diff --git "a/checkpoint-10000/trainer_state.json" "b/checkpoint-10000/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-10000/trainer_state.json"
@@ -0,0 +1,7021 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.8622190667954701,
+  "eval_steps": 500,
+  "global_step": 10000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "grad_norm": 0.9869178533554077,
+      "learning_rate": 9.99997860104108e-06,
+      "loss": 1.4113,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 0.7985073924064636,
+      "learning_rate": 9.999914404347487e-06,
+      "loss": 1.3054,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.5533589720726013,
+      "learning_rate": 9.999807410468713e-06,
+      "loss": 1.2698,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.527658224105835,
+      "learning_rate": 9.999657620320587e-06,
+      "loss": 1.1817,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.5116757750511169,
+      "learning_rate": 9.999465035185248e-06,
+      "loss": 1.1681,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.40034493803977966,
+      "learning_rate": 9.999229656711143e-06,
+      "loss": 1.1709,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.41553106904029846,
+      "learning_rate": 9.998951486913015e-06,
+      "loss": 1.133,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.38187628984451294,
+      "learning_rate": 9.998630528171882e-06,
+      "loss": 1.0915,
+      "step": 80
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.39334940910339355,
+      "learning_rate": 9.998266783235018e-06,
+      "loss": 1.1154,
+      "step": 90
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.38754093647003174,
+      "learning_rate": 9.997860255215926e-06,
+      "loss": 1.0616,
+      "step": 100
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.3934289216995239,
+      "learning_rate": 9.997410947594317e-06,
+      "loss": 1.1031,
+      "step": 110
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.44751420617103577,
+      "learning_rate": 9.996918864216081e-06,
+      "loss": 1.0547,
+      "step": 120
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.4995039701461792,
+      "learning_rate": 9.99638400929324e-06,
+      "loss": 1.0634,
+      "step": 130
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.519430935382843,
+      "learning_rate": 9.995806387403935e-06,
+      "loss": 1.0366,
+      "step": 140
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.4063739478588104,
+      "learning_rate": 9.995186003492366e-06,
+      "loss": 1.0329,
+      "step": 150
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.47834908962249756,
+      "learning_rate": 9.994522862868763e-06,
+      "loss": 1.0168,
+      "step": 160
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.4666057825088501,
+      "learning_rate": 9.993816971209332e-06,
+      "loss": 1.044,
+      "step": 170
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5252623558044434,
+      "learning_rate": 9.99306833455621e-06,
+      "loss": 1.0311,
+      "step": 180
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.41999587416648865,
+      "learning_rate": 9.992276959317419e-06,
+      "loss": 1.0273,
+      "step": 190
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4322344660758972,
+      "learning_rate": 9.9914428522668e-06,
+      "loss": 1.0174,
+      "step": 200
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.44230663776397705,
+      "learning_rate": 9.99056602054396e-06,
+      "loss": 0.9849,
+      "step": 210
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.48284780979156494,
+      "learning_rate": 9.989646471654216e-06,
+      "loss": 1.0219,
+      "step": 220
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.478462815284729,
+      "learning_rate": 9.988684213468521e-06,
+      "loss": 1.01,
+      "step": 230
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4927475154399872,
+      "learning_rate": 9.987679254223405e-06,
+      "loss": 1.0199,
+      "step": 240
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4281039237976074,
+      "learning_rate": 9.986631602520904e-06,
+      "loss": 1.0072,
+      "step": 250
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.5018512010574341,
+      "learning_rate": 9.985541267328479e-06,
+      "loss": 0.9704,
+      "step": 260
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4972495138645172,
+      "learning_rate": 9.98440825797894e-06,
+      "loss": 0.9858,
+      "step": 270
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.5815178751945496,
+      "learning_rate": 9.98323258417038e-06,
+      "loss": 0.9752,
+      "step": 280
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.6393250823020935,
+      "learning_rate": 9.982014255966078e-06,
+      "loss": 1.0034,
+      "step": 290
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.5562496781349182,
+      "learning_rate": 9.980753283794414e-06,
+      "loss": 1.0004,
+      "step": 300
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.48535385727882385,
+      "learning_rate": 9.979449678448785e-06,
+      "loss": 0.995,
+      "step": 310
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.6010926365852356,
+      "learning_rate": 9.978103451087511e-06,
+      "loss": 0.9669,
+      "step": 320
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.5188761949539185,
+      "learning_rate": 9.976714613233736e-06,
+      "loss": 0.978,
+      "step": 330
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.7140072584152222,
+      "learning_rate": 9.975283176775332e-06,
+      "loss": 0.9778,
+      "step": 340
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.5040972232818604,
+      "learning_rate": 9.973809153964803e-06,
+      "loss": 0.984,
+      "step": 350
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.557113528251648,
+      "learning_rate": 9.97229255741917e-06,
+      "loss": 0.9871,
+      "step": 360
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.6509271860122681,
+      "learning_rate": 9.970733400119865e-06,
+      "loss": 0.9797,
+      "step": 370
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.6498264670372009,
+      "learning_rate": 9.969131695412629e-06,
+      "loss": 0.9665,
+      "step": 380
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.6095420718193054,
+      "learning_rate": 9.967487457007382e-06,
+      "loss": 0.9589,
+      "step": 390
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.5704444646835327,
+      "learning_rate": 9.965800698978126e-06,
+      "loss": 0.9666,
+      "step": 400
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.5877268314361572,
+      "learning_rate": 9.964071435762802e-06,
+      "loss": 0.9999,
+      "step": 410
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.5909533500671387,
+      "learning_rate": 9.962299682163185e-06,
+      "loss": 0.9381,
+      "step": 420
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.5712386965751648,
+      "learning_rate": 9.96048545334475e-06,
+      "loss": 0.972,
+      "step": 430
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.737307608127594,
+      "learning_rate": 9.95862876483654e-06,
+      "loss": 0.9754,
+      "step": 440
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.7787168025970459,
+      "learning_rate": 9.956729632531032e-06,
+      "loss": 1.0197,
+      "step": 450
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.5988175868988037,
+      "learning_rate": 9.954788072684011e-06,
+      "loss": 0.9596,
+      "step": 460
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.981139600276947,
+      "learning_rate": 9.952804101914418e-06,
+      "loss": 0.9876,
+      "step": 470
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.6938815712928772,
+      "learning_rate": 9.95077773720422e-06,
+      "loss": 0.9533,
+      "step": 480
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.671062707901001,
+      "learning_rate": 9.948708995898251e-06,
+      "loss": 0.9812,
+      "step": 490
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.6615872979164124,
+      "learning_rate": 9.94659789570408e-06,
+      "loss": 0.9522,
+      "step": 500
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.5831172466278076,
+      "learning_rate": 9.94444445469184e-06,
+      "loss": 0.9454,
+      "step": 510
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.6584324836730957,
+      "learning_rate": 9.942248691294092e-06,
+      "loss": 0.951,
+      "step": 520
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.6748599410057068,
+      "learning_rate": 9.940010624305658e-06,
+      "loss": 0.927,
+      "step": 530
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.5425313115119934,
+      "learning_rate": 9.937730272883458e-06,
+      "loss": 0.9542,
+      "step": 540
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.551030158996582,
+      "learning_rate": 9.93540765654635e-06,
+      "loss": 0.933,
+      "step": 550
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.743243396282196,
+      "learning_rate": 9.933042795174964e-06,
+      "loss": 0.9573,
+      "step": 560
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.6725818514823914,
+      "learning_rate": 9.930635709011524e-06,
+      "loss": 0.9745,
+      "step": 570
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.743486225605011,
+      "learning_rate": 9.928186418659693e-06,
+      "loss": 0.9705,
+      "step": 580
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.6531779170036316,
+      "learning_rate": 9.925694945084369e-06,
+      "loss": 0.9526,
+      "step": 590
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.7471463680267334,
+      "learning_rate": 9.923161309611534e-06,
+      "loss": 0.9718,
+      "step": 600
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.6551310420036316,
+      "learning_rate": 9.920585533928049e-06,
+      "loss": 0.9738,
+      "step": 610
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.835099458694458,
+      "learning_rate": 9.917967640081482e-06,
+      "loss": 0.9559,
+      "step": 620
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.6788685321807861,
+      "learning_rate": 9.915307650479915e-06,
+      "loss": 0.9448,
+      "step": 630
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.640802800655365,
+      "learning_rate": 9.912605587891748e-06,
+      "loss": 0.9632,
+      "step": 640
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.7521352767944336,
+      "learning_rate": 9.909861475445517e-06,
+      "loss": 0.9947,
+      "step": 650
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.6148037314414978,
+      "learning_rate": 9.907075336629679e-06,
+      "loss": 0.8992,
+      "step": 660
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.7000744938850403,
+      "learning_rate": 9.90424719529242e-06,
+      "loss": 0.9978,
+      "step": 670
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.6232642531394958,
+      "learning_rate": 9.901377075641457e-06,
+      "loss": 0.9916,
+      "step": 680
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.6425381898880005,
+      "learning_rate": 9.898465002243813e-06,
+      "loss": 0.9478,
+      "step": 690
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.6344175934791565,
+      "learning_rate": 9.89551100002563e-06,
+      "loss": 0.9749,
+      "step": 700
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.677696704864502,
+      "learning_rate": 9.892515094271931e-06,
+      "loss": 0.9815,
+      "step": 710
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.6174888610839844,
+      "learning_rate": 9.889477310626426e-06,
+      "loss": 0.9444,
+      "step": 720
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.7845975756645203,
+      "learning_rate": 9.886397675091275e-06,
+      "loss": 0.9751,
+      "step": 730
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.844460666179657,
+      "learning_rate": 9.883276214026877e-06,
+      "loss": 0.9412,
+      "step": 740
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.6752843260765076,
+      "learning_rate": 9.880112954151639e-06,
+      "loss": 0.95,
+      "step": 750
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.7257810831069946,
+      "learning_rate": 9.876907922541748e-06,
+      "loss": 0.9468,
+      "step": 760
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.6597475409507751,
+      "learning_rate": 9.87366114663094e-06,
+      "loss": 0.9219,
+      "step": 770
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.6881552934646606,
+      "learning_rate": 9.870372654210265e-06,
+      "loss": 0.9571,
+      "step": 780
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8410021662712097,
+      "learning_rate": 9.867042473427848e-06,
+      "loss": 0.9429,
+      "step": 790
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.6908501982688904,
+      "learning_rate": 9.863670632788652e-06,
+      "loss": 0.9618,
+      "step": 800
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.6102294921875,
+      "learning_rate": 9.860257161154224e-06,
+      "loss": 0.9539,
+      "step": 810
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8168625831604004,
+      "learning_rate": 9.856802087742463e-06,
+      "loss": 0.9384,
+      "step": 820
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.6453863978385925,
+      "learning_rate": 9.85330544212736e-06,
+      "loss": 0.9341,
+      "step": 830
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.6028039455413818,
+      "learning_rate": 9.849767254238741e-06,
+      "loss": 0.95,
+      "step": 840
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.6181658506393433,
+      "learning_rate": 9.846187554362026e-06,
+      "loss": 0.9184,
+      "step": 850
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.7346929907798767,
+      "learning_rate": 9.842566373137949e-06,
+      "loss": 0.9714,
+      "step": 860
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.6204361915588379,
+      "learning_rate": 9.83890374156232e-06,
+      "loss": 0.9161,
+      "step": 870
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.7579419016838074,
+      "learning_rate": 9.835199690985737e-06,
+      "loss": 0.943,
+      "step": 880
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.6947230100631714,
+      "learning_rate": 9.831454253113328e-06,
+      "loss": 0.9092,
+      "step": 890
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.887786865234375,
+      "learning_rate": 9.827667460004487e-06,
+      "loss": 0.9149,
+      "step": 900
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.7788590788841248,
+      "learning_rate": 9.823839344072582e-06,
+      "loss": 0.9187,
+      "step": 910
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.6010196208953857,
+      "learning_rate": 9.81996993808469e-06,
+      "loss": 0.9445,
+      "step": 920
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.6567085385322571,
+      "learning_rate": 9.816059275161318e-06,
+      "loss": 0.9492,
+      "step": 930
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.8992276191711426,
+      "learning_rate": 9.812107388776113e-06,
+      "loss": 0.8785,
+      "step": 940
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.6781167387962341,
+      "learning_rate": 9.808114312755574e-06,
+      "loss": 0.902,
+      "step": 950
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.6629174947738647,
+      "learning_rate": 9.804080081278768e-06,
+      "loss": 0.9416,
+      "step": 960
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.8017804026603699,
+      "learning_rate": 9.800004728877042e-06,
+      "loss": 0.9273,
+      "step": 970
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.6197686195373535,
+      "learning_rate": 9.795888290433709e-06,
+      "loss": 0.9335,
+      "step": 980
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.7496291995048523,
+      "learning_rate": 9.791730801183772e-06,
+      "loss": 0.9036,
+      "step": 990
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.6641804575920105,
+      "learning_rate": 9.787532296713605e-06,
+      "loss": 0.9768,
+      "step": 1000
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.7570294737815857,
+      "learning_rate": 9.78329281296066e-06,
+      "loss": 0.9162,
+      "step": 1010
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.7247979044914246,
+      "learning_rate": 9.779012386213151e-06,
+      "loss": 0.9261,
+      "step": 1020
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.7128037214279175,
+      "learning_rate": 9.77469105310975e-06,
+      "loss": 0.9403,
+      "step": 1030
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.7633296847343445,
+      "learning_rate": 9.770328850639268e-06,
+      "loss": 0.9184,
+      "step": 1040
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.7282195687294006,
+      "learning_rate": 9.76592581614034e-06,
+      "loss": 0.9144,
+      "step": 1050
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.7485966086387634,
+      "learning_rate": 9.761481987301111e-06,
+      "loss": 0.9307,
+      "step": 1060
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.7050462961196899,
+      "learning_rate": 9.756997402158904e-06,
+      "loss": 0.9269,
+      "step": 1070
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.6218549609184265,
+      "learning_rate": 9.752472099099897e-06,
+      "loss": 0.9497,
+      "step": 1080
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.019995927810669,
+      "learning_rate": 9.747906116858805e-06,
+      "loss": 0.9218,
+      "step": 1090
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.6491205096244812,
+      "learning_rate": 9.743299494518532e-06,
+      "loss": 0.9447,
+      "step": 1100
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.6476668119430542,
+      "learning_rate": 9.738652271509846e-06,
+      "loss": 0.9189,
+      "step": 1110
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.6570378541946411,
+      "learning_rate": 9.733964487611044e-06,
+      "loss": 0.938,
+      "step": 1120
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.6387359499931335,
+      "learning_rate": 9.729236182947597e-06,
+      "loss": 0.933,
+      "step": 1130
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.5747265815734863,
+      "learning_rate": 9.724467397991833e-06,
+      "loss": 0.9155,
+      "step": 1140
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.6549525260925293,
+      "learning_rate": 9.719658173562558e-06,
+      "loss": 0.9293,
+      "step": 1150
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.6779614686965942,
+      "learning_rate": 9.714808550824735e-06,
+      "loss": 0.9377,
+      "step": 1160
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.6967174410820007,
+      "learning_rate": 9.709918571289114e-06,
+      "loss": 0.9253,
+      "step": 1170
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.7676512598991394,
+      "learning_rate": 9.704988276811883e-06,
+      "loss": 0.9454,
+      "step": 1180
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.655295193195343,
+      "learning_rate": 9.70001770959431e-06,
+      "loss": 0.9125,
+      "step": 1190
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.7563107013702393,
+      "learning_rate": 9.695006912182379e-06,
+      "loss": 0.9405,
+      "step": 1200
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.8399559259414673,
+      "learning_rate": 9.68995592746643e-06,
+      "loss": 0.9427,
+      "step": 1210
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.7510868906974792,
+      "learning_rate": 9.684864798680789e-06,
+      "loss": 0.9303,
+      "step": 1220
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.9126793146133423,
+      "learning_rate": 9.679733569403398e-06,
+      "loss": 0.9218,
+      "step": 1230
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.6488207578659058,
+      "learning_rate": 9.674562283555445e-06,
+      "loss": 0.943,
+      "step": 1240
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.9493321776390076,
+      "learning_rate": 9.669350985400979e-06,
+      "loss": 0.9429,
+      "step": 1250
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.7507259845733643,
+      "learning_rate": 9.664099719546547e-06,
+      "loss": 0.9041,
+      "step": 1260
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.7052815556526184,
+      "learning_rate": 9.658808530940794e-06,
+      "loss": 0.9214,
+      "step": 1270
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.7954320311546326,
+      "learning_rate": 9.653477464874093e-06,
+      "loss": 0.9178,
+      "step": 1280
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.0493558645248413,
+      "learning_rate": 9.648106566978149e-06,
+      "loss": 0.9019,
+      "step": 1290
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.767572820186615,
+      "learning_rate": 9.64269588322561e-06,
+      "loss": 0.9551,
+      "step": 1300
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.7655388712882996,
+      "learning_rate": 9.637245459929678e-06,
+      "loss": 0.9281,
+      "step": 1310
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.2685116529464722,
+      "learning_rate": 9.631755343743707e-06,
+      "loss": 0.9443,
+      "step": 1320
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.6689302325248718,
+      "learning_rate": 9.626225581660802e-06,
+      "loss": 0.9128,
+      "step": 1330
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.8155603408813477,
+      "learning_rate": 9.620656221013428e-06,
+      "loss": 0.9437,
+      "step": 1340
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.7464457154273987,
+      "learning_rate": 9.61504730947299e-06,
+      "loss": 0.891,
+      "step": 1350
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.629400908946991,
+      "learning_rate": 9.609398895049435e-06,
+      "loss": 0.8955,
+      "step": 1360
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.7425062656402588,
+      "learning_rate": 9.60371102609084e-06,
+      "loss": 0.8861,
+      "step": 1370
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.6443877816200256,
+      "learning_rate": 9.597983751282993e-06,
+      "loss": 0.9051,
+      "step": 1380
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.7057684063911438,
+      "learning_rate": 9.592217119648982e-06,
+      "loss": 0.8885,
+      "step": 1390
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.8697768449783325,
+      "learning_rate": 9.586411180548771e-06,
+      "loss": 0.914,
+      "step": 1400
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.7217336893081665,
+      "learning_rate": 9.580565983678784e-06,
+      "loss": 0.9314,
+      "step": 1410
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.609015703201294,
+      "learning_rate": 9.57468157907147e-06,
+      "loss": 0.9134,
+      "step": 1420
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.7643072605133057,
+      "learning_rate": 9.568758017094884e-06,
+      "loss": 0.9209,
+      "step": 1430
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.6023324728012085,
+      "learning_rate": 9.562795348452245e-06,
+      "loss": 0.929,
+      "step": 1440
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.6700206398963928,
+      "learning_rate": 9.55679362418152e-06,
+      "loss": 0.9361,
+      "step": 1450
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.6590617895126343,
+      "learning_rate": 9.550752895654963e-06,
+      "loss": 0.8946,
+      "step": 1460
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.6463683843612671,
+      "learning_rate": 9.544673214578699e-06,
+      "loss": 0.9069,
+      "step": 1470
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.6586316823959351,
+      "learning_rate": 9.538554632992265e-06,
+      "loss": 0.9218,
+      "step": 1480
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.6628296375274658,
+      "learning_rate": 9.532397203268172e-06,
+      "loss": 0.8845,
+      "step": 1490
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.6902598142623901,
+      "learning_rate": 9.526200978111452e-06,
+      "loss": 0.9196,
+      "step": 1500
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.6551186442375183,
+      "learning_rate": 9.519966010559217e-06,
+      "loss": 0.9147,
+      "step": 1510
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.6625112295150757,
+      "learning_rate": 9.513692353980186e-06,
+      "loss": 0.912,
+      "step": 1520
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.7705367207527161,
+      "learning_rate": 9.507380062074252e-06,
+      "loss": 0.9298,
+      "step": 1530
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.6840114593505859,
+      "learning_rate": 9.501029188872005e-06,
+      "loss": 0.9199,
+      "step": 1540
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.6210880875587463,
+      "learning_rate": 9.49463978873427e-06,
+      "loss": 0.9077,
+      "step": 1550
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.7556228637695312,
+      "learning_rate": 9.488211916351656e-06,
+      "loss": 0.9089,
+      "step": 1560
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.735604465007782,
+      "learning_rate": 9.481745626744071e-06,
+      "loss": 0.9305,
+      "step": 1570
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.7971885800361633,
+      "learning_rate": 9.475240975260266e-06,
+      "loss": 0.9319,
+      "step": 1580
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.6014077067375183,
+      "learning_rate": 9.468698017577344e-06,
+      "loss": 0.9292,
+      "step": 1590
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.7462084889411926,
+      "learning_rate": 9.4621168097003e-06,
+      "loss": 0.8942,
+      "step": 1600
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.7277187705039978,
+      "learning_rate": 9.455497407961533e-06,
+      "loss": 0.8978,
+      "step": 1610
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.7430045008659363,
+      "learning_rate": 9.448839869020364e-06,
+      "loss": 0.9175,
+      "step": 1620
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.693151593208313,
+      "learning_rate": 9.442144249862555e-06,
+      "loss": 0.9416,
+      "step": 1630
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.6412799954414368,
+      "learning_rate": 9.435410607799818e-06,
+      "loss": 0.8838,
+      "step": 1640
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.8170580863952637,
+      "learning_rate": 9.428639000469326e-06,
+      "loss": 0.924,
+      "step": 1650
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.699052095413208,
+      "learning_rate": 9.421829485833214e-06,
+      "loss": 0.9344,
+      "step": 1660
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.7370940446853638,
+      "learning_rate": 9.414982122178095e-06,
+      "loss": 0.935,
+      "step": 1670
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.7777985334396362,
+      "learning_rate": 9.40809696811455e-06,
+      "loss": 0.933,
+      "step": 1680
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.7101726531982422,
+      "learning_rate": 9.40117408257663e-06,
+      "loss": 0.9002,
+      "step": 1690
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.6398364305496216,
+      "learning_rate": 9.394213524821351e-06,
+      "loss": 0.9449,
+      "step": 1700
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.8372628688812256,
+      "learning_rate": 9.387215354428192e-06,
+      "loss": 0.916,
+      "step": 1710
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.6209415793418884,
+      "learning_rate": 9.380179631298573e-06,
+      "loss": 0.9391,
+      "step": 1720
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.6791089177131653,
+      "learning_rate": 9.373106415655359e-06,
+      "loss": 0.8643,
+      "step": 1730
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.681890070438385,
+      "learning_rate": 9.365995768042327e-06,
+      "loss": 0.9468,
+      "step": 1740
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.7921962738037109,
+      "learning_rate": 9.35884774932366e-06,
+      "loss": 0.9127,
+      "step": 1750
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.7955833077430725,
+      "learning_rate": 9.351662420683421e-06,
+      "loss": 0.9091,
+      "step": 1760
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.7748980522155762,
+      "learning_rate": 9.344439843625034e-06,
+      "loss": 0.897,
+      "step": 1770
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.8075698614120483,
+      "learning_rate": 9.337180079970747e-06,
+      "loss": 0.9111,
+      "step": 1780
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.6480504274368286,
+      "learning_rate": 9.329883191861116e-06,
+      "loss": 0.9003,
+      "step": 1790
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.6438207626342773,
+      "learning_rate": 9.322549241754465e-06,
+      "loss": 0.8922,
+      "step": 1800
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.7411364912986755,
+      "learning_rate": 9.31517829242635e-06,
+      "loss": 0.8988,
+      "step": 1810
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.864782989025116,
+      "learning_rate": 9.307770406969032e-06,
+      "loss": 0.9395,
+      "step": 1820
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.7258287668228149,
+      "learning_rate": 9.300325648790922e-06,
+      "loss": 0.9025,
+      "step": 1830
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.7146714925765991,
+      "learning_rate": 9.29284408161605e-06,
+      "loss": 0.9223,
+      "step": 1840
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.6053746342658997,
+      "learning_rate": 9.285325769483517e-06,
+      "loss": 0.9332,
+      "step": 1850
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.8638799786567688,
+      "learning_rate": 9.277770776746944e-06,
+      "loss": 0.942,
+      "step": 1860
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.8022944927215576,
+      "learning_rate": 9.270179168073923e-06,
+      "loss": 0.909,
+      "step": 1870
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.7210090756416321,
+      "learning_rate": 9.26255100844546e-06,
+      "loss": 0.9314,
+      "step": 1880
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.650475263595581,
+      "learning_rate": 9.254886363155429e-06,
+      "loss": 0.9355,
+      "step": 1890
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.8607627153396606,
+      "learning_rate": 9.247185297809997e-06,
+      "loss": 0.9138,
+      "step": 1900
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.9340272545814514,
+      "learning_rate": 9.239447878327081e-06,
+      "loss": 0.9417,
+      "step": 1910
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.7494912147521973,
+      "learning_rate": 9.231674170935767e-06,
+      "loss": 0.8857,
+      "step": 1920
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.7654171586036682,
+      "learning_rate": 9.223864242175756e-06,
+      "loss": 0.8908,
+      "step": 1930
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.7642711400985718,
+      "learning_rate": 9.21601815889678e-06,
+      "loss": 0.8765,
+      "step": 1940
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.696348249912262,
+      "learning_rate": 9.20813598825805e-06,
+      "loss": 0.9323,
+      "step": 1950
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.6663697957992554,
+      "learning_rate": 9.200217797727663e-06,
+      "loss": 0.8775,
+      "step": 1960
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.7656813859939575,
+      "learning_rate": 9.192263655082033e-06,
+      "loss": 0.8997,
+      "step": 1970
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.6377949714660645,
+      "learning_rate": 9.184273628405304e-06,
+      "loss": 0.8842,
+      "step": 1980
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.7093709707260132,
+      "learning_rate": 9.176247786088783e-06,
+      "loss": 0.9002,
+      "step": 1990
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.875128984451294,
+      "learning_rate": 9.168186196830336e-06,
+      "loss": 0.9209,
+      "step": 2000
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.7550728917121887,
+      "learning_rate": 9.16008892963381e-06,
+      "loss": 0.9035,
+      "step": 2010
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.0108016729354858,
+      "learning_rate": 9.15195605380844e-06,
+      "loss": 0.875,
+      "step": 2020
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.7439371943473816,
+      "learning_rate": 9.143787638968255e-06,
+      "loss": 0.8722,
+      "step": 2030
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.6964495182037354,
+      "learning_rate": 9.135583755031486e-06,
+      "loss": 0.8726,
+      "step": 2040
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.9313227534294128,
+      "learning_rate": 9.127344472219964e-06,
+      "loss": 0.9107,
+      "step": 2050
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.4056764841079712,
+      "learning_rate": 9.119069861058516e-06,
+      "loss": 0.9061,
+      "step": 2060
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.680668830871582,
+      "learning_rate": 9.110759992374369e-06,
+      "loss": 0.9099,
+      "step": 2070
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.7828279733657837,
+      "learning_rate": 9.102414937296542e-06,
+      "loss": 0.8803,
+      "step": 2080
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.6008317470550537,
+      "learning_rate": 9.094034767255225e-06,
+      "loss": 0.8987,
+      "step": 2090
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.8024834990501404,
+      "learning_rate": 9.085619553981186e-06,
+      "loss": 0.9235,
+      "step": 2100
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.8021408915519714,
+      "learning_rate": 9.07716936950515e-06,
+      "loss": 0.8767,
+      "step": 2110
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.7997848391532898,
+      "learning_rate": 9.068684286157174e-06,
+      "loss": 0.8941,
+      "step": 2120
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.8140115141868591,
+      "learning_rate": 9.060164376566037e-06,
+      "loss": 0.8886,
+      "step": 2130
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.7906537652015686,
+      "learning_rate": 9.051609713658619e-06,
+      "loss": 0.8951,
+      "step": 2140
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.6860573291778564,
+      "learning_rate": 9.04302037065927e-06,
+      "loss": 0.8922,
+      "step": 2150
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.7657009363174438,
+      "learning_rate": 9.034396421089192e-06,
+      "loss": 0.9203,
+      "step": 2160
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.9350467920303345,
+      "learning_rate": 9.025737938765803e-06,
+      "loss": 0.9592,
+      "step": 2170
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.8163906931877136,
+      "learning_rate": 9.0170449978021e-06,
+      "loss": 0.9156,
+      "step": 2180
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.6962751150131226,
+      "learning_rate": 9.008317672606044e-06,
+      "loss": 0.8692,
+      "step": 2190
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.6299543976783752,
+      "learning_rate": 8.9995560378799e-06,
+      "loss": 0.9052,
+      "step": 2200
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.6928566694259644,
+      "learning_rate": 8.990760168619616e-06,
+      "loss": 0.9231,
+      "step": 2210
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.8014935851097107,
+      "learning_rate": 8.981930140114167e-06,
+      "loss": 0.9165,
+      "step": 2220
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.7766875624656677,
+      "learning_rate": 8.97306602794492e-06,
+      "loss": 0.9003,
+      "step": 2230
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.7637115120887756,
+      "learning_rate": 8.964167907984989e-06,
+      "loss": 0.8775,
+      "step": 2240
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.8758956789970398,
+      "learning_rate": 8.955235856398568e-06,
+      "loss": 0.8945,
+      "step": 2250
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.8250870704650879,
+      "learning_rate": 8.946269949640306e-06,
+      "loss": 0.8865,
+      "step": 2260
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.6754716634750366,
+      "learning_rate": 8.937270264454624e-06,
+      "loss": 0.8994,
+      "step": 2270
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.6988954544067383,
+      "learning_rate": 8.928236877875084e-06,
+      "loss": 0.9221,
+      "step": 2280
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.7391093373298645,
+      "learning_rate": 8.919169867223713e-06,
+      "loss": 0.9124,
+      "step": 2290
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.7323642373085022,
+      "learning_rate": 8.910069310110346e-06,
+      "loss": 0.887,
+      "step": 2300
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.9379021525382996,
+      "learning_rate": 8.900935284431962e-06,
+      "loss": 0.8925,
+      "step": 2310
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.7654469609260559,
+      "learning_rate": 8.891767868372016e-06,
+      "loss": 0.9569,
+      "step": 2320
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.6094922423362732,
+      "learning_rate": 8.882567140399775e-06,
+      "loss": 0.878,
+      "step": 2330
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.7786396741867065,
+      "learning_rate": 8.873333179269635e-06,
+      "loss": 0.8933,
+      "step": 2340
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.6561703681945801,
+      "learning_rate": 8.864066064020462e-06,
+      "loss": 0.8821,
+      "step": 2350
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.6152118444442749,
+      "learning_rate": 8.854765873974898e-06,
+      "loss": 0.8905,
+      "step": 2360
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.8001115322113037,
+      "learning_rate": 8.845432688738703e-06,
+      "loss": 0.9002,
+      "step": 2370
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.709892988204956,
+      "learning_rate": 8.836066588200052e-06,
+      "loss": 0.887,
+      "step": 2380
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.6732577085494995,
+      "learning_rate": 8.826667652528866e-06,
+      "loss": 0.8866,
+      "step": 2390
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.9450706839561462,
+      "learning_rate": 8.817235962176121e-06,
+      "loss": 0.8808,
+      "step": 2400
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.9471026659011841,
+      "learning_rate": 8.807771597873159e-06,
+      "loss": 0.8943,
+      "step": 2410
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.7772449851036072,
+      "learning_rate": 8.798274640630997e-06,
+      "loss": 0.8859,
+      "step": 2420
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.8146265745162964,
+      "learning_rate": 8.788745171739632e-06,
+      "loss": 0.8869,
+      "step": 2430
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.7327932715415955,
+      "learning_rate": 8.779183272767353e-06,
+      "loss": 0.874,
+      "step": 2440
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.6848121881484985,
+      "learning_rate": 8.76958902556003e-06,
+      "loss": 0.8677,
+      "step": 2450
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.679011881351471,
+      "learning_rate": 8.759962512240426e-06,
+      "loss": 0.9292,
+      "step": 2460
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.8341182470321655,
+      "learning_rate": 8.750303815207487e-06,
+      "loss": 0.9098,
+      "step": 2470
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.874434769153595,
+      "learning_rate": 8.740613017135634e-06,
+      "loss": 0.9081,
+      "step": 2480
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.7139870524406433,
+      "learning_rate": 8.730890200974065e-06,
+      "loss": 0.896,
+      "step": 2490
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.7836432456970215,
+      "learning_rate": 8.721135449946037e-06,
+      "loss": 0.8939,
+      "step": 2500
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.6674822568893433,
+      "learning_rate": 8.711348847548157e-06,
+      "loss": 0.9039,
+      "step": 2510
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.6287670135498047,
+      "learning_rate": 8.701530477549666e-06,
+      "loss": 0.8786,
+      "step": 2520
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.7027100324630737,
+      "learning_rate": 8.691680423991722e-06,
+      "loss": 0.8887,
+      "step": 2530
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.713030993938446,
+      "learning_rate": 8.681798771186683e-06,
+      "loss": 0.9048,
+      "step": 2540
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.7505326867103577,
+      "learning_rate": 8.67188560371738e-06,
+      "loss": 0.868,
+      "step": 2550
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.6765055060386658,
+      "learning_rate": 8.661941006436401e-06,
+      "loss": 0.8686,
+      "step": 2560
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.7267132997512817,
+      "learning_rate": 8.651965064465355e-06,
+      "loss": 0.9243,
+      "step": 2570
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.6484735608100891,
+      "learning_rate": 8.641957863194153e-06,
+      "loss": 0.8846,
+      "step": 2580
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.6731644868850708,
+      "learning_rate": 8.631919488280267e-06,
+      "loss": 0.914,
+      "step": 2590
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.6963881850242615,
+      "learning_rate": 8.621850025648008e-06,
+      "loss": 0.8604,
+      "step": 2600
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.9081583619117737,
+      "learning_rate": 8.611749561487785e-06,
+      "loss": 0.9331,
+      "step": 2610
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.7570102214813232,
+      "learning_rate": 8.601618182255364e-06,
+      "loss": 0.8996,
+      "step": 2620
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.820863664150238,
+      "learning_rate": 8.59145597467113e-06,
+      "loss": 0.8707,
+      "step": 2630
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.8625741004943848,
+      "learning_rate": 8.581263025719352e-06,
+      "loss": 0.8806,
+      "step": 2640
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.1321467161178589,
+      "learning_rate": 8.571039422647423e-06,
+      "loss": 0.9222,
+      "step": 2650
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.7913120985031128,
+      "learning_rate": 8.560785252965131e-06,
+      "loss": 0.9212,
+      "step": 2660
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.7801851034164429,
+      "learning_rate": 8.5505006044439e-06,
+      "loss": 0.9001,
+      "step": 2670
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.689361035823822,
+      "learning_rate": 8.540185565116034e-06,
+      "loss": 0.9105,
+      "step": 2680
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.6706151366233826,
+      "learning_rate": 8.52984022327398e-06,
+      "loss": 0.8955,
+      "step": 2690
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.8290680050849915,
+      "learning_rate": 8.51946466746955e-06,
+      "loss": 0.9029,
+      "step": 2700
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.74539715051651,
+      "learning_rate": 8.509058986513185e-06,
+      "loss": 0.9072,
+      "step": 2710
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.6989156603813171,
+      "learning_rate": 8.498623269473178e-06,
+      "loss": 0.9111,
+      "step": 2720
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.8335017561912537,
+      "learning_rate": 8.488157605674924e-06,
+      "loss": 0.9043,
+      "step": 2730
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.8350189924240112,
+      "learning_rate": 8.477662084700145e-06,
+      "loss": 0.9496,
+      "step": 2740
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.7352144122123718,
+      "learning_rate": 8.46713679638613e-06,
+      "loss": 0.8805,
+      "step": 2750
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.8180497884750366,
+      "learning_rate": 8.456581830824966e-06,
+      "loss": 0.8715,
+      "step": 2760
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.8565731644630432,
+      "learning_rate": 8.445997278362759e-06,
+      "loss": 0.881,
+      "step": 2770
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.7524616122245789,
+      "learning_rate": 8.435383229598872e-06,
+      "loss": 0.873,
+      "step": 2780
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.8238971829414368,
+      "learning_rate": 8.424739775385145e-06,
+      "loss": 0.8766,
+      "step": 2790
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.7271744608879089,
+      "learning_rate": 8.414067006825108e-06,
+      "loss": 0.8659,
+      "step": 2800
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.6863881945610046,
+      "learning_rate": 8.403365015273223e-06,
+      "loss": 0.9122,
+      "step": 2810
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.7746860384941101,
+      "learning_rate": 8.392633892334074e-06,
+      "loss": 0.8661,
+      "step": 2820
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.7054237723350525,
+      "learning_rate": 8.381873729861609e-06,
+      "loss": 0.8884,
+      "step": 2830
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.6736589074134827,
+      "learning_rate": 8.371084619958337e-06,
+      "loss": 0.8776,
+      "step": 2840
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.671820878982544,
+      "learning_rate": 8.360266654974543e-06,
+      "loss": 0.866,
+      "step": 2850
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.8050073981285095,
+      "learning_rate": 8.349419927507505e-06,
+      "loss": 0.8884,
+      "step": 2860
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.8177492618560791,
+      "learning_rate": 8.338544530400693e-06,
+      "loss": 0.8678,
+      "step": 2870
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.8012502193450928,
+      "learning_rate": 8.327640556742978e-06,
+      "loss": 0.8792,
+      "step": 2880
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.6939083337783813,
+      "learning_rate": 8.316708099867834e-06,
+      "loss": 0.8928,
+      "step": 2890
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.7548695206642151,
+      "learning_rate": 8.305747253352534e-06,
+      "loss": 0.8564,
+      "step": 2900
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.7304048538208008,
+      "learning_rate": 8.294758111017367e-06,
+      "loss": 0.8836,
+      "step": 2910
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.8671132922172546,
+      "learning_rate": 8.28374076692481e-06,
+      "loss": 0.8864,
+      "step": 2920
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.7430797219276428,
+      "learning_rate": 8.272695315378744e-06,
+      "loss": 0.874,
+      "step": 2930
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.6934476494789124,
+      "learning_rate": 8.261621850923634e-06,
+      "loss": 0.8583,
+      "step": 2940
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.7404267191886902,
+      "learning_rate": 8.250520468343722e-06,
+      "loss": 0.8734,
+      "step": 2950
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.6465054154396057,
+      "learning_rate": 8.239391262662221e-06,
+      "loss": 0.8734,
+      "step": 2960
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.6825598478317261,
+      "learning_rate": 8.228234329140498e-06,
+      "loss": 0.8666,
+      "step": 2970
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.6966167688369751,
+      "learning_rate": 8.217049763277257e-06,
+      "loss": 0.8823,
+      "step": 2980
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.8441101908683777,
+      "learning_rate": 8.205837660807726e-06,
+      "loss": 0.8876,
+      "step": 2990
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.78465336561203,
+      "learning_rate": 8.194598117702828e-06,
+      "loss": 0.8991,
+      "step": 3000
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.7731233835220337,
+      "learning_rate": 8.183331230168376e-06,
+      "loss": 0.8763,
+      "step": 3010
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.7912877202033997,
+      "learning_rate": 8.172037094644236e-06,
+      "loss": 0.867,
+      "step": 3020
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.7123367190361023,
+      "learning_rate": 8.160715807803503e-06,
+      "loss": 0.9098,
+      "step": 3030
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.689221203327179,
+      "learning_rate": 8.149367466551678e-06,
+      "loss": 0.8859,
+      "step": 3040
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.7518553137779236,
+      "learning_rate": 8.137992168025835e-06,
+      "loss": 0.9198,
+      "step": 3050
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.7416417598724365,
+      "learning_rate": 8.126590009593794e-06,
+      "loss": 0.9011,
+      "step": 3060
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.6670682430267334,
+      "learning_rate": 8.115161088853283e-06,
+      "loss": 0.8884,
+      "step": 3070
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.7208391427993774,
+      "learning_rate": 8.103705503631103e-06,
+      "loss": 0.8903,
+      "step": 3080
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.635645866394043,
+      "learning_rate": 8.092223351982296e-06,
+      "loss": 0.8773,
+      "step": 3090
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.7762060761451721,
+      "learning_rate": 8.080714732189295e-06,
+      "loss": 0.9039,
+      "step": 3100
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.6757723689079285,
+      "learning_rate": 8.069179742761092e-06,
+      "loss": 0.8745,
+      "step": 3110
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.7416348457336426,
+      "learning_rate": 8.057618482432399e-06,
+      "loss": 0.9077,
+      "step": 3120
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.8212820887565613,
+      "learning_rate": 8.046031050162784e-06,
+      "loss": 0.8753,
+      "step": 3130
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.6839666962623596,
+      "learning_rate": 8.034417545135846e-06,
+      "loss": 0.8759,
+      "step": 3140
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.767983078956604,
+      "learning_rate": 8.022778066758348e-06,
+      "loss": 0.8778,
+      "step": 3150
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.6773496270179749,
+      "learning_rate": 8.01111271465938e-06,
+      "loss": 0.868,
+      "step": 3160
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.793113648891449,
+      "learning_rate": 7.999421588689498e-06,
+      "loss": 0.9105,
+      "step": 3170
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.7550264000892639,
+      "learning_rate": 7.987704788919872e-06,
+      "loss": 0.9105,
+      "step": 3180
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.7124478220939636,
+      "learning_rate": 7.975962415641428e-06,
+      "loss": 0.8728,
+      "step": 3190
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.805101215839386,
+      "learning_rate": 7.96419456936399e-06,
+      "loss": 0.9488,
+      "step": 3200
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.8259230852127075,
+      "learning_rate": 7.952401350815427e-06,
+      "loss": 0.8665,
+      "step": 3210
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.7200030088424683,
+      "learning_rate": 7.940582860940771e-06,
+      "loss": 0.9036,
+      "step": 3220
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.6903145909309387,
+      "learning_rate": 7.928739200901381e-06,
+      "loss": 0.8781,
+      "step": 3230
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.6554933786392212,
+      "learning_rate": 7.91687047207405e-06,
+      "loss": 0.8773,
+      "step": 3240
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.7486357688903809,
+      "learning_rate": 7.904976776050156e-06,
+      "loss": 0.8914,
+      "step": 3250
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.7948868870735168,
+      "learning_rate": 7.893058214634785e-06,
+      "loss": 0.8497,
+      "step": 3260
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.6839557886123657,
+      "learning_rate": 7.881114889845858e-06,
+      "loss": 0.9194,
+      "step": 3270
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.8078474998474121,
+      "learning_rate": 7.869146903913261e-06,
+      "loss": 0.8988,
+      "step": 3280
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.7275850176811218,
+      "learning_rate": 7.857154359277972e-06,
+      "loss": 0.8843,
+      "step": 3290
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.638208270072937,
+      "learning_rate": 7.845137358591176e-06,
+      "loss": 0.8942,
+      "step": 3300
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.7862017750740051,
+      "learning_rate": 7.833096004713398e-06,
+      "loss": 0.8712,
+      "step": 3310
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.7157966494560242,
+      "learning_rate": 7.82103040071361e-06,
+      "loss": 0.8902,
+      "step": 3320
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.7477262020111084,
+      "learning_rate": 7.80894064986836e-06,
+      "loss": 0.9196,
+      "step": 3330
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.9382122755050659,
+      "learning_rate": 7.796826855660876e-06,
+      "loss": 0.8779,
+      "step": 3340
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.6241675615310669,
+      "learning_rate": 7.7846891217802e-06,
+      "loss": 0.891,
+      "step": 3350
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.7218158841133118,
+      "learning_rate": 7.772527552120274e-06,
+      "loss": 0.8948,
+      "step": 3360
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.9833104610443115,
+      "learning_rate": 7.760342250779069e-06,
+      "loss": 0.8914,
+      "step": 3370
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.8101766109466553,
+      "learning_rate": 7.748133322057693e-06,
+      "loss": 0.8875,
+      "step": 3380
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.8606131672859192,
+      "learning_rate": 7.73590087045949e-06,
+      "loss": 0.9178,
+      "step": 3390
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.6935533881187439,
+      "learning_rate": 7.723645000689153e-06,
+      "loss": 0.9003,
+      "step": 3400
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.8851600289344788,
+      "learning_rate": 7.711365817651822e-06,
+      "loss": 0.8694,
+      "step": 3410
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.7385109663009644,
+      "learning_rate": 7.699063426452191e-06,
+      "loss": 0.8655,
+      "step": 3420
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.7656209468841553,
+      "learning_rate": 7.686737932393606e-06,
+      "loss": 0.8576,
+      "step": 3430
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.6485370397567749,
+      "learning_rate": 7.67438944097716e-06,
+      "loss": 0.8662,
+      "step": 3440
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.7411097884178162,
+      "learning_rate": 7.662018057900805e-06,
+      "loss": 0.8384,
+      "step": 3450
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.6974339485168457,
+      "learning_rate": 7.649623889058423e-06,
+      "loss": 0.8955,
+      "step": 3460
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.7264565825462341,
+      "learning_rate": 7.637207040538937e-06,
+      "loss": 0.8896,
+      "step": 3470
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.7551647424697876,
+      "learning_rate": 7.6247676186253995e-06,
+      "loss": 0.8735,
+      "step": 3480
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.8099594116210938,
+      "learning_rate": 7.612305729794085e-06,
+      "loss": 0.8714,
+      "step": 3490
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.8042902946472168,
+      "learning_rate": 7.599821480713571e-06,
+      "loss": 0.8959,
+      "step": 3500
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.7295987010002136,
+      "learning_rate": 7.5873149782438295e-06,
+      "loss": 0.8784,
+      "step": 3510
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.663959264755249,
+      "learning_rate": 7.574786329435317e-06,
+      "loss": 0.8603,
+      "step": 3520
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.7810430526733398,
+      "learning_rate": 7.5622356415280465e-06,
+      "loss": 0.8701,
+      "step": 3530
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.7424361109733582,
+      "learning_rate": 7.5496630219506805e-06,
+      "loss": 0.8502,
+      "step": 3540
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.8611142039299011,
+      "learning_rate": 7.537068578319609e-06,
+      "loss": 0.8859,
+      "step": 3550
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.643858015537262,
+      "learning_rate": 7.524452418438021e-06,
+      "loss": 0.9419,
+      "step": 3560
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.8085283637046814,
+      "learning_rate": 7.511814650294994e-06,
+      "loss": 0.8847,
+      "step": 3570
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.7591559886932373,
+      "learning_rate": 7.499155382064562e-06,
+      "loss": 0.8805,
+      "step": 3580
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.7097784876823425,
+      "learning_rate": 7.486474722104786e-06,
+      "loss": 0.8886,
+      "step": 3590
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.6670699119567871,
+      "learning_rate": 7.473772778956837e-06,
+      "loss": 0.9047,
+      "step": 3600
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.727094292640686,
+      "learning_rate": 7.461049661344058e-06,
+      "loss": 0.8449,
+      "step": 3610
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.7498268485069275,
+      "learning_rate": 7.448305478171036e-06,
+      "loss": 0.8638,
+      "step": 3620
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.8697443604469299,
+      "learning_rate": 7.435540338522673e-06,
+      "loss": 0.8804,
+      "step": 3630
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.8494949340820312,
+      "learning_rate": 7.422754351663252e-06,
+      "loss": 0.8498,
+      "step": 3640
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.6735732555389404,
+      "learning_rate": 7.40994762703549e-06,
+      "loss": 0.8724,
+      "step": 3650
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.7467610836029053,
+      "learning_rate": 7.397120274259622e-06,
+      "loss": 0.8946,
+      "step": 3660
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.8220069408416748,
+      "learning_rate": 7.384272403132442e-06,
+      "loss": 0.8735,
+      "step": 3670
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.7360656261444092,
+      "learning_rate": 7.371404123626379e-06,
+      "loss": 0.8829,
+      "step": 3680
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.8948329091072083,
+      "learning_rate": 7.358515545888545e-06,
+      "loss": 0.874,
+      "step": 3690
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.7232267260551453,
+      "learning_rate": 7.345606780239799e-06,
+      "loss": 0.9013,
+      "step": 3700
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.7717054486274719,
+      "learning_rate": 7.3326779371738e-06,
+      "loss": 0.8868,
+      "step": 3710
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.830051600933075,
+      "learning_rate": 7.319729127356061e-06,
+      "loss": 0.8763,
+      "step": 3720
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.7295044660568237,
+      "learning_rate": 7.306760461623e-06,
+      "loss": 0.8699,
+      "step": 3730
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.7495072484016418,
+      "learning_rate": 7.2937720509809975e-06,
+      "loss": 0.8866,
+      "step": 3740
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.7179144024848938,
+      "learning_rate": 7.280764006605438e-06,
+      "loss": 0.8871,
+      "step": 3750
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.7208660840988159,
+      "learning_rate": 7.267736439839763e-06,
+      "loss": 0.9242,
+      "step": 3760
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.7881584167480469,
+      "learning_rate": 7.254689462194522e-06,
+      "loss": 0.8718,
+      "step": 3770
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.6987658739089966,
+      "learning_rate": 7.241623185346409e-06,
+      "loss": 0.8837,
+      "step": 3780
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.6674745678901672,
+      "learning_rate": 7.228537721137313e-06,
+      "loss": 0.8701,
+      "step": 3790
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.6959546804428101,
+      "learning_rate": 7.215433181573358e-06,
+      "loss": 0.8853,
+      "step": 3800
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.7006931900978088,
+      "learning_rate": 7.2023096788239464e-06,
+      "loss": 0.9244,
+      "step": 3810
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.7189576029777527,
+      "learning_rate": 7.189167325220795e-06,
+      "loss": 0.8586,
+      "step": 3820
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.8226879239082336,
+      "learning_rate": 7.176006233256979e-06,
+      "loss": 0.8693,
+      "step": 3830
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.7610313892364502,
+      "learning_rate": 7.162826515585965e-06,
+      "loss": 0.8677,
+      "step": 3840
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.7577977776527405,
+      "learning_rate": 7.149628285020647e-06,
+      "loss": 0.8495,
+      "step": 3850
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.7412477135658264,
+      "learning_rate": 7.136411654532383e-06,
+      "loss": 0.8834,
+      "step": 3860
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.8123847246170044,
+      "learning_rate": 7.1231767372500285e-06,
+      "loss": 0.874,
+      "step": 3870
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.8736533522605896,
+      "learning_rate": 7.109923646458958e-06,
+      "loss": 0.8557,
+      "step": 3880
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.6327398419380188,
+      "learning_rate": 7.0966524956001155e-06,
+      "loss": 0.863,
+      "step": 3890
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.7053546905517578,
+      "learning_rate": 7.083363398269022e-06,
+      "loss": 0.8953,
+      "step": 3900
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.6791098117828369,
+      "learning_rate": 7.07005646821482e-06,
+      "loss": 0.8765,
+      "step": 3910
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.8618676066398621,
+      "learning_rate": 7.056731819339287e-06,
+      "loss": 0.9007,
+      "step": 3920
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.7667632102966309,
+      "learning_rate": 7.043389565695868e-06,
+      "loss": 0.8981,
+      "step": 3930
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.7833120822906494,
+      "learning_rate": 7.0300298214887e-06,
+      "loss": 0.8799,
+      "step": 3940
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.8504341244697571,
+      "learning_rate": 7.01665270107163e-06,
+      "loss": 0.8985,
+      "step": 3950
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.8204311728477478,
+      "learning_rate": 7.003258318947236e-06,
+      "loss": 0.8814,
+      "step": 3960
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.9083641171455383,
+      "learning_rate": 6.989846789765853e-06,
+      "loss": 0.8762,
+      "step": 3970
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.7540957927703857,
+      "learning_rate": 6.976418228324584e-06,
+      "loss": 0.8429,
+      "step": 3980
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.741821825504303,
+      "learning_rate": 6.9629727495663265e-06,
+      "loss": 0.8875,
+      "step": 3990
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.8164342641830444,
+      "learning_rate": 6.9495104685787755e-06,
+      "loss": 0.8647,
+      "step": 4000
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.6732923984527588,
+      "learning_rate": 6.936031500593453e-06,
+      "loss": 0.905,
+      "step": 4010
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.826184093952179,
+      "learning_rate": 6.922535960984709e-06,
+      "loss": 0.8723,
+      "step": 4020
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.8277553915977478,
+      "learning_rate": 6.909023965268746e-06,
+      "loss": 0.8756,
+      "step": 4030
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.9936055541038513,
+      "learning_rate": 6.895495629102617e-06,
+      "loss": 0.9035,
+      "step": 4040
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.7322288751602173,
+      "learning_rate": 6.881951068283248e-06,
+      "loss": 0.886,
+      "step": 4050
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.7935315370559692,
+      "learning_rate": 6.86839039874644e-06,
+      "loss": 0.9276,
+      "step": 4060
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.8838260173797607,
+      "learning_rate": 6.854813736565874e-06,
+      "loss": 0.888,
+      "step": 4070
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.781810462474823,
+      "learning_rate": 6.841221197952129e-06,
+      "loss": 0.8386,
+      "step": 4080
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.9165651202201843,
+      "learning_rate": 6.827612899251671e-06,
+      "loss": 0.8816,
+      "step": 4090
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.751937210559845,
+      "learning_rate": 6.813988956945872e-06,
+      "loss": 0.8935,
+      "step": 4100
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.8413047790527344,
+      "learning_rate": 6.800349487650004e-06,
+      "loss": 0.8916,
+      "step": 4110
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.7908408045768738,
+      "learning_rate": 6.786694608112247e-06,
+      "loss": 0.8579,
+      "step": 4120
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.7976765632629395,
+      "learning_rate": 6.773024435212678e-06,
+      "loss": 0.8694,
+      "step": 4130
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.7089449763298035,
+      "learning_rate": 6.75933908596229e-06,
+      "loss": 0.907,
+      "step": 4140
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.6339461207389832,
+      "learning_rate": 6.74563867750197e-06,
+      "loss": 0.8886,
+      "step": 4150
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.7009597420692444,
+      "learning_rate": 6.7319233271015104e-06,
+      "loss": 0.8765,
+      "step": 4160
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.6806443929672241,
+      "learning_rate": 6.718193152158598e-06,
+      "loss": 0.8662,
+      "step": 4170
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.8072966933250427,
+      "learning_rate": 6.704448270197814e-06,
+      "loss": 0.8329,
+      "step": 4180
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.9389317631721497,
+      "learning_rate": 6.690688798869624e-06,
+      "loss": 0.8661,
+      "step": 4190
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.6365747451782227,
+      "learning_rate": 6.676914855949372e-06,
+      "loss": 0.864,
+      "step": 4200
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.6249781847000122,
+      "learning_rate": 6.663126559336274e-06,
+      "loss": 0.8993,
+      "step": 4210
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.823241114616394,
+      "learning_rate": 6.6493240270524085e-06,
+      "loss": 0.8868,
+      "step": 4220
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.6719274520874023,
+      "learning_rate": 6.6355073772417e-06,
+      "loss": 0.9074,
+      "step": 4230
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.6730551719665527,
+      "learning_rate": 6.6216767281689195e-06,
+      "loss": 0.917,
+      "step": 4240
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.6049341559410095,
+      "learning_rate": 6.6078321982186656e-06,
+      "loss": 0.8892,
+      "step": 4250
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.7270164489746094,
+      "learning_rate": 6.593973905894348e-06,
+      "loss": 0.8542,
+      "step": 4260
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.7423452734947205,
+      "learning_rate": 6.580101969817176e-06,
+      "loss": 0.865,
+      "step": 4270
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.7351382970809937,
+      "learning_rate": 6.5662165087251475e-06,
+      "loss": 0.896,
+      "step": 4280
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.7507643103599548,
+      "learning_rate": 6.552317641472027e-06,
+      "loss": 0.8574,
+      "step": 4290
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.7578912377357483,
+      "learning_rate": 6.538405487026329e-06,
+      "loss": 0.8816,
+      "step": 4300
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.742656409740448,
+      "learning_rate": 6.524480164470304e-06,
+      "loss": 0.8853,
+      "step": 4310
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.8041390776634216,
+      "learning_rate": 6.510541792998914e-06,
+      "loss": 0.8546,
+      "step": 4320
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.7750957012176514,
+      "learning_rate": 6.496590491918813e-06,
+      "loss": 0.9019,
+      "step": 4330
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.833149790763855,
+      "learning_rate": 6.48262638064733e-06,
+      "loss": 0.8584,
+      "step": 4340
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.7976315021514893,
+      "learning_rate": 6.46864957871144e-06,
+      "loss": 0.8547,
+      "step": 4350
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.9380526542663574,
+      "learning_rate": 6.454660205746749e-06,
+      "loss": 0.8739,
+      "step": 4360
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.7742275595664978,
+      "learning_rate": 6.440658381496464e-06,
+      "loss": 0.8877,
+      "step": 4370
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.977070689201355,
+      "learning_rate": 6.426644225810369e-06,
+      "loss": 0.8436,
+      "step": 4380
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.8549525737762451,
+      "learning_rate": 6.412617858643802e-06,
+      "loss": 0.8907,
+      "step": 4390
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.8932984471321106,
+      "learning_rate": 6.398579400056624e-06,
+      "loss": 0.8897,
+      "step": 4400
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.7153633236885071,
+      "learning_rate": 6.384528970212196e-06,
+      "loss": 0.8611,
+      "step": 4410
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.740108847618103,
+      "learning_rate": 6.370466689376343e-06,
+      "loss": 0.8528,
+      "step": 4420
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.8926063776016235,
+      "learning_rate": 6.356392677916336e-06,
+      "loss": 0.8284,
+      "step": 4430
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.6496686339378357,
+      "learning_rate": 6.342307056299849e-06,
+      "loss": 0.8808,
+      "step": 4440
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.6834713816642761,
+      "learning_rate": 6.328209945093942e-06,
+      "loss": 0.9055,
+      "step": 4450
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.8347498774528503,
+      "learning_rate": 6.314101464964012e-06,
+      "loss": 0.891,
+      "step": 4460
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 1.0568692684173584,
+      "learning_rate": 6.299981736672778e-06,
+      "loss": 0.882,
+      "step": 4470
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.6839845776557922,
+      "learning_rate": 6.285850881079229e-06,
+      "loss": 0.8823,
+      "step": 4480
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.6580236554145813,
+      "learning_rate": 6.271709019137611e-06,
+      "loss": 0.8858,
+      "step": 4490
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.8284133076667786,
+      "learning_rate": 6.257556271896365e-06,
+      "loss": 0.8659,
+      "step": 4500
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.655863881111145,
+      "learning_rate": 6.243392760497121e-06,
+      "loss": 0.8646,
+      "step": 4510
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.7806566953659058,
+      "learning_rate": 6.2292186061736345e-06,
+      "loss": 0.8559,
+      "step": 4520
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.7328823208808899,
+      "learning_rate": 6.215033930250766e-06,
+      "loss": 0.8822,
+      "step": 4530
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.7945101261138916,
+      "learning_rate": 6.200838854143432e-06,
+      "loss": 0.8774,
+      "step": 4540
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.6925386786460876,
+      "learning_rate": 6.186633499355576e-06,
+      "loss": 0.9191,
+      "step": 4550
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.729263186454773,
+      "learning_rate": 6.172417987479115e-06,
+      "loss": 0.8812,
+      "step": 4560
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.871244490146637,
+      "learning_rate": 6.158192440192915e-06,
+      "loss": 0.826,
+      "step": 4570
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.8375679850578308,
+      "learning_rate": 6.143956979261734e-06,
+      "loss": 0.8814,
+      "step": 4580
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.8106005191802979,
+      "learning_rate": 6.129711726535192e-06,
+      "loss": 0.8983,
+      "step": 4590
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.8209431171417236,
+      "learning_rate": 6.115456803946719e-06,
+      "loss": 0.8877,
+      "step": 4600
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.6159712076187134,
+      "learning_rate": 6.101192333512515e-06,
+      "loss": 0.8786,
+      "step": 4610
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.7372955083847046,
+      "learning_rate": 6.086918437330508e-06,
+      "loss": 0.8746,
+      "step": 4620
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.7908628582954407,
+      "learning_rate": 6.072635237579306e-06,
+      "loss": 0.8709,
+      "step": 4630
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.7661042809486389,
+      "learning_rate": 6.058342856517147e-06,
+      "loss": 0.8793,
+      "step": 4640
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.7974932789802551,
+      "learning_rate": 6.044041416480866e-06,
+      "loss": 0.8709,
+      "step": 4650
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.7971488237380981,
+      "learning_rate": 6.0297310398848315e-06,
+      "loss": 0.8446,
+      "step": 4660
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.8158445358276367,
+      "learning_rate": 6.0154118492199085e-06,
+      "loss": 0.8463,
+      "step": 4670
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.6840909123420715,
+      "learning_rate": 6.001083967052408e-06,
+      "loss": 0.8843,
+      "step": 4680
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.8997191786766052,
+      "learning_rate": 5.986747516023031e-06,
+      "loss": 0.888,
+      "step": 4690
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.7533353567123413,
+      "learning_rate": 5.97240261884583e-06,
+      "loss": 0.8888,
+      "step": 4700
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.7168859839439392,
+      "learning_rate": 5.9580493983071505e-06,
+      "loss": 0.8388,
+      "step": 4710
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.817183792591095,
+      "learning_rate": 5.943687977264584e-06,
+      "loss": 0.8496,
+      "step": 4720
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.8092912435531616,
+      "learning_rate": 5.929318478645913e-06,
+      "loss": 0.8696,
+      "step": 4730
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.9155735969543457,
+      "learning_rate": 5.914941025448065e-06,
+      "loss": 0.8756,
+      "step": 4740
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.6442350745201111,
+      "learning_rate": 5.9005557407360495e-06,
+      "loss": 0.8459,
+      "step": 4750
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.6828013062477112,
+      "learning_rate": 5.886162747641912e-06,
+      "loss": 0.8731,
+      "step": 4760
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.8492785692214966,
+      "learning_rate": 5.871762169363682e-06,
+      "loss": 0.8899,
+      "step": 4770
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.6571186780929565,
+      "learning_rate": 5.857354129164311e-06,
+      "loss": 0.8822,
+      "step": 4780
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.7441083192825317,
+      "learning_rate": 5.842938750370623e-06,
+      "loss": 0.871,
+      "step": 4790
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.6670249104499817,
+      "learning_rate": 5.828516156372262e-06,
+      "loss": 0.8633,
+      "step": 4800
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.7199469208717346,
+      "learning_rate": 5.81408647062062e-06,
+      "loss": 0.8746,
+      "step": 4810
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.6860167384147644,
+      "learning_rate": 5.799649816627803e-06,
+      "loss": 0.8937,
+      "step": 4820
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.707673192024231,
+      "learning_rate": 5.785206317965554e-06,
+      "loss": 0.8898,
+      "step": 4830
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.6907834410667419,
+      "learning_rate": 5.770756098264208e-06,
+      "loss": 0.8704,
+      "step": 4840
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.7484092712402344,
+      "learning_rate": 5.756299281211628e-06,
+      "loss": 0.8732,
+      "step": 4850
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.7744404673576355,
+      "learning_rate": 5.7418359905521496e-06,
+      "loss": 0.8943,
+      "step": 4860
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.7255435585975647,
+      "learning_rate": 5.727366350085514e-06,
+      "loss": 0.8783,
+      "step": 4870
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.8677462339401245,
+      "learning_rate": 5.712890483665822e-06,
+      "loss": 0.9054,
+      "step": 4880
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 1.0392132997512817,
+      "learning_rate": 5.69840851520046e-06,
+      "loss": 0.8524,
+      "step": 4890
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.7953920960426331,
+      "learning_rate": 5.6839205686490474e-06,
+      "loss": 0.8826,
+      "step": 4900
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.770919144153595,
+      "learning_rate": 5.669426768022371e-06,
+      "loss": 0.8565,
+      "step": 4910
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.735884428024292,
+      "learning_rate": 5.654927237381334e-06,
+      "loss": 0.8882,
+      "step": 4920
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.8747037053108215,
+      "learning_rate": 5.640422100835876e-06,
+      "loss": 0.9044,
+      "step": 4930
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.9188621640205383,
+      "learning_rate": 5.625911482543928e-06,
+      "loss": 0.8937,
+      "step": 4940
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.8039751052856445,
+      "learning_rate": 5.6113955067103385e-06,
+      "loss": 0.8714,
+      "step": 4950
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.9061981439590454,
+      "learning_rate": 5.596874297585816e-06,
+      "loss": 0.8798,
+      "step": 4960
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.9334414005279541,
+      "learning_rate": 5.582347979465864e-06,
+      "loss": 0.9003,
+      "step": 4970
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 1.0764784812927246,
+      "learning_rate": 5.5678166766897154e-06,
+      "loss": 0.8611,
+      "step": 4980
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.8570699691772461,
+      "learning_rate": 5.553280513639272e-06,
+      "loss": 0.8703,
+      "step": 4990
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.7193588614463806,
+      "learning_rate": 5.538739614738035e-06,
+      "loss": 0.8393,
+      "step": 5000
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.8715471029281616,
+      "learning_rate": 5.524194104450045e-06,
+      "loss": 0.8685,
+      "step": 5010
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.7570303082466125,
+      "learning_rate": 5.509644107278811e-06,
+      "loss": 0.8661,
+      "step": 5020
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.6701875925064087,
+      "learning_rate": 5.4950897477662526e-06,
+      "loss": 0.9007,
+      "step": 5030
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.761762797832489,
+      "learning_rate": 5.480531150491622e-06,
+      "loss": 0.8702,
+      "step": 5040
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.7878777384757996,
+      "learning_rate": 5.4659684400704525e-06,
+      "loss": 0.8881,
+      "step": 5050
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.7229417562484741,
+      "learning_rate": 5.45140174115348e-06,
+      "loss": 0.8684,
+      "step": 5060
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.6083467602729797,
+      "learning_rate": 5.436831178425582e-06,
+      "loss": 0.887,
+      "step": 5070
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.7734194993972778,
+      "learning_rate": 5.4222568766047066e-06,
+      "loss": 0.8589,
+      "step": 5080
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.8098832368850708,
+      "learning_rate": 5.40767896044081e-06,
+      "loss": 0.8861,
+      "step": 5090
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.7349828481674194,
+      "learning_rate": 5.393097554714781e-06,
+      "loss": 0.8761,
+      "step": 5100
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.8485532402992249,
+      "learning_rate": 5.378512784237382e-06,
+      "loss": 0.8867,
+      "step": 5110
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.7467071413993835,
+      "learning_rate": 5.363924773848175e-06,
+      "loss": 0.8697,
+      "step": 5120
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.9522390961647034,
+      "learning_rate": 5.349333648414454e-06,
+      "loss": 0.8573,
+      "step": 5130
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.8301889896392822,
+      "learning_rate": 5.334739532830177e-06,
+      "loss": 0.8714,
+      "step": 5140
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.7364034652709961,
+      "learning_rate": 5.3201425520148945e-06,
+      "loss": 0.8278,
+      "step": 5150
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.7919098734855652,
+      "learning_rate": 5.305542830912683e-06,
+      "loss": 0.8582,
+      "step": 5160
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.7982082366943359,
+      "learning_rate": 5.290940494491079e-06,
+      "loss": 0.8466,
+      "step": 5170
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.8373641967773438,
+      "learning_rate": 5.276335667739998e-06,
+      "loss": 0.8487,
+      "step": 5180
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.8268640637397766,
+      "learning_rate": 5.261728475670676e-06,
+      "loss": 0.8431,
+      "step": 5190
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.7479544878005981,
+      "learning_rate": 5.247119043314592e-06,
+      "loss": 0.8763,
+      "step": 5200
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.7835308909416199,
+      "learning_rate": 5.23250749572241e-06,
+      "loss": 0.9,
+      "step": 5210
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.6818840503692627,
+      "learning_rate": 5.2178939579628855e-06,
+      "loss": 0.8657,
+      "step": 5220
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.8019084334373474,
+      "learning_rate": 5.203278555121818e-06,
+      "loss": 0.8499,
+      "step": 5230
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.8876912593841553,
+      "learning_rate": 5.188661412300973e-06,
+      "loss": 0.8507,
+      "step": 5240
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.7216961979866028,
+      "learning_rate": 5.174042654617001e-06,
+      "loss": 0.8796,
+      "step": 5250
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.778347909450531,
+      "learning_rate": 5.159422407200383e-06,
+      "loss": 0.8912,
+      "step": 5260
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.8412684798240662,
+      "learning_rate": 5.144800795194348e-06,
+      "loss": 0.8335,
+      "step": 5270
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.7514455914497375,
+      "learning_rate": 5.1301779437538065e-06,
+      "loss": 0.9007,
+      "step": 5280
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.7360111474990845,
+      "learning_rate": 5.115553978044276e-06,
+      "loss": 0.8789,
+      "step": 5290
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.849012017250061,
+      "learning_rate": 5.1009290232408145e-06,
+      "loss": 0.8932,
+      "step": 5300
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.9734262228012085,
+      "learning_rate": 5.0863032045269435e-06,
+      "loss": 0.8918,
+      "step": 5310
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.8109739422798157,
+      "learning_rate": 5.071676647093581e-06,
+      "loss": 0.9057,
+      "step": 5320
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.9420107007026672,
+      "learning_rate": 5.057049476137968e-06,
+      "loss": 0.8732,
+      "step": 5330
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.8012482523918152,
+      "learning_rate": 5.042421816862597e-06,
+      "loss": 0.8609,
+      "step": 5340
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.6558821797370911,
+      "learning_rate": 5.027793794474138e-06,
+      "loss": 0.8914,
+      "step": 5350
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.7077735066413879,
+      "learning_rate": 5.013165534182375e-06,
+      "loss": 0.8635,
+      "step": 5360
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.9742336869239807,
+      "learning_rate": 4.998537161199121e-06,
+      "loss": 0.8852,
+      "step": 5370
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.7346500754356384,
+      "learning_rate": 4.983908800737156e-06,
+      "loss": 0.8413,
+      "step": 5380
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.7610775828361511,
+      "learning_rate": 4.969280578009157e-06,
+      "loss": 0.8966,
+      "step": 5390
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.7269995212554932,
+      "learning_rate": 4.954652618226618e-06,
+      "loss": 0.9104,
+      "step": 5400
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.8161409497261047,
+      "learning_rate": 4.940025046598784e-06,
+      "loss": 0.8841,
+      "step": 5410
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.7914517521858215,
+      "learning_rate": 4.925397988331575e-06,
+      "loss": 0.8776,
+      "step": 5420
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.8652795553207397,
+      "learning_rate": 4.910771568626518e-06,
+      "loss": 0.8902,
+      "step": 5430
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.7037116289138794,
+      "learning_rate": 4.896145912679676e-06,
+      "loss": 0.882,
+      "step": 5440
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 1.025122880935669,
+      "learning_rate": 4.881521145680575e-06,
+      "loss": 0.8067,
+      "step": 5450
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 0.7079437971115112,
+      "learning_rate": 4.866897392811127e-06,
+      "loss": 0.8758,
+      "step": 5460
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 0.7855519652366638,
+      "learning_rate": 4.852274779244568e-06,
+      "loss": 0.8574,
+      "step": 5470
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 0.7936415672302246,
+      "learning_rate": 4.837653430144382e-06,
+      "loss": 0.8464,
+      "step": 5480
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 0.6861563920974731,
+      "learning_rate": 4.82303347066323e-06,
+      "loss": 0.8861,
+      "step": 5490
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 0.8690369129180908,
+      "learning_rate": 4.808415025941872e-06,
+      "loss": 0.8992,
+      "step": 5500
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.8693572878837585,
+      "learning_rate": 4.79379822110811e-06,
+      "loss": 0.8632,
+      "step": 5510
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.7840404510498047,
+      "learning_rate": 4.7791831812757065e-06,
+      "loss": 0.8487,
+      "step": 5520
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.763825535774231,
+      "learning_rate": 4.764570031543316e-06,
+      "loss": 0.8355,
+      "step": 5530
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.5975529551506042,
+      "learning_rate": 4.749958896993416e-06,
+      "loss": 0.8837,
+      "step": 5540
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.7438409328460693,
+      "learning_rate": 4.735349902691231e-06,
+      "loss": 0.8575,
+      "step": 5550
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 0.9720832705497742,
+      "learning_rate": 4.720743173683671e-06,
+      "loss": 0.8701,
+      "step": 5560
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 0.6886677145957947,
+      "learning_rate": 4.706138834998253e-06,
+      "loss": 0.886,
+      "step": 5570
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 0.725154459476471,
+      "learning_rate": 4.6915370116420345e-06,
+      "loss": 0.8411,
+      "step": 5580
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 0.7134535908699036,
+      "learning_rate": 4.676937828600542e-06,
+      "loss": 0.8656,
+      "step": 5590
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 0.7429702877998352,
+      "learning_rate": 4.662341410836703e-06,
+      "loss": 0.875,
+      "step": 5600
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 0.690294623374939,
+      "learning_rate": 4.647747883289775e-06,
+      "loss": 0.8728,
+      "step": 5610
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.821758508682251,
+      "learning_rate": 4.6331573708742796e-06,
+      "loss": 0.8541,
+      "step": 5620
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.8202694654464722,
+      "learning_rate": 4.6185699984789225e-06,
+      "loss": 0.878,
+      "step": 5630
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.9476158618927002,
+      "learning_rate": 4.603985890965539e-06,
+      "loss": 0.832,
+      "step": 5640
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.8102102875709534,
+      "learning_rate": 4.589405173168016e-06,
+      "loss": 0.8946,
+      "step": 5650
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.7010157108306885,
+      "learning_rate": 4.574827969891227e-06,
+      "loss": 0.8449,
+      "step": 5660
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 0.7295305728912354,
+      "learning_rate": 4.560254405909959e-06,
+      "loss": 0.8596,
+      "step": 5670
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 0.7125367522239685,
+      "learning_rate": 4.545684605967852e-06,
+      "loss": 0.8996,
+      "step": 5680
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 0.9055048823356628,
+      "learning_rate": 4.531118694776328e-06,
+      "loss": 0.8962,
+      "step": 5690
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 0.8343817591667175,
+      "learning_rate": 4.51655679701352e-06,
+      "loss": 0.8512,
+      "step": 5700
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 0.714158296585083,
+      "learning_rate": 4.5019990373232075e-06,
+      "loss": 0.9047,
+      "step": 5710
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.6660721898078918,
+      "learning_rate": 4.487445540313752e-06,
+      "loss": 0.8745,
+      "step": 5720
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.7838519215583801,
+      "learning_rate": 4.472896430557027e-06,
+      "loss": 0.8468,
+      "step": 5730
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.8021497130393982,
+      "learning_rate": 4.458351832587354e-06,
+      "loss": 0.8909,
+      "step": 5740
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.7863754630088806,
+      "learning_rate": 4.443811870900435e-06,
+      "loss": 0.8544,
+      "step": 5750
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.7674603462219238,
+      "learning_rate": 4.429276669952285e-06,
+      "loss": 0.8792,
+      "step": 5760
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.9132154583930969,
+      "learning_rate": 4.414746354158175e-06,
+      "loss": 0.8715,
+      "step": 5770
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 0.8174626231193542,
+      "learning_rate": 4.400221047891554e-06,
+      "loss": 0.8167,
+      "step": 5780
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 0.8001808524131775,
+      "learning_rate": 4.385700875482996e-06,
+      "loss": 0.8609,
+      "step": 5790
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 0.7907263040542603,
+      "learning_rate": 4.37118596121913e-06,
+      "loss": 0.8525,
+      "step": 5800
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 0.7097474336624146,
+      "learning_rate": 4.356676429341577e-06,
+      "loss": 0.871,
+      "step": 5810
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 0.7247830629348755,
+      "learning_rate": 4.34217240404589e-06,
+      "loss": 0.8759,
+      "step": 5820
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.8027610778808594,
+      "learning_rate": 4.327674009480485e-06,
+      "loss": 0.8762,
+      "step": 5830
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.6493086218833923,
+      "learning_rate": 4.313181369745578e-06,
+      "loss": 0.8519,
+      "step": 5840
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 1.0388668775558472,
+      "learning_rate": 4.298694608892134e-06,
+      "loss": 0.8479,
+      "step": 5850
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.6935440301895142,
+      "learning_rate": 4.284213850920792e-06,
+      "loss": 0.8903,
+      "step": 5860
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.6928813457489014,
+      "learning_rate": 4.269739219780811e-06,
+      "loss": 0.8566,
+      "step": 5870
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.7724248170852661,
+      "learning_rate": 4.2552708393690035e-06,
+      "loss": 0.8658,
+      "step": 5880
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 0.8958523273468018,
+      "learning_rate": 4.240808833528682e-06,
+      "loss": 0.8542,
+      "step": 5890
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 0.678366482257843,
+      "learning_rate": 4.226353326048594e-06,
+      "loss": 0.8689,
+      "step": 5900
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 0.7881646156311035,
+      "learning_rate": 4.211904440661866e-06,
+      "loss": 0.8607,
+      "step": 5910
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 0.7571308016777039,
+      "learning_rate": 4.1974623010449355e-06,
+      "loss": 0.8456,
+      "step": 5920
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 0.7976707816123962,
+      "learning_rate": 4.183027030816505e-06,
+      "loss": 0.9029,
+      "step": 5930
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 0.909136176109314,
+      "learning_rate": 4.168598753536479e-06,
+      "loss": 0.8436,
+      "step": 5940
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 0.7112350463867188,
+      "learning_rate": 4.154177592704902e-06,
+      "loss": 0.8683,
+      "step": 5950
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 1.074589490890503,
+      "learning_rate": 4.139763671760901e-06,
+      "loss": 0.8731,
+      "step": 5960
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 0.9114029407501221,
+      "learning_rate": 4.125357114081642e-06,
+      "loss": 0.8755,
+      "step": 5970
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 0.7599102854728699,
+      "learning_rate": 4.110958042981256e-06,
+      "loss": 0.8908,
+      "step": 5980
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.8223927021026611,
+      "learning_rate": 4.096566581709798e-06,
+      "loss": 0.8497,
+      "step": 5990
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.6141775250434875,
+      "learning_rate": 4.082182853452181e-06,
+      "loss": 0.8372,
+      "step": 6000
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.827046275138855,
+      "learning_rate": 4.067806981327129e-06,
+      "loss": 0.8725,
+      "step": 6010
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.6700268983840942,
+      "learning_rate": 4.053439088386124e-06,
+      "loss": 0.9119,
+      "step": 6020
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.8427640199661255,
+      "learning_rate": 4.039079297612345e-06,
+      "loss": 0.8389,
+      "step": 6030
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.8254772424697876,
+      "learning_rate": 4.024727731919619e-06,
+      "loss": 0.8533,
+      "step": 6040
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 0.7223808169364929,
+      "learning_rate": 4.010384514151373e-06,
+      "loss": 0.8899,
+      "step": 6050
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 0.8630271553993225,
+      "learning_rate": 3.996049767079578e-06,
+      "loss": 0.8805,
+      "step": 6060
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 0.8118922114372253,
+      "learning_rate": 3.981723613403702e-06,
+      "loss": 0.8622,
+      "step": 6070
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 0.7789329290390015,
+      "learning_rate": 3.967406175749651e-06,
+      "loss": 0.8827,
+      "step": 6080
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 0.7576073408126831,
+      "learning_rate": 3.95309757666873e-06,
+      "loss": 0.8647,
+      "step": 6090
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.785173237323761,
+      "learning_rate": 3.93879793863659e-06,
+      "loss": 0.9079,
+      "step": 6100
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.7168451547622681,
+      "learning_rate": 3.924507384052177e-06,
+      "loss": 0.8408,
+      "step": 6110
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.808817446231842,
+      "learning_rate": 3.910226035236687e-06,
+      "loss": 0.879,
+      "step": 6120
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.8001656532287598,
+      "learning_rate": 3.8959540144325195e-06,
+      "loss": 0.8916,
+      "step": 6130
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.8590776920318604,
+      "learning_rate": 3.881691443802227e-06,
+      "loss": 0.8426,
+      "step": 6140
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 0.9599938988685608,
+      "learning_rate": 3.867438445427476e-06,
+      "loss": 0.8663,
+      "step": 6150
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 0.8696479201316833,
+      "learning_rate": 3.853195141308001e-06,
+      "loss": 0.8373,
+      "step": 6160
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 0.8265348076820374,
+      "learning_rate": 3.838961653360548e-06,
+      "loss": 0.8652,
+      "step": 6170
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 0.8333196640014648,
+      "learning_rate": 3.824738103417848e-06,
+      "loss": 0.8515,
+      "step": 6180
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 0.7337362766265869,
+      "learning_rate": 3.8105246132275664e-06,
+      "loss": 0.8449,
+      "step": 6190
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 0.9051868319511414,
+      "learning_rate": 3.7963213044512616e-06,
+      "loss": 0.8771,
+      "step": 6200
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.7232071161270142,
+      "learning_rate": 3.782128298663339e-06,
+      "loss": 0.8506,
+      "step": 6210
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.8423312306404114,
+      "learning_rate": 3.7679457173500195e-06,
+      "loss": 0.8638,
+      "step": 6220
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.8243368864059448,
+      "learning_rate": 3.7537736819082926e-06,
+      "loss": 0.8716,
+      "step": 6230
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.8379287123680115,
+      "learning_rate": 3.7396123136448824e-06,
+      "loss": 0.89,
+      "step": 6240
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.830217182636261,
+      "learning_rate": 3.7254617337751996e-06,
+      "loss": 0.8837,
+      "step": 6250
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 0.9521846771240234,
+      "learning_rate": 3.711322063422318e-06,
+      "loss": 0.8953,
+      "step": 6260
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 0.803010106086731,
+      "learning_rate": 3.6971934236159263e-06,
+      "loss": 0.8474,
+      "step": 6270
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 0.6727975606918335,
+      "learning_rate": 3.6830759352912993e-06,
+      "loss": 0.8854,
+      "step": 6280
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 0.8023139834403992,
+      "learning_rate": 3.6689697192882562e-06,
+      "loss": 0.8766,
+      "step": 6290
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 0.756571888923645,
+      "learning_rate": 3.6548748963501324e-06,
+      "loss": 0.8544,
+      "step": 6300
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.8170919418334961,
+      "learning_rate": 3.6407915871227427e-06,
+      "loss": 0.8712,
+      "step": 6310
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.9085463285446167,
+      "learning_rate": 3.6267199121533513e-06,
+      "loss": 0.8483,
+      "step": 6320
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.8478624820709229,
+      "learning_rate": 3.6126599918896327e-06,
+      "loss": 0.8642,
+      "step": 6330
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.8010087013244629,
+      "learning_rate": 3.5986119466786505e-06,
+      "loss": 0.8643,
+      "step": 6340
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.7449839115142822,
+      "learning_rate": 3.584575896765823e-06,
+      "loss": 0.8563,
+      "step": 6350
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.793870747089386,
+      "learning_rate": 3.5705519622938935e-06,
+      "loss": 0.8717,
+      "step": 6360
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 0.7289552092552185,
+      "learning_rate": 3.5565402633018963e-06,
+      "loss": 0.8677,
+      "step": 6370
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 0.7323423624038696,
+      "learning_rate": 3.5425409197241427e-06,
+      "loss": 0.8396,
+      "step": 6380
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 0.8906660079956055,
+      "learning_rate": 3.528554051389184e-06,
+      "loss": 0.8962,
+      "step": 6390
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 0.73725426197052,
+      "learning_rate": 3.5145797780187884e-06,
+      "loss": 0.8698,
+      "step": 6400
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 0.6689397692680359,
+      "learning_rate": 3.5006182192269177e-06,
+      "loss": 0.9035,
+      "step": 6410
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.9590892195701599,
+      "learning_rate": 3.486669494518699e-06,
+      "loss": 0.8808,
+      "step": 6420
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.8461219668388367,
+      "learning_rate": 3.4727337232894086e-06,
+      "loss": 0.8629,
+      "step": 6430
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.766094982624054,
+      "learning_rate": 3.458811024823444e-06,
+      "loss": 0.8759,
+      "step": 6440
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.8102118968963623,
+      "learning_rate": 3.444901518293309e-06,
+      "loss": 0.862,
+      "step": 6450
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.6815525889396667,
+      "learning_rate": 3.4310053227585827e-06,
+      "loss": 0.84,
+      "step": 6460
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.725439727306366,
+      "learning_rate": 3.4171225571649138e-06,
+      "loss": 0.8576,
+      "step": 6470
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.7938675880432129,
+      "learning_rate": 3.403253340342994e-06,
+      "loss": 0.8758,
+      "step": 6480
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.7491623759269714,
+      "learning_rate": 3.389397791007548e-06,
+      "loss": 0.9085,
+      "step": 6490
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.6669076681137085,
+      "learning_rate": 3.3755560277563028e-06,
+      "loss": 0.8659,
+      "step": 6500
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.6785613298416138,
+      "learning_rate": 3.3617281690689895e-06,
+      "loss": 0.8627,
+      "step": 6510
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.8891999125480652,
+      "learning_rate": 3.3479143333063214e-06,
+      "loss": 0.8481,
+      "step": 6520
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 0.8349424600601196,
+      "learning_rate": 3.334114638708979e-06,
+      "loss": 0.8342,
+      "step": 6530
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 0.6928399801254272,
+      "learning_rate": 3.3203292033966007e-06,
+      "loss": 0.8485,
+      "step": 6540
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 0.9773241877555847,
+      "learning_rate": 3.306558145366773e-06,
+      "loss": 0.8419,
+      "step": 6550
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 0.7637768983840942,
+      "learning_rate": 3.2928015824940172e-06,
+      "loss": 0.8542,
+      "step": 6560
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 0.7400500774383545,
+      "learning_rate": 3.2790596325287868e-06,
+      "loss": 0.8662,
+      "step": 6570
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.8206118941307068,
+      "learning_rate": 3.265332413096444e-06,
+      "loss": 0.8609,
+      "step": 6580
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 1.1378952264785767,
+      "learning_rate": 3.251620041696275e-06,
+      "loss": 0.8682,
+      "step": 6590
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.7858985066413879,
+      "learning_rate": 3.237922635700468e-06,
+      "loss": 0.8648,
+      "step": 6600
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.6994051933288574,
+      "learning_rate": 3.224240312353115e-06,
+      "loss": 0.8391,
+      "step": 6610
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.8151301145553589,
+      "learning_rate": 3.210573188769205e-06,
+      "loss": 0.8717,
+      "step": 6620
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.7834542393684387,
+      "learning_rate": 3.196921381933624e-06,
+      "loss": 0.8603,
+      "step": 6630
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 0.7387751340866089,
+      "learning_rate": 3.1832850087001544e-06,
+      "loss": 0.8434,
+      "step": 6640
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 0.7852590084075928,
+      "learning_rate": 3.1696641857904743e-06,
+      "loss": 0.8406,
+      "step": 6650
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 0.748166561126709,
+      "learning_rate": 3.1560590297931516e-06,
+      "loss": 0.8714,
+      "step": 6660
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 0.8609456419944763,
+      "learning_rate": 3.142469657162657e-06,
+      "loss": 0.8732,
+      "step": 6670
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 0.9150477051734924,
+      "learning_rate": 3.1288961842183633e-06,
+      "loss": 0.8677,
+      "step": 6680
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.7367973327636719,
+      "learning_rate": 3.115338727143546e-06,
+      "loss": 0.872,
+      "step": 6690
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.7203314304351807,
+      "learning_rate": 3.1017974019843928e-06,
+      "loss": 0.8553,
+      "step": 6700
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.7875660061836243,
+      "learning_rate": 3.0882723246490044e-06,
+      "loss": 0.8713,
+      "step": 6710
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.6855483055114746,
+      "learning_rate": 3.0747636109064126e-06,
+      "loss": 0.8795,
+      "step": 6720
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.7277706861495972,
+      "learning_rate": 3.0612713763855823e-06,
+      "loss": 0.8694,
+      "step": 6730
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 0.9142410159111023,
+      "learning_rate": 3.0477957365744227e-06,
+      "loss": 0.8812,
+      "step": 6740
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 0.7891936898231506,
+      "learning_rate": 3.034336806818796e-06,
+      "loss": 0.8685,
+      "step": 6750
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 0.7860237956047058,
+      "learning_rate": 3.020894702321539e-06,
+      "loss": 0.8615,
+      "step": 6760
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 0.8407242894172668,
+      "learning_rate": 3.0074695381414675e-06,
+      "loss": 0.8759,
+      "step": 6770
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 0.7908905744552612,
+      "learning_rate": 2.994061429192398e-06,
+      "loss": 0.8581,
+      "step": 6780
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 0.8444909453392029,
+      "learning_rate": 2.9806704902421557e-06,
+      "loss": 0.8608,
+      "step": 6790
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 0.7494134306907654,
+      "learning_rate": 2.9672968359116044e-06,
+      "loss": 0.8604,
+      "step": 6800
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 0.7267681956291199,
+      "learning_rate": 2.9539405806736555e-06,
+      "loss": 0.8537,
+      "step": 6810
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 0.8014705181121826,
+      "learning_rate": 2.9406018388522917e-06,
+      "loss": 0.8563,
+      "step": 6820
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 0.9584213495254517,
+      "learning_rate": 2.927280724621588e-06,
+      "loss": 0.8444,
+      "step": 6830
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 0.9992091059684753,
+      "learning_rate": 2.9139773520047353e-06,
+      "loss": 0.8893,
+      "step": 6840
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.9005075097084045,
+      "learning_rate": 2.900691834873062e-06,
+      "loss": 0.8971,
+      "step": 6850
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.912613034248352,
+      "learning_rate": 2.8874242869450655e-06,
+      "loss": 0.831,
+      "step": 6860
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.9488005638122559,
+      "learning_rate": 2.8741748217854282e-06,
+      "loss": 0.9023,
+      "step": 6870
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.8542575240135193,
+      "learning_rate": 2.860943552804053e-06,
+      "loss": 0.8598,
+      "step": 6880
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.8006412982940674,
+      "learning_rate": 2.847730593255097e-06,
+      "loss": 0.8447,
+      "step": 6890
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.9463589191436768,
+      "learning_rate": 2.834536056235989e-06,
+      "loss": 0.8537,
+      "step": 6900
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 0.5892205238342285,
+      "learning_rate": 2.8213600546864693e-06,
+      "loss": 0.8485,
+      "step": 6910
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 0.8081678152084351,
+      "learning_rate": 2.808202701387629e-06,
+      "loss": 0.8941,
+      "step": 6920
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 0.7223058938980103,
+      "learning_rate": 2.7950641089609275e-06,
+      "loss": 0.8667,
+      "step": 6930
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 0.8346781730651855,
+      "learning_rate": 2.781944389867253e-06,
+      "loss": 0.8538,
+      "step": 6940
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 0.968744695186615,
+      "learning_rate": 2.7688436564059274e-06,
+      "loss": 0.8429,
+      "step": 6950
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.774070143699646,
+      "learning_rate": 2.7557620207137805e-06,
+      "loss": 0.8753,
+      "step": 6960
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.7462638020515442,
+      "learning_rate": 2.7426995947641614e-06,
+      "loss": 0.8611,
+      "step": 6970
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.7983014583587646,
+      "learning_rate": 2.7296564903660023e-06,
+      "loss": 0.8616,
+      "step": 6980
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.776932418346405,
+      "learning_rate": 2.7166328191628418e-06,
+      "loss": 0.8848,
+      "step": 6990
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.7690081596374512,
+      "learning_rate": 2.70362869263188e-06,
+      "loss": 0.8459,
+      "step": 7000
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 0.8364967703819275,
+      "learning_rate": 2.69064422208303e-06,
+      "loss": 0.8406,
+      "step": 7010
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 1.0267677307128906,
+      "learning_rate": 2.6776795186579466e-06,
+      "loss": 0.8707,
+      "step": 7020
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 0.8048702478408813,
+      "learning_rate": 2.6647346933290994e-06,
+      "loss": 0.8784,
+      "step": 7030
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 0.7151132225990295,
+      "learning_rate": 2.651809856898795e-06,
+      "loss": 0.8508,
+      "step": 7040
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 0.7352768182754517,
+      "learning_rate": 2.638905119998254e-06,
+      "loss": 0.854,
+      "step": 7050
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 0.8499438166618347,
+      "learning_rate": 2.626020593086653e-06,
+      "loss": 0.8799,
+      "step": 7060
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.7417687177658081,
+      "learning_rate": 2.613156386450174e-06,
+      "loss": 0.8393,
+      "step": 7070
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.8973614573478699,
+      "learning_rate": 2.6003126102010696e-06,
+      "loss": 0.8689,
+      "step": 7080
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.784491777420044,
+      "learning_rate": 2.587489374276714e-06,
+      "loss": 0.8635,
+      "step": 7090
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.736359715461731,
+      "learning_rate": 2.574686788438669e-06,
+      "loss": 0.8618,
+      "step": 7100
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.854882001876831,
+      "learning_rate": 2.561904962271738e-06,
+      "loss": 0.8721,
+      "step": 7110
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 0.7691269516944885,
+      "learning_rate": 2.5491440051830275e-06,
+      "loss": 0.8443,
+      "step": 7120
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 0.6664737462997437,
+      "learning_rate": 2.5364040264010197e-06,
+      "loss": 0.8711,
+      "step": 7130
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 0.7332617044448853,
+      "learning_rate": 2.5236851349746242e-06,
+      "loss": 0.8666,
+      "step": 7140
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 0.9961148500442505,
+      "learning_rate": 2.510987439772261e-06,
+      "loss": 0.8338,
+      "step": 7150
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 0.8273669481277466,
+      "learning_rate": 2.4983110494809047e-06,
+      "loss": 0.8788,
+      "step": 7160
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.8583030700683594,
+      "learning_rate": 2.485656072605185e-06,
+      "loss": 0.8676,
+      "step": 7170
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.7933262586593628,
+      "learning_rate": 2.473022617466428e-06,
+      "loss": 0.8353,
+      "step": 7180
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.7235906720161438,
+      "learning_rate": 2.4604107922017535e-06,
+      "loss": 0.8155,
+      "step": 7190
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.7979265451431274,
+      "learning_rate": 2.447820704763132e-06,
+      "loss": 0.8608,
+      "step": 7200
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.6730958223342896,
+      "learning_rate": 2.435252462916467e-06,
+      "loss": 0.8744,
+      "step": 7210
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 1.0072416067123413,
+      "learning_rate": 2.4227061742406788e-06,
+      "loss": 0.879,
+      "step": 7220
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 0.8248384594917297,
+      "learning_rate": 2.4101819461267718e-06,
+      "loss": 0.8338,
+      "step": 7230
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 0.8090803027153015,
+      "learning_rate": 2.397679885776921e-06,
+      "loss": 0.8186,
+      "step": 7240
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 0.854955792427063,
+      "learning_rate": 2.3852001002035606e-06,
+      "loss": 0.8518,
+      "step": 7250
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 0.6975179314613342,
+      "learning_rate": 2.3727426962284554e-06,
+      "loss": 0.8472,
+      "step": 7260
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 0.7336934208869934,
+      "learning_rate": 2.360307780481798e-06,
+      "loss": 0.8534,
+      "step": 7270
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 0.7961238026618958,
+      "learning_rate": 2.3478954594012884e-06,
+      "loss": 0.8634,
+      "step": 7280
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 0.836816668510437,
+      "learning_rate": 2.3355058392312248e-06,
+      "loss": 0.8657,
+      "step": 7290
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 1.0481915473937988,
+      "learning_rate": 2.3231390260215953e-06,
+      "loss": 0.8551,
+      "step": 7300
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 0.8657959699630737,
+      "learning_rate": 2.310795125627175e-06,
+      "loss": 0.8998,
+      "step": 7310
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 0.745989978313446,
+      "learning_rate": 2.2984742437066084e-06,
+      "loss": 0.8527,
+      "step": 7320
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.8489542007446289,
+      "learning_rate": 2.2861764857215124e-06,
+      "loss": 0.8801,
+      "step": 7330
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.7227764129638672,
+      "learning_rate": 2.2739019569355782e-06,
+      "loss": 0.8396,
+      "step": 7340
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.9913476705551147,
+      "learning_rate": 2.2616507624136564e-06,
+      "loss": 0.8199,
+      "step": 7350
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.8966109156608582,
+      "learning_rate": 2.249423007020876e-06,
+      "loss": 0.8779,
+      "step": 7360
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.7755988240242004,
+      "learning_rate": 2.2372187954217238e-06,
+      "loss": 0.8842,
+      "step": 7370
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.790214478969574,
+      "learning_rate": 2.225038232079175e-06,
+      "loss": 0.8157,
+      "step": 7380
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 0.8576365113258362,
+      "learning_rate": 2.212881421253775e-06,
+      "loss": 0.8756,
+      "step": 7390
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 0.7593070268630981,
+      "learning_rate": 2.2007484670027657e-06,
+      "loss": 0.8895,
+      "step": 7400
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 1.0803626775741577,
+      "learning_rate": 2.1886394731791814e-06,
+      "loss": 0.8568,
+      "step": 7410
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 0.769171953201294,
+      "learning_rate": 2.176554543430965e-06,
+      "loss": 0.8388,
+      "step": 7420
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 0.8300871849060059,
+      "learning_rate": 2.164493781200086e-06,
+      "loss": 0.8678,
+      "step": 7430
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 0.8364355564117432,
+      "learning_rate": 2.1524572897216447e-06,
+      "loss": 0.8737,
+      "step": 7440
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 0.7527965307235718,
+      "learning_rate": 2.1404451720229934e-06,
+      "loss": 0.8597,
+      "step": 7450
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 0.7414934635162354,
+      "learning_rate": 2.1284575309228622e-06,
+      "loss": 0.8642,
+      "step": 7460
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 0.8682785630226135,
+      "learning_rate": 2.1164944690304624e-06,
+      "loss": 0.8859,
+      "step": 7470
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 0.8084505200386047,
+      "learning_rate": 2.104556088744628e-06,
+      "loss": 0.9021,
+      "step": 7480
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 0.7078831791877747,
+      "learning_rate": 2.092642492252915e-06,
+      "loss": 0.8803,
+      "step": 7490
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 0.7991881966590881,
+      "learning_rate": 2.080753781530754e-06,
+      "loss": 0.8772,
+      "step": 7500
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 0.7577025890350342,
+      "learning_rate": 2.0688900583405538e-06,
+      "loss": 0.8809,
+      "step": 7510
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 0.6723018884658813,
+      "learning_rate": 2.057051424230849e-06,
+      "loss": 0.864,
+      "step": 7520
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 0.8017199039459229,
+      "learning_rate": 2.0452379805354147e-06,
+      "loss": 0.8718,
+      "step": 7530
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 0.7686904072761536,
+      "learning_rate": 2.033449828372408e-06,
+      "loss": 0.8806,
+      "step": 7540
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 0.848595380783081,
+      "learning_rate": 2.0216870686435063e-06,
+      "loss": 0.8774,
+      "step": 7550
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 1.001030683517456,
+      "learning_rate": 2.0099498020330305e-06,
+      "loss": 0.8655,
+      "step": 7560
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 0.7667485475540161,
+      "learning_rate": 1.9982381290071014e-06,
+      "loss": 0.8583,
+      "step": 7570
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 0.6792320013046265,
+      "learning_rate": 1.9865521498127545e-06,
+      "loss": 0.8611,
+      "step": 7580
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 0.8859857320785522,
+      "learning_rate": 1.97489196447711e-06,
+      "loss": 0.8624,
+      "step": 7590
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 0.8147785663604736,
+      "learning_rate": 1.9632576728064996e-06,
+      "loss": 0.8337,
+      "step": 7600
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 0.8163339495658875,
+      "learning_rate": 1.9516493743856137e-06,
+      "loss": 0.8659,
+      "step": 7610
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 0.7094168663024902,
+      "learning_rate": 1.9400671685766526e-06,
+      "loss": 0.8401,
+      "step": 7620
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 0.7262287139892578,
+      "learning_rate": 1.928511154518473e-06,
+      "loss": 0.8682,
+      "step": 7630
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 1.0086379051208496,
+      "learning_rate": 1.9169814311257472e-06,
+      "loss": 0.8726,
+      "step": 7640
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 0.8723043203353882,
+      "learning_rate": 1.9054780970881038e-06,
+      "loss": 0.8675,
+      "step": 7650
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 0.8167620301246643,
+      "learning_rate": 1.8940012508692895e-06,
+      "loss": 0.8347,
+      "step": 7660
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 0.7233433723449707,
+      "learning_rate": 1.8825509907063328e-06,
+      "loss": 0.8652,
+      "step": 7670
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 0.7926831245422363,
+      "learning_rate": 1.8711274146086888e-06,
+      "loss": 0.8528,
+      "step": 7680
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 0.7207840085029602,
+      "learning_rate": 1.8597306203574173e-06,
+      "loss": 0.8954,
+      "step": 7690
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 0.8239450454711914,
+      "learning_rate": 1.8483607055043234e-06,
+      "loss": 0.8884,
+      "step": 7700
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 0.7212558388710022,
+      "learning_rate": 1.8370177673711492e-06,
+      "loss": 0.8557,
+      "step": 7710
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 0.8253349661827087,
+      "learning_rate": 1.8257019030487177e-06,
+      "loss": 0.8663,
+      "step": 7720
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 0.85089111328125,
+      "learning_rate": 1.8144132093961193e-06,
+      "loss": 0.8198,
+      "step": 7730
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 0.7277496457099915,
+      "learning_rate": 1.8031517830398682e-06,
+      "loss": 0.9007,
+      "step": 7740
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 0.8520780205726624,
+      "learning_rate": 1.7919177203730831e-06,
+      "loss": 0.8658,
+      "step": 7750
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 0.7603586316108704,
+      "learning_rate": 1.780711117554666e-06,
+      "loss": 0.8811,
+      "step": 7760
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 0.8584238886833191,
+      "learning_rate": 1.7695320705084678e-06,
+      "loss": 0.8759,
+      "step": 7770
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 0.705485999584198,
+      "learning_rate": 1.7583806749224762e-06,
+      "loss": 0.8794,
+      "step": 7780
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 0.6853485107421875,
+      "learning_rate": 1.7472570262479926e-06,
+      "loss": 0.8674,
+      "step": 7790
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 0.9203032851219177,
+      "learning_rate": 1.7361612196988177e-06,
+      "loss": 0.8801,
+      "step": 7800
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 0.8290138244628906,
+      "learning_rate": 1.7250933502504397e-06,
+      "loss": 0.8267,
+      "step": 7810
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 0.8048394322395325,
+      "learning_rate": 1.7140535126392045e-06,
+      "loss": 0.8836,
+      "step": 7820
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 0.7928733825683594,
+      "learning_rate": 1.7030418013615297e-06,
+      "loss": 0.88,
+      "step": 7830
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 0.8748371601104736,
+      "learning_rate": 1.6920583106730749e-06,
+      "loss": 0.8836,
+      "step": 7840
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 0.8439584970474243,
+      "learning_rate": 1.6811031345879504e-06,
+      "loss": 0.8687,
+      "step": 7850
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 0.7450461387634277,
+      "learning_rate": 1.6701763668778992e-06,
+      "loss": 0.8903,
+      "step": 7860
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 0.7663140892982483,
+      "learning_rate": 1.6592781010715015e-06,
+      "loss": 0.8537,
+      "step": 7870
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 1.0831756591796875,
+      "learning_rate": 1.6484084304533766e-06,
+      "loss": 0.8242,
+      "step": 7880
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 0.7070683836936951,
+      "learning_rate": 1.6375674480633764e-06,
+      "loss": 0.8665,
+      "step": 7890
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 0.7162163853645325,
+      "learning_rate": 1.6267552466957998e-06,
+      "loss": 0.8622,
+      "step": 7900
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 0.8870092630386353,
+      "learning_rate": 1.615971918898581e-06,
+      "loss": 0.8849,
+      "step": 7910
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 0.7890207171440125,
+      "learning_rate": 1.605217556972522e-06,
+      "loss": 0.8798,
+      "step": 7920
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 0.8524425029754639,
+      "learning_rate": 1.5944922529704776e-06,
+      "loss": 0.8672,
+      "step": 7930
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 0.7908310890197754,
+      "learning_rate": 1.5837960986965872e-06,
+      "loss": 0.8487,
+      "step": 7940
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 0.8728739619255066,
+      "learning_rate": 1.5731291857054759e-06,
+      "loss": 0.8506,
+      "step": 7950
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 1.0538735389709473,
+      "learning_rate": 1.562491605301475e-06,
+      "loss": 0.8517,
+      "step": 7960
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 0.8153250217437744,
+      "learning_rate": 1.5518834485378465e-06,
+      "loss": 0.8832,
+      "step": 7970
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 0.9526890516281128,
+      "learning_rate": 1.541304806215993e-06,
+      "loss": 0.8408,
+      "step": 7980
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 0.6733323931694031,
+      "learning_rate": 1.5307557688846853e-06,
+      "loss": 0.8594,
+      "step": 7990
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 0.7504012584686279,
+      "learning_rate": 1.5202364268392939e-06,
+      "loss": 0.9048,
+      "step": 8000
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 0.7233144640922546,
+      "learning_rate": 1.5097468701210028e-06,
+      "loss": 0.8857,
+      "step": 8010
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 0.9594971537590027,
+      "learning_rate": 1.4992871885160536e-06,
+      "loss": 0.8693,
+      "step": 8020
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 1.0271095037460327,
+      "learning_rate": 1.4888574715549597e-06,
+      "loss": 0.8573,
+      "step": 8030
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.7665346264839172,
+      "learning_rate": 1.478457808511759e-06,
+      "loss": 0.8463,
+      "step": 8040
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.8240995407104492,
+      "learning_rate": 1.4680882884032333e-06,
+      "loss": 0.8497,
+      "step": 8050
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.8238330483436584,
+      "learning_rate": 1.45774899998816e-06,
+      "loss": 0.8678,
+      "step": 8060
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.9149099588394165,
+      "learning_rate": 1.4474400317665404e-06,
+      "loss": 0.852,
+      "step": 8070
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.8823344111442566,
+      "learning_rate": 1.4371614719788485e-06,
+      "loss": 0.8596,
+      "step": 8080
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 0.8889569640159607,
+      "learning_rate": 1.4269134086052783e-06,
+      "loss": 0.8506,
+      "step": 8090
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 0.9761025309562683,
+      "learning_rate": 1.416695929364983e-06,
+      "loss": 0.849,
+      "step": 8100
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 0.8224256038665771,
+      "learning_rate": 1.406509121715331e-06,
+      "loss": 0.8686,
+      "step": 8110
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 0.8096652030944824,
+      "learning_rate": 1.396353072851151e-06,
+      "loss": 0.8359,
+      "step": 8120
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 0.8475440144538879,
+      "learning_rate": 1.3862278697039949e-06,
+      "loss": 0.8923,
+      "step": 8130
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 0.7281385660171509,
+      "learning_rate": 1.3761335989413837e-06,
+      "loss": 0.8356,
+      "step": 8140
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 0.8342397212982178,
+      "learning_rate": 1.366070346966073e-06,
+      "loss": 0.8817,
+      "step": 8150
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 0.8839266896247864,
+      "learning_rate": 1.3560381999153082e-06,
+      "loss": 0.8593,
+      "step": 8160
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 0.7736698389053345,
+      "learning_rate": 1.3460372436600888e-06,
+      "loss": 0.8717,
+      "step": 8170
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 0.8016160130500793,
+      "learning_rate": 1.336067563804439e-06,
+      "loss": 0.866,
+      "step": 8180
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 0.7394517064094543,
+      "learning_rate": 1.3261292456846648e-06,
+      "loss": 0.8525,
+      "step": 8190
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 0.7359172105789185,
+      "learning_rate": 1.3162223743686286e-06,
+      "loss": 0.8295,
+      "step": 8200
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 0.8422078490257263,
+      "learning_rate": 1.3063470346550271e-06,
+      "loss": 0.8652,
+      "step": 8210
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 0.8203555941581726,
+      "learning_rate": 1.2965033110726526e-06,
+      "loss": 0.8846,
+      "step": 8220
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 0.7276378870010376,
+      "learning_rate": 1.2866912878796845e-06,
+      "loss": 0.8604,
+      "step": 8230
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 0.7511932253837585,
+      "learning_rate": 1.276911049062949e-06,
+      "loss": 0.8645,
+      "step": 8240
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 0.7461341619491577,
+      "learning_rate": 1.267162678337222e-06,
+      "loss": 0.8491,
+      "step": 8250
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 0.8160814642906189,
+      "learning_rate": 1.257446259144494e-06,
+      "loss": 0.8928,
+      "step": 8260
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 0.8415591716766357,
+      "learning_rate": 1.2477618746532715e-06,
+      "loss": 0.8731,
+      "step": 8270
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 0.6584311723709106,
+      "learning_rate": 1.2381096077578498e-06,
+      "loss": 0.8553,
+      "step": 8280
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 0.8987860679626465,
+      "learning_rate": 1.2284895410776132e-06,
+      "loss": 0.8804,
+      "step": 8290
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 0.8643062710762024,
+      "learning_rate": 1.2189017569563293e-06,
+      "loss": 0.8668,
+      "step": 8300
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 0.820304811000824,
+      "learning_rate": 1.209346337461436e-06,
+      "loss": 0.83,
+      "step": 8310
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 1.3529876470565796,
+      "learning_rate": 1.1998233643833457e-06,
+      "loss": 0.8994,
+      "step": 8320
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 0.9008579850196838,
+      "learning_rate": 1.1903329192347397e-06,
+      "loss": 0.8418,
+      "step": 8330
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 0.6985065340995789,
+      "learning_rate": 1.1808750832498778e-06,
+      "loss": 0.8243,
+      "step": 8340
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 0.6970599889755249,
+      "learning_rate": 1.171449937383901e-06,
+      "loss": 0.8313,
+      "step": 8350
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 0.7619720697402954,
+      "learning_rate": 1.1620575623121277e-06,
+      "loss": 0.8422,
+      "step": 8360
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 0.7389846444129944,
+      "learning_rate": 1.1526980384293801e-06,
+      "loss": 0.8287,
+      "step": 8370
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 0.7180916666984558,
+      "learning_rate": 1.1433714458492833e-06,
+      "loss": 0.8821,
+      "step": 8380
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 0.7150819301605225,
+      "learning_rate": 1.1340778644035882e-06,
+      "loss": 0.871,
+      "step": 8390
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 0.87416011095047,
+      "learning_rate": 1.1248173736414807e-06,
+      "loss": 0.8713,
+      "step": 8400
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 0.7086650729179382,
+      "learning_rate": 1.1155900528289044e-06,
+      "loss": 0.8757,
+      "step": 8410
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 0.8164438605308533,
+      "learning_rate": 1.1063959809478846e-06,
+      "loss": 0.8674,
+      "step": 8420
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 0.7536283731460571,
+      "learning_rate": 1.0972352366958456e-06,
+      "loss": 0.8469,
+      "step": 8430
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 0.6917878985404968,
+      "learning_rate": 1.088107898484948e-06,
+      "loss": 0.861,
+      "step": 8440
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 0.9259957671165466,
+      "learning_rate": 1.0790140444414e-06,
+      "loss": 0.8364,
+      "step": 8450
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 0.6649090647697449,
+      "learning_rate": 1.0699537524048092e-06,
+      "loss": 0.8791,
+      "step": 8460
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 0.9272850751876831,
+      "learning_rate": 1.0609270999275e-06,
+      "loss": 0.8526,
+      "step": 8470
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 0.6726030111312866,
+      "learning_rate": 1.051934164273861e-06,
+      "loss": 0.8206,
+      "step": 8480
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 0.7841635942459106,
+      "learning_rate": 1.0429750224196754e-06,
+      "loss": 0.8756,
+      "step": 8490
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 0.7904772758483887,
+      "learning_rate": 1.0340497510514658e-06,
+      "loss": 0.8637,
+      "step": 8500
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 0.8979971408843994,
+      "learning_rate": 1.0251584265658404e-06,
+      "loss": 0.8616,
+      "step": 8510
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 0.7169375419616699,
+      "learning_rate": 1.016301125068832e-06,
+      "loss": 0.8538,
+      "step": 8520
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 0.8624107241630554,
+      "learning_rate": 1.0074779223752534e-06,
+      "loss": 0.8997,
+      "step": 8530
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 0.766612708568573,
+      "learning_rate": 9.986888940080468e-07,
+      "loss": 0.8202,
+      "step": 8540
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 0.7679712772369385,
+      "learning_rate": 9.899341151976327e-07,
+      "loss": 0.8521,
+      "step": 8550
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 0.7928770780563354,
+      "learning_rate": 9.812136608812762e-07,
+      "loss": 0.8616,
+      "step": 8560
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 0.9012945890426636,
+      "learning_rate": 9.725276057024286e-07,
+      "loss": 0.8502,
+      "step": 8570
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 0.742287814617157,
+      "learning_rate": 9.638760240101102e-07,
+      "loss": 0.8504,
+      "step": 8580
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 0.8516942858695984,
+      "learning_rate": 9.552589898582538e-07,
+      "loss": 0.8798,
+      "step": 8590
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 0.7784149050712585,
+      "learning_rate": 9.466765770050868e-07,
+      "loss": 0.8645,
+      "step": 8600
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 0.770591139793396,
+      "learning_rate": 9.381288589124877e-07,
+      "loss": 0.8388,
+      "step": 8610
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 0.7950923442840576,
+      "learning_rate": 9.296159087453615e-07,
+      "loss": 0.8499,
+      "step": 8620
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 0.8978777527809143,
+      "learning_rate": 9.211377993710207e-07,
+      "loss": 0.8402,
+      "step": 8630
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 0.7035576105117798,
+      "learning_rate": 9.126946033585493e-07,
+      "loss": 0.8526,
+      "step": 8640
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 1.0000416040420532,
+      "learning_rate": 9.042863929781886e-07,
+      "loss": 0.8707,
+      "step": 8650
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 0.8697690367698669,
+      "learning_rate": 8.95913240200717e-07,
+      "loss": 0.8184,
+      "step": 8660
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 0.7483598589897156,
+      "learning_rate": 8.875752166968382e-07,
+      "loss": 0.835,
+      "step": 8670
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 0.7968469858169556,
+      "learning_rate": 8.792723938365599e-07,
+      "loss": 0.8641,
+      "step": 8680
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 0.8039857745170593,
+      "learning_rate": 8.710048426885864e-07,
+      "loss": 0.8739,
+      "step": 8690
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 0.8848390579223633,
+      "learning_rate": 8.627726340197157e-07,
+      "loss": 0.8556,
+      "step": 8700
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 0.7995756268501282,
+      "learning_rate": 8.545758382942232e-07,
+      "loss": 0.8573,
+      "step": 8710
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 0.8631070256233215,
+      "learning_rate": 8.464145256732692e-07,
+      "loss": 0.8586,
+      "step": 8720
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 1.1139453649520874,
+      "learning_rate": 8.382887660142902e-07,
+      "loss": 0.8566,
+      "step": 8730
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 0.7360924482345581,
+      "learning_rate": 8.30198628870404e-07,
+      "loss": 0.8924,
+      "step": 8740
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 0.7692661881446838,
+      "learning_rate": 8.221441834898175e-07,
+      "loss": 0.8336,
+      "step": 8750
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 0.7830842733383179,
+      "learning_rate": 8.141254988152264e-07,
+      "loss": 0.8753,
+      "step": 8760
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 0.709661602973938,
+      "learning_rate": 8.061426434832371e-07,
+      "loss": 0.8361,
+      "step": 8770
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 0.7729387879371643,
+      "learning_rate": 7.981956858237605e-07,
+      "loss": 0.8434,
+      "step": 8780
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 0.7739190459251404,
+      "learning_rate": 7.902846938594494e-07,
+      "loss": 0.8736,
+      "step": 8790
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 0.8497752547264099,
+      "learning_rate": 7.824097353050986e-07,
+      "loss": 0.8587,
+      "step": 8800
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 0.7665298581123352,
+      "learning_rate": 7.745708775670752e-07,
+      "loss": 0.866,
+      "step": 8810
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 0.726333498954773,
+      "learning_rate": 7.667681877427363e-07,
+      "loss": 0.8446,
+      "step": 8820
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 1.0447399616241455,
+      "learning_rate": 7.590017326198568e-07,
+      "loss": 0.8398,
+      "step": 8830
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 0.8425661325454712,
+      "learning_rate": 7.512715786760604e-07,
+      "loss": 0.8618,
+      "step": 8840
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 0.771229088306427,
+      "learning_rate": 7.435777920782444e-07,
+      "loss": 0.8896,
+      "step": 8850
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 0.8381311297416687,
+      "learning_rate": 7.359204386820185e-07,
+      "loss": 0.8455,
+      "step": 8860
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 0.7867329120635986,
+      "learning_rate": 7.282995840311379e-07,
+      "loss": 0.8412,
+      "step": 8870
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 0.7634229063987732,
+      "learning_rate": 7.207152933569455e-07,
+      "loss": 0.8413,
+      "step": 8880
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 0.73702073097229,
+      "learning_rate": 7.131676315778136e-07,
+      "loss": 0.8605,
+      "step": 8890
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 0.8490579724311829,
+      "learning_rate": 7.056566632985801e-07,
+      "loss": 0.843,
+      "step": 8900
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 0.7541806101799011,
+      "learning_rate": 6.981824528100079e-07,
+      "loss": 0.863,
+      "step": 8910
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 0.9212934374809265,
+      "learning_rate": 6.907450640882246e-07,
+      "loss": 0.8736,
+      "step": 8920
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 1.2079908847808838,
+      "learning_rate": 6.83344560794183e-07,
+      "loss": 0.9028,
+      "step": 8930
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 0.7188634276390076,
+      "learning_rate": 6.759810062731076e-07,
+      "loss": 0.8343,
+      "step": 8940
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 0.8919622898101807,
+      "learning_rate": 6.686544635539582e-07,
+      "loss": 0.858,
+      "step": 8950
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 0.7763935923576355,
+      "learning_rate": 6.613649953488921e-07,
+      "loss": 0.8932,
+      "step": 8960
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 0.8691978454589844,
+      "learning_rate": 6.541126640527195e-07,
+      "loss": 0.8408,
+      "step": 8970
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 0.8207686543464661,
+      "learning_rate": 6.468975317423765e-07,
+      "loss": 0.8581,
+      "step": 8980
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 0.8498722314834595,
+      "learning_rate": 6.3971966017639e-07,
+      "loss": 0.8668,
+      "step": 8990
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 0.7766693830490112,
+      "learning_rate": 6.325791107943541e-07,
+      "loss": 0.9057,
+      "step": 9000
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 0.8322445750236511,
+      "learning_rate": 6.254759447163955e-07,
+      "loss": 0.8646,
+      "step": 9010
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 0.9103628396987915,
+      "learning_rate": 6.184102227426597e-07,
+      "loss": 0.8277,
+      "step": 9020
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 0.9469191431999207,
+      "learning_rate": 6.113820053527835e-07,
+      "loss": 0.868,
+      "step": 9030
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 0.7680910229682922,
+      "learning_rate": 6.043913527053802e-07,
+      "loss": 0.8611,
+      "step": 9040
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 0.8795392513275146,
+      "learning_rate": 5.974383246375276e-07,
+      "loss": 0.8764,
+      "step": 9050
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 0.7836953401565552,
+      "learning_rate": 5.905229806642488e-07,
+      "loss": 0.8443,
+      "step": 9060
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 1.0021981000900269,
+      "learning_rate": 5.836453799780084e-07,
+      "loss": 0.8784,
+      "step": 9070
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 0.8484304547309875,
+      "learning_rate": 5.768055814482037e-07,
+      "loss": 0.8697,
+      "step": 9080
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 0.783854603767395,
+      "learning_rate": 5.70003643620663e-07,
+      "loss": 0.8593,
+      "step": 9090
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 1.0196940898895264,
+      "learning_rate": 5.632396247171429e-07,
+      "loss": 0.8377,
+      "step": 9100
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 0.8246765732765198,
+      "learning_rate": 5.565135826348251e-07,
+      "loss": 0.8628,
+      "step": 9110
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 1.347435474395752,
+      "learning_rate": 5.498255749458314e-07,
+      "loss": 0.8629,
+      "step": 9120
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 0.7431601881980896,
+      "learning_rate": 5.431756588967202e-07,
+      "loss": 0.8397,
+      "step": 9130
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 0.8778267502784729,
+      "learning_rate": 5.365638914080068e-07,
+      "loss": 0.8178,
+      "step": 9140
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 0.8058485984802246,
+      "learning_rate": 5.299903290736658e-07,
+      "loss": 0.8556,
+      "step": 9150
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 0.7546533346176147,
+      "learning_rate": 5.23455028160652e-07,
+      "loss": 0.837,
+      "step": 9160
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 0.6787422895431519,
+      "learning_rate": 5.169580446084226e-07,
+      "loss": 0.8733,
+      "step": 9170
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 0.8343322277069092,
+      "learning_rate": 5.104994340284497e-07,
+      "loss": 0.9156,
+      "step": 9180
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 0.8385430574417114,
+      "learning_rate": 5.040792517037502e-07,
+      "loss": 0.8289,
+      "step": 9190
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 0.9122020602226257,
+      "learning_rate": 4.97697552588411e-07,
+      "loss": 0.8742,
+      "step": 9200
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 0.7369406223297119,
+      "learning_rate": 4.913543913071194e-07,
+      "loss": 0.8744,
+      "step": 9210
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 0.8071365356445312,
+      "learning_rate": 4.850498221546951e-07,
+      "loss": 0.8887,
+      "step": 9220
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 0.7439571619033813,
+      "learning_rate": 4.787838990956228e-07,
+      "loss": 0.8794,
+      "step": 9230
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 0.8147474527359009,
+      "learning_rate": 4.7255667576359687e-07,
+      "loss": 0.8621,
+      "step": 9240
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 0.90129554271698,
+      "learning_rate": 4.6636820546105413e-07,
+      "loss": 0.8603,
+      "step": 9250
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 0.7907412648200989,
+      "learning_rate": 4.602185411587251e-07,
+      "loss": 0.8334,
+      "step": 9260
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 0.7743856310844421,
+      "learning_rate": 4.541077354951723e-07,
+      "loss": 0.8593,
+      "step": 9270
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 0.72720867395401,
+      "learning_rate": 4.4803584077634967e-07,
+      "loss": 0.8442,
+      "step": 9280
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 0.8847765922546387,
+      "learning_rate": 4.4200290897514766e-07,
+      "loss": 0.8629,
+      "step": 9290
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 0.7650594115257263,
+      "learning_rate": 4.3600899173094845e-07,
+      "loss": 0.8536,
+      "step": 9300
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 0.8924193382263184,
+      "learning_rate": 4.300541403491909e-07,
+      "loss": 0.8768,
+      "step": 9310
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 0.7987867593765259,
+      "learning_rate": 4.241384058009185e-07,
+      "loss": 0.8197,
+      "step": 9320
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 0.7926146984100342,
+      "learning_rate": 4.1826183872235916e-07,
+      "loss": 0.8918,
+      "step": 9330
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 0.7198899388313293,
+      "learning_rate": 4.1242448941447743e-07,
+      "loss": 0.8798,
+      "step": 9340
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 0.7921025156974792,
+      "learning_rate": 4.066264078425547e-07,
+      "loss": 0.8952,
+      "step": 9350
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 0.8365041613578796,
+      "learning_rate": 4.008676436357539e-07,
+      "loss": 0.8706,
+      "step": 9360
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 0.6742424368858337,
+      "learning_rate": 3.9514824608669676e-07,
+      "loss": 0.876,
+      "step": 9370
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 0.9962718486785889,
+      "learning_rate": 3.894682641510478e-07,
+      "loss": 0.832,
+      "step": 9380
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 1.0092328786849976,
+      "learning_rate": 3.8382774644708463e-07,
+      "loss": 0.8799,
+      "step": 9390
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 0.8240341544151306,
+      "learning_rate": 3.782267412552909e-07,
+      "loss": 0.8486,
+      "step": 9400
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 0.8725962042808533,
+      "learning_rate": 3.7266529651793725e-07,
+      "loss": 0.8162,
+      "step": 9410
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 0.8800333738327026,
+      "learning_rate": 3.6714345983867696e-07,
+      "loss": 0.8507,
+      "step": 9420
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 0.8511844873428345,
+      "learning_rate": 3.6166127848213105e-07,
+      "loss": 0.8313,
+      "step": 9430
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 0.827422022819519,
+      "learning_rate": 3.5621879937348836e-07,
+      "loss": 0.8571,
+      "step": 9440
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 0.9481132626533508,
+      "learning_rate": 3.508160690981055e-07,
+      "loss": 0.8688,
+      "step": 9450
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 0.7517793774604797,
+      "learning_rate": 3.454531339011019e-07,
+      "loss": 0.8787,
+      "step": 9460
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 0.8055853247642517,
+      "learning_rate": 3.4013003968697167e-07,
+      "loss": 0.8613,
+      "step": 9470
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 0.9458131790161133,
+      "learning_rate": 3.3484683201918276e-07,
+      "loss": 0.861,
+      "step": 9480
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 0.7651836276054382,
+      "learning_rate": 3.2960355611979245e-07,
+      "loss": 0.8467,
+      "step": 9490
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 0.7675281167030334,
+      "learning_rate": 3.244002568690613e-07,
+      "loss": 0.8539,
+      "step": 9500
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 0.8231499195098877,
+      "learning_rate": 3.19236978805062e-07,
+      "loss": 0.8681,
+      "step": 9510
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 0.8024943470954895,
+      "learning_rate": 3.1411376612330513e-07,
+      "loss": 0.8717,
+      "step": 9520
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 0.7540535926818848,
+      "learning_rate": 3.0903066267635727e-07,
+      "loss": 0.8461,
+      "step": 9530
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 1.0400876998901367,
+      "learning_rate": 3.039877119734691e-07,
+      "loss": 0.8188,
+      "step": 9540
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 0.773931086063385,
+      "learning_rate": 2.989849571801967e-07,
+      "loss": 0.8466,
+      "step": 9550
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 1.0147455930709839,
+      "learning_rate": 2.940224411180376e-07,
+      "loss": 0.8794,
+      "step": 9560
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 0.8649349808692932,
+      "learning_rate": 2.891002062640641e-07,
+      "loss": 0.8618,
+      "step": 9570
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 0.7081354856491089,
+      "learning_rate": 2.842182947505556e-07,
+      "loss": 0.8599,
+      "step": 9580
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 0.7580844163894653,
+      "learning_rate": 2.7937674836464256e-07,
+      "loss": 0.8286,
+      "step": 9590
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 0.858424186706543,
+      "learning_rate": 2.7457560854794616e-07,
+      "loss": 0.8558,
+      "step": 9600
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 0.9366195201873779,
+      "learning_rate": 2.698149163962227e-07,
+      "loss": 0.8387,
+      "step": 9610
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 0.7247710227966309,
+      "learning_rate": 2.6509471265901476e-07,
+      "loss": 0.833,
+      "step": 9620
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 1.0109031200408936,
+      "learning_rate": 2.604150377393011e-07,
+      "loss": 0.8317,
+      "step": 9630
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 0.7899563908576965,
+      "learning_rate": 2.557759316931513e-07,
+      "loss": 0.8511,
+      "step": 9640
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 0.8207105994224548,
+      "learning_rate": 2.5117743422937823e-07,
+      "loss": 0.883,
+      "step": 9650
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 0.7731668949127197,
+      "learning_rate": 2.4661958470920845e-07,
+      "loss": 0.8562,
+      "step": 9660
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 0.7830813527107239,
+      "learning_rate": 2.421024221459334e-07,
+      "loss": 0.8603,
+      "step": 9670
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 0.8500844836235046,
+      "learning_rate": 2.3762598520458536e-07,
+      "loss": 0.8858,
+      "step": 9680
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 0.6461296081542969,
+      "learning_rate": 2.3319031220159994e-07,
+      "loss": 0.8787,
+      "step": 9690
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 0.745896577835083,
+      "learning_rate": 2.287954411044896e-07,
+      "loss": 0.8686,
+      "step": 9700
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 0.8797740936279297,
+      "learning_rate": 2.2444140953152227e-07,
+      "loss": 0.8693,
+      "step": 9710
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 0.9192375540733337,
+      "learning_rate": 2.2012825475139498e-07,
+      "loss": 0.8895,
+      "step": 9720
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 0.706160306930542,
+      "learning_rate": 2.1585601368291575e-07,
+      "loss": 0.8317,
+      "step": 9730
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 0.7271298170089722,
+      "learning_rate": 2.1162472289468884e-07,
+      "loss": 0.8621,
+      "step": 9740
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 0.8096296787261963,
+      "learning_rate": 2.0743441860480218e-07,
+      "loss": 0.8647,
+      "step": 9750
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 0.7598336338996887,
+      "learning_rate": 2.0328513668051553e-07,
+      "loss": 0.8649,
+      "step": 9760
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 0.7457004189491272,
+      "learning_rate": 1.9917691263795336e-07,
+      "loss": 0.8271,
+      "step": 9770
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 0.8716655373573303,
+      "learning_rate": 1.9510978164180404e-07,
+      "loss": 0.8513,
+      "step": 9780
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 0.8340162038803101,
+      "learning_rate": 1.9108377850501347e-07,
+      "loss": 0.8391,
+      "step": 9790
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 0.7895581722259521,
+      "learning_rate": 1.8709893768849406e-07,
+      "loss": 0.8619,
+      "step": 9800
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 0.7233043313026428,
+      "learning_rate": 1.8315529330082182e-07,
+      "loss": 0.89,
+      "step": 9810
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 0.7920761108398438,
+      "learning_rate": 1.7925287909795087e-07,
+      "loss": 0.8384,
+      "step": 9820
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 0.7956525683403015,
+      "learning_rate": 1.7539172848292263e-07,
+      "loss": 0.8274,
+      "step": 9830
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 1.2218879461288452,
+      "learning_rate": 1.7157187450557834e-07,
+      "loss": 0.8782,
+      "step": 9840
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 0.7387891411781311,
+      "learning_rate": 1.6779334986227636e-07,
+      "loss": 0.846,
+      "step": 9850
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 0.9012438058853149,
+      "learning_rate": 1.6405618689561474e-07,
+      "loss": 0.8535,
+      "step": 9860
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 0.8426781892776489,
+      "learning_rate": 1.6036041759415255e-07,
+      "loss": 0.8693,
+      "step": 9870
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 0.845955491065979,
+      "learning_rate": 1.5670607359213442e-07,
+      "loss": 0.8467,
+      "step": 9880
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 0.8698325753211975,
+      "learning_rate": 1.5309318616922587e-07,
+      "loss": 0.8336,
+      "step": 9890
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 1.0371123552322388,
+      "learning_rate": 1.4952178625023627e-07,
+      "loss": 0.8585,
+      "step": 9900
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 0.7943004369735718,
+      "learning_rate": 1.45991904404863e-07,
+      "loss": 0.8686,
+      "step": 9910
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 1.2979638576507568,
+      "learning_rate": 1.4250357084742427e-07,
+      "loss": 0.8768,
+      "step": 9920
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 0.6948923468589783,
+      "learning_rate": 1.3905681543660288e-07,
+      "loss": 0.8787,
+      "step": 9930
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 0.8536478281021118,
+      "learning_rate": 1.3565166767519012e-07,
+      "loss": 0.8317,
+      "step": 9940
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 0.8086930513381958,
+      "learning_rate": 1.322881567098311e-07,
+      "loss": 0.875,
+      "step": 9950
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 1.0101767778396606,
+      "learning_rate": 1.2896631133078151e-07,
+      "loss": 0.8404,
+      "step": 9960
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 0.7897870540618896,
+      "learning_rate": 1.2568615997165246e-07,
+      "loss": 0.8691,
+      "step": 9970
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 0.7759121656417847,
+      "learning_rate": 1.2244773070917427e-07,
+      "loss": 0.8452,
+      "step": 9980
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 1.095391035079956,
+      "learning_rate": 1.1925105126295356e-07,
+      "loss": 0.9201,
+      "step": 9990
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 0.8890740275382996,
+      "learning_rate": 1.1609614899523447e-07,
+      "loss": 0.8028,
+      "step": 10000
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 10738,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 2,
+  "save_steps": 500,
+  "total_flos": 3.600007041152503e+19,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}