diff --git "a/checkpoints/checkpoint-14000/trainer_state.json" "b/checkpoints/checkpoint-14000/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoints/checkpoint-14000/trainer_state.json"
@@ -0,0 +1,9834 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.6867921288716373,
+  "eval_steps": 500,
+  "global_step": 14000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0004905658063368838,
+      "grad_norm": 48.75,
+      "learning_rate": 1.471550032701112e-07,
+      "loss": 103.1825,
+      "step": 10
+    },
+    {
+      "epoch": 0.0009811316126737675,
+      "grad_norm": 49.5,
+      "learning_rate": 3.106605624591236e-07,
+      "loss": 103.9151,
+      "step": 20
+    },
+    {
+      "epoch": 0.0014716974190106514,
+      "grad_norm": 47.0,
+      "learning_rate": 4.7416612164813603e-07,
+      "loss": 104.2889,
+      "step": 30
+    },
+    {
+      "epoch": 0.001962263225347535,
+      "grad_norm": 40.0,
+      "learning_rate": 6.376716808371485e-07,
+      "loss": 104.1391,
+      "step": 40
+    },
+    {
+      "epoch": 0.002452829031684419,
+      "grad_norm": 42.25,
+      "learning_rate": 8.011772400261609e-07,
+      "loss": 104.4931,
+      "step": 50
+    },
+    {
+      "epoch": 0.0029433948380213027,
+      "grad_norm": 41.75,
+      "learning_rate": 9.646827992151733e-07,
+      "loss": 104.123,
+      "step": 60
+    },
+    {
+      "epoch": 0.003433960644358187,
+      "grad_norm": 43.0,
+      "learning_rate": 1.1281883584041859e-06,
+      "loss": 103.4803,
+      "step": 70
+    },
+    {
+      "epoch": 0.00392452645069507,
+      "grad_norm": 46.5,
+      "learning_rate": 1.2916939175931983e-06,
+      "loss": 103.4754,
+      "step": 80
+    },
+    {
+      "epoch": 0.0044150922570319545,
+      "grad_norm": 44.5,
+      "learning_rate": 1.4551994767822106e-06,
+      "loss": 103.3514,
+      "step": 90
+    },
+    {
+      "epoch": 0.004905658063368838,
+      "grad_norm": 43.0,
+      "learning_rate": 1.618705035971223e-06,
+      "loss": 103.8501,
+      "step": 100
+    },
+    {
+      "epoch": 0.005396223869705722,
+      "grad_norm": 50.25,
+      "learning_rate": 1.7822105951602354e-06,
+      "loss": 102.954,
+      "step": 110
+    },
+    {
+      "epoch": 0.0058867896760426055,
+      "grad_norm": 41.75,
+      "learning_rate": 1.945716154349248e-06,
+      "loss": 104.1795,
+      "step": 120
+    },
+    {
+      "epoch": 0.006377355482379489,
+      "grad_norm": 40.0,
+      "learning_rate": 2.1092217135382606e-06,
+      "loss": 104.1945,
+      "step": 130
+    },
+    {
+      "epoch": 0.006867921288716374,
+      "grad_norm": 42.5,
+      "learning_rate": 2.2727272727272728e-06,
+      "loss": 103.516,
+      "step": 140
+    },
+    {
+      "epoch": 0.007358487095053257,
+      "grad_norm": 43.5,
+      "learning_rate": 2.4362328319162854e-06,
+      "loss": 103.7904,
+      "step": 150
+    },
+    {
+      "epoch": 0.00784905290139014,
+      "grad_norm": 40.5,
+      "learning_rate": 2.5997383911052975e-06,
+      "loss": 103.0723,
+      "step": 160
+    },
+    {
+      "epoch": 0.008339618707727025,
+      "grad_norm": 43.5,
+      "learning_rate": 2.76324395029431e-06,
+      "loss": 104.0213,
+      "step": 170
+    },
+    {
+      "epoch": 0.008830184514063909,
+      "grad_norm": 42.5,
+      "learning_rate": 2.9267495094833227e-06,
+      "loss": 103.3656,
+      "step": 180
+    },
+    {
+      "epoch": 0.009320750320400792,
+      "grad_norm": 44.5,
+      "learning_rate": 3.090255068672335e-06,
+      "loss": 103.67,
+      "step": 190
+    },
+    {
+      "epoch": 0.009811316126737676,
+      "grad_norm": 44.75,
+      "learning_rate": 3.253760627861348e-06,
+      "loss": 104.2247,
+      "step": 200
+    },
+    {
+      "epoch": 0.01030188193307456,
+      "grad_norm": 47.75,
+      "learning_rate": 3.41726618705036e-06,
+      "loss": 103.4264,
+      "step": 210
+    },
+    {
+      "epoch": 0.010792447739411444,
+      "grad_norm": 46.5,
+      "learning_rate": 3.5807717462393727e-06,
+      "loss": 103.728,
+      "step": 220
+    },
+    {
+      "epoch": 0.011283013545748328,
+      "grad_norm": 41.0,
+      "learning_rate": 3.744277305428385e-06,
+      "loss": 103.5934,
+      "step": 230
+    },
+    {
+      "epoch": 0.011773579352085211,
+      "grad_norm": 46.25,
+      "learning_rate": 3.907782864617397e-06,
+      "loss": 102.8712,
+      "step": 240
+    },
+    {
+      "epoch": 0.012264145158422095,
+      "grad_norm": 49.5,
+      "learning_rate": 4.07128842380641e-06,
+      "loss": 103.0208,
+      "step": 250
+    },
+    {
+      "epoch": 0.012754710964758978,
+      "grad_norm": 48.75,
+      "learning_rate": 4.234793982995422e-06,
+      "loss": 103.3193,
+      "step": 260
+    },
+    {
+      "epoch": 0.013245276771095863,
+      "grad_norm": 42.75,
+      "learning_rate": 4.398299542184434e-06,
+      "loss": 101.6977,
+      "step": 270
+    },
+    {
+      "epoch": 0.013735842577432747,
+      "grad_norm": 44.5,
+      "learning_rate": 4.5618051013734465e-06,
+      "loss": 102.7333,
+      "step": 280
+    },
+    {
+      "epoch": 0.01422640838376963,
+      "grad_norm": 46.75,
+      "learning_rate": 4.725310660562459e-06,
+      "loss": 102.2653,
+      "step": 290
+    },
+    {
+      "epoch": 0.014716974190106515,
+      "grad_norm": 49.5,
+      "learning_rate": 4.888816219751472e-06,
+      "loss": 102.4468,
+      "step": 300
+    },
+    {
+      "epoch": 0.015207539996443397,
+      "grad_norm": 41.25,
+      "learning_rate": 5.052321778940484e-06,
+      "loss": 101.577,
+      "step": 310
+    },
+    {
+      "epoch": 0.01569810580278028,
+      "grad_norm": 46.25,
+      "learning_rate": 5.215827338129497e-06,
+      "loss": 102.1117,
+      "step": 320
+    },
+    {
+      "epoch": 0.016188671609117165,
+      "grad_norm": 44.5,
+      "learning_rate": 5.3793328973185095e-06,
+      "loss": 102.6705,
+      "step": 330
+    },
+    {
+      "epoch": 0.01667923741545405,
+      "grad_norm": 43.5,
+      "learning_rate": 5.542838456507521e-06,
+      "loss": 102.1149,
+      "step": 340
+    },
+    {
+      "epoch": 0.017169803221790934,
+      "grad_norm": 45.0,
+      "learning_rate": 5.706344015696534e-06,
+      "loss": 102.8788,
+      "step": 350
+    },
+    {
+      "epoch": 0.017660369028127818,
+      "grad_norm": 47.5,
+      "learning_rate": 5.869849574885546e-06,
+      "loss": 102.119,
+      "step": 360
+    },
+    {
+      "epoch": 0.0181509348344647,
+      "grad_norm": 43.75,
+      "learning_rate": 6.033355134074559e-06,
+      "loss": 101.3416,
+      "step": 370
+    },
+    {
+      "epoch": 0.018641500640801584,
+      "grad_norm": 44.5,
+      "learning_rate": 6.196860693263571e-06,
+      "loss": 101.6792,
+      "step": 380
+    },
+    {
+      "epoch": 0.01913206644713847,
+      "grad_norm": 43.0,
+      "learning_rate": 6.360366252452583e-06,
+      "loss": 100.9905,
+      "step": 390
+    },
+    {
+      "epoch": 0.019622632253475353,
+      "grad_norm": 55.25,
+      "learning_rate": 6.523871811641597e-06,
+      "loss": 100.5244,
+      "step": 400
+    },
+    {
+      "epoch": 0.020113198059812237,
+      "grad_norm": 56.75,
+      "learning_rate": 6.687377370830608e-06,
+      "loss": 100.8768,
+      "step": 410
+    },
+    {
+      "epoch": 0.02060376386614912,
+      "grad_norm": 48.75,
+      "learning_rate": 6.850882930019621e-06,
+      "loss": 101.3346,
+      "step": 420
+    },
+    {
+      "epoch": 0.021094329672486003,
+      "grad_norm": 43.25,
+      "learning_rate": 7.014388489208633e-06,
+      "loss": 100.5578,
+      "step": 430
+    },
+    {
+      "epoch": 0.021584895478822887,
+      "grad_norm": 44.0,
+      "learning_rate": 7.177894048397646e-06,
+      "loss": 99.5179,
+      "step": 440
+    },
+    {
+      "epoch": 0.022075461285159772,
+      "grad_norm": 48.75,
+      "learning_rate": 7.341399607586659e-06,
+      "loss": 101.0797,
+      "step": 450
+    },
+    {
+      "epoch": 0.022566027091496656,
+      "grad_norm": 44.5,
+      "learning_rate": 7.504905166775671e-06,
+      "loss": 99.5985,
+      "step": 460
+    },
+    {
+      "epoch": 0.023056592897833537,
+      "grad_norm": 60.5,
+      "learning_rate": 7.668410725964683e-06,
+      "loss": 99.062,
+      "step": 470
+    },
+    {
+      "epoch": 0.023547158704170422,
+      "grad_norm": 53.5,
+      "learning_rate": 7.831916285153695e-06,
+      "loss": 99.0438,
+      "step": 480
+    },
+    {
+      "epoch": 0.024037724510507306,
+      "grad_norm": 51.0,
+      "learning_rate": 7.995421844342708e-06,
+      "loss": 98.5337,
+      "step": 490
+    },
+    {
+      "epoch": 0.02452829031684419,
+      "grad_norm": 48.25,
+      "learning_rate": 8.15892740353172e-06,
+      "loss": 97.4177,
+      "step": 500
+    },
+    {
+      "epoch": 0.025018856123181075,
+      "grad_norm": 55.25,
+      "learning_rate": 8.322432962720734e-06,
+      "loss": 97.3938,
+      "step": 510
+    },
+    {
+      "epoch": 0.025509421929517957,
+      "grad_norm": 81.0,
+      "learning_rate": 8.485938521909745e-06,
+      "loss": 96.0514,
+      "step": 520
+    },
+    {
+      "epoch": 0.02599998773585484,
+      "grad_norm": 70.5,
+      "learning_rate": 8.649444081098757e-06,
+      "loss": 96.4423,
+      "step": 530
+    },
+    {
+      "epoch": 0.026490553542191726,
+      "grad_norm": 87.5,
+      "learning_rate": 8.812949640287769e-06,
+      "loss": 95.6965,
+      "step": 540
+    },
+    {
+      "epoch": 0.02698111934852861,
+      "grad_norm": 113.0,
+      "learning_rate": 8.976455199476782e-06,
+      "loss": 94.126,
+      "step": 550
+    },
+    {
+      "epoch": 0.027471685154865495,
+      "grad_norm": 147.0,
+      "learning_rate": 9.139960758665796e-06,
+      "loss": 92.2339,
+      "step": 560
+    },
+    {
+      "epoch": 0.027962250961202376,
+      "grad_norm": 163.0,
+      "learning_rate": 9.303466317854807e-06,
+      "loss": 88.7373,
+      "step": 570
+    },
+    {
+      "epoch": 0.02845281676753926,
+      "grad_norm": 274.0,
+      "learning_rate": 9.466971877043821e-06,
+      "loss": 85.5932,
+      "step": 580
+    },
+    {
+      "epoch": 0.028943382573876145,
+      "grad_norm": 424.0,
+      "learning_rate": 9.630477436232833e-06,
+      "loss": 81.2647,
+      "step": 590
+    },
+    {
+      "epoch": 0.02943394838021303,
+      "grad_norm": 356.0,
+      "learning_rate": 9.793982995421844e-06,
+      "loss": 79.4912,
+      "step": 600
+    },
+    {
+      "epoch": 0.029924514186549914,
+      "grad_norm": 338.0,
+      "learning_rate": 9.957488554610856e-06,
+      "loss": 72.7454,
+      "step": 610
+    },
+    {
+      "epoch": 0.030415079992886795,
+      "grad_norm": 398.0,
+      "learning_rate": 1.012099411379987e-05,
+      "loss": 66.9605,
+      "step": 620
+    },
+    {
+      "epoch": 0.03090564579922368,
+      "grad_norm": 342.0,
+      "learning_rate": 1.0284499672988881e-05,
+      "loss": 63.5135,
+      "step": 630
+    },
+    {
+      "epoch": 0.03139621160556056,
+      "grad_norm": 372.0,
+      "learning_rate": 1.0448005232177895e-05,
+      "loss": 57.5055,
+      "step": 640
+    },
+    {
+      "epoch": 0.031886777411897445,
+      "grad_norm": 334.0,
+      "learning_rate": 1.0611510791366908e-05,
+      "loss": 51.0307,
+      "step": 650
+    },
+    {
+      "epoch": 0.03237734321823433,
+      "grad_norm": 360.0,
+      "learning_rate": 1.077501635055592e-05,
+      "loss": 47.7223,
+      "step": 660
+    },
+    {
+      "epoch": 0.032867909024571214,
+      "grad_norm": 270.0,
+      "learning_rate": 1.0938521909744932e-05,
+      "loss": 43.6193,
+      "step": 670
+    },
+    {
+      "epoch": 0.0333584748309081,
+      "grad_norm": 252.0,
+      "learning_rate": 1.1102027468933943e-05,
+      "loss": 39.8644,
+      "step": 680
+    },
+    {
+      "epoch": 0.03384904063724498,
+      "grad_norm": 232.0,
+      "learning_rate": 1.1265533028122957e-05,
+      "loss": 41.45,
+      "step": 690
+    },
+    {
+      "epoch": 0.03433960644358187,
+      "grad_norm": 134.0,
+      "learning_rate": 1.1429038587311969e-05,
+      "loss": 34.4322,
+      "step": 700
+    },
+    {
+      "epoch": 0.03483017224991875,
+      "grad_norm": 113.5,
+      "learning_rate": 1.1592544146500982e-05,
+      "loss": 39.6195,
+      "step": 710
+    },
+    {
+      "epoch": 0.035320738056255636,
+      "grad_norm": 220.0,
+      "learning_rate": 1.1756049705689994e-05,
+      "loss": 35.0983,
+      "step": 720
+    },
+    {
+      "epoch": 0.03581130386259252,
+      "grad_norm": 109.0,
+      "learning_rate": 1.1919555264879007e-05,
+      "loss": 34.4757,
+      "step": 730
+    },
+    {
+      "epoch": 0.0363018696689294,
+      "grad_norm": 149.0,
+      "learning_rate": 1.2083060824068019e-05,
+      "loss": 34.2608,
+      "step": 740
+    },
+    {
+      "epoch": 0.03679243547526628,
+      "grad_norm": 66.5,
+      "learning_rate": 1.224656638325703e-05,
+      "loss": 31.0653,
+      "step": 750
+    },
+    {
+      "epoch": 0.03728300128160317,
+      "grad_norm": 96.5,
+      "learning_rate": 1.2410071942446044e-05,
+      "loss": 32.5303,
+      "step": 760
+    },
+    {
+      "epoch": 0.03777356708794005,
+      "grad_norm": 82.5,
+      "learning_rate": 1.2573577501635056e-05,
+      "loss": 32.1891,
+      "step": 770
+    },
+    {
+      "epoch": 0.03826413289427694,
+      "grad_norm": 72.0,
+      "learning_rate": 1.2737083060824068e-05,
+      "loss": 32.1906,
+      "step": 780
+    },
+    {
+      "epoch": 0.03875469870061382,
+      "grad_norm": 58.25,
+      "learning_rate": 1.2900588620013083e-05,
+      "loss": 32.5546,
+      "step": 790
+    },
+    {
+      "epoch": 0.039245264506950706,
+      "grad_norm": 141.0,
+      "learning_rate": 1.3064094179202095e-05,
+      "loss": 29.5903,
+      "step": 800
+    },
+    {
+      "epoch": 0.03973583031328759,
+      "grad_norm": 87.5,
+      "learning_rate": 1.3227599738391105e-05,
+      "loss": 28.0366,
+      "step": 810
+    },
+    {
+      "epoch": 0.040226396119624475,
+      "grad_norm": 89.5,
+      "learning_rate": 1.3391105297580116e-05,
+      "loss": 28.156,
+      "step": 820
+    },
+    {
+      "epoch": 0.04071696192596136,
+      "grad_norm": 57.0,
+      "learning_rate": 1.3554610856769132e-05,
+      "loss": 28.1194,
+      "step": 830
+    },
+    {
+      "epoch": 0.04120752773229824,
+      "grad_norm": 58.0,
+      "learning_rate": 1.3718116415958143e-05,
+      "loss": 24.1945,
+      "step": 840
+    },
+    {
+      "epoch": 0.04169809353863512,
+      "grad_norm": 36.0,
+      "learning_rate": 1.3881621975147155e-05,
+      "loss": 28.0072,
+      "step": 850
+    },
+    {
+      "epoch": 0.042188659344972006,
+      "grad_norm": 49.75,
+      "learning_rate": 1.404512753433617e-05,
+      "loss": 26.4452,
+      "step": 860
+    },
+    {
+      "epoch": 0.04267922515130889,
+      "grad_norm": 49.5,
+      "learning_rate": 1.420863309352518e-05,
+      "loss": 27.5915,
+      "step": 870
+    },
+    {
+      "epoch": 0.043169790957645775,
+      "grad_norm": 45.5,
+      "learning_rate": 1.4372138652714192e-05,
+      "loss": 25.8689,
+      "step": 880
+    },
+    {
+      "epoch": 0.04366035676398266,
+      "grad_norm": 75.0,
+      "learning_rate": 1.4535644211903204e-05,
+      "loss": 26.5536,
+      "step": 890
+    },
+    {
+      "epoch": 0.044150922570319544,
+      "grad_norm": 38.0,
+      "learning_rate": 1.4699149771092219e-05,
+      "loss": 29.0372,
+      "step": 900
+    },
+    {
+      "epoch": 0.04464148837665643,
+      "grad_norm": 36.25,
+      "learning_rate": 1.486265533028123e-05,
+      "loss": 25.1781,
+      "step": 910
+    },
+    {
+      "epoch": 0.04513205418299331,
+      "grad_norm": 34.5,
+      "learning_rate": 1.5026160889470242e-05,
+      "loss": 25.2917,
+      "step": 920
+    },
+    {
+      "epoch": 0.04562261998933019,
+      "grad_norm": 30.0,
+      "learning_rate": 1.5189666448659254e-05,
+      "loss": 24.7534,
+      "step": 930
+    },
+    {
+      "epoch": 0.046113185795667075,
+      "grad_norm": 20.75,
+      "learning_rate": 1.535317200784827e-05,
+      "loss": 24.7541,
+      "step": 940
+    },
+    {
+      "epoch": 0.04660375160200396,
+      "grad_norm": 111.5,
+      "learning_rate": 1.551667756703728e-05,
+      "loss": 25.0763,
+      "step": 950
+    },
+    {
+      "epoch": 0.047094317408340844,
+      "grad_norm": 35.75,
+      "learning_rate": 1.5680183126226293e-05,
+      "loss": 23.9253,
+      "step": 960
+    },
+    {
+      "epoch": 0.04758488321467773,
+      "grad_norm": 23.375,
+      "learning_rate": 1.5843688685415306e-05,
+      "loss": 22.4641,
+      "step": 970
+    },
+    {
+      "epoch": 0.04807544902101461,
+      "grad_norm": 18.125,
+      "learning_rate": 1.6007194244604316e-05,
+      "loss": 21.8752,
+      "step": 980
+    },
+    {
+      "epoch": 0.0485660148273515,
+      "grad_norm": 19.375,
+      "learning_rate": 1.617069980379333e-05,
+      "loss": 20.6978,
+      "step": 990
+    },
+    {
+      "epoch": 0.04905658063368838,
+      "grad_norm": 15.5625,
+      "learning_rate": 1.633420536298234e-05,
+      "loss": 19.1208,
+      "step": 1000
+    },
+    {
+      "epoch": 0.049547146440025266,
+      "grad_norm": 31.0,
+      "learning_rate": 1.6497710922171357e-05,
+      "loss": 19.9687,
+      "step": 1010
+    },
+    {
+      "epoch": 0.05003771224636215,
+      "grad_norm": 94.5,
+      "learning_rate": 1.6661216481360367e-05,
+      "loss": 19.8846,
+      "step": 1020
+    },
+    {
+      "epoch": 0.05052827805269903,
+      "grad_norm": 167.0,
+      "learning_rate": 1.682472204054938e-05,
+      "loss": 20.3119,
+      "step": 1030
+    },
+    {
+      "epoch": 0.05101884385903591,
+      "grad_norm": 31.875,
+      "learning_rate": 1.6988227599738393e-05,
+      "loss": 19.763,
+      "step": 1040
+    },
+    {
+      "epoch": 0.0515094096653728,
+      "grad_norm": 30.125,
+      "learning_rate": 1.7151733158927403e-05,
+      "loss": 18.4537,
+      "step": 1050
+    },
+    {
+      "epoch": 0.05199997547170968,
+      "grad_norm": 28.75,
+      "learning_rate": 1.7315238718116417e-05,
+      "loss": 18.8536,
+      "step": 1060
+    },
+    {
+      "epoch": 0.05249054127804657,
+      "grad_norm": 47.5,
+      "learning_rate": 1.7478744277305427e-05,
+      "loss": 19.0843,
+      "step": 1070
+    },
+    {
+      "epoch": 0.05298110708438345,
+      "grad_norm": 10.875,
+      "learning_rate": 1.7642249836494444e-05,
+      "loss": 18.1357,
+      "step": 1080
+    },
+    {
+      "epoch": 0.053471672890720336,
+      "grad_norm": 12.375,
+      "learning_rate": 1.7805755395683454e-05,
+      "loss": 17.805,
+      "step": 1090
+    },
+    {
+      "epoch": 0.05396223869705722,
+      "grad_norm": 23.5,
+      "learning_rate": 1.7969260954872467e-05,
+      "loss": 17.2343,
+      "step": 1100
+    },
+    {
+      "epoch": 0.054452804503394105,
+      "grad_norm": 16.0,
+      "learning_rate": 1.8132766514061477e-05,
+      "loss": 19.1025,
+      "step": 1110
+    },
+    {
+      "epoch": 0.05494337030973099,
+      "grad_norm": 13.4375,
+      "learning_rate": 1.829627207325049e-05,
+      "loss": 17.2674,
+      "step": 1120
+    },
+    {
+      "epoch": 0.05543393611606787,
+      "grad_norm": 34.25,
+      "learning_rate": 1.8459777632439504e-05,
+      "loss": 17.5344,
+      "step": 1130
+    },
+    {
+      "epoch": 0.05592450192240475,
+      "grad_norm": 23.375,
+      "learning_rate": 1.8623283191628514e-05,
+      "loss": 16.709,
+      "step": 1140
+    },
+    {
+      "epoch": 0.056415067728741636,
+      "grad_norm": 27.125,
+      "learning_rate": 1.878678875081753e-05,
+      "loss": 17.5936,
+      "step": 1150
+    },
+    {
+      "epoch": 0.05690563353507852,
+      "grad_norm": 8.6875,
+      "learning_rate": 1.895029431000654e-05,
+      "loss": 18.3802,
+      "step": 1160
+    },
+    {
+      "epoch": 0.057396199341415405,
+      "grad_norm": 14.5625,
+      "learning_rate": 1.9113799869195555e-05,
+      "loss": 16.4857,
+      "step": 1170
+    },
+    {
+      "epoch": 0.05788676514775229,
+      "grad_norm": 23.25,
+      "learning_rate": 1.9277305428384565e-05,
+      "loss": 15.4465,
+      "step": 1180
+    },
+    {
+      "epoch": 0.058377330954089174,
+      "grad_norm": 13.6875,
+      "learning_rate": 1.9440810987573578e-05,
+      "loss": 15.4513,
+      "step": 1190
+    },
+    {
+      "epoch": 0.05886789676042606,
+      "grad_norm": 45.5,
+      "learning_rate": 1.960431654676259e-05,
+      "loss": 15.0515,
+      "step": 1200
+    },
+    {
+      "epoch": 0.05935846256676294,
+      "grad_norm": 17.625,
+      "learning_rate": 1.97678221059516e-05,
+      "loss": 17.6995,
+      "step": 1210
+    },
+    {
+      "epoch": 0.05984902837309983,
+      "grad_norm": 26.375,
+      "learning_rate": 1.9931327665140615e-05,
+      "loss": 15.2606,
+      "step": 1220
+    },
+    {
+      "epoch": 0.060339594179436705,
+      "grad_norm": 11.0625,
+      "learning_rate": 2.009483322432963e-05,
+      "loss": 15.909,
+      "step": 1230
+    },
+    {
+      "epoch": 0.06083015998577359,
+      "grad_norm": 260.0,
+      "learning_rate": 2.0258338783518642e-05,
+      "loss": 14.311,
+      "step": 1240
+    },
+    {
+      "epoch": 0.061320725792110474,
+      "grad_norm": 9.4375,
+      "learning_rate": 2.0421844342707652e-05,
+      "loss": 13.3078,
+      "step": 1250
+    },
+    {
+      "epoch": 0.06181129159844736,
+      "grad_norm": 113.0,
+      "learning_rate": 2.0585349901896665e-05,
+      "loss": 14.7056,
+      "step": 1260
+    },
+    {
+      "epoch": 0.06230185740478424,
+      "grad_norm": 10.5625,
+      "learning_rate": 2.074885546108568e-05,
+      "loss": 13.4457,
+      "step": 1270
+    },
+    {
+      "epoch": 0.06279242321112112,
+      "grad_norm": 10.5625,
+      "learning_rate": 2.091236102027469e-05,
+      "loss": 13.3094,
+      "step": 1280
+    },
+    {
+      "epoch": 0.063282989017458,
+      "grad_norm": 22.625,
+      "learning_rate": 2.1075866579463702e-05,
+      "loss": 13.7567,
+      "step": 1290
+    },
+    {
+      "epoch": 0.06377355482379489,
+      "grad_norm": 12.625,
+      "learning_rate": 2.1239372138652716e-05,
+      "loss": 14.2653,
+      "step": 1300
+    },
+    {
+      "epoch": 0.06426412063013177,
+      "grad_norm": 143.0,
+      "learning_rate": 2.140287769784173e-05,
+      "loss": 14.4901,
+      "step": 1310
+    },
+    {
+      "epoch": 0.06475468643646866,
+      "grad_norm": 13.1875,
+      "learning_rate": 2.156638325703074e-05,
+      "loss": 13.5675,
+      "step": 1320
+    },
+    {
+      "epoch": 0.06524525224280554,
+      "grad_norm": 4.34375,
+      "learning_rate": 2.1729888816219753e-05,
+      "loss": 13.0833,
+      "step": 1330
+    },
+    {
+      "epoch": 0.06573581804914243,
+      "grad_norm": 63.25,
+      "learning_rate": 2.1893394375408766e-05,
+      "loss": 14.6539,
+      "step": 1340
+    },
+    {
+      "epoch": 0.06622638385547931,
+      "grad_norm": 6.78125,
+      "learning_rate": 2.2056899934597776e-05,
+      "loss": 12.8027,
+      "step": 1350
+    },
+    {
+      "epoch": 0.0667169496618162,
+      "grad_norm": 318.0,
+      "learning_rate": 2.222040549378679e-05,
+      "loss": 12.5888,
+      "step": 1360
+    },
+    {
+      "epoch": 0.06720751546815308,
+      "grad_norm": 5.5625,
+      "learning_rate": 2.2383911052975803e-05,
+      "loss": 12.7075,
+      "step": 1370
+    },
+    {
+      "epoch": 0.06769808127448997,
+      "grad_norm": 57.75,
+      "learning_rate": 2.2547416612164817e-05,
+      "loss": 13.2399,
+      "step": 1380
+    },
+    {
+      "epoch": 0.06818864708082685,
+      "grad_norm": 6.78125,
+      "learning_rate": 2.2710922171353827e-05,
+      "loss": 13.3551,
+      "step": 1390
+    },
+    {
+      "epoch": 0.06867921288716373,
+      "grad_norm": 4.78125,
+      "learning_rate": 2.287442773054284e-05,
+      "loss": 12.7103,
+      "step": 1400
+    },
+    {
+      "epoch": 0.06916977869350062,
+      "grad_norm": 5.09375,
+      "learning_rate": 2.3037933289731853e-05,
+      "loss": 12.9241,
+      "step": 1410
+    },
+    {
+      "epoch": 0.0696603444998375,
+      "grad_norm": 10.5625,
+      "learning_rate": 2.3201438848920864e-05,
+      "loss": 12.9225,
+      "step": 1420
+    },
+    {
+      "epoch": 0.07015091030617439,
+      "grad_norm": 5.8125,
+      "learning_rate": 2.3364944408109877e-05,
+      "loss": 12.943,
+      "step": 1430
+    },
+    {
+      "epoch": 0.07064147611251127,
+      "grad_norm": 6.75,
+      "learning_rate": 2.352844996729889e-05,
+      "loss": 14.0352,
+      "step": 1440
+    },
+    {
+      "epoch": 0.07113204191884816,
+      "grad_norm": 5.53125,
+      "learning_rate": 2.3691955526487904e-05,
+      "loss": 12.5482,
+      "step": 1450
+    },
+    {
+      "epoch": 0.07162260772518504,
+      "grad_norm": 3.625,
+      "learning_rate": 2.3855461085676914e-05,
+      "loss": 11.5144,
+      "step": 1460
+    },
+    {
+      "epoch": 0.07211317353152191,
+      "grad_norm": 164.0,
+      "learning_rate": 2.4018966644865924e-05,
+      "loss": 12.4646,
+      "step": 1470
+    },
+    {
+      "epoch": 0.0726037393378588,
+      "grad_norm": 40.0,
+      "learning_rate": 2.418247220405494e-05,
+      "loss": 12.4292,
+      "step": 1480
+    },
+    {
+      "epoch": 0.07309430514419568,
+      "grad_norm": 19.125,
+      "learning_rate": 2.434597776324395e-05,
+      "loss": 12.3798,
+      "step": 1490
+    },
+    {
+      "epoch": 0.07358487095053257,
+      "grad_norm": 229.0,
+      "learning_rate": 2.4509483322432964e-05,
+      "loss": 12.9266,
+      "step": 1500
+    },
+    {
+      "epoch": 0.07407543675686945,
+      "grad_norm": 12.0,
+      "learning_rate": 2.4672988881621974e-05,
+      "loss": 11.541,
+      "step": 1510
+    },
+    {
+      "epoch": 0.07456600256320634,
+      "grad_norm": 10.125,
+      "learning_rate": 2.483649444081099e-05,
+      "loss": 12.6567,
+      "step": 1520
+    },
+    {
+      "epoch": 0.07505656836954322,
+      "grad_norm": 4.03125,
+      "learning_rate": 2.5e-05,
+      "loss": 11.67,
+      "step": 1530
+    },
+    {
+      "epoch": 0.0755471341758801,
+      "grad_norm": 25.625,
+      "learning_rate": 2.5163505559189015e-05,
+      "loss": 11.3946,
+      "step": 1540
+    },
+    {
+      "epoch": 0.07603769998221699,
+      "grad_norm": 13.5,
+      "learning_rate": 2.5327011118378025e-05,
+      "loss": 11.414,
+      "step": 1550
+    },
+    {
+      "epoch": 0.07652826578855387,
+      "grad_norm": 4.96875,
+      "learning_rate": 2.5490516677567038e-05,
+      "loss": 11.6602,
+      "step": 1560
+    },
+    {
+      "epoch": 0.07701883159489076,
+      "grad_norm": 3.09375,
+      "learning_rate": 2.5654022236756055e-05,
+      "loss": 11.4031,
+      "step": 1570
+    },
+    {
+      "epoch": 0.07750939740122764,
+      "grad_norm": 6.96875,
+      "learning_rate": 2.581752779594506e-05,
+      "loss": 10.9844,
+      "step": 1580
+    },
+    {
+      "epoch": 0.07799996320756453,
+      "grad_norm": 3.28125,
+      "learning_rate": 2.598103335513408e-05,
+      "loss": 12.9034,
+      "step": 1590
+    },
+    {
+      "epoch": 0.07849052901390141,
+      "grad_norm": 5.59375,
+      "learning_rate": 2.6144538914323085e-05,
+      "loss": 11.7687,
+      "step": 1600
+    },
+    {
+      "epoch": 0.0789810948202383,
+      "grad_norm": 133.0,
+      "learning_rate": 2.63080444735121e-05,
+      "loss": 12.0474,
+      "step": 1610
+    },
+    {
+      "epoch": 0.07947166062657518,
+      "grad_norm": 8.1875,
+      "learning_rate": 2.6471550032701115e-05,
+      "loss": 11.8777,
+      "step": 1620
+    },
+    {
+      "epoch": 0.07996222643291206,
+      "grad_norm": 3.78125,
+      "learning_rate": 2.6635055591890122e-05,
+      "loss": 11.8737,
+      "step": 1630
+    },
+    {
+      "epoch": 0.08045279223924895,
+      "grad_norm": 3.53125,
+      "learning_rate": 2.679856115107914e-05,
+      "loss": 12.319,
+      "step": 1640
+    },
+    {
+      "epoch": 0.08094335804558583,
+      "grad_norm": 4.59375,
+      "learning_rate": 2.6962066710268152e-05,
+      "loss": 11.686,
+      "step": 1650
+    },
+    {
+      "epoch": 0.08143392385192272,
+      "grad_norm": 2.765625,
+      "learning_rate": 2.7125572269457162e-05,
+      "loss": 10.6974,
+      "step": 1660
+    },
+    {
+      "epoch": 0.08192448965825959,
+      "grad_norm": 4.40625,
+      "learning_rate": 2.7289077828646176e-05,
+      "loss": 12.4035,
+      "step": 1670
+    },
+    {
+      "epoch": 0.08241505546459647,
+      "grad_norm": 3.109375,
+      "learning_rate": 2.745258338783519e-05,
+      "loss": 11.7501,
+      "step": 1680
+    },
+    {
+      "epoch": 0.08290562127093336,
+      "grad_norm": 5.34375,
+      "learning_rate": 2.76160889470242e-05,
+      "loss": 11.4395,
+      "step": 1690
+    },
+    {
+      "epoch": 0.08339618707727024,
+      "grad_norm": 12.8125,
+      "learning_rate": 2.7779594506213213e-05,
+      "loss": 11.7575,
+      "step": 1700
+    },
+    {
+      "epoch": 0.08388675288360713,
+      "grad_norm": 7.1875,
+      "learning_rate": 2.7943100065402223e-05,
+      "loss": 11.2958,
+      "step": 1710
+    },
+    {
+      "epoch": 0.08437731868994401,
+      "grad_norm": 2.328125,
+      "learning_rate": 2.8106605624591236e-05,
+      "loss": 11.4801,
+      "step": 1720
+    },
+    {
+      "epoch": 0.0848678844962809,
+      "grad_norm": 6.5625,
+      "learning_rate": 2.827011118378025e-05,
+      "loss": 11.5054,
+      "step": 1730
+    },
+    {
+      "epoch": 0.08535845030261778,
+      "grad_norm": 2.546875,
+      "learning_rate": 2.843361674296926e-05,
+      "loss": 11.1001,
+      "step": 1740
+    },
+    {
+      "epoch": 0.08584901610895466,
+      "grad_norm": 2.359375,
+      "learning_rate": 2.8597122302158273e-05,
+      "loss": 12.427,
+      "step": 1750
+    },
+    {
+      "epoch": 0.08633958191529155,
+      "grad_norm": 7.40625,
+      "learning_rate": 2.876062786134729e-05,
+      "loss": 11.2234,
+      "step": 1760
+    },
+    {
+      "epoch": 0.08683014772162843,
+      "grad_norm": 248.0,
+      "learning_rate": 2.8924133420536297e-05,
+      "loss": 11.7346,
+      "step": 1770
+    },
+    {
+      "epoch": 0.08732071352796532,
+      "grad_norm": 10.125,
+      "learning_rate": 2.9087638979725313e-05,
+      "loss": 12.5661,
+      "step": 1780
+    },
+    {
+      "epoch": 0.0878112793343022,
+      "grad_norm": 2.484375,
+      "learning_rate": 2.9251144538914327e-05,
+      "loss": 10.1607,
+      "step": 1790
+    },
+    {
+      "epoch": 0.08830184514063909,
+      "grad_norm": 2.65625,
+      "learning_rate": 2.9414650098103337e-05,
+      "loss": 11.1169,
+      "step": 1800
+    },
+    {
+      "epoch": 0.08879241094697597,
+      "grad_norm": 2.625,
+      "learning_rate": 2.957815565729235e-05,
+      "loss": 10.876,
+      "step": 1810
+    },
+    {
+      "epoch": 0.08928297675331286,
+      "grad_norm": 3.046875,
+      "learning_rate": 2.974166121648136e-05,
+      "loss": 11.7584,
+      "step": 1820
+    },
+    {
+      "epoch": 0.08977354255964974,
+      "grad_norm": 2.296875,
+      "learning_rate": 2.9905166775670374e-05,
+      "loss": 11.4279,
+      "step": 1830
+    },
+    {
+      "epoch": 0.09026410836598663,
+      "grad_norm": 4.125,
+      "learning_rate": 3.0068672334859387e-05,
+      "loss": 10.9975,
+      "step": 1840
+    },
+    {
+      "epoch": 0.09075467417232351,
+      "grad_norm": 3.234375,
+      "learning_rate": 3.0232177894048397e-05,
+      "loss": 10.9377,
+      "step": 1850
+    },
+    {
+      "epoch": 0.09124523997866038,
+      "grad_norm": 4.34375,
+      "learning_rate": 3.039568345323741e-05,
+      "loss": 11.1918,
+      "step": 1860
+    },
+    {
+      "epoch": 0.09173580578499727,
+      "grad_norm": 2.59375,
+      "learning_rate": 3.055918901242643e-05,
+      "loss": 10.8162,
+      "step": 1870
+    },
+    {
+      "epoch": 0.09222637159133415,
+      "grad_norm": 3.84375,
+      "learning_rate": 3.0722694571615434e-05,
+      "loss": 11.1515,
+      "step": 1880
+    },
+    {
+      "epoch": 0.09271693739767103,
+      "grad_norm": 5.46875,
+      "learning_rate": 3.088620013080445e-05,
+      "loss": 11.5057,
+      "step": 1890
+    },
+    {
+      "epoch": 0.09320750320400792,
+      "grad_norm": 3.328125,
+      "learning_rate": 3.104970568999346e-05,
+      "loss": 11.1651,
+      "step": 1900
+    },
+    {
+      "epoch": 0.0936980690103448,
+      "grad_norm": 2.40625,
+      "learning_rate": 3.1213211249182475e-05,
+      "loss": 11.0284,
+      "step": 1910
+    },
+    {
+      "epoch": 0.09418863481668169,
+      "grad_norm": 3.765625,
+      "learning_rate": 3.137671680837149e-05,
+      "loss": 11.2881,
+      "step": 1920
+    },
+    {
+      "epoch": 0.09467920062301857,
+      "grad_norm": 3.03125,
+      "learning_rate": 3.15402223675605e-05,
+      "loss": 11.6501,
+      "step": 1930
+    },
+    {
+      "epoch": 0.09516976642935546,
+      "grad_norm": 2.5,
+      "learning_rate": 3.170372792674951e-05,
+      "loss": 11.2798,
+      "step": 1940
+    },
+    {
+      "epoch": 0.09566033223569234,
+      "grad_norm": 3.390625,
+      "learning_rate": 3.186723348593852e-05,
+      "loss": 11.6165,
+      "step": 1950
+    },
+    {
+      "epoch": 0.09615089804202923,
+      "grad_norm": 1.2734375,
+      "learning_rate": 3.2030739045127535e-05,
+      "loss": 10.3709,
+      "step": 1960
+    },
+    {
+      "epoch": 0.09664146384836611,
+      "grad_norm": 2.1875,
+      "learning_rate": 3.219424460431655e-05,
+      "loss": 10.8632,
+      "step": 1970
+    },
+    {
+      "epoch": 0.097132029654703,
+      "grad_norm": 7.0625,
+      "learning_rate": 3.235775016350556e-05,
+      "loss": 11.4351,
+      "step": 1980
+    },
+    {
+      "epoch": 0.09762259546103988,
+      "grad_norm": 1.46875,
+      "learning_rate": 3.252125572269457e-05,
+      "loss": 10.9888,
+      "step": 1990
+    },
+    {
+      "epoch": 0.09811316126737676,
+      "grad_norm": 1.578125,
+      "learning_rate": 3.268476128188359e-05,
+      "loss": 9.9517,
+      "step": 2000
+    },
+    {
+      "epoch": 0.09860372707371365,
+      "grad_norm": 1.2421875,
+      "learning_rate": 3.28482668410726e-05,
+      "loss": 11.0291,
+      "step": 2010
+    },
+    {
+      "epoch": 0.09909429288005053,
+      "grad_norm": 2.546875,
+      "learning_rate": 3.301177240026161e-05,
+      "loss": 10.9602,
+      "step": 2020
+    },
+    {
+      "epoch": 0.09958485868638742,
+      "grad_norm": 2.265625,
+      "learning_rate": 3.317527795945062e-05,
+      "loss": 10.4633,
+      "step": 2030
+    },
+    {
+      "epoch": 0.1000754244927243,
+      "grad_norm": 3.46875,
+      "learning_rate": 3.3338783518639636e-05,
+      "loss": 10.1375,
+      "step": 2040
+    },
+    {
+      "epoch": 0.10056599029906119,
+      "grad_norm": 1.5,
+      "learning_rate": 3.350228907782865e-05,
+      "loss": 10.5169,
+      "step": 2050
+    },
+    {
+      "epoch": 0.10105655610539806,
+      "grad_norm": 2.171875,
+      "learning_rate": 3.366579463701766e-05,
+      "loss": 10.2923,
+      "step": 2060
+    },
+    {
+      "epoch": 0.10154712191173494,
+      "grad_norm": 3.109375,
+      "learning_rate": 3.382930019620667e-05,
+      "loss": 10.584,
+      "step": 2070
+    },
+    {
+      "epoch": 0.10203768771807183,
+      "grad_norm": 1.875,
+      "learning_rate": 3.399280575539568e-05,
+      "loss": 10.6741,
+      "step": 2080
+    },
+    {
+      "epoch": 0.10252825352440871,
+      "grad_norm": 1.984375,
+      "learning_rate": 3.4156311314584696e-05,
+      "loss": 10.9653,
+      "step": 2090
+    },
+    {
+      "epoch": 0.1030188193307456,
+      "grad_norm": 2.65625,
+      "learning_rate": 3.431981687377371e-05,
+      "loss": 11.4489,
+      "step": 2100
+    },
+    {
+      "epoch": 0.10350938513708248,
+      "grad_norm": 2.390625,
+      "learning_rate": 3.448332243296272e-05,
+      "loss": 11.0648,
+      "step": 2110
+    },
+    {
+      "epoch": 0.10399995094341936,
+      "grad_norm": 3.0625,
+      "learning_rate": 3.4646827992151737e-05,
+      "loss": 10.2566,
+      "step": 2120
+    },
+    {
+      "epoch": 0.10449051674975625,
+      "grad_norm": 5.59375,
+      "learning_rate": 3.481033355134074e-05,
+      "loss": 10.9612,
+      "step": 2130
+    },
+    {
+      "epoch": 0.10498108255609313,
+      "grad_norm": 2.546875,
+      "learning_rate": 3.4973839110529763e-05,
+      "loss": 10.8974,
+      "step": 2140
+    },
+    {
+      "epoch": 0.10547164836243002,
+      "grad_norm": 1.234375,
+      "learning_rate": 3.513734466971878e-05,
+      "loss": 9.9906,
+      "step": 2150
+    },
+    {
+      "epoch": 0.1059622141687669,
+      "grad_norm": 2.75,
+      "learning_rate": 3.5300850228907784e-05,
+      "loss": 10.4103,
+      "step": 2160
+    },
+    {
+      "epoch": 0.10645277997510379,
+      "grad_norm": 74.5,
+      "learning_rate": 3.54643557880968e-05,
+      "loss": 10.8823,
+      "step": 2170
+    },
+    {
+      "epoch": 0.10694334578144067,
+      "grad_norm": 1.4375,
+      "learning_rate": 3.5627861347285804e-05,
+      "loss": 10.3752,
+      "step": 2180
+    },
+    {
+      "epoch": 0.10743391158777756,
+      "grad_norm": 1.5390625,
+      "learning_rate": 3.5791366906474824e-05,
+      "loss": 11.5171,
+      "step": 2190
+    },
+    {
+      "epoch": 0.10792447739411444,
+      "grad_norm": 1.9921875,
+      "learning_rate": 3.595487246566384e-05,
+      "loss": 11.0651,
+      "step": 2200
+    },
+    {
+      "epoch": 0.10841504320045132,
+      "grad_norm": 2.09375,
+      "learning_rate": 3.6118378024852844e-05,
+      "loss": 10.148,
+      "step": 2210
+    },
+    {
+      "epoch": 0.10890560900678821,
+      "grad_norm": 1.921875,
+      "learning_rate": 3.628188358404186e-05,
+      "loss": 10.6235,
+      "step": 2220
+    },
+    {
+      "epoch": 0.1093961748131251,
+      "grad_norm": 1.7890625,
+      "learning_rate": 3.644538914323087e-05,
+      "loss": 9.6748,
+      "step": 2230
+    },
+    {
+      "epoch": 0.10988674061946198,
+      "grad_norm": 26.25,
+      "learning_rate": 3.6608894702419884e-05,
+      "loss": 10.9376,
+      "step": 2240
+    },
+    {
+      "epoch": 0.11037730642579885,
+      "grad_norm": 2.1875,
+      "learning_rate": 3.67724002616089e-05,
+      "loss": 10.5539,
+      "step": 2250
+    },
+    {
+      "epoch": 0.11086787223213573,
+      "grad_norm": 2.671875,
+      "learning_rate": 3.693590582079791e-05,
+      "loss": 11.0931,
+      "step": 2260
+    },
+    {
+      "epoch": 0.11135843803847262,
+      "grad_norm": 4.8125,
+      "learning_rate": 3.709941137998692e-05,
+      "loss": 10.6696,
+      "step": 2270
+    },
+    {
+      "epoch": 0.1118490038448095,
+      "grad_norm": 5.4375,
+      "learning_rate": 3.726291693917594e-05,
+      "loss": 10.9604,
+      "step": 2280
+    },
+    {
+      "epoch": 0.11233956965114639,
+      "grad_norm": 3.71875,
+      "learning_rate": 3.7426422498364945e-05,
+      "loss": 10.2374,
+      "step": 2290
+    },
+    {
+      "epoch": 0.11283013545748327,
+      "grad_norm": 3.078125,
+      "learning_rate": 3.758992805755396e-05,
+      "loss": 10.9767,
+      "step": 2300
+    },
+    {
+      "epoch": 0.11332070126382016,
+      "grad_norm": 68.0,
+      "learning_rate": 3.775343361674297e-05,
+      "loss": 11.8492,
+      "step": 2310
+    },
+    {
+      "epoch": 0.11381126707015704,
+      "grad_norm": 2.703125,
+      "learning_rate": 3.791693917593198e-05,
+      "loss": 10.3107,
+      "step": 2320
+    },
+    {
+      "epoch": 0.11430183287649393,
+      "grad_norm": 1.3046875,
+      "learning_rate": 3.8080444735121e-05,
+      "loss": 10.0719,
+      "step": 2330
+    },
+    {
+      "epoch": 0.11479239868283081,
+      "grad_norm": 4.75,
+      "learning_rate": 3.824395029431001e-05,
+      "loss": 11.837,
+      "step": 2340
+    },
+    {
+      "epoch": 0.1152829644891677,
+      "grad_norm": 1.8671875,
+      "learning_rate": 3.840745585349902e-05,
+      "loss": 10.6051,
+      "step": 2350
+    },
+    {
+      "epoch": 0.11577353029550458,
+      "grad_norm": 2.09375,
+      "learning_rate": 3.857096141268803e-05,
+      "loss": 10.7695,
+      "step": 2360
+    },
+    {
+      "epoch": 0.11626409610184146,
+      "grad_norm": 2.421875,
+      "learning_rate": 3.8734466971877045e-05,
+      "loss": 10.7775,
+      "step": 2370
+    },
+    {
+      "epoch": 0.11675466190817835,
+      "grad_norm": 1.1953125,
+      "learning_rate": 3.889797253106606e-05,
+      "loss": 10.3568,
+      "step": 2380
+    },
+    {
+      "epoch": 0.11724522771451523,
+      "grad_norm": 2.375,
+      "learning_rate": 3.906147809025507e-05,
+      "loss": 10.2089,
+      "step": 2390
+    },
+    {
+      "epoch": 0.11773579352085212,
+      "grad_norm": 2.71875,
+      "learning_rate": 3.922498364944408e-05,
+      "loss": 10.4171,
+      "step": 2400
+    },
+    {
+      "epoch": 0.118226359327189,
+      "grad_norm": 1.8125,
+      "learning_rate": 3.938848920863309e-05,
+      "loss": 10.1651,
+      "step": 2410
+    },
+    {
+      "epoch": 0.11871692513352589,
+      "grad_norm": 3.734375,
+      "learning_rate": 3.9551994767822106e-05,
+      "loss": 10.1019,
+      "step": 2420
+    },
+    {
+      "epoch": 0.11920749093986277,
+      "grad_norm": 1.484375,
+      "learning_rate": 3.971550032701112e-05,
+      "loss": 10.4909,
+      "step": 2430
+    },
+    {
+      "epoch": 0.11969805674619965,
+      "grad_norm": 2.484375,
+      "learning_rate": 3.987900588620013e-05,
+      "loss": 9.6771,
+      "step": 2440
+    },
+    {
+      "epoch": 0.12018862255253653,
+      "grad_norm": 1.46875,
+      "learning_rate": 4.0042511445389146e-05,
+      "loss": 10.4348,
+      "step": 2450
+    },
+    {
+      "epoch": 0.12067918835887341,
+      "grad_norm": 3.671875,
+      "learning_rate": 4.020601700457815e-05,
+      "loss": 9.7455,
+      "step": 2460
+    },
+    {
+      "epoch": 0.1211697541652103,
+      "grad_norm": 2.34375,
+      "learning_rate": 4.036952256376717e-05,
+      "loss": 9.5623,
+      "step": 2470
+    },
+    {
+      "epoch": 0.12166031997154718,
+      "grad_norm": 2.921875,
+      "learning_rate": 4.0533028122956187e-05,
+      "loss": 10.2971,
+      "step": 2480
+    },
+    {
+      "epoch": 0.12215088577788406,
+      "grad_norm": 3.796875,
+      "learning_rate": 4.069653368214519e-05,
+      "loss": 10.3839,
+      "step": 2490
+    },
+    {
+      "epoch": 0.12264145158422095,
+      "grad_norm": 3.3125,
+      "learning_rate": 4.086003924133421e-05,
+      "loss": 10.9838,
+      "step": 2500
+    },
+    {
+      "epoch": 0.12313201739055783,
+      "grad_norm": 3.34375,
+      "learning_rate": 4.102354480052322e-05,
+      "loss": 10.1035,
+      "step": 2510
+    },
+    {
+      "epoch": 0.12362258319689472,
+      "grad_norm": 1.765625,
+      "learning_rate": 4.1187050359712234e-05,
+      "loss": 10.796,
+      "step": 2520
+    },
+    {
+      "epoch": 0.1241131490032316,
+      "grad_norm": 6.0,
+      "learning_rate": 4.135055591890125e-05,
+      "loss": 10.1963,
+      "step": 2530
+    },
+    {
+      "epoch": 0.12460371480956849,
+      "grad_norm": 2.640625,
+      "learning_rate": 4.1514061478090254e-05,
+      "loss": 10.7691,
+      "step": 2540
+    },
+    {
+      "epoch": 0.12509428061590536,
+      "grad_norm": 2.84375,
+      "learning_rate": 4.167756703727927e-05,
+      "loss": 10.3571,
+      "step": 2550
+    },
+    {
+      "epoch": 0.12558484642224224,
+      "grad_norm": 1.5078125,
+      "learning_rate": 4.184107259646828e-05,
+      "loss": 9.9979,
+      "step": 2560
+    },
+    {
+      "epoch": 0.12607541222857913,
+      "grad_norm": 3.703125,
+      "learning_rate": 4.2004578155657294e-05,
+      "loss": 10.3162,
+      "step": 2570
+    },
+    {
+      "epoch": 0.126565978034916,
+      "grad_norm": 2.140625,
+      "learning_rate": 4.216808371484631e-05,
+      "loss": 10.3054,
+      "step": 2580
+    },
+    {
+      "epoch": 0.1270565438412529,
+      "grad_norm": 2.453125,
+      "learning_rate": 4.233158927403532e-05,
+      "loss": 10.1395,
+      "step": 2590
+    },
+    {
+      "epoch": 0.12754710964758978,
+      "grad_norm": 1.9140625,
+      "learning_rate": 4.249509483322433e-05,
+      "loss": 10.7519,
+      "step": 2600
+    },
+    {
+      "epoch": 0.12803767545392666,
+      "grad_norm": 2.546875,
+      "learning_rate": 4.265860039241335e-05,
+      "loss": 10.3744,
+      "step": 2610
+    },
+    {
+      "epoch": 0.12852824126026355,
+      "grad_norm": 15.625,
+      "learning_rate": 4.282210595160236e-05,
+      "loss": 10.748,
+      "step": 2620
+    },
+    {
+      "epoch": 0.12901880706660043,
+      "grad_norm": 4.78125,
+      "learning_rate": 4.298561151079137e-05,
+      "loss": 10.0749,
+      "step": 2630
+    },
+    {
+      "epoch": 0.12950937287293732,
+      "grad_norm": 2.859375,
+      "learning_rate": 4.314911706998038e-05,
+      "loss": 10.3348,
+      "step": 2640
+    },
+    {
+      "epoch": 0.1299999386792742,
+      "grad_norm": 2.109375,
+      "learning_rate": 4.3312622629169395e-05,
+      "loss": 10.9035,
+      "step": 2650
+    },
+    {
+      "epoch": 0.1304905044856111,
+      "grad_norm": 1.8046875,
+      "learning_rate": 4.347612818835841e-05,
+      "loss": 10.4322,
+      "step": 2660
+    },
+    {
+      "epoch": 0.13098107029194797,
+      "grad_norm": 1.34375,
+      "learning_rate": 4.363963374754742e-05,
+      "loss": 10.4895,
+      "step": 2670
+    },
+    {
+      "epoch": 0.13147163609828486,
+      "grad_norm": 2.84375,
+      "learning_rate": 4.380313930673643e-05,
+      "loss": 10.1848,
+      "step": 2680
+    },
+    {
+      "epoch": 0.13196220190462174,
+      "grad_norm": 1.53125,
+      "learning_rate": 4.396664486592544e-05,
+      "loss": 9.8962,
+      "step": 2690
+    },
+    {
+      "epoch": 0.13245276771095862,
+      "grad_norm": 4.1875,
+      "learning_rate": 4.4130150425114455e-05,
+      "loss": 9.9074,
+      "step": 2700
+    },
+    {
+      "epoch": 0.1329433335172955,
+      "grad_norm": 2.578125,
+      "learning_rate": 4.429365598430347e-05,
+      "loss": 10.318,
+      "step": 2710
+    },
+    {
+      "epoch": 0.1334338993236324,
+      "grad_norm": 1.96875,
+      "learning_rate": 4.445716154349248e-05,
+      "loss": 10.3435,
+      "step": 2720
+    },
+    {
+      "epoch": 0.13392446512996928,
+      "grad_norm": 5.15625,
+      "learning_rate": 4.4620667102681495e-05,
+      "loss": 10.297,
+      "step": 2730
+    },
+    {
+      "epoch": 0.13441503093630616,
+      "grad_norm": 9.3125,
+      "learning_rate": 4.47841726618705e-05,
+      "loss": 11.1889,
+      "step": 2740
+    },
+    {
+      "epoch": 0.13490559674264305,
+      "grad_norm": 7.03125,
+      "learning_rate": 4.494767822105952e-05,
+      "loss": 10.6022,
+      "step": 2750
+    },
+    {
+      "epoch": 0.13539616254897993,
+      "grad_norm": 1.8359375,
+      "learning_rate": 4.511118378024853e-05,
+      "loss": 10.1121,
+      "step": 2760
+    },
+    {
+      "epoch": 0.13588672835531682,
+      "grad_norm": 2.828125,
+      "learning_rate": 4.527468933943754e-05,
+      "loss": 10.7248,
+      "step": 2770
+    },
+    {
+      "epoch": 0.1363772941616537,
+      "grad_norm": 3.203125,
+      "learning_rate": 4.5438194898626556e-05,
+      "loss": 11.1469,
+      "step": 2780
+    },
+    {
+      "epoch": 0.13686785996799058,
+      "grad_norm": 1.2890625,
+      "learning_rate": 4.560170045781557e-05,
+      "loss": 9.9668,
+      "step": 2790
+    },
+    {
+      "epoch": 0.13735842577432747,
+      "grad_norm": 2.4375,
+      "learning_rate": 4.576520601700458e-05,
+      "loss": 10.4493,
+      "step": 2800
+    },
+    {
+      "epoch": 0.13784899158066435,
+      "grad_norm": 1.921875,
+      "learning_rate": 4.5928711576193596e-05,
+      "loss": 10.109,
+      "step": 2810
+    },
+    {
+      "epoch": 0.13833955738700124,
+      "grad_norm": 1.546875,
+      "learning_rate": 4.60922171353826e-05,
+      "loss": 10.2227,
+      "step": 2820
+    },
+    {
+      "epoch": 0.13883012319333812,
+      "grad_norm": 3.90625,
+      "learning_rate": 4.6255722694571616e-05,
+      "loss": 10.1969,
+      "step": 2830
+    },
+    {
+      "epoch": 0.139320688999675,
+      "grad_norm": 1.828125,
+      "learning_rate": 4.641922825376063e-05,
+      "loss": 9.6371,
+      "step": 2840
+    },
+    {
+      "epoch": 0.1398112548060119,
+      "grad_norm": 2.9375,
+      "learning_rate": 4.658273381294964e-05,
+      "loss": 9.7864,
+      "step": 2850
+    },
+    {
+      "epoch": 0.14030182061234878,
+      "grad_norm": 2.1875,
+      "learning_rate": 4.674623937213866e-05,
+      "loss": 10.174,
+      "step": 2860
+    },
+    {
+      "epoch": 0.14079238641868566,
+      "grad_norm": 1.9609375,
+      "learning_rate": 4.690974493132766e-05,
+      "loss": 9.888,
+      "step": 2870
+    },
+    {
+      "epoch": 0.14128295222502255,
+      "grad_norm": 2.21875,
+      "learning_rate": 4.707325049051668e-05,
+      "loss": 10.6881,
+      "step": 2880
+    },
+    {
+      "epoch": 0.14177351803135943,
+      "grad_norm": 1.7109375,
+      "learning_rate": 4.72367560497057e-05,
+      "loss": 9.7418,
+      "step": 2890
+    },
+    {
+      "epoch": 0.14226408383769631,
+      "grad_norm": 1.2265625,
+      "learning_rate": 4.7400261608894704e-05,
+      "loss": 10.3103,
+      "step": 2900
+    },
+    {
+      "epoch": 0.1427546496440332,
+      "grad_norm": 1.65625,
+      "learning_rate": 4.756376716808372e-05,
+      "loss": 9.8839,
+      "step": 2910
+    },
+    {
+      "epoch": 0.14324521545037008,
+      "grad_norm": 2.578125,
+      "learning_rate": 4.772727272727273e-05,
+      "loss": 9.5034,
+      "step": 2920
+    },
+    {
+      "epoch": 0.14373578125670697,
+      "grad_norm": 3.921875,
+      "learning_rate": 4.789077828646174e-05,
+      "loss": 10.5557,
+      "step": 2930
+    },
+    {
+      "epoch": 0.14422634706304382,
+      "grad_norm": 2.234375,
+      "learning_rate": 4.805428384565076e-05,
+      "loss": 9.9611,
+      "step": 2940
+    },
+    {
+      "epoch": 0.1447169128693807,
+      "grad_norm": 1.6796875,
+      "learning_rate": 4.821778940483977e-05,
+      "loss": 9.7333,
+      "step": 2950
+    },
+    {
+      "epoch": 0.1452074786757176,
+      "grad_norm": 1.0703125,
+      "learning_rate": 4.838129496402878e-05,
+      "loss": 10.8158,
+      "step": 2960
+    },
+    {
+      "epoch": 0.14569804448205448,
+      "grad_norm": 11.75,
+      "learning_rate": 4.854480052321779e-05,
+      "loss": 10.7542,
+      "step": 2970
+    },
+    {
+      "epoch": 0.14618861028839136,
+      "grad_norm": 2.96875,
+      "learning_rate": 4.8708306082406804e-05,
+      "loss": 9.8826,
+      "step": 2980
+    },
+    {
+      "epoch": 0.14667917609472825,
+      "grad_norm": 3.203125,
+      "learning_rate": 4.887181164159582e-05,
+      "loss": 10.8316,
+      "step": 2990
+    },
+    {
+      "epoch": 0.14716974190106513,
+      "grad_norm": 3.84375,
+      "learning_rate": 4.903531720078483e-05,
+      "loss": 9.4779,
+      "step": 3000
+    },
+    {
+      "epoch": 0.14766030770740202,
+      "grad_norm": 1.859375,
+      "learning_rate": 4.919882275997384e-05,
+      "loss": 9.9817,
+      "step": 3010
+    },
+    {
+      "epoch": 0.1481508735137389,
+      "grad_norm": 1.8359375,
+      "learning_rate": 4.936232831916285e-05,
+      "loss": 10.2961,
+      "step": 3020
+    },
+    {
+      "epoch": 0.14864143932007579,
+      "grad_norm": 4.625,
+      "learning_rate": 4.952583387835187e-05,
+      "loss": 10.9962,
+      "step": 3030
+    },
+    {
+      "epoch": 0.14913200512641267,
+      "grad_norm": 1.859375,
+      "learning_rate": 4.968933943754088e-05,
+      "loss": 10.0193,
+      "step": 3040
+    },
+    {
+      "epoch": 0.14962257093274955,
+      "grad_norm": 4.8125,
+      "learning_rate": 4.985284499672989e-05,
+      "loss": 10.0264,
+      "step": 3050
+    },
+    {
+      "epoch": 0.15011313673908644,
+      "grad_norm": 2.40625,
+      "learning_rate": 5.00163505559189e-05,
+      "loss": 9.8969,
+      "step": 3060
+    },
+    {
+      "epoch": 0.15060370254542332,
+      "grad_norm": 1.96875,
+      "learning_rate": 5.017985611510791e-05,
+      "loss": 9.876,
+      "step": 3070
+    },
+    {
+      "epoch": 0.1510942683517602,
+      "grad_norm": 1.890625,
+      "learning_rate": 5.034336167429693e-05,
+      "loss": 10.2774,
+      "step": 3080
+    },
+    {
+      "epoch": 0.1515848341580971,
+      "grad_norm": 7.96875,
+      "learning_rate": 5.0506867233485945e-05,
+      "loss": 10.3209,
+      "step": 3090
+    },
+    {
+      "epoch": 0.15207539996443398,
+      "grad_norm": 4.125,
+      "learning_rate": 5.067037279267496e-05,
+      "loss": 9.86,
+      "step": 3100
+    },
+    {
+      "epoch": 0.15256596577077086,
+      "grad_norm": 3.53125,
+      "learning_rate": 5.083387835186396e-05,
+      "loss": 9.9964,
+      "step": 3110
+    },
+    {
+      "epoch": 0.15305653157710775,
+      "grad_norm": 1.78125,
+      "learning_rate": 5.099738391105298e-05,
+      "loss": 10.6453,
+      "step": 3120
+    },
+    {
+      "epoch": 0.15354709738344463,
+      "grad_norm": 1.4921875,
+      "learning_rate": 5.116088947024199e-05,
+      "loss": 9.8406,
+      "step": 3130
+    },
+    {
+      "epoch": 0.15403766318978152,
+      "grad_norm": 3.40625,
+      "learning_rate": 5.1324395029431006e-05,
+      "loss": 10.2586,
+      "step": 3140
+    },
+    {
+      "epoch": 0.1545282289961184,
+      "grad_norm": 1.9765625,
+      "learning_rate": 5.148790058862002e-05,
+      "loss": 9.4454,
+      "step": 3150
+    },
+    {
+      "epoch": 0.15501879480245528,
+      "grad_norm": 3.203125,
+      "learning_rate": 5.1651406147809026e-05,
+      "loss": 9.6739,
+      "step": 3160
+    },
+    {
+      "epoch": 0.15550936060879217,
+      "grad_norm": 10.375,
+      "learning_rate": 5.181491170699804e-05,
+      "loss": 10.6741,
+      "step": 3170
+    },
+    {
+      "epoch": 0.15599992641512905,
+      "grad_norm": 4.6875,
+      "learning_rate": 5.197841726618705e-05,
+      "loss": 9.6606,
+      "step": 3180
+    },
+    {
+      "epoch": 0.15649049222146594,
+      "grad_norm": 3.75,
+      "learning_rate": 5.2141922825376066e-05,
+      "loss": 10.0526,
+      "step": 3190
+    },
+    {
+      "epoch": 0.15698105802780282,
+      "grad_norm": 2.359375,
+      "learning_rate": 5.230542838456508e-05,
+      "loss": 10.6605,
+      "step": 3200
+    },
+    {
+      "epoch": 0.1574716238341397,
+      "grad_norm": 1.859375,
+      "learning_rate": 5.246893394375409e-05,
+      "loss": 10.2658,
+      "step": 3210
+    },
+    {
+      "epoch": 0.1579621896404766,
+      "grad_norm": 7.5,
+      "learning_rate": 5.26324395029431e-05,
+      "loss": 9.532,
+      "step": 3220
+    },
+    {
+      "epoch": 0.15845275544681348,
+      "grad_norm": 1.4296875,
+      "learning_rate": 5.279594506213211e-05,
+      "loss": 10.1919,
+      "step": 3230
+    },
+    {
+      "epoch": 0.15894332125315036,
+      "grad_norm": 2.171875,
+      "learning_rate": 5.295945062132113e-05,
+      "loss": 9.6622,
+      "step": 3240
+    },
+    {
+      "epoch": 0.15943388705948724,
+      "grad_norm": 3.375,
+      "learning_rate": 5.312295618051014e-05,
+      "loss": 10.4818,
+      "step": 3250
+    },
+    {
+      "epoch": 0.15992445286582413,
+      "grad_norm": 4.40625,
+      "learning_rate": 5.3286461739699154e-05,
+      "loss": 10.8637,
+      "step": 3260
+    },
+    {
+      "epoch": 0.160415018672161,
+      "grad_norm": 2.09375,
+      "learning_rate": 5.344996729888816e-05,
+      "loss": 10.2733,
+      "step": 3270
+    },
+    {
+      "epoch": 0.1609055844784979,
+      "grad_norm": 1.8984375,
+      "learning_rate": 5.3613472858077174e-05,
+      "loss": 10.1647,
+      "step": 3280
+    },
+    {
+      "epoch": 0.16139615028483478,
+      "grad_norm": 1.9921875,
+      "learning_rate": 5.377697841726619e-05,
+      "loss": 9.9118,
+      "step": 3290
+    },
+    {
+      "epoch": 0.16188671609117167,
+      "grad_norm": 24.75,
+      "learning_rate": 5.39404839764552e-05,
+      "loss": 10.5719,
+      "step": 3300
+    },
+    {
+      "epoch": 0.16237728189750855,
+      "grad_norm": 1.25,
+      "learning_rate": 5.4103989535644214e-05,
+      "loss": 10.1434,
+      "step": 3310
+    },
+    {
+      "epoch": 0.16286784770384544,
+      "grad_norm": 1.7890625,
+      "learning_rate": 5.4267495094833234e-05,
+      "loss": 9.9758,
+      "step": 3320
+    },
+    {
+      "epoch": 0.1633584135101823,
+      "grad_norm": 1.34375,
+      "learning_rate": 5.4431000654022234e-05,
+      "loss": 10.4058,
+      "step": 3330
+    },
+    {
+      "epoch": 0.16384897931651918,
+      "grad_norm": 3.484375,
+      "learning_rate": 5.459450621321125e-05,
+      "loss": 10.481,
+      "step": 3340
+    },
+    {
+      "epoch": 0.16433954512285606,
+      "grad_norm": 1.546875,
+      "learning_rate": 5.475801177240026e-05,
+      "loss": 10.4166,
+      "step": 3350
+    },
+    {
+      "epoch": 0.16483011092919295,
+      "grad_norm": 1.453125,
+      "learning_rate": 5.492151733158928e-05,
+      "loss": 9.8992,
+      "step": 3360
+    },
+    {
+      "epoch": 0.16532067673552983,
+      "grad_norm": 2.25,
+      "learning_rate": 5.5085022890778295e-05,
+      "loss": 11.0622,
+      "step": 3370
+    },
+    {
+      "epoch": 0.16581124254186672,
+      "grad_norm": 2.21875,
+      "learning_rate": 5.5248528449967295e-05,
+      "loss": 9.4652,
+      "step": 3380
+    },
+    {
+      "epoch": 0.1663018083482036,
+      "grad_norm": 3.796875,
+      "learning_rate": 5.541203400915631e-05,
+      "loss": 10.0684,
+      "step": 3390
+    },
+    {
+      "epoch": 0.16679237415454048,
+      "grad_norm": 25.625,
+      "learning_rate": 5.557553956834533e-05,
+      "loss": 10.1285,
+      "step": 3400
+    },
+    {
+      "epoch": 0.16728293996087737,
+      "grad_norm": 2.6875,
+      "learning_rate": 5.573904512753434e-05,
+      "loss": 9.3848,
+      "step": 3410
+    },
+    {
+      "epoch": 0.16777350576721425,
+      "grad_norm": 1.953125,
+      "learning_rate": 5.5902550686723355e-05,
+      "loss": 10.651,
+      "step": 3420
+    },
+    {
+      "epoch": 0.16826407157355114,
+      "grad_norm": 3.21875,
+      "learning_rate": 5.606605624591237e-05,
+      "loss": 10.7245,
+      "step": 3430
+    },
+    {
+      "epoch": 0.16875463737988802,
+      "grad_norm": 2.40625,
+      "learning_rate": 5.622956180510137e-05,
+      "loss": 10.3454,
+      "step": 3440
+    },
+    {
+      "epoch": 0.1692452031862249,
+      "grad_norm": 2.265625,
+      "learning_rate": 5.639306736429039e-05,
+      "loss": 10.0013,
+      "step": 3450
+    },
+    {
+      "epoch": 0.1697357689925618,
+      "grad_norm": 2.140625,
+      "learning_rate": 5.65565729234794e-05,
+      "loss": 9.8554,
+      "step": 3460
+    },
+    {
+      "epoch": 0.17022633479889868,
+      "grad_norm": 2.296875,
+      "learning_rate": 5.6720078482668415e-05,
+      "loss": 9.5814,
+      "step": 3470
+    },
+    {
+      "epoch": 0.17071690060523556,
+      "grad_norm": 1.6796875,
+      "learning_rate": 5.688358404185743e-05,
+      "loss": 9.7779,
+      "step": 3480
+    },
+    {
+      "epoch": 0.17120746641157245,
+      "grad_norm": 2.4375,
+      "learning_rate": 5.7047089601046436e-05,
+      "loss": 9.8782,
+      "step": 3490
+    },
+    {
+      "epoch": 0.17169803221790933,
+      "grad_norm": 1.6640625,
+      "learning_rate": 5.721059516023545e-05,
+      "loss": 10.2612,
+      "step": 3500
+    },
+    {
+      "epoch": 0.17218859802424621,
+      "grad_norm": 2.1875,
+      "learning_rate": 5.737410071942446e-05,
+      "loss": 9.6275,
+      "step": 3510
+    },
+    {
+      "epoch": 0.1726791638305831,
+      "grad_norm": 2.515625,
+      "learning_rate": 5.7537606278613476e-05,
+      "loss": 10.2306,
+      "step": 3520
+    },
+    {
+      "epoch": 0.17316972963691998,
+      "grad_norm": 2.0,
+      "learning_rate": 5.770111183780249e-05,
+      "loss": 10.1461,
+      "step": 3530
+    },
+    {
+      "epoch": 0.17366029544325687,
+      "grad_norm": 1.5234375,
+      "learning_rate": 5.78646173969915e-05,
+      "loss": 9.5698,
+      "step": 3540
+    },
+    {
+      "epoch": 0.17415086124959375,
+      "grad_norm": 1.546875,
+      "learning_rate": 5.802812295618051e-05,
+      "loss": 10.0161,
+      "step": 3550
+    },
+    {
+      "epoch": 0.17464142705593064,
+      "grad_norm": 1.1796875,
+      "learning_rate": 5.819162851536952e-05,
+      "loss": 10.4343,
+      "step": 3560
+    },
+    {
+      "epoch": 0.17513199286226752,
+      "grad_norm": 1.9375,
+      "learning_rate": 5.8355134074558536e-05,
+      "loss": 10.5189,
+      "step": 3570
+    },
+    {
+      "epoch": 0.1756225586686044,
+      "grad_norm": 1.7890625,
+      "learning_rate": 5.851863963374755e-05,
+      "loss": 9.1473,
+      "step": 3580
+    },
+    {
+      "epoch": 0.1761131244749413,
+      "grad_norm": 1.9453125,
+      "learning_rate": 5.868214519293656e-05,
+      "loss": 10.1085,
+      "step": 3590
+    },
+    {
+      "epoch": 0.17660369028127818,
+      "grad_norm": 2.21875,
+      "learning_rate": 5.884565075212557e-05,
+      "loss": 9.6962,
+      "step": 3600
+    },
+    {
+      "epoch": 0.17709425608761506,
+      "grad_norm": 2.265625,
+      "learning_rate": 5.900915631131458e-05,
+      "loss": 10.2332,
+      "step": 3610
+    },
+    {
+      "epoch": 0.17758482189395194,
+      "grad_norm": 1.34375,
+      "learning_rate": 5.91726618705036e-05,
+      "loss": 9.834,
+      "step": 3620
+    },
+    {
+      "epoch": 0.17807538770028883,
+      "grad_norm": 1.625,
+      "learning_rate": 5.933616742969261e-05,
+      "loss": 9.6503,
+      "step": 3630
+    },
+    {
+      "epoch": 0.1785659535066257,
+      "grad_norm": 1.375,
+      "learning_rate": 5.949967298888163e-05,
+      "loss": 9.9655,
+      "step": 3640
+    },
+    {
+      "epoch": 0.1790565193129626,
+      "grad_norm": 3.59375,
+      "learning_rate": 5.9663178548070644e-05,
+      "loss": 10.0095,
+      "step": 3650
+    },
+    {
+      "epoch": 0.17954708511929948,
+      "grad_norm": 1.8984375,
+      "learning_rate": 5.9826684107259644e-05,
+      "loss": 9.9373,
+      "step": 3660
+    },
+    {
+      "epoch": 0.18003765092563637,
+      "grad_norm": 2.640625,
+      "learning_rate": 5.999018966644866e-05,
+      "loss": 9.9908,
+      "step": 3670
+    },
+    {
+      "epoch": 0.18052821673197325,
+      "grad_norm": 3.0625,
+      "learning_rate": 6.015369522563767e-05,
+      "loss": 9.8641,
+      "step": 3680
+    },
+    {
+      "epoch": 0.18101878253831014,
+      "grad_norm": 1.8046875,
+      "learning_rate": 6.031720078482669e-05,
+      "loss": 9.8284,
+      "step": 3690
+    },
+    {
+      "epoch": 0.18150934834464702,
+      "grad_norm": 3.96875,
+      "learning_rate": 6.0480706344015704e-05,
+      "loss": 9.807,
+      "step": 3700
+    },
+    {
+      "epoch": 0.1819999141509839,
+      "grad_norm": 1.3125,
+      "learning_rate": 6.064421190320472e-05,
+      "loss": 9.0827,
+      "step": 3710
+    },
+    {
+      "epoch": 0.18249047995732076,
+      "grad_norm": 1.4375,
+      "learning_rate": 6.080771746239372e-05,
+      "loss": 9.5714,
+      "step": 3720
+    },
+    {
+      "epoch": 0.18298104576365765,
+      "grad_norm": 5.09375,
+      "learning_rate": 6.097122302158274e-05,
+      "loss": 10.1513,
+      "step": 3730
+    },
+    {
+      "epoch": 0.18347161156999453,
+      "grad_norm": 1.4765625,
+      "learning_rate": 6.113472858077175e-05,
+      "loss": 9.7627,
+      "step": 3740
+    },
+    {
+      "epoch": 0.18396217737633141,
+      "grad_norm": 5.0,
+      "learning_rate": 6.129823413996076e-05,
+      "loss": 10.0912,
+      "step": 3750
+    },
+    {
+      "epoch": 0.1844527431826683,
+      "grad_norm": 2.984375,
+      "learning_rate": 6.146173969914978e-05,
+      "loss": 9.4106,
+      "step": 3760
+    },
+    {
+      "epoch": 0.18494330898900518,
+      "grad_norm": 1.5625,
+      "learning_rate": 6.162524525833878e-05,
+      "loss": 9.6704,
+      "step": 3770
+    },
+    {
+      "epoch": 0.18543387479534207,
+      "grad_norm": 1.5703125,
+      "learning_rate": 6.178875081752779e-05,
+      "loss": 9.7668,
+      "step": 3780
+    },
+    {
+      "epoch": 0.18592444060167895,
+      "grad_norm": 3.25,
+      "learning_rate": 6.19522563767168e-05,
+      "loss": 10.349,
+      "step": 3790
+    },
+    {
+      "epoch": 0.18641500640801584,
+      "grad_norm": 1.1953125,
+      "learning_rate": 6.211576193590583e-05,
+      "loss": 10.3616,
+      "step": 3800
+    },
+    {
+      "epoch": 0.18690557221435272,
+      "grad_norm": 0.78125,
+      "learning_rate": 6.227926749509485e-05,
+      "loss": 9.7588,
+      "step": 3810
+    },
+    {
+      "epoch": 0.1873961380206896,
+      "grad_norm": 2.296875,
+      "learning_rate": 6.244277305428386e-05,
+      "loss": 9.8407,
+      "step": 3820
+    },
+    {
+      "epoch": 0.1878867038270265,
+      "grad_norm": 1.90625,
+      "learning_rate": 6.260627861347286e-05,
+      "loss": 9.8291,
+      "step": 3830
+    },
+    {
+      "epoch": 0.18837726963336338,
+      "grad_norm": 1.7578125,
+      "learning_rate": 6.276978417266187e-05,
+      "loss": 9.4544,
+      "step": 3840
+    },
+    {
+      "epoch": 0.18886783543970026,
+      "grad_norm": 3.046875,
+      "learning_rate": 6.293328973185089e-05,
+      "loss": 9.7936,
+      "step": 3850
+    },
+    {
+      "epoch": 0.18935840124603714,
+      "grad_norm": 1.2109375,
+      "learning_rate": 6.30967952910399e-05,
+      "loss": 9.5762,
+      "step": 3860
+    },
+    {
+      "epoch": 0.18984896705237403,
+      "grad_norm": 1.2578125,
+      "learning_rate": 6.326030085022891e-05,
+      "loss": 9.6703,
+      "step": 3870
+    },
+    {
+      "epoch": 0.1903395328587109,
+      "grad_norm": 1.734375,
+      "learning_rate": 6.342380640941791e-05,
+      "loss": 9.7873,
+      "step": 3880
+    },
+    {
+      "epoch": 0.1908300986650478,
+      "grad_norm": 1.359375,
+      "learning_rate": 6.358731196860694e-05,
+      "loss": 10.6224,
+      "step": 3890
+    },
+    {
+      "epoch": 0.19132066447138468,
+      "grad_norm": 1.890625,
+      "learning_rate": 6.375081752779595e-05,
+      "loss": 9.1861,
+      "step": 3900
+    },
+    {
+      "epoch": 0.19181123027772157,
+      "grad_norm": 1.1640625,
+      "learning_rate": 6.391432308698497e-05,
+      "loss": 9.3092,
+      "step": 3910
+    },
+    {
+      "epoch": 0.19230179608405845,
+      "grad_norm": 1.890625,
+      "learning_rate": 6.407782864617398e-05,
+      "loss": 9.6731,
+      "step": 3920
+    },
+    {
+      "epoch": 0.19279236189039534,
+      "grad_norm": 4.375,
+      "learning_rate": 6.424133420536299e-05,
+      "loss": 10.9559,
+      "step": 3930
+    },
+    {
+      "epoch": 0.19328292769673222,
+      "grad_norm": 1.9453125,
+      "learning_rate": 6.440483976455199e-05,
+      "loss": 10.1029,
+      "step": 3940
+    },
+    {
+      "epoch": 0.1937734935030691,
+      "grad_norm": 1.8984375,
+      "learning_rate": 6.4568345323741e-05,
+      "loss": 9.7778,
+      "step": 3950
+    },
+    {
+      "epoch": 0.194264059309406,
+      "grad_norm": 2.484375,
+      "learning_rate": 6.473185088293002e-05,
+      "loss": 9.7748,
+      "step": 3960
+    },
+    {
+      "epoch": 0.19475462511574287,
+      "grad_norm": 4.4375,
+      "learning_rate": 6.489535644211903e-05,
+      "loss": 10.2811,
+      "step": 3970
+    },
+    {
+      "epoch": 0.19524519092207976,
+      "grad_norm": 1.28125,
+      "learning_rate": 6.505886200130805e-05,
+      "loss": 9.9864,
+      "step": 3980
+    },
+    {
+      "epoch": 0.19573575672841664,
+      "grad_norm": 1.453125,
+      "learning_rate": 6.522236756049706e-05,
+      "loss": 9.8218,
+      "step": 3990
+    },
+    {
+      "epoch": 0.19622632253475353,
+      "grad_norm": 1.3125,
+      "learning_rate": 6.538587311968607e-05,
+      "loss": 9.7552,
+      "step": 4000
+    },
+    {
+      "epoch": 0.1967168883410904,
+      "grad_norm": 1.5,
+      "learning_rate": 6.554937867887509e-05,
+      "loss": 9.769,
+      "step": 4010
+    },
+    {
+      "epoch": 0.1972074541474273,
+      "grad_norm": 1.6015625,
+      "learning_rate": 6.57128842380641e-05,
+      "loss": 9.4737,
+      "step": 4020
+    },
+    {
+      "epoch": 0.19769801995376418,
+      "grad_norm": 2.03125,
+      "learning_rate": 6.587638979725311e-05,
+      "loss": 9.6332,
+      "step": 4030
+    },
+    {
+      "epoch": 0.19818858576010107,
+      "grad_norm": 1.8203125,
+      "learning_rate": 6.603989535644213e-05,
+      "loss": 9.7534,
+      "step": 4040
+    },
+    {
+      "epoch": 0.19867915156643795,
+      "grad_norm": 3.015625,
+      "learning_rate": 6.620340091563113e-05,
+      "loss": 9.9509,
+      "step": 4050
+    },
+    {
+      "epoch": 0.19916971737277483,
+      "grad_norm": 2.4375,
+      "learning_rate": 6.636690647482014e-05,
+      "loss": 9.6021,
+      "step": 4060
+    },
+    {
+      "epoch": 0.19966028317911172,
+      "grad_norm": 1.1875,
+      "learning_rate": 6.653041203400915e-05,
+      "loss": 10.1229,
+      "step": 4070
+    },
+    {
+      "epoch": 0.2001508489854486,
+      "grad_norm": 1.703125,
+      "learning_rate": 6.669391759319817e-05,
+      "loss": 9.6246,
+      "step": 4080
+    },
+    {
+      "epoch": 0.2006414147917855,
+      "grad_norm": 2.140625,
+      "learning_rate": 6.68574231523872e-05,
+      "loss": 9.8894,
+      "step": 4090
+    },
+    {
+      "epoch": 0.20113198059812237,
+      "grad_norm": 2.390625,
+      "learning_rate": 6.70209287115762e-05,
+      "loss": 9.8453,
+      "step": 4100
+    },
+    {
+      "epoch": 0.20162254640445923,
+      "grad_norm": 2.015625,
+      "learning_rate": 6.718443427076521e-05,
+      "loss": 9.7062,
+      "step": 4110
+    },
+    {
+      "epoch": 0.20211311221079611,
+      "grad_norm": 1.109375,
+      "learning_rate": 6.734793982995422e-05,
+      "loss": 9.5229,
+      "step": 4120
+    },
+    {
+      "epoch": 0.202603678017133,
+      "grad_norm": 3.203125,
+      "learning_rate": 6.751144538914323e-05,
+      "loss": 9.4862,
+      "step": 4130
+    },
+    {
+      "epoch": 0.20309424382346988,
+      "grad_norm": 2.03125,
+      "learning_rate": 6.767495094833225e-05,
+      "loss": 9.6983,
+      "step": 4140
+    },
+    {
+      "epoch": 0.20358480962980677,
+      "grad_norm": 1.7734375,
+      "learning_rate": 6.783845650752126e-05,
+      "loss": 9.6236,
+      "step": 4150
+    },
+    {
+      "epoch": 0.20407537543614365,
+      "grad_norm": 1.625,
+      "learning_rate": 6.800196206671026e-05,
+      "loss": 10.7213,
+      "step": 4160
+    },
+    {
+      "epoch": 0.20456594124248054,
+      "grad_norm": 2.640625,
+      "learning_rate": 6.816546762589929e-05,
+      "loss": 8.9305,
+      "step": 4170
+    },
+    {
+      "epoch": 0.20505650704881742,
+      "grad_norm": 2.015625,
+      "learning_rate": 6.83289731850883e-05,
+      "loss": 10.5615,
+      "step": 4180
+    },
+    {
+      "epoch": 0.2055470728551543,
+      "grad_norm": 2.8125,
+      "learning_rate": 6.849247874427732e-05,
+      "loss": 10.1414,
+      "step": 4190
+    },
+    {
+      "epoch": 0.2060376386614912,
+      "grad_norm": 2.0625,
+      "learning_rate": 6.865598430346633e-05,
+      "loss": 9.8064,
+      "step": 4200
+    },
+    {
+      "epoch": 0.20652820446782807,
+      "grad_norm": 1.0859375,
+      "learning_rate": 6.881948986265533e-05,
+      "loss": 9.9652,
+      "step": 4210
+    },
+    {
+      "epoch": 0.20701877027416496,
+      "grad_norm": 2.21875,
+      "learning_rate": 6.898299542184434e-05,
+      "loss": 9.9365,
+      "step": 4220
+    },
+    {
+      "epoch": 0.20750933608050184,
+      "grad_norm": 1.7109375,
+      "learning_rate": 6.914650098103336e-05,
+      "loss": 9.7281,
+      "step": 4230
+    },
+    {
+      "epoch": 0.20799990188683873,
+      "grad_norm": 2.75,
+      "learning_rate": 6.931000654022237e-05,
+      "loss": 9.6273,
+      "step": 4240
+    },
+    {
+      "epoch": 0.2084904676931756,
+      "grad_norm": 1.640625,
+      "learning_rate": 6.947351209941138e-05,
+      "loss": 10.1488,
+      "step": 4250
+    },
+    {
+      "epoch": 0.2089810334995125,
+      "grad_norm": 1.71875,
+      "learning_rate": 6.96370176586004e-05,
+      "loss": 9.1238,
+      "step": 4260
+    },
+    {
+      "epoch": 0.20947159930584938,
+      "grad_norm": 1.859375,
+      "learning_rate": 6.980052321778941e-05,
+      "loss": 9.6721,
+      "step": 4270
+    },
+    {
+      "epoch": 0.20996216511218627,
+      "grad_norm": 1.921875,
+      "learning_rate": 6.996402877697842e-05,
+      "loss": 9.5141,
+      "step": 4280
+    },
+    {
+      "epoch": 0.21045273091852315,
+      "grad_norm": 1.59375,
+      "learning_rate": 7.012753433616744e-05,
+      "loss": 9.595,
+      "step": 4290
+    },
+    {
+      "epoch": 0.21094329672486004,
+      "grad_norm": 1.59375,
+      "learning_rate": 7.029103989535645e-05,
+      "loss": 10.3595,
+      "step": 4300
+    },
+    {
+      "epoch": 0.21143386253119692,
+      "grad_norm": 2.390625,
+      "learning_rate": 7.045454545454546e-05,
+      "loss": 9.3384,
+      "step": 4310
+    },
+    {
+      "epoch": 0.2119244283375338,
+      "grad_norm": 1.2265625,
+      "learning_rate": 7.061805101373446e-05,
+      "loss": 9.2934,
+      "step": 4320
+    },
+    {
+      "epoch": 0.2124149941438707,
+      "grad_norm": 1.1796875,
+      "learning_rate": 7.078155657292348e-05,
+      "loss": 10.1621,
+      "step": 4330
+    },
+    {
+      "epoch": 0.21290555995020757,
+      "grad_norm": 2.609375,
+      "learning_rate": 7.094506213211249e-05,
+      "loss": 9.8211,
+      "step": 4340
+    },
+    {
+      "epoch": 0.21339612575654446,
+      "grad_norm": 1.1171875,
+      "learning_rate": 7.11085676913015e-05,
+      "loss": 9.5687,
+      "step": 4350
+    },
+    {
+      "epoch": 0.21388669156288134,
+      "grad_norm": 1.6171875,
+      "learning_rate": 7.127207325049052e-05,
+      "loss": 9.4998,
+      "step": 4360
+    },
+    {
+      "epoch": 0.21437725736921823,
+      "grad_norm": 2.859375,
+      "learning_rate": 7.143557880967954e-05,
+      "loss": 10.0604,
+      "step": 4370
+    },
+    {
+      "epoch": 0.2148678231755551,
+      "grad_norm": 2.25,
+      "learning_rate": 7.159908436886854e-05,
+      "loss": 9.5754,
+      "step": 4380
+    },
+    {
+      "epoch": 0.215358388981892,
+      "grad_norm": 4.40625,
+      "learning_rate": 7.176258992805756e-05,
+      "loss": 10.112,
+      "step": 4390
+    },
+    {
+      "epoch": 0.21584895478822888,
+      "grad_norm": 2.0,
+      "learning_rate": 7.192609548724657e-05,
+      "loss": 10.0719,
+      "step": 4400
+    },
+    {
+      "epoch": 0.21633952059456577,
+      "grad_norm": 2.6875,
+      "learning_rate": 7.208960104643558e-05,
+      "loss": 10.1256,
+      "step": 4410
+    },
+    {
+      "epoch": 0.21683008640090265,
+      "grad_norm": 1.078125,
+      "learning_rate": 7.22531066056246e-05,
+      "loss": 9.739,
+      "step": 4420
+    },
+    {
+      "epoch": 0.21732065220723953,
+      "grad_norm": 1.1875,
+      "learning_rate": 7.24166121648136e-05,
+      "loss": 9.6756,
+      "step": 4430
+    },
+    {
+      "epoch": 0.21781121801357642,
+      "grad_norm": 1.4296875,
+      "learning_rate": 7.258011772400261e-05,
+      "loss": 9.6539,
+      "step": 4440
+    },
+    {
+      "epoch": 0.2183017838199133,
+      "grad_norm": 2.59375,
+      "learning_rate": 7.274362328319162e-05,
+      "loss": 10.0042,
+      "step": 4450
+    },
+    {
+      "epoch": 0.2187923496262502,
+      "grad_norm": 2.3125,
+      "learning_rate": 7.290712884238065e-05,
+      "loss": 10.0242,
+      "step": 4460
+    },
+    {
+      "epoch": 0.21928291543258707,
+      "grad_norm": 3.890625,
+      "learning_rate": 7.307063440156966e-05,
+      "loss": 10.1849,
+      "step": 4470
+    },
+    {
+      "epoch": 0.21977348123892396,
+      "grad_norm": 2.90625,
+      "learning_rate": 7.323413996075868e-05,
+      "loss": 10.4992,
+      "step": 4480
+    },
+    {
+      "epoch": 0.22026404704526084,
+      "grad_norm": 2.125,
+      "learning_rate": 7.339764551994768e-05,
+      "loss": 10.2533,
+      "step": 4490
+    },
+    {
+      "epoch": 0.2207546128515977,
+      "grad_norm": 1.1171875,
+      "learning_rate": 7.356115107913669e-05,
+      "loss": 9.4999,
+      "step": 4500
+    },
+    {
+      "epoch": 0.22124517865793458,
+      "grad_norm": 2.4375,
+      "learning_rate": 7.37246566383257e-05,
+      "loss": 10.0668,
+      "step": 4510
+    },
+    {
+      "epoch": 0.22173574446427147,
+      "grad_norm": 3.609375,
+      "learning_rate": 7.388816219751472e-05,
+      "loss": 10.6657,
+      "step": 4520
+    },
+    {
+      "epoch": 0.22222631027060835,
+      "grad_norm": 1.671875,
+      "learning_rate": 7.405166775670373e-05,
+      "loss": 9.792,
+      "step": 4530
+    },
+    {
+      "epoch": 0.22271687607694524,
+      "grad_norm": 2.390625,
+      "learning_rate": 7.421517331589275e-05,
+      "loss": 10.2273,
+      "step": 4540
+    },
+    {
+      "epoch": 0.22320744188328212,
+      "grad_norm": 1.5625,
+      "learning_rate": 7.437867887508176e-05,
+      "loss": 9.523,
+      "step": 4550
+    },
+    {
+      "epoch": 0.223698007689619,
+      "grad_norm": 2.0,
+      "learning_rate": 7.454218443427077e-05,
+      "loss": 9.5333,
+      "step": 4560
+    },
+    {
+      "epoch": 0.2241885734959559,
+      "grad_norm": 1.6484375,
+      "learning_rate": 7.470568999345979e-05,
+      "loss": 9.6024,
+      "step": 4570
+    },
+    {
+      "epoch": 0.22467913930229277,
+      "grad_norm": 1.625,
+      "learning_rate": 7.48691955526488e-05,
+      "loss": 10.0717,
+      "step": 4580
+    },
+    {
+      "epoch": 0.22516970510862966,
+      "grad_norm": 24.0,
+      "learning_rate": 7.503270111183781e-05,
+      "loss": 9.9712,
+      "step": 4590
+    },
+    {
+      "epoch": 0.22566027091496654,
+      "grad_norm": 3.0625,
+      "learning_rate": 7.519620667102681e-05,
+      "loss": 9.6542,
+      "step": 4600
+    },
+    {
+      "epoch": 0.22615083672130343,
+      "grad_norm": 1.7265625,
+      "learning_rate": 7.535971223021583e-05,
+      "loss": 9.5918,
+      "step": 4610
+    },
+    {
+      "epoch": 0.2266414025276403,
+      "grad_norm": 2.15625,
+      "learning_rate": 7.552321778940484e-05,
+      "loss": 9.8799,
+      "step": 4620
+    },
+    {
+      "epoch": 0.2271319683339772,
+      "grad_norm": 1.2578125,
+      "learning_rate": 7.568672334859385e-05,
+      "loss": 9.9606,
+      "step": 4630
+    },
+    {
+      "epoch": 0.22762253414031408,
+      "grad_norm": 1.703125,
+      "learning_rate": 7.585022890778287e-05,
+      "loss": 9.1418,
+      "step": 4640
+    },
+    {
+      "epoch": 0.22811309994665097,
+      "grad_norm": 2.859375,
+      "learning_rate": 7.601373446697188e-05,
+      "loss": 10.0271,
+      "step": 4650
+    },
+    {
+      "epoch": 0.22860366575298785,
+      "grad_norm": 2.234375,
+      "learning_rate": 7.617724002616089e-05,
+      "loss": 10.0253,
+      "step": 4660
+    },
+    {
+      "epoch": 0.22909423155932473,
+      "grad_norm": 2.609375,
+      "learning_rate": 7.63407455853499e-05,
+      "loss": 10.1575,
+      "step": 4670
+    },
+    {
+      "epoch": 0.22958479736566162,
+      "grad_norm": 1.6796875,
+      "learning_rate": 7.650425114453892e-05,
+      "loss": 10.0217,
+      "step": 4680
+    },
+    {
+      "epoch": 0.2300753631719985,
+      "grad_norm": 3.265625,
+      "learning_rate": 7.666775670372793e-05,
+      "loss": 9.1447,
+      "step": 4690
+    },
+    {
+      "epoch": 0.2305659289783354,
+      "grad_norm": 1.8359375,
+      "learning_rate": 7.683126226291695e-05,
+      "loss": 9.787,
+      "step": 4700
+    },
+    {
+      "epoch": 0.23105649478467227,
+      "grad_norm": 2.359375,
+      "learning_rate": 7.699476782210595e-05,
+      "loss": 9.4155,
+      "step": 4710
+    },
+    {
+      "epoch": 0.23154706059100916,
+      "grad_norm": 1.6015625,
+      "learning_rate": 7.715827338129496e-05,
+      "loss": 10.074,
+      "step": 4720
+    },
+    {
+      "epoch": 0.23203762639734604,
+      "grad_norm": 2.375,
+      "learning_rate": 7.732177894048397e-05,
+      "loss": 10.0346,
+      "step": 4730
+    },
+    {
+      "epoch": 0.23252819220368293,
+      "grad_norm": 1.828125,
+      "learning_rate": 7.7485284499673e-05,
+      "loss": 10.0144,
+      "step": 4740
+    },
+    {
+      "epoch": 0.2330187580100198,
+      "grad_norm": 1.8828125,
+      "learning_rate": 7.764879005886201e-05,
+      "loss": 10.3287,
+      "step": 4750
+    },
+    {
+      "epoch": 0.2335093238163567,
+      "grad_norm": 2.015625,
+      "learning_rate": 7.781229561805101e-05,
+      "loss": 10.148,
+      "step": 4760
+    },
+    {
+      "epoch": 0.23399988962269358,
+      "grad_norm": 3.421875,
+      "learning_rate": 7.797580117724003e-05,
+      "loss": 10.0125,
+      "step": 4770
+    },
+    {
+      "epoch": 0.23449045542903046,
+      "grad_norm": 7.46875,
+      "learning_rate": 7.813930673642904e-05,
+      "loss": 9.1771,
+      "step": 4780
+    },
+    {
+      "epoch": 0.23498102123536735,
+      "grad_norm": 3.515625,
+      "learning_rate": 7.830281229561805e-05,
+      "loss": 9.8328,
+      "step": 4790
+    },
+    {
+      "epoch": 0.23547158704170423,
+      "grad_norm": 0.90625,
+      "learning_rate": 7.846631785480707e-05,
+      "loss": 9.3142,
+      "step": 4800
+    },
+    {
+      "epoch": 0.23596215284804112,
+      "grad_norm": 2.15625,
+      "learning_rate": 7.862982341399608e-05,
+      "loss": 9.6561,
+      "step": 4810
+    },
+    {
+      "epoch": 0.236452718654378,
+      "grad_norm": 1.09375,
+      "learning_rate": 7.87933289731851e-05,
+      "loss": 9.6921,
+      "step": 4820
+    },
+    {
+      "epoch": 0.2369432844607149,
+      "grad_norm": 2.625,
+      "learning_rate": 7.895683453237411e-05,
+      "loss": 10.3582,
+      "step": 4830
+    },
+    {
+      "epoch": 0.23743385026705177,
+      "grad_norm": 1.7265625,
+      "learning_rate": 7.912034009156312e-05,
+      "loss": 10.134,
+      "step": 4840
+    },
+    {
+      "epoch": 0.23792441607338866,
+      "grad_norm": 2.4375,
+      "learning_rate": 7.928384565075213e-05,
+      "loss": 9.2376,
+      "step": 4850
+    },
+    {
+      "epoch": 0.23841498187972554,
+      "grad_norm": 2.921875,
+      "learning_rate": 7.944735120994115e-05,
+      "loss": 9.6234,
+      "step": 4860
+    },
+    {
+      "epoch": 0.23890554768606242,
+      "grad_norm": 1.6953125,
+      "learning_rate": 7.961085676913015e-05,
+      "loss": 10.0118,
+      "step": 4870
+    },
+    {
+      "epoch": 0.2393961134923993,
+      "grad_norm": 1.390625,
+      "learning_rate": 7.977436232831916e-05,
+      "loss": 10.1091,
+      "step": 4880
+    },
+    {
+      "epoch": 0.23988667929873617,
+      "grad_norm": 1.171875,
+      "learning_rate": 7.993786788750817e-05,
+      "loss": 9.9576,
+      "step": 4890
+    },
+    {
+      "epoch": 0.24037724510507305,
+      "grad_norm": 2.078125,
+      "learning_rate": 8.010137344669719e-05,
+      "loss": 9.7576,
+      "step": 4900
+    },
+    {
+      "epoch": 0.24086781091140994,
+      "grad_norm": 1.2265625,
+      "learning_rate": 8.02648790058862e-05,
+      "loss": 9.7052,
+      "step": 4910
+    },
+    {
+      "epoch": 0.24135837671774682,
+      "grad_norm": 2.75,
+      "learning_rate": 8.042838456507522e-05,
+      "loss": 9.8565,
+      "step": 4920
+    },
+    {
+      "epoch": 0.2418489425240837,
+      "grad_norm": 1.484375,
+      "learning_rate": 8.059189012426423e-05,
+      "loss": 10.0658,
+      "step": 4930
+    },
+    {
+      "epoch": 0.2423395083304206,
+      "grad_norm": 1.3046875,
+      "learning_rate": 8.075539568345324e-05,
+      "loss": 9.2865,
+      "step": 4940
+    },
+    {
+      "epoch": 0.24283007413675747,
+      "grad_norm": 1.9765625,
+      "learning_rate": 8.091890124264226e-05,
+      "loss": 9.2832,
+      "step": 4950
+    },
+    {
+      "epoch": 0.24332063994309436,
+      "grad_norm": 1.734375,
+      "learning_rate": 8.108240680183127e-05,
+      "loss": 9.1876,
+      "step": 4960
+    },
+    {
+      "epoch": 0.24381120574943124,
+      "grad_norm": 1.4375,
+      "learning_rate": 8.124591236102028e-05,
+      "loss": 9.7646,
+      "step": 4970
+    },
+    {
+      "epoch": 0.24430177155576813,
+      "grad_norm": 1.78125,
+      "learning_rate": 8.140941792020928e-05,
+      "loss": 9.5622,
+      "step": 4980
+    },
+    {
+      "epoch": 0.244792337362105,
+      "grad_norm": 2.5,
+      "learning_rate": 8.15729234793983e-05,
+      "loss": 9.3843,
+      "step": 4990
+    },
+    {
+      "epoch": 0.2452829031684419,
+      "grad_norm": 1.2578125,
+      "learning_rate": 8.173642903858731e-05,
+      "loss": 9.8428,
+      "step": 5000
+    },
+    {
+      "epoch": 0.24577346897477878,
+      "grad_norm": 0.9609375,
+      "learning_rate": 8.189993459777632e-05,
+      "loss": 9.0422,
+      "step": 5010
+    },
+    {
+      "epoch": 0.24626403478111566,
+      "grad_norm": 10.625,
+      "learning_rate": 8.206344015696535e-05,
+      "loss": 9.801,
+      "step": 5020
+    },
+    {
+      "epoch": 0.24675460058745255,
+      "grad_norm": 1.0625,
+      "learning_rate": 8.222694571615436e-05,
+      "loss": 9.4362,
+      "step": 5030
+    },
+    {
+      "epoch": 0.24724516639378943,
+      "grad_norm": 1.359375,
+      "learning_rate": 8.239045127534336e-05,
+      "loss": 9.9827,
+      "step": 5040
+    },
+    {
+      "epoch": 0.24773573220012632,
+      "grad_norm": 2.046875,
+      "learning_rate": 8.255395683453238e-05,
+      "loss": 9.8378,
+      "step": 5050
+    },
+    {
+      "epoch": 0.2482262980064632,
+      "grad_norm": 2.609375,
+      "learning_rate": 8.271746239372139e-05,
+      "loss": 9.683,
+      "step": 5060
+    },
+    {
+      "epoch": 0.2487168638128001,
+      "grad_norm": 3.34375,
+      "learning_rate": 8.28809679529104e-05,
+      "loss": 9.5704,
+      "step": 5070
+    },
+    {
+      "epoch": 0.24920742961913697,
+      "grad_norm": 2.4375,
+      "learning_rate": 8.304447351209942e-05,
+      "loss": 9.2374,
+      "step": 5080
+    },
+    {
+      "epoch": 0.24969799542547386,
+      "grad_norm": 1.34375,
+      "learning_rate": 8.320797907128843e-05,
+      "loss": 8.9982,
+      "step": 5090
+    },
+    {
+      "epoch": 0.2501885612318107,
+      "grad_norm": 2.0,
+      "learning_rate": 8.337148463047743e-05,
+      "loss": 9.6409,
+      "step": 5100
+    },
+    {
+      "epoch": 0.2506791270381476,
+      "grad_norm": 5.375,
+      "learning_rate": 8.353499018966646e-05,
+      "loss": 10.2844,
+      "step": 5110
+    },
+    {
+      "epoch": 0.2511696928444845,
+      "grad_norm": 1.296875,
+      "learning_rate": 8.369849574885547e-05,
+      "loss": 9.4956,
+      "step": 5120
+    },
+    {
+      "epoch": 0.25166025865082137,
+      "grad_norm": 1.3125,
+      "learning_rate": 8.386200130804448e-05,
+      "loss": 9.3721,
+      "step": 5130
+    },
+    {
+      "epoch": 0.25215082445715825,
+      "grad_norm": 1.34375,
+      "learning_rate": 8.40255068672335e-05,
+      "loss": 9.0161,
+      "step": 5140
+    },
+    {
+      "epoch": 0.25264139026349514,
+      "grad_norm": 2.921875,
+      "learning_rate": 8.41890124264225e-05,
+      "loss": 9.804,
+      "step": 5150
+    },
+    {
+      "epoch": 0.253131956069832,
+      "grad_norm": 1.9921875,
+      "learning_rate": 8.435251798561151e-05,
+      "loss": 9.7011,
+      "step": 5160
+    },
+    {
+      "epoch": 0.2536225218761689,
+      "grad_norm": 1.9140625,
+      "learning_rate": 8.451602354480052e-05,
+      "loss": 8.8055,
+      "step": 5170
+    },
+    {
+      "epoch": 0.2541130876825058,
+      "grad_norm": 2.375,
+      "learning_rate": 8.467952910398954e-05,
+      "loss": 9.9802,
+      "step": 5180
+    },
+    {
+      "epoch": 0.2546036534888427,
+      "grad_norm": 1.5078125,
+      "learning_rate": 8.484303466317855e-05,
+      "loss": 9.7401,
+      "step": 5190
+    },
+    {
+      "epoch": 0.25509421929517956,
+      "grad_norm": 2.25,
+      "learning_rate": 8.500654022236756e-05,
+      "loss": 9.9389,
+      "step": 5200
+    },
+    {
+      "epoch": 0.25558478510151644,
+      "grad_norm": 1.84375,
+      "learning_rate": 8.517004578155658e-05,
+      "loss": 10.0407,
+      "step": 5210
+    },
+    {
+      "epoch": 0.2560753509078533,
+      "grad_norm": 2.078125,
+      "learning_rate": 8.533355134074559e-05,
+      "loss": 8.9066,
+      "step": 5220
+    },
+    {
+      "epoch": 0.2565659167141902,
+      "grad_norm": 1.5703125,
+      "learning_rate": 8.54970568999346e-05,
+      "loss": 9.8056,
+      "step": 5230
+    },
+    {
+      "epoch": 0.2570564825205271,
+      "grad_norm": 1.3984375,
+      "learning_rate": 8.566056245912362e-05,
+      "loss": 10.5467,
+      "step": 5240
+    },
+    {
+      "epoch": 0.257547048326864,
+      "grad_norm": 1.7109375,
+      "learning_rate": 8.582406801831263e-05,
+      "loss": 10.1704,
+      "step": 5250
+    },
+    {
+      "epoch": 0.25803761413320087,
+      "grad_norm": 0.96484375,
+      "learning_rate": 8.598757357750163e-05,
+      "loss": 9.6242,
+      "step": 5260
+    },
+    {
+      "epoch": 0.25852817993953775,
+      "grad_norm": 1.5703125,
+      "learning_rate": 8.615107913669065e-05,
+      "loss": 10.1155,
+      "step": 5270
+    },
+    {
+      "epoch": 0.25901874574587463,
+      "grad_norm": 1.4453125,
+      "learning_rate": 8.631458469587966e-05,
+      "loss": 9.9926,
+      "step": 5280
+    },
+    {
+      "epoch": 0.2595093115522115,
+      "grad_norm": 1.109375,
+      "learning_rate": 8.647809025506867e-05,
+      "loss": 8.9682,
+      "step": 5290
+    },
+    {
+      "epoch": 0.2599998773585484,
+      "grad_norm": 1.578125,
+      "learning_rate": 8.66415958142577e-05,
+      "loss": 9.5974,
+      "step": 5300
+    },
+    {
+      "epoch": 0.2604904431648853,
+      "grad_norm": 1.8515625,
+      "learning_rate": 8.680510137344671e-05,
+      "loss": 9.6436,
+      "step": 5310
+    },
+    {
+      "epoch": 0.2609810089712222,
+      "grad_norm": 1.3828125,
+      "learning_rate": 8.696860693263571e-05,
+      "loss": 9.8851,
+      "step": 5320
+    },
+    {
+      "epoch": 0.26147157477755906,
+      "grad_norm": 1.1484375,
+      "learning_rate": 8.713211249182473e-05,
+      "loss": 10.6433,
+      "step": 5330
+    },
+    {
+      "epoch": 0.26196214058389594,
+      "grad_norm": 1.3046875,
+      "learning_rate": 8.729561805101374e-05,
+      "loss": 9.5067,
+      "step": 5340
+    },
+    {
+      "epoch": 0.2624527063902328,
+      "grad_norm": 2.328125,
+      "learning_rate": 8.745912361020275e-05,
+      "loss": 10.0498,
+      "step": 5350
+    },
+    {
+      "epoch": 0.2629432721965697,
+      "grad_norm": 2.328125,
+      "learning_rate": 8.762262916939177e-05,
+      "loss": 9.9077,
+      "step": 5360
+    },
+    {
+      "epoch": 0.2634338380029066,
+      "grad_norm": 1.4375,
+      "learning_rate": 8.778613472858077e-05,
+      "loss": 9.6394,
+      "step": 5370
+    },
+    {
+      "epoch": 0.2639244038092435,
+      "grad_norm": 1.296875,
+      "learning_rate": 8.794964028776978e-05,
+      "loss": 9.61,
+      "step": 5380
+    },
+    {
+      "epoch": 0.26441496961558036,
+      "grad_norm": 1.625,
+      "learning_rate": 8.81131458469588e-05,
+      "loss": 9.3544,
+      "step": 5390
+    },
+    {
+      "epoch": 0.26490553542191725,
+      "grad_norm": 2.265625,
+      "learning_rate": 8.827665140614782e-05,
+      "loss": 9.2304,
+      "step": 5400
+    },
+    {
+      "epoch": 0.26539610122825413,
+      "grad_norm": 1.7109375,
+      "learning_rate": 8.844015696533683e-05,
+      "loss": 9.7015,
+      "step": 5410
+    },
+    {
+      "epoch": 0.265886667034591,
+      "grad_norm": 1.546875,
+      "learning_rate": 8.860366252452585e-05,
+      "loss": 9.8647,
+      "step": 5420
+    },
+    {
+      "epoch": 0.2663772328409279,
+      "grad_norm": 1.015625,
+      "learning_rate": 8.876716808371485e-05,
+      "loss": 9.2876,
+      "step": 5430
+    },
+    {
+      "epoch": 0.2668677986472648,
+      "grad_norm": 0.83203125,
+      "learning_rate": 8.893067364290386e-05,
+      "loss": 8.8778,
+      "step": 5440
+    },
+    {
+      "epoch": 0.26735836445360167,
+      "grad_norm": 0.91015625,
+      "learning_rate": 8.909417920209287e-05,
+      "loss": 8.6977,
+      "step": 5450
+    },
+    {
+      "epoch": 0.26784893025993856,
+      "grad_norm": 0.66015625,
+      "learning_rate": 8.925768476128189e-05,
+      "loss": 9.7907,
+      "step": 5460
+    },
+    {
+      "epoch": 0.26833949606627544,
+      "grad_norm": 2.03125,
+      "learning_rate": 8.94211903204709e-05,
+      "loss": 8.9295,
+      "step": 5470
+    },
+    {
+      "epoch": 0.2688300618726123,
+      "grad_norm": 1.6015625,
+      "learning_rate": 8.958469587965991e-05,
+      "loss": 9.778,
+      "step": 5480
+    },
+    {
+      "epoch": 0.2693206276789492,
+      "grad_norm": 5.9375,
+      "learning_rate": 8.974820143884893e-05,
+      "loss": 9.4321,
+      "step": 5490
+    },
+    {
+      "epoch": 0.2698111934852861,
+      "grad_norm": 1.8984375,
+      "learning_rate": 8.991170699803794e-05,
+      "loss": 10.1971,
+      "step": 5500
+    },
+    {
+      "epoch": 0.270301759291623,
+      "grad_norm": 1.828125,
+      "learning_rate": 9.007521255722695e-05,
+      "loss": 9.8334,
+      "step": 5510
+    },
+    {
+      "epoch": 0.27079232509795986,
+      "grad_norm": 1.296875,
+      "learning_rate": 9.023871811641597e-05,
+      "loss": 9.423,
+      "step": 5520
+    },
+    {
+      "epoch": 0.27128289090429675,
+      "grad_norm": 101.5,
+      "learning_rate": 9.040222367560498e-05,
+      "loss": 9.5153,
+      "step": 5530
+    },
+    {
+      "epoch": 0.27177345671063363,
+      "grad_norm": 1.1875,
+      "learning_rate": 9.056572923479398e-05,
+      "loss": 9.9508,
+      "step": 5540
+    },
+    {
+      "epoch": 0.2722640225169705,
+      "grad_norm": 3.796875,
+      "learning_rate": 9.0729234793983e-05,
+      "loss": 8.8877,
+      "step": 5550
+    },
+    {
+      "epoch": 0.2727545883233074,
+      "grad_norm": 0.9921875,
+      "learning_rate": 9.089274035317201e-05,
+      "loss": 9.1971,
+      "step": 5560
+    },
+    {
+      "epoch": 0.2732451541296443,
+      "grad_norm": 1.4140625,
+      "learning_rate": 9.105624591236102e-05,
+      "loss": 10.0009,
+      "step": 5570
+    },
+    {
+      "epoch": 0.27373571993598117,
+      "grad_norm": 274.0,
+      "learning_rate": 9.121975147155003e-05,
+      "loss": 9.683,
+      "step": 5580
+    },
+    {
+      "epoch": 0.27422628574231805,
+      "grad_norm": 1.484375,
+      "learning_rate": 9.138325703073905e-05,
+      "loss": 9.4862,
+      "step": 5590
+    },
+    {
+      "epoch": 0.27471685154865494,
+      "grad_norm": 2.28125,
+      "learning_rate": 9.154676258992806e-05,
+      "loss": 10.0901,
+      "step": 5600
+    },
+    {
+      "epoch": 0.2752074173549918,
+      "grad_norm": 5.1875,
+      "learning_rate": 9.171026814911707e-05,
+      "loss": 9.5661,
+      "step": 5610
+    },
+    {
+      "epoch": 0.2756979831613287,
+      "grad_norm": 2.71875,
+      "learning_rate": 9.187377370830609e-05,
+      "loss": 10.3501,
+      "step": 5620
+    },
+    {
+      "epoch": 0.2761885489676656,
+      "grad_norm": 1.8515625,
+      "learning_rate": 9.20372792674951e-05,
+      "loss": 10.0153,
+      "step": 5630
+    },
+    {
+      "epoch": 0.2766791147740025,
+      "grad_norm": 1.46875,
+      "learning_rate": 9.220078482668412e-05,
+      "loss": 9.851,
+      "step": 5640
+    },
+    {
+      "epoch": 0.27716968058033936,
+      "grad_norm": 2.640625,
+      "learning_rate": 9.236429038587312e-05,
+      "loss": 9.4078,
+      "step": 5650
+    },
+    {
+      "epoch": 0.27766024638667625,
+      "grad_norm": 1.2421875,
+      "learning_rate": 9.252779594506213e-05,
+      "loss": 9.6189,
+      "step": 5660
+    },
+    {
+      "epoch": 0.27815081219301313,
+      "grad_norm": 1.5390625,
+      "learning_rate": 9.269130150425116e-05,
+      "loss": 8.8418,
+      "step": 5670
+    },
+    {
+      "epoch": 0.27864137799935,
+      "grad_norm": 2.734375,
+      "learning_rate": 9.285480706344017e-05,
+      "loss": 9.6745,
+      "step": 5680
+    },
+    {
+      "epoch": 0.2791319438056869,
+      "grad_norm": 1.5234375,
+      "learning_rate": 9.301831262262918e-05,
+      "loss": 9.664,
+      "step": 5690
+    },
+    {
+      "epoch": 0.2796225096120238,
+      "grad_norm": 1.703125,
+      "learning_rate": 9.318181818181818e-05,
+      "loss": 9.5209,
+      "step": 5700
+    },
+    {
+      "epoch": 0.28011307541836067,
+      "grad_norm": 1.34375,
+      "learning_rate": 9.33453237410072e-05,
+      "loss": 9.571,
+      "step": 5710
+    },
+    {
+      "epoch": 0.28060364122469755,
+      "grad_norm": 1.6328125,
+      "learning_rate": 9.350882930019621e-05,
+      "loss": 9.9315,
+      "step": 5720
+    },
+    {
+      "epoch": 0.28109420703103444,
+      "grad_norm": 1.1328125,
+      "learning_rate": 9.367233485938522e-05,
+      "loss": 9.7774,
+      "step": 5730
+    },
+    {
+      "epoch": 0.2815847728373713,
+      "grad_norm": 2.5,
+      "learning_rate": 9.383584041857424e-05,
+      "loss": 9.6759,
+      "step": 5740
+    },
+    {
+      "epoch": 0.2820753386437082,
+      "grad_norm": 1.9296875,
+      "learning_rate": 9.399934597776325e-05,
+      "loss": 9.6031,
+      "step": 5750
+    },
+    {
+      "epoch": 0.2825659044500451,
+      "grad_norm": 2.140625,
+      "learning_rate": 9.416285153695226e-05,
+      "loss": 10.1924,
+      "step": 5760
+    },
+    {
+      "epoch": 0.283056470256382,
+      "grad_norm": 1.4375,
+      "learning_rate": 9.432635709614128e-05,
+      "loss": 9.7358,
+      "step": 5770
+    },
+    {
+      "epoch": 0.28354703606271886,
+      "grad_norm": 1.5,
+      "learning_rate": 9.448986265533029e-05,
+      "loss": 10.2525,
+      "step": 5780
+    },
+    {
+      "epoch": 0.28403760186905574,
+      "grad_norm": 1.984375,
+      "learning_rate": 9.46533682145193e-05,
+      "loss": 9.5691,
+      "step": 5790
+    },
+    {
+      "epoch": 0.28452816767539263,
+      "grad_norm": 1.7890625,
+      "learning_rate": 9.481687377370832e-05,
+      "loss": 10.362,
+      "step": 5800
+    },
+    {
+      "epoch": 0.2850187334817295,
+      "grad_norm": 1.5,
+      "learning_rate": 9.498037933289732e-05,
+      "loss": 9.9536,
+      "step": 5810
+    },
+    {
+      "epoch": 0.2855092992880664,
+      "grad_norm": 1.890625,
+      "learning_rate": 9.514388489208633e-05,
+      "loss": 9.7241,
+      "step": 5820
+    },
+    {
+      "epoch": 0.2859998650944033,
+      "grad_norm": 1.765625,
+      "learning_rate": 9.530739045127534e-05,
+      "loss": 9.5508,
+      "step": 5830
+    },
+    {
+      "epoch": 0.28649043090074017,
+      "grad_norm": 1.09375,
+      "learning_rate": 9.547089601046436e-05,
+      "loss": 9.5517,
+      "step": 5840
+    },
+    {
+      "epoch": 0.28698099670707705,
+      "grad_norm": 1.515625,
+      "learning_rate": 9.563440156965337e-05,
+      "loss": 8.86,
+      "step": 5850
+    },
+    {
+      "epoch": 0.28747156251341394,
+      "grad_norm": 1.578125,
+      "learning_rate": 9.579790712884238e-05,
+      "loss": 9.6284,
+      "step": 5860
+    },
+    {
+      "epoch": 0.28796212831975077,
+      "grad_norm": 1.9140625,
+      "learning_rate": 9.59614126880314e-05,
+      "loss": 9.4105,
+      "step": 5870
+    },
+    {
+      "epoch": 0.28845269412608765,
+      "grad_norm": 1.9921875,
+      "learning_rate": 9.612491824722041e-05,
+      "loss": 9.9665,
+      "step": 5880
+    },
+    {
+      "epoch": 0.28894325993242453,
+      "grad_norm": 1.8828125,
+      "learning_rate": 9.628842380640942e-05,
+      "loss": 9.958,
+      "step": 5890
+    },
+    {
+      "epoch": 0.2894338257387614,
+      "grad_norm": 2.8125,
+      "learning_rate": 9.645192936559844e-05,
+      "loss": 9.5919,
+      "step": 5900
+    },
+    {
+      "epoch": 0.2899243915450983,
+      "grad_norm": 2.21875,
+      "learning_rate": 9.661543492478745e-05,
+      "loss": 10.0782,
+      "step": 5910
+    },
+    {
+      "epoch": 0.2904149573514352,
+      "grad_norm": 1.453125,
+      "learning_rate": 9.677894048397645e-05,
+      "loss": 8.6973,
+      "step": 5920
+    },
+    {
+      "epoch": 0.2909055231577721,
+      "grad_norm": 1.328125,
+      "learning_rate": 9.694244604316546e-05,
+      "loss": 9.348,
+      "step": 5930
+    },
+    {
+      "epoch": 0.29139608896410896,
+      "grad_norm": 1.7109375,
+      "learning_rate": 9.710595160235448e-05,
+      "loss": 9.3799,
+      "step": 5940
+    },
+    {
+      "epoch": 0.29188665477044584,
+      "grad_norm": 1.3515625,
+      "learning_rate": 9.72694571615435e-05,
+      "loss": 9.2056,
+      "step": 5950
+    },
+    {
+      "epoch": 0.2923772205767827,
+      "grad_norm": 1.2109375,
+      "learning_rate": 9.743296272073252e-05,
+      "loss": 9.3729,
+      "step": 5960
+    },
+    {
+      "epoch": 0.2928677863831196,
+      "grad_norm": 1.5390625,
+      "learning_rate": 9.759646827992153e-05,
+      "loss": 8.9693,
+      "step": 5970
+    },
+    {
+      "epoch": 0.2933583521894565,
+      "grad_norm": 2.0,
+      "learning_rate": 9.775997383911053e-05,
+      "loss": 9.5931,
+      "step": 5980
+    },
+    {
+      "epoch": 0.2938489179957934,
+      "grad_norm": 2.609375,
+      "learning_rate": 9.792347939829954e-05,
+      "loss": 8.7752,
+      "step": 5990
+    },
+    {
+      "epoch": 0.29433948380213026,
+      "grad_norm": 1.34375,
+      "learning_rate": 9.808698495748856e-05,
+      "loss": 9.6542,
+      "step": 6000
+    },
+    {
+      "epoch": 0.29483004960846715,
+      "grad_norm": 1.75,
+      "learning_rate": 9.825049051667757e-05,
+      "loss": 9.5273,
+      "step": 6010
+    },
+    {
+      "epoch": 0.29532061541480403,
+      "grad_norm": 2.796875,
+      "learning_rate": 9.841399607586659e-05,
+      "loss": 9.1164,
+      "step": 6020
+    },
+    {
+      "epoch": 0.2958111812211409,
+      "grad_norm": 1.46875,
+      "learning_rate": 9.857750163505559e-05,
+      "loss": 9.6354,
+      "step": 6030
+    },
+    {
+      "epoch": 0.2963017470274778,
+      "grad_norm": 1.4921875,
+      "learning_rate": 9.874100719424461e-05,
+      "loss": 9.8704,
+      "step": 6040
+    },
+    {
+      "epoch": 0.2967923128338147,
+      "grad_norm": 1.25,
+      "learning_rate": 9.890451275343363e-05,
+      "loss": 9.1349,
+      "step": 6050
+    },
+    {
+      "epoch": 0.29728287864015157,
+      "grad_norm": 1.8203125,
+      "learning_rate": 9.906801831262264e-05,
+      "loss": 9.3073,
+      "step": 6060
+    },
+    {
+      "epoch": 0.29777344444648846,
+      "grad_norm": 1.53125,
+      "learning_rate": 9.923152387181165e-05,
+      "loss": 9.479,
+      "step": 6070
+    },
+    {
+      "epoch": 0.29826401025282534,
+      "grad_norm": 3.1875,
+      "learning_rate": 9.939502943100067e-05,
+      "loss": 9.6812,
+      "step": 6080
+    },
+    {
+      "epoch": 0.2987545760591622,
+      "grad_norm": 1.3125,
+      "learning_rate": 9.955853499018967e-05,
+      "loss": 8.847,
+      "step": 6090
+    },
+    {
+      "epoch": 0.2992451418654991,
+      "grad_norm": 1.609375,
+      "learning_rate": 9.972204054937868e-05,
+      "loss": 9.6771,
+      "step": 6100
+    },
+    {
+      "epoch": 0.299735707671836,
+      "grad_norm": 2.296875,
+      "learning_rate": 9.988554610856769e-05,
+      "loss": 9.7602,
+      "step": 6110
+    },
+    {
+      "epoch": 0.3002262734781729,
+      "grad_norm": 0.984375,
+      "learning_rate": 9.999999926693718e-05,
+      "loss": 9.519,
+      "step": 6120
+    },
+    {
+      "epoch": 0.30071683928450976,
+      "grad_norm": 2.015625,
+      "learning_rate": 9.999998623470987e-05,
+      "loss": 9.9142,
+      "step": 6130
+    },
+    {
+      "epoch": 0.30120740509084665,
+      "grad_norm": 1.2109375,
+      "learning_rate": 9.999995691220257e-05,
+      "loss": 9.5174,
+      "step": 6140
+    },
+    {
+      "epoch": 0.30169797089718353,
+      "grad_norm": 3.046875,
+      "learning_rate": 9.999991129942481e-05,
+      "loss": 9.6071,
+      "step": 6150
+    },
+    {
+      "epoch": 0.3021885367035204,
+      "grad_norm": 1.921875,
+      "learning_rate": 9.999984939639145e-05,
+      "loss": 9.5535,
+      "step": 6160
+    },
+    {
+      "epoch": 0.3026791025098573,
+      "grad_norm": 1.6484375,
+      "learning_rate": 9.99997712031227e-05,
+      "loss": 9.0336,
+      "step": 6170
+    },
+    {
+      "epoch": 0.3031696683161942,
+      "grad_norm": 1.3828125,
+      "learning_rate": 9.999967671964399e-05,
+      "loss": 9.4898,
+      "step": 6180
+    },
+    {
+      "epoch": 0.30366023412253107,
+      "grad_norm": 1.8203125,
+      "learning_rate": 9.999956594598612e-05,
+      "loss": 9.3706,
+      "step": 6190
+    },
+    {
+      "epoch": 0.30415079992886795,
+      "grad_norm": 2.875,
+      "learning_rate": 9.99994388821852e-05,
+      "loss": 10.0704,
+      "step": 6200
+    },
+    {
+      "epoch": 0.30464136573520484,
+      "grad_norm": 2.0625,
+      "learning_rate": 9.99992955282826e-05,
+      "loss": 9.6955,
+      "step": 6210
+    },
+    {
+      "epoch": 0.3051319315415417,
+      "grad_norm": 1.21875,
+      "learning_rate": 9.999913588432502e-05,
+      "loss": 9.3493,
+      "step": 6220
+    },
+    {
+      "epoch": 0.3056224973478786,
+      "grad_norm": 194.0,
+      "learning_rate": 9.99989599503645e-05,
+      "loss": 9.1016,
+      "step": 6230
+    },
+    {
+      "epoch": 0.3061130631542155,
+      "grad_norm": 1.1328125,
+      "learning_rate": 9.999876772645835e-05,
+      "loss": 9.5685,
+      "step": 6240
+    },
+    {
+      "epoch": 0.3066036289605524,
+      "grad_norm": 2.28125,
+      "learning_rate": 9.999855921266921e-05,
+      "loss": 9.7715,
+      "step": 6250
+    },
+    {
+      "epoch": 0.30709419476688926,
+      "grad_norm": 2.046875,
+      "learning_rate": 9.999833440906499e-05,
+      "loss": 10.0416,
+      "step": 6260
+    },
+    {
+      "epoch": 0.30758476057322615,
+      "grad_norm": 1.3515625,
+      "learning_rate": 9.999809331571895e-05,
+      "loss": 9.3091,
+      "step": 6270
+    },
+    {
+      "epoch": 0.30807532637956303,
+      "grad_norm": 2.1875,
+      "learning_rate": 9.999783593270962e-05,
+      "loss": 9.6715,
+      "step": 6280
+    },
+    {
+      "epoch": 0.3085658921858999,
+      "grad_norm": 1.5703125,
+      "learning_rate": 9.999756226012087e-05,
+      "loss": 9.9269,
+      "step": 6290
+    },
+    {
+      "epoch": 0.3090564579922368,
+      "grad_norm": 1.7421875,
+      "learning_rate": 9.999727229804187e-05,
+      "loss": 9.7873,
+      "step": 6300
+    },
+    {
+      "epoch": 0.3095470237985737,
+      "grad_norm": 1.2578125,
+      "learning_rate": 9.999696604656709e-05,
+      "loss": 9.9722,
+      "step": 6310
+    },
+    {
+      "epoch": 0.31003758960491057,
+      "grad_norm": 1.3828125,
+      "learning_rate": 9.999664350579631e-05,
+      "loss": 9.2905,
+      "step": 6320
+    },
+    {
+      "epoch": 0.31052815541124745,
+      "grad_norm": 1.1640625,
+      "learning_rate": 9.99963046758346e-05,
+      "loss": 9.6233,
+      "step": 6330
+    },
+    {
+      "epoch": 0.31101872121758434,
+      "grad_norm": 2.4375,
+      "learning_rate": 9.999594955679236e-05,
+      "loss": 9.7319,
+      "step": 6340
+    },
+    {
+      "epoch": 0.3115092870239212,
+      "grad_norm": 1.8828125,
+      "learning_rate": 9.999557814878531e-05,
+      "loss": 10.4584,
+      "step": 6350
+    },
+    {
+      "epoch": 0.3119998528302581,
+      "grad_norm": 1.234375,
+      "learning_rate": 9.999519045193442e-05,
+      "loss": 9.4544,
+      "step": 6360
+    },
+    {
+      "epoch": 0.312490418636595,
+      "grad_norm": 1.7578125,
+      "learning_rate": 9.999478646636602e-05,
+      "loss": 9.6401,
+      "step": 6370
+    },
+    {
+      "epoch": 0.3129809844429319,
+      "grad_norm": 1.7109375,
+      "learning_rate": 9.999436619221175e-05,
+      "loss": 9.2365,
+      "step": 6380
+    },
+    {
+      "epoch": 0.31347155024926876,
+      "grad_norm": 1.6328125,
+      "learning_rate": 9.999392962960851e-05,
+      "loss": 9.4004,
+      "step": 6390
+    },
+    {
+      "epoch": 0.31396211605560564,
+      "grad_norm": 0.8515625,
+      "learning_rate": 9.999347677869854e-05,
+      "loss": 9.4101,
+      "step": 6400
+    },
+    {
+      "epoch": 0.31445268186194253,
+      "grad_norm": 1.8125,
+      "learning_rate": 9.99930076396294e-05,
+      "loss": 9.4732,
+      "step": 6410
+    },
+    {
+      "epoch": 0.3149432476682794,
+      "grad_norm": 1.765625,
+      "learning_rate": 9.999252221255393e-05,
+      "loss": 9.2319,
+      "step": 6420
+    },
+    {
+      "epoch": 0.3154338134746163,
+      "grad_norm": 0.9765625,
+      "learning_rate": 9.999202049763027e-05,
+      "loss": 9.4702,
+      "step": 6430
+    },
+    {
+      "epoch": 0.3159243792809532,
+      "grad_norm": 0.7578125,
+      "learning_rate": 9.99915024950219e-05,
+      "loss": 9.2764,
+      "step": 6440
+    },
+    {
+      "epoch": 0.31641494508729007,
+      "grad_norm": 1.1953125,
+      "learning_rate": 9.999096820489757e-05,
+      "loss": 9.2799,
+      "step": 6450
+    },
+    {
+      "epoch": 0.31690551089362695,
+      "grad_norm": 2.9375,
+      "learning_rate": 9.999041762743137e-05,
+      "loss": 9.2077,
+      "step": 6460
+    },
+    {
+      "epoch": 0.31739607669996384,
+      "grad_norm": 1.234375,
+      "learning_rate": 9.998985076280268e-05,
+      "loss": 9.0989,
+      "step": 6470
+    },
+    {
+      "epoch": 0.3178866425063007,
+      "grad_norm": 0.8515625,
+      "learning_rate": 9.998926761119618e-05,
+      "loss": 9.5567,
+      "step": 6480
+    },
+    {
+      "epoch": 0.3183772083126376,
+      "grad_norm": 1.984375,
+      "learning_rate": 9.998866817280187e-05,
+      "loss": 9.6093,
+      "step": 6490
+    },
+    {
+      "epoch": 0.3188677741189745,
+      "grad_norm": 1.0234375,
+      "learning_rate": 9.998805244781504e-05,
+      "loss": 9.0293,
+      "step": 6500
+    },
+    {
+      "epoch": 0.3193583399253114,
+      "grad_norm": 1.3515625,
+      "learning_rate": 9.998742043643632e-05,
+      "loss": 9.4566,
+      "step": 6510
+    },
+    {
+      "epoch": 0.31984890573164826,
+      "grad_norm": 1.359375,
+      "learning_rate": 9.99867721388716e-05,
+      "loss": 9.5419,
+      "step": 6520
+    },
+    {
+      "epoch": 0.32033947153798514,
+      "grad_norm": 1.2578125,
+      "learning_rate": 9.998610755533211e-05,
+      "loss": 9.3771,
+      "step": 6530
+    },
+    {
+      "epoch": 0.320830037344322,
+      "grad_norm": 1.8203125,
+      "learning_rate": 9.998542668603436e-05,
+      "loss": 10.0576,
+      "step": 6540
+    },
+    {
+      "epoch": 0.3213206031506589,
+      "grad_norm": 0.99609375,
+      "learning_rate": 9.998472953120022e-05,
+      "loss": 8.9065,
+      "step": 6550
+    },
+    {
+      "epoch": 0.3218111689569958,
+      "grad_norm": 1.15625,
+      "learning_rate": 9.998401609105677e-05,
+      "loss": 9.5357,
+      "step": 6560
+    },
+    {
+      "epoch": 0.3223017347633327,
+      "grad_norm": 1.46875,
+      "learning_rate": 9.99832863658365e-05,
+      "loss": 9.4754,
+      "step": 6570
+    },
+    {
+      "epoch": 0.32279230056966957,
+      "grad_norm": 1.7890625,
+      "learning_rate": 9.998254035577714e-05,
+      "loss": 10.0797,
+      "step": 6580
+    },
+    {
+      "epoch": 0.32328286637600645,
+      "grad_norm": 1.0,
+      "learning_rate": 9.998177806112175e-05,
+      "loss": 9.659,
+      "step": 6590
+    },
+    {
+      "epoch": 0.32377343218234333,
+      "grad_norm": 1.4140625,
+      "learning_rate": 9.998099948211868e-05,
+      "loss": 9.5543,
+      "step": 6600
+    },
+    {
+      "epoch": 0.3242639979886802,
+      "grad_norm": 1.25,
+      "learning_rate": 9.99802046190216e-05,
+      "loss": 9.3885,
+      "step": 6610
+    },
+    {
+      "epoch": 0.3247545637950171,
+      "grad_norm": 1.171875,
+      "learning_rate": 9.997939347208949e-05,
+      "loss": 8.6886,
+      "step": 6620
+    },
+    {
+      "epoch": 0.325245129601354,
+      "grad_norm": 0.8515625,
+      "learning_rate": 9.997856604158662e-05,
+      "loss": 10.1025,
+      "step": 6630
+    },
+    {
+      "epoch": 0.3257356954076909,
+      "grad_norm": 1.8671875,
+      "learning_rate": 9.997772232778256e-05,
+      "loss": 9.5884,
+      "step": 6640
+    },
+    {
+      "epoch": 0.3262262612140277,
+      "grad_norm": 1.4375,
+      "learning_rate": 9.997686233095222e-05,
+      "loss": 9.5029,
+      "step": 6650
+    },
+    {
+      "epoch": 0.3267168270203646,
+      "grad_norm": 1.765625,
+      "learning_rate": 9.997598605137577e-05,
+      "loss": 9.418,
+      "step": 6660
+    },
+    {
+      "epoch": 0.32720739282670147,
+      "grad_norm": 1.1640625,
+      "learning_rate": 9.997509348933871e-05,
+      "loss": 9.5372,
+      "step": 6670
+    },
+    {
+      "epoch": 0.32769795863303836,
+      "grad_norm": 2.25,
+      "learning_rate": 9.997418464513187e-05,
+      "loss": 9.3032,
+      "step": 6680
+    },
+    {
+      "epoch": 0.32818852443937524,
+      "grad_norm": 1.71875,
+      "learning_rate": 9.997325951905131e-05,
+      "loss": 9.7677,
+      "step": 6690
+    },
+    {
+      "epoch": 0.3286790902457121,
+      "grad_norm": 2.5625,
+      "learning_rate": 9.997231811139848e-05,
+      "loss": 9.8891,
+      "step": 6700
+    },
+    {
+      "epoch": 0.329169656052049,
+      "grad_norm": 1.5078125,
+      "learning_rate": 9.997136042248008e-05,
+      "loss": 9.475,
+      "step": 6710
+    },
+    {
+      "epoch": 0.3296602218583859,
+      "grad_norm": 1.9453125,
+      "learning_rate": 9.997038645260812e-05,
+      "loss": 9.2955,
+      "step": 6720
+    },
+    {
+      "epoch": 0.3301507876647228,
+      "grad_norm": 1.578125,
+      "learning_rate": 9.996939620209995e-05,
+      "loss": 10.0806,
+      "step": 6730
+    },
+    {
+      "epoch": 0.33064135347105966,
+      "grad_norm": 1.078125,
+      "learning_rate": 9.996838967127818e-05,
+      "loss": 8.6916,
+      "step": 6740
+    },
+    {
+      "epoch": 0.33113191927739655,
+      "grad_norm": 1.734375,
+      "learning_rate": 9.996736686047075e-05,
+      "loss": 9.1876,
+      "step": 6750
+    },
+    {
+      "epoch": 0.33162248508373343,
+      "grad_norm": 1.2578125,
+      "learning_rate": 9.99663277700109e-05,
+      "loss": 9.1853,
+      "step": 6760
+    },
+    {
+      "epoch": 0.3321130508900703,
+      "grad_norm": 1.421875,
+      "learning_rate": 9.996527240023717e-05,
+      "loss": 9.2008,
+      "step": 6770
+    },
+    {
+      "epoch": 0.3326036166964072,
+      "grad_norm": 1.75,
+      "learning_rate": 9.996420075149341e-05,
+      "loss": 9.2655,
+      "step": 6780
+    },
+    {
+      "epoch": 0.3330941825027441,
+      "grad_norm": 1.7265625,
+      "learning_rate": 9.996311282412876e-05,
+      "loss": 9.4802,
+      "step": 6790
+    },
+    {
+      "epoch": 0.33358474830908097,
+      "grad_norm": 1.5234375,
+      "learning_rate": 9.996200861849767e-05,
+      "loss": 9.3536,
+      "step": 6800
+    },
+    {
+      "epoch": 0.33407531411541785,
+      "grad_norm": 1.390625,
+      "learning_rate": 9.99608881349599e-05,
+      "loss": 9.6765,
+      "step": 6810
+    },
+    {
+      "epoch": 0.33456587992175474,
+      "grad_norm": 2.484375,
+      "learning_rate": 9.995975137388052e-05,
+      "loss": 9.1743,
+      "step": 6820
+    },
+    {
+      "epoch": 0.3350564457280916,
+      "grad_norm": 1.171875,
+      "learning_rate": 9.995859833562988e-05,
+      "loss": 8.9971,
+      "step": 6830
+    },
+    {
+      "epoch": 0.3355470115344285,
+      "grad_norm": 2.140625,
+      "learning_rate": 9.995742902058367e-05,
+      "loss": 9.143,
+      "step": 6840
+    },
+    {
+      "epoch": 0.3360375773407654,
+      "grad_norm": 1.859375,
+      "learning_rate": 9.995624342912282e-05,
+      "loss": 9.8555,
+      "step": 6850
+    },
+    {
+      "epoch": 0.3365281431471023,
+      "grad_norm": 2.203125,
+      "learning_rate": 9.995504156163364e-05,
+      "loss": 9.659,
+      "step": 6860
+    },
+    {
+      "epoch": 0.33701870895343916,
+      "grad_norm": 1.9765625,
+      "learning_rate": 9.995382341850769e-05,
+      "loss": 9.178,
+      "step": 6870
+    },
+    {
+      "epoch": 0.33750927475977605,
+      "grad_norm": 1.4765625,
+      "learning_rate": 9.995258900014184e-05,
+      "loss": 9.9946,
+      "step": 6880
+    },
+    {
+      "epoch": 0.33799984056611293,
+      "grad_norm": 1.3046875,
+      "learning_rate": 9.995133830693828e-05,
+      "loss": 9.1998,
+      "step": 6890
+    },
+    {
+      "epoch": 0.3384904063724498,
+      "grad_norm": 1.2265625,
+      "learning_rate": 9.995007133930449e-05,
+      "loss": 9.3834,
+      "step": 6900
+    },
+    {
+      "epoch": 0.3389809721787867,
+      "grad_norm": 1.5078125,
+      "learning_rate": 9.994878809765327e-05,
+      "loss": 9.5709,
+      "step": 6910
+    },
+    {
+      "epoch": 0.3394715379851236,
+      "grad_norm": 1.5859375,
+      "learning_rate": 9.994748858240268e-05,
+      "loss": 9.3729,
+      "step": 6920
+    },
+    {
+      "epoch": 0.33996210379146047,
+      "grad_norm": 1.1484375,
+      "learning_rate": 9.994617279397612e-05,
+      "loss": 10.3316,
+      "step": 6930
+    },
+    {
+      "epoch": 0.34045266959779735,
+      "grad_norm": 1.8125,
+      "learning_rate": 9.99448407328023e-05,
+      "loss": 10.2478,
+      "step": 6940
+    },
+    {
+      "epoch": 0.34094323540413424,
+      "grad_norm": 1.421875,
+      "learning_rate": 9.994349239931517e-05,
+      "loss": 9.4121,
+      "step": 6950
+    },
+    {
+      "epoch": 0.3414338012104711,
+      "grad_norm": 1.8359375,
+      "learning_rate": 9.994212779395409e-05,
+      "loss": 9.5804,
+      "step": 6960
+    },
+    {
+      "epoch": 0.341924367016808,
+      "grad_norm": 1.359375,
+      "learning_rate": 9.99407469171636e-05,
+      "loss": 9.8077,
+      "step": 6970
+    },
+    {
+      "epoch": 0.3424149328231449,
+      "grad_norm": 1.6484375,
+      "learning_rate": 9.99393497693936e-05,
+      "loss": 9.1917,
+      "step": 6980
+    },
+    {
+      "epoch": 0.3429054986294818,
+      "grad_norm": 1.609375,
+      "learning_rate": 9.993793635109933e-05,
+      "loss": 9.6918,
+      "step": 6990
+    },
+    {
+      "epoch": 0.34339606443581866,
+      "grad_norm": 1.1015625,
+      "learning_rate": 9.993650666274126e-05,
+      "loss": 10.0207,
+      "step": 7000
+    },
+    {
+      "epoch": 0.34388663024215554,
+      "grad_norm": 1.5234375,
+      "learning_rate": 9.993506070478519e-05,
+      "loss": 9.6955,
+      "step": 7010
+    },
+    {
+      "epoch": 0.34437719604849243,
+      "grad_norm": 5.125,
+      "learning_rate": 9.993359847770223e-05,
+      "loss": 9.4934,
+      "step": 7020
+    },
+    {
+      "epoch": 0.3448677618548293,
+      "grad_norm": 1.2421875,
+      "learning_rate": 9.993211998196877e-05,
+      "loss": 9.6029,
+      "step": 7030
+    },
+    {
+      "epoch": 0.3453583276611662,
+      "grad_norm": 1.5078125,
+      "learning_rate": 9.993062521806653e-05,
+      "loss": 10.157,
+      "step": 7040
+    },
+    {
+      "epoch": 0.3458488934675031,
+      "grad_norm": 1.4453125,
+      "learning_rate": 9.99291141864825e-05,
+      "loss": 9.3385,
+      "step": 7050
+    },
+    {
+      "epoch": 0.34633945927383997,
+      "grad_norm": 2.125,
+      "learning_rate": 9.9927586887709e-05,
+      "loss": 9.7968,
+      "step": 7060
+    },
+    {
+      "epoch": 0.34683002508017685,
+      "grad_norm": 1.3515625,
+      "learning_rate": 9.99260433222436e-05,
+      "loss": 9.5476,
+      "step": 7070
+    },
+    {
+      "epoch": 0.34732059088651374,
+      "grad_norm": 1.09375,
+      "learning_rate": 9.992448349058923e-05,
+      "loss": 9.7711,
+      "step": 7080
+    },
+    {
+      "epoch": 0.3478111566928506,
+      "grad_norm": 1.296875,
+      "learning_rate": 9.992290739325408e-05,
+      "loss": 9.3846,
+      "step": 7090
+    },
+    {
+      "epoch": 0.3483017224991875,
+      "grad_norm": 1.5,
+      "learning_rate": 9.992131503075166e-05,
+      "loss": 9.5349,
+      "step": 7100
+    },
+    {
+      "epoch": 0.3487922883055244,
+      "grad_norm": 1.5546875,
+      "learning_rate": 9.991970640360075e-05,
+      "loss": 9.5095,
+      "step": 7110
+    },
+    {
+      "epoch": 0.3492828541118613,
+      "grad_norm": 1.28125,
+      "learning_rate": 9.99180815123255e-05,
+      "loss": 9.7473,
+      "step": 7120
+    },
+    {
+      "epoch": 0.34977341991819816,
+      "grad_norm": 2.359375,
+      "learning_rate": 9.991644035745524e-05,
+      "loss": 9.3305,
+      "step": 7130
+    },
+    {
+      "epoch": 0.35026398572453504,
+      "grad_norm": 2.390625,
+      "learning_rate": 9.991478293952472e-05,
+      "loss": 9.5504,
+      "step": 7140
+    },
+    {
+      "epoch": 0.3507545515308719,
+      "grad_norm": 1.9140625,
+      "learning_rate": 9.99131092590739e-05,
+      "loss": 9.4789,
+      "step": 7150
+    },
+    {
+      "epoch": 0.3512451173372088,
+      "grad_norm": 1.2734375,
+      "learning_rate": 9.991141931664811e-05,
+      "loss": 9.5481,
+      "step": 7160
+    },
+    {
+      "epoch": 0.3517356831435457,
+      "grad_norm": 2.328125,
+      "learning_rate": 9.99097131127979e-05,
+      "loss": 9.7359,
+      "step": 7170
+    },
+    {
+      "epoch": 0.3522262489498826,
+      "grad_norm": 1.921875,
+      "learning_rate": 9.990799064807921e-05,
+      "loss": 9.6452,
+      "step": 7180
+    },
+    {
+      "epoch": 0.35271681475621947,
+      "grad_norm": 1.28125,
+      "learning_rate": 9.990625192305321e-05,
+      "loss": 9.7301,
+      "step": 7190
+    },
+    {
+      "epoch": 0.35320738056255635,
+      "grad_norm": 0.81640625,
+      "learning_rate": 9.990449693828636e-05,
+      "loss": 9.1542,
+      "step": 7200
+    },
+    {
+      "epoch": 0.35369794636889323,
+      "grad_norm": 1.5390625,
+      "learning_rate": 9.990272569435048e-05,
+      "loss": 9.6957,
+      "step": 7210
+    },
+    {
+      "epoch": 0.3541885121752301,
+      "grad_norm": 1.4765625,
+      "learning_rate": 9.990093819182264e-05,
+      "loss": 10.0123,
+      "step": 7220
+    },
+    {
+      "epoch": 0.354679077981567,
+      "grad_norm": 1.0,
+      "learning_rate": 9.989913443128521e-05,
+      "loss": 9.1995,
+      "step": 7230
+    },
+    {
+      "epoch": 0.3551696437879039,
+      "grad_norm": 1.046875,
+      "learning_rate": 9.989731441332587e-05,
+      "loss": 9.1959,
+      "step": 7240
+    },
+    {
+      "epoch": 0.3556602095942408,
+      "grad_norm": 1.3828125,
+      "learning_rate": 9.989547813853758e-05,
+      "loss": 9.8696,
+      "step": 7250
+    },
+    {
+      "epoch": 0.35615077540057766,
+      "grad_norm": 1.6328125,
+      "learning_rate": 9.989362560751864e-05,
+      "loss": 9.9184,
+      "step": 7260
+    },
+    {
+      "epoch": 0.35664134120691454,
+      "grad_norm": 1.53125,
+      "learning_rate": 9.989175682087261e-05,
+      "loss": 8.766,
+      "step": 7270
+    },
+    {
+      "epoch": 0.3571319070132514,
+      "grad_norm": 1.3046875,
+      "learning_rate": 9.988987177920832e-05,
+      "loss": 9.919,
+      "step": 7280
+    },
+    {
+      "epoch": 0.3576224728195883,
+      "grad_norm": 2.03125,
+      "learning_rate": 9.988797048313995e-05,
+      "loss": 9.6369,
+      "step": 7290
+    },
+    {
+      "epoch": 0.3581130386259252,
+      "grad_norm": 1.484375,
+      "learning_rate": 9.988605293328694e-05,
+      "loss": 9.3527,
+      "step": 7300
+    },
+    {
+      "epoch": 0.3586036044322621,
+      "grad_norm": 1.0703125,
+      "learning_rate": 9.988411913027405e-05,
+      "loss": 9.7346,
+      "step": 7310
+    },
+    {
+      "epoch": 0.35909417023859896,
+      "grad_norm": 0.9921875,
+      "learning_rate": 9.988216907473134e-05,
+      "loss": 9.8303,
+      "step": 7320
+    },
+    {
+      "epoch": 0.35958473604493585,
+      "grad_norm": 1.03125,
+      "learning_rate": 9.988020276729411e-05,
+      "loss": 9.6438,
+      "step": 7330
+    },
+    {
+      "epoch": 0.36007530185127273,
+      "grad_norm": 1.1015625,
+      "learning_rate": 9.987822020860303e-05,
+      "loss": 9.2781,
+      "step": 7340
+    },
+    {
+      "epoch": 0.3605658676576096,
+      "grad_norm": 1.3515625,
+      "learning_rate": 9.987622139930399e-05,
+      "loss": 9.7961,
+      "step": 7350
+    },
+    {
+      "epoch": 0.3610564334639465,
+      "grad_norm": 2.078125,
+      "learning_rate": 9.987420634004826e-05,
+      "loss": 9.5671,
+      "step": 7360
+    },
+    {
+      "epoch": 0.3615469992702834,
+      "grad_norm": 1.53125,
+      "learning_rate": 9.987217503149233e-05,
+      "loss": 9.6844,
+      "step": 7370
+    },
+    {
+      "epoch": 0.36203756507662027,
+      "grad_norm": 1.8203125,
+      "learning_rate": 9.987012747429801e-05,
+      "loss": 10.0758,
+      "step": 7380
+    },
+    {
+      "epoch": 0.36252813088295716,
+      "grad_norm": 1.8828125,
+      "learning_rate": 9.986806366913241e-05,
+      "loss": 9.2339,
+      "step": 7390
+    },
+    {
+      "epoch": 0.36301869668929404,
+      "grad_norm": 2.1875,
+      "learning_rate": 9.986598361666794e-05,
+      "loss": 9.7534,
+      "step": 7400
+    },
+    {
+      "epoch": 0.3635092624956309,
+      "grad_norm": 1.4453125,
+      "learning_rate": 9.98638873175823e-05,
+      "loss": 9.531,
+      "step": 7410
+    },
+    {
+      "epoch": 0.3639998283019678,
+      "grad_norm": 1.125,
+      "learning_rate": 9.986177477255843e-05,
+      "loss": 9.281,
+      "step": 7420
+    },
+    {
+      "epoch": 0.36449039410830464,
+      "grad_norm": 4.5,
+      "learning_rate": 9.985964598228465e-05,
+      "loss": 9.3941,
+      "step": 7430
+    },
+    {
+      "epoch": 0.3649809599146415,
+      "grad_norm": 2.640625,
+      "learning_rate": 9.985750094745454e-05,
+      "loss": 9.8157,
+      "step": 7440
+    },
+    {
+      "epoch": 0.3654715257209784,
+      "grad_norm": 1.3515625,
+      "learning_rate": 9.985533966876693e-05,
+      "loss": 9.3986,
+      "step": 7450
+    },
+    {
+      "epoch": 0.3659620915273153,
+      "grad_norm": 2.578125,
+      "learning_rate": 9.9853162146926e-05,
+      "loss": 10.3461,
+      "step": 7460
+    },
+    {
+      "epoch": 0.3664526573336522,
+      "grad_norm": 1.5703125,
+      "learning_rate": 9.985096838264118e-05,
+      "loss": 9.5886,
+      "step": 7470
+    },
+    {
+      "epoch": 0.36694322313998906,
+      "grad_norm": 1.1015625,
+      "learning_rate": 9.984875837662725e-05,
+      "loss": 9.5306,
+      "step": 7480
+    },
+    {
+      "epoch": 0.36743378894632595,
+      "grad_norm": 1.8515625,
+      "learning_rate": 9.98465321296042e-05,
+      "loss": 9.589,
+      "step": 7490
+    },
+    {
+      "epoch": 0.36792435475266283,
+      "grad_norm": 2.078125,
+      "learning_rate": 9.984428964229737e-05,
+      "loss": 9.4794,
+      "step": 7500
+    },
+    {
+      "epoch": 0.3684149205589997,
+      "grad_norm": 1.640625,
+      "learning_rate": 9.984203091543738e-05,
+      "loss": 9.4859,
+      "step": 7510
+    },
+    {
+      "epoch": 0.3689054863653366,
+      "grad_norm": 0.84375,
+      "learning_rate": 9.983975594976012e-05,
+      "loss": 9.3247,
+      "step": 7520
+    },
+    {
+      "epoch": 0.3693960521716735,
+      "grad_norm": 1.40625,
+      "learning_rate": 9.983746474600682e-05,
+      "loss": 9.3625,
+      "step": 7530
+    },
+    {
+      "epoch": 0.36988661797801037,
+      "grad_norm": 1.6640625,
+      "learning_rate": 9.983515730492392e-05,
+      "loss": 8.9516,
+      "step": 7540
+    },
+    {
+      "epoch": 0.37037718378434725,
+      "grad_norm": 1.421875,
+      "learning_rate": 9.983283362726324e-05,
+      "loss": 9.6106,
+      "step": 7550
+    },
+    {
+      "epoch": 0.37086774959068414,
+      "grad_norm": 1.4921875,
+      "learning_rate": 9.983049371378182e-05,
+      "loss": 9.892,
+      "step": 7560
+    },
+    {
+      "epoch": 0.371358315397021,
+      "grad_norm": 0.9921875,
+      "learning_rate": 9.982813756524203e-05,
+      "loss": 9.1626,
+      "step": 7570
+    },
+    {
+      "epoch": 0.3718488812033579,
+      "grad_norm": 1.3671875,
+      "learning_rate": 9.98257651824115e-05,
+      "loss": 9.6846,
+      "step": 7580
+    },
+    {
+      "epoch": 0.3723394470096948,
+      "grad_norm": 1.0390625,
+      "learning_rate": 9.982337656606323e-05,
+      "loss": 9.525,
+      "step": 7590
+    },
+    {
+      "epoch": 0.3728300128160317,
+      "grad_norm": 286.0,
+      "learning_rate": 9.982097171697535e-05,
+      "loss": 9.5155,
+      "step": 7600
+    },
+    {
+      "epoch": 0.37332057862236856,
+      "grad_norm": 1.7265625,
+      "learning_rate": 9.981855063593142e-05,
+      "loss": 10.0883,
+      "step": 7610
+    },
+    {
+      "epoch": 0.37381114442870544,
+      "grad_norm": 1.2265625,
+      "learning_rate": 9.981611332372025e-05,
+      "loss": 8.9297,
+      "step": 7620
+    },
+    {
+      "epoch": 0.37430171023504233,
+      "grad_norm": 1.5234375,
+      "learning_rate": 9.981365978113592e-05,
+      "loss": 9.2481,
+      "step": 7630
+    },
+    {
+      "epoch": 0.3747922760413792,
+      "grad_norm": 2.21875,
+      "learning_rate": 9.98111900089778e-05,
+      "loss": 9.2244,
+      "step": 7640
+    },
+    {
+      "epoch": 0.3752828418477161,
+      "grad_norm": 1.5859375,
+      "learning_rate": 9.980870400805058e-05,
+      "loss": 9.7969,
+      "step": 7650
+    },
+    {
+      "epoch": 0.375773407654053,
+      "grad_norm": 1.015625,
+      "learning_rate": 9.980620177916419e-05,
+      "loss": 9.0249,
+      "step": 7660
+    },
+    {
+      "epoch": 0.37626397346038987,
+      "grad_norm": 1.59375,
+      "learning_rate": 9.980368332313388e-05,
+      "loss": 9.4258,
+      "step": 7670
+    },
+    {
+      "epoch": 0.37675453926672675,
+      "grad_norm": 1.71875,
+      "learning_rate": 9.980114864078016e-05,
+      "loss": 9.5474,
+      "step": 7680
+    },
+    {
+      "epoch": 0.37724510507306364,
+      "grad_norm": 2.28125,
+      "learning_rate": 9.979859773292888e-05,
+      "loss": 9.557,
+      "step": 7690
+    },
+    {
+      "epoch": 0.3777356708794005,
+      "grad_norm": 0.98046875,
+      "learning_rate": 9.979603060041111e-05,
+      "loss": 9.2212,
+      "step": 7700
+    },
+    {
+      "epoch": 0.3782262366857374,
+      "grad_norm": 1.265625,
+      "learning_rate": 9.979344724406323e-05,
+      "loss": 9.5588,
+      "step": 7710
+    },
+    {
+      "epoch": 0.3787168024920743,
+      "grad_norm": 1.34375,
+      "learning_rate": 9.979084766472695e-05,
+      "loss": 9.7479,
+      "step": 7720
+    },
+    {
+      "epoch": 0.3792073682984112,
+      "grad_norm": 1.53125,
+      "learning_rate": 9.978823186324921e-05,
+      "loss": 8.9423,
+      "step": 7730
+    },
+    {
+      "epoch": 0.37969793410474806,
+      "grad_norm": 0.9921875,
+      "learning_rate": 9.978559984048224e-05,
+      "loss": 9.62,
+      "step": 7740
+    },
+    {
+      "epoch": 0.38018849991108494,
+      "grad_norm": 1.421875,
+      "learning_rate": 9.978295159728357e-05,
+      "loss": 9.4276,
+      "step": 7750
+    },
+    {
+      "epoch": 0.3806790657174218,
+      "grad_norm": 1.4140625,
+      "learning_rate": 9.978028713451602e-05,
+      "loss": 9.5612,
+      "step": 7760
+    },
+    {
+      "epoch": 0.3811696315237587,
+      "grad_norm": 1.0859375,
+      "learning_rate": 9.97776064530477e-05,
+      "loss": 9.5241,
+      "step": 7770
+    },
+    {
+      "epoch": 0.3816601973300956,
+      "grad_norm": 1.4140625,
+      "learning_rate": 9.977490955375195e-05,
+      "loss": 9.4522,
+      "step": 7780
+    },
+    {
+      "epoch": 0.3821507631364325,
+      "grad_norm": 1.0390625,
+      "learning_rate": 9.977219643750747e-05,
+      "loss": 9.0719,
+      "step": 7790
+    },
+    {
+      "epoch": 0.38264132894276937,
+      "grad_norm": 0.98046875,
+      "learning_rate": 9.976946710519823e-05,
+      "loss": 9.0643,
+      "step": 7800
+    },
+    {
+      "epoch": 0.38313189474910625,
+      "grad_norm": 1.3984375,
+      "learning_rate": 9.976672155771342e-05,
+      "loss": 9.7797,
+      "step": 7810
+    },
+    {
+      "epoch": 0.38362246055544313,
+      "grad_norm": 1.0546875,
+      "learning_rate": 9.976395979594755e-05,
+      "loss": 9.2137,
+      "step": 7820
+    },
+    {
+      "epoch": 0.38411302636178,
+      "grad_norm": 1.375,
+      "learning_rate": 9.976118182080045e-05,
+      "loss": 9.1207,
+      "step": 7830
+    },
+    {
+      "epoch": 0.3846035921681169,
+      "grad_norm": 1.578125,
+      "learning_rate": 9.97583876331772e-05,
+      "loss": 9.828,
+      "step": 7840
+    },
+    {
+      "epoch": 0.3850941579744538,
+      "grad_norm": 1.265625,
+      "learning_rate": 9.975557723398814e-05,
+      "loss": 9.2545,
+      "step": 7850
+    },
+    {
+      "epoch": 0.3855847237807907,
+      "grad_norm": 96.0,
+      "learning_rate": 9.975275062414891e-05,
+      "loss": 9.3358,
+      "step": 7860
+    },
+    {
+      "epoch": 0.38607528958712756,
+      "grad_norm": 0.88671875,
+      "learning_rate": 9.974990780458047e-05,
+      "loss": 9.3523,
+      "step": 7870
+    },
+    {
+      "epoch": 0.38656585539346444,
+      "grad_norm": 1.0625,
+      "learning_rate": 9.974704877620899e-05,
+      "loss": 8.9798,
+      "step": 7880
+    },
+    {
+      "epoch": 0.3870564211998013,
+      "grad_norm": 1.8203125,
+      "learning_rate": 9.974417353996597e-05,
+      "loss": 9.2332,
+      "step": 7890
+    },
+    {
+      "epoch": 0.3875469870061382,
+      "grad_norm": 1.828125,
+      "learning_rate": 9.97412820967882e-05,
+      "loss": 9.6917,
+      "step": 7900
+    },
+    {
+      "epoch": 0.3880375528124751,
+      "grad_norm": 1.8125,
+      "learning_rate": 9.973837444761767e-05,
+      "loss": 10.2114,
+      "step": 7910
+    },
+    {
+      "epoch": 0.388528118618812,
+      "grad_norm": 1.1328125,
+      "learning_rate": 9.973545059340179e-05,
+      "loss": 8.8373,
+      "step": 7920
+    },
+    {
+      "epoch": 0.38901868442514886,
+      "grad_norm": 2.125,
+      "learning_rate": 9.973251053509311e-05,
+      "loss": 9.2673,
+      "step": 7930
+    },
+    {
+      "epoch": 0.38950925023148575,
+      "grad_norm": 1.4609375,
+      "learning_rate": 9.972955427364953e-05,
+      "loss": 9.5442,
+      "step": 7940
+    },
+    {
+      "epoch": 0.38999981603782263,
+      "grad_norm": 1.2421875,
+      "learning_rate": 9.972658181003422e-05,
+      "loss": 9.3631,
+      "step": 7950
+    },
+    {
+      "epoch": 0.3904903818441595,
+      "grad_norm": 1.46875,
+      "learning_rate": 9.972359314521565e-05,
+      "loss": 9.2575,
+      "step": 7960
+    },
+    {
+      "epoch": 0.3909809476504964,
+      "grad_norm": 1.4453125,
+      "learning_rate": 9.972058828016749e-05,
+      "loss": 9.5496,
+      "step": 7970
+    },
+    {
+      "epoch": 0.3914715134568333,
+      "grad_norm": 2.21875,
+      "learning_rate": 9.971756721586879e-05,
+      "loss": 9.5364,
+      "step": 7980
+    },
+    {
+      "epoch": 0.39196207926317017,
+      "grad_norm": 1.3359375,
+      "learning_rate": 9.971452995330382e-05,
+      "loss": 9.7381,
+      "step": 7990
+    },
+    {
+      "epoch": 0.39245264506950706,
+      "grad_norm": 1.328125,
+      "learning_rate": 9.971147649346211e-05,
+      "loss": 9.1089,
+      "step": 8000
+    },
+    {
+      "epoch": 0.39294321087584394,
+      "grad_norm": 1.0078125,
+      "learning_rate": 9.970840683733853e-05,
+      "loss": 9.3147,
+      "step": 8010
+    },
+    {
+      "epoch": 0.3934337766821808,
+      "grad_norm": 1.2421875,
+      "learning_rate": 9.970532098593317e-05,
+      "loss": 9.3678,
+      "step": 8020
+    },
+    {
+      "epoch": 0.3939243424885177,
+      "grad_norm": 1.453125,
+      "learning_rate": 9.970221894025143e-05,
+      "loss": 8.9748,
+      "step": 8030
+    },
+    {
+      "epoch": 0.3944149082948546,
+      "grad_norm": 0.92578125,
+      "learning_rate": 9.969910070130395e-05,
+      "loss": 9.4,
+      "step": 8040
+    },
+    {
+      "epoch": 0.3949054741011915,
+      "grad_norm": 0.76171875,
+      "learning_rate": 9.969596627010671e-05,
+      "loss": 9.2384,
+      "step": 8050
+    },
+    {
+      "epoch": 0.39539603990752836,
+      "grad_norm": 1.53125,
+      "learning_rate": 9.969281564768089e-05,
+      "loss": 9.7778,
+      "step": 8060
+    },
+    {
+      "epoch": 0.39588660571386525,
+      "grad_norm": 1.0234375,
+      "learning_rate": 9.9689648835053e-05,
+      "loss": 9.2588,
+      "step": 8070
+    },
+    {
+      "epoch": 0.39637717152020213,
+      "grad_norm": 1.5078125,
+      "learning_rate": 9.96864658332548e-05,
+      "loss": 9.7844,
+      "step": 8080
+    },
+    {
+      "epoch": 0.396867737326539,
+      "grad_norm": 1.09375,
+      "learning_rate": 9.968326664332333e-05,
+      "loss": 8.7478,
+      "step": 8090
+    },
+    {
+      "epoch": 0.3973583031328759,
+      "grad_norm": 2.40625,
+      "learning_rate": 9.96800512663009e-05,
+      "loss": 10.1442,
+      "step": 8100
+    },
+    {
+      "epoch": 0.3978488689392128,
+      "grad_norm": 1.40625,
+      "learning_rate": 9.967681970323512e-05,
+      "loss": 9.3243,
+      "step": 8110
+    },
+    {
+      "epoch": 0.39833943474554967,
+      "grad_norm": 1.3359375,
+      "learning_rate": 9.967357195517881e-05,
+      "loss": 9.4967,
+      "step": 8120
+    },
+    {
+      "epoch": 0.39883000055188655,
+      "grad_norm": 1.453125,
+      "learning_rate": 9.967030802319015e-05,
+      "loss": 9.3689,
+      "step": 8130
+    },
+    {
+      "epoch": 0.39932056635822344,
+      "grad_norm": 1.2578125,
+      "learning_rate": 9.966702790833253e-05,
+      "loss": 9.0353,
+      "step": 8140
+    },
+    {
+      "epoch": 0.3998111321645603,
+      "grad_norm": 0.76171875,
+      "learning_rate": 9.96637316116746e-05,
+      "loss": 9.3215,
+      "step": 8150
+    },
+    {
+      "epoch": 0.4003016979708972,
+      "grad_norm": 1.15625,
+      "learning_rate": 9.966041913429036e-05,
+      "loss": 10.0265,
+      "step": 8160
+    },
+    {
+      "epoch": 0.4007922637772341,
+      "grad_norm": 1.1640625,
+      "learning_rate": 9.965709047725901e-05,
+      "loss": 8.743,
+      "step": 8170
+    },
+    {
+      "epoch": 0.401282829583571,
+      "grad_norm": 0.82421875,
+      "learning_rate": 9.965374564166505e-05,
+      "loss": 8.9593,
+      "step": 8180
+    },
+    {
+      "epoch": 0.40177339538990786,
+      "grad_norm": 1.0390625,
+      "learning_rate": 9.965038462859824e-05,
+      "loss": 8.961,
+      "step": 8190
+    },
+    {
+      "epoch": 0.40226396119624475,
+      "grad_norm": 1.2890625,
+      "learning_rate": 9.964700743915361e-05,
+      "loss": 9.808,
+      "step": 8200
+    },
+    {
+      "epoch": 0.4027545270025816,
+      "grad_norm": 8.375,
+      "learning_rate": 9.96436140744315e-05,
+      "loss": 9.4756,
+      "step": 8210
+    },
+    {
+      "epoch": 0.40324509280891846,
+      "grad_norm": 0.99609375,
+      "learning_rate": 9.964020453553746e-05,
+      "loss": 9.3352,
+      "step": 8220
+    },
+    {
+      "epoch": 0.40373565861525534,
+      "grad_norm": 1.7421875,
+      "learning_rate": 9.963677882358233e-05,
+      "loss": 9.3511,
+      "step": 8230
+    },
+    {
+      "epoch": 0.40422622442159223,
+      "grad_norm": 1.203125,
+      "learning_rate": 9.963333693968226e-05,
+      "loss": 9.5166,
+      "step": 8240
+    },
+    {
+      "epoch": 0.4047167902279291,
+      "grad_norm": 1.53125,
+      "learning_rate": 9.962987888495862e-05,
+      "loss": 9.3433,
+      "step": 8250
+    },
+    {
+      "epoch": 0.405207356034266,
+      "grad_norm": 2.59375,
+      "learning_rate": 9.962640466053804e-05,
+      "loss": 9.9907,
+      "step": 8260
+    },
+    {
+      "epoch": 0.4056979218406029,
+      "grad_norm": 1.625,
+      "learning_rate": 9.962291426755248e-05,
+      "loss": 9.5791,
+      "step": 8270
+    },
+    {
+      "epoch": 0.40618848764693977,
+      "grad_norm": 1.125,
+      "learning_rate": 9.96194077071391e-05,
+      "loss": 9.2634,
+      "step": 8280
+    },
+    {
+      "epoch": 0.40667905345327665,
+      "grad_norm": 1.1484375,
+      "learning_rate": 9.961588498044037e-05,
+      "loss": 9.2259,
+      "step": 8290
+    },
+    {
+      "epoch": 0.40716961925961354,
+      "grad_norm": 1.6875,
+      "learning_rate": 9.961234608860402e-05,
+      "loss": 9.1666,
+      "step": 8300
+    },
+    {
+      "epoch": 0.4076601850659504,
+      "grad_norm": 1.3359375,
+      "learning_rate": 9.960879103278303e-05,
+      "loss": 9.4266,
+      "step": 8310
+    },
+    {
+      "epoch": 0.4081507508722873,
+      "grad_norm": 1.0703125,
+      "learning_rate": 9.960521981413566e-05,
+      "loss": 9.2596,
+      "step": 8320
+    },
+    {
+      "epoch": 0.4086413166786242,
+      "grad_norm": 1.9296875,
+      "learning_rate": 9.960163243382545e-05,
+      "loss": 8.4992,
+      "step": 8330
+    },
+    {
+      "epoch": 0.4091318824849611,
+      "grad_norm": 1.5234375,
+      "learning_rate": 9.959802889302117e-05,
+      "loss": 9.1112,
+      "step": 8340
+    },
+    {
+      "epoch": 0.40962244829129796,
+      "grad_norm": 1.421875,
+      "learning_rate": 9.959440919289686e-05,
+      "loss": 9.6092,
+      "step": 8350
+    },
+    {
+      "epoch": 0.41011301409763484,
+      "grad_norm": 2.09375,
+      "learning_rate": 9.959077333463187e-05,
+      "loss": 9.1144,
+      "step": 8360
+    },
+    {
+      "epoch": 0.4106035799039717,
+      "grad_norm": 1.265625,
+      "learning_rate": 9.958712131941077e-05,
+      "loss": 9.007,
+      "step": 8370
+    },
+    {
+      "epoch": 0.4110941457103086,
+      "grad_norm": 1.1328125,
+      "learning_rate": 9.95834531484234e-05,
+      "loss": 9.4561,
+      "step": 8380
+    },
+    {
+      "epoch": 0.4115847115166455,
+      "grad_norm": 1.6484375,
+      "learning_rate": 9.95797688228649e-05,
+      "loss": 9.0578,
+      "step": 8390
+    },
+    {
+      "epoch": 0.4120752773229824,
+      "grad_norm": 1.5703125,
+      "learning_rate": 9.957606834393561e-05,
+      "loss": 9.0047,
+      "step": 8400
+    },
+    {
+      "epoch": 0.41256584312931927,
+      "grad_norm": 1.1328125,
+      "learning_rate": 9.957235171284118e-05,
+      "loss": 9.0116,
+      "step": 8410
+    },
+    {
+      "epoch": 0.41305640893565615,
+      "grad_norm": 1.09375,
+      "learning_rate": 9.956861893079253e-05,
+      "loss": 9.5539,
+      "step": 8420
+    },
+    {
+      "epoch": 0.41354697474199303,
+      "grad_norm": 2.375,
+      "learning_rate": 9.956486999900578e-05,
+      "loss": 9.4898,
+      "step": 8430
+    },
+    {
+      "epoch": 0.4140375405483299,
+      "grad_norm": 1.0703125,
+      "learning_rate": 9.956110491870237e-05,
+      "loss": 9.4607,
+      "step": 8440
+    },
+    {
+      "epoch": 0.4145281063546668,
+      "grad_norm": 0.9765625,
+      "learning_rate": 9.9557323691109e-05,
+      "loss": 9.0116,
+      "step": 8450
+    },
+    {
+      "epoch": 0.4150186721610037,
+      "grad_norm": 1.0625,
+      "learning_rate": 9.955352631745761e-05,
+      "loss": 8.9376,
+      "step": 8460
+    },
+    {
+      "epoch": 0.4155092379673406,
+      "grad_norm": 0.84375,
+      "learning_rate": 9.954971279898538e-05,
+      "loss": 9.4654,
+      "step": 8470
+    },
+    {
+      "epoch": 0.41599980377367746,
+      "grad_norm": 1.65625,
+      "learning_rate": 9.954588313693482e-05,
+      "loss": 9.6625,
+      "step": 8480
+    },
+    {
+      "epoch": 0.41649036958001434,
+      "grad_norm": 1.84375,
+      "learning_rate": 9.954203733255362e-05,
+      "loss": 9.5381,
+      "step": 8490
+    },
+    {
+      "epoch": 0.4169809353863512,
+      "grad_norm": 5.375,
+      "learning_rate": 9.953817538709478e-05,
+      "loss": 9.4922,
+      "step": 8500
+    },
+    {
+      "epoch": 0.4174715011926881,
+      "grad_norm": 0.9921875,
+      "learning_rate": 9.953429730181653e-05,
+      "loss": 9.4195,
+      "step": 8510
+    },
+    {
+      "epoch": 0.417962066999025,
+      "grad_norm": 1.71875,
+      "learning_rate": 9.95304030779824e-05,
+      "loss": 9.0456,
+      "step": 8520
+    },
+    {
+      "epoch": 0.4184526328053619,
+      "grad_norm": 1.3671875,
+      "learning_rate": 9.952649271686114e-05,
+      "loss": 9.5672,
+      "step": 8530
+    },
+    {
+      "epoch": 0.41894319861169876,
+      "grad_norm": 0.83203125,
+      "learning_rate": 9.952256621972676e-05,
+      "loss": 8.7467,
+      "step": 8540
+    },
+    {
+      "epoch": 0.41943376441803565,
+      "grad_norm": 1.515625,
+      "learning_rate": 9.951862358785852e-05,
+      "loss": 9.6638,
+      "step": 8550
+    },
+    {
+      "epoch": 0.41992433022437253,
+      "grad_norm": 1.28125,
+      "learning_rate": 9.951466482254097e-05,
+      "loss": 9.5221,
+      "step": 8560
+    },
+    {
+      "epoch": 0.4204148960307094,
+      "grad_norm": 1.6015625,
+      "learning_rate": 9.951068992506391e-05,
+      "loss": 9.3728,
+      "step": 8570
+    },
+    {
+      "epoch": 0.4209054618370463,
+      "grad_norm": 1.6640625,
+      "learning_rate": 9.950669889672238e-05,
+      "loss": 9.0309,
+      "step": 8580
+    },
+    {
+      "epoch": 0.4213960276433832,
+      "grad_norm": 0.77734375,
+      "learning_rate": 9.950269173881663e-05,
+      "loss": 9.546,
+      "step": 8590
+    },
+    {
+      "epoch": 0.42188659344972007,
+      "grad_norm": 1.390625,
+      "learning_rate": 9.94986684526523e-05,
+      "loss": 9.3854,
+      "step": 8600
+    },
+    {
+      "epoch": 0.42237715925605696,
+      "grad_norm": 1.734375,
+      "learning_rate": 9.949462903954014e-05,
+      "loss": 10.0409,
+      "step": 8610
+    },
+    {
+      "epoch": 0.42286772506239384,
+      "grad_norm": 1.1875,
+      "learning_rate": 9.949057350079622e-05,
+      "loss": 10.0401,
+      "step": 8620
+    },
+    {
+      "epoch": 0.4233582908687307,
+      "grad_norm": 1.171875,
+      "learning_rate": 9.948650183774187e-05,
+      "loss": 9.3531,
+      "step": 8630
+    },
+    {
+      "epoch": 0.4238488566750676,
+      "grad_norm": 1.046875,
+      "learning_rate": 9.948241405170367e-05,
+      "loss": 9.6503,
+      "step": 8640
+    },
+    {
+      "epoch": 0.4243394224814045,
+      "grad_norm": 1.03125,
+      "learning_rate": 9.947831014401342e-05,
+      "loss": 10.1025,
+      "step": 8650
+    },
+    {
+      "epoch": 0.4248299882877414,
+      "grad_norm": 1.8515625,
+      "learning_rate": 9.94741901160082e-05,
+      "loss": 9.1217,
+      "step": 8660
+    },
+    {
+      "epoch": 0.42532055409407826,
+      "grad_norm": 1.296875,
+      "learning_rate": 9.947005396903036e-05,
+      "loss": 9.4203,
+      "step": 8670
+    },
+    {
+      "epoch": 0.42581111990041515,
+      "grad_norm": 1.3046875,
+      "learning_rate": 9.946590170442747e-05,
+      "loss": 8.9508,
+      "step": 8680
+    },
+    {
+      "epoch": 0.42630168570675203,
+      "grad_norm": 1.78125,
+      "learning_rate": 9.946173332355236e-05,
+      "loss": 9.3826,
+      "step": 8690
+    },
+    {
+      "epoch": 0.4267922515130889,
+      "grad_norm": 1.1796875,
+      "learning_rate": 9.945754882776311e-05,
+      "loss": 9.6903,
+      "step": 8700
+    },
+    {
+      "epoch": 0.4272828173194258,
+      "grad_norm": 0.984375,
+      "learning_rate": 9.945334821842303e-05,
+      "loss": 9.1157,
+      "step": 8710
+    },
+    {
+      "epoch": 0.4277733831257627,
+      "grad_norm": 3.171875,
+      "learning_rate": 9.944913149690075e-05,
+      "loss": 9.7155,
+      "step": 8720
+    },
+    {
+      "epoch": 0.42826394893209957,
+      "grad_norm": 1.421875,
+      "learning_rate": 9.944489866457008e-05,
+      "loss": 9.4667,
+      "step": 8730
+    },
+    {
+      "epoch": 0.42875451473843645,
+      "grad_norm": 1.53125,
+      "learning_rate": 9.944064972281007e-05,
+      "loss": 9.5808,
+      "step": 8740
+    },
+    {
+      "epoch": 0.42924508054477334,
+      "grad_norm": 1.1328125,
+      "learning_rate": 9.943638467300511e-05,
+      "loss": 9.3587,
+      "step": 8750
+    },
+    {
+      "epoch": 0.4297356463511102,
+      "grad_norm": 2.921875,
+      "learning_rate": 9.943210351654473e-05,
+      "loss": 9.6425,
+      "step": 8760
+    },
+    {
+      "epoch": 0.4302262121574471,
+      "grad_norm": 1.40625,
+      "learning_rate": 9.942780625482376e-05,
+      "loss": 9.6145,
+      "step": 8770
+    },
+    {
+      "epoch": 0.430716777963784,
+      "grad_norm": 1.703125,
+      "learning_rate": 9.942349288924229e-05,
+      "loss": 9.6646,
+      "step": 8780
+    },
+    {
+      "epoch": 0.4312073437701209,
+      "grad_norm": 1.546875,
+      "learning_rate": 9.941916342120564e-05,
+      "loss": 9.7029,
+      "step": 8790
+    },
+    {
+      "epoch": 0.43169790957645776,
+      "grad_norm": 1.71875,
+      "learning_rate": 9.941481785212437e-05,
+      "loss": 9.1658,
+      "step": 8800
+    },
+    {
+      "epoch": 0.43218847538279465,
+      "grad_norm": 1.09375,
+      "learning_rate": 9.941045618341427e-05,
+      "loss": 9.9594,
+      "step": 8810
+    },
+    {
+      "epoch": 0.43267904118913153,
+      "grad_norm": 1.0390625,
+      "learning_rate": 9.940607841649643e-05,
+      "loss": 9.3896,
+      "step": 8820
+    },
+    {
+      "epoch": 0.4331696069954684,
+      "grad_norm": 1.4375,
+      "learning_rate": 9.940168455279713e-05,
+      "loss": 8.9524,
+      "step": 8830
+    },
+    {
+      "epoch": 0.4336601728018053,
+      "grad_norm": 1.203125,
+      "learning_rate": 9.939727459374792e-05,
+      "loss": 9.8094,
+      "step": 8840
+    },
+    {
+      "epoch": 0.4341507386081422,
+      "grad_norm": 2.734375,
+      "learning_rate": 9.93928485407856e-05,
+      "loss": 9.6122,
+      "step": 8850
+    },
+    {
+      "epoch": 0.43464130441447907,
+      "grad_norm": 0.88671875,
+      "learning_rate": 9.938840639535219e-05,
+      "loss": 9.5027,
+      "step": 8860
+    },
+    {
+      "epoch": 0.43513187022081595,
+      "grad_norm": 1.1015625,
+      "learning_rate": 9.938394815889497e-05,
+      "loss": 8.9665,
+      "step": 8870
+    },
+    {
+      "epoch": 0.43562243602715284,
+      "grad_norm": 0.96484375,
+      "learning_rate": 9.937947383286646e-05,
+      "loss": 9.6251,
+      "step": 8880
+    },
+    {
+      "epoch": 0.4361130018334897,
+      "grad_norm": 1.7265625,
+      "learning_rate": 9.937498341872443e-05,
+      "loss": 10.0964,
+      "step": 8890
+    },
+    {
+      "epoch": 0.4366035676398266,
+      "grad_norm": 0.91796875,
+      "learning_rate": 9.937047691793186e-05,
+      "loss": 9.6408,
+      "step": 8900
+    },
+    {
+      "epoch": 0.4370941334461635,
+      "grad_norm": 121.5,
+      "learning_rate": 9.936595433195701e-05,
+      "loss": 9.911,
+      "step": 8910
+    },
+    {
+      "epoch": 0.4375846992525004,
+      "grad_norm": 1.328125,
+      "learning_rate": 9.936141566227335e-05,
+      "loss": 9.3583,
+      "step": 8920
+    },
+    {
+      "epoch": 0.43807526505883726,
+      "grad_norm": 1.1171875,
+      "learning_rate": 9.935686091035963e-05,
+      "loss": 9.074,
+      "step": 8930
+    },
+    {
+      "epoch": 0.43856583086517414,
+      "grad_norm": 1.4765625,
+      "learning_rate": 9.93522900776998e-05,
+      "loss": 9.409,
+      "step": 8940
+    },
+    {
+      "epoch": 0.43905639667151103,
+      "grad_norm": 1.3515625,
+      "learning_rate": 9.934770316578306e-05,
+      "loss": 9.5203,
+      "step": 8950
+    },
+    {
+      "epoch": 0.4395469624778479,
+      "grad_norm": 1.2421875,
+      "learning_rate": 9.934310017610385e-05,
+      "loss": 9.6801,
+      "step": 8960
+    },
+    {
+      "epoch": 0.4400375282841848,
+      "grad_norm": 1.1015625,
+      "learning_rate": 9.933848111016186e-05,
+      "loss": 9.5165,
+      "step": 8970
+    },
+    {
+      "epoch": 0.4405280940905217,
+      "grad_norm": 1.5234375,
+      "learning_rate": 9.933384596946201e-05,
+      "loss": 9.6044,
+      "step": 8980
+    },
+    {
+      "epoch": 0.44101865989685857,
+      "grad_norm": 1.1875,
+      "learning_rate": 9.932919475551443e-05,
+      "loss": 9.2709,
+      "step": 8990
+    },
+    {
+      "epoch": 0.4415092257031954,
+      "grad_norm": 1.234375,
+      "learning_rate": 9.932452746983455e-05,
+      "loss": 9.0887,
+      "step": 9000
+    },
+    {
+      "epoch": 0.4419997915095323,
+      "grad_norm": 0.94140625,
+      "learning_rate": 9.931984411394297e-05,
+      "loss": 8.9249,
+      "step": 9010
+    },
+    {
+      "epoch": 0.44249035731586916,
+      "grad_norm": 1.234375,
+      "learning_rate": 9.931514468936556e-05,
+      "loss": 9.3311,
+      "step": 9020
+    },
+    {
+      "epoch": 0.44298092312220605,
+      "grad_norm": 1.5625,
+      "learning_rate": 9.931042919763343e-05,
+      "loss": 9.1204,
+      "step": 9030
+    },
+    {
+      "epoch": 0.44347148892854293,
+      "grad_norm": 1.0078125,
+      "learning_rate": 9.930569764028289e-05,
+      "loss": 8.936,
+      "step": 9040
+    },
+    {
+      "epoch": 0.4439620547348798,
+      "grad_norm": 1.5078125,
+      "learning_rate": 9.930095001885554e-05,
+      "loss": 9.8463,
+      "step": 9050
+    },
+    {
+      "epoch": 0.4444526205412167,
+      "grad_norm": 1.3125,
+      "learning_rate": 9.929618633489815e-05,
+      "loss": 9.4598,
+      "step": 9060
+    },
+    {
+      "epoch": 0.4449431863475536,
+      "grad_norm": 1.390625,
+      "learning_rate": 9.929140658996278e-05,
+      "loss": 9.1565,
+      "step": 9070
+    },
+    {
+      "epoch": 0.44543375215389047,
+      "grad_norm": 1.0625,
+      "learning_rate": 9.928661078560669e-05,
+      "loss": 9.5713,
+      "step": 9080
+    },
+    {
+      "epoch": 0.44592431796022736,
+      "grad_norm": 1.265625,
+      "learning_rate": 9.928179892339238e-05,
+      "loss": 9.5154,
+      "step": 9090
+    },
+    {
+      "epoch": 0.44641488376656424,
+      "grad_norm": 2.671875,
+      "learning_rate": 9.927697100488757e-05,
+      "loss": 9.8792,
+      "step": 9100
+    },
+    {
+      "epoch": 0.4469054495729011,
+      "grad_norm": 1.1640625,
+      "learning_rate": 9.927212703166526e-05,
+      "loss": 8.8977,
+      "step": 9110
+    },
+    {
+      "epoch": 0.447396015379238,
+      "grad_norm": 1.2109375,
+      "learning_rate": 9.92672670053036e-05,
+      "loss": 9.3305,
+      "step": 9120
+    },
+    {
+      "epoch": 0.4478865811855749,
+      "grad_norm": 1.5390625,
+      "learning_rate": 9.926239092738606e-05,
+      "loss": 9.4327,
+      "step": 9130
+    },
+    {
+      "epoch": 0.4483771469919118,
+      "grad_norm": 1.0859375,
+      "learning_rate": 9.925749879950123e-05,
+      "loss": 9.3707,
+      "step": 9140
+    },
+    {
+      "epoch": 0.44886771279824866,
+      "grad_norm": 3.578125,
+      "learning_rate": 9.925259062324305e-05,
+      "loss": 9.2381,
+      "step": 9150
+    },
+    {
+      "epoch": 0.44935827860458555,
+      "grad_norm": 6.03125,
+      "learning_rate": 9.924766640021061e-05,
+      "loss": 8.6538,
+      "step": 9160
+    },
+    {
+      "epoch": 0.44984884441092243,
+      "grad_norm": 2.25,
+      "learning_rate": 9.924272613200825e-05,
+      "loss": 9.581,
+      "step": 9170
+    },
+    {
+      "epoch": 0.4503394102172593,
+      "grad_norm": 1.9921875,
+      "learning_rate": 9.923776982024554e-05,
+      "loss": 9.4617,
+      "step": 9180
+    },
+    {
+      "epoch": 0.4508299760235962,
+      "grad_norm": 1.59375,
+      "learning_rate": 9.923279746653728e-05,
+      "loss": 9.5088,
+      "step": 9190
+    },
+    {
+      "epoch": 0.4513205418299331,
+      "grad_norm": 1.3125,
+      "learning_rate": 9.922780907250348e-05,
+      "loss": 8.9247,
+      "step": 9200
+    },
+    {
+      "epoch": 0.45181110763626997,
+      "grad_norm": 1.4921875,
+      "learning_rate": 9.922280463976938e-05,
+      "loss": 9.1817,
+      "step": 9210
+    },
+    {
+      "epoch": 0.45230167344260686,
+      "grad_norm": 1.6953125,
+      "learning_rate": 9.921778416996549e-05,
+      "loss": 9.3479,
+      "step": 9220
+    },
+    {
+      "epoch": 0.45279223924894374,
+      "grad_norm": 1.3125,
+      "learning_rate": 9.921274766472748e-05,
+      "loss": 10.0808,
+      "step": 9230
+    },
+    {
+      "epoch": 0.4532828050552806,
+      "grad_norm": 1.2421875,
+      "learning_rate": 9.920769512569625e-05,
+      "loss": 9.1088,
+      "step": 9240
+    },
+    {
+      "epoch": 0.4537733708616175,
+      "grad_norm": 1.171875,
+      "learning_rate": 9.9202626554518e-05,
+      "loss": 10.0015,
+      "step": 9250
+    },
+    {
+      "epoch": 0.4542639366679544,
+      "grad_norm": 1.8046875,
+      "learning_rate": 9.919754195284406e-05,
+      "loss": 10.0798,
+      "step": 9260
+    },
+    {
+      "epoch": 0.4547545024742913,
+      "grad_norm": 1.09375,
+      "learning_rate": 9.919244132233104e-05,
+      "loss": 9.0094,
+      "step": 9270
+    },
+    {
+      "epoch": 0.45524506828062816,
+      "grad_norm": 2.640625,
+      "learning_rate": 9.918732466464072e-05,
+      "loss": 9.6669,
+      "step": 9280
+    },
+    {
+      "epoch": 0.45573563408696505,
+      "grad_norm": 1.484375,
+      "learning_rate": 9.918219198144019e-05,
+      "loss": 9.5382,
+      "step": 9290
+    },
+    {
+      "epoch": 0.45622619989330193,
+      "grad_norm": 1.421875,
+      "learning_rate": 9.917704327440166e-05,
+      "loss": 9.4784,
+      "step": 9300
+    },
+    {
+      "epoch": 0.4567167656996388,
+      "grad_norm": 1.1484375,
+      "learning_rate": 9.917187854520264e-05,
+      "loss": 10.2777,
+      "step": 9310
+    },
+    {
+      "epoch": 0.4572073315059757,
+      "grad_norm": 1.1953125,
+      "learning_rate": 9.916669779552581e-05,
+      "loss": 9.0473,
+      "step": 9320
+    },
+    {
+      "epoch": 0.4576978973123126,
+      "grad_norm": 1.46875,
+      "learning_rate": 9.91615010270591e-05,
+      "loss": 9.1827,
+      "step": 9330
+    },
+    {
+      "epoch": 0.45818846311864947,
+      "grad_norm": 1.8828125,
+      "learning_rate": 9.915628824149564e-05,
+      "loss": 9.3418,
+      "step": 9340
+    },
+    {
+      "epoch": 0.45867902892498635,
+      "grad_norm": 1.0546875,
+      "learning_rate": 9.91510594405338e-05,
+      "loss": 9.5231,
+      "step": 9350
+    },
+    {
+      "epoch": 0.45916959473132324,
+      "grad_norm": 0.93359375,
+      "learning_rate": 9.914581462587712e-05,
+      "loss": 9.2647,
+      "step": 9360
+    },
+    {
+      "epoch": 0.4596601605376601,
+      "grad_norm": 1.1484375,
+      "learning_rate": 9.914055379923442e-05,
+      "loss": 9.6544,
+      "step": 9370
+    },
+    {
+      "epoch": 0.460150726343997,
+      "grad_norm": 2.015625,
+      "learning_rate": 9.913527696231969e-05,
+      "loss": 9.1867,
+      "step": 9380
+    },
+    {
+      "epoch": 0.4606412921503339,
+      "grad_norm": 1.125,
+      "learning_rate": 9.912998411685216e-05,
+      "loss": 9.5706,
+      "step": 9390
+    },
+    {
+      "epoch": 0.4611318579566708,
+      "grad_norm": 0.9375,
+      "learning_rate": 9.912467526455626e-05,
+      "loss": 9.2712,
+      "step": 9400
+    },
+    {
+      "epoch": 0.46162242376300766,
+      "grad_norm": 1.15625,
+      "learning_rate": 9.911935040716169e-05,
+      "loss": 9.6751,
+      "step": 9410
+    },
+    {
+      "epoch": 0.46211298956934455,
+      "grad_norm": 1.265625,
+      "learning_rate": 9.911400954640325e-05,
+      "loss": 9.2709,
+      "step": 9420
+    },
+    {
+      "epoch": 0.46260355537568143,
+      "grad_norm": 1.6953125,
+      "learning_rate": 9.910865268402106e-05,
+      "loss": 9.6717,
+      "step": 9430
+    },
+    {
+      "epoch": 0.4630941211820183,
+      "grad_norm": 1.0625,
+      "learning_rate": 9.910327982176042e-05,
+      "loss": 8.833,
+      "step": 9440
+    },
+    {
+      "epoch": 0.4635846869883552,
+      "grad_norm": 1.3125,
+      "learning_rate": 9.909789096137184e-05,
+      "loss": 9.7037,
+      "step": 9450
+    },
+    {
+      "epoch": 0.4640752527946921,
+      "grad_norm": 1.21875,
+      "learning_rate": 9.909248610461101e-05,
+      "loss": 9.3282,
+      "step": 9460
+    },
+    {
+      "epoch": 0.46456581860102897,
+      "grad_norm": 1.4921875,
+      "learning_rate": 9.90870652532389e-05,
+      "loss": 9.7322,
+      "step": 9470
+    },
+    {
+      "epoch": 0.46505638440736585,
+      "grad_norm": 2.609375,
+      "learning_rate": 9.908162840902163e-05,
+      "loss": 9.5577,
+      "step": 9480
+    },
+    {
+      "epoch": 0.46554695021370274,
+      "grad_norm": 1.3671875,
+      "learning_rate": 9.907617557373057e-05,
+      "loss": 9.5126,
+      "step": 9490
+    },
+    {
+      "epoch": 0.4660375160200396,
+      "grad_norm": 1.5546875,
+      "learning_rate": 9.907070674914228e-05,
+      "loss": 9.4535,
+      "step": 9500
+    },
+    {
+      "epoch": 0.4665280818263765,
+      "grad_norm": 1.0390625,
+      "learning_rate": 9.906522193703853e-05,
+      "loss": 9.1627,
+      "step": 9510
+    },
+    {
+      "epoch": 0.4670186476327134,
+      "grad_norm": 1.15625,
+      "learning_rate": 9.905972113920632e-05,
+      "loss": 8.9282,
+      "step": 9520
+    },
+    {
+      "epoch": 0.4675092134390503,
+      "grad_norm": 1.8828125,
+      "learning_rate": 9.905420435743782e-05,
+      "loss": 9.3116,
+      "step": 9530
+    },
+    {
+      "epoch": 0.46799977924538716,
+      "grad_norm": 1.1875,
+      "learning_rate": 9.904867159353041e-05,
+      "loss": 9.3445,
+      "step": 9540
+    },
+    {
+      "epoch": 0.46849034505172404,
+      "grad_norm": 1.3984375,
+      "learning_rate": 9.904312284928677e-05,
+      "loss": 8.8062,
+      "step": 9550
+    },
+    {
+      "epoch": 0.46898091085806093,
+      "grad_norm": 1.7109375,
+      "learning_rate": 9.903755812651463e-05,
+      "loss": 9.8019,
+      "step": 9560
+    },
+    {
+      "epoch": 0.4694714766643978,
+      "grad_norm": 1.4140625,
+      "learning_rate": 9.903197742702706e-05,
+      "loss": 9.7276,
+      "step": 9570
+    },
+    {
+      "epoch": 0.4699620424707347,
+      "grad_norm": 1.4140625,
+      "learning_rate": 9.902638075264226e-05,
+      "loss": 8.9828,
+      "step": 9580
+    },
+    {
+      "epoch": 0.4704526082770716,
+      "grad_norm": 1.0234375,
+      "learning_rate": 9.902076810518366e-05,
+      "loss": 9.4312,
+      "step": 9590
+    },
+    {
+      "epoch": 0.47094317408340847,
+      "grad_norm": 1.6875,
+      "learning_rate": 9.90151394864799e-05,
+      "loss": 9.4076,
+      "step": 9600
+    },
+    {
+      "epoch": 0.47143373988974535,
+      "grad_norm": 1.265625,
+      "learning_rate": 9.900949489836482e-05,
+      "loss": 9.443,
+      "step": 9610
+    },
+    {
+      "epoch": 0.47192430569608224,
+      "grad_norm": 0.9765625,
+      "learning_rate": 9.900383434267745e-05,
+      "loss": 9.1485,
+      "step": 9620
+    },
+    {
+      "epoch": 0.4724148715024191,
+      "grad_norm": 1.390625,
+      "learning_rate": 9.899815782126203e-05,
+      "loss": 9.0047,
+      "step": 9630
+    },
+    {
+      "epoch": 0.472905437308756,
+      "grad_norm": 1.2421875,
+      "learning_rate": 9.8992465335968e-05,
+      "loss": 9.5106,
+      "step": 9640
+    },
+    {
+      "epoch": 0.4733960031150929,
+      "grad_norm": 1.2734375,
+      "learning_rate": 9.898675688865004e-05,
+      "loss": 8.8625,
+      "step": 9650
+    },
+    {
+      "epoch": 0.4738865689214298,
+      "grad_norm": 1.578125,
+      "learning_rate": 9.898103248116795e-05,
+      "loss": 9.7804,
+      "step": 9660
+    },
+    {
+      "epoch": 0.47437713472776666,
+      "grad_norm": 1.3203125,
+      "learning_rate": 9.897529211538678e-05,
+      "loss": 9.1752,
+      "step": 9670
+    },
+    {
+      "epoch": 0.47486770053410354,
+      "grad_norm": 1.6015625,
+      "learning_rate": 9.896953579317679e-05,
+      "loss": 9.4863,
+      "step": 9680
+    },
+    {
+      "epoch": 0.4753582663404404,
+      "grad_norm": 0.9296875,
+      "learning_rate": 9.89637635164134e-05,
+      "loss": 9.5613,
+      "step": 9690
+    },
+    {
+      "epoch": 0.4758488321467773,
+      "grad_norm": 1.6796875,
+      "learning_rate": 9.89579752869773e-05,
+      "loss": 9.3063,
+      "step": 9700
+    },
+    {
+      "epoch": 0.4763393979531142,
+      "grad_norm": 1.28125,
+      "learning_rate": 9.895217110675428e-05,
+      "loss": 9.356,
+      "step": 9710
+    },
+    {
+      "epoch": 0.4768299637594511,
+      "grad_norm": 1.4609375,
+      "learning_rate": 9.894635097763538e-05,
+      "loss": 9.4519,
+      "step": 9720
+    },
+    {
+      "epoch": 0.47732052956578797,
+      "grad_norm": 1.1484375,
+      "learning_rate": 9.894051490151686e-05,
+      "loss": 9.8951,
+      "step": 9730
+    },
+    {
+      "epoch": 0.47781109537212485,
+      "grad_norm": 1.703125,
+      "learning_rate": 9.893466288030011e-05,
+      "loss": 9.0165,
+      "step": 9740
+    },
+    {
+      "epoch": 0.47830166117846173,
+      "grad_norm": 1.625,
+      "learning_rate": 9.892879491589179e-05,
+      "loss": 9.2397,
+      "step": 9750
+    },
+    {
+      "epoch": 0.4787922269847986,
+      "grad_norm": 1.875,
+      "learning_rate": 9.892291101020368e-05,
+      "loss": 9.1272,
+      "step": 9760
+    },
+    {
+      "epoch": 0.4792827927911355,
+      "grad_norm": 1.1796875,
+      "learning_rate": 9.891701116515282e-05,
+      "loss": 9.2885,
+      "step": 9770
+    },
+    {
+      "epoch": 0.47977335859747233,
+      "grad_norm": 1.5390625,
+      "learning_rate": 9.89110953826614e-05,
+      "loss": 10.1661,
+      "step": 9780
+    },
+    {
+      "epoch": 0.4802639244038092,
+      "grad_norm": 0.8515625,
+      "learning_rate": 9.890516366465678e-05,
+      "loss": 9.3322,
+      "step": 9790
+    },
+    {
+      "epoch": 0.4807544902101461,
+      "grad_norm": 1.28125,
+      "learning_rate": 9.889921601307161e-05,
+      "loss": 9.2185,
+      "step": 9800
+    },
+    {
+      "epoch": 0.481245056016483,
+      "grad_norm": 1.046875,
+      "learning_rate": 9.889325242984365e-05,
+      "loss": 9.7809,
+      "step": 9810
+    },
+    {
+      "epoch": 0.48173562182281987,
+      "grad_norm": 0.9765625,
+      "learning_rate": 9.888727291691584e-05,
+      "loss": 9.2243,
+      "step": 9820
+    },
+    {
+      "epoch": 0.48222618762915676,
+      "grad_norm": 1.6875,
+      "learning_rate": 9.888127747623637e-05,
+      "loss": 9.6819,
+      "step": 9830
+    },
+    {
+      "epoch": 0.48271675343549364,
+      "grad_norm": 1.7578125,
+      "learning_rate": 9.887526610975857e-05,
+      "loss": 9.172,
+      "step": 9840
+    },
+    {
+      "epoch": 0.4832073192418305,
+      "grad_norm": 1.1171875,
+      "learning_rate": 9.8869238819441e-05,
+      "loss": 9.2786,
+      "step": 9850
+    },
+    {
+      "epoch": 0.4836978850481674,
+      "grad_norm": 1.9296875,
+      "learning_rate": 9.886319560724735e-05,
+      "loss": 9.4504,
+      "step": 9860
+    },
+    {
+      "epoch": 0.4841884508545043,
+      "grad_norm": 1.6640625,
+      "learning_rate": 9.885713647514658e-05,
+      "loss": 10.0407,
+      "step": 9870
+    },
+    {
+      "epoch": 0.4846790166608412,
+      "grad_norm": 0.9765625,
+      "learning_rate": 9.885106142511275e-05,
+      "loss": 8.9902,
+      "step": 9880
+    },
+    {
+      "epoch": 0.48516958246717806,
+      "grad_norm": 1.0234375,
+      "learning_rate": 9.884497045912515e-05,
+      "loss": 8.8224,
+      "step": 9890
+    },
+    {
+      "epoch": 0.48566014827351495,
+      "grad_norm": 1.4453125,
+      "learning_rate": 9.883886357916828e-05,
+      "loss": 9.893,
+      "step": 9900
+    },
+    {
+      "epoch": 0.48615071407985183,
+      "grad_norm": 1.2890625,
+      "learning_rate": 9.883274078723177e-05,
+      "loss": 9.0176,
+      "step": 9910
+    },
+    {
+      "epoch": 0.4866412798861887,
+      "grad_norm": 3.359375,
+      "learning_rate": 9.882660208531046e-05,
+      "loss": 9.8512,
+      "step": 9920
+    },
+    {
+      "epoch": 0.4871318456925256,
+      "grad_norm": 1.1640625,
+      "learning_rate": 9.882044747540439e-05,
+      "loss": 9.3034,
+      "step": 9930
+    },
+    {
+      "epoch": 0.4876224114988625,
+      "grad_norm": 1.265625,
+      "learning_rate": 9.881427695951875e-05,
+      "loss": 9.5191,
+      "step": 9940
+    },
+    {
+      "epoch": 0.48811297730519937,
+      "grad_norm": 1.1796875,
+      "learning_rate": 9.880809053966395e-05,
+      "loss": 9.4207,
+      "step": 9950
+    },
+    {
+      "epoch": 0.48860354311153625,
+      "grad_norm": 1.3828125,
+      "learning_rate": 9.880188821785554e-05,
+      "loss": 9.2694,
+      "step": 9960
+    },
+    {
+      "epoch": 0.48909410891787314,
+      "grad_norm": 1.609375,
+      "learning_rate": 9.879566999611429e-05,
+      "loss": 9.7303,
+      "step": 9970
+    },
+    {
+      "epoch": 0.48958467472421,
+      "grad_norm": 1.3515625,
+      "learning_rate": 9.878943587646611e-05,
+      "loss": 9.449,
+      "step": 9980
+    },
+    {
+      "epoch": 0.4900752405305469,
+      "grad_norm": 1.6484375,
+      "learning_rate": 9.878318586094213e-05,
+      "loss": 9.3069,
+      "step": 9990
+    },
+    {
+      "epoch": 0.4905658063368838,
+      "grad_norm": 1.2734375,
+      "learning_rate": 9.877691995157862e-05,
+      "loss": 9.0935,
+      "step": 10000
+    },
+    {
+      "epoch": 0.4910563721432207,
+      "grad_norm": 1.640625,
+      "learning_rate": 9.877063815041709e-05,
+      "loss": 9.6356,
+      "step": 10010
+    },
+    {
+      "epoch": 0.49154693794955756,
+      "grad_norm": 1.25,
+      "learning_rate": 9.876434045950414e-05,
+      "loss": 9.878,
+      "step": 10020
+    },
+    {
+      "epoch": 0.49203750375589445,
+      "grad_norm": 1.5390625,
+      "learning_rate": 9.875802688089163e-05,
+      "loss": 9.4215,
+      "step": 10030
+    },
+    {
+      "epoch": 0.49252806956223133,
+      "grad_norm": 1.25,
+      "learning_rate": 9.875169741663653e-05,
+      "loss": 9.108,
+      "step": 10040
+    },
+    {
+      "epoch": 0.4930186353685682,
+      "grad_norm": 1.2734375,
+      "learning_rate": 9.874535206880105e-05,
+      "loss": 9.5489,
+      "step": 10050
+    },
+    {
+      "epoch": 0.4935092011749051,
+      "grad_norm": 1.203125,
+      "learning_rate": 9.873899083945248e-05,
+      "loss": 9.9763,
+      "step": 10060
+    },
+    {
+      "epoch": 0.493999766981242,
+      "grad_norm": 1.234375,
+      "learning_rate": 9.873261373066342e-05,
+      "loss": 9.0065,
+      "step": 10070
+    },
+    {
+      "epoch": 0.49449033278757887,
+      "grad_norm": 1.4609375,
+      "learning_rate": 9.872622074451153e-05,
+      "loss": 9.4474,
+      "step": 10080
+    },
+    {
+      "epoch": 0.49498089859391575,
+      "grad_norm": 1.359375,
+      "learning_rate": 9.871981188307966e-05,
+      "loss": 9.4922,
+      "step": 10090
+    },
+    {
+      "epoch": 0.49547146440025264,
+      "grad_norm": 1.09375,
+      "learning_rate": 9.871338714845589e-05,
+      "loss": 9.3891,
+      "step": 10100
+    },
+    {
+      "epoch": 0.4959620302065895,
+      "grad_norm": 0.98828125,
+      "learning_rate": 9.870694654273342e-05,
+      "loss": 9.0705,
+      "step": 10110
+    },
+    {
+      "epoch": 0.4964525960129264,
+      "grad_norm": 1.1015625,
+      "learning_rate": 9.870049006801065e-05,
+      "loss": 9.4599,
+      "step": 10120
+    },
+    {
+      "epoch": 0.4969431618192633,
+      "grad_norm": 1.234375,
+      "learning_rate": 9.86940177263911e-05,
+      "loss": 9.3144,
+      "step": 10130
+    },
+    {
+      "epoch": 0.4974337276256002,
+      "grad_norm": 1.421875,
+      "learning_rate": 9.868752951998355e-05,
+      "loss": 9.1926,
+      "step": 10140
+    },
+    {
+      "epoch": 0.49792429343193706,
+      "grad_norm": 1.0546875,
+      "learning_rate": 9.868102545090186e-05,
+      "loss": 9.8184,
+      "step": 10150
+    },
+    {
+      "epoch": 0.49841485923827394,
+      "grad_norm": 1.3984375,
+      "learning_rate": 9.867450552126506e-05,
+      "loss": 9.8685,
+      "step": 10160
+    },
+    {
+      "epoch": 0.49890542504461083,
+      "grad_norm": 0.84375,
+      "learning_rate": 9.866796973319747e-05,
+      "loss": 9.5477,
+      "step": 10170
+    },
+    {
+      "epoch": 0.4993959908509477,
+      "grad_norm": 1.4140625,
+      "learning_rate": 9.86614180888284e-05,
+      "loss": 9.5079,
+      "step": 10180
+    },
+    {
+      "epoch": 0.4998865566572846,
+      "grad_norm": 1.21875,
+      "learning_rate": 9.865485059029246e-05,
+      "loss": 9.8399,
+      "step": 10190
+    },
+    {
+      "epoch": 0.5003771224636214,
+      "grad_norm": 1.015625,
+      "learning_rate": 9.864826723972938e-05,
+      "loss": 9.1246,
+      "step": 10200
+    },
+    {
+      "epoch": 0.5008676882699583,
+      "grad_norm": 1.21875,
+      "learning_rate": 9.864166803928401e-05,
+      "loss": 8.9986,
+      "step": 10210
+    },
+    {
+      "epoch": 0.5013582540762952,
+      "grad_norm": 1.390625,
+      "learning_rate": 9.863505299110645e-05,
+      "loss": 8.951,
+      "step": 10220
+    },
+    {
+      "epoch": 0.5018488198826321,
+      "grad_norm": 1.296875,
+      "learning_rate": 9.862842209735191e-05,
+      "loss": 9.4867,
+      "step": 10230
+    },
+    {
+      "epoch": 0.502339385688969,
+      "grad_norm": 1.0234375,
+      "learning_rate": 9.862177536018075e-05,
+      "loss": 9.658,
+      "step": 10240
+    },
+    {
+      "epoch": 0.5028299514953058,
+      "grad_norm": 1.0625,
+      "learning_rate": 9.861511278175857e-05,
+      "loss": 9.2603,
+      "step": 10250
+    },
+    {
+      "epoch": 0.5033205173016427,
+      "grad_norm": 0.8359375,
+      "learning_rate": 9.8608434364256e-05,
+      "loss": 9.1361,
+      "step": 10260
+    },
+    {
+      "epoch": 0.5038110831079796,
+      "grad_norm": 1.1640625,
+      "learning_rate": 9.860174010984898e-05,
+      "loss": 9.3492,
+      "step": 10270
+    },
+    {
+      "epoch": 0.5043016489143165,
+      "grad_norm": 1.140625,
+      "learning_rate": 9.859503002071848e-05,
+      "loss": 9.3052,
+      "step": 10280
+    },
+    {
+      "epoch": 0.5047922147206534,
+      "grad_norm": 1.984375,
+      "learning_rate": 9.858830409905072e-05,
+      "loss": 9.3797,
+      "step": 10290
+    },
+    {
+      "epoch": 0.5052827805269903,
+      "grad_norm": 1.28125,
+      "learning_rate": 9.858156234703704e-05,
+      "loss": 9.5779,
+      "step": 10300
+    },
+    {
+      "epoch": 0.5057733463333272,
+      "grad_norm": 0.70703125,
+      "learning_rate": 9.85748047668739e-05,
+      "loss": 8.8646,
+      "step": 10310
+    },
+    {
+      "epoch": 0.506263912139664,
+      "grad_norm": 1.546875,
+      "learning_rate": 9.856803136076302e-05,
+      "loss": 9.3479,
+      "step": 10320
+    },
+    {
+      "epoch": 0.5067544779460009,
+      "grad_norm": 0.9609375,
+      "learning_rate": 9.856124213091118e-05,
+      "loss": 9.5916,
+      "step": 10330
+    },
+    {
+      "epoch": 0.5072450437523378,
+      "grad_norm": 1.5859375,
+      "learning_rate": 9.855443707953034e-05,
+      "loss": 10.0713,
+      "step": 10340
+    },
+    {
+      "epoch": 0.5077356095586747,
+      "grad_norm": 1.1875,
+      "learning_rate": 9.854761620883766e-05,
+      "loss": 9.7502,
+      "step": 10350
+    },
+    {
+      "epoch": 0.5082261753650116,
+      "grad_norm": 1.078125,
+      "learning_rate": 9.854077952105539e-05,
+      "loss": 9.5492,
+      "step": 10360
+    },
+    {
+      "epoch": 0.5087167411713485,
+      "grad_norm": 1.3671875,
+      "learning_rate": 9.853392701841097e-05,
+      "loss": 9.0267,
+      "step": 10370
+    },
+    {
+      "epoch": 0.5092073069776853,
+      "grad_norm": 1.4375,
+      "learning_rate": 9.852705870313697e-05,
+      "loss": 9.5468,
+      "step": 10380
+    },
+    {
+      "epoch": 0.5096978727840222,
+      "grad_norm": 1.0390625,
+      "learning_rate": 9.852017457747116e-05,
+      "loss": 9.4679,
+      "step": 10390
+    },
+    {
+      "epoch": 0.5101884385903591,
+      "grad_norm": 1.375,
+      "learning_rate": 9.851327464365639e-05,
+      "loss": 9.2376,
+      "step": 10400
+    },
+    {
+      "epoch": 0.510679004396696,
+      "grad_norm": 1.5546875,
+      "learning_rate": 9.850635890394073e-05,
+      "loss": 8.5749,
+      "step": 10410
+    },
+    {
+      "epoch": 0.5111695702030329,
+      "grad_norm": 1.0625,
+      "learning_rate": 9.849942736057734e-05,
+      "loss": 8.9378,
+      "step": 10420
+    },
+    {
+      "epoch": 0.5116601360093698,
+      "grad_norm": 1.484375,
+      "learning_rate": 9.849248001582458e-05,
+      "loss": 9.6109,
+      "step": 10430
+    },
+    {
+      "epoch": 0.5121507018157067,
+      "grad_norm": 1.2890625,
+      "learning_rate": 9.848551687194591e-05,
+      "loss": 9.7468,
+      "step": 10440
+    },
+    {
+      "epoch": 0.5126412676220435,
+      "grad_norm": 1.0390625,
+      "learning_rate": 9.847853793121e-05,
+      "loss": 9.2302,
+      "step": 10450
+    },
+    {
+      "epoch": 0.5131318334283804,
+      "grad_norm": 0.87109375,
+      "learning_rate": 9.847154319589058e-05,
+      "loss": 9.3679,
+      "step": 10460
+    },
+    {
+      "epoch": 0.5136223992347173,
+      "grad_norm": 1.59375,
+      "learning_rate": 9.846453266826659e-05,
+      "loss": 9.499,
+      "step": 10470
+    },
+    {
+      "epoch": 0.5141129650410542,
+      "grad_norm": 1.5234375,
+      "learning_rate": 9.845750635062213e-05,
+      "loss": 9.6534,
+      "step": 10480
+    },
+    {
+      "epoch": 0.5146035308473911,
+      "grad_norm": 1.53125,
+      "learning_rate": 9.845046424524638e-05,
+      "loss": 9.8588,
+      "step": 10490
+    },
+    {
+      "epoch": 0.515094096653728,
+      "grad_norm": 1.4609375,
+      "learning_rate": 9.844340635443372e-05,
+      "loss": 10.2558,
+      "step": 10500
+    },
+    {
+      "epoch": 0.5155846624600648,
+      "grad_norm": 2.40625,
+      "learning_rate": 9.843633268048363e-05,
+      "loss": 9.9006,
+      "step": 10510
+    },
+    {
+      "epoch": 0.5160752282664017,
+      "grad_norm": 1.390625,
+      "learning_rate": 9.842924322570077e-05,
+      "loss": 9.0815,
+      "step": 10520
+    },
+    {
+      "epoch": 0.5165657940727386,
+      "grad_norm": 1.28125,
+      "learning_rate": 9.842213799239492e-05,
+      "loss": 10.4063,
+      "step": 10530
+    },
+    {
+      "epoch": 0.5170563598790755,
+      "grad_norm": 1.328125,
+      "learning_rate": 9.8415016982881e-05,
+      "loss": 8.8923,
+      "step": 10540
+    },
+    {
+      "epoch": 0.5175469256854124,
+      "grad_norm": 1.8359375,
+      "learning_rate": 9.840788019947908e-05,
+      "loss": 9.8298,
+      "step": 10550
+    },
+    {
+      "epoch": 0.5180374914917493,
+      "grad_norm": 1.2421875,
+      "learning_rate": 9.840072764451438e-05,
+      "loss": 8.764,
+      "step": 10560
+    },
+    {
+      "epoch": 0.5185280572980862,
+      "grad_norm": 1.765625,
+      "learning_rate": 9.839355932031721e-05,
+      "loss": 9.384,
+      "step": 10570
+    },
+    {
+      "epoch": 0.519018623104423,
+      "grad_norm": 1.6015625,
+      "learning_rate": 9.838637522922308e-05,
+      "loss": 9.458,
+      "step": 10580
+    },
+    {
+      "epoch": 0.5195091889107599,
+      "grad_norm": 1.265625,
+      "learning_rate": 9.837917537357257e-05,
+      "loss": 9.5669,
+      "step": 10590
+    },
+    {
+      "epoch": 0.5199997547170968,
+      "grad_norm": 1.3203125,
+      "learning_rate": 9.83719597557115e-05,
+      "loss": 9.0019,
+      "step": 10600
+    },
+    {
+      "epoch": 0.5204903205234337,
+      "grad_norm": 1.484375,
+      "learning_rate": 9.836472837799069e-05,
+      "loss": 9.708,
+      "step": 10610
+    },
+    {
+      "epoch": 0.5209808863297706,
+      "grad_norm": 1.46875,
+      "learning_rate": 9.83574812427662e-05,
+      "loss": 9.3858,
+      "step": 10620
+    },
+    {
+      "epoch": 0.5214714521361075,
+      "grad_norm": 1.8984375,
+      "learning_rate": 9.835021835239918e-05,
+      "loss": 8.7211,
+      "step": 10630
+    },
+    {
+      "epoch": 0.5219620179424443,
+      "grad_norm": 1.09375,
+      "learning_rate": 9.834293970925593e-05,
+      "loss": 9.141,
+      "step": 10640
+    },
+    {
+      "epoch": 0.5224525837487812,
+      "grad_norm": 1.640625,
+      "learning_rate": 9.833564531570785e-05,
+      "loss": 9.6482,
+      "step": 10650
+    },
+    {
+      "epoch": 0.5229431495551181,
+      "grad_norm": 1.5078125,
+      "learning_rate": 9.832833517413153e-05,
+      "loss": 8.9195,
+      "step": 10660
+    },
+    {
+      "epoch": 0.523433715361455,
+      "grad_norm": 1.125,
+      "learning_rate": 9.832100928690864e-05,
+      "loss": 9.2983,
+      "step": 10670
+    },
+    {
+      "epoch": 0.5239242811677919,
+      "grad_norm": 1.140625,
+      "learning_rate": 9.831366765642597e-05,
+      "loss": 9.4973,
+      "step": 10680
+    },
+    {
+      "epoch": 0.5244148469741288,
+      "grad_norm": 1.4375,
+      "learning_rate": 9.830631028507551e-05,
+      "loss": 10.2198,
+      "step": 10690
+    },
+    {
+      "epoch": 0.5249054127804657,
+      "grad_norm": 1.4609375,
+      "learning_rate": 9.829893717525428e-05,
+      "loss": 8.9836,
+      "step": 10700
+    },
+    {
+      "epoch": 0.5253959785868025,
+      "grad_norm": 1.0078125,
+      "learning_rate": 9.829154832936454e-05,
+      "loss": 9.1102,
+      "step": 10710
+    },
+    {
+      "epoch": 0.5258865443931394,
+      "grad_norm": 1.65625,
+      "learning_rate": 9.828414374981359e-05,
+      "loss": 9.444,
+      "step": 10720
+    },
+    {
+      "epoch": 0.5263771101994763,
+      "grad_norm": 1.3515625,
+      "learning_rate": 9.827672343901386e-05,
+      "loss": 8.9337,
+      "step": 10730
+    },
+    {
+      "epoch": 0.5268676760058132,
+      "grad_norm": 1.2578125,
+      "learning_rate": 9.826928739938297e-05,
+      "loss": 9.6987,
+      "step": 10740
+    },
+    {
+      "epoch": 0.5273582418121501,
+      "grad_norm": 1.2890625,
+      "learning_rate": 9.826183563334361e-05,
+      "loss": 9.2521,
+      "step": 10750
+    },
+    {
+      "epoch": 0.527848807618487,
+      "grad_norm": 0.984375,
+      "learning_rate": 9.825436814332359e-05,
+      "loss": 9.3628,
+      "step": 10760
+    },
+    {
+      "epoch": 0.5283393734248238,
+      "grad_norm": 0.7421875,
+      "learning_rate": 9.824688493175589e-05,
+      "loss": 9.5574,
+      "step": 10770
+    },
+    {
+      "epoch": 0.5288299392311607,
+      "grad_norm": 1.265625,
+      "learning_rate": 9.823938600107856e-05,
+      "loss": 8.8906,
+      "step": 10780
+    },
+    {
+      "epoch": 0.5293205050374976,
+      "grad_norm": 0.9296875,
+      "learning_rate": 9.823187135373481e-05,
+      "loss": 9.2323,
+      "step": 10790
+    },
+    {
+      "epoch": 0.5298110708438345,
+      "grad_norm": 1.109375,
+      "learning_rate": 9.822434099217295e-05,
+      "loss": 9.0068,
+      "step": 10800
+    },
+    {
+      "epoch": 0.5303016366501714,
+      "grad_norm": 1.0234375,
+      "learning_rate": 9.821679491884641e-05,
+      "loss": 9.5788,
+      "step": 10810
+    },
+    {
+      "epoch": 0.5307922024565083,
+      "grad_norm": 1.1171875,
+      "learning_rate": 9.820923313621374e-05,
+      "loss": 9.619,
+      "step": 10820
+    },
+    {
+      "epoch": 0.5312827682628452,
+      "grad_norm": 1.046875,
+      "learning_rate": 9.820165564673862e-05,
+      "loss": 9.9449,
+      "step": 10830
+    },
+    {
+      "epoch": 0.531773334069182,
+      "grad_norm": 1.2734375,
+      "learning_rate": 9.819406245288984e-05,
+      "loss": 9.6994,
+      "step": 10840
+    },
+    {
+      "epoch": 0.5322638998755189,
+      "grad_norm": 1.1953125,
+      "learning_rate": 9.81864535571413e-05,
+      "loss": 9.9071,
+      "step": 10850
+    },
+    {
+      "epoch": 0.5327544656818558,
+      "grad_norm": 1.3984375,
+      "learning_rate": 9.817882896197203e-05,
+      "loss": 9.1392,
+      "step": 10860
+    },
+    {
+      "epoch": 0.5332450314881927,
+      "grad_norm": 1.140625,
+      "learning_rate": 9.817118866986615e-05,
+      "loss": 9.3059,
+      "step": 10870
+    },
+    {
+      "epoch": 0.5337355972945296,
+      "grad_norm": 1.2109375,
+      "learning_rate": 9.816353268331293e-05,
+      "loss": 9.3637,
+      "step": 10880
+    },
+    {
+      "epoch": 0.5342261631008665,
+      "grad_norm": 1.140625,
+      "learning_rate": 9.815586100480674e-05,
+      "loss": 8.9703,
+      "step": 10890
+    },
+    {
+      "epoch": 0.5347167289072033,
+      "grad_norm": 1.5703125,
+      "learning_rate": 9.8148173636847e-05,
+      "loss": 10.3108,
+      "step": 10900
+    },
+    {
+      "epoch": 0.5352072947135402,
+      "grad_norm": 1.2578125,
+      "learning_rate": 9.814047058193839e-05,
+      "loss": 9.5247,
+      "step": 10910
+    },
+    {
+      "epoch": 0.5356978605198771,
+      "grad_norm": 1.15625,
+      "learning_rate": 9.813275184259052e-05,
+      "loss": 9.4193,
+      "step": 10920
+    },
+    {
+      "epoch": 0.536188426326214,
+      "grad_norm": 0.94921875,
+      "learning_rate": 9.812501742131825e-05,
+      "loss": 9.0993,
+      "step": 10930
+    },
+    {
+      "epoch": 0.5366789921325509,
+      "grad_norm": 1.234375,
+      "learning_rate": 9.81172673206415e-05,
+      "loss": 9.294,
+      "step": 10940
+    },
+    {
+      "epoch": 0.5371695579388878,
+      "grad_norm": 14.8125,
+      "learning_rate": 9.810950154308528e-05,
+      "loss": 9.8395,
+      "step": 10950
+    },
+    {
+      "epoch": 0.5376601237452246,
+      "grad_norm": 1.5234375,
+      "learning_rate": 9.810172009117973e-05,
+      "loss": 10.1739,
+      "step": 10960
+    },
+    {
+      "epoch": 0.5381506895515615,
+      "grad_norm": 1.1953125,
+      "learning_rate": 9.809392296746007e-05,
+      "loss": 8.8649,
+      "step": 10970
+    },
+    {
+      "epoch": 0.5386412553578984,
+      "grad_norm": 1.0703125,
+      "learning_rate": 9.80861101744667e-05,
+      "loss": 9.1894,
+      "step": 10980
+    },
+    {
+      "epoch": 0.5391318211642353,
+      "grad_norm": 1.03125,
+      "learning_rate": 9.807828171474502e-05,
+      "loss": 9.866,
+      "step": 10990
+    },
+    {
+      "epoch": 0.5396223869705722,
+      "grad_norm": 1.8984375,
+      "learning_rate": 9.80704375908456e-05,
+      "loss": 9.6843,
+      "step": 11000
+    },
+    {
+      "epoch": 0.5401129527769091,
+      "grad_norm": 1.53125,
+      "learning_rate": 9.80625778053241e-05,
+      "loss": 9.3401,
+      "step": 11010
+    },
+    {
+      "epoch": 0.540603518583246,
+      "grad_norm": 1.0546875,
+      "learning_rate": 9.805470236074131e-05,
+      "loss": 9.4714,
+      "step": 11020
+    },
+    {
+      "epoch": 0.5410940843895828,
+      "grad_norm": 1.6171875,
+      "learning_rate": 9.804681125966306e-05,
+      "loss": 9.2955,
+      "step": 11030
+    },
+    {
+      "epoch": 0.5415846501959197,
+      "grad_norm": 1.3671875,
+      "learning_rate": 9.803890450466035e-05,
+      "loss": 9.4521,
+      "step": 11040
+    },
+    {
+      "epoch": 0.5420752160022566,
+      "grad_norm": 1.546875,
+      "learning_rate": 9.80309820983092e-05,
+      "loss": 10.1109,
+      "step": 11050
+    },
+    {
+      "epoch": 0.5425657818085935,
+      "grad_norm": 1.53125,
+      "learning_rate": 9.802304404319081e-05,
+      "loss": 9.9055,
+      "step": 11060
+    },
+    {
+      "epoch": 0.5430563476149304,
+      "grad_norm": 1.234375,
+      "learning_rate": 9.801509034189144e-05,
+      "loss": 9.3527,
+      "step": 11070
+    },
+    {
+      "epoch": 0.5435469134212673,
+      "grad_norm": 0.87890625,
+      "learning_rate": 9.800712099700244e-05,
+      "loss": 9.4049,
+      "step": 11080
+    },
+    {
+      "epoch": 0.5440374792276041,
+      "grad_norm": 1.25,
+      "learning_rate": 9.799913601112028e-05,
+      "loss": 9.8058,
+      "step": 11090
+    },
+    {
+      "epoch": 0.544528045033941,
+      "grad_norm": 1.25,
+      "learning_rate": 9.79911353868465e-05,
+      "loss": 9.1666,
+      "step": 11100
+    },
+    {
+      "epoch": 0.5450186108402779,
+      "grad_norm": 1.21875,
+      "learning_rate": 9.798311912678775e-05,
+      "loss": 9.0768,
+      "step": 11110
+    },
+    {
+      "epoch": 0.5455091766466148,
+      "grad_norm": 0.91796875,
+      "learning_rate": 9.797508723355578e-05,
+      "loss": 8.8543,
+      "step": 11120
+    },
+    {
+      "epoch": 0.5459997424529517,
+      "grad_norm": 1.9609375,
+      "learning_rate": 9.796703970976744e-05,
+      "loss": 9.5889,
+      "step": 11130
+    },
+    {
+      "epoch": 0.5464903082592886,
+      "grad_norm": 1.546875,
+      "learning_rate": 9.795897655804462e-05,
+      "loss": 9.3185,
+      "step": 11140
+    },
+    {
+      "epoch": 0.5469808740656255,
+      "grad_norm": 1.125,
+      "learning_rate": 9.795089778101439e-05,
+      "loss": 9.3281,
+      "step": 11150
+    },
+    {
+      "epoch": 0.5474714398719623,
+      "grad_norm": 1.046875,
+      "learning_rate": 9.794280338130882e-05,
+      "loss": 9.5044,
+      "step": 11160
+    },
+    {
+      "epoch": 0.5479620056782992,
+      "grad_norm": 0.828125,
+      "learning_rate": 9.793469336156513e-05,
+      "loss": 9.1314,
+      "step": 11170
+    },
+    {
+      "epoch": 0.5484525714846361,
+      "grad_norm": 0.984375,
+      "learning_rate": 9.792656772442561e-05,
+      "loss": 9.2528,
+      "step": 11180
+    },
+    {
+      "epoch": 0.548943137290973,
+      "grad_norm": 1.4765625,
+      "learning_rate": 9.791842647253764e-05,
+      "loss": 9.0135,
+      "step": 11190
+    },
+    {
+      "epoch": 0.5494337030973099,
+      "grad_norm": 1.4609375,
+      "learning_rate": 9.791026960855369e-05,
+      "loss": 8.8728,
+      "step": 11200
+    },
+    {
+      "epoch": 0.5499242689036468,
+      "grad_norm": 0.90625,
+      "learning_rate": 9.79020971351313e-05,
+      "loss": 9.2327,
+      "step": 11210
+    },
+    {
+      "epoch": 0.5504148347099836,
+      "grad_norm": 1.3671875,
+      "learning_rate": 9.789390905493311e-05,
+      "loss": 9.3236,
+      "step": 11220
+    },
+    {
+      "epoch": 0.5509054005163205,
+      "grad_norm": 1.3046875,
+      "learning_rate": 9.788570537062685e-05,
+      "loss": 9.4911,
+      "step": 11230
+    },
+    {
+      "epoch": 0.5513959663226574,
+      "grad_norm": 1.203125,
+      "learning_rate": 9.787748608488533e-05,
+      "loss": 8.5407,
+      "step": 11240
+    },
+    {
+      "epoch": 0.5518865321289943,
+      "grad_norm": 1.125,
+      "learning_rate": 9.786925120038644e-05,
+      "loss": 9.4687,
+      "step": 11250
+    },
+    {
+      "epoch": 0.5523770979353312,
+      "grad_norm": 0.9921875,
+      "learning_rate": 9.786100071981313e-05,
+      "loss": 9.7501,
+      "step": 11260
+    },
+    {
+      "epoch": 0.5528676637416681,
+      "grad_norm": 1.3046875,
+      "learning_rate": 9.785273464585347e-05,
+      "loss": 9.2585,
+      "step": 11270
+    },
+    {
+      "epoch": 0.553358229548005,
+      "grad_norm": 1.5078125,
+      "learning_rate": 9.784445298120063e-05,
+      "loss": 9.5025,
+      "step": 11280
+    },
+    {
+      "epoch": 0.5538487953543418,
+      "grad_norm": 1.765625,
+      "learning_rate": 9.783615572855274e-05,
+      "loss": 10.4089,
+      "step": 11290
+    },
+    {
+      "epoch": 0.5543393611606787,
+      "grad_norm": 1.09375,
+      "learning_rate": 9.782784289061317e-05,
+      "loss": 9.3678,
+      "step": 11300
+    },
+    {
+      "epoch": 0.5548299269670156,
+      "grad_norm": 1.4453125,
+      "learning_rate": 9.781951447009025e-05,
+      "loss": 9.685,
+      "step": 11310
+    },
+    {
+      "epoch": 0.5553204927733525,
+      "grad_norm": 1.5234375,
+      "learning_rate": 9.781117046969743e-05,
+      "loss": 9.4025,
+      "step": 11320
+    },
+    {
+      "epoch": 0.5558110585796894,
+      "grad_norm": 1.5546875,
+      "learning_rate": 9.780281089215325e-05,
+      "loss": 9.1134,
+      "step": 11330
+    },
+    {
+      "epoch": 0.5563016243860263,
+      "grad_norm": 1.3125,
+      "learning_rate": 9.77944357401813e-05,
+      "loss": 9.4538,
+      "step": 11340
+    },
+    {
+      "epoch": 0.5567921901923631,
+      "grad_norm": 1.359375,
+      "learning_rate": 9.778604501651024e-05,
+      "loss": 9.0207,
+      "step": 11350
+    },
+    {
+      "epoch": 0.5572827559987,
+      "grad_norm": 1.2109375,
+      "learning_rate": 9.777763872387384e-05,
+      "loss": 9.7341,
+      "step": 11360
+    },
+    {
+      "epoch": 0.5577733218050369,
+      "grad_norm": 2.46875,
+      "learning_rate": 9.776921686501089e-05,
+      "loss": 9.9912,
+      "step": 11370
+    },
+    {
+      "epoch": 0.5582638876113738,
+      "grad_norm": 1.21875,
+      "learning_rate": 9.776077944266529e-05,
+      "loss": 8.8371,
+      "step": 11380
+    },
+    {
+      "epoch": 0.5587544534177107,
+      "grad_norm": 0.98828125,
+      "learning_rate": 9.7752326459586e-05,
+      "loss": 9.3957,
+      "step": 11390
+    },
+    {
+      "epoch": 0.5592450192240476,
+      "grad_norm": 1.0859375,
+      "learning_rate": 9.774385791852708e-05,
+      "loss": 9.0049,
+      "step": 11400
+    },
+    {
+      "epoch": 0.5597355850303845,
+      "grad_norm": 1.84375,
+      "learning_rate": 9.773537382224758e-05,
+      "loss": 9.448,
+      "step": 11410
+    },
+    {
+      "epoch": 0.5602261508367213,
+      "grad_norm": 1.9140625,
+      "learning_rate": 9.772687417351169e-05,
+      "loss": 9.2995,
+      "step": 11420
+    },
+    {
+      "epoch": 0.5607167166430582,
+      "grad_norm": 1.015625,
+      "learning_rate": 9.771835897508864e-05,
+      "loss": 8.856,
+      "step": 11430
+    },
+    {
+      "epoch": 0.5612072824493951,
+      "grad_norm": 1.4375,
+      "learning_rate": 9.770982822975275e-05,
+      "loss": 9.4798,
+      "step": 11440
+    },
+    {
+      "epoch": 0.561697848255732,
+      "grad_norm": 1.4296875,
+      "learning_rate": 9.770128194028334e-05,
+      "loss": 10.1526,
+      "step": 11450
+    },
+    {
+      "epoch": 0.5621884140620689,
+      "grad_norm": 1.8203125,
+      "learning_rate": 9.769272010946488e-05,
+      "loss": 9.4594,
+      "step": 11460
+    },
+    {
+      "epoch": 0.5626789798684058,
+      "grad_norm": 1.109375,
+      "learning_rate": 9.768414274008686e-05,
+      "loss": 8.7307,
+      "step": 11470
+    },
+    {
+      "epoch": 0.5631695456747426,
+      "grad_norm": 0.921875,
+      "learning_rate": 9.767554983494383e-05,
+      "loss": 9.1459,
+      "step": 11480
+    },
+    {
+      "epoch": 0.5636601114810795,
+      "grad_norm": 2.390625,
+      "learning_rate": 9.766694139683539e-05,
+      "loss": 9.6059,
+      "step": 11490
+    },
+    {
+      "epoch": 0.5641506772874164,
+      "grad_norm": 1.5703125,
+      "learning_rate": 9.765831742856624e-05,
+      "loss": 9.5779,
+      "step": 11500
+    },
+    {
+      "epoch": 0.5646412430937533,
+      "grad_norm": 1.40625,
+      "learning_rate": 9.764967793294611e-05,
+      "loss": 10.1103,
+      "step": 11510
+    },
+    {
+      "epoch": 0.5651318089000902,
+      "grad_norm": 1.3359375,
+      "learning_rate": 9.764102291278978e-05,
+      "loss": 9.0898,
+      "step": 11520
+    },
+    {
+      "epoch": 0.5656223747064271,
+      "grad_norm": 1.515625,
+      "learning_rate": 9.763235237091714e-05,
+      "loss": 9.6021,
+      "step": 11530
+    },
+    {
+      "epoch": 0.566112940512764,
+      "grad_norm": 1.3125,
+      "learning_rate": 9.762366631015307e-05,
+      "loss": 9.3611,
+      "step": 11540
+    },
+    {
+      "epoch": 0.5666035063191008,
+      "grad_norm": 1.1640625,
+      "learning_rate": 9.761496473332756e-05,
+      "loss": 8.6628,
+      "step": 11550
+    },
+    {
+      "epoch": 0.5670940721254377,
+      "grad_norm": 2.765625,
+      "learning_rate": 9.760624764327564e-05,
+      "loss": 9.3294,
+      "step": 11560
+    },
+    {
+      "epoch": 0.5675846379317746,
+      "grad_norm": 1.2421875,
+      "learning_rate": 9.759751504283735e-05,
+      "loss": 9.7098,
+      "step": 11570
+    },
+    {
+      "epoch": 0.5680752037381115,
+      "grad_norm": 1.1484375,
+      "learning_rate": 9.758876693485785e-05,
+      "loss": 9.3732,
+      "step": 11580
+    },
+    {
+      "epoch": 0.5685657695444484,
+      "grad_norm": 1.0234375,
+      "learning_rate": 9.758000332218732e-05,
+      "loss": 9.2857,
+      "step": 11590
+    },
+    {
+      "epoch": 0.5690563353507853,
+      "grad_norm": 1.171875,
+      "learning_rate": 9.757122420768099e-05,
+      "loss": 8.9375,
+      "step": 11600
+    },
+    {
+      "epoch": 0.5695469011571221,
+      "grad_norm": 1.765625,
+      "learning_rate": 9.756242959419915e-05,
+      "loss": 9.5477,
+      "step": 11610
+    },
+    {
+      "epoch": 0.570037466963459,
+      "grad_norm": 1.3125,
+      "learning_rate": 9.755361948460713e-05,
+      "loss": 9.0618,
+      "step": 11620
+    },
+    {
+      "epoch": 0.5705280327697959,
+      "grad_norm": 2.09375,
+      "learning_rate": 9.754479388177531e-05,
+      "loss": 9.4139,
+      "step": 11630
+    },
+    {
+      "epoch": 0.5710185985761328,
+      "grad_norm": 1.2265625,
+      "learning_rate": 9.753595278857912e-05,
+      "loss": 9.6154,
+      "step": 11640
+    },
+    {
+      "epoch": 0.5715091643824697,
+      "grad_norm": 1.0078125,
+      "learning_rate": 9.752709620789906e-05,
+      "loss": 8.8043,
+      "step": 11650
+    },
+    {
+      "epoch": 0.5719997301888066,
+      "grad_norm": 1.265625,
+      "learning_rate": 9.751822414262064e-05,
+      "loss": 9.5509,
+      "step": 11660
+    },
+    {
+      "epoch": 0.5724902959951435,
+      "grad_norm": 1.6875,
+      "learning_rate": 9.750933659563444e-05,
+      "loss": 9.223,
+      "step": 11670
+    },
+    {
+      "epoch": 0.5729808618014803,
+      "grad_norm": 1.7890625,
+      "learning_rate": 9.750043356983605e-05,
+      "loss": 9.4112,
+      "step": 11680
+    },
+    {
+      "epoch": 0.5734714276078172,
+      "grad_norm": 1.3515625,
+      "learning_rate": 9.749151506812612e-05,
+      "loss": 8.8999,
+      "step": 11690
+    },
+    {
+      "epoch": 0.5739619934141541,
+      "grad_norm": 1.2265625,
+      "learning_rate": 9.74825810934104e-05,
+      "loss": 9.2815,
+      "step": 11700
+    },
+    {
+      "epoch": 0.574452559220491,
+      "grad_norm": 1.0,
+      "learning_rate": 9.747363164859959e-05,
+      "loss": 9.3178,
+      "step": 11710
+    },
+    {
+      "epoch": 0.5749431250268279,
+      "grad_norm": 1.1015625,
+      "learning_rate": 9.746466673660947e-05,
+      "loss": 9.0209,
+      "step": 11720
+    },
+    {
+      "epoch": 0.5754336908331646,
+      "grad_norm": 2.0625,
+      "learning_rate": 9.745568636036087e-05,
+      "loss": 9.3736,
+      "step": 11730
+    },
+    {
+      "epoch": 0.5759242566395015,
+      "grad_norm": 1.140625,
+      "learning_rate": 9.744669052277966e-05,
+      "loss": 9.5949,
+      "step": 11740
+    },
+    {
+      "epoch": 0.5764148224458384,
+      "grad_norm": 0.91796875,
+      "learning_rate": 9.74376792267967e-05,
+      "loss": 9.2923,
+      "step": 11750
+    },
+    {
+      "epoch": 0.5769053882521753,
+      "grad_norm": 1.09375,
+      "learning_rate": 9.742865247534795e-05,
+      "loss": 9.1704,
+      "step": 11760
+    },
+    {
+      "epoch": 0.5773959540585122,
+      "grad_norm": 1.7734375,
+      "learning_rate": 9.741961027137437e-05,
+      "loss": 9.6,
+      "step": 11770
+    },
+    {
+      "epoch": 0.5778865198648491,
+      "grad_norm": 0.9375,
+      "learning_rate": 9.741055261782195e-05,
+      "loss": 9.1026,
+      "step": 11780
+    },
+    {
+      "epoch": 0.578377085671186,
+      "grad_norm": 1.2109375,
+      "learning_rate": 9.740147951764175e-05,
+      "loss": 9.5726,
+      "step": 11790
+    },
+    {
+      "epoch": 0.5788676514775228,
+      "grad_norm": 1.1953125,
+      "learning_rate": 9.739239097378981e-05,
+      "loss": 8.7005,
+      "step": 11800
+    },
+    {
+      "epoch": 0.5793582172838597,
+      "grad_norm": 1.1953125,
+      "learning_rate": 9.738328698922725e-05,
+      "loss": 9.0097,
+      "step": 11810
+    },
+    {
+      "epoch": 0.5798487830901966,
+      "grad_norm": 1.515625,
+      "learning_rate": 9.73741675669202e-05,
+      "loss": 9.0148,
+      "step": 11820
+    },
+    {
+      "epoch": 0.5803393488965335,
+      "grad_norm": 1.3359375,
+      "learning_rate": 9.736503270983979e-05,
+      "loss": 9.1574,
+      "step": 11830
+    },
+    {
+      "epoch": 0.5808299147028704,
+      "grad_norm": 1.109375,
+      "learning_rate": 9.735588242096225e-05,
+      "loss": 9.2741,
+      "step": 11840
+    },
+    {
+      "epoch": 0.5813204805092073,
+      "grad_norm": 1.625,
+      "learning_rate": 9.734671670326875e-05,
+      "loss": 9.3417,
+      "step": 11850
+    },
+    {
+      "epoch": 0.5818110463155441,
+      "grad_norm": 1.3203125,
+      "learning_rate": 9.733753555974558e-05,
+      "loss": 9.5609,
+      "step": 11860
+    },
+    {
+      "epoch": 0.582301612121881,
+      "grad_norm": 1.3515625,
+      "learning_rate": 9.732833899338398e-05,
+      "loss": 9.3026,
+      "step": 11870
+    },
+    {
+      "epoch": 0.5827921779282179,
+      "grad_norm": 1.1953125,
+      "learning_rate": 9.731912700718027e-05,
+      "loss": 9.1469,
+      "step": 11880
+    },
+    {
+      "epoch": 0.5832827437345548,
+      "grad_norm": 1.140625,
+      "learning_rate": 9.730989960413572e-05,
+      "loss": 9.0359,
+      "step": 11890
+    },
+    {
+      "epoch": 0.5837733095408917,
+      "grad_norm": 1.515625,
+      "learning_rate": 9.730065678725672e-05,
+      "loss": 9.7309,
+      "step": 11900
+    },
+    {
+      "epoch": 0.5842638753472286,
+      "grad_norm": 1.9140625,
+      "learning_rate": 9.729139855955461e-05,
+      "loss": 9.5244,
+      "step": 11910
+    },
+    {
+      "epoch": 0.5847544411535655,
+      "grad_norm": 1.171875,
+      "learning_rate": 9.728212492404578e-05,
+      "loss": 8.8631,
+      "step": 11920
+    },
+    {
+      "epoch": 0.5852450069599023,
+      "grad_norm": 0.78125,
+      "learning_rate": 9.727283588375162e-05,
+      "loss": 9.0954,
+      "step": 11930
+    },
+    {
+      "epoch": 0.5857355727662392,
+      "grad_norm": 1.09375,
+      "learning_rate": 9.726353144169856e-05,
+      "loss": 9.5136,
+      "step": 11940
+    },
+    {
+      "epoch": 0.5862261385725761,
+      "grad_norm": 1.7265625,
+      "learning_rate": 9.725421160091805e-05,
+      "loss": 9.4926,
+      "step": 11950
+    },
+    {
+      "epoch": 0.586716704378913,
+      "grad_norm": 0.9375,
+      "learning_rate": 9.724487636444653e-05,
+      "loss": 8.8427,
+      "step": 11960
+    },
+    {
+      "epoch": 0.5872072701852499,
+      "grad_norm": 1.53125,
+      "learning_rate": 9.723552573532549e-05,
+      "loss": 9.2992,
+      "step": 11970
+    },
+    {
+      "epoch": 0.5876978359915868,
+      "grad_norm": 1.4609375,
+      "learning_rate": 9.722615971660139e-05,
+      "loss": 8.8732,
+      "step": 11980
+    },
+    {
+      "epoch": 0.5881884017979236,
+      "grad_norm": 1.390625,
+      "learning_rate": 9.721677831132576e-05,
+      "loss": 9.6405,
+      "step": 11990
+    },
+    {
+      "epoch": 0.5886789676042605,
+      "grad_norm": 1.328125,
+      "learning_rate": 9.72073815225551e-05,
+      "loss": 9.0488,
+      "step": 12000
+    },
+    {
+      "epoch": 0.5891695334105974,
+      "grad_norm": 1.78125,
+      "learning_rate": 9.719796935335098e-05,
+      "loss": 9.5136,
+      "step": 12010
+    },
+    {
+      "epoch": 0.5896600992169343,
+      "grad_norm": 0.90234375,
+      "learning_rate": 9.718854180677985e-05,
+      "loss": 9.6142,
+      "step": 12020
+    },
+    {
+      "epoch": 0.5901506650232712,
+      "grad_norm": 1.4921875,
+      "learning_rate": 9.717909888591335e-05,
+      "loss": 10.0236,
+      "step": 12030
+    },
+    {
+      "epoch": 0.5906412308296081,
+      "grad_norm": 0.7734375,
+      "learning_rate": 9.716964059382799e-05,
+      "loss": 8.9046,
+      "step": 12040
+    },
+    {
+      "epoch": 0.591131796635945,
+      "grad_norm": 1.125,
+      "learning_rate": 9.716016693360532e-05,
+      "loss": 9.8064,
+      "step": 12050
+    },
+    {
+      "epoch": 0.5916223624422818,
+      "grad_norm": 1.6015625,
+      "learning_rate": 9.715067790833197e-05,
+      "loss": 9.2575,
+      "step": 12060
+    },
+    {
+      "epoch": 0.5921129282486187,
+      "grad_norm": 1.0859375,
+      "learning_rate": 9.714117352109946e-05,
+      "loss": 9.0638,
+      "step": 12070
+    },
+    {
+      "epoch": 0.5926034940549556,
+      "grad_norm": 1.671875,
+      "learning_rate": 9.713165377500439e-05,
+      "loss": 9.6969,
+      "step": 12080
+    },
+    {
+      "epoch": 0.5930940598612925,
+      "grad_norm": 1.21875,
+      "learning_rate": 9.712211867314837e-05,
+      "loss": 9.5435,
+      "step": 12090
+    },
+    {
+      "epoch": 0.5935846256676294,
+      "grad_norm": 1.3203125,
+      "learning_rate": 9.7112568218638e-05,
+      "loss": 9.5468,
+      "step": 12100
+    },
+    {
+      "epoch": 0.5940751914739663,
+      "grad_norm": 1.609375,
+      "learning_rate": 9.71030024145848e-05,
+      "loss": 9.1927,
+      "step": 12110
+    },
+    {
+      "epoch": 0.5945657572803031,
+      "grad_norm": 1.71875,
+      "learning_rate": 9.709342126410543e-05,
+      "loss": 9.1248,
+      "step": 12120
+    },
+    {
+      "epoch": 0.59505632308664,
+      "grad_norm": 1.0625,
+      "learning_rate": 9.708382477032148e-05,
+      "loss": 9.2016,
+      "step": 12130
+    },
+    {
+      "epoch": 0.5955468888929769,
+      "grad_norm": 1.2421875,
+      "learning_rate": 9.707421293635952e-05,
+      "loss": 9.842,
+      "step": 12140
+    },
+    {
+      "epoch": 0.5960374546993138,
+      "grad_norm": 0.9140625,
+      "learning_rate": 9.706458576535113e-05,
+      "loss": 8.3227,
+      "step": 12150
+    },
+    {
+      "epoch": 0.5965280205056507,
+      "grad_norm": 0.953125,
+      "learning_rate": 9.705494326043293e-05,
+      "loss": 9.3225,
+      "step": 12160
+    },
+    {
+      "epoch": 0.5970185863119876,
+      "grad_norm": 1.0546875,
+      "learning_rate": 9.704528542474649e-05,
+      "loss": 9.0988,
+      "step": 12170
+    },
+    {
+      "epoch": 0.5975091521183244,
+      "grad_norm": 1.3671875,
+      "learning_rate": 9.703561226143838e-05,
+      "loss": 9.8386,
+      "step": 12180
+    },
+    {
+      "epoch": 0.5979997179246613,
+      "grad_norm": 1.5390625,
+      "learning_rate": 9.702592377366017e-05,
+      "loss": 9.0532,
+      "step": 12190
+    },
+    {
+      "epoch": 0.5984902837309982,
+      "grad_norm": 168.0,
+      "learning_rate": 9.701621996456843e-05,
+      "loss": 9.7543,
+      "step": 12200
+    },
+    {
+      "epoch": 0.5989808495373351,
+      "grad_norm": 1.1796875,
+      "learning_rate": 9.700650083732472e-05,
+      "loss": 9.6631,
+      "step": 12210
+    },
+    {
+      "epoch": 0.599471415343672,
+      "grad_norm": 1.59375,
+      "learning_rate": 9.69967663950956e-05,
+      "loss": 9.3424,
+      "step": 12220
+    },
+    {
+      "epoch": 0.5999619811500089,
+      "grad_norm": 1.265625,
+      "learning_rate": 9.698701664105256e-05,
+      "loss": 9.8517,
+      "step": 12230
+    },
+    {
+      "epoch": 0.6004525469563458,
+      "grad_norm": 1.4296875,
+      "learning_rate": 9.697725157837219e-05,
+      "loss": 9.0243,
+      "step": 12240
+    },
+    {
+      "epoch": 0.6009431127626826,
+      "grad_norm": 1.40625,
+      "learning_rate": 9.696747121023595e-05,
+      "loss": 9.3301,
+      "step": 12250
+    },
+    {
+      "epoch": 0.6014336785690195,
+      "grad_norm": 0.87890625,
+      "learning_rate": 9.695767553983036e-05,
+      "loss": 9.409,
+      "step": 12260
+    },
+    {
+      "epoch": 0.6019242443753564,
+      "grad_norm": 1.5625,
+      "learning_rate": 9.69478645703469e-05,
+      "loss": 9.3312,
+      "step": 12270
+    },
+    {
+      "epoch": 0.6024148101816933,
+      "grad_norm": 1.25,
+      "learning_rate": 9.693803830498206e-05,
+      "loss": 9.0197,
+      "step": 12280
+    },
+    {
+      "epoch": 0.6029053759880302,
+      "grad_norm": 1.375,
+      "learning_rate": 9.692819674693726e-05,
+      "loss": 9.6851,
+      "step": 12290
+    },
+    {
+      "epoch": 0.6033959417943671,
+      "grad_norm": 2.25,
+      "learning_rate": 9.691833989941896e-05,
+      "loss": 9.1445,
+      "step": 12300
+    },
+    {
+      "epoch": 0.603886507600704,
+      "grad_norm": 1.4765625,
+      "learning_rate": 9.690846776563858e-05,
+      "loss": 9.6592,
+      "step": 12310
+    },
+    {
+      "epoch": 0.6043770734070408,
+      "grad_norm": 1.5546875,
+      "learning_rate": 9.689858034881248e-05,
+      "loss": 9.1707,
+      "step": 12320
+    },
+    {
+      "epoch": 0.6048676392133777,
+      "grad_norm": 0.91015625,
+      "learning_rate": 9.688867765216207e-05,
+      "loss": 8.7805,
+      "step": 12330
+    },
+    {
+      "epoch": 0.6053582050197146,
+      "grad_norm": 1.34375,
+      "learning_rate": 9.68787596789137e-05,
+      "loss": 9.0361,
+      "step": 12340
+    },
+    {
+      "epoch": 0.6058487708260515,
+      "grad_norm": 1.71875,
+      "learning_rate": 9.686882643229872e-05,
+      "loss": 9.5252,
+      "step": 12350
+    },
+    {
+      "epoch": 0.6063393366323884,
+      "grad_norm": 2.125,
+      "learning_rate": 9.685887791555342e-05,
+      "loss": 10.1536,
+      "step": 12360
+    },
+    {
+      "epoch": 0.6068299024387253,
+      "grad_norm": 1.5703125,
+      "learning_rate": 9.684891413191905e-05,
+      "loss": 9.2695,
+      "step": 12370
+    },
+    {
+      "epoch": 0.6073204682450621,
+      "grad_norm": 1.234375,
+      "learning_rate": 9.683893508464193e-05,
+      "loss": 9.6399,
+      "step": 12380
+    },
+    {
+      "epoch": 0.607811034051399,
+      "grad_norm": 1.5078125,
+      "learning_rate": 9.682894077697324e-05,
+      "loss": 10.4395,
+      "step": 12390
+    },
+    {
+      "epoch": 0.6083015998577359,
+      "grad_norm": 1.65625,
+      "learning_rate": 9.68189312121692e-05,
+      "loss": 9.7236,
+      "step": 12400
+    },
+    {
+      "epoch": 0.6087921656640728,
+      "grad_norm": 1.265625,
+      "learning_rate": 9.680890639349101e-05,
+      "loss": 9.2564,
+      "step": 12410
+    },
+    {
+      "epoch": 0.6092827314704097,
+      "grad_norm": 1.3125,
+      "learning_rate": 9.679886632420476e-05,
+      "loss": 10.0029,
+      "step": 12420
+    },
+    {
+      "epoch": 0.6097732972767466,
+      "grad_norm": 1.234375,
+      "learning_rate": 9.67888110075816e-05,
+      "loss": 8.9355,
+      "step": 12430
+    },
+    {
+      "epoch": 0.6102638630830834,
+      "grad_norm": 1.0703125,
+      "learning_rate": 9.67787404468976e-05,
+      "loss": 9.3857,
+      "step": 12440
+    },
+    {
+      "epoch": 0.6107544288894203,
+      "grad_norm": 0.98046875,
+      "learning_rate": 9.676865464543382e-05,
+      "loss": 9.0286,
+      "step": 12450
+    },
+    {
+      "epoch": 0.6112449946957572,
+      "grad_norm": 1.078125,
+      "learning_rate": 9.675855360647624e-05,
+      "loss": 8.9718,
+      "step": 12460
+    },
+    {
+      "epoch": 0.6117355605020941,
+      "grad_norm": 1.484375,
+      "learning_rate": 9.674843733331585e-05,
+      "loss": 9.7685,
+      "step": 12470
+    },
+    {
+      "epoch": 0.612226126308431,
+      "grad_norm": 1.515625,
+      "learning_rate": 9.67383058292486e-05,
+      "loss": 9.6544,
+      "step": 12480
+    },
+    {
+      "epoch": 0.6127166921147679,
+      "grad_norm": 1.6953125,
+      "learning_rate": 9.672815909757538e-05,
+      "loss": 8.5495,
+      "step": 12490
+    },
+    {
+      "epoch": 0.6132072579211048,
+      "grad_norm": 1.609375,
+      "learning_rate": 9.671799714160205e-05,
+      "loss": 9.1717,
+      "step": 12500
+    },
+    {
+      "epoch": 0.6136978237274416,
+      "grad_norm": 0.96484375,
+      "learning_rate": 9.670781996463945e-05,
+      "loss": 8.6525,
+      "step": 12510
+    },
+    {
+      "epoch": 0.6141883895337785,
+      "grad_norm": 1.1015625,
+      "learning_rate": 9.669762757000336e-05,
+      "loss": 9.138,
+      "step": 12520
+    },
+    {
+      "epoch": 0.6146789553401154,
+      "grad_norm": 1.1484375,
+      "learning_rate": 9.66874199610145e-05,
+      "loss": 9.006,
+      "step": 12530
+    },
+    {
+      "epoch": 0.6151695211464523,
+      "grad_norm": 1.5546875,
+      "learning_rate": 9.667719714099858e-05,
+      "loss": 9.2872,
+      "step": 12540
+    },
+    {
+      "epoch": 0.6156600869527892,
+      "grad_norm": 1.15625,
+      "learning_rate": 9.666695911328625e-05,
+      "loss": 9.4924,
+      "step": 12550
+    },
+    {
+      "epoch": 0.6161506527591261,
+      "grad_norm": 1.1640625,
+      "learning_rate": 9.665670588121312e-05,
+      "loss": 9.2584,
+      "step": 12560
+    },
+    {
+      "epoch": 0.616641218565463,
+      "grad_norm": 1.125,
+      "learning_rate": 9.664643744811976e-05,
+      "loss": 9.0898,
+      "step": 12570
+    },
+    {
+      "epoch": 0.6171317843717998,
+      "grad_norm": 1.40625,
+      "learning_rate": 9.663615381735167e-05,
+      "loss": 9.7375,
+      "step": 12580
+    },
+    {
+      "epoch": 0.6176223501781367,
+      "grad_norm": 0.8203125,
+      "learning_rate": 9.662585499225931e-05,
+      "loss": 9.363,
+      "step": 12590
+    },
+    {
+      "epoch": 0.6181129159844736,
+      "grad_norm": 0.95703125,
+      "learning_rate": 9.661554097619812e-05,
+      "loss": 9.2414,
+      "step": 12600
+    },
+    {
+      "epoch": 0.6186034817908105,
+      "grad_norm": 1.328125,
+      "learning_rate": 9.660521177252844e-05,
+      "loss": 9.6223,
+      "step": 12610
+    },
+    {
+      "epoch": 0.6190940475971474,
+      "grad_norm": 1.2421875,
+      "learning_rate": 9.659486738461561e-05,
+      "loss": 9.2775,
+      "step": 12620
+    },
+    {
+      "epoch": 0.6195846134034843,
+      "grad_norm": 1.6875,
+      "learning_rate": 9.658450781582986e-05,
+      "loss": 9.1536,
+      "step": 12630
+    },
+    {
+      "epoch": 0.6200751792098211,
+      "grad_norm": 1.2265625,
+      "learning_rate": 9.657413306954642e-05,
+      "loss": 9.8214,
+      "step": 12640
+    },
+    {
+      "epoch": 0.620565745016158,
+      "grad_norm": 1.234375,
+      "learning_rate": 9.656374314914543e-05,
+      "loss": 9.451,
+      "step": 12650
+    },
+    {
+      "epoch": 0.6210563108224949,
+      "grad_norm": 1.3984375,
+      "learning_rate": 9.655333805801199e-05,
+      "loss": 9.1744,
+      "step": 12660
+    },
+    {
+      "epoch": 0.6215468766288318,
+      "grad_norm": 1.1328125,
+      "learning_rate": 9.654291779953613e-05,
+      "loss": 9.3551,
+      "step": 12670
+    },
+    {
+      "epoch": 0.6220374424351687,
+      "grad_norm": 1.5,
+      "learning_rate": 9.653248237711284e-05,
+      "loss": 9.3635,
+      "step": 12680
+    },
+    {
+      "epoch": 0.6225280082415056,
+      "grad_norm": 1.1171875,
+      "learning_rate": 9.652203179414203e-05,
+      "loss": 9.1314,
+      "step": 12690
+    },
+    {
+      "epoch": 0.6230185740478424,
+      "grad_norm": 1.8203125,
+      "learning_rate": 9.651156605402859e-05,
+      "loss": 9.1049,
+      "step": 12700
+    },
+    {
+      "epoch": 0.6235091398541793,
+      "grad_norm": 1.2578125,
+      "learning_rate": 9.650108516018225e-05,
+      "loss": 9.111,
+      "step": 12710
+    },
+    {
+      "epoch": 0.6239997056605162,
+      "grad_norm": 1.078125,
+      "learning_rate": 9.649058911601783e-05,
+      "loss": 9.8787,
+      "step": 12720
+    },
+    {
+      "epoch": 0.6244902714668531,
+      "grad_norm": 1.4609375,
+      "learning_rate": 9.648007792495493e-05,
+      "loss": 9.5325,
+      "step": 12730
+    },
+    {
+      "epoch": 0.62498083727319,
+      "grad_norm": 1.9296875,
+      "learning_rate": 9.64695515904182e-05,
+      "loss": 9.6468,
+      "step": 12740
+    },
+    {
+      "epoch": 0.6254714030795269,
+      "grad_norm": 1.6875,
+      "learning_rate": 9.645901011583716e-05,
+      "loss": 9.3853,
+      "step": 12750
+    },
+    {
+      "epoch": 0.6259619688858638,
+      "grad_norm": 1.46875,
+      "learning_rate": 9.644845350464628e-05,
+      "loss": 10.0464,
+      "step": 12760
+    },
+    {
+      "epoch": 0.6264525346922006,
+      "grad_norm": 1.515625,
+      "learning_rate": 9.643788176028497e-05,
+      "loss": 8.9483,
+      "step": 12770
+    },
+    {
+      "epoch": 0.6269431004985375,
+      "grad_norm": 1.109375,
+      "learning_rate": 9.642729488619758e-05,
+      "loss": 9.1678,
+      "step": 12780
+    },
+    {
+      "epoch": 0.6274336663048744,
+      "grad_norm": 1.296875,
+      "learning_rate": 9.641669288583334e-05,
+      "loss": 9.0915,
+      "step": 12790
+    },
+    {
+      "epoch": 0.6279242321112113,
+      "grad_norm": 2.4375,
+      "learning_rate": 9.640607576264647e-05,
+      "loss": 9.3949,
+      "step": 12800
+    },
+    {
+      "epoch": 0.6284147979175482,
+      "grad_norm": 1.5078125,
+      "learning_rate": 9.639544352009606e-05,
+      "loss": 9.4251,
+      "step": 12810
+    },
+    {
+      "epoch": 0.6289053637238851,
+      "grad_norm": 1.0859375,
+      "learning_rate": 9.638479616164619e-05,
+      "loss": 8.8592,
+      "step": 12820
+    },
+    {
+      "epoch": 0.6293959295302219,
+      "grad_norm": 1.421875,
+      "learning_rate": 9.637413369076582e-05,
+      "loss": 9.2557,
+      "step": 12830
+    },
+    {
+      "epoch": 0.6298864953365588,
+      "grad_norm": 1.0546875,
+      "learning_rate": 9.636345611092885e-05,
+      "loss": 9.2183,
+      "step": 12840
+    },
+    {
+      "epoch": 0.6303770611428957,
+      "grad_norm": 0.74609375,
+      "learning_rate": 9.635276342561405e-05,
+      "loss": 9.2622,
+      "step": 12850
+    },
+    {
+      "epoch": 0.6308676269492326,
+      "grad_norm": 1.15625,
+      "learning_rate": 9.634205563830522e-05,
+      "loss": 8.6132,
+      "step": 12860
+    },
+    {
+      "epoch": 0.6313581927555695,
+      "grad_norm": 1.2890625,
+      "learning_rate": 9.633133275249099e-05,
+      "loss": 9.559,
+      "step": 12870
+    },
+    {
+      "epoch": 0.6318487585619064,
+      "grad_norm": 1.3046875,
+      "learning_rate": 9.632059477166493e-05,
+      "loss": 9.7625,
+      "step": 12880
+    },
+    {
+      "epoch": 0.6323393243682432,
+      "grad_norm": 1.1640625,
+      "learning_rate": 9.630984169932554e-05,
+      "loss": 9.831,
+      "step": 12890
+    },
+    {
+      "epoch": 0.6328298901745801,
+      "grad_norm": 1.1796875,
+      "learning_rate": 9.629907353897625e-05,
+      "loss": 9.5791,
+      "step": 12900
+    },
+    {
+      "epoch": 0.633320455980917,
+      "grad_norm": 1.078125,
+      "learning_rate": 9.628829029412536e-05,
+      "loss": 9.3434,
+      "step": 12910
+    },
+    {
+      "epoch": 0.6338110217872539,
+      "grad_norm": 1.1328125,
+      "learning_rate": 9.627749196828612e-05,
+      "loss": 8.9186,
+      "step": 12920
+    },
+    {
+      "epoch": 0.6343015875935908,
+      "grad_norm": 1.046875,
+      "learning_rate": 9.626667856497671e-05,
+      "loss": 9.0165,
+      "step": 12930
+    },
+    {
+      "epoch": 0.6347921533999277,
+      "grad_norm": 0.98828125,
+      "learning_rate": 9.625585008772015e-05,
+      "loss": 9.2345,
+      "step": 12940
+    },
+    {
+      "epoch": 0.6352827192062646,
+      "grad_norm": 1.0234375,
+      "learning_rate": 9.624500654004448e-05,
+      "loss": 9.4892,
+      "step": 12950
+    },
+    {
+      "epoch": 0.6357732850126014,
+      "grad_norm": 0.828125,
+      "learning_rate": 9.623414792548255e-05,
+      "loss": 9.1177,
+      "step": 12960
+    },
+    {
+      "epoch": 0.6362638508189383,
+      "grad_norm": 1.375,
+      "learning_rate": 9.622327424757217e-05,
+      "loss": 9.765,
+      "step": 12970
+    },
+    {
+      "epoch": 0.6367544166252752,
+      "grad_norm": 1.453125,
+      "learning_rate": 9.621238550985603e-05,
+      "loss": 9.4108,
+      "step": 12980
+    },
+    {
+      "epoch": 0.6372449824316121,
+      "grad_norm": 1.3828125,
+      "learning_rate": 9.620148171588175e-05,
+      "loss": 9.0686,
+      "step": 12990
+    },
+    {
+      "epoch": 0.637735548237949,
+      "grad_norm": 1.4765625,
+      "learning_rate": 9.619056286920188e-05,
+      "loss": 8.926,
+      "step": 13000
+    },
+    {
+      "epoch": 0.6382261140442859,
+      "grad_norm": 0.93359375,
+      "learning_rate": 9.61796289733738e-05,
+      "loss": 9.4987,
+      "step": 13010
+    },
+    {
+      "epoch": 0.6387166798506227,
+      "grad_norm": 1.5078125,
+      "learning_rate": 9.616868003195985e-05,
+      "loss": 8.8955,
+      "step": 13020
+    },
+    {
+      "epoch": 0.6392072456569596,
+      "grad_norm": 0.921875,
+      "learning_rate": 9.615771604852728e-05,
+      "loss": 9.7532,
+      "step": 13030
+    },
+    {
+      "epoch": 0.6396978114632965,
+      "grad_norm": 1.6640625,
+      "learning_rate": 9.614673702664818e-05,
+      "loss": 9.6133,
+      "step": 13040
+    },
+    {
+      "epoch": 0.6401883772696334,
+      "grad_norm": 1.09375,
+      "learning_rate": 9.61357429698996e-05,
+      "loss": 9.5092,
+      "step": 13050
+    },
+    {
+      "epoch": 0.6406789430759703,
+      "grad_norm": 1.296875,
+      "learning_rate": 9.612473388186346e-05,
+      "loss": 9.2628,
+      "step": 13060
+    },
+    {
+      "epoch": 0.6411695088823072,
+      "grad_norm": 1.5234375,
+      "learning_rate": 9.611370976612659e-05,
+      "loss": 8.9858,
+      "step": 13070
+    },
+    {
+      "epoch": 0.641660074688644,
+      "grad_norm": 1.859375,
+      "learning_rate": 9.610267062628069e-05,
+      "loss": 9.6127,
+      "step": 13080
+    },
+    {
+      "epoch": 0.6421506404949809,
+      "grad_norm": 0.8671875,
+      "learning_rate": 9.60916164659224e-05,
+      "loss": 9.2645,
+      "step": 13090
+    },
+    {
+      "epoch": 0.6426412063013178,
+      "grad_norm": 1.53125,
+      "learning_rate": 9.608054728865323e-05,
+      "loss": 9.8776,
+      "step": 13100
+    },
+    {
+      "epoch": 0.6431317721076547,
+      "grad_norm": 0.91015625,
+      "learning_rate": 9.606946309807956e-05,
+      "loss": 9.572,
+      "step": 13110
+    },
+    {
+      "epoch": 0.6436223379139916,
+      "grad_norm": 2.28125,
+      "learning_rate": 9.605836389781269e-05,
+      "loss": 9.4712,
+      "step": 13120
+    },
+    {
+      "epoch": 0.6441129037203285,
+      "grad_norm": 1.75,
+      "learning_rate": 9.604724969146879e-05,
+      "loss": 9.7115,
+      "step": 13130
+    },
+    {
+      "epoch": 0.6446034695266654,
+      "grad_norm": 1.1875,
+      "learning_rate": 9.603612048266895e-05,
+      "loss": 9.3763,
+      "step": 13140
+    },
+    {
+      "epoch": 0.6450940353330022,
+      "grad_norm": 1.0703125,
+      "learning_rate": 9.602497627503915e-05,
+      "loss": 9.0467,
+      "step": 13150
+    },
+    {
+      "epoch": 0.6455846011393391,
+      "grad_norm": 0.77734375,
+      "learning_rate": 9.601381707221017e-05,
+      "loss": 9.2757,
+      "step": 13160
+    },
+    {
+      "epoch": 0.646075166945676,
+      "grad_norm": 1.6796875,
+      "learning_rate": 9.600264287781781e-05,
+      "loss": 9.5238,
+      "step": 13170
+    },
+    {
+      "epoch": 0.6465657327520129,
+      "grad_norm": 1.203125,
+      "learning_rate": 9.599145369550266e-05,
+      "loss": 9.5609,
+      "step": 13180
+    },
+    {
+      "epoch": 0.6470562985583498,
+      "grad_norm": 1.3828125,
+      "learning_rate": 9.598024952891022e-05,
+      "loss": 9.3848,
+      "step": 13190
+    },
+    {
+      "epoch": 0.6475468643646867,
+      "grad_norm": 1.3203125,
+      "learning_rate": 9.596903038169086e-05,
+      "loss": 9.0239,
+      "step": 13200
+    },
+    {
+      "epoch": 0.6480374301710236,
+      "grad_norm": 0.90234375,
+      "learning_rate": 9.595779625749985e-05,
+      "loss": 8.9481,
+      "step": 13210
+    },
+    {
+      "epoch": 0.6485279959773604,
+      "grad_norm": 1.4609375,
+      "learning_rate": 9.594654715999733e-05,
+      "loss": 9.719,
+      "step": 13220
+    },
+    {
+      "epoch": 0.6490185617836973,
+      "grad_norm": 1.3515625,
+      "learning_rate": 9.593528309284834e-05,
+      "loss": 8.9989,
+      "step": 13230
+    },
+    {
+      "epoch": 0.6495091275900342,
+      "grad_norm": 1.3046875,
+      "learning_rate": 9.592400405972274e-05,
+      "loss": 9.6057,
+      "step": 13240
+    },
+    {
+      "epoch": 0.6499996933963711,
+      "grad_norm": 1.5859375,
+      "learning_rate": 9.591271006429534e-05,
+      "loss": 8.8013,
+      "step": 13250
+    },
+    {
+      "epoch": 0.650490259202708,
+      "grad_norm": 1.5234375,
+      "learning_rate": 9.590140111024577e-05,
+      "loss": 9.5656,
+      "step": 13260
+    },
+    {
+      "epoch": 0.6509808250090449,
+      "grad_norm": 1.203125,
+      "learning_rate": 9.589007720125856e-05,
+      "loss": 9.2253,
+      "step": 13270
+    },
+    {
+      "epoch": 0.6514713908153817,
+      "grad_norm": 1.4765625,
+      "learning_rate": 9.587873834102308e-05,
+      "loss": 9.3081,
+      "step": 13280
+    },
+    {
+      "epoch": 0.6519619566217186,
+      "grad_norm": 1.2734375,
+      "learning_rate": 9.58673845332336e-05,
+      "loss": 8.5528,
+      "step": 13290
+    },
+    {
+      "epoch": 0.6524525224280554,
+      "grad_norm": 1.6796875,
+      "learning_rate": 9.585601578158929e-05,
+      "loss": 9.5058,
+      "step": 13300
+    },
+    {
+      "epoch": 0.6529430882343923,
+      "grad_norm": 1.171875,
+      "learning_rate": 9.584463208979413e-05,
+      "loss": 9.0481,
+      "step": 13310
+    },
+    {
+      "epoch": 0.6534336540407292,
+      "grad_norm": 1.4765625,
+      "learning_rate": 9.583323346155699e-05,
+      "loss": 9.5199,
+      "step": 13320
+    },
+    {
+      "epoch": 0.6539242198470661,
+      "grad_norm": 1.28125,
+      "learning_rate": 9.582181990059161e-05,
+      "loss": 9.136,
+      "step": 13330
+    },
+    {
+      "epoch": 0.6544147856534029,
+      "grad_norm": 1.140625,
+      "learning_rate": 9.581039141061658e-05,
+      "loss": 9.2259,
+      "step": 13340
+    },
+    {
+      "epoch": 0.6549053514597398,
+      "grad_norm": 1.09375,
+      "learning_rate": 9.57989479953554e-05,
+      "loss": 9.7394,
+      "step": 13350
+    },
+    {
+      "epoch": 0.6553959172660767,
+      "grad_norm": 1.6015625,
+      "learning_rate": 9.578748965853638e-05,
+      "loss": 9.112,
+      "step": 13360
+    },
+    {
+      "epoch": 0.6558864830724136,
+      "grad_norm": 0.98828125,
+      "learning_rate": 9.577601640389272e-05,
+      "loss": 9.7379,
+      "step": 13370
+    },
+    {
+      "epoch": 0.6563770488787505,
+      "grad_norm": 2.109375,
+      "learning_rate": 9.576452823516243e-05,
+      "loss": 9.1594,
+      "step": 13380
+    },
+    {
+      "epoch": 0.6568676146850874,
+      "grad_norm": 1.1796875,
+      "learning_rate": 9.575302515608848e-05,
+      "loss": 9.2677,
+      "step": 13390
+    },
+    {
+      "epoch": 0.6573581804914242,
+      "grad_norm": 0.97265625,
+      "learning_rate": 9.574150717041861e-05,
+      "loss": 9.517,
+      "step": 13400
+    },
+    {
+      "epoch": 0.6578487462977611,
+      "grad_norm": 1.2578125,
+      "learning_rate": 9.572997428190544e-05,
+      "loss": 9.2462,
+      "step": 13410
+    },
+    {
+      "epoch": 0.658339312104098,
+      "grad_norm": 1.59375,
+      "learning_rate": 9.571842649430647e-05,
+      "loss": 9.1855,
+      "step": 13420
+    },
+    {
+      "epoch": 0.6588298779104349,
+      "grad_norm": 1.21875,
+      "learning_rate": 9.5706863811384e-05,
+      "loss": 9.2533,
+      "step": 13430
+    },
+    {
+      "epoch": 0.6593204437167718,
+      "grad_norm": 1.4375,
+      "learning_rate": 9.569528623690527e-05,
+      "loss": 9.0064,
+      "step": 13440
+    },
+    {
+      "epoch": 0.6598110095231087,
+      "grad_norm": 1.265625,
+      "learning_rate": 9.568369377464228e-05,
+      "loss": 9.4393,
+      "step": 13450
+    },
+    {
+      "epoch": 0.6603015753294456,
+      "grad_norm": 1.1171875,
+      "learning_rate": 9.567208642837195e-05,
+      "loss": 8.9572,
+      "step": 13460
+    },
+    {
+      "epoch": 0.6607921411357824,
+      "grad_norm": 0.89453125,
+      "learning_rate": 9.566046420187598e-05,
+      "loss": 8.8403,
+      "step": 13470
+    },
+    {
+      "epoch": 0.6612827069421193,
+      "grad_norm": 1.2109375,
+      "learning_rate": 9.564882709894099e-05,
+      "loss": 9.4499,
+      "step": 13480
+    },
+    {
+      "epoch": 0.6617732727484562,
+      "grad_norm": 1.15625,
+      "learning_rate": 9.56371751233584e-05,
+      "loss": 9.8409,
+      "step": 13490
+    },
+    {
+      "epoch": 0.6622638385547931,
+      "grad_norm": 1.28125,
+      "learning_rate": 9.562550827892446e-05,
+      "loss": 9.0476,
+      "step": 13500
+    },
+    {
+      "epoch": 0.66275440436113,
+      "grad_norm": 1.515625,
+      "learning_rate": 9.561382656944037e-05,
+      "loss": 9.5756,
+      "step": 13510
+    },
+    {
+      "epoch": 0.6632449701674669,
+      "grad_norm": 1.4921875,
+      "learning_rate": 9.560212999871203e-05,
+      "loss": 8.5325,
+      "step": 13520
+    },
+    {
+      "epoch": 0.6637355359738037,
+      "grad_norm": 1.0390625,
+      "learning_rate": 9.559041857055029e-05,
+      "loss": 9.41,
+      "step": 13530
+    },
+    {
+      "epoch": 0.6642261017801406,
+      "grad_norm": 1.1328125,
+      "learning_rate": 9.557869228877078e-05,
+      "loss": 9.5957,
+      "step": 13540
+    },
+    {
+      "epoch": 0.6647166675864775,
+      "grad_norm": 0.921875,
+      "learning_rate": 9.556695115719398e-05,
+      "loss": 9.0518,
+      "step": 13550
+    },
+    {
+      "epoch": 0.6652072333928144,
+      "grad_norm": 0.98046875,
+      "learning_rate": 9.555519517964523e-05,
+      "loss": 10.0197,
+      "step": 13560
+    },
+    {
+      "epoch": 0.6656977991991513,
+      "grad_norm": 1.234375,
+      "learning_rate": 9.554342435995473e-05,
+      "loss": 8.9941,
+      "step": 13570
+    },
+    {
+      "epoch": 0.6661883650054882,
+      "grad_norm": 3.296875,
+      "learning_rate": 9.553163870195742e-05,
+      "loss": 9.2883,
+      "step": 13580
+    },
+    {
+      "epoch": 0.666678930811825,
+      "grad_norm": 1.4375,
+      "learning_rate": 9.551983820949315e-05,
+      "loss": 9.4084,
+      "step": 13590
+    },
+    {
+      "epoch": 0.6671694966181619,
+      "grad_norm": 0.8671875,
+      "learning_rate": 9.550802288640662e-05,
+      "loss": 8.7519,
+      "step": 13600
+    },
+    {
+      "epoch": 0.6676600624244988,
+      "grad_norm": 1.203125,
+      "learning_rate": 9.549619273654728e-05,
+      "loss": 9.3373,
+      "step": 13610
+    },
+    {
+      "epoch": 0.6681506282308357,
+      "grad_norm": 1.3203125,
+      "learning_rate": 9.54843477637695e-05,
+      "loss": 9.5479,
+      "step": 13620
+    },
+    {
+      "epoch": 0.6686411940371726,
+      "grad_norm": 1.03125,
+      "learning_rate": 9.547248797193243e-05,
+      "loss": 9.2869,
+      "step": 13630
+    },
+    {
+      "epoch": 0.6691317598435095,
+      "grad_norm": 1.953125,
+      "learning_rate": 9.546061336490006e-05,
+      "loss": 9.7029,
+      "step": 13640
+    },
+    {
+      "epoch": 0.6696223256498464,
+      "grad_norm": 1.3046875,
+      "learning_rate": 9.544872394654118e-05,
+      "loss": 9.2265,
+      "step": 13650
+    },
+    {
+      "epoch": 0.6701128914561832,
+      "grad_norm": 0.9765625,
+      "learning_rate": 9.543681972072946e-05,
+      "loss": 8.9653,
+      "step": 13660
+    },
+    {
+      "epoch": 0.6706034572625201,
+      "grad_norm": 1.2734375,
+      "learning_rate": 9.542490069134335e-05,
+      "loss": 9.4043,
+      "step": 13670
+    },
+    {
+      "epoch": 0.671094023068857,
+      "grad_norm": 0.8359375,
+      "learning_rate": 9.541296686226614e-05,
+      "loss": 9.6635,
+      "step": 13680
+    },
+    {
+      "epoch": 0.6715845888751939,
+      "grad_norm": 1.078125,
+      "learning_rate": 9.540101823738594e-05,
+      "loss": 9.6812,
+      "step": 13690
+    },
+    {
+      "epoch": 0.6720751546815308,
+      "grad_norm": 0.8984375,
+      "learning_rate": 9.538905482059569e-05,
+      "loss": 9.147,
+      "step": 13700
+    },
+    {
+      "epoch": 0.6725657204878677,
+      "grad_norm": 1.5,
+      "learning_rate": 9.537707661579311e-05,
+      "loss": 9.4605,
+      "step": 13710
+    },
+    {
+      "epoch": 0.6730562862942046,
+      "grad_norm": 1.1953125,
+      "learning_rate": 9.536508362688079e-05,
+      "loss": 8.9682,
+      "step": 13720
+    },
+    {
+      "epoch": 0.6735468521005414,
+      "grad_norm": 1.1328125,
+      "learning_rate": 9.535307585776612e-05,
+      "loss": 9.2662,
+      "step": 13730
+    },
+    {
+      "epoch": 0.6740374179068783,
+      "grad_norm": 1.34375,
+      "learning_rate": 9.53410533123613e-05,
+      "loss": 9.3428,
+      "step": 13740
+    },
+    {
+      "epoch": 0.6745279837132152,
+      "grad_norm": 1.3203125,
+      "learning_rate": 9.53290159945833e-05,
+      "loss": 9.3854,
+      "step": 13750
+    },
+    {
+      "epoch": 0.6750185495195521,
+      "grad_norm": 0.9375,
+      "learning_rate": 9.531696390835401e-05,
+      "loss": 9.006,
+      "step": 13760
+    },
+    {
+      "epoch": 0.675509115325889,
+      "grad_norm": 0.8828125,
+      "learning_rate": 9.530489705760003e-05,
+      "loss": 9.3028,
+      "step": 13770
+    },
+    {
+      "epoch": 0.6759996811322259,
+      "grad_norm": 1.5703125,
+      "learning_rate": 9.529281544625282e-05,
+      "loss": 10.081,
+      "step": 13780
+    },
+    {
+      "epoch": 0.6764902469385627,
+      "grad_norm": 87.5,
+      "learning_rate": 9.528071907824863e-05,
+      "loss": 9.882,
+      "step": 13790
+    },
+    {
+      "epoch": 0.6769808127448996,
+      "grad_norm": 1.1796875,
+      "learning_rate": 9.526860795752853e-05,
+      "loss": 9.4508,
+      "step": 13800
+    },
+    {
+      "epoch": 0.6774713785512365,
+      "grad_norm": 0.8671875,
+      "learning_rate": 9.525648208803839e-05,
+      "loss": 9.4287,
+      "step": 13810
+    },
+    {
+      "epoch": 0.6779619443575734,
+      "grad_norm": 2.0625,
+      "learning_rate": 9.524434147372889e-05,
+      "loss": 9.2851,
+      "step": 13820
+    },
+    {
+      "epoch": 0.6784525101639103,
+      "grad_norm": 1.6328125,
+      "learning_rate": 9.523218611855552e-05,
+      "loss": 9.198,
+      "step": 13830
+    },
+    {
+      "epoch": 0.6789430759702472,
+      "grad_norm": 1.0546875,
+      "learning_rate": 9.522001602647855e-05,
+      "loss": 8.844,
+      "step": 13840
+    },
+    {
+      "epoch": 0.679433641776584,
+      "grad_norm": 1.4140625,
+      "learning_rate": 9.520783120146307e-05,
+      "loss": 10.2174,
+      "step": 13850
+    },
+    {
+      "epoch": 0.6799242075829209,
+      "grad_norm": 1.03125,
+      "learning_rate": 9.519563164747896e-05,
+      "loss": 9.1452,
+      "step": 13860
+    },
+    {
+      "epoch": 0.6804147733892578,
+      "grad_norm": 1.125,
+      "learning_rate": 9.518341736850091e-05,
+      "loss": 9.3896,
+      "step": 13870
+    },
+    {
+      "epoch": 0.6809053391955947,
+      "grad_norm": 0.99609375,
+      "learning_rate": 9.517118836850843e-05,
+      "loss": 9.0489,
+      "step": 13880
+    },
+    {
+      "epoch": 0.6813959050019316,
+      "grad_norm": 1.4296875,
+      "learning_rate": 9.515894465148574e-05,
+      "loss": 9.5612,
+      "step": 13890
+    },
+    {
+      "epoch": 0.6818864708082685,
+      "grad_norm": 1.328125,
+      "learning_rate": 9.514668622142193e-05,
+      "loss": 9.1809,
+      "step": 13900
+    },
+    {
+      "epoch": 0.6823770366146054,
+      "grad_norm": 1.3671875,
+      "learning_rate": 9.51344130823109e-05,
+      "loss": 9.2658,
+      "step": 13910
+    },
+    {
+      "epoch": 0.6828676024209422,
+      "grad_norm": 0.9921875,
+      "learning_rate": 9.512212523815129e-05,
+      "loss": 8.724,
+      "step": 13920
+    },
+    {
+      "epoch": 0.6833581682272791,
+      "grad_norm": 1.7890625,
+      "learning_rate": 9.510982269294652e-05,
+      "loss": 9.9933,
+      "step": 13930
+    },
+    {
+      "epoch": 0.683848734033616,
+      "grad_norm": 2.15625,
+      "learning_rate": 9.509750545070488e-05,
+      "loss": 9.8759,
+      "step": 13940
+    },
+    {
+      "epoch": 0.6843392998399529,
+      "grad_norm": 1.1328125,
+      "learning_rate": 9.508517351543938e-05,
+      "loss": 9.3409,
+      "step": 13950
+    },
+    {
+      "epoch": 0.6848298656462898,
+      "grad_norm": 1.9296875,
+      "learning_rate": 9.50728268911678e-05,
+      "loss": 9.4605,
+      "step": 13960
+    },
+    {
+      "epoch": 0.6853204314526267,
+      "grad_norm": 1.34375,
+      "learning_rate": 9.506046558191277e-05,
+      "loss": 9.2771,
+      "step": 13970
+    },
+    {
+      "epoch": 0.6858109972589636,
+      "grad_norm": 1.359375,
+      "learning_rate": 9.504808959170168e-05,
+      "loss": 9.2664,
+      "step": 13980
+    },
+    {
+      "epoch": 0.6863015630653004,
+      "grad_norm": 0.83984375,
+      "learning_rate": 9.50356989245667e-05,
+      "loss": 8.6913,
+      "step": 13990
+    },
+    {
+      "epoch": 0.6867921288716373,
+      "grad_norm": 1.1953125,
+      "learning_rate": 9.502329358454477e-05,
+      "loss": 9.3056,
+      "step": 14000
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 61155,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "save_steps": 2000,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 3.766227025526784e+19,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}