": 0.8755,
       "step": 91850
     },
     {
       "epoch": 275.86,
-      "grad_norm": 7.499328136444092,
+      "grad_norm": 9.192970275878906,
       "learning_rate": 8.048048048048049e-07,
-      "loss": 0.959,
+      "loss": 0.9602,
       "step": 91860
     },
     {
       "epoch": 275.89,
-      "grad_norm": 6.475515842437744,
+      "grad_norm": 6.759606838226318,
       "learning_rate": 8.038038038038038e-07,
-      "loss": 0.8287,
+      "loss": 0.8343,
       "step": 91870
     },
     {
       "epoch": 275.92,
-      "grad_norm": 5.442521572113037,
+      "grad_norm": 5.977389812469482,
       "learning_rate": 8.028028028028029e-07,
-      "loss": 0.9242,
+      "loss": 0.9093,
       "step": 91880
     },
     {
       "epoch": 275.95,
-      "grad_norm": 6.747589111328125,
+      "grad_norm": 6.354434013366699,
       "learning_rate": 8.018018018018019e-07,
-      "loss": 0.9662,
+      "loss": 0.9553,
       "step": 91890
     },
     {
       "epoch": 275.98,
-      "grad_norm": 6.112606525421143,
+      "grad_norm": 6.667107582092285,
       "learning_rate": 8.008008008008009e-07,
-      "loss": 0.8345,
+      "loss": 0.832,
       "step": 91900
     },
     {
       "epoch": 276.0,
-      "eval_accuracy": 0.8299,
-      "eval_loss": 0.5695052146911621,
-      "eval_runtime": 5.6167,
-      "eval_samples_per_second": 1780.397,
-      "eval_steps_per_second": 7.122,
+      "eval_accuracy": 0.8306,
+      "eval_loss": 0.5694773197174072,
+      "eval_runtime": 5.4626,
+      "eval_samples_per_second": 1830.622,
+      "eval_steps_per_second": 7.322,
       "step": 91908
     },
     {
       "epoch": 276.01,
-      "grad_norm": 7.711673736572266,
+      "grad_norm": 9.850144386291504,
       "learning_rate": 7.997997997997998e-07,
-      "loss": 1.0431,
+      "loss": 1.0468,
       "step": 91910
     },
     {
       "epoch": 276.04,
-      "grad_norm": 4.840076446533203,
+      "grad_norm": 4.851872444152832,
       "learning_rate": 7.987987987987988e-07,
-      "loss": 0.9439,
+      "loss": 0.9478,
       "step": 91920
     },
     {
       "epoch": 276.07,
-      "grad_norm": 9.561906814575195,
+      "grad_norm": 9.330443382263184,
       "learning_rate": 7.977977977977978e-07,
-      "loss": 0.9427,
+      "loss": 0.9391,
       "step": 91930
     },
     {
       "epoch": 276.1,
-      "grad_norm": 7.928176403045654,
+      "grad_norm": 6.291606903076172,
       "learning_rate": 7.967967967967969e-07,
-      "loss": 0.877,
+      "loss": 0.8749,
       "step": 91940
     },
     {
       "epoch": 276.13,
-      "grad_norm": 6.258967876434326,
+      "grad_norm": 5.494486331939697,
       "learning_rate": 7.957957957957958e-07,
-      "loss": 0.9025,
+      "loss": 0.9044,
       "step": 91950
     },
     {
       "epoch": 276.16,
-      "grad_norm": 6.958043098449707,
+      "grad_norm": 8.817266464233398,
       "learning_rate": 7.947947947947948e-07,
-      "loss": 0.9228,
+      "loss": 0.9366,
       "step": 91960
     },
     {
       "epoch": 276.19,
-      "grad_norm": 6.862864017486572,
+      "grad_norm": 4.926501750946045,
       "learning_rate": 7.937937937937938e-07,
-      "loss": 0.924,
+      "loss": 0.9255,
       "step": 91970
     },
     {
       "epoch": 276.22,
-      "grad_norm": 4.827759742736816,
+      "grad_norm": 7.516117095947266,
       "learning_rate": 7.927927927927928e-07,
-      "loss": 0.8789,
+      "loss": 0.8814,
       "step": 91980
     },
     {
       "epoch": 276.25,
-      "grad_norm": 6.918020725250244,
+      "grad_norm": 6.900108337402344,
       "learning_rate": 7.917917917917919e-07,
-      "loss": 0.9216,
+      "loss": 0.9125,
       "step": 91990
     },
     {
       "epoch": 276.28,
-      "grad_norm": 7.662712097167969,
+      "grad_norm": 8.787199974060059,
       "learning_rate": 7.907907907907908e-07,
-      "loss": 0.8911,
+      "loss": 0.8972,
       "step": 92000
     },
     {
       "epoch": 276.31,
-      "grad_norm": 7.315223693847656,
+      "grad_norm": 7.61363410949707,
       "learning_rate": 7.897897897897898e-07,
-      "loss": 0.9376,
+      "loss": 0.9211,
       "step": 92010
     },
     {
       "epoch": 276.34,
-      "grad_norm": 6.128088474273682,
+      "grad_norm": 5.217662811279297,
       "learning_rate": 7.887887887887888e-07,
-      "loss": 0.9181,
+      "loss": 0.9139,
       "step": 92020
     },
     {
       "epoch": 276.37,
-      "grad_norm": 8.083415031433105,
+      "grad_norm": 7.8714704513549805,
       "learning_rate": 7.877877877877879e-07,
-      "loss": 0.9338,
+      "loss": 0.9354,
       "step": 92030
     },
     {
       "epoch": 276.4,
-      "grad_norm": 5.155091285705566,
+      "grad_norm": 6.924942970275879,
       "learning_rate": 7.867867867867869e-07,
-      "loss": 0.8492,
+      "loss": 0.8413,
       "step": 92040
     },
     {
       "epoch": 276.43,
-      "grad_norm": 5.929826259613037,
+      "grad_norm": 5.801425457000732,
       "learning_rate": 7.857857857857859e-07,
-      "loss": 0.9406,
+      "loss": 0.9376,
       "step": 92050
     },
     {
       "epoch": 276.46,
-      "grad_norm": 5.315683364868164,
+      "grad_norm": 5.305128574371338,
       "learning_rate": 7.847847847847849e-07,
-      "loss": 1.0024,
+      "loss": 0.9896,
       "step": 92060
     },
     {
       "epoch": 276.49,
-      "grad_norm": 6.050452709197998,
+      "grad_norm": 6.880484104156494,
       "learning_rate": 7.837837837837839e-07,
-      "loss": 0.9305,
+      "loss": 0.9245,
       "step": 92070
     },
     {
       "epoch": 276.52,
-      "grad_norm": 5.573220252990723,
+      "grad_norm": 6.726749897003174,
       "learning_rate": 7.827827827827829e-07,
-      "loss": 0.8897,
+      "loss": 0.8917,
       "step": 92080
     },
     {
       "epoch": 276.55,
-      "grad_norm": 5.458828926086426,
+      "grad_norm": 6.790647029876709,
       "learning_rate": 7.817817817817819e-07,
-      "loss": 0.9766,
+      "loss": 0.9784,
       "step": 92090
     },
     {
       "epoch": 276.58,
-      "grad_norm": 4.9438066482543945,
+      "grad_norm": 6.49752950668335,
       "learning_rate": 7.807807807807808e-07,
-      "loss": 0.8977,
+      "loss": 0.8961,
       "step": 92100
     },
     {
       "epoch": 276.61,
-      "grad_norm": 7.091030120849609,
+      "grad_norm": 5.947443008422852,
       "learning_rate": 7.797797797797799e-07,
-      "loss": 0.8733,
+      "loss": 0.8651,
       "step": 92110
     },
     {
       "epoch": 276.64,
-      "grad_norm": 7.754078388214111,
+      "grad_norm": 6.390049934387207,
       "learning_rate": 7.787787787787789e-07,
-      "loss": 0.922,
+      "loss": 0.9183,
       "step": 92120
     },
     {
       "epoch": 276.67,
-      "grad_norm": 6.075771808624268,
+      "grad_norm": 6.244369029998779,
       "learning_rate": 7.777777777777779e-07,
-      "loss": 1.0103,
+      "loss": 0.9918,
       "step": 92130
     },
     {
       "epoch": 276.7,
-      "grad_norm": 9.245996475219727,
+      "grad_norm": 10.935163497924805,
       "learning_rate": 7.767767767767768e-07,
-      "loss": 0.9182,
+      "loss": 0.9296,
       "step": 92140
     },
     {
       "epoch": 276.73,
-      "grad_norm": 7.6787333488464355,
+      "grad_norm": 9.561530113220215,
       "learning_rate": 7.757757757757758e-07,
-      "loss": 0.8885,
+      "loss": 0.8744,
       "step": 92150
     },
     {
       "epoch": 276.76,
-      "grad_norm": 5.738522529602051,
+      "grad_norm": 6.083709239959717,
       "learning_rate": 7.747747747747748e-07,
-      "loss": 0.9476,
+      "loss": 0.9382,
       "step": 92160
     },
     {
       "epoch": 276.79,
-      "grad_norm": 5.292880535125732,
+      "grad_norm": 6.30128812789917,
       "learning_rate": 7.737737737737739e-07,
-      "loss": 0.875,
+      "loss": 0.8793,
       "step": 92170
     },
     {
       "epoch": 276.82,
-      "grad_norm": 10.013197898864746,
+      "grad_norm": 10.34902572631836,
       "learning_rate": 7.727727727727728e-07,
-      "loss": 0.919,
+      "loss": 0.9148,
       "step": 92180
     },
     {
       "epoch": 276.85,
-      "grad_norm": 6.668144226074219,
+      "grad_norm": 7.2575249671936035,
       "learning_rate": 7.717717717717718e-07,
-      "loss": 0.9051,
+      "loss": 0.9016,
       "step": 92190
     },
     {
       "epoch": 276.88,
-      "grad_norm": 10.429415702819824,
+      "grad_norm": 11.543721199035645,
       "learning_rate": 7.707707707707708e-07,
-      "loss": 0.8874,
+      "loss": 0.8952,
       "step": 92200
     },
     {
       "epoch": 276.91,
-      "grad_norm": 6.363533020019531,
+      "grad_norm": 6.285833835601807,
       "learning_rate": 7.697697697697698e-07,
-      "loss": 0.9274,
+      "loss": 0.9325,
       "step": 92210
     },
     {
       "epoch": 276.94,
-      "grad_norm": 5.221539497375488,
+      "grad_norm": 5.582337379455566,
       "learning_rate": 7.687687687687688e-07,
-      "loss": 0.903,
+      "loss": 0.9087,
       "step": 92220
     },
     {
       "epoch": 276.97,
-      "grad_norm": 5.936734676361084,
+      "grad_norm": 8.242130279541016,
       "learning_rate": 7.677677677677678e-07,
-      "loss": 0.9034,
+      "loss": 0.9064,
       "step": 92230
     },
     {
       "epoch": 277.0,
-      "grad_norm": 6.485191345214844,
+      "grad_norm": 6.754052639007568,
       "learning_rate": 7.667667667667668e-07,
-      "loss": 0.8627,
+      "loss": 0.8592,
       "step": 92240
     },
     {
       "epoch": 277.0,
-      "eval_accuracy": 0.8277,
-      "eval_loss": 0.574047327041626,
-      "eval_runtime": 5.8789,
-      "eval_samples_per_second": 1701.008,
-      "eval_steps_per_second": 6.804,
+      "eval_accuracy": 0.829,
+      "eval_loss": 0.5733513236045837,
+      "eval_runtime": 5.3761,
+      "eval_samples_per_second": 1860.1,
+      "eval_steps_per_second": 7.44,
       "step": 92241
     },
     {
       "epoch": 277.03,
-      "grad_norm": 6.626804828643799,
+      "grad_norm": 7.051092147827148,
       "learning_rate": 7.657657657657658e-07,
-      "loss": 0.973,
+      "loss": 0.9529,
       "step": 92250
     },
     {
       "epoch": 277.06,
-      "grad_norm": 7.199199199676514,
+      "grad_norm": 8.25096321105957,
       "learning_rate": 7.647647647647648e-07,
-      "loss": 0.9505,
+      "loss": 0.9466,
       "step": 92260
     },
     {
       "epoch": 277.09,
-      "grad_norm": 8.531537055969238,
+      "grad_norm": 6.6200361251831055,
       "learning_rate": 7.637637637637637e-07,
-      "loss": 0.9095,
+      "loss": 0.9041,
       "step": 92270
     },
     {
       "epoch": 277.12,
-      "grad_norm": 6.138494491577148,
+      "grad_norm": 6.897741317749023,
       "learning_rate": 7.627627627627627e-07,
-      "loss": 0.8457,
+      "loss": 0.8448,
       "step": 92280
     },
     {
       "epoch": 277.15,
-      "grad_norm": 5.326183319091797,
+      "grad_norm": 4.906522750854492,
       "learning_rate": 7.617617617617619e-07,
-      "loss": 0.8841,
+      "loss": 0.8966,
       "step": 92290
     },
     {
       "epoch": 277.18,
-      "grad_norm": 6.7263875007629395,
+      "grad_norm": 7.63535737991333,
       "learning_rate": 7.607607607607609e-07,
-      "loss": 0.9299,
+      "loss": 0.9216,
       "step": 92300
     },
     {
       "epoch": 277.21,
-      "grad_norm": 7.116785049438477,
+      "grad_norm": 7.837902545928955,
       "learning_rate": 7.597597597597599e-07,
-      "loss": 0.9275,
+      "loss": 0.9313,
       "step": 92310
     },
     {
       "epoch": 277.24,
-      "grad_norm": 6.687285900115967,
+      "grad_norm": 5.695587158203125,
       "learning_rate": 7.587587587587588e-07,
-      "loss": 0.9187,
+      "loss": 0.9123,
       "step": 92320
     },
     {
       "epoch": 277.27,
-      "grad_norm": 6.874638080596924,
+      "grad_norm": 6.064999580383301,
       "learning_rate": 7.577577577577578e-07,
-      "loss": 0.9055,
+      "loss": 0.9013,
       "step": 92330
     },
     {
       "epoch": 277.3,
-      "grad_norm": 5.5549116134643555,
+      "grad_norm": 5.544204235076904,
       "learning_rate": 7.567567567567569e-07,
-      "loss": 0.8699,
+      "loss": 0.8745,
       "step": 92340
     },
     {
       "epoch": 277.33,
-      "grad_norm": 9.166231155395508,
+      "grad_norm": 11.73093032836914,
       "learning_rate": 7.557557557557559e-07,
-      "loss": 0.9744,
+      "loss": 0.9708,
       "step": 92350
     },
     {
       "epoch": 277.36,
-      "grad_norm": 6.968006134033203,
+      "grad_norm": 5.4327778816223145,
       "learning_rate": 7.547547547547549e-07,
-      "loss": 0.9177,
+      "loss": 0.9227,
       "step": 92360
     },
     {
       "epoch": 277.39,
-      "grad_norm": 6.9218363761901855,
+      "grad_norm": 6.821206569671631,
       "learning_rate": 7.537537537537538e-07,
-      "loss": 0.9327,
+      "loss": 0.9404,
       "step": 92370
     },
     {
       "epoch": 277.42,
-      "grad_norm": 6.202635288238525,
+      "grad_norm": 5.069740295410156,
       "learning_rate": 7.527527527527528e-07,
-      "loss": 0.8834,
+      "loss": 0.8846,
       "step": 92380
     },
     {
       "epoch": 277.45,
-      "grad_norm": 6.147881031036377,
+      "grad_norm": 5.627203464508057,
       "learning_rate": 7.517517517517518e-07,
-      "loss": 0.9246,
+      "loss": 0.9284,
       "step": 92390
     },
     {
       "epoch": 277.48,
-      "grad_norm": 6.396838188171387,
+      "grad_norm": 8.253003120422363,
       "learning_rate": 7.507507507507509e-07,
-      "loss": 0.8734,
+      "loss": 0.8739,
       "step": 92400
     },
     {
       "epoch": 277.51,
-      "grad_norm": 5.425735950469971,
+      "grad_norm": 7.240413665771484,
       "learning_rate": 7.497497497497498e-07,
-      "loss": 0.8901,
+      "loss": 0.8925,
       "step": 92410
     },
     {
       "epoch": 277.54,
-      "grad_norm": 8.668984413146973,
+      "grad_norm": 11.43804931640625,
       "learning_rate": 7.487487487487488e-07,
-      "loss": 0.9097,
+      "loss": 0.9027,
       "step": 92420
     },
     {
       "epoch": 277.57,
-      "grad_norm": 9.347290992736816,
+      "grad_norm": 9.300599098205566,
       "learning_rate": 7.477477477477478e-07,
-      "loss": 0.891,
+      "loss": 0.895,
       "step": 92430
     },
     {
       "epoch": 277.6,
-      "grad_norm": 6.495796203613281,
+      "grad_norm": 7.674130439758301,
       "learning_rate": 7.467467467467468e-07,
-      "loss": 0.8708,
+      "loss": 0.8869,
       "step": 92440
     },
     {
       "epoch": 277.63,
-      "grad_norm": 8.76785945892334,
+      "grad_norm": 9.26746940612793,
       "learning_rate": 7.457457457457457e-07,
-      "loss": 0.9159,
+      "loss": 0.9171,
       "step": 92450
     },
     {
       "epoch": 277.66,
-      "grad_norm": 4.3415303230285645,
+      "grad_norm": 5.835110664367676,
       "learning_rate": 7.447447447447448e-07,
-      "loss": 0.9158,
+      "loss": 0.9168,
       "step": 92460
     },
     {
       "epoch": 277.69,
-      "grad_norm": 5.338366508483887,
+      "grad_norm": 5.77410364151001,
       "learning_rate": 7.437437437437438e-07,
-      "loss": 0.8744,
+      "loss": 0.8699,
       "step": 92470
     },
     {
       "epoch": 277.72,
-      "grad_norm": 7.556757926940918,
+      "grad_norm": 7.790188789367676,
       "learning_rate": 7.427427427427428e-07,
-      "loss": 0.9279,
+      "loss": 0.9233,
       "step": 92480
     },
     {
       "epoch": 277.75,
-      "grad_norm": 7.207361221313477,
+      "grad_norm": 7.349185943603516,
       "learning_rate": 7.417417417417418e-07,
-      "loss": 0.9875,
+      "loss": 0.9806,
       "step": 92490
     },
     {
       "epoch": 277.78,
-      "grad_norm": 9.594274520874023,
+      "grad_norm": 7.212000370025635,
       "learning_rate": 7.407407407407407e-07,
-      "loss": 0.8769,
+      "loss": 0.8645,
       "step": 92500
     },
     {
       "epoch": 277.81,
-      "grad_norm": 4.245636940002441,
+      "grad_norm": 5.628396034240723,
       "learning_rate": 7.397397397397397e-07,
-      "loss": 0.9074,
+      "loss": 0.9051,
       "step": 92510
     },
     {
       "epoch": 277.84,
-      "grad_norm": 4.495979309082031,
+      "grad_norm": 5.294569492340088,
       "learning_rate": 7.387387387387388e-07,
-      "loss": 0.8677,
+      "loss": 0.8565,
       "step": 92520
     },
     {
       "epoch": 277.87,
-      "grad_norm": 5.150935649871826,
+      "grad_norm": 5.815654754638672,
       "learning_rate": 7.377377377377378e-07,
-      "loss": 0.9127,
+      "loss": 0.9087,
       "step": 92530
     },
     {
       "epoch": 277.9,
-      "grad_norm": 5.5675764083862305,
+      "grad_norm": 5.105222225189209,
       "learning_rate": 7.367367367367367e-07,
-      "loss": 0.889,
+      "loss": 0.886,
       "step": 92540
     },
     {
       "epoch": 277.93,
-      "grad_norm": 5.665716171264648,
+      "grad_norm": 6.02232551574707,
       "learning_rate": 7.357357357357357e-07,
-      "loss": 0.9213,
+      "loss": 0.9299,
       "step": 92550
     },
     {
       "epoch": 277.96,
-      "grad_norm": 4.7215352058410645,
+      "grad_norm": 6.253346920013428,
       "learning_rate": 7.347347347347348e-07,
-      "loss": 0.9104,
+      "loss": 0.9176,
       "step": 92560
     },
     {
       "epoch": 277.99,
-      "grad_norm": 8.501802444458008,
+      "grad_norm": 6.866567134857178,
       "learning_rate": 7.337337337337339e-07,
-      "loss": 0.8964,
+      "loss": 0.8842,
       "step": 92570
     },
     {
       "epoch": 278.0,
-      "eval_accuracy": 0.8262,
-      "eval_loss": 0.5756428241729736,
-      "eval_runtime": 5.5201,
-      "eval_samples_per_second": 1811.559,
-      "eval_steps_per_second": 7.246,
+      "eval_accuracy": 0.8264,
+      "eval_loss": 0.5765002965927124,
+      "eval_runtime": 5.2177,
+      "eval_samples_per_second": 1916.539,
+      "eval_steps_per_second": 7.666,
       "step": 92574
     },
     {
       "epoch": 278.02,
-      "grad_norm": 7.797743320465088,
+      "grad_norm": 7.848457336425781,
       "learning_rate": 7.327327327327329e-07,
-      "loss": 1.1314,
+      "loss": 1.1271,
       "step": 92580
     },
     {
       "epoch": 278.05,
-      "grad_norm": 5.223755836486816,
+      "grad_norm": 7.26680326461792,
       "learning_rate": 7.317317317317319e-07,
-      "loss": 0.9176,
+      "loss": 0.9062,
       "step": 92590
     },
     {
       "epoch": 278.08,
-      "grad_norm": 5.957810878753662,
+      "grad_norm": 6.9440107345581055,
       "learning_rate": 7.307307307307308e-07,
-      "loss": 0.8683,
+      "loss": 0.8741,
       "step": 92600
     },
     {
       "epoch": 278.11,
-      "grad_norm": 4.564215183258057,
+      "grad_norm": 5.263193607330322,
       "learning_rate": 7.297297297297298e-07,
-      "loss": 0.932,
+      "loss": 0.9352,
       "step": 92610
     },
     {
       "epoch": 278.14,
-      "grad_norm": 7.135761737823486,
+      "grad_norm": 6.654391288757324,
       "learning_rate": 7.287287287287288e-07,
-      "loss": 0.8806,
+      "loss": 0.8912,
       "step": 92620
     },
     {
       "epoch": 278.17,
-      "grad_norm": 7.657742977142334,
+      "grad_norm": 6.866096496582031,
       "learning_rate": 7.277277277277279e-07,
-      "loss": 0.8655,
+      "loss": 0.858,
       "step": 92630
     },
     {
       "epoch": 278.2,
-      "grad_norm": 5.876004219055176,
+      "grad_norm": 6.461132526397705,
       "learning_rate": 7.267267267267268e-07,
-      "loss": 0.9075,
+      "loss": 0.9089,
       "step": 92640
     },
     {
       "epoch": 278.23,
-      "grad_norm": 6.536590099334717,
+      "grad_norm": 6.773262977600098,
       "learning_rate": 7.257257257257258e-07,
-      "loss": 0.9114,
+      "loss": 0.9099,
       "step": 92650
     },
     {
       "epoch": 278.26,
-      "grad_norm": 5.825421333312988,
+      "grad_norm": 5.199424743652344,
       "learning_rate": 7.247247247247248e-07,
-      "loss": 0.8701,
+      "loss": 0.8693,
       "step": 92660
     },
     {
       "epoch": 278.29,
-      "grad_norm": 7.303572177886963,
+      "grad_norm": 8.763754844665527,
       "learning_rate": 7.237237237237238e-07,
-      "loss": 0.964,
+      "loss": 0.9632,
       "step": 92670
     },
     {
       "epoch": 278.32,
-      "grad_norm": 5.007962703704834,
+      "grad_norm": 5.791506767272949,
       "learning_rate": 7.227227227227227e-07,
-      "loss": 0.9685,
+      "loss": 0.9583,
       "step": 92680
     },
     {
       "epoch": 278.35,
-      "grad_norm": 4.913329601287842,
+      "grad_norm": 5.882763385772705,
       "learning_rate": 7.217217217217218e-07,
-      "loss": 0.8962,
+      "loss": 0.8924,
       "step": 92690
     },
     {
       "epoch": 278.38,
-      "grad_norm": 7.3726654052734375,
+      "grad_norm": 7.813642501831055,
       "learning_rate": 7.207207207207208e-07,
-      "loss": 0.8763,
+      "loss": 0.8704,
       "step": 92700
     },
     {
       "epoch": 278.41,
-      "grad_norm": 6.276536464691162,
+      "grad_norm": 7.618997097015381,
       "learning_rate": 7.197197197197198e-07,
-      "loss": 0.9085,
+      "loss": 0.9038,
       "step": 92710
     },
     {
       "epoch": 278.44,
-      "grad_norm": 6.260035037994385,
+      "grad_norm": 6.137224197387695,
       "learning_rate": 7.187187187187188e-07,
-      "loss": 0.8998,
+      "loss": 0.9051,
       "step": 92720
     },
     {
       "epoch": 278.47,
-      "grad_norm": 7.673048496246338,
+      "grad_norm": 8.06644058227539,
       "learning_rate": 7.177177177177177e-07,
-      "loss": 0.898,
+      "loss": 0.9016,
       "step": 92730
     },
     {
       "epoch": 278.5,
-      "grad_norm": 8.48145866394043,
+      "grad_norm": 6.957803249359131,
       "learning_rate": 7.167167167167167e-07,
-      "loss": 0.9298,
+      "loss": 0.9357,
       "step": 92740
     },
     {
       "epoch": 278.53,
-      "grad_norm": 6.921023368835449,
+      "grad_norm": 7.689302444458008,
       "learning_rate": 7.157157157157158e-07,
-      "loss": 0.9519,
+      "loss": 0.9621,
       "step": 92750
     },
     {
       "epoch": 278.56,
-      "grad_norm": 6.476722240447998,
+      "grad_norm": 6.562351703643799,
       "learning_rate": 7.147147147147148e-07,
-      "loss": 0.9244,
+      "loss": 0.9202,
       "step": 92760
     },
     {
       "epoch": 278.59,
-      "grad_norm": 6.6869401931762695,
+      "grad_norm": 7.383624076843262,
       "learning_rate": 7.137137137137137e-07,
-      "loss": 0.88,
+      "loss": 0.8745,
       "step": 92770
     },
     {
       "epoch": 278.62,
-      "grad_norm": 9.452567100524902,
+      "grad_norm": 9.855730056762695,
       "learning_rate": 7.127127127127127e-07,
-      "loss": 0.9168,
+      "loss": 0.9214,
       "step": 92780
     },
     {
       "epoch": 278.65,
-      "grad_norm": 4.347983360290527,
+      "grad_norm": 6.06299352645874,
       "learning_rate": 7.117117117117117e-07,
-      "loss": 0.8843,
+      "loss": 0.8836,
       "step": 92790
     },
     {
       "epoch": 278.68,
-      "grad_norm": 11.6673002243042,
+      "grad_norm": 10.155706405639648,
       "learning_rate": 7.107107107107107e-07,
-      "loss": 0.9234,
+      "loss": 0.9215,
       "step": 92800
     },
     {
       "epoch": 278.71,
-      "grad_norm": 6.34547233581543,
+      "grad_norm": 9.828014373779297,
       "learning_rate": 7.097097097097097e-07,
-      "loss": 0.9415,
+      "loss": 0.932,
       "step": 92810
     },
     {
       "epoch": 278.74,
-      "grad_norm": 7.654984951019287,
+      "grad_norm": 5.991550445556641,
       "learning_rate": 7.087087087087087e-07,
-      "loss": 0.9673,
+      "loss": 0.9606,
       "step": 92820
     },
     {
       "epoch": 278.77,
-      "grad_norm": 6.456099510192871,
+      "grad_norm": 8.43188190460205,
       "learning_rate": 7.077077077077078e-07,
-      "loss": 0.9325,
+      "loss": 0.9344,
       "step": 92830
     },
     {
       "epoch": 278.8,
-      "grad_norm": 7.821959972381592,
+      "grad_norm": 8.1044921875,
       "learning_rate": 7.067067067067068e-07,
-      "loss": 0.9339,
+      "loss": 0.9416,
       "step": 92840
     },
     {
       "epoch": 278.83,
-      "grad_norm": 6.744688034057617,
+      "grad_norm": 10.701412200927734,
       "learning_rate": 7.057057057057058e-07,
-      "loss": 0.9352,
+      "loss": 0.9368,
       "step": 92850
     },
     {
       "epoch": 278.86,
-      "grad_norm": 6.477091312408447,
+      "grad_norm": 6.637618541717529,
       "learning_rate": 7.047047047047049e-07,
-      "loss": 0.9143,
+      "loss": 0.9089,
       "step": 92860
     },
     {
       "epoch": 278.89,
-      "grad_norm": 6.456478595733643,
+      "grad_norm": 6.4607343673706055,
       "learning_rate": 7.037037037037038e-07,
-      "loss": 0.9069,
+      "loss": 0.9028,
       "step": 92870
     },
     {
       "epoch": 278.92,
-      "grad_norm": 5.784519195556641,
+      "grad_norm": 5.606046199798584,
       "learning_rate": 7.027027027027028e-07,
-      "loss": 0.8898,
+      "loss": 0.8959,
       "step": 92880
     },
     {
       "epoch": 278.95,
-      "grad_norm": 6.217902183532715,
+      "grad_norm": 7.702358722686768,
       "learning_rate": 7.017017017017018e-07,
-      "loss": 0.8873,
+      "loss": 0.8806,
       "step": 92890
     },
     {
       "epoch": 278.98,
-      "grad_norm": 6.10832405090332,
+      "grad_norm": 6.020671367645264,
       "learning_rate": 7.007007007007008e-07,
-      "loss": 0.8793,
+      "loss": 0.872,
       "step": 92900
     },
     {
       "epoch": 279.0,
-      "eval_accuracy": 0.8268,
-      "eval_loss": 0.5728374123573303,
-      "eval_runtime": 5.376,
-      "eval_samples_per_second": 1860.116,
-      "eval_steps_per_second": 7.44,
+      "eval_accuracy": 0.827,
+      "eval_loss": 0.5729128122329712,
+      "eval_runtime": 5.1248,
+      "eval_samples_per_second": 1951.301,
+      "eval_steps_per_second": 7.805,
       "step": 92907
     },
     {
       "epoch": 279.01,
-      "grad_norm": 8.26775074005127,
+      "grad_norm": 6.780515670776367,
       "learning_rate": 6.996996996996997e-07,
-      "loss": 1.0001,
+      "loss": 0.9907,
       "step": 92910
     },
     {
       "epoch": 279.04,
-      "grad_norm": 6.212910175323486,
+      "grad_norm": 6.184480667114258,
       "learning_rate": 6.986986986986988e-07,
-      "loss": 0.8681,
+      "loss": 0.8677,
       "step": 92920
     },
     {
       "epoch": 279.07,
-      "grad_norm": 6.158387184143066,
+      "grad_norm": 6.8586578369140625,
       "learning_rate": 6.976976976976978e-07,
-      "loss": 0.8888,
+      "loss": 0.8793,
       "step": 92930
     },
     {
       "epoch": 279.1,
-      "grad_norm": 5.856427192687988,
+      "grad_norm": 7.602867603302002,
       "learning_rate": 6.966966966966968e-07,
-      "loss": 0.8612,
+      "loss": 0.8578,
       "step": 92940
     },
     {
       "epoch": 279.13,
-      "grad_norm": 7.270031452178955,
+      "grad_norm": 7.61496114730835,
       "learning_rate": 6.956956956956957e-07,
-      "loss": 0.9879,
+      "loss": 0.9897,
       "step": 92950
     },
     {
       "epoch": 279.16,
-      "grad_norm": 5.879837512969971,
+      "grad_norm": 5.3027849197387695,
       "learning_rate": 6.946946946946947e-07,
-      "loss": 0.8505,
+      "loss": 0.8609,
       "step": 92960
     },
     {
       "epoch": 279.19,
-      "grad_norm": 6.241171360015869,
+      "grad_norm": 6.009478569030762,
       "learning_rate": 6.936936936936937e-07,
-      "loss": 0.9014,
+      "loss": 0.8924,
       "step": 92970
     },
     {
       "epoch": 279.22,
-      "grad_norm": 6.2854323387146,
+      "grad_norm": 6.730966091156006,
       "learning_rate": 6.926926926926928e-07,
-      "loss": 0.8926,
+      "loss": 0.8832,
       "step": 92980
     },
     {
       "epoch": 279.25,
-      "grad_norm": 6.686382293701172,
+      "grad_norm": 6.033196449279785,
       "learning_rate": 6.916916916916918e-07,
-      "loss": 0.9305,
+      "loss": 0.9225,
       "step": 92990
     },
     {
       "epoch": 279.28,
-      "grad_norm": 5.9485273361206055,
+      "grad_norm": 6.232922077178955,
       "learning_rate": 6.906906906906907e-07,
-      "loss": 0.8693,
+      "loss": 0.8683,
       "step": 93000
     },
     {
       "epoch": 279.31,
-      "grad_norm": 6.808274269104004,
+      "grad_norm": 6.975637435913086,
       "learning_rate": 6.896896896896897e-07,
-      "loss": 0.8863,
+      "loss": 0.8867,
       "step": 93010
     },
     {
       "epoch": 279.34,
-      "grad_norm": 12.413859367370605,
+      "grad_norm": 11.975049018859863,
       "learning_rate": 6.886886886886887e-07,
-      "loss": 0.979,
+      "loss": 0.994,
       "step": 93020
     },
     {
       "epoch": 279.37,
-      "grad_norm": 5.142533779144287,
+      "grad_norm": 6.216423511505127,
       "learning_rate": 6.876876876876877e-07,
-      "loss": 0.8997,
+      "loss": 0.9046,
       "step": 93030
     },
     {
       "epoch": 279.4,
-      "grad_norm": 10.158255577087402,
+      "grad_norm": 8.982991218566895,
       "learning_rate": 6.866866866866867e-07,
-      "loss": 0.882,
+      "loss": 0.8858,
       "step": 93040
     },
     {
       "epoch": 279.43,
-      "grad_norm": 6.44525671005249,
+      "grad_norm": 5.670457363128662,
       "learning_rate": 6.856856856856857e-07,
-      "loss": 0.929,
+      "loss": 0.9374,
       "step": 93050
     },
     {
       "epoch": 279.46,
-      "grad_norm": 7.2013092041015625,
+      "grad_norm": 8.824944496154785,
       "learning_rate": 6.846846846846847e-07,
-      "loss": 0.9088,
+      "loss": 0.9083,
       "step": 93060
     },
     {
       "epoch": 279.49,
-      "grad_norm": 6.051013469696045,
+      "grad_norm": 7.239628314971924,
       "learning_rate": 6.836836836836837e-07,
-      "loss": 0.9111,
+      "loss": 0.9033,
       "step": 93070
     },
     {
       "epoch": 279.52,
-      "grad_norm": 6.34689998626709,
+      "grad_norm": 5.164305686950684,
       "learning_rate": 6.826826826826826e-07,
-      "loss": 0.9125,
+      "loss": 0.9106,
       "step": 93080
     },
     {
       "epoch": 279.55,
-      "grad_norm": 5.518311500549316,
+      "grad_norm": 7.322073936462402,
       "learning_rate": 6.816816816816819e-07,
-      "loss": 0.9949,
+      "loss": 0.9875,
       "step": 93090
     },
     {
       "epoch": 279.58,
-      "grad_norm": 9.47513484954834,
+      "grad_norm": 7.81954288482666,
       "learning_rate": 6.806806806806808e-07,
-      "loss": 0.8707,
+      "loss": 0.8639,
       "step": 93100
     },
     {
       "epoch": 279.61,
-      "grad_norm": 7.803071975708008,
+      "grad_norm": 6.913333415985107,
       "learning_rate": 6.796796796796798e-07,
-      "loss": 0.9232,
+      "loss": 0.9218,
       "step": 93110
     },
     {
       "epoch": 279.64,
-      "grad_norm": 5.2681355476379395,
+      "grad_norm": 5.555294513702393,
       "learning_rate": 6.786786786786788e-07,
-      "loss": 0.9256,
+      "loss": 0.9325,
       "step": 93120
     },
     {
       "epoch": 279.67,
-      "grad_norm": 6.942198753356934,
+      "grad_norm": 6.8005242347717285,
       "learning_rate": 6.776776776776778e-07,
-      "loss": 0.9479,
+      "loss": 0.9345,
       "step": 93130
     },
     {
       "epoch": 279.7,
-      "grad_norm": 5.63667106628418,
+      "grad_norm": 4.981520175933838,
       "learning_rate": 6.766766766766767e-07,
-      "loss": 0.8821,
+      "loss": 0.8844,
       "step": 93140
     },
     {
       "epoch": 279.73,
-      "grad_norm": 6.087000370025635,
+      "grad_norm": 6.582069396972656,
       "learning_rate": 6.756756756756758e-07,
-      "loss": 0.9248,
+      "loss": 0.9286,
       "step": 93150
     },
     {
       "epoch": 279.76,
-      "grad_norm": 7.471983909606934,
+      "grad_norm": 7.143840789794922,
       "learning_rate": 6.746746746746748e-07,
-      "loss": 0.931,
+      "loss": 0.9212,
       "step": 93160
     },
     {
       "epoch": 279.79,
-      "grad_norm": 5.144480228424072,
+      "grad_norm": 5.5217204093933105,
       "learning_rate": 6.736736736736738e-07,
-      "loss": 0.9087,
+      "loss": 0.9105,
       "step": 93170
     },
     {
       "epoch": 279.82,
-      "grad_norm": 8.116826057434082,
+      "grad_norm": 7.316493988037109,
       "learning_rate": 6.726726726726727e-07,
-      "loss": 0.9582,
+      "loss": 0.9607,
       "step": 93180
     },
     {
       "epoch": 279.85,
-      "grad_norm": 6.056633949279785,
+      "grad_norm": 6.431314945220947,
       "learning_rate": 6.716716716716717e-07,
-      "loss": 0.9624,
+      "loss": 0.9562,
       "step": 93190
     },
     {
       "epoch": 279.88,
-      "grad_norm": 6.001691818237305,
+      "grad_norm": 6.442792892456055,
       "learning_rate": 6.706706706706707e-07,
-      "loss": 0.9387,
+      "loss": 0.9356,
       "step": 93200
     },
     {
       "epoch": 279.91,
-      "grad_norm": 6.985274791717529,
+      "grad_norm": 5.495701789855957,
       "learning_rate": 6.696696696696698e-07,
-      "loss": 0.8487,
+      "loss": 0.8466,
       "step": 93210
     },
     {
       "epoch": 279.94,
-      "grad_norm": 6.944366455078125,
+      "grad_norm": 7.279847145080566,
       "learning_rate": 6.686686686686688e-07,
-      "loss": 0.9385,
+      "loss": 0.9426,
       "step": 93220
     },
     {
       "epoch": 279.97,
-      "grad_norm": 4.652657508850098,
+      "grad_norm": 5.377726078033447,
       "learning_rate": 6.676676676676677e-07,
-      "loss": 0.9166,
+      "loss": 0.9219,
       "step": 93230
     },
     {
       "epoch": 280.0,
-      "grad_norm": 68.19041442871094,
+      "grad_norm": 125.95001220703125,
       "learning_rate": 6.666666666666667e-07,
-      "loss": 1.0857,
+      "loss": 1.044,
       "step": 93240
     },
     {
       "epoch": 280.0,
-      "eval_accuracy": 0.8306,
-      "eval_loss": 0.5670732259750366,
-      "eval_runtime": 5.515,
-      "eval_samples_per_second": 1813.247,
-      "eval_steps_per_second": 7.253,
+      "eval_accuracy": 0.8319,
+      "eval_loss": 0.565922200679779,
+      "eval_runtime": 5.408,
+      "eval_samples_per_second": 1849.103,
+      "eval_steps_per_second": 7.396,
       "step": 93240
     },
     {
       "epoch": 280.03,
-      "grad_norm": 6.0536112785339355,
+      "grad_norm": 7.097991466522217,
       "learning_rate": 6.656656656656657e-07,
-      "loss": 0.9519,
+      "loss": 0.965,
       "step": 93250
     },
     {
       "epoch": 280.06,
-      "grad_norm": 7.860095024108887,
+      "grad_norm": 8.956608772277832,
       "learning_rate": 6.646646646646647e-07,
-      "loss": 0.9155,
+      "loss": 0.9047,
       "step": 93260
     },
     {
       "epoch": 280.09,
-      "grad_norm": 12.075226783752441,
+      "grad_norm": 13.509387969970703,
       "learning_rate": 6.636636636636637e-07,
-      "loss": 0.8653,
+      "loss": 0.8675,
       "step": 93270
     },
     {
       "epoch": 280.12,
-      "grad_norm": 5.978463172912598,
+      "grad_norm": 6.041293144226074,
       "learning_rate": 6.626626626626627e-07,
-      "loss": 0.9175,
+      "loss": 0.9113,
       "step": 93280
     },
     {
       "epoch": 280.15,
-      "grad_norm": 5.644857883453369,
+      "grad_norm": 5.954202175140381,
       "learning_rate": 6.616616616616617e-07,
-      "loss": 0.9037,
+      "loss": 0.8977,
       "step": 93290
     },
     {
       "epoch": 280.18,
-      "grad_norm": 7.039139270782471,
+      "grad_norm": 7.576687335968018,
       "learning_rate": 6.606606606606607e-07,
-      "loss": 0.9255,
+      "loss": 0.925,
       "step": 93300
     },
     {
       "epoch": 280.21,
-      "grad_norm": 5.245366096496582,
+      "grad_norm": 5.763840675354004,
       "learning_rate": 6.596596596596596e-07,
-      "loss": 0.9094,
+      "loss": 0.9043,
       "step": 93310
     },
     {
       "epoch": 280.24,
-      "grad_norm": 5.7134881019592285,
+      "grad_norm": 7.51145076751709,
       "learning_rate": 6.586586586586586e-07,
-      "loss": 0.8744,
+      "loss": 0.8807,
       "step": 93320
     },
     {
       "epoch": 280.27,
-      "grad_norm": 5.885521411895752,
+      "grad_norm": 6.9774394035339355,
       "learning_rate": 6.576576576576577e-07,
-      "loss": 0.9453,
+      "loss": 0.9375,
       "step": 93330
     },
     {
       "epoch": 280.3,
-      "grad_norm": 6.692439079284668,
+      "grad_norm": 9.262008666992188,
       "learning_rate": 6.566566566566567e-07,
-      "loss": 0.8994,
+      "loss": 0.9028,
       "step": 93340
     },
     {
       "epoch": 280.33,
-      "grad_norm": 5.566783905029297,
+      "grad_norm": 6.0443549156188965,
       "learning_rate": 6.556556556556556e-07,
       "loss": 0.9316,
       "step": 93350
     },
     {
       "epoch": 280.36,
-      "grad_norm": 7.7347636222839355,
+      "grad_norm": 7.321676254272461,
       "learning_rate": 6.546546546546548e-07,
-      "loss": 0.9474,
+      "loss": 0.939,
       "step": 93360
     },
     {
       "epoch": 280.39,
-      "grad_norm": 10.281641960144043,
+      "grad_norm": 8.594127655029297,
       "learning_rate": 6.536536536536537e-07,
       "loss": 0.9227,
       "step": 93370
     },
     {
       "epoch": 280.42,
-      "grad_norm": 6.5554327964782715,
+      "grad_norm": 5.327077388763428,
       "learning_rate": 6.526526526526528e-07,
-      "loss": 0.9406,
+      "loss": 0.9315,
       "step": 93380
     },
     {
       "epoch": 280.45,
-      "grad_norm": 7.25815486907959,
+      "grad_norm": 8.254833221435547,
       "learning_rate": 6.516516516516518e-07,
-      "loss": 0.8537,
+      "loss": 0.846,
       "step": 93390
     },
     {
       "epoch": 280.48,
-      "grad_norm": 6.471640586853027,
+      "grad_norm": 5.689449310302734,
       "learning_rate": 6.506506506506508e-07,
-      "loss": 0.9334,
+      "loss": 0.9273,
       "step": 93400
     },
     {
       "epoch": 280.51,
-      "grad_norm": 5.826311111450195,
+      "grad_norm": 7.922422885894775,
       "learning_rate": 6.496496496496497e-07,
-      "loss": 0.9199,
+      "loss": 0.9207,
       "step": 93410
     },
     {
       "epoch": 280.54,
-      "grad_norm": 7.436545372009277,
+      "grad_norm": 8.967936515808105,
       "learning_rate": 6.486486486486487e-07,
-      "loss": 0.884,
+      "loss": 0.8813,
       "step": 93420
     },
     {
       "epoch": 280.57,
-      "grad_norm": 9.757994651794434,
+      "grad_norm": 12.026484489440918,
       "learning_rate": 6.476476476476477e-07,
-      "loss": 0.9492,
+      "loss": 0.9379,
       "step": 93430
     },
     {
       "epoch": 280.6,
-      "grad_norm": 6.409905433654785,
+      "grad_norm": 7.341688632965088,
       "learning_rate": 6.466466466466468e-07,
-      "loss": 0.8634,
+      "loss": 0.8664,
       "step": 93440
     },
     {
       "epoch": 280.63,
-      "grad_norm": 6.097116470336914,
+      "grad_norm": 7.7083587646484375,
       "learning_rate": 6.456456456456457e-07,
-      "loss": 0.9383,
+      "loss": 0.9375,
       "step": 93450
     },
     {
       "epoch": 280.66,
-      "grad_norm": 6.3898701667785645,
+      "grad_norm": 6.1095194816589355,
       "learning_rate": 6.446446446446447e-07,
-      "loss": 0.86,
+      "loss": 0.8598,
       "step": 93460
     },
     {
       "epoch": 280.69,
-      "grad_norm": 8.181117057800293,
+      "grad_norm": 6.753756046295166,
       "learning_rate": 6.436436436436437e-07,
-      "loss": 0.8428,
+      "loss": 0.8561,
       "step": 93470
     },
     {
       "epoch": 280.72,
-      "grad_norm": 6.769516944885254,
+      "grad_norm": 7.620142459869385,
       "learning_rate": 6.426426426426427e-07,
-      "loss": 0.8616,
+      "loss": 0.8625,
       "step": 93480
     },
     {
       "epoch": 280.75,
-      "grad_norm": 6.3391313552856445,
+      "grad_norm": 8.296645164489746,
       "learning_rate": 6.416416416416417e-07,
-      "loss": 0.9143,
+      "loss": 0.9119,
       "step": 93490
     },
     {
       "epoch": 280.78,
-      "grad_norm": 6.15891695022583,
+      "grad_norm": 6.239481449127197,
       "learning_rate": 6.406406406406407e-07,
-      "loss": 0.8883,
+      "loss": 0.8937,
       "step": 93500
     },
     {
       "epoch": 280.81,
-      "grad_norm": 8.34593391418457,
+      "grad_norm": 7.144046306610107,
       "learning_rate": 6.396396396396397e-07,
-      "loss": 0.8784,
+      "loss": 0.8688,
       "step": 93510
     },
     {
       "epoch": 280.84,
-      "grad_norm": 4.631336688995361,
+      "grad_norm": 6.1563920974731445,
       "learning_rate": 6.386386386386387e-07,
-      "loss": 0.9558,
+      "loss": 0.9597,
       "step": 93520
     },
     {
       "epoch": 280.87,
-      "grad_norm": 5.529867649078369,
+      "grad_norm": 5.16436767578125,
       "learning_rate": 6.376376376376377e-07,
-      "loss": 0.8775,
+      "loss": 0.8793,
       "step": 93530
     },
     {
       "epoch": 280.9,
-      "grad_norm": 8.648277282714844,
+      "grad_norm": 8.998621940612793,
       "learning_rate": 6.366366366366366e-07,
-      "loss": 0.9335,
+      "loss": 0.9279,
       "step": 93540
     },
     {
       "epoch": 280.93,
-      "grad_norm": 6.450811862945557,
+      "grad_norm": 8.426607131958008,
       "learning_rate": 6.356356356356356e-07,
-      "loss": 0.8692,
+      "loss": 0.8728,
       "step": 93550
     },
     {
       "epoch": 280.96,
-      "grad_norm": 5.39520788192749,
+      "grad_norm": 6.065493583679199,
       "learning_rate": 6.346346346346347e-07,
-      "loss": 0.9048,
+      "loss": 0.9006,
       "step": 93560
     },
     {
       "epoch": 280.99,
-      "grad_norm": 5.921318054199219,
+      "grad_norm": 7.306640148162842,
       "learning_rate": 6.336336336336337e-07,
-      "loss": 0.8668,
+      "loss": 0.8633,
       "step": 93570
     },
     {
       "epoch": 281.0,
-      "eval_accuracy": 0.8305,
-      "eval_loss": 0.568498432636261,
-      "eval_runtime": 5.61,
-      "eval_samples_per_second": 1782.519,
-      "eval_steps_per_second": 7.13,
+      "eval_accuracy": 0.8295,
+      "eval_loss": 0.5693560838699341,
+      "eval_runtime": 5.3829,
+      "eval_samples_per_second": 1857.748,
+      "eval_steps_per_second": 7.431,
       "step": 93573
     },
     {
       "epoch": 281.02,
-      "grad_norm": 8.450757026672363,
+      "grad_norm": 7.15805196762085,
       "learning_rate": 6.326326326326326e-07,
-      "loss": 0.9506,
+      "loss": 0.9522,
       "step": 93580
     },
     {
       "epoch": 281.05,
-      "grad_norm": 7.35972261428833,
+      "grad_norm": 9.252472877502441,
       "learning_rate": 6.316316316316316e-07,
-      "loss": 0.9296,
+      "loss": 0.9257,
       "step": 93590
     },
     {
       "epoch": 281.08,
-      "grad_norm": 7.6909050941467285,
+      "grad_norm": 5.97149133682251,
       "learning_rate": 6.306306306306306e-07,
-      "loss": 0.9646,
+      "loss": 0.9728,
       "step": 93600
     },
     {
       "epoch": 281.11,
-      "grad_norm": 7.871725082397461,
+      "grad_norm": 7.633782863616943,
       "learning_rate": 6.296296296296296e-07,
-      "loss": 0.9288,
+      "loss": 0.9252,
       "step": 93610
     },
     {
       "epoch": 281.14,
-      "grad_norm": 7.3215837478637695,
+      "grad_norm": 6.750473976135254,
       "learning_rate": 6.286286286286287e-07,
-      "loss": 0.9731,
+      "loss": 0.978,
       "step": 93620
     },
     {
       "epoch": 281.17,
-      "grad_norm": 7.651861667633057,
+      "grad_norm": 11.549907684326172,
       "learning_rate": 6.276276276276278e-07,
-      "loss": 0.9837,
+      "loss": 0.9821,
       "step": 93630
     },
     {
       "epoch": 281.2,
-      "grad_norm": 5.53886604309082,
+      "grad_norm": 6.852136135101318,
       "learning_rate": 6.266266266266267e-07,
-      "loss": 0.9433,
+      "loss": 0.94,
       "step": 93640
     },
     {
       "epoch": 281.23,
-      "grad_norm": 7.393007278442383,
+      "grad_norm": 6.770543098449707,
       "learning_rate": 6.256256256256257e-07,
-      "loss": 0.8565,
+      "loss": 0.8642,
       "step": 93650
     },
     {
       "epoch": 281.26,
-      "grad_norm": 5.351471424102783,
+      "grad_norm": 7.460578441619873,
       "learning_rate": 6.246246246246246e-07,
-      "loss": 0.9018,
+      "loss": 0.8946,
       "step": 93660
     },
     {
       "epoch": 281.29,
-      "grad_norm": 6.591458320617676,
+      "grad_norm": 6.217329502105713,
       "learning_rate": 6.236236236236236e-07,
-      "loss": 0.9527,
+      "loss": 0.9425,
       "step": 93670
     },
     {
       "epoch": 281.32,
-      "grad_norm": 6.555788040161133,
+      "grad_norm": 6.338205337524414,
       "learning_rate": 6.226226226226227e-07,
-      "loss": 0.8487,
+      "loss": 0.8597,
       "step": 93680
     },
     {
       "epoch": 281.35,
-      "grad_norm": 5.170680522918701,
+      "grad_norm": 5.404399394989014,
       "learning_rate": 6.216216216216217e-07,
-      "loss": 0.909,
+      "loss": 0.904,
       "step": 93690
     },
     {
       "epoch": 281.38,
-      "grad_norm": 6.78475284576416,
+      "grad_norm": 7.429164886474609,
       "learning_rate": 6.206206206206207e-07,
-      "loss": 0.9664,
+      "loss": 0.9628,
       "step": 93700
     },
     {
       "epoch": 281.41,
-      "grad_norm": 8.08737564086914,
+      "grad_norm": 7.398869037628174,
       "learning_rate": 6.196196196196197e-07,
-      "loss": 0.933,
+      "loss": 0.9294,
       "step": 93710
     },
     {
       "epoch": 281.44,
-      "grad_norm": 5.809135437011719,
+      "grad_norm": 7.686655521392822,
       "learning_rate": 6.186186186186187e-07,
-      "loss": 0.9134,
+      "loss": 0.9159,
       "step": 93720
     },
     {
       "epoch": 281.47,
-      "grad_norm": 5.404476165771484,
+      "grad_norm": 5.751033782958984,
       "learning_rate": 6.176176176176177e-07,
-      "loss": 0.937,
+      "loss": 0.9461,
       "step": 93730
     },
     {
       "epoch": 281.5,
-      "grad_norm": 7.165302753448486,
+      "grad_norm": 6.8058762550354,
       "learning_rate": 6.166166166166167e-07,
-      "loss": 0.9107,
+      "loss": 0.9155,
       "step": 93740
     },
     {
       "epoch": 281.53,
-      "grad_norm": 6.44708776473999,
+      "grad_norm": 6.137560844421387,
       "learning_rate": 6.156156156156157e-07,
-      "loss": 0.9219,
+      "loss": 0.9228,
       "step": 93750
     },
     {
       "epoch": 281.56,
-      "grad_norm": 7.21232795715332,
+      "grad_norm": 6.045130252838135,
       "learning_rate": 6.146146146146147e-07,
-      "loss": 0.8915,
+      "loss": 0.8907,
       "step": 93760
     },
     {
       "epoch": 281.59,
-      "grad_norm": 6.708644866943359,
+      "grad_norm": 9.903508186340332,
       "learning_rate": 6.136136136136136e-07,
-      "loss": 0.9083,
+      "loss": 0.9119,
       "step": 93770
     },
     {
       "epoch": 281.62,
-      "grad_norm": 5.535583019256592,
+      "grad_norm": 10.221904754638672,
       "learning_rate": 6.126126126126126e-07,
-      "loss": 0.8869,
+      "loss": 0.8879,
       "step": 93780
     },
     {
       "epoch": 281.65,
-      "grad_norm": 5.95030403137207,
+      "grad_norm": 5.73384428024292,
       "learning_rate": 6.116116116116117e-07,
-      "loss": 0.8769,
+      "loss": 0.8793,
       "step": 93790
     },
     {
       "epoch": 281.68,
-      "grad_norm": 6.790555953979492,
+      "grad_norm": 5.938068389892578,
       "learning_rate": 6.106106106106107e-07,
-      "loss": 0.9163,
+      "loss": 0.9192,
       "step": 93800
     },
     {
       "epoch": 281.71,
-      "grad_norm": 6.3720855712890625,
+      "grad_norm": 5.131904602050781,
       "learning_rate": 6.096096096096096e-07,
-      "loss": 0.8842,
+      "loss": 0.8826,
       "step": 93810
     },
     {
       "epoch": 281.74,
-      "grad_norm": 7.360748291015625,
+      "grad_norm": 7.019989013671875,
       "learning_rate": 6.086086086086086e-07,
-      "loss": 0.8403,
+      "loss": 0.8293,
       "step": 93820
     },
     {
       "epoch": 281.77,
-      "grad_norm": 5.5954461097717285,
+      "grad_norm": 5.624490737915039,
       "learning_rate": 6.076076076076077e-07,
-      "loss": 0.8798,
+      "loss": 0.8756,
       "step": 93830
     },
     {
       "epoch": 281.8,
-      "grad_norm": 6.110034465789795,
+      "grad_norm": 6.223542213439941,
       "learning_rate": 6.066066066066066e-07,
-      "loss": 0.9728,
+      "loss": 0.977,
       "step": 93840
     },
     {
       "epoch": 281.83,
-      "grad_norm": 7.619960784912109,
+      "grad_norm": 7.071920871734619,
       "learning_rate": 6.056056056056057e-07,
-      "loss": 0.8404,
+      "loss": 0.8507,
       "step": 93850
     },
     {
       "epoch": 281.86,
-      "grad_norm": 6.240673065185547,
+      "grad_norm": 10.702765464782715,
       "learning_rate": 6.046046046046047e-07,
-      "loss": 0.8409,
+      "loss": 0.8355,
       "step": 93860
     },
     {
       "epoch": 281.89,
-      "grad_norm": 5.501859188079834,
+      "grad_norm": 5.0864129066467285,
       "learning_rate": 6.036036036036037e-07,
-      "loss": 0.8393,
+      "loss": 0.8397,
       "step": 93870
     },
     {
       "epoch": 281.92,
-      "grad_norm": 7.441924571990967,
+      "grad_norm": 7.798623085021973,
       "learning_rate": 6.026026026026026e-07,
-      "loss": 0.8927,
+      "loss": 0.8846,
       "step": 93880
     },
     {
       "epoch": 281.95,
-      "grad_norm": 7.066494941711426,
+      "grad_norm": 8.227519035339355,
       "learning_rate": 6.016016016016016e-07,
-      "loss": 1.0089,
+      "loss": 1.0232,
       "step": 93890
     },
     {
       "epoch": 281.98,
-      "grad_norm": 6.482019424438477,
+      "grad_norm": 6.733935832977295,
       "learning_rate": 6.006006006006006e-07,
-      "loss": 0.9051,
+      "loss": 0.904,
       "step": 93900
     },
     {
       "epoch": 282.0,
-      "eval_accuracy": 0.8299,
-      "eval_loss": 0.5751441121101379,
-      "eval_runtime": 5.3634,
-      "eval_samples_per_second": 1864.481,
-      "eval_steps_per_second": 7.458,
+      "eval_accuracy": 0.8293,
+      "eval_loss": 0.574241042137146,
+      "eval_runtime": 5.3681,
+      "eval_samples_per_second": 1862.869,
+      "eval_steps_per_second": 7.451,
       "step": 93906
     },
     {
       "epoch": 282.01,
-      "grad_norm": 8.809532165527344,
+      "grad_norm": 10.815971374511719,
       "learning_rate": 5.995995995995997e-07,
-      "loss": 1.1852,
+      "loss": 1.2047,
       "step": 93910
     },
     {
       "epoch": 282.04,
-      "grad_norm": 6.12144660949707,
+      "grad_norm": 5.11206579208374,
       "learning_rate": 5.985985985985987e-07,
-      "loss": 0.891,
+      "loss": 0.8808,
       "step": 93920
     },
     {
       "epoch": 282.07,
-      "grad_norm": 7.729508399963379,
+      "grad_norm": 6.464010238647461,
       "learning_rate": 5.975975975975976e-07,
-      "loss": 0.8699,
+      "loss": 0.8683,
       "step": 93930
     },
     {
       "epoch": 282.1,
-      "grad_norm": 4.629786491394043,
+      "grad_norm": 5.022777557373047,
       "learning_rate": 5.965965965965966e-07,
-      "loss": 0.9384,
+      "loss": 0.9366,
       "step": 93940
     },
     {
       "epoch": 282.13,
-      "grad_norm": 6.41082763671875,
+      "grad_norm": 9.425188064575195,
       "learning_rate": 5.955955955955956e-07,
-      "loss": 0.8399,
+      "loss": 0.8539,
       "step": 93950
     },
     {
       "epoch": 282.16,
-      "grad_norm": 5.567385196685791,
+      "grad_norm": 6.858564853668213,
       "learning_rate": 5.945945945945947e-07,
-      "loss": 0.8993,
+      "loss": 0.9027,
       "step": 93960
     },
     {
       "epoch": 282.19,
-      "grad_norm": 4.831497669219971,
+      "grad_norm": 5.754809856414795,
       "learning_rate": 5.935935935935937e-07,
-      "loss": 0.8517,
+      "loss": 0.8448,
       "step": 93970
     },
     {
       "epoch": 282.22,
-      "grad_norm": 13.747218132019043,
+      "grad_norm": 19.970314025878906,
       "learning_rate": 5.925925925925927e-07,
-      "loss": 0.8901,
+      "loss": 0.8935,
       "step": 93980
     },
     {
       "epoch": 282.25,
-      "grad_norm": 5.367348670959473,
+      "grad_norm": 7.441779136657715,
       "learning_rate": 5.915915915915917e-07,
-      "loss": 0.8981,
+      "loss": 0.8965,
       "step": 93990
     },
     {
       "epoch": 282.28,
-      "grad_norm": 8.350503921508789,
+      "grad_norm": 10.066197395324707,
       "learning_rate": 5.905905905905906e-07,
-      "loss": 0.8684,
+      "loss": 0.873,
       "step": 94000
     },
     {
       "epoch": 282.31,
-      "grad_norm": 5.585697650909424,
+      "grad_norm": 6.194280624389648,
       "learning_rate": 5.895895895895896e-07,
-      "loss": 0.8919,
+      "loss": 0.8908,
       "step": 94010
     },
     {
       "epoch": 282.34,
-      "grad_norm": 6.308382511138916,
+      "grad_norm": 6.351343154907227,
       "learning_rate": 5.885885885885887e-07,
-      "loss": 0.9081,
+      "loss": 0.9113,
       "step": 94020
     },
     {
       "epoch": 282.37,
-      "grad_norm": 6.397364139556885,
+      "grad_norm": 7.4982781410217285,
       "learning_rate": 5.875875875875877e-07,
-      "loss": 0.9535,
+      "loss": 0.9575,
       "step": 94030
     },
     {
       "epoch": 282.4,
-      "grad_norm": 5.464698791503906,
+      "grad_norm": 6.089322090148926,
       "learning_rate": 5.865865865865866e-07,
-      "loss": 0.9819,
+      "loss": 0.9881,
       "step": 94040
     },
     {
       "epoch": 282.43,
-      "grad_norm": 5.756587505340576,
+      "grad_norm": 7.492433071136475,
       "learning_rate": 5.855855855855856e-07,
-      "loss": 0.9735,
+      "loss": 0.9807,
       "step": 94050
     },
     {
       "epoch": 282.46,
-      "grad_norm": 7.1845011711120605,
+      "grad_norm": 8.039434432983398,
       "learning_rate": 5.845845845845846e-07,
-      "loss": 0.9538,
+      "loss": 0.9507,
       "step": 94060
     },
     {
       "epoch": 282.49,
-      "grad_norm": 6.517018795013428,
+      "grad_norm": 5.769461154937744,
       "learning_rate": 5.835835835835836e-07,
-      "loss": 0.8968,
+      "loss": 0.904,
       "step": 94070
     },
     {
       "epoch": 282.52,
-      "grad_norm": 6.154367923736572,
+      "grad_norm": 9.793501853942871,
       "learning_rate": 5.825825825825826e-07,
-      "loss": 0.9111,
+      "loss": 0.903,
       "step": 94080
     },
     {
       "epoch": 282.55,
-      "grad_norm": 6.047812461853027,
+      "grad_norm": 6.346325874328613,
       "learning_rate": 5.815815815815816e-07,
-      "loss": 0.9333,
+      "loss": 0.9311,
       "step": 94090
     },
     {
       "epoch": 282.58,
-      "grad_norm": 4.929807186126709,
+      "grad_norm": 5.195594310760498,
       "learning_rate": 5.805805805805807e-07,
-      "loss": 0.9253,
+      "loss": 0.9215,
       "step": 94100
     },
     {
       "epoch": 282.61,
-      "grad_norm": 6.533315658569336,
+      "grad_norm": 6.469295501708984,
       "learning_rate": 5.795795795795796e-07,
-      "loss": 0.9123,
+      "loss": 0.9102,
       "step": 94110
     },
     {
       "epoch": 282.64,
-      "grad_norm": 11.7636137008667,
+      "grad_norm": 12.096847534179688,
       "learning_rate": 5.785785785785786e-07,
-      "loss": 1.0163,
+      "loss": 1.0119,
       "step": 94120
     },
     {
       "epoch": 282.67,
-      "grad_norm": 6.712940692901611,
+      "grad_norm": 7.744287490844727,
       "learning_rate": 5.775775775775776e-07,
-      "loss": 0.8929,
+      "loss": 0.8988,
       "step": 94130
     },
     {
       "epoch": 282.7,
-      "grad_norm": 6.075592041015625,
+      "grad_norm": 5.9449639320373535,
       "learning_rate": 5.765765765765767e-07,
-      "loss": 0.9324,
+      "loss": 0.9266,
       "step": 94140
     },
     {
       "epoch": 282.73,
-      "grad_norm": 7.258015155792236,
+      "grad_norm": 7.2620954513549805,
       "learning_rate": 5.755755755755756e-07,
-      "loss": 0.915,
+      "loss": 0.9236,
       "step": 94150
     },
     {
       "epoch": 282.76,
-      "grad_norm": 7.903081893920898,
+      "grad_norm": 9.12565803527832,
       "learning_rate": 5.745745745745746e-07,
-      "loss": 0.9282,
+      "loss": 0.926,
       "step": 94160
     },
     {
       "epoch": 282.79,
-      "grad_norm": 10.644553184509277,
+      "grad_norm": 11.879918098449707,
       "learning_rate": 5.735735735735736e-07,
-      "loss": 0.9337,
+      "loss": 0.9238,
       "step": 94170
     },
     {
       "epoch": 282.82,
-      "grad_norm": 6.8167877197265625,
+      "grad_norm": 9.64084529876709,
       "learning_rate": 5.725725725725726e-07,
-      "loss": 0.8118,
+      "loss": 0.8151,
       "step": 94180
     },
     {
       "epoch": 282.85,
-      "grad_norm": 5.61602783203125,
+      "grad_norm": 7.325475215911865,
       "learning_rate": 5.715715715715716e-07,
-      "loss": 0.8923,
+      "loss": 0.8965,
       "step": 94190
     },
     {
       "epoch": 282.88,
-      "grad_norm": 4.936233997344971,
+      "grad_norm": 4.827665328979492,
       "learning_rate": 5.705705705705706e-07,
-      "loss": 0.8589,
+      "loss": 0.8473,
       "step": 94200
     },
     {
       "epoch": 282.91,
-      "grad_norm": 7.34697151184082,
+      "grad_norm": 8.072908401489258,
       "learning_rate": 5.695695695695696e-07,
-      "loss": 0.9439,
+      "loss": 0.9507,
       "step": 94210
     },
     {
       "epoch": 282.94,
-      "grad_norm": 9.056435585021973,
+      "grad_norm": 7.953220367431641,
       "learning_rate": 5.685685685685686e-07,
-      "loss": 0.9576,
+      "loss": 0.9746,
       "step": 94220
     },
     {
       "epoch": 282.97,
-      "grad_norm": 6.079002380371094,
+      "grad_norm": 5.993223667144775,
       "learning_rate": 5.675675675675676e-07,
-      "loss": 0.9183,
+      "loss": 0.9237,
       "step": 94230
     },
     {
       "epoch": 283.0,
-      "eval_accuracy": 0.8267,
-      "eval_loss": 0.5746187567710876,
-      "eval_runtime": 5.8052,
-      "eval_samples_per_second": 1722.598,
-      "eval_steps_per_second": 6.89,
+      "eval_accuracy": 0.8271,
+      "eval_loss": 0.5733911395072937,
+      "eval_runtime": 5.3557,
+      "eval_samples_per_second": 1867.183,
+      "eval_steps_per_second": 7.469,
       "step": 94239
     },
     {
       "epoch": 283.0,
-      "grad_norm": 7.707002639770508,
+      "grad_norm": 9.28600788116455,
       "learning_rate": 5.665665665665666e-07,
-      "loss": 1.174,
+      "loss": 1.1861,
       "step": 94240
     },
     {
       "epoch": 283.03,
-      "grad_norm": 8.086526870727539,
+      "grad_norm": 10.681473731994629,
       "learning_rate": 5.655655655655657e-07,
-      "loss": 0.9241,
+      "loss": 0.9113,
       "step": 94250
     },
     {
       "epoch": 283.06,
-      "grad_norm": 6.862698078155518,
+      "grad_norm": 6.2181830406188965,
       "learning_rate": 5.645645645645647e-07,
-      "loss": 0.8773,
+      "loss": 0.8783,
       "step": 94260
     },
     {
       "epoch": 283.09,
-      "grad_norm": 7.2408270835876465,
+      "grad_norm": 5.787871360778809,
       "learning_rate": 5.635635635635636e-07,
-      "loss": 0.9185,
+      "loss": 0.9109,
       "step": 94270
     },
     {
       "epoch": 283.12,
-      "grad_norm": 7.246132850646973,
+      "grad_norm": 7.090926647186279,
       "learning_rate": 5.625625625625626e-07,
-      "loss": 0.8692,
+      "loss": 0.867,
       "step": 94280
     },
     {
       "epoch": 283.15,
-      "grad_norm": 6.6554059982299805,
+      "grad_norm": 6.678389072418213,
       "learning_rate": 5.615615615615616e-07,
-      "loss": 0.9176,
+      "loss": 0.915,
       "step": 94290
     },
     {
       "epoch": 283.18,
-      "grad_norm": 9.379473686218262,
+      "grad_norm": 7.585777759552002,
       "learning_rate": 5.605605605605606e-07,
-      "loss": 0.9085,
+      "loss": 0.9076,
       "step": 94300
     },
     {
       "epoch": 283.21,
-      "grad_norm": 11.143158912658691,
+      "grad_norm": 8.526387214660645,
       "learning_rate": 5.595595595595596e-07,
-      "loss": 0.9523,
+      "loss": 0.9453,
       "step": 94310
     },
     {
       "epoch": 283.24,
-      "grad_norm": 6.192355155944824,
+      "grad_norm": 6.308413982391357,
       "learning_rate": 5.585585585585586e-07,
-      "loss": 0.9954,
+      "loss": 0.9795,
       "step": 94320
     },
     {
       "epoch": 283.27,
-      "grad_norm": 6.106219291687012,
+      "grad_norm": 6.38983154296875,
       "learning_rate": 5.575575575575576e-07,
-      "loss": 0.9316,
+      "loss": 0.9273,
       "step": 94330
     },
     {
       "epoch": 283.3,
-      "grad_norm": 6.910588264465332,
+      "grad_norm": 7.951139450073242,
       "learning_rate": 5.565565565565566e-07,
-      "loss": 0.8989,
+      "loss": 0.9011,
       "step": 94340
     },
     {
       "epoch": 283.33,
-      "grad_norm": 6.7348151206970215,
+      "grad_norm": 10.161040306091309,
       "learning_rate": 5.555555555555555e-07,
-      "loss": 0.9373,
+      "loss": 0.9284,
       "step": 94350
     },
     {
       "epoch": 283.36,
-      "grad_norm": 6.295235633850098,
+      "grad_norm": 8.16911506652832,
       "learning_rate": 5.545545545545546e-07,
-      "loss": 0.8778,
+      "loss": 0.8827,
       "step": 94360
     },
     {
       "epoch": 283.39,
-      "grad_norm": 7.240963935852051,
+      "grad_norm": 4.984744071960449,
       "learning_rate": 5.535535535535537e-07,
-      "loss": 0.8718,
+      "loss": 0.8755,
       "step": 94370
     },
     {
       "epoch": 283.42,
-      "grad_norm": 7.530025959014893,
+      "grad_norm": 7.015360355377197,
       "learning_rate": 5.525525525525526e-07,
-      "loss": 0.8284,
+      "loss": 0.8193,
       "step": 94380
     },
     {
       "epoch": 283.45,
-      "grad_norm": 11.76689338684082,
+      "grad_norm": 11.659465789794922,
       "learning_rate": 5.515515515515516e-07,
-      "loss": 0.9053,
+      "loss": 0.8972,
       "step": 94390
     },
     {
       "epoch": 283.48,
-      "grad_norm": 9.460384368896484,
+      "grad_norm": 5.8816728591918945,
       "learning_rate": 5.505505505505506e-07,
-      "loss": 0.876,
+      "loss": 0.8663,
       "step": 94400
     },
     {
       "epoch": 283.51,
-      "grad_norm": 5.989537239074707,
+      "grad_norm": 6.553326606750488,
       "learning_rate": 5.495495495495496e-07,
-      "loss": 0.8426,
+      "loss": 0.8406,
       "step": 94410
     },
     {
       "epoch": 283.54,
-      "grad_norm": 5.024631023406982,
+      "grad_norm": 6.401606559753418,
       "learning_rate": 5.485485485485486e-07,
-      "loss": 0.8683,
+      "loss": 0.8668,
       "step": 94420
     },
     {
       "epoch": 283.57,
-      "grad_norm": 6.634443283081055,
+      "grad_norm": 6.588178634643555,
       "learning_rate": 5.475475475475476e-07,
-      "loss": 0.8967,
+      "loss": 0.9045,
       "step": 94430
     },
     {
       "epoch": 283.6,
-      "grad_norm": 6.580499649047852,
+      "grad_norm": 6.2017645835876465,
       "learning_rate": 5.465465465465466e-07,
-      "loss": 0.8599,
+      "loss": 0.8521,
       "step": 94440
     },
     {
       "epoch": 283.63,
-      "grad_norm": 8.13507080078125,
+      "grad_norm": 8.090546607971191,
       "learning_rate": 5.455455455455456e-07,
-      "loss": 0.8847,
+      "loss": 0.8892,
       "step": 94450
     },
     {
       "epoch": 283.66,
-      "grad_norm": 5.141096591949463,
+      "grad_norm": 5.421953201293945,
       "learning_rate": 5.445445445445446e-07,
-      "loss": 0.8473,
+      "loss": 0.8459,
       "step": 94460
     },
     {
       "epoch": 283.69,
-      "grad_norm": 7.445930004119873,
+      "grad_norm": 5.892432689666748,
       "learning_rate": 5.435435435435435e-07,
-      "loss": 0.9313,
+      "loss": 0.9256,
       "step": 94470
     },
     {
       "epoch": 283.72,
-      "grad_norm": 4.708615303039551,
+      "grad_norm": 5.309521675109863,
       "learning_rate": 5.425425425425425e-07,
-      "loss": 0.9104,
+      "loss": 0.9262,
       "step": 94480
     },
     {
       "epoch": 283.75,
-      "grad_norm": 5.75912618637085,
+      "grad_norm": 7.094065189361572,
       "learning_rate": 5.415415415415416e-07,
-      "loss": 0.8707,
+      "loss": 0.8769,
       "step": 94490
     },
     {
       "epoch": 283.78,
-      "grad_norm": 9.296241760253906,
+      "grad_norm": 8.291787147521973,
       "learning_rate": 5.405405405405406e-07,
-      "loss": 0.9064,
+      "loss": 0.9001,
       "step": 94500
     },
     {
       "epoch": 283.81,
-      "grad_norm": 7.3343706130981445,
+      "grad_norm": 8.522994041442871,
       "learning_rate": 5.395395395395396e-07,
-      "loss": 0.8633,
+      "loss": 0.8586,
       "step": 94510
     },
     {
       "epoch": 283.84,
-      "grad_norm": 4.960142612457275,
+      "grad_norm": 5.845592975616455,
       "learning_rate": 5.385385385385386e-07,
-      "loss": 0.9472,
+      "loss": 0.9636,
       "step": 94520
     },
     {
       "epoch": 283.87,
-      "grad_norm": 6.440370082855225,
+      "grad_norm": 7.1242475509643555,
       "learning_rate": 5.375375375375376e-07,
-      "loss": 0.8726,
+      "loss": 0.8665,
       "step": 94530
     },
     {
       "epoch": 283.9,
-      "grad_norm": 4.743549823760986,
+      "grad_norm": 5.374210357666016,
       "learning_rate": 5.365365365365365e-07,
-      "loss": 0.9261,
+      "loss": 0.927,
       "step": 94540
     },
     {
       "epoch": 283.93,
-      "grad_norm": 7.6255364418029785,
+      "grad_norm": 6.286463260650635,
       "learning_rate": 5.355355355355356e-07,
-      "loss": 0.8906,
+      "loss": 0.8883,
       "step": 94550
     },
     {
       "epoch": 283.96,
-      "grad_norm": 6.854985237121582,
+      "grad_norm": 6.49800968170166,
       "learning_rate": 5.345345345345346e-07,
-      "loss": 0.8533,
+      "loss": 0.8532,
       "step": 94560
     },
     {
       "epoch": 283.99,
-      "grad_norm": 7.767961025238037,
+      "grad_norm": 7.606026649475098,
       "learning_rate": 5.335335335335336e-07,
-      "loss": 0.9507,
+      "loss": 0.9444,
       "step": 94570
     },
     {
       "epoch": 284.0,
-      "eval_accuracy": 0.8262,
-      "eval_loss": 0.56987464427948,
-      "eval_runtime": 5.3033,
-      "eval_samples_per_second": 1885.608,
-      "eval_steps_per_second": 7.542,
+      "eval_accuracy": 0.8285,
+      "eval_loss": 0.5704050064086914,
+      "eval_runtime": 5.3674,
+      "eval_samples_per_second": 1863.112,
+      "eval_steps_per_second": 7.452,
       "step": 94572
     },
     {
       "epoch": 284.02,
-      "grad_norm": 5.690791606903076,
+      "grad_norm": 7.541325092315674,
       "learning_rate": 5.325325325325325e-07,
-      "loss": 1.246,
+      "loss": 1.2454,
       "step": 94580
     },
     {
       "epoch": 284.05,
-      "grad_norm": 6.641265869140625,
+      "grad_norm": 6.043653964996338,
       "learning_rate": 5.315315315315315e-07,
-      "loss": 0.9534,
+      "loss": 0.9407,
       "step": 94590
     },
     {
       "epoch": 284.08,
-      "grad_norm": 8.06477165222168,
+      "grad_norm": 9.100628852844238,
       "learning_rate": 5.305305305305306e-07,
-      "loss": 0.9248,
+      "loss": 0.9262,
       "step": 94600
     },
     {
       "epoch": 284.11,
-      "grad_norm": 5.722204685211182,
+      "grad_norm": 5.973348140716553,
       "learning_rate": 5.295295295295296e-07,
-      "loss": 0.914,
+      "loss": 0.9103,
       "step": 94610
     },
     {
       "epoch": 284.14,
-      "grad_norm": 6.034558296203613,
+      "grad_norm": 6.188824653625488,
       "learning_rate": 5.285285285285286e-07,
-      "loss": 0.9335,
+      "loss": 0.9431,
       "step": 94620
     },
     {
       "epoch": 284.17,
-      "grad_norm": 6.249390602111816,
+      "grad_norm": 8.80744457244873,
       "learning_rate": 5.275275275275276e-07,
-      "loss": 0.9323,
+      "loss": 0.9341,
       "step": 94630
     },
     {
       "epoch": 284.2,
-      "grad_norm": 5.771833419799805,
+      "grad_norm": 7.046574592590332,
       "learning_rate": 5.265265265265266e-07,
-      "loss": 0.9397,
+      "loss": 0.9473,
       "step": 94640
     },
     {
       "epoch": 284.23,
-      "grad_norm": 6.152585029602051,
+      "grad_norm": 5.026200294494629,
       "learning_rate": 5.255255255255255e-07,
-      "loss": 0.8869,
+      "loss": 0.8884,
       "step": 94650
     },
     {
       "epoch": 284.26,
-      "grad_norm": 4.86219596862793,
+      "grad_norm": 9.456332206726074,
       "learning_rate": 5.245245245245246e-07,
-      "loss": 0.81,
+      "loss": 0.8121,
       "step": 94660
     },
     {
       "epoch": 284.29,
-      "grad_norm": 10.129953384399414,
+      "grad_norm": 9.725746154785156,
       "learning_rate": 5.235235235235236e-07,
-      "loss": 0.93,
+      "loss": 0.9273,
       "step": 94670
     },
     {
       "epoch": 284.32,
-      "grad_norm": 8.265799522399902,
+      "grad_norm": 9.130484580993652,
       "learning_rate": 5.225225225225226e-07,
-      "loss": 0.9157,
+      "loss": 0.9066,
       "step": 94680
     },
     {
       "epoch": 284.35,
-      "grad_norm": 6.982344627380371,
+      "grad_norm": 8.705439567565918,
       "learning_rate": 5.215215215215216e-07,
-      "loss": 0.9282,
+      "loss": 0.9294,
       "step": 94690
     },
     {
       "epoch": 284.38,
-      "grad_norm": 5.628733158111572,
+      "grad_norm": 6.336816310882568,
       "learning_rate": 5.205205205205205e-07,
-      "loss": 0.8801,
+      "loss": 0.8656,
       "step": 94700
     },
     {
       "epoch": 284.41,
-      "grad_norm": 7.8927788734436035,
+      "grad_norm": 8.533949851989746,
       "learning_rate": 5.195195195195195e-07,
-      "loss": 0.9807,
+      "loss": 0.9732,
       "step": 94710
     },
     {
       "epoch": 284.44,
-      "grad_norm": 7.11502742767334,
+      "grad_norm": 6.4058756828308105,
       "learning_rate": 5.185185185185186e-07,
-      "loss": 0.8806,
+      "loss": 0.8893,
       "step": 94720
     },
     {
       "epoch": 284.47,
-      "grad_norm": 6.257608413696289,
+      "grad_norm": 5.245304584503174,
       "learning_rate": 5.175175175175176e-07,
-      "loss": 0.9189,
+      "loss": 0.925,
       "step": 94730
     },
     {
       "epoch": 284.5,
-      "grad_norm": 13.471474647521973,
+      "grad_norm": 10.874311447143555,
       "learning_rate": 5.165165165165165e-07,
-      "loss": 0.8593,
+      "loss": 0.8584,
       "step": 94740
     },
     {
       "epoch": 284.53,
-      "grad_norm": 7.890151500701904,
+      "grad_norm": 7.420746803283691,
       "learning_rate": 5.155155155155155e-07,
-      "loss": 0.8515,
+      "loss": 0.8569,
       "step": 94750
     },
     {
       "epoch": 284.56,
-      "grad_norm": 6.721907615661621,
+      "grad_norm": 8.019206047058105,
       "learning_rate": 5.145145145145146e-07,
-      "loss": 0.9064,
+      "loss": 0.8962,
       "step": 94760
     },
     {
       "epoch": 284.59,
-      "grad_norm": 6.610014915466309,
+      "grad_norm": 7.312765598297119,
       "learning_rate": 5.135135135135135e-07,
-      "loss": 0.9621,
+      "loss": 0.9497,
       "step": 94770
     },
     {
       "epoch": 284.62,
-      "grad_norm": 12.095010757446289,
+      "grad_norm": 12.722189903259277,
       "learning_rate": 5.125125125125126e-07,
-      "loss": 0.8384,
+      "loss": 0.8442,
       "step": 94780
     },
     {
       "epoch": 284.65,
-      "grad_norm": 5.415124893188477,
+      "grad_norm": 5.0456390380859375,
       "learning_rate": 5.115115115115116e-07,
-      "loss": 0.9259,
+      "loss": 0.9249,
       "step": 94790
     },
     {
       "epoch": 284.68,
-      "grad_norm": 5.795363426208496,
+      "grad_norm": 7.681943416595459,
       "learning_rate": 5.105105105105106e-07,
-      "loss": 0.8661,
+      "loss": 0.868,
       "step": 94800
     },
     {
       "epoch": 284.71,
-      "grad_norm": 7.66715145111084,
+      "grad_norm": 9.246806144714355,
       "learning_rate": 5.095095095095095e-07,
-      "loss": 0.8821,
+      "loss": 0.8737,
       "step": 94810
     },
     {
       "epoch": 284.74,
-      "grad_norm": 8.01939868927002,
+      "grad_norm": 9.425219535827637,
       "learning_rate": 5.085085085085085e-07,
-      "loss": 0.8661,
+      "loss": 0.8417,
       "step": 94820
     },
     {
       "epoch": 284.77,
-      "grad_norm": 6.678418159484863,
+      "grad_norm": 6.610201358795166,
       "learning_rate": 5.075075075075076e-07,
-      "loss": 0.8682,
+      "loss": 0.8599,
       "step": 94830
     },
     {
       "epoch": 284.8,
-      "grad_norm": 5.674825191497803,
+      "grad_norm": 6.954248428344727,
       "learning_rate": 5.065065065065066e-07,
-      "loss": 0.9221,
+      "loss": 0.926,
       "step": 94840
     },
     {
       "epoch": 284.83,
-      "grad_norm": 5.155121803283691,
+      "grad_norm": 4.937582492828369,
       "learning_rate": 5.055055055055055e-07,
-      "loss": 0.8433,
+      "loss": 0.8442,
       "step": 94850
     },
     {
       "epoch": 284.86,
-      "grad_norm": 7.5911545753479,
+      "grad_norm": 9.4218111038208,
       "learning_rate": 5.045045045045045e-07,
-      "loss": 0.9249,
+      "loss": 0.926,
       "step": 94860
     },
     {
       "epoch": 284.89,
-      "grad_norm": 6.073254108428955,
+      "grad_norm": 5.9308600425720215,
       "learning_rate": 5.035035035035035e-07,
-      "loss": 0.8819,
+      "loss": 0.8895,
       "step": 94870
     },
     {
       "epoch": 284.92,
-      "grad_norm": 8.747124671936035,
+      "grad_norm": 8.66191577911377,
       "learning_rate": 5.025025025025025e-07,
-      "loss": 0.9028,
+      "loss": 0.9169,
       "step": 94880
     },
     {
       "epoch": 284.95,
-      "grad_norm": 5.194583892822266,
+      "grad_norm": 6.165359020233154,
       "learning_rate": 5.015015015015016e-07,
-      "loss": 0.8996,
+      "loss": 0.9027,
       "step": 94890
     },
     {
       "epoch": 284.98,
-      "grad_norm": 7.3105292320251465,
+      "grad_norm": 6.820361614227295,
       "learning_rate": 5.005005005005006e-07,
-      "loss": 0.8783,
+      "loss": 0.8772,
       "step": 94900
     },
     {
       "epoch": 285.0,
-      "eval_accuracy": 0.8289,
-      "eval_loss": 0.5745196342468262,
-      "eval_runtime": 5.3164,
-      "eval_samples_per_second": 1880.97,
-      "eval_steps_per_second": 7.524,
+      "eval_accuracy": 0.8298,
+      "eval_loss": 0.5748004913330078,
+      "eval_runtime": 5.3813,
+      "eval_samples_per_second": 1858.29,
+      "eval_steps_per_second": 7.433,
       "step": 94905
     },
     {
       "epoch": 285.02,
-      "grad_norm": 6.137405872344971,
+      "grad_norm": 5.593204498291016,
       "learning_rate": 4.994994994994996e-07,
-      "loss": 1.0325,
+      "loss": 1.017,
       "step": 94910
     },
     {
       "epoch": 285.05,
-      "grad_norm": 8.578142166137695,
+      "grad_norm": 9.948431015014648,
       "learning_rate": 4.984984984984986e-07,
-      "loss": 0.9049,
+      "loss": 0.8933,
       "step": 94920
     },
     {
       "epoch": 285.08,
-      "grad_norm": 6.668851375579834,
+      "grad_norm": 8.068488121032715,
       "learning_rate": 4.974974974974975e-07,
-      "loss": 0.9107,
+      "loss": 0.9065,
       "step": 94930
     },
     {
       "epoch": 285.11,
-      "grad_norm": 7.2388105392456055,
+      "grad_norm": 7.58504056930542,
       "learning_rate": 4.964964964964965e-07,
-      "loss": 0.9389,
+      "loss": 0.9296,
       "step": 94940
     },
     {
       "epoch": 285.14,
-      "grad_norm": 9.173542976379395,
+      "grad_norm": 7.121247291564941,
       "learning_rate": 4.954954954954956e-07,
-      "loss": 0.8901,
+      "loss": 0.8846,
       "step": 94950
     },
     {
       "epoch": 285.17,
-      "grad_norm": 4.9385666847229,
+      "grad_norm": 6.381187915802002,
       "learning_rate": 4.944944944944946e-07,
-      "loss": 0.9307,
+      "loss": 0.9201,
       "step": 94960
     },
     {
       "epoch": 285.2,
-      "grad_norm": 7.009274482727051,
+      "grad_norm": 6.358853816986084,
       "learning_rate": 4.934934934934935e-07,
-      "loss": 0.835,
+      "loss": 0.8322,
       "step": 94970
     },
     {
       "epoch": 285.23,
-      "grad_norm": 5.835934638977051,
+      "grad_norm": 7.087692737579346,
       "learning_rate": 4.924924924924925e-07,
-      "loss": 0.7888,
+      "loss": 0.7995,
       "step": 94980
     },
     {
       "epoch": 285.26,
-      "grad_norm": 11.057921409606934,
+      "grad_norm": 9.118692398071289,
       "learning_rate": 4.914914914914915e-07,
-      "loss": 0.976,
+      "loss": 0.9787,
       "step": 94990
     },
     {
       "epoch": 285.29,
-      "grad_norm": 6.087952613830566,
+      "grad_norm": 5.336687088012695,
       "learning_rate": 4.904904904904905e-07,
-      "loss": 0.921,
+      "loss": 0.9127,
       "step": 95000
     },
     {
       "epoch": 285.32,
-      "grad_norm": 4.797694206237793,
+      "grad_norm": 6.55172061920166,
       "learning_rate": 4.894894894894895e-07,
-      "loss": 0.9206,
+      "loss": 0.913,
       "step": 95010
     },
     {
       "epoch": 285.35,
-      "grad_norm": 10.75818920135498,
+      "grad_norm": 11.544368743896484,
       "learning_rate": 4.884884884884885e-07,
-      "loss": 0.9032,
+      "loss": 0.9023,
       "step": 95020
     },
     {
       "epoch": 285.38,
-      "grad_norm": 7.703462600708008,
+      "grad_norm": 6.546451568603516,
       "learning_rate": 4.874874874874876e-07,
-      "loss": 0.8509,
+      "loss": 0.8472,
       "step": 95030
     },
     {
       "epoch": 285.41,
-      "grad_norm": 7.807680606842041,
+      "grad_norm": 9.102458000183105,
       "learning_rate": 4.864864864864865e-07,
-      "loss": 0.9134,
+      "loss": 0.9111,
       "step": 95040
     },
     {
       "epoch": 285.44,
-      "grad_norm": 6.562778472900391,
+      "grad_norm": 7.275708198547363,
       "learning_rate": 4.854854854854855e-07,
-      "loss": 0.9136,
+      "loss": 0.9003,
       "step": 95050
     },
     {
       "epoch": 285.47,
-      "grad_norm": 8.378222465515137,
+      "grad_norm": 7.751875400543213,
       "learning_rate": 4.844844844844845e-07,
-      "loss": 0.9011,
+      "loss": 0.895,
       "step": 95060
     },
     {
       "epoch": 285.5,
-      "grad_norm": 6.297860145568848,
+      "grad_norm": 9.074051856994629,
       "learning_rate": 4.834834834834836e-07,
-      "loss": 0.903,
+      "loss": 0.9091,
       "step": 95070
     },
     {
       "epoch": 285.53,
-      "grad_norm": 8.53394603729248,
+      "grad_norm": 8.466559410095215,
       "learning_rate": 4.824824824824825e-07,
-      "loss": 0.9091,
+      "loss": 0.9021,
       "step": 95080
     },
     {
       "epoch": 285.56,
-      "grad_norm": 8.820236206054688,
+      "grad_norm": 8.975911140441895,
       "learning_rate": 4.814814814814815e-07,
-      "loss": 0.9175,
+      "loss": 0.9026,
       "step": 95090
     },
     {
       "epoch": 285.59,
-      "grad_norm": 5.979088306427002,
+      "grad_norm": 6.542499542236328,
       "learning_rate": 4.804804804804805e-07,
-      "loss": 0.8774,
+      "loss": 0.8812,
       "step": 95100
     },
     {
       "epoch": 285.62,
-      "grad_norm": 7.557089328765869,
+      "grad_norm": 7.92050313949585,
       "learning_rate": 4.794794794794795e-07,
       "loss": 0.9026,
       "step": 95110
     },
     {
       "epoch": 285.65,
-      "grad_norm": 5.805126667022705,
+      "grad_norm": 5.3177924156188965,
       "learning_rate": 4.784784784784786e-07,
-      "loss": 0.8774,
+      "loss": 0.8814,
       "step": 95120
     },
     {
       "epoch": 285.68,
-      "grad_norm": 5.666385173797607,
+      "grad_norm": 7.120809078216553,
       "learning_rate": 4.774774774774775e-07,
-      "loss": 0.8864,
+      "loss": 0.8795,
       "step": 95130
     },
     {
       "epoch": 285.71,
-      "grad_norm": 8.773850440979004,
+      "grad_norm": 8.603256225585938,
       "learning_rate": 4.764764764764765e-07,
-      "loss": 0.9631,
+      "loss": 0.9575,
       "step": 95140
     },
     {
       "epoch": 285.74,
-      "grad_norm": 7.080404758453369,
+      "grad_norm": 6.021905422210693,
       "learning_rate": 4.754754754754755e-07,
-      "loss": 0.8582,
+      "loss": 0.8561,
       "step": 95150
     },
     {
       "epoch": 285.77,
-      "grad_norm": 6.363414287567139,
+      "grad_norm": 7.243905067443848,
       "learning_rate": 4.744744744744745e-07,
-      "loss": 0.9042,
+      "loss": 0.9091,
       "step": 95160
     },
     {
       "epoch": 285.8,
-      "grad_norm": 4.90008020401001,
+      "grad_norm": 7.302052974700928,
       "learning_rate": 4.7347347347347354e-07,
-      "loss": 0.9196,
+      "loss": 0.909,
       "step": 95170
     },
     {
       "epoch": 285.83,
-      "grad_norm": 6.356344223022461,
+      "grad_norm": 6.915886402130127,
       "learning_rate": 4.7247247247247256e-07,
-      "loss": 0.8995,
+      "loss": 0.8943,
       "step": 95180
     },
     {
       "epoch": 285.86,
-      "grad_norm": 8.39130687713623,
+      "grad_norm": 8.856454849243164,
       "learning_rate": 4.714714714714715e-07,
-      "loss": 0.8494,
+      "loss": 0.852,
       "step": 95190
     },
     {
       "epoch": 285.89,
-      "grad_norm": 7.553210258483887,
+      "grad_norm": 7.917201519012451,
       "learning_rate": 4.7047047047047054e-07,
-      "loss": 0.8268,
+      "loss": 0.8298,
       "step": 95200
     },
     {
       "epoch": 285.92,
-      "grad_norm": 5.5426025390625,
+      "grad_norm": 6.056844234466553,
       "learning_rate": 4.694694694694695e-07,
-      "loss": 0.9499,
+      "loss": 0.9421,
       "step": 95210
     },
     {
       "epoch": 285.95,
-      "grad_norm": 7.92521858215332,
+      "grad_norm": 6.84312105178833,
       "learning_rate": 4.684684684684685e-07,
-      "loss": 0.8524,
+      "loss": 0.847,
       "step": 95220
     },
     {
       "epoch": 285.98,
-      "grad_norm": 7.104458332061768,
+      "grad_norm": 7.407921314239502,
       "learning_rate": 4.674674674674675e-07,
-      "loss": 0.8942,
+      "loss": 0.8895,
       "step": 95230
     },
     {
       "epoch": 286.0,
-      "eval_accuracy": 0.8262,
-      "eval_loss": 0.5745360851287842,
-      "eval_runtime": 5.6924,
-      "eval_samples_per_second": 1756.741,
-      "eval_steps_per_second": 7.027,
+      "eval_accuracy": 0.8275,
+      "eval_loss": 0.5721040964126587,
+      "eval_runtime": 5.3832,
+      "eval_samples_per_second": 1857.639,
+      "eval_steps_per_second": 7.431,
       "step": 95238
     },
     {
       "epoch": 286.01,
-      "grad_norm": 6.349193096160889,
+      "grad_norm": 4.785063743591309,
       "learning_rate": 4.664664664664665e-07,
-      "loss": 1.0094,
+      "loss": 0.9997,
       "step": 95240
     },
     {
       "epoch": 286.04,
-      "grad_norm": 6.223058223724365,
+      "grad_norm": 6.420033931732178,
       "learning_rate": 4.654654654654655e-07,
-      "loss": 0.9271,
+      "loss": 0.9126,
       "step": 95250
     },
     {
       "epoch": 286.07,
-      "grad_norm": 6.181783676147461,
+      "grad_norm": 5.990813255310059,
       "learning_rate": 4.644644644644645e-07,
-      "loss": 0.8719,
+      "loss": 0.8653,
       "step": 95260
     },
     {
       "epoch": 286.1,
-      "grad_norm": 6.118484020233154,
+      "grad_norm": 9.301092147827148,
       "learning_rate": 4.6346346346346346e-07,
-      "loss": 0.9181,
+      "loss": 0.9165,
       "step": 95270
     },
     {
       "epoch": 286.13,
-      "grad_norm": 6.364919662475586,
+      "grad_norm": 13.168231010437012,
       "learning_rate": 4.624624624624625e-07,
-      "loss": 0.8549,
+      "loss": 0.8599,
       "step": 95280
     },
     {
       "epoch": 286.16,
-      "grad_norm": 7.690777778625488,
+      "grad_norm": 7.716577529907227,
       "learning_rate": 4.6146146146146144e-07,
-      "loss": 1.0017,
+      "loss": 0.9949,
       "step": 95290
     },
     {
       "epoch": 286.19,
-      "grad_norm": 5.658433437347412,
+      "grad_norm": 6.217874526977539,
       "learning_rate": 4.604604604604605e-07,
-      "loss": 0.8538,
+      "loss": 0.8492,
       "step": 95300
     },
     {
       "epoch": 286.22,
-      "grad_norm": 7.770297527313232,
+      "grad_norm": 8.500850677490234,
       "learning_rate": 4.5945945945945953e-07,
-      "loss": 0.9459,
+      "loss": 0.9377,
       "step": 95310
     },
     {
       "epoch": 286.25,
-      "grad_norm": 5.805601119995117,
+      "grad_norm": 7.491369247436523,
       "learning_rate": 4.584584584584585e-07,
-      "loss": 0.8505,
+      "loss": 0.8495,
       "step": 95320
     },
     {
       "epoch": 286.28,
-      "grad_norm": 5.836670398712158,
+      "grad_norm": 7.464807033538818,
       "learning_rate": 4.574574574574575e-07,
-      "loss": 0.8576,
+      "loss": 0.8665,
       "step": 95330
     },
     {
       "epoch": 286.31,
-      "grad_norm": 5.93828010559082,
+      "grad_norm": 5.493793487548828,
       "learning_rate": 4.564564564564565e-07,
-      "loss": 0.8874,
+      "loss": 0.8855,
       "step": 95340
     },
     {
       "epoch": 286.34,
-      "grad_norm": 7.836377143859863,
+      "grad_norm": 7.382957458496094,
       "learning_rate": 4.554554554554555e-07,
-      "loss": 0.877,
+      "loss": 0.8823,
       "step": 95350
     },
     {
       "epoch": 286.37,
-      "grad_norm": 6.3243584632873535,
+      "grad_norm": 7.452633380889893,
       "learning_rate": 4.544544544544545e-07,
-      "loss": 0.9337,
+      "loss": 0.9282,
       "step": 95360
     },
     {
       "epoch": 286.4,
-      "grad_norm": 7.644233703613281,
+      "grad_norm": 7.797518253326416,
       "learning_rate": 4.534534534534535e-07,
-      "loss": 0.9362,
+      "loss": 0.9345,
       "step": 95370
     },
     {
       "epoch": 286.43,
-      "grad_norm": 6.77482795715332,
+      "grad_norm": 8.1951904296875,
       "learning_rate": 4.524524524524525e-07,
-      "loss": 0.956,
+      "loss": 0.9585,
       "step": 95380
     },
     {
       "epoch": 286.46,
-      "grad_norm": 5.901970863342285,
+      "grad_norm": 6.953071117401123,
       "learning_rate": 4.5145145145145147e-07,
-      "loss": 0.9013,
+      "loss": 0.9004,
       "step": 95390
     },
     {
       "epoch": 286.49,
-      "grad_norm": 5.058960437774658,
+      "grad_norm": 4.295292854309082,
       "learning_rate": 4.504504504504505e-07,
-      "loss": 0.8939,
+      "loss": 0.9032,
       "step": 95400
     },
     {
       "epoch": 286.52,
-      "grad_norm": 8.95838737487793,
+      "grad_norm": 11.413030624389648,
       "learning_rate": 4.4944944944944945e-07,
-      "loss": 0.9594,
+      "loss": 0.9557,
       "step": 95410
     },
     {
       "epoch": 286.55,
-      "grad_norm": 4.649218559265137,
+      "grad_norm": 5.840286731719971,
       "learning_rate": 4.4844844844844847e-07,
-      "loss": 0.8265,
+      "loss": 0.8255,
       "step": 95420
     },
     {
       "epoch": 286.58,
-      "grad_norm": 5.349297046661377,
+      "grad_norm": 7.038581371307373,
       "learning_rate": 4.4744744744744754e-07,
-      "loss": 1.0075,
+      "loss": 1.006,
       "step": 95430
     },
     {
       "epoch": 286.61,
-      "grad_norm": 5.2084736824035645,
+      "grad_norm": 6.329689025878906,
       "learning_rate": 4.464464464464465e-07,
-      "loss": 0.8851,
+      "loss": 0.8831,
       "step": 95440
     },
     {
       "epoch": 286.64,
-      "grad_norm": 4.804886817932129,
+      "grad_norm": 5.417207717895508,
       "learning_rate": 4.454454454454455e-07,
-      "loss": 0.8702,
+      "loss": 0.8649,
       "step": 95450
     },
     {
       "epoch": 286.67,
-      "grad_norm": 5.409758567810059,
+      "grad_norm": 5.5656256675720215,
       "learning_rate": 4.444444444444445e-07,
-      "loss": 0.8656,
+      "loss": 0.8725,
       "step": 95460
     },
     {
       "epoch": 286.7,
-      "grad_norm": 10.44924259185791,
+      "grad_norm": 7.273180961608887,
       "learning_rate": 4.434434434434435e-07,
-      "loss": 0.8849,
+      "loss": 0.879,
       "step": 95470
     },
     {
       "epoch": 286.73,
-      "grad_norm": 9.018890380859375,
+      "grad_norm": 10.221141815185547,
       "learning_rate": 4.4244244244244247e-07,
-      "loss": 0.8779,
+      "loss": 0.8899,
       "step": 95480
     },
     {
       "epoch": 286.76,
-      "grad_norm": 5.787815570831299,
+      "grad_norm": 7.235655784606934,
       "learning_rate": 4.414414414414415e-07,
-      "loss": 0.8761,
+      "loss": 0.88,
       "step": 95490
     },
     {
       "epoch": 286.79,
-      "grad_norm": 5.888273239135742,
+      "grad_norm": 7.28717565536499,
       "learning_rate": 4.4044044044044046e-07,
-      "loss": 0.889,
+      "loss": 0.888,
       "step": 95500
     },
     {
       "epoch": 286.82,
-      "grad_norm": 6.760576248168945,
+      "grad_norm": 5.479304313659668,
       "learning_rate": 4.394394394394395e-07,
       "loss": 0.9312,
       "step": 95510
     },
     {
       "epoch": 286.85,
-      "grad_norm": 6.9745073318481445,
+      "grad_norm": 7.05747127532959,
       "learning_rate": 4.3843843843843844e-07,
-      "loss": 0.8948,
+      "loss": 0.8891,
       "step": 95520
     },
     {
       "epoch": 286.88,
-      "grad_norm": 7.326995372772217,
+      "grad_norm": 6.746333122253418,
       "learning_rate": 4.3743743743743746e-07,
-      "loss": 0.8706,
+      "loss": 0.8657,
       "step": 95530
     },
     {
       "epoch": 286.91,
-      "grad_norm": 7.018818378448486,
+      "grad_norm": 6.577698230743408,
       "learning_rate": 4.364364364364364e-07,
-      "loss": 0.9301,
+      "loss": 0.9149,
       "step": 95540
     },
     {
       "epoch": 286.94,
-      "grad_norm": 5.421653747558594,
+      "grad_norm": 6.0160417556762695,
       "learning_rate": 4.3543543543543544e-07,
-      "loss": 0.9028,
+      "loss": 0.8946,
       "step": 95550
     },
     {
       "epoch": 286.97,
-      "grad_norm": 7.30696439743042,
+      "grad_norm": 7.539275169372559,
       "learning_rate": 4.344344344344345e-07,
-      "loss": 0.9457,
+      "loss": 0.9372,
       "step": 95560
     },
     {
       "epoch": 287.0,
-      "grad_norm": 7.556680679321289,
+      "grad_norm": 8.509405136108398,
       "learning_rate": 4.334334334334335e-07,
-      "loss": 0.8369,
+      "loss": 0.8373,
       "step": 95570
     },
     {
       "epoch": 287.0,
-      "eval_accuracy": 0.8288,
-      "eval_loss": 0.5688017010688782,
-      "eval_runtime": 5.5393,
-      "eval_samples_per_second": 1805.28,
-      "eval_steps_per_second": 7.221,
+      "eval_accuracy": 0.8286,
+      "eval_loss": 0.570774257183075,
+      "eval_runtime": 5.1716,
+      "eval_samples_per_second": 1933.62,
+      "eval_steps_per_second": 7.734,
       "step": 95571
     },
     {
       "epoch": 287.03,
-      "grad_norm": 8.331501960754395,
+      "grad_norm": 8.272597312927246,
       "learning_rate": 4.324324324324325e-07,
-      "loss": 1.2353,
+      "loss": 1.2567,
       "step": 95580
     },
     {
       "epoch": 287.06,
-      "grad_norm": 5.730729103088379,
+      "grad_norm": 5.175628185272217,
       "learning_rate": 4.3143143143143146e-07,
-      "loss": 0.9406,
+      "loss": 0.9551,
       "step": 95590
     },
     {
       "epoch": 287.09,
-      "grad_norm": 5.4140825271606445,
+      "grad_norm": 6.92178201675415,
       "learning_rate": 4.304304304304305e-07,
-      "loss": 0.8326,
+      "loss": 0.8245,
       "step": 95600
     },
     {
       "epoch": 287.12,
-      "grad_norm": 6.119048595428467,
+      "grad_norm": 6.833115100860596,
       "learning_rate": 4.294294294294295e-07,
-      "loss": 0.8564,
+      "loss": 0.8598,
       "step": 95610
     },
     {
       "epoch": 287.15,
-      "grad_norm": 7.526758670806885,
+      "grad_norm": 5.908480644226074,
       "learning_rate": 4.2842842842842846e-07,
-      "loss": 0.8823,
+      "loss": 0.8785,
       "step": 95620
     },
     {
       "epoch": 287.18,
-      "grad_norm": 6.373589515686035,
+      "grad_norm": 7.328498363494873,
       "learning_rate": 4.274274274274275e-07,
-      "loss": 0.8956,
+      "loss": 0.8935,
       "step": 95630
     },
     {
       "epoch": 287.21,
-      "grad_norm": 7.604051113128662,
+      "grad_norm": 8.014592170715332,
       "learning_rate": 4.2642642642642645e-07,
-      "loss": 0.8665,
+      "loss": 0.862,
       "step": 95640
     },
     {
       "epoch": 287.24,
-      "grad_norm": 5.50390100479126,
+      "grad_norm": 4.931750297546387,
       "learning_rate": 4.2542542542542547e-07,
-      "loss": 0.9424,
+      "loss": 0.9285,
       "step": 95650
     },
     {
       "epoch": 287.27,
-      "grad_norm": 6.4763593673706055,
+      "grad_norm": 7.388879776000977,
       "learning_rate": 4.2442442442442443e-07,
-      "loss": 0.8694,
+      "loss": 0.8556,
       "step": 95660
     },
     {
       "epoch": 287.3,
-      "grad_norm": 6.913318634033203,
+      "grad_norm": 5.964560031890869,
       "learning_rate": 4.2342342342342345e-07,
-      "loss": 0.9128,
+      "loss": 0.9132,
       "step": 95670
     },
     {
       "epoch": 287.33,
-      "grad_norm": 6.026805877685547,
+      "grad_norm": 6.719400405883789,
       "learning_rate": 4.224224224224224e-07,
-      "loss": 0.8991,
+      "loss": 0.8944,
       "step": 95680
     },
     {
       "epoch": 287.36,
-      "grad_norm": 9.47110366821289,
+      "grad_norm": 9.596646308898926,
       "learning_rate": 4.2142142142142143e-07,
-      "loss": 0.8403,
+      "loss": 0.8384,
       "step": 95690
     },
     {
       "epoch": 287.39,
-      "grad_norm": 9.567365646362305,
+      "grad_norm": 6.8760199546813965,
       "learning_rate": 4.204204204204205e-07,
-      "loss": 0.8095,
+      "loss": 0.8157,
       "step": 95700
     },
     {
       "epoch": 287.42,
-      "grad_norm": 8.807852745056152,
+      "grad_norm": 8.449357032775879,
       "learning_rate": 4.1941941941941947e-07,
-      "loss": 0.896,
+      "loss": 0.8907,
       "step": 95710
     },
     {
       "epoch": 287.45,
-      "grad_norm": 8.792410850524902,
+      "grad_norm": 10.792881965637207,
       "learning_rate": 4.184184184184185e-07,
-      "loss": 0.8832,
+      "loss": 0.8809,
       "step": 95720
     },
     {
       "epoch": 287.48,
-      "grad_norm": 6.459597587585449,
+      "grad_norm": 8.439273834228516,
       "learning_rate": 4.1741741741741745e-07,
-      "loss": 0.9326,
+      "loss": 0.9272,
       "step": 95730
     },
     {
       "epoch": 287.51,
-      "grad_norm": 6.334410190582275,
+      "grad_norm": 6.965673923492432,
       "learning_rate": 4.1641641641641647e-07,
-      "loss": 0.9131,
+      "loss": 0.917,
       "step": 95740
     },
     {
       "epoch": 287.54,
-      "grad_norm": 8.29755687713623,
+      "grad_norm": 6.500332355499268,
       "learning_rate": 4.1541541541541544e-07,
-      "loss": 0.9086,
+      "loss": 0.906,
       "step": 95750
     },
     {
       "epoch": 287.57,
-      "grad_norm": 7.801623821258545,
+      "grad_norm": 7.087130069732666,
       "learning_rate": 4.1441441441441446e-07,
-      "loss": 0.8904,
+      "loss": 0.9017,
       "step": 95760
     },
     {
       "epoch": 287.6,
-      "grad_norm": 7.1463942527771,
+      "grad_norm": 6.953995227813721,
       "learning_rate": 4.134134134134134e-07,
-      "loss": 0.8924,
+      "loss": 0.8972,
       "step": 95770
     },
     {
       "epoch": 287.63,
-      "grad_norm": 7.122136116027832,
+      "grad_norm": 8.784906387329102,
       "learning_rate": 4.1241241241241244e-07,
-      "loss": 0.984,
+      "loss": 0.9843,
       "step": 95780
     },
     {
       "epoch": 287.66,
-      "grad_norm": 4.875283241271973,
+      "grad_norm": 5.9809489250183105,
       "learning_rate": 4.114114114114114e-07,
-      "loss": 0.8991,
+      "loss": 0.8997,
       "step": 95790
     },
     {
       "epoch": 287.69,
-      "grad_norm": 6.142414093017578,
+      "grad_norm": 7.644022464752197,
       "learning_rate": 4.104104104104104e-07,
-      "loss": 0.8473,
+      "loss": 0.8438,
       "step": 95800
     },
     {
       "epoch": 287.72,
-      "grad_norm": 6.077850818634033,
+      "grad_norm": 9.654455184936523,
       "learning_rate": 4.0940940940940944e-07,
-      "loss": 0.9159,
+      "loss": 0.9171,
       "step": 95810
     },
     {
       "epoch": 287.75,
-      "grad_norm": 7.95081901550293,
+      "grad_norm": 7.469677925109863,
       "learning_rate": 4.084084084084084e-07,
-      "loss": 0.9639,
+      "loss": 0.9515,
       "step": 95820
     },
     {
       "epoch": 287.78,
-      "grad_norm": 7.451082229614258,
+      "grad_norm": 8.946952819824219,
       "learning_rate": 4.074074074074075e-07,
-      "loss": 0.9045,
+      "loss": 0.8961,
       "step": 95830
     },
     {
       "epoch": 287.81,
-      "grad_norm": 5.759469032287598,
+      "grad_norm": 9.530072212219238,
       "learning_rate": 4.0640640640640644e-07,
-      "loss": 0.9376,
+      "loss": 0.9468,
       "step": 95840
     },
     {
       "epoch": 287.84,
-      "grad_norm": 6.313235282897949,
+      "grad_norm": 6.3218584060668945,
       "learning_rate": 4.0540540540540546e-07,
-      "loss": 0.8888,
+      "loss": 0.8927,
       "step": 95850
     },
     {
       "epoch": 287.87,
-      "grad_norm": 5.4835524559021,
+      "grad_norm": 6.188301086425781,
       "learning_rate": 4.044044044044045e-07,
-      "loss": 0.914,
+      "loss": 0.9013,
       "step": 95860
     },
     {
       "epoch": 287.9,
-      "grad_norm": 6.343493938446045,
+      "grad_norm": 7.452801704406738,
       "learning_rate": 4.0340340340340345e-07,
-      "loss": 0.9672,
+      "loss": 0.9716,
       "step": 95870
     },
     {
       "epoch": 287.93,
-      "grad_norm": 6.0688605308532715,
+      "grad_norm": 5.399874687194824,
       "learning_rate": 4.0240240240240246e-07,
-      "loss": 0.9016,
+      "loss": 0.8951,
       "step": 95880
     },
     {
       "epoch": 287.96,
-      "grad_norm": 7.867003917694092,
+      "grad_norm": 9.677435874938965,
       "learning_rate": 4.0140140140140143e-07,
-      "loss": 0.9549,
+      "loss": 0.9445,
       "step": 95890
     },
     {
       "epoch": 287.99,
-      "grad_norm": 4.2346110343933105,
+      "grad_norm": 5.8136725425720215,
       "learning_rate": 4.0040040040040045e-07,
-      "loss": 0.9278,
+      "loss": 0.9203,
       "step": 95900
     },
     {
       "epoch": 288.0,
-      "eval_accuracy": 0.8287,
-      "eval_loss": 0.566037118434906,
-      "eval_runtime": 5.4573,
-      "eval_samples_per_second": 1832.398,
-      "eval_steps_per_second": 7.33,
+      "eval_accuracy": 0.8303,
+      "eval_loss": 0.5663820505142212,
+      "eval_runtime": 5.3722,
+      "eval_samples_per_second": 1861.444,
+      "eval_steps_per_second": 7.446,
       "step": 95904
     },
     {
       "epoch": 288.02,
-      "grad_norm": 5.90889835357666,
+      "grad_norm": 6.134928226470947,
       "learning_rate": 3.993993993993994e-07,
-      "loss": 1.0811,
+      "loss": 1.0568,
       "step": 95910
     },
     {
       "epoch": 288.05,
-      "grad_norm": 13.530159950256348,
+      "grad_norm": 15.262358665466309,
       "learning_rate": 3.9839839839839843e-07,
-      "loss": 0.8951,
+      "loss": 0.8953,
       "step": 95920
     },
     {
       "epoch": 288.08,
-      "grad_norm": 6.55112361907959,
+      "grad_norm": 6.471493721008301,
       "learning_rate": 3.973973973973974e-07,
-      "loss": 0.8885,
+      "loss": 0.8872,
       "step": 95930
     },
     {
       "epoch": 288.11,
-      "grad_norm": 6.379058361053467,
+      "grad_norm": 6.582591533660889,
       "learning_rate": 3.963963963963964e-07,
-      "loss": 0.8908,
+      "loss": 0.8773,
       "step": 95940
     },
     {
       "epoch": 288.14,
-      "grad_norm": 7.269066333770752,
+      "grad_norm": 6.2243266105651855,
       "learning_rate": 3.953953953953954e-07,
-      "loss": 0.8888,
+      "loss": 0.8928,
       "step": 95950
     },
     {
       "epoch": 288.17,
-      "grad_norm": 6.6817946434021,
+      "grad_norm": 7.676078796386719,
       "learning_rate": 3.943943943943944e-07,
-      "loss": 0.8294,
+      "loss": 0.8284,
       "step": 95960
     },
     {
       "epoch": 288.2,
-      "grad_norm": 6.489509582519531,
+      "grad_norm": 6.712618350982666,
       "learning_rate": 3.9339339339339347e-07,
-      "loss": 0.9469,
+      "loss": 0.9456,
       "step": 95970
     },
     {
       "epoch": 288.23,
-      "grad_norm": 7.520278453826904,
+      "grad_norm": 9.366049766540527,
       "learning_rate": 3.9239239239239244e-07,
-      "loss": 0.8696,
+      "loss": 0.8661,
       "step": 95980
     },
     {
       "epoch": 288.26,
-      "grad_norm": 6.550919055938721,
+      "grad_norm": 7.311176300048828,
       "learning_rate": 3.9139139139139145e-07,
-      "loss": 0.9027,
+      "loss": 0.9006,
       "step": 95990
     },
     {
       "epoch": 288.29,
-      "grad_norm": 7.760721206665039,
+      "grad_norm": 6.984836101531982,
       "learning_rate": 3.903903903903904e-07,
-      "loss": 0.9225,
+      "loss": 0.9169,
       "step": 96000
     },
     {
       "epoch": 288.32,
-      "grad_norm": 5.067615509033203,
+      "grad_norm": 5.307101249694824,
       "learning_rate": 3.8938938938938944e-07,
-      "loss": 0.9176,
+      "loss": 0.9202,
       "step": 96010
     },
     {
       "epoch": 288.35,
-      "grad_norm": 6.992642402648926,
+      "grad_norm": 9.491991996765137,
       "learning_rate": 3.883883883883884e-07,
-      "loss": 0.9643,
+      "loss": 0.9731,
       "step": 96020
     },
     {
       "epoch": 288.38,
-      "grad_norm": 5.9921698570251465,
+      "grad_norm": 5.471489906311035,
       "learning_rate": 3.873873873873874e-07,
-      "loss": 0.9155,
+      "loss": 0.9207,
       "step": 96030
     },
     {
       "epoch": 288.41,
-      "grad_norm": 4.904808521270752,
+      "grad_norm": 5.188915729522705,
       "learning_rate": 3.863863863863864e-07,
-      "loss": 0.8182,
+      "loss": 0.8189,
       "step": 96040
     },
     {
       "epoch": 288.44,
-      "grad_norm": 5.2711358070373535,
+      "grad_norm": 5.211928844451904,
       "learning_rate": 3.853853853853854e-07,
-      "loss": 0.9243,
+      "loss": 0.9277,
       "step": 96050
     },
     {
       "epoch": 288.47,
-      "grad_norm": 5.620598316192627,
+      "grad_norm": 6.076613426208496,
       "learning_rate": 3.843843843843844e-07,
-      "loss": 0.884,
+      "loss": 0.8841,
       "step": 96060
     },
     {
       "epoch": 288.5,
-      "grad_norm": 5.275291919708252,
+      "grad_norm": 5.310138702392578,
       "learning_rate": 3.833833833833834e-07,
-      "loss": 0.8387,
+      "loss": 0.8206,
       "step": 96070
     },
     {
       "epoch": 288.53,
-      "grad_norm": 7.374592304229736,
+      "grad_norm": 8.373382568359375,
       "learning_rate": 3.823823823823824e-07,
-      "loss": 0.8905,
+      "loss": 0.8917,
       "step": 96080
     },
     {
       "epoch": 288.56,
-      "grad_norm": 11.533805847167969,
+      "grad_norm": 8.570422172546387,
       "learning_rate": 3.8138138138138137e-07,
-      "loss": 0.9141,
+      "loss": 0.903,
       "step": 96090
     },
     {
       "epoch": 288.59,
-      "grad_norm": 6.758476734161377,
+      "grad_norm": 6.9485955238342285,
       "learning_rate": 3.8038038038038044e-07,
-      "loss": 0.949,
+      "loss": 0.946,
       "step": 96100
     },
     {
       "epoch": 288.62,
-      "grad_norm": 7.791934490203857,
+      "grad_norm": 5.8270158767700195,
       "learning_rate": 3.793793793793794e-07,
-      "loss": 0.8578,
+      "loss": 0.8467,
       "step": 96110
     },
     {
       "epoch": 288.65,
-      "grad_norm": 5.405816555023193,
+      "grad_norm": 6.4875359535217285,
       "learning_rate": 3.7837837837837843e-07,
-      "loss": 0.9446,
+      "loss": 0.9519,
       "step": 96120
     },
     {
       "epoch": 288.68,
-      "grad_norm": 6.810920238494873,
+      "grad_norm": 9.41608715057373,
       "learning_rate": 3.7737737737737745e-07,
-      "loss": 0.8713,
+      "loss": 0.8662,
       "step": 96130
     },
     {
       "epoch": 288.71,
-      "grad_norm": 10.678221702575684,
+      "grad_norm": 7.97383451461792,
       "learning_rate": 3.763763763763764e-07,
-      "loss": 0.9498,
+      "loss": 0.9506,
       "step": 96140
     },
     {
       "epoch": 288.74,
-      "grad_norm": 7.311709403991699,
+      "grad_norm": 8.50930404663086,
       "learning_rate": 3.7537537537537543e-07,
-      "loss": 0.8823,
+      "loss": 0.8879,
       "step": 96150
     },
     {
       "epoch": 288.77,
-      "grad_norm": 7.497401714324951,
+      "grad_norm": 7.612275123596191,
       "learning_rate": 3.743743743743744e-07,
-      "loss": 0.9104,
+      "loss": 0.9157,
       "step": 96160
     },
     {
       "epoch": 288.8,
-      "grad_norm": 5.681319236755371,
+      "grad_norm": 6.595509052276611,
       "learning_rate": 3.733733733733734e-07,
-      "loss": 0.8978,
+      "loss": 0.8938,
       "step": 96170
     },
     {
       "epoch": 288.83,
-      "grad_norm": 8.420674324035645,
+      "grad_norm": 7.333737850189209,
       "learning_rate": 3.723723723723724e-07,
-      "loss": 0.8326,
+      "loss": 0.8357,
       "step": 96180
     },
     {
       "epoch": 288.86,
-      "grad_norm": 9.911273002624512,
+      "grad_norm": 9.854785919189453,
       "learning_rate": 3.713713713713714e-07,
-      "loss": 0.901,
+      "loss": 0.891,
       "step": 96190
     },
     {
       "epoch": 288.89,
-      "grad_norm": 6.59670877456665,
+      "grad_norm": 7.83756685256958,
       "learning_rate": 3.7037037037037036e-07,
-      "loss": 0.8906,
+      "loss": 0.8797,
       "step": 96200
     },
     {
       "epoch": 288.92,
-      "grad_norm": 5.7453155517578125,
+      "grad_norm": 5.359286308288574,
       "learning_rate": 3.693693693693694e-07,
-      "loss": 0.9816,
+      "loss": 0.9686,
       "step": 96210
     },
     {
       "epoch": 288.95,
-      "grad_norm": 6.713436603546143,
+      "grad_norm": 8.173161506652832,
       "learning_rate": 3.6836836836836835e-07,
-      "loss": 0.8947,
+      "loss": 0.8875,
       "step": 96220
     },
     {
       "epoch": 288.98,
-      "grad_norm": 10.41745662689209,
+      "grad_norm": 11.54624080657959,
       "learning_rate": 3.673673673673674e-07,
-      "loss": 0.9285,
+      "loss": 0.9336,
       "step": 96230
     },
     {
       "epoch": 289.0,
-      "eval_accuracy": 0.8285,
-      "eval_loss": 0.5706514716148376,
-      "eval_runtime": 5.6835,
-      "eval_samples_per_second": 1759.477,
-      "eval_steps_per_second": 7.038,
+      "eval_accuracy": 0.8293,
+      "eval_loss": 0.5706751942634583,
+      "eval_runtime": 5.3086,
+      "eval_samples_per_second": 1883.721,
+      "eval_steps_per_second": 7.535,
       "step": 96237
     },
     {
       "epoch": 289.01,
-      "grad_norm": 6.5139875411987305,
+      "grad_norm": 7.957648277282715,
       "learning_rate": 3.6636636636636644e-07,
-      "loss": 1.0329,
+      "loss": 1.0113,
       "step": 96240
     },
     {
       "epoch": 289.04,
-      "grad_norm": 8.677268981933594,
+      "grad_norm": 8.971043586730957,
       "learning_rate": 3.653653653653654e-07,
-      "loss": 0.9002,
+      "loss": 0.8925,
       "step": 96250
     },
     {
       "epoch": 289.07,
-      "grad_norm": 9.230862617492676,
+      "grad_norm": 8.563596725463867,
       "learning_rate": 3.643643643643644e-07,
-      "loss": 0.9295,
+      "loss": 0.9262,
       "step": 96260
     },
     {
       "epoch": 289.1,
-      "grad_norm": 6.869067668914795,
+      "grad_norm": 10.040695190429688,
       "learning_rate": 3.633633633633634e-07,
-      "loss": 0.9079,
+      "loss": 0.903,
       "step": 96270
     },
     {
       "epoch": 289.13,
-      "grad_norm": 9.748868942260742,
+      "grad_norm": 11.02414608001709,
       "learning_rate": 3.623623623623624e-07,
-      "loss": 0.8787,
+      "loss": 0.8843,
       "step": 96280
     },
     {
       "epoch": 289.16,
-      "grad_norm": 6.382780075073242,
+      "grad_norm": 9.248641967773438,
       "learning_rate": 3.6136136136136137e-07,
-      "loss": 0.8671,
+      "loss": 0.8674,
       "step": 96290
     },
     {
       "epoch": 289.19,
-      "grad_norm": 5.203657627105713,
+      "grad_norm": 7.297589302062988,
       "learning_rate": 3.603603603603604e-07,
-      "loss": 0.8791,
+      "loss": 0.8698,
       "step": 96300
     },
     {
       "epoch": 289.22,
-      "grad_norm": 9.175958633422852,
+      "grad_norm": 9.760224342346191,
       "learning_rate": 3.593593593593594e-07,
-      "loss": 0.8353,
+      "loss": 0.8373,
       "step": 96310
     },
     {
       "epoch": 289.25,
-      "grad_norm": 4.99818229675293,
+      "grad_norm": 8.584593772888184,
       "learning_rate": 3.5835835835835837e-07,
-      "loss": 0.9239,
+      "loss": 0.9359,
       "step": 96320
     },
     {
       "epoch": 289.28,
-      "grad_norm": 8.07487678527832,
+      "grad_norm": 8.70361042022705,
       "learning_rate": 3.573573573573574e-07,
-      "loss": 0.888,
+      "loss": 0.883,
       "step": 96330
     },
     {
       "epoch": 289.31,
-      "grad_norm": 8.528456687927246,
+      "grad_norm": 6.9436726570129395,
       "learning_rate": 3.5635635635635636e-07,
-      "loss": 0.9352,
+      "loss": 0.9331,
       "step": 96340
     },
     {
       "epoch": 289.34,
-      "grad_norm": 7.363620758056641,
+      "grad_norm": 7.815725803375244,
       "learning_rate": 3.553553553553554e-07,
-      "loss": 0.8805,
+      "loss": 0.8864,
       "step": 96350
     },
     {
       "epoch": 289.37,
-      "grad_norm": 7.7861785888671875,
+      "grad_norm": 9.40410327911377,
       "learning_rate": 3.5435435435435434e-07,
       "loss": 0.8482,
       "step": 96360
     },
     {
       "epoch": 289.4,
-      "grad_norm": 6.139326095581055,
+      "grad_norm": 6.962365627288818,
       "learning_rate": 3.533533533533534e-07,
-      "loss": 0.8426,
+      "loss": 0.8494,
       "step": 96370
     },
     {
       "epoch": 289.43,
-      "grad_norm": 6.51355504989624,
+      "grad_norm": 7.422054767608643,
       "learning_rate": 3.5235235235235243e-07,
-      "loss": 0.9138,
+      "loss": 0.916,
       "step": 96380
     },
     {
       "epoch": 289.46,
-      "grad_norm": 5.9675140380859375,
+      "grad_norm": 5.481865882873535,
       "learning_rate": 3.513513513513514e-07,
-      "loss": 0.8644,
+      "loss": 0.8626,
       "step": 96390
     },
     {
       "epoch": 289.49,
-      "grad_norm": 5.845391750335693,
+      "grad_norm": 6.450154781341553,
       "learning_rate": 3.503503503503504e-07,
-      "loss": 0.9347,
+      "loss": 0.929,
       "step": 96400
     },
     {
       "epoch": 289.52,
-      "grad_norm": 5.146974563598633,
+      "grad_norm": 6.277951240539551,
       "learning_rate": 3.493493493493494e-07,
-      "loss": 0.9552,
+      "loss": 0.9573,
       "step": 96410
     },
     {
       "epoch": 289.55,
-      "grad_norm": 9.40829849243164,
+      "grad_norm": 9.105733871459961,
       "learning_rate": 3.483483483483484e-07,
-      "loss": 0.9211,
+      "loss": 0.9241,
       "step": 96420
     },
     {
       "epoch": 289.58,
-      "grad_norm": 7.1239471435546875,
+      "grad_norm": 7.685225963592529,
       "learning_rate": 3.4734734734734736e-07,
-      "loss": 0.9349,
+      "loss": 0.9321,
       "step": 96430
     },
     {
       "epoch": 289.61,
-      "grad_norm": 7.842417240142822,
+      "grad_norm": 7.97860050201416,
       "learning_rate": 3.463463463463464e-07,
-      "loss": 0.8781,
+      "loss": 0.8811,
       "step": 96440
     },
     {
       "epoch": 289.64,
-      "grad_norm": 5.093599796295166,
+      "grad_norm": 5.728398323059082,
       "learning_rate": 3.4534534534534535e-07,
-      "loss": 0.8906,
+      "loss": 0.8765,
       "step": 96450
     },
     {
       "epoch": 289.67,
-      "grad_norm": 5.135491371154785,
+      "grad_norm": 5.9992780685424805,
       "learning_rate": 3.4434434434434436e-07,
-      "loss": 0.9329,
+      "loss": 0.9395,
       "step": 96460
     },
     {
       "epoch": 289.7,
-      "grad_norm": 8.590882301330566,
+      "grad_norm": 6.856700420379639,
       "learning_rate": 3.4334334334334333e-07,
-      "loss": 0.883,
+      "loss": 0.8758,
       "step": 96470
     },
     {
       "epoch": 289.73,
-      "grad_norm": 9.05753231048584,
+      "grad_norm": 10.55618667602539,
       "learning_rate": 3.4234234234234235e-07,
-      "loss": 0.9269,
+      "loss": 0.9296,
       "step": 96480
     },
     {
       "epoch": 289.76,
-      "grad_norm": 6.46863317489624,
+      "grad_norm": 6.6747145652771,
       "learning_rate": 3.413413413413413e-07,
-      "loss": 0.8551,
+      "loss": 0.8516,
       "step": 96490
     },
     {
       "epoch": 289.79,
-      "grad_norm": 9.479883193969727,
+      "grad_norm": 6.6967453956604,
       "learning_rate": 3.403403403403404e-07,
-      "loss": 0.9184,
+      "loss": 0.912,
       "step": 96500
     },
     {
       "epoch": 289.82,
-      "grad_norm": 6.234785556793213,
+      "grad_norm": 6.001014709472656,
       "learning_rate": 3.393393393393394e-07,
-      "loss": 0.8858,
+      "loss": 0.8806,
       "step": 96510
     },
     {
       "epoch": 289.85,
-      "grad_norm": 7.543824195861816,
+      "grad_norm": 7.262793064117432,
       "learning_rate": 3.3833833833833837e-07,
-      "loss": 0.8998,
+      "loss": 0.895,
       "step": 96520
     },
     {
       "epoch": 289.88,
-      "grad_norm": 6.384433269500732,
+      "grad_norm": 7.002352714538574,
       "learning_rate": 3.373373373373374e-07,
-      "loss": 0.9534,
+      "loss": 0.9427,
       "step": 96530
     },
     {
       "epoch": 289.91,
-      "grad_norm": 6.736269474029541,
+      "grad_norm": 8.610376358032227,
       "learning_rate": 3.3633633633633635e-07,
-      "loss": 0.85,
+      "loss": 0.8476,
       "step": 96540
     },
     {
       "epoch": 289.94,
-      "grad_norm": 6.046384334564209,
+      "grad_norm": 6.9864044189453125,
       "learning_rate": 3.3533533533533537e-07,
-      "loss": 0.8555,
+      "loss": 0.8521,
       "step": 96550
     },
     {
       "epoch": 289.97,
-      "grad_norm": 6.087065696716309,
+      "grad_norm": 7.792603015899658,
       "learning_rate": 3.343343343343344e-07,
-      "loss": 0.8936,
+      "loss": 0.8888,
       "step": 96560
     },
     {
       "epoch": 290.0,
-      "grad_norm": 23.689327239990234,
+      "grad_norm": 27.83856964111328,
       "learning_rate": 3.3333333333333335e-07,
-      "loss": 0.9858,
+      "loss": 1.0155,
       "step": 96570
     },
     {
       "epoch": 290.0,
-      "eval_accuracy": 0.829,
-      "eval_loss": 0.566765546798706,
-      "eval_runtime": 5.551,
-      "eval_samples_per_second": 1801.469,
-      "eval_steps_per_second": 7.206,
+      "eval_accuracy": 0.8326,
+      "eval_loss": 0.5661998391151428,
+      "eval_runtime": 5.2027,
+      "eval_samples_per_second": 1922.064,
+      "eval_steps_per_second": 7.688,
       "step": 96570
     },
     {
       "epoch": 290.03,
-      "grad_norm": 5.162928581237793,
+      "grad_norm": 5.918946743011475,
       "learning_rate": 3.3233233233233237e-07,
-      "loss": 0.9337,
+      "loss": 0.929,
       "step": 96580
     },
     {
       "epoch": 290.06,
-      "grad_norm": 8.403775215148926,
+      "grad_norm": 10.902996063232422,
       "learning_rate": 3.3133133133133134e-07,
-      "loss": 0.8657,
+      "loss": 0.8619,
       "step": 96590
     },
     {
       "epoch": 290.09,
-      "grad_norm": 6.613644123077393,
+      "grad_norm": 7.742284297943115,
       "learning_rate": 3.3033033033033036e-07,
-      "loss": 0.9149,
+      "loss": 0.9071,
       "step": 96600
     },
     {
       "epoch": 290.12,
-      "grad_norm": 5.358506679534912,
+      "grad_norm": 5.691558837890625,
       "learning_rate": 3.293293293293293e-07,
-      "loss": 0.8728,
+      "loss": 0.8843,
       "step": 96610
     },
     {
       "epoch": 290.15,
-      "grad_norm": 6.320145606994629,
+      "grad_norm": 5.593356132507324,
       "learning_rate": 3.2832832832832834e-07,
-      "loss": 0.9508,
+      "loss": 0.9453,
       "step": 96620
     },
     {
       "epoch": 290.18,
-      "grad_norm": 6.831081867218018,
+      "grad_norm": 6.052535533905029,
       "learning_rate": 3.273273273273274e-07,
-      "loss": 0.8831,
+      "loss": 0.8841,
       "step": 96630
     },
     {
       "epoch": 290.21,
-      "grad_norm": 5.515520095825195,
+      "grad_norm": 6.034413814544678,
       "learning_rate": 3.263263263263264e-07,
-      "loss": 0.8912,
+      "loss": 0.8882,
       "step": 96640
     },
     {
       "epoch": 290.24,
-      "grad_norm": 7.101522922515869,
+      "grad_norm": 6.070439338684082,
       "learning_rate": 3.253253253253254e-07,
-      "loss": 0.8776,
+      "loss": 0.8724,
       "step": 96650
     },
     {
       "epoch": 290.27,
-      "grad_norm": 6.713544845581055,
+      "grad_norm": 9.742959022521973,
       "learning_rate": 3.2432432432432436e-07,
-      "loss": 0.9023,
+      "loss": 0.9005,
       "step": 96660
     },
     {
       "epoch": 290.3,
-      "grad_norm": 6.2447028160095215,
+      "grad_norm": 7.215478420257568,
       "learning_rate": 3.233233233233234e-07,
-      "loss": 0.9655,
+      "loss": 0.9548,
       "step": 96670
     },
     {
       "epoch": 290.33,
-      "grad_norm": 5.59855842590332,
+      "grad_norm": 4.704133033752441,
       "learning_rate": 3.2232232232232234e-07,
-      "loss": 0.9115,
+      "loss": 0.9158,
       "step": 96680
     },
     {
       "epoch": 290.36,
-      "grad_norm": 9.983766555786133,
+      "grad_norm": 11.349839210510254,
       "learning_rate": 3.2132132132132136e-07,
-      "loss": 0.878,
+      "loss": 0.8705,
       "step": 96690
     },
     {
       "epoch": 290.39,
-      "grad_norm": 8.553966522216797,
+      "grad_norm": 9.029533386230469,
       "learning_rate": 3.2032032032032033e-07,
-      "loss": 0.9443,
+      "loss": 0.9405,
       "step": 96700
     },
     {
       "epoch": 290.42,
-      "grad_norm": 5.969021320343018,
+      "grad_norm": 6.143709182739258,
       "learning_rate": 3.1931931931931935e-07,
-      "loss": 0.9245,
+      "loss": 0.9344,
       "step": 96710
     },
     {
       "epoch": 290.45,
-      "grad_norm": 6.258532524108887,
+      "grad_norm": 7.200097560882568,
       "learning_rate": 3.183183183183183e-07,
-      "loss": 0.922,
+      "loss": 0.9271,
       "step": 96720
     },
     {
       "epoch": 290.48,
-      "grad_norm": 4.38950777053833,
+      "grad_norm": 4.452569961547852,
       "learning_rate": 3.1731731731731733e-07,
-      "loss": 0.9151,
+      "loss": 0.9143,
       "step": 96730
     },
     {
       "epoch": 290.51,
-      "grad_norm": 6.5272603034973145,
+      "grad_norm": 5.745941162109375,
       "learning_rate": 3.163163163163163e-07,
-      "loss": 0.9199,
+      "loss": 0.9252,
       "step": 96740
     },
     {
       "epoch": 290.54,
-      "grad_norm": 5.739402770996094,
+      "grad_norm": 5.607093334197998,
       "learning_rate": 3.153153153153153e-07,
-      "loss": 0.9333,
+      "loss": 0.9303,
       "step": 96750
     },
     {
       "epoch": 290.57,
-      "grad_norm": 8.744824409484863,
+      "grad_norm": 7.52081823348999,
       "learning_rate": 3.1431431431431433e-07,
-      "loss": 0.92,
+      "loss": 0.9312,
       "step": 96760
     },
     {
       "epoch": 290.6,
-      "grad_norm": 5.053074836730957,
+      "grad_norm": 4.780002593994141,
       "learning_rate": 3.1331331331331335e-07,
-      "loss": 0.8503,
+      "loss": 0.8546,
       "step": 96770
     },
     {
       "epoch": 290.63,
-      "grad_norm": 4.2511067390441895,
+      "grad_norm": 5.647804260253906,
       "learning_rate": 3.123123123123123e-07,
-      "loss": 0.9045,
+      "loss": 0.9008,
       "step": 96780
     },
     {
       "epoch": 290.66,
-      "grad_norm": 6.175032615661621,
+      "grad_norm": 9.184822082519531,
       "learning_rate": 3.1131131131131133e-07,
-      "loss": 0.9191,
+      "loss": 0.9164,
       "step": 96790
     },
     {
       "epoch": 290.69,
-      "grad_norm": 7.6070661544799805,
+      "grad_norm": 12.028594017028809,
       "learning_rate": 3.1031031031031035e-07,
-      "loss": 0.8809,
+      "loss": 0.8882,
       "step": 96800
     },
     {
       "epoch": 290.72,
-      "grad_norm": 6.924339294433594,
+      "grad_norm": 7.166258335113525,
       "learning_rate": 3.0930930930930937e-07,
-      "loss": 0.8894,
+      "loss": 0.8872,
       "step": 96810
     },
     {
       "epoch": 290.75,
-      "grad_norm": 8.669384956359863,
+      "grad_norm": 7.572995662689209,
       "learning_rate": 3.0830830830830834e-07,
-      "loss": 0.8928,
+      "loss": 0.8875,
       "step": 96820
     },
     {
       "epoch": 290.78,
-      "grad_norm": 5.394067764282227,
+      "grad_norm": 7.300982475280762,
       "learning_rate": 3.0730730730730735e-07,
-      "loss": 0.951,
+      "loss": 0.9517,
       "step": 96830
     },
     {
       "epoch": 290.81,
-      "grad_norm": 7.578121185302734,
+      "grad_norm": 7.535978317260742,
       "learning_rate": 3.063063063063063e-07,
-      "loss": 1.0016,
+      "loss": 0.9879,
       "step": 96840
     },
     {
       "epoch": 290.84,
-      "grad_norm": 5.974545955657959,
+      "grad_norm": 9.00899887084961,
       "learning_rate": 3.0530530530530534e-07,
-      "loss": 0.9112,
+      "loss": 0.9004,
       "step": 96850
     },
     {
       "epoch": 290.87,
-      "grad_norm": 7.124532699584961,
+      "grad_norm": 8.033156394958496,
       "learning_rate": 3.043043043043043e-07,
-      "loss": 0.9407,
+      "loss": 0.9564,
       "step": 96860
     },
     {
       "epoch": 290.9,
-      "grad_norm": 5.388976097106934,
+      "grad_norm": 5.550476551055908,
       "learning_rate": 3.033033033033033e-07,
-      "loss": 0.8244,
+      "loss": 0.8375,
       "step": 96870
     },
     {
       "epoch": 290.93,
-      "grad_norm": 6.682571887969971,
+      "grad_norm": 7.688784122467041,
       "learning_rate": 3.0230230230230234e-07,
-      "loss": 0.9325,
+      "loss": 0.9336,
       "step": 96880
     },
     {
       "epoch": 290.96,
-      "grad_norm": 7.258574962615967,
+      "grad_norm": 7.634803771972656,
       "learning_rate": 3.013013013013013e-07,
-      "loss": 0.8862,
+      "loss": 0.8793,
       "step": 96890
     },
     {
       "epoch": 290.99,
-      "grad_norm": 5.891548156738281,
+      "grad_norm": 6.952076435089111,
       "learning_rate": 3.003003003003003e-07,
-      "loss": 0.8783,
+      "loss": 0.8775,
       "step": 96900
     },
     {
       "epoch": 291.0,
-      "eval_accuracy": 0.8261,
-      "eval_loss": 0.5781397819519043,
-      "eval_runtime": 5.7365,
-      "eval_samples_per_second": 1743.223,
-      "eval_steps_per_second": 6.973,
+      "eval_accuracy": 0.8267,
+      "eval_loss": 0.5773621201515198,
+      "eval_runtime": 5.1385,
+      "eval_samples_per_second": 1946.078,
+      "eval_steps_per_second": 7.784,
       "step": 96903
     },
     {
       "epoch": 291.02,
-      "grad_norm": 6.4347734451293945,
+      "grad_norm": 6.368497848510742,
       "learning_rate": 2.9929929929929934e-07,
-      "loss": 0.9335,
+      "loss": 0.9415,
       "step": 96910
     },
     {
       "epoch": 291.05,
-      "grad_norm": 6.039548397064209,
+      "grad_norm": 6.110110282897949,
       "learning_rate": 2.982982982982983e-07,
-      "loss": 0.9455,
+      "loss": 0.9369,
       "step": 96920
     },
     {
       "epoch": 291.08,
-      "grad_norm": 5.672606945037842,
+      "grad_norm": 5.988052845001221,
       "learning_rate": 2.972972972972973e-07,
-      "loss": 0.889,
+      "loss": 0.8889,
       "step": 96930
     },
     {
       "epoch": 291.11,
-      "grad_norm": 7.051268100738525,
+      "grad_norm": 6.609304904937744,
       "learning_rate": 2.9629629629629634e-07,
-      "loss": 0.873,
+      "loss": 0.8705,
       "step": 96940
     },
     {
       "epoch": 291.14,
-      "grad_norm": 6.605191707611084,
+      "grad_norm": 6.615848064422607,
       "learning_rate": 2.952952952952953e-07,
-      "loss": 0.8634,
+      "loss": 0.8633,
       "step": 96950
     },
     {
       "epoch": 291.17,
-      "grad_norm": 7.048513412475586,
+      "grad_norm": 18.798877716064453,
       "learning_rate": 2.9429429429429433e-07,
-      "loss": 0.8535,
+      "loss": 0.8624,
       "step": 96960
     },
     {
       "epoch": 291.2,
-      "grad_norm": 7.626347064971924,
+      "grad_norm": 7.812747478485107,
       "learning_rate": 2.932932932932933e-07,
-      "loss": 0.8257,
+      "loss": 0.8192,
       "step": 96970
     },
     {
       "epoch": 291.23,
-      "grad_norm": 10.838584899902344,
+      "grad_norm": 8.50556468963623,
       "learning_rate": 2.922922922922923e-07,
-      "loss": 0.8837,
+      "loss": 0.8918,
       "step": 96980
     },
     {
       "epoch": 291.26,
-      "grad_norm": 7.726558208465576,
+      "grad_norm": 7.239748954772949,
       "learning_rate": 2.912912912912913e-07,
-      "loss": 0.9046,
+      "loss": 0.901,
       "step": 96990
     },
     {
       "epoch": 291.29,
-      "grad_norm": 7.2872843742370605,
+      "grad_norm": 7.468454360961914,
       "learning_rate": 2.9029029029029035e-07,
-      "loss": 0.8348,
+      "loss": 0.8406,
       "step": 97000
     },
     {
       "epoch": 291.32,
-      "grad_norm": 7.552936553955078,
+      "grad_norm": 4.89129638671875,
       "learning_rate": 2.892892892892893e-07,
-      "loss": 0.9721,
+      "loss": 0.9719,
       "step": 97010
     },
     {
       "epoch": 291.35,
-      "grad_norm": 5.111980438232422,
+      "grad_norm": 6.64334774017334,
       "learning_rate": 2.8828828828828833e-07,
-      "loss": 0.8843,
+      "loss": 0.878,
       "step": 97020
     },
     {
       "epoch": 291.38,
-      "grad_norm": 4.969394683837891,
+      "grad_norm": 5.868730068206787,
       "learning_rate": 2.872872872872873e-07,
-      "loss": 0.9012,
+      "loss": 0.8962,
       "step": 97030
     },
     {
       "epoch": 291.41,
-      "grad_norm": 10.907661437988281,
+      "grad_norm": 11.010114669799805,
       "learning_rate": 2.862862862862863e-07,
-      "loss": 0.9066,
+      "loss": 0.9096,
       "step": 97040
     },
     {
       "epoch": 291.44,
-      "grad_norm": 5.135157108306885,
+      "grad_norm": 5.149742603302002,
       "learning_rate": 2.852852852852853e-07,
-      "loss": 0.9544,
+      "loss": 0.9487,
       "step": 97050
     },
     {
       "epoch": 291.47,
-      "grad_norm": 6.66072940826416,
+      "grad_norm": 6.375172138214111,
       "learning_rate": 2.842842842842843e-07,
-      "loss": 0.91,
+      "loss": 0.9072,
       "step": 97060
     },
     {
       "epoch": 291.5,
-      "grad_norm": 9.379136085510254,
+      "grad_norm": 7.5333428382873535,
       "learning_rate": 2.832832832832833e-07,
-      "loss": 0.7825,
+      "loss": 0.7842,
       "step": 97070
     },
     {
       "epoch": 291.53,
-      "grad_norm": 5.607841968536377,
+      "grad_norm": 6.653573036193848,
       "learning_rate": 2.8228228228228234e-07,
-      "loss": 1.06,
+      "loss": 1.066,
       "step": 97080
     },
     {
       "epoch": 291.56,
-      "grad_norm": 5.467153549194336,
+      "grad_norm": 5.846202850341797,
       "learning_rate": 2.812812812812813e-07,
-      "loss": 0.9055,
+      "loss": 0.91,
       "step": 97090
     },
     {
       "epoch": 291.59,
-      "grad_norm": 5.527655601501465,
+      "grad_norm": 5.534663677215576,
       "learning_rate": 2.802802802802803e-07,
-      "loss": 0.8786,
+      "loss": 0.8803,
       "step": 97100
     },
     {
       "epoch": 291.62,
-      "grad_norm": 8.38955020904541,
+      "grad_norm": 9.892864227294922,
       "learning_rate": 2.792792792792793e-07,
-      "loss": 0.9086,
+      "loss": 0.8999,
       "step": 97110
     },
     {
       "epoch": 291.65,
-      "grad_norm": 7.306517601013184,
+      "grad_norm": 6.589744567871094,
       "learning_rate": 2.782782782782783e-07,
-      "loss": 0.9205,
+      "loss": 0.9186,
       "step": 97120
     },
     {
       "epoch": 291.68,
-      "grad_norm": 4.722521781921387,
+      "grad_norm": 7.066751956939697,
       "learning_rate": 2.772772772772773e-07,
-      "loss": 0.8469,
+      "loss": 0.8447,
       "step": 97130
     },
     {
       "epoch": 291.71,
-      "grad_norm": 7.043709754943848,
+      "grad_norm": 9.972186088562012,
       "learning_rate": 2.762762762762763e-07,
-      "loss": 0.8575,
+      "loss": 0.8564,
       "step": 97140
     },
     {
       "epoch": 291.74,
-      "grad_norm": 7.658958435058594,
+      "grad_norm": 13.006203651428223,
       "learning_rate": 2.752752752752753e-07,
-      "loss": 0.925,
+      "loss": 0.9136,
       "step": 97150
     },
     {
       "epoch": 291.77,
-      "grad_norm": 5.728487491607666,
+      "grad_norm": 5.598698139190674,
       "learning_rate": 2.742742742742743e-07,
-      "loss": 0.8471,
+      "loss": 0.8393,
       "step": 97160
     },
     {
       "epoch": 291.8,
-      "grad_norm": 5.268272876739502,
+      "grad_norm": 6.140029430389404,
       "learning_rate": 2.732732732732733e-07,
-      "loss": 0.8537,
+      "loss": 0.8542,
       "step": 97170
     },
     {
       "epoch": 291.83,
-      "grad_norm": 4.934046745300293,
+      "grad_norm": 6.267948627471924,
       "learning_rate": 2.722722722722723e-07,
-      "loss": 0.9144,
+      "loss": 0.916,
       "step": 97180
     },
     {
       "epoch": 291.86,
-      "grad_norm": 6.57564115524292,
+      "grad_norm": 9.023906707763672,
       "learning_rate": 2.7127127127127127e-07,
-      "loss": 0.9253,
+      "loss": 0.9363,
       "step": 97190
     },
     {
       "epoch": 291.89,
-      "grad_norm": 7.4078145027160645,
+      "grad_norm": 5.515810489654541,
       "learning_rate": 2.702702702702703e-07,
-      "loss": 0.9117,
+      "loss": 0.9087,
       "step": 97200
     },
     {
       "epoch": 291.92,
-      "grad_norm": 6.532509803771973,
+      "grad_norm": 10.288102149963379,
       "learning_rate": 2.692692692692693e-07,
-      "loss": 0.8843,
+      "loss": 0.8904,
       "step": 97210
     },
     {
       "epoch": 291.95,
-      "grad_norm": 7.081875801086426,
+      "grad_norm": 7.3626532554626465,
       "learning_rate": 2.682682682682683e-07,
-      "loss": 0.8899,
+      "loss": 0.8972,
       "step": 97220
     },
     {
       "epoch": 291.98,
-      "grad_norm": 6.4939470291137695,
+      "grad_norm": 8.550332069396973,
       "learning_rate": 2.672672672672673e-07,
-      "loss": 0.9269,
+      "loss": 0.9309,
       "step": 97230
     },
     {
       "epoch": 292.0,
-      "eval_accuracy": 0.8277,
-      "eval_loss": 0.5680537819862366,
-      "eval_runtime": 5.4735,
-      "eval_samples_per_second": 1826.987,
-      "eval_steps_per_second": 7.308,
+      "eval_accuracy": 0.8305,
+      "eval_loss": 0.5701309442520142,
+      "eval_runtime": 5.3899,
+      "eval_samples_per_second": 1855.318,
+      "eval_steps_per_second": 7.421,
       "step": 97236
     },
     {
       "epoch": 292.01,
-      "grad_norm": 6.426619529724121,
+      "grad_norm": 6.01282262802124,
       "learning_rate": 2.6626626626626626e-07,
-      "loss": 1.0911,
+      "loss": 1.0722,
       "step": 97240
     },
     {
       "epoch": 292.04,
-      "grad_norm": 6.38612174987793,
+      "grad_norm": 7.01643705368042,
       "learning_rate": 2.652652652652653e-07,
-      "loss": 0.8384,
+      "loss": 0.8423,
       "step": 97250
     },
     {
       "epoch": 292.07,
-      "grad_norm": 6.472698211669922,
+      "grad_norm": 5.994320392608643,
       "learning_rate": 2.642642642642643e-07,
-      "loss": 0.9518,
+      "loss": 0.9611,
       "step": 97260
     },
     {
       "epoch": 292.1,
-      "grad_norm": 5.494879722595215,
+      "grad_norm": 6.606977462768555,
       "learning_rate": 2.632632632632633e-07,
-      "loss": 0.8516,
+      "loss": 0.8468,
       "step": 97270
     },
     {
       "epoch": 292.13,
-      "grad_norm": 8.737200736999512,
+      "grad_norm": 8.87147331237793,
       "learning_rate": 2.622622622622623e-07,
-      "loss": 0.9598,
+      "loss": 0.9722,
       "step": 97280
     },
     {
       "epoch": 292.16,
-      "grad_norm": 6.033303737640381,
+      "grad_norm": 6.9385175704956055,
       "learning_rate": 2.612612612612613e-07,
-      "loss": 0.9157,
+      "loss": 0.9231,
       "step": 97290
     },
     {
       "epoch": 292.19,
-      "grad_norm": 8.717066764831543,
+      "grad_norm": 7.484282970428467,
       "learning_rate": 2.6026026026026026e-07,
-      "loss": 0.9302,
+      "loss": 0.945,
       "step": 97300
     },
     {
       "epoch": 292.22,
-      "grad_norm": 9.8933744430542,
+      "grad_norm": 8.921442031860352,
       "learning_rate": 2.592592592592593e-07,
-      "loss": 0.9312,
+      "loss": 0.9278,
       "step": 97310
     },
     {
       "epoch": 292.25,
-      "grad_norm": 7.7617692947387695,
+      "grad_norm": 9.145984649658203,
       "learning_rate": 2.5825825825825825e-07,
-      "loss": 0.9073,
+      "loss": 0.9025,
       "step": 97320
     },
     {
       "epoch": 292.28,
-      "grad_norm": 7.786377906799316,
+      "grad_norm": 7.296932697296143,
       "learning_rate": 2.572572572572573e-07,
-      "loss": 0.9317,
+      "loss": 0.9354,
       "step": 97330
     },
     {
       "epoch": 292.31,
-      "grad_norm": 8.18445110321045,
+      "grad_norm": 8.839743614196777,
       "learning_rate": 2.562562562562563e-07,
       "loss": 0.9577,
       "step": 97340
     },
     {
       "epoch": 292.34,
-      "grad_norm": 9.376206398010254,
+      "grad_norm": 9.597140312194824,
       "learning_rate": 2.552552552552553e-07,
-      "loss": 0.9094,
+      "loss": 0.8983,
       "step": 97350
     },
     {
       "epoch": 292.37,
-      "grad_norm": 6.635852336883545,
+      "grad_norm": 6.1591691970825195,
       "learning_rate": 2.5425425425425427e-07,
-      "loss": 0.9281,
+      "loss": 0.9165,
       "step": 97360
     },
     {
       "epoch": 292.4,
-      "grad_norm": 7.368305683135986,
+      "grad_norm": 6.505436420440674,
       "learning_rate": 2.532532532532533e-07,
-      "loss": 0.8963,
+      "loss": 0.8916,
       "step": 97370
     },
     {
       "epoch": 292.43,
-      "grad_norm": 6.800168514251709,
+      "grad_norm": 6.9030046463012695,
       "learning_rate": 2.5225225225225225e-07,
-      "loss": 0.8361,
+      "loss": 0.8322,
       "step": 97380
     },
     {
       "epoch": 292.46,
-      "grad_norm": 7.5222954750061035,
+      "grad_norm": 6.124162197113037,
       "learning_rate": 2.5125125125125127e-07,
-      "loss": 0.8508,
+      "loss": 0.8518,
       "step": 97390
     },
     {
       "epoch": 292.49,
-      "grad_norm": 7.608657360076904,
+      "grad_norm": 7.317683219909668,
       "learning_rate": 2.502502502502503e-07,
-      "loss": 0.9854,
+      "loss": 0.9865,
       "step": 97400
     },
     {
       "epoch": 292.52,
-      "grad_norm": 6.415729522705078,
+      "grad_norm": 6.114566802978516,
       "learning_rate": 2.492492492492493e-07,
-      "loss": 0.9041,
+      "loss": 0.906,
       "step": 97410
     },
     {
       "epoch": 292.55,
-      "grad_norm": 6.961464881896973,
+      "grad_norm": 8.66525650024414,
       "learning_rate": 2.4824824824824827e-07,
-      "loss": 1.0021,
+      "loss": 1.0146,
       "step": 97420
     },
     {
       "epoch": 292.58,
-      "grad_norm": 5.6046037673950195,
+      "grad_norm": 6.3956499099731445,
       "learning_rate": 2.472472472472473e-07,
-      "loss": 0.9097,
+      "loss": 0.9106,
       "step": 97430
     },
     {
       "epoch": 292.61,
-      "grad_norm": 5.797962665557861,
+      "grad_norm": 6.871274471282959,
       "learning_rate": 2.4624624624624625e-07,
-      "loss": 0.8716,
+      "loss": 0.8691,
       "step": 97440
     },
     {
       "epoch": 292.64,
-      "grad_norm": 5.237704753875732,
+      "grad_norm": 7.573976039886475,
       "learning_rate": 2.4524524524524527e-07,
-      "loss": 0.8423,
+      "loss": 0.845,
       "step": 97450
     },
     {
       "epoch": 292.67,
-      "grad_norm": 6.435274124145508,
+      "grad_norm": 9.593302726745605,
       "learning_rate": 2.4424424424424424e-07,
-      "loss": 0.9266,
+      "loss": 0.9172,
       "step": 97460
     },
     {
       "epoch": 292.7,
-      "grad_norm": 5.816596508026123,
+      "grad_norm": 7.433988571166992,
       "learning_rate": 2.4324324324324326e-07,
-      "loss": 0.899,
+      "loss": 0.9022,
       "step": 97470
     },
     {
       "epoch": 292.73,
-      "grad_norm": 6.51927375793457,
+      "grad_norm": 8.007139205932617,
       "learning_rate": 2.422422422422423e-07,
-      "loss": 0.8521,
+      "loss": 0.8373,
       "step": 97480
     },
     {
       "epoch": 292.76,
-      "grad_norm": 6.857600212097168,
+      "grad_norm": 6.909301280975342,
       "learning_rate": 2.4124124124124124e-07,
-      "loss": 0.8893,
+      "loss": 0.8826,
       "step": 97490
     },
     {
       "epoch": 292.79,
-      "grad_norm": 8.952960014343262,
+      "grad_norm": 9.305088996887207,
       "learning_rate": 2.4024024024024026e-07,
-      "loss": 0.8806,
+      "loss": 0.8765,
       "step": 97500
     },
     {
       "epoch": 292.82,
-      "grad_norm": 5.400733470916748,
+      "grad_norm": 7.418209552764893,
       "learning_rate": 2.392392392392393e-07,
-      "loss": 0.9572,
+      "loss": 0.9651,
       "step": 97510
     },
     {
       "epoch": 292.85,
-      "grad_norm": 7.713247299194336,
+      "grad_norm": 9.34345531463623,
       "learning_rate": 2.3823823823823824e-07,
-      "loss": 0.8896,
+      "loss": 0.8793,
       "step": 97520
     },
     {
       "epoch": 292.88,
-      "grad_norm": 7.793484210968018,
+      "grad_norm": 7.010451316833496,
       "learning_rate": 2.3723723723723726e-07,
-      "loss": 0.9105,
+      "loss": 0.8964,
       "step": 97530
     },
     {
       "epoch": 292.91,
-      "grad_norm": 5.1670732498168945,
+      "grad_norm": 6.122264385223389,
       "learning_rate": 2.3623623623623628e-07,
-      "loss": 0.8681,
+      "loss": 0.8601,
       "step": 97540
     },
     {
       "epoch": 292.94,
-      "grad_norm": 4.523153305053711,
+      "grad_norm": 5.324338912963867,
       "learning_rate": 2.3523523523523527e-07,
-      "loss": 0.9119,
+      "loss": 0.9075,
       "step": 97550
     },
     {
       "epoch": 292.97,
-      "grad_norm": 7.330749034881592,
+      "grad_norm": 8.024182319641113,
       "learning_rate": 2.3423423423423426e-07,
-      "loss": 0.9153,
+      "loss": 0.9014,
       "step": 97560
     },
     {
       "epoch": 293.0,
-      "eval_accuracy": 0.8271,
-      "eval_loss": 0.5690792798995972,
-      "eval_runtime": 5.5713,
-      "eval_samples_per_second": 1794.91,
-      "eval_steps_per_second": 7.18,
+      "eval_accuracy": 0.83,
+      "eval_loss": 0.5686560273170471,
+      "eval_runtime": 5.1662,
+      "eval_samples_per_second": 1935.655,
+      "eval_steps_per_second": 7.743,
       "step": 97569
     },
     {
       "epoch": 293.0,
-      "grad_norm": 7.011456489562988,
+      "grad_norm": 7.568058967590332,
       "learning_rate": 2.3323323323323325e-07,
-      "loss": 1.0043,
+      "loss": 1.0049,
       "step": 97570
     },
     {
       "epoch": 293.03,
-      "grad_norm": 5.389587879180908,
+      "grad_norm": 7.248111724853516,
       "learning_rate": 2.3223223223223225e-07,
-      "loss": 0.8576,
+      "loss": 0.8474,
       "step": 97580
     },
     {
       "epoch": 293.06,
-      "grad_norm": 6.532996654510498,
+      "grad_norm": 6.761630535125732,
       "learning_rate": 2.3123123123123124e-07,
-      "loss": 0.8246,
+      "loss": 0.8182,
       "step": 97590
     },
     {
       "epoch": 293.09,
-      "grad_norm": 9.757499694824219,
+      "grad_norm": 12.385995864868164,
       "learning_rate": 2.3023023023023026e-07,
-      "loss": 0.9151,
+      "loss": 0.9194,
       "step": 97600
     },
     {
       "epoch": 293.12,
-      "grad_norm": 6.377275466918945,
+      "grad_norm": 7.010016441345215,
       "learning_rate": 2.2922922922922925e-07,
-      "loss": 0.8902,
+      "loss": 0.8804,
       "step": 97610
     },
     {
       "epoch": 293.15,
-      "grad_norm": 8.050729751586914,
+      "grad_norm": 11.455182075500488,
       "learning_rate": 2.2822822822822824e-07,
-      "loss": 0.8566,
+      "loss": 0.8491,
       "step": 97620
     },
     {
       "epoch": 293.18,
-      "grad_norm": 9.259918212890625,
+      "grad_norm": 11.208486557006836,
       "learning_rate": 2.2722722722722726e-07,
-      "loss": 0.9908,
+      "loss": 0.9815,
       "step": 97630
     },
     {
       "epoch": 293.21,
-      "grad_norm": 5.4577202796936035,
+      "grad_norm": 6.553277969360352,
       "learning_rate": 2.2622622622622625e-07,
-      "loss": 0.9217,
+      "loss": 0.9229,
       "step": 97640
     },
     {
       "epoch": 293.24,
-      "grad_norm": 5.69035005569458,
+      "grad_norm": 5.181638717651367,
       "learning_rate": 2.2522522522522524e-07,
-      "loss": 0.9019,
+      "loss": 0.9017,
       "step": 97650
     },
     {
       "epoch": 293.27,
-      "grad_norm": 7.393981456756592,
+      "grad_norm": 9.513397216796875,
       "learning_rate": 2.2422422422422423e-07,
-      "loss": 0.944,
+      "loss": 0.9462,
       "step": 97660
     },
     {
       "epoch": 293.3,
-      "grad_norm": 7.8229851722717285,
+      "grad_norm": 8.832915306091309,
       "learning_rate": 2.2322322322322325e-07,
-      "loss": 0.9367,
+      "loss": 0.9344,
       "step": 97670
     },
     {
       "epoch": 293.33,
-      "grad_norm": 8.979887008666992,
+      "grad_norm": 8.953526496887207,
       "learning_rate": 2.2222222222222224e-07,
-      "loss": 0.921,
+      "loss": 0.9152,
       "step": 97680
     },
     {
       "epoch": 293.36,
-      "grad_norm": 6.400908946990967,
+      "grad_norm": 7.211710453033447,
       "learning_rate": 2.2122122122122124e-07,
-      "loss": 0.8497,
+      "loss": 0.8414,
       "step": 97690
     },
     {
       "epoch": 293.39,
-      "grad_norm": 5.4380784034729,
+      "grad_norm": 7.878899097442627,
       "learning_rate": 2.2022022022022023e-07,
-      "loss": 0.8844,
+      "loss": 0.8817,
       "step": 97700
     },
     {
       "epoch": 293.42,
-      "grad_norm": 5.700555801391602,
+      "grad_norm": 5.159241199493408,
       "learning_rate": 2.1921921921921922e-07,
-      "loss": 0.929,
+      "loss": 0.93,
       "step": 97710
     },
     {
       "epoch": 293.45,
-      "grad_norm": 7.510645866394043,
+      "grad_norm": 7.421966552734375,
       "learning_rate": 2.182182182182182e-07,
-      "loss": 0.938,
+      "loss": 0.9372,
       "step": 97720
     },
     {
       "epoch": 293.48,
-      "grad_norm": 8.172004699707031,
+      "grad_norm": 7.8214616775512695,
       "learning_rate": 2.1721721721721726e-07,
-      "loss": 0.9706,
+      "loss": 0.978,
       "step": 97730
     },
     {
       "epoch": 293.51,
-      "grad_norm": 8.15710735321045,
+      "grad_norm": 6.8256916999816895,
       "learning_rate": 2.1621621621621625e-07,
-      "loss": 0.9405,
+      "loss": 0.9439,
       "step": 97740
     },
     {
       "epoch": 293.54,
-      "grad_norm": 5.571786403656006,
+      "grad_norm": 7.25612211227417,
       "learning_rate": 2.1521521521521524e-07,
-      "loss": 0.9296,
+      "loss": 0.9235,
       "step": 97750
     },
     {
       "epoch": 293.57,
-      "grad_norm": 6.150842666625977,
+      "grad_norm": 6.426498889923096,
       "learning_rate": 2.1421421421421423e-07,
-      "loss": 0.8844,
+      "loss": 0.8748,
       "step": 97760
     },
     {
       "epoch": 293.6,
-      "grad_norm": 4.390723705291748,
+      "grad_norm": 4.844997882843018,
       "learning_rate": 2.1321321321321322e-07,
-      "loss": 0.8882,
+      "loss": 0.8945,
       "step": 97770
     },
     {
       "epoch": 293.63,
-      "grad_norm": 9.761536598205566,
+      "grad_norm": 7.717916488647461,
       "learning_rate": 2.1221221221221222e-07,
-      "loss": 0.923,
+      "loss": 0.9184,
       "step": 97780
     },
     {
       "epoch": 293.66,
-      "grad_norm": 5.533973217010498,
+      "grad_norm": 5.070001125335693,
       "learning_rate": 2.112112112112112e-07,
-      "loss": 0.8558,
+      "loss": 0.8482,
       "step": 97790
     },
     {
       "epoch": 293.69,
-      "grad_norm": 7.926022052764893,
+      "grad_norm": 7.807445526123047,
       "learning_rate": 2.1021021021021025e-07,
-      "loss": 0.9058,
+      "loss": 0.8953,
       "step": 97800
     },
     {
       "epoch": 293.72,
-      "grad_norm": 7.757513999938965,
+      "grad_norm": 6.571441173553467,
       "learning_rate": 2.0920920920920924e-07,
-      "loss": 0.8755,
+      "loss": 0.8766,
       "step": 97810
     },
     {
       "epoch": 293.75,
-      "grad_norm": 6.728062152862549,
+      "grad_norm": 6.921735763549805,
       "learning_rate": 2.0820820820820824e-07,
-      "loss": 0.882,
+      "loss": 0.885,
       "step": 97820
     },
     {
       "epoch": 293.78,
-      "grad_norm": 8.659940719604492,
+      "grad_norm": 13.83038330078125,
       "learning_rate": 2.0720720720720723e-07,
-      "loss": 0.9341,
+      "loss": 0.939,
       "step": 97830
     },
     {
       "epoch": 293.81,
-      "grad_norm": 4.771610260009766,
+      "grad_norm": 5.131385803222656,
       "learning_rate": 2.0620620620620622e-07,
-      "loss": 0.8136,
+      "loss": 0.8156,
       "step": 97840
     },
     {
       "epoch": 293.84,
-      "grad_norm": 6.450154781341553,
+      "grad_norm": 5.953934669494629,
       "learning_rate": 2.052052052052052e-07,
-      "loss": 0.8848,
+      "loss": 0.8849,
       "step": 97850
     },
     {
       "epoch": 293.87,
-      "grad_norm": 7.299238681793213,
+      "grad_norm": 8.05733871459961,
       "learning_rate": 2.042042042042042e-07,
-      "loss": 0.8464,
+      "loss": 0.8434,
       "step": 97860
     },
     {
       "epoch": 293.9,
-      "grad_norm": 5.8163957595825195,
+      "grad_norm": 5.343295097351074,
       "learning_rate": 2.0320320320320322e-07,
-      "loss": 0.9697,
+      "loss": 0.9613,
       "step": 97870
     },
     {
       "epoch": 293.93,
-      "grad_norm": 7.096786022186279,
+      "grad_norm": 6.216550350189209,
       "learning_rate": 2.0220220220220224e-07,
-      "loss": 0.8554,
+      "loss": 0.8613,
       "step": 97880
     },
     {
       "epoch": 293.96,
-      "grad_norm": 5.58074951171875,
+      "grad_norm": 6.856605052947998,
       "learning_rate": 2.0120120120120123e-07,
-      "loss": 0.8991,
+      "loss": 0.8877,
       "step": 97890
     },
     {
       "epoch": 293.99,
-      "grad_norm": 8.246901512145996,
+      "grad_norm": 7.4229416847229,
       "learning_rate": 2.0020020020020022e-07,
-      "loss": 0.8814,
+      "loss": 0.878,
       "step": 97900
     },
     {
       "epoch": 294.0,
-      "eval_accuracy": 0.8273,
-      "eval_loss": 0.569220244884491,
-      "eval_runtime": 5.5872,
-      "eval_samples_per_second": 1789.798,
-      "eval_steps_per_second": 7.159,
+      "eval_accuracy": 0.8287,
+      "eval_loss": 0.5698645710945129,
+      "eval_runtime": 5.203,
+      "eval_samples_per_second": 1921.977,
+      "eval_steps_per_second": 7.688,
       "step": 97902
     },
     {
       "epoch": 294.02,
-      "grad_norm": 5.726893901824951,
+      "grad_norm": 9.18761157989502,
       "learning_rate": 1.9919919919919922e-07,
-      "loss": 0.9664,
+      "loss": 0.9795,
       "step": 97910
     },
     {
       "epoch": 294.05,
-      "grad_norm": 5.6355485916137695,
+      "grad_norm": 5.424155235290527,
       "learning_rate": 1.981981981981982e-07,
-      "loss": 0.8715,
+      "loss": 0.8749,
       "step": 97920
     },
     {
       "epoch": 294.08,
-      "grad_norm": 5.047061920166016,
+      "grad_norm": 6.0034050941467285,
       "learning_rate": 1.971971971971972e-07,
-      "loss": 0.8918,
+      "loss": 0.8903,
       "step": 97930
     },
     {
       "epoch": 294.11,
-      "grad_norm": 6.550655841827393,
+      "grad_norm": 6.509764194488525,
       "learning_rate": 1.9619619619619622e-07,
-      "loss": 0.8894,
+      "loss": 0.8884,
       "step": 97940
     },
     {
       "epoch": 294.14,
-      "grad_norm": 4.311676502227783,
+      "grad_norm": 5.383185863494873,
       "learning_rate": 1.951951951951952e-07,
-      "loss": 0.876,
+      "loss": 0.882,
       "step": 97950
     },
     {
       "epoch": 294.17,
-      "grad_norm": 6.5771307945251465,
+      "grad_norm": 6.880011558532715,
       "learning_rate": 1.941941941941942e-07,
-      "loss": 0.8679,
+      "loss": 0.8603,
       "step": 97960
     },
     {
       "epoch": 294.2,
-      "grad_norm": 5.84274959564209,
+      "grad_norm": 8.527591705322266,
       "learning_rate": 1.931931931931932e-07,
-      "loss": 0.9079,
+      "loss": 0.9081,
       "step": 97970
     },
     {
       "epoch": 294.23,
-      "grad_norm": 8.563691139221191,
+      "grad_norm": 6.130377769470215,
       "learning_rate": 1.921921921921922e-07,
-      "loss": 0.8987,
+      "loss": 0.8893,
       "step": 97980
     },
     {
       "epoch": 294.26,
-      "grad_norm": 6.0965118408203125,
+      "grad_norm": 6.683150768280029,
       "learning_rate": 1.911911911911912e-07,
-      "loss": 0.894,
+      "loss": 0.8906,
       "step": 97990
     },
     {
       "epoch": 294.29,
-      "grad_norm": 6.504528522491455,
+      "grad_norm": 8.972715377807617,
       "learning_rate": 1.9019019019019022e-07,
-      "loss": 0.8644,
+      "loss": 0.8643,
       "step": 98000
     },
     {
       "epoch": 294.32,
-      "grad_norm": 6.378507137298584,
+      "grad_norm": 7.366360187530518,
       "learning_rate": 1.8918918918918921e-07,
-      "loss": 0.8509,
+      "loss": 0.8556,
       "step": 98010
     },
     {
       "epoch": 294.35,
-      "grad_norm": 5.893850326538086,
+      "grad_norm": 6.60482120513916,
       "learning_rate": 1.881881881881882e-07,
-      "loss": 0.893,
+      "loss": 0.8999,
       "step": 98020
     },
     {
       "epoch": 294.38,
-      "grad_norm": 7.361412525177002,
+      "grad_norm": 6.160721778869629,
       "learning_rate": 1.871871871871872e-07,
-      "loss": 0.8602,
+      "loss": 0.8441,
       "step": 98030
     },
     {
       "epoch": 294.41,
-      "grad_norm": 9.0660982131958,
+      "grad_norm": 5.952327728271484,
       "learning_rate": 1.861861861861862e-07,
-      "loss": 0.9353,
+      "loss": 0.9348,
       "step": 98040
     },
     {
       "epoch": 294.44,
-      "grad_norm": 5.818506717681885,
+      "grad_norm": 6.734945774078369,
       "learning_rate": 1.8518518518518518e-07,
-      "loss": 0.9105,
+      "loss": 0.9027,
       "step": 98050
     },
     {
       "epoch": 294.47,
-      "grad_norm": 6.31784725189209,
+      "grad_norm": 5.301029205322266,
       "learning_rate": 1.8418418418418417e-07,
-      "loss": 0.8758,
+      "loss": 0.8567,
       "step": 98060
     },
     {
       "epoch": 294.5,
-      "grad_norm": 7.968667507171631,
+      "grad_norm": 7.530423641204834,
       "learning_rate": 1.8318318318318322e-07,
-      "loss": 0.8945,
+      "loss": 0.8959,
       "step": 98070
     },
     {
       "epoch": 294.53,
-      "grad_norm": 7.294238567352295,
+      "grad_norm": 8.585826873779297,
       "learning_rate": 1.821821821821822e-07,
-      "loss": 0.9387,
+      "loss": 0.9292,
       "step": 98080
     },
     {
       "epoch": 294.56,
-      "grad_norm": 12.315067291259766,
+      "grad_norm": 11.638001441955566,
       "learning_rate": 1.811811811811812e-07,
-      "loss": 0.8834,
+      "loss": 0.8854,
       "step": 98090
     },
     {
       "epoch": 294.59,
-      "grad_norm": 4.3012824058532715,
+      "grad_norm": 5.490418910980225,
       "learning_rate": 1.801801801801802e-07,
-      "loss": 0.863,
+      "loss": 0.8708,
       "step": 98100
     },
     {
       "epoch": 294.62,
-      "grad_norm": 4.673679351806641,
+      "grad_norm": 5.042558670043945,
       "learning_rate": 1.7917917917917919e-07,
-      "loss": 0.8948,
+      "loss": 0.8933,
       "step": 98110
     },
     {
       "epoch": 294.65,
-      "grad_norm": 10.309165000915527,
+      "grad_norm": 9.73974609375,
       "learning_rate": 1.7817817817817818e-07,
-      "loss": 0.888,
+      "loss": 0.8866,
       "step": 98120
     },
     {
       "epoch": 294.68,
-      "grad_norm": 6.0395097732543945,
+      "grad_norm": 5.405859470367432,
       "learning_rate": 1.7717717717717717e-07,
-      "loss": 0.8905,
+      "loss": 0.8986,
       "step": 98130
     },
     {
       "epoch": 294.71,
-      "grad_norm": 5.768089294433594,
+      "grad_norm": 7.075048923492432,
       "learning_rate": 1.7617617617617621e-07,
-      "loss": 0.8717,
+      "loss": 0.8692,
       "step": 98140
     },
     {
       "epoch": 294.74,
-      "grad_norm": 8.730615615844727,
+      "grad_norm": 6.755312919616699,
       "learning_rate": 1.751751751751752e-07,
-      "loss": 0.9642,
+      "loss": 0.9703,
       "step": 98150
     },
     {
       "epoch": 294.77,
-      "grad_norm": 5.214756965637207,
+      "grad_norm": 6.242851257324219,
       "learning_rate": 1.741741741741742e-07,
-      "loss": 0.9261,
+      "loss": 0.9271,
       "step": 98160
     },
     {
       "epoch": 294.8,
-      "grad_norm": 6.945539474487305,
+      "grad_norm": 5.487273216247559,
       "learning_rate": 1.731731731731732e-07,
-      "loss": 0.8841,
+      "loss": 0.8832,
       "step": 98170
     },
     {
       "epoch": 294.83,
-      "grad_norm": 8.12825870513916,
+      "grad_norm": 6.401918411254883,
       "learning_rate": 1.7217217217217218e-07,
-      "loss": 0.9114,
+      "loss": 0.9162,
       "step": 98180
     },
     {
       "epoch": 294.86,
-      "grad_norm": 5.774511814117432,
+      "grad_norm": 5.408132553100586,
       "learning_rate": 1.7117117117117117e-07,
-      "loss": 0.849,
+      "loss": 0.8537,
       "step": 98190
     },
     {
       "epoch": 294.89,
-      "grad_norm": 8.114261627197266,
+      "grad_norm": 7.494041442871094,
       "learning_rate": 1.701701701701702e-07,
-      "loss": 1.0149,
+      "loss": 1.0084,
       "step": 98200
     },
     {
       "epoch": 294.92,
-      "grad_norm": 9.25036334991455,
+      "grad_norm": 8.250554084777832,
       "learning_rate": 1.6916916916916918e-07,
-      "loss": 0.9474,
+      "loss": 0.9493,
       "step": 98210
     },
     {
       "epoch": 294.95,
-      "grad_norm": 7.054948806762695,
+      "grad_norm": 12.807097434997559,
       "learning_rate": 1.6816816816816818e-07,
-      "loss": 0.9082,
+      "loss": 0.902,
       "step": 98220
     },
     {
       "epoch": 294.98,
-      "grad_norm": 9.114188194274902,
+      "grad_norm": 8.222254753112793,
       "learning_rate": 1.671671671671672e-07,
-      "loss": 0.915,
+      "loss": 0.9219,
       "step": 98230
     },
     {
       "epoch": 295.0,
-      "eval_accuracy": 0.8269,
-      "eval_loss": 0.575465738773346,
-      "eval_runtime": 5.3647,
-      "eval_samples_per_second": 1864.024,
-      "eval_steps_per_second": 7.456,
+      "eval_accuracy": 0.8296,
+      "eval_loss": 0.575412392616272,
+      "eval_runtime": 5.3277,
+      "eval_samples_per_second": 1876.987,
+      "eval_steps_per_second": 7.508,
       "step": 98235
     },
     {
       "epoch": 295.02,
-      "grad_norm": 7.524016857147217,
+      "grad_norm": 6.235860347747803,
       "learning_rate": 1.6616616616616619e-07,
-      "loss": 0.9672,
+      "loss": 0.9636,
       "step": 98240
     },
     {
       "epoch": 295.05,
-      "grad_norm": 7.12274694442749,
+      "grad_norm": 9.943619728088379,
       "learning_rate": 1.6516516516516518e-07,
-      "loss": 0.8251,
+      "loss": 0.8287,
       "step": 98250
     },
     {
       "epoch": 295.08,
-      "grad_norm": 5.841676712036133,
+      "grad_norm": 6.524040699005127,
       "learning_rate": 1.6416416416416417e-07,
-      "loss": 0.8926,
+      "loss": 0.8975,
       "step": 98260
     },
     {
       "epoch": 295.11,
-      "grad_norm": 5.076598644256592,
+      "grad_norm": 6.303874492645264,
       "learning_rate": 1.631631631631632e-07,
-      "loss": 1.0142,
+      "loss": 1.008,
       "step": 98270
     },
     {
       "epoch": 295.14,
-      "grad_norm": 4.674306869506836,
+      "grad_norm": 6.899210453033447,
       "learning_rate": 1.6216216216216218e-07,
-      "loss": 0.8632,
+      "loss": 0.8606,
       "step": 98280
     },
     {
       "epoch": 295.17,
-      "grad_norm": 8.002975463867188,
+      "grad_norm": 7.207037925720215,
       "learning_rate": 1.6116116116116117e-07,
-      "loss": 0.9173,
+      "loss": 0.9144,
       "step": 98290
     },
     {
       "epoch": 295.2,
-      "grad_norm": 5.195116996765137,
+      "grad_norm": 5.270463466644287,
       "learning_rate": 1.6016016016016016e-07,
-      "loss": 0.8786,
+      "loss": 0.8784,
       "step": 98300
     },
     {
       "epoch": 295.23,
-      "grad_norm": 6.069470405578613,
+      "grad_norm": 6.2862138748168945,
       "learning_rate": 1.5915915915915916e-07,
-      "loss": 0.9216,
+      "loss": 0.9239,
       "step": 98310
     },
     {
       "epoch": 295.26,
-      "grad_norm": 6.679693698883057,
+      "grad_norm": 7.716737747192383,
       "learning_rate": 1.5815815815815815e-07,
-      "loss": 0.8899,
+      "loss": 0.8882,
       "step": 98320
     },
     {
       "epoch": 295.29,
-      "grad_norm": 7.20912504196167,
+      "grad_norm": 7.592684268951416,
       "learning_rate": 1.5715715715715717e-07,
-      "loss": 0.9661,
+      "loss": 0.958,
       "step": 98330
     },
     {
       "epoch": 295.32,
-      "grad_norm": 5.078705787658691,
+      "grad_norm": 5.4384355545043945,
       "learning_rate": 1.5615615615615616e-07,
-      "loss": 0.8461,
+      "loss": 0.8502,
       "step": 98340
     },
     {
       "epoch": 295.35,
-      "grad_norm": 6.910800933837891,
+      "grad_norm": 6.225616455078125,
       "learning_rate": 1.5515515515515518e-07,
-      "loss": 0.9413,
+      "loss": 0.9309,
       "step": 98350
     },
     {
       "epoch": 295.38,
-      "grad_norm": 7.670577526092529,
+      "grad_norm": 6.071933269500732,
       "learning_rate": 1.5415415415415417e-07,
-      "loss": 0.9282,
+      "loss": 0.9205,
       "step": 98360
     },
     {
       "epoch": 295.41,
-      "grad_norm": 8.78126049041748,
+      "grad_norm": 8.752120018005371,
       "learning_rate": 1.5315315315315316e-07,
-      "loss": 0.9282,
+      "loss": 0.9285,
       "step": 98370
     },
     {
       "epoch": 295.44,
-      "grad_norm": 5.8206281661987305,
+      "grad_norm": 6.496067523956299,
       "learning_rate": 1.5215215215215215e-07,
-      "loss": 0.9112,
+      "loss": 0.9133,
       "step": 98380
     },
     {
       "epoch": 295.47,
-      "grad_norm": 5.534880638122559,
+      "grad_norm": 6.084729194641113,
       "learning_rate": 1.5115115115115117e-07,
-      "loss": 0.8495,
+      "loss": 0.8402,
       "step": 98390
     },
     {
       "epoch": 295.5,
-      "grad_norm": 8.05062484741211,
+      "grad_norm": 9.95107650756836,
       "learning_rate": 1.5015015015015016e-07,
-      "loss": 0.9163,
+      "loss": 0.9128,
       "step": 98400
     },
     {
       "epoch": 295.53,
-      "grad_norm": 8.196608543395996,
+      "grad_norm": 7.5813446044921875,
       "learning_rate": 1.4914914914914915e-07,
-      "loss": 0.8967,
+      "loss": 0.8956,
       "step": 98410
     },
     {
       "epoch": 295.56,
-      "grad_norm": 5.0741801261901855,
+      "grad_norm": 4.540396690368652,
       "learning_rate": 1.4814814814814817e-07,
-      "loss": 0.905,
+      "loss": 0.9057,
       "step": 98420
     },
     {
       "epoch": 295.59,
-      "grad_norm": 5.8667988777160645,
+      "grad_norm": 9.409579277038574,
       "learning_rate": 1.4714714714714716e-07,
-      "loss": 0.916,
+      "loss": 0.9122,
       "step": 98430
     },
     {
       "epoch": 295.62,
-      "grad_norm": 11.556714057922363,
+      "grad_norm": 7.350164413452148,
       "learning_rate": 1.4614614614614616e-07,
-      "loss": 0.878,
+      "loss": 0.8686,
       "step": 98440
     },
     {
       "epoch": 295.65,
-      "grad_norm": 5.295146942138672,
+      "grad_norm": 5.213441848754883,
       "learning_rate": 1.4514514514514517e-07,
-      "loss": 0.886,
+      "loss": 0.8846,
       "step": 98450
     },
     {
       "epoch": 295.68,
-      "grad_norm": 6.66457986831665,
+      "grad_norm": 6.332927227020264,
       "learning_rate": 1.4414414414414417e-07,
-      "loss": 0.9099,
+      "loss": 0.8945,
       "step": 98460
     },
     {
       "epoch": 295.71,
-      "grad_norm": 8.593743324279785,
+      "grad_norm": 7.878532409667969,
       "learning_rate": 1.4314314314314316e-07,
-      "loss": 0.9098,
+      "loss": 0.9077,
       "step": 98470
     },
     {
       "epoch": 295.74,
-      "grad_norm": 5.72769832611084,
+      "grad_norm": 7.6684184074401855,
       "learning_rate": 1.4214214214214215e-07,
-      "loss": 0.9119,
+      "loss": 0.9084,
       "step": 98480
     },
     {
       "epoch": 295.77,
-      "grad_norm": 5.379662036895752,
+      "grad_norm": 5.035165786743164,
       "learning_rate": 1.4114114114114117e-07,
-      "loss": 0.926,
+      "loss": 0.9237,
       "step": 98490
     },
     {
       "epoch": 295.8,
-      "grad_norm": 4.801339626312256,
+      "grad_norm": 5.123696804046631,
       "learning_rate": 1.4014014014014016e-07,
-      "loss": 0.9019,
+      "loss": 0.8965,
       "step": 98500
     },
     {
       "epoch": 295.83,
-      "grad_norm": 7.9495697021484375,
+      "grad_norm": 7.406163215637207,
       "learning_rate": 1.3913913913913915e-07,
-      "loss": 0.9284,
+      "loss": 0.9171,
       "step": 98510
     },
     {
       "epoch": 295.86,
-      "grad_norm": 7.009015083312988,
+      "grad_norm": 6.3559250831604,
       "learning_rate": 1.3813813813813814e-07,
-      "loss": 0.9206,
+      "loss": 0.9298,
       "step": 98520
     },
     {
       "epoch": 295.89,
-      "grad_norm": 6.8108344078063965,
+      "grad_norm": 5.6796064376831055,
       "learning_rate": 1.3713713713713716e-07,
-      "loss": 0.902,
+      "loss": 0.8984,
       "step": 98530
     },
     {
       "epoch": 295.92,
-      "grad_norm": 7.25115966796875,
+      "grad_norm": 7.309746742248535,
       "learning_rate": 1.3613613613613615e-07,
-      "loss": 0.8931,
+      "loss": 0.8967,
       "step": 98540
     },
     {
       "epoch": 295.95,
-      "grad_norm": 7.309234619140625,
+      "grad_norm": 7.728320598602295,
       "learning_rate": 1.3513513513513515e-07,
-      "loss": 0.8354,
+      "loss": 0.8324,
       "step": 98550
     },
     {
       "epoch": 295.98,
-      "grad_norm": 6.898353576660156,
+      "grad_norm": 9.847877502441406,
       "learning_rate": 1.3413413413413414e-07,
-      "loss": 0.9032,
+      "loss": 0.9007,
       "step": 98560
     },
     {
       "epoch": 296.0,
-      "eval_accuracy": 0.8228,
-      "eval_loss": 0.5805512070655823,
-      "eval_runtime": 5.6241,
-      "eval_samples_per_second": 1778.059,
-      "eval_steps_per_second": 7.112,
+      "eval_accuracy": 0.8243,
+      "eval_loss": 0.5818559527397156,
+      "eval_runtime": 5.1698,
+      "eval_samples_per_second": 1934.316,
+      "eval_steps_per_second": 7.737,
       "step": 98568
     },
     {
       "epoch": 296.01,
-      "grad_norm": 6.156898498535156,
+      "grad_norm": 8.738362312316895,
       "learning_rate": 1.3313313313313313e-07,
-      "loss": 1.1027,
+      "loss": 1.0912,
       "step": 98570
     },
     {
       "epoch": 296.04,
-      "grad_norm": 5.640598297119141,
+      "grad_norm": 5.450796127319336,
       "learning_rate": 1.3213213213213215e-07,
-      "loss": 0.9017,
+      "loss": 0.8933,
       "step": 98580
     },
     {
       "epoch": 296.07,
-      "grad_norm": 5.871313571929932,
+      "grad_norm": 5.084892749786377,
       "learning_rate": 1.3113113113113114e-07,
-      "loss": 0.9915,
+      "loss": 0.9868,
       "step": 98590
     },
     {
       "epoch": 296.1,
-      "grad_norm": 6.760378837585449,
+      "grad_norm": 7.174663066864014,
       "learning_rate": 1.3013013013013013e-07,
-      "loss": 0.9543,
+      "loss": 0.9423,
       "step": 98600
     },
     {
       "epoch": 296.13,
-      "grad_norm": 5.284930229187012,
+      "grad_norm": 4.684029579162598,
       "learning_rate": 1.2912912912912912e-07,
-      "loss": 0.9327,
+      "loss": 0.9306,
       "step": 98610
     },
     {
       "epoch": 296.16,
-      "grad_norm": 6.394189834594727,
+      "grad_norm": 7.8613786697387695,
       "learning_rate": 1.2812812812812814e-07,
-      "loss": 0.9523,
+      "loss": 0.9442,
       "step": 98620
     },
     {
       "epoch": 296.19,
-      "grad_norm": 6.485101222991943,
+      "grad_norm": 8.105913162231445,
       "learning_rate": 1.2712712712712713e-07,
-      "loss": 0.9733,
+      "loss": 0.9587,
       "step": 98630
     },
     {
       "epoch": 296.22,
-      "grad_norm": 5.145962715148926,
+      "grad_norm": 8.392313003540039,
       "learning_rate": 1.2612612612612613e-07,
-      "loss": 0.8515,
+      "loss": 0.857,
       "step": 98640
     },
     {
       "epoch": 296.25,
-      "grad_norm": 8.710779190063477,
+      "grad_norm": 10.649443626403809,
       "learning_rate": 1.2512512512512514e-07,
-      "loss": 0.9091,
+      "loss": 0.8992,
       "step": 98650
     },
     {
       "epoch": 296.28,
-      "grad_norm": 6.305322647094727,
+      "grad_norm": 7.455546855926514,
       "learning_rate": 1.2412412412412414e-07,
-      "loss": 0.9006,
+      "loss": 0.9024,
       "step": 98660
     },
     {
       "epoch": 296.31,
-      "grad_norm": 9.307201385498047,
+      "grad_norm": 8.645292282104492,
       "learning_rate": 1.2312312312312313e-07,
-      "loss": 0.8387,
+      "loss": 0.8403,
       "step": 98670
     },
     {
       "epoch": 296.34,
-      "grad_norm": 5.854321479797363,
+      "grad_norm": 8.878702163696289,
       "learning_rate": 1.2212212212212212e-07,
-      "loss": 0.942,
+      "loss": 0.9427,
       "step": 98680
     },
     {
       "epoch": 296.37,
-      "grad_norm": 4.888759136199951,
+      "grad_norm": 4.923945426940918,
       "learning_rate": 1.2112112112112114e-07,
-      "loss": 0.8906,
+      "loss": 0.8937,
       "step": 98690
     },
     {
       "epoch": 296.4,
-      "grad_norm": 5.127511024475098,
+      "grad_norm": 5.4294304847717285,
       "learning_rate": 1.2012012012012013e-07,
-      "loss": 0.9235,
+      "loss": 0.9288,
       "step": 98700
     },
     {
       "epoch": 296.43,
-      "grad_norm": 8.489784240722656,
+      "grad_norm": 6.712618827819824,
       "learning_rate": 1.1911911911911912e-07,
-      "loss": 0.9546,
+      "loss": 0.9423,
       "step": 98710
     },
     {
       "epoch": 296.46,
-      "grad_norm": 5.536572456359863,
+      "grad_norm": 5.248286247253418,
       "learning_rate": 1.1811811811811814e-07,
-      "loss": 0.8513,
+      "loss": 0.8474,
       "step": 98720
     },
     {
       "epoch": 296.49,
-      "grad_norm": 7.290252685546875,
+      "grad_norm": 6.185132026672363,
       "learning_rate": 1.1711711711711713e-07,
-      "loss": 0.8912,
+      "loss": 0.8845,
       "step": 98730
     },
     {
       "epoch": 296.52,
-      "grad_norm": 9.693050384521484,
+      "grad_norm": 9.42718505859375,
       "learning_rate": 1.1611611611611612e-07,
-      "loss": 0.9183,
+      "loss": 0.9293,
       "step": 98740
     },
     {
       "epoch": 296.55,
-      "grad_norm": 5.239686965942383,
+      "grad_norm": 7.834132194519043,
       "learning_rate": 1.1511511511511513e-07,
-      "loss": 0.8594,
+      "loss": 0.8556,
       "step": 98750
     },
     {
       "epoch": 296.58,
-      "grad_norm": 7.103659629821777,
+      "grad_norm": 6.923455715179443,
       "learning_rate": 1.1411411411411412e-07,
-      "loss": 0.9222,
+      "loss": 0.9109,
       "step": 98760
     },
     {
       "epoch": 296.61,
-      "grad_norm": 5.729584217071533,
+      "grad_norm": 4.83266019821167,
       "learning_rate": 1.1311311311311313e-07,
-      "loss": 0.9289,
+      "loss": 0.9227,
       "step": 98770
     },
     {
       "epoch": 296.64,
-      "grad_norm": 7.891506195068359,
+      "grad_norm": 8.230085372924805,
       "learning_rate": 1.1211211211211212e-07,
-      "loss": 0.9313,
+      "loss": 0.9291,
       "step": 98780
     },
     {
       "epoch": 296.67,
-      "grad_norm": 4.760802268981934,
+      "grad_norm": 5.303246974945068,
       "learning_rate": 1.1111111111111112e-07,
-      "loss": 0.9363,
+      "loss": 0.921,
       "step": 98790
     },
     {
       "epoch": 296.7,
-      "grad_norm": 5.562394142150879,
+      "grad_norm": 5.472437381744385,
       "learning_rate": 1.1011011011011011e-07,
-      "loss": 0.8651,
+      "loss": 0.862,
       "step": 98800
     },
     {
       "epoch": 296.73,
-      "grad_norm": 6.774091720581055,
+      "grad_norm": 5.803252696990967,
       "learning_rate": 1.091091091091091e-07,
-      "loss": 0.9272,
+      "loss": 0.9312,
       "step": 98810
     },
     {
       "epoch": 296.76,
-      "grad_norm": 5.885730743408203,
+      "grad_norm": 6.83626651763916,
       "learning_rate": 1.0810810810810812e-07,
-      "loss": 0.8726,
+      "loss": 0.8764,
       "step": 98820
     },
     {
       "epoch": 296.79,
-      "grad_norm": 8.066106796264648,
+      "grad_norm": 7.416439533233643,
       "learning_rate": 1.0710710710710712e-07,
-      "loss": 0.959,
+      "loss": 0.9577,
       "step": 98830
     },
     {
       "epoch": 296.82,
-      "grad_norm": 6.368505001068115,
+      "grad_norm": 6.918694019317627,
       "learning_rate": 1.0610610610610611e-07,
-      "loss": 0.905,
+      "loss": 0.8949,
       "step": 98840
     },
     {
       "epoch": 296.85,
-      "grad_norm": 5.76190185546875,
+      "grad_norm": 6.658125877380371,
       "learning_rate": 1.0510510510510513e-07,
-      "loss": 0.8564,
+      "loss": 0.8558,
       "step": 98850
     },
     {
       "epoch": 296.88,
-      "grad_norm": 6.153330326080322,
+      "grad_norm": 6.5581793785095215,
       "learning_rate": 1.0410410410410412e-07,
-      "loss": 0.9478,
+      "loss": 0.9241,
       "step": 98860
     },
     {
       "epoch": 296.91,
-      "grad_norm": 7.150406837463379,
+      "grad_norm": 7.3488688468933105,
       "learning_rate": 1.0310310310310311e-07,
-      "loss": 0.9263,
+      "loss": 0.9199,
       "step": 98870
     },
     {
       "epoch": 296.94,
-      "grad_norm": 5.159527778625488,
+      "grad_norm": 5.228977680206299,
       "learning_rate": 1.021021021021021e-07,
-      "loss": 0.9053,
+      "loss": 0.8983,
       "step": 98880
     },
     {
       "epoch": 296.97,
-      "grad_norm": 6.45241641998291,
+      "grad_norm": 6.544425964355469,
       "learning_rate": 1.0110110110110112e-07,
-      "loss": 0.9075,
+      "loss": 0.9059,
       "step": 98890
     },
     {
       "epoch": 297.0,
-      "grad_norm": 9.044663429260254,
+      "grad_norm": 6.903194427490234,
       "learning_rate": 1.0010010010010011e-07,
-      "loss": 0.9006,
+      "loss": 0.9016,
       "step": 98900
     },
     {
       "epoch": 297.0,
-      "eval_accuracy": 0.8273,
-      "eval_loss": 0.5719346404075623,
-      "eval_runtime": 5.6842,
-      "eval_samples_per_second": 1759.271,
-      "eval_steps_per_second": 7.037,
+      "eval_accuracy": 0.8295,
+      "eval_loss": 0.5727095007896423,
+      "eval_runtime": 5.475,
+      "eval_samples_per_second": 1826.493,
+      "eval_steps_per_second": 7.306,
       "step": 98901
     },
     {
       "epoch": 297.03,
-      "grad_norm": 7.243643283843994,
+      "grad_norm": 6.475955963134766,
       "learning_rate": 9.90990990990991e-08,
-      "loss": 1.1014,
+      "loss": 1.1076,
       "step": 98910
     },
     {
       "epoch": 297.06,
-      "grad_norm": 9.414041519165039,
+      "grad_norm": 9.700071334838867,
       "learning_rate": 9.809809809809811e-08,
-      "loss": 0.8901,
+      "loss": 0.8813,
       "step": 98920
     },
     {
       "epoch": 297.09,
-      "grad_norm": 4.783314228057861,
+      "grad_norm": 4.903450965881348,
       "learning_rate": 9.70970970970971e-08,
-      "loss": 0.9393,
+      "loss": 0.9523,
       "step": 98930
     },
     {
       "epoch": 297.12,
-      "grad_norm": 5.4052605628967285,
+      "grad_norm": 4.344940185546875,
       "learning_rate": 9.60960960960961e-08,
-      "loss": 0.9114,
+      "loss": 0.9157,
       "step": 98940
     },
     {
       "epoch": 297.15,
-      "grad_norm": 6.81166410446167,
+      "grad_norm": 8.27981185913086,
       "learning_rate": 9.509509509509511e-08,
-      "loss": 0.9389,
+      "loss": 0.9369,
       "step": 98950
     },
     {
       "epoch": 297.18,
-      "grad_norm": 12.393120765686035,
+      "grad_norm": 14.610191345214844,
       "learning_rate": 9.40940940940941e-08,
-      "loss": 0.8476,
+      "loss": 0.8414,
       "step": 98960
     },
     {
       "epoch": 297.21,
-      "grad_norm": 6.856719493865967,
+      "grad_norm": 7.674727916717529,
       "learning_rate": 9.30930930930931e-08,
-      "loss": 0.9369,
+      "loss": 0.946,
       "step": 98970
     },
     {
       "epoch": 297.24,
-      "grad_norm": 4.524720191955566,
+      "grad_norm": 6.3353190422058105,
       "learning_rate": 9.209209209209209e-08,
-      "loss": 0.8973,
+      "loss": 0.8987,
       "step": 98980
     },
     {
       "epoch": 297.27,
-      "grad_norm": 4.543455600738525,
+      "grad_norm": 5.766744613647461,
       "learning_rate": 9.10910910910911e-08,
-      "loss": 0.9354,
+      "loss": 0.9382,
       "step": 98990
     },
     {
       "epoch": 297.3,
-      "grad_norm": 6.365161418914795,
+      "grad_norm": 7.679034233093262,
       "learning_rate": 9.00900900900901e-08,
-      "loss": 0.9145,
+      "loss": 0.9223,
       "step": 99000
     },
     {
       "epoch": 297.33,
-      "grad_norm": 7.331628322601318,
+      "grad_norm": 10.691654205322266,
       "learning_rate": 8.908908908908909e-08,
-      "loss": 0.9418,
+      "loss": 0.9482,
       "step": 99010
     },
     {
       "epoch": 297.36,
-      "grad_norm": 6.313700199127197,
+      "grad_norm": 6.5469465255737305,
       "learning_rate": 8.808808808808811e-08,
-      "loss": 0.8985,
+      "loss": 0.8876,
       "step": 99020
     },
     {
       "epoch": 297.39,
-      "grad_norm": 5.338436603546143,
+      "grad_norm": 6.552236557006836,
       "learning_rate": 8.70870870870871e-08,
-      "loss": 0.9539,
+      "loss": 0.9444,
       "step": 99030
     },
     {
       "epoch": 297.42,
-      "grad_norm": 5.639608383178711,
+      "grad_norm": 6.015292167663574,
       "learning_rate": 8.608608608608609e-08,
-      "loss": 0.8744,
+      "loss": 0.8771,
       "step": 99040
     },
     {
       "epoch": 297.45,
-      "grad_norm": 8.83679485321045,
+      "grad_norm": 8.983518600463867,
       "learning_rate": 8.50850850850851e-08,
-      "loss": 0.8625,
+      "loss": 0.8559,
       "step": 99050
     },
     {
       "epoch": 297.48,
-      "grad_norm": 5.839866638183594,
+      "grad_norm": 6.951140403747559,
       "learning_rate": 8.408408408408409e-08,
-      "loss": 0.9543,
+      "loss": 0.9447,
       "step": 99060
     },
     {
       "epoch": 297.51,
-      "grad_norm": 5.543236255645752,
+      "grad_norm": 6.991142749786377,
       "learning_rate": 8.308308308308309e-08,
-      "loss": 0.9387,
+      "loss": 0.9412,
       "step": 99070
     },
     {
       "epoch": 297.54,
-      "grad_norm": 7.26675271987915,
+      "grad_norm": 4.711069107055664,
       "learning_rate": 8.208208208208208e-08,
-      "loss": 0.954,
+      "loss": 0.9553,
       "step": 99080
     },
     {
       "epoch": 297.57,
-      "grad_norm": 7.6672563552856445,
+      "grad_norm": 8.618901252746582,
       "learning_rate": 8.108108108108109e-08,
-      "loss": 0.9597,
+      "loss": 0.9662,
       "step": 99090
     },
     {
       "epoch": 297.6,
-      "grad_norm": 5.714897155761719,
+      "grad_norm": 5.293499946594238,
       "learning_rate": 8.008008008008008e-08,
-      "loss": 0.9086,
+      "loss": 0.9084,
       "step": 99100
     },
     {
       "epoch": 297.63,
-      "grad_norm": 8.582242012023926,
+      "grad_norm": 6.790743350982666,
       "learning_rate": 7.907907907907907e-08,
-      "loss": 0.9094,
+      "loss": 0.8965,
       "step": 99110
     },
     {
       "epoch": 297.66,
-      "grad_norm": 4.860990047454834,
+      "grad_norm": 5.337958812713623,
       "learning_rate": 7.807807807807808e-08,
-      "loss": 0.8525,
+      "loss": 0.8517,
       "step": 99120
     },
     {
       "epoch": 297.69,
-      "grad_norm": 6.652520179748535,
+      "grad_norm": 8.264336585998535,
       "learning_rate": 7.707707707707708e-08,
-      "loss": 0.8993,
+      "loss": 0.9068,
       "step": 99130
     },
     {
       "epoch": 297.72,
-      "grad_norm": 9.60055160522461,
+      "grad_norm": 6.707715034484863,
       "learning_rate": 7.607607607607608e-08,
-      "loss": 0.8705,
+      "loss": 0.8663,
       "step": 99140
     },
     {
       "epoch": 297.75,
-      "grad_norm": 7.623078346252441,
+      "grad_norm": 9.000467300415039,
       "learning_rate": 7.507507507507508e-08,
-      "loss": 0.9637,
+      "loss": 0.9663,
       "step": 99150
     },
     {
       "epoch": 297.78,
-      "grad_norm": 6.410462856292725,
+      "grad_norm": 5.120025157928467,
       "learning_rate": 7.407407407407409e-08,
-      "loss": 0.8345,
+      "loss": 0.8354,
       "step": 99160
     },
     {
       "epoch": 297.81,
-      "grad_norm": 7.780923843383789,
+      "grad_norm": 7.918667793273926,
       "learning_rate": 7.307307307307308e-08,
-      "loss": 0.8491,
+      "loss": 0.8445,
       "step": 99170
     },
     {
       "epoch": 297.84,
-      "grad_norm": 4.788844585418701,
+      "grad_norm": 6.461482048034668,
       "learning_rate": 7.207207207207208e-08,
-      "loss": 0.9288,
+      "loss": 0.9254,
       "step": 99180
     },
     {
       "epoch": 297.87,
-      "grad_norm": 6.786112308502197,
+      "grad_norm": 7.990549564361572,
       "learning_rate": 7.107107107107107e-08,
-      "loss": 0.9228,
+      "loss": 0.9208,
       "step": 99190
     },
     {
       "epoch": 297.9,
-      "grad_norm": 5.183621406555176,
+      "grad_norm": 5.793527603149414,
       "learning_rate": 7.007007007007008e-08,
-      "loss": 0.9363,
+      "loss": 0.9349,
       "step": 99200
     },
     {
       "epoch": 297.93,
-      "grad_norm": 8.80527400970459,
+      "grad_norm": 5.486868381500244,
       "learning_rate": 6.906906906906907e-08,
-      "loss": 0.9656,
+      "loss": 0.9613,
       "step": 99210
     },
     {
       "epoch": 297.96,
-      "grad_norm": 4.973134517669678,
+      "grad_norm": 5.9121317863464355,
       "learning_rate": 6.806806806806808e-08,
-      "loss": 0.8771,
+      "loss": 0.8724,
       "step": 99220
     },
     {
       "epoch": 297.99,
-      "grad_norm": 5.043403148651123,
+      "grad_norm": 7.262234687805176,
       "learning_rate": 6.706706706706707e-08,
-      "loss": 0.8629,
+      "loss": 0.8547,
       "step": 99230
     },
     {
       "epoch": 298.0,
-      "eval_accuracy": 0.8262,
-      "eval_loss": 0.5732461810112,
-      "eval_runtime": 5.9343,
-      "eval_samples_per_second": 1685.111,
-      "eval_steps_per_second": 6.74,
+      "eval_accuracy": 0.8283,
+      "eval_loss": 0.5713414549827576,
+      "eval_runtime": 5.5077,
+      "eval_samples_per_second": 1815.627,
+      "eval_steps_per_second": 7.263,
       "step": 99234
     },
     {
       "epoch": 298.02,
-      "grad_norm": 6.396614074707031,
+      "grad_norm": 7.921267509460449,
       "learning_rate": 6.606606606606607e-08,
-      "loss": 1.1166,
+      "loss": 1.1257,
       "step": 99240
     },
     {
       "epoch": 298.05,
-      "grad_norm": 7.166237831115723,
+      "grad_norm": 4.585839748382568,
       "learning_rate": 6.506506506506507e-08,
-      "loss": 0.9422,
+      "loss": 0.939,
       "step": 99250
     },
     {
       "epoch": 298.08,
-      "grad_norm": 4.982211112976074,
+      "grad_norm": 6.477282524108887,
       "learning_rate": 6.406406406406407e-08,
-      "loss": 0.8959,
+      "loss": 0.8993,
       "step": 99260
     },
     {
       "epoch": 298.11,
-      "grad_norm": 4.090776443481445,
+      "grad_norm": 5.345349311828613,
       "learning_rate": 6.306306306306306e-08,
-      "loss": 0.9157,
+      "loss": 0.9219,
       "step": 99270
     },
     {
       "epoch": 298.14,
-      "grad_norm": 8.059222221374512,
+      "grad_norm": 11.773028373718262,
       "learning_rate": 6.206206206206207e-08,
-      "loss": 0.9006,
+      "loss": 0.9036,
       "step": 99280
     },
     {
       "epoch": 298.17,
-      "grad_norm": 7.697545528411865,
+      "grad_norm": 9.012651443481445,
       "learning_rate": 6.106106106106106e-08,
-      "loss": 0.9527,
+      "loss": 0.9499,
       "step": 99290
     },
     {
       "epoch": 298.2,
-      "grad_norm": 5.117954730987549,
+      "grad_norm": 5.583630084991455,
       "learning_rate": 6.006006006006006e-08,
-      "loss": 0.9031,
+      "loss": 0.9065,
       "step": 99300
     },
     {
       "epoch": 298.23,
-      "grad_norm": 4.851950645446777,
+      "grad_norm": 7.401758670806885,
       "learning_rate": 5.905905905905907e-08,
-      "loss": 0.9105,
+      "loss": 0.9063,
       "step": 99310
     },
     {
       "epoch": 298.26,
-      "grad_norm": 5.400831699371338,
+      "grad_norm": 6.402074337005615,
       "learning_rate": 5.805805805805806e-08,
-      "loss": 0.8788,
+      "loss": 0.8734,
       "step": 99320
     },
     {
       "epoch": 298.29,
-      "grad_norm": 7.512354850769043,
+      "grad_norm": 8.964186668395996,
       "learning_rate": 5.705705705705706e-08,
       "loss": 0.8942,
       "step": 99330
     },
     {
       "epoch": 298.32,
-      "grad_norm": 5.959722518920898,
+      "grad_norm": 6.425734996795654,
       "learning_rate": 5.605605605605606e-08,
-      "loss": 0.8814,
+      "loss": 0.8848,
       "step": 99340
     },
     {
       "epoch": 298.35,
-      "grad_norm": 7.578411102294922,
+      "grad_norm": 8.006776809692383,
       "learning_rate": 5.505505505505506e-08,
-      "loss": 0.8737,
+      "loss": 0.8808,
       "step": 99350
     },
     {
       "epoch": 298.38,
-      "grad_norm": 7.265635967254639,
+      "grad_norm": 7.227714538574219,
       "learning_rate": 5.405405405405406e-08,
-      "loss": 0.8421,
+      "loss": 0.8405,
       "step": 99360
     },
     {
       "epoch": 298.41,
-      "grad_norm": 4.803159236907959,
+      "grad_norm": 6.5481038093566895,
       "learning_rate": 5.3053053053053054e-08,
       "loss": 0.8739,
       "step": 99370
     },
     {
       "epoch": 298.44,
-      "grad_norm": 6.4690470695495605,
+      "grad_norm": 7.928037166595459,
       "learning_rate": 5.205205205205206e-08,
-      "loss": 0.9384,
+      "loss": 0.9364,
       "step": 99380
     },
     {
       "epoch": 298.47,
-      "grad_norm": 7.429672718048096,
+      "grad_norm": 6.0007452964782715,
       "learning_rate": 5.105105105105105e-08,
-      "loss": 0.8506,
+      "loss": 0.8436,
       "step": 99390
     },
     {
       "epoch": 298.5,
-      "grad_norm": 7.213605880737305,
+      "grad_norm": 6.875518321990967,
       "learning_rate": 5.0050050050050056e-08,
       "loss": 0.9329,
       "step": 99400
     },
     {
       "epoch": 298.53,
-      "grad_norm": 6.537362098693848,
+      "grad_norm": 9.358150482177734,
       "learning_rate": 4.9049049049049055e-08,
-      "loss": 0.9022,
+      "loss": 0.9025,
       "step": 99410
     },
     {
       "epoch": 298.56,
-      "grad_norm": 5.838995933532715,
+      "grad_norm": 7.701679229736328,
       "learning_rate": 4.804804804804805e-08,
-      "loss": 0.8514,
+      "loss": 0.8468,
       "step": 99420
     },
     {
       "epoch": 298.59,
-      "grad_norm": 8.816514015197754,
+      "grad_norm": 7.96947717666626,
       "learning_rate": 4.704704704704705e-08,
-      "loss": 0.8497,
+      "loss": 0.8355,
       "step": 99430
     },
     {
       "epoch": 298.62,
-      "grad_norm": 5.383877277374268,
+      "grad_norm": 5.8224711418151855,
       "learning_rate": 4.6046046046046043e-08,
-      "loss": 0.8775,
+      "loss": 0.8725,
       "step": 99440
     },
     {
       "epoch": 298.65,
-      "grad_norm": 9.927471160888672,
+      "grad_norm": 7.4014787673950195,
       "learning_rate": 4.504504504504505e-08,
-      "loss": 0.9669,
+      "loss": 0.9653,
       "step": 99450
     },
     {
       "epoch": 298.68,
-      "grad_norm": 7.0839524269104,
+      "grad_norm": 5.89166784286499,
       "learning_rate": 4.4044044044044054e-08,
-      "loss": 0.8232,
+      "loss": 0.8218,
       "step": 99460
     },
     {
       "epoch": 298.71,
-      "grad_norm": 8.043771743774414,
+      "grad_norm": 9.76279067993164,
       "learning_rate": 4.3043043043043045e-08,
-      "loss": 0.9186,
+      "loss": 0.9182,
       "step": 99470
     },
     {
       "epoch": 298.74,
-      "grad_norm": 7.054793834686279,
+      "grad_norm": 9.320079803466797,
       "learning_rate": 4.2042042042042044e-08,
-      "loss": 0.9079,
+      "loss": 0.9046,
       "step": 99480
     },
     {
       "epoch": 298.77,
-      "grad_norm": 6.015527725219727,
+      "grad_norm": 7.047538757324219,
       "learning_rate": 4.104104104104104e-08,
-      "loss": 0.8823,
+      "loss": 0.8638,
       "step": 99490
     },
     {
       "epoch": 298.8,
-      "grad_norm": 6.494959831237793,
+      "grad_norm": 7.721532821655273,
       "learning_rate": 4.004004004004004e-08,
-      "loss": 0.9576,
+      "loss": 0.9435,
       "step": 99500
     },
     {
       "epoch": 298.83,
-      "grad_norm": 5.466606616973877,
+      "grad_norm": 5.88856840133667,
       "learning_rate": 3.903903903903904e-08,
-      "loss": 0.8717,
+      "loss": 0.8743,
       "step": 99510
     },
     {
       "epoch": 298.86,
-      "grad_norm": 10.476134300231934,
+      "grad_norm": 11.054788589477539,
       "learning_rate": 3.803803803803804e-08,
-      "loss": 0.928,
+      "loss": 0.9292,
       "step": 99520
     },
     {
       "epoch": 298.89,
-      "grad_norm": 8.686444282531738,
+      "grad_norm": 12.428742408752441,
       "learning_rate": 3.703703703703704e-08,
-      "loss": 0.95,
+      "loss": 0.9428,
       "step": 99530
     },
     {
       "epoch": 298.92,
-      "grad_norm": 9.539712905883789,
+      "grad_norm": 12.257431983947754,
       "learning_rate": 3.603603603603604e-08,
-      "loss": 0.9685,
+      "loss": 0.9641,
       "step": 99540
     },
     {
       "epoch": 298.95,
-      "grad_norm": 7.552163600921631,
+      "grad_norm": 7.652799606323242,
       "learning_rate": 3.503503503503504e-08,
-      "loss": 0.8531,
+      "loss": 0.8551,
       "step": 99550
     },
     {
       "epoch": 298.98,
-      "grad_norm": 7.647386074066162,
+      "grad_norm": 13.74491024017334,
       "learning_rate": 3.403403403403404e-08,
-      "loss": 0.887,
+      "loss": 0.8904,
       "step": 99560
     },
     {
       "epoch": 299.0,
-      "eval_accuracy": 0.828,
-      "eval_loss": 0.5658428072929382,
-      "eval_runtime": 5.7002,
-      "eval_samples_per_second": 1754.309,
-      "eval_steps_per_second": 7.017,
+      "eval_accuracy": 0.8328,
+      "eval_loss": 0.5651447772979736,
+      "eval_runtime": 5.5268,
+      "eval_samples_per_second": 1809.355,
+      "eval_steps_per_second": 7.237,
       "step": 99567
     },
     {
       "epoch": 299.01,
-      "grad_norm": 8.459427833557129,
+      "grad_norm": 7.46140193939209,
       "learning_rate": 3.303303303303304e-08,
-      "loss": 1.0884,
+      "loss": 1.0861,
       "step": 99570
     },
     {
       "epoch": 299.04,
-      "grad_norm": 6.025773525238037,
+      "grad_norm": 8.382796287536621,
       "learning_rate": 3.2032032032032035e-08,
-      "loss": 0.9783,
+      "loss": 0.9767,
       "step": 99580
     },
     {
       "epoch": 299.07,
-      "grad_norm": 6.32681131362915,
+      "grad_norm": 7.4561381340026855,
       "learning_rate": 3.1031031031031034e-08,
-      "loss": 0.859,
+      "loss": 0.8598,
       "step": 99590
     },
     {
       "epoch": 299.1,
-      "grad_norm": 5.891650676727295,
+      "grad_norm": 5.297982692718506,
       "learning_rate": 3.003003003003003e-08,
-      "loss": 0.8707,
+      "loss": 0.8808,
       "step": 99600
     },
     {
       "epoch": 299.13,
-      "grad_norm": 5.688296794891357,
+      "grad_norm": 6.959763050079346,
       "learning_rate": 2.902902902902903e-08,
-      "loss": 0.8993,
+      "loss": 0.8928,
       "step": 99610
     },
     {
       "epoch": 299.16,
-      "grad_norm": 4.279796600341797,
+      "grad_norm": 5.013470649719238,
       "learning_rate": 2.802802802802803e-08,
-      "loss": 0.894,
+      "loss": 0.8787,
       "step": 99620
     },
     {
       "epoch": 299.19,
-      "grad_norm": 6.340704917907715,
+      "grad_norm": 5.614914417266846,
       "learning_rate": 2.702702702702703e-08,
-      "loss": 0.8465,
+      "loss": 0.8452,
       "step": 99630
     },
     {
       "epoch": 299.22,
-      "grad_norm": 6.150396347045898,
+      "grad_norm": 7.214180946350098,
       "learning_rate": 2.602602602602603e-08,
-      "loss": 0.9153,
+      "loss": 0.9068,
       "step": 99640
     },
     {
       "epoch": 299.25,
-      "grad_norm": 5.902482032775879,
+      "grad_norm": 6.8756208419799805,
       "learning_rate": 2.5025025025025028e-08,
-      "loss": 0.8265,
+      "loss": 0.8299,
       "step": 99650
     },
     {
       "epoch": 299.28,
-      "grad_norm": 8.286025047302246,
+      "grad_norm": 11.92014217376709,
       "learning_rate": 2.4024024024024027e-08,
-      "loss": 0.8757,
+      "loss": 0.8836,
       "step": 99660
     },
     {
       "epoch": 299.31,
-      "grad_norm": 6.070191860198975,
+      "grad_norm": 6.162424087524414,
       "learning_rate": 2.3023023023023022e-08,
-      "loss": 0.8951,
+      "loss": 0.8992,
       "step": 99670
     },
     {
       "epoch": 299.34,
-      "grad_norm": 9.538198471069336,
+      "grad_norm": 9.208022117614746,
       "learning_rate": 2.2022022022022027e-08,
-      "loss": 0.9389,
+      "loss": 0.9442,
       "step": 99680
     },
     {
       "epoch": 299.37,
-      "grad_norm": 7.474327087402344,
+      "grad_norm": 8.076813697814941,
       "learning_rate": 2.1021021021021022e-08,
-      "loss": 0.9599,
+      "loss": 0.9545,
       "step": 99690
     },
     {
       "epoch": 299.4,
-      "grad_norm": 7.08575963973999,
+      "grad_norm": 6.862570285797119,
       "learning_rate": 2.002002002002002e-08,
-      "loss": 0.8497,
+      "loss": 0.8525,
       "step": 99700
     },
     {
       "epoch": 299.43,
-      "grad_norm": 6.244769096374512,
+      "grad_norm": 5.814998149871826,
       "learning_rate": 1.901901901901902e-08,
-      "loss": 0.8883,
+      "loss": 0.8872,
       "step": 99710
     },
     {
       "epoch": 299.46,
-      "grad_norm": 7.262641906738281,
+      "grad_norm": 7.172391414642334,
       "learning_rate": 1.801801801801802e-08,
-      "loss": 0.8244,
+      "loss": 0.8311,
       "step": 99720
     },
     {
       "epoch": 299.49,
-      "grad_norm": 8.70535659790039,
+      "grad_norm": 8.238947868347168,
       "learning_rate": 1.701701701701702e-08,
-      "loss": 0.9263,
+      "loss": 0.9239,
       "step": 99730
     },
     {
       "epoch": 299.52,
-      "grad_norm": 10.411171913146973,
+      "grad_norm": 10.633116722106934,
       "learning_rate": 1.6016016016016018e-08,
-      "loss": 0.9361,
+      "loss": 0.9369,
       "step": 99740
     },
     {
       "epoch": 299.55,
-      "grad_norm": 6.143913745880127,
+      "grad_norm": 6.939387321472168,
       "learning_rate": 1.5015015015015016e-08,
-      "loss": 0.8829,
+      "loss": 0.879,
       "step": 99750
     },
     {
       "epoch": 299.58,
-      "grad_norm": 5.337463855743408,
+      "grad_norm": 7.942529678344727,
       "learning_rate": 1.4014014014014015e-08,
-      "loss": 0.8565,
+      "loss": 0.8572,
       "step": 99760
     },
     {
       "epoch": 299.61,
-      "grad_norm": 9.014431953430176,
+      "grad_norm": 9.416084289550781,
       "learning_rate": 1.3013013013013015e-08,
-      "loss": 0.9151,
+      "loss": 0.9057,
       "step": 99770
     },
     {
       "epoch": 299.64,
-      "grad_norm": 6.923699378967285,
+      "grad_norm": 7.586679458618164,
       "learning_rate": 1.2012012012012013e-08,
-      "loss": 0.8827,
+      "loss": 0.8855,
       "step": 99780
     },
     {
       "epoch": 299.67,
-      "grad_norm": 6.877496242523193,
+      "grad_norm": 5.9510273933410645,
       "learning_rate": 1.1011011011011013e-08,
       "loss": 0.9035,
       "step": 99790
     },
     {
       "epoch": 299.7,
-      "grad_norm": 7.297639846801758,
+      "grad_norm": 9.52353572845459,
       "learning_rate": 1.001001001001001e-08,
-      "loss": 0.8767,
+      "loss": 0.8728,
       "step": 99800
     },
     {
       "epoch": 299.73,
-      "grad_norm": 5.776065349578857,
+      "grad_norm": 7.854809284210205,
       "learning_rate": 9.00900900900901e-09,
-      "loss": 0.8981,
+      "loss": 0.9017,
       "step": 99810
     },
     {
       "epoch": 299.76,
-      "grad_norm": 5.922935962677002,
+      "grad_norm": 5.274312973022461,
       "learning_rate": 8.008008008008009e-09,
-      "loss": 0.8846,
+      "loss": 0.8826,
       "step": 99820
     },
     {
       "epoch": 299.79,
-      "grad_norm": 6.165457725524902,
+      "grad_norm": 8.946617126464844,
       "learning_rate": 7.007007007007007e-09,
-      "loss": 0.9079,
+      "loss": 0.9106,
       "step": 99830
     },
     {
       "epoch": 299.82,
-      "grad_norm": 5.6747636795043945,
+      "grad_norm": 6.330191612243652,
       "learning_rate": 6.006006006006007e-09,
-      "loss": 0.8922,
+      "loss": 0.8878,
       "step": 99840
     },
     {
       "epoch": 299.85,
-      "grad_norm": 7.47860050201416,
+      "grad_norm": 7.810037612915039,
       "learning_rate": 5.005005005005005e-09,
-      "loss": 0.9176,
+      "loss": 0.9058,
       "step": 99850
     },
     {
       "epoch": 299.88,
-      "grad_norm": 4.776383876800537,
+      "grad_norm": 4.56422233581543,
       "learning_rate": 4.0040040040040044e-09,
-      "loss": 0.8705,
+      "loss": 0.879,
       "step": 99860
     },
     {
       "epoch": 299.91,
-      "grad_norm": 8.419172286987305,
+      "grad_norm": 9.196592330932617,
       "learning_rate": 3.0030030030030033e-09,
-      "loss": 0.9626,
+      "loss": 0.9569,
       "step": 99870
     },
     {
       "epoch": 299.94,
-      "grad_norm": 8.259916305541992,
+      "grad_norm": 9.700933456420898,
       "learning_rate": 2.0020020020020022e-09,
-      "loss": 0.9802,
+      "loss": 0.9695,
       "step": 99880
     },
     {
       "epoch": 299.97,
-      "grad_norm": 5.5089311599731445,
+      "grad_norm": 4.170505523681641,
       "learning_rate": 1.0010010010010011e-09,
-      "loss": 0.8332,
+      "loss": 0.8317,
       "step": 99890
     },
     {
       "epoch": 300.0,
-      "grad_norm": 95.11062622070312,
+      "grad_norm": 131.86001586914062,
       "learning_rate": 0.0,
-      "loss": 1.0707,
+      "loss": 1.0262,
       "step": 99900
     },
     {
       "epoch": 300.0,
-      "eval_accuracy": 0.8286,
-      "eval_loss": 0.5731849670410156,
-      "eval_runtime": 5.7684,
-      "eval_samples_per_second": 1733.57,
-      "eval_steps_per_second": 6.934,
+      "eval_accuracy": 0.8276,
+      "eval_loss": 0.5721720457077026,
+      "eval_runtime": 5.4844,
+      "eval_samples_per_second": 1823.338,
+      "eval_steps_per_second": 7.293,
       "step": 99900
     },
     {
       "epoch": 300.0,
       "step": 99900,
       "total_flos": 2.73063707449344e+20,
-      "train_loss": 1.232413830227322,
-      "train_runtime": 17888.831,
-      "train_samples_per_second": 712.735,
-      "train_steps_per_second": 5.584
+      "train_loss": 1.2348953081155802,
+      "train_runtime": 17779.0058,
+      "train_samples_per_second": 717.138,
+      "train_steps_per_second": 5.619
     }
   ],
   "logging_steps": 10,