{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.0,
  "eval_steps": 500,
  "global_step": 38334,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0007825950853028644,
      "grad_norm": 8.824877738952637,
      "learning_rate": 4.5e-06,
      "loss": 1.3738,
      "step": 10
    },
    {
      "epoch": 0.0015651901706057287,
      "grad_norm": 7.243584632873535,
      "learning_rate": 9.5e-06,
      "loss": 1.3581,
      "step": 20
    },
    {
      "epoch": 0.0023477852559085927,
      "grad_norm": 5.590239524841309,
      "learning_rate": 1.45e-05,
      "loss": 1.0634,
      "step": 30
    },
    {
      "epoch": 0.0031303803412114574,
      "grad_norm": 4.286197662353516,
      "learning_rate": 1.9500000000000003e-05,
      "loss": 1.0189,
      "step": 40
    },
    {
      "epoch": 0.003912975426514321,
      "grad_norm": 3.216794013977051,
      "learning_rate": 2.45e-05,
      "loss": 1.1141,
      "step": 50
    },
    {
      "epoch": 0.0046955705118171855,
      "grad_norm": 8.824544906616211,
      "learning_rate": 2.95e-05,
      "loss": 1.0233,
      "step": 60
    },
    {
      "epoch": 0.00547816559712005,
      "grad_norm": 5.246798992156982,
      "learning_rate": 3.45e-05,
      "loss": 1.1139,
      "step": 70
    },
    {
      "epoch": 0.006260760682422915,
      "grad_norm": 9.952336311340332,
      "learning_rate": 3.9500000000000005e-05,
      "loss": 0.8867,
      "step": 80
    },
    {
      "epoch": 0.007043355767725779,
      "grad_norm": 4.7690300941467285,
      "learning_rate": 4.4500000000000004e-05,
      "loss": 1.0453,
      "step": 90
    },
    {
      "epoch": 0.007825950853028642,
      "grad_norm": 8.076007843017578,
      "learning_rate": 4.9500000000000004e-05,
      "loss": 1.2364,
      "step": 100
    },
    {
      "epoch": 0.008608545938331507,
      "grad_norm": 4.618903636932373,
      "learning_rate": 4.998823037087409e-05,
      "loss": 0.8758,
      "step": 110
    },
    {
      "epoch": 0.009391141023634371,
      "grad_norm": 1.8372598886489868,
      "learning_rate": 4.997515300517864e-05,
      "loss": 0.658,
      "step": 120
    },
    {
      "epoch": 0.010173736108937235,
      "grad_norm": 1.834579348564148,
      "learning_rate": 4.996207563948318e-05,
      "loss": 0.93,
      "step": 130
    },
    {
      "epoch": 0.0109563311942401,
      "grad_norm": 6.350945472717285,
      "learning_rate": 4.994899827378773e-05,
      "loss": 0.7286,
      "step": 140
    },
    {
      "epoch": 0.011738926279542964,
      "grad_norm": 4.728095054626465,
      "learning_rate": 4.993592090809228e-05,
      "loss": 0.8131,
      "step": 150
    },
    {
      "epoch": 0.01252152136484583,
      "grad_norm": 7.766292095184326,
      "learning_rate": 4.992284354239682e-05,
      "loss": 0.9469,
      "step": 160
    },
    {
      "epoch": 0.013304116450148694,
      "grad_norm": 8.075868606567383,
      "learning_rate": 4.990976617670137e-05,
      "loss": 0.999,
      "step": 170
    },
    {
      "epoch": 0.014086711535451558,
      "grad_norm": 6.597504615783691,
      "learning_rate": 4.989668881100591e-05,
      "loss": 0.894,
      "step": 180
    },
    {
      "epoch": 0.014869306620754422,
      "grad_norm": 16.1438045501709,
      "learning_rate": 4.9883611445310457e-05,
      "loss": 1.0953,
      "step": 190
    },
    {
      "epoch": 0.015651901706057285,
      "grad_norm": 7.803539276123047,
      "learning_rate": 4.987053407961501e-05,
      "loss": 1.1391,
      "step": 200
    },
    {
      "epoch": 0.01643449679136015,
      "grad_norm": 7.40112829208374,
      "learning_rate": 4.9857456713919546e-05,
      "loss": 0.8126,
      "step": 210
    },
    {
      "epoch": 0.017217091876663013,
      "grad_norm": 93.28771209716797,
      "learning_rate": 4.98443793482241e-05,
      "loss": 1.0908,
      "step": 220
    },
    {
      "epoch": 0.01799968696196588,
      "grad_norm": 7.3326735496521,
      "learning_rate": 4.983130198252864e-05,
      "loss": 1.001,
      "step": 230
    },
    {
      "epoch": 0.018782282047268742,
      "grad_norm": 10.945075035095215,
      "learning_rate": 4.9818224616833186e-05,
      "loss": 0.9465,
      "step": 240
    },
    {
      "epoch": 0.019564877132571608,
      "grad_norm": 8.686254501342773,
      "learning_rate": 4.980514725113774e-05,
      "loss": 0.8682,
      "step": 250
    },
    {
      "epoch": 0.02034747221787447,
      "grad_norm": 7.731786727905273,
      "learning_rate": 4.9792069885442275e-05,
      "loss": 1.1012,
      "step": 260
    },
    {
      "epoch": 0.021130067303177336,
      "grad_norm": 5.565306186676025,
      "learning_rate": 4.977899251974683e-05,
      "loss": 0.6625,
      "step": 270
    },
    {
      "epoch": 0.0219126623884802,
      "grad_norm": 7.283803939819336,
      "learning_rate": 4.976591515405137e-05,
      "loss": 1.211,
      "step": 280
    },
    {
      "epoch": 0.022695257473783065,
      "grad_norm": 8.42110824584961,
      "learning_rate": 4.9752837788355916e-05,
      "loss": 0.7628,
      "step": 290
    },
    {
      "epoch": 0.023477852559085927,
      "grad_norm": 17.0642032623291,
      "learning_rate": 4.973976042266047e-05,
      "loss": 1.0262,
      "step": 300
    },
    {
      "epoch": 0.024260447644388793,
      "grad_norm": 2.226909637451172,
      "learning_rate": 4.9726683056965005e-05,
      "loss": 0.9703,
      "step": 310
    },
    {
      "epoch": 0.02504304272969166,
      "grad_norm": 2.0942604541778564,
      "learning_rate": 4.9713605691269556e-05,
      "loss": 1.3457,
      "step": 320
    },
    {
      "epoch": 0.025825637814994522,
      "grad_norm": 8.412962913513184,
      "learning_rate": 4.9700528325574094e-05,
      "loss": 0.9652,
      "step": 330
    },
    {
      "epoch": 0.026608232900297388,
      "grad_norm": 12.943373680114746,
      "learning_rate": 4.9687450959878645e-05,
      "loss": 0.937,
      "step": 340
    },
    {
      "epoch": 0.02739082798560025,
      "grad_norm": 3.5882763862609863,
      "learning_rate": 4.967437359418319e-05,
      "loss": 0.8246,
      "step": 350
    },
    {
      "epoch": 0.028173423070903116,
      "grad_norm": 6.775402545928955,
      "learning_rate": 4.9661296228487735e-05,
      "loss": 1.0439,
      "step": 360
    },
    {
      "epoch": 0.02895601815620598,
      "grad_norm": 5.154697418212891,
      "learning_rate": 4.9648218862792286e-05,
      "loss": 0.6955,
      "step": 370
    },
    {
      "epoch": 0.029738613241508845,
      "grad_norm": 8.044297218322754,
      "learning_rate": 4.9635141497096824e-05,
      "loss": 1.2601,
      "step": 380
    },
    {
      "epoch": 0.030521208326811707,
      "grad_norm": 2.093432903289795,
      "learning_rate": 4.9622064131401375e-05,
      "loss": 0.9007,
      "step": 390
    },
    {
      "epoch": 0.03130380341211457,
      "grad_norm": 5.379454135894775,
      "learning_rate": 4.960898676570592e-05,
      "loss": 0.7426,
      "step": 400
    },
    {
      "epoch": 0.032086398497417436,
      "grad_norm": 5.305158615112305,
      "learning_rate": 4.9595909400010464e-05,
      "loss": 0.4897,
      "step": 410
    },
    {
      "epoch": 0.0328689935827203,
      "grad_norm": 12.261921882629395,
      "learning_rate": 4.958283203431501e-05,
      "loss": 1.0303,
      "step": 420
    },
    {
      "epoch": 0.03365158866802317,
      "grad_norm": 4.6564483642578125,
      "learning_rate": 4.956975466861955e-05,
      "loss": 0.8591,
      "step": 430
    },
    {
      "epoch": 0.03443418375332603,
      "grad_norm": 7.278219223022461,
      "learning_rate": 4.9556677302924105e-05,
      "loss": 1.0999,
      "step": 440
    },
    {
      "epoch": 0.03521677883862889,
      "grad_norm": 20.723955154418945,
      "learning_rate": 4.954359993722865e-05,
      "loss": 0.7756,
      "step": 450
    },
    {
      "epoch": 0.03599937392393176,
      "grad_norm": 7.310290336608887,
      "learning_rate": 4.9530522571533194e-05,
      "loss": 1.0384,
      "step": 460
    },
    {
      "epoch": 0.036781969009234625,
      "grad_norm": 6.938417911529541,
      "learning_rate": 4.951744520583774e-05,
      "loss": 1.1484,
      "step": 470
    },
    {
      "epoch": 0.037564564094537484,
      "grad_norm": 10.381139755249023,
      "learning_rate": 4.950436784014228e-05,
      "loss": 0.6538,
      "step": 480
    },
    {
      "epoch": 0.03834715917984035,
      "grad_norm": 9.069252967834473,
      "learning_rate": 4.949129047444683e-05,
      "loss": 1.3562,
      "step": 490
    },
    {
      "epoch": 0.039129754265143216,
      "grad_norm": 8.355674743652344,
      "learning_rate": 4.947821310875138e-05,
      "loss": 0.7836,
      "step": 500
    },
    {
      "epoch": 0.03991234935044608,
      "grad_norm": 11.738659858703613,
      "learning_rate": 4.9465135743055924e-05,
      "loss": 1.0835,
      "step": 510
    },
    {
      "epoch": 0.04069494443574894,
      "grad_norm": 10.903166770935059,
      "learning_rate": 4.945205837736047e-05,
      "loss": 1.1446,
      "step": 520
    },
    {
      "epoch": 0.04147753952105181,
      "grad_norm": 8.311982154846191,
      "learning_rate": 4.943898101166501e-05,
      "loss": 0.7397,
      "step": 530
    },
    {
      "epoch": 0.04226013460635467,
      "grad_norm": 5.930412769317627,
      "learning_rate": 4.942590364596956e-05,
      "loss": 0.8393,
      "step": 540
    },
    {
      "epoch": 0.04304272969165754,
      "grad_norm": 6.0082221031188965,
      "learning_rate": 4.94128262802741e-05,
      "loss": 0.8647,
      "step": 550
    },
    {
      "epoch": 0.0438253247769604,
      "grad_norm": 7.8167500495910645,
      "learning_rate": 4.9399748914578646e-05,
      "loss": 0.6437,
      "step": 560
    },
    {
      "epoch": 0.044607919862263264,
      "grad_norm": 4.5714430809021,
      "learning_rate": 4.93866715488832e-05,
      "loss": 0.8295,
      "step": 570
    },
    {
      "epoch": 0.04539051494756613,
      "grad_norm": 8.749919891357422,
      "learning_rate": 4.937359418318774e-05,
      "loss": 0.9843,
      "step": 580
    },
    {
      "epoch": 0.046173110032868996,
      "grad_norm": 5.6613006591796875,
      "learning_rate": 4.936051681749229e-05,
      "loss": 0.811,
      "step": 590
    },
    {
      "epoch": 0.046955705118171855,
      "grad_norm": 9.800485610961914,
      "learning_rate": 4.934743945179683e-05,
      "loss": 1.2225,
      "step": 600
    },
    {
      "epoch": 0.04773830020347472,
      "grad_norm": 2.9545469284057617,
      "learning_rate": 4.9334362086101376e-05,
      "loss": 0.9004,
      "step": 610
    },
    {
      "epoch": 0.04852089528877759,
      "grad_norm": 5.178247451782227,
      "learning_rate": 4.932128472040593e-05,
      "loss": 0.7071,
      "step": 620
    },
    {
      "epoch": 0.04930349037408045,
      "grad_norm": 8.115721702575684,
      "learning_rate": 4.9308207354710465e-05,
      "loss": 0.9897,
      "step": 630
    },
    {
      "epoch": 0.05008608545938332,
      "grad_norm": 5.324657917022705,
      "learning_rate": 4.9295129989015017e-05,
      "loss": 0.9439,
      "step": 640
    },
    {
      "epoch": 0.05086868054468618,
      "grad_norm": 8.37894344329834,
      "learning_rate": 4.928205262331956e-05,
      "loss": 0.6966,
      "step": 650
    },
    {
      "epoch": 0.051651275629989044,
      "grad_norm": 8.153889656066895,
      "learning_rate": 4.9268975257624106e-05,
      "loss": 0.7854,
      "step": 660
    },
    {
      "epoch": 0.05243387071529191,
      "grad_norm": 3.3832263946533203,
      "learning_rate": 4.925589789192866e-05,
      "loss": 1.0184,
      "step": 670
    },
    {
      "epoch": 0.053216465800594775,
      "grad_norm": 31.724245071411133,
      "learning_rate": 4.9242820526233195e-05,
      "loss": 0.6219,
      "step": 680
    },
    {
      "epoch": 0.053999060885897635,
      "grad_norm": 12.828802108764648,
      "learning_rate": 4.9229743160537746e-05,
      "loss": 0.621,
      "step": 690
    },
    {
      "epoch": 0.0547816559712005,
      "grad_norm": 6.971835613250732,
      "learning_rate": 4.9216665794842284e-05,
      "loss": 1.1402,
      "step": 700
    },
    {
      "epoch": 0.055564251056503366,
      "grad_norm": 3.5057766437530518,
      "learning_rate": 4.9203588429146835e-05,
      "loss": 1.0843,
      "step": 710
    },
    {
      "epoch": 0.05634684614180623,
      "grad_norm": 7.382892608642578,
      "learning_rate": 4.919051106345139e-05,
      "loss": 0.8703,
      "step": 720
    },
    {
      "epoch": 0.05712944122710909,
      "grad_norm": 9.188787460327148,
      "learning_rate": 4.9177433697755924e-05,
      "loss": 1.0394,
      "step": 730
    },
    {
      "epoch": 0.05791203631241196,
      "grad_norm": 8.636894226074219,
      "learning_rate": 4.9164356332060476e-05,
      "loss": 0.8996,
      "step": 740
    },
    {
      "epoch": 0.05869463139771482,
      "grad_norm": 10.068634033203125,
      "learning_rate": 4.9151278966365014e-05,
      "loss": 1.4501,
      "step": 750
    },
    {
      "epoch": 0.05947722648301769,
      "grad_norm": 8.645387649536133,
      "learning_rate": 4.9138201600669565e-05,
      "loss": 0.9979,
      "step": 760
    },
    {
      "epoch": 0.06025982156832055,
      "grad_norm": 6.682597637176514,
      "learning_rate": 4.912512423497411e-05,
      "loss": 0.9332,
      "step": 770
    },
    {
      "epoch": 0.061042416653623414,
      "grad_norm": 1.738028883934021,
      "learning_rate": 4.9112046869278654e-05,
      "loss": 0.6189,
      "step": 780
    },
    {
      "epoch": 0.06182501173892628,
      "grad_norm": 10.897079467773438,
      "learning_rate": 4.9098969503583205e-05,
      "loss": 1.1003,
      "step": 790
    },
    {
      "epoch": 0.06260760682422914,
      "grad_norm": 7.05759859085083,
      "learning_rate": 4.908589213788774e-05,
      "loss": 1.0562,
      "step": 800
    },
    {
      "epoch": 0.06339020190953201,
      "grad_norm": 6.998068332672119,
      "learning_rate": 4.9072814772192295e-05,
      "loss": 1.0437,
      "step": 810
    },
    {
      "epoch": 0.06417279699483487,
      "grad_norm": 7.598848819732666,
      "learning_rate": 4.905973740649684e-05,
      "loss": 0.8614,
      "step": 820
    },
    {
      "epoch": 0.06495539208013773,
      "grad_norm": 4.302680492401123,
      "learning_rate": 4.9046660040801384e-05,
      "loss": 0.9334,
      "step": 830
    },
    {
      "epoch": 0.0657379871654406,
      "grad_norm": 7.4813690185546875,
      "learning_rate": 4.903358267510593e-05,
      "loss": 1.1129,
      "step": 840
    },
    {
      "epoch": 0.06652058225074346,
      "grad_norm": 6.9071526527404785,
      "learning_rate": 4.902050530941047e-05,
      "loss": 0.7529,
      "step": 850
    },
    {
      "epoch": 0.06730317733604634,
      "grad_norm": 7.265669822692871,
      "learning_rate": 4.9007427943715024e-05,
      "loss": 1.0385,
      "step": 860
    },
    {
      "epoch": 0.0680857724213492,
      "grad_norm": 5.417905330657959,
      "learning_rate": 4.899435057801957e-05,
      "loss": 0.8374,
      "step": 870
    },
    {
      "epoch": 0.06886836750665205,
      "grad_norm": 6.535093307495117,
      "learning_rate": 4.898127321232411e-05,
      "loss": 0.9643,
      "step": 880
    },
    {
      "epoch": 0.06965096259195493,
      "grad_norm": 10.06280517578125,
      "learning_rate": 4.896819584662866e-05,
      "loss": 0.7333,
      "step": 890
    },
    {
      "epoch": 0.07043355767725779,
      "grad_norm": 4.503497123718262,
      "learning_rate": 4.89551184809332e-05,
      "loss": 1.0591,
      "step": 900
    },
    {
      "epoch": 0.07121615276256064,
      "grad_norm": 7.74230432510376,
      "learning_rate": 4.894204111523775e-05,
      "loss": 0.955,
      "step": 910
    },
    {
      "epoch": 0.07199874784786352,
      "grad_norm": 5.705163955688477,
      "learning_rate": 4.892896374954229e-05,
      "loss": 0.7766,
      "step": 920
    },
    {
      "epoch": 0.07278134293316638,
      "grad_norm": 8.884862899780273,
      "learning_rate": 4.891588638384684e-05,
      "loss": 1.0704,
      "step": 930
    },
    {
      "epoch": 0.07356393801846925,
      "grad_norm": 7.701562404632568,
      "learning_rate": 4.890280901815139e-05,
      "loss": 0.7437,
      "step": 940
    },
    {
      "epoch": 0.07434653310377211,
      "grad_norm": 5.332188606262207,
      "learning_rate": 4.888973165245593e-05,
      "loss": 0.918,
      "step": 950
    },
    {
      "epoch": 0.07512912818907497,
      "grad_norm": 5.704586505889893,
      "learning_rate": 4.887665428676048e-05,
      "loss": 1.455,
      "step": 960
    },
    {
      "epoch": 0.07591172327437784,
      "grad_norm": 11.023058891296387,
      "learning_rate": 4.886357692106502e-05,
      "loss": 1.0372,
      "step": 970
    },
    {
      "epoch": 0.0766943183596807,
      "grad_norm": 5.437788486480713,
      "learning_rate": 4.8850499555369566e-05,
      "loss": 1.0157,
      "step": 980
    },
    {
      "epoch": 0.07747691344498357,
      "grad_norm": 6.113413333892822,
      "learning_rate": 4.883742218967412e-05,
      "loss": 0.8535,
      "step": 990
    },
    {
      "epoch": 0.07825950853028643,
      "grad_norm": 5.188055992126465,
      "learning_rate": 4.882434482397866e-05,
      "loss": 0.9289,
      "step": 1000
    },
    {
      "epoch": 0.07904210361558929,
      "grad_norm": 5.30755090713501,
      "learning_rate": 4.8811267458283206e-05,
      "loss": 0.7754,
      "step": 1010
    },
    {
      "epoch": 0.07982469870089216,
      "grad_norm": 4.937777519226074,
      "learning_rate": 4.879819009258775e-05,
      "loss": 0.8724,
      "step": 1020
    },
    {
      "epoch": 0.08060729378619502,
      "grad_norm": 5.260272026062012,
      "learning_rate": 4.8785112726892295e-05,
      "loss": 1.2297,
      "step": 1030
    },
    {
      "epoch": 0.08138988887149788,
      "grad_norm": 5.706908702850342,
      "learning_rate": 4.877203536119685e-05,
      "loss": 1.1276,
      "step": 1040
    },
    {
      "epoch": 0.08217248395680075,
      "grad_norm": 5.526937484741211,
      "learning_rate": 4.8758957995501385e-05,
      "loss": 0.7496,
      "step": 1050
    },
    {
      "epoch": 0.08295507904210361,
      "grad_norm": 10.0362548828125,
      "learning_rate": 4.8745880629805936e-05,
      "loss": 0.773,
      "step": 1060
    },
    {
      "epoch": 0.08373767412740649,
      "grad_norm": 3.359102964401245,
      "learning_rate": 4.873280326411048e-05,
      "loss": 0.6729,
      "step": 1070
    },
    {
      "epoch": 0.08452026921270935,
      "grad_norm": 8.838607788085938,
      "learning_rate": 4.8719725898415025e-05,
      "loss": 1.2747,
      "step": 1080
    },
    {
      "epoch": 0.0853028642980122,
      "grad_norm": 4.321221351623535,
      "learning_rate": 4.8706648532719576e-05,
      "loss": 0.8138,
      "step": 1090
    },
    {
      "epoch": 0.08608545938331508,
      "grad_norm": 10.322368621826172,
      "learning_rate": 4.8693571167024114e-05,
      "loss": 0.9185,
      "step": 1100
    },
    {
      "epoch": 0.08686805446861794,
      "grad_norm": 2.670121908187866,
      "learning_rate": 4.8680493801328666e-05,
      "loss": 0.7382,
      "step": 1110
    },
    {
      "epoch": 0.0876506495539208,
      "grad_norm": 30.149219512939453,
      "learning_rate": 4.8667416435633203e-05,
      "loss": 0.9991,
      "step": 1120
    },
    {
      "epoch": 0.08843324463922367,
      "grad_norm": 12.900429725646973,
      "learning_rate": 4.8654339069937755e-05,
      "loss": 0.6006,
      "step": 1130
    },
    {
      "epoch": 0.08921583972452653,
      "grad_norm": 6.6820068359375,
      "learning_rate": 4.86412617042423e-05,
      "loss": 0.6391,
      "step": 1140
    },
    {
      "epoch": 0.0899984348098294,
      "grad_norm": 11.987547874450684,
      "learning_rate": 4.8628184338546844e-05,
      "loss": 1.2152,
      "step": 1150
    },
    {
      "epoch": 0.09078102989513226,
      "grad_norm": 1.6027709245681763,
      "learning_rate": 4.8615106972851395e-05,
      "loss": 1.1554,
      "step": 1160
    },
    {
      "epoch": 0.09156362498043512,
      "grad_norm": 10.42098331451416,
      "learning_rate": 4.860202960715593e-05,
      "loss": 0.6806,
      "step": 1170
    },
    {
      "epoch": 0.09234622006573799,
      "grad_norm": 1.23116135597229,
      "learning_rate": 4.8588952241460484e-05,
      "loss": 0.635,
      "step": 1180
    },
    {
      "epoch": 0.09312881515104085,
      "grad_norm": 9.524988174438477,
      "learning_rate": 4.857587487576503e-05,
      "loss": 0.7024,
      "step": 1190
    },
    {
      "epoch": 0.09391141023634371,
      "grad_norm": 7.468221664428711,
      "learning_rate": 4.8562797510069574e-05,
      "loss": 0.7544,
      "step": 1200
    },
    {
      "epoch": 0.09469400532164658,
      "grad_norm": 18.16954231262207,
      "learning_rate": 4.8549720144374125e-05,
      "loss": 1.0721,
      "step": 1210
    },
    {
      "epoch": 0.09547660040694944,
      "grad_norm": 10.648679733276367,
      "learning_rate": 4.853664277867866e-05,
      "loss": 0.845,
      "step": 1220
    },
    {
      "epoch": 0.09625919549225231,
      "grad_norm": 5.016397476196289,
      "learning_rate": 4.8523565412983214e-05,
      "loss": 0.9784,
      "step": 1230
    },
    {
      "epoch": 0.09704179057755517,
      "grad_norm": 13.620104789733887,
      "learning_rate": 4.851048804728776e-05,
      "loss": 0.8142,
      "step": 1240
    },
    {
      "epoch": 0.09782438566285803,
      "grad_norm": 1.6859623193740845,
      "learning_rate": 4.84974106815923e-05,
      "loss": 0.743,
      "step": 1250
    },
    {
      "epoch": 0.0986069807481609,
      "grad_norm": 12.822032928466797,
      "learning_rate": 4.848433331589685e-05,
      "loss": 1.3022,
      "step": 1260
    },
    {
      "epoch": 0.09938957583346376,
      "grad_norm": 20.77387237548828,
      "learning_rate": 4.847125595020139e-05,
      "loss": 1.0182,
      "step": 1270
    },
    {
      "epoch": 0.10017217091876664,
      "grad_norm": 1.7113702297210693,
      "learning_rate": 4.8458178584505944e-05,
      "loss": 1.002,
      "step": 1280
    },
    {
      "epoch": 0.1009547660040695,
      "grad_norm": 5.503539562225342,
      "learning_rate": 4.844510121881048e-05,
      "loss": 1.2507,
      "step": 1290
    },
    {
      "epoch": 0.10173736108937236,
      "grad_norm": 11.611311912536621,
      "learning_rate": 4.843202385311503e-05,
      "loss": 0.816,
      "step": 1300
    },
    {
      "epoch": 0.10251995617467523,
      "grad_norm": 4.955935955047607,
      "learning_rate": 4.841894648741958e-05,
      "loss": 0.8376,
      "step": 1310
    },
    {
      "epoch": 0.10330255125997809,
      "grad_norm": 4.007413864135742,
      "learning_rate": 4.840586912172412e-05,
      "loss": 0.748,
      "step": 1320
    },
    {
      "epoch": 0.10408514634528095,
      "grad_norm": 12.696671485900879,
      "learning_rate": 4.8392791756028667e-05,
      "loss": 0.888,
      "step": 1330
    },
    {
      "epoch": 0.10486774143058382,
      "grad_norm": 1.6879503726959229,
      "learning_rate": 4.837971439033321e-05,
      "loss": 0.7054,
      "step": 1340
    },
    {
      "epoch": 0.10565033651588668,
      "grad_norm": 4.341312408447266,
      "learning_rate": 4.836663702463776e-05,
      "loss": 0.9209,
      "step": 1350
    },
    {
      "epoch": 0.10643293160118955,
      "grad_norm": 3.5054709911346436,
      "learning_rate": 4.835355965894231e-05,
      "loss": 0.7145,
      "step": 1360
    },
    {
      "epoch": 0.10721552668649241,
      "grad_norm": 13.475558280944824,
      "learning_rate": 4.834048229324685e-05,
      "loss": 0.8157,
      "step": 1370
    },
    {
      "epoch": 0.10799812177179527,
      "grad_norm": 11.342016220092773,
      "learning_rate": 4.8327404927551396e-05,
      "loss": 0.8103,
      "step": 1380
    },
    {
      "epoch": 0.10878071685709814,
      "grad_norm": 15.065884590148926,
      "learning_rate": 4.831432756185594e-05,
      "loss": 0.9404,
      "step": 1390
    },
    {
      "epoch": 0.109563311942401,
      "grad_norm": 1.9674978256225586,
      "learning_rate": 4.8301250196160485e-05,
      "loss": 0.8777,
      "step": 1400
    },
    {
      "epoch": 0.11034590702770386,
      "grad_norm": 8.363801956176758,
      "learning_rate": 4.828817283046504e-05,
      "loss": 0.7516,
      "step": 1410
    },
    {
      "epoch": 0.11112850211300673,
      "grad_norm": 1.406518578529358,
      "learning_rate": 4.827509546476958e-05,
      "loss": 0.8184,
      "step": 1420
    },
    {
      "epoch": 0.11191109719830959,
      "grad_norm": 6.014228343963623,
      "learning_rate": 4.8262018099074126e-05,
      "loss": 1.1715,
      "step": 1430
    },
    {
      "epoch": 0.11269369228361246,
      "grad_norm": 8.163867950439453,
      "learning_rate": 4.824894073337867e-05,
      "loss": 1.1901,
      "step": 1440
    },
    {
      "epoch": 0.11347628736891532,
      "grad_norm": 5.873777866363525,
      "learning_rate": 4.8235863367683215e-05,
      "loss": 0.5119,
      "step": 1450
    },
    {
      "epoch": 0.11425888245421818,
      "grad_norm": 19.948497772216797,
      "learning_rate": 4.8222786001987766e-05,
      "loss": 0.8981,
      "step": 1460
    },
    {
      "epoch": 0.11504147753952106,
      "grad_norm": 12.005247116088867,
      "learning_rate": 4.8209708636292304e-05,
      "loss": 0.7426,
      "step": 1470
    },
    {
      "epoch": 0.11582407262482391,
      "grad_norm": 4.2786407470703125,
      "learning_rate": 4.8196631270596855e-05,
      "loss": 1.0124,
      "step": 1480
    },
    {
      "epoch": 0.11660666771012677,
      "grad_norm": 5.5871381759643555,
      "learning_rate": 4.818355390490139e-05,
      "loss": 1.1837,
      "step": 1490
    },
    {
      "epoch": 0.11738926279542965,
      "grad_norm": 4.274723529815674,
      "learning_rate": 4.8170476539205945e-05,
      "loss": 0.748,
      "step": 1500
    },
    {
      "epoch": 0.1181718578807325,
      "grad_norm": 12.916749000549316,
      "learning_rate": 4.815739917351049e-05,
      "loss": 1.2706,
      "step": 1510
    },
    {
      "epoch": 0.11895445296603538,
      "grad_norm": 10.5656156539917,
      "learning_rate": 4.8144321807815034e-05,
      "loss": 1.0388,
      "step": 1520
    },
    {
      "epoch": 0.11973704805133824,
      "grad_norm": 2.746166229248047,
      "learning_rate": 4.8131244442119585e-05,
      "loss": 1.1119,
      "step": 1530
    },
    {
      "epoch": 0.1205196431366411,
      "grad_norm": 12.125962257385254,
      "learning_rate": 4.811816707642412e-05,
      "loss": 0.8528,
      "step": 1540
    },
    {
      "epoch": 0.12130223822194397,
      "grad_norm": 5.979024887084961,
      "learning_rate": 4.8105089710728674e-05,
      "loss": 0.7995,
      "step": 1550
    },
    {
      "epoch": 0.12208483330724683,
      "grad_norm": 22.572675704956055,
      "learning_rate": 4.809201234503322e-05,
      "loss": 0.8757,
      "step": 1560
    },
    {
      "epoch": 0.12286742839254969,
      "grad_norm": 6.074213027954102,
      "learning_rate": 4.807893497933776e-05,
      "loss": 0.9717,
      "step": 1570
    },
    {
      "epoch": 0.12365002347785256,
      "grad_norm": 8.571769714355469,
      "learning_rate": 4.8065857613642315e-05,
      "loss": 0.8973,
      "step": 1580
    },
    {
      "epoch": 0.12443261856315542,
      "grad_norm": 3.804872751235962,
      "learning_rate": 4.805278024794685e-05,
      "loss": 0.5681,
      "step": 1590
    },
    {
      "epoch": 0.12521521364845828,
      "grad_norm": 2.985506057739258,
      "learning_rate": 4.8039702882251404e-05,
      "loss": 0.6882,
      "step": 1600
    },
    {
      "epoch": 0.12599780873376115,
      "grad_norm": 2.2083003520965576,
      "learning_rate": 4.802662551655595e-05,
      "loss": 0.746,
      "step": 1610
    },
    {
      "epoch": 0.12678040381906402,
      "grad_norm": 2.118117332458496,
      "learning_rate": 4.801354815086049e-05,
      "loss": 0.8963,
      "step": 1620
    },
    {
      "epoch": 0.12756299890436687,
      "grad_norm": 2.6880273818969727,
      "learning_rate": 4.800047078516504e-05,
      "loss": 0.757,
      "step": 1630
    },
    {
      "epoch": 0.12834559398966974,
      "grad_norm": 4.991662502288818,
      "learning_rate": 4.798739341946958e-05,
      "loss": 0.784,
      "step": 1640
    },
    {
      "epoch": 0.12912818907497262,
      "grad_norm": 4.601889610290527,
      "learning_rate": 4.7974316053774134e-05,
      "loss": 0.6763,
      "step": 1650
    },
    {
      "epoch": 0.12991078416027546,
      "grad_norm": 4.33494234085083,
      "learning_rate": 4.796123868807867e-05,
      "loss": 0.9798,
      "step": 1660
    },
    {
      "epoch": 0.13069337924557833,
      "grad_norm": 12.356074333190918,
      "learning_rate": 4.794816132238322e-05,
      "loss": 1.1545,
      "step": 1670
    },
    {
      "epoch": 0.1314759743308812,
      "grad_norm": 7.004703044891357,
      "learning_rate": 4.793508395668777e-05,
      "loss": 0.7624,
      "step": 1680
    },
    {
      "epoch": 0.13225856941618408,
      "grad_norm": 5.557196140289307,
      "learning_rate": 4.792200659099231e-05,
      "loss": 0.782,
      "step": 1690
    },
    {
      "epoch": 0.13304116450148692,
      "grad_norm": 6.410905361175537,
      "learning_rate": 4.7908929225296856e-05,
      "loss": 0.8536,
      "step": 1700
    },
    {
      "epoch": 0.1338237595867898,
      "grad_norm": 9.07260513305664,
      "learning_rate": 4.78958518596014e-05,
      "loss": 0.8879,
      "step": 1710
    },
    {
      "epoch": 0.13460635467209267,
      "grad_norm": 10.27880573272705,
      "learning_rate": 4.788277449390595e-05,
      "loss": 0.8183,
      "step": 1720
    },
    {
      "epoch": 0.13538894975739552,
      "grad_norm": 7.777210712432861,
      "learning_rate": 4.78696971282105e-05,
      "loss": 1.0635,
      "step": 1730
    },
    {
      "epoch": 0.1361715448426984,
      "grad_norm": 7.529254913330078,
      "learning_rate": 4.785661976251504e-05,
      "loss": 1.0126,
      "step": 1740
    },
    {
      "epoch": 0.13695413992800126,
      "grad_norm": 6.8362650871276855,
      "learning_rate": 4.7843542396819586e-05,
      "loss": 0.977,
      "step": 1750
    },
    {
      "epoch": 0.1377367350133041,
      "grad_norm": 6.567052364349365,
      "learning_rate": 4.783046503112413e-05,
      "loss": 0.8691,
      "step": 1760
    },
    {
      "epoch": 0.13851933009860698,
      "grad_norm": 5.154236316680908,
      "learning_rate": 4.7817387665428675e-05,
      "loss": 0.6714,
      "step": 1770
    },
    {
      "epoch": 0.13930192518390985,
      "grad_norm": 9.093849182128906,
      "learning_rate": 4.7804310299733226e-05,
      "loss": 1.0161,
      "step": 1780
    },
    {
      "epoch": 0.1400845202692127,
      "grad_norm": 5.8644022941589355,
      "learning_rate": 4.779123293403777e-05,
      "loss": 0.9167,
      "step": 1790
    },
    {
      "epoch": 0.14086711535451557,
      "grad_norm": 10.522290229797363,
      "learning_rate": 4.7778155568342316e-05,
      "loss": 0.8447,
      "step": 1800
    },
    {
      "epoch": 0.14164971043981844,
      "grad_norm": 6.840420722961426,
      "learning_rate": 4.776507820264686e-05,
      "loss": 0.9769,
      "step": 1810
    },
    {
      "epoch": 0.1424323055251213,
      "grad_norm": 4.706451416015625,
      "learning_rate": 4.7752000836951405e-05,
      "loss": 0.8274,
      "step": 1820
    },
    {
      "epoch": 0.14321490061042416,
      "grad_norm": 3.5274391174316406,
      "learning_rate": 4.7738923471255956e-05,
      "loss": 0.8992,
      "step": 1830
    },
    {
      "epoch": 0.14399749569572703,
      "grad_norm": 2.3817696571350098,
      "learning_rate": 4.7725846105560494e-05,
      "loss": 0.6918,
      "step": 1840
    },
    {
      "epoch": 0.1447800907810299,
      "grad_norm": 5.326117992401123,
      "learning_rate": 4.7712768739865045e-05,
      "loss": 0.7262,
      "step": 1850
    },
    {
      "epoch": 0.14556268586633275,
      "grad_norm": 3.4415576457977295,
      "learning_rate": 4.769969137416959e-05,
      "loss": 0.7401,
      "step": 1860
    },
    {
      "epoch": 0.14634528095163563,
      "grad_norm": 4.568815231323242,
      "learning_rate": 4.7686614008474134e-05,
      "loss": 0.9085,
      "step": 1870
    },
    {
      "epoch": 0.1471278760369385,
      "grad_norm": 6.495962142944336,
      "learning_rate": 4.767353664277868e-05,
      "loss": 0.8737,
      "step": 1880
    },
    {
      "epoch": 0.14791047112224134,
      "grad_norm": 3.9560654163360596,
      "learning_rate": 4.7660459277083224e-05,
      "loss": 0.7522,
      "step": 1890
    },
    {
      "epoch": 0.14869306620754422,
      "grad_norm": 4.286434650421143,
      "learning_rate": 4.7647381911387775e-05,
      "loss": 0.6752,
      "step": 1900
    },
    {
      "epoch": 0.1494756612928471,
      "grad_norm": 10.119564056396484,
      "learning_rate": 4.763430454569231e-05,
      "loss": 0.9227,
      "step": 1910
    },
    {
      "epoch": 0.15025825637814993,
      "grad_norm": 27.79133415222168,
      "learning_rate": 4.7621227179996864e-05,
      "loss": 0.7156,
      "step": 1920
    },
    {
      "epoch": 0.1510408514634528,
      "grad_norm": 4.925629615783691,
      "learning_rate": 4.760814981430141e-05,
      "loss": 1.0409,
      "step": 1930
    },
    {
      "epoch": 0.15182344654875568,
      "grad_norm": 11.785629272460938,
      "learning_rate": 4.759507244860595e-05,
      "loss": 0.9478,
      "step": 1940
    },
    {
      "epoch": 0.15260604163405853,
      "grad_norm": 11.736627578735352,
      "learning_rate": 4.7581995082910505e-05,
      "loss": 1.0268,
      "step": 1950
    },
    {
      "epoch": 0.1533886367193614,
      "grad_norm": 4.255064010620117,
      "learning_rate": 4.756891771721504e-05,
      "loss": 1.2208,
      "step": 1960
    },
    {
      "epoch": 0.15417123180466427,
      "grad_norm": 1.3686619997024536,
      "learning_rate": 4.7555840351519594e-05,
      "loss": 0.7867,
      "step": 1970
    },
    {
      "epoch": 0.15495382688996714,
      "grad_norm": 33.7691650390625,
      "learning_rate": 4.754276298582414e-05,
      "loss": 0.9792,
      "step": 1980
    },
    {
      "epoch": 0.15573642197527,
      "grad_norm": 1.9538439512252808,
      "learning_rate": 4.752968562012868e-05,
      "loss": 0.8176,
      "step": 1990
    },
    {
      "epoch": 0.15651901706057286,
      "grad_norm": 7.140876293182373,
      "learning_rate": 4.7516608254433234e-05,
      "loss": 0.7125,
      "step": 2000
    },
    {
      "epoch": 0.15730161214587574,
      "grad_norm": 7.9959516525268555,
      "learning_rate": 4.750353088873777e-05,
      "loss": 1.0487,
      "step": 2010
    },
    {
      "epoch": 0.15808420723117858,
      "grad_norm": 5.557621955871582,
      "learning_rate": 4.749045352304232e-05,
      "loss": 0.9908,
      "step": 2020
    },
    {
      "epoch": 0.15886680231648145,
      "grad_norm": 3.401766300201416,
      "learning_rate": 4.747737615734686e-05,
      "loss": 0.8413,
      "step": 2030
    },
    {
      "epoch": 0.15964939740178433,
      "grad_norm": 1.9561461210250854,
      "learning_rate": 4.746429879165141e-05,
      "loss": 0.9273,
      "step": 2040
    },
    {
      "epoch": 0.16043199248708717,
      "grad_norm": 4.945536136627197,
      "learning_rate": 4.745122142595596e-05,
      "loss": 1.0679,
      "step": 2050
    },
    {
      "epoch": 0.16121458757239004,
      "grad_norm": 1.5364329814910889,
      "learning_rate": 4.74381440602605e-05,
      "loss": 1.0704,
      "step": 2060
    },
    {
      "epoch": 0.16199718265769292,
      "grad_norm": 4.17298698425293,
      "learning_rate": 4.742506669456505e-05,
      "loss": 0.9088,
      "step": 2070
    },
    {
      "epoch": 0.16277977774299576,
      "grad_norm": 10.479344367980957,
      "learning_rate": 4.741198932886959e-05,
      "loss": 0.9891,
      "step": 2080
    },
    {
      "epoch": 0.16356237282829864,
      "grad_norm": 3.378692388534546,
      "learning_rate": 4.739891196317414e-05,
      "loss": 0.8288,
      "step": 2090
    },
    {
      "epoch": 0.1643449679136015,
      "grad_norm": 5.812997341156006,
      "learning_rate": 4.738583459747869e-05,
      "loss": 0.8598,
      "step": 2100
    },
    {
      "epoch": 0.16512756299890435,
      "grad_norm": 3.5841050148010254,
      "learning_rate": 4.737275723178323e-05,
      "loss": 0.9635,
      "step": 2110
    },
    {
      "epoch": 0.16591015808420723,
      "grad_norm": 5.1485209465026855,
      "learning_rate": 4.7359679866087776e-05,
      "loss": 0.5112,
      "step": 2120
    },
    {
      "epoch": 0.1666927531695101,
      "grad_norm": 6.404463291168213,
      "learning_rate": 4.734660250039232e-05,
      "loss": 0.8861,
      "step": 2130
    },
    {
      "epoch": 0.16747534825481297,
      "grad_norm": 25.026241302490234,
      "learning_rate": 4.733352513469687e-05,
      "loss": 0.8734,
      "step": 2140
    },
    {
      "epoch": 0.16825794334011582,
      "grad_norm": 3.0007646083831787,
      "learning_rate": 4.7320447769001416e-05,
      "loss": 0.9908,
      "step": 2150
    },
    {
      "epoch": 0.1690405384254187,
      "grad_norm": 2.865469217300415,
      "learning_rate": 4.730737040330596e-05,
      "loss": 0.9741,
      "step": 2160
    },
    {
      "epoch": 0.16982313351072156,
      "grad_norm": 12.156944274902344,
      "learning_rate": 4.7294293037610505e-05,
      "loss": 0.9607,
      "step": 2170
    },
    {
      "epoch": 0.1706057285960244,
      "grad_norm": 21.078413009643555,
      "learning_rate": 4.728121567191505e-05,
      "loss": 0.766,
      "step": 2180
    },
    {
      "epoch": 0.17138832368132728,
      "grad_norm": 13.807541847229004,
      "learning_rate": 4.7268138306219595e-05,
      "loss": 1.1055,
      "step": 2190
    },
    {
      "epoch": 0.17217091876663015,
      "grad_norm": 14.438948631286621,
      "learning_rate": 4.7255060940524146e-05,
      "loss": 0.7465,
      "step": 2200
    },
    {
      "epoch": 0.172953513851933,
      "grad_norm": 3.7213642597198486,
      "learning_rate": 4.724198357482869e-05,
      "loss": 0.9525,
      "step": 2210
    },
    {
      "epoch": 0.17373610893723587,
      "grad_norm": 13.071233749389648,
      "learning_rate": 4.7228906209133235e-05,
      "loss": 0.7032,
      "step": 2220
    },
    {
      "epoch": 0.17451870402253875,
      "grad_norm": 6.262630462646484,
      "learning_rate": 4.721582884343778e-05,
      "loss": 1.3304,
      "step": 2230
    },
    {
      "epoch": 0.1753012991078416,
      "grad_norm": 16.427715301513672,
      "learning_rate": 4.7202751477742324e-05,
      "loss": 1.2157,
      "step": 2240
    },
    {
      "epoch": 0.17608389419314446,
      "grad_norm": 4.990220546722412,
      "learning_rate": 4.718967411204687e-05,
      "loss": 0.8671,
      "step": 2250
    },
    {
      "epoch": 0.17686648927844734,
      "grad_norm": 3.1963694095611572,
      "learning_rate": 4.7176596746351413e-05,
      "loss": 0.8913,
      "step": 2260
    },
    {
      "epoch": 0.1776490843637502,
      "grad_norm": 2.061413526535034,
      "learning_rate": 4.7163519380655965e-05,
      "loss": 0.6844,
      "step": 2270
    },
    {
      "epoch": 0.17843167944905305,
      "grad_norm": 4.668086051940918,
      "learning_rate": 4.715044201496051e-05,
      "loss": 1.0351,
      "step": 2280
    },
    {
      "epoch": 0.17921427453435593,
      "grad_norm": 8.851571083068848,
      "learning_rate": 4.7137364649265054e-05,
      "loss": 0.9137,
      "step": 2290
    },
    {
      "epoch": 0.1799968696196588,
      "grad_norm": 6.870872497558594,
      "learning_rate": 4.71242872835696e-05,
      "loss": 1.0831,
      "step": 2300
    },
    {
      "epoch": 0.18077946470496165,
      "grad_norm": 10.239887237548828,
      "learning_rate": 4.711120991787414e-05,
      "loss": 0.8808,
      "step": 2310
    },
    {
      "epoch": 0.18156205979026452,
      "grad_norm": 4.945154666900635,
      "learning_rate": 4.7098132552178694e-05,
      "loss": 0.9164,
      "step": 2320
    },
    {
      "epoch": 0.1823446548755674,
      "grad_norm": 5.1290998458862305,
      "learning_rate": 4.708505518648323e-05,
      "loss": 0.8717,
      "step": 2330
    },
    {
      "epoch": 0.18312724996087024,
      "grad_norm": 1.5949681997299194,
      "learning_rate": 4.7071977820787784e-05,
      "loss": 0.73,
      "step": 2340
    },
    {
      "epoch": 0.1839098450461731,
      "grad_norm": 10.23507022857666,
      "learning_rate": 4.705890045509233e-05,
      "loss": 1.3583,
      "step": 2350
    },
    {
      "epoch": 0.18469244013147598,
      "grad_norm": 13.008855819702148,
      "learning_rate": 4.704582308939687e-05,
      "loss": 0.7403,
      "step": 2360
    },
    {
      "epoch": 0.18547503521677883,
      "grad_norm": 5.918224811553955,
      "learning_rate": 4.7032745723701424e-05,
      "loss": 0.7977,
      "step": 2370
    },
    {
      "epoch": 0.1862576303020817,
      "grad_norm": 4.380965709686279,
      "learning_rate": 4.701966835800596e-05,
      "loss": 1.0987,
      "step": 2380
    },
    {
      "epoch": 0.18704022538738457,
      "grad_norm": 12.563697814941406,
      "learning_rate": 4.700659099231051e-05,
      "loss": 1.0774,
      "step": 2390
    },
    {
      "epoch": 0.18782282047268742,
      "grad_norm": 4.6697096824646,
      "learning_rate": 4.699351362661506e-05,
      "loss": 0.8629,
      "step": 2400
    },
    {
      "epoch": 0.1886054155579903,
      "grad_norm": 2.661947727203369,
      "learning_rate": 4.69804362609196e-05,
      "loss": 0.6285,
      "step": 2410
    },
    {
      "epoch": 0.18938801064329316,
      "grad_norm": 4.7253265380859375,
      "learning_rate": 4.6967358895224154e-05,
      "loss": 0.7703,
      "step": 2420
    },
    {
      "epoch": 0.19017060572859604,
      "grad_norm": 7.107388973236084,
      "learning_rate": 4.695428152952869e-05,
      "loss": 1.1848,
      "step": 2430
    },
    {
      "epoch": 0.19095320081389888,
      "grad_norm": 8.729486465454102,
      "learning_rate": 4.694120416383324e-05,
      "loss": 0.9243,
      "step": 2440
    },
    {
      "epoch": 0.19173579589920176,
      "grad_norm": 9.02910041809082,
      "learning_rate": 4.692812679813778e-05,
      "loss": 0.9284,
      "step": 2450
    },
    {
      "epoch": 0.19251839098450463,
      "grad_norm": 10.413541793823242,
      "learning_rate": 4.691504943244233e-05,
      "loss": 0.8921,
      "step": 2460
    },
    {
      "epoch": 0.19330098606980747,
      "grad_norm": 1.7911694049835205,
      "learning_rate": 4.6901972066746877e-05,
      "loss": 0.9405,
      "step": 2470
    },
    {
      "epoch": 0.19408358115511035,
      "grad_norm": 8.156761169433594,
      "learning_rate": 4.688889470105142e-05,
      "loss": 0.9637,
      "step": 2480
    },
    {
      "epoch": 0.19486617624041322,
      "grad_norm": 8.125580787658691,
      "learning_rate": 4.687581733535597e-05,
      "loss": 0.8374,
      "step": 2490
    },
    {
      "epoch": 0.19564877132571606,
      "grad_norm": 18.39687728881836,
      "learning_rate": 4.686273996966051e-05,
      "loss": 0.7913,
      "step": 2500
    },
    {
      "epoch": 0.19643136641101894,
      "grad_norm": 12.097225189208984,
      "learning_rate": 4.684966260396506e-05,
      "loss": 1.0305,
      "step": 2510
    },
    {
      "epoch": 0.1972139614963218,
      "grad_norm": 6.462371349334717,
      "learning_rate": 4.6836585238269606e-05,
      "loss": 0.6533,
      "step": 2520
    },
    {
      "epoch": 0.19799655658162466,
      "grad_norm": 7.59140157699585,
      "learning_rate": 4.682350787257415e-05,
      "loss": 0.9584,
      "step": 2530
    },
    {
      "epoch": 0.19877915166692753,
      "grad_norm": 1.8001086711883545,
      "learning_rate": 4.6810430506878695e-05,
      "loss": 0.6217,
      "step": 2540
    },
    {
      "epoch": 0.1995617467522304,
      "grad_norm": 2.8310940265655518,
      "learning_rate": 4.679735314118324e-05,
      "loss": 0.7707,
      "step": 2550
    },
    {
      "epoch": 0.20034434183753327,
      "grad_norm": 12.313705444335938,
      "learning_rate": 4.678427577548779e-05,
      "loss": 0.7026,
      "step": 2560
    },
    {
      "epoch": 0.20112693692283612,
      "grad_norm": 7.751932144165039,
      "learning_rate": 4.6771198409792336e-05,
      "loss": 1.1653,
      "step": 2570
    },
    {
      "epoch": 0.201909532008139,
      "grad_norm": 0.8290034532546997,
      "learning_rate": 4.675812104409688e-05,
      "loss": 1.2214,
      "step": 2580
    },
    {
      "epoch": 0.20269212709344187,
      "grad_norm": 9.922255516052246,
      "learning_rate": 4.6745043678401425e-05,
      "loss": 0.9118,
      "step": 2590
    },
    {
      "epoch": 0.2034747221787447,
      "grad_norm": 5.491465091705322,
      "learning_rate": 4.673196631270597e-05,
      "loss": 0.7696,
      "step": 2600
    },
    {
      "epoch": 0.20425731726404758,
      "grad_norm": 3.960170030593872,
      "learning_rate": 4.6718888947010514e-05,
      "loss": 1.0239,
      "step": 2610
    },
    {
      "epoch": 0.20503991234935046,
      "grad_norm": 6.319471836090088,
      "learning_rate": 4.6705811581315065e-05,
      "loss": 0.9603,
      "step": 2620
    },
    {
      "epoch": 0.2058225074346533,
      "grad_norm": 3.989011526107788,
      "learning_rate": 4.669273421561961e-05,
      "loss": 0.8044,
      "step": 2630
    },
    {
      "epoch": 0.20660510251995617,
      "grad_norm": 5.9908127784729,
      "learning_rate": 4.6679656849924155e-05,
      "loss": 0.8895,
      "step": 2640
    },
    {
      "epoch": 0.20738769760525905,
      "grad_norm": 17.62709617614746,
      "learning_rate": 4.66665794842287e-05,
      "loss": 0.8315,
      "step": 2650
    },
    {
      "epoch": 0.2081702926905619,
      "grad_norm": 3.1965339183807373,
      "learning_rate": 4.6653502118533244e-05,
      "loss": 0.8242,
      "step": 2660
    },
    {
      "epoch": 0.20895288777586477,
      "grad_norm": 4.125606060028076,
      "learning_rate": 4.664042475283779e-05,
      "loss": 0.9565,
      "step": 2670
    },
    {
      "epoch": 0.20973548286116764,
      "grad_norm": 1.8301092386245728,
      "learning_rate": 4.662734738714233e-05,
      "loss": 0.8072,
      "step": 2680
    },
    {
      "epoch": 0.21051807794647048,
      "grad_norm": 9.636066436767578,
      "learning_rate": 4.6614270021446884e-05,
      "loss": 0.6691,
      "step": 2690
    },
    {
      "epoch": 0.21130067303177336,
      "grad_norm": 6.180354595184326,
      "learning_rate": 4.660119265575143e-05,
      "loss": 1.2949,
      "step": 2700
    },
    {
      "epoch": 0.21208326811707623,
      "grad_norm": 5.820437908172607,
      "learning_rate": 4.658811529005597e-05,
      "loss": 1.1351,
      "step": 2710
    },
    {
      "epoch": 0.2128658632023791,
      "grad_norm": 3.0123584270477295,
      "learning_rate": 4.657503792436052e-05,
      "loss": 1.0554,
      "step": 2720
    },
    {
      "epoch": 0.21364845828768195,
      "grad_norm": 5.682946681976318,
      "learning_rate": 4.656196055866506e-05,
      "loss": 1.0787,
      "step": 2730
    },
    {
      "epoch": 0.21443105337298482,
      "grad_norm": 6.117527961730957,
      "learning_rate": 4.6548883192969614e-05,
      "loss": 0.9349,
      "step": 2740
    },
    {
      "epoch": 0.2152136484582877,
      "grad_norm": 3.2545790672302246,
      "learning_rate": 4.653580582727415e-05,
      "loss": 0.7788,
      "step": 2750
    },
    {
      "epoch": 0.21599624354359054,
      "grad_norm": 9.487701416015625,
      "learning_rate": 4.65227284615787e-05,
      "loss": 0.92,
      "step": 2760
    },
    {
      "epoch": 0.2167788386288934,
      "grad_norm": 4.6170172691345215,
      "learning_rate": 4.650965109588325e-05,
      "loss": 0.9769,
      "step": 2770
    },
    {
      "epoch": 0.21756143371419628,
      "grad_norm": 4.835626125335693,
      "learning_rate": 4.649657373018779e-05,
      "loss": 0.8333,
      "step": 2780
    },
    {
      "epoch": 0.21834402879949913,
      "grad_norm": 6.326450824737549,
      "learning_rate": 4.6483496364492343e-05,
      "loss": 0.8629,
      "step": 2790
    },
    {
      "epoch": 0.219126623884802,
      "grad_norm": 3.6024186611175537,
      "learning_rate": 4.647041899879688e-05,
      "loss": 0.7685,
      "step": 2800
    },
    {
      "epoch": 0.21990921897010487,
      "grad_norm": 5.848554611206055,
      "learning_rate": 4.645734163310143e-05,
      "loss": 0.8862,
      "step": 2810
    },
    {
      "epoch": 0.22069181405540772,
      "grad_norm": 8.191306114196777,
      "learning_rate": 4.644426426740597e-05,
      "loss": 0.9227,
      "step": 2820
    },
    {
      "epoch": 0.2214744091407106,
      "grad_norm": 8.240391731262207,
      "learning_rate": 4.643118690171052e-05,
      "loss": 0.7796,
      "step": 2830
    },
    {
      "epoch": 0.22225700422601347,
      "grad_norm": 20.118391036987305,
      "learning_rate": 4.6418109536015066e-05,
      "loss": 1.0307,
      "step": 2840
    },
    {
      "epoch": 0.2230395993113163,
      "grad_norm": 5.32141637802124,
      "learning_rate": 4.640503217031961e-05,
      "loss": 1.013,
      "step": 2850
    },
    {
      "epoch": 0.22382219439661918,
      "grad_norm": 5.300339221954346,
      "learning_rate": 4.639195480462416e-05,
      "loss": 0.8123,
      "step": 2860
    },
    {
      "epoch": 0.22460478948192206,
      "grad_norm": 6.741914749145508,
      "learning_rate": 4.63788774389287e-05,
      "loss": 1.0021,
      "step": 2870
    },
    {
      "epoch": 0.22538738456722493,
      "grad_norm": 6.9500508308410645,
      "learning_rate": 4.636580007323325e-05,
      "loss": 0.7968,
      "step": 2880
    },
    {
      "epoch": 0.22616997965252778,
      "grad_norm": 21.05146598815918,
      "learning_rate": 4.6352722707537796e-05,
      "loss": 1.1365,
      "step": 2890
    },
    {
      "epoch": 0.22695257473783065,
      "grad_norm": 7.512876987457275,
      "learning_rate": 4.633964534184234e-05,
      "loss": 0.9293,
      "step": 2900
    },
    {
      "epoch": 0.22773516982313352,
      "grad_norm": 5.362032890319824,
      "learning_rate": 4.632656797614689e-05,
      "loss": 1.1768,
      "step": 2910
    },
    {
      "epoch": 0.22851776490843637,
      "grad_norm": 7.874504566192627,
      "learning_rate": 4.631349061045143e-05,
      "loss": 1.1222,
      "step": 2920
    },
    {
      "epoch": 0.22930035999373924,
      "grad_norm": 5.077906131744385,
      "learning_rate": 4.630041324475598e-05,
      "loss": 0.7554,
      "step": 2930
    },
    {
      "epoch": 0.2300829550790421,
      "grad_norm": 13.081748008728027,
      "learning_rate": 4.6287335879060526e-05,
      "loss": 0.7857,
      "step": 2940
    },
    {
      "epoch": 0.23086555016434496,
      "grad_norm": 4.204710483551025,
      "learning_rate": 4.627425851336507e-05,
      "loss": 0.6259,
      "step": 2950
    },
    {
      "epoch": 0.23164814524964783,
      "grad_norm": 4.149248123168945,
      "learning_rate": 4.6261181147669615e-05,
      "loss": 0.8102,
      "step": 2960
    },
    {
      "epoch": 0.2324307403349507,
      "grad_norm": 2.9448940753936768,
      "learning_rate": 4.624810378197416e-05,
      "loss": 0.3323,
      "step": 2970
    },
    {
      "epoch": 0.23321333542025355,
      "grad_norm": 3.0060842037200928,
      "learning_rate": 4.623502641627871e-05,
      "loss": 0.7558,
      "step": 2980
    },
    {
      "epoch": 0.23399593050555642,
      "grad_norm": 5.069430828094482,
      "learning_rate": 4.6221949050583255e-05,
      "loss": 1.7604,
      "step": 2990
    },
    {
      "epoch": 0.2347785255908593,
      "grad_norm": 12.806517601013184,
      "learning_rate": 4.62088716848878e-05,
      "loss": 0.6135,
      "step": 3000
    },
    {
      "epoch": 0.23556112067616217,
      "grad_norm": 1.2452585697174072,
      "learning_rate": 4.6195794319192344e-05,
      "loss": 0.6796,
      "step": 3010
    },
    {
      "epoch": 0.236343715761465,
      "grad_norm": 10.799848556518555,
      "learning_rate": 4.618271695349689e-05,
      "loss": 1.0904,
      "step": 3020
    },
    {
      "epoch": 0.23712631084676788,
      "grad_norm": 16.27574348449707,
      "learning_rate": 4.6169639587801434e-05,
      "loss": 0.7476,
      "step": 3030
    },
    {
      "epoch": 0.23790890593207076,
      "grad_norm": 0.7114291191101074,
      "learning_rate": 4.615656222210598e-05,
      "loss": 0.7251,
      "step": 3040
    },
    {
      "epoch": 0.2386915010173736,
      "grad_norm": 4.574992656707764,
      "learning_rate": 4.614348485641053e-05,
      "loss": 0.9076,
      "step": 3050
    },
    {
      "epoch": 0.23947409610267648,
      "grad_norm": 4.558896541595459,
      "learning_rate": 4.6130407490715074e-05,
      "loss": 1.0599,
      "step": 3060
    },
    {
      "epoch": 0.24025669118797935,
      "grad_norm": 2.2301275730133057,
      "learning_rate": 4.611733012501962e-05,
      "loss": 1.2519,
      "step": 3070
    },
    {
      "epoch": 0.2410392862732822,
      "grad_norm": 11.03345775604248,
      "learning_rate": 4.610425275932416e-05,
      "loss": 1.215,
      "step": 3080
    },
    {
      "epoch": 0.24182188135858507,
      "grad_norm": 4.903374195098877,
      "learning_rate": 4.609117539362871e-05,
      "loss": 0.7111,
      "step": 3090
    },
    {
      "epoch": 0.24260447644388794,
      "grad_norm": 8.505735397338867,
      "learning_rate": 4.607809802793325e-05,
      "loss": 0.5905,
      "step": 3100
    },
    {
      "epoch": 0.24338707152919078,
      "grad_norm": 9.631453514099121,
      "learning_rate": 4.6065020662237804e-05,
      "loss": 1.3352,
      "step": 3110
    },
    {
      "epoch": 0.24416966661449366,
      "grad_norm": 15.109895706176758,
      "learning_rate": 4.605194329654235e-05,
      "loss": 1.072,
      "step": 3120
    },
    {
      "epoch": 0.24495226169979653,
      "grad_norm": 3.5323593616485596,
      "learning_rate": 4.603886593084689e-05,
      "loss": 0.8187,
      "step": 3130
    },
    {
      "epoch": 0.24573485678509938,
      "grad_norm": 3.473493814468384,
      "learning_rate": 4.602578856515144e-05,
      "loss": 0.7049,
      "step": 3140
    },
    {
      "epoch": 0.24651745187040225,
      "grad_norm": 13.174273490905762,
      "learning_rate": 4.601271119945598e-05,
      "loss": 1.1379,
      "step": 3150
    },
    {
      "epoch": 0.24730004695570512,
      "grad_norm": 7.59885835647583,
      "learning_rate": 4.599963383376053e-05,
      "loss": 0.8686,
      "step": 3160
    },
    {
      "epoch": 0.248082642041008,
      "grad_norm": 6.904193878173828,
      "learning_rate": 4.598655646806507e-05,
      "loss": 0.6044,
      "step": 3170
    },
    {
      "epoch": 0.24886523712631084,
      "grad_norm": 5.149388790130615,
      "learning_rate": 4.597347910236962e-05,
      "loss": 1.001,
      "step": 3180
    },
    {
      "epoch": 0.2496478322116137,
      "grad_norm": 5.271276473999023,
      "learning_rate": 4.596040173667417e-05,
      "loss": 1.0277,
      "step": 3190
    },
    {
      "epoch": 0.25043042729691656,
      "grad_norm": 5.769474506378174,
      "learning_rate": 4.594732437097871e-05,
      "loss": 0.8189,
      "step": 3200
    },
    {
      "epoch": 0.25121302238221943,
      "grad_norm": 5.48963737487793,
      "learning_rate": 4.593424700528326e-05,
      "loss": 0.8752,
      "step": 3210
    },
    {
      "epoch": 0.2519956174675223,
      "grad_norm": 4.530579566955566,
      "learning_rate": 4.59211696395878e-05,
      "loss": 0.9609,
      "step": 3220
    },
    {
      "epoch": 0.2527782125528252,
      "grad_norm": 6.283682346343994,
      "learning_rate": 4.590809227389235e-05,
      "loss": 0.8396,
      "step": 3230
    },
    {
      "epoch": 0.25356080763812805,
      "grad_norm": 9.943126678466797,
      "learning_rate": 4.589501490819689e-05,
      "loss": 0.5559,
      "step": 3240
    },
    {
      "epoch": 0.2543434027234309,
      "grad_norm": 16.064008712768555,
      "learning_rate": 4.588193754250144e-05,
      "loss": 1.2824,
      "step": 3250
    },
    {
      "epoch": 0.25512599780873374,
      "grad_norm": 17.29366683959961,
      "learning_rate": 4.5868860176805986e-05,
      "loss": 1.1087,
      "step": 3260
    },
    {
      "epoch": 0.2559085928940366,
      "grad_norm": 4.302084922790527,
      "learning_rate": 4.585578281111053e-05,
      "loss": 0.7062,
      "step": 3270
    },
    {
      "epoch": 0.2566911879793395,
      "grad_norm": 6.408489227294922,
      "learning_rate": 4.584270544541508e-05,
      "loss": 1.0779,
      "step": 3280
    },
    {
      "epoch": 0.25747378306464236,
      "grad_norm": 2.888685941696167,
      "learning_rate": 4.582962807971962e-05,
      "loss": 1.0337,
      "step": 3290
    },
    {
      "epoch": 0.25825637814994523,
      "grad_norm": 1.909744143486023,
      "learning_rate": 4.581655071402417e-05,
      "loss": 0.8997,
      "step": 3300
    },
    {
      "epoch": 0.2590389732352481,
      "grad_norm": 7.126741409301758,
      "learning_rate": 4.5803473348328715e-05,
      "loss": 1.0384,
      "step": 3310
    },
    {
      "epoch": 0.2598215683205509,
      "grad_norm": 10.55949592590332,
      "learning_rate": 4.579039598263326e-05,
      "loss": 0.7422,
      "step": 3320
    },
    {
      "epoch": 0.2606041634058538,
      "grad_norm": 1.480275273323059,
      "learning_rate": 4.577731861693781e-05,
      "loss": 0.7087,
      "step": 3330
    },
    {
      "epoch": 0.26138675849115667,
      "grad_norm": 2.9011356830596924,
      "learning_rate": 4.576424125124235e-05,
      "loss": 1.0107,
      "step": 3340
    },
    {
      "epoch": 0.26216935357645954,
      "grad_norm": 7.247684478759766,
      "learning_rate": 4.57511638855469e-05,
      "loss": 0.573,
      "step": 3350
    },
    {
      "epoch": 0.2629519486617624,
      "grad_norm": 5.109511852264404,
      "learning_rate": 4.5738086519851445e-05,
      "loss": 0.5095,
      "step": 3360
    },
    {
      "epoch": 0.2637345437470653,
      "grad_norm": 10.802379608154297,
      "learning_rate": 4.572500915415599e-05,
      "loss": 0.7536,
      "step": 3370
    },
    {
      "epoch": 0.26451713883236816,
      "grad_norm": 19.47506332397461,
      "learning_rate": 4.5711931788460534e-05,
      "loss": 1.3027,
      "step": 3380
    },
    {
      "epoch": 0.265299733917671,
      "grad_norm": 1.0734502077102661,
      "learning_rate": 4.569885442276508e-05,
      "loss": 0.9189,
      "step": 3390
    },
    {
      "epoch": 0.26608232900297385,
      "grad_norm": 1.288217544555664,
      "learning_rate": 4.568577705706963e-05,
      "loss": 0.602,
      "step": 3400
    },
    {
      "epoch": 0.2668649240882767,
      "grad_norm": 1.8219399452209473,
      "learning_rate": 4.567269969137417e-05,
      "loss": 1.1577,
      "step": 3410
    },
    {
      "epoch": 0.2676475191735796,
      "grad_norm": 7.18124532699585,
      "learning_rate": 4.565962232567872e-05,
      "loss": 1.2643,
      "step": 3420
    },
    {
      "epoch": 0.26843011425888247,
      "grad_norm": 18.866384506225586,
      "learning_rate": 4.5646544959983264e-05,
      "loss": 0.8101,
      "step": 3430
    },
    {
      "epoch": 0.26921270934418534,
      "grad_norm": 5.988043785095215,
      "learning_rate": 4.563346759428781e-05,
      "loss": 1.0372,
      "step": 3440
    },
    {
      "epoch": 0.26999530442948816,
      "grad_norm": 13.536476135253906,
      "learning_rate": 4.562039022859235e-05,
      "loss": 1.0546,
      "step": 3450
    },
    {
      "epoch": 0.27077789951479103,
      "grad_norm": 1.69777512550354,
      "learning_rate": 4.56073128628969e-05,
      "loss": 0.949,
      "step": 3460
    },
    {
      "epoch": 0.2715604946000939,
      "grad_norm": 9.163593292236328,
      "learning_rate": 4.559423549720145e-05,
      "loss": 1.0426,
      "step": 3470
    },
    {
      "epoch": 0.2723430896853968,
      "grad_norm": 4.334875583648682,
      "learning_rate": 4.5581158131505994e-05,
      "loss": 1.1235,
      "step": 3480
    },
    {
      "epoch": 0.27312568477069965,
      "grad_norm": 6.928515911102295,
      "learning_rate": 4.556808076581054e-05,
      "loss": 1.0622,
      "step": 3490
    },
    {
      "epoch": 0.2739082798560025,
      "grad_norm": 4.518532752990723,
      "learning_rate": 4.555500340011508e-05,
      "loss": 0.8035,
      "step": 3500
    },
    {
      "epoch": 0.2746908749413054,
      "grad_norm": 4.834164142608643,
      "learning_rate": 4.554192603441963e-05,
      "loss": 1.0125,
      "step": 3510
    },
    {
      "epoch": 0.2754734700266082,
      "grad_norm": 5.751556396484375,
      "learning_rate": 4.552884866872417e-05,
      "loss": 0.668,
      "step": 3520
    },
    {
      "epoch": 0.2762560651119111,
      "grad_norm": 1.6642838716506958,
      "learning_rate": 4.551577130302872e-05,
      "loss": 0.8938,
      "step": 3530
    },
    {
      "epoch": 0.27703866019721396,
      "grad_norm": 26.209014892578125,
      "learning_rate": 4.550269393733327e-05,
      "loss": 0.8267,
      "step": 3540
    },
    {
      "epoch": 0.27782125528251683,
      "grad_norm": 10.160623550415039,
      "learning_rate": 4.548961657163781e-05,
      "loss": 1.1346,
      "step": 3550
    },
    {
      "epoch": 0.2786038503678197,
      "grad_norm": 5.363959312438965,
      "learning_rate": 4.547653920594236e-05,
      "loss": 1.0051,
      "step": 3560
    },
    {
      "epoch": 0.2793864454531226,
      "grad_norm": 3.2373569011688232,
      "learning_rate": 4.54634618402469e-05,
      "loss": 1.0206,
      "step": 3570
    },
    {
      "epoch": 0.2801690405384254,
      "grad_norm": 9.569706916809082,
      "learning_rate": 4.545038447455145e-05,
      "loss": 0.7854,
      "step": 3580
    },
    {
      "epoch": 0.28095163562372827,
      "grad_norm": 7.251800537109375,
      "learning_rate": 4.543730710885599e-05,
      "loss": 0.8779,
      "step": 3590
    },
    {
      "epoch": 0.28173423070903114,
      "grad_norm": 5.713415622711182,
      "learning_rate": 4.542422974316054e-05,
      "loss": 0.8045,
      "step": 3600
    },
    {
      "epoch": 0.282516825794334,
      "grad_norm": 2.4142863750457764,
      "learning_rate": 4.5411152377465087e-05,
      "loss": 0.914,
      "step": 3610
    },
    {
      "epoch": 0.2832994208796369,
      "grad_norm": 7.011434555053711,
      "learning_rate": 4.539807501176963e-05,
      "loss": 0.8237,
      "step": 3620
    },
    {
      "epoch": 0.28408201596493976,
      "grad_norm": 13.55313777923584,
      "learning_rate": 4.5384997646074176e-05,
      "loss": 1.0518,
      "step": 3630
    },
    {
      "epoch": 0.2848646110502426,
      "grad_norm": 7.908879280090332,
      "learning_rate": 4.537192028037872e-05,
      "loss": 0.9868,
      "step": 3640
    },
    {
      "epoch": 0.28564720613554545,
      "grad_norm": 10.656712532043457,
      "learning_rate": 4.535884291468327e-05,
      "loss": 0.8023,
      "step": 3650
    },
    {
      "epoch": 0.2864298012208483,
      "grad_norm": 6.37989616394043,
      "learning_rate": 4.534576554898781e-05,
      "loss": 0.7005,
      "step": 3660
    },
    {
      "epoch": 0.2872123963061512,
      "grad_norm": 6.416524887084961,
      "learning_rate": 4.533268818329236e-05,
      "loss": 0.7668,
      "step": 3670
    },
    {
      "epoch": 0.28799499139145407,
      "grad_norm": 3.598287343978882,
      "learning_rate": 4.5319610817596905e-05,
      "loss": 1.0337,
      "step": 3680
    },
    {
      "epoch": 0.28877758647675694,
      "grad_norm": 5.156396865844727,
      "learning_rate": 4.530653345190145e-05,
      "loss": 0.9394,
      "step": 3690
    },
    {
      "epoch": 0.2895601815620598,
      "grad_norm": 11.683138847351074,
      "learning_rate": 4.5293456086206e-05,
      "loss": 0.9813,
      "step": 3700
    },
    {
      "epoch": 0.29034277664736263,
      "grad_norm": 18.248300552368164,
      "learning_rate": 4.528037872051054e-05,
      "loss": 0.7193,
      "step": 3710
    },
    {
      "epoch": 0.2911253717326655,
      "grad_norm": 15.677491188049316,
      "learning_rate": 4.526730135481509e-05,
      "loss": 1.2949,
      "step": 3720
    },
    {
      "epoch": 0.2919079668179684,
      "grad_norm": 8.293972969055176,
      "learning_rate": 4.5254223989119635e-05,
      "loss": 0.9781,
      "step": 3730
    },
    {
      "epoch": 0.29269056190327125,
      "grad_norm": 9.9196138381958,
      "learning_rate": 4.524114662342418e-05,
      "loss": 0.781,
      "step": 3740
    },
    {
      "epoch": 0.2934731569885741,
      "grad_norm": 4.724277496337891,
      "learning_rate": 4.522806925772873e-05,
      "loss": 0.7992,
      "step": 3750
    },
    {
      "epoch": 0.294255752073877,
      "grad_norm": 2.4277870655059814,
      "learning_rate": 4.521499189203327e-05,
      "loss": 0.6236,
      "step": 3760
    },
    {
      "epoch": 0.2950383471591798,
      "grad_norm": 2.2673211097717285,
      "learning_rate": 4.520191452633782e-05,
      "loss": 1.1281,
      "step": 3770
    },
    {
      "epoch": 0.2958209422444827,
      "grad_norm": 3.4653267860412598,
      "learning_rate": 4.518883716064236e-05,
      "loss": 0.7146,
      "step": 3780
    },
    {
      "epoch": 0.29660353732978556,
      "grad_norm": 8.272395133972168,
      "learning_rate": 4.517575979494691e-05,
      "loss": 1.1187,
      "step": 3790
    },
    {
      "epoch": 0.29738613241508843,
      "grad_norm": 9.940991401672363,
      "learning_rate": 4.5162682429251454e-05,
      "loss": 0.5503,
      "step": 3800
    },
    {
      "epoch": 0.2981687275003913,
      "grad_norm": 4.770025730133057,
      "learning_rate": 4.5149605063556e-05,
      "loss": 0.7396,
      "step": 3810
    },
    {
      "epoch": 0.2989513225856942,
      "grad_norm": 9.120471000671387,
      "learning_rate": 4.513652769786055e-05,
      "loss": 0.8367,
      "step": 3820
    },
    {
      "epoch": 0.29973391767099705,
      "grad_norm": 5.21095609664917,
      "learning_rate": 4.512345033216509e-05,
      "loss": 0.7605,
      "step": 3830
    },
    {
      "epoch": 0.30051651275629987,
      "grad_norm": 4.611201763153076,
      "learning_rate": 4.511037296646964e-05,
      "loss": 0.6173,
      "step": 3840
    },
    {
      "epoch": 0.30129910784160274,
      "grad_norm": 12.684550285339355,
      "learning_rate": 4.509729560077418e-05,
      "loss": 1.201,
      "step": 3850
    },
    {
      "epoch": 0.3020817029269056,
      "grad_norm": 4.040058135986328,
      "learning_rate": 4.508421823507873e-05,
      "loss": 0.9748,
      "step": 3860
    },
    {
      "epoch": 0.3028642980122085,
      "grad_norm": 11.91784381866455,
      "learning_rate": 4.507114086938327e-05,
      "loss": 1.0553,
      "step": 3870
    },
    {
      "epoch": 0.30364689309751136,
      "grad_norm": 9.336968421936035,
      "learning_rate": 4.505806350368782e-05,
      "loss": 1.3157,
      "step": 3880
    },
    {
      "epoch": 0.30442948818281423,
      "grad_norm": 7.9654412269592285,
      "learning_rate": 4.504498613799237e-05,
      "loss": 0.7663,
      "step": 3890
    },
    {
      "epoch": 0.30521208326811705,
      "grad_norm": 8.888216972351074,
      "learning_rate": 4.503190877229691e-05,
      "loss": 0.7301,
      "step": 3900
    },
    {
      "epoch": 0.3059946783534199,
      "grad_norm": 23.653072357177734,
      "learning_rate": 4.501883140660146e-05,
      "loss": 0.8122,
      "step": 3910
    },
    {
      "epoch": 0.3067772734387228,
      "grad_norm": 4.7662553787231445,
      "learning_rate": 4.5005754040906e-05,
      "loss": 0.9137,
      "step": 3920
    },
    {
      "epoch": 0.30755986852402567,
      "grad_norm": 2.8691177368164062,
      "learning_rate": 4.499267667521055e-05,
      "loss": 0.7599,
      "step": 3930
    },
    {
      "epoch": 0.30834246360932854,
      "grad_norm": 24.311439514160156,
      "learning_rate": 4.497959930951509e-05,
      "loss": 0.8491,
      "step": 3940
    },
    {
      "epoch": 0.3091250586946314,
      "grad_norm": 5.708023548126221,
      "learning_rate": 4.496652194381964e-05,
      "loss": 0.9815,
      "step": 3950
    },
    {
      "epoch": 0.3099076537799343,
      "grad_norm": 4.055660724639893,
      "learning_rate": 4.495344457812419e-05,
      "loss": 0.9646,
      "step": 3960
    },
    {
      "epoch": 0.3106902488652371,
      "grad_norm": 12.111351013183594,
      "learning_rate": 4.494036721242873e-05,
      "loss": 0.683,
      "step": 3970
    },
    {
      "epoch": 0.31147284395054,
      "grad_norm": 6.234447002410889,
      "learning_rate": 4.4927289846733276e-05,
      "loss": 0.8839,
      "step": 3980
    },
    {
      "epoch": 0.31225543903584285,
      "grad_norm": 6.414594650268555,
      "learning_rate": 4.491421248103782e-05,
      "loss": 0.8336,
      "step": 3990
    },
    {
      "epoch": 0.3130380341211457,
      "grad_norm": 6.55598258972168,
      "learning_rate": 4.4901135115342365e-05,
      "loss": 1.1688,
      "step": 4000
    },
    {
      "epoch": 0.3138206292064486,
      "grad_norm": 5.831319808959961,
      "learning_rate": 4.488805774964691e-05,
      "loss": 1.0471,
      "step": 4010
    },
    {
      "epoch": 0.31460322429175147,
      "grad_norm": 9.01219367980957,
      "learning_rate": 4.487498038395146e-05,
      "loss": 1.0237,
      "step": 4020
    },
    {
      "epoch": 0.3153858193770543,
      "grad_norm": 6.871992588043213,
      "learning_rate": 4.4861903018256006e-05,
      "loss": 0.6685,
      "step": 4030
    },
    {
      "epoch": 0.31616841446235716,
      "grad_norm": 3.960470676422119,
      "learning_rate": 4.484882565256055e-05,
      "loss": 0.6945,
      "step": 4040
    },
    {
      "epoch": 0.31695100954766003,
      "grad_norm": 3.9110944271087646,
      "learning_rate": 4.4835748286865095e-05,
      "loss": 0.8744,
      "step": 4050
    },
    {
      "epoch": 0.3177336046329629,
      "grad_norm": 4.5652313232421875,
      "learning_rate": 4.482267092116964e-05,
      "loss": 0.6673,
      "step": 4060
    },
    {
      "epoch": 0.3185161997182658,
      "grad_norm": 17.10112953186035,
      "learning_rate": 4.480959355547419e-05,
      "loss": 0.8755,
      "step": 4070
    },
    {
      "epoch": 0.31929879480356865,
      "grad_norm": 8.332122802734375,
      "learning_rate": 4.479651618977873e-05,
      "loss": 0.9151,
      "step": 4080
    },
    {
      "epoch": 0.32008138988887147,
      "grad_norm": 9.392374038696289,
      "learning_rate": 4.478343882408328e-05,
      "loss": 1.211,
      "step": 4090
    },
    {
      "epoch": 0.32086398497417434,
      "grad_norm": 1.6295779943466187,
      "learning_rate": 4.4770361458387825e-05,
      "loss": 0.7518,
      "step": 4100
    },
    {
      "epoch": 0.3216465800594772,
      "grad_norm": 26.37354278564453,
      "learning_rate": 4.475728409269237e-05,
      "loss": 0.925,
      "step": 4110
    },
    {
      "epoch": 0.3224291751447801,
      "grad_norm": 27.758852005004883,
      "learning_rate": 4.474420672699692e-05,
      "loss": 1.0307,
      "step": 4120
    },
    {
      "epoch": 0.32321177023008296,
      "grad_norm": 1.464961290359497,
      "learning_rate": 4.473112936130146e-05,
      "loss": 0.8733,
      "step": 4130
    },
    {
      "epoch": 0.32399436531538583,
      "grad_norm": 10.887391090393066,
      "learning_rate": 4.471805199560601e-05,
      "loss": 0.7483,
      "step": 4140
    },
    {
      "epoch": 0.3247769604006887,
      "grad_norm": 7.526797771453857,
      "learning_rate": 4.470497462991055e-05,
      "loss": 0.7396,
      "step": 4150
    },
    {
      "epoch": 0.3255595554859915,
      "grad_norm": 9.462432861328125,
      "learning_rate": 4.46918972642151e-05,
      "loss": 0.6619,
      "step": 4160
    },
    {
      "epoch": 0.3263421505712944,
      "grad_norm": 7.175870895385742,
      "learning_rate": 4.467881989851965e-05,
      "loss": 1.0097,
      "step": 4170
    },
    {
      "epoch": 0.32712474565659727,
      "grad_norm": 1.8711260557174683,
      "learning_rate": 4.466574253282419e-05,
      "loss": 0.61,
      "step": 4180
    },
    {
      "epoch": 0.32790734074190014,
      "grad_norm": 9.878432273864746,
      "learning_rate": 4.465266516712874e-05,
      "loss": 0.8016,
      "step": 4190
    },
    {
      "epoch": 0.328689935827203,
      "grad_norm": 5.068032264709473,
      "learning_rate": 4.463958780143328e-05,
      "loss": 1.029,
      "step": 4200
    },
    {
      "epoch": 0.3294725309125059,
      "grad_norm": 21.201108932495117,
      "learning_rate": 4.462651043573783e-05,
      "loss": 1.0855,
      "step": 4210
    },
    {
      "epoch": 0.3302551259978087,
      "grad_norm": 6.089317321777344,
      "learning_rate": 4.461343307004237e-05,
      "loss": 0.8731,
      "step": 4220
    },
    {
      "epoch": 0.3310377210831116,
      "grad_norm": 2.123523473739624,
      "learning_rate": 4.460035570434692e-05,
      "loss": 0.8776,
      "step": 4230
    },
    {
      "epoch": 0.33182031616841445,
      "grad_norm": 4.153194427490234,
      "learning_rate": 4.458727833865146e-05,
      "loss": 0.8841,
      "step": 4240
    },
    {
      "epoch": 0.3326029112537173,
      "grad_norm": 1.6861412525177002,
      "learning_rate": 4.457420097295601e-05,
      "loss": 0.7059,
      "step": 4250
    },
    {
      "epoch": 0.3333855063390202,
      "grad_norm": 2.498950719833374,
      "learning_rate": 4.456112360726056e-05,
      "loss": 0.6836,
      "step": 4260
    },
    {
      "epoch": 0.33416810142432307,
      "grad_norm": 4.471656322479248,
      "learning_rate": 4.45480462415651e-05,
      "loss": 1.0306,
      "step": 4270
    },
    {
      "epoch": 0.33495069650962594,
      "grad_norm": 9.220391273498535,
      "learning_rate": 4.453496887586965e-05,
      "loss": 0.8343,
      "step": 4280
    },
    {
      "epoch": 0.33573329159492876,
      "grad_norm": 3.7967326641082764,
      "learning_rate": 4.452189151017419e-05,
      "loss": 0.7613,
      "step": 4290
    },
    {
      "epoch": 0.33651588668023164,
      "grad_norm": 7.448493003845215,
      "learning_rate": 4.4508814144478737e-05,
      "loss": 0.8609,
      "step": 4300
    },
    {
      "epoch": 0.3372984817655345,
      "grad_norm": 1.721092939376831,
      "learning_rate": 4.449573677878328e-05,
      "loss": 0.8494,
      "step": 4310
    },
    {
      "epoch": 0.3380810768508374,
      "grad_norm": 5.445475101470947,
      "learning_rate": 4.448265941308783e-05,
      "loss": 1.1637,
      "step": 4320
    },
    {
      "epoch": 0.33886367193614025,
      "grad_norm": 15.984429359436035,
      "learning_rate": 4.446958204739238e-05,
      "loss": 0.983,
      "step": 4330
    },
    {
      "epoch": 0.3396462670214431,
      "grad_norm": 5.98968505859375,
      "learning_rate": 4.445650468169692e-05,
      "loss": 0.7121,
      "step": 4340
    },
    {
      "epoch": 0.34042886210674594,
      "grad_norm": 1.9178190231323242,
      "learning_rate": 4.4443427316001466e-05,
      "loss": 0.8464,
      "step": 4350
    },
    {
      "epoch": 0.3412114571920488,
      "grad_norm": 2.4100534915924072,
      "learning_rate": 4.443034995030601e-05,
      "loss": 0.8048,
      "step": 4360
    },
    {
      "epoch": 0.3419940522773517,
      "grad_norm": 13.346720695495605,
      "learning_rate": 4.4417272584610555e-05,
      "loss": 1.0026,
      "step": 4370
    },
    {
      "epoch": 0.34277664736265456,
      "grad_norm": 6.768532752990723,
      "learning_rate": 4.44041952189151e-05,
      "loss": 0.841,
      "step": 4380
    },
    {
      "epoch": 0.34355924244795744,
      "grad_norm": 13.153164863586426,
      "learning_rate": 4.439111785321965e-05,
      "loss": 1.0783,
      "step": 4390
    },
    {
      "epoch": 0.3443418375332603,
      "grad_norm": 5.562099933624268,
      "learning_rate": 4.4378040487524196e-05,
      "loss": 0.9963,
      "step": 4400
    },
    {
      "epoch": 0.3451244326185632,
      "grad_norm": 5.71734619140625,
      "learning_rate": 4.436496312182874e-05,
      "loss": 0.8858,
      "step": 4410
    },
    {
      "epoch": 0.345907027703866,
      "grad_norm": 8.115924835205078,
      "learning_rate": 4.4351885756133285e-05,
      "loss": 0.7101,
      "step": 4420
    },
    {
      "epoch": 0.34668962278916887,
      "grad_norm": 4.748242378234863,
      "learning_rate": 4.433880839043783e-05,
      "loss": 0.767,
      "step": 4430
    },
    {
      "epoch": 0.34747221787447174,
      "grad_norm": 2.413039207458496,
      "learning_rate": 4.432573102474238e-05,
      "loss": 0.8093,
      "step": 4440
    },
    {
      "epoch": 0.3482548129597746,
      "grad_norm": 8.299149513244629,
      "learning_rate": 4.431265365904692e-05,
      "loss": 0.9007,
      "step": 4450
    },
    {
      "epoch": 0.3490374080450775,
      "grad_norm": 3.99772047996521,
      "learning_rate": 4.429957629335147e-05,
      "loss": 0.7088,
      "step": 4460
    },
    {
      "epoch": 0.34982000313038036,
      "grad_norm": 8.929529190063477,
      "learning_rate": 4.4286498927656015e-05,
      "loss": 1.0174,
      "step": 4470
    },
    {
      "epoch": 0.3506025982156832,
      "grad_norm": 8.684158325195312,
      "learning_rate": 4.427342156196056e-05,
      "loss": 0.9022,
      "step": 4480
    },
    {
      "epoch": 0.35138519330098605,
      "grad_norm": 14.087525367736816,
      "learning_rate": 4.426034419626511e-05,
      "loss": 0.772,
      "step": 4490
    },
    {
      "epoch": 0.3521677883862889,
      "grad_norm": 11.703319549560547,
      "learning_rate": 4.424726683056965e-05,
      "loss": 0.9162,
      "step": 4500
    },
    {
      "epoch": 0.3529503834715918,
      "grad_norm": 3.7111408710479736,
      "learning_rate": 4.42341894648742e-05,
      "loss": 0.6951,
      "step": 4510
    },
    {
      "epoch": 0.3537329785568947,
      "grad_norm": 3.8791961669921875,
      "learning_rate": 4.4221112099178744e-05,
      "loss": 0.752,
      "step": 4520
    },
    {
      "epoch": 0.35451557364219755,
      "grad_norm": 11.605985641479492,
      "learning_rate": 4.420803473348329e-05,
      "loss": 0.8943,
      "step": 4530
    },
    {
      "epoch": 0.3552981687275004,
      "grad_norm": 7.224593639373779,
      "learning_rate": 4.419495736778784e-05,
      "loss": 1.031,
      "step": 4540
    },
    {
      "epoch": 0.35608076381280324,
      "grad_norm": 12.535014152526855,
      "learning_rate": 4.418188000209238e-05,
      "loss": 0.6609,
      "step": 4550
    },
    {
      "epoch": 0.3568633588981061,
      "grad_norm": 8.325557708740234,
      "learning_rate": 4.416880263639693e-05,
      "loss": 0.8063,
      "step": 4560
    },
    {
      "epoch": 0.357645953983409,
      "grad_norm": 8.68255615234375,
      "learning_rate": 4.415572527070147e-05,
      "loss": 0.9023,
      "step": 4570
    },
    {
      "epoch": 0.35842854906871185,
      "grad_norm": 4.318446159362793,
      "learning_rate": 4.414264790500602e-05,
      "loss": 0.8609,
      "step": 4580
    },
    {
      "epoch": 0.3592111441540147,
      "grad_norm": 12.712624549865723,
      "learning_rate": 4.412957053931056e-05,
      "loss": 1.2192,
      "step": 4590
    },
    {
      "epoch": 0.3599937392393176,
      "grad_norm": 11.502581596374512,
      "learning_rate": 4.411649317361511e-05,
      "loss": 0.6417,
      "step": 4600
    },
    {
      "epoch": 0.3607763343246204,
      "grad_norm": 8.325228691101074,
      "learning_rate": 4.410341580791966e-05,
      "loss": 1.1118,
      "step": 4610
    },
    {
      "epoch": 0.3615589294099233,
      "grad_norm": 1.5091371536254883,
      "learning_rate": 4.40903384422242e-05,
      "loss": 0.8623,
      "step": 4620
    },
    {
      "epoch": 0.36234152449522616,
      "grad_norm": 7.073453426361084,
      "learning_rate": 4.407726107652875e-05,
      "loss": 0.8437,
      "step": 4630
    },
    {
      "epoch": 0.36312411958052904,
      "grad_norm": 4.275660514831543,
      "learning_rate": 4.406418371083329e-05,
      "loss": 0.6839,
      "step": 4640
    },
    {
      "epoch": 0.3639067146658319,
      "grad_norm": 1.7065638303756714,
      "learning_rate": 4.405110634513784e-05,
      "loss": 0.6172,
      "step": 4650
    },
    {
      "epoch": 0.3646893097511348,
      "grad_norm": 7.723235607147217,
      "learning_rate": 4.403802897944238e-05,
      "loss": 0.7889,
      "step": 4660
    },
    {
      "epoch": 0.3654719048364376,
      "grad_norm": 8.03349494934082,
      "learning_rate": 4.4024951613746926e-05,
      "loss": 1.3809,
      "step": 4670
    },
    {
      "epoch": 0.3662544999217405,
      "grad_norm": 1.3818224668502808,
      "learning_rate": 4.401187424805148e-05,
      "loss": 1.1016,
      "step": 4680
    },
    {
      "epoch": 0.36703709500704335,
      "grad_norm": 5.216217517852783,
      "learning_rate": 4.399879688235602e-05,
      "loss": 1.0748,
      "step": 4690
    },
    {
      "epoch": 0.3678196900923462,
      "grad_norm": 4.537940502166748,
      "learning_rate": 4.398571951666057e-05,
      "loss": 0.6561,
      "step": 4700
    },
    {
      "epoch": 0.3686022851776491,
      "grad_norm": 4.30039119720459,
      "learning_rate": 4.397264215096511e-05,
      "loss": 0.8285,
      "step": 4710
    },
    {
      "epoch": 0.36938488026295196,
      "grad_norm": 5.825339317321777,
      "learning_rate": 4.3959564785269656e-05,
      "loss": 0.5904,
      "step": 4720
    },
    {
      "epoch": 0.37016747534825484,
      "grad_norm": 12.709004402160645,
      "learning_rate": 4.39464874195742e-05,
      "loss": 0.7155,
      "step": 4730
    },
    {
      "epoch": 0.37095007043355765,
      "grad_norm": 5.101252555847168,
      "learning_rate": 4.3933410053878745e-05,
      "loss": 0.773,
      "step": 4740
    },
    {
      "epoch": 0.37173266551886053,
      "grad_norm": 12.49058723449707,
      "learning_rate": 4.3920332688183297e-05,
      "loss": 0.6535,
      "step": 4750
    },
    {
      "epoch": 0.3725152606041634,
      "grad_norm": 4.1231770515441895,
      "learning_rate": 4.390725532248784e-05,
      "loss": 1.3727,
      "step": 4760
    },
    {
      "epoch": 0.3732978556894663,
      "grad_norm": 8.208578109741211,
      "learning_rate": 4.3894177956792386e-05,
      "loss": 1.0827,
      "step": 4770
    },
    {
      "epoch": 0.37408045077476915,
      "grad_norm": 4.281179904937744,
      "learning_rate": 4.388110059109693e-05,
      "loss": 0.786,
      "step": 4780
    },
    {
      "epoch": 0.374863045860072,
      "grad_norm": 1.853075623512268,
      "learning_rate": 4.3868023225401475e-05,
      "loss": 0.777,
      "step": 4790
    },
    {
      "epoch": 0.37564564094537484,
      "grad_norm": 0.9749950170516968,
      "learning_rate": 4.385494585970602e-05,
      "loss": 1.011,
      "step": 4800
    },
    {
      "epoch": 0.3764282360306777,
      "grad_norm": 1.3098162412643433,
      "learning_rate": 4.384186849401057e-05,
      "loss": 0.8023,
      "step": 4810
    },
    {
      "epoch": 0.3772108311159806,
      "grad_norm": 12.025623321533203,
      "learning_rate": 4.3828791128315115e-05,
      "loss": 0.9554,
      "step": 4820
    },
    {
      "epoch": 0.37799342620128346,
      "grad_norm": 4.926311492919922,
      "learning_rate": 4.381571376261966e-05,
      "loss": 0.7996,
      "step": 4830
    },
    {
      "epoch": 0.37877602128658633,
      "grad_norm": 7.87861442565918,
      "learning_rate": 4.3802636396924204e-05,
      "loss": 1.0401,
      "step": 4840
    },
    {
      "epoch": 0.3795586163718892,
      "grad_norm": 2.469330072402954,
      "learning_rate": 4.378955903122875e-05,
      "loss": 0.7648,
      "step": 4850
    },
    {
      "epoch": 0.3803412114571921,
      "grad_norm": 7.958139419555664,
      "learning_rate": 4.37764816655333e-05,
      "loss": 0.5369,
      "step": 4860
    },
    {
      "epoch": 0.3811238065424949,
      "grad_norm": 2.3533060550689697,
      "learning_rate": 4.376340429983784e-05,
      "loss": 0.7438,
      "step": 4870
    },
    {
      "epoch": 0.38190640162779776,
      "grad_norm": 20.174997329711914,
      "learning_rate": 4.375032693414239e-05,
      "loss": 0.8248,
      "step": 4880
    },
    {
      "epoch": 0.38268899671310064,
      "grad_norm": 4.372502326965332,
      "learning_rate": 4.3737249568446934e-05,
      "loss": 0.9441,
      "step": 4890
    },
    {
      "epoch": 0.3834715917984035,
      "grad_norm": 4.614714622497559,
      "learning_rate": 4.372417220275148e-05,
      "loss": 0.7557,
      "step": 4900
    },
    {
      "epoch": 0.3842541868837064,
      "grad_norm": 3.5219814777374268,
      "learning_rate": 4.371109483705603e-05,
      "loss": 0.8383,
      "step": 4910
    },
    {
      "epoch": 0.38503678196900926,
      "grad_norm": 5.484682083129883,
      "learning_rate": 4.369801747136057e-05,
      "loss": 0.8654,
      "step": 4920
    },
    {
      "epoch": 0.3858193770543121,
      "grad_norm": 5.502137660980225,
      "learning_rate": 4.368494010566512e-05,
      "loss": 1.1253,
      "step": 4930
    },
    {
      "epoch": 0.38660197213961495,
      "grad_norm": 9.793045997619629,
      "learning_rate": 4.367186273996966e-05,
      "loss": 0.3216,
      "step": 4940
    },
    {
      "epoch": 0.3873845672249178,
      "grad_norm": 20.375972747802734,
      "learning_rate": 4.365878537427421e-05,
      "loss": 0.7279,
      "step": 4950
    },
    {
      "epoch": 0.3881671623102207,
      "grad_norm": 4.498573303222656,
      "learning_rate": 4.364570800857875e-05,
      "loss": 0.9626,
      "step": 4960
    },
    {
      "epoch": 0.38894975739552357,
      "grad_norm": 5.550812721252441,
      "learning_rate": 4.36326306428833e-05,
      "loss": 0.9459,
      "step": 4970
    },
    {
      "epoch": 0.38973235248082644,
      "grad_norm": 5.718824863433838,
      "learning_rate": 4.361955327718785e-05,
      "loss": 0.6522,
      "step": 4980
    },
    {
      "epoch": 0.3905149475661293,
      "grad_norm": 4.715776443481445,
      "learning_rate": 4.3606475911492387e-05,
      "loss": 0.9819,
      "step": 4990
    },
    {
      "epoch": 0.39129754265143213,
      "grad_norm": 8.381905555725098,
      "learning_rate": 4.359339854579694e-05,
      "loss": 0.8428,
      "step": 5000
    },
    {
      "epoch": 0.392080137736735,
      "grad_norm": 10.693737030029297,
      "learning_rate": 4.358032118010148e-05,
      "loss": 1.0696,
      "step": 5010
    },
    {
      "epoch": 0.3928627328220379,
      "grad_norm": 3.3298988342285156,
      "learning_rate": 4.356724381440603e-05,
      "loss": 0.7439,
      "step": 5020
    },
    {
      "epoch": 0.39364532790734075,
      "grad_norm": 9.761030197143555,
      "learning_rate": 4.355416644871058e-05,
      "loss": 0.6654,
      "step": 5030
    },
    {
      "epoch": 0.3944279229926436,
      "grad_norm": 1.3037787675857544,
      "learning_rate": 4.3541089083015116e-05,
      "loss": 0.5571,
      "step": 5040
    },
    {
      "epoch": 0.3952105180779465,
      "grad_norm": 9.149327278137207,
      "learning_rate": 4.352801171731967e-05,
      "loss": 0.5131,
      "step": 5050
    },
    {
      "epoch": 0.3959931131632493,
      "grad_norm": 4.249685287475586,
      "learning_rate": 4.351493435162421e-05,
      "loss": 1.1117,
      "step": 5060
    },
    {
      "epoch": 0.3967757082485522,
      "grad_norm": 3.654939889907837,
      "learning_rate": 4.350185698592876e-05,
      "loss": 0.7362,
      "step": 5070
    },
    {
      "epoch": 0.39755830333385506,
      "grad_norm": 18.665395736694336,
      "learning_rate": 4.34887796202333e-05,
      "loss": 0.8601,
      "step": 5080
    },
    {
      "epoch": 0.39834089841915793,
      "grad_norm": 4.335116386413574,
      "learning_rate": 4.3475702254537846e-05,
      "loss": 0.7595,
      "step": 5090
    },
    {
      "epoch": 0.3991234935044608,
      "grad_norm": 8.50953197479248,
      "learning_rate": 4.34626248888424e-05,
      "loss": 0.9619,
      "step": 5100
    },
    {
      "epoch": 0.3999060885897637,
      "grad_norm": 12.90100383758545,
      "learning_rate": 4.344954752314694e-05,
      "loss": 0.9423,
      "step": 5110
    },
    {
      "epoch": 0.40068868367506655,
      "grad_norm": 4.149703025817871,
      "learning_rate": 4.3436470157451486e-05,
      "loss": 0.5267,
      "step": 5120
    },
    {
      "epoch": 0.40147127876036937,
      "grad_norm": 1.2713582515716553,
      "learning_rate": 4.342339279175603e-05,
      "loss": 0.6908,
      "step": 5130
    },
    {
      "epoch": 0.40225387384567224,
      "grad_norm": 12.91207504272461,
      "learning_rate": 4.3410315426060575e-05,
      "loss": 1.1873,
      "step": 5140
    },
    {
      "epoch": 0.4030364689309751,
      "grad_norm": 4.0377373695373535,
      "learning_rate": 4.339723806036512e-05,
      "loss": 0.85,
      "step": 5150
    },
    {
      "epoch": 0.403819064016278,
      "grad_norm": 5.970839500427246,
      "learning_rate": 4.3384160694669665e-05,
      "loss": 0.5893,
      "step": 5160
    },
    {
      "epoch": 0.40460165910158086,
      "grad_norm": 8.37652587890625,
      "learning_rate": 4.3371083328974216e-05,
      "loss": 1.1267,
      "step": 5170
    },
    {
      "epoch": 0.40538425418688373,
      "grad_norm": 5.100812911987305,
      "learning_rate": 4.335800596327876e-05,
      "loss": 0.7317,
      "step": 5180
    },
    {
      "epoch": 0.40616684927218655,
      "grad_norm": 7.9662861824035645,
      "learning_rate": 4.3344928597583305e-05,
      "loss": 0.5729,
      "step": 5190
    },
    {
      "epoch": 0.4069494443574894,
      "grad_norm": 7.416536808013916,
      "learning_rate": 4.333185123188785e-05,
      "loss": 0.6342,
      "step": 5200
    },
    {
      "epoch": 0.4077320394427923,
      "grad_norm": 5.159666538238525,
      "learning_rate": 4.3318773866192394e-05,
      "loss": 0.8068,
      "step": 5210
    },
    {
      "epoch": 0.40851463452809517,
      "grad_norm": 5.698326110839844,
      "learning_rate": 4.330569650049694e-05,
      "loss": 1.3247,
      "step": 5220
    },
    {
      "epoch": 0.40929722961339804,
      "grad_norm": 19.01287841796875,
      "learning_rate": 4.329261913480149e-05,
      "loss": 1.081,
      "step": 5230
    },
    {
      "epoch": 0.4100798246987009,
      "grad_norm": 4.0934624671936035,
      "learning_rate": 4.3279541769106035e-05,
      "loss": 0.7588,
      "step": 5240
    },
    {
      "epoch": 0.41086241978400373,
      "grad_norm": 3.2454323768615723,
      "learning_rate": 4.326646440341058e-05,
      "loss": 0.8275,
      "step": 5250
    },
    {
      "epoch": 0.4116450148693066,
      "grad_norm": 3.9217965602874756,
      "learning_rate": 4.3253387037715124e-05,
      "loss": 0.9803,
      "step": 5260
    },
    {
      "epoch": 0.4124276099546095,
      "grad_norm": 5.756573677062988,
      "learning_rate": 4.324030967201967e-05,
      "loss": 0.5852,
      "step": 5270
    },
    {
      "epoch": 0.41321020503991235,
      "grad_norm": 7.040308952331543,
      "learning_rate": 4.322723230632422e-05,
      "loss": 0.7233,
      "step": 5280
    },
    {
      "epoch": 0.4139928001252152,
      "grad_norm": 1.3038653135299683,
      "learning_rate": 4.321415494062876e-05,
      "loss": 1.0247,
      "step": 5290
    },
    {
      "epoch": 0.4147753952105181,
      "grad_norm": 13.954681396484375,
      "learning_rate": 4.320107757493331e-05,
      "loss": 0.8088,
      "step": 5300
    },
    {
      "epoch": 0.41555799029582097,
      "grad_norm": 4.800546169281006,
      "learning_rate": 4.3188000209237854e-05,
      "loss": 0.6506,
      "step": 5310
    },
    {
      "epoch": 0.4163405853811238,
      "grad_norm": 12.754584312438965,
      "learning_rate": 4.31749228435424e-05,
      "loss": 0.9028,
      "step": 5320
    },
    {
      "epoch": 0.41712318046642666,
      "grad_norm": 7.179859638214111,
      "learning_rate": 4.316184547784695e-05,
      "loss": 0.8788,
      "step": 5330
    },
    {
      "epoch": 0.41790577555172953,
      "grad_norm": 3.098322868347168,
      "learning_rate": 4.314876811215149e-05,
      "loss": 0.8621,
      "step": 5340
    },
    {
      "epoch": 0.4186883706370324,
      "grad_norm": 3.9998879432678223,
      "learning_rate": 4.313569074645604e-05,
      "loss": 0.632,
      "step": 5350
    },
    {
      "epoch": 0.4194709657223353,
      "grad_norm": 8.211560249328613,
      "learning_rate": 4.3122613380760576e-05,
      "loss": 0.7339,
      "step": 5360
    },
    {
      "epoch": 0.42025356080763815,
      "grad_norm": 7.214698791503906,
      "learning_rate": 4.310953601506513e-05,
      "loss": 1.3738,
      "step": 5370
    },
    {
      "epoch": 0.42103615589294097,
      "grad_norm": 10.63038158416748,
      "learning_rate": 4.309645864936967e-05,
      "loss": 1.1848,
      "step": 5380
    },
    {
      "epoch": 0.42181875097824384,
      "grad_norm": 5.349085807800293,
      "learning_rate": 4.308338128367422e-05,
      "loss": 0.8479,
      "step": 5390
    },
    {
      "epoch": 0.4226013460635467,
      "grad_norm": 1.9422141313552856,
      "learning_rate": 4.307030391797877e-05,
      "loss": 0.823,
      "step": 5400
    },
    {
      "epoch": 0.4233839411488496,
      "grad_norm": 8.411735534667969,
      "learning_rate": 4.3057226552283306e-05,
      "loss": 0.9487,
      "step": 5410
    },
    {
      "epoch": 0.42416653623415246,
      "grad_norm": 9.314790725708008,
      "learning_rate": 4.304414918658786e-05,
      "loss": 0.5545,
      "step": 5420
    },
    {
      "epoch": 0.42494913131945533,
      "grad_norm": 7.006775856018066,
      "learning_rate": 4.30310718208924e-05,
      "loss": 0.8145,
      "step": 5430
    },
    {
      "epoch": 0.4257317264047582,
      "grad_norm": 7.959507465362549,
      "learning_rate": 4.3017994455196947e-05,
      "loss": 0.7338,
      "step": 5440
    },
    {
      "epoch": 0.426514321490061,
      "grad_norm": 7.3458476066589355,
      "learning_rate": 4.30049170895015e-05,
      "loss": 1.0543,
      "step": 5450
    },
    {
      "epoch": 0.4272969165753639,
      "grad_norm": 4.901602268218994,
      "learning_rate": 4.2991839723806036e-05,
      "loss": 0.9533,
      "step": 5460
    },
    {
      "epoch": 0.42807951166066677,
      "grad_norm": 7.371197700500488,
      "learning_rate": 4.297876235811059e-05,
      "loss": 0.5822,
      "step": 5470
    },
    {
      "epoch": 0.42886210674596964,
      "grad_norm": 4.84442138671875,
      "learning_rate": 4.296568499241513e-05,
      "loss": 0.8858,
      "step": 5480
    },
    {
      "epoch": 0.4296447018312725,
      "grad_norm": 2.7441470623016357,
      "learning_rate": 4.2952607626719676e-05,
      "loss": 1.0043,
      "step": 5490
    },
    {
      "epoch": 0.4304272969165754,
      "grad_norm": 4.679873943328857,
      "learning_rate": 4.293953026102422e-05,
      "loss": 1.0494,
      "step": 5500
    },
    {
      "epoch": 0.4312098920018782,
      "grad_norm": 18.664464950561523,
      "learning_rate": 4.2926452895328765e-05,
      "loss": 0.7783,
      "step": 5510
    },
    {
      "epoch": 0.4319924870871811,
      "grad_norm": 6.715144157409668,
      "learning_rate": 4.291337552963332e-05,
      "loss": 0.905,
      "step": 5520
    },
    {
      "epoch": 0.43277508217248395,
      "grad_norm": 3.7708868980407715,
      "learning_rate": 4.2900298163937854e-05,
      "loss": 0.9154,
      "step": 5530
    },
    {
      "epoch": 0.4335576772577868,
      "grad_norm": 3.8173434734344482,
      "learning_rate": 4.2887220798242406e-05,
      "loss": 0.7823,
      "step": 5540
    },
    {
      "epoch": 0.4343402723430897,
      "grad_norm": 8.107263565063477,
      "learning_rate": 4.287414343254695e-05,
      "loss": 1.0085,
      "step": 5550
    },
    {
      "epoch": 0.43512286742839257,
      "grad_norm": 4.369537830352783,
      "learning_rate": 4.2861066066851495e-05,
      "loss": 0.9796,
      "step": 5560
    },
    {
      "epoch": 0.43590546251369544,
      "grad_norm": 7.889426231384277,
      "learning_rate": 4.284798870115604e-05,
      "loss": 0.7638,
      "step": 5570
    },
    {
      "epoch": 0.43668805759899826,
      "grad_norm": 1.6684634685516357,
      "learning_rate": 4.2834911335460584e-05,
      "loss": 0.7725,
      "step": 5580
    },
    {
      "epoch": 0.43747065268430113,
      "grad_norm": 5.122396945953369,
      "learning_rate": 4.2821833969765135e-05,
      "loss": 0.7234,
      "step": 5590
    },
    {
      "epoch": 0.438253247769604,
      "grad_norm": 9.798338890075684,
      "learning_rate": 4.280875660406968e-05,
      "loss": 1.0717,
      "step": 5600
    },
    {
      "epoch": 0.4390358428549069,
      "grad_norm": 4.712523937225342,
      "learning_rate": 4.2795679238374225e-05,
      "loss": 0.7866,
      "step": 5610
    },
    {
      "epoch": 0.43981843794020975,
      "grad_norm": 5.201725959777832,
      "learning_rate": 4.278260187267877e-05,
      "loss": 0.4902,
      "step": 5620
    },
    {
      "epoch": 0.4406010330255126,
      "grad_norm": 4.470975875854492,
      "learning_rate": 4.2769524506983314e-05,
      "loss": 1.1571,
      "step": 5630
    },
    {
      "epoch": 0.44138362811081544,
      "grad_norm": 7.608837127685547,
      "learning_rate": 4.275644714128786e-05,
      "loss": 0.8757,
      "step": 5640
    },
    {
      "epoch": 0.4421662231961183,
      "grad_norm": 7.587765216827393,
      "learning_rate": 4.274336977559241e-05,
      "loss": 0.806,
      "step": 5650
    },
    {
      "epoch": 0.4429488182814212,
      "grad_norm": 5.839571475982666,
      "learning_rate": 4.2730292409896954e-05,
      "loss": 0.7567,
      "step": 5660
    },
    {
      "epoch": 0.44373141336672406,
      "grad_norm": 7.573368072509766,
      "learning_rate": 4.27172150442015e-05,
      "loss": 0.6052,
      "step": 5670
    },
    {
      "epoch": 0.44451400845202693,
      "grad_norm": 13.171113967895508,
      "learning_rate": 4.270413767850604e-05,
      "loss": 1.1772,
      "step": 5680
    },
    {
      "epoch": 0.4452966035373298,
      "grad_norm": 8.026895523071289,
      "learning_rate": 4.269106031281059e-05,
      "loss": 0.8674,
      "step": 5690
    },
    {
      "epoch": 0.4460791986226326,
      "grad_norm": 3.2458009719848633,
      "learning_rate": 4.267798294711514e-05,
      "loss": 0.9804,
      "step": 5700
    },
    {
      "epoch": 0.4468617937079355,
      "grad_norm": 6.227912425994873,
      "learning_rate": 4.266490558141968e-05,
      "loss": 0.866,
      "step": 5710
    },
    {
      "epoch": 0.44764438879323837,
      "grad_norm": 8.100434303283691,
      "learning_rate": 4.265182821572423e-05,
      "loss": 1.1335,
      "step": 5720
    },
    {
      "epoch": 0.44842698387854124,
      "grad_norm": 6.692510604858398,
      "learning_rate": 4.263875085002877e-05,
      "loss": 1.0506,
      "step": 5730
    },
    {
      "epoch": 0.4492095789638441,
      "grad_norm": 6.5808000564575195,
      "learning_rate": 4.262567348433332e-05,
      "loss": 1.0801,
      "step": 5740
    },
    {
      "epoch": 0.449992174049147,
      "grad_norm": 5.460785865783691,
      "learning_rate": 4.261259611863786e-05,
      "loss": 0.9518,
      "step": 5750
    },
    {
      "epoch": 0.45077476913444986,
      "grad_norm": 1.568529486656189,
      "learning_rate": 4.259951875294241e-05,
      "loss": 0.5632,
      "step": 5760
    },
    {
      "epoch": 0.4515573642197527,
      "grad_norm": 7.855735778808594,
      "learning_rate": 4.258644138724696e-05,
      "loss": 0.79,
      "step": 5770
    },
    {
      "epoch": 0.45233995930505555,
      "grad_norm": 10.289188385009766,
      "learning_rate": 4.2573364021551496e-05,
      "loss": 0.8663,
      "step": 5780
    },
    {
      "epoch": 0.4531225543903584,
      "grad_norm": 2.500807523727417,
      "learning_rate": 4.256028665585605e-05,
      "loss": 0.7722,
      "step": 5790
    },
    {
      "epoch": 0.4539051494756613,
      "grad_norm": 4.384109973907471,
      "learning_rate": 4.254720929016059e-05,
      "loss": 1.2004,
      "step": 5800
    },
    {
      "epoch": 0.45468774456096417,
      "grad_norm": 9.316767692565918,
      "learning_rate": 4.2534131924465136e-05,
      "loss": 1.0572,
      "step": 5810
    },
    {
      "epoch": 0.45547033964626704,
      "grad_norm": 3.3972270488739014,
      "learning_rate": 4.252105455876969e-05,
      "loss": 1.0277,
      "step": 5820
    },
    {
      "epoch": 0.45625293473156986,
      "grad_norm": 8.7282133102417,
      "learning_rate": 4.2507977193074226e-05,
      "loss": 1.29,
      "step": 5830
    },
    {
      "epoch": 0.45703552981687273,
      "grad_norm": 9.969352722167969,
      "learning_rate": 4.249489982737878e-05,
      "loss": 0.7479,
      "step": 5840
    },
    {
      "epoch": 0.4578181249021756,
      "grad_norm": 6.853831768035889,
      "learning_rate": 4.248182246168332e-05,
      "loss": 1.0817,
      "step": 5850
    },
    {
      "epoch": 0.4586007199874785,
      "grad_norm": 3.179537773132324,
      "learning_rate": 4.2468745095987866e-05,
      "loss": 0.6744,
      "step": 5860
    },
    {
      "epoch": 0.45938331507278135,
      "grad_norm": 5.318263053894043,
      "learning_rate": 4.245566773029242e-05,
      "loss": 0.9519,
      "step": 5870
    },
    {
      "epoch": 0.4601659101580842,
      "grad_norm": 4.587821960449219,
      "learning_rate": 4.2442590364596955e-05,
      "loss": 0.7541,
      "step": 5880
    },
    {
      "epoch": 0.4609485052433871,
      "grad_norm": 1.308416485786438,
      "learning_rate": 4.2429512998901506e-05,
      "loss": 0.8088,
      "step": 5890
    },
    {
      "epoch": 0.4617311003286899,
      "grad_norm": 5.100376129150391,
      "learning_rate": 4.2416435633206044e-05,
      "loss": 0.7451,
      "step": 5900
    },
    {
      "epoch": 0.4625136954139928,
      "grad_norm": 1.8653109073638916,
      "learning_rate": 4.2403358267510596e-05,
      "loss": 0.6831,
      "step": 5910
    },
    {
      "epoch": 0.46329629049929566,
      "grad_norm": 5.50417947769165,
      "learning_rate": 4.239028090181514e-05,
      "loss": 0.9182,
      "step": 5920
    },
    {
      "epoch": 0.46407888558459853,
      "grad_norm": 9.242417335510254,
      "learning_rate": 4.2377203536119685e-05,
      "loss": 0.9053,
      "step": 5930
    },
    {
      "epoch": 0.4648614806699014,
      "grad_norm": 4.24540901184082,
      "learning_rate": 4.2364126170424236e-05,
      "loss": 1.0467,
      "step": 5940
    },
    {
      "epoch": 0.4656440757552043,
      "grad_norm": 8.809833526611328,
      "learning_rate": 4.2351048804728774e-05,
      "loss": 0.8238,
      "step": 5950
    },
    {
      "epoch": 0.4664266708405071,
      "grad_norm": 0.8909691572189331,
      "learning_rate": 4.2337971439033325e-05,
      "loss": 0.8654,
      "step": 5960
    },
    {
      "epoch": 0.46720926592580997,
      "grad_norm": 6.090839862823486,
      "learning_rate": 4.232489407333787e-05,
      "loss": 0.8935,
      "step": 5970
    },
    {
      "epoch": 0.46799186101111284,
      "grad_norm": 2.231489419937134,
      "learning_rate": 4.2311816707642414e-05,
      "loss": 0.6473,
      "step": 5980
    },
    {
      "epoch": 0.4687744560964157,
      "grad_norm": 0.8318849802017212,
      "learning_rate": 4.229873934194696e-05,
      "loss": 0.8164,
      "step": 5990
    },
    {
      "epoch": 0.4695570511817186,
      "grad_norm": 5.023898601531982,
      "learning_rate": 4.2285661976251504e-05,
      "loss": 0.6729,
      "step": 6000
    },
    {
      "epoch": 0.47033964626702146,
      "grad_norm": 7.08870267868042,
      "learning_rate": 4.2272584610556055e-05,
      "loss": 0.6985,
      "step": 6010
    },
    {
      "epoch": 0.47112224135232433,
      "grad_norm": 6.056117534637451,
      "learning_rate": 4.22595072448606e-05,
      "loss": 0.605,
      "step": 6020
    },
    {
      "epoch": 0.47190483643762715,
      "grad_norm": 2.5088109970092773,
      "learning_rate": 4.2246429879165144e-05,
      "loss": 0.8572,
      "step": 6030
    },
    {
      "epoch": 0.47268743152293,
      "grad_norm": 1.9034944772720337,
      "learning_rate": 4.223335251346969e-05,
      "loss": 0.812,
      "step": 6040
    },
    {
      "epoch": 0.4734700266082329,
      "grad_norm": 9.28056526184082,
      "learning_rate": 4.222027514777423e-05,
      "loss": 1.0801,
      "step": 6050
    },
    {
      "epoch": 0.47425262169353577,
      "grad_norm": 1.6649211645126343,
      "learning_rate": 4.220719778207878e-05,
      "loss": 0.579,
      "step": 6060
    },
    {
      "epoch": 0.47503521677883864,
      "grad_norm": 4.8080830574035645,
      "learning_rate": 4.219412041638333e-05,
      "loss": 0.5903,
      "step": 6070
    },
    {
      "epoch": 0.4758178118641415,
      "grad_norm": 9.352557182312012,
      "learning_rate": 4.2181043050687874e-05,
      "loss": 0.7024,
      "step": 6080
    },
    {
      "epoch": 0.47660040694944433,
      "grad_norm": 2.6933300495147705,
      "learning_rate": 4.216796568499242e-05,
      "loss": 0.6504,
      "step": 6090
    },
    {
      "epoch": 0.4773830020347472,
      "grad_norm": 10.936189651489258,
      "learning_rate": 4.215488831929696e-05,
      "loss": 1.2889,
      "step": 6100
    },
    {
      "epoch": 0.4781655971200501,
      "grad_norm": 1.0524879693984985,
      "learning_rate": 4.214181095360151e-05,
      "loss": 0.8204,
      "step": 6110
    },
    {
      "epoch": 0.47894819220535295,
      "grad_norm": 3.820434093475342,
      "learning_rate": 4.212873358790605e-05,
      "loss": 0.9388,
      "step": 6120
    },
    {
      "epoch": 0.4797307872906558,
      "grad_norm": 19.879465103149414,
      "learning_rate": 4.2115656222210597e-05,
      "loss": 0.9098,
      "step": 6130
    },
    {
      "epoch": 0.4805133823759587,
      "grad_norm": 6.985306739807129,
      "learning_rate": 4.210257885651515e-05,
      "loss": 0.6145,
      "step": 6140
    },
    {
      "epoch": 0.48129597746126157,
      "grad_norm": 3.9043712615966797,
      "learning_rate": 4.208950149081969e-05,
      "loss": 0.4707,
      "step": 6150
    },
    {
      "epoch": 0.4820785725465644,
      "grad_norm": 3.002317190170288,
      "learning_rate": 4.207642412512424e-05,
      "loss": 0.9738,
      "step": 6160
    },
    {
      "epoch": 0.48286116763186726,
      "grad_norm": 7.402571201324463,
      "learning_rate": 4.206334675942878e-05,
      "loss": 1.149,
      "step": 6170
    },
    {
      "epoch": 0.48364376271717013,
      "grad_norm": 9.954267501831055,
      "learning_rate": 4.2050269393733326e-05,
      "loss": 0.6372,
      "step": 6180
    },
    {
      "epoch": 0.484426357802473,
      "grad_norm": 6.13657808303833,
      "learning_rate": 4.203719202803788e-05,
      "loss": 0.741,
      "step": 6190
    },
    {
      "epoch": 0.4852089528877759,
      "grad_norm": 29.618146896362305,
      "learning_rate": 4.2024114662342415e-05,
      "loss": 0.7731,
      "step": 6200
    },
    {
      "epoch": 0.48599154797307875,
      "grad_norm": 0.6891242861747742,
      "learning_rate": 4.201103729664697e-05,
      "loss": 0.9102,
      "step": 6210
    },
    {
      "epoch": 0.48677414305838157,
      "grad_norm": 6.880820274353027,
      "learning_rate": 4.199795993095151e-05,
      "loss": 0.7561,
      "step": 6220
    },
    {
      "epoch": 0.48755673814368444,
      "grad_norm": 9.175710678100586,
      "learning_rate": 4.1984882565256056e-05,
      "loss": 1.0944,
      "step": 6230
    },
    {
      "epoch": 0.4883393332289873,
      "grad_norm": 3.079404830932617,
      "learning_rate": 4.197180519956061e-05,
      "loss": 0.6762,
      "step": 6240
    },
    {
      "epoch": 0.4891219283142902,
      "grad_norm": 4.204259395599365,
      "learning_rate": 4.1958727833865145e-05,
      "loss": 1.2591,
      "step": 6250
    },
    {
      "epoch": 0.48990452339959306,
      "grad_norm": 13.256816864013672,
      "learning_rate": 4.1945650468169696e-05,
      "loss": 0.9979,
      "step": 6260
    },
    {
      "epoch": 0.49068711848489593,
      "grad_norm": 10.875786781311035,
      "learning_rate": 4.1932573102474234e-05,
      "loss": 1.0764,
      "step": 6270
    },
    {
      "epoch": 0.49146971357019875,
      "grad_norm": 4.268348217010498,
      "learning_rate": 4.1919495736778785e-05,
      "loss": 0.8091,
      "step": 6280
    },
    {
      "epoch": 0.4922523086555016,
      "grad_norm": 7.001514911651611,
      "learning_rate": 4.190641837108334e-05,
      "loss": 0.9622,
      "step": 6290
    },
    {
      "epoch": 0.4930349037408045,
      "grad_norm": 9.879840850830078,
      "learning_rate": 4.1893341005387875e-05,
      "loss": 0.7132,
      "step": 6300
    },
    {
      "epoch": 0.49381749882610737,
      "grad_norm": 5.099706649780273,
      "learning_rate": 4.1880263639692426e-05,
      "loss": 0.8669,
      "step": 6310
    },
    {
      "epoch": 0.49460009391141024,
      "grad_norm": 9.640771865844727,
      "learning_rate": 4.1867186273996964e-05,
      "loss": 1.1774,
      "step": 6320
    },
    {
      "epoch": 0.4953826889967131,
      "grad_norm": 4.465900897979736,
      "learning_rate": 4.1854108908301515e-05,
      "loss": 1.1188,
      "step": 6330
    },
    {
      "epoch": 0.496165284082016,
      "grad_norm": 2.9091076850891113,
      "learning_rate": 4.184103154260606e-05,
      "loss": 0.8245,
      "step": 6340
    },
    {
      "epoch": 0.4969478791673188,
      "grad_norm": 5.486077785491943,
      "learning_rate": 4.1827954176910604e-05,
      "loss": 0.7245,
      "step": 6350
    },
    {
      "epoch": 0.4977304742526217,
      "grad_norm": 2.82596492767334,
      "learning_rate": 4.1814876811215156e-05,
      "loss": 0.8598,
      "step": 6360
    },
    {
      "epoch": 0.49851306933792455,
      "grad_norm": 3.2534055709838867,
      "learning_rate": 4.1801799445519693e-05,
      "loss": 1.0666,
      "step": 6370
    },
    {
      "epoch": 0.4992956644232274,
      "grad_norm": 3.977182388305664,
      "learning_rate": 4.1788722079824245e-05,
      "loss": 0.6384,
      "step": 6380
    },
    {
      "epoch": 0.5000782595085302,
      "grad_norm": 6.799421787261963,
      "learning_rate": 4.177564471412879e-05,
      "loss": 1.0839,
      "step": 6390
    },
    {
      "epoch": 0.5008608545938331,
      "grad_norm": 3.1868929862976074,
      "learning_rate": 4.1762567348433334e-05,
      "loss": 0.9354,
      "step": 6400
    },
    {
      "epoch": 0.501643449679136,
      "grad_norm": 2.262669801712036,
      "learning_rate": 4.174948998273788e-05,
      "loss": 0.8898,
      "step": 6410
    },
    {
      "epoch": 0.5024260447644389,
      "grad_norm": 7.541187763214111,
      "learning_rate": 4.173641261704242e-05,
      "loss": 0.9433,
      "step": 6420
    },
    {
      "epoch": 0.5032086398497417,
      "grad_norm": 7.670241355895996,
      "learning_rate": 4.1723335251346974e-05,
      "loss": 1.1123,
      "step": 6430
    },
    {
      "epoch": 0.5039912349350446,
      "grad_norm": 4.806732654571533,
      "learning_rate": 4.171025788565152e-05,
      "loss": 0.7624,
      "step": 6440
    },
    {
      "epoch": 0.5047738300203475,
      "grad_norm": 2.4296107292175293,
      "learning_rate": 4.1697180519956064e-05,
      "loss": 0.9113,
      "step": 6450
    },
    {
      "epoch": 0.5055564251056504,
      "grad_norm": 12.619784355163574,
      "learning_rate": 4.168410315426061e-05,
      "loss": 0.7343,
      "step": 6460
    },
    {
      "epoch": 0.5063390201909532,
      "grad_norm": 5.003669738769531,
      "learning_rate": 4.167102578856515e-05,
      "loss": 0.7921,
      "step": 6470
    },
    {
      "epoch": 0.5071216152762561,
      "grad_norm": 3.6604151725769043,
      "learning_rate": 4.16579484228697e-05,
      "loss": 0.8914,
      "step": 6480
    },
    {
      "epoch": 0.507904210361559,
      "grad_norm": 3.455329656600952,
      "learning_rate": 4.164487105717424e-05,
      "loss": 0.6749,
      "step": 6490
    },
    {
      "epoch": 0.5086868054468618,
      "grad_norm": 1.7722100019454956,
      "learning_rate": 4.163179369147879e-05,
      "loss": 0.6487,
      "step": 6500
    },
    {
      "epoch": 0.5094694005321646,
      "grad_norm": 9.74009895324707,
      "learning_rate": 4.161871632578334e-05,
      "loss": 0.7673,
      "step": 6510
    },
    {
      "epoch": 0.5102519956174675,
      "grad_norm": 17.43278694152832,
      "learning_rate": 4.160563896008788e-05,
      "loss": 0.9189,
      "step": 6520
    },
    {
      "epoch": 0.5110345907027704,
      "grad_norm": 8.360116004943848,
      "learning_rate": 4.159256159439243e-05,
      "loss": 1.0359,
      "step": 6530
    },
    {
      "epoch": 0.5118171857880732,
      "grad_norm": 7.835478782653809,
      "learning_rate": 4.157948422869697e-05,
      "loss": 1.2056,
      "step": 6540
    },
    {
      "epoch": 0.5125997808733761,
      "grad_norm": 2.389543294906616,
      "learning_rate": 4.1566406863001516e-05,
      "loss": 0.8126,
      "step": 6550
    },
    {
      "epoch": 0.513382375958679,
      "grad_norm": 1.9873762130737305,
      "learning_rate": 4.155332949730607e-05,
      "loss": 0.6306,
      "step": 6560
    },
    {
      "epoch": 0.5141649710439818,
      "grad_norm": 8.879312515258789,
      "learning_rate": 4.154025213161061e-05,
      "loss": 0.8691,
      "step": 6570
    },
    {
      "epoch": 0.5149475661292847,
      "grad_norm": 12.108166694641113,
      "learning_rate": 4.1527174765915157e-05,
      "loss": 0.7999,
      "step": 6580
    },
    {
      "epoch": 0.5157301612145876,
      "grad_norm": 7.578449726104736,
      "learning_rate": 4.15140974002197e-05,
      "loss": 0.7275,
      "step": 6590
    },
    {
      "epoch": 0.5165127562998905,
      "grad_norm": 3.6329076290130615,
      "learning_rate": 4.1501020034524246e-05,
      "loss": 1.1944,
      "step": 6600
    },
    {
      "epoch": 0.5172953513851933,
      "grad_norm": 5.156015396118164,
      "learning_rate": 4.14879426688288e-05,
      "loss": 0.8748,
      "step": 6610
    },
    {
      "epoch": 0.5180779464704962,
      "grad_norm": 9.462631225585938,
      "learning_rate": 4.1474865303133335e-05,
      "loss": 0.9658,
      "step": 6620
    },
    {
      "epoch": 0.5188605415557991,
      "grad_norm": 6.577657699584961,
      "learning_rate": 4.1461787937437886e-05,
      "loss": 0.7307,
      "step": 6630
    },
    {
      "epoch": 0.5196431366411018,
      "grad_norm": 5.839386940002441,
      "learning_rate": 4.144871057174243e-05,
      "loss": 0.6848,
      "step": 6640
    },
    {
      "epoch": 0.5204257317264047,
      "grad_norm": 2.0847902297973633,
      "learning_rate": 4.1435633206046975e-05,
      "loss": 0.7109,
      "step": 6650
    },
    {
      "epoch": 0.5212083268117076,
      "grad_norm": 8.381388664245605,
      "learning_rate": 4.142255584035153e-05,
      "loss": 1.0366,
      "step": 6660
    },
    {
      "epoch": 0.5219909218970105,
      "grad_norm": 4.458929538726807,
      "learning_rate": 4.1409478474656064e-05,
      "loss": 0.8372,
      "step": 6670
    },
    {
      "epoch": 0.5227735169823133,
      "grad_norm": 9.136251449584961,
      "learning_rate": 4.1396401108960616e-05,
      "loss": 1.0537,
      "step": 6680
    },
    {
      "epoch": 0.5235561120676162,
      "grad_norm": 2.092806100845337,
      "learning_rate": 4.1383323743265154e-05,
      "loss": 0.8706,
      "step": 6690
    },
    {
      "epoch": 0.5243387071529191,
      "grad_norm": 3.753567695617676,
      "learning_rate": 4.1370246377569705e-05,
      "loss": 0.8737,
      "step": 6700
    },
    {
      "epoch": 0.525121302238222,
      "grad_norm": 4.9837141036987305,
      "learning_rate": 4.135716901187425e-05,
      "loss": 0.9368,
      "step": 6710
    },
    {
      "epoch": 0.5259038973235248,
      "grad_norm": 4.6572651863098145,
      "learning_rate": 4.1344091646178794e-05,
      "loss": 0.9227,
      "step": 6720
    },
    {
      "epoch": 0.5266864924088277,
      "grad_norm": 4.516173839569092,
      "learning_rate": 4.1331014280483345e-05,
      "loss": 0.9388,
      "step": 6730
    },
    {
      "epoch": 0.5274690874941306,
      "grad_norm": 6.859401226043701,
      "learning_rate": 4.131793691478788e-05,
      "loss": 0.7125,
      "step": 6740
    },
    {
      "epoch": 0.5282516825794334,
      "grad_norm": 7.8523359298706055,
      "learning_rate": 4.1304859549092435e-05,
      "loss": 0.8351,
      "step": 6750
    },
    {
      "epoch": 0.5290342776647363,
      "grad_norm": 8.20600414276123,
      "learning_rate": 4.129178218339698e-05,
      "loss": 0.958,
      "step": 6760
    },
    {
      "epoch": 0.5298168727500391,
      "grad_norm": 5.009696006774902,
      "learning_rate": 4.1278704817701524e-05,
      "loss": 0.8577,
      "step": 6770
    },
    {
      "epoch": 0.530599467835342,
      "grad_norm": 7.609869956970215,
      "learning_rate": 4.1265627452006075e-05,
      "loss": 0.8644,
      "step": 6780
    },
    {
      "epoch": 0.5313820629206448,
      "grad_norm": 5.101411819458008,
      "learning_rate": 4.125255008631061e-05,
      "loss": 0.776,
      "step": 6790
    },
    {
      "epoch": 0.5321646580059477,
      "grad_norm": 7.355860233306885,
      "learning_rate": 4.1239472720615164e-05,
      "loss": 0.9497,
      "step": 6800
    },
    {
      "epoch": 0.5329472530912506,
      "grad_norm": 5.411699295043945,
      "learning_rate": 4.122639535491971e-05,
      "loss": 0.9666,
      "step": 6810
    },
    {
      "epoch": 0.5337298481765534,
      "grad_norm": 4.4015302658081055,
      "learning_rate": 4.121331798922425e-05,
      "loss": 0.7601,
      "step": 6820
    },
    {
      "epoch": 0.5345124432618563,
      "grad_norm": 2.0549137592315674,
      "learning_rate": 4.12002406235288e-05,
      "loss": 0.8668,
      "step": 6830
    },
    {
      "epoch": 0.5352950383471592,
      "grad_norm": 7.405129432678223,
      "learning_rate": 4.118716325783334e-05,
      "loss": 1.0562,
      "step": 6840
    },
    {
      "epoch": 0.5360776334324621,
      "grad_norm": 5.602811813354492,
      "learning_rate": 4.1174085892137894e-05,
      "loss": 0.5263,
      "step": 6850
    },
    {
      "epoch": 0.5368602285177649,
      "grad_norm": 4.994288921356201,
      "learning_rate": 4.116100852644243e-05,
      "loss": 1.0705,
      "step": 6860
    },
    {
      "epoch": 0.5376428236030678,
      "grad_norm": 7.019219875335693,
      "learning_rate": 4.114793116074698e-05,
      "loss": 1.053,
      "step": 6870
    },
    {
      "epoch": 0.5384254186883707,
      "grad_norm": 10.013219833374023,
      "learning_rate": 4.113485379505153e-05,
      "loss": 1.229,
      "step": 6880
    },
    {
      "epoch": 0.5392080137736736,
      "grad_norm": 3.13301944732666,
      "learning_rate": 4.112177642935607e-05,
      "loss": 0.8034,
      "step": 6890
    },
    {
      "epoch": 0.5399906088589763,
      "grad_norm": 6.059565544128418,
      "learning_rate": 4.110869906366062e-05,
      "loss": 1.0193,
      "step": 6900
    },
    {
      "epoch": 0.5407732039442792,
      "grad_norm": 5.790078639984131,
      "learning_rate": 4.109562169796516e-05,
      "loss": 0.491,
      "step": 6910
    },
    {
      "epoch": 0.5415557990295821,
      "grad_norm": 8.443180084228516,
      "learning_rate": 4.108254433226971e-05,
      "loss": 0.8935,
      "step": 6920
    },
    {
      "epoch": 0.5423383941148849,
      "grad_norm": 9.563846588134766,
      "learning_rate": 4.106946696657426e-05,
      "loss": 0.8208,
      "step": 6930
    },
    {
      "epoch": 0.5431209892001878,
      "grad_norm": 6.550536155700684,
      "learning_rate": 4.10563896008788e-05,
      "loss": 0.7168,
      "step": 6940
    },
    {
      "epoch": 0.5439035842854907,
      "grad_norm": 1.621096134185791,
      "learning_rate": 4.1043312235183346e-05,
      "loss": 0.8407,
      "step": 6950
    },
    {
      "epoch": 0.5446861793707936,
      "grad_norm": 6.381199836730957,
      "learning_rate": 4.103023486948789e-05,
      "loss": 1.0467,
      "step": 6960
    },
    {
      "epoch": 0.5454687744560964,
      "grad_norm": 2.989492654800415,
      "learning_rate": 4.1017157503792436e-05,
      "loss": 0.7224,
      "step": 6970
    },
    {
      "epoch": 0.5462513695413993,
      "grad_norm": 1.3734077215194702,
      "learning_rate": 4.100408013809699e-05,
      "loss": 0.7701,
      "step": 6980
    },
    {
      "epoch": 0.5470339646267022,
      "grad_norm": 15.767515182495117,
      "learning_rate": 4.0991002772401525e-05,
      "loss": 1.1411,
      "step": 6990
    },
    {
      "epoch": 0.547816559712005,
      "grad_norm": 5.225765228271484,
      "learning_rate": 4.0977925406706076e-05,
      "loss": 0.8272,
      "step": 7000
    },
    {
      "epoch": 0.5485991547973079,
      "grad_norm": 5.506872653961182,
      "learning_rate": 4.096484804101062e-05,
      "loss": 0.7649,
      "step": 7010
    },
    {
      "epoch": 0.5493817498826108,
      "grad_norm": 2.5691516399383545,
      "learning_rate": 4.0951770675315165e-05,
      "loss": 0.5469,
      "step": 7020
    },
    {
      "epoch": 0.5501643449679136,
      "grad_norm": 11.37523078918457,
      "learning_rate": 4.0938693309619716e-05,
      "loss": 0.5729,
      "step": 7030
    },
    {
      "epoch": 0.5509469400532164,
      "grad_norm": 20.04637908935547,
      "learning_rate": 4.0925615943924254e-05,
      "loss": 0.935,
      "step": 7040
    },
    {
      "epoch": 0.5517295351385193,
      "grad_norm": 3.848532199859619,
      "learning_rate": 4.0912538578228806e-05,
      "loss": 0.8966,
      "step": 7050
    },
    {
      "epoch": 0.5525121302238222,
      "grad_norm": 4.016798496246338,
      "learning_rate": 4.0899461212533343e-05,
      "loss": 0.765,
      "step": 7060
    },
    {
      "epoch": 0.553294725309125,
      "grad_norm": 2.296889543533325,
      "learning_rate": 4.0886383846837895e-05,
      "loss": 0.8442,
      "step": 7070
    },
    {
      "epoch": 0.5540773203944279,
      "grad_norm": 6.586348533630371,
      "learning_rate": 4.087330648114244e-05,
      "loss": 0.8741,
      "step": 7080
    },
    {
      "epoch": 0.5548599154797308,
      "grad_norm": 6.29811954498291,
      "learning_rate": 4.0860229115446984e-05,
      "loss": 0.7246,
      "step": 7090
    },
    {
      "epoch": 0.5556425105650337,
      "grad_norm": 9.012500762939453,
      "learning_rate": 4.0847151749751535e-05,
      "loss": 0.8175,
      "step": 7100
    },
    {
      "epoch": 0.5564251056503365,
      "grad_norm": 6.962828636169434,
      "learning_rate": 4.083407438405607e-05,
      "loss": 0.5927,
      "step": 7110
    },
    {
      "epoch": 0.5572077007356394,
      "grad_norm": 0.6537289619445801,
      "learning_rate": 4.0820997018360624e-05,
      "loss": 0.5156,
      "step": 7120
    },
    {
      "epoch": 0.5579902958209423,
      "grad_norm": 17.705307006835938,
      "learning_rate": 4.080791965266517e-05,
      "loss": 0.6636,
      "step": 7130
    },
    {
      "epoch": 0.5587728909062452,
      "grad_norm": 34.3797721862793,
      "learning_rate": 4.0794842286969714e-05,
      "loss": 1.1391,
      "step": 7140
    },
    {
      "epoch": 0.5595554859915479,
      "grad_norm": 13.07193660736084,
      "learning_rate": 4.0781764921274265e-05,
      "loss": 0.981,
      "step": 7150
    },
    {
      "epoch": 0.5603380810768508,
      "grad_norm": 14.050215721130371,
      "learning_rate": 4.07686875555788e-05,
      "loss": 0.8724,
      "step": 7160
    },
    {
      "epoch": 0.5611206761621537,
      "grad_norm": 0.9045813083648682,
      "learning_rate": 4.0755610189883354e-05,
      "loss": 0.7424,
      "step": 7170
    },
    {
      "epoch": 0.5619032712474565,
      "grad_norm": 0.9760553240776062,
      "learning_rate": 4.07425328241879e-05,
      "loss": 0.6682,
      "step": 7180
    },
    {
      "epoch": 0.5626858663327594,
      "grad_norm": 6.105494499206543,
      "learning_rate": 4.072945545849244e-05,
      "loss": 0.969,
      "step": 7190
    },
    {
      "epoch": 0.5634684614180623,
      "grad_norm": 11.819249153137207,
      "learning_rate": 4.071637809279699e-05,
      "loss": 0.9312,
      "step": 7200
    },
    {
      "epoch": 0.5642510565033652,
      "grad_norm": 8.315714836120605,
      "learning_rate": 4.070330072710153e-05,
      "loss": 0.9955,
      "step": 7210
    },
    {
      "epoch": 0.565033651588668,
      "grad_norm": 7.560182094573975,
      "learning_rate": 4.0690223361406084e-05,
      "loss": 0.9901,
      "step": 7220
    },
    {
      "epoch": 0.5658162466739709,
      "grad_norm": 7.071794033050537,
      "learning_rate": 4.067714599571063e-05,
      "loss": 0.8023,
      "step": 7230
    },
    {
      "epoch": 0.5665988417592738,
      "grad_norm": 5.295083999633789,
      "learning_rate": 4.066406863001517e-05,
      "loss": 1.1322,
      "step": 7240
    },
    {
      "epoch": 0.5673814368445766,
      "grad_norm": 6.615595817565918,
      "learning_rate": 4.065099126431972e-05,
      "loss": 0.7708,
      "step": 7250
    },
    {
      "epoch": 0.5681640319298795,
      "grad_norm": 5.040798187255859,
      "learning_rate": 4.063791389862426e-05,
      "loss": 0.7418,
      "step": 7260
    },
    {
      "epoch": 0.5689466270151824,
      "grad_norm": 0.8150774240493774,
      "learning_rate": 4.0624836532928807e-05,
      "loss": 0.8378,
      "step": 7270
    },
    {
      "epoch": 0.5697292221004852,
      "grad_norm": 4.4758734703063965,
      "learning_rate": 4.061175916723335e-05,
      "loss": 1.2278,
      "step": 7280
    },
    {
      "epoch": 0.570511817185788,
      "grad_norm": 1.8130991458892822,
      "learning_rate": 4.05986818015379e-05,
      "loss": 0.9242,
      "step": 7290
    },
    {
      "epoch": 0.5712944122710909,
      "grad_norm": 1.2587727308273315,
      "learning_rate": 4.058560443584245e-05,
      "loss": 0.5757,
      "step": 7300
    },
    {
      "epoch": 0.5720770073563938,
      "grad_norm": 5.881790637969971,
      "learning_rate": 4.057252707014699e-05,
      "loss": 0.6734,
      "step": 7310
    },
    {
      "epoch": 0.5728596024416966,
      "grad_norm": 5.996326446533203,
      "learning_rate": 4.0559449704451536e-05,
      "loss": 0.9887,
      "step": 7320
    },
    {
      "epoch": 0.5736421975269995,
      "grad_norm": 1.812027096748352,
      "learning_rate": 4.054637233875608e-05,
      "loss": 0.7092,
      "step": 7330
    },
    {
      "epoch": 0.5744247926123024,
      "grad_norm": 5.868380546569824,
      "learning_rate": 4.0533294973060625e-05,
      "loss": 0.9867,
      "step": 7340
    },
    {
      "epoch": 0.5752073876976053,
      "grad_norm": 1.1364696025848389,
      "learning_rate": 4.052021760736518e-05,
      "loss": 0.8679,
      "step": 7350
    },
    {
      "epoch": 0.5759899827829081,
      "grad_norm": 9.977632522583008,
      "learning_rate": 4.050714024166972e-05,
      "loss": 1.1375,
      "step": 7360
    },
    {
      "epoch": 0.576772577868211,
      "grad_norm": 2.1670806407928467,
      "learning_rate": 4.0494062875974266e-05,
      "loss": 0.8436,
      "step": 7370
    },
    {
      "epoch": 0.5775551729535139,
      "grad_norm": 2.8490488529205322,
      "learning_rate": 4.048098551027881e-05,
      "loss": 0.6112,
      "step": 7380
    },
    {
      "epoch": 0.5783377680388168,
      "grad_norm": 3.8483452796936035,
      "learning_rate": 4.0467908144583355e-05,
      "loss": 1.0292,
      "step": 7390
    },
    {
      "epoch": 0.5791203631241196,
      "grad_norm": 4.73075532913208,
      "learning_rate": 4.0454830778887906e-05,
      "loss": 0.8924,
      "step": 7400
    },
    {
      "epoch": 0.5799029582094224,
      "grad_norm": 6.014554023742676,
      "learning_rate": 4.0441753413192444e-05,
      "loss": 1.1257,
      "step": 7410
    },
    {
      "epoch": 0.5806855532947253,
      "grad_norm": 6.086801052093506,
      "learning_rate": 4.0428676047496995e-05,
      "loss": 1.19,
      "step": 7420
    },
    {
      "epoch": 0.5814681483800281,
      "grad_norm": 4.827360153198242,
      "learning_rate": 4.041559868180154e-05,
      "loss": 0.9972,
      "step": 7430
    },
    {
      "epoch": 0.582250743465331,
      "grad_norm": 5.746564865112305,
      "learning_rate": 4.0402521316106085e-05,
      "loss": 0.8571,
      "step": 7440
    },
    {
      "epoch": 0.5830333385506339,
      "grad_norm": 7.889925479888916,
      "learning_rate": 4.038944395041063e-05,
      "loss": 0.9606,
      "step": 7450
    },
    {
      "epoch": 0.5838159336359368,
      "grad_norm": 5.5271124839782715,
      "learning_rate": 4.0376366584715174e-05,
      "loss": 0.903,
      "step": 7460
    },
    {
      "epoch": 0.5845985287212396,
      "grad_norm": 3.415358304977417,
      "learning_rate": 4.0363289219019725e-05,
      "loss": 0.7725,
      "step": 7470
    },
    {
      "epoch": 0.5853811238065425,
      "grad_norm": 2.46823787689209,
      "learning_rate": 4.035021185332426e-05,
      "loss": 0.4273,
      "step": 7480
    },
    {
      "epoch": 0.5861637188918454,
      "grad_norm": 8.98774242401123,
      "learning_rate": 4.0337134487628814e-05,
      "loss": 0.8989,
      "step": 7490
    },
    {
      "epoch": 0.5869463139771482,
      "grad_norm": 1.5800225734710693,
      "learning_rate": 4.032405712193336e-05,
      "loss": 0.8059,
      "step": 7500
    },
    {
      "epoch": 0.5877289090624511,
      "grad_norm": 2.008143663406372,
      "learning_rate": 4.0310979756237903e-05,
      "loss": 1.0188,
      "step": 7510
    },
    {
      "epoch": 0.588511504147754,
      "grad_norm": 4.741922378540039,
      "learning_rate": 4.0297902390542455e-05,
      "loss": 0.8306,
      "step": 7520
    },
    {
      "epoch": 0.5892940992330569,
      "grad_norm": 11.266371726989746,
      "learning_rate": 4.028482502484699e-05,
      "loss": 0.9082,
      "step": 7530
    },
    {
      "epoch": 0.5900766943183596,
      "grad_norm": 6.009096145629883,
      "learning_rate": 4.0271747659151544e-05,
      "loss": 1.0794,
      "step": 7540
    },
    {
      "epoch": 0.5908592894036625,
      "grad_norm": 8.112404823303223,
      "learning_rate": 4.025867029345609e-05,
      "loss": 1.1258,
      "step": 7550
    },
    {
      "epoch": 0.5916418844889654,
      "grad_norm": 7.974365234375,
      "learning_rate": 4.024559292776063e-05,
      "loss": 0.7961,
      "step": 7560
    },
    {
      "epoch": 0.5924244795742682,
      "grad_norm": 5.3536295890808105,
      "learning_rate": 4.0232515562065184e-05,
      "loss": 0.7742,
      "step": 7570
    },
    {
      "epoch": 0.5932070746595711,
      "grad_norm": 7.270729064941406,
      "learning_rate": 4.021943819636972e-05,
      "loss": 0.842,
      "step": 7580
    },
    {
      "epoch": 0.593989669744874,
      "grad_norm": 9.920305252075195,
      "learning_rate": 4.0206360830674274e-05,
      "loss": 0.9887,
      "step": 7590
    },
    {
      "epoch": 0.5947722648301769,
      "grad_norm": 9.178397178649902,
      "learning_rate": 4.019328346497882e-05,
      "loss": 0.7124,
      "step": 7600
    },
    {
      "epoch": 0.5955548599154797,
      "grad_norm": 5.195706367492676,
      "learning_rate": 4.018020609928336e-05,
      "loss": 0.8778,
      "step": 7610
    },
    {
      "epoch": 0.5963374550007826,
      "grad_norm": 5.53049373626709,
      "learning_rate": 4.016712873358791e-05,
      "loss": 0.72,
      "step": 7620
    },
    {
      "epoch": 0.5971200500860855,
      "grad_norm": 10.406530380249023,
      "learning_rate": 4.015405136789245e-05,
      "loss": 0.8835,
      "step": 7630
    },
    {
      "epoch": 0.5979026451713884,
      "grad_norm": 2.3972623348236084,
      "learning_rate": 4.0140974002197e-05,
      "loss": 0.61,
      "step": 7640
    },
    {
      "epoch": 0.5986852402566912,
      "grad_norm": 3.2227351665496826,
      "learning_rate": 4.012789663650154e-05,
      "loss": 0.658,
      "step": 7650
    },
    {
      "epoch": 0.5994678353419941,
      "grad_norm": 8.290729522705078,
      "learning_rate": 4.011481927080609e-05,
      "loss": 0.9089,
      "step": 7660
    },
    {
      "epoch": 0.6002504304272969,
      "grad_norm": 8.77101993560791,
      "learning_rate": 4.010174190511064e-05,
      "loss": 0.7607,
      "step": 7670
    },
    {
      "epoch": 0.6010330255125997,
      "grad_norm": 4.051955699920654,
      "learning_rate": 4.008866453941518e-05,
      "loss": 1.3316,
      "step": 7680
    },
    {
      "epoch": 0.6018156205979026,
      "grad_norm": 4.124317646026611,
      "learning_rate": 4.0075587173719726e-05,
      "loss": 0.8881,
      "step": 7690
    },
    {
      "epoch": 0.6025982156832055,
      "grad_norm": 14.274579048156738,
      "learning_rate": 4.006250980802427e-05,
      "loss": 1.0219,
      "step": 7700
    },
    {
      "epoch": 0.6033808107685084,
      "grad_norm": 7.567116737365723,
      "learning_rate": 4.004943244232882e-05,
      "loss": 0.5201,
      "step": 7710
    },
    {
      "epoch": 0.6041634058538112,
      "grad_norm": 13.264985084533691,
      "learning_rate": 4.0036355076633367e-05,
      "loss": 0.8217,
      "step": 7720
    },
    {
      "epoch": 0.6049460009391141,
      "grad_norm": 7.211345672607422,
      "learning_rate": 4.002327771093791e-05,
      "loss": 0.8324,
      "step": 7730
    },
    {
      "epoch": 0.605728596024417,
      "grad_norm": 2.173584461212158,
      "learning_rate": 4.0010200345242456e-05,
      "loss": 0.6128,
      "step": 7740
    },
    {
      "epoch": 0.6065111911097198,
      "grad_norm": 2.842360496520996,
      "learning_rate": 3.9997122979547e-05,
      "loss": 0.703,
      "step": 7750
    },
    {
      "epoch": 0.6072937861950227,
      "grad_norm": 4.671855449676514,
      "learning_rate": 3.9984045613851545e-05,
      "loss": 1.2442,
      "step": 7760
    },
    {
      "epoch": 0.6080763812803256,
      "grad_norm": 3.841827154159546,
      "learning_rate": 3.9970968248156096e-05,
      "loss": 1.3196,
      "step": 7770
    },
    {
      "epoch": 0.6088589763656285,
      "grad_norm": 15.082938194274902,
      "learning_rate": 3.995789088246064e-05,
      "loss": 0.7336,
      "step": 7780
    },
    {
      "epoch": 0.6096415714509313,
      "grad_norm": 20.764631271362305,
      "learning_rate": 3.9944813516765185e-05,
      "loss": 0.8225,
      "step": 7790
    },
    {
      "epoch": 0.6104241665362341,
      "grad_norm": 6.369302272796631,
      "learning_rate": 3.993173615106973e-05,
      "loss": 1.0145,
      "step": 7800
    },
    {
      "epoch": 0.611206761621537,
      "grad_norm": 7.531883239746094,
      "learning_rate": 3.9918658785374274e-05,
      "loss": 0.8714,
      "step": 7810
    },
    {
      "epoch": 0.6119893567068398,
      "grad_norm": 4.655513286590576,
      "learning_rate": 3.9905581419678826e-05,
      "loss": 0.8423,
      "step": 7820
    },
    {
      "epoch": 0.6127719517921427,
      "grad_norm": 4.002667427062988,
      "learning_rate": 3.9892504053983364e-05,
      "loss": 0.9085,
      "step": 7830
    },
    {
      "epoch": 0.6135545468774456,
      "grad_norm": 3.6337437629699707,
      "learning_rate": 3.9879426688287915e-05,
      "loss": 0.8629,
      "step": 7840
    },
    {
      "epoch": 0.6143371419627485,
      "grad_norm": 5.2091474533081055,
      "learning_rate": 3.986634932259246e-05,
      "loss": 0.8377,
      "step": 7850
    },
    {
      "epoch": 0.6151197370480513,
      "grad_norm": 7.961836814880371,
      "learning_rate": 3.9853271956897004e-05,
      "loss": 0.9662,
      "step": 7860
    },
    {
      "epoch": 0.6159023321333542,
      "grad_norm": 5.6108078956604,
      "learning_rate": 3.984019459120155e-05,
      "loss": 0.766,
      "step": 7870
    },
    {
      "epoch": 0.6166849272186571,
      "grad_norm": 6.1843037605285645,
      "learning_rate": 3.982711722550609e-05,
      "loss": 0.7331,
      "step": 7880
    },
    {
      "epoch": 0.61746752230396,
      "grad_norm": 4.770885467529297,
      "learning_rate": 3.9814039859810645e-05,
      "loss": 1.0845,
      "step": 7890
    },
    {
      "epoch": 0.6182501173892628,
      "grad_norm": 7.661377429962158,
      "learning_rate": 3.980096249411518e-05,
      "loss": 1.0498,
      "step": 7900
    },
    {
      "epoch": 0.6190327124745657,
      "grad_norm": 5.5640549659729,
      "learning_rate": 3.9787885128419734e-05,
      "loss": 0.8036,
      "step": 7910
    },
    {
      "epoch": 0.6198153075598686,
      "grad_norm": 3.8442351818084717,
      "learning_rate": 3.977480776272428e-05,
      "loss": 1.3385,
      "step": 7920
    },
    {
      "epoch": 0.6205979026451713,
      "grad_norm": 3.0587193965911865,
      "learning_rate": 3.976173039702882e-05,
      "loss": 0.9,
      "step": 7930
    },
    {
      "epoch": 0.6213804977304742,
      "grad_norm": 3.4369826316833496,
      "learning_rate": 3.9748653031333374e-05,
      "loss": 0.5814,
      "step": 7940
    },
    {
      "epoch": 0.6221630928157771,
      "grad_norm": 8.10569953918457,
      "learning_rate": 3.973557566563791e-05,
      "loss": 0.5707,
      "step": 7950
    },
    {
      "epoch": 0.62294568790108,
      "grad_norm": 3.8327856063842773,
      "learning_rate": 3.972249829994246e-05,
      "loss": 0.9111,
      "step": 7960
    },
    {
      "epoch": 0.6237282829863828,
      "grad_norm": 4.4550676345825195,
      "learning_rate": 3.970942093424701e-05,
      "loss": 1.1978,
      "step": 7970
    },
    {
      "epoch": 0.6245108780716857,
      "grad_norm": 3.88533878326416,
      "learning_rate": 3.969634356855155e-05,
      "loss": 0.7118,
      "step": 7980
    },
    {
      "epoch": 0.6252934731569886,
      "grad_norm": 4.269775390625,
      "learning_rate": 3.9683266202856104e-05,
      "loss": 0.8105,
      "step": 7990
    },
    {
      "epoch": 0.6260760682422915,
      "grad_norm": 7.454277515411377,
      "learning_rate": 3.967018883716064e-05,
      "loss": 0.6156,
      "step": 8000
    },
    {
      "epoch": 0.6268586633275943,
      "grad_norm": 8.546407699584961,
      "learning_rate": 3.965711147146519e-05,
      "loss": 0.9779,
      "step": 8010
    },
    {
      "epoch": 0.6276412584128972,
      "grad_norm": 8.373024940490723,
      "learning_rate": 3.964403410576973e-05,
      "loss": 1.0092,
      "step": 8020
    },
    {
      "epoch": 0.6284238534982001,
      "grad_norm": 7.303249835968018,
      "learning_rate": 3.963095674007428e-05,
      "loss": 0.8831,
      "step": 8030
    },
    {
      "epoch": 0.6292064485835029,
      "grad_norm": 12.274666786193848,
      "learning_rate": 3.961787937437883e-05,
      "loss": 0.9299,
      "step": 8040
    },
    {
      "epoch": 0.6299890436688058,
      "grad_norm": 6.561683177947998,
      "learning_rate": 3.960480200868337e-05,
      "loss": 1.0307,
      "step": 8050
    },
    {
      "epoch": 0.6307716387541086,
      "grad_norm": 6.4463629722595215,
      "learning_rate": 3.959172464298792e-05,
      "loss": 1.0049,
      "step": 8060
    },
    {
      "epoch": 0.6315542338394114,
      "grad_norm": 12.442595481872559,
      "learning_rate": 3.957864727729246e-05,
      "loss": 0.8747,
      "step": 8070
    },
    {
      "epoch": 0.6323368289247143,
      "grad_norm": 4.198892116546631,
      "learning_rate": 3.956556991159701e-05,
      "loss": 0.7991,
      "step": 8080
    },
    {
      "epoch": 0.6331194240100172,
      "grad_norm": 5.992439270019531,
      "learning_rate": 3.9552492545901556e-05,
      "loss": 0.6455,
      "step": 8090
    },
    {
      "epoch": 0.6339020190953201,
      "grad_norm": 4.231416702270508,
      "learning_rate": 3.95394151802061e-05,
      "loss": 0.8993,
      "step": 8100
    },
    {
      "epoch": 0.6346846141806229,
      "grad_norm": 7.246849060058594,
      "learning_rate": 3.9526337814510645e-05,
      "loss": 0.6839,
      "step": 8110
    },
    {
      "epoch": 0.6354672092659258,
      "grad_norm": 5.8323163986206055,
      "learning_rate": 3.951326044881519e-05,
      "loss": 0.7103,
      "step": 8120
    },
    {
      "epoch": 0.6362498043512287,
      "grad_norm": 7.429286003112793,
      "learning_rate": 3.950018308311974e-05,
      "loss": 0.9718,
      "step": 8130
    },
    {
      "epoch": 0.6370323994365316,
      "grad_norm": 8.574079513549805,
      "learning_rate": 3.9487105717424286e-05,
      "loss": 0.6747,
      "step": 8140
    },
    {
      "epoch": 0.6378149945218344,
      "grad_norm": 37.70345687866211,
      "learning_rate": 3.947402835172883e-05,
      "loss": 0.9549,
      "step": 8150
    },
    {
      "epoch": 0.6385975896071373,
      "grad_norm": 1.2692748308181763,
      "learning_rate": 3.9460950986033375e-05,
      "loss": 0.6274,
      "step": 8160
    },
    {
      "epoch": 0.6393801846924402,
      "grad_norm": 5.124848365783691,
      "learning_rate": 3.944787362033792e-05,
      "loss": 0.7259,
      "step": 8170
    },
    {
      "epoch": 0.6401627797777429,
      "grad_norm": 4.178142547607422,
      "learning_rate": 3.9434796254642464e-05,
      "loss": 0.7535,
      "step": 8180
    },
    {
      "epoch": 0.6409453748630458,
      "grad_norm": 17.53697395324707,
      "learning_rate": 3.9421718888947016e-05,
      "loss": 0.9198,
      "step": 8190
    },
    {
      "epoch": 0.6417279699483487,
      "grad_norm": 7.360964775085449,
      "learning_rate": 3.940864152325156e-05,
      "loss": 0.7657,
      "step": 8200
    },
    {
      "epoch": 0.6425105650336516,
      "grad_norm": 10.883098602294922,
      "learning_rate": 3.9395564157556105e-05,
      "loss": 1.0551,
      "step": 8210
    },
    {
      "epoch": 0.6432931601189544,
      "grad_norm": 9.776481628417969,
      "learning_rate": 3.938248679186065e-05,
      "loss": 0.6848,
      "step": 8220
    },
    {
      "epoch": 0.6440757552042573,
      "grad_norm": 2.9045660495758057,
      "learning_rate": 3.9369409426165194e-05,
      "loss": 0.7699,
      "step": 8230
    },
    {
      "epoch": 0.6448583502895602,
      "grad_norm": 3.065711498260498,
      "learning_rate": 3.935633206046974e-05,
      "loss": 0.9373,
      "step": 8240
    },
    {
      "epoch": 0.645640945374863,
      "grad_norm": 10.685598373413086,
      "learning_rate": 3.934325469477428e-05,
      "loss": 1.086,
      "step": 8250
    },
    {
      "epoch": 0.6464235404601659,
      "grad_norm": 10.905689239501953,
      "learning_rate": 3.9330177329078834e-05,
      "loss": 0.8322,
      "step": 8260
    },
    {
      "epoch": 0.6472061355454688,
      "grad_norm": 1.8690558671951294,
      "learning_rate": 3.931709996338338e-05,
      "loss": 0.8125,
      "step": 8270
    },
    {
      "epoch": 0.6479887306307717,
      "grad_norm": 8.079241752624512,
      "learning_rate": 3.9304022597687924e-05,
      "loss": 1.031,
      "step": 8280
    },
    {
      "epoch": 0.6487713257160745,
      "grad_norm": 2.5291178226470947,
      "learning_rate": 3.929094523199247e-05,
      "loss": 0.997,
      "step": 8290
    },
    {
      "epoch": 0.6495539208013774,
      "grad_norm": 9.047258377075195,
      "learning_rate": 3.927786786629701e-05,
      "loss": 0.8552,
      "step": 8300
    },
    {
      "epoch": 0.6503365158866802,
      "grad_norm": 5.699478626251221,
      "learning_rate": 3.9264790500601564e-05,
      "loss": 0.6647,
      "step": 8310
    },
    {
      "epoch": 0.651119110971983,
      "grad_norm": 4.0365800857543945,
      "learning_rate": 3.92517131349061e-05,
      "loss": 0.892,
      "step": 8320
    },
    {
      "epoch": 0.6519017060572859,
      "grad_norm": 15.259045600891113,
      "learning_rate": 3.923863576921065e-05,
      "loss": 0.6862,
      "step": 8330
    },
    {
      "epoch": 0.6526843011425888,
      "grad_norm": 10.10246753692627,
      "learning_rate": 3.92255584035152e-05,
      "loss": 1.0159,
      "step": 8340
    },
    {
      "epoch": 0.6534668962278917,
      "grad_norm": 4.875640392303467,
      "learning_rate": 3.921248103781974e-05,
      "loss": 0.8815,
      "step": 8350
    },
    {
      "epoch": 0.6542494913131945,
      "grad_norm": 3.466360330581665,
      "learning_rate": 3.9199403672124294e-05,
      "loss": 0.642,
      "step": 8360
    },
    {
      "epoch": 0.6550320863984974,
      "grad_norm": 5.00355863571167,
      "learning_rate": 3.918632630642883e-05,
      "loss": 0.9294,
      "step": 8370
    },
    {
      "epoch": 0.6558146814838003,
      "grad_norm": 7.477866172790527,
      "learning_rate": 3.917324894073338e-05,
      "loss": 0.8317,
      "step": 8380
    },
    {
      "epoch": 0.6565972765691032,
      "grad_norm": 5.42965030670166,
      "learning_rate": 3.916017157503792e-05,
      "loss": 0.8093,
      "step": 8390
    },
    {
      "epoch": 0.657379871654406,
      "grad_norm": 2.232001304626465,
      "learning_rate": 3.914709420934247e-05,
      "loss": 0.7553,
      "step": 8400
    },
    {
      "epoch": 0.6581624667397089,
      "grad_norm": 3.3870415687561035,
      "learning_rate": 3.913401684364702e-05,
      "loss": 0.8512,
      "step": 8410
    },
    {
      "epoch": 0.6589450618250118,
      "grad_norm": 3.486584186553955,
      "learning_rate": 3.912093947795156e-05,
      "loss": 0.6386,
      "step": 8420
    },
    {
      "epoch": 0.6597276569103147,
      "grad_norm": 1.318729281425476,
      "learning_rate": 3.910786211225611e-05,
      "loss": 0.7513,
      "step": 8430
    },
    {
      "epoch": 0.6605102519956174,
      "grad_norm": 4.359373569488525,
      "learning_rate": 3.909478474656065e-05,
      "loss": 0.7276,
      "step": 8440
    },
    {
      "epoch": 0.6612928470809203,
      "grad_norm": 4.5334672927856445,
      "learning_rate": 3.90817073808652e-05,
      "loss": 0.9675,
      "step": 8450
    },
    {
      "epoch": 0.6620754421662232,
      "grad_norm": 3.5145533084869385,
      "learning_rate": 3.9068630015169746e-05,
      "loss": 0.9209,
      "step": 8460
    },
    {
      "epoch": 0.662858037251526,
      "grad_norm": 6.361563682556152,
      "learning_rate": 3.905555264947429e-05,
      "loss": 1.1597,
      "step": 8470
    },
    {
      "epoch": 0.6636406323368289,
      "grad_norm": 5.506303310394287,
      "learning_rate": 3.904247528377884e-05,
      "loss": 0.7539,
      "step": 8480
    },
    {
      "epoch": 0.6644232274221318,
      "grad_norm": 4.357978343963623,
      "learning_rate": 3.902939791808338e-05,
      "loss": 0.9237,
      "step": 8490
    },
    {
      "epoch": 0.6652058225074347,
      "grad_norm": 4.345588207244873,
      "learning_rate": 3.901632055238793e-05,
      "loss": 0.9312,
      "step": 8500
    },
    {
      "epoch": 0.6659884175927375,
      "grad_norm": 5.054117202758789,
      "learning_rate": 3.9003243186692476e-05,
      "loss": 1.1273,
      "step": 8510
    },
    {
      "epoch": 0.6667710126780404,
      "grad_norm": 4.986987590789795,
      "learning_rate": 3.899016582099702e-05,
      "loss": 0.8362,
      "step": 8520
    },
    {
      "epoch": 0.6675536077633433,
      "grad_norm": 4.389319896697998,
      "learning_rate": 3.8977088455301565e-05,
      "loss": 0.8538,
      "step": 8530
    },
    {
      "epoch": 0.6683362028486461,
      "grad_norm": 5.841724872589111,
      "learning_rate": 3.896401108960611e-05,
      "loss": 0.583,
      "step": 8540
    },
    {
      "epoch": 0.669118797933949,
      "grad_norm": 6.915998935699463,
      "learning_rate": 3.895093372391066e-05,
      "loss": 0.5487,
      "step": 8550
    },
    {
      "epoch": 0.6699013930192519,
      "grad_norm": 30.626752853393555,
      "learning_rate": 3.8937856358215205e-05,
      "loss": 0.8502,
      "step": 8560
    },
    {
      "epoch": 0.6706839881045547,
      "grad_norm": 15.640740394592285,
      "learning_rate": 3.892477899251975e-05,
      "loss": 1.1156,
      "step": 8570
    },
    {
      "epoch": 0.6714665831898575,
      "grad_norm": 8.39625072479248,
      "learning_rate": 3.8911701626824295e-05,
      "loss": 1.149,
      "step": 8580
    },
    {
      "epoch": 0.6722491782751604,
      "grad_norm": 11.482784271240234,
      "learning_rate": 3.889862426112884e-05,
      "loss": 0.8417,
      "step": 8590
    },
    {
      "epoch": 0.6730317733604633,
      "grad_norm": 7.895070552825928,
      "learning_rate": 3.8885546895433384e-05,
      "loss": 0.7574,
      "step": 8600
    },
    {
      "epoch": 0.6738143684457661,
      "grad_norm": 3.285834789276123,
      "learning_rate": 3.887246952973793e-05,
      "loss": 1.0415,
      "step": 8610
    },
    {
      "epoch": 0.674596963531069,
      "grad_norm": 1.252188801765442,
      "learning_rate": 3.885939216404248e-05,
      "loss": 0.7518,
      "step": 8620
    },
    {
      "epoch": 0.6753795586163719,
      "grad_norm": 3.9072813987731934,
      "learning_rate": 3.8846314798347024e-05,
      "loss": 0.9285,
      "step": 8630
    },
    {
      "epoch": 0.6761621537016748,
      "grad_norm": 13.482275009155273,
      "learning_rate": 3.883323743265157e-05,
      "loss": 0.9774,
      "step": 8640
    },
    {
      "epoch": 0.6769447487869776,
      "grad_norm": 3.7827744483947754,
      "learning_rate": 3.882016006695611e-05,
      "loss": 1.0574,
      "step": 8650
    },
    {
      "epoch": 0.6777273438722805,
      "grad_norm": 5.040523529052734,
      "learning_rate": 3.880708270126066e-05,
      "loss": 0.8951,
      "step": 8660
    },
    {
      "epoch": 0.6785099389575834,
      "grad_norm": 8.14504623413086,
      "learning_rate": 3.87940053355652e-05,
      "loss": 1.2289,
      "step": 8670
    },
    {
      "epoch": 0.6792925340428863,
      "grad_norm": 6.661283493041992,
      "learning_rate": 3.8780927969869754e-05,
      "loss": 0.6626,
      "step": 8680
    },
    {
      "epoch": 0.6800751291281891,
      "grad_norm": 7.322127342224121,
      "learning_rate": 3.87678506041743e-05,
      "loss": 0.7105,
      "step": 8690
    },
    {
      "epoch": 0.6808577242134919,
      "grad_norm": 3.489060640335083,
      "learning_rate": 3.875477323847884e-05,
      "loss": 0.9561,
      "step": 8700
    },
    {
      "epoch": 0.6816403192987948,
      "grad_norm": 1.2218551635742188,
      "learning_rate": 3.874169587278339e-05,
      "loss": 0.7331,
      "step": 8710
    },
    {
      "epoch": 0.6824229143840976,
      "grad_norm": 5.532180309295654,
      "learning_rate": 3.872861850708793e-05,
      "loss": 0.6582,
      "step": 8720
    },
    {
      "epoch": 0.6832055094694005,
      "grad_norm": 2.4819753170013428,
      "learning_rate": 3.8715541141392484e-05,
      "loss": 0.9414,
      "step": 8730
    },
    {
      "epoch": 0.6839881045547034,
      "grad_norm": 5.580556869506836,
      "learning_rate": 3.870246377569702e-05,
      "loss": 0.8151,
      "step": 8740
    },
    {
      "epoch": 0.6847706996400063,
      "grad_norm": 1.4750864505767822,
      "learning_rate": 3.868938641000157e-05,
      "loss": 0.723,
      "step": 8750
    },
    {
      "epoch": 0.6855532947253091,
      "grad_norm": 12.116117477416992,
      "learning_rate": 3.867630904430612e-05,
      "loss": 1.0143,
      "step": 8760
    },
    {
      "epoch": 0.686335889810612,
      "grad_norm": 5.472472190856934,
      "learning_rate": 3.866323167861066e-05,
      "loss": 0.7797,
      "step": 8770
    },
    {
      "epoch": 0.6871184848959149,
      "grad_norm": 2.2424354553222656,
      "learning_rate": 3.865015431291521e-05,
      "loss": 0.8306,
      "step": 8780
    },
    {
      "epoch": 0.6879010799812177,
      "grad_norm": 4.965039253234863,
      "learning_rate": 3.863707694721975e-05,
      "loss": 0.5525,
      "step": 8790
    },
    {
      "epoch": 0.6886836750665206,
      "grad_norm": 3.9194388389587402,
      "learning_rate": 3.86239995815243e-05,
      "loss": 0.7412,
      "step": 8800
    },
    {
      "epoch": 0.6894662701518235,
      "grad_norm": 18.769512176513672,
      "learning_rate": 3.861092221582884e-05,
      "loss": 1.2767,
      "step": 8810
    },
    {
      "epoch": 0.6902488652371264,
      "grad_norm": 3.259730815887451,
      "learning_rate": 3.859784485013339e-05,
      "loss": 0.9105,
      "step": 8820
    },
    {
      "epoch": 0.6910314603224291,
      "grad_norm": 5.308115482330322,
      "learning_rate": 3.8584767484437936e-05,
      "loss": 0.9216,
      "step": 8830
    },
    {
      "epoch": 0.691814055407732,
      "grad_norm": 15.051186561584473,
      "learning_rate": 3.857169011874248e-05,
      "loss": 0.7102,
      "step": 8840
    },
    {
      "epoch": 0.6925966504930349,
      "grad_norm": 1.050187110900879,
      "learning_rate": 3.855861275304703e-05,
      "loss": 0.9872,
      "step": 8850
    },
    {
      "epoch": 0.6933792455783377,
      "grad_norm": 1.2644296884536743,
      "learning_rate": 3.854553538735157e-05,
      "loss": 0.7608,
      "step": 8860
    },
    {
      "epoch": 0.6941618406636406,
      "grad_norm": 12.766301155090332,
      "learning_rate": 3.853245802165612e-05,
      "loss": 0.5484,
      "step": 8870
    },
    {
      "epoch": 0.6949444357489435,
      "grad_norm": 12.443408012390137,
      "learning_rate": 3.8519380655960666e-05,
      "loss": 0.5641,
      "step": 8880
    },
    {
      "epoch": 0.6957270308342464,
      "grad_norm": 15.832154273986816,
      "learning_rate": 3.850630329026521e-05,
      "loss": 0.994,
      "step": 8890
    },
    {
      "epoch": 0.6965096259195492,
      "grad_norm": 4.372637748718262,
      "learning_rate": 3.849322592456976e-05,
      "loss": 0.5101,
      "step": 8900
    },
    {
      "epoch": 0.6972922210048521,
      "grad_norm": 0.7082036733627319,
      "learning_rate": 3.84801485588743e-05,
      "loss": 1.3531,
      "step": 8910
    },
    {
      "epoch": 0.698074816090155,
      "grad_norm": 4.521925449371338,
      "learning_rate": 3.846707119317885e-05,
      "loss": 0.6277,
      "step": 8920
    },
    {
      "epoch": 0.6988574111754579,
      "grad_norm": 4.903423309326172,
      "learning_rate": 3.8453993827483395e-05,
      "loss": 0.9935,
      "step": 8930
    },
    {
      "epoch": 0.6996400062607607,
      "grad_norm": 12.721101760864258,
      "learning_rate": 3.844091646178794e-05,
      "loss": 1.127,
      "step": 8940
    },
    {
      "epoch": 0.7004226013460636,
      "grad_norm": 26.866106033325195,
      "learning_rate": 3.8427839096092484e-05,
      "loss": 1.1423,
      "step": 8950
    },
    {
      "epoch": 0.7012051964313664,
      "grad_norm": 4.475317478179932,
      "learning_rate": 3.841476173039703e-05,
      "loss": 0.7251,
      "step": 8960
    },
    {
      "epoch": 0.7019877915166692,
      "grad_norm": 14.420462608337402,
      "learning_rate": 3.840168436470158e-05,
      "loss": 1.0441,
      "step": 8970
    },
    {
      "epoch": 0.7027703866019721,
      "grad_norm": 4.142608642578125,
      "learning_rate": 3.838860699900612e-05,
      "loss": 1.0354,
      "step": 8980
    },
    {
      "epoch": 0.703552981687275,
      "grad_norm": 6.299661159515381,
      "learning_rate": 3.837552963331067e-05,
      "loss": 0.7773,
      "step": 8990
    },
    {
      "epoch": 0.7043355767725779,
      "grad_norm": 7.491670608520508,
      "learning_rate": 3.8362452267615214e-05,
      "loss": 1.1789,
      "step": 9000
    },
    {
      "epoch": 0.7051181718578807,
      "grad_norm": 4.930359363555908,
      "learning_rate": 3.834937490191976e-05,
      "loss": 0.9229,
      "step": 9010
    },
    {
      "epoch": 0.7059007669431836,
      "grad_norm": 3.3752729892730713,
      "learning_rate": 3.83362975362243e-05,
      "loss": 0.9724,
      "step": 9020
    },
    {
      "epoch": 0.7066833620284865,
      "grad_norm": 13.789896011352539,
      "learning_rate": 3.832322017052885e-05,
      "loss": 0.9148,
      "step": 9030
    },
    {
      "epoch": 0.7074659571137893,
      "grad_norm": 6.980893135070801,
      "learning_rate": 3.83101428048334e-05,
      "loss": 0.6447,
      "step": 9040
    },
    {
      "epoch": 0.7082485521990922,
      "grad_norm": 4.548523902893066,
      "learning_rate": 3.8297065439137944e-05,
      "loss": 0.8022,
      "step": 9050
    },
    {
      "epoch": 0.7090311472843951,
      "grad_norm": 7.349812030792236,
      "learning_rate": 3.828398807344249e-05,
      "loss": 0.8728,
      "step": 9060
    },
    {
      "epoch": 0.709813742369698,
      "grad_norm": 3.2322158813476562,
      "learning_rate": 3.827091070774703e-05,
      "loss": 0.5906,
      "step": 9070
    },
    {
      "epoch": 0.7105963374550008,
      "grad_norm": 1.0035629272460938,
      "learning_rate": 3.825783334205158e-05,
      "loss": 0.7483,
      "step": 9080
    },
    {
      "epoch": 0.7113789325403036,
      "grad_norm": 2.894554853439331,
      "learning_rate": 3.824475597635612e-05,
      "loss": 1.0506,
      "step": 9090
    },
    {
      "epoch": 0.7121615276256065,
      "grad_norm": 7.798361778259277,
      "learning_rate": 3.823167861066067e-05,
      "loss": 0.6554,
      "step": 9100
    },
    {
      "epoch": 0.7129441227109093,
      "grad_norm": 1.0737498998641968,
      "learning_rate": 3.821860124496522e-05,
      "loss": 0.8849,
      "step": 9110
    },
    {
      "epoch": 0.7137267177962122,
      "grad_norm": 9.232399940490723,
      "learning_rate": 3.820552387926976e-05,
      "loss": 0.8448,
      "step": 9120
    },
    {
      "epoch": 0.7145093128815151,
      "grad_norm": 10.180237770080566,
      "learning_rate": 3.819244651357431e-05,
      "loss": 0.804,
      "step": 9130
    },
    {
      "epoch": 0.715291907966818,
      "grad_norm": 5.827654838562012,
      "learning_rate": 3.817936914787885e-05,
      "loss": 1.0289,
      "step": 9140
    },
    {
      "epoch": 0.7160745030521208,
      "grad_norm": 4.43319845199585,
      "learning_rate": 3.81662917821834e-05,
      "loss": 0.6195,
      "step": 9150
    },
    {
      "epoch": 0.7168570981374237,
      "grad_norm": 1.7918765544891357,
      "learning_rate": 3.815321441648794e-05,
      "loss": 0.7195,
      "step": 9160
    },
    {
      "epoch": 0.7176396932227266,
      "grad_norm": 3.8808138370513916,
      "learning_rate": 3.814013705079249e-05,
      "loss": 0.5104,
      "step": 9170
    },
    {
      "epoch": 0.7184222883080295,
      "grad_norm": 1.1242902278900146,
      "learning_rate": 3.812705968509704e-05,
      "loss": 0.5629,
      "step": 9180
    },
    {
      "epoch": 0.7192048833933323,
      "grad_norm": 10.47758674621582,
      "learning_rate": 3.811398231940158e-05,
      "loss": 1.1876,
      "step": 9190
    },
    {
      "epoch": 0.7199874784786352,
      "grad_norm": 7.005457878112793,
      "learning_rate": 3.8100904953706126e-05,
      "loss": 0.6841,
      "step": 9200
    },
    {
      "epoch": 0.720770073563938,
      "grad_norm": 27.591657638549805,
      "learning_rate": 3.808782758801067e-05,
      "loss": 1.1133,
      "step": 9210
    },
    {
      "epoch": 0.7215526686492408,
      "grad_norm": 5.632648944854736,
      "learning_rate": 3.807475022231522e-05,
      "loss": 0.7802,
      "step": 9220
    },
    {
      "epoch": 0.7223352637345437,
      "grad_norm": 4.281847953796387,
      "learning_rate": 3.806167285661976e-05,
      "loss": 0.8675,
      "step": 9230
    },
    {
      "epoch": 0.7231178588198466,
      "grad_norm": 7.942632675170898,
      "learning_rate": 3.804859549092431e-05,
      "loss": 0.618,
      "step": 9240
    },
    {
      "epoch": 0.7239004539051495,
      "grad_norm": 8.70691204071045,
      "learning_rate": 3.8035518125228855e-05,
      "loss": 0.6268,
      "step": 9250
    },
    {
      "epoch": 0.7246830489904523,
      "grad_norm": 1.7314077615737915,
      "learning_rate": 3.80224407595334e-05,
      "loss": 0.5347,
      "step": 9260
    },
    {
      "epoch": 0.7254656440757552,
      "grad_norm": 4.125974178314209,
      "learning_rate": 3.800936339383795e-05,
      "loss": 0.8899,
      "step": 9270
    },
    {
      "epoch": 0.7262482391610581,
      "grad_norm": 6.036971569061279,
      "learning_rate": 3.799628602814249e-05,
      "loss": 1.1183,
      "step": 9280
    },
    {
      "epoch": 0.727030834246361,
      "grad_norm": 6.218247413635254,
      "learning_rate": 3.798320866244704e-05,
      "loss": 1.0016,
      "step": 9290
    },
    {
      "epoch": 0.7278134293316638,
      "grad_norm": 7.430245876312256,
      "learning_rate": 3.7970131296751585e-05,
      "loss": 0.831,
      "step": 9300
    },
    {
      "epoch": 0.7285960244169667,
      "grad_norm": 8.069963455200195,
      "learning_rate": 3.795705393105613e-05,
      "loss": 0.9822,
      "step": 9310
    },
    {
      "epoch": 0.7293786195022696,
      "grad_norm": 10.461420059204102,
      "learning_rate": 3.794397656536068e-05,
      "loss": 0.759,
      "step": 9320
    },
    {
      "epoch": 0.7301612145875724,
      "grad_norm": 4.242290019989014,
      "learning_rate": 3.793089919966522e-05,
      "loss": 0.9689,
      "step": 9330
    },
    {
      "epoch": 0.7309438096728752,
      "grad_norm": 5.22723913192749,
      "learning_rate": 3.791782183396977e-05,
      "loss": 0.9139,
      "step": 9340
    },
    {
      "epoch": 0.7317264047581781,
      "grad_norm": 8.378803253173828,
      "learning_rate": 3.790474446827431e-05,
      "loss": 0.7636,
      "step": 9350
    },
    {
      "epoch": 0.732508999843481,
      "grad_norm": 5.3998565673828125,
      "learning_rate": 3.789166710257886e-05,
      "loss": 0.8347,
      "step": 9360
    },
    {
      "epoch": 0.7332915949287838,
      "grad_norm": 7.94550895690918,
      "learning_rate": 3.7878589736883404e-05,
      "loss": 0.9406,
      "step": 9370
    },
    {
      "epoch": 0.7340741900140867,
      "grad_norm": 9.198704719543457,
      "learning_rate": 3.786551237118795e-05,
      "loss": 0.8072,
      "step": 9380
    },
    {
      "epoch": 0.7348567850993896,
      "grad_norm": 8.282915115356445,
      "learning_rate": 3.78524350054925e-05,
      "loss": 0.8997,
      "step": 9390
    },
    {
      "epoch": 0.7356393801846924,
      "grad_norm": 11.969487190246582,
      "learning_rate": 3.783935763979704e-05,
      "loss": 0.8958,
      "step": 9400
    },
    {
      "epoch": 0.7364219752699953,
      "grad_norm": 3.8165900707244873,
      "learning_rate": 3.782628027410159e-05,
      "loss": 0.8986,
      "step": 9410
    },
    {
      "epoch": 0.7372045703552982,
      "grad_norm": 37.40110397338867,
      "learning_rate": 3.7813202908406134e-05,
      "loss": 1.1135,
      "step": 9420
    },
    {
      "epoch": 0.7379871654406011,
      "grad_norm": 11.692840576171875,
      "learning_rate": 3.780012554271068e-05,
      "loss": 0.7456,
      "step": 9430
    },
    {
      "epoch": 0.7387697605259039,
      "grad_norm": 8.747519493103027,
      "learning_rate": 3.778704817701522e-05,
      "loss": 0.6946,
      "step": 9440
    },
    {
      "epoch": 0.7395523556112068,
      "grad_norm": 2.548922061920166,
      "learning_rate": 3.777397081131977e-05,
      "loss": 0.9097,
      "step": 9450
    },
    {
      "epoch": 0.7403349506965097,
      "grad_norm": 1.003528118133545,
      "learning_rate": 3.776089344562432e-05,
      "loss": 0.4344,
      "step": 9460
    },
    {
      "epoch": 0.7411175457818124,
      "grad_norm": 5.278439521789551,
      "learning_rate": 3.774781607992886e-05,
      "loss": 0.6904,
      "step": 9470
    },
    {
      "epoch": 0.7419001408671153,
      "grad_norm": 11.276946067810059,
      "learning_rate": 3.773473871423341e-05,
      "loss": 0.9419,
      "step": 9480
    },
    {
      "epoch": 0.7426827359524182,
      "grad_norm": 0.983633816242218,
      "learning_rate": 3.772166134853795e-05,
      "loss": 0.7707,
      "step": 9490
    },
    {
      "epoch": 0.7434653310377211,
      "grad_norm": 0.8894138336181641,
      "learning_rate": 3.77085839828425e-05,
      "loss": 0.9367,
      "step": 9500
    },
    {
      "epoch": 0.7442479261230239,
      "grad_norm": 6.310152053833008,
      "learning_rate": 3.769550661714704e-05,
      "loss": 0.8635,
      "step": 9510
    },
    {
      "epoch": 0.7450305212083268,
      "grad_norm": 3.871520519256592,
      "learning_rate": 3.768242925145159e-05,
      "loss": 1.0465,
      "step": 9520
    },
    {
      "epoch": 0.7458131162936297,
      "grad_norm": 5.212246417999268,
      "learning_rate": 3.766935188575614e-05,
      "loss": 0.5772,
      "step": 9530
    },
    {
      "epoch": 0.7465957113789325,
      "grad_norm": 6.283555030822754,
      "learning_rate": 3.765627452006068e-05,
      "loss": 0.9283,
      "step": 9540
    },
    {
      "epoch": 0.7473783064642354,
      "grad_norm": 13.624217987060547,
      "learning_rate": 3.7643197154365227e-05,
      "loss": 0.4396,
      "step": 9550
    },
    {
      "epoch": 0.7481609015495383,
      "grad_norm": 7.971993446350098,
      "learning_rate": 3.763011978866977e-05,
      "loss": 0.9457,
      "step": 9560
    },
    {
      "epoch": 0.7489434966348412,
      "grad_norm": 0.7513082027435303,
      "learning_rate": 3.7617042422974316e-05,
      "loss": 0.6036,
      "step": 9570
    },
    {
      "epoch": 0.749726091720144,
      "grad_norm": 19.42154312133789,
      "learning_rate": 3.760396505727886e-05,
      "loss": 1.532,
      "step": 9580
    },
    {
      "epoch": 0.7505086868054469,
      "grad_norm": 8.422746658325195,
      "learning_rate": 3.759088769158341e-05,
      "loss": 0.939,
      "step": 9590
    },
    {
      "epoch": 0.7512912818907497,
      "grad_norm": 3.8456811904907227,
      "learning_rate": 3.7577810325887956e-05,
      "loss": 1.1079,
      "step": 9600
    },
    {
      "epoch": 0.7520738769760525,
      "grad_norm": 7.188963890075684,
      "learning_rate": 3.75647329601925e-05,
      "loss": 1.0373,
      "step": 9610
    },
    {
      "epoch": 0.7528564720613554,
      "grad_norm": 6.120348930358887,
      "learning_rate": 3.7551655594497045e-05,
      "loss": 1.0224,
      "step": 9620
    },
    {
      "epoch": 0.7536390671466583,
      "grad_norm": 6.81748104095459,
      "learning_rate": 3.753857822880159e-05,
      "loss": 0.8087,
      "step": 9630
    },
    {
      "epoch": 0.7544216622319612,
      "grad_norm": 14.95832633972168,
      "learning_rate": 3.752550086310614e-05,
      "loss": 0.759,
      "step": 9640
    },
    {
      "epoch": 0.755204257317264,
      "grad_norm": 2.6390271186828613,
      "learning_rate": 3.751242349741068e-05,
      "loss": 0.8151,
      "step": 9650
    },
    {
      "epoch": 0.7559868524025669,
      "grad_norm": 1.4763137102127075,
      "learning_rate": 3.749934613171523e-05,
      "loss": 0.5496,
      "step": 9660
    },
    {
      "epoch": 0.7567694474878698,
      "grad_norm": 4.592081069946289,
      "learning_rate": 3.7486268766019775e-05,
      "loss": 0.8472,
      "step": 9670
    },
    {
      "epoch": 0.7575520425731727,
      "grad_norm": 1.482389211654663,
      "learning_rate": 3.747319140032432e-05,
      "loss": 1.09,
      "step": 9680
    },
    {
      "epoch": 0.7583346376584755,
      "grad_norm": 4.770016670227051,
      "learning_rate": 3.746011403462887e-05,
      "loss": 0.9178,
      "step": 9690
    },
    {
      "epoch": 0.7591172327437784,
      "grad_norm": 6.179487228393555,
      "learning_rate": 3.744703666893341e-05,
      "loss": 0.7273,
      "step": 9700
    },
    {
      "epoch": 0.7598998278290813,
      "grad_norm": 6.565785884857178,
      "learning_rate": 3.743395930323796e-05,
      "loss": 0.8274,
      "step": 9710
    },
    {
      "epoch": 0.7606824229143841,
      "grad_norm": 3.916203498840332,
      "learning_rate": 3.7420881937542505e-05,
      "loss": 1.1538,
      "step": 9720
    },
    {
      "epoch": 0.7614650179996869,
      "grad_norm": 24.689058303833008,
      "learning_rate": 3.740780457184705e-05,
      "loss": 0.8927,
      "step": 9730
    },
    {
      "epoch": 0.7622476130849898,
      "grad_norm": 4.631040573120117,
      "learning_rate": 3.7394727206151594e-05,
      "loss": 1.0314,
      "step": 9740
    },
    {
      "epoch": 0.7630302081702927,
      "grad_norm": 3.2966184616088867,
      "learning_rate": 3.738164984045614e-05,
      "loss": 1.023,
      "step": 9750
    },
    {
      "epoch": 0.7638128032555955,
      "grad_norm": 3.7500574588775635,
      "learning_rate": 3.736857247476069e-05,
      "loss": 1.0678,
      "step": 9760
    },
    {
      "epoch": 0.7645953983408984,
      "grad_norm": 13.31845760345459,
      "learning_rate": 3.735549510906523e-05,
      "loss": 0.6984,
      "step": 9770
    },
    {
      "epoch": 0.7653779934262013,
      "grad_norm": 2.469182014465332,
      "learning_rate": 3.734241774336978e-05,
      "loss": 0.4922,
      "step": 9780
    },
    {
      "epoch": 0.7661605885115041,
      "grad_norm": 20.336883544921875,
      "learning_rate": 3.732934037767432e-05,
      "loss": 0.6928,
      "step": 9790
    },
    {
      "epoch": 0.766943183596807,
      "grad_norm": 8.114266395568848,
      "learning_rate": 3.731626301197887e-05,
      "loss": 1.0819,
      "step": 9800
    },
    {
      "epoch": 0.7677257786821099,
      "grad_norm": 8.66950798034668,
      "learning_rate": 3.730318564628341e-05,
      "loss": 1.2087,
      "step": 9810
    },
    {
      "epoch": 0.7685083737674128,
      "grad_norm": 13.178998947143555,
      "learning_rate": 3.729010828058796e-05,
      "loss": 0.7941,
      "step": 9820
    },
    {
      "epoch": 0.7692909688527156,
      "grad_norm": 4.626152992248535,
      "learning_rate": 3.727703091489251e-05,
      "loss": 0.6085,
      "step": 9830
    },
    {
      "epoch": 0.7700735639380185,
      "grad_norm": 7.5283660888671875,
      "learning_rate": 3.726395354919705e-05,
      "loss": 1.0769,
      "step": 9840
    },
    {
      "epoch": 0.7708561590233214,
      "grad_norm": 2.433025598526001,
      "learning_rate": 3.72508761835016e-05,
      "loss": 0.7287,
      "step": 9850
    },
    {
      "epoch": 0.7716387541086241,
      "grad_norm": 8.39775276184082,
      "learning_rate": 3.723779881780614e-05,
      "loss": 0.7709,
      "step": 9860
    },
    {
      "epoch": 0.772421349193927,
      "grad_norm": 14.26219367980957,
      "learning_rate": 3.722472145211069e-05,
      "loss": 0.8046,
      "step": 9870
    },
    {
      "epoch": 0.7732039442792299,
      "grad_norm": 2.0247738361358643,
      "learning_rate": 3.721164408641523e-05,
      "loss": 0.7578,
      "step": 9880
    },
    {
      "epoch": 0.7739865393645328,
      "grad_norm": 5.331399917602539,
      "learning_rate": 3.719856672071978e-05,
      "loss": 0.9136,
      "step": 9890
    },
    {
      "epoch": 0.7747691344498356,
      "grad_norm": 3.968198299407959,
      "learning_rate": 3.718548935502433e-05,
      "loss": 0.5967,
      "step": 9900
    },
    {
      "epoch": 0.7755517295351385,
      "grad_norm": 11.547032356262207,
      "learning_rate": 3.717241198932887e-05,
      "loss": 1.1023,
      "step": 9910
    },
    {
      "epoch": 0.7763343246204414,
      "grad_norm": 6.5445990562438965,
      "learning_rate": 3.7159334623633416e-05,
      "loss": 0.6064,
      "step": 9920
    },
    {
      "epoch": 0.7771169197057443,
      "grad_norm": 15.941474914550781,
      "learning_rate": 3.714625725793796e-05,
      "loss": 1.0777,
      "step": 9930
    },
    {
      "epoch": 0.7778995147910471,
      "grad_norm": 30.76422882080078,
      "learning_rate": 3.713317989224251e-05,
      "loss": 0.8349,
      "step": 9940
    },
    {
      "epoch": 0.77868210987635,
      "grad_norm": 4.633932590484619,
      "learning_rate": 3.712010252654705e-05,
      "loss": 0.8281,
      "step": 9950
    },
    {
      "epoch": 0.7794647049616529,
      "grad_norm": 5.232914924621582,
      "learning_rate": 3.71070251608516e-05,
      "loss": 0.8473,
      "step": 9960
    },
    {
      "epoch": 0.7802473000469557,
      "grad_norm": 3.8370039463043213,
      "learning_rate": 3.7093947795156146e-05,
      "loss": 1.0476,
      "step": 9970
    },
    {
      "epoch": 0.7810298951322586,
      "grad_norm": 11.700311660766602,
      "learning_rate": 3.708087042946069e-05,
      "loss": 0.8183,
      "step": 9980
    },
    {
      "epoch": 0.7818124902175614,
      "grad_norm": 3.6448049545288086,
      "learning_rate": 3.7067793063765235e-05,
      "loss": 0.8783,
      "step": 9990
    },
    {
      "epoch": 0.7825950853028643,
      "grad_norm": 5.594269752502441,
      "learning_rate": 3.705471569806978e-05,
      "loss": 0.7489,
      "step": 10000
    },
    {
      "epoch": 0.7833776803881671,
      "grad_norm": 1.8851486444473267,
      "learning_rate": 3.704163833237433e-05,
      "loss": 0.6986,
      "step": 10010
    },
    {
      "epoch": 0.78416027547347,
      "grad_norm": 4.303359508514404,
      "learning_rate": 3.702856096667887e-05,
      "loss": 0.7558,
      "step": 10020
    },
    {
      "epoch": 0.7849428705587729,
      "grad_norm": 2.122544288635254,
      "learning_rate": 3.701548360098342e-05,
      "loss": 1.0747,
      "step": 10030
    },
    {
      "epoch": 0.7857254656440757,
      "grad_norm": 3.381225824356079,
      "learning_rate": 3.7002406235287965e-05,
      "loss": 0.8602,
      "step": 10040
    },
    {
      "epoch": 0.7865080607293786,
      "grad_norm": 4.713757038116455,
      "learning_rate": 3.698932886959251e-05,
      "loss": 0.9867,
      "step": 10050
    },
    {
      "epoch": 0.7872906558146815,
      "grad_norm": 6.000115871429443,
      "learning_rate": 3.697625150389706e-05,
      "loss": 0.6765,
      "step": 10060
    },
    {
      "epoch": 0.7880732508999844,
      "grad_norm": 4.810426235198975,
      "learning_rate": 3.69631741382016e-05,
      "loss": 0.8077,
      "step": 10070
    },
    {
      "epoch": 0.7888558459852872,
      "grad_norm": 6.62565803527832,
      "learning_rate": 3.695009677250615e-05,
      "loss": 0.7944,
      "step": 10080
    },
    {
      "epoch": 0.7896384410705901,
      "grad_norm": 16.192970275878906,
      "learning_rate": 3.6937019406810694e-05,
      "loss": 0.8114,
      "step": 10090
    },
    {
      "epoch": 0.790421036155893,
      "grad_norm": 4.13571834564209,
      "learning_rate": 3.692394204111524e-05,
      "loss": 0.8765,
      "step": 10100
    },
    {
      "epoch": 0.7912036312411959,
      "grad_norm": 3.9994308948516846,
      "learning_rate": 3.691086467541979e-05,
      "loss": 0.9674,
      "step": 10110
    },
    {
      "epoch": 0.7919862263264986,
      "grad_norm": 8.671630859375,
      "learning_rate": 3.689778730972433e-05,
      "loss": 0.6679,
      "step": 10120
    },
    {
      "epoch": 0.7927688214118015,
      "grad_norm": 7.721748352050781,
      "learning_rate": 3.688470994402888e-05,
      "loss": 0.9417,
      "step": 10130
    },
    {
      "epoch": 0.7935514164971044,
      "grad_norm": 13.646540641784668,
      "learning_rate": 3.687163257833342e-05,
      "loss": 0.8865,
      "step": 10140
    },
    {
      "epoch": 0.7943340115824072,
      "grad_norm": 3.5141353607177734,
      "learning_rate": 3.685855521263797e-05,
      "loss": 0.6626,
      "step": 10150
    },
    {
      "epoch": 0.7951166066677101,
      "grad_norm": 6.034914493560791,
      "learning_rate": 3.684547784694251e-05,
      "loss": 0.561,
      "step": 10160
    },
    {
      "epoch": 0.795899201753013,
      "grad_norm": 4.47662353515625,
      "learning_rate": 3.683240048124706e-05,
      "loss": 0.6882,
      "step": 10170
    },
    {
      "epoch": 0.7966817968383159,
      "grad_norm": 2.4077489376068115,
      "learning_rate": 3.681932311555161e-05,
      "loss": 1.0312,
      "step": 10180
    },
    {
      "epoch": 0.7974643919236187,
      "grad_norm": 11.261221885681152,
      "learning_rate": 3.680624574985615e-05,
      "loss": 0.9641,
      "step": 10190
    },
    {
      "epoch": 0.7982469870089216,
      "grad_norm": 4.636811256408691,
      "learning_rate": 3.67931683841607e-05,
      "loss": 1.0307,
      "step": 10200
    },
    {
      "epoch": 0.7990295820942245,
      "grad_norm": 6.250582695007324,
      "learning_rate": 3.678009101846524e-05,
      "loss": 1.0749,
      "step": 10210
    },
    {
      "epoch": 0.7998121771795274,
      "grad_norm": 1.855982780456543,
      "learning_rate": 3.676701365276979e-05,
      "loss": 0.7429,
      "step": 10220
    },
    {
      "epoch": 0.8005947722648302,
      "grad_norm": 12.746696472167969,
      "learning_rate": 3.675393628707433e-05,
      "loss": 0.8538,
      "step": 10230
    },
    {
      "epoch": 0.8013773673501331,
      "grad_norm": 5.151258945465088,
      "learning_rate": 3.6740858921378877e-05,
      "loss": 0.5362,
      "step": 10240
    },
    {
      "epoch": 0.8021599624354359,
      "grad_norm": 4.88902473449707,
      "learning_rate": 3.672778155568343e-05,
      "loss": 0.9518,
      "step": 10250
    },
    {
      "epoch": 0.8029425575207387,
      "grad_norm": 12.849262237548828,
      "learning_rate": 3.671470418998797e-05,
      "loss": 1.0623,
      "step": 10260
    },
    {
      "epoch": 0.8037251526060416,
      "grad_norm": 4.634492874145508,
      "learning_rate": 3.670162682429252e-05,
      "loss": 0.8879,
      "step": 10270
    },
    {
      "epoch": 0.8045077476913445,
      "grad_norm": 10.906535148620605,
      "learning_rate": 3.668854945859706e-05,
      "loss": 0.851,
      "step": 10280
    },
    {
      "epoch": 0.8052903427766473,
      "grad_norm": 10.888851165771484,
      "learning_rate": 3.6675472092901606e-05,
      "loss": 0.9617,
      "step": 10290
    },
    {
      "epoch": 0.8060729378619502,
      "grad_norm": 11.071433067321777,
      "learning_rate": 3.666239472720615e-05,
      "loss": 0.6783,
      "step": 10300
    },
    {
      "epoch": 0.8068555329472531,
      "grad_norm": 4.086836338043213,
      "learning_rate": 3.66493173615107e-05,
      "loss": 0.8607,
      "step": 10310
    },
    {
      "epoch": 0.807638128032556,
      "grad_norm": 4.503959655761719,
      "learning_rate": 3.663623999581525e-05,
      "loss": 0.6994,
      "step": 10320
    },
    {
      "epoch": 0.8084207231178588,
      "grad_norm": 7.322079658508301,
      "learning_rate": 3.662316263011979e-05,
      "loss": 0.664,
      "step": 10330
    },
    {
      "epoch": 0.8092033182031617,
      "grad_norm": 5.085352420806885,
      "learning_rate": 3.6610085264424336e-05,
      "loss": 1.1925,
      "step": 10340
    },
    {
      "epoch": 0.8099859132884646,
      "grad_norm": 2.6267688274383545,
      "learning_rate": 3.659700789872888e-05,
      "loss": 0.5812,
      "step": 10350
    },
    {
      "epoch": 0.8107685083737675,
      "grad_norm": 6.729258060455322,
      "learning_rate": 3.6583930533033425e-05,
      "loss": 0.5385,
      "step": 10360
    },
    {
      "epoch": 0.8115511034590702,
      "grad_norm": 5.728621006011963,
      "learning_rate": 3.657085316733797e-05,
      "loss": 0.8976,
      "step": 10370
    },
    {
      "epoch": 0.8123336985443731,
      "grad_norm": 5.400430679321289,
      "learning_rate": 3.655777580164252e-05,
      "loss": 0.5853,
      "step": 10380
    },
    {
      "epoch": 0.813116293629676,
      "grad_norm": 5.039793968200684,
      "learning_rate": 3.6544698435947065e-05,
      "loss": 0.5568,
      "step": 10390
    },
    {
      "epoch": 0.8138988887149788,
      "grad_norm": 8.553184509277344,
      "learning_rate": 3.653162107025161e-05,
      "loss": 0.953,
      "step": 10400
    },
    {
      "epoch": 0.8146814838002817,
      "grad_norm": 5.668429374694824,
      "learning_rate": 3.6518543704556155e-05,
      "loss": 0.6773,
      "step": 10410
    },
    {
      "epoch": 0.8154640788855846,
      "grad_norm": 1.2841883897781372,
      "learning_rate": 3.65054663388607e-05,
      "loss": 0.808,
      "step": 10420
    },
    {
      "epoch": 0.8162466739708875,
      "grad_norm": 11.50239372253418,
      "learning_rate": 3.649238897316525e-05,
      "loss": 0.955,
      "step": 10430
    },
    {
      "epoch": 0.8170292690561903,
      "grad_norm": 9.88693904876709,
      "learning_rate": 3.647931160746979e-05,
      "loss": 1.0384,
      "step": 10440
    },
    {
      "epoch": 0.8178118641414932,
      "grad_norm": 5.970240592956543,
      "learning_rate": 3.646623424177434e-05,
      "loss": 0.8175,
      "step": 10450
    },
    {
      "epoch": 0.8185944592267961,
      "grad_norm": 19.090686798095703,
      "learning_rate": 3.6453156876078884e-05,
      "loss": 0.7178,
      "step": 10460
    },
    {
      "epoch": 0.819377054312099,
      "grad_norm": 3.014803647994995,
      "learning_rate": 3.644007951038343e-05,
      "loss": 0.9667,
      "step": 10470
    },
    {
      "epoch": 0.8201596493974018,
      "grad_norm": 8.407426834106445,
      "learning_rate": 3.642700214468798e-05,
      "loss": 1.12,
      "step": 10480
    },
    {
      "epoch": 0.8209422444827047,
      "grad_norm": 6.075494766235352,
      "learning_rate": 3.641392477899252e-05,
      "loss": 0.7193,
      "step": 10490
    },
    {
      "epoch": 0.8217248395680075,
      "grad_norm": 7.446828842163086,
      "learning_rate": 3.640084741329707e-05,
      "loss": 1.1026,
      "step": 10500
    },
    {
      "epoch": 0.8225074346533103,
      "grad_norm": 2.5052731037139893,
      "learning_rate": 3.638777004760161e-05,
      "loss": 0.5184,
      "step": 10510
    },
    {
      "epoch": 0.8232900297386132,
      "grad_norm": 9.594279289245605,
      "learning_rate": 3.637469268190616e-05,
      "loss": 0.7818,
      "step": 10520
    },
    {
      "epoch": 0.8240726248239161,
      "grad_norm": 6.815733909606934,
      "learning_rate": 3.636161531621071e-05,
      "loss": 0.6129,
      "step": 10530
    },
    {
      "epoch": 0.824855219909219,
      "grad_norm": 3.2187747955322266,
      "learning_rate": 3.634853795051525e-05,
      "loss": 0.7717,
      "step": 10540
    },
    {
      "epoch": 0.8256378149945218,
      "grad_norm": 13.635519027709961,
      "learning_rate": 3.63354605848198e-05,
      "loss": 0.9649,
      "step": 10550
    },
    {
      "epoch": 0.8264204100798247,
      "grad_norm": 6.028953552246094,
      "learning_rate": 3.632238321912434e-05,
      "loss": 0.739,
      "step": 10560
    },
    {
      "epoch": 0.8272030051651276,
      "grad_norm": 3.571211576461792,
      "learning_rate": 3.630930585342889e-05,
      "loss": 0.5042,
      "step": 10570
    },
    {
      "epoch": 0.8279856002504304,
      "grad_norm": 1.4583101272583008,
      "learning_rate": 3.629622848773343e-05,
      "loss": 0.9001,
      "step": 10580
    },
    {
      "epoch": 0.8287681953357333,
      "grad_norm": 6.762120246887207,
      "learning_rate": 3.628315112203798e-05,
      "loss": 0.9965,
      "step": 10590
    },
    {
      "epoch": 0.8295507904210362,
      "grad_norm": 5.120227336883545,
      "learning_rate": 3.627007375634253e-05,
      "loss": 0.7437,
      "step": 10600
    },
    {
      "epoch": 0.8303333855063391,
      "grad_norm": 12.414810180664062,
      "learning_rate": 3.6256996390647066e-05,
      "loss": 0.6291,
      "step": 10610
    },
    {
      "epoch": 0.8311159805916419,
      "grad_norm": 26.107145309448242,
      "learning_rate": 3.624391902495162e-05,
      "loss": 0.674,
      "step": 10620
    },
    {
      "epoch": 0.8318985756769447,
      "grad_norm": 1.9007712602615356,
      "learning_rate": 3.623084165925616e-05,
      "loss": 0.9711,
      "step": 10630
    },
    {
      "epoch": 0.8326811707622476,
      "grad_norm": 1.1280664205551147,
      "learning_rate": 3.621776429356071e-05,
      "loss": 0.4964,
      "step": 10640
    },
    {
      "epoch": 0.8334637658475504,
      "grad_norm": 9.071521759033203,
      "learning_rate": 3.620468692786525e-05,
      "loss": 0.8479,
      "step": 10650
    },
    {
      "epoch": 0.8342463609328533,
      "grad_norm": 7.848548412322998,
      "learning_rate": 3.6191609562169796e-05,
      "loss": 1.049,
      "step": 10660
    },
    {
      "epoch": 0.8350289560181562,
      "grad_norm": 7.424879550933838,
      "learning_rate": 3.617853219647435e-05,
      "loss": 0.9272,
      "step": 10670
    },
    {
      "epoch": 0.8358115511034591,
      "grad_norm": 4.661067962646484,
      "learning_rate": 3.616545483077889e-05,
      "loss": 0.5639,
      "step": 10680
    },
    {
      "epoch": 0.8365941461887619,
      "grad_norm": 8.482114791870117,
      "learning_rate": 3.6152377465083437e-05,
      "loss": 1.1394,
      "step": 10690
    },
    {
      "epoch": 0.8373767412740648,
      "grad_norm": 5.792028427124023,
      "learning_rate": 3.613930009938798e-05,
      "loss": 1.0093,
      "step": 10700
    },
    {
      "epoch": 0.8381593363593677,
      "grad_norm": 18.442520141601562,
      "learning_rate": 3.6126222733692526e-05,
      "loss": 0.9296,
      "step": 10710
    },
    {
      "epoch": 0.8389419314446706,
      "grad_norm": 2.8579583168029785,
      "learning_rate": 3.611314536799707e-05,
      "loss": 0.7334,
      "step": 10720
    },
    {
      "epoch": 0.8397245265299734,
      "grad_norm": 13.648813247680664,
      "learning_rate": 3.6100068002301615e-05,
      "loss": 0.96,
      "step": 10730
    },
    {
      "epoch": 0.8405071216152763,
      "grad_norm": 15.56800651550293,
      "learning_rate": 3.6086990636606166e-05,
      "loss": 0.8682,
      "step": 10740
    },
    {
      "epoch": 0.8412897167005792,
      "grad_norm": 4.822196006774902,
      "learning_rate": 3.607391327091071e-05,
      "loss": 0.9104,
      "step": 10750
    },
    {
      "epoch": 0.8420723117858819,
      "grad_norm": 8.859773635864258,
      "learning_rate": 3.6060835905215255e-05,
      "loss": 1.0493,
      "step": 10760
    },
    {
      "epoch": 0.8428549068711848,
      "grad_norm": 8.34626293182373,
      "learning_rate": 3.60477585395198e-05,
      "loss": 0.7663,
      "step": 10770
    },
    {
      "epoch": 0.8436375019564877,
      "grad_norm": 5.600551605224609,
      "learning_rate": 3.6034681173824344e-05,
      "loss": 0.5737,
      "step": 10780
    },
    {
      "epoch": 0.8444200970417906,
      "grad_norm": 9.783740997314453,
      "learning_rate": 3.602160380812889e-05,
      "loss": 0.9168,
      "step": 10790
    },
    {
      "epoch": 0.8452026921270934,
      "grad_norm": 6.234668254852295,
      "learning_rate": 3.600852644243344e-05,
      "loss": 0.7063,
      "step": 10800
    },
    {
      "epoch": 0.8459852872123963,
      "grad_norm": 6.626399517059326,
      "learning_rate": 3.5995449076737985e-05,
      "loss": 0.8895,
      "step": 10810
    },
    {
      "epoch": 0.8467678822976992,
      "grad_norm": 24.69268226623535,
      "learning_rate": 3.598237171104253e-05,
      "loss": 0.5736,
      "step": 10820
    },
    {
      "epoch": 0.847550477383002,
      "grad_norm": 6.0261969566345215,
      "learning_rate": 3.5969294345347074e-05,
      "loss": 0.8957,
      "step": 10830
    },
    {
      "epoch": 0.8483330724683049,
      "grad_norm": 10.660802841186523,
      "learning_rate": 3.595621697965162e-05,
      "loss": 0.9292,
      "step": 10840
    },
    {
      "epoch": 0.8491156675536078,
      "grad_norm": 6.673320770263672,
      "learning_rate": 3.594313961395617e-05,
      "loss": 0.6514,
      "step": 10850
    },
    {
      "epoch": 0.8498982626389107,
      "grad_norm": 10.375528335571289,
      "learning_rate": 3.593006224826071e-05,
      "loss": 0.843,
      "step": 10860
    },
    {
      "epoch": 0.8506808577242135,
      "grad_norm": 6.132887363433838,
      "learning_rate": 3.591698488256526e-05,
      "loss": 0.7347,
      "step": 10870
    },
    {
      "epoch": 0.8514634528095164,
      "grad_norm": 3.303619146347046,
      "learning_rate": 3.5903907516869804e-05,
      "loss": 0.7317,
      "step": 10880
    },
    {
      "epoch": 0.8522460478948192,
      "grad_norm": 1.2219334840774536,
      "learning_rate": 3.589083015117435e-05,
      "loss": 0.6631,
      "step": 10890
    },
    {
      "epoch": 0.853028642980122,
      "grad_norm": 1.58870530128479,
      "learning_rate": 3.58777527854789e-05,
      "loss": 0.9663,
      "step": 10900
    },
    {
      "epoch": 0.8538112380654249,
      "grad_norm": 6.987854957580566,
      "learning_rate": 3.586467541978344e-05,
      "loss": 1.0693,
      "step": 10910
    },
    {
      "epoch": 0.8545938331507278,
      "grad_norm": 3.152799129486084,
      "learning_rate": 3.585159805408799e-05,
      "loss": 0.6371,
      "step": 10920
    },
    {
      "epoch": 0.8553764282360307,
      "grad_norm": 8.551745414733887,
      "learning_rate": 3.5838520688392527e-05,
      "loss": 0.7452,
      "step": 10930
    },
    {
      "epoch": 0.8561590233213335,
      "grad_norm": 5.619504928588867,
      "learning_rate": 3.582544332269708e-05,
      "loss": 1.1262,
      "step": 10940
    },
    {
      "epoch": 0.8569416184066364,
      "grad_norm": 7.77957820892334,
      "learning_rate": 3.581236595700162e-05,
      "loss": 0.7286,
      "step": 10950
    },
    {
      "epoch": 0.8577242134919393,
      "grad_norm": 7.110109806060791,
      "learning_rate": 3.579928859130617e-05,
      "loss": 0.7619,
      "step": 10960
    },
    {
      "epoch": 0.8585068085772422,
      "grad_norm": 2.2537777423858643,
      "learning_rate": 3.578621122561072e-05,
      "loss": 0.826,
      "step": 10970
    },
    {
      "epoch": 0.859289403662545,
      "grad_norm": 7.485612869262695,
      "learning_rate": 3.5773133859915256e-05,
      "loss": 0.8245,
      "step": 10980
    },
    {
      "epoch": 0.8600719987478479,
      "grad_norm": 0.6913577318191528,
      "learning_rate": 3.576005649421981e-05,
      "loss": 1.0054,
      "step": 10990
    },
    {
      "epoch": 0.8608545938331508,
      "grad_norm": 6.375449180603027,
      "learning_rate": 3.574697912852435e-05,
      "loss": 0.7661,
      "step": 11000
    },
    {
      "epoch": 0.8616371889184536,
      "grad_norm": 7.860613822937012,
      "learning_rate": 3.57339017628289e-05,
      "loss": 0.872,
      "step": 11010
    },
    {
      "epoch": 0.8624197840037564,
      "grad_norm": 5.204077243804932,
      "learning_rate": 3.572082439713345e-05,
      "loss": 0.5816,
      "step": 11020
    },
    {
      "epoch": 0.8632023790890593,
      "grad_norm": 4.138625621795654,
      "learning_rate": 3.5707747031437986e-05,
      "loss": 1.003,
      "step": 11030
    },
    {
      "epoch": 0.8639849741743622,
      "grad_norm": 10.49350357055664,
      "learning_rate": 3.569466966574254e-05,
      "loss": 0.9276,
      "step": 11040
    },
    {
      "epoch": 0.864767569259665,
      "grad_norm": 4.171073913574219,
      "learning_rate": 3.568159230004708e-05,
      "loss": 0.6475,
      "step": 11050
    },
    {
      "epoch": 0.8655501643449679,
      "grad_norm": 6.7842841148376465,
      "learning_rate": 3.5668514934351626e-05,
      "loss": 0.8633,
      "step": 11060
    },
    {
      "epoch": 0.8663327594302708,
      "grad_norm": 7.551939964294434,
      "learning_rate": 3.565543756865617e-05,
      "loss": 1.0577,
      "step": 11070
    },
    {
      "epoch": 0.8671153545155736,
      "grad_norm": 4.101297855377197,
      "learning_rate": 3.5642360202960716e-05,
      "loss": 1.0196,
      "step": 11080
    },
    {
      "epoch": 0.8678979496008765,
      "grad_norm": 1.2399321794509888,
      "learning_rate": 3.562928283726527e-05,
      "loss": 0.8262,
      "step": 11090
    },
    {
      "epoch": 0.8686805446861794,
      "grad_norm": 3.779043674468994,
      "learning_rate": 3.5616205471569805e-05,
      "loss": 0.9328,
      "step": 11100
    },
    {
      "epoch": 0.8694631397714823,
      "grad_norm": 3.655642509460449,
      "learning_rate": 3.5603128105874356e-05,
      "loss": 0.3786,
      "step": 11110
    },
    {
      "epoch": 0.8702457348567851,
      "grad_norm": 9.589274406433105,
      "learning_rate": 3.55900507401789e-05,
      "loss": 1.0415,
      "step": 11120
    },
    {
      "epoch": 0.871028329942088,
      "grad_norm": 5.463377952575684,
      "learning_rate": 3.5576973374483445e-05,
      "loss": 0.8886,
      "step": 11130
    },
    {
      "epoch": 0.8718109250273909,
      "grad_norm": 14.04687213897705,
      "learning_rate": 3.556389600878799e-05,
      "loss": 0.9082,
      "step": 11140
    },
    {
      "epoch": 0.8725935201126936,
      "grad_norm": 9.215266227722168,
      "learning_rate": 3.5550818643092534e-05,
      "loss": 0.6144,
      "step": 11150
    },
    {
      "epoch": 0.8733761151979965,
      "grad_norm": 7.178060531616211,
      "learning_rate": 3.5537741277397086e-05,
      "loss": 0.6992,
      "step": 11160
    },
    {
      "epoch": 0.8741587102832994,
      "grad_norm": 11.43918514251709,
      "learning_rate": 3.552466391170163e-05,
      "loss": 0.7117,
      "step": 11170
    },
    {
      "epoch": 0.8749413053686023,
      "grad_norm": 10.544267654418945,
      "learning_rate": 3.5511586546006175e-05,
      "loss": 0.9928,
      "step": 11180
    },
    {
      "epoch": 0.8757239004539051,
      "grad_norm": 4.563891887664795,
      "learning_rate": 3.549850918031072e-05,
      "loss": 0.6514,
      "step": 11190
    },
    {
      "epoch": 0.876506495539208,
      "grad_norm": 4.212701797485352,
      "learning_rate": 3.5485431814615264e-05,
      "loss": 0.9144,
      "step": 11200
    },
    {
      "epoch": 0.8772890906245109,
      "grad_norm": 0.8957701921463013,
      "learning_rate": 3.547235444891981e-05,
      "loss": 0.7607,
      "step": 11210
    },
    {
      "epoch": 0.8780716857098138,
      "grad_norm": 12.570056915283203,
      "learning_rate": 3.545927708322436e-05,
      "loss": 0.8321,
      "step": 11220
    },
    {
      "epoch": 0.8788542807951166,
      "grad_norm": 10.646146774291992,
      "learning_rate": 3.5446199717528904e-05,
      "loss": 0.7585,
      "step": 11230
    },
    {
      "epoch": 0.8796368758804195,
      "grad_norm": 14.058634757995605,
      "learning_rate": 3.543312235183345e-05,
      "loss": 0.7811,
      "step": 11240
    },
    {
      "epoch": 0.8804194709657224,
      "grad_norm": 7.458741188049316,
      "learning_rate": 3.5420044986137994e-05,
      "loss": 0.8115,
      "step": 11250
    },
    {
      "epoch": 0.8812020660510252,
      "grad_norm": 2.37280011177063,
      "learning_rate": 3.540696762044254e-05,
      "loss": 0.5458,
      "step": 11260
    },
    {
      "epoch": 0.8819846611363281,
      "grad_norm": 3.860011100769043,
      "learning_rate": 3.539389025474709e-05,
      "loss": 0.6914,
      "step": 11270
    },
    {
      "epoch": 0.8827672562216309,
      "grad_norm": 5.6228156089782715,
      "learning_rate": 3.538081288905163e-05,
      "loss": 0.9698,
      "step": 11280
    },
    {
      "epoch": 0.8835498513069338,
      "grad_norm": 13.426071166992188,
      "learning_rate": 3.536773552335618e-05,
      "loss": 0.6297,
      "step": 11290
    },
    {
      "epoch": 0.8843324463922366,
      "grad_norm": 5.350366592407227,
      "learning_rate": 3.535465815766072e-05,
      "loss": 0.8667,
      "step": 11300
    },
    {
      "epoch": 0.8851150414775395,
      "grad_norm": 9.123900413513184,
      "learning_rate": 3.534158079196527e-05,
      "loss": 1.0265,
      "step": 11310
    },
    {
      "epoch": 0.8858976365628424,
      "grad_norm": 1.06675124168396,
      "learning_rate": 3.532850342626981e-05,
      "loss": 0.7876,
      "step": 11320
    },
    {
      "epoch": 0.8866802316481452,
      "grad_norm": 9.339900970458984,
      "learning_rate": 3.531542606057436e-05,
      "loss": 1.0014,
      "step": 11330
    },
    {
      "epoch": 0.8874628267334481,
      "grad_norm": 8.362929344177246,
      "learning_rate": 3.530234869487891e-05,
      "loss": 0.9635,
      "step": 11340
    },
    {
      "epoch": 0.888245421818751,
      "grad_norm": 2.205980062484741,
      "learning_rate": 3.5289271329183446e-05,
      "loss": 0.795,
      "step": 11350
    },
    {
      "epoch": 0.8890280169040539,
      "grad_norm": 2.7704968452453613,
      "learning_rate": 3.5276193963488e-05,
      "loss": 0.6705,
      "step": 11360
    },
    {
      "epoch": 0.8898106119893567,
      "grad_norm": 3.5042600631713867,
      "learning_rate": 3.526311659779254e-05,
      "loss": 0.7997,
      "step": 11370
    },
    {
      "epoch": 0.8905932070746596,
      "grad_norm": 1.9923834800720215,
      "learning_rate": 3.5250039232097087e-05,
      "loss": 0.4462,
      "step": 11380
    },
    {
      "epoch": 0.8913758021599625,
      "grad_norm": 5.433088302612305,
      "learning_rate": 3.523696186640164e-05,
      "loss": 1.1243,
      "step": 11390
    },
    {
      "epoch": 0.8921583972452652,
      "grad_norm": 15.210296630859375,
      "learning_rate": 3.5223884500706176e-05,
      "loss": 0.683,
      "step": 11400
    },
    {
      "epoch": 0.8929409923305681,
      "grad_norm": 0.7778974771499634,
      "learning_rate": 3.521080713501073e-05,
      "loss": 0.9579,
      "step": 11410
    },
    {
      "epoch": 0.893723587415871,
      "grad_norm": 10.186505317687988,
      "learning_rate": 3.519772976931527e-05,
      "loss": 0.7321,
      "step": 11420
    },
    {
      "epoch": 0.8945061825011739,
      "grad_norm": 3.609571933746338,
      "learning_rate": 3.5184652403619816e-05,
      "loss": 0.8581,
      "step": 11430
    },
    {
      "epoch": 0.8952887775864767,
      "grad_norm": 7.004932880401611,
      "learning_rate": 3.517157503792437e-05,
      "loss": 1.0937,
      "step": 11440
    },
    {
      "epoch": 0.8960713726717796,
      "grad_norm": 1.1994749307632446,
      "learning_rate": 3.5158497672228905e-05,
      "loss": 0.5834,
      "step": 11450
    },
    {
      "epoch": 0.8968539677570825,
      "grad_norm": 8.21371078491211,
      "learning_rate": 3.514542030653346e-05,
      "loss": 0.8117,
      "step": 11460
    },
    {
      "epoch": 0.8976365628423854,
      "grad_norm": 23.49414825439453,
      "learning_rate": 3.5132342940837994e-05,
      "loss": 0.518,
      "step": 11470
    },
    {
      "epoch": 0.8984191579276882,
      "grad_norm": 9.8228178024292,
      "learning_rate": 3.5119265575142546e-05,
      "loss": 1.0051,
      "step": 11480
    },
    {
      "epoch": 0.8992017530129911,
      "grad_norm": 10.133712768554688,
      "learning_rate": 3.510618820944709e-05,
      "loss": 0.7794,
      "step": 11490
    },
    {
      "epoch": 0.899984348098294,
      "grad_norm": 4.849931240081787,
      "learning_rate": 3.5093110843751635e-05,
      "loss": 1.1454,
      "step": 11500
    },
    {
      "epoch": 0.9007669431835968,
      "grad_norm": 10.140023231506348,
      "learning_rate": 3.5080033478056186e-05,
      "loss": 0.8717,
      "step": 11510
    },
    {
      "epoch": 0.9015495382688997,
      "grad_norm": 2.7324447631835938,
      "learning_rate": 3.5066956112360724e-05,
      "loss": 1.0125,
      "step": 11520
    },
    {
      "epoch": 0.9023321333542025,
      "grad_norm": 6.915637016296387,
      "learning_rate": 3.5053878746665275e-05,
      "loss": 0.6915,
      "step": 11530
    },
    {
      "epoch": 0.9031147284395054,
      "grad_norm": 9.463434219360352,
      "learning_rate": 3.504080138096982e-05,
      "loss": 0.9816,
      "step": 11540
    },
    {
      "epoch": 0.9038973235248082,
      "grad_norm": 5.309085845947266,
      "learning_rate": 3.5027724015274365e-05,
      "loss": 0.8567,
      "step": 11550
    },
    {
      "epoch": 0.9046799186101111,
      "grad_norm": 2.1555042266845703,
      "learning_rate": 3.501464664957891e-05,
      "loss": 0.753,
      "step": 11560
    },
    {
      "epoch": 0.905462513695414,
      "grad_norm": 6.905576705932617,
      "learning_rate": 3.5001569283883454e-05,
      "loss": 0.8016,
      "step": 11570
    },
    {
      "epoch": 0.9062451087807168,
      "grad_norm": 7.598255634307861,
      "learning_rate": 3.4988491918188005e-05,
      "loss": 1.0286,
      "step": 11580
    },
    {
      "epoch": 0.9070277038660197,
      "grad_norm": 6.0048980712890625,
      "learning_rate": 3.497541455249255e-05,
      "loss": 0.7693,
      "step": 11590
    },
    {
      "epoch": 0.9078102989513226,
      "grad_norm": 4.856121063232422,
      "learning_rate": 3.4962337186797094e-05,
      "loss": 0.5874,
      "step": 11600
    },
    {
      "epoch": 0.9085928940366255,
      "grad_norm": 7.813376426696777,
      "learning_rate": 3.494925982110164e-05,
      "loss": 0.6017,
      "step": 11610
    },
    {
      "epoch": 0.9093754891219283,
      "grad_norm": 6.94638729095459,
      "learning_rate": 3.4936182455406183e-05,
      "loss": 0.7024,
      "step": 11620
    },
    {
      "epoch": 0.9101580842072312,
      "grad_norm": 10.164712905883789,
      "learning_rate": 3.492310508971073e-05,
      "loss": 0.9132,
      "step": 11630
    },
    {
      "epoch": 0.9109406792925341,
      "grad_norm": 1.1712931394577026,
      "learning_rate": 3.491002772401528e-05,
      "loss": 0.7547,
      "step": 11640
    },
    {
      "epoch": 0.911723274377837,
      "grad_norm": 1.240386724472046,
      "learning_rate": 3.4896950358319824e-05,
      "loss": 0.5933,
      "step": 11650
    },
    {
      "epoch": 0.9125058694631397,
      "grad_norm": 3.7581727504730225,
      "learning_rate": 3.488387299262437e-05,
      "loss": 0.6442,
      "step": 11660
    },
    {
      "epoch": 0.9132884645484426,
      "grad_norm": 4.248022079467773,
      "learning_rate": 3.487079562692891e-05,
      "loss": 0.9498,
      "step": 11670
    },
    {
      "epoch": 0.9140710596337455,
      "grad_norm": 4.522243499755859,
      "learning_rate": 3.485771826123346e-05,
      "loss": 0.9588,
      "step": 11680
    },
    {
      "epoch": 0.9148536547190483,
      "grad_norm": 3.487558126449585,
      "learning_rate": 3.4844640895538e-05,
      "loss": 0.8439,
      "step": 11690
    },
    {
      "epoch": 0.9156362498043512,
      "grad_norm": 6.033092021942139,
      "learning_rate": 3.483156352984255e-05,
      "loss": 0.9251,
      "step": 11700
    },
    {
      "epoch": 0.9164188448896541,
      "grad_norm": 4.021429538726807,
      "learning_rate": 3.48184861641471e-05,
      "loss": 0.4442,
      "step": 11710
    },
    {
      "epoch": 0.917201439974957,
      "grad_norm": 5.726035118103027,
      "learning_rate": 3.480540879845164e-05,
      "loss": 0.9211,
      "step": 11720
    },
    {
      "epoch": 0.9179840350602598,
      "grad_norm": 14.468080520629883,
      "learning_rate": 3.479233143275619e-05,
      "loss": 0.9856,
      "step": 11730
    },
    {
      "epoch": 0.9187666301455627,
      "grad_norm": 11.844073295593262,
      "learning_rate": 3.477925406706073e-05,
      "loss": 0.8183,
      "step": 11740
    },
    {
      "epoch": 0.9195492252308656,
      "grad_norm": 19.281150817871094,
      "learning_rate": 3.4766176701365276e-05,
      "loss": 0.7925,
      "step": 11750
    },
    {
      "epoch": 0.9203318203161684,
      "grad_norm": 3.8624112606048584,
      "learning_rate": 3.475309933566983e-05,
      "loss": 1.2411,
      "step": 11760
    },
    {
      "epoch": 0.9211144154014713,
      "grad_norm": 1.2650091648101807,
      "learning_rate": 3.4740021969974366e-05,
      "loss": 0.5362,
      "step": 11770
    },
    {
      "epoch": 0.9218970104867742,
      "grad_norm": 2.2171061038970947,
      "learning_rate": 3.472694460427892e-05,
      "loss": 0.7298,
      "step": 11780
    },
    {
      "epoch": 0.922679605572077,
      "grad_norm": 4.107991695404053,
      "learning_rate": 3.471386723858346e-05,
      "loss": 0.8515,
      "step": 11790
    },
    {
      "epoch": 0.9234622006573798,
      "grad_norm": 5.685413360595703,
      "learning_rate": 3.4700789872888006e-05,
      "loss": 0.7024,
      "step": 11800
    },
    {
      "epoch": 0.9242447957426827,
      "grad_norm": 4.032734394073486,
      "learning_rate": 3.468771250719256e-05,
      "loss": 0.9588,
      "step": 11810
    },
    {
      "epoch": 0.9250273908279856,
      "grad_norm": 3.8399784564971924,
      "learning_rate": 3.4674635141497095e-05,
      "loss": 0.6165,
      "step": 11820
    },
    {
      "epoch": 0.9258099859132884,
      "grad_norm": 4.293371200561523,
      "learning_rate": 3.4661557775801647e-05,
      "loss": 0.8421,
      "step": 11830
    },
    {
      "epoch": 0.9265925809985913,
      "grad_norm": 5.428767204284668,
      "learning_rate": 3.464848041010619e-05,
      "loss": 1.1149,
      "step": 11840
    },
    {
      "epoch": 0.9273751760838942,
      "grad_norm": 2.8092947006225586,
      "learning_rate": 3.4635403044410736e-05,
      "loss": 0.7712,
      "step": 11850
    },
    {
      "epoch": 0.9281577711691971,
      "grad_norm": 6.400421142578125,
      "learning_rate": 3.462232567871529e-05,
      "loss": 1.31,
      "step": 11860
    },
    {
      "epoch": 0.9289403662544999,
      "grad_norm": 4.97222900390625,
      "learning_rate": 3.4609248313019825e-05,
      "loss": 0.8649,
      "step": 11870
    },
    {
      "epoch": 0.9297229613398028,
      "grad_norm": 2.7045905590057373,
      "learning_rate": 3.4596170947324376e-05,
      "loss": 0.6218,
      "step": 11880
    },
    {
      "epoch": 0.9305055564251057,
      "grad_norm": 10.887977600097656,
      "learning_rate": 3.4583093581628914e-05,
      "loss": 1.0222,
      "step": 11890
    },
    {
      "epoch": 0.9312881515104086,
      "grad_norm": 7.447041034698486,
      "learning_rate": 3.4570016215933465e-05,
      "loss": 0.9743,
      "step": 11900
    },
    {
      "epoch": 0.9320707465957114,
      "grad_norm": 2.4817514419555664,
      "learning_rate": 3.455693885023801e-05,
      "loss": 0.8288,
      "step": 11910
    },
    {
      "epoch": 0.9328533416810142,
      "grad_norm": 7.068378448486328,
      "learning_rate": 3.4543861484542554e-05,
      "loss": 0.6148,
      "step": 11920
    },
    {
      "epoch": 0.9336359367663171,
      "grad_norm": 3.0792009830474854,
      "learning_rate": 3.4530784118847106e-05,
      "loss": 0.8655,
      "step": 11930
    },
    {
      "epoch": 0.9344185318516199,
      "grad_norm": 8.4252347946167,
      "learning_rate": 3.4517706753151644e-05,
      "loss": 0.8406,
      "step": 11940
    },
    {
      "epoch": 0.9352011269369228,
      "grad_norm": 6.500677585601807,
      "learning_rate": 3.4504629387456195e-05,
      "loss": 0.9855,
      "step": 11950
    },
    {
      "epoch": 0.9359837220222257,
      "grad_norm": 6.721009731292725,
      "learning_rate": 3.449155202176074e-05,
      "loss": 0.8905,
      "step": 11960
    },
    {
      "epoch": 0.9367663171075286,
      "grad_norm": 3.5871284008026123,
      "learning_rate": 3.4478474656065284e-05,
      "loss": 0.9179,
      "step": 11970
    },
    {
      "epoch": 0.9375489121928314,
      "grad_norm": 5.363833904266357,
      "learning_rate": 3.446539729036983e-05,
      "loss": 0.8921,
      "step": 11980
    },
    {
      "epoch": 0.9383315072781343,
      "grad_norm": 10.35180377960205,
      "learning_rate": 3.445231992467437e-05,
      "loss": 0.7991,
      "step": 11990
    },
    {
      "epoch": 0.9391141023634372,
      "grad_norm": 2.513490915298462,
      "learning_rate": 3.4439242558978925e-05,
      "loss": 0.6113,
      "step": 12000
    },
    {
      "epoch": 0.93989669744874,
      "grad_norm": 5.3392558097839355,
      "learning_rate": 3.442616519328347e-05,
      "loss": 0.8911,
      "step": 12010
    },
    {
      "epoch": 0.9406792925340429,
      "grad_norm": 3.96268892288208,
      "learning_rate": 3.4413087827588014e-05,
      "loss": 0.5592,
      "step": 12020
    },
    {
      "epoch": 0.9414618876193458,
      "grad_norm": 7.213372230529785,
      "learning_rate": 3.440001046189256e-05,
      "loss": 0.6888,
      "step": 12030
    },
    {
      "epoch": 0.9422444827046487,
      "grad_norm": 15.021231651306152,
      "learning_rate": 3.43869330961971e-05,
      "loss": 0.8322,
      "step": 12040
    },
    {
      "epoch": 0.9430270777899514,
      "grad_norm": 2.3253257274627686,
      "learning_rate": 3.437385573050165e-05,
      "loss": 0.6476,
      "step": 12050
    },
    {
      "epoch": 0.9438096728752543,
      "grad_norm": 0.8980764746665955,
      "learning_rate": 3.436077836480619e-05,
      "loss": 0.756,
      "step": 12060
    },
    {
      "epoch": 0.9445922679605572,
      "grad_norm": 19.388641357421875,
      "learning_rate": 3.434770099911074e-05,
      "loss": 1.0922,
      "step": 12070
    },
    {
      "epoch": 0.94537486304586,
      "grad_norm": 7.800928115844727,
      "learning_rate": 3.433462363341529e-05,
      "loss": 0.6618,
      "step": 12080
    },
    {
      "epoch": 0.9461574581311629,
      "grad_norm": 11.819890975952148,
      "learning_rate": 3.432154626771983e-05,
      "loss": 0.5937,
      "step": 12090
    },
    {
      "epoch": 0.9469400532164658,
      "grad_norm": 4.127297401428223,
      "learning_rate": 3.430846890202438e-05,
      "loss": 0.5629,
      "step": 12100
    },
    {
      "epoch": 0.9477226483017687,
      "grad_norm": 3.9273183345794678,
      "learning_rate": 3.429539153632892e-05,
      "loss": 0.9652,
      "step": 12110
    },
    {
      "epoch": 0.9485052433870715,
      "grad_norm": 9.410789489746094,
      "learning_rate": 3.4282314170633466e-05,
      "loss": 1.2749,
      "step": 12120
    },
    {
      "epoch": 0.9492878384723744,
      "grad_norm": 6.090137481689453,
      "learning_rate": 3.426923680493802e-05,
      "loss": 0.8789,
      "step": 12130
    },
    {
      "epoch": 0.9500704335576773,
      "grad_norm": 13.6611967086792,
      "learning_rate": 3.425615943924256e-05,
      "loss": 0.801,
      "step": 12140
    },
    {
      "epoch": 0.9508530286429802,
      "grad_norm": 16.564332962036133,
      "learning_rate": 3.424308207354711e-05,
      "loss": 0.7265,
      "step": 12150
    },
    {
      "epoch": 0.951635623728283,
      "grad_norm": 6.528677463531494,
      "learning_rate": 3.423000470785165e-05,
      "loss": 0.9773,
      "step": 12160
    },
    {
      "epoch": 0.9524182188135859,
      "grad_norm": 8.16634464263916,
      "learning_rate": 3.4216927342156196e-05,
      "loss": 0.867,
      "step": 12170
    },
    {
      "epoch": 0.9532008138988887,
      "grad_norm": 2.8427419662475586,
      "learning_rate": 3.420384997646075e-05,
      "loss": 0.8184,
      "step": 12180
    },
    {
      "epoch": 0.9539834089841915,
      "grad_norm": 4.514340400695801,
      "learning_rate": 3.4190772610765285e-05,
      "loss": 0.7513,
      "step": 12190
    },
    {
      "epoch": 0.9547660040694944,
      "grad_norm": 5.544354438781738,
      "learning_rate": 3.4177695245069836e-05,
      "loss": 1.1189,
      "step": 12200
    },
    {
      "epoch": 0.9555485991547973,
      "grad_norm": 4.246341705322266,
      "learning_rate": 3.416461787937438e-05,
      "loss": 0.8083,
      "step": 12210
    },
    {
      "epoch": 0.9563311942401002,
      "grad_norm": 0.856589138507843,
      "learning_rate": 3.4151540513678925e-05,
      "loss": 0.662,
      "step": 12220
    },
    {
      "epoch": 0.957113789325403,
      "grad_norm": 8.42383098602295,
      "learning_rate": 3.413846314798348e-05,
      "loss": 1.1416,
      "step": 12230
    },
    {
      "epoch": 0.9578963844107059,
      "grad_norm": 6.359172821044922,
      "learning_rate": 3.4125385782288015e-05,
      "loss": 1.0002,
      "step": 12240
    },
    {
      "epoch": 0.9586789794960088,
      "grad_norm": 4.44343900680542,
      "learning_rate": 3.4112308416592566e-05,
      "loss": 0.978,
      "step": 12250
    },
    {
      "epoch": 0.9594615745813116,
      "grad_norm": 4.113702774047852,
      "learning_rate": 3.4099231050897104e-05,
      "loss": 0.8704,
      "step": 12260
    },
    {
      "epoch": 0.9602441696666145,
      "grad_norm": 8.135980606079102,
      "learning_rate": 3.4086153685201655e-05,
      "loss": 0.7274,
      "step": 12270
    },
    {
      "epoch": 0.9610267647519174,
      "grad_norm": 86.19222259521484,
      "learning_rate": 3.40730763195062e-05,
      "loss": 1.0817,
      "step": 12280
    },
    {
      "epoch": 0.9618093598372203,
      "grad_norm": 8.527544975280762,
      "learning_rate": 3.4059998953810744e-05,
      "loss": 0.9385,
      "step": 12290
    },
    {
      "epoch": 0.9625919549225231,
      "grad_norm": 5.870935440063477,
      "learning_rate": 3.4046921588115296e-05,
      "loss": 0.7213,
      "step": 12300
    },
    {
      "epoch": 0.9633745500078259,
      "grad_norm": 2.9078922271728516,
      "learning_rate": 3.4033844222419833e-05,
      "loss": 0.8056,
      "step": 12310
    },
    {
      "epoch": 0.9641571450931288,
      "grad_norm": 12.643360137939453,
      "learning_rate": 3.4020766856724385e-05,
      "loss": 0.8185,
      "step": 12320
    },
    {
      "epoch": 0.9649397401784316,
      "grad_norm": 3.632298707962036,
      "learning_rate": 3.400768949102893e-05,
      "loss": 0.8258,
      "step": 12330
    },
    {
      "epoch": 0.9657223352637345,
      "grad_norm": 8.43837833404541,
      "learning_rate": 3.3994612125333474e-05,
      "loss": 0.714,
      "step": 12340
    },
    {
      "epoch": 0.9665049303490374,
      "grad_norm": 1.6649950742721558,
      "learning_rate": 3.3981534759638025e-05,
      "loss": 0.5351,
      "step": 12350
    },
    {
      "epoch": 0.9672875254343403,
      "grad_norm": 20.268205642700195,
      "learning_rate": 3.396845739394256e-05,
      "loss": 0.7903,
      "step": 12360
    },
    {
      "epoch": 0.9680701205196431,
      "grad_norm": 1.283799171447754,
      "learning_rate": 3.3955380028247114e-05,
      "loss": 0.3101,
      "step": 12370
    },
    {
      "epoch": 0.968852715604946,
      "grad_norm": 2.1902029514312744,
      "learning_rate": 3.394230266255166e-05,
      "loss": 0.9646,
      "step": 12380
    },
    {
      "epoch": 0.9696353106902489,
      "grad_norm": 5.173718452453613,
      "learning_rate": 3.3929225296856204e-05,
      "loss": 0.936,
      "step": 12390
    },
    {
      "epoch": 0.9704179057755518,
      "grad_norm": 0.7088520526885986,
      "learning_rate": 3.391614793116075e-05,
      "loss": 0.9327,
      "step": 12400
    },
    {
      "epoch": 0.9712005008608546,
      "grad_norm": 0.9154521822929382,
      "learning_rate": 3.390307056546529e-05,
      "loss": 0.8619,
      "step": 12410
    },
    {
      "epoch": 0.9719830959461575,
      "grad_norm": 8.815054893493652,
      "learning_rate": 3.388999319976984e-05,
      "loss": 0.8928,
      "step": 12420
    },
    {
      "epoch": 0.9727656910314604,
      "grad_norm": 0.7634473443031311,
      "learning_rate": 3.387691583407439e-05,
      "loss": 0.7432,
      "step": 12430
    },
    {
      "epoch": 0.9735482861167631,
      "grad_norm": 7.639765739440918,
      "learning_rate": 3.386383846837893e-05,
      "loss": 0.9485,
      "step": 12440
    },
    {
      "epoch": 0.974330881202066,
      "grad_norm": 14.40683364868164,
      "learning_rate": 3.385076110268348e-05,
      "loss": 1.0001,
      "step": 12450
    },
    {
      "epoch": 0.9751134762873689,
      "grad_norm": 15.210894584655762,
      "learning_rate": 3.383768373698802e-05,
      "loss": 1.0562,
      "step": 12460
    },
    {
      "epoch": 0.9758960713726718,
      "grad_norm": 7.205379962921143,
      "learning_rate": 3.382460637129257e-05,
      "loss": 0.8556,
      "step": 12470
    },
    {
      "epoch": 0.9766786664579746,
      "grad_norm": 6.309723377227783,
      "learning_rate": 3.381152900559711e-05,
      "loss": 0.7174,
      "step": 12480
    },
    {
      "epoch": 0.9774612615432775,
      "grad_norm": 5.31593132019043,
      "learning_rate": 3.3798451639901656e-05,
      "loss": 1.1234,
      "step": 12490
    },
    {
      "epoch": 0.9782438566285804,
      "grad_norm": 1.1455459594726562,
      "learning_rate": 3.378537427420621e-05,
      "loss": 0.7538,
      "step": 12500
    },
    {
      "epoch": 0.9790264517138833,
      "grad_norm": 0.8121974468231201,
      "learning_rate": 3.377229690851075e-05,
      "loss": 0.4264,
      "step": 12510
    },
    {
      "epoch": 0.9798090467991861,
      "grad_norm": 17.719966888427734,
      "learning_rate": 3.3759219542815297e-05,
      "loss": 0.9446,
      "step": 12520
    },
    {
      "epoch": 0.980591641884489,
      "grad_norm": 10.493967056274414,
      "learning_rate": 3.374614217711984e-05,
      "loss": 1.2618,
      "step": 12530
    },
    {
      "epoch": 0.9813742369697919,
      "grad_norm": 5.243232727050781,
      "learning_rate": 3.3733064811424386e-05,
      "loss": 1.0812,
      "step": 12540
    },
    {
      "epoch": 0.9821568320550947,
      "grad_norm": 8.974884986877441,
      "learning_rate": 3.371998744572894e-05,
      "loss": 0.7182,
      "step": 12550
    },
    {
      "epoch": 0.9829394271403975,
      "grad_norm": 8.383482933044434,
      "learning_rate": 3.3706910080033475e-05,
      "loss": 0.9543,
      "step": 12560
    },
    {
      "epoch": 0.9837220222257004,
      "grad_norm": 4.43451452255249,
      "learning_rate": 3.3693832714338026e-05,
      "loss": 0.8486,
      "step": 12570
    },
    {
      "epoch": 0.9845046173110032,
      "grad_norm": 3.6539413928985596,
      "learning_rate": 3.368075534864257e-05,
      "loss": 0.8844,
      "step": 12580
    },
    {
      "epoch": 0.9852872123963061,
      "grad_norm": 10.653773307800293,
      "learning_rate": 3.3667677982947115e-05,
      "loss": 0.4564,
      "step": 12590
    },
    {
      "epoch": 0.986069807481609,
      "grad_norm": 9.804265022277832,
      "learning_rate": 3.365460061725167e-05,
      "loss": 1.1837,
      "step": 12600
    },
    {
      "epoch": 0.9868524025669119,
      "grad_norm": 7.625603675842285,
      "learning_rate": 3.3641523251556204e-05,
      "loss": 0.8658,
      "step": 12610
    },
    {
      "epoch": 0.9876349976522147,
      "grad_norm": 10.99460220336914,
      "learning_rate": 3.3628445885860756e-05,
      "loss": 0.7303,
      "step": 12620
    },
    {
      "epoch": 0.9884175927375176,
      "grad_norm": 2.686492681503296,
      "learning_rate": 3.3615368520165294e-05,
      "loss": 0.7699,
      "step": 12630
    },
    {
      "epoch": 0.9892001878228205,
      "grad_norm": 13.371170997619629,
      "learning_rate": 3.3602291154469845e-05,
      "loss": 0.8791,
      "step": 12640
    },
    {
      "epoch": 0.9899827829081234,
      "grad_norm": 9.744256973266602,
      "learning_rate": 3.3589213788774396e-05,
      "loss": 0.7502,
      "step": 12650
    },
    {
      "epoch": 0.9907653779934262,
      "grad_norm": 1.8037083148956299,
      "learning_rate": 3.3576136423078934e-05,
      "loss": 1.1442,
      "step": 12660
    },
    {
      "epoch": 0.9915479730787291,
      "grad_norm": 7.6590256690979,
      "learning_rate": 3.3563059057383485e-05,
      "loss": 0.7333,
      "step": 12670
    },
    {
      "epoch": 0.992330568164032,
      "grad_norm": 8.28674030303955,
      "learning_rate": 3.354998169168802e-05,
      "loss": 0.9158,
      "step": 12680
    },
    {
      "epoch": 0.9931131632493347,
      "grad_norm": 5.525173664093018,
      "learning_rate": 3.3536904325992575e-05,
      "loss": 0.8037,
      "step": 12690
    },
    {
      "epoch": 0.9938957583346376,
      "grad_norm": 4.220394611358643,
      "learning_rate": 3.352382696029712e-05,
      "loss": 0.6638,
      "step": 12700
    },
    {
      "epoch": 0.9946783534199405,
      "grad_norm": 5.03434419631958,
      "learning_rate": 3.3510749594601664e-05,
      "loss": 0.6424,
      "step": 12710
    },
    {
      "epoch": 0.9954609485052434,
      "grad_norm": 8.16191291809082,
      "learning_rate": 3.3497672228906215e-05,
      "loss": 1.0027,
      "step": 12720
    },
    {
      "epoch": 0.9962435435905462,
      "grad_norm": 1.308840274810791,
      "learning_rate": 3.348459486321075e-05,
      "loss": 0.8715,
      "step": 12730
    },
    {
      "epoch": 0.9970261386758491,
      "grad_norm": 3.9774765968322754,
      "learning_rate": 3.3471517497515304e-05,
      "loss": 0.9284,
      "step": 12740
    },
    {
      "epoch": 0.997808733761152,
      "grad_norm": 4.672857284545898,
      "learning_rate": 3.345844013181985e-05,
      "loss": 0.879,
      "step": 12750
    },
    {
      "epoch": 0.9985913288464549,
      "grad_norm": 9.048922538757324,
      "learning_rate": 3.344536276612439e-05,
      "loss": 0.9643,
      "step": 12760
    },
    {
      "epoch": 0.9993739239317577,
      "grad_norm": 8.266080856323242,
      "learning_rate": 3.343228540042894e-05,
      "loss": 0.7097,
      "step": 12770
    },
    {
      "epoch": 1.0001565190170605,
      "grad_norm": 4.558141231536865,
      "learning_rate": 3.341920803473348e-05,
      "loss": 1.0076,
      "step": 12780
    },
    {
      "epoch": 1.0009391141023634,
      "grad_norm": 3.7503702640533447,
      "learning_rate": 3.3406130669038034e-05,
      "loss": 0.825,
      "step": 12790
    },
    {
      "epoch": 1.0017217091876662,
      "grad_norm": 1.5131771564483643,
      "learning_rate": 3.339305330334258e-05,
      "loss": 0.5686,
      "step": 12800
    },
    {
      "epoch": 1.002504304272969,
      "grad_norm": 10.810714721679688,
      "learning_rate": 3.337997593764712e-05,
      "loss": 0.8279,
      "step": 12810
    },
    {
      "epoch": 1.003286899358272,
      "grad_norm": 14.753837585449219,
      "learning_rate": 3.336689857195167e-05,
      "loss": 0.7759,
      "step": 12820
    },
    {
      "epoch": 1.0040694944435749,
      "grad_norm": 9.5933198928833,
      "learning_rate": 3.335382120625621e-05,
      "loss": 0.9791,
      "step": 12830
    },
    {
      "epoch": 1.0048520895288777,
      "grad_norm": 5.956554889678955,
      "learning_rate": 3.334074384056076e-05,
      "loss": 0.6099,
      "step": 12840
    },
    {
      "epoch": 1.0056346846141806,
      "grad_norm": 5.397317409515381,
      "learning_rate": 3.33276664748653e-05,
      "loss": 0.9008,
      "step": 12850
    },
    {
      "epoch": 1.0064172796994835,
      "grad_norm": 8.963834762573242,
      "learning_rate": 3.331458910916985e-05,
      "loss": 1.1085,
      "step": 12860
    },
    {
      "epoch": 1.0071998747847863,
      "grad_norm": 6.256855487823486,
      "learning_rate": 3.33015117434744e-05,
      "loss": 0.8749,
      "step": 12870
    },
    {
      "epoch": 1.0079824698700892,
      "grad_norm": 16.48638343811035,
      "learning_rate": 3.328843437777894e-05,
      "loss": 0.8758,
      "step": 12880
    },
    {
      "epoch": 1.008765064955392,
      "grad_norm": 7.118346214294434,
      "learning_rate": 3.3275357012083486e-05,
      "loss": 1.0352,
      "step": 12890
    },
    {
      "epoch": 1.009547660040695,
      "grad_norm": 3.3117823600769043,
      "learning_rate": 3.326227964638803e-05,
      "loss": 0.8755,
      "step": 12900
    },
    {
      "epoch": 1.0103302551259978,
      "grad_norm": 9.302314758300781,
      "learning_rate": 3.3249202280692576e-05,
      "loss": 0.5801,
      "step": 12910
    },
    {
      "epoch": 1.0111128502113007,
      "grad_norm": 9.233630180358887,
      "learning_rate": 3.323612491499713e-05,
      "loss": 0.8405,
      "step": 12920
    },
    {
      "epoch": 1.0118954452966036,
      "grad_norm": 2.0376780033111572,
      "learning_rate": 3.322304754930167e-05,
      "loss": 1.1386,
      "step": 12930
    },
    {
      "epoch": 1.0126780403819065,
      "grad_norm": 5.999090671539307,
      "learning_rate": 3.3209970183606216e-05,
      "loss": 0.8194,
      "step": 12940
    },
    {
      "epoch": 1.0134606354672093,
      "grad_norm": 6.736708641052246,
      "learning_rate": 3.319689281791076e-05,
      "loss": 0.6024,
      "step": 12950
    },
    {
      "epoch": 1.0142432305525122,
      "grad_norm": 8.859903335571289,
      "learning_rate": 3.3183815452215305e-05,
      "loss": 0.6854,
      "step": 12960
    },
    {
      "epoch": 1.015025825637815,
      "grad_norm": 5.0429205894470215,
      "learning_rate": 3.3170738086519857e-05,
      "loss": 0.9517,
      "step": 12970
    },
    {
      "epoch": 1.015808420723118,
      "grad_norm": 2.5933470726013184,
      "learning_rate": 3.3157660720824394e-05,
      "loss": 0.768,
      "step": 12980
    },
    {
      "epoch": 1.0165910158084208,
      "grad_norm": 7.679603576660156,
      "learning_rate": 3.3144583355128946e-05,
      "loss": 0.8891,
      "step": 12990
    },
    {
      "epoch": 1.0173736108937237,
      "grad_norm": 8.114143371582031,
      "learning_rate": 3.313150598943349e-05,
      "loss": 0.9916,
      "step": 13000
    },
    {
      "epoch": 1.0181562059790266,
      "grad_norm": 5.110938549041748,
      "learning_rate": 3.3118428623738035e-05,
      "loss": 0.5929,
      "step": 13010
    },
    {
      "epoch": 1.0189388010643292,
      "grad_norm": 17.189193725585938,
      "learning_rate": 3.3105351258042586e-05,
      "loss": 1.1916,
      "step": 13020
    },
    {
      "epoch": 1.019721396149632,
      "grad_norm": 1.5004600286483765,
      "learning_rate": 3.3092273892347124e-05,
      "loss": 0.7222,
      "step": 13030
    },
    {
      "epoch": 1.020503991234935,
      "grad_norm": 7.415332317352295,
      "learning_rate": 3.3079196526651675e-05,
      "loss": 1.1256,
      "step": 13040
    },
    {
      "epoch": 1.0212865863202378,
      "grad_norm": 7.4548115730285645,
      "learning_rate": 3.306611916095621e-05,
      "loss": 0.8343,
      "step": 13050
    },
    {
      "epoch": 1.0220691814055407,
      "grad_norm": 1.6195958852767944,
      "learning_rate": 3.3053041795260764e-05,
      "loss": 1.0715,
      "step": 13060
    },
    {
      "epoch": 1.0228517764908436,
      "grad_norm": 3.5696194171905518,
      "learning_rate": 3.303996442956531e-05,
      "loss": 0.9592,
      "step": 13070
    },
    {
      "epoch": 1.0236343715761465,
      "grad_norm": 5.654525279998779,
      "learning_rate": 3.3026887063869854e-05,
      "loss": 0.5307,
      "step": 13080
    },
    {
      "epoch": 1.0244169666614493,
      "grad_norm": 6.307244300842285,
      "learning_rate": 3.3013809698174405e-05,
      "loss": 0.7186,
      "step": 13090
    },
    {
      "epoch": 1.0251995617467522,
      "grad_norm": 10.143659591674805,
      "learning_rate": 3.300073233247894e-05,
      "loss": 0.7621,
      "step": 13100
    },
    {
      "epoch": 1.025982156832055,
      "grad_norm": 4.604301452636719,
      "learning_rate": 3.2987654966783494e-05,
      "loss": 0.5627,
      "step": 13110
    },
    {
      "epoch": 1.026764751917358,
      "grad_norm": 6.994559288024902,
      "learning_rate": 3.297457760108804e-05,
      "loss": 0.6414,
      "step": 13120
    },
    {
      "epoch": 1.0275473470026608,
      "grad_norm": 4.4918084144592285,
      "learning_rate": 3.296150023539258e-05,
      "loss": 0.8008,
      "step": 13130
    },
    {
      "epoch": 1.0283299420879637,
      "grad_norm": 18.3920955657959,
      "learning_rate": 3.2948422869697135e-05,
      "loss": 0.7741,
      "step": 13140
    },
    {
      "epoch": 1.0291125371732666,
      "grad_norm": 16.14258575439453,
      "learning_rate": 3.293534550400167e-05,
      "loss": 0.7839,
      "step": 13150
    },
    {
      "epoch": 1.0298951322585694,
      "grad_norm": 5.319972038269043,
      "learning_rate": 3.2922268138306224e-05,
      "loss": 0.7377,
      "step": 13160
    },
    {
      "epoch": 1.0306777273438723,
      "grad_norm": 5.142157554626465,
      "learning_rate": 3.290919077261077e-05,
      "loss": 0.5664,
      "step": 13170
    },
    {
      "epoch": 1.0314603224291752,
      "grad_norm": 19.907926559448242,
      "learning_rate": 3.289611340691531e-05,
      "loss": 0.7002,
      "step": 13180
    },
    {
      "epoch": 1.032242917514478,
      "grad_norm": 17.50787353515625,
      "learning_rate": 3.288303604121986e-05,
      "loss": 1.1241,
      "step": 13190
    },
    {
      "epoch": 1.033025512599781,
      "grad_norm": 8.171710014343262,
      "learning_rate": 3.28699586755244e-05,
      "loss": 0.8221,
      "step": 13200
    },
    {
      "epoch": 1.0338081076850838,
      "grad_norm": 9.997758865356445,
      "learning_rate": 3.285688130982895e-05,
      "loss": 0.8061,
      "step": 13210
    },
    {
      "epoch": 1.0345907027703867,
      "grad_norm": 6.855379581451416,
      "learning_rate": 3.284380394413349e-05,
      "loss": 0.8867,
      "step": 13220
    },
    {
      "epoch": 1.0353732978556895,
      "grad_norm": 20.468605041503906,
      "learning_rate": 3.283072657843804e-05,
      "loss": 0.772,
      "step": 13230
    },
    {
      "epoch": 1.0361558929409924,
      "grad_norm": 5.409748554229736,
      "learning_rate": 3.281764921274259e-05,
      "loss": 0.5416,
      "step": 13240
    },
    {
      "epoch": 1.0369384880262953,
      "grad_norm": 4.824619293212891,
      "learning_rate": 3.280457184704713e-05,
      "loss": 1.0409,
      "step": 13250
    },
    {
      "epoch": 1.0377210831115982,
      "grad_norm": 7.001192092895508,
      "learning_rate": 3.2791494481351676e-05,
      "loss": 0.7725,
      "step": 13260
    },
    {
      "epoch": 1.0385036781969008,
      "grad_norm": 2.1187210083007812,
      "learning_rate": 3.277841711565622e-05,
      "loss": 0.6184,
      "step": 13270
    },
    {
      "epoch": 1.0392862732822037,
      "grad_norm": 5.170557498931885,
      "learning_rate": 3.276533974996077e-05,
      "loss": 0.4598,
      "step": 13280
    },
    {
      "epoch": 1.0400688683675066,
      "grad_norm": 6.07156229019165,
      "learning_rate": 3.275226238426532e-05,
      "loss": 0.5997,
      "step": 13290
    },
    {
      "epoch": 1.0408514634528094,
      "grad_norm": 0.9804114103317261,
      "learning_rate": 3.273918501856986e-05,
      "loss": 0.7176,
      "step": 13300
    },
    {
      "epoch": 1.0416340585381123,
      "grad_norm": 8.109137535095215,
      "learning_rate": 3.2726107652874406e-05,
      "loss": 0.7626,
      "step": 13310
    },
    {
      "epoch": 1.0424166536234152,
      "grad_norm": 8.575441360473633,
      "learning_rate": 3.271303028717895e-05,
      "loss": 0.8509,
      "step": 13320
    },
    {
      "epoch": 1.043199248708718,
      "grad_norm": 3.0686333179473877,
      "learning_rate": 3.2699952921483495e-05,
      "loss": 0.6717,
      "step": 13330
    },
    {
      "epoch": 1.043981843794021,
      "grad_norm": 0.8798385262489319,
      "learning_rate": 3.2686875555788046e-05,
      "loss": 0.7805,
      "step": 13340
    },
    {
      "epoch": 1.0447644388793238,
      "grad_norm": 2.457923650741577,
      "learning_rate": 3.267379819009259e-05,
      "loss": 0.8093,
      "step": 13350
    },
    {
      "epoch": 1.0455470339646267,
      "grad_norm": 5.038333892822266,
      "learning_rate": 3.2660720824397135e-05,
      "loss": 0.8663,
      "step": 13360
    },
    {
      "epoch": 1.0463296290499295,
      "grad_norm": 6.735020637512207,
      "learning_rate": 3.264764345870168e-05,
      "loss": 0.6717,
      "step": 13370
    },
    {
      "epoch": 1.0471122241352324,
      "grad_norm": 2.642568349838257,
      "learning_rate": 3.2634566093006225e-05,
      "loss": 0.779,
      "step": 13380
    },
    {
      "epoch": 1.0478948192205353,
      "grad_norm": 0.869793176651001,
      "learning_rate": 3.2621488727310776e-05,
      "loss": 0.7028,
      "step": 13390
    },
    {
      "epoch": 1.0486774143058382,
      "grad_norm": 2.2345774173736572,
      "learning_rate": 3.2608411361615314e-05,
      "loss": 0.9041,
      "step": 13400
    },
    {
      "epoch": 1.049460009391141,
      "grad_norm": 0.8774744272232056,
      "learning_rate": 3.2595333995919865e-05,
      "loss": 0.7321,
      "step": 13410
    },
    {
      "epoch": 1.050242604476444,
      "grad_norm": 8.501220703125,
      "learning_rate": 3.258225663022441e-05,
      "loss": 0.9583,
      "step": 13420
    },
    {
      "epoch": 1.0510251995617468,
      "grad_norm": 4.635941505432129,
      "learning_rate": 3.2569179264528954e-05,
      "loss": 0.9406,
      "step": 13430
    },
    {
      "epoch": 1.0518077946470497,
      "grad_norm": 4.838067054748535,
      "learning_rate": 3.25561018988335e-05,
      "loss": 0.7253,
      "step": 13440
    },
    {
      "epoch": 1.0525903897323525,
      "grad_norm": 16.428955078125,
      "learning_rate": 3.2543024533138043e-05,
      "loss": 0.9371,
      "step": 13450
    },
    {
      "epoch": 1.0533729848176554,
      "grad_norm": 8.612907409667969,
      "learning_rate": 3.2529947167442595e-05,
      "loss": 0.7186,
      "step": 13460
    },
    {
      "epoch": 1.0541555799029583,
      "grad_norm": 12.15218448638916,
      "learning_rate": 3.251686980174713e-05,
      "loss": 0.5892,
      "step": 13470
    },
    {
      "epoch": 1.0549381749882611,
      "grad_norm": 19.347126007080078,
      "learning_rate": 3.2503792436051684e-05,
      "loss": 1.2021,
      "step": 13480
    },
    {
      "epoch": 1.055720770073564,
      "grad_norm": 2.4843411445617676,
      "learning_rate": 3.249071507035623e-05,
      "loss": 0.8404,
      "step": 13490
    },
    {
      "epoch": 1.056503365158867,
      "grad_norm": 3.1366217136383057,
      "learning_rate": 3.247763770466077e-05,
      "loss": 0.4325,
      "step": 13500
    },
    {
      "epoch": 1.0572859602441698,
      "grad_norm": 16.16063690185547,
      "learning_rate": 3.2464560338965324e-05,
      "loss": 1.0,
      "step": 13510
    },
    {
      "epoch": 1.0580685553294726,
      "grad_norm": 0.6046148538589478,
      "learning_rate": 3.245148297326986e-05,
      "loss": 0.4769,
      "step": 13520
    },
    {
      "epoch": 1.0588511504147753,
      "grad_norm": 2.006985664367676,
      "learning_rate": 3.2438405607574414e-05,
      "loss": 0.6498,
      "step": 13530
    },
    {
      "epoch": 1.0596337455000782,
      "grad_norm": 12.313105583190918,
      "learning_rate": 3.242532824187896e-05,
      "loss": 0.6483,
      "step": 13540
    },
    {
      "epoch": 1.060416340585381,
      "grad_norm": 10.8492431640625,
      "learning_rate": 3.24122508761835e-05,
      "loss": 0.7455,
      "step": 13550
    },
    {
      "epoch": 1.061198935670684,
      "grad_norm": 0.9784603118896484,
      "learning_rate": 3.2399173510488054e-05,
      "loss": 0.6869,
      "step": 13560
    },
    {
      "epoch": 1.0619815307559868,
      "grad_norm": 12.044229507446289,
      "learning_rate": 3.238609614479259e-05,
      "loss": 1.1638,
      "step": 13570
    },
    {
      "epoch": 1.0627641258412897,
      "grad_norm": 2.0092766284942627,
      "learning_rate": 3.237301877909714e-05,
      "loss": 0.8055,
      "step": 13580
    },
    {
      "epoch": 1.0635467209265925,
      "grad_norm": 10.826544761657715,
      "learning_rate": 3.235994141340168e-05,
      "loss": 0.6989,
      "step": 13590
    },
    {
      "epoch": 1.0643293160118954,
      "grad_norm": 9.704537391662598,
      "learning_rate": 3.234686404770623e-05,
      "loss": 0.7069,
      "step": 13600
    },
    {
      "epoch": 1.0651119110971983,
      "grad_norm": 1.6798218488693237,
      "learning_rate": 3.233378668201078e-05,
      "loss": 0.636,
      "step": 13610
    },
    {
      "epoch": 1.0658945061825011,
      "grad_norm": 9.630789756774902,
      "learning_rate": 3.232070931631532e-05,
      "loss": 0.6767,
      "step": 13620
    },
    {
      "epoch": 1.066677101267804,
      "grad_norm": 4.369169235229492,
      "learning_rate": 3.230763195061987e-05,
      "loss": 0.7982,
      "step": 13630
    },
    {
      "epoch": 1.067459696353107,
      "grad_norm": 5.251848220825195,
      "learning_rate": 3.229455458492441e-05,
      "loss": 0.8999,
      "step": 13640
    },
    {
      "epoch": 1.0682422914384098,
      "grad_norm": 1.469886064529419,
      "learning_rate": 3.228147721922896e-05,
      "loss": 0.4345,
      "step": 13650
    },
    {
      "epoch": 1.0690248865237126,
      "grad_norm": 3.6854872703552246,
      "learning_rate": 3.2268399853533507e-05,
      "loss": 0.6803,
      "step": 13660
    },
    {
      "epoch": 1.0698074816090155,
      "grad_norm": 6.131328582763672,
      "learning_rate": 3.225532248783805e-05,
      "loss": 0.5839,
      "step": 13670
    },
    {
      "epoch": 1.0705900766943184,
      "grad_norm": 6.766135215759277,
      "learning_rate": 3.2242245122142596e-05,
      "loss": 0.7147,
      "step": 13680
    },
    {
      "epoch": 1.0713726717796213,
      "grad_norm": 45.26825714111328,
      "learning_rate": 3.222916775644714e-05,
      "loss": 1.0832,
      "step": 13690
    },
    {
      "epoch": 1.0721552668649241,
      "grad_norm": 14.496328353881836,
      "learning_rate": 3.221609039075169e-05,
      "loss": 0.7184,
      "step": 13700
    },
    {
      "epoch": 1.072937861950227,
      "grad_norm": 10.388596534729004,
      "learning_rate": 3.2203013025056236e-05,
      "loss": 0.7608,
      "step": 13710
    },
    {
      "epoch": 1.0737204570355299,
      "grad_norm": 6.7745208740234375,
      "learning_rate": 3.218993565936078e-05,
      "loss": 0.7178,
      "step": 13720
    },
    {
      "epoch": 1.0745030521208327,
      "grad_norm": 5.867278099060059,
      "learning_rate": 3.2176858293665325e-05,
      "loss": 0.6861,
      "step": 13730
    },
    {
      "epoch": 1.0752856472061356,
      "grad_norm": 2.3179659843444824,
      "learning_rate": 3.216378092796987e-05,
      "loss": 0.7816,
      "step": 13740
    },
    {
      "epoch": 1.0760682422914385,
      "grad_norm": 1.0148940086364746,
      "learning_rate": 3.2150703562274414e-05,
      "loss": 0.5044,
      "step": 13750
    },
    {
      "epoch": 1.0768508373767414,
      "grad_norm": 7.125001907348633,
      "learning_rate": 3.2137626196578966e-05,
      "loss": 0.5552,
      "step": 13760
    },
    {
      "epoch": 1.0776334324620442,
      "grad_norm": 18.84972381591797,
      "learning_rate": 3.212454883088351e-05,
      "loss": 0.6494,
      "step": 13770
    },
    {
      "epoch": 1.078416027547347,
      "grad_norm": 4.548697471618652,
      "learning_rate": 3.2111471465188055e-05,
      "loss": 0.879,
      "step": 13780
    },
    {
      "epoch": 1.0791986226326498,
      "grad_norm": 8.818333625793457,
      "learning_rate": 3.20983940994926e-05,
      "loss": 0.6301,
      "step": 13790
    },
    {
      "epoch": 1.0799812177179526,
      "grad_norm": 6.193119525909424,
      "learning_rate": 3.2085316733797144e-05,
      "loss": 0.8381,
      "step": 13800
    },
    {
      "epoch": 1.0807638128032555,
      "grad_norm": 1.0661730766296387,
      "learning_rate": 3.207223936810169e-05,
      "loss": 0.9507,
      "step": 13810
    },
    {
      "epoch": 1.0815464078885584,
      "grad_norm": 22.62188148498535,
      "learning_rate": 3.205916200240623e-05,
      "loss": 0.9038,
      "step": 13820
    },
    {
      "epoch": 1.0823290029738613,
      "grad_norm": 2.1725914478302,
      "learning_rate": 3.2046084636710785e-05,
      "loss": 0.895,
      "step": 13830
    },
    {
      "epoch": 1.0831115980591641,
      "grad_norm": 8.905023574829102,
      "learning_rate": 3.203300727101533e-05,
      "loss": 0.7426,
      "step": 13840
    },
    {
      "epoch": 1.083894193144467,
      "grad_norm": 10.994120597839355,
      "learning_rate": 3.2019929905319874e-05,
      "loss": 0.7086,
      "step": 13850
    },
    {
      "epoch": 1.0846767882297699,
      "grad_norm": 18.242061614990234,
      "learning_rate": 3.200685253962442e-05,
      "loss": 0.9795,
      "step": 13860
    },
    {
      "epoch": 1.0854593833150727,
      "grad_norm": 10.723325729370117,
      "learning_rate": 3.199377517392896e-05,
      "loss": 0.6835,
      "step": 13870
    },
    {
      "epoch": 1.0862419784003756,
      "grad_norm": 11.821295738220215,
      "learning_rate": 3.1980697808233514e-05,
      "loss": 0.8522,
      "step": 13880
    },
    {
      "epoch": 1.0870245734856785,
      "grad_norm": 7.602274417877197,
      "learning_rate": 3.196762044253805e-05,
      "loss": 0.5868,
      "step": 13890
    },
    {
      "epoch": 1.0878071685709814,
      "grad_norm": 6.295817852020264,
      "learning_rate": 3.19545430768426e-05,
      "loss": 0.7021,
      "step": 13900
    },
    {
      "epoch": 1.0885897636562842,
      "grad_norm": 4.081143856048584,
      "learning_rate": 3.194146571114715e-05,
      "loss": 0.7545,
      "step": 13910
    },
    {
      "epoch": 1.089372358741587,
      "grad_norm": 14.858965873718262,
      "learning_rate": 3.192838834545169e-05,
      "loss": 0.8806,
      "step": 13920
    },
    {
      "epoch": 1.09015495382689,
      "grad_norm": 7.7110137939453125,
      "learning_rate": 3.1915310979756244e-05,
      "loss": 0.5628,
      "step": 13930
    },
    {
      "epoch": 1.0909375489121929,
      "grad_norm": 35.933982849121094,
      "learning_rate": 3.190223361406078e-05,
      "loss": 0.51,
      "step": 13940
    },
    {
      "epoch": 1.0917201439974957,
      "grad_norm": 9.242915153503418,
      "learning_rate": 3.188915624836533e-05,
      "loss": 1.2348,
      "step": 13950
    },
    {
      "epoch": 1.0925027390827986,
      "grad_norm": 1.3667802810668945,
      "learning_rate": 3.187607888266987e-05,
      "loss": 0.7601,
      "step": 13960
    },
    {
      "epoch": 1.0932853341681015,
      "grad_norm": 18.395681381225586,
      "learning_rate": 3.186300151697442e-05,
      "loss": 0.5988,
      "step": 13970
    },
    {
      "epoch": 1.0940679292534043,
      "grad_norm": 0.6689698696136475,
      "learning_rate": 3.1849924151278974e-05,
      "loss": 0.5384,
      "step": 13980
    },
    {
      "epoch": 1.0948505243387072,
      "grad_norm": 8.682588577270508,
      "learning_rate": 3.183684678558351e-05,
      "loss": 0.8649,
      "step": 13990
    },
    {
      "epoch": 1.09563311942401,
      "grad_norm": 12.447678565979004,
      "learning_rate": 3.182376941988806e-05,
      "loss": 1.2047,
      "step": 14000
    },
    {
      "epoch": 1.096415714509313,
      "grad_norm": 1.6601954698562622,
      "learning_rate": 3.18106920541926e-05,
      "loss": 0.793,
      "step": 14010
    },
    {
      "epoch": 1.0971983095946158,
      "grad_norm": 8.750977516174316,
      "learning_rate": 3.179761468849715e-05,
      "loss": 0.843,
      "step": 14020
    },
    {
      "epoch": 1.0979809046799187,
      "grad_norm": 7.0161895751953125,
      "learning_rate": 3.1784537322801696e-05,
      "loss": 0.8807,
      "step": 14030
    },
    {
      "epoch": 1.0987634997652216,
      "grad_norm": 11.845017433166504,
      "learning_rate": 3.177145995710624e-05,
      "loss": 0.9645,
      "step": 14040
    },
    {
      "epoch": 1.0995460948505242,
      "grad_norm": 2.5921998023986816,
      "learning_rate": 3.175838259141079e-05,
      "loss": 1.0479,
      "step": 14050
    },
    {
      "epoch": 1.100328689935827,
      "grad_norm": 6.621883392333984,
      "learning_rate": 3.174530522571533e-05,
      "loss": 0.6658,
      "step": 14060
    },
    {
      "epoch": 1.10111128502113,
      "grad_norm": 2.5733375549316406,
      "learning_rate": 3.173222786001988e-05,
      "loss": 0.606,
      "step": 14070
    },
    {
      "epoch": 1.1018938801064329,
      "grad_norm": 18.924177169799805,
      "learning_rate": 3.1719150494324426e-05,
      "loss": 0.8531,
      "step": 14080
    },
    {
      "epoch": 1.1026764751917357,
      "grad_norm": 15.449837684631348,
      "learning_rate": 3.170607312862897e-05,
      "loss": 0.9055,
      "step": 14090
    },
    {
      "epoch": 1.1034590702770386,
      "grad_norm": 13.931599617004395,
      "learning_rate": 3.1692995762933515e-05,
      "loss": 0.8854,
      "step": 14100
    },
    {
      "epoch": 1.1042416653623415,
      "grad_norm": 14.948174476623535,
      "learning_rate": 3.167991839723806e-05,
      "loss": 1.2732,
      "step": 14110
    },
    {
      "epoch": 1.1050242604476443,
      "grad_norm": 4.089398384094238,
      "learning_rate": 3.166684103154261e-05,
      "loss": 0.745,
      "step": 14120
    },
    {
      "epoch": 1.1058068555329472,
      "grad_norm": 4.59906530380249,
      "learning_rate": 3.1653763665847156e-05,
      "loss": 0.5663,
      "step": 14130
    },
    {
      "epoch": 1.10658945061825,
      "grad_norm": 3.201033115386963,
      "learning_rate": 3.16406863001517e-05,
      "loss": 0.6342,
      "step": 14140
    },
    {
      "epoch": 1.107372045703553,
      "grad_norm": 6.922934055328369,
      "learning_rate": 3.1627608934456245e-05,
      "loss": 0.6389,
      "step": 14150
    },
    {
      "epoch": 1.1081546407888558,
      "grad_norm": 13.867130279541016,
      "learning_rate": 3.161453156876079e-05,
      "loss": 0.7221,
      "step": 14160
    },
    {
      "epoch": 1.1089372358741587,
      "grad_norm": 9.951213836669922,
      "learning_rate": 3.1601454203065334e-05,
      "loss": 0.6794,
      "step": 14170
    },
    {
      "epoch": 1.1097198309594616,
      "grad_norm": 16.712305068969727,
      "learning_rate": 3.158837683736988e-05,
      "loss": 0.5178,
      "step": 14180
    },
    {
      "epoch": 1.1105024260447645,
      "grad_norm": 13.532806396484375,
      "learning_rate": 3.157529947167443e-05,
      "loss": 1.1808,
      "step": 14190
    },
    {
      "epoch": 1.1112850211300673,
      "grad_norm": 1.8434150218963623,
      "learning_rate": 3.1562222105978974e-05,
      "loss": 0.4419,
      "step": 14200
    },
    {
      "epoch": 1.1120676162153702,
      "grad_norm": 7.14523983001709,
      "learning_rate": 3.154914474028352e-05,
      "loss": 1.2017,
      "step": 14210
    },
    {
      "epoch": 1.112850211300673,
      "grad_norm": 4.471436500549316,
      "learning_rate": 3.1536067374588064e-05,
      "loss": 0.6325,
      "step": 14220
    },
    {
      "epoch": 1.113632806385976,
      "grad_norm": 4.067905902862549,
      "learning_rate": 3.152299000889261e-05,
      "loss": 0.9524,
      "step": 14230
    },
    {
      "epoch": 1.1144154014712788,
      "grad_norm": 11.119894981384277,
      "learning_rate": 3.150991264319715e-05,
      "loss": 0.9086,
      "step": 14240
    },
    {
      "epoch": 1.1151979965565817,
      "grad_norm": 4.083450794219971,
      "learning_rate": 3.1496835277501704e-05,
      "loss": 0.8188,
      "step": 14250
    },
    {
      "epoch": 1.1159805916418846,
      "grad_norm": 8.468968391418457,
      "learning_rate": 3.148375791180625e-05,
      "loss": 1.2904,
      "step": 14260
    },
    {
      "epoch": 1.1167631867271874,
      "grad_norm": 5.648281574249268,
      "learning_rate": 3.147068054611079e-05,
      "loss": 0.5586,
      "step": 14270
    },
    {
      "epoch": 1.1175457818124903,
      "grad_norm": 6.402794361114502,
      "learning_rate": 3.145760318041534e-05,
      "loss": 1.2292,
      "step": 14280
    },
    {
      "epoch": 1.118328376897793,
      "grad_norm": 7.014227390289307,
      "learning_rate": 3.144452581471988e-05,
      "loss": 0.8994,
      "step": 14290
    },
    {
      "epoch": 1.1191109719830958,
      "grad_norm": 4.1913580894470215,
      "learning_rate": 3.1431448449024434e-05,
      "loss": 0.5994,
      "step": 14300
    },
    {
      "epoch": 1.1198935670683987,
      "grad_norm": 12.3535795211792,
      "learning_rate": 3.141837108332897e-05,
      "loss": 0.915,
      "step": 14310
    },
    {
      "epoch": 1.1206761621537016,
      "grad_norm": 13.684656143188477,
      "learning_rate": 3.140529371763352e-05,
      "loss": 0.8126,
      "step": 14320
    },
    {
      "epoch": 1.1214587572390045,
      "grad_norm": 6.796149730682373,
      "learning_rate": 3.139221635193807e-05,
      "loss": 0.8701,
      "step": 14330
    },
    {
      "epoch": 1.1222413523243073,
      "grad_norm": 6.6552300453186035,
      "learning_rate": 3.137913898624261e-05,
      "loss": 0.8017,
      "step": 14340
    },
    {
      "epoch": 1.1230239474096102,
      "grad_norm": 9.002670288085938,
      "learning_rate": 3.136606162054716e-05,
      "loss": 0.6583,
      "step": 14350
    },
    {
      "epoch": 1.123806542494913,
      "grad_norm": 4.361416816711426,
      "learning_rate": 3.13529842548517e-05,
      "loss": 0.8087,
      "step": 14360
    },
    {
      "epoch": 1.124589137580216,
      "grad_norm": 6.140327453613281,
      "learning_rate": 3.133990688915625e-05,
      "loss": 0.7438,
      "step": 14370
    },
    {
      "epoch": 1.1253717326655188,
      "grad_norm": 4.343984127044678,
      "learning_rate": 3.132682952346079e-05,
      "loss": 1.0177,
      "step": 14380
    },
    {
      "epoch": 1.1261543277508217,
      "grad_norm": 2.5007712841033936,
      "learning_rate": 3.131375215776534e-05,
      "loss": 0.5851,
      "step": 14390
    },
    {
      "epoch": 1.1269369228361246,
      "grad_norm": 8.695201873779297,
      "learning_rate": 3.1300674792069886e-05,
      "loss": 1.11,
      "step": 14400
    },
    {
      "epoch": 1.1277195179214274,
      "grad_norm": 2.1847329139709473,
      "learning_rate": 3.128759742637443e-05,
      "loss": 0.785,
      "step": 14410
    },
    {
      "epoch": 1.1285021130067303,
      "grad_norm": 20.139345169067383,
      "learning_rate": 3.127452006067898e-05,
      "loss": 0.8009,
      "step": 14420
    },
    {
      "epoch": 1.1292847080920332,
      "grad_norm": 6.507087230682373,
      "learning_rate": 3.126144269498352e-05,
      "loss": 0.7756,
      "step": 14430
    },
    {
      "epoch": 1.130067303177336,
      "grad_norm": 7.025235176086426,
      "learning_rate": 3.124836532928807e-05,
      "loss": 0.9252,
      "step": 14440
    },
    {
      "epoch": 1.130849898262639,
      "grad_norm": 9.239069938659668,
      "learning_rate": 3.1235287963592616e-05,
      "loss": 0.6674,
      "step": 14450
    },
    {
      "epoch": 1.1316324933479418,
      "grad_norm": 7.568877696990967,
      "learning_rate": 3.122221059789716e-05,
      "loss": 0.7837,
      "step": 14460
    },
    {
      "epoch": 1.1324150884332447,
      "grad_norm": 4.107082843780518,
      "learning_rate": 3.120913323220171e-05,
      "loss": 0.5976,
      "step": 14470
    },
    {
      "epoch": 1.1331976835185475,
      "grad_norm": 4.798367023468018,
      "learning_rate": 3.119605586650625e-05,
      "loss": 0.6167,
      "step": 14480
    },
    {
      "epoch": 1.1339802786038504,
      "grad_norm": 3.988173246383667,
      "learning_rate": 3.11829785008108e-05,
      "loss": 0.7124,
      "step": 14490
    },
    {
      "epoch": 1.1347628736891533,
      "grad_norm": 5.218972682952881,
      "learning_rate": 3.1169901135115345e-05,
      "loss": 0.8399,
      "step": 14500
    },
    {
      "epoch": 1.1355454687744562,
      "grad_norm": 5.297395706176758,
      "learning_rate": 3.115682376941989e-05,
      "loss": 0.7088,
      "step": 14510
    },
    {
      "epoch": 1.136328063859759,
      "grad_norm": 18.102771759033203,
      "learning_rate": 3.1143746403724435e-05,
      "loss": 0.8221,
      "step": 14520
    },
    {
      "epoch": 1.137110658945062,
      "grad_norm": 1.68930184841156,
      "learning_rate": 3.113066903802898e-05,
      "loss": 0.5609,
      "step": 14530
    },
    {
      "epoch": 1.1378932540303648,
      "grad_norm": 18.196561813354492,
      "learning_rate": 3.111759167233353e-05,
      "loss": 0.6461,
      "step": 14540
    },
    {
      "epoch": 1.1386758491156677,
      "grad_norm": 2.043034315109253,
      "learning_rate": 3.1104514306638075e-05,
      "loss": 0.7994,
      "step": 14550
    },
    {
      "epoch": 1.1394584442009705,
      "grad_norm": 16.53839683532715,
      "learning_rate": 3.109143694094262e-05,
      "loss": 1.0608,
      "step": 14560
    },
    {
      "epoch": 1.1402410392862734,
      "grad_norm": 0.6766161322593689,
      "learning_rate": 3.1078359575247164e-05,
      "loss": 0.746,
      "step": 14570
    },
    {
      "epoch": 1.141023634371576,
      "grad_norm": 4.460118770599365,
      "learning_rate": 3.106528220955171e-05,
      "loss": 0.7765,
      "step": 14580
    },
    {
      "epoch": 1.141806229456879,
      "grad_norm": 7.417885780334473,
      "learning_rate": 3.1052204843856253e-05,
      "loss": 1.1742,
      "step": 14590
    },
    {
      "epoch": 1.1425888245421818,
      "grad_norm": 3.4931159019470215,
      "learning_rate": 3.10391274781608e-05,
      "loss": 0.961,
      "step": 14600
    },
    {
      "epoch": 1.1433714196274847,
      "grad_norm": 7.992660045623779,
      "learning_rate": 3.102605011246535e-05,
      "loss": 0.7698,
      "step": 14610
    },
    {
      "epoch": 1.1441540147127875,
      "grad_norm": 6.174659252166748,
      "learning_rate": 3.1012972746769894e-05,
      "loss": 0.7211,
      "step": 14620
    },
    {
      "epoch": 1.1449366097980904,
      "grad_norm": 2.3168723583221436,
      "learning_rate": 3.099989538107444e-05,
      "loss": 0.7665,
      "step": 14630
    },
    {
      "epoch": 1.1457192048833933,
      "grad_norm": 5.7286763191223145,
      "learning_rate": 3.098681801537898e-05,
      "loss": 0.6396,
      "step": 14640
    },
    {
      "epoch": 1.1465017999686962,
      "grad_norm": 7.80592155456543,
      "learning_rate": 3.097374064968353e-05,
      "loss": 0.8088,
      "step": 14650
    },
    {
      "epoch": 1.147284395053999,
      "grad_norm": 6.175577163696289,
      "learning_rate": 3.096066328398807e-05,
      "loss": 0.7215,
      "step": 14660
    },
    {
      "epoch": 1.148066990139302,
      "grad_norm": 9.497570991516113,
      "learning_rate": 3.0947585918292624e-05,
      "loss": 0.728,
      "step": 14670
    },
    {
      "epoch": 1.1488495852246048,
      "grad_norm": 5.021395683288574,
      "learning_rate": 3.093450855259717e-05,
      "loss": 0.7622,
      "step": 14680
    },
    {
      "epoch": 1.1496321803099077,
      "grad_norm": 4.741960048675537,
      "learning_rate": 3.092143118690171e-05,
      "loss": 0.6365,
      "step": 14690
    },
    {
      "epoch": 1.1504147753952105,
      "grad_norm": 7.776442527770996,
      "learning_rate": 3.090835382120626e-05,
      "loss": 0.8865,
      "step": 14700
    },
    {
      "epoch": 1.1511973704805134,
      "grad_norm": 6.464144706726074,
      "learning_rate": 3.08952764555108e-05,
      "loss": 0.7219,
      "step": 14710
    },
    {
      "epoch": 1.1519799655658163,
      "grad_norm": 1.8215070962905884,
      "learning_rate": 3.088219908981535e-05,
      "loss": 0.6076,
      "step": 14720
    },
    {
      "epoch": 1.1527625606511192,
      "grad_norm": 24.33786964416504,
      "learning_rate": 3.086912172411989e-05,
      "loss": 0.9094,
      "step": 14730
    },
    {
      "epoch": 1.153545155736422,
      "grad_norm": 2.469613552093506,
      "learning_rate": 3.085604435842444e-05,
      "loss": 0.8243,
      "step": 14740
    },
    {
      "epoch": 1.154327750821725,
      "grad_norm": 1.3821479082107544,
      "learning_rate": 3.084296699272899e-05,
      "loss": 0.8277,
      "step": 14750
    },
    {
      "epoch": 1.1551103459070278,
      "grad_norm": 10.650533676147461,
      "learning_rate": 3.082988962703353e-05,
      "loss": 0.8354,
      "step": 14760
    },
    {
      "epoch": 1.1558929409923306,
      "grad_norm": 19.54239845275879,
      "learning_rate": 3.0816812261338076e-05,
      "loss": 0.9261,
      "step": 14770
    },
    {
      "epoch": 1.1566755360776335,
      "grad_norm": 3.0714704990386963,
      "learning_rate": 3.080373489564262e-05,
      "loss": 1.2016,
      "step": 14780
    },
    {
      "epoch": 1.1574581311629364,
      "grad_norm": 4.848270893096924,
      "learning_rate": 3.079065752994717e-05,
      "loss": 1.0701,
      "step": 14790
    },
    {
      "epoch": 1.158240726248239,
      "grad_norm": 4.5985107421875,
      "learning_rate": 3.077758016425171e-05,
      "loss": 0.6659,
      "step": 14800
    },
    {
      "epoch": 1.159023321333542,
      "grad_norm": 1.410330057144165,
      "learning_rate": 3.076450279855626e-05,
      "loss": 0.8809,
      "step": 14810
    },
    {
      "epoch": 1.1598059164188448,
      "grad_norm": 7.233081340789795,
      "learning_rate": 3.0751425432860806e-05,
      "loss": 1.0286,
      "step": 14820
    },
    {
      "epoch": 1.1605885115041477,
      "grad_norm": 3.732213258743286,
      "learning_rate": 3.073834806716535e-05,
      "loss": 0.6064,
      "step": 14830
    },
    {
      "epoch": 1.1613711065894505,
      "grad_norm": 14.53012466430664,
      "learning_rate": 3.07252707014699e-05,
      "loss": 0.7007,
      "step": 14840
    },
    {
      "epoch": 1.1621537016747534,
      "grad_norm": 8.245583534240723,
      "learning_rate": 3.071219333577444e-05,
      "loss": 1.1521,
      "step": 14850
    },
    {
      "epoch": 1.1629362967600563,
      "grad_norm": 1.4045220613479614,
      "learning_rate": 3.069911597007899e-05,
      "loss": 0.7193,
      "step": 14860
    },
    {
      "epoch": 1.1637188918453591,
      "grad_norm": 2.660977602005005,
      "learning_rate": 3.0686038604383535e-05,
      "loss": 0.8807,
      "step": 14870
    },
    {
      "epoch": 1.164501486930662,
      "grad_norm": 1.9380470514297485,
      "learning_rate": 3.067296123868808e-05,
      "loss": 0.374,
      "step": 14880
    },
    {
      "epoch": 1.165284082015965,
      "grad_norm": 7.9445576667785645,
      "learning_rate": 3.065988387299263e-05,
      "loss": 0.8137,
      "step": 14890
    },
    {
      "epoch": 1.1660666771012678,
      "grad_norm": 4.407632827758789,
      "learning_rate": 3.064680650729717e-05,
      "loss": 1.1445,
      "step": 14900
    },
    {
      "epoch": 1.1668492721865706,
      "grad_norm": 2.805987596511841,
      "learning_rate": 3.063372914160172e-05,
      "loss": 0.8166,
      "step": 14910
    },
    {
      "epoch": 1.1676318672718735,
      "grad_norm": 3.975192070007324,
      "learning_rate": 3.0620651775906265e-05,
      "loss": 0.8461,
      "step": 14920
    },
    {
      "epoch": 1.1684144623571764,
      "grad_norm": 4.681769371032715,
      "learning_rate": 3.060757441021081e-05,
      "loss": 0.5692,
      "step": 14930
    },
    {
      "epoch": 1.1691970574424793,
      "grad_norm": 4.448126316070557,
      "learning_rate": 3.0594497044515354e-05,
      "loss": 0.7068,
      "step": 14940
    },
    {
      "epoch": 1.1699796525277821,
      "grad_norm": 17.486492156982422,
      "learning_rate": 3.05814196788199e-05,
      "loss": 0.9881,
      "step": 14950
    },
    {
      "epoch": 1.170762247613085,
      "grad_norm": 11.61962604522705,
      "learning_rate": 3.056834231312445e-05,
      "loss": 0.631,
      "step": 14960
    },
    {
      "epoch": 1.1715448426983879,
      "grad_norm": 5.153378486633301,
      "learning_rate": 3.055526494742899e-05,
      "loss": 0.604,
      "step": 14970
    },
    {
      "epoch": 1.1723274377836908,
      "grad_norm": 9.415731430053711,
      "learning_rate": 3.054218758173354e-05,
      "loss": 0.7575,
      "step": 14980
    },
    {
      "epoch": 1.1731100328689936,
      "grad_norm": 10.82515811920166,
      "learning_rate": 3.0529110216038084e-05,
      "loss": 0.57,
      "step": 14990
    },
    {
      "epoch": 1.1738926279542965,
      "grad_norm": 33.8817024230957,
      "learning_rate": 3.0516032850342628e-05,
      "loss": 0.6233,
      "step": 15000
    },
    {
      "epoch": 1.1746752230395994,
      "grad_norm": 11.555156707763672,
      "learning_rate": 3.0502955484647173e-05,
      "loss": 0.7676,
      "step": 15010
    },
    {
      "epoch": 1.1754578181249022,
      "grad_norm": 10.734020233154297,
      "learning_rate": 3.048987811895172e-05,
      "loss": 0.8358,
      "step": 15020
    },
    {
      "epoch": 1.1762404132102051,
      "grad_norm": 1.0914133787155151,
      "learning_rate": 3.047680075325627e-05,
      "loss": 1.1203,
      "step": 15030
    },
    {
      "epoch": 1.177023008295508,
      "grad_norm": 6.945030689239502,
      "learning_rate": 3.046372338756081e-05,
      "loss": 0.9614,
      "step": 15040
    },
    {
      "epoch": 1.1778056033808109,
      "grad_norm": 9.829622268676758,
      "learning_rate": 3.0450646021865358e-05,
      "loss": 0.6683,
      "step": 15050
    },
    {
      "epoch": 1.1785881984661137,
      "grad_norm": 0.9769691824913025,
      "learning_rate": 3.0437568656169903e-05,
      "loss": 0.7411,
      "step": 15060
    },
    {
      "epoch": 1.1793707935514166,
      "grad_norm": 1.7882674932479858,
      "learning_rate": 3.042449129047445e-05,
      "loss": 0.6273,
      "step": 15070
    },
    {
      "epoch": 1.1801533886367195,
      "grad_norm": 1.0762916803359985,
      "learning_rate": 3.041141392477899e-05,
      "loss": 0.5242,
      "step": 15080
    },
    {
      "epoch": 1.1809359837220221,
      "grad_norm": 7.706854820251465,
      "learning_rate": 3.039833655908354e-05,
      "loss": 0.8449,
      "step": 15090
    },
    {
      "epoch": 1.181718578807325,
      "grad_norm": 7.555969715118408,
      "learning_rate": 3.0385259193388088e-05,
      "loss": 1.0046,
      "step": 15100
    },
    {
      "epoch": 1.1825011738926279,
      "grad_norm": 28.185285568237305,
      "learning_rate": 3.0372181827692632e-05,
      "loss": 0.7834,
      "step": 15110
    },
    {
      "epoch": 1.1832837689779307,
      "grad_norm": 13.14148235321045,
      "learning_rate": 3.035910446199718e-05,
      "loss": 0.5621,
      "step": 15120
    },
    {
      "epoch": 1.1840663640632336,
      "grad_norm": 3.406726121902466,
      "learning_rate": 3.034602709630172e-05,
      "loss": 0.9244,
      "step": 15130
    },
    {
      "epoch": 1.1848489591485365,
      "grad_norm": 7.162285327911377,
      "learning_rate": 3.033294973060627e-05,
      "loss": 1.0609,
      "step": 15140
    },
    {
      "epoch": 1.1856315542338394,
      "grad_norm": 2.1518702507019043,
      "learning_rate": 3.031987236491081e-05,
      "loss": 1.0664,
      "step": 15150
    },
    {
      "epoch": 1.1864141493191422,
      "grad_norm": 0.40068504214286804,
      "learning_rate": 3.030679499921536e-05,
      "loss": 0.5585,
      "step": 15160
    },
    {
      "epoch": 1.1871967444044451,
      "grad_norm": 1.528444528579712,
      "learning_rate": 3.0293717633519903e-05,
      "loss": 0.6668,
      "step": 15170
    },
    {
      "epoch": 1.187979339489748,
      "grad_norm": 12.458999633789062,
      "learning_rate": 3.028064026782445e-05,
      "loss": 0.8954,
      "step": 15180
    },
    {
      "epoch": 1.1887619345750509,
      "grad_norm": 19.069040298461914,
      "learning_rate": 3.0267562902129e-05,
      "loss": 0.654,
      "step": 15190
    },
    {
      "epoch": 1.1895445296603537,
      "grad_norm": 4.577025413513184,
      "learning_rate": 3.025448553643354e-05,
      "loss": 0.8035,
      "step": 15200
    },
    {
      "epoch": 1.1903271247456566,
      "grad_norm": 9.983968734741211,
      "learning_rate": 3.0241408170738088e-05,
      "loss": 0.5476,
      "step": 15210
    },
    {
      "epoch": 1.1911097198309595,
      "grad_norm": 6.77061128616333,
      "learning_rate": 3.0228330805042633e-05,
      "loss": 0.9863,
      "step": 15220
    },
    {
      "epoch": 1.1918923149162624,
      "grad_norm": 9.639410972595215,
      "learning_rate": 3.021525343934718e-05,
      "loss": 0.7936,
      "step": 15230
    },
    {
      "epoch": 1.1926749100015652,
      "grad_norm": 9.970869064331055,
      "learning_rate": 3.0202176073651722e-05,
      "loss": 0.7635,
      "step": 15240
    },
    {
      "epoch": 1.193457505086868,
      "grad_norm": 14.093001365661621,
      "learning_rate": 3.018909870795627e-05,
      "loss": 0.9769,
      "step": 15250
    },
    {
      "epoch": 1.194240100172171,
      "grad_norm": 9.298399925231934,
      "learning_rate": 3.0176021342260818e-05,
      "loss": 0.6687,
      "step": 15260
    },
    {
      "epoch": 1.1950226952574738,
      "grad_norm": 9.80194091796875,
      "learning_rate": 3.0162943976565362e-05,
      "loss": 1.1069,
      "step": 15270
    },
    {
      "epoch": 1.1958052903427767,
      "grad_norm": 2.5877583026885986,
      "learning_rate": 3.014986661086991e-05,
      "loss": 0.8541,
      "step": 15280
    },
    {
      "epoch": 1.1965878854280796,
      "grad_norm": 8.11124324798584,
      "learning_rate": 3.013678924517445e-05,
      "loss": 0.7006,
      "step": 15290
    },
    {
      "epoch": 1.1973704805133825,
      "grad_norm": 5.729658126831055,
      "learning_rate": 3.0123711879479e-05,
      "loss": 0.4609,
      "step": 15300
    },
    {
      "epoch": 1.1981530755986853,
      "grad_norm": 0.6629953980445862,
      "learning_rate": 3.011063451378354e-05,
      "loss": 0.5956,
      "step": 15310
    },
    {
      "epoch": 1.198935670683988,
      "grad_norm": 5.032694339752197,
      "learning_rate": 3.0097557148088092e-05,
      "loss": 0.7707,
      "step": 15320
    },
    {
      "epoch": 1.1997182657692909,
      "grad_norm": 0.7197624444961548,
      "learning_rate": 3.008447978239264e-05,
      "loss": 0.6185,
      "step": 15330
    },
    {
      "epoch": 1.2005008608545937,
      "grad_norm": 7.651124954223633,
      "learning_rate": 3.007140241669718e-05,
      "loss": 0.9271,
      "step": 15340
    },
    {
      "epoch": 1.2012834559398966,
      "grad_norm": 14.596146583557129,
      "learning_rate": 3.005832505100173e-05,
      "loss": 1.2834,
      "step": 15350
    },
    {
      "epoch": 1.2020660510251995,
      "grad_norm": 11.155280113220215,
      "learning_rate": 3.004524768530627e-05,
      "loss": 0.8819,
      "step": 15360
    },
    {
      "epoch": 1.2028486461105024,
      "grad_norm": 13.321341514587402,
      "learning_rate": 3.0032170319610818e-05,
      "loss": 0.5221,
      "step": 15370
    },
    {
      "epoch": 1.2036312411958052,
      "grad_norm": 6.941053867340088,
      "learning_rate": 3.0019092953915363e-05,
      "loss": 0.4997,
      "step": 15380
    },
    {
      "epoch": 1.204413836281108,
      "grad_norm": 8.819761276245117,
      "learning_rate": 3.000601558821991e-05,
      "loss": 0.7262,
      "step": 15390
    },
    {
      "epoch": 1.205196431366411,
      "grad_norm": 5.51768159866333,
      "learning_rate": 2.999293822252446e-05,
      "loss": 0.7409,
      "step": 15400
    },
    {
      "epoch": 1.2059790264517138,
      "grad_norm": 1.2082017660140991,
      "learning_rate": 2.9979860856829e-05,
      "loss": 0.914,
      "step": 15410
    },
    {
      "epoch": 1.2067616215370167,
      "grad_norm": 1.9012688398361206,
      "learning_rate": 2.9966783491133548e-05,
      "loss": 0.5661,
      "step": 15420
    },
    {
      "epoch": 1.2075442166223196,
      "grad_norm": 8.125970840454102,
      "learning_rate": 2.9953706125438092e-05,
      "loss": 0.6176,
      "step": 15430
    },
    {
      "epoch": 1.2083268117076225,
      "grad_norm": 9.019189834594727,
      "learning_rate": 2.994062875974264e-05,
      "loss": 0.6321,
      "step": 15440
    },
    {
      "epoch": 1.2091094067929253,
      "grad_norm": 6.978433132171631,
      "learning_rate": 2.992755139404718e-05,
      "loss": 0.8323,
      "step": 15450
    },
    {
      "epoch": 1.2098920018782282,
      "grad_norm": 8.657017707824707,
      "learning_rate": 2.991447402835173e-05,
      "loss": 0.5667,
      "step": 15460
    },
    {
      "epoch": 1.210674596963531,
      "grad_norm": 5.395399570465088,
      "learning_rate": 2.9901396662656277e-05,
      "loss": 0.9166,
      "step": 15470
    },
    {
      "epoch": 1.211457192048834,
      "grad_norm": 9.849907875061035,
      "learning_rate": 2.9888319296960822e-05,
      "loss": 0.9587,
      "step": 15480
    },
    {
      "epoch": 1.2122397871341368,
      "grad_norm": 4.676992893218994,
      "learning_rate": 2.987524193126537e-05,
      "loss": 0.7543,
      "step": 15490
    },
    {
      "epoch": 1.2130223822194397,
      "grad_norm": 2.3024911880493164,
      "learning_rate": 2.986216456556991e-05,
      "loss": 1.1768,
      "step": 15500
    },
    {
      "epoch": 1.2138049773047426,
      "grad_norm": 2.500763416290283,
      "learning_rate": 2.984908719987446e-05,
      "loss": 0.3537,
      "step": 15510
    },
    {
      "epoch": 1.2145875723900454,
      "grad_norm": 4.894447326660156,
      "learning_rate": 2.9836009834179e-05,
      "loss": 0.9143,
      "step": 15520
    },
    {
      "epoch": 1.2153701674753483,
      "grad_norm": 6.812446117401123,
      "learning_rate": 2.9822932468483548e-05,
      "loss": 0.9146,
      "step": 15530
    },
    {
      "epoch": 1.2161527625606512,
      "grad_norm": 5.78674840927124,
      "learning_rate": 2.98098551027881e-05,
      "loss": 0.8085,
      "step": 15540
    },
    {
      "epoch": 1.216935357645954,
      "grad_norm": 9.266759872436523,
      "learning_rate": 2.979677773709264e-05,
      "loss": 1.0687,
      "step": 15550
    },
    {
      "epoch": 1.217717952731257,
      "grad_norm": 2.767092704772949,
      "learning_rate": 2.978370037139719e-05,
      "loss": 0.6719,
      "step": 15560
    },
    {
      "epoch": 1.2185005478165598,
      "grad_norm": 12.031109809875488,
      "learning_rate": 2.977062300570173e-05,
      "loss": 0.7575,
      "step": 15570
    },
    {
      "epoch": 1.2192831429018627,
      "grad_norm": 3.6429831981658936,
      "learning_rate": 2.9757545640006278e-05,
      "loss": 0.9169,
      "step": 15580
    },
    {
      "epoch": 1.2200657379871656,
      "grad_norm": 5.938225269317627,
      "learning_rate": 2.9744468274310822e-05,
      "loss": 1.1786,
      "step": 15590
    },
    {
      "epoch": 1.2208483330724684,
      "grad_norm": 4.013261795043945,
      "learning_rate": 2.973139090861537e-05,
      "loss": 0.7499,
      "step": 15600
    },
    {
      "epoch": 1.221630928157771,
      "grad_norm": 2.5703365802764893,
      "learning_rate": 2.971831354291992e-05,
      "loss": 0.3785,
      "step": 15610
    },
    {
      "epoch": 1.222413523243074,
      "grad_norm": 12.347562789916992,
      "learning_rate": 2.970523617722446e-05,
      "loss": 0.9038,
      "step": 15620
    },
    {
      "epoch": 1.2231961183283768,
      "grad_norm": 7.312799453735352,
      "learning_rate": 2.9692158811529008e-05,
      "loss": 0.8009,
      "step": 15630
    },
    {
      "epoch": 1.2239787134136797,
      "grad_norm": 18.49422264099121,
      "learning_rate": 2.9679081445833552e-05,
      "loss": 0.7257,
      "step": 15640
    },
    {
      "epoch": 1.2247613084989826,
      "grad_norm": 9.06771183013916,
      "learning_rate": 2.96660040801381e-05,
      "loss": 0.8354,
      "step": 15650
    },
    {
      "epoch": 1.2255439035842854,
      "grad_norm": 0.8526217937469482,
      "learning_rate": 2.965292671444264e-05,
      "loss": 0.5582,
      "step": 15660
    },
    {
      "epoch": 1.2263264986695883,
      "grad_norm": 5.609523296356201,
      "learning_rate": 2.963984934874719e-05,
      "loss": 0.9538,
      "step": 15670
    },
    {
      "epoch": 1.2271090937548912,
      "grad_norm": 4.343968868255615,
      "learning_rate": 2.9626771983051737e-05,
      "loss": 0.4514,
      "step": 15680
    },
    {
      "epoch": 1.227891688840194,
      "grad_norm": 3.613077402114868,
      "learning_rate": 2.9613694617356282e-05,
      "loss": 1.0443,
      "step": 15690
    },
    {
      "epoch": 1.228674283925497,
      "grad_norm": 1.7593401670455933,
      "learning_rate": 2.960061725166083e-05,
      "loss": 0.5053,
      "step": 15700
    },
    {
      "epoch": 1.2294568790107998,
      "grad_norm": 0.9026181101799011,
      "learning_rate": 2.958753988596537e-05,
      "loss": 0.8868,
      "step": 15710
    },
    {
      "epoch": 1.2302394740961027,
      "grad_norm": 9.358074188232422,
      "learning_rate": 2.957446252026992e-05,
      "loss": 0.7199,
      "step": 15720
    },
    {
      "epoch": 1.2310220691814056,
      "grad_norm": 4.70798921585083,
      "learning_rate": 2.956138515457446e-05,
      "loss": 0.7034,
      "step": 15730
    },
    {
      "epoch": 1.2318046642667084,
      "grad_norm": 7.896917343139648,
      "learning_rate": 2.9548307788879008e-05,
      "loss": 0.624,
      "step": 15740
    },
    {
      "epoch": 1.2325872593520113,
      "grad_norm": 10.66620922088623,
      "learning_rate": 2.9535230423183556e-05,
      "loss": 1.1342,
      "step": 15750
    },
    {
      "epoch": 1.2333698544373142,
      "grad_norm": 21.562175750732422,
      "learning_rate": 2.95221530574881e-05,
      "loss": 0.6859,
      "step": 15760
    },
    {
      "epoch": 1.234152449522617,
      "grad_norm": 10.537701606750488,
      "learning_rate": 2.950907569179265e-05,
      "loss": 0.599,
      "step": 15770
    },
    {
      "epoch": 1.23493504460792,
      "grad_norm": 7.475683212280273,
      "learning_rate": 2.949599832609719e-05,
      "loss": 0.9433,
      "step": 15780
    },
    {
      "epoch": 1.2357176396932228,
      "grad_norm": 13.218484878540039,
      "learning_rate": 2.9482920960401738e-05,
      "loss": 0.7791,
      "step": 15790
    },
    {
      "epoch": 1.2365002347785257,
      "grad_norm": 0.795686662197113,
      "learning_rate": 2.9469843594706282e-05,
      "loss": 0.9595,
      "step": 15800
    },
    {
      "epoch": 1.2372828298638285,
      "grad_norm": 5.731076240539551,
      "learning_rate": 2.945676622901083e-05,
      "loss": 1.0692,
      "step": 15810
    },
    {
      "epoch": 1.2380654249491314,
      "grad_norm": 8.259038925170898,
      "learning_rate": 2.9443688863315378e-05,
      "loss": 1.1462,
      "step": 15820
    },
    {
      "epoch": 1.238848020034434,
      "grad_norm": 7.878340721130371,
      "learning_rate": 2.943061149761992e-05,
      "loss": 0.9222,
      "step": 15830
    },
    {
      "epoch": 1.239630615119737,
      "grad_norm": 4.304504871368408,
      "learning_rate": 2.9417534131924467e-05,
      "loss": 0.8796,
      "step": 15840
    },
    {
      "epoch": 1.2404132102050398,
      "grad_norm": 3.3958654403686523,
      "learning_rate": 2.9404456766229012e-05,
      "loss": 0.9734,
      "step": 15850
    },
    {
      "epoch": 1.2411958052903427,
      "grad_norm": 10.502890586853027,
      "learning_rate": 2.939137940053356e-05,
      "loss": 0.7005,
      "step": 15860
    },
    {
      "epoch": 1.2419784003756456,
      "grad_norm": 8.100044250488281,
      "learning_rate": 2.93783020348381e-05,
      "loss": 0.7641,
      "step": 15870
    },
    {
      "epoch": 1.2427609954609484,
      "grad_norm": 1.421992301940918,
      "learning_rate": 2.936522466914265e-05,
      "loss": 0.5673,
      "step": 15880
    },
    {
      "epoch": 1.2435435905462513,
      "grad_norm": 4.890364646911621,
      "learning_rate": 2.9352147303447197e-05,
      "loss": 0.9332,
      "step": 15890
    },
    {
      "epoch": 1.2443261856315542,
      "grad_norm": 8.71719741821289,
      "learning_rate": 2.9339069937751738e-05,
      "loss": 0.4195,
      "step": 15900
    },
    {
      "epoch": 1.245108780716857,
      "grad_norm": 6.988454341888428,
      "learning_rate": 2.932599257205629e-05,
      "loss": 0.5368,
      "step": 15910
    },
    {
      "epoch": 1.24589137580216,
      "grad_norm": 2.9225213527679443,
      "learning_rate": 2.931291520636083e-05,
      "loss": 0.6331,
      "step": 15920
    },
    {
      "epoch": 1.2466739708874628,
      "grad_norm": 11.154273986816406,
      "learning_rate": 2.929983784066538e-05,
      "loss": 0.9798,
      "step": 15930
    },
    {
      "epoch": 1.2474565659727657,
      "grad_norm": 12.400888442993164,
      "learning_rate": 2.928676047496992e-05,
      "loss": 0.8487,
      "step": 15940
    },
    {
      "epoch": 1.2482391610580685,
      "grad_norm": 10.76899528503418,
      "learning_rate": 2.9273683109274468e-05,
      "loss": 0.6437,
      "step": 15950
    },
    {
      "epoch": 1.2490217561433714,
      "grad_norm": 3.0618896484375,
      "learning_rate": 2.9260605743579016e-05,
      "loss": 1.144,
      "step": 15960
    },
    {
      "epoch": 1.2498043512286743,
      "grad_norm": 27.164220809936523,
      "learning_rate": 2.924752837788356e-05,
      "loss": 0.8853,
      "step": 15970
    },
    {
      "epoch": 1.2505869463139772,
      "grad_norm": 5.144565582275391,
      "learning_rate": 2.9234451012188108e-05,
      "loss": 0.6135,
      "step": 15980
    },
    {
      "epoch": 1.25136954139928,
      "grad_norm": 5.7737884521484375,
      "learning_rate": 2.922137364649265e-05,
      "loss": 0.6653,
      "step": 15990
    },
    {
      "epoch": 1.252152136484583,
      "grad_norm": 6.812036991119385,
      "learning_rate": 2.9208296280797197e-05,
      "loss": 1.0801,
      "step": 16000
    },
    {
      "epoch": 1.2529347315698858,
      "grad_norm": 7.811410903930664,
      "learning_rate": 2.9195218915101742e-05,
      "loss": 1.1994,
      "step": 16010
    },
    {
      "epoch": 1.2537173266551886,
      "grad_norm": 1.294587254524231,
      "learning_rate": 2.918214154940629e-05,
      "loss": 0.4797,
      "step": 16020
    },
    {
      "epoch": 1.2544999217404915,
      "grad_norm": 8.139857292175293,
      "learning_rate": 2.9169064183710838e-05,
      "loss": 0.6217,
      "step": 16030
    },
    {
      "epoch": 1.2552825168257944,
      "grad_norm": 14.346275329589844,
      "learning_rate": 2.915598681801538e-05,
      "loss": 0.8393,
      "step": 16040
    },
    {
      "epoch": 1.2560651119110973,
      "grad_norm": 2.7643864154815674,
      "learning_rate": 2.9142909452319927e-05,
      "loss": 1.0951,
      "step": 16050
    },
    {
      "epoch": 1.2568477069964001,
      "grad_norm": 11.251521110534668,
      "learning_rate": 2.912983208662447e-05,
      "loss": 1.0366,
      "step": 16060
    },
    {
      "epoch": 1.257630302081703,
      "grad_norm": 1.640625,
      "learning_rate": 2.911675472092902e-05,
      "loss": 0.4742,
      "step": 16070
    },
    {
      "epoch": 1.2584128971670059,
      "grad_norm": 8.082651138305664,
      "learning_rate": 2.910367735523356e-05,
      "loss": 0.7383,
      "step": 16080
    },
    {
      "epoch": 1.2591954922523088,
      "grad_norm": 11.309863090515137,
      "learning_rate": 2.909059998953811e-05,
      "loss": 0.9401,
      "step": 16090
    },
    {
      "epoch": 1.2599780873376116,
      "grad_norm": 9.650496482849121,
      "learning_rate": 2.9077522623842657e-05,
      "loss": 0.7981,
      "step": 16100
    },
    {
      "epoch": 1.2607606824229145,
      "grad_norm": 0.8719072341918945,
      "learning_rate": 2.9064445258147198e-05,
      "loss": 0.8464,
      "step": 16110
    },
    {
      "epoch": 1.2615432775082174,
      "grad_norm": 4.393743515014648,
      "learning_rate": 2.9051367892451746e-05,
      "loss": 0.9766,
      "step": 16120
    },
    {
      "epoch": 1.2623258725935202,
      "grad_norm": 22.52047348022461,
      "learning_rate": 2.903829052675629e-05,
      "loss": 0.7589,
      "step": 16130
    },
    {
      "epoch": 1.263108467678823,
      "grad_norm": 5.104656219482422,
      "learning_rate": 2.9025213161060838e-05,
      "loss": 0.962,
      "step": 16140
    },
    {
      "epoch": 1.2638910627641258,
      "grad_norm": 2.2536892890930176,
      "learning_rate": 2.901213579536538e-05,
      "loss": 0.7271,
      "step": 16150
    },
    {
      "epoch": 1.2646736578494286,
      "grad_norm": 6.358922004699707,
      "learning_rate": 2.8999058429669927e-05,
      "loss": 1.0643,
      "step": 16160
    },
    {
      "epoch": 1.2654562529347315,
      "grad_norm": 13.57130241394043,
      "learning_rate": 2.8985981063974475e-05,
      "loss": 0.9232,
      "step": 16170
    },
    {
      "epoch": 1.2662388480200344,
      "grad_norm": 8.537301063537598,
      "learning_rate": 2.897290369827902e-05,
      "loss": 0.8156,
      "step": 16180
    },
    {
      "epoch": 1.2670214431053373,
      "grad_norm": 4.230781555175781,
      "learning_rate": 2.8959826332583568e-05,
      "loss": 0.5996,
      "step": 16190
    },
    {
      "epoch": 1.2678040381906401,
      "grad_norm": 5.309373378753662,
      "learning_rate": 2.894674896688811e-05,
      "loss": 0.8081,
      "step": 16200
    },
    {
      "epoch": 1.268586633275943,
      "grad_norm": 6.214583396911621,
      "learning_rate": 2.8933671601192657e-05,
      "loss": 0.612,
      "step": 16210
    },
    {
      "epoch": 1.2693692283612459,
      "grad_norm": 2.7156760692596436,
      "learning_rate": 2.89205942354972e-05,
      "loss": 0.6513,
      "step": 16220
    },
    {
      "epoch": 1.2701518234465488,
      "grad_norm": 17.646570205688477,
      "learning_rate": 2.890751686980175e-05,
      "loss": 0.4411,
      "step": 16230
    },
    {
      "epoch": 1.2709344185318516,
      "grad_norm": 1.9079283475875854,
      "learning_rate": 2.8894439504106298e-05,
      "loss": 1.0108,
      "step": 16240
    },
    {
      "epoch": 1.2717170136171545,
      "grad_norm": 2.5087153911590576,
      "learning_rate": 2.888136213841084e-05,
      "loss": 1.2651,
      "step": 16250
    },
    {
      "epoch": 1.2724996087024574,
      "grad_norm": 9.55866527557373,
      "learning_rate": 2.8868284772715387e-05,
      "loss": 0.977,
      "step": 16260
    },
    {
      "epoch": 1.2732822037877602,
      "grad_norm": 1.1118944883346558,
      "learning_rate": 2.885520740701993e-05,
      "loss": 0.6044,
      "step": 16270
    },
    {
      "epoch": 1.2740647988730631,
      "grad_norm": 14.167759895324707,
      "learning_rate": 2.884213004132448e-05,
      "loss": 0.6948,
      "step": 16280
    },
    {
      "epoch": 1.274847393958366,
      "grad_norm": 5.541069507598877,
      "learning_rate": 2.882905267562902e-05,
      "loss": 0.6644,
      "step": 16290
    },
    {
      "epoch": 1.2756299890436689,
      "grad_norm": 4.974329471588135,
      "learning_rate": 2.881597530993357e-05,
      "loss": 0.4757,
      "step": 16300
    },
    {
      "epoch": 1.2764125841289717,
      "grad_norm": 6.223014831542969,
      "learning_rate": 2.8802897944238116e-05,
      "loss": 0.9122,
      "step": 16310
    },
    {
      "epoch": 1.2771951792142746,
      "grad_norm": 12.30238151550293,
      "learning_rate": 2.8789820578542658e-05,
      "loss": 0.8005,
      "step": 16320
    },
    {
      "epoch": 1.2779777742995773,
      "grad_norm": 26.706138610839844,
      "learning_rate": 2.8776743212847205e-05,
      "loss": 1.2546,
      "step": 16330
    },
    {
      "epoch": 1.2787603693848801,
      "grad_norm": 13.593615531921387,
      "learning_rate": 2.876366584715175e-05,
      "loss": 0.7089,
      "step": 16340
    },
    {
      "epoch": 1.279542964470183,
      "grad_norm": 5.0694451332092285,
      "learning_rate": 2.8750588481456298e-05,
      "loss": 0.6868,
      "step": 16350
    },
    {
      "epoch": 1.2803255595554859,
      "grad_norm": 11.858141899108887,
      "learning_rate": 2.873751111576084e-05,
      "loss": 0.7542,
      "step": 16360
    },
    {
      "epoch": 1.2811081546407888,
      "grad_norm": 10.689255714416504,
      "learning_rate": 2.8724433750065387e-05,
      "loss": 0.8947,
      "step": 16370
    },
    {
      "epoch": 1.2818907497260916,
      "grad_norm": 2.6941916942596436,
      "learning_rate": 2.8711356384369935e-05,
      "loss": 0.8416,
      "step": 16380
    },
    {
      "epoch": 1.2826733448113945,
      "grad_norm": 13.524312019348145,
      "learning_rate": 2.869827901867448e-05,
      "loss": 0.6044,
      "step": 16390
    },
    {
      "epoch": 1.2834559398966974,
      "grad_norm": 6.455661296844482,
      "learning_rate": 2.8685201652979028e-05,
      "loss": 0.8056,
      "step": 16400
    },
    {
      "epoch": 1.2842385349820002,
      "grad_norm": 1.3092318773269653,
      "learning_rate": 2.867212428728357e-05,
      "loss": 1.0438,
      "step": 16410
    },
    {
      "epoch": 1.2850211300673031,
      "grad_norm": 7.055698394775391,
      "learning_rate": 2.8659046921588117e-05,
      "loss": 0.9109,
      "step": 16420
    },
    {
      "epoch": 1.285803725152606,
      "grad_norm": 6.349701881408691,
      "learning_rate": 2.864596955589266e-05,
      "loss": 0.7996,
      "step": 16430
    },
    {
      "epoch": 1.2865863202379089,
      "grad_norm": 8.48344612121582,
      "learning_rate": 2.863289219019721e-05,
      "loss": 0.9273,
      "step": 16440
    },
    {
      "epoch": 1.2873689153232117,
      "grad_norm": 3.474637269973755,
      "learning_rate": 2.8619814824501757e-05,
      "loss": 0.7238,
      "step": 16450
    },
    {
      "epoch": 1.2881515104085146,
      "grad_norm": 9.412572860717773,
      "learning_rate": 2.86067374588063e-05,
      "loss": 0.5755,
      "step": 16460
    },
    {
      "epoch": 1.2889341054938175,
      "grad_norm": 6.036607265472412,
      "learning_rate": 2.8593660093110846e-05,
      "loss": 0.7254,
      "step": 16470
    },
    {
      "epoch": 1.2897167005791204,
      "grad_norm": 4.508155345916748,
      "learning_rate": 2.8580582727415388e-05,
      "loss": 0.763,
      "step": 16480
    },
    {
      "epoch": 1.2904992956644232,
      "grad_norm": 9.607423782348633,
      "learning_rate": 2.856750536171994e-05,
      "loss": 0.8331,
      "step": 16490
    },
    {
      "epoch": 1.291281890749726,
      "grad_norm": 34.68032455444336,
      "learning_rate": 2.855442799602448e-05,
      "loss": 0.8558,
      "step": 16500
    },
    {
      "epoch": 1.292064485835029,
      "grad_norm": 9.170145988464355,
      "learning_rate": 2.8541350630329028e-05,
      "loss": 0.9168,
      "step": 16510
    },
    {
      "epoch": 1.2928470809203318,
      "grad_norm": 19.83042335510254,
      "learning_rate": 2.8528273264633576e-05,
      "loss": 1.0664,
      "step": 16520
    },
    {
      "epoch": 1.2936296760056347,
      "grad_norm": 3.963723659515381,
      "learning_rate": 2.8515195898938117e-05,
      "loss": 0.8042,
      "step": 16530
    },
    {
      "epoch": 1.2944122710909376,
      "grad_norm": 10.148459434509277,
      "learning_rate": 2.8502118533242665e-05,
      "loss": 1.0284,
      "step": 16540
    },
    {
      "epoch": 1.2951948661762405,
      "grad_norm": 12.75153923034668,
      "learning_rate": 2.848904116754721e-05,
      "loss": 0.7138,
      "step": 16550
    },
    {
      "epoch": 1.2959774612615433,
      "grad_norm": 15.433623313903809,
      "learning_rate": 2.8475963801851758e-05,
      "loss": 0.9179,
      "step": 16560
    },
    {
      "epoch": 1.2967600563468462,
      "grad_norm": 1.100906252861023,
      "learning_rate": 2.84628864361563e-05,
      "loss": 0.784,
      "step": 16570
    },
    {
      "epoch": 1.297542651432149,
      "grad_norm": 1.5830748081207275,
      "learning_rate": 2.8449809070460847e-05,
      "loss": 0.373,
      "step": 16580
    },
    {
      "epoch": 1.298325246517452,
      "grad_norm": 11.89136791229248,
      "learning_rate": 2.8436731704765395e-05,
      "loss": 0.9696,
      "step": 16590
    },
    {
      "epoch": 1.2991078416027548,
      "grad_norm": 10.603907585144043,
      "learning_rate": 2.842365433906994e-05,
      "loss": 1.2744,
      "step": 16600
    },
    {
      "epoch": 1.2998904366880577,
      "grad_norm": 11.99843978881836,
      "learning_rate": 2.8410576973374487e-05,
      "loss": 0.8305,
      "step": 16610
    },
    {
      "epoch": 1.3006730317733606,
      "grad_norm": 9.91014575958252,
      "learning_rate": 2.839749960767903e-05,
      "loss": 0.8029,
      "step": 16620
    },
    {
      "epoch": 1.3014556268586635,
      "grad_norm": 1.1170716285705566,
      "learning_rate": 2.8384422241983577e-05,
      "loss": 0.6473,
      "step": 16630
    },
    {
      "epoch": 1.3022382219439663,
      "grad_norm": 3.7874815464019775,
      "learning_rate": 2.837134487628812e-05,
      "loss": 0.4936,
      "step": 16640
    },
    {
      "epoch": 1.303020817029269,
      "grad_norm": 11.998238563537598,
      "learning_rate": 2.835826751059267e-05,
      "loss": 0.722,
      "step": 16650
    },
    {
      "epoch": 1.3038034121145718,
      "grad_norm": 6.836818218231201,
      "learning_rate": 2.8345190144897217e-05,
      "loss": 0.9026,
      "step": 16660
    },
    {
      "epoch": 1.3045860071998747,
      "grad_norm": 10.998945236206055,
      "learning_rate": 2.8332112779201758e-05,
      "loss": 0.5435,
      "step": 16670
    },
    {
      "epoch": 1.3053686022851776,
      "grad_norm": 12.765204429626465,
      "learning_rate": 2.8319035413506306e-05,
      "loss": 0.8692,
      "step": 16680
    },
    {
      "epoch": 1.3061511973704805,
      "grad_norm": 5.536992073059082,
      "learning_rate": 2.8305958047810847e-05,
      "loss": 1.0891,
      "step": 16690
    },
    {
      "epoch": 1.3069337924557833,
      "grad_norm": 2.899336576461792,
      "learning_rate": 2.8292880682115395e-05,
      "loss": 0.6466,
      "step": 16700
    },
    {
      "epoch": 1.3077163875410862,
      "grad_norm": 5.334604263305664,
      "learning_rate": 2.827980331641994e-05,
      "loss": 1.0199,
      "step": 16710
    },
    {
      "epoch": 1.308498982626389,
      "grad_norm": 1.5473582744598389,
      "learning_rate": 2.8266725950724488e-05,
      "loss": 0.7292,
      "step": 16720
    },
    {
      "epoch": 1.309281577711692,
      "grad_norm": 2.1517181396484375,
      "learning_rate": 2.8253648585029036e-05,
      "loss": 0.9299,
      "step": 16730
    },
    {
      "epoch": 1.3100641727969948,
      "grad_norm": 1.354984164237976,
      "learning_rate": 2.8240571219333577e-05,
      "loss": 0.747,
      "step": 16740
    },
    {
      "epoch": 1.3108467678822977,
      "grad_norm": 7.708607196807861,
      "learning_rate": 2.8227493853638125e-05,
      "loss": 0.6409,
      "step": 16750
    },
    {
      "epoch": 1.3116293629676006,
      "grad_norm": 11.132925987243652,
      "learning_rate": 2.821441648794267e-05,
      "loss": 0.9912,
      "step": 16760
    },
    {
      "epoch": 1.3124119580529034,
      "grad_norm": 13.234508514404297,
      "learning_rate": 2.8201339122247217e-05,
      "loss": 0.8009,
      "step": 16770
    },
    {
      "epoch": 1.3131945531382063,
      "grad_norm": 5.807067394256592,
      "learning_rate": 2.818826175655176e-05,
      "loss": 0.8001,
      "step": 16780
    },
    {
      "epoch": 1.3139771482235092,
      "grad_norm": 3.1485202312469482,
      "learning_rate": 2.8175184390856307e-05,
      "loss": 0.6569,
      "step": 16790
    },
    {
      "epoch": 1.314759743308812,
      "grad_norm": 2.555232048034668,
      "learning_rate": 2.8162107025160855e-05,
      "loss": 0.6503,
      "step": 16800
    },
    {
      "epoch": 1.315542338394115,
      "grad_norm": 1.2698845863342285,
      "learning_rate": 2.81490296594654e-05,
      "loss": 0.893,
      "step": 16810
    },
    {
      "epoch": 1.3163249334794178,
      "grad_norm": 4.798511981964111,
      "learning_rate": 2.8135952293769947e-05,
      "loss": 0.5923,
      "step": 16820
    },
    {
      "epoch": 1.3171075285647207,
      "grad_norm": 2.107064962387085,
      "learning_rate": 2.812287492807449e-05,
      "loss": 0.8047,
      "step": 16830
    },
    {
      "epoch": 1.3178901236500236,
      "grad_norm": 10.044811248779297,
      "learning_rate": 2.8109797562379036e-05,
      "loss": 1.2158,
      "step": 16840
    },
    {
      "epoch": 1.3186727187353262,
      "grad_norm": 9.402920722961426,
      "learning_rate": 2.8096720196683577e-05,
      "loss": 0.9359,
      "step": 16850
    },
    {
      "epoch": 1.319455313820629,
      "grad_norm": 5.767270565032959,
      "learning_rate": 2.808364283098813e-05,
      "loss": 0.6189,
      "step": 16860
    },
    {
      "epoch": 1.320237908905932,
      "grad_norm": 6.6527533531188965,
      "learning_rate": 2.8070565465292677e-05,
      "loss": 1.015,
      "step": 16870
    },
    {
      "epoch": 1.3210205039912348,
      "grad_norm": 8.265036582946777,
      "learning_rate": 2.8057488099597218e-05,
      "loss": 0.4595,
      "step": 16880
    },
    {
      "epoch": 1.3218030990765377,
      "grad_norm": 12.769767761230469,
      "learning_rate": 2.8044410733901766e-05,
      "loss": 1.339,
      "step": 16890
    },
    {
      "epoch": 1.3225856941618406,
      "grad_norm": 0.9287058711051941,
      "learning_rate": 2.8031333368206307e-05,
      "loss": 0.7196,
      "step": 16900
    },
    {
      "epoch": 1.3233682892471434,
      "grad_norm": 4.175753116607666,
      "learning_rate": 2.8018256002510855e-05,
      "loss": 0.6742,
      "step": 16910
    },
    {
      "epoch": 1.3241508843324463,
      "grad_norm": 6.573429107666016,
      "learning_rate": 2.80051786368154e-05,
      "loss": 0.7937,
      "step": 16920
    },
    {
      "epoch": 1.3249334794177492,
      "grad_norm": 8.91126823425293,
      "learning_rate": 2.7992101271119948e-05,
      "loss": 0.7581,
      "step": 16930
    },
    {
      "epoch": 1.325716074503052,
      "grad_norm": 7.225174903869629,
      "learning_rate": 2.7979023905424496e-05,
      "loss": 1.0089,
      "step": 16940
    },
    {
      "epoch": 1.326498669588355,
      "grad_norm": 3.9276673793792725,
      "learning_rate": 2.7965946539729037e-05,
      "loss": 0.756,
      "step": 16950
    },
    {
      "epoch": 1.3272812646736578,
      "grad_norm": 8.111257553100586,
      "learning_rate": 2.7952869174033585e-05,
      "loss": 0.4455,
      "step": 16960
    },
    {
      "epoch": 1.3280638597589607,
      "grad_norm": 6.034783363342285,
      "learning_rate": 2.793979180833813e-05,
      "loss": 0.5555,
      "step": 16970
    },
    {
      "epoch": 1.3288464548442636,
      "grad_norm": 6.902520656585693,
      "learning_rate": 2.7926714442642677e-05,
      "loss": 1.0342,
      "step": 16980
    },
    {
      "epoch": 1.3296290499295664,
      "grad_norm": 9.604551315307617,
      "learning_rate": 2.791363707694722e-05,
      "loss": 0.7393,
      "step": 16990
    },
    {
      "epoch": 1.3304116450148693,
      "grad_norm": 6.528283596038818,
      "learning_rate": 2.7900559711251766e-05,
      "loss": 1.099,
      "step": 17000
    },
    {
      "epoch": 1.3311942401001722,
      "grad_norm": 3.699901580810547,
      "learning_rate": 2.7887482345556314e-05,
      "loss": 1.0943,
      "step": 17010
    },
    {
      "epoch": 1.331976835185475,
      "grad_norm": 4.722621440887451,
      "learning_rate": 2.787440497986086e-05,
      "loss": 0.5904,
      "step": 17020
    },
    {
      "epoch": 1.332759430270778,
      "grad_norm": 12.418442726135254,
      "learning_rate": 2.7861327614165407e-05,
      "loss": 0.8601,
      "step": 17030
    },
    {
      "epoch": 1.3335420253560808,
      "grad_norm": 17.65366554260254,
      "learning_rate": 2.7848250248469948e-05,
      "loss": 0.8866,
      "step": 17040
    },
    {
      "epoch": 1.3343246204413837,
      "grad_norm": 13.960923194885254,
      "learning_rate": 2.7835172882774496e-05,
      "loss": 1.0336,
      "step": 17050
    },
    {
      "epoch": 1.3351072155266865,
      "grad_norm": 6.151979446411133,
      "learning_rate": 2.7822095517079037e-05,
      "loss": 0.7561,
      "step": 17060
    },
    {
      "epoch": 1.3358898106119894,
      "grad_norm": 7.058422565460205,
      "learning_rate": 2.7809018151383585e-05,
      "loss": 0.6166,
      "step": 17070
    },
    {
      "epoch": 1.3366724056972923,
      "grad_norm": 6.182403564453125,
      "learning_rate": 2.7795940785688137e-05,
      "loss": 0.681,
      "step": 17080
    },
    {
      "epoch": 1.3374550007825952,
      "grad_norm": 31.880441665649414,
      "learning_rate": 2.7782863419992678e-05,
      "loss": 1.1232,
      "step": 17090
    },
    {
      "epoch": 1.338237595867898,
      "grad_norm": 13.407164573669434,
      "learning_rate": 2.7769786054297226e-05,
      "loss": 0.8913,
      "step": 17100
    },
    {
      "epoch": 1.339020190953201,
      "grad_norm": 5.0224609375,
      "learning_rate": 2.7756708688601767e-05,
      "loss": 1.0502,
      "step": 17110
    },
    {
      "epoch": 1.3398027860385038,
      "grad_norm": 9.912830352783203,
      "learning_rate": 2.7743631322906315e-05,
      "loss": 0.8706,
      "step": 17120
    },
    {
      "epoch": 1.3405853811238067,
      "grad_norm": 1.1200066804885864,
      "learning_rate": 2.773055395721086e-05,
      "loss": 0.5042,
      "step": 17130
    },
    {
      "epoch": 1.3413679762091095,
      "grad_norm": 2.0975797176361084,
      "learning_rate": 2.7717476591515407e-05,
      "loss": 0.4809,
      "step": 17140
    },
    {
      "epoch": 1.3421505712944124,
      "grad_norm": 1.899957299232483,
      "learning_rate": 2.7704399225819955e-05,
      "loss": 0.7438,
      "step": 17150
    },
    {
      "epoch": 1.3429331663797153,
      "grad_norm": 4.647892951965332,
      "learning_rate": 2.7691321860124496e-05,
      "loss": 0.7036,
      "step": 17160
    },
    {
      "epoch": 1.343715761465018,
      "grad_norm": 3.2041850090026855,
      "learning_rate": 2.7678244494429044e-05,
      "loss": 0.5695,
      "step": 17170
    },
    {
      "epoch": 1.3444983565503208,
      "grad_norm": 4.674700736999512,
      "learning_rate": 2.766516712873359e-05,
      "loss": 0.8934,
      "step": 17180
    },
    {
      "epoch": 1.3452809516356237,
      "grad_norm": 9.801416397094727,
      "learning_rate": 2.7652089763038137e-05,
      "loss": 1.0232,
      "step": 17190
    },
    {
      "epoch": 1.3460635467209265,
      "grad_norm": 3.513523578643799,
      "learning_rate": 2.7639012397342678e-05,
      "loss": 0.7009,
      "step": 17200
    },
    {
      "epoch": 1.3468461418062294,
      "grad_norm": 13.430389404296875,
      "learning_rate": 2.7625935031647226e-05,
      "loss": 0.7791,
      "step": 17210
    },
    {
      "epoch": 1.3476287368915323,
      "grad_norm": 12.010319709777832,
      "learning_rate": 2.7612857665951774e-05,
      "loss": 1.079,
      "step": 17220
    },
    {
      "epoch": 1.3484113319768352,
      "grad_norm": 13.90014362335205,
      "learning_rate": 2.759978030025632e-05,
      "loss": 0.8906,
      "step": 17230
    },
    {
      "epoch": 1.349193927062138,
      "grad_norm": 6.8910369873046875,
      "learning_rate": 2.7586702934560867e-05,
      "loss": 1.0031,
      "step": 17240
    },
    {
      "epoch": 1.349976522147441,
      "grad_norm": 14.473077774047852,
      "learning_rate": 2.7573625568865408e-05,
      "loss": 0.9465,
      "step": 17250
    },
    {
      "epoch": 1.3507591172327438,
      "grad_norm": 2.938900947570801,
      "learning_rate": 2.7560548203169956e-05,
      "loss": 0.9858,
      "step": 17260
    },
    {
      "epoch": 1.3515417123180467,
      "grad_norm": 7.263138771057129,
      "learning_rate": 2.7547470837474497e-05,
      "loss": 0.5624,
      "step": 17270
    },
    {
      "epoch": 1.3523243074033495,
      "grad_norm": 6.078798294067383,
      "learning_rate": 2.7534393471779045e-05,
      "loss": 0.8505,
      "step": 17280
    },
    {
      "epoch": 1.3531069024886524,
      "grad_norm": 4.198497295379639,
      "learning_rate": 2.7521316106083593e-05,
      "loss": 0.7864,
      "step": 17290
    },
    {
      "epoch": 1.3538894975739553,
      "grad_norm": 1.552048683166504,
      "learning_rate": 2.7508238740388137e-05,
      "loss": 0.7334,
      "step": 17300
    },
    {
      "epoch": 1.3546720926592581,
      "grad_norm": 5.205622673034668,
      "learning_rate": 2.7495161374692685e-05,
      "loss": 0.563,
      "step": 17310
    },
    {
      "epoch": 1.355454687744561,
      "grad_norm": 6.002500534057617,
      "learning_rate": 2.7482084008997227e-05,
      "loss": 0.7307,
      "step": 17320
    },
    {
      "epoch": 1.3562372828298639,
      "grad_norm": 3.5631818771362305,
      "learning_rate": 2.7469006643301775e-05,
      "loss": 1.0371,
      "step": 17330
    },
    {
      "epoch": 1.3570198779151668,
      "grad_norm": 12.533166885375977,
      "learning_rate": 2.745592927760632e-05,
      "loss": 0.7142,
      "step": 17340
    },
    {
      "epoch": 1.3578024730004696,
      "grad_norm": 10.810503005981445,
      "learning_rate": 2.7442851911910867e-05,
      "loss": 0.8219,
      "step": 17350
    },
    {
      "epoch": 1.3585850680857723,
      "grad_norm": 10.968480110168457,
      "learning_rate": 2.7429774546215415e-05,
      "loss": 0.5712,
      "step": 17360
    },
    {
      "epoch": 1.3593676631710752,
      "grad_norm": 2.5195844173431396,
      "learning_rate": 2.7416697180519956e-05,
      "loss": 0.6105,
      "step": 17370
    },
    {
      "epoch": 1.360150258256378,
      "grad_norm": 3.708937168121338,
      "learning_rate": 2.7403619814824504e-05,
      "loss": 0.5273,
      "step": 17380
    },
    {
      "epoch": 1.360932853341681,
      "grad_norm": 12.1312255859375,
      "learning_rate": 2.739054244912905e-05,
      "loss": 0.5928,
      "step": 17390
    },
    {
      "epoch": 1.3617154484269838,
      "grad_norm": 1.902864694595337,
      "learning_rate": 2.7377465083433597e-05,
      "loss": 1.0438,
      "step": 17400
    },
    {
      "epoch": 1.3624980435122866,
      "grad_norm": 4.130918025970459,
      "learning_rate": 2.7364387717738138e-05,
      "loss": 0.7896,
      "step": 17410
    },
    {
      "epoch": 1.3632806385975895,
      "grad_norm": 8.001883506774902,
      "learning_rate": 2.7351310352042686e-05,
      "loss": 0.7977,
      "step": 17420
    },
    {
      "epoch": 1.3640632336828924,
      "grad_norm": 7.101633548736572,
      "learning_rate": 2.7338232986347234e-05,
      "loss": 0.7441,
      "step": 17430
    },
    {
      "epoch": 1.3648458287681953,
      "grad_norm": 11.34511947631836,
      "learning_rate": 2.732515562065178e-05,
      "loss": 0.8509,
      "step": 17440
    },
    {
      "epoch": 1.3656284238534981,
      "grad_norm": 6.0388336181640625,
      "learning_rate": 2.7312078254956326e-05,
      "loss": 1.0643,
      "step": 17450
    },
    {
      "epoch": 1.366411018938801,
      "grad_norm": 7.288684844970703,
      "learning_rate": 2.7299000889260868e-05,
      "loss": 0.6871,
      "step": 17460
    },
    {
      "epoch": 1.3671936140241039,
      "grad_norm": 6.316847801208496,
      "learning_rate": 2.7285923523565415e-05,
      "loss": 0.5933,
      "step": 17470
    },
    {
      "epoch": 1.3679762091094068,
      "grad_norm": 2.2967870235443115,
      "learning_rate": 2.7272846157869957e-05,
      "loss": 0.7964,
      "step": 17480
    },
    {
      "epoch": 1.3687588041947096,
      "grad_norm": 8.264652252197266,
      "learning_rate": 2.7259768792174505e-05,
      "loss": 0.6922,
      "step": 17490
    },
    {
      "epoch": 1.3695413992800125,
      "grad_norm": 2.361069917678833,
      "learning_rate": 2.7246691426479053e-05,
      "loss": 1.0682,
      "step": 17500
    },
    {
      "epoch": 1.3703239943653154,
      "grad_norm": 6.146968841552734,
      "learning_rate": 2.7233614060783597e-05,
      "loss": 0.7468,
      "step": 17510
    },
    {
      "epoch": 1.3711065894506183,
      "grad_norm": 13.312433242797852,
      "learning_rate": 2.7220536695088145e-05,
      "loss": 0.7543,
      "step": 17520
    },
    {
      "epoch": 1.3718891845359211,
      "grad_norm": 3.7335801124572754,
      "learning_rate": 2.7207459329392686e-05,
      "loss": 1.1331,
      "step": 17530
    },
    {
      "epoch": 1.372671779621224,
      "grad_norm": 7.973845481872559,
      "learning_rate": 2.7194381963697234e-05,
      "loss": 0.8104,
      "step": 17540
    },
    {
      "epoch": 1.3734543747065269,
      "grad_norm": 8.52013874053955,
      "learning_rate": 2.718130459800178e-05,
      "loss": 0.9928,
      "step": 17550
    },
    {
      "epoch": 1.3742369697918297,
      "grad_norm": 9.166069984436035,
      "learning_rate": 2.7168227232306327e-05,
      "loss": 0.9756,
      "step": 17560
    },
    {
      "epoch": 1.3750195648771326,
      "grad_norm": 7.303042411804199,
      "learning_rate": 2.7155149866610875e-05,
      "loss": 0.6258,
      "step": 17570
    },
    {
      "epoch": 1.3758021599624355,
      "grad_norm": 4.854603290557861,
      "learning_rate": 2.7142072500915416e-05,
      "loss": 0.9982,
      "step": 17580
    },
    {
      "epoch": 1.3765847550477384,
      "grad_norm": 7.7962327003479,
      "learning_rate": 2.7128995135219964e-05,
      "loss": 0.6508,
      "step": 17590
    },
    {
      "epoch": 1.3773673501330412,
      "grad_norm": 9.495028495788574,
      "learning_rate": 2.711591776952451e-05,
      "loss": 0.8024,
      "step": 17600
    },
    {
      "epoch": 1.378149945218344,
      "grad_norm": 5.923781394958496,
      "learning_rate": 2.7102840403829056e-05,
      "loss": 0.6083,
      "step": 17610
    },
    {
      "epoch": 1.378932540303647,
      "grad_norm": 8.948060035705566,
      "learning_rate": 2.7089763038133598e-05,
      "loss": 0.8993,
      "step": 17620
    },
    {
      "epoch": 1.3797151353889499,
      "grad_norm": 7.916760444641113,
      "learning_rate": 2.7076685672438146e-05,
      "loss": 0.7453,
      "step": 17630
    },
    {
      "epoch": 1.3804977304742527,
      "grad_norm": 5.996373176574707,
      "learning_rate": 2.7063608306742694e-05,
      "loss": 0.6793,
      "step": 17640
    },
    {
      "epoch": 1.3812803255595556,
      "grad_norm": 9.501396179199219,
      "learning_rate": 2.7050530941047235e-05,
      "loss": 0.7578,
      "step": 17650
    },
    {
      "epoch": 1.3820629206448585,
      "grad_norm": 7.6660261154174805,
      "learning_rate": 2.7037453575351783e-05,
      "loss": 0.7011,
      "step": 17660
    },
    {
      "epoch": 1.3828455157301613,
      "grad_norm": 13.302132606506348,
      "learning_rate": 2.7024376209656327e-05,
      "loss": 1.277,
      "step": 17670
    },
    {
      "epoch": 1.383628110815464,
      "grad_norm": 1.8029171228408813,
      "learning_rate": 2.7011298843960875e-05,
      "loss": 0.4317,
      "step": 17680
    },
    {
      "epoch": 1.3844107059007669,
      "grad_norm": 11.360743522644043,
      "learning_rate": 2.6998221478265416e-05,
      "loss": 1.1565,
      "step": 17690
    },
    {
      "epoch": 1.3851933009860697,
      "grad_norm": 2.0121405124664307,
      "learning_rate": 2.6985144112569964e-05,
      "loss": 0.5426,
      "step": 17700
    },
    {
      "epoch": 1.3859758960713726,
      "grad_norm": 7.48969841003418,
      "learning_rate": 2.6972066746874512e-05,
      "loss": 0.9371,
      "step": 17710
    },
    {
      "epoch": 1.3867584911566755,
      "grad_norm": 9.184112548828125,
      "learning_rate": 2.6958989381179057e-05,
      "loss": 0.6682,
      "step": 17720
    },
    {
      "epoch": 1.3875410862419784,
      "grad_norm": 1.441542387008667,
      "learning_rate": 2.6945912015483605e-05,
      "loss": 0.5971,
      "step": 17730
    },
    {
      "epoch": 1.3883236813272812,
      "grad_norm": 0.8104006052017212,
      "learning_rate": 2.6932834649788146e-05,
      "loss": 0.5813,
      "step": 17740
    },
    {
      "epoch": 1.389106276412584,
      "grad_norm": 5.18435001373291,
      "learning_rate": 2.6919757284092694e-05,
      "loss": 0.7261,
      "step": 17750
    },
    {
      "epoch": 1.389888871497887,
      "grad_norm": 8.634261131286621,
      "learning_rate": 2.690667991839724e-05,
      "loss": 0.8633,
      "step": 17760
    },
    {
      "epoch": 1.3906714665831899,
      "grad_norm": 2.0563526153564453,
      "learning_rate": 2.6893602552701787e-05,
      "loss": 0.6896,
      "step": 17770
    },
    {
      "epoch": 1.3914540616684927,
      "grad_norm": 10.164725303649902,
      "learning_rate": 2.6880525187006334e-05,
      "loss": 0.9393,
      "step": 17780
    },
    {
      "epoch": 1.3922366567537956,
      "grad_norm": 4.818479537963867,
      "learning_rate": 2.6867447821310876e-05,
      "loss": 0.8211,
      "step": 17790
    },
    {
      "epoch": 1.3930192518390985,
      "grad_norm": 6.462790012359619,
      "learning_rate": 2.6854370455615424e-05,
      "loss": 0.8788,
      "step": 17800
    },
    {
      "epoch": 1.3938018469244013,
      "grad_norm": 7.934448719024658,
      "learning_rate": 2.6841293089919968e-05,
      "loss": 0.9475,
      "step": 17810
    },
    {
      "epoch": 1.3945844420097042,
      "grad_norm": 12.516240119934082,
      "learning_rate": 2.6828215724224516e-05,
      "loss": 0.719,
      "step": 17820
    },
    {
      "epoch": 1.395367037095007,
      "grad_norm": 29.658275604248047,
      "learning_rate": 2.6815138358529057e-05,
      "loss": 0.8836,
      "step": 17830
    },
    {
      "epoch": 1.39614963218031,
      "grad_norm": 6.180906772613525,
      "learning_rate": 2.6802060992833605e-05,
      "loss": 0.8227,
      "step": 17840
    },
    {
      "epoch": 1.3969322272656128,
      "grad_norm": 7.193913459777832,
      "learning_rate": 2.6788983627138153e-05,
      "loss": 0.8417,
      "step": 17850
    },
    {
      "epoch": 1.3977148223509157,
      "grad_norm": 10.692030906677246,
      "learning_rate": 2.6775906261442694e-05,
      "loss": 0.9221,
      "step": 17860
    },
    {
      "epoch": 1.3984974174362186,
      "grad_norm": 3.901637554168701,
      "learning_rate": 2.6762828895747242e-05,
      "loss": 0.7621,
      "step": 17870
    },
    {
      "epoch": 1.3992800125215212,
      "grad_norm": 11.899907112121582,
      "learning_rate": 2.6749751530051787e-05,
      "loss": 0.7165,
      "step": 17880
    },
    {
      "epoch": 1.400062607606824,
      "grad_norm": 8.645288467407227,
      "learning_rate": 2.6736674164356335e-05,
      "loss": 0.6468,
      "step": 17890
    },
    {
      "epoch": 1.400845202692127,
      "grad_norm": 13.222161293029785,
      "learning_rate": 2.6723596798660876e-05,
      "loss": 0.6818,
      "step": 17900
    },
    {
      "epoch": 1.4016277977774299,
      "grad_norm": 13.590484619140625,
      "learning_rate": 2.6710519432965424e-05,
      "loss": 0.6458,
      "step": 17910
    },
    {
      "epoch": 1.4024103928627327,
      "grad_norm": 12.015575408935547,
      "learning_rate": 2.669744206726997e-05,
      "loss": 0.8026,
      "step": 17920
    },
    {
      "epoch": 1.4031929879480356,
      "grad_norm": 2.272465229034424,
      "learning_rate": 2.6684364701574517e-05,
      "loss": 0.7203,
      "step": 17930
    },
    {
      "epoch": 1.4039755830333385,
      "grad_norm": 8.91211986541748,
      "learning_rate": 2.6671287335879065e-05,
      "loss": 0.6279,
      "step": 17940
    },
    {
      "epoch": 1.4047581781186413,
      "grad_norm": 4.915283203125,
      "learning_rate": 2.6658209970183606e-05,
      "loss": 0.8344,
      "step": 17950
    },
    {
      "epoch": 1.4055407732039442,
      "grad_norm": 16.887100219726562,
      "learning_rate": 2.6645132604488154e-05,
      "loss": 0.9535,
      "step": 17960
    },
    {
      "epoch": 1.406323368289247,
      "grad_norm": 10.82038688659668,
      "learning_rate": 2.66320552387927e-05,
      "loss": 0.7878,
      "step": 17970
    },
    {
      "epoch": 1.40710596337455,
      "grad_norm": 1.1912857294082642,
      "learning_rate": 2.6618977873097246e-05,
      "loss": 0.7351,
      "step": 17980
    },
    {
      "epoch": 1.4078885584598528,
      "grad_norm": 1.1585042476654053,
      "learning_rate": 2.6605900507401787e-05,
      "loss": 0.821,
      "step": 17990
    },
    {
      "epoch": 1.4086711535451557,
      "grad_norm": 3.255173921585083,
      "learning_rate": 2.6592823141706335e-05,
      "loss": 0.6774,
      "step": 18000
    },
    {
      "epoch": 1.4094537486304586,
      "grad_norm": 10.036865234375,
      "learning_rate": 2.6579745776010883e-05,
      "loss": 0.7543,
      "step": 18010
    },
    {
      "epoch": 1.4102363437157615,
      "grad_norm": 4.045093059539795,
      "learning_rate": 2.6566668410315425e-05,
      "loss": 0.4775,
      "step": 18020
    },
    {
      "epoch": 1.4110189388010643,
      "grad_norm": 10.892631530761719,
      "learning_rate": 2.6553591044619976e-05,
      "loss": 0.71,
      "step": 18030
    },
    {
      "epoch": 1.4118015338863672,
      "grad_norm": 1.4389020204544067,
      "learning_rate": 2.6540513678924517e-05,
      "loss": 1.0378,
      "step": 18040
    },
    {
      "epoch": 1.41258412897167,
      "grad_norm": 8.514484405517578,
      "learning_rate": 2.6527436313229065e-05,
      "loss": 0.918,
      "step": 18050
    },
    {
      "epoch": 1.413366724056973,
      "grad_norm": 13.020790100097656,
      "learning_rate": 2.6514358947533606e-05,
      "loss": 0.6432,
      "step": 18060
    },
    {
      "epoch": 1.4141493191422758,
      "grad_norm": 24.1947078704834,
      "learning_rate": 2.6501281581838154e-05,
      "loss": 0.6591,
      "step": 18070
    },
    {
      "epoch": 1.4149319142275787,
      "grad_norm": 13.207395553588867,
      "learning_rate": 2.6488204216142702e-05,
      "loss": 1.1729,
      "step": 18080
    },
    {
      "epoch": 1.4157145093128816,
      "grad_norm": 7.325832366943359,
      "learning_rate": 2.6475126850447247e-05,
      "loss": 0.6862,
      "step": 18090
    },
    {
      "epoch": 1.4164971043981844,
      "grad_norm": 1.9712107181549072,
      "learning_rate": 2.6462049484751795e-05,
      "loss": 0.5425,
      "step": 18100
    },
    {
      "epoch": 1.4172796994834873,
      "grad_norm": 4.6793742179870605,
      "learning_rate": 2.6448972119056336e-05,
      "loss": 0.9167,
      "step": 18110
    },
    {
      "epoch": 1.4180622945687902,
      "grad_norm": 0.8497974276542664,
      "learning_rate": 2.6435894753360884e-05,
      "loss": 0.7352,
      "step": 18120
    },
    {
      "epoch": 1.418844889654093,
      "grad_norm": 2.760479211807251,
      "learning_rate": 2.642281738766543e-05,
      "loss": 0.8255,
      "step": 18130
    },
    {
      "epoch": 1.419627484739396,
      "grad_norm": 9.215371131896973,
      "learning_rate": 2.6409740021969976e-05,
      "loss": 0.926,
      "step": 18140
    },
    {
      "epoch": 1.4204100798246988,
      "grad_norm": 19.523902893066406,
      "learning_rate": 2.6396662656274524e-05,
      "loss": 0.8937,
      "step": 18150
    },
    {
      "epoch": 1.4211926749100017,
      "grad_norm": 6.497433185577393,
      "learning_rate": 2.6383585290579066e-05,
      "loss": 0.7407,
      "step": 18160
    },
    {
      "epoch": 1.4219752699953045,
      "grad_norm": 7.969742774963379,
      "learning_rate": 2.6370507924883613e-05,
      "loss": 0.8423,
      "step": 18170
    },
    {
      "epoch": 1.4227578650806074,
      "grad_norm": 2.6403822898864746,
      "learning_rate": 2.6357430559188158e-05,
      "loss": 1.3147,
      "step": 18180
    },
    {
      "epoch": 1.4235404601659103,
      "grad_norm": 3.652294635772705,
      "learning_rate": 2.6344353193492706e-05,
      "loss": 0.6593,
      "step": 18190
    },
    {
      "epoch": 1.424323055251213,
      "grad_norm": 14.53834342956543,
      "learning_rate": 2.6331275827797247e-05,
      "loss": 0.9213,
      "step": 18200
    },
    {
      "epoch": 1.4251056503365158,
      "grad_norm": 12.434102058410645,
      "learning_rate": 2.6318198462101795e-05,
      "loss": 0.9218,
      "step": 18210
    },
    {
      "epoch": 1.4258882454218187,
      "grad_norm": 1.2343978881835938,
      "learning_rate": 2.6305121096406343e-05,
      "loss": 1.0335,
      "step": 18220
    },
    {
      "epoch": 1.4266708405071216,
      "grad_norm": 2.1116576194763184,
      "learning_rate": 2.6292043730710884e-05,
      "loss": 0.5046,
      "step": 18230
    },
    {
      "epoch": 1.4274534355924244,
      "grad_norm": 4.492133140563965,
      "learning_rate": 2.6278966365015432e-05,
      "loss": 0.7002,
      "step": 18240
    },
    {
      "epoch": 1.4282360306777273,
      "grad_norm": 5.0913848876953125,
      "learning_rate": 2.6265888999319977e-05,
      "loss": 0.628,
      "step": 18250
    },
    {
      "epoch": 1.4290186257630302,
      "grad_norm": 5.290317058563232,
      "learning_rate": 2.6252811633624525e-05,
      "loss": 0.7237,
      "step": 18260
    },
    {
      "epoch": 1.429801220848333,
      "grad_norm": 11.296941757202148,
      "learning_rate": 2.6239734267929066e-05,
      "loss": 0.7681,
      "step": 18270
    },
    {
      "epoch": 1.430583815933636,
      "grad_norm": 3.8759896755218506,
      "learning_rate": 2.6226656902233614e-05,
      "loss": 1.0442,
      "step": 18280
    },
    {
      "epoch": 1.4313664110189388,
      "grad_norm": 7.233345985412598,
      "learning_rate": 2.6213579536538162e-05,
      "loss": 0.6959,
      "step": 18290
    },
    {
      "epoch": 1.4321490061042417,
      "grad_norm": 1.4508079290390015,
      "learning_rate": 2.6200502170842706e-05,
      "loss": 0.8516,
      "step": 18300
    },
    {
      "epoch": 1.4329316011895445,
      "grad_norm": 3.3517355918884277,
      "learning_rate": 2.6187424805147254e-05,
      "loss": 0.3021,
      "step": 18310
    },
    {
      "epoch": 1.4337141962748474,
      "grad_norm": 6.400518894195557,
      "learning_rate": 2.6174347439451796e-05,
      "loss": 0.9174,
      "step": 18320
    },
    {
      "epoch": 1.4344967913601503,
      "grad_norm": 8.03769588470459,
      "learning_rate": 2.6161270073756344e-05,
      "loss": 0.6589,
      "step": 18330
    },
    {
      "epoch": 1.4352793864454532,
      "grad_norm": 5.155874729156494,
      "learning_rate": 2.6148192708060888e-05,
      "loss": 0.6634,
      "step": 18340
    },
    {
      "epoch": 1.436061981530756,
      "grad_norm": 3.581977367401123,
      "learning_rate": 2.6135115342365436e-05,
      "loss": 0.9179,
      "step": 18350
    },
    {
      "epoch": 1.436844576616059,
      "grad_norm": 10.089953422546387,
      "learning_rate": 2.6122037976669984e-05,
      "loss": 0.8615,
      "step": 18360
    },
    {
      "epoch": 1.4376271717013618,
      "grad_norm": 9.516461372375488,
      "learning_rate": 2.6108960610974525e-05,
      "loss": 0.8899,
      "step": 18370
    },
    {
      "epoch": 1.4384097667866647,
      "grad_norm": 0.9461118578910828,
      "learning_rate": 2.6095883245279073e-05,
      "loss": 0.6568,
      "step": 18380
    },
    {
      "epoch": 1.4391923618719673,
      "grad_norm": 12.669405937194824,
      "learning_rate": 2.6082805879583618e-05,
      "loss": 0.675,
      "step": 18390
    },
    {
      "epoch": 1.4399749569572702,
      "grad_norm": 4.93573522567749,
      "learning_rate": 2.6069728513888166e-05,
      "loss": 1.1611,
      "step": 18400
    },
    {
      "epoch": 1.440757552042573,
      "grad_norm": 14.645889282226562,
      "learning_rate": 2.6056651148192707e-05,
      "loss": 0.7167,
      "step": 18410
    },
    {
      "epoch": 1.441540147127876,
      "grad_norm": 3.5778534412384033,
      "learning_rate": 2.6043573782497255e-05,
      "loss": 0.8196,
      "step": 18420
    },
    {
      "epoch": 1.4423227422131788,
      "grad_norm": 5.28636360168457,
      "learning_rate": 2.6030496416801803e-05,
      "loss": 0.7736,
      "step": 18430
    },
    {
      "epoch": 1.4431053372984817,
      "grad_norm": 14.62289047241211,
      "learning_rate": 2.6017419051106344e-05,
      "loss": 1.1296,
      "step": 18440
    },
    {
      "epoch": 1.4438879323837845,
      "grad_norm": 8.1622953414917,
      "learning_rate": 2.6004341685410892e-05,
      "loss": 1.4046,
      "step": 18450
    },
    {
      "epoch": 1.4446705274690874,
      "grad_norm": 6.461987018585205,
      "learning_rate": 2.5991264319715437e-05,
      "loss": 1.0298,
      "step": 18460
    },
    {
      "epoch": 1.4454531225543903,
      "grad_norm": 5.500799179077148,
      "learning_rate": 2.5978186954019985e-05,
      "loss": 0.6462,
      "step": 18470
    },
    {
      "epoch": 1.4462357176396932,
      "grad_norm": 11.830198287963867,
      "learning_rate": 2.5965109588324526e-05,
      "loss": 1.0191,
      "step": 18480
    },
    {
      "epoch": 1.447018312724996,
      "grad_norm": 16.479045867919922,
      "learning_rate": 2.5952032222629074e-05,
      "loss": 0.773,
      "step": 18490
    },
    {
      "epoch": 1.447800907810299,
      "grad_norm": 4.908631324768066,
      "learning_rate": 2.593895485693362e-05,
      "loss": 0.8912,
      "step": 18500
    },
    {
      "epoch": 1.4485835028956018,
      "grad_norm": 2.39898681640625,
      "learning_rate": 2.5925877491238166e-05,
      "loss": 0.7252,
      "step": 18510
    },
    {
      "epoch": 1.4493660979809047,
      "grad_norm": 15.881580352783203,
      "learning_rate": 2.5912800125542714e-05,
      "loss": 1.3465,
      "step": 18520
    },
    {
      "epoch": 1.4501486930662075,
      "grad_norm": 12.574956893920898,
      "learning_rate": 2.5899722759847255e-05,
      "loss": 0.5744,
      "step": 18530
    },
    {
      "epoch": 1.4509312881515104,
      "grad_norm": 2.3377225399017334,
      "learning_rate": 2.5886645394151803e-05,
      "loss": 0.8618,
      "step": 18540
    },
    {
      "epoch": 1.4517138832368133,
      "grad_norm": 29.150312423706055,
      "learning_rate": 2.5873568028456348e-05,
      "loss": 0.9137,
      "step": 18550
    },
    {
      "epoch": 1.4524964783221161,
      "grad_norm": 9.707600593566895,
      "learning_rate": 2.5860490662760896e-05,
      "loss": 0.6624,
      "step": 18560
    },
    {
      "epoch": 1.453279073407419,
      "grad_norm": 12.325383186340332,
      "learning_rate": 2.5847413297065444e-05,
      "loss": 0.6448,
      "step": 18570
    },
    {
      "epoch": 1.454061668492722,
      "grad_norm": 0.7494577765464783,
      "learning_rate": 2.5834335931369985e-05,
      "loss": 0.8663,
      "step": 18580
    },
    {
      "epoch": 1.4548442635780248,
      "grad_norm": 10.404990196228027,
      "learning_rate": 2.5821258565674533e-05,
      "loss": 1.1296,
      "step": 18590
    },
    {
      "epoch": 1.4556268586633276,
      "grad_norm": 4.678487777709961,
      "learning_rate": 2.5808181199979074e-05,
      "loss": 1.0632,
      "step": 18600
    },
    {
      "epoch": 1.4564094537486305,
      "grad_norm": 4.0322184562683105,
      "learning_rate": 2.5795103834283622e-05,
      "loss": 0.7246,
      "step": 18610
    },
    {
      "epoch": 1.4571920488339334,
      "grad_norm": 3.3383290767669678,
      "learning_rate": 2.5782026468588167e-05,
      "loss": 0.3746,
      "step": 18620
    },
    {
      "epoch": 1.4579746439192363,
      "grad_norm": 0.785410463809967,
      "learning_rate": 2.5768949102892715e-05,
      "loss": 0.5628,
      "step": 18630
    },
    {
      "epoch": 1.4587572390045391,
      "grad_norm": 10.228160858154297,
      "learning_rate": 2.5755871737197263e-05,
      "loss": 1.1556,
      "step": 18640
    },
    {
      "epoch": 1.459539834089842,
      "grad_norm": 9.33395767211914,
      "learning_rate": 2.5742794371501804e-05,
      "loss": 1.0157,
      "step": 18650
    },
    {
      "epoch": 1.4603224291751449,
      "grad_norm": 8.022296905517578,
      "learning_rate": 2.5729717005806352e-05,
      "loss": 0.5238,
      "step": 18660
    },
    {
      "epoch": 1.4611050242604477,
      "grad_norm": 7.05710506439209,
      "learning_rate": 2.5716639640110896e-05,
      "loss": 0.785,
      "step": 18670
    },
    {
      "epoch": 1.4618876193457506,
      "grad_norm": 6.35413932800293,
      "learning_rate": 2.5703562274415444e-05,
      "loss": 0.8725,
      "step": 18680
    },
    {
      "epoch": 1.4626702144310535,
      "grad_norm": 2.345409393310547,
      "learning_rate": 2.5690484908719985e-05,
      "loss": 0.8313,
      "step": 18690
    },
    {
      "epoch": 1.4634528095163564,
      "grad_norm": 9.662617683410645,
      "learning_rate": 2.5677407543024533e-05,
      "loss": 0.9821,
      "step": 18700
    },
    {
      "epoch": 1.464235404601659,
      "grad_norm": 11.102802276611328,
      "learning_rate": 2.566433017732908e-05,
      "loss": 0.942,
      "step": 18710
    },
    {
      "epoch": 1.465017999686962,
      "grad_norm": 14.024301528930664,
      "learning_rate": 2.5651252811633626e-05,
      "loss": 0.9959,
      "step": 18720
    },
    {
      "epoch": 1.4658005947722648,
      "grad_norm": 3.3596816062927246,
      "learning_rate": 2.5638175445938174e-05,
      "loss": 0.833,
      "step": 18730
    },
    {
      "epoch": 1.4665831898575676,
      "grad_norm": 8.642980575561523,
      "learning_rate": 2.5625098080242715e-05,
      "loss": 0.5842,
      "step": 18740
    },
    {
      "epoch": 1.4673657849428705,
      "grad_norm": 6.945165634155273,
      "learning_rate": 2.5612020714547263e-05,
      "loss": 0.8708,
      "step": 18750
    },
    {
      "epoch": 1.4681483800281734,
      "grad_norm": 11.196599960327148,
      "learning_rate": 2.5598943348851808e-05,
      "loss": 0.9242,
      "step": 18760
    },
    {
      "epoch": 1.4689309751134763,
      "grad_norm": 8.654644966125488,
      "learning_rate": 2.5585865983156356e-05,
      "loss": 0.9592,
      "step": 18770
    },
    {
      "epoch": 1.4697135701987791,
      "grad_norm": 9.469658851623535,
      "learning_rate": 2.5572788617460904e-05,
      "loss": 0.8242,
      "step": 18780
    },
    {
      "epoch": 1.470496165284082,
      "grad_norm": 5.45184850692749,
      "learning_rate": 2.5559711251765445e-05,
      "loss": 0.5062,
      "step": 18790
    },
    {
      "epoch": 1.4712787603693849,
      "grad_norm": 14.156975746154785,
      "learning_rate": 2.5546633886069993e-05,
      "loss": 0.8727,
      "step": 18800
    },
    {
      "epoch": 1.4720613554546877,
      "grad_norm": 6.620410442352295,
      "learning_rate": 2.5533556520374534e-05,
      "loss": 0.6562,
      "step": 18810
    },
    {
      "epoch": 1.4728439505399906,
      "grad_norm": 7.667637348175049,
      "learning_rate": 2.5520479154679082e-05,
      "loss": 0.6079,
      "step": 18820
    },
    {
      "epoch": 1.4736265456252935,
      "grad_norm": 12.301952362060547,
      "learning_rate": 2.5507401788983626e-05,
      "loss": 0.7787,
      "step": 18830
    },
    {
      "epoch": 1.4744091407105964,
      "grad_norm": 4.449300289154053,
      "learning_rate": 2.5494324423288174e-05,
      "loss": 0.6631,
      "step": 18840
    },
    {
      "epoch": 1.4751917357958992,
      "grad_norm": 9.99222183227539,
      "learning_rate": 2.5481247057592722e-05,
      "loss": 1.241,
      "step": 18850
    },
    {
      "epoch": 1.4759743308812021,
      "grad_norm": 0.6596569418907166,
      "learning_rate": 2.5468169691897263e-05,
      "loss": 1.1178,
      "step": 18860
    },
    {
      "epoch": 1.476756925966505,
      "grad_norm": 10.468120574951172,
      "learning_rate": 2.545509232620181e-05,
      "loss": 0.6664,
      "step": 18870
    },
    {
      "epoch": 1.4775395210518079,
      "grad_norm": 11.286723136901855,
      "learning_rate": 2.5442014960506356e-05,
      "loss": 0.994,
      "step": 18880
    },
    {
      "epoch": 1.4783221161371107,
      "grad_norm": 4.971017360687256,
      "learning_rate": 2.5428937594810904e-05,
      "loss": 0.4525,
      "step": 18890
    },
    {
      "epoch": 1.4791047112224136,
      "grad_norm": 5.0735182762146,
      "learning_rate": 2.5415860229115445e-05,
      "loss": 0.7518,
      "step": 18900
    },
    {
      "epoch": 1.4798873063077163,
      "grad_norm": 7.419325351715088,
      "learning_rate": 2.5402782863419993e-05,
      "loss": 0.7304,
      "step": 18910
    },
    {
      "epoch": 1.4806699013930191,
      "grad_norm": 8.503355026245117,
      "learning_rate": 2.538970549772454e-05,
      "loss": 1.1914,
      "step": 18920
    },
    {
      "epoch": 1.481452496478322,
      "grad_norm": 11.816587448120117,
      "learning_rate": 2.5376628132029086e-05,
      "loss": 0.5745,
      "step": 18930
    },
    {
      "epoch": 1.4822350915636249,
      "grad_norm": 6.067477703094482,
      "learning_rate": 2.5363550766333634e-05,
      "loss": 0.4843,
      "step": 18940
    },
    {
      "epoch": 1.4830176866489277,
      "grad_norm": 10.03332805633545,
      "learning_rate": 2.5350473400638175e-05,
      "loss": 0.8037,
      "step": 18950
    },
    {
      "epoch": 1.4838002817342306,
      "grad_norm": 8.315699577331543,
      "learning_rate": 2.5337396034942723e-05,
      "loss": 0.8335,
      "step": 18960
    },
    {
      "epoch": 1.4845828768195335,
      "grad_norm": 13.55517578125,
      "learning_rate": 2.5324318669247264e-05,
      "loss": 0.8405,
      "step": 18970
    },
    {
      "epoch": 1.4853654719048364,
      "grad_norm": 9.77314567565918,
      "learning_rate": 2.5311241303551815e-05,
      "loss": 0.9108,
      "step": 18980
    },
    {
      "epoch": 1.4861480669901392,
      "grad_norm": 1.7688748836517334,
      "learning_rate": 2.5298163937856363e-05,
      "loss": 0.5332,
      "step": 18990
    },
    {
      "epoch": 1.4869306620754421,
      "grad_norm": 0.9942969083786011,
      "learning_rate": 2.5285086572160904e-05,
      "loss": 0.2961,
      "step": 19000
    },
    {
      "epoch": 1.487713257160745,
      "grad_norm": 4.920785903930664,
      "learning_rate": 2.5272009206465452e-05,
      "loss": 0.5359,
      "step": 19010
    },
    {
      "epoch": 1.4884958522460479,
      "grad_norm": 4.496872901916504,
      "learning_rate": 2.5258931840769994e-05,
      "loss": 0.6212,
      "step": 19020
    },
    {
      "epoch": 1.4892784473313507,
      "grad_norm": 4.483256816864014,
      "learning_rate": 2.524585447507454e-05,
      "loss": 1.0326,
      "step": 19030
    },
    {
      "epoch": 1.4900610424166536,
      "grad_norm": 0.4632585346698761,
      "learning_rate": 2.5232777109379086e-05,
      "loss": 0.6184,
      "step": 19040
    },
    {
      "epoch": 1.4908436375019565,
      "grad_norm": 15.521585464477539,
      "learning_rate": 2.5219699743683634e-05,
      "loss": 0.9514,
      "step": 19050
    },
    {
      "epoch": 1.4916262325872593,
      "grad_norm": 7.657164573669434,
      "learning_rate": 2.5206622377988182e-05,
      "loss": 0.8913,
      "step": 19060
    },
    {
      "epoch": 1.4924088276725622,
      "grad_norm": 3.1610119342803955,
      "learning_rate": 2.5193545012292723e-05,
      "loss": 0.6824,
      "step": 19070
    },
    {
      "epoch": 1.493191422757865,
      "grad_norm": 8.980730056762695,
      "learning_rate": 2.518046764659727e-05,
      "loss": 0.8165,
      "step": 19080
    },
    {
      "epoch": 1.493974017843168,
      "grad_norm": 2.370962142944336,
      "learning_rate": 2.5167390280901816e-05,
      "loss": 0.6871,
      "step": 19090
    },
    {
      "epoch": 1.4947566129284708,
      "grad_norm": 9.10927963256836,
      "learning_rate": 2.5154312915206364e-05,
      "loss": 1.1372,
      "step": 19100
    },
    {
      "epoch": 1.4955392080137737,
      "grad_norm": 17.87879180908203,
      "learning_rate": 2.5141235549510905e-05,
      "loss": 0.9615,
      "step": 19110
    },
    {
      "epoch": 1.4963218030990766,
      "grad_norm": 3.009150743484497,
      "learning_rate": 2.5128158183815453e-05,
      "loss": 0.5363,
      "step": 19120
    },
    {
      "epoch": 1.4971043981843795,
      "grad_norm": 16.865381240844727,
      "learning_rate": 2.511508081812e-05,
      "loss": 1.0191,
      "step": 19130
    },
    {
      "epoch": 1.4978869932696823,
      "grad_norm": 0.9950406551361084,
      "learning_rate": 2.5102003452424545e-05,
      "loss": 0.473,
      "step": 19140
    },
    {
      "epoch": 1.4986695883549852,
      "grad_norm": 4.7827959060668945,
      "learning_rate": 2.5088926086729093e-05,
      "loss": 0.9703,
      "step": 19150
    },
    {
      "epoch": 1.499452183440288,
      "grad_norm": 4.919632911682129,
      "learning_rate": 2.5075848721033635e-05,
      "loss": 0.6393,
      "step": 19160
    },
    {
      "epoch": 1.500234778525591,
      "grad_norm": 7.4306488037109375,
      "learning_rate": 2.5062771355338183e-05,
      "loss": 0.7152,
      "step": 19170
    },
    {
      "epoch": 1.5010173736108938,
      "grad_norm": 4.4506120681762695,
      "learning_rate": 2.5049693989642724e-05,
      "loss": 0.8183,
      "step": 19180
    },
    {
      "epoch": 1.5017999686961967,
      "grad_norm": 4.41422700881958,
      "learning_rate": 2.503661662394727e-05,
      "loss": 0.7172,
      "step": 19190
    },
    {
      "epoch": 1.5025825637814996,
      "grad_norm": 6.697972774505615,
      "learning_rate": 2.5023539258251823e-05,
      "loss": 0.6243,
      "step": 19200
    },
    {
      "epoch": 1.5033651588668024,
      "grad_norm": 17.344009399414062,
      "learning_rate": 2.5010461892556364e-05,
      "loss": 0.8259,
      "step": 19210
    },
    {
      "epoch": 1.5041477539521053,
      "grad_norm": 15.168281555175781,
      "learning_rate": 2.499738452686091e-05,
      "loss": 0.9054,
      "step": 19220
    },
    {
      "epoch": 1.5049303490374082,
      "grad_norm": 5.563377380371094,
      "learning_rate": 2.4984307161165457e-05,
      "loss": 0.7888,
      "step": 19230
    },
    {
      "epoch": 1.505712944122711,
      "grad_norm": 12.7642183303833,
      "learning_rate": 2.497122979547e-05,
      "loss": 0.7639,
      "step": 19240
    },
    {
      "epoch": 1.506495539208014,
      "grad_norm": 1.1031168699264526,
      "learning_rate": 2.495815242977455e-05,
      "loss": 0.6744,
      "step": 19250
    },
    {
      "epoch": 1.5072781342933166,
      "grad_norm": 3.2368249893188477,
      "learning_rate": 2.4945075064079094e-05,
      "loss": 0.9091,
      "step": 19260
    },
    {
      "epoch": 1.5080607293786195,
      "grad_norm": 7.3180623054504395,
      "learning_rate": 2.493199769838364e-05,
      "loss": 0.8611,
      "step": 19270
    },
    {
      "epoch": 1.5088433244639223,
      "grad_norm": 5.035080432891846,
      "learning_rate": 2.4918920332688183e-05,
      "loss": 0.6808,
      "step": 19280
    },
    {
      "epoch": 1.5096259195492252,
      "grad_norm": 5.068151950836182,
      "learning_rate": 2.4905842966992728e-05,
      "loss": 0.6274,
      "step": 19290
    },
    {
      "epoch": 1.510408514634528,
      "grad_norm": 35.25727462768555,
      "learning_rate": 2.4892765601297276e-05,
      "loss": 0.6424,
      "step": 19300
    },
    {
      "epoch": 1.511191109719831,
      "grad_norm": 3.9121453762054443,
      "learning_rate": 2.4879688235601823e-05,
      "loss": 0.6819,
      "step": 19310
    },
    {
      "epoch": 1.5119737048051338,
      "grad_norm": 7.930941104888916,
      "learning_rate": 2.4866610869906368e-05,
      "loss": 0.7778,
      "step": 19320
    },
    {
      "epoch": 1.5127562998904367,
      "grad_norm": 3.780491828918457,
      "learning_rate": 2.4853533504210913e-05,
      "loss": 0.7684,
      "step": 19330
    },
    {
      "epoch": 1.5135388949757396,
      "grad_norm": 8.71904182434082,
      "learning_rate": 2.4840456138515457e-05,
      "loss": 1.0614,
      "step": 19340
    },
    {
      "epoch": 1.5143214900610424,
      "grad_norm": 5.72932767868042,
      "learning_rate": 2.4827378772820005e-05,
      "loss": 0.4208,
      "step": 19350
    },
    {
      "epoch": 1.5151040851463453,
      "grad_norm": 8.352985382080078,
      "learning_rate": 2.481430140712455e-05,
      "loss": 0.8096,
      "step": 19360
    },
    {
      "epoch": 1.5158866802316482,
      "grad_norm": 6.010400295257568,
      "learning_rate": 2.4801224041429098e-05,
      "loss": 0.8137,
      "step": 19370
    },
    {
      "epoch": 1.516669275316951,
      "grad_norm": 1.238579511642456,
      "learning_rate": 2.4788146675733642e-05,
      "loss": 0.8548,
      "step": 19380
    },
    {
      "epoch": 1.5174518704022537,
      "grad_norm": 20.04273223876953,
      "learning_rate": 2.4775069310038187e-05,
      "loss": 0.796,
      "step": 19390
    },
    {
      "epoch": 1.5182344654875566,
      "grad_norm": 9.148284912109375,
      "learning_rate": 2.476199194434273e-05,
      "loss": 0.6052,
      "step": 19400
    },
    {
      "epoch": 1.5190170605728595,
      "grad_norm": 6.862086772918701,
      "learning_rate": 2.474891457864728e-05,
      "loss": 0.9477,
      "step": 19410
    },
    {
      "epoch": 1.5197996556581623,
      "grad_norm": 13.698674201965332,
      "learning_rate": 2.4735837212951824e-05,
      "loss": 0.7981,
      "step": 19420
    },
    {
      "epoch": 1.5205822507434652,
      "grad_norm": 8.718360900878906,
      "learning_rate": 2.472275984725637e-05,
      "loss": 0.7809,
      "step": 19430
    },
    {
      "epoch": 1.521364845828768,
      "grad_norm": 4.186866283416748,
      "learning_rate": 2.4709682481560916e-05,
      "loss": 0.7392,
      "step": 19440
    },
    {
      "epoch": 1.522147440914071,
      "grad_norm": 6.426671981811523,
      "learning_rate": 2.469660511586546e-05,
      "loss": 0.7531,
      "step": 19450
    },
    {
      "epoch": 1.5229300359993738,
      "grad_norm": 4.425591468811035,
      "learning_rate": 2.468352775017001e-05,
      "loss": 0.913,
      "step": 19460
    },
    {
      "epoch": 1.5237126310846767,
      "grad_norm": 5.4722418785095215,
      "learning_rate": 2.4670450384474554e-05,
      "loss": 0.738,
      "step": 19470
    },
    {
      "epoch": 1.5244952261699796,
      "grad_norm": 10.503229141235352,
      "learning_rate": 2.4657373018779098e-05,
      "loss": 0.6016,
      "step": 19480
    },
    {
      "epoch": 1.5252778212552824,
      "grad_norm": 14.99528694152832,
      "learning_rate": 2.4644295653083643e-05,
      "loss": 0.6193,
      "step": 19490
    },
    {
      "epoch": 1.5260604163405853,
      "grad_norm": 11.173569679260254,
      "learning_rate": 2.4631218287388187e-05,
      "loss": 1.0247,
      "step": 19500
    },
    {
      "epoch": 1.5268430114258882,
      "grad_norm": 1.8646844625473022,
      "learning_rate": 2.4618140921692735e-05,
      "loss": 0.8838,
      "step": 19510
    },
    {
      "epoch": 1.527625606511191,
      "grad_norm": 4.467301368713379,
      "learning_rate": 2.4605063555997283e-05,
      "loss": 0.676,
      "step": 19520
    },
    {
      "epoch": 1.528408201596494,
      "grad_norm": 13.702070236206055,
      "learning_rate": 2.4591986190301828e-05,
      "loss": 0.4141,
      "step": 19530
    },
    {
      "epoch": 1.5291907966817968,
      "grad_norm": 26.530975341796875,
      "learning_rate": 2.4578908824606372e-05,
      "loss": 0.5477,
      "step": 19540
    },
    {
      "epoch": 1.5299733917670997,
      "grad_norm": 11.720985412597656,
      "learning_rate": 2.4565831458910917e-05,
      "loss": 0.8143,
      "step": 19550
    },
    {
      "epoch": 1.5307559868524026,
      "grad_norm": 3.531208038330078,
      "learning_rate": 2.455275409321546e-05,
      "loss": 0.7718,
      "step": 19560
    },
    {
      "epoch": 1.5315385819377054,
      "grad_norm": 5.517270088195801,
      "learning_rate": 2.453967672752001e-05,
      "loss": 0.5031,
      "step": 19570
    },
    {
      "epoch": 1.5323211770230083,
      "grad_norm": 4.397434711456299,
      "learning_rate": 2.4526599361824557e-05,
      "loss": 0.7596,
      "step": 19580
    },
    {
      "epoch": 1.5331037721083112,
      "grad_norm": 12.220331192016602,
      "learning_rate": 2.4513521996129102e-05,
      "loss": 0.8539,
      "step": 19590
    },
    {
      "epoch": 1.533886367193614,
      "grad_norm": 10.373991966247559,
      "learning_rate": 2.4500444630433647e-05,
      "loss": 0.6526,
      "step": 19600
    },
    {
      "epoch": 1.534668962278917,
      "grad_norm": 12.066545486450195,
      "learning_rate": 2.448736726473819e-05,
      "loss": 0.7428,
      "step": 19610
    },
    {
      "epoch": 1.5354515573642198,
      "grad_norm": 8.046629905700684,
      "learning_rate": 2.447428989904274e-05,
      "loss": 0.7545,
      "step": 19620
    },
    {
      "epoch": 1.5362341524495227,
      "grad_norm": 1.765020489692688,
      "learning_rate": 2.4461212533347284e-05,
      "loss": 0.6213,
      "step": 19630
    },
    {
      "epoch": 1.5370167475348255,
      "grad_norm": 3.997912883758545,
      "learning_rate": 2.4448135167651828e-05,
      "loss": 0.9871,
      "step": 19640
    },
    {
      "epoch": 1.5377993426201284,
      "grad_norm": 9.606061935424805,
      "learning_rate": 2.4435057801956376e-05,
      "loss": 0.6432,
      "step": 19650
    },
    {
      "epoch": 1.5385819377054313,
      "grad_norm": 3.5107696056365967,
      "learning_rate": 2.442198043626092e-05,
      "loss": 0.6806,
      "step": 19660
    },
    {
      "epoch": 1.5393645327907342,
      "grad_norm": 6.338613986968994,
      "learning_rate": 2.4408903070565465e-05,
      "loss": 0.566,
      "step": 19670
    },
    {
      "epoch": 1.540147127876037,
      "grad_norm": 5.061330795288086,
      "learning_rate": 2.4395825704870013e-05,
      "loss": 0.9743,
      "step": 19680
    },
    {
      "epoch": 1.54092972296134,
      "grad_norm": 6.124874591827393,
      "learning_rate": 2.4382748339174558e-05,
      "loss": 0.9071,
      "step": 19690
    },
    {
      "epoch": 1.5417123180466428,
      "grad_norm": 12.31804084777832,
      "learning_rate": 2.4369670973479102e-05,
      "loss": 0.9477,
      "step": 19700
    },
    {
      "epoch": 1.5424949131319456,
      "grad_norm": 11.410151481628418,
      "learning_rate": 2.4356593607783647e-05,
      "loss": 1.0218,
      "step": 19710
    },
    {
      "epoch": 1.5432775082172485,
      "grad_norm": 6.529988765716553,
      "learning_rate": 2.4343516242088195e-05,
      "loss": 1.0344,
      "step": 19720
    },
    {
      "epoch": 1.5440601033025514,
      "grad_norm": 13.492829322814941,
      "learning_rate": 2.4330438876392743e-05,
      "loss": 0.9372,
      "step": 19730
    },
    {
      "epoch": 1.5448426983878543,
      "grad_norm": 33.18148422241211,
      "learning_rate": 2.4317361510697288e-05,
      "loss": 0.649,
      "step": 19740
    },
    {
      "epoch": 1.5456252934731571,
      "grad_norm": 6.827428817749023,
      "learning_rate": 2.4304284145001832e-05,
      "loss": 0.6975,
      "step": 19750
    },
    {
      "epoch": 1.54640788855846,
      "grad_norm": 13.31991958618164,
      "learning_rate": 2.4291206779306377e-05,
      "loss": 0.4329,
      "step": 19760
    },
    {
      "epoch": 1.5471904836437629,
      "grad_norm": 2.005964517593384,
      "learning_rate": 2.427812941361092e-05,
      "loss": 0.6788,
      "step": 19770
    },
    {
      "epoch": 1.5479730787290655,
      "grad_norm": 14.026615142822266,
      "learning_rate": 2.426505204791547e-05,
      "loss": 0.8904,
      "step": 19780
    },
    {
      "epoch": 1.5487556738143684,
      "grad_norm": 13.49682903289795,
      "learning_rate": 2.4251974682220017e-05,
      "loss": 0.7525,
      "step": 19790
    },
    {
      "epoch": 1.5495382688996713,
      "grad_norm": 11.233592987060547,
      "learning_rate": 2.4238897316524562e-05,
      "loss": 0.4871,
      "step": 19800
    },
    {
      "epoch": 1.5503208639849742,
      "grad_norm": 6.452766418457031,
      "learning_rate": 2.4225819950829106e-05,
      "loss": 0.7403,
      "step": 19810
    },
    {
      "epoch": 1.551103459070277,
      "grad_norm": 2.843595504760742,
      "learning_rate": 2.421274258513365e-05,
      "loss": 0.9103,
      "step": 19820
    },
    {
      "epoch": 1.55188605415558,
      "grad_norm": 1.1563888788223267,
      "learning_rate": 2.41996652194382e-05,
      "loss": 0.8852,
      "step": 19830
    },
    {
      "epoch": 1.5526686492408828,
      "grad_norm": 8.249170303344727,
      "learning_rate": 2.4186587853742743e-05,
      "loss": 0.7983,
      "step": 19840
    },
    {
      "epoch": 1.5534512443261856,
      "grad_norm": 6.359381675720215,
      "learning_rate": 2.4173510488047288e-05,
      "loss": 0.7567,
      "step": 19850
    },
    {
      "epoch": 1.5542338394114885,
      "grad_norm": 10.136160850524902,
      "learning_rate": 2.4160433122351836e-05,
      "loss": 0.8509,
      "step": 19860
    },
    {
      "epoch": 1.5550164344967914,
      "grad_norm": 14.01846981048584,
      "learning_rate": 2.414735575665638e-05,
      "loss": 0.4818,
      "step": 19870
    },
    {
      "epoch": 1.5557990295820943,
      "grad_norm": 8.055374145507812,
      "learning_rate": 2.4134278390960925e-05,
      "loss": 0.7166,
      "step": 19880
    },
    {
      "epoch": 1.5565816246673971,
      "grad_norm": 10.074016571044922,
      "learning_rate": 2.4121201025265473e-05,
      "loss": 0.6878,
      "step": 19890
    },
    {
      "epoch": 1.5573642197527,
      "grad_norm": 0.9911321997642517,
      "learning_rate": 2.4108123659570018e-05,
      "loss": 0.7649,
      "step": 19900
    },
    {
      "epoch": 1.5581468148380027,
      "grad_norm": 10.318758010864258,
      "learning_rate": 2.4095046293874562e-05,
      "loss": 0.7451,
      "step": 19910
    },
    {
      "epoch": 1.5589294099233055,
      "grad_norm": 4.284429550170898,
      "learning_rate": 2.4081968928179107e-05,
      "loss": 0.8378,
      "step": 19920
    },
    {
      "epoch": 1.5597120050086084,
      "grad_norm": 6.598245143890381,
      "learning_rate": 2.4068891562483655e-05,
      "loss": 0.5224,
      "step": 19930
    },
    {
      "epoch": 1.5604946000939113,
      "grad_norm": 7.8659234046936035,
      "learning_rate": 2.4055814196788203e-05,
      "loss": 0.5488,
      "step": 19940
    },
    {
      "epoch": 1.5612771951792142,
      "grad_norm": 1.983095645904541,
      "learning_rate": 2.4042736831092747e-05,
      "loss": 1.061,
      "step": 19950
    },
    {
      "epoch": 1.562059790264517,
      "grad_norm": 0.6626718640327454,
      "learning_rate": 2.4029659465397292e-05,
      "loss": 0.6955,
      "step": 19960
    },
    {
      "epoch": 1.56284238534982,
      "grad_norm": 13.388556480407715,
      "learning_rate": 2.4016582099701836e-05,
      "loss": 0.9838,
      "step": 19970
    },
    {
      "epoch": 1.5636249804351228,
      "grad_norm": 6.437559604644775,
      "learning_rate": 2.400350473400638e-05,
      "loss": 0.8324,
      "step": 19980
    },
    {
      "epoch": 1.5644075755204256,
      "grad_norm": 7.523143768310547,
      "learning_rate": 2.399042736831093e-05,
      "loss": 0.8577,
      "step": 19990
    },
    {
      "epoch": 1.5651901706057285,
      "grad_norm": 13.151342391967773,
      "learning_rate": 2.3977350002615473e-05,
      "loss": 1.0593,
      "step": 20000
    },
    {
      "epoch": 1.5659727656910314,
      "grad_norm": 16.736045837402344,
      "learning_rate": 2.396427263692002e-05,
      "loss": 0.9376,
      "step": 20010
    },
    {
      "epoch": 1.5667553607763343,
      "grad_norm": 5.390063285827637,
      "learning_rate": 2.3951195271224566e-05,
      "loss": 0.7256,
      "step": 20020
    },
    {
      "epoch": 1.5675379558616371,
      "grad_norm": 8.15056324005127,
      "learning_rate": 2.393811790552911e-05,
      "loss": 0.8202,
      "step": 20030
    },
    {
      "epoch": 1.56832055094694,
      "grad_norm": 3.020688533782959,
      "learning_rate": 2.392504053983366e-05,
      "loss": 0.7763,
      "step": 20040
    },
    {
      "epoch": 1.5691031460322429,
      "grad_norm": 3.355116367340088,
      "learning_rate": 2.3911963174138203e-05,
      "loss": 0.6682,
      "step": 20050
    },
    {
      "epoch": 1.5698857411175458,
      "grad_norm": 6.480714797973633,
      "learning_rate": 2.3898885808442748e-05,
      "loss": 0.6977,
      "step": 20060
    },
    {
      "epoch": 1.5706683362028486,
      "grad_norm": 9.62375259399414,
      "learning_rate": 2.3885808442747292e-05,
      "loss": 0.7447,
      "step": 20070
    },
    {
      "epoch": 1.5714509312881515,
      "grad_norm": 9.04086971282959,
      "learning_rate": 2.387273107705184e-05,
      "loss": 0.7715,
      "step": 20080
    },
    {
      "epoch": 1.5722335263734544,
      "grad_norm": 3.0661797523498535,
      "learning_rate": 2.3859653711356385e-05,
      "loss": 0.6251,
      "step": 20090
    },
    {
      "epoch": 1.5730161214587572,
      "grad_norm": 4.084636211395264,
      "learning_rate": 2.3846576345660933e-05,
      "loss": 0.7183,
      "step": 20100
    },
    {
      "epoch": 1.5737987165440601,
      "grad_norm": 13.316582679748535,
      "learning_rate": 2.3833498979965477e-05,
      "loss": 0.77,
      "step": 20110
    },
    {
      "epoch": 1.574581311629363,
      "grad_norm": 4.889891147613525,
      "learning_rate": 2.3820421614270022e-05,
      "loss": 0.7097,
      "step": 20120
    },
    {
      "epoch": 1.5753639067146659,
      "grad_norm": 5.359167575836182,
      "learning_rate": 2.3807344248574566e-05,
      "loss": 0.9158,
      "step": 20130
    },
    {
      "epoch": 1.5761465017999687,
      "grad_norm": 5.012209892272949,
      "learning_rate": 2.379426688287911e-05,
      "loss": 0.6869,
      "step": 20140
    },
    {
      "epoch": 1.5769290968852716,
      "grad_norm": 4.747707843780518,
      "learning_rate": 2.3781189517183662e-05,
      "loss": 0.8063,
      "step": 20150
    },
    {
      "epoch": 1.5777116919705745,
      "grad_norm": 3.437080144882202,
      "learning_rate": 2.3768112151488207e-05,
      "loss": 0.8032,
      "step": 20160
    },
    {
      "epoch": 1.5784942870558774,
      "grad_norm": 10.857711791992188,
      "learning_rate": 2.375503478579275e-05,
      "loss": 0.5483,
      "step": 20170
    },
    {
      "epoch": 1.5792768821411802,
      "grad_norm": 4.088388442993164,
      "learning_rate": 2.3741957420097296e-05,
      "loss": 0.4,
      "step": 20180
    },
    {
      "epoch": 1.580059477226483,
      "grad_norm": 8.135491371154785,
      "learning_rate": 2.372888005440184e-05,
      "loss": 0.9138,
      "step": 20190
    },
    {
      "epoch": 1.580842072311786,
      "grad_norm": 6.258747100830078,
      "learning_rate": 2.371580268870639e-05,
      "loss": 0.758,
      "step": 20200
    },
    {
      "epoch": 1.5816246673970888,
      "grad_norm": 7.605544090270996,
      "learning_rate": 2.3702725323010933e-05,
      "loss": 0.8143,
      "step": 20210
    },
    {
      "epoch": 1.5824072624823917,
      "grad_norm": 10.306143760681152,
      "learning_rate": 2.368964795731548e-05,
      "loss": 0.9073,
      "step": 20220
    },
    {
      "epoch": 1.5831898575676946,
      "grad_norm": 5.060551166534424,
      "learning_rate": 2.3676570591620026e-05,
      "loss": 0.7511,
      "step": 20230
    },
    {
      "epoch": 1.5839724526529975,
      "grad_norm": 9.770401000976562,
      "learning_rate": 2.366349322592457e-05,
      "loss": 0.8579,
      "step": 20240
    },
    {
      "epoch": 1.5847550477383003,
      "grad_norm": 1.6828912496566772,
      "learning_rate": 2.3650415860229115e-05,
      "loss": 0.6583,
      "step": 20250
    },
    {
      "epoch": 1.5855376428236032,
      "grad_norm": 5.8892669677734375,
      "learning_rate": 2.3637338494533663e-05,
      "loss": 0.4187,
      "step": 20260
    },
    {
      "epoch": 1.586320237908906,
      "grad_norm": 8.474958419799805,
      "learning_rate": 2.3624261128838207e-05,
      "loss": 0.7659,
      "step": 20270
    },
    {
      "epoch": 1.587102832994209,
      "grad_norm": 9.815071105957031,
      "learning_rate": 2.3611183763142752e-05,
      "loss": 1.1255,
      "step": 20280
    },
    {
      "epoch": 1.5878854280795116,
      "grad_norm": 3.29111647605896,
      "learning_rate": 2.35981063974473e-05,
      "loss": 0.7697,
      "step": 20290
    },
    {
      "epoch": 1.5886680231648145,
      "grad_norm": 4.7818098068237305,
      "learning_rate": 2.3585029031751845e-05,
      "loss": 0.7684,
      "step": 20300
    },
    {
      "epoch": 1.5894506182501174,
      "grad_norm": 6.466434955596924,
      "learning_rate": 2.3571951666056392e-05,
      "loss": 0.6533,
      "step": 20310
    },
    {
      "epoch": 1.5902332133354202,
      "grad_norm": 5.470525741577148,
      "learning_rate": 2.3558874300360937e-05,
      "loss": 0.9342,
      "step": 20320
    },
    {
      "epoch": 1.591015808420723,
      "grad_norm": 5.92152214050293,
      "learning_rate": 2.354579693466548e-05,
      "loss": 0.5385,
      "step": 20330
    },
    {
      "epoch": 1.591798403506026,
      "grad_norm": 6.243565082550049,
      "learning_rate": 2.3532719568970026e-05,
      "loss": 0.8572,
      "step": 20340
    },
    {
      "epoch": 1.5925809985913288,
      "grad_norm": 4.617930889129639,
      "learning_rate": 2.351964220327457e-05,
      "loss": 0.6347,
      "step": 20350
    },
    {
      "epoch": 1.5933635936766317,
      "grad_norm": 5.072941303253174,
      "learning_rate": 2.350656483757912e-05,
      "loss": 0.8249,
      "step": 20360
    },
    {
      "epoch": 1.5941461887619346,
      "grad_norm": 2.2251815795898438,
      "learning_rate": 2.3493487471883667e-05,
      "loss": 0.5903,
      "step": 20370
    },
    {
      "epoch": 1.5949287838472375,
      "grad_norm": 12.08675765991211,
      "learning_rate": 2.348041010618821e-05,
      "loss": 0.5301,
      "step": 20380
    },
    {
      "epoch": 1.5957113789325403,
      "grad_norm": 16.778545379638672,
      "learning_rate": 2.3467332740492756e-05,
      "loss": 0.7815,
      "step": 20390
    },
    {
      "epoch": 1.5964939740178432,
      "grad_norm": 1.2624973058700562,
      "learning_rate": 2.34542553747973e-05,
      "loss": 0.8007,
      "step": 20400
    },
    {
      "epoch": 1.597276569103146,
      "grad_norm": 13.04897689819336,
      "learning_rate": 2.344117800910185e-05,
      "loss": 1.2245,
      "step": 20410
    },
    {
      "epoch": 1.5980591641884487,
      "grad_norm": 15.48123836517334,
      "learning_rate": 2.3428100643406393e-05,
      "loss": 0.8423,
      "step": 20420
    },
    {
      "epoch": 1.5988417592737516,
      "grad_norm": 4.090372085571289,
      "learning_rate": 2.341502327771094e-05,
      "loss": 0.596,
      "step": 20430
    },
    {
      "epoch": 1.5996243543590545,
      "grad_norm": 0.6069270372390747,
      "learning_rate": 2.3401945912015485e-05,
      "loss": 0.8518,
      "step": 20440
    },
    {
      "epoch": 1.6004069494443574,
      "grad_norm": 5.6707072257995605,
      "learning_rate": 2.338886854632003e-05,
      "loss": 0.7239,
      "step": 20450
    },
    {
      "epoch": 1.6011895445296602,
      "grad_norm": 2.4384067058563232,
      "learning_rate": 2.3375791180624575e-05,
      "loss": 0.6857,
      "step": 20460
    },
    {
      "epoch": 1.601972139614963,
      "grad_norm": 8.115840911865234,
      "learning_rate": 2.3362713814929123e-05,
      "loss": 0.8459,
      "step": 20470
    },
    {
      "epoch": 1.602754734700266,
      "grad_norm": 3.2026963233947754,
      "learning_rate": 2.3349636449233667e-05,
      "loss": 0.9038,
      "step": 20480
    },
    {
      "epoch": 1.6035373297855688,
      "grad_norm": 2.5028879642486572,
      "learning_rate": 2.3336559083538212e-05,
      "loss": 0.6215,
      "step": 20490
    },
    {
      "epoch": 1.6043199248708717,
      "grad_norm": 7.38784122467041,
      "learning_rate": 2.332348171784276e-05,
      "loss": 0.4896,
      "step": 20500
    },
    {
      "epoch": 1.6051025199561746,
      "grad_norm": 7.045618534088135,
      "learning_rate": 2.3310404352147304e-05,
      "loss": 0.8067,
      "step": 20510
    },
    {
      "epoch": 1.6058851150414775,
      "grad_norm": 14.172012329101562,
      "learning_rate": 2.3297326986451852e-05,
      "loss": 0.8599,
      "step": 20520
    },
    {
      "epoch": 1.6066677101267803,
      "grad_norm": 1.0999268293380737,
      "learning_rate": 2.3284249620756397e-05,
      "loss": 0.4864,
      "step": 20530
    },
    {
      "epoch": 1.6074503052120832,
      "grad_norm": 17.632755279541016,
      "learning_rate": 2.327117225506094e-05,
      "loss": 0.9369,
      "step": 20540
    },
    {
      "epoch": 1.608232900297386,
      "grad_norm": 5.9907546043396,
      "learning_rate": 2.3258094889365486e-05,
      "loss": 0.7103,
      "step": 20550
    },
    {
      "epoch": 1.609015495382689,
      "grad_norm": 8.430891036987305,
      "learning_rate": 2.324501752367003e-05,
      "loss": 1.1994,
      "step": 20560
    },
    {
      "epoch": 1.6097980904679918,
      "grad_norm": 0.671229362487793,
      "learning_rate": 2.323194015797458e-05,
      "loss": 0.9069,
      "step": 20570
    },
    {
      "epoch": 1.6105806855532947,
      "grad_norm": 19.41168785095215,
      "learning_rate": 2.3218862792279126e-05,
      "loss": 1.0067,
      "step": 20580
    },
    {
      "epoch": 1.6113632806385976,
      "grad_norm": 6.75885534286499,
      "learning_rate": 2.320578542658367e-05,
      "loss": 0.8823,
      "step": 20590
    },
    {
      "epoch": 1.6121458757239004,
      "grad_norm": 7.760869979858398,
      "learning_rate": 2.3192708060888216e-05,
      "loss": 0.7486,
      "step": 20600
    },
    {
      "epoch": 1.6129284708092033,
      "grad_norm": 20.582530975341797,
      "learning_rate": 2.317963069519276e-05,
      "loss": 1.0413,
      "step": 20610
    },
    {
      "epoch": 1.6137110658945062,
      "grad_norm": 3.8748583793640137,
      "learning_rate": 2.3166553329497305e-05,
      "loss": 0.8325,
      "step": 20620
    },
    {
      "epoch": 1.614493660979809,
      "grad_norm": 4.648910999298096,
      "learning_rate": 2.3153475963801853e-05,
      "loss": 0.6188,
      "step": 20630
    },
    {
      "epoch": 1.615276256065112,
      "grad_norm": 47.88683319091797,
      "learning_rate": 2.31403985981064e-05,
      "loss": 0.7478,
      "step": 20640
    },
    {
      "epoch": 1.6160588511504148,
      "grad_norm": 6.098818778991699,
      "learning_rate": 2.3127321232410945e-05,
      "loss": 0.6032,
      "step": 20650
    },
    {
      "epoch": 1.6168414462357177,
      "grad_norm": 8.369281768798828,
      "learning_rate": 2.311424386671549e-05,
      "loss": 0.908,
      "step": 20660
    },
    {
      "epoch": 1.6176240413210206,
      "grad_norm": 6.431020259857178,
      "learning_rate": 2.3101166501020034e-05,
      "loss": 0.9844,
      "step": 20670
    },
    {
      "epoch": 1.6184066364063234,
      "grad_norm": 8.388540267944336,
      "learning_rate": 2.3088089135324582e-05,
      "loss": 0.956,
      "step": 20680
    },
    {
      "epoch": 1.6191892314916263,
      "grad_norm": 8.34304428100586,
      "learning_rate": 2.3075011769629127e-05,
      "loss": 0.8993,
      "step": 20690
    },
    {
      "epoch": 1.6199718265769292,
      "grad_norm": 4.830888271331787,
      "learning_rate": 2.306193440393367e-05,
      "loss": 0.5513,
      "step": 20700
    },
    {
      "epoch": 1.620754421662232,
      "grad_norm": 15.154412269592285,
      "learning_rate": 2.304885703823822e-05,
      "loss": 0.6681,
      "step": 20710
    },
    {
      "epoch": 1.621537016747535,
      "grad_norm": 9.322186470031738,
      "learning_rate": 2.3035779672542764e-05,
      "loss": 0.6393,
      "step": 20720
    },
    {
      "epoch": 1.6223196118328378,
      "grad_norm": 0.6010481715202332,
      "learning_rate": 2.302270230684731e-05,
      "loss": 0.5486,
      "step": 20730
    },
    {
      "epoch": 1.6231022069181407,
      "grad_norm": 8.608955383300781,
      "learning_rate": 2.3009624941151857e-05,
      "loss": 0.3883,
      "step": 20740
    },
    {
      "epoch": 1.6238848020034435,
      "grad_norm": 7.38911771774292,
      "learning_rate": 2.29965475754564e-05,
      "loss": 0.671,
      "step": 20750
    },
    {
      "epoch": 1.6246673970887464,
      "grad_norm": 5.183398723602295,
      "learning_rate": 2.2983470209760946e-05,
      "loss": 0.7175,
      "step": 20760
    },
    {
      "epoch": 1.6254499921740493,
      "grad_norm": 1.9231162071228027,
      "learning_rate": 2.297039284406549e-05,
      "loss": 0.4149,
      "step": 20770
    },
    {
      "epoch": 1.6262325872593522,
      "grad_norm": 5.724416255950928,
      "learning_rate": 2.2957315478370038e-05,
      "loss": 0.9182,
      "step": 20780
    },
    {
      "epoch": 1.627015182344655,
      "grad_norm": 28.55622673034668,
      "learning_rate": 2.2944238112674586e-05,
      "loss": 0.6849,
      "step": 20790
    },
    {
      "epoch": 1.627797777429958,
      "grad_norm": 4.702361106872559,
      "learning_rate": 2.293116074697913e-05,
      "loss": 0.8383,
      "step": 20800
    },
    {
      "epoch": 1.6285803725152606,
      "grad_norm": 2.0235321521759033,
      "learning_rate": 2.2918083381283675e-05,
      "loss": 0.8116,
      "step": 20810
    },
    {
      "epoch": 1.6293629676005634,
      "grad_norm": 5.782841205596924,
      "learning_rate": 2.290500601558822e-05,
      "loss": 1.0554,
      "step": 20820
    },
    {
      "epoch": 1.6301455626858663,
      "grad_norm": 18.762319564819336,
      "learning_rate": 2.2891928649892764e-05,
      "loss": 0.9302,
      "step": 20830
    },
    {
      "epoch": 1.6309281577711692,
      "grad_norm": 1.9226845502853394,
      "learning_rate": 2.2878851284197312e-05,
      "loss": 0.5856,
      "step": 20840
    },
    {
      "epoch": 1.631710752856472,
      "grad_norm": 6.762653827667236,
      "learning_rate": 2.286577391850186e-05,
      "loss": 0.9733,
      "step": 20850
    },
    {
      "epoch": 1.632493347941775,
      "grad_norm": 10.515716552734375,
      "learning_rate": 2.2852696552806405e-05,
      "loss": 1.0432,
      "step": 20860
    },
    {
      "epoch": 1.6332759430270778,
      "grad_norm": 3.658238410949707,
      "learning_rate": 2.283961918711095e-05,
      "loss": 0.7528,
      "step": 20870
    },
    {
      "epoch": 1.6340585381123807,
      "grad_norm": 4.770216464996338,
      "learning_rate": 2.2826541821415494e-05,
      "loss": 0.8269,
      "step": 20880
    },
    {
      "epoch": 1.6348411331976835,
      "grad_norm": 2.1927719116210938,
      "learning_rate": 2.2813464455720042e-05,
      "loss": 0.5411,
      "step": 20890
    },
    {
      "epoch": 1.6356237282829864,
      "grad_norm": 6.749607563018799,
      "learning_rate": 2.2800387090024587e-05,
      "loss": 1.0468,
      "step": 20900
    },
    {
      "epoch": 1.6364063233682893,
      "grad_norm": 4.4417314529418945,
      "learning_rate": 2.278730972432913e-05,
      "loss": 0.8218,
      "step": 20910
    },
    {
      "epoch": 1.6371889184535922,
      "grad_norm": 6.933020114898682,
      "learning_rate": 2.277423235863368e-05,
      "loss": 0.7425,
      "step": 20920
    },
    {
      "epoch": 1.637971513538895,
      "grad_norm": 9.781383514404297,
      "learning_rate": 2.2761154992938224e-05,
      "loss": 0.8605,
      "step": 20930
    },
    {
      "epoch": 1.6387541086241977,
      "grad_norm": 9.740824699401855,
      "learning_rate": 2.274807762724277e-05,
      "loss": 0.9492,
      "step": 20940
    },
    {
      "epoch": 1.6395367037095006,
      "grad_norm": 10.359014511108398,
      "learning_rate": 2.2735000261547316e-05,
      "loss": 0.8693,
      "step": 20950
    },
    {
      "epoch": 1.6403192987948034,
      "grad_norm": 7.69581937789917,
      "learning_rate": 2.272192289585186e-05,
      "loss": 0.8257,
      "step": 20960
    },
    {
      "epoch": 1.6411018938801063,
      "grad_norm": 1.8388097286224365,
      "learning_rate": 2.2708845530156405e-05,
      "loss": 0.961,
      "step": 20970
    },
    {
      "epoch": 1.6418844889654092,
      "grad_norm": 16.127275466918945,
      "learning_rate": 2.269576816446095e-05,
      "loss": 0.7758,
      "step": 20980
    },
    {
      "epoch": 1.642667084050712,
      "grad_norm": 5.231544017791748,
      "learning_rate": 2.2682690798765498e-05,
      "loss": 0.8958,
      "step": 20990
    },
    {
      "epoch": 1.643449679136015,
      "grad_norm": 10.003687858581543,
      "learning_rate": 2.2669613433070046e-05,
      "loss": 0.8815,
      "step": 21000
    },
    {
      "epoch": 1.6442322742213178,
      "grad_norm": 8.429361343383789,
      "learning_rate": 2.265653606737459e-05,
      "loss": 0.9845,
      "step": 21010
    },
    {
      "epoch": 1.6450148693066207,
      "grad_norm": 0.8681591153144836,
      "learning_rate": 2.2643458701679135e-05,
      "loss": 0.5476,
      "step": 21020
    },
    {
      "epoch": 1.6457974643919235,
      "grad_norm": 9.575735092163086,
      "learning_rate": 2.263038133598368e-05,
      "loss": 0.9117,
      "step": 21030
    },
    {
      "epoch": 1.6465800594772264,
      "grad_norm": 3.355271100997925,
      "learning_rate": 2.2617303970288224e-05,
      "loss": 0.7127,
      "step": 21040
    },
    {
      "epoch": 1.6473626545625293,
      "grad_norm": 3.7823095321655273,
      "learning_rate": 2.2604226604592772e-05,
      "loss": 0.9216,
      "step": 21050
    },
    {
      "epoch": 1.6481452496478322,
      "grad_norm": 6.967016220092773,
      "learning_rate": 2.259114923889732e-05,
      "loss": 0.714,
      "step": 21060
    },
    {
      "epoch": 1.648927844733135,
      "grad_norm": 3.4683027267456055,
      "learning_rate": 2.2578071873201865e-05,
      "loss": 0.6713,
      "step": 21070
    },
    {
      "epoch": 1.649710439818438,
      "grad_norm": 4.0077738761901855,
      "learning_rate": 2.256499450750641e-05,
      "loss": 0.8007,
      "step": 21080
    },
    {
      "epoch": 1.6504930349037408,
      "grad_norm": 9.132213592529297,
      "learning_rate": 2.2551917141810954e-05,
      "loss": 0.9134,
      "step": 21090
    },
    {
      "epoch": 1.6512756299890436,
      "grad_norm": 4.193911075592041,
      "learning_rate": 2.2538839776115502e-05,
      "loss": 0.8581,
      "step": 21100
    },
    {
      "epoch": 1.6520582250743465,
      "grad_norm": 26.271717071533203,
      "learning_rate": 2.2525762410420046e-05,
      "loss": 0.5748,
      "step": 21110
    },
    {
      "epoch": 1.6528408201596494,
      "grad_norm": 11.94161605834961,
      "learning_rate": 2.251268504472459e-05,
      "loss": 0.7842,
      "step": 21120
    },
    {
      "epoch": 1.6536234152449523,
      "grad_norm": 15.609495162963867,
      "learning_rate": 2.249960767902914e-05,
      "loss": 1.1214,
      "step": 21130
    },
    {
      "epoch": 1.6544060103302551,
      "grad_norm": 0.5881310701370239,
      "learning_rate": 2.2486530313333683e-05,
      "loss": 0.8989,
      "step": 21140
    },
    {
      "epoch": 1.655188605415558,
      "grad_norm": 2.157937526702881,
      "learning_rate": 2.2473452947638228e-05,
      "loss": 1.0215,
      "step": 21150
    },
    {
      "epoch": 1.6559712005008609,
      "grad_norm": 10.33519458770752,
      "learning_rate": 2.2460375581942776e-05,
      "loss": 0.7876,
      "step": 21160
    },
    {
      "epoch": 1.6567537955861638,
      "grad_norm": 8.212785720825195,
      "learning_rate": 2.244729821624732e-05,
      "loss": 0.6264,
      "step": 21170
    },
    {
      "epoch": 1.6575363906714666,
      "grad_norm": 5.035301685333252,
      "learning_rate": 2.2434220850551865e-05,
      "loss": 0.8633,
      "step": 21180
    },
    {
      "epoch": 1.6583189857567695,
      "grad_norm": 9.68832778930664,
      "learning_rate": 2.242114348485641e-05,
      "loss": 0.986,
      "step": 21190
    },
    {
      "epoch": 1.6591015808420724,
      "grad_norm": 8.666345596313477,
      "learning_rate": 2.2408066119160958e-05,
      "loss": 0.5802,
      "step": 21200
    },
    {
      "epoch": 1.6598841759273752,
      "grad_norm": 4.95186185836792,
      "learning_rate": 2.2394988753465506e-05,
      "loss": 0.6909,
      "step": 21210
    },
    {
      "epoch": 1.6606667710126781,
      "grad_norm": 5.6905717849731445,
      "learning_rate": 2.238191138777005e-05,
      "loss": 0.7222,
      "step": 21220
    },
    {
      "epoch": 1.661449366097981,
      "grad_norm": 9.105822563171387,
      "learning_rate": 2.2368834022074595e-05,
      "loss": 0.9991,
      "step": 21230
    },
    {
      "epoch": 1.6622319611832839,
      "grad_norm": 5.960890769958496,
      "learning_rate": 2.235575665637914e-05,
      "loss": 0.752,
      "step": 21240
    },
    {
      "epoch": 1.6630145562685867,
      "grad_norm": 8.561434745788574,
      "learning_rate": 2.2342679290683684e-05,
      "loss": 0.8759,
      "step": 21250
    },
    {
      "epoch": 1.6637971513538896,
      "grad_norm": 25.01967430114746,
      "learning_rate": 2.2329601924988232e-05,
      "loss": 1.0734,
      "step": 21260
    },
    {
      "epoch": 1.6645797464391925,
      "grad_norm": 8.416291236877441,
      "learning_rate": 2.2316524559292776e-05,
      "loss": 1.0517,
      "step": 21270
    },
    {
      "epoch": 1.6653623415244954,
      "grad_norm": 9.009618759155273,
      "learning_rate": 2.2303447193597324e-05,
      "loss": 0.8335,
      "step": 21280
    },
    {
      "epoch": 1.6661449366097982,
      "grad_norm": 3.2850663661956787,
      "learning_rate": 2.229036982790187e-05,
      "loss": 0.7159,
      "step": 21290
    },
    {
      "epoch": 1.666927531695101,
      "grad_norm": 6.480772018432617,
      "learning_rate": 2.2277292462206414e-05,
      "loss": 0.8137,
      "step": 21300
    },
    {
      "epoch": 1.667710126780404,
      "grad_norm": 4.898353576660156,
      "learning_rate": 2.2264215096510958e-05,
      "loss": 0.814,
      "step": 21310
    },
    {
      "epoch": 1.6684927218657066,
      "grad_norm": 8.886615753173828,
      "learning_rate": 2.2251137730815506e-05,
      "loss": 0.4694,
      "step": 21320
    },
    {
      "epoch": 1.6692753169510095,
      "grad_norm": 16.841367721557617,
      "learning_rate": 2.223806036512005e-05,
      "loss": 1.0728,
      "step": 21330
    },
    {
      "epoch": 1.6700579120363124,
      "grad_norm": 1.8575127124786377,
      "learning_rate": 2.2224982999424595e-05,
      "loss": 0.9912,
      "step": 21340
    },
    {
      "epoch": 1.6708405071216152,
      "grad_norm": 9.336725234985352,
      "learning_rate": 2.2211905633729143e-05,
      "loss": 0.418,
      "step": 21350
    },
    {
      "epoch": 1.6716231022069181,
      "grad_norm": 19.496456146240234,
      "learning_rate": 2.2198828268033688e-05,
      "loss": 0.8535,
      "step": 21360
    },
    {
      "epoch": 1.672405697292221,
      "grad_norm": 5.557608127593994,
      "learning_rate": 2.2185750902338236e-05,
      "loss": 0.6655,
      "step": 21370
    },
    {
      "epoch": 1.6731882923775239,
      "grad_norm": 13.436347961425781,
      "learning_rate": 2.217267353664278e-05,
      "loss": 0.7846,
      "step": 21380
    },
    {
      "epoch": 1.6739708874628267,
      "grad_norm": 5.314496040344238,
      "learning_rate": 2.2159596170947325e-05,
      "loss": 0.9104,
      "step": 21390
    },
    {
      "epoch": 1.6747534825481296,
      "grad_norm": 5.255336761474609,
      "learning_rate": 2.214651880525187e-05,
      "loss": 0.7613,
      "step": 21400
    },
    {
      "epoch": 1.6755360776334325,
      "grad_norm": 11.782753944396973,
      "learning_rate": 2.2133441439556414e-05,
      "loss": 0.743,
      "step": 21410
    },
    {
      "epoch": 1.6763186727187354,
      "grad_norm": 1.0165871381759644,
      "learning_rate": 2.2120364073860962e-05,
      "loss": 1.1479,
      "step": 21420
    },
    {
      "epoch": 1.6771012678040382,
      "grad_norm": 4.841526985168457,
      "learning_rate": 2.210728670816551e-05,
      "loss": 0.7975,
      "step": 21430
    },
    {
      "epoch": 1.677883862889341,
      "grad_norm": 8.632345199584961,
      "learning_rate": 2.2094209342470055e-05,
      "loss": 0.7946,
      "step": 21440
    },
    {
      "epoch": 1.6786664579746438,
      "grad_norm": 3.2662274837493896,
      "learning_rate": 2.20811319767746e-05,
      "loss": 0.72,
      "step": 21450
    },
    {
      "epoch": 1.6794490530599466,
      "grad_norm": 8.437361717224121,
      "learning_rate": 2.2068054611079144e-05,
      "loss": 0.8187,
      "step": 21460
    },
    {
      "epoch": 1.6802316481452495,
      "grad_norm": 3.9259533882141113,
      "learning_rate": 2.205497724538369e-05,
      "loss": 0.6944,
      "step": 21470
    },
    {
      "epoch": 1.6810142432305524,
      "grad_norm": 2.9823412895202637,
      "learning_rate": 2.2041899879688236e-05,
      "loss": 0.8645,
      "step": 21480
    },
    {
      "epoch": 1.6817968383158552,
      "grad_norm": 11.067787170410156,
      "learning_rate": 2.2028822513992784e-05,
      "loss": 0.6037,
      "step": 21490
    },
    {
      "epoch": 1.6825794334011581,
      "grad_norm": 2.235625743865967,
      "learning_rate": 2.201574514829733e-05,
      "loss": 0.6831,
      "step": 21500
    },
    {
      "epoch": 1.683362028486461,
      "grad_norm": 17.416568756103516,
      "learning_rate": 2.2002667782601873e-05,
      "loss": 0.907,
      "step": 21510
    },
    {
      "epoch": 1.6841446235717639,
      "grad_norm": 6.040261745452881,
      "learning_rate": 2.1989590416906418e-05,
      "loss": 0.7584,
      "step": 21520
    },
    {
      "epoch": 1.6849272186570667,
      "grad_norm": 8.22166633605957,
      "learning_rate": 2.1976513051210966e-05,
      "loss": 0.7679,
      "step": 21530
    },
    {
      "epoch": 1.6857098137423696,
      "grad_norm": 19.199665069580078,
      "learning_rate": 2.196343568551551e-05,
      "loss": 0.6676,
      "step": 21540
    },
    {
      "epoch": 1.6864924088276725,
      "grad_norm": 5.967693328857422,
      "learning_rate": 2.1950358319820055e-05,
      "loss": 0.6647,
      "step": 21550
    },
    {
      "epoch": 1.6872750039129754,
      "grad_norm": 4.679856777191162,
      "learning_rate": 2.1937280954124603e-05,
      "loss": 0.7943,
      "step": 21560
    },
    {
      "epoch": 1.6880575989982782,
      "grad_norm": 11.992086410522461,
      "learning_rate": 2.1924203588429148e-05,
      "loss": 0.6243,
      "step": 21570
    },
    {
      "epoch": 1.688840194083581,
      "grad_norm": 4.90042781829834,
      "learning_rate": 2.1911126222733695e-05,
      "loss": 0.8139,
      "step": 21580
    },
    {
      "epoch": 1.689622789168884,
      "grad_norm": 3.5325093269348145,
      "learning_rate": 2.189804885703824e-05,
      "loss": 0.977,
      "step": 21590
    },
    {
      "epoch": 1.6904053842541868,
      "grad_norm": 19.224172592163086,
      "learning_rate": 2.1884971491342785e-05,
      "loss": 1.0434,
      "step": 21600
    },
    {
      "epoch": 1.6911879793394897,
      "grad_norm": 4.8249125480651855,
      "learning_rate": 2.187189412564733e-05,
      "loss": 0.5567,
      "step": 21610
    },
    {
      "epoch": 1.6919705744247926,
      "grad_norm": 4.486117362976074,
      "learning_rate": 2.1858816759951874e-05,
      "loss": 0.7576,
      "step": 21620
    },
    {
      "epoch": 1.6927531695100955,
      "grad_norm": 1.410689115524292,
      "learning_rate": 2.1845739394256422e-05,
      "loss": 0.8642,
      "step": 21630
    },
    {
      "epoch": 1.6935357645953983,
      "grad_norm": 2.1602742671966553,
      "learning_rate": 2.183266202856097e-05,
      "loss": 0.7162,
      "step": 21640
    },
    {
      "epoch": 1.6943183596807012,
      "grad_norm": 13.51518440246582,
      "learning_rate": 2.1819584662865514e-05,
      "loss": 0.5898,
      "step": 21650
    },
    {
      "epoch": 1.695100954766004,
      "grad_norm": 8.609230995178223,
      "learning_rate": 2.180650729717006e-05,
      "loss": 0.7958,
      "step": 21660
    },
    {
      "epoch": 1.695883549851307,
      "grad_norm": 28.894607543945312,
      "learning_rate": 2.1793429931474603e-05,
      "loss": 0.8129,
      "step": 21670
    },
    {
      "epoch": 1.6966661449366098,
      "grad_norm": 3.1056528091430664,
      "learning_rate": 2.1780352565779148e-05,
      "loss": 0.5974,
      "step": 21680
    },
    {
      "epoch": 1.6974487400219127,
      "grad_norm": 1.476617693901062,
      "learning_rate": 2.1767275200083696e-05,
      "loss": 0.59,
      "step": 21690
    },
    {
      "epoch": 1.6982313351072156,
      "grad_norm": 41.16594314575195,
      "learning_rate": 2.1754197834388244e-05,
      "loss": 0.7921,
      "step": 21700
    },
    {
      "epoch": 1.6990139301925185,
      "grad_norm": 12.60340690612793,
      "learning_rate": 2.174112046869279e-05,
      "loss": 0.9554,
      "step": 21710
    },
    {
      "epoch": 1.6997965252778213,
      "grad_norm": 2.005171775817871,
      "learning_rate": 2.1728043102997333e-05,
      "loss": 0.5712,
      "step": 21720
    },
    {
      "epoch": 1.7005791203631242,
      "grad_norm": 7.106380939483643,
      "learning_rate": 2.1714965737301878e-05,
      "loss": 0.7243,
      "step": 21730
    },
    {
      "epoch": 1.701361715448427,
      "grad_norm": 9.540108680725098,
      "learning_rate": 2.1701888371606426e-05,
      "loss": 0.747,
      "step": 21740
    },
    {
      "epoch": 1.70214431053373,
      "grad_norm": 5.150615692138672,
      "learning_rate": 2.168881100591097e-05,
      "loss": 0.7672,
      "step": 21750
    },
    {
      "epoch": 1.7029269056190328,
      "grad_norm": 14.827492713928223,
      "learning_rate": 2.1675733640215515e-05,
      "loss": 0.6155,
      "step": 21760
    },
    {
      "epoch": 1.7037095007043357,
      "grad_norm": 7.372093200683594,
      "learning_rate": 2.1662656274520063e-05,
      "loss": 1.0259,
      "step": 21770
    },
    {
      "epoch": 1.7044920957896386,
      "grad_norm": 5.891328811645508,
      "learning_rate": 2.1649578908824607e-05,
      "loss": 0.703,
      "step": 21780
    },
    {
      "epoch": 1.7052746908749414,
      "grad_norm": 2.145221710205078,
      "learning_rate": 2.1636501543129152e-05,
      "loss": 0.8683,
      "step": 21790
    },
    {
      "epoch": 1.7060572859602443,
      "grad_norm": 13.106571197509766,
      "learning_rate": 2.16234241774337e-05,
      "loss": 0.6941,
      "step": 21800
    },
    {
      "epoch": 1.7068398810455472,
      "grad_norm": 6.904123306274414,
      "learning_rate": 2.1610346811738244e-05,
      "loss": 0.8701,
      "step": 21810
    },
    {
      "epoch": 1.70762247613085,
      "grad_norm": 8.95477294921875,
      "learning_rate": 2.159726944604279e-05,
      "loss": 0.7665,
      "step": 21820
    },
    {
      "epoch": 1.708405071216153,
      "grad_norm": 1.1992980241775513,
      "learning_rate": 2.1584192080347334e-05,
      "loss": 0.9885,
      "step": 21830
    },
    {
      "epoch": 1.7091876663014556,
      "grad_norm": 7.602221965789795,
      "learning_rate": 2.157111471465188e-05,
      "loss": 0.8594,
      "step": 21840
    },
    {
      "epoch": 1.7099702613867584,
      "grad_norm": 2.3350656032562256,
      "learning_rate": 2.155803734895643e-05,
      "loss": 0.6226,
      "step": 21850
    },
    {
      "epoch": 1.7107528564720613,
      "grad_norm": 3.7207226753234863,
      "learning_rate": 2.1544959983260974e-05,
      "loss": 0.6695,
      "step": 21860
    },
    {
      "epoch": 1.7115354515573642,
      "grad_norm": 8.243067741394043,
      "learning_rate": 2.153188261756552e-05,
      "loss": 0.7286,
      "step": 21870
    },
    {
      "epoch": 1.712318046642667,
      "grad_norm": 5.5627760887146,
      "learning_rate": 2.1518805251870063e-05,
      "loss": 0.6949,
      "step": 21880
    },
    {
      "epoch": 1.71310064172797,
      "grad_norm": 12.262612342834473,
      "learning_rate": 2.1505727886174608e-05,
      "loss": 1.052,
      "step": 21890
    },
    {
      "epoch": 1.7138832368132728,
      "grad_norm": 14.223176002502441,
      "learning_rate": 2.1492650520479156e-05,
      "loss": 0.6519,
      "step": 21900
    },
    {
      "epoch": 1.7146658318985757,
      "grad_norm": 5.55275297164917,
      "learning_rate": 2.1479573154783704e-05,
      "loss": 1.0557,
      "step": 21910
    },
    {
      "epoch": 1.7154484269838786,
      "grad_norm": 4.93870210647583,
      "learning_rate": 2.1466495789088248e-05,
      "loss": 0.3837,
      "step": 21920
    },
    {
      "epoch": 1.7162310220691814,
      "grad_norm": 1.7418746948242188,
      "learning_rate": 2.1453418423392793e-05,
      "loss": 0.8715,
      "step": 21930
    },
    {
      "epoch": 1.7170136171544843,
      "grad_norm": 4.061482906341553,
      "learning_rate": 2.1440341057697337e-05,
      "loss": 0.5046,
      "step": 21940
    },
    {
      "epoch": 1.7177962122397872,
      "grad_norm": 7.342411518096924,
      "learning_rate": 2.1427263692001885e-05,
      "loss": 0.5328,
      "step": 21950
    },
    {
      "epoch": 1.71857880732509,
      "grad_norm": 10.922207832336426,
      "learning_rate": 2.141418632630643e-05,
      "loss": 0.902,
      "step": 21960
    },
    {
      "epoch": 1.7193614024103927,
      "grad_norm": 11.858112335205078,
      "learning_rate": 2.1401108960610974e-05,
      "loss": 1.234,
      "step": 21970
    },
    {
      "epoch": 1.7201439974956956,
      "grad_norm": 6.827694892883301,
      "learning_rate": 2.1388031594915522e-05,
      "loss": 0.8009,
      "step": 21980
    },
    {
      "epoch": 1.7209265925809984,
      "grad_norm": 1.0594213008880615,
      "learning_rate": 2.1374954229220067e-05,
      "loss": 0.7353,
      "step": 21990
    },
    {
      "epoch": 1.7217091876663013,
      "grad_norm": 6.630974292755127,
      "learning_rate": 2.136187686352461e-05,
      "loss": 1.0401,
      "step": 22000
    },
    {
      "epoch": 1.7224917827516042,
      "grad_norm": 10.25693130493164,
      "learning_rate": 2.134879949782916e-05,
      "loss": 0.7686,
      "step": 22010
    },
    {
      "epoch": 1.723274377836907,
      "grad_norm": 11.25400447845459,
      "learning_rate": 2.1335722132133704e-05,
      "loss": 0.736,
      "step": 22020
    },
    {
      "epoch": 1.72405697292221,
      "grad_norm": 11.077977180480957,
      "learning_rate": 2.132264476643825e-05,
      "loss": 0.8373,
      "step": 22030
    },
    {
      "epoch": 1.7248395680075128,
      "grad_norm": 7.1665496826171875,
      "learning_rate": 2.1309567400742793e-05,
      "loss": 0.8962,
      "step": 22040
    },
    {
      "epoch": 1.7256221630928157,
      "grad_norm": 9.739192008972168,
      "learning_rate": 2.129649003504734e-05,
      "loss": 0.8056,
      "step": 22050
    },
    {
      "epoch": 1.7264047581781186,
      "grad_norm": 2.762864828109741,
      "learning_rate": 2.128341266935189e-05,
      "loss": 0.8007,
      "step": 22060
    },
    {
      "epoch": 1.7271873532634214,
      "grad_norm": 5.947878837585449,
      "learning_rate": 2.1270335303656434e-05,
      "loss": 0.7348,
      "step": 22070
    },
    {
      "epoch": 1.7279699483487243,
      "grad_norm": 17.1767520904541,
      "learning_rate": 2.125725793796098e-05,
      "loss": 0.8687,
      "step": 22080
    },
    {
      "epoch": 1.7287525434340272,
      "grad_norm": 1.5349773168563843,
      "learning_rate": 2.1244180572265523e-05,
      "loss": 0.3472,
      "step": 22090
    },
    {
      "epoch": 1.72953513851933,
      "grad_norm": 15.688610076904297,
      "learning_rate": 2.1231103206570067e-05,
      "loss": 0.8704,
      "step": 22100
    },
    {
      "epoch": 1.730317733604633,
      "grad_norm": 6.6746015548706055,
      "learning_rate": 2.1218025840874615e-05,
      "loss": 0.8383,
      "step": 22110
    },
    {
      "epoch": 1.7311003286899358,
      "grad_norm": 8.48005485534668,
      "learning_rate": 2.1204948475179163e-05,
      "loss": 0.9146,
      "step": 22120
    },
    {
      "epoch": 1.7318829237752387,
      "grad_norm": 0.9756578207015991,
      "learning_rate": 2.1191871109483708e-05,
      "loss": 0.7512,
      "step": 22130
    },
    {
      "epoch": 1.7326655188605415,
      "grad_norm": 10.128151893615723,
      "learning_rate": 2.1178793743788253e-05,
      "loss": 1.1539,
      "step": 22140
    },
    {
      "epoch": 1.7334481139458444,
      "grad_norm": 6.0727949142456055,
      "learning_rate": 2.1165716378092797e-05,
      "loss": 0.7105,
      "step": 22150
    },
    {
      "epoch": 1.7342307090311473,
      "grad_norm": 4.416561126708984,
      "learning_rate": 2.1152639012397345e-05,
      "loss": 0.7592,
      "step": 22160
    },
    {
      "epoch": 1.7350133041164502,
      "grad_norm": 3.740788221359253,
      "learning_rate": 2.113956164670189e-05,
      "loss": 0.6551,
      "step": 22170
    },
    {
      "epoch": 1.735795899201753,
      "grad_norm": 6.45063591003418,
      "learning_rate": 2.1126484281006434e-05,
      "loss": 0.6019,
      "step": 22180
    },
    {
      "epoch": 1.736578494287056,
      "grad_norm": 5.176023483276367,
      "learning_rate": 2.1113406915310982e-05,
      "loss": 0.7189,
      "step": 22190
    },
    {
      "epoch": 1.7373610893723588,
      "grad_norm": 0.6704067587852478,
      "learning_rate": 2.1100329549615527e-05,
      "loss": 0.66,
      "step": 22200
    },
    {
      "epoch": 1.7381436844576617,
      "grad_norm": 1.2391045093536377,
      "learning_rate": 2.108725218392007e-05,
      "loss": 0.4853,
      "step": 22210
    },
    {
      "epoch": 1.7389262795429645,
      "grad_norm": 7.163005828857422,
      "learning_rate": 2.107417481822462e-05,
      "loss": 0.8966,
      "step": 22220
    },
    {
      "epoch": 1.7397088746282674,
      "grad_norm": 6.852952480316162,
      "learning_rate": 2.1061097452529164e-05,
      "loss": 0.8918,
      "step": 22230
    },
    {
      "epoch": 1.7404914697135703,
      "grad_norm": 13.181251525878906,
      "learning_rate": 2.104802008683371e-05,
      "loss": 0.7639,
      "step": 22240
    },
    {
      "epoch": 1.7412740647988731,
      "grad_norm": 6.521900653839111,
      "learning_rate": 2.1034942721138253e-05,
      "loss": 0.8176,
      "step": 22250
    },
    {
      "epoch": 1.742056659884176,
      "grad_norm": 13.775201797485352,
      "learning_rate": 2.10218653554428e-05,
      "loss": 0.8754,
      "step": 22260
    },
    {
      "epoch": 1.742839254969479,
      "grad_norm": 10.21781063079834,
      "learning_rate": 2.1008787989747346e-05,
      "loss": 0.8419,
      "step": 22270
    },
    {
      "epoch": 1.7436218500547818,
      "grad_norm": 4.173689365386963,
      "learning_rate": 2.0995710624051893e-05,
      "loss": 0.6336,
      "step": 22280
    },
    {
      "epoch": 1.7444044451400846,
      "grad_norm": 10.051905632019043,
      "learning_rate": 2.0982633258356438e-05,
      "loss": 0.5897,
      "step": 22290
    },
    {
      "epoch": 1.7451870402253875,
      "grad_norm": 3.155581474304199,
      "learning_rate": 2.0969555892660983e-05,
      "loss": 0.8134,
      "step": 22300
    },
    {
      "epoch": 1.7459696353106904,
      "grad_norm": 9.249658584594727,
      "learning_rate": 2.0956478526965527e-05,
      "loss": 0.5418,
      "step": 22310
    },
    {
      "epoch": 1.7467522303959933,
      "grad_norm": 5.119617462158203,
      "learning_rate": 2.0943401161270075e-05,
      "loss": 0.7631,
      "step": 22320
    },
    {
      "epoch": 1.7475348254812961,
      "grad_norm": 7.612769603729248,
      "learning_rate": 2.0930323795574623e-05,
      "loss": 1.0039,
      "step": 22330
    },
    {
      "epoch": 1.748317420566599,
      "grad_norm": 7.742544651031494,
      "learning_rate": 2.0917246429879168e-05,
      "loss": 0.9407,
      "step": 22340
    },
    {
      "epoch": 1.7491000156519017,
      "grad_norm": 5.696972846984863,
      "learning_rate": 2.0904169064183712e-05,
      "loss": 0.8494,
      "step": 22350
    },
    {
      "epoch": 1.7498826107372045,
      "grad_norm": 5.840358257293701,
      "learning_rate": 2.0891091698488257e-05,
      "loss": 0.6202,
      "step": 22360
    },
    {
      "epoch": 1.7506652058225074,
      "grad_norm": 2.404650926589966,
      "learning_rate": 2.08780143327928e-05,
      "loss": 0.9129,
      "step": 22370
    },
    {
      "epoch": 1.7514478009078103,
      "grad_norm": 2.471250295639038,
      "learning_rate": 2.086493696709735e-05,
      "loss": 0.8733,
      "step": 22380
    },
    {
      "epoch": 1.7522303959931131,
      "grad_norm": 5.44634485244751,
      "learning_rate": 2.0851859601401894e-05,
      "loss": 0.6227,
      "step": 22390
    },
    {
      "epoch": 1.753012991078416,
      "grad_norm": 3.1709930896759033,
      "learning_rate": 2.0838782235706442e-05,
      "loss": 0.8382,
      "step": 22400
    },
    {
      "epoch": 1.753795586163719,
      "grad_norm": 2.7234952449798584,
      "learning_rate": 2.0825704870010986e-05,
      "loss": 0.5639,
      "step": 22410
    },
    {
      "epoch": 1.7545781812490218,
      "grad_norm": 4.502098560333252,
      "learning_rate": 2.081262750431553e-05,
      "loss": 0.4979,
      "step": 22420
    },
    {
      "epoch": 1.7553607763343246,
      "grad_norm": 8.832462310791016,
      "learning_rate": 2.079955013862008e-05,
      "loss": 0.8373,
      "step": 22430
    },
    {
      "epoch": 1.7561433714196275,
      "grad_norm": 3.511338233947754,
      "learning_rate": 2.0786472772924624e-05,
      "loss": 0.9303,
      "step": 22440
    },
    {
      "epoch": 1.7569259665049304,
      "grad_norm": 7.021348476409912,
      "learning_rate": 2.0773395407229168e-05,
      "loss": 0.7389,
      "step": 22450
    },
    {
      "epoch": 1.7577085615902333,
      "grad_norm": 9.02792739868164,
      "learning_rate": 2.0760318041533713e-05,
      "loss": 0.9696,
      "step": 22460
    },
    {
      "epoch": 1.7584911566755361,
      "grad_norm": 3.937958240509033,
      "learning_rate": 2.074724067583826e-05,
      "loss": 0.9131,
      "step": 22470
    },
    {
      "epoch": 1.7592737517608388,
      "grad_norm": 20.325332641601562,
      "learning_rate": 2.0734163310142805e-05,
      "loss": 0.9458,
      "step": 22480
    },
    {
      "epoch": 1.7600563468461417,
      "grad_norm": 0.9961243867874146,
      "learning_rate": 2.0721085944447353e-05,
      "loss": 0.5131,
      "step": 22490
    },
    {
      "epoch": 1.7608389419314445,
      "grad_norm": 13.234907150268555,
      "learning_rate": 2.0708008578751898e-05,
      "loss": 0.9622,
      "step": 22500
    },
    {
      "epoch": 1.7616215370167474,
      "grad_norm": 1.7734347581863403,
      "learning_rate": 2.0694931213056442e-05,
      "loss": 0.9853,
      "step": 22510
    },
    {
      "epoch": 1.7624041321020503,
      "grad_norm": 8.941816329956055,
      "learning_rate": 2.0681853847360987e-05,
      "loss": 0.7143,
      "step": 22520
    },
    {
      "epoch": 1.7631867271873531,
      "grad_norm": 9.519405364990234,
      "learning_rate": 2.0668776481665535e-05,
      "loss": 0.6136,
      "step": 22530
    },
    {
      "epoch": 1.763969322272656,
      "grad_norm": 8.802802085876465,
      "learning_rate": 2.0655699115970083e-05,
      "loss": 0.979,
      "step": 22540
    },
    {
      "epoch": 1.7647519173579589,
      "grad_norm": 2.9852428436279297,
      "learning_rate": 2.0642621750274627e-05,
      "loss": 0.8648,
      "step": 22550
    },
    {
      "epoch": 1.7655345124432618,
      "grad_norm": 8.315165519714355,
      "learning_rate": 2.0629544384579172e-05,
      "loss": 1.2894,
      "step": 22560
    },
    {
      "epoch": 1.7663171075285646,
      "grad_norm": 4.418003082275391,
      "learning_rate": 2.0616467018883717e-05,
      "loss": 0.8916,
      "step": 22570
    },
    {
      "epoch": 1.7670997026138675,
      "grad_norm": 3.493744134902954,
      "learning_rate": 2.060338965318826e-05,
      "loss": 0.9049,
      "step": 22580
    },
    {
      "epoch": 1.7678822976991704,
      "grad_norm": 7.965389728546143,
      "learning_rate": 2.059031228749281e-05,
      "loss": 0.922,
      "step": 22590
    },
    {
      "epoch": 1.7686648927844733,
      "grad_norm": 3.5176966190338135,
      "learning_rate": 2.0577234921797354e-05,
      "loss": 0.513,
      "step": 22600
    },
    {
      "epoch": 1.7694474878697761,
      "grad_norm": 2.1263675689697266,
      "learning_rate": 2.05641575561019e-05,
      "loss": 0.6975,
      "step": 22610
    },
    {
      "epoch": 1.770230082955079,
      "grad_norm": 9.632349014282227,
      "learning_rate": 2.0551080190406446e-05,
      "loss": 0.991,
      "step": 22620
    },
    {
      "epoch": 1.7710126780403819,
      "grad_norm": 3.2448787689208984,
      "learning_rate": 2.053800282471099e-05,
      "loss": 0.869,
      "step": 22630
    },
    {
      "epoch": 1.7717952731256847,
      "grad_norm": 7.239644527435303,
      "learning_rate": 2.052492545901554e-05,
      "loss": 0.6533,
      "step": 22640
    },
    {
      "epoch": 1.7725778682109876,
      "grad_norm": 5.273192882537842,
      "learning_rate": 2.0511848093320083e-05,
      "loss": 0.5623,
      "step": 22650
    },
    {
      "epoch": 1.7733604632962905,
      "grad_norm": 8.263648986816406,
      "learning_rate": 2.0498770727624628e-05,
      "loss": 0.4813,
      "step": 22660
    },
    {
      "epoch": 1.7741430583815934,
      "grad_norm": 3.426119089126587,
      "learning_rate": 2.0485693361929172e-05,
      "loss": 0.8178,
      "step": 22670
    },
    {
      "epoch": 1.7749256534668962,
      "grad_norm": 13.191840171813965,
      "learning_rate": 2.0472615996233717e-05,
      "loss": 0.891,
      "step": 22680
    },
    {
      "epoch": 1.775708248552199,
      "grad_norm": 5.519374847412109,
      "learning_rate": 2.0459538630538265e-05,
      "loss": 0.6269,
      "step": 22690
    },
    {
      "epoch": 1.776490843637502,
      "grad_norm": 2.17944598197937,
      "learning_rate": 2.0446461264842813e-05,
      "loss": 0.4937,
      "step": 22700
    },
    {
      "epoch": 1.7772734387228049,
      "grad_norm": 5.906281471252441,
      "learning_rate": 2.0433383899147358e-05,
      "loss": 0.6746,
      "step": 22710
    },
    {
      "epoch": 1.7780560338081077,
      "grad_norm": 2.056196928024292,
      "learning_rate": 2.0420306533451902e-05,
      "loss": 0.4066,
      "step": 22720
    },
    {
      "epoch": 1.7788386288934106,
      "grad_norm": 11.540655136108398,
      "learning_rate": 2.0407229167756447e-05,
      "loss": 0.5484,
      "step": 22730
    },
    {
      "epoch": 1.7796212239787135,
      "grad_norm": 0.599817156791687,
      "learning_rate": 2.039415180206099e-05,
      "loss": 0.7196,
      "step": 22740
    },
    {
      "epoch": 1.7804038190640163,
      "grad_norm": 21.80187225341797,
      "learning_rate": 2.038107443636554e-05,
      "loss": 0.4772,
      "step": 22750
    },
    {
      "epoch": 1.7811864141493192,
      "grad_norm": 1.4702868461608887,
      "learning_rate": 2.0367997070670087e-05,
      "loss": 0.7,
      "step": 22760
    },
    {
      "epoch": 1.781969009234622,
      "grad_norm": 0.6858302354812622,
      "learning_rate": 2.0354919704974632e-05,
      "loss": 1.3605,
      "step": 22770
    },
    {
      "epoch": 1.782751604319925,
      "grad_norm": 12.074968338012695,
      "learning_rate": 2.0341842339279176e-05,
      "loss": 0.9372,
      "step": 22780
    },
    {
      "epoch": 1.7835341994052278,
      "grad_norm": 4.758157253265381,
      "learning_rate": 2.032876497358372e-05,
      "loss": 0.4611,
      "step": 22790
    },
    {
      "epoch": 1.7843167944905307,
      "grad_norm": 7.259366989135742,
      "learning_rate": 2.031568760788827e-05,
      "loss": 1.2069,
      "step": 22800
    },
    {
      "epoch": 1.7850993895758336,
      "grad_norm": 2.8948140144348145,
      "learning_rate": 2.0302610242192813e-05,
      "loss": 1.0229,
      "step": 22810
    },
    {
      "epoch": 1.7858819846611365,
      "grad_norm": 20.217927932739258,
      "learning_rate": 2.0289532876497358e-05,
      "loss": 0.7773,
      "step": 22820
    },
    {
      "epoch": 1.7866645797464393,
      "grad_norm": 10.481877326965332,
      "learning_rate": 2.0276455510801906e-05,
      "loss": 0.5566,
      "step": 22830
    },
    {
      "epoch": 1.7874471748317422,
      "grad_norm": 33.533241271972656,
      "learning_rate": 2.026337814510645e-05,
      "loss": 0.6063,
      "step": 22840
    },
    {
      "epoch": 1.788229769917045,
      "grad_norm": 9.199257850646973,
      "learning_rate": 2.0250300779410995e-05,
      "loss": 1.0104,
      "step": 22850
    },
    {
      "epoch": 1.789012365002348,
      "grad_norm": 6.992598056793213,
      "learning_rate": 2.0237223413715543e-05,
      "loss": 0.5897,
      "step": 22860
    },
    {
      "epoch": 1.7897949600876506,
      "grad_norm": 10.849595069885254,
      "learning_rate": 2.0224146048020088e-05,
      "loss": 0.8911,
      "step": 22870
    },
    {
      "epoch": 1.7905775551729535,
      "grad_norm": 1.218825340270996,
      "learning_rate": 2.0211068682324632e-05,
      "loss": 1.0199,
      "step": 22880
    },
    {
      "epoch": 1.7913601502582563,
      "grad_norm": 13.134035110473633,
      "learning_rate": 2.0197991316629177e-05,
      "loss": 0.7937,
      "step": 22890
    },
    {
      "epoch": 1.7921427453435592,
      "grad_norm": 0.517034649848938,
      "learning_rate": 2.0184913950933725e-05,
      "loss": 0.7396,
      "step": 22900
    },
    {
      "epoch": 1.792925340428862,
      "grad_norm": 18.224374771118164,
      "learning_rate": 2.0171836585238273e-05,
      "loss": 0.9758,
      "step": 22910
    },
    {
      "epoch": 1.793707935514165,
      "grad_norm": 8.817658424377441,
      "learning_rate": 2.0158759219542817e-05,
      "loss": 0.5734,
      "step": 22920
    },
    {
      "epoch": 1.7944905305994678,
      "grad_norm": 15.969501495361328,
      "learning_rate": 2.0145681853847362e-05,
      "loss": 0.9488,
      "step": 22930
    },
    {
      "epoch": 1.7952731256847707,
      "grad_norm": 5.523311138153076,
      "learning_rate": 2.0132604488151906e-05,
      "loss": 0.5154,
      "step": 22940
    },
    {
      "epoch": 1.7960557207700736,
      "grad_norm": 7.043122291564941,
      "learning_rate": 2.011952712245645e-05,
      "loss": 1.0183,
      "step": 22950
    },
    {
      "epoch": 1.7968383158553765,
      "grad_norm": 8.249909400939941,
      "learning_rate": 2.0106449756761e-05,
      "loss": 0.6311,
      "step": 22960
    },
    {
      "epoch": 1.7976209109406793,
      "grad_norm": 0.7932335734367371,
      "learning_rate": 2.0093372391065547e-05,
      "loss": 1.1109,
      "step": 22970
    },
    {
      "epoch": 1.7984035060259822,
      "grad_norm": 7.757596969604492,
      "learning_rate": 2.008029502537009e-05,
      "loss": 0.6605,
      "step": 22980
    },
    {
      "epoch": 1.799186101111285,
      "grad_norm": 14.178553581237793,
      "learning_rate": 2.0067217659674636e-05,
      "loss": 0.5977,
      "step": 22990
    },
    {
      "epoch": 1.7999686961965877,
      "grad_norm": 9.697897911071777,
      "learning_rate": 2.005414029397918e-05,
      "loss": 0.7615,
      "step": 23000
    },
    {
      "epoch": 1.8007512912818906,
      "grad_norm": 5.798877239227295,
      "learning_rate": 2.004106292828373e-05,
      "loss": 0.7249,
      "step": 23010
    },
    {
      "epoch": 1.8015338863671935,
      "grad_norm": 11.297525405883789,
      "learning_rate": 2.0027985562588273e-05,
      "loss": 1.0591,
      "step": 23020
    },
    {
      "epoch": 1.8023164814524963,
      "grad_norm": 10.25127124786377,
      "learning_rate": 2.0014908196892818e-05,
      "loss": 0.9543,
      "step": 23030
    },
    {
      "epoch": 1.8030990765377992,
      "grad_norm": 2.8508098125457764,
      "learning_rate": 2.0001830831197366e-05,
      "loss": 0.7866,
      "step": 23040
    },
    {
      "epoch": 1.803881671623102,
      "grad_norm": 20.964500427246094,
      "learning_rate": 1.998875346550191e-05,
      "loss": 0.8086,
      "step": 23050
    },
    {
      "epoch": 1.804664266708405,
      "grad_norm": 26.15506362915039,
      "learning_rate": 1.9975676099806455e-05,
      "loss": 1.0519,
      "step": 23060
    },
    {
      "epoch": 1.8054468617937078,
      "grad_norm": 6.771284103393555,
      "learning_rate": 1.9962598734111003e-05,
      "loss": 0.6562,
      "step": 23070
    },
    {
      "epoch": 1.8062294568790107,
      "grad_norm": 10.643194198608398,
      "learning_rate": 1.9949521368415547e-05,
      "loss": 0.9917,
      "step": 23080
    },
    {
      "epoch": 1.8070120519643136,
      "grad_norm": 12.835536003112793,
      "learning_rate": 1.9936444002720092e-05,
      "loss": 1.0294,
      "step": 23090
    },
    {
      "epoch": 1.8077946470496165,
      "grad_norm": 9.197125434875488,
      "learning_rate": 1.9923366637024636e-05,
      "loss": 0.5002,
      "step": 23100
    },
    {
      "epoch": 1.8085772421349193,
      "grad_norm": 13.356221199035645,
      "learning_rate": 1.9910289271329184e-05,
      "loss": 0.8639,
      "step": 23110
    },
    {
      "epoch": 1.8093598372202222,
      "grad_norm": 2.478785991668701,
      "learning_rate": 1.9897211905633732e-05,
      "loss": 0.4825,
      "step": 23120
    },
    {
      "epoch": 1.810142432305525,
      "grad_norm": 8.04997730255127,
      "learning_rate": 1.9884134539938277e-05,
      "loss": 0.5452,
      "step": 23130
    },
    {
      "epoch": 1.810925027390828,
      "grad_norm": 5.9365997314453125,
      "learning_rate": 1.987105717424282e-05,
      "loss": 0.8873,
      "step": 23140
    },
    {
      "epoch": 1.8117076224761308,
      "grad_norm": 5.273404598236084,
      "learning_rate": 1.9857979808547366e-05,
      "loss": 0.7582,
      "step": 23150
    },
    {
      "epoch": 1.8124902175614337,
      "grad_norm": 2.1371829509735107,
      "learning_rate": 1.984490244285191e-05,
      "loss": 0.8527,
      "step": 23160
    },
    {
      "epoch": 1.8132728126467366,
      "grad_norm": 16.187349319458008,
      "learning_rate": 1.983182507715646e-05,
      "loss": 0.9984,
      "step": 23170
    },
    {
      "epoch": 1.8140554077320394,
      "grad_norm": 14.993395805358887,
      "learning_rate": 1.9818747711461007e-05,
      "loss": 1.0492,
      "step": 23180
    },
    {
      "epoch": 1.8148380028173423,
      "grad_norm": 8.237442970275879,
      "learning_rate": 1.980567034576555e-05,
      "loss": 0.7708,
      "step": 23190
    },
    {
      "epoch": 1.8156205979026452,
      "grad_norm": 0.8713195323944092,
      "learning_rate": 1.9792592980070096e-05,
      "loss": 0.7957,
      "step": 23200
    },
    {
      "epoch": 1.816403192987948,
      "grad_norm": 4.170684814453125,
      "learning_rate": 1.977951561437464e-05,
      "loss": 0.5072,
      "step": 23210
    },
    {
      "epoch": 1.817185788073251,
      "grad_norm": 11.119500160217285,
      "learning_rate": 1.9766438248679185e-05,
      "loss": 1.2055,
      "step": 23220
    },
    {
      "epoch": 1.8179683831585538,
      "grad_norm": 8.848735809326172,
      "learning_rate": 1.9753360882983733e-05,
      "loss": 0.5742,
      "step": 23230
    },
    {
      "epoch": 1.8187509782438567,
      "grad_norm": 3.6243531703948975,
      "learning_rate": 1.9740283517288277e-05,
      "loss": 0.704,
      "step": 23240
    },
    {
      "epoch": 1.8195335733291595,
      "grad_norm": 1.365095615386963,
      "learning_rate": 1.9727206151592825e-05,
      "loss": 0.7033,
      "step": 23250
    },
    {
      "epoch": 1.8203161684144624,
      "grad_norm": 6.85057258605957,
      "learning_rate": 1.971412878589737e-05,
      "loss": 1.2085,
      "step": 23260
    },
    {
      "epoch": 1.8210987634997653,
      "grad_norm": 8.62144947052002,
      "learning_rate": 1.9701051420201915e-05,
      "loss": 0.6844,
      "step": 23270
    },
    {
      "epoch": 1.8218813585850682,
      "grad_norm": 5.643413543701172,
      "learning_rate": 1.9687974054506463e-05,
      "loss": 0.9197,
      "step": 23280
    },
    {
      "epoch": 1.822663953670371,
      "grad_norm": 7.657876491546631,
      "learning_rate": 1.9674896688811007e-05,
      "loss": 0.6824,
      "step": 23290
    },
    {
      "epoch": 1.823446548755674,
      "grad_norm": 8.764404296875,
      "learning_rate": 1.966181932311555e-05,
      "loss": 1.036,
      "step": 23300
    },
    {
      "epoch": 1.8242291438409768,
      "grad_norm": 5.023722171783447,
      "learning_rate": 1.9648741957420096e-05,
      "loss": 0.7255,
      "step": 23310
    },
    {
      "epoch": 1.8250117389262797,
      "grad_norm": 3.9335403442382812,
      "learning_rate": 1.9635664591724644e-05,
      "loss": 0.5459,
      "step": 23320
    },
    {
      "epoch": 1.8257943340115825,
      "grad_norm": 1.6663060188293457,
      "learning_rate": 1.962258722602919e-05,
      "loss": 0.6206,
      "step": 23330
    },
    {
      "epoch": 1.8265769290968854,
      "grad_norm": 6.2383317947387695,
      "learning_rate": 1.9609509860333737e-05,
      "loss": 0.5859,
      "step": 23340
    },
    {
      "epoch": 1.8273595241821883,
      "grad_norm": 1.4167931079864502,
      "learning_rate": 1.959643249463828e-05,
      "loss": 0.6244,
      "step": 23350
    },
    {
      "epoch": 1.8281421192674912,
      "grad_norm": 9.228361129760742,
      "learning_rate": 1.9583355128942826e-05,
      "loss": 0.8254,
      "step": 23360
    },
    {
      "epoch": 1.828924714352794,
      "grad_norm": 1.5492888689041138,
      "learning_rate": 1.957027776324737e-05,
      "loss": 0.4608,
      "step": 23370
    },
    {
      "epoch": 1.8297073094380967,
      "grad_norm": 1.7753559350967407,
      "learning_rate": 1.955720039755192e-05,
      "loss": 1.3263,
      "step": 23380
    },
    {
      "epoch": 1.8304899045233995,
      "grad_norm": 9.652050971984863,
      "learning_rate": 1.9544123031856466e-05,
      "loss": 0.98,
      "step": 23390
    },
    {
      "epoch": 1.8312724996087024,
      "grad_norm": 3.5658788681030273,
      "learning_rate": 1.953104566616101e-05,
      "loss": 0.6701,
      "step": 23400
    },
    {
      "epoch": 1.8320550946940053,
      "grad_norm": 8.075160026550293,
      "learning_rate": 1.9517968300465555e-05,
      "loss": 0.8423,
      "step": 23410
    },
    {
      "epoch": 1.8328376897793082,
      "grad_norm": 1.2914469242095947,
      "learning_rate": 1.95048909347701e-05,
      "loss": 1.2348,
      "step": 23420
    },
    {
      "epoch": 1.833620284864611,
      "grad_norm": 5.562222003936768,
      "learning_rate": 1.9491813569074645e-05,
      "loss": 0.598,
      "step": 23430
    },
    {
      "epoch": 1.834402879949914,
      "grad_norm": 6.995854377746582,
      "learning_rate": 1.9478736203379193e-05,
      "loss": 0.7245,
      "step": 23440
    },
    {
      "epoch": 1.8351854750352168,
      "grad_norm": 1.5724756717681885,
      "learning_rate": 1.9465658837683737e-05,
      "loss": 0.6577,
      "step": 23450
    },
    {
      "epoch": 1.8359680701205197,
      "grad_norm": 5.490779876708984,
      "learning_rate": 1.9452581471988285e-05,
      "loss": 0.6647,
      "step": 23460
    },
    {
      "epoch": 1.8367506652058225,
      "grad_norm": 8.320096969604492,
      "learning_rate": 1.943950410629283e-05,
      "loss": 0.7468,
      "step": 23470
    },
    {
      "epoch": 1.8375332602911254,
      "grad_norm": 5.044735431671143,
      "learning_rate": 1.9426426740597374e-05,
      "loss": 0.7322,
      "step": 23480
    },
    {
      "epoch": 1.8383158553764283,
      "grad_norm": 6.855508327484131,
      "learning_rate": 1.9413349374901922e-05,
      "loss": 0.964,
      "step": 23490
    },
    {
      "epoch": 1.8390984504617311,
      "grad_norm": 11.671350479125977,
      "learning_rate": 1.9400272009206467e-05,
      "loss": 1.0374,
      "step": 23500
    },
    {
      "epoch": 1.8398810455470338,
      "grad_norm": 4.395977973937988,
      "learning_rate": 1.938719464351101e-05,
      "loss": 0.6942,
      "step": 23510
    },
    {
      "epoch": 1.8406636406323367,
      "grad_norm": 8.509678840637207,
      "learning_rate": 1.9374117277815556e-05,
      "loss": 0.5662,
      "step": 23520
    },
    {
      "epoch": 1.8414462357176395,
      "grad_norm": 5.672106742858887,
      "learning_rate": 1.9361039912120104e-05,
      "loss": 0.6284,
      "step": 23530
    },
    {
      "epoch": 1.8422288308029424,
      "grad_norm": 4.924073219299316,
      "learning_rate": 1.934796254642465e-05,
      "loss": 0.6924,
      "step": 23540
    },
    {
      "epoch": 1.8430114258882453,
      "grad_norm": 17.050735473632812,
      "learning_rate": 1.9334885180729196e-05,
      "loss": 0.852,
      "step": 23550
    },
    {
      "epoch": 1.8437940209735482,
      "grad_norm": 10.444417953491211,
      "learning_rate": 1.932180781503374e-05,
      "loss": 1.1202,
      "step": 23560
    },
    {
      "epoch": 1.844576616058851,
      "grad_norm": 4.6727986335754395,
      "learning_rate": 1.9308730449338286e-05,
      "loss": 0.7787,
      "step": 23570
    },
    {
      "epoch": 1.845359211144154,
      "grad_norm": 7.480617523193359,
      "learning_rate": 1.929565308364283e-05,
      "loss": 0.8882,
      "step": 23580
    },
    {
      "epoch": 1.8461418062294568,
      "grad_norm": 3.0243403911590576,
      "learning_rate": 1.9282575717947378e-05,
      "loss": 0.6956,
      "step": 23590
    },
    {
      "epoch": 1.8469244013147597,
      "grad_norm": 6.935908317565918,
      "learning_rate": 1.9269498352251926e-05,
      "loss": 0.829,
      "step": 23600
    },
    {
      "epoch": 1.8477069964000625,
      "grad_norm": 4.402510166168213,
      "learning_rate": 1.925642098655647e-05,
      "loss": 0.6255,
      "step": 23610
    },
    {
      "epoch": 1.8484895914853654,
      "grad_norm": 2.0141005516052246,
      "learning_rate": 1.9243343620861015e-05,
      "loss": 0.6088,
      "step": 23620
    },
    {
      "epoch": 1.8492721865706683,
      "grad_norm": 4.69474458694458,
      "learning_rate": 1.923026625516556e-05,
      "loss": 0.8822,
      "step": 23630
    },
    {
      "epoch": 1.8500547816559711,
      "grad_norm": 0.8279262185096741,
      "learning_rate": 1.9217188889470104e-05,
      "loss": 0.5959,
      "step": 23640
    },
    {
      "epoch": 1.850837376741274,
      "grad_norm": 4.240744113922119,
      "learning_rate": 1.9204111523774652e-05,
      "loss": 0.7293,
      "step": 23650
    },
    {
      "epoch": 1.851619971826577,
      "grad_norm": 6.15984582901001,
      "learning_rate": 1.9191034158079197e-05,
      "loss": 0.7363,
      "step": 23660
    },
    {
      "epoch": 1.8524025669118798,
      "grad_norm": 4.27946662902832,
      "learning_rate": 1.9177956792383745e-05,
      "loss": 1.0655,
      "step": 23670
    },
    {
      "epoch": 1.8531851619971826,
      "grad_norm": 3.034358263015747,
      "learning_rate": 1.916487942668829e-05,
      "loss": 0.4217,
      "step": 23680
    },
    {
      "epoch": 1.8539677570824855,
      "grad_norm": 2.0383963584899902,
      "learning_rate": 1.9151802060992834e-05,
      "loss": 0.707,
      "step": 23690
    },
    {
      "epoch": 1.8547503521677884,
      "grad_norm": 1.2238657474517822,
      "learning_rate": 1.9138724695297382e-05,
      "loss": 0.7635,
      "step": 23700
    },
    {
      "epoch": 1.8555329472530913,
      "grad_norm": 6.901283264160156,
      "learning_rate": 1.9125647329601927e-05,
      "loss": 0.7218,
      "step": 23710
    },
    {
      "epoch": 1.8563155423383941,
      "grad_norm": 8.000720024108887,
      "learning_rate": 1.911256996390647e-05,
      "loss": 1.1043,
      "step": 23720
    },
    {
      "epoch": 1.857098137423697,
      "grad_norm": 29.50677490234375,
      "learning_rate": 1.9099492598211016e-05,
      "loss": 0.5644,
      "step": 23730
    },
    {
      "epoch": 1.8578807325089999,
      "grad_norm": 11.273759841918945,
      "learning_rate": 1.9086415232515564e-05,
      "loss": 0.6803,
      "step": 23740
    },
    {
      "epoch": 1.8586633275943027,
      "grad_norm": 8.937383651733398,
      "learning_rate": 1.9073337866820108e-05,
      "loss": 0.7458,
      "step": 23750
    },
    {
      "epoch": 1.8594459226796056,
      "grad_norm": 6.717219352722168,
      "learning_rate": 1.9060260501124656e-05,
      "loss": 0.6355,
      "step": 23760
    },
    {
      "epoch": 1.8602285177649085,
      "grad_norm": 14.883992195129395,
      "learning_rate": 1.90471831354292e-05,
      "loss": 0.9604,
      "step": 23770
    },
    {
      "epoch": 1.8610111128502114,
      "grad_norm": 0.9538350105285645,
      "learning_rate": 1.9034105769733745e-05,
      "loss": 0.8877,
      "step": 23780
    },
    {
      "epoch": 1.8617937079355142,
      "grad_norm": 1.3885371685028076,
      "learning_rate": 1.902102840403829e-05,
      "loss": 0.9893,
      "step": 23790
    },
    {
      "epoch": 1.8625763030208171,
      "grad_norm": 3.73441481590271,
      "learning_rate": 1.9007951038342834e-05,
      "loss": 0.4048,
      "step": 23800
    },
    {
      "epoch": 1.86335889810612,
      "grad_norm": 8.21196460723877,
      "learning_rate": 1.8994873672647386e-05,
      "loss": 0.8256,
      "step": 23810
    },
    {
      "epoch": 1.8641414931914229,
      "grad_norm": 28.797164916992188,
      "learning_rate": 1.898179630695193e-05,
      "loss": 1.0863,
      "step": 23820
    },
    {
      "epoch": 1.8649240882767257,
      "grad_norm": 0.4742733836174011,
      "learning_rate": 1.8968718941256475e-05,
      "loss": 0.4827,
      "step": 23830
    },
    {
      "epoch": 1.8657066833620286,
      "grad_norm": 11.527212142944336,
      "learning_rate": 1.895564157556102e-05,
      "loss": 0.6297,
      "step": 23840
    },
    {
      "epoch": 1.8664892784473315,
      "grad_norm": 38.41826248168945,
      "learning_rate": 1.8942564209865564e-05,
      "loss": 0.4683,
      "step": 23850
    },
    {
      "epoch": 1.8672718735326344,
      "grad_norm": 10.132311820983887,
      "learning_rate": 1.8929486844170112e-05,
      "loss": 1.0844,
      "step": 23860
    },
    {
      "epoch": 1.8680544686179372,
      "grad_norm": 23.62603187561035,
      "learning_rate": 1.8916409478474657e-05,
      "loss": 0.7415,
      "step": 23870
    },
    {
      "epoch": 1.86883706370324,
      "grad_norm": 14.863296508789062,
      "learning_rate": 1.8903332112779205e-05,
      "loss": 1.1762,
      "step": 23880
    },
    {
      "epoch": 1.869619658788543,
      "grad_norm": 13.659287452697754,
      "learning_rate": 1.889025474708375e-05,
      "loss": 0.7721,
      "step": 23890
    },
    {
      "epoch": 1.8704022538738456,
      "grad_norm": 7.976527214050293,
      "learning_rate": 1.8877177381388294e-05,
      "loss": 0.6254,
      "step": 23900
    },
    {
      "epoch": 1.8711848489591485,
      "grad_norm": 11.320398330688477,
      "learning_rate": 1.886410001569284e-05,
      "loss": 0.7323,
      "step": 23910
    },
    {
      "epoch": 1.8719674440444514,
      "grad_norm": 0.9704088568687439,
      "learning_rate": 1.8851022649997386e-05,
      "loss": 0.7874,
      "step": 23920
    },
    {
      "epoch": 1.8727500391297542,
      "grad_norm": 8.839340209960938,
      "learning_rate": 1.883794528430193e-05,
      "loss": 0.9522,
      "step": 23930
    },
    {
      "epoch": 1.8735326342150571,
      "grad_norm": 14.233475685119629,
      "learning_rate": 1.8824867918606475e-05,
      "loss": 0.6402,
      "step": 23940
    },
    {
      "epoch": 1.87431522930036,
      "grad_norm": 1.035098910331726,
      "learning_rate": 1.8811790552911023e-05,
      "loss": 0.7878,
      "step": 23950
    },
    {
      "epoch": 1.8750978243856629,
      "grad_norm": 1.9485267400741577,
      "learning_rate": 1.8798713187215568e-05,
      "loss": 0.6144,
      "step": 23960
    },
    {
      "epoch": 1.8758804194709657,
      "grad_norm": 16.35893440246582,
      "learning_rate": 1.8785635821520116e-05,
      "loss": 0.761,
      "step": 23970
    },
    {
      "epoch": 1.8766630145562686,
      "grad_norm": 9.892289161682129,
      "learning_rate": 1.877255845582466e-05,
      "loss": 0.8506,
      "step": 23980
    },
    {
      "epoch": 1.8774456096415715,
      "grad_norm": 7.338379383087158,
      "learning_rate": 1.8759481090129205e-05,
      "loss": 1.0889,
      "step": 23990
    },
    {
      "epoch": 1.8782282047268744,
      "grad_norm": 4.1190876960754395,
      "learning_rate": 1.874640372443375e-05,
      "loss": 0.5791,
      "step": 24000
    },
    {
      "epoch": 1.8790107998121772,
      "grad_norm": 6.18887186050415,
      "learning_rate": 1.8733326358738294e-05,
      "loss": 0.8602,
      "step": 24010
    },
    {
      "epoch": 1.87979339489748,
      "grad_norm": 0.7922699451446533,
      "learning_rate": 1.8720248993042842e-05,
      "loss": 0.4351,
      "step": 24020
    },
    {
      "epoch": 1.8805759899827827,
      "grad_norm": 0.8942759037017822,
      "learning_rate": 1.870717162734739e-05,
      "loss": 0.6408,
      "step": 24030
    },
    {
      "epoch": 1.8813585850680856,
      "grad_norm": 5.793100357055664,
      "learning_rate": 1.8694094261651935e-05,
      "loss": 0.8884,
      "step": 24040
    },
    {
      "epoch": 1.8821411801533885,
      "grad_norm": 20.133939743041992,
      "learning_rate": 1.868101689595648e-05,
      "loss": 0.5982,
      "step": 24050
    },
    {
      "epoch": 1.8829237752386914,
      "grad_norm": 5.050759315490723,
      "learning_rate": 1.8667939530261024e-05,
      "loss": 0.6287,
      "step": 24060
    },
    {
      "epoch": 1.8837063703239942,
      "grad_norm": 1.4979051351547241,
      "learning_rate": 1.8654862164565572e-05,
      "loss": 0.6544,
      "step": 24070
    },
    {
      "epoch": 1.8844889654092971,
      "grad_norm": 6.128724098205566,
      "learning_rate": 1.8641784798870116e-05,
      "loss": 0.9013,
      "step": 24080
    },
    {
      "epoch": 1.8852715604946,
      "grad_norm": 6.884725093841553,
      "learning_rate": 1.862870743317466e-05,
      "loss": 0.8733,
      "step": 24090
    },
    {
      "epoch": 1.8860541555799029,
      "grad_norm": 6.197681903839111,
      "learning_rate": 1.861563006747921e-05,
      "loss": 0.8975,
      "step": 24100
    },
    {
      "epoch": 1.8868367506652057,
      "grad_norm": 19.020109176635742,
      "learning_rate": 1.8602552701783753e-05,
      "loss": 0.8152,
      "step": 24110
    },
    {
      "epoch": 1.8876193457505086,
      "grad_norm": 0.9438962936401367,
      "learning_rate": 1.8589475336088298e-05,
      "loss": 0.7818,
      "step": 24120
    },
    {
      "epoch": 1.8884019408358115,
      "grad_norm": 4.233465194702148,
      "learning_rate": 1.8576397970392846e-05,
      "loss": 0.6109,
      "step": 24130
    },
    {
      "epoch": 1.8891845359211143,
      "grad_norm": 6.405465602874756,
      "learning_rate": 1.856332060469739e-05,
      "loss": 0.8649,
      "step": 24140
    },
    {
      "epoch": 1.8899671310064172,
      "grad_norm": 12.206430435180664,
      "learning_rate": 1.8550243239001935e-05,
      "loss": 0.6673,
      "step": 24150
    },
    {
      "epoch": 1.89074972609172,
      "grad_norm": 12.567753791809082,
      "learning_rate": 1.853716587330648e-05,
      "loss": 1.0794,
      "step": 24160
    },
    {
      "epoch": 1.891532321177023,
      "grad_norm": 3.2156906127929688,
      "learning_rate": 1.8524088507611028e-05,
      "loss": 0.5588,
      "step": 24170
    },
    {
      "epoch": 1.8923149162623258,
      "grad_norm": 1.2809009552001953,
      "learning_rate": 1.8511011141915576e-05,
      "loss": 0.6075,
      "step": 24180
    },
    {
      "epoch": 1.8930975113476287,
      "grad_norm": 7.122217178344727,
      "learning_rate": 1.849793377622012e-05,
      "loss": 0.7651,
      "step": 24190
    },
    {
      "epoch": 1.8938801064329316,
      "grad_norm": 11.724427223205566,
      "learning_rate": 1.8484856410524665e-05,
      "loss": 0.5806,
      "step": 24200
    },
    {
      "epoch": 1.8946627015182345,
      "grad_norm": 10.122334480285645,
      "learning_rate": 1.847177904482921e-05,
      "loss": 0.8263,
      "step": 24210
    },
    {
      "epoch": 1.8954452966035373,
      "grad_norm": 7.191051006317139,
      "learning_rate": 1.8458701679133754e-05,
      "loss": 0.8035,
      "step": 24220
    },
    {
      "epoch": 1.8962278916888402,
      "grad_norm": 27.009241104125977,
      "learning_rate": 1.8445624313438302e-05,
      "loss": 0.8574,
      "step": 24230
    },
    {
      "epoch": 1.897010486774143,
      "grad_norm": 1.6179457902908325,
      "learning_rate": 1.843254694774285e-05,
      "loss": 0.9458,
      "step": 24240
    },
    {
      "epoch": 1.897793081859446,
      "grad_norm": 11.52603530883789,
      "learning_rate": 1.8419469582047394e-05,
      "loss": 0.6188,
      "step": 24250
    },
    {
      "epoch": 1.8985756769447488,
      "grad_norm": 1.4866626262664795,
      "learning_rate": 1.840639221635194e-05,
      "loss": 0.6434,
      "step": 24260
    },
    {
      "epoch": 1.8993582720300517,
      "grad_norm": 5.479170322418213,
      "learning_rate": 1.8393314850656484e-05,
      "loss": 0.972,
      "step": 24270
    },
    {
      "epoch": 1.9001408671153546,
      "grad_norm": 13.938703536987305,
      "learning_rate": 1.8380237484961028e-05,
      "loss": 0.9001,
      "step": 24280
    },
    {
      "epoch": 1.9009234622006574,
      "grad_norm": 57.85873031616211,
      "learning_rate": 1.8367160119265576e-05,
      "loss": 0.7344,
      "step": 24290
    },
    {
      "epoch": 1.9017060572859603,
      "grad_norm": 8.725378036499023,
      "learning_rate": 1.835408275357012e-05,
      "loss": 0.6542,
      "step": 24300
    },
    {
      "epoch": 1.9024886523712632,
      "grad_norm": 12.553061485290527,
      "learning_rate": 1.834100538787467e-05,
      "loss": 0.7804,
      "step": 24310
    },
    {
      "epoch": 1.903271247456566,
      "grad_norm": 11.111990928649902,
      "learning_rate": 1.8327928022179213e-05,
      "loss": 0.8199,
      "step": 24320
    },
    {
      "epoch": 1.904053842541869,
      "grad_norm": 1.4458671808242798,
      "learning_rate": 1.8314850656483758e-05,
      "loss": 0.8225,
      "step": 24330
    },
    {
      "epoch": 1.9048364376271718,
      "grad_norm": 8.192154884338379,
      "learning_rate": 1.8301773290788306e-05,
      "loss": 0.6051,
      "step": 24340
    },
    {
      "epoch": 1.9056190327124747,
      "grad_norm": 11.247943878173828,
      "learning_rate": 1.828869592509285e-05,
      "loss": 1.0022,
      "step": 24350
    },
    {
      "epoch": 1.9064016277977776,
      "grad_norm": 19.24875259399414,
      "learning_rate": 1.8275618559397395e-05,
      "loss": 0.8854,
      "step": 24360
    },
    {
      "epoch": 1.9071842228830804,
      "grad_norm": 9.002843856811523,
      "learning_rate": 1.826254119370194e-05,
      "loss": 0.5746,
      "step": 24370
    },
    {
      "epoch": 1.9079668179683833,
      "grad_norm": 5.661463737487793,
      "learning_rate": 1.8249463828006487e-05,
      "loss": 0.5546,
      "step": 24380
    },
    {
      "epoch": 1.9087494130536862,
      "grad_norm": 20.632993698120117,
      "learning_rate": 1.8236386462311032e-05,
      "loss": 0.3641,
      "step": 24390
    },
    {
      "epoch": 1.909532008138989,
      "grad_norm": 20.37727928161621,
      "learning_rate": 1.822330909661558e-05,
      "loss": 0.6277,
      "step": 24400
    },
    {
      "epoch": 1.9103146032242917,
      "grad_norm": 6.7909159660339355,
      "learning_rate": 1.8210231730920125e-05,
      "loss": 1.0981,
      "step": 24410
    },
    {
      "epoch": 1.9110971983095946,
      "grad_norm": 0.6375157237052917,
      "learning_rate": 1.819715436522467e-05,
      "loss": 0.4274,
      "step": 24420
    },
    {
      "epoch": 1.9118797933948974,
      "grad_norm": 15.4984130859375,
      "learning_rate": 1.8184076999529214e-05,
      "loss": 0.8013,
      "step": 24430
    },
    {
      "epoch": 1.9126623884802003,
      "grad_norm": 7.787580966949463,
      "learning_rate": 1.817099963383376e-05,
      "loss": 0.8859,
      "step": 24440
    },
    {
      "epoch": 1.9134449835655032,
      "grad_norm": 17.329402923583984,
      "learning_rate": 1.815792226813831e-05,
      "loss": 0.7449,
      "step": 24450
    },
    {
      "epoch": 1.914227578650806,
      "grad_norm": 8.013160705566406,
      "learning_rate": 1.8144844902442854e-05,
      "loss": 1.2486,
      "step": 24460
    },
    {
      "epoch": 1.915010173736109,
      "grad_norm": 18.694644927978516,
      "learning_rate": 1.81317675367474e-05,
      "loss": 0.4386,
      "step": 24470
    },
    {
      "epoch": 1.9157927688214118,
      "grad_norm": 6.742979049682617,
      "learning_rate": 1.8118690171051943e-05,
      "loss": 0.9057,
      "step": 24480
    },
    {
      "epoch": 1.9165753639067147,
      "grad_norm": 10.124967575073242,
      "learning_rate": 1.8105612805356488e-05,
      "loss": 0.6432,
      "step": 24490
    },
    {
      "epoch": 1.9173579589920176,
      "grad_norm": 6.466037750244141,
      "learning_rate": 1.8092535439661036e-05,
      "loss": 0.7367,
      "step": 24500
    },
    {
      "epoch": 1.9181405540773204,
      "grad_norm": 11.196810722351074,
      "learning_rate": 1.807945807396558e-05,
      "loss": 0.9553,
      "step": 24510
    },
    {
      "epoch": 1.9189231491626233,
      "grad_norm": 14.218509674072266,
      "learning_rate": 1.806638070827013e-05,
      "loss": 0.5415,
      "step": 24520
    },
    {
      "epoch": 1.9197057442479262,
      "grad_norm": 7.938785076141357,
      "learning_rate": 1.8053303342574673e-05,
      "loss": 0.5667,
      "step": 24530
    },
    {
      "epoch": 1.9204883393332288,
      "grad_norm": 7.730929374694824,
      "learning_rate": 1.8040225976879218e-05,
      "loss": 0.8085,
      "step": 24540
    },
    {
      "epoch": 1.9212709344185317,
      "grad_norm": 7.006549835205078,
      "learning_rate": 1.8027148611183765e-05,
      "loss": 0.756,
      "step": 24550
    },
    {
      "epoch": 1.9220535295038346,
      "grad_norm": 5.373427391052246,
      "learning_rate": 1.801407124548831e-05,
      "loss": 0.8388,
      "step": 24560
    },
    {
      "epoch": 1.9228361245891374,
      "grad_norm": 6.263729572296143,
      "learning_rate": 1.8000993879792855e-05,
      "loss": 0.8731,
      "step": 24570
    },
    {
      "epoch": 1.9236187196744403,
      "grad_norm": 5.810124397277832,
      "learning_rate": 1.79879165140974e-05,
      "loss": 0.694,
      "step": 24580
    },
    {
      "epoch": 1.9244013147597432,
      "grad_norm": 7.045419692993164,
      "learning_rate": 1.7974839148401947e-05,
      "loss": 0.6729,
      "step": 24590
    },
    {
      "epoch": 1.925183909845046,
      "grad_norm": 0.9184361696243286,
      "learning_rate": 1.7961761782706492e-05,
      "loss": 0.8987,
      "step": 24600
    },
    {
      "epoch": 1.925966504930349,
      "grad_norm": 6.64227294921875,
      "learning_rate": 1.794868441701104e-05,
      "loss": 1.0971,
      "step": 24610
    },
    {
      "epoch": 1.9267491000156518,
      "grad_norm": 1.9795728921890259,
      "learning_rate": 1.7935607051315584e-05,
      "loss": 0.8587,
      "step": 24620
    },
    {
      "epoch": 1.9275316951009547,
      "grad_norm": 7.195316314697266,
      "learning_rate": 1.792252968562013e-05,
      "loss": 0.8755,
      "step": 24630
    },
    {
      "epoch": 1.9283142901862576,
      "grad_norm": 6.9714579582214355,
      "learning_rate": 1.7909452319924673e-05,
      "loss": 0.7588,
      "step": 24640
    },
    {
      "epoch": 1.9290968852715604,
      "grad_norm": 7.5026068687438965,
      "learning_rate": 1.789637495422922e-05,
      "loss": 0.6119,
      "step": 24650
    },
    {
      "epoch": 1.9298794803568633,
      "grad_norm": 19.107507705688477,
      "learning_rate": 1.788329758853377e-05,
      "loss": 0.6944,
      "step": 24660
    },
    {
      "epoch": 1.9306620754421662,
      "grad_norm": 8.85046672821045,
      "learning_rate": 1.7870220222838314e-05,
      "loss": 0.8125,
      "step": 24670
    },
    {
      "epoch": 1.931444670527469,
      "grad_norm": 6.427289962768555,
      "learning_rate": 1.785714285714286e-05,
      "loss": 0.7412,
      "step": 24680
    },
    {
      "epoch": 1.932227265612772,
      "grad_norm": 3.2617099285125732,
      "learning_rate": 1.7844065491447403e-05,
      "loss": 0.5656,
      "step": 24690
    },
    {
      "epoch": 1.9330098606980748,
      "grad_norm": 7.4547200202941895,
      "learning_rate": 1.7830988125751948e-05,
      "loss": 0.5271,
      "step": 24700
    },
    {
      "epoch": 1.9337924557833777,
      "grad_norm": 13.843535423278809,
      "learning_rate": 1.7817910760056496e-05,
      "loss": 1.3763,
      "step": 24710
    },
    {
      "epoch": 1.9345750508686805,
      "grad_norm": 8.032967567443848,
      "learning_rate": 1.780483339436104e-05,
      "loss": 0.9023,
      "step": 24720
    },
    {
      "epoch": 1.9353576459539834,
      "grad_norm": 9.051257133483887,
      "learning_rate": 1.7791756028665588e-05,
      "loss": 0.967,
      "step": 24730
    },
    {
      "epoch": 1.9361402410392863,
      "grad_norm": 5.771273136138916,
      "learning_rate": 1.7778678662970133e-05,
      "loss": 0.7628,
      "step": 24740
    },
    {
      "epoch": 1.9369228361245892,
      "grad_norm": 4.542823314666748,
      "learning_rate": 1.7765601297274677e-05,
      "loss": 0.9043,
      "step": 24750
    },
    {
      "epoch": 1.937705431209892,
      "grad_norm": 1.1710137128829956,
      "learning_rate": 1.7752523931579225e-05,
      "loss": 0.7274,
      "step": 24760
    },
    {
      "epoch": 1.938488026295195,
      "grad_norm": 12.623332977294922,
      "learning_rate": 1.773944656588377e-05,
      "loss": 0.7785,
      "step": 24770
    },
    {
      "epoch": 1.9392706213804978,
      "grad_norm": 5.706684112548828,
      "learning_rate": 1.7726369200188314e-05,
      "loss": 0.7305,
      "step": 24780
    },
    {
      "epoch": 1.9400532164658006,
      "grad_norm": 5.985862731933594,
      "learning_rate": 1.771329183449286e-05,
      "loss": 0.6921,
      "step": 24790
    },
    {
      "epoch": 1.9408358115511035,
      "grad_norm": 8.603960037231445,
      "learning_rate": 1.7700214468797407e-05,
      "loss": 0.4535,
      "step": 24800
    },
    {
      "epoch": 1.9416184066364064,
      "grad_norm": 4.134652137756348,
      "learning_rate": 1.768713710310195e-05,
      "loss": 0.885,
      "step": 24810
    },
    {
      "epoch": 1.9424010017217093,
      "grad_norm": 10.97968578338623,
      "learning_rate": 1.76740597374065e-05,
      "loss": 0.854,
      "step": 24820
    },
    {
      "epoch": 1.9431835968070121,
      "grad_norm": 11.716464042663574,
      "learning_rate": 1.7660982371711044e-05,
      "loss": 0.8552,
      "step": 24830
    },
    {
      "epoch": 1.943966191892315,
      "grad_norm": 4.674681663513184,
      "learning_rate": 1.764790500601559e-05,
      "loss": 0.9223,
      "step": 24840
    },
    {
      "epoch": 1.9447487869776179,
      "grad_norm": 10.966365814208984,
      "learning_rate": 1.7634827640320133e-05,
      "loss": 0.5059,
      "step": 24850
    },
    {
      "epoch": 1.9455313820629208,
      "grad_norm": 6.245600700378418,
      "learning_rate": 1.7621750274624678e-05,
      "loss": 0.733,
      "step": 24860
    },
    {
      "epoch": 1.9463139771482236,
      "grad_norm": 2.140138626098633,
      "learning_rate": 1.760867290892923e-05,
      "loss": 0.6349,
      "step": 24870
    },
    {
      "epoch": 1.9470965722335265,
      "grad_norm": 2.419879674911499,
      "learning_rate": 1.7595595543233774e-05,
      "loss": 0.5622,
      "step": 24880
    },
    {
      "epoch": 1.9478791673188294,
      "grad_norm": 9.233438491821289,
      "learning_rate": 1.7582518177538318e-05,
      "loss": 0.7982,
      "step": 24890
    },
    {
      "epoch": 1.9486617624041322,
      "grad_norm": 7.977922439575195,
      "learning_rate": 1.7569440811842863e-05,
      "loss": 0.6835,
      "step": 24900
    },
    {
      "epoch": 1.9494443574894351,
      "grad_norm": 7.151707172393799,
      "learning_rate": 1.7556363446147407e-05,
      "loss": 1.235,
      "step": 24910
    },
    {
      "epoch": 1.950226952574738,
      "grad_norm": 5.4568071365356445,
      "learning_rate": 1.7543286080451955e-05,
      "loss": 0.5078,
      "step": 24920
    },
    {
      "epoch": 1.9510095476600406,
      "grad_norm": 12.243608474731445,
      "learning_rate": 1.75302087147565e-05,
      "loss": 0.598,
      "step": 24930
    },
    {
      "epoch": 1.9517921427453435,
      "grad_norm": 8.057435035705566,
      "learning_rate": 1.7517131349061048e-05,
      "loss": 0.8254,
      "step": 24940
    },
    {
      "epoch": 1.9525747378306464,
      "grad_norm": 1.0648618936538696,
      "learning_rate": 1.7504053983365592e-05,
      "loss": 0.8755,
      "step": 24950
    },
    {
      "epoch": 1.9533573329159493,
      "grad_norm": 29.923450469970703,
      "learning_rate": 1.7490976617670137e-05,
      "loss": 0.7155,
      "step": 24960
    },
    {
      "epoch": 1.9541399280012521,
      "grad_norm": 1.019179105758667,
      "learning_rate": 1.747789925197468e-05,
      "loss": 0.8585,
      "step": 24970
    },
    {
      "epoch": 1.954922523086555,
      "grad_norm": 0.7632762789726257,
      "learning_rate": 1.746482188627923e-05,
      "loss": 1.2192,
      "step": 24980
    },
    {
      "epoch": 1.9557051181718579,
      "grad_norm": 22.63650131225586,
      "learning_rate": 1.7451744520583774e-05,
      "loss": 0.8799,
      "step": 24990
    },
    {
      "epoch": 1.9564877132571608,
      "grad_norm": 5.567646503448486,
      "learning_rate": 1.743866715488832e-05,
      "loss": 0.7341,
      "step": 25000
    },
    {
      "epoch": 1.9572703083424636,
      "grad_norm": 12.822020530700684,
      "learning_rate": 1.7425589789192867e-05,
      "loss": 0.8983,
      "step": 25010
    },
    {
      "epoch": 1.9580529034277665,
      "grad_norm": 1.1644936800003052,
      "learning_rate": 1.741251242349741e-05,
      "loss": 0.6001,
      "step": 25020
    },
    {
      "epoch": 1.9588354985130694,
      "grad_norm": 7.061683654785156,
      "learning_rate": 1.739943505780196e-05,
      "loss": 0.6601,
      "step": 25030
    },
    {
      "epoch": 1.9596180935983722,
      "grad_norm": 5.430846214294434,
      "learning_rate": 1.7386357692106504e-05,
      "loss": 0.5416,
      "step": 25040
    },
    {
      "epoch": 1.9604006886836751,
      "grad_norm": 7.669902801513672,
      "learning_rate": 1.737328032641105e-05,
      "loss": 0.7013,
      "step": 25050
    },
    {
      "epoch": 1.9611832837689778,
      "grad_norm": 10.214205741882324,
      "learning_rate": 1.7360202960715593e-05,
      "loss": 1.1638,
      "step": 25060
    },
    {
      "epoch": 1.9619658788542806,
      "grad_norm": 10.038572311401367,
      "learning_rate": 1.7347125595020137e-05,
      "loss": 0.8103,
      "step": 25070
    },
    {
      "epoch": 1.9627484739395835,
      "grad_norm": 1.2127084732055664,
      "learning_rate": 1.7334048229324685e-05,
      "loss": 0.6753,
      "step": 25080
    },
    {
      "epoch": 1.9635310690248864,
      "grad_norm": 9.09495735168457,
      "learning_rate": 1.7320970863629233e-05,
      "loss": 1.216,
      "step": 25090
    },
    {
      "epoch": 1.9643136641101893,
      "grad_norm": 4.490250587463379,
      "learning_rate": 1.7307893497933778e-05,
      "loss": 0.6897,
      "step": 25100
    },
    {
      "epoch": 1.9650962591954921,
      "grad_norm": 8.38504695892334,
      "learning_rate": 1.7294816132238323e-05,
      "loss": 0.8961,
      "step": 25110
    },
    {
      "epoch": 1.965878854280795,
      "grad_norm": 5.883788108825684,
      "learning_rate": 1.7281738766542867e-05,
      "loss": 0.7243,
      "step": 25120
    },
    {
      "epoch": 1.9666614493660979,
      "grad_norm": 2.4146220684051514,
      "learning_rate": 1.7268661400847415e-05,
      "loss": 0.6765,
      "step": 25130
    },
    {
      "epoch": 1.9674440444514008,
      "grad_norm": 14.493239402770996,
      "learning_rate": 1.725558403515196e-05,
      "loss": 0.525,
      "step": 25140
    },
    {
      "epoch": 1.9682266395367036,
      "grad_norm": 9.767292976379395,
      "learning_rate": 1.7242506669456508e-05,
      "loss": 0.7883,
      "step": 25150
    },
    {
      "epoch": 1.9690092346220065,
      "grad_norm": 9.645112991333008,
      "learning_rate": 1.7229429303761052e-05,
      "loss": 0.5296,
      "step": 25160
    },
    {
      "epoch": 1.9697918297073094,
      "grad_norm": 8.444754600524902,
      "learning_rate": 1.7216351938065597e-05,
      "loss": 0.6841,
      "step": 25170
    },
    {
      "epoch": 1.9705744247926122,
      "grad_norm": 4.423091411590576,
      "learning_rate": 1.720327457237014e-05,
      "loss": 0.8984,
      "step": 25180
    },
    {
      "epoch": 1.9713570198779151,
      "grad_norm": 1.4006595611572266,
      "learning_rate": 1.719019720667469e-05,
      "loss": 0.9831,
      "step": 25190
    },
    {
      "epoch": 1.972139614963218,
      "grad_norm": 6.540835857391357,
      "learning_rate": 1.7177119840979234e-05,
      "loss": 0.8016,
      "step": 25200
    },
    {
      "epoch": 1.9729222100485209,
      "grad_norm": 1.8392066955566406,
      "learning_rate": 1.716404247528378e-05,
      "loss": 0.6865,
      "step": 25210
    },
    {
      "epoch": 1.9737048051338237,
      "grad_norm": 8.221014022827148,
      "learning_rate": 1.7150965109588326e-05,
      "loss": 0.7339,
      "step": 25220
    },
    {
      "epoch": 1.9744874002191266,
      "grad_norm": 4.5739240646362305,
      "learning_rate": 1.713788774389287e-05,
      "loss": 0.5657,
      "step": 25230
    },
    {
      "epoch": 1.9752699953044295,
      "grad_norm": 4.0027031898498535,
      "learning_rate": 1.712481037819742e-05,
      "loss": 0.7003,
      "step": 25240
    },
    {
      "epoch": 1.9760525903897324,
      "grad_norm": 6.7804789543151855,
      "learning_rate": 1.7111733012501963e-05,
      "loss": 0.756,
      "step": 25250
    },
    {
      "epoch": 1.9768351854750352,
      "grad_norm": 10.052289009094238,
      "learning_rate": 1.7098655646806508e-05,
      "loss": 0.8111,
      "step": 25260
    },
    {
      "epoch": 1.977617780560338,
      "grad_norm": 4.044284343719482,
      "learning_rate": 1.7085578281111053e-05,
      "loss": 0.4933,
      "step": 25270
    },
    {
      "epoch": 1.978400375645641,
      "grad_norm": 8.88404369354248,
      "learning_rate": 1.7072500915415597e-05,
      "loss": 0.9994,
      "step": 25280
    },
    {
      "epoch": 1.9791829707309438,
      "grad_norm": 1.3912770748138428,
      "learning_rate": 1.7059423549720145e-05,
      "loss": 0.4543,
      "step": 25290
    },
    {
      "epoch": 1.9799655658162467,
      "grad_norm": 8.334857940673828,
      "learning_rate": 1.7046346184024693e-05,
      "loss": 0.7617,
      "step": 25300
    },
    {
      "epoch": 1.9807481609015496,
      "grad_norm": 8.540970802307129,
      "learning_rate": 1.7033268818329238e-05,
      "loss": 0.8241,
      "step": 25310
    },
    {
      "epoch": 1.9815307559868525,
      "grad_norm": 8.965055465698242,
      "learning_rate": 1.7020191452633782e-05,
      "loss": 0.7439,
      "step": 25320
    },
    {
      "epoch": 1.9823133510721553,
      "grad_norm": 5.22905158996582,
      "learning_rate": 1.7007114086938327e-05,
      "loss": 0.5722,
      "step": 25330
    },
    {
      "epoch": 1.9830959461574582,
      "grad_norm": 1.1053564548492432,
      "learning_rate": 1.699403672124287e-05,
      "loss": 0.5377,
      "step": 25340
    },
    {
      "epoch": 1.983878541242761,
      "grad_norm": 11.045722007751465,
      "learning_rate": 1.698095935554742e-05,
      "loss": 0.6284,
      "step": 25350
    },
    {
      "epoch": 1.984661136328064,
      "grad_norm": 5.840368747711182,
      "learning_rate": 1.6967881989851967e-05,
      "loss": 0.9204,
      "step": 25360
    },
    {
      "epoch": 1.9854437314133668,
      "grad_norm": 0.7054376602172852,
      "learning_rate": 1.6954804624156512e-05,
      "loss": 0.8872,
      "step": 25370
    },
    {
      "epoch": 1.9862263264986697,
      "grad_norm": 21.96257972717285,
      "learning_rate": 1.6941727258461056e-05,
      "loss": 0.9403,
      "step": 25380
    },
    {
      "epoch": 1.9870089215839726,
      "grad_norm": 15.434897422790527,
      "learning_rate": 1.69286498927656e-05,
      "loss": 0.721,
      "step": 25390
    },
    {
      "epoch": 1.9877915166692754,
      "grad_norm": 12.664597511291504,
      "learning_rate": 1.691557252707015e-05,
      "loss": 0.8986,
      "step": 25400
    },
    {
      "epoch": 1.9885741117545783,
      "grad_norm": 12.699058532714844,
      "learning_rate": 1.6902495161374694e-05,
      "loss": 0.6837,
      "step": 25410
    },
    {
      "epoch": 1.9893567068398812,
      "grad_norm": 18.498661041259766,
      "learning_rate": 1.6889417795679238e-05,
      "loss": 0.6095,
      "step": 25420
    },
    {
      "epoch": 1.990139301925184,
      "grad_norm": 7.893812656402588,
      "learning_rate": 1.6876340429983783e-05,
      "loss": 0.6869,
      "step": 25430
    },
    {
      "epoch": 1.990921897010487,
      "grad_norm": 14.542349815368652,
      "learning_rate": 1.686326306428833e-05,
      "loss": 0.976,
      "step": 25440
    },
    {
      "epoch": 1.9917044920957896,
      "grad_norm": 11.50949478149414,
      "learning_rate": 1.6850185698592875e-05,
      "loss": 0.8875,
      "step": 25450
    },
    {
      "epoch": 1.9924870871810925,
      "grad_norm": 4.380439758300781,
      "learning_rate": 1.6837108332897423e-05,
      "loss": 0.6673,
      "step": 25460
    },
    {
      "epoch": 1.9932696822663953,
      "grad_norm": 11.190052032470703,
      "learning_rate": 1.6824030967201968e-05,
      "loss": 0.7291,
      "step": 25470
    },
    {
      "epoch": 1.9940522773516982,
      "grad_norm": 1.1678557395935059,
      "learning_rate": 1.6810953601506512e-05,
      "loss": 0.5352,
      "step": 25480
    },
    {
      "epoch": 1.994834872437001,
      "grad_norm": 2.5017154216766357,
      "learning_rate": 1.6797876235811057e-05,
      "loss": 0.4033,
      "step": 25490
    },
    {
      "epoch": 1.995617467522304,
      "grad_norm": 9.645197868347168,
      "learning_rate": 1.6784798870115605e-05,
      "loss": 0.8094,
      "step": 25500
    },
    {
      "epoch": 1.9964000626076068,
      "grad_norm": 13.466177940368652,
      "learning_rate": 1.6771721504420153e-05,
      "loss": 0.9934,
      "step": 25510
    },
    {
      "epoch": 1.9971826576929097,
      "grad_norm": 8.420391082763672,
      "learning_rate": 1.6758644138724697e-05,
      "loss": 0.6728,
      "step": 25520
    },
    {
      "epoch": 1.9979652527782126,
      "grad_norm": 0.9046197533607483,
      "learning_rate": 1.6745566773029242e-05,
      "loss": 1.0707,
      "step": 25530
    },
    {
      "epoch": 1.9987478478635154,
      "grad_norm": 14.365519523620605,
      "learning_rate": 1.6732489407333787e-05,
      "loss": 1.0351,
      "step": 25540
    },
    {
      "epoch": 1.9995304429488183,
      "grad_norm": 14.952195167541504,
      "learning_rate": 1.671941204163833e-05,
      "loss": 0.86,
      "step": 25550
    },
    {
      "epoch": 2.000313038034121,
      "grad_norm": 11.284976959228516,
      "learning_rate": 1.670633467594288e-05,
      "loss": 0.7356,
      "step": 25560
    },
    {
      "epoch": 2.001095633119424,
      "grad_norm": 20.481517791748047,
      "learning_rate": 1.6693257310247424e-05,
      "loss": 0.9171,
      "step": 25570
    },
    {
      "epoch": 2.0018782282047267,
      "grad_norm": 9.60076904296875,
      "learning_rate": 1.668017994455197e-05,
      "loss": 0.8,
      "step": 25580
    },
    {
      "epoch": 2.0026608232900296,
      "grad_norm": 7.574582576751709,
      "learning_rate": 1.6667102578856516e-05,
      "loss": 0.5705,
      "step": 25590
    },
    {
      "epoch": 2.0034434183753325,
      "grad_norm": 4.862351417541504,
      "learning_rate": 1.665402521316106e-05,
      "loss": 0.8314,
      "step": 25600
    },
    {
      "epoch": 2.0042260134606353,
      "grad_norm": 9.92888069152832,
      "learning_rate": 1.664094784746561e-05,
      "loss": 0.3921,
      "step": 25610
    },
    {
      "epoch": 2.005008608545938,
      "grad_norm": 26.494823455810547,
      "learning_rate": 1.6627870481770153e-05,
      "loss": 0.4808,
      "step": 25620
    },
    {
      "epoch": 2.005791203631241,
      "grad_norm": 28.09138298034668,
      "learning_rate": 1.6614793116074698e-05,
      "loss": 0.8012,
      "step": 25630
    },
    {
      "epoch": 2.006573798716544,
      "grad_norm": 1.039027214050293,
      "learning_rate": 1.6601715750379242e-05,
      "loss": 0.2504,
      "step": 25640
    },
    {
      "epoch": 2.007356393801847,
      "grad_norm": 0.9664219617843628,
      "learning_rate": 1.658863838468379e-05,
      "loss": 0.8868,
      "step": 25650
    },
    {
      "epoch": 2.0081389888871497,
      "grad_norm": 5.172874927520752,
      "learning_rate": 1.6575561018988335e-05,
      "loss": 0.7963,
      "step": 25660
    },
    {
      "epoch": 2.0089215839724526,
      "grad_norm": 5.766213893890381,
      "learning_rate": 1.6562483653292883e-05,
      "loss": 0.5926,
      "step": 25670
    },
    {
      "epoch": 2.0097041790577554,
      "grad_norm": 10.66617202758789,
      "learning_rate": 1.6549406287597428e-05,
      "loss": 0.6087,
      "step": 25680
    },
    {
      "epoch": 2.0104867741430583,
      "grad_norm": 5.7942328453063965,
      "learning_rate": 1.6536328921901972e-05,
      "loss": 0.9995,
      "step": 25690
    },
    {
      "epoch": 2.011269369228361,
      "grad_norm": 5.991124153137207,
      "learning_rate": 1.6523251556206517e-05,
      "loss": 0.8445,
      "step": 25700
    },
    {
      "epoch": 2.012051964313664,
      "grad_norm": 5.200613498687744,
      "learning_rate": 1.6510174190511065e-05,
      "loss": 0.4738,
      "step": 25710
    },
    {
      "epoch": 2.012834559398967,
      "grad_norm": 28.848329544067383,
      "learning_rate": 1.6497096824815613e-05,
      "loss": 0.6617,
      "step": 25720
    },
    {
      "epoch": 2.01361715448427,
      "grad_norm": 11.892016410827637,
      "learning_rate": 1.6484019459120157e-05,
      "loss": 0.7185,
      "step": 25730
    },
    {
      "epoch": 2.0143997495695727,
      "grad_norm": 11.727056503295898,
      "learning_rate": 1.6470942093424702e-05,
      "loss": 0.6683,
      "step": 25740
    },
    {
      "epoch": 2.0151823446548756,
      "grad_norm": 0.8492801785469055,
      "learning_rate": 1.6457864727729246e-05,
      "loss": 1.0339,
      "step": 25750
    },
    {
      "epoch": 2.0159649397401784,
      "grad_norm": 6.839691162109375,
      "learning_rate": 1.644478736203379e-05,
      "loss": 0.9492,
      "step": 25760
    },
    {
      "epoch": 2.0167475348254813,
      "grad_norm": 4.980136394500732,
      "learning_rate": 1.643170999633834e-05,
      "loss": 0.6009,
      "step": 25770
    },
    {
      "epoch": 2.017530129910784,
      "grad_norm": 9.131011009216309,
      "learning_rate": 1.6418632630642883e-05,
      "loss": 0.6824,
      "step": 25780
    },
    {
      "epoch": 2.018312724996087,
      "grad_norm": 10.256357192993164,
      "learning_rate": 1.640555526494743e-05,
      "loss": 0.5893,
      "step": 25790
    },
    {
      "epoch": 2.01909532008139,
      "grad_norm": 15.386486053466797,
      "learning_rate": 1.6392477899251976e-05,
      "loss": 1.1391,
      "step": 25800
    },
    {
      "epoch": 2.019877915166693,
      "grad_norm": 16.300949096679688,
      "learning_rate": 1.637940053355652e-05,
      "loss": 0.7034,
      "step": 25810
    },
    {
      "epoch": 2.0206605102519957,
      "grad_norm": 16.920703887939453,
      "learning_rate": 1.636632316786107e-05,
      "loss": 0.9682,
      "step": 25820
    },
    {
      "epoch": 2.0214431053372985,
      "grad_norm": 0.8947649598121643,
      "learning_rate": 1.6353245802165613e-05,
      "loss": 0.7654,
      "step": 25830
    },
    {
      "epoch": 2.0222257004226014,
      "grad_norm": 15.256795883178711,
      "learning_rate": 1.6340168436470158e-05,
      "loss": 0.4929,
      "step": 25840
    },
    {
      "epoch": 2.0230082955079043,
      "grad_norm": 18.551897048950195,
      "learning_rate": 1.6327091070774702e-05,
      "loss": 0.8247,
      "step": 25850
    },
    {
      "epoch": 2.023790890593207,
      "grad_norm": 6.337754726409912,
      "learning_rate": 1.631401370507925e-05,
      "loss": 0.7138,
      "step": 25860
    },
    {
      "epoch": 2.02457348567851,
      "grad_norm": 6.712518215179443,
      "learning_rate": 1.6300936339383795e-05,
      "loss": 0.7995,
      "step": 25870
    },
    {
      "epoch": 2.025356080763813,
      "grad_norm": 50.981529235839844,
      "learning_rate": 1.6287858973688343e-05,
      "loss": 0.8525,
      "step": 25880
    },
    {
      "epoch": 2.0261386758491158,
      "grad_norm": 7.637388229370117,
      "learning_rate": 1.6274781607992887e-05,
      "loss": 0.4495,
      "step": 25890
    },
    {
      "epoch": 2.0269212709344187,
      "grad_norm": 18.750091552734375,
      "learning_rate": 1.6261704242297432e-05,
      "loss": 0.641,
      "step": 25900
    },
    {
      "epoch": 2.0277038660197215,
      "grad_norm": 3.197986364364624,
      "learning_rate": 1.6248626876601976e-05,
      "loss": 1.1007,
      "step": 25910
    },
    {
      "epoch": 2.0284864611050244,
      "grad_norm": 3.0514261722564697,
      "learning_rate": 1.623554951090652e-05,
      "loss": 0.7976,
      "step": 25920
    },
    {
      "epoch": 2.0292690561903273,
      "grad_norm": 0.861305296421051,
      "learning_rate": 1.622247214521107e-05,
      "loss": 0.5024,
      "step": 25930
    },
    {
      "epoch": 2.03005165127563,
      "grad_norm": 8.591322898864746,
      "learning_rate": 1.6209394779515617e-05,
      "loss": 0.6448,
      "step": 25940
    },
    {
      "epoch": 2.030834246360933,
      "grad_norm": 7.225265979766846,
      "learning_rate": 1.619631741382016e-05,
      "loss": 0.4697,
      "step": 25950
    },
    {
      "epoch": 2.031616841446236,
      "grad_norm": 1.6502920389175415,
      "learning_rate": 1.6183240048124706e-05,
      "loss": 0.3545,
      "step": 25960
    },
    {
      "epoch": 2.0323994365315388,
      "grad_norm": 6.69855260848999,
      "learning_rate": 1.617016268242925e-05,
      "loss": 0.41,
      "step": 25970
    },
    {
      "epoch": 2.0331820316168416,
      "grad_norm": 19.02666473388672,
      "learning_rate": 1.61570853167338e-05,
      "loss": 0.704,
      "step": 25980
    },
    {
      "epoch": 2.0339646267021445,
      "grad_norm": 14.051426887512207,
      "learning_rate": 1.6144007951038343e-05,
      "loss": 0.6725,
      "step": 25990
    },
    {
      "epoch": 2.0347472217874474,
      "grad_norm": 8.831354141235352,
      "learning_rate": 1.613093058534289e-05,
      "loss": 1.2932,
      "step": 26000
    },
    {
      "epoch": 2.0355298168727503,
      "grad_norm": 21.920621871948242,
      "learning_rate": 1.6117853219647436e-05,
      "loss": 0.9508,
      "step": 26010
    },
    {
      "epoch": 2.036312411958053,
      "grad_norm": 4.376535415649414,
      "learning_rate": 1.610477585395198e-05,
      "loss": 0.8632,
      "step": 26020
    },
    {
      "epoch": 2.0370950070433556,
      "grad_norm": 19.6162166595459,
      "learning_rate": 1.6091698488256525e-05,
      "loss": 1.1749,
      "step": 26030
    },
    {
      "epoch": 2.0378776021286584,
      "grad_norm": 8.316838264465332,
      "learning_rate": 1.6078621122561073e-05,
      "loss": 0.6995,
      "step": 26040
    },
    {
      "epoch": 2.0386601972139613,
      "grad_norm": 14.489377975463867,
      "learning_rate": 1.6065543756865617e-05,
      "loss": 0.5411,
      "step": 26050
    },
    {
      "epoch": 2.039442792299264,
      "grad_norm": 0.9984341859817505,
      "learning_rate": 1.6052466391170162e-05,
      "loss": 0.757,
      "step": 26060
    },
    {
      "epoch": 2.040225387384567,
      "grad_norm": 2.5882925987243652,
      "learning_rate": 1.603938902547471e-05,
      "loss": 0.901,
      "step": 26070
    },
    {
      "epoch": 2.04100798246987,
      "grad_norm": 2.4812119007110596,
      "learning_rate": 1.6026311659779254e-05,
      "loss": 0.4714,
      "step": 26080
    },
    {
      "epoch": 2.041790577555173,
      "grad_norm": 7.741962909698486,
      "learning_rate": 1.6013234294083802e-05,
      "loss": 0.6613,
      "step": 26090
    },
    {
      "epoch": 2.0425731726404757,
      "grad_norm": 11.597372055053711,
      "learning_rate": 1.6000156928388347e-05,
      "loss": 0.6358,
      "step": 26100
    },
    {
      "epoch": 2.0433557677257785,
      "grad_norm": 6.784639358520508,
      "learning_rate": 1.598707956269289e-05,
      "loss": 0.7495,
      "step": 26110
    },
    {
      "epoch": 2.0441383628110814,
      "grad_norm": 6.173268795013428,
      "learning_rate": 1.5974002196997436e-05,
      "loss": 0.6733,
      "step": 26120
    },
    {
      "epoch": 2.0449209578963843,
      "grad_norm": 5.741412162780762,
      "learning_rate": 1.596092483130198e-05,
      "loss": 0.6308,
      "step": 26130
    },
    {
      "epoch": 2.045703552981687,
      "grad_norm": 12.254070281982422,
      "learning_rate": 1.594784746560653e-05,
      "loss": 0.876,
      "step": 26140
    },
    {
      "epoch": 2.04648614806699,
      "grad_norm": 12.799230575561523,
      "learning_rate": 1.5934770099911077e-05,
      "loss": 1.0755,
      "step": 26150
    },
    {
      "epoch": 2.047268743152293,
      "grad_norm": 4.918544769287109,
      "learning_rate": 1.592169273421562e-05,
      "loss": 0.6887,
      "step": 26160
    },
    {
      "epoch": 2.0480513382375958,
      "grad_norm": 10.988869667053223,
      "learning_rate": 1.5908615368520166e-05,
      "loss": 0.7025,
      "step": 26170
    },
    {
      "epoch": 2.0488339333228986,
      "grad_norm": 7.832058429718018,
      "learning_rate": 1.589553800282471e-05,
      "loss": 0.6417,
      "step": 26180
    },
    {
      "epoch": 2.0496165284082015,
      "grad_norm": 1.5952489376068115,
      "learning_rate": 1.5882460637129258e-05,
      "loss": 0.5206,
      "step": 26190
    },
    {
      "epoch": 2.0503991234935044,
      "grad_norm": 7.820155143737793,
      "learning_rate": 1.5869383271433803e-05,
      "loss": 0.9149,
      "step": 26200
    },
    {
      "epoch": 2.0511817185788073,
      "grad_norm": 24.868810653686523,
      "learning_rate": 1.585630590573835e-05,
      "loss": 0.7751,
      "step": 26210
    },
    {
      "epoch": 2.05196431366411,
      "grad_norm": 7.005501747131348,
      "learning_rate": 1.5843228540042895e-05,
      "loss": 0.6366,
      "step": 26220
    },
    {
      "epoch": 2.052746908749413,
      "grad_norm": 7.512907028198242,
      "learning_rate": 1.583015117434744e-05,
      "loss": 0.604,
      "step": 26230
    },
    {
      "epoch": 2.053529503834716,
      "grad_norm": 5.657446384429932,
      "learning_rate": 1.5817073808651985e-05,
      "loss": 0.6716,
      "step": 26240
    },
    {
      "epoch": 2.0543120989200188,
      "grad_norm": 5.063937664031982,
      "learning_rate": 1.5803996442956533e-05,
      "loss": 0.6998,
      "step": 26250
    },
    {
      "epoch": 2.0550946940053216,
      "grad_norm": 11.866458892822266,
      "learning_rate": 1.5790919077261077e-05,
      "loss": 0.6685,
      "step": 26260
    },
    {
      "epoch": 2.0558772890906245,
      "grad_norm": 13.181941032409668,
      "learning_rate": 1.577784171156562e-05,
      "loss": 1.017,
      "step": 26270
    },
    {
      "epoch": 2.0566598841759274,
      "grad_norm": 0.9448202848434448,
      "learning_rate": 1.576476434587017e-05,
      "loss": 0.7605,
      "step": 26280
    },
    {
      "epoch": 2.0574424792612303,
      "grad_norm": 15.40060043334961,
      "learning_rate": 1.5751686980174714e-05,
      "loss": 0.6852,
      "step": 26290
    },
    {
      "epoch": 2.058225074346533,
      "grad_norm": 6.514410018920898,
      "learning_rate": 1.5738609614479262e-05,
      "loss": 0.9345,
      "step": 26300
    },
    {
      "epoch": 2.059007669431836,
      "grad_norm": 16.644725799560547,
      "learning_rate": 1.5725532248783807e-05,
      "loss": 0.7018,
      "step": 26310
    },
    {
      "epoch": 2.059790264517139,
      "grad_norm": 2.9845893383026123,
      "learning_rate": 1.571245488308835e-05,
      "loss": 0.4975,
      "step": 26320
    },
    {
      "epoch": 2.0605728596024417,
      "grad_norm": 19.66400909423828,
      "learning_rate": 1.5699377517392896e-05,
      "loss": 0.919,
      "step": 26330
    },
    {
      "epoch": 2.0613554546877446,
      "grad_norm": 19.03360366821289,
      "learning_rate": 1.568630015169744e-05,
      "loss": 0.8449,
      "step": 26340
    },
    {
      "epoch": 2.0621380497730475,
      "grad_norm": 8.46846866607666,
      "learning_rate": 1.567322278600199e-05,
      "loss": 0.8054,
      "step": 26350
    },
    {
      "epoch": 2.0629206448583504,
      "grad_norm": 1.3813949823379517,
      "learning_rate": 1.5660145420306536e-05,
      "loss": 0.4392,
      "step": 26360
    },
    {
      "epoch": 2.0637032399436532,
      "grad_norm": 35.599815368652344,
      "learning_rate": 1.564706805461108e-05,
      "loss": 0.6413,
      "step": 26370
    },
    {
      "epoch": 2.064485835028956,
      "grad_norm": 6.871073246002197,
      "learning_rate": 1.5633990688915626e-05,
      "loss": 0.8446,
      "step": 26380
    },
    {
      "epoch": 2.065268430114259,
      "grad_norm": 18.574886322021484,
      "learning_rate": 1.562091332322017e-05,
      "loss": 0.4749,
      "step": 26390
    },
    {
      "epoch": 2.066051025199562,
      "grad_norm": 15.296669006347656,
      "learning_rate": 1.5607835957524715e-05,
      "loss": 0.7521,
      "step": 26400
    },
    {
      "epoch": 2.0668336202848647,
      "grad_norm": 7.355377674102783,
      "learning_rate": 1.5594758591829263e-05,
      "loss": 0.6066,
      "step": 26410
    },
    {
      "epoch": 2.0676162153701676,
      "grad_norm": 0.6190397143363953,
      "learning_rate": 1.558168122613381e-05,
      "loss": 0.3655,
      "step": 26420
    },
    {
      "epoch": 2.0683988104554705,
      "grad_norm": 5.672061443328857,
      "learning_rate": 1.5568603860438355e-05,
      "loss": 0.4754,
      "step": 26430
    },
    {
      "epoch": 2.0691814055407733,
      "grad_norm": 7.040503978729248,
      "learning_rate": 1.55555264947429e-05,
      "loss": 0.5246,
      "step": 26440
    },
    {
      "epoch": 2.069964000626076,
      "grad_norm": 15.722984313964844,
      "learning_rate": 1.5542449129047444e-05,
      "loss": 0.747,
      "step": 26450
    },
    {
      "epoch": 2.070746595711379,
      "grad_norm": 0.39616528153419495,
      "learning_rate": 1.5529371763351992e-05,
      "loss": 0.737,
      "step": 26460
    },
    {
      "epoch": 2.071529190796682,
      "grad_norm": 2.1874725818634033,
      "learning_rate": 1.5516294397656537e-05,
      "loss": 0.7428,
      "step": 26470
    },
    {
      "epoch": 2.072311785881985,
      "grad_norm": 5.8409295082092285,
      "learning_rate": 1.550321703196108e-05,
      "loss": 0.8347,
      "step": 26480
    },
    {
      "epoch": 2.0730943809672877,
      "grad_norm": 0.9718887805938721,
      "learning_rate": 1.549013966626563e-05,
      "loss": 0.6268,
      "step": 26490
    },
    {
      "epoch": 2.0738769760525906,
      "grad_norm": 19.386890411376953,
      "learning_rate": 1.5477062300570174e-05,
      "loss": 0.7558,
      "step": 26500
    },
    {
      "epoch": 2.0746595711378935,
      "grad_norm": 1.057795524597168,
      "learning_rate": 1.546398493487472e-05,
      "loss": 0.7351,
      "step": 26510
    },
    {
      "epoch": 2.0754421662231963,
      "grad_norm": 9.253228187561035,
      "learning_rate": 1.5450907569179266e-05,
      "loss": 0.8827,
      "step": 26520
    },
    {
      "epoch": 2.0762247613084988,
      "grad_norm": 7.977609634399414,
      "learning_rate": 1.543783020348381e-05,
      "loss": 0.7697,
      "step": 26530
    },
    {
      "epoch": 2.0770073563938016,
      "grad_norm": 7.592144012451172,
      "learning_rate": 1.5424752837788356e-05,
      "loss": 0.9842,
      "step": 26540
    },
    {
      "epoch": 2.0777899514791045,
      "grad_norm": 12.128246307373047,
      "learning_rate": 1.54116754720929e-05,
      "loss": 0.5156,
      "step": 26550
    },
    {
      "epoch": 2.0785725465644074,
      "grad_norm": 10.54001522064209,
      "learning_rate": 1.5398598106397448e-05,
      "loss": 0.8202,
      "step": 26560
    },
    {
      "epoch": 2.0793551416497102,
      "grad_norm": 1.9544382095336914,
      "learning_rate": 1.5385520740701996e-05,
      "loss": 0.8025,
      "step": 26570
    },
    {
      "epoch": 2.080137736735013,
      "grad_norm": 5.8736395835876465,
      "learning_rate": 1.537244337500654e-05,
      "loss": 0.6843,
      "step": 26580
    },
    {
      "epoch": 2.080920331820316,
      "grad_norm": 2.625613212585449,
      "learning_rate": 1.5359366009311085e-05,
      "loss": 0.8121,
      "step": 26590
    },
    {
      "epoch": 2.081702926905619,
      "grad_norm": 10.443289756774902,
      "learning_rate": 1.534628864361563e-05,
      "loss": 0.8222,
      "step": 26600
    },
    {
      "epoch": 2.0824855219909217,
      "grad_norm": 4.960959434509277,
      "learning_rate": 1.5333211277920174e-05,
      "loss": 0.5575,
      "step": 26610
    },
    {
      "epoch": 2.0832681170762246,
      "grad_norm": 24.10548973083496,
      "learning_rate": 1.5320133912224722e-05,
      "loss": 0.806,
      "step": 26620
    },
    {
      "epoch": 2.0840507121615275,
      "grad_norm": 2.607389450073242,
      "learning_rate": 1.530705654652927e-05,
      "loss": 0.4705,
      "step": 26630
    },
    {
      "epoch": 2.0848333072468304,
      "grad_norm": 13.477851867675781,
      "learning_rate": 1.5293979180833815e-05,
      "loss": 0.7803,
      "step": 26640
    },
    {
      "epoch": 2.0856159023321332,
      "grad_norm": 7.330872058868408,
      "learning_rate": 1.528090181513836e-05,
      "loss": 0.8369,
      "step": 26650
    },
    {
      "epoch": 2.086398497417436,
      "grad_norm": 5.793092250823975,
      "learning_rate": 1.5267824449442904e-05,
      "loss": 0.653,
      "step": 26660
    },
    {
      "epoch": 2.087181092502739,
      "grad_norm": 7.394784450531006,
      "learning_rate": 1.525474708374745e-05,
      "loss": 0.493,
      "step": 26670
    },
    {
      "epoch": 2.087963687588042,
      "grad_norm": 1.116746425628662,
      "learning_rate": 1.5241669718051997e-05,
      "loss": 0.3627,
      "step": 26680
    },
    {
      "epoch": 2.0887462826733447,
      "grad_norm": 13.180914878845215,
      "learning_rate": 1.5228592352356541e-05,
      "loss": 0.9117,
      "step": 26690
    },
    {
      "epoch": 2.0895288777586476,
      "grad_norm": 23.096704483032227,
      "learning_rate": 1.5215514986661089e-05,
      "loss": 0.8873,
      "step": 26700
    },
    {
      "epoch": 2.0903114728439505,
      "grad_norm": 11.765103340148926,
      "learning_rate": 1.5202437620965634e-05,
      "loss": 0.8631,
      "step": 26710
    },
    {
      "epoch": 2.0910940679292533,
      "grad_norm": 21.70179557800293,
      "learning_rate": 1.518936025527018e-05,
      "loss": 0.8749,
      "step": 26720
    },
    {
      "epoch": 2.091876663014556,
      "grad_norm": 6.450163841247559,
      "learning_rate": 1.5176282889574725e-05,
      "loss": 0.8915,
      "step": 26730
    },
    {
      "epoch": 2.092659258099859,
      "grad_norm": 3.3924319744110107,
      "learning_rate": 1.516320552387927e-05,
      "loss": 0.6055,
      "step": 26740
    },
    {
      "epoch": 2.093441853185162,
      "grad_norm": 5.62240743637085,
      "learning_rate": 1.5150128158183815e-05,
      "loss": 0.5959,
      "step": 26750
    },
    {
      "epoch": 2.094224448270465,
      "grad_norm": 15.91279411315918,
      "learning_rate": 1.5137050792488362e-05,
      "loss": 0.643,
      "step": 26760
    },
    {
      "epoch": 2.0950070433557677,
      "grad_norm": 12.832975387573242,
      "learning_rate": 1.5123973426792906e-05,
      "loss": 0.5445,
      "step": 26770
    },
    {
      "epoch": 2.0957896384410706,
      "grad_norm": 6.409086227416992,
      "learning_rate": 1.5110896061097454e-05,
      "loss": 0.7828,
      "step": 26780
    },
    {
      "epoch": 2.0965722335263735,
      "grad_norm": 8.790616035461426,
      "learning_rate": 1.5097818695401999e-05,
      "loss": 0.7954,
      "step": 26790
    },
    {
      "epoch": 2.0973548286116763,
      "grad_norm": 6.2133097648620605,
      "learning_rate": 1.5084741329706545e-05,
      "loss": 0.6973,
      "step": 26800
    },
    {
      "epoch": 2.098137423696979,
      "grad_norm": 0.9145981669425964,
      "learning_rate": 1.507166396401109e-05,
      "loss": 0.7008,
      "step": 26810
    },
    {
      "epoch": 2.098920018782282,
      "grad_norm": 9.8286714553833,
      "learning_rate": 1.5058586598315636e-05,
      "loss": 0.7064,
      "step": 26820
    },
    {
      "epoch": 2.099702613867585,
      "grad_norm": 8.839984893798828,
      "learning_rate": 1.504550923262018e-05,
      "loss": 0.8091,
      "step": 26830
    },
    {
      "epoch": 2.100485208952888,
      "grad_norm": 1.6428234577178955,
      "learning_rate": 1.5032431866924727e-05,
      "loss": 1.2528,
      "step": 26840
    },
    {
      "epoch": 2.1012678040381907,
      "grad_norm": 1.1154688596725464,
      "learning_rate": 1.5019354501229275e-05,
      "loss": 0.4879,
      "step": 26850
    },
    {
      "epoch": 2.1020503991234936,
      "grad_norm": 15.73962688446045,
      "learning_rate": 1.500627713553382e-05,
      "loss": 0.8479,
      "step": 26860
    },
    {
      "epoch": 2.1028329942087964,
      "grad_norm": 21.01934814453125,
      "learning_rate": 1.4993199769838365e-05,
      "loss": 0.833,
      "step": 26870
    },
    {
      "epoch": 2.1036155892940993,
      "grad_norm": 7.300323963165283,
      "learning_rate": 1.498012240414291e-05,
      "loss": 0.9513,
      "step": 26880
    },
    {
      "epoch": 2.104398184379402,
      "grad_norm": 0.5027827620506287,
      "learning_rate": 1.4967045038447455e-05,
      "loss": 0.4177,
      "step": 26890
    },
    {
      "epoch": 2.105180779464705,
      "grad_norm": 10.323197364807129,
      "learning_rate": 1.4953967672752001e-05,
      "loss": 0.5828,
      "step": 26900
    },
    {
      "epoch": 2.105963374550008,
      "grad_norm": 3.686628818511963,
      "learning_rate": 1.4940890307056545e-05,
      "loss": 0.8523,
      "step": 26910
    },
    {
      "epoch": 2.106745969635311,
      "grad_norm": 8.017386436462402,
      "learning_rate": 1.4927812941361093e-05,
      "loss": 0.6783,
      "step": 26920
    },
    {
      "epoch": 2.1075285647206137,
      "grad_norm": 11.5304594039917,
      "learning_rate": 1.491473557566564e-05,
      "loss": 0.5479,
      "step": 26930
    },
    {
      "epoch": 2.1083111598059165,
      "grad_norm": 4.189701080322266,
      "learning_rate": 1.4901658209970184e-05,
      "loss": 0.9227,
      "step": 26940
    },
    {
      "epoch": 2.1090937548912194,
      "grad_norm": 17.709531784057617,
      "learning_rate": 1.488858084427473e-05,
      "loss": 0.5455,
      "step": 26950
    },
    {
      "epoch": 2.1098763499765223,
      "grad_norm": 12.155518531799316,
      "learning_rate": 1.4875503478579275e-05,
      "loss": 0.8647,
      "step": 26960
    },
    {
      "epoch": 2.110658945061825,
      "grad_norm": 2.1802234649658203,
      "learning_rate": 1.486242611288382e-05,
      "loss": 0.7881,
      "step": 26970
    },
    {
      "epoch": 2.111441540147128,
      "grad_norm": 8.09160041809082,
      "learning_rate": 1.4849348747188366e-05,
      "loss": 0.8943,
      "step": 26980
    },
    {
      "epoch": 2.112224135232431,
      "grad_norm": 10.030418395996094,
      "learning_rate": 1.4836271381492914e-05,
      "loss": 0.7728,
      "step": 26990
    },
    {
      "epoch": 2.113006730317734,
      "grad_norm": 10.256857872009277,
      "learning_rate": 1.4823194015797458e-05,
      "loss": 0.6604,
      "step": 27000
    },
    {
      "epoch": 2.1137893254030367,
      "grad_norm": 10.670951843261719,
      "learning_rate": 1.4810116650102005e-05,
      "loss": 0.6534,
      "step": 27010
    },
    {
      "epoch": 2.1145719204883395,
      "grad_norm": 0.9598550796508789,
      "learning_rate": 1.479703928440655e-05,
      "loss": 0.5359,
      "step": 27020
    },
    {
      "epoch": 2.1153545155736424,
      "grad_norm": 48.854835510253906,
      "learning_rate": 1.4783961918711096e-05,
      "loss": 0.9861,
      "step": 27030
    },
    {
      "epoch": 2.1161371106589453,
      "grad_norm": 2.4441351890563965,
      "learning_rate": 1.477088455301564e-05,
      "loss": 0.4124,
      "step": 27040
    },
    {
      "epoch": 2.116919705744248,
      "grad_norm": 18.872650146484375,
      "learning_rate": 1.4757807187320186e-05,
      "loss": 0.5386,
      "step": 27050
    },
    {
      "epoch": 2.1177023008295506,
      "grad_norm": 1.989243745803833,
      "learning_rate": 1.4744729821624734e-05,
      "loss": 0.6133,
      "step": 27060
    },
    {
      "epoch": 2.1184848959148534,
      "grad_norm": 6.247564792633057,
      "learning_rate": 1.4731652455929279e-05,
      "loss": 0.7865,
      "step": 27070
    },
    {
      "epoch": 2.1192674910001563,
      "grad_norm": 3.2262465953826904,
      "learning_rate": 1.4718575090233823e-05,
      "loss": 0.8231,
      "step": 27080
    },
    {
      "epoch": 2.120050086085459,
      "grad_norm": 1.8706382513046265,
      "learning_rate": 1.470549772453837e-05,
      "loss": 0.6964,
      "step": 27090
    },
    {
      "epoch": 2.120832681170762,
      "grad_norm": 4.315087795257568,
      "learning_rate": 1.4692420358842914e-05,
      "loss": 0.4562,
      "step": 27100
    },
    {
      "epoch": 2.121615276256065,
      "grad_norm": 18.745153427124023,
      "learning_rate": 1.467934299314746e-05,
      "loss": 0.8073,
      "step": 27110
    },
    {
      "epoch": 2.122397871341368,
      "grad_norm": 1.823157548904419,
      "learning_rate": 1.4666265627452005e-05,
      "loss": 0.4448,
      "step": 27120
    },
    {
      "epoch": 2.1231804664266707,
      "grad_norm": 0.7687124013900757,
      "learning_rate": 1.4653188261756553e-05,
      "loss": 0.8272,
      "step": 27130
    },
    {
      "epoch": 2.1239630615119736,
      "grad_norm": 10.660720825195312,
      "learning_rate": 1.46401108960611e-05,
      "loss": 1.0299,
      "step": 27140
    },
    {
      "epoch": 2.1247456565972764,
      "grad_norm": 34.86052703857422,
      "learning_rate": 1.4627033530365644e-05,
      "loss": 0.7472,
      "step": 27150
    },
    {
      "epoch": 2.1255282516825793,
      "grad_norm": 20.17201805114746,
      "learning_rate": 1.461395616467019e-05,
      "loss": 0.7276,
      "step": 27160
    },
    {
      "epoch": 2.126310846767882,
      "grad_norm": 3.4193267822265625,
      "learning_rate": 1.4600878798974735e-05,
      "loss": 0.7678,
      "step": 27170
    },
    {
      "epoch": 2.127093441853185,
      "grad_norm": 9.641719818115234,
      "learning_rate": 1.458780143327928e-05,
      "loss": 0.839,
      "step": 27180
    },
    {
      "epoch": 2.127876036938488,
      "grad_norm": 57.91522216796875,
      "learning_rate": 1.4574724067583826e-05,
      "loss": 0.7528,
      "step": 27190
    },
    {
      "epoch": 2.128658632023791,
      "grad_norm": 6.700779438018799,
      "learning_rate": 1.4561646701888374e-05,
      "loss": 0.7001,
      "step": 27200
    },
    {
      "epoch": 2.1294412271090937,
      "grad_norm": 6.451566219329834,
      "learning_rate": 1.4548569336192918e-05,
      "loss": 0.9892,
      "step": 27210
    },
    {
      "epoch": 2.1302238221943965,
      "grad_norm": 12.978513717651367,
      "learning_rate": 1.4535491970497464e-05,
      "loss": 0.8291,
      "step": 27220
    },
    {
      "epoch": 2.1310064172796994,
      "grad_norm": 4.730929851531982,
      "learning_rate": 1.4522414604802009e-05,
      "loss": 0.7982,
      "step": 27230
    },
    {
      "epoch": 2.1317890123650023,
      "grad_norm": 6.227127552032471,
      "learning_rate": 1.4509337239106555e-05,
      "loss": 0.8324,
      "step": 27240
    },
    {
      "epoch": 2.132571607450305,
      "grad_norm": 18.510761260986328,
      "learning_rate": 1.44962598734111e-05,
      "loss": 0.595,
      "step": 27250
    },
    {
      "epoch": 2.133354202535608,
      "grad_norm": 6.648828029632568,
      "learning_rate": 1.4483182507715644e-05,
      "loss": 0.4764,
      "step": 27260
    },
    {
      "epoch": 2.134136797620911,
      "grad_norm": 1.582897424697876,
      "learning_rate": 1.4470105142020194e-05,
      "loss": 0.6135,
      "step": 27270
    },
    {
      "epoch": 2.134919392706214,
      "grad_norm": 1.9581129550933838,
      "learning_rate": 1.4457027776324739e-05,
      "loss": 0.4252,
      "step": 27280
    },
    {
      "epoch": 2.1357019877915167,
      "grad_norm": 1.5512436628341675,
      "learning_rate": 1.4443950410629283e-05,
      "loss": 0.4044,
      "step": 27290
    },
    {
      "epoch": 2.1364845828768195,
      "grad_norm": 6.79715633392334,
      "learning_rate": 1.443087304493383e-05,
      "loss": 0.7898,
      "step": 27300
    },
    {
      "epoch": 2.1372671779621224,
      "grad_norm": 56.16928482055664,
      "learning_rate": 1.4417795679238374e-05,
      "loss": 0.4847,
      "step": 27310
    },
    {
      "epoch": 2.1380497730474253,
      "grad_norm": 12.65277099609375,
      "learning_rate": 1.440471831354292e-05,
      "loss": 1.1532,
      "step": 27320
    },
    {
      "epoch": 2.138832368132728,
      "grad_norm": 2.201775312423706,
      "learning_rate": 1.4391640947847465e-05,
      "loss": 0.6333,
      "step": 27330
    },
    {
      "epoch": 2.139614963218031,
      "grad_norm": 0.4568299651145935,
      "learning_rate": 1.4378563582152013e-05,
      "loss": 0.838,
      "step": 27340
    },
    {
      "epoch": 2.140397558303334,
      "grad_norm": 13.891705513000488,
      "learning_rate": 1.4365486216456559e-05,
      "loss": 0.7574,
      "step": 27350
    },
    {
      "epoch": 2.1411801533886368,
      "grad_norm": 9.814606666564941,
      "learning_rate": 1.4352408850761104e-05,
      "loss": 0.5728,
      "step": 27360
    },
    {
      "epoch": 2.1419627484739396,
      "grad_norm": 14.384461402893066,
      "learning_rate": 1.4339331485065648e-05,
      "loss": 0.4636,
      "step": 27370
    },
    {
      "epoch": 2.1427453435592425,
      "grad_norm": 4.691298484802246,
      "learning_rate": 1.4326254119370195e-05,
      "loss": 0.8101,
      "step": 27380
    },
    {
      "epoch": 2.1435279386445454,
      "grad_norm": 45.819942474365234,
      "learning_rate": 1.4313176753674739e-05,
      "loss": 0.8358,
      "step": 27390
    },
    {
      "epoch": 2.1443105337298483,
      "grad_norm": 4.602461814880371,
      "learning_rate": 1.4300099387979285e-05,
      "loss": 0.8796,
      "step": 27400
    },
    {
      "epoch": 2.145093128815151,
      "grad_norm": 9.980877876281738,
      "learning_rate": 1.4287022022283833e-05,
      "loss": 0.919,
      "step": 27410
    },
    {
      "epoch": 2.145875723900454,
      "grad_norm": 10.89567756652832,
      "learning_rate": 1.4273944656588378e-05,
      "loss": 0.826,
      "step": 27420
    },
    {
      "epoch": 2.146658318985757,
      "grad_norm": 5.545185089111328,
      "learning_rate": 1.4260867290892924e-05,
      "loss": 0.7742,
      "step": 27430
    },
    {
      "epoch": 2.1474409140710597,
      "grad_norm": 5.855535984039307,
      "learning_rate": 1.4247789925197469e-05,
      "loss": 0.6041,
      "step": 27440
    },
    {
      "epoch": 2.1482235091563626,
      "grad_norm": 9.502318382263184,
      "learning_rate": 1.4234712559502015e-05,
      "loss": 0.7063,
      "step": 27450
    },
    {
      "epoch": 2.1490061042416655,
      "grad_norm": 4.544643402099609,
      "learning_rate": 1.422163519380656e-05,
      "loss": 0.8586,
      "step": 27460
    },
    {
      "epoch": 2.1497886993269684,
      "grad_norm": 12.757966041564941,
      "learning_rate": 1.4208557828111104e-05,
      "loss": 0.7847,
      "step": 27470
    },
    {
      "epoch": 2.1505712944122712,
      "grad_norm": 21.283077239990234,
      "learning_rate": 1.4195480462415652e-05,
      "loss": 1.0304,
      "step": 27480
    },
    {
      "epoch": 2.151353889497574,
      "grad_norm": 4.786354064941406,
      "learning_rate": 1.4182403096720198e-05,
      "loss": 0.5824,
      "step": 27490
    },
    {
      "epoch": 2.152136484582877,
      "grad_norm": 6.804817199707031,
      "learning_rate": 1.4169325731024743e-05,
      "loss": 0.7939,
      "step": 27500
    },
    {
      "epoch": 2.15291907966818,
      "grad_norm": 5.134428977966309,
      "learning_rate": 1.415624836532929e-05,
      "loss": 0.5898,
      "step": 27510
    },
    {
      "epoch": 2.1537016747534827,
      "grad_norm": 8.174689292907715,
      "learning_rate": 1.4143170999633834e-05,
      "loss": 0.6385,
      "step": 27520
    },
    {
      "epoch": 2.1544842698387856,
      "grad_norm": 12.405238151550293,
      "learning_rate": 1.413009363393838e-05,
      "loss": 0.9141,
      "step": 27530
    },
    {
      "epoch": 2.1552668649240885,
      "grad_norm": 1.4031028747558594,
      "learning_rate": 1.4117016268242925e-05,
      "loss": 0.7763,
      "step": 27540
    },
    {
      "epoch": 2.156049460009391,
      "grad_norm": 13.917854309082031,
      "learning_rate": 1.4103938902547473e-05,
      "loss": 0.947,
      "step": 27550
    },
    {
      "epoch": 2.156832055094694,
      "grad_norm": 33.147071838378906,
      "learning_rate": 1.4090861536852019e-05,
      "loss": 0.5056,
      "step": 27560
    },
    {
      "epoch": 2.1576146501799967,
      "grad_norm": 1.9224683046340942,
      "learning_rate": 1.4077784171156563e-05,
      "loss": 0.5928,
      "step": 27570
    },
    {
      "epoch": 2.1583972452652995,
      "grad_norm": 3.3488619327545166,
      "learning_rate": 1.4064706805461108e-05,
      "loss": 0.4068,
      "step": 27580
    },
    {
      "epoch": 2.1591798403506024,
      "grad_norm": 15.014608383178711,
      "learning_rate": 1.4051629439765654e-05,
      "loss": 0.5554,
      "step": 27590
    },
    {
      "epoch": 2.1599624354359053,
      "grad_norm": 10.953495025634766,
      "learning_rate": 1.4038552074070199e-05,
      "loss": 0.9059,
      "step": 27600
    },
    {
      "epoch": 2.160745030521208,
      "grad_norm": 10.364972114562988,
      "learning_rate": 1.4025474708374745e-05,
      "loss": 0.3889,
      "step": 27610
    },
    {
      "epoch": 2.161527625606511,
      "grad_norm": 7.3353800773620605,
      "learning_rate": 1.4012397342679293e-05,
      "loss": 0.6752,
      "step": 27620
    },
    {
      "epoch": 2.162310220691814,
      "grad_norm": 16.542028427124023,
      "learning_rate": 1.3999319976983838e-05,
      "loss": 1.4518,
      "step": 27630
    },
    {
      "epoch": 2.1630928157771168,
      "grad_norm": 9.876993179321289,
      "learning_rate": 1.3986242611288384e-05,
      "loss": 0.8557,
      "step": 27640
    },
    {
      "epoch": 2.1638754108624196,
      "grad_norm": 0.8241252899169922,
      "learning_rate": 1.3973165245592928e-05,
      "loss": 0.5438,
      "step": 27650
    },
    {
      "epoch": 2.1646580059477225,
      "grad_norm": 15.983797073364258,
      "learning_rate": 1.3960087879897473e-05,
      "loss": 0.6478,
      "step": 27660
    },
    {
      "epoch": 2.1654406010330254,
      "grad_norm": 14.6947021484375,
      "learning_rate": 1.394701051420202e-05,
      "loss": 0.7778,
      "step": 27670
    },
    {
      "epoch": 2.1662231961183283,
      "grad_norm": 0.45581933856010437,
      "learning_rate": 1.3933933148506564e-05,
      "loss": 0.3981,
      "step": 27680
    },
    {
      "epoch": 2.167005791203631,
      "grad_norm": 4.059055805206299,
      "learning_rate": 1.3920855782811112e-05,
      "loss": 0.5909,
      "step": 27690
    },
    {
      "epoch": 2.167788386288934,
      "grad_norm": 11.105546951293945,
      "learning_rate": 1.3907778417115658e-05,
      "loss": 1.1921,
      "step": 27700
    },
    {
      "epoch": 2.168570981374237,
      "grad_norm": 10.329076766967773,
      "learning_rate": 1.3894701051420203e-05,
      "loss": 0.7478,
      "step": 27710
    },
    {
      "epoch": 2.1693535764595397,
      "grad_norm": 13.252687454223633,
      "learning_rate": 1.3881623685724749e-05,
      "loss": 0.6695,
      "step": 27720
    },
    {
      "epoch": 2.1701361715448426,
      "grad_norm": 2.881356954574585,
      "learning_rate": 1.3868546320029294e-05,
      "loss": 0.7294,
      "step": 27730
    },
    {
      "epoch": 2.1709187666301455,
      "grad_norm": 10.337958335876465,
      "learning_rate": 1.3855468954333838e-05,
      "loss": 0.7195,
      "step": 27740
    },
    {
      "epoch": 2.1717013617154484,
      "grad_norm": 1.2436773777008057,
      "learning_rate": 1.3842391588638384e-05,
      "loss": 0.4209,
      "step": 27750
    },
    {
      "epoch": 2.1724839568007512,
      "grad_norm": 3.0573532581329346,
      "learning_rate": 1.3829314222942932e-05,
      "loss": 0.6648,
      "step": 27760
    },
    {
      "epoch": 2.173266551886054,
      "grad_norm": 6.896702766418457,
      "learning_rate": 1.3816236857247477e-05,
      "loss": 0.6936,
      "step": 27770
    },
    {
      "epoch": 2.174049146971357,
      "grad_norm": 8.274828910827637,
      "learning_rate": 1.3803159491552023e-05,
      "loss": 1.125,
      "step": 27780
    },
    {
      "epoch": 2.17483174205666,
      "grad_norm": 16.464719772338867,
      "learning_rate": 1.3790082125856568e-05,
      "loss": 0.42,
      "step": 27790
    },
    {
      "epoch": 2.1756143371419627,
      "grad_norm": 11.58179759979248,
      "learning_rate": 1.3777004760161114e-05,
      "loss": 0.6645,
      "step": 27800
    },
    {
      "epoch": 2.1763969322272656,
      "grad_norm": 4.694297790527344,
      "learning_rate": 1.3763927394465659e-05,
      "loss": 0.7221,
      "step": 27810
    },
    {
      "epoch": 2.1771795273125685,
      "grad_norm": 2.252152442932129,
      "learning_rate": 1.3750850028770205e-05,
      "loss": 0.7026,
      "step": 27820
    },
    {
      "epoch": 2.1779621223978713,
      "grad_norm": 6.135369300842285,
      "learning_rate": 1.3737772663074753e-05,
      "loss": 1.1494,
      "step": 27830
    },
    {
      "epoch": 2.178744717483174,
      "grad_norm": 10.136958122253418,
      "learning_rate": 1.3724695297379297e-05,
      "loss": 0.7846,
      "step": 27840
    },
    {
      "epoch": 2.179527312568477,
      "grad_norm": 7.667915344238281,
      "learning_rate": 1.3711617931683842e-05,
      "loss": 0.6936,
      "step": 27850
    },
    {
      "epoch": 2.18030990765378,
      "grad_norm": 16.039321899414062,
      "learning_rate": 1.3698540565988388e-05,
      "loss": 0.4818,
      "step": 27860
    },
    {
      "epoch": 2.181092502739083,
      "grad_norm": 1.710989236831665,
      "learning_rate": 1.3685463200292933e-05,
      "loss": 0.5458,
      "step": 27870
    },
    {
      "epoch": 2.1818750978243857,
      "grad_norm": 9.833703994750977,
      "learning_rate": 1.3672385834597479e-05,
      "loss": 0.8441,
      "step": 27880
    },
    {
      "epoch": 2.1826576929096886,
      "grad_norm": 31.291494369506836,
      "learning_rate": 1.3659308468902024e-05,
      "loss": 0.9514,
      "step": 27890
    },
    {
      "epoch": 2.1834402879949915,
      "grad_norm": 15.828462600708008,
      "learning_rate": 1.3646231103206572e-05,
      "loss": 0.7344,
      "step": 27900
    },
    {
      "epoch": 2.1842228830802943,
      "grad_norm": 2.6696314811706543,
      "learning_rate": 1.3633153737511118e-05,
      "loss": 0.4494,
      "step": 27910
    },
    {
      "epoch": 2.185005478165597,
      "grad_norm": 1.071773886680603,
      "learning_rate": 1.3620076371815662e-05,
      "loss": 0.6727,
      "step": 27920
    },
    {
      "epoch": 2.1857880732509,
      "grad_norm": 10.682232856750488,
      "learning_rate": 1.3606999006120209e-05,
      "loss": 0.6081,
      "step": 27930
    },
    {
      "epoch": 2.186570668336203,
      "grad_norm": 15.563626289367676,
      "learning_rate": 1.3593921640424753e-05,
      "loss": 0.811,
      "step": 27940
    },
    {
      "epoch": 2.187353263421506,
      "grad_norm": 19.947080612182617,
      "learning_rate": 1.3580844274729298e-05,
      "loss": 0.501,
      "step": 27950
    },
    {
      "epoch": 2.1881358585068087,
      "grad_norm": 18.10178565979004,
      "learning_rate": 1.3567766909033844e-05,
      "loss": 0.5861,
      "step": 27960
    },
    {
      "epoch": 2.1889184535921116,
      "grad_norm": 10.642863273620605,
      "learning_rate": 1.3554689543338392e-05,
      "loss": 0.6568,
      "step": 27970
    },
    {
      "epoch": 2.1897010486774144,
      "grad_norm": 17.06035804748535,
      "learning_rate": 1.3541612177642937e-05,
      "loss": 0.8942,
      "step": 27980
    },
    {
      "epoch": 2.1904836437627173,
      "grad_norm": 1.7717318534851074,
      "learning_rate": 1.3528534811947483e-05,
      "loss": 0.5667,
      "step": 27990
    },
    {
      "epoch": 2.19126623884802,
      "grad_norm": 8.330546379089355,
      "learning_rate": 1.3515457446252027e-05,
      "loss": 0.5555,
      "step": 28000
    },
    {
      "epoch": 2.192048833933323,
      "grad_norm": 7.759912014007568,
      "learning_rate": 1.3502380080556574e-05,
      "loss": 0.8189,
      "step": 28010
    },
    {
      "epoch": 2.192831429018626,
      "grad_norm": 8.03850269317627,
      "learning_rate": 1.3489302714861118e-05,
      "loss": 0.8633,
      "step": 28020
    },
    {
      "epoch": 2.193614024103929,
      "grad_norm": 11.419326782226562,
      "learning_rate": 1.3476225349165663e-05,
      "loss": 0.6952,
      "step": 28030
    },
    {
      "epoch": 2.1943966191892317,
      "grad_norm": 9.798736572265625,
      "learning_rate": 1.3463147983470213e-05,
      "loss": 1.1603,
      "step": 28040
    },
    {
      "epoch": 2.1951792142745346,
      "grad_norm": 3.6351473331451416,
      "learning_rate": 1.3450070617774757e-05,
      "loss": 0.4789,
      "step": 28050
    },
    {
      "epoch": 2.1959618093598374,
      "grad_norm": 12.652070999145508,
      "learning_rate": 1.3436993252079302e-05,
      "loss": 0.8379,
      "step": 28060
    },
    {
      "epoch": 2.1967444044451403,
      "grad_norm": 11.94027328491211,
      "learning_rate": 1.3423915886383848e-05,
      "loss": 1.3018,
      "step": 28070
    },
    {
      "epoch": 2.197526999530443,
      "grad_norm": 21.78432846069336,
      "learning_rate": 1.3410838520688393e-05,
      "loss": 0.5549,
      "step": 28080
    },
    {
      "epoch": 2.198309594615746,
      "grad_norm": 15.984980583190918,
      "learning_rate": 1.3397761154992939e-05,
      "loss": 0.6468,
      "step": 28090
    },
    {
      "epoch": 2.1990921897010485,
      "grad_norm": 7.493800640106201,
      "learning_rate": 1.3384683789297483e-05,
      "loss": 0.7229,
      "step": 28100
    },
    {
      "epoch": 2.1998747847863513,
      "grad_norm": 8.410491943359375,
      "learning_rate": 1.337160642360203e-05,
      "loss": 0.4159,
      "step": 28110
    },
    {
      "epoch": 2.200657379871654,
      "grad_norm": 37.21291732788086,
      "learning_rate": 1.3358529057906578e-05,
      "loss": 0.7881,
      "step": 28120
    },
    {
      "epoch": 2.201439974956957,
      "grad_norm": 25.852121353149414,
      "learning_rate": 1.3345451692211122e-05,
      "loss": 0.7701,
      "step": 28130
    },
    {
      "epoch": 2.20222257004226,
      "grad_norm": 0.9387451410293579,
      "learning_rate": 1.3332374326515667e-05,
      "loss": 0.4224,
      "step": 28140
    },
    {
      "epoch": 2.203005165127563,
      "grad_norm": 7.491085052490234,
      "learning_rate": 1.3319296960820213e-05,
      "loss": 0.8089,
      "step": 28150
    },
    {
      "epoch": 2.2037877602128657,
      "grad_norm": 1.5017385482788086,
      "learning_rate": 1.3306219595124758e-05,
      "loss": 0.5019,
      "step": 28160
    },
    {
      "epoch": 2.2045703552981686,
      "grad_norm": 11.055264472961426,
      "learning_rate": 1.3293142229429304e-05,
      "loss": 0.9963,
      "step": 28170
    },
    {
      "epoch": 2.2053529503834715,
      "grad_norm": 16.322643280029297,
      "learning_rate": 1.3280064863733848e-05,
      "loss": 0.9624,
      "step": 28180
    },
    {
      "epoch": 2.2061355454687743,
      "grad_norm": 3.495830535888672,
      "learning_rate": 1.3266987498038396e-05,
      "loss": 0.7294,
      "step": 28190
    },
    {
      "epoch": 2.206918140554077,
      "grad_norm": 19.248231887817383,
      "learning_rate": 1.3253910132342943e-05,
      "loss": 0.5328,
      "step": 28200
    },
    {
      "epoch": 2.20770073563938,
      "grad_norm": 3.865118980407715,
      "learning_rate": 1.3240832766647487e-05,
      "loss": 0.6587,
      "step": 28210
    },
    {
      "epoch": 2.208483330724683,
      "grad_norm": 39.83748245239258,
      "learning_rate": 1.3227755400952033e-05,
      "loss": 1.1805,
      "step": 28220
    },
    {
      "epoch": 2.209265925809986,
      "grad_norm": 3.2664458751678467,
      "learning_rate": 1.3214678035256578e-05,
      "loss": 0.5933,
      "step": 28230
    },
    {
      "epoch": 2.2100485208952887,
      "grad_norm": 8.842866897583008,
      "learning_rate": 1.3201600669561123e-05,
      "loss": 0.9244,
      "step": 28240
    },
    {
      "epoch": 2.2108311159805916,
      "grad_norm": 2.328725576400757,
      "learning_rate": 1.3188523303865669e-05,
      "loss": 0.3924,
      "step": 28250
    },
    {
      "epoch": 2.2116137110658944,
      "grad_norm": 6.246313571929932,
      "learning_rate": 1.3175445938170217e-05,
      "loss": 0.9241,
      "step": 28260
    },
    {
      "epoch": 2.2123963061511973,
      "grad_norm": 4.482760429382324,
      "learning_rate": 1.3162368572474761e-05,
      "loss": 0.5366,
      "step": 28270
    },
    {
      "epoch": 2.2131789012365,
      "grad_norm": 3.89351749420166,
      "learning_rate": 1.3149291206779308e-05,
      "loss": 0.7913,
      "step": 28280
    },
    {
      "epoch": 2.213961496321803,
      "grad_norm": 5.157376766204834,
      "learning_rate": 1.3136213841083852e-05,
      "loss": 0.6597,
      "step": 28290
    },
    {
      "epoch": 2.214744091407106,
      "grad_norm": 23.566638946533203,
      "learning_rate": 1.3123136475388399e-05,
      "loss": 0.7061,
      "step": 28300
    },
    {
      "epoch": 2.215526686492409,
      "grad_norm": 13.590498924255371,
      "learning_rate": 1.3110059109692943e-05,
      "loss": 0.7491,
      "step": 28310
    },
    {
      "epoch": 2.2163092815777117,
      "grad_norm": 8.758382797241211,
      "learning_rate": 1.3096981743997488e-05,
      "loss": 0.6387,
      "step": 28320
    },
    {
      "epoch": 2.2170918766630145,
      "grad_norm": 1.766776442527771,
      "learning_rate": 1.3083904378302037e-05,
      "loss": 0.8073,
      "step": 28330
    },
    {
      "epoch": 2.2178744717483174,
      "grad_norm": 10.18277645111084,
      "learning_rate": 1.3070827012606582e-05,
      "loss": 0.7991,
      "step": 28340
    },
    {
      "epoch": 2.2186570668336203,
      "grad_norm": 3.40946626663208,
      "learning_rate": 1.3057749646911126e-05,
      "loss": 0.5539,
      "step": 28350
    },
    {
      "epoch": 2.219439661918923,
      "grad_norm": 1.1049573421478271,
      "learning_rate": 1.3044672281215673e-05,
      "loss": 0.9004,
      "step": 28360
    },
    {
      "epoch": 2.220222257004226,
      "grad_norm": 6.324769973754883,
      "learning_rate": 1.3031594915520217e-05,
      "loss": 0.5644,
      "step": 28370
    },
    {
      "epoch": 2.221004852089529,
      "grad_norm": 10.904841423034668,
      "learning_rate": 1.3018517549824764e-05,
      "loss": 0.6594,
      "step": 28380
    },
    {
      "epoch": 2.221787447174832,
      "grad_norm": 17.154254913330078,
      "learning_rate": 1.3005440184129308e-05,
      "loss": 0.9462,
      "step": 28390
    },
    {
      "epoch": 2.2225700422601347,
      "grad_norm": 6.291401386260986,
      "learning_rate": 1.2992362818433856e-05,
      "loss": 0.6154,
      "step": 28400
    },
    {
      "epoch": 2.2233526373454375,
      "grad_norm": 4.2689008712768555,
      "learning_rate": 1.2979285452738402e-05,
      "loss": 0.621,
      "step": 28410
    },
    {
      "epoch": 2.2241352324307404,
      "grad_norm": 3.8880388736724854,
      "learning_rate": 1.2966208087042947e-05,
      "loss": 0.5587,
      "step": 28420
    },
    {
      "epoch": 2.2249178275160433,
      "grad_norm": 16.027786254882812,
      "learning_rate": 1.2953130721347492e-05,
      "loss": 0.5,
      "step": 28430
    },
    {
      "epoch": 2.225700422601346,
      "grad_norm": 1.183531403541565,
      "learning_rate": 1.2940053355652038e-05,
      "loss": 0.6305,
      "step": 28440
    },
    {
      "epoch": 2.226483017686649,
      "grad_norm": 11.637067794799805,
      "learning_rate": 1.2926975989956582e-05,
      "loss": 0.6538,
      "step": 28450
    },
    {
      "epoch": 2.227265612771952,
      "grad_norm": 2.6959407329559326,
      "learning_rate": 1.2913898624261129e-05,
      "loss": 0.9201,
      "step": 28460
    },
    {
      "epoch": 2.2280482078572548,
      "grad_norm": 0.760138988494873,
      "learning_rate": 1.2900821258565677e-05,
      "loss": 0.7629,
      "step": 28470
    },
    {
      "epoch": 2.2288308029425576,
      "grad_norm": 9.873852729797363,
      "learning_rate": 1.2887743892870221e-05,
      "loss": 0.9993,
      "step": 28480
    },
    {
      "epoch": 2.2296133980278605,
      "grad_norm": 2.4817087650299072,
      "learning_rate": 1.2874666527174767e-05,
      "loss": 0.7541,
      "step": 28490
    },
    {
      "epoch": 2.2303959931131634,
      "grad_norm": 7.417629241943359,
      "learning_rate": 1.2861589161479312e-05,
      "loss": 0.49,
      "step": 28500
    },
    {
      "epoch": 2.2311785881984663,
      "grad_norm": 8.362085342407227,
      "learning_rate": 1.2848511795783858e-05,
      "loss": 0.763,
      "step": 28510
    },
    {
      "epoch": 2.231961183283769,
      "grad_norm": 3.292586326599121,
      "learning_rate": 1.2835434430088403e-05,
      "loss": 0.7591,
      "step": 28520
    },
    {
      "epoch": 2.232743778369072,
      "grad_norm": 3.947648286819458,
      "learning_rate": 1.2822357064392947e-05,
      "loss": 0.6714,
      "step": 28530
    },
    {
      "epoch": 2.233526373454375,
      "grad_norm": 1.1809580326080322,
      "learning_rate": 1.2809279698697495e-05,
      "loss": 0.5226,
      "step": 28540
    },
    {
      "epoch": 2.2343089685396778,
      "grad_norm": 14.341992378234863,
      "learning_rate": 1.2796202333002042e-05,
      "loss": 0.881,
      "step": 28550
    },
    {
      "epoch": 2.2350915636249806,
      "grad_norm": 10.657034873962402,
      "learning_rate": 1.2783124967306586e-05,
      "loss": 0.7525,
      "step": 28560
    },
    {
      "epoch": 2.2358741587102835,
      "grad_norm": 10.231977462768555,
      "learning_rate": 1.2770047601611132e-05,
      "loss": 0.5333,
      "step": 28570
    },
    {
      "epoch": 2.236656753795586,
      "grad_norm": 3.1532201766967773,
      "learning_rate": 1.2756970235915677e-05,
      "loss": 0.4861,
      "step": 28580
    },
    {
      "epoch": 2.237439348880889,
      "grad_norm": 2.62473464012146,
      "learning_rate": 1.2743892870220223e-05,
      "loss": 0.4585,
      "step": 28590
    },
    {
      "epoch": 2.2382219439661917,
      "grad_norm": 15.046574592590332,
      "learning_rate": 1.2730815504524768e-05,
      "loss": 0.7546,
      "step": 28600
    },
    {
      "epoch": 2.2390045390514945,
      "grad_norm": 7.216394424438477,
      "learning_rate": 1.2717738138829316e-05,
      "loss": 0.6762,
      "step": 28610
    },
    {
      "epoch": 2.2397871341367974,
      "grad_norm": 1.6292506456375122,
      "learning_rate": 1.2704660773133862e-05,
      "loss": 0.5861,
      "step": 28620
    },
    {
      "epoch": 2.2405697292221003,
      "grad_norm": 4.089292049407959,
      "learning_rate": 1.2691583407438407e-05,
      "loss": 0.6322,
      "step": 28630
    },
    {
      "epoch": 2.241352324307403,
      "grad_norm": 7.410955905914307,
      "learning_rate": 1.2678506041742951e-05,
      "loss": 0.7916,
      "step": 28640
    },
    {
      "epoch": 2.242134919392706,
      "grad_norm": 5.809022426605225,
      "learning_rate": 1.2665428676047498e-05,
      "loss": 0.9225,
      "step": 28650
    },
    {
      "epoch": 2.242917514478009,
      "grad_norm": 6.024724960327148,
      "learning_rate": 1.2652351310352042e-05,
      "loss": 0.4766,
      "step": 28660
    },
    {
      "epoch": 2.243700109563312,
      "grad_norm": 8.712727546691895,
      "learning_rate": 1.2639273944656588e-05,
      "loss": 0.8524,
      "step": 28670
    },
    {
      "epoch": 2.2444827046486147,
      "grad_norm": 9.616090774536133,
      "learning_rate": 1.2626196578961136e-05,
      "loss": 0.7285,
      "step": 28680
    },
    {
      "epoch": 2.2452652997339175,
      "grad_norm": 2.5103859901428223,
      "learning_rate": 1.2613119213265681e-05,
      "loss": 0.5482,
      "step": 28690
    },
    {
      "epoch": 2.2460478948192204,
      "grad_norm": 24.364755630493164,
      "learning_rate": 1.2600041847570227e-05,
      "loss": 0.7384,
      "step": 28700
    },
    {
      "epoch": 2.2468304899045233,
      "grad_norm": 15.924491882324219,
      "learning_rate": 1.2586964481874772e-05,
      "loss": 0.939,
      "step": 28710
    },
    {
      "epoch": 2.247613084989826,
      "grad_norm": 7.254472732543945,
      "learning_rate": 1.2573887116179316e-05,
      "loss": 0.8357,
      "step": 28720
    },
    {
      "epoch": 2.248395680075129,
      "grad_norm": 15.541780471801758,
      "learning_rate": 1.2560809750483863e-05,
      "loss": 0.6778,
      "step": 28730
    },
    {
      "epoch": 2.249178275160432,
      "grad_norm": 0.4969427287578583,
      "learning_rate": 1.2547732384788407e-05,
      "loss": 0.7173,
      "step": 28740
    },
    {
      "epoch": 2.2499608702457348,
      "grad_norm": 1.04001784324646,
      "learning_rate": 1.2534655019092955e-05,
      "loss": 0.4297,
      "step": 28750
    },
    {
      "epoch": 2.2507434653310376,
      "grad_norm": 1.132002353668213,
      "learning_rate": 1.2521577653397501e-05,
      "loss": 0.6981,
      "step": 28760
    },
    {
      "epoch": 2.2515260604163405,
      "grad_norm": 4.089123249053955,
      "learning_rate": 1.2508500287702046e-05,
      "loss": 0.612,
      "step": 28770
    },
    {
      "epoch": 2.2523086555016434,
      "grad_norm": 7.9485554695129395,
      "learning_rate": 1.2495422922006592e-05,
      "loss": 0.4132,
      "step": 28780
    },
    {
      "epoch": 2.2530912505869463,
      "grad_norm": 5.991998195648193,
      "learning_rate": 1.2482345556311137e-05,
      "loss": 1.1367,
      "step": 28790
    },
    {
      "epoch": 2.253873845672249,
      "grad_norm": 11.455936431884766,
      "learning_rate": 1.2469268190615683e-05,
      "loss": 1.16,
      "step": 28800
    },
    {
      "epoch": 2.254656440757552,
      "grad_norm": 22.021892547607422,
      "learning_rate": 1.245619082492023e-05,
      "loss": 0.5046,
      "step": 28810
    },
    {
      "epoch": 2.255439035842855,
      "grad_norm": 5.294410705566406,
      "learning_rate": 1.2443113459224774e-05,
      "loss": 0.7633,
      "step": 28820
    },
    {
      "epoch": 2.2562216309281578,
      "grad_norm": 4.921542167663574,
      "learning_rate": 1.243003609352932e-05,
      "loss": 0.8762,
      "step": 28830
    },
    {
      "epoch": 2.2570042260134606,
      "grad_norm": 1.0527578592300415,
      "learning_rate": 1.2416958727833866e-05,
      "loss": 0.5374,
      "step": 28840
    },
    {
      "epoch": 2.2577868210987635,
      "grad_norm": 16.926298141479492,
      "learning_rate": 1.2403881362138411e-05,
      "loss": 0.611,
      "step": 28850
    },
    {
      "epoch": 2.2585694161840664,
      "grad_norm": 14.423622131347656,
      "learning_rate": 1.2390803996442957e-05,
      "loss": 0.9833,
      "step": 28860
    },
    {
      "epoch": 2.2593520112693692,
      "grad_norm": 27.036569595336914,
      "learning_rate": 1.2377726630747504e-05,
      "loss": 0.9971,
      "step": 28870
    },
    {
      "epoch": 2.260134606354672,
      "grad_norm": 10.269298553466797,
      "learning_rate": 1.2364649265052048e-05,
      "loss": 0.9267,
      "step": 28880
    },
    {
      "epoch": 2.260917201439975,
      "grad_norm": 7.765970706939697,
      "learning_rate": 1.2351571899356594e-05,
      "loss": 0.5896,
      "step": 28890
    },
    {
      "epoch": 2.261699796525278,
      "grad_norm": 6.190877437591553,
      "learning_rate": 1.2338494533661139e-05,
      "loss": 0.8428,
      "step": 28900
    },
    {
      "epoch": 2.2624823916105807,
      "grad_norm": 20.483137130737305,
      "learning_rate": 1.2325417167965685e-05,
      "loss": 0.68,
      "step": 28910
    },
    {
      "epoch": 2.2632649866958836,
      "grad_norm": 8.916142463684082,
      "learning_rate": 1.2312339802270231e-05,
      "loss": 1.3102,
      "step": 28920
    },
    {
      "epoch": 2.2640475817811865,
      "grad_norm": 1.1019705533981323,
      "learning_rate": 1.2299262436574776e-05,
      "loss": 0.7228,
      "step": 28930
    },
    {
      "epoch": 2.2648301768664894,
      "grad_norm": 15.586950302124023,
      "learning_rate": 1.2286185070879324e-05,
      "loss": 0.6408,
      "step": 28940
    },
    {
      "epoch": 2.2656127719517922,
      "grad_norm": 8.181303977966309,
      "learning_rate": 1.2273107705183869e-05,
      "loss": 0.8865,
      "step": 28950
    },
    {
      "epoch": 2.266395367037095,
      "grad_norm": 17.24395179748535,
      "learning_rate": 1.2260030339488413e-05,
      "loss": 0.7763,
      "step": 28960
    },
    {
      "epoch": 2.267177962122398,
      "grad_norm": 11.958258628845215,
      "learning_rate": 1.224695297379296e-05,
      "loss": 0.6154,
      "step": 28970
    },
    {
      "epoch": 2.267960557207701,
      "grad_norm": 1.5090694427490234,
      "learning_rate": 1.2233875608097506e-05,
      "loss": 0.7337,
      "step": 28980
    },
    {
      "epoch": 2.2687431522930037,
      "grad_norm": 19.631576538085938,
      "learning_rate": 1.2220798242402052e-05,
      "loss": 0.6779,
      "step": 28990
    },
    {
      "epoch": 2.2695257473783066,
      "grad_norm": 3.5952510833740234,
      "learning_rate": 1.2207720876706597e-05,
      "loss": 0.6031,
      "step": 29000
    },
    {
      "epoch": 2.2703083424636095,
      "grad_norm": 17.54590606689453,
      "learning_rate": 1.2194643511011143e-05,
      "loss": 1.0061,
      "step": 29010
    },
    {
      "epoch": 2.2710909375489123,
      "grad_norm": 9.74372386932373,
      "learning_rate": 1.2181566145315689e-05,
      "loss": 0.688,
      "step": 29020
    },
    {
      "epoch": 2.271873532634215,
      "grad_norm": 11.582658767700195,
      "learning_rate": 1.2168488779620234e-05,
      "loss": 0.8634,
      "step": 29030
    },
    {
      "epoch": 2.272656127719518,
      "grad_norm": 0.841009795665741,
      "learning_rate": 1.215541141392478e-05,
      "loss": 0.5541,
      "step": 29040
    },
    {
      "epoch": 2.273438722804821,
      "grad_norm": 13.743673324584961,
      "learning_rate": 1.2142334048229326e-05,
      "loss": 0.745,
      "step": 29050
    },
    {
      "epoch": 2.274221317890124,
      "grad_norm": 3.477416515350342,
      "learning_rate": 1.212925668253387e-05,
      "loss": 0.5268,
      "step": 29060
    },
    {
      "epoch": 2.2750039129754267,
      "grad_norm": 13.247010231018066,
      "learning_rate": 1.2116179316838417e-05,
      "loss": 0.5187,
      "step": 29070
    },
    {
      "epoch": 2.2757865080607296,
      "grad_norm": 10.655842781066895,
      "learning_rate": 1.2103101951142963e-05,
      "loss": 0.7214,
      "step": 29080
    },
    {
      "epoch": 2.2765691031460324,
      "grad_norm": 1.3123672008514404,
      "learning_rate": 1.2090024585447508e-05,
      "loss": 0.6965,
      "step": 29090
    },
    {
      "epoch": 2.2773516982313353,
      "grad_norm": 7.525054931640625,
      "learning_rate": 1.2076947219752054e-05,
      "loss": 0.8982,
      "step": 29100
    },
    {
      "epoch": 2.278134293316638,
      "grad_norm": 2.812131643295288,
      "learning_rate": 1.2063869854056599e-05,
      "loss": 0.9436,
      "step": 29110
    },
    {
      "epoch": 2.278916888401941,
      "grad_norm": 5.6227593421936035,
      "learning_rate": 1.2050792488361145e-05,
      "loss": 0.6986,
      "step": 29120
    },
    {
      "epoch": 2.279699483487244,
      "grad_norm": 1.0272597074508667,
      "learning_rate": 1.2037715122665691e-05,
      "loss": 0.7272,
      "step": 29130
    },
    {
      "epoch": 2.280482078572547,
      "grad_norm": 15.854442596435547,
      "learning_rate": 1.2024637756970236e-05,
      "loss": 1.055,
      "step": 29140
    },
    {
      "epoch": 2.2812646736578492,
      "grad_norm": 1.2382537126541138,
      "learning_rate": 1.2011560391274782e-05,
      "loss": 0.6823,
      "step": 29150
    },
    {
      "epoch": 2.282047268743152,
      "grad_norm": 0.819065272808075,
      "learning_rate": 1.1998483025579328e-05,
      "loss": 0.504,
      "step": 29160
    },
    {
      "epoch": 2.282829863828455,
      "grad_norm": 9.866506576538086,
      "learning_rate": 1.1985405659883873e-05,
      "loss": 0.6335,
      "step": 29170
    },
    {
      "epoch": 2.283612458913758,
      "grad_norm": 12.050588607788086,
      "learning_rate": 1.197232829418842e-05,
      "loss": 0.7846,
      "step": 29180
    },
    {
      "epoch": 2.2843950539990607,
      "grad_norm": 16.188634872436523,
      "learning_rate": 1.1959250928492965e-05,
      "loss": 0.5684,
      "step": 29190
    },
    {
      "epoch": 2.2851776490843636,
      "grad_norm": 1.1101096868515015,
      "learning_rate": 1.194617356279751e-05,
      "loss": 0.7483,
      "step": 29200
    },
    {
      "epoch": 2.2859602441696665,
      "grad_norm": 53.595924377441406,
      "learning_rate": 1.1933096197102056e-05,
      "loss": 0.5106,
      "step": 29210
    },
    {
      "epoch": 2.2867428392549694,
      "grad_norm": 1.7351657152175903,
      "learning_rate": 1.19200188314066e-05,
      "loss": 0.3712,
      "step": 29220
    },
    {
      "epoch": 2.2875254343402722,
      "grad_norm": 14.273153305053711,
      "learning_rate": 1.1906941465711149e-05,
      "loss": 0.8796,
      "step": 29230
    },
    {
      "epoch": 2.288308029425575,
      "grad_norm": 11.936266899108887,
      "learning_rate": 1.1893864100015693e-05,
      "loss": 0.6146,
      "step": 29240
    },
    {
      "epoch": 2.289090624510878,
      "grad_norm": 4.541723728179932,
      "learning_rate": 1.1880786734320238e-05,
      "loss": 0.541,
      "step": 29250
    },
    {
      "epoch": 2.289873219596181,
      "grad_norm": 9.277892112731934,
      "learning_rate": 1.1867709368624786e-05,
      "loss": 0.9653,
      "step": 29260
    },
    {
      "epoch": 2.2906558146814837,
      "grad_norm": 3.8254334926605225,
      "learning_rate": 1.185463200292933e-05,
      "loss": 0.5203,
      "step": 29270
    },
    {
      "epoch": 2.2914384097667866,
      "grad_norm": 27.12392234802246,
      "learning_rate": 1.1841554637233877e-05,
      "loss": 0.5501,
      "step": 29280
    },
    {
      "epoch": 2.2922210048520895,
      "grad_norm": 2.868198871612549,
      "learning_rate": 1.1828477271538421e-05,
      "loss": 0.8233,
      "step": 29290
    },
    {
      "epoch": 2.2930035999373923,
      "grad_norm": 9.49418830871582,
      "learning_rate": 1.1815399905842968e-05,
      "loss": 0.6257,
      "step": 29300
    },
    {
      "epoch": 2.293786195022695,
      "grad_norm": 37.48750305175781,
      "learning_rate": 1.1802322540147514e-05,
      "loss": 0.5663,
      "step": 29310
    },
    {
      "epoch": 2.294568790107998,
      "grad_norm": 1.8494004011154175,
      "learning_rate": 1.1789245174452058e-05,
      "loss": 0.4179,
      "step": 29320
    },
    {
      "epoch": 2.295351385193301,
      "grad_norm": 12.947622299194336,
      "learning_rate": 1.1776167808756605e-05,
      "loss": 0.9562,
      "step": 29330
    },
    {
      "epoch": 2.296133980278604,
      "grad_norm": 8.248499870300293,
      "learning_rate": 1.1763090443061151e-05,
      "loss": 0.6381,
      "step": 29340
    },
    {
      "epoch": 2.2969165753639067,
      "grad_norm": 23.19195556640625,
      "learning_rate": 1.1750013077365696e-05,
      "loss": 1.0775,
      "step": 29350
    },
    {
      "epoch": 2.2976991704492096,
      "grad_norm": 7.171169281005859,
      "learning_rate": 1.1736935711670242e-05,
      "loss": 0.9386,
      "step": 29360
    },
    {
      "epoch": 2.2984817655345124,
      "grad_norm": 2.1315693855285645,
      "learning_rate": 1.1723858345974788e-05,
      "loss": 0.7435,
      "step": 29370
    },
    {
      "epoch": 2.2992643606198153,
      "grad_norm": 5.2050676345825195,
      "learning_rate": 1.1710780980279333e-05,
      "loss": 0.4975,
      "step": 29380
    },
    {
      "epoch": 2.300046955705118,
      "grad_norm": 9.218902587890625,
      "learning_rate": 1.1697703614583879e-05,
      "loss": 1.1662,
      "step": 29390
    },
    {
      "epoch": 2.300829550790421,
      "grad_norm": 10.843720436096191,
      "learning_rate": 1.1684626248888425e-05,
      "loss": 0.6319,
      "step": 29400
    },
    {
      "epoch": 2.301612145875724,
      "grad_norm": 10.812329292297363,
      "learning_rate": 1.167154888319297e-05,
      "loss": 0.6361,
      "step": 29410
    },
    {
      "epoch": 2.302394740961027,
      "grad_norm": 11.236583709716797,
      "learning_rate": 1.1658471517497516e-05,
      "loss": 0.9957,
      "step": 29420
    },
    {
      "epoch": 2.3031773360463297,
      "grad_norm": 19.36318016052246,
      "learning_rate": 1.164539415180206e-05,
      "loss": 0.9733,
      "step": 29430
    },
    {
      "epoch": 2.3039599311316326,
      "grad_norm": 10.600651741027832,
      "learning_rate": 1.1632316786106607e-05,
      "loss": 0.679,
      "step": 29440
    },
    {
      "epoch": 2.3047425262169354,
      "grad_norm": 2.944840431213379,
      "learning_rate": 1.1619239420411153e-05,
      "loss": 0.5436,
      "step": 29450
    },
    {
      "epoch": 2.3055251213022383,
      "grad_norm": 18.760663986206055,
      "learning_rate": 1.1606162054715698e-05,
      "loss": 1.2758,
      "step": 29460
    },
    {
      "epoch": 2.306307716387541,
      "grad_norm": 1.0447930097579956,
      "learning_rate": 1.1593084689020246e-05,
      "loss": 0.5389,
      "step": 29470
    },
    {
      "epoch": 2.307090311472844,
      "grad_norm": 9.507171630859375,
      "learning_rate": 1.158000732332479e-05,
      "loss": 0.5284,
      "step": 29480
    },
    {
      "epoch": 2.307872906558147,
      "grad_norm": 0.7498610615730286,
      "learning_rate": 1.1566929957629335e-05,
      "loss": 0.632,
      "step": 29490
    },
    {
      "epoch": 2.30865550164345,
      "grad_norm": 1.510491132736206,
      "learning_rate": 1.1553852591933881e-05,
      "loss": 0.6021,
      "step": 29500
    },
    {
      "epoch": 2.3094380967287527,
      "grad_norm": 9.872903823852539,
      "learning_rate": 1.1540775226238427e-05,
      "loss": 0.7331,
      "step": 29510
    },
    {
      "epoch": 2.3102206918140555,
      "grad_norm": 2.446087598800659,
      "learning_rate": 1.1527697860542974e-05,
      "loss": 0.793,
      "step": 29520
    },
    {
      "epoch": 2.3110032868993584,
      "grad_norm": 9.441887855529785,
      "learning_rate": 1.1514620494847518e-05,
      "loss": 0.8336,
      "step": 29530
    },
    {
      "epoch": 2.3117858819846613,
      "grad_norm": 41.94770431518555,
      "learning_rate": 1.1501543129152064e-05,
      "loss": 0.6735,
      "step": 29540
    },
    {
      "epoch": 2.312568477069964,
      "grad_norm": 10.9691162109375,
      "learning_rate": 1.148846576345661e-05,
      "loss": 0.9557,
      "step": 29550
    },
    {
      "epoch": 2.313351072155267,
      "grad_norm": 8.776360511779785,
      "learning_rate": 1.1475388397761155e-05,
      "loss": 1.616,
      "step": 29560
    },
    {
      "epoch": 2.31413366724057,
      "grad_norm": 2.795468807220459,
      "learning_rate": 1.1462311032065702e-05,
      "loss": 0.6143,
      "step": 29570
    },
    {
      "epoch": 2.3149162623258728,
      "grad_norm": 3.0981969833374023,
      "learning_rate": 1.1449233666370248e-05,
      "loss": 0.6225,
      "step": 29580
    },
    {
      "epoch": 2.3156988574111756,
      "grad_norm": 19.74159049987793,
      "learning_rate": 1.1436156300674792e-05,
      "loss": 0.6784,
      "step": 29590
    },
    {
      "epoch": 2.316481452496478,
      "grad_norm": 19.183551788330078,
      "learning_rate": 1.1423078934979339e-05,
      "loss": 0.5971,
      "step": 29600
    },
    {
      "epoch": 2.317264047581781,
      "grad_norm": 36.348724365234375,
      "learning_rate": 1.1410001569283885e-05,
      "loss": 0.9318,
      "step": 29610
    },
    {
      "epoch": 2.318046642667084,
      "grad_norm": 3.7115092277526855,
      "learning_rate": 1.139692420358843e-05,
      "loss": 0.5924,
      "step": 29620
    },
    {
      "epoch": 2.3188292377523867,
      "grad_norm": 10.651322364807129,
      "learning_rate": 1.1383846837892976e-05,
      "loss": 0.6781,
      "step": 29630
    },
    {
      "epoch": 2.3196118328376896,
      "grad_norm": 10.631875991821289,
      "learning_rate": 1.137076947219752e-05,
      "loss": 0.8133,
      "step": 29640
    },
    {
      "epoch": 2.3203944279229924,
      "grad_norm": 7.216742515563965,
      "learning_rate": 1.1357692106502067e-05,
      "loss": 0.8643,
      "step": 29650
    },
    {
      "epoch": 2.3211770230082953,
      "grad_norm": 13.12137222290039,
      "learning_rate": 1.1344614740806613e-05,
      "loss": 0.8336,
      "step": 29660
    },
    {
      "epoch": 2.321959618093598,
      "grad_norm": 9.115492820739746,
      "learning_rate": 1.1331537375111157e-05,
      "loss": 0.5355,
      "step": 29670
    },
    {
      "epoch": 2.322742213178901,
      "grad_norm": 18.722692489624023,
      "learning_rate": 1.1318460009415704e-05,
      "loss": 0.6691,
      "step": 29680
    },
    {
      "epoch": 2.323524808264204,
      "grad_norm": 12.554488182067871,
      "learning_rate": 1.130538264372025e-05,
      "loss": 0.6658,
      "step": 29690
    },
    {
      "epoch": 2.324307403349507,
      "grad_norm": 1.620331048965454,
      "learning_rate": 1.1292305278024795e-05,
      "loss": 0.649,
      "step": 29700
    },
    {
      "epoch": 2.3250899984348097,
      "grad_norm": 11.083174705505371,
      "learning_rate": 1.127922791232934e-05,
      "loss": 0.9448,
      "step": 29710
    },
    {
      "epoch": 2.3258725935201126,
      "grad_norm": 2.021794557571411,
      "learning_rate": 1.1266150546633887e-05,
      "loss": 0.5234,
      "step": 29720
    },
    {
      "epoch": 2.3266551886054154,
      "grad_norm": 4.626170635223389,
      "learning_rate": 1.1253073180938432e-05,
      "loss": 0.9323,
      "step": 29730
    },
    {
      "epoch": 2.3274377836907183,
      "grad_norm": 0.8002862930297852,
      "learning_rate": 1.1239995815242978e-05,
      "loss": 0.5163,
      "step": 29740
    },
    {
      "epoch": 2.328220378776021,
      "grad_norm": 0.7993191480636597,
      "learning_rate": 1.1226918449547524e-05,
      "loss": 0.4052,
      "step": 29750
    },
    {
      "epoch": 2.329002973861324,
      "grad_norm": 0.9969700574874878,
      "learning_rate": 1.121384108385207e-05,
      "loss": 0.9183,
      "step": 29760
    },
    {
      "epoch": 2.329785568946627,
      "grad_norm": 35.88633346557617,
      "learning_rate": 1.1200763718156615e-05,
      "loss": 0.3847,
      "step": 29770
    },
    {
      "epoch": 2.33056816403193,
      "grad_norm": 2.1834824085235596,
      "learning_rate": 1.118768635246116e-05,
      "loss": 0.5519,
      "step": 29780
    },
    {
      "epoch": 2.3313507591172327,
      "grad_norm": 12.117508888244629,
      "learning_rate": 1.1174608986765708e-05,
      "loss": 0.8888,
      "step": 29790
    },
    {
      "epoch": 2.3321333542025355,
      "grad_norm": 2.7807860374450684,
      "learning_rate": 1.1161531621070252e-05,
      "loss": 0.4693,
      "step": 29800
    },
    {
      "epoch": 2.3329159492878384,
      "grad_norm": 0.8564630150794983,
      "learning_rate": 1.1148454255374798e-05,
      "loss": 0.4906,
      "step": 29810
    },
    {
      "epoch": 2.3336985443731413,
      "grad_norm": 0.9275245070457458,
      "learning_rate": 1.1135376889679343e-05,
      "loss": 0.4531,
      "step": 29820
    },
    {
      "epoch": 2.334481139458444,
      "grad_norm": 7.12086820602417,
      "learning_rate": 1.112229952398389e-05,
      "loss": 0.5855,
      "step": 29830
    },
    {
      "epoch": 2.335263734543747,
      "grad_norm": 0.33474865555763245,
      "learning_rate": 1.1109222158288435e-05,
      "loss": 0.5227,
      "step": 29840
    },
    {
      "epoch": 2.33604632962905,
      "grad_norm": 10.774924278259277,
      "learning_rate": 1.109614479259298e-05,
      "loss": 0.6207,
      "step": 29850
    },
    {
      "epoch": 2.3368289247143528,
      "grad_norm": 11.722686767578125,
      "learning_rate": 1.1083067426897526e-05,
      "loss": 0.5109,
      "step": 29860
    },
    {
      "epoch": 2.3376115197996556,
      "grad_norm": 29.595596313476562,
      "learning_rate": 1.1069990061202073e-05,
      "loss": 1.0046,
      "step": 29870
    },
    {
      "epoch": 2.3383941148849585,
      "grad_norm": 15.055028915405273,
      "learning_rate": 1.1056912695506617e-05,
      "loss": 1.2526,
      "step": 29880
    },
    {
      "epoch": 2.3391767099702614,
      "grad_norm": 10.024129867553711,
      "learning_rate": 1.1043835329811163e-05,
      "loss": 0.9182,
      "step": 29890
    },
    {
      "epoch": 2.3399593050555643,
      "grad_norm": 0.8820858597755432,
      "learning_rate": 1.103075796411571e-05,
      "loss": 0.782,
      "step": 29900
    },
    {
      "epoch": 2.340741900140867,
      "grad_norm": 22.637250900268555,
      "learning_rate": 1.1017680598420254e-05,
      "loss": 0.9596,
      "step": 29910
    },
    {
      "epoch": 2.34152449522617,
      "grad_norm": 1.2055853605270386,
      "learning_rate": 1.10046032327248e-05,
      "loss": 0.8283,
      "step": 29920
    },
    {
      "epoch": 2.342307090311473,
      "grad_norm": 14.550863265991211,
      "learning_rate": 1.0991525867029347e-05,
      "loss": 0.9412,
      "step": 29930
    },
    {
      "epoch": 2.3430896853967758,
      "grad_norm": 8.756558418273926,
      "learning_rate": 1.0978448501333891e-05,
      "loss": 0.8243,
      "step": 29940
    },
    {
      "epoch": 2.3438722804820786,
      "grad_norm": 6.258150577545166,
      "learning_rate": 1.0965371135638438e-05,
      "loss": 0.5969,
      "step": 29950
    },
    {
      "epoch": 2.3446548755673815,
      "grad_norm": 6.4616241455078125,
      "learning_rate": 1.0952293769942982e-05,
      "loss": 0.8766,
      "step": 29960
    },
    {
      "epoch": 2.3454374706526844,
      "grad_norm": 2.6577212810516357,
      "learning_rate": 1.0939216404247528e-05,
      "loss": 0.6108,
      "step": 29970
    },
    {
      "epoch": 2.3462200657379872,
      "grad_norm": 8.526130676269531,
      "learning_rate": 1.0926139038552075e-05,
      "loss": 0.586,
      "step": 29980
    },
    {
      "epoch": 2.34700266082329,
      "grad_norm": 2.159850835800171,
      "learning_rate": 1.091306167285662e-05,
      "loss": 0.6928,
      "step": 29990
    },
    {
      "epoch": 2.347785255908593,
      "grad_norm": 11.420709609985352,
      "learning_rate": 1.0899984307161167e-05,
      "loss": 0.6936,
      "step": 30000
    },
    {
      "epoch": 2.348567850993896,
      "grad_norm": 15.549372673034668,
      "learning_rate": 1.0886906941465712e-05,
      "loss": 0.8571,
      "step": 30010
    },
    {
      "epoch": 2.3493504460791987,
      "grad_norm": 15.926445007324219,
      "learning_rate": 1.0873829575770256e-05,
      "loss": 0.8686,
      "step": 30020
    },
    {
      "epoch": 2.3501330411645016,
      "grad_norm": 8.2820463180542,
      "learning_rate": 1.0860752210074803e-05,
      "loss": 0.9724,
      "step": 30030
    },
    {
      "epoch": 2.3509156362498045,
      "grad_norm": 17.990325927734375,
      "learning_rate": 1.0847674844379349e-05,
      "loss": 0.7721,
      "step": 30040
    },
    {
      "epoch": 2.3516982313351074,
      "grad_norm": 0.9953508973121643,
      "learning_rate": 1.0834597478683895e-05,
      "loss": 0.3863,
      "step": 30050
    },
    {
      "epoch": 2.3524808264204102,
      "grad_norm": 0.5159671902656555,
      "learning_rate": 1.082152011298844e-05,
      "loss": 1.1485,
      "step": 30060
    },
    {
      "epoch": 2.353263421505713,
      "grad_norm": 12.904888153076172,
      "learning_rate": 1.0808442747292986e-05,
      "loss": 0.9822,
      "step": 30070
    },
    {
      "epoch": 2.354046016591016,
      "grad_norm": 1.6344401836395264,
      "learning_rate": 1.0795365381597532e-05,
      "loss": 0.505,
      "step": 30080
    },
    {
      "epoch": 2.354828611676319,
      "grad_norm": 0.9323446750640869,
      "learning_rate": 1.0782288015902077e-05,
      "loss": 1.0486,
      "step": 30090
    },
    {
      "epoch": 2.3556112067616217,
      "grad_norm": 18.58977508544922,
      "learning_rate": 1.0769210650206621e-05,
      "loss": 0.922,
      "step": 30100
    },
    {
      "epoch": 2.3563938018469246,
      "grad_norm": 13.530488967895508,
      "learning_rate": 1.075613328451117e-05,
      "loss": 0.8301,
      "step": 30110
    },
    {
      "epoch": 2.3571763969322275,
      "grad_norm": 8.301910400390625,
      "learning_rate": 1.0743055918815714e-05,
      "loss": 0.9058,
      "step": 30120
    },
    {
      "epoch": 2.3579589920175303,
      "grad_norm": 11.303254127502441,
      "learning_rate": 1.072997855312026e-05,
      "loss": 0.8506,
      "step": 30130
    },
    {
      "epoch": 2.358741587102833,
      "grad_norm": 13.609940528869629,
      "learning_rate": 1.0716901187424807e-05,
      "loss": 0.8398,
      "step": 30140
    },
    {
      "epoch": 2.359524182188136,
      "grad_norm": 16.862964630126953,
      "learning_rate": 1.0703823821729351e-05,
      "loss": 0.5264,
      "step": 30150
    },
    {
      "epoch": 2.360306777273439,
      "grad_norm": 7.446914196014404,
      "learning_rate": 1.0690746456033897e-05,
      "loss": 0.5317,
      "step": 30160
    },
    {
      "epoch": 2.361089372358742,
      "grad_norm": 3.245384931564331,
      "learning_rate": 1.0677669090338442e-05,
      "loss": 0.5263,
      "step": 30170
    },
    {
      "epoch": 2.3618719674440443,
      "grad_norm": 1.7723047733306885,
      "learning_rate": 1.0664591724642988e-05,
      "loss": 1.017,
      "step": 30180
    },
    {
      "epoch": 2.362654562529347,
      "grad_norm": 9.990551948547363,
      "learning_rate": 1.0651514358947534e-05,
      "loss": 0.6334,
      "step": 30190
    },
    {
      "epoch": 2.36343715761465,
      "grad_norm": 1.3746365308761597,
      "learning_rate": 1.0638436993252079e-05,
      "loss": 0.5743,
      "step": 30200
    },
    {
      "epoch": 2.364219752699953,
      "grad_norm": 13.314827919006348,
      "learning_rate": 1.0625359627556625e-05,
      "loss": 0.7907,
      "step": 30210
    },
    {
      "epoch": 2.3650023477852558,
      "grad_norm": 7.002245903015137,
      "learning_rate": 1.0612282261861172e-05,
      "loss": 0.9562,
      "step": 30220
    },
    {
      "epoch": 2.3657849428705586,
      "grad_norm": 15.965538024902344,
      "learning_rate": 1.0599204896165716e-05,
      "loss": 0.7522,
      "step": 30230
    },
    {
      "epoch": 2.3665675379558615,
      "grad_norm": 16.37049102783203,
      "learning_rate": 1.0586127530470262e-05,
      "loss": 0.4769,
      "step": 30240
    },
    {
      "epoch": 2.3673501330411644,
      "grad_norm": 5.697165012359619,
      "learning_rate": 1.0573050164774809e-05,
      "loss": 0.6676,
      "step": 30250
    },
    {
      "epoch": 2.3681327281264672,
      "grad_norm": 14.043044090270996,
      "learning_rate": 1.0559972799079353e-05,
      "loss": 0.91,
      "step": 30260
    },
    {
      "epoch": 2.36891532321177,
      "grad_norm": 22.480989456176758,
      "learning_rate": 1.05468954333839e-05,
      "loss": 0.6839,
      "step": 30270
    },
    {
      "epoch": 2.369697918297073,
      "grad_norm": 7.655246734619141,
      "learning_rate": 1.0533818067688446e-05,
      "loss": 0.3205,
      "step": 30280
    },
    {
      "epoch": 2.370480513382376,
      "grad_norm": 20.034330368041992,
      "learning_rate": 1.0520740701992992e-05,
      "loss": 0.8971,
      "step": 30290
    },
    {
      "epoch": 2.3712631084676787,
      "grad_norm": 10.673073768615723,
      "learning_rate": 1.0507663336297537e-05,
      "loss": 1.005,
      "step": 30300
    },
    {
      "epoch": 2.3720457035529816,
      "grad_norm": 24.30335807800293,
      "learning_rate": 1.0494585970602081e-05,
      "loss": 0.6126,
      "step": 30310
    },
    {
      "epoch": 2.3728282986382845,
      "grad_norm": 7.116356372833252,
      "learning_rate": 1.0481508604906629e-05,
      "loss": 0.5573,
      "step": 30320
    },
    {
      "epoch": 2.3736108937235874,
      "grad_norm": 4.947970867156982,
      "learning_rate": 1.0468431239211174e-05,
      "loss": 0.7757,
      "step": 30330
    },
    {
      "epoch": 2.3743934888088902,
      "grad_norm": 10.108707427978516,
      "learning_rate": 1.045535387351572e-05,
      "loss": 1.0565,
      "step": 30340
    },
    {
      "epoch": 2.375176083894193,
      "grad_norm": 6.84506368637085,
      "learning_rate": 1.0442276507820266e-05,
      "loss": 0.5065,
      "step": 30350
    },
    {
      "epoch": 2.375958678979496,
      "grad_norm": 3.7498886585235596,
      "learning_rate": 1.042919914212481e-05,
      "loss": 0.6824,
      "step": 30360
    },
    {
      "epoch": 2.376741274064799,
      "grad_norm": 2.4262101650238037,
      "learning_rate": 1.0416121776429357e-05,
      "loss": 0.8574,
      "step": 30370
    },
    {
      "epoch": 2.3775238691501017,
      "grad_norm": 28.85550308227539,
      "learning_rate": 1.0403044410733902e-05,
      "loss": 0.4415,
      "step": 30380
    },
    {
      "epoch": 2.3783064642354046,
      "grad_norm": 28.086658477783203,
      "learning_rate": 1.0389967045038448e-05,
      "loss": 0.6915,
      "step": 30390
    },
    {
      "epoch": 2.3790890593207075,
      "grad_norm": 4.331473350524902,
      "learning_rate": 1.0376889679342994e-05,
      "loss": 0.4848,
      "step": 30400
    },
    {
      "epoch": 2.3798716544060103,
      "grad_norm": 25.73353385925293,
      "learning_rate": 1.0363812313647539e-05,
      "loss": 0.7028,
      "step": 30410
    },
    {
      "epoch": 2.380654249491313,
      "grad_norm": 19.817739486694336,
      "learning_rate": 1.0350734947952085e-05,
      "loss": 0.9444,
      "step": 30420
    },
    {
      "epoch": 2.381436844576616,
      "grad_norm": 29.54033660888672,
      "learning_rate": 1.0337657582256631e-05,
      "loss": 0.6268,
      "step": 30430
    },
    {
      "epoch": 2.382219439661919,
      "grad_norm": 8.186518669128418,
      "learning_rate": 1.0324580216561176e-05,
      "loss": 0.4549,
      "step": 30440
    },
    {
      "epoch": 2.383002034747222,
      "grad_norm": 16.787803649902344,
      "learning_rate": 1.0311502850865722e-05,
      "loss": 0.799,
      "step": 30450
    },
    {
      "epoch": 2.3837846298325247,
      "grad_norm": 0.9046483039855957,
      "learning_rate": 1.0298425485170268e-05,
      "loss": 0.6456,
      "step": 30460
    },
    {
      "epoch": 2.3845672249178276,
      "grad_norm": 7.019650936126709,
      "learning_rate": 1.0285348119474813e-05,
      "loss": 0.6882,
      "step": 30470
    },
    {
      "epoch": 2.3853498200031304,
      "grad_norm": 4.331158638000488,
      "learning_rate": 1.027227075377936e-05,
      "loss": 0.6444,
      "step": 30480
    },
    {
      "epoch": 2.3861324150884333,
      "grad_norm": 12.30126953125,
      "learning_rate": 1.0259193388083904e-05,
      "loss": 0.5744,
      "step": 30490
    },
    {
      "epoch": 2.386915010173736,
      "grad_norm": 5.394369602203369,
      "learning_rate": 1.024611602238845e-05,
      "loss": 0.6208,
      "step": 30500
    },
    {
      "epoch": 2.387697605259039,
      "grad_norm": 24.337810516357422,
      "learning_rate": 1.0233038656692996e-05,
      "loss": 0.6756,
      "step": 30510
    },
    {
      "epoch": 2.388480200344342,
      "grad_norm": 14.254026412963867,
      "learning_rate": 1.0219961290997541e-05,
      "loss": 0.8241,
      "step": 30520
    },
    {
      "epoch": 2.389262795429645,
      "grad_norm": 2.3426499366760254,
      "learning_rate": 1.0206883925302089e-05,
      "loss": 0.6728,
      "step": 30530
    },
    {
      "epoch": 2.3900453905149477,
      "grad_norm": 23.190404891967773,
      "learning_rate": 1.0193806559606633e-05,
      "loss": 0.7039,
      "step": 30540
    },
    {
      "epoch": 2.3908279856002506,
      "grad_norm": 0.342405766248703,
      "learning_rate": 1.0180729193911178e-05,
      "loss": 0.9739,
      "step": 30550
    },
    {
      "epoch": 2.3916105806855534,
      "grad_norm": 1.6985507011413574,
      "learning_rate": 1.0167651828215724e-05,
      "loss": 0.8269,
      "step": 30560
    },
    {
      "epoch": 2.3923931757708563,
      "grad_norm": 0.7223021984100342,
      "learning_rate": 1.015457446252027e-05,
      "loss": 1.0287,
      "step": 30570
    },
    {
      "epoch": 2.393175770856159,
      "grad_norm": 8.799676895141602,
      "learning_rate": 1.0141497096824817e-05,
      "loss": 0.7823,
      "step": 30580
    },
    {
      "epoch": 2.393958365941462,
      "grad_norm": 6.3330464363098145,
      "learning_rate": 1.0128419731129361e-05,
      "loss": 0.9309,
      "step": 30590
    },
    {
      "epoch": 2.394740961026765,
      "grad_norm": 25.61558723449707,
      "learning_rate": 1.0115342365433908e-05,
      "loss": 0.8752,
      "step": 30600
    },
    {
      "epoch": 2.395523556112068,
      "grad_norm": 16.980920791625977,
      "learning_rate": 1.0102264999738454e-05,
      "loss": 0.5697,
      "step": 30610
    },
    {
      "epoch": 2.3963061511973707,
      "grad_norm": 3.0881783962249756,
      "learning_rate": 1.0089187634042999e-05,
      "loss": 0.6193,
      "step": 30620
    },
    {
      "epoch": 2.397088746282673,
      "grad_norm": 9.166165351867676,
      "learning_rate": 1.0076110268347543e-05,
      "loss": 0.6058,
      "step": 30630
    },
    {
      "epoch": 2.397871341367976,
      "grad_norm": 18.767494201660156,
      "learning_rate": 1.0063032902652091e-05,
      "loss": 0.5567,
      "step": 30640
    },
    {
      "epoch": 2.398653936453279,
      "grad_norm": 5.304967403411865,
      "learning_rate": 1.0049955536956636e-05,
      "loss": 0.7314,
      "step": 30650
    },
    {
      "epoch": 2.3994365315385817,
      "grad_norm": 16.05694007873535,
      "learning_rate": 1.0036878171261182e-05,
      "loss": 0.6154,
      "step": 30660
    },
    {
      "epoch": 2.4002191266238846,
      "grad_norm": 4.893183708190918,
      "learning_rate": 1.0023800805565728e-05,
      "loss": 0.3556,
      "step": 30670
    },
    {
      "epoch": 2.4010017217091875,
      "grad_norm": 7.981976509094238,
      "learning_rate": 1.0010723439870273e-05,
      "loss": 0.5534,
      "step": 30680
    },
    {
      "epoch": 2.4017843167944903,
      "grad_norm": 9.21424674987793,
      "learning_rate": 9.997646074174819e-06,
      "loss": 0.8817,
      "step": 30690
    },
    {
      "epoch": 2.402566911879793,
      "grad_norm": 16.677879333496094,
      "learning_rate": 9.984568708479364e-06,
      "loss": 0.9429,
      "step": 30700
    },
    {
      "epoch": 2.403349506965096,
      "grad_norm": 21.53274154663086,
      "learning_rate": 9.97149134278391e-06,
      "loss": 0.6443,
      "step": 30710
    },
    {
      "epoch": 2.404132102050399,
      "grad_norm": 15.453154563903809,
      "learning_rate": 9.958413977088456e-06,
      "loss": 0.7618,
      "step": 30720
    },
    {
      "epoch": 2.404914697135702,
      "grad_norm": 8.596860885620117,
      "learning_rate": 9.945336611393e-06,
      "loss": 0.4479,
      "step": 30730
    },
    {
      "epoch": 2.4056972922210047,
      "grad_norm": 59.1745719909668,
      "learning_rate": 9.932259245697547e-06,
      "loss": 1.0558,
      "step": 30740
    },
    {
      "epoch": 2.4064798873063076,
      "grad_norm": 2.5243632793426514,
      "learning_rate": 9.919181880002093e-06,
      "loss": 0.6799,
      "step": 30750
    },
    {
      "epoch": 2.4072624823916104,
      "grad_norm": 7.344128131866455,
      "learning_rate": 9.906104514306638e-06,
      "loss": 0.6445,
      "step": 30760
    },
    {
      "epoch": 2.4080450774769133,
      "grad_norm": 1.1467002630233765,
      "learning_rate": 9.893027148611184e-06,
      "loss": 0.4255,
      "step": 30770
    },
    {
      "epoch": 2.408827672562216,
      "grad_norm": 23.99819564819336,
      "learning_rate": 9.87994978291573e-06,
      "loss": 0.695,
      "step": 30780
    },
    {
      "epoch": 2.409610267647519,
      "grad_norm": 15.483481407165527,
      "learning_rate": 9.866872417220275e-06,
      "loss": 0.3885,
      "step": 30790
    },
    {
      "epoch": 2.410392862732822,
      "grad_norm": 8.849390029907227,
      "learning_rate": 9.853795051524821e-06,
      "loss": 0.5758,
      "step": 30800
    },
    {
      "epoch": 2.411175457818125,
      "grad_norm": 15.042736053466797,
      "learning_rate": 9.840717685829367e-06,
      "loss": 0.5772,
      "step": 30810
    },
    {
      "epoch": 2.4119580529034277,
      "grad_norm": 32.002410888671875,
      "learning_rate": 9.827640320133914e-06,
      "loss": 1.2631,
      "step": 30820
    },
    {
      "epoch": 2.4127406479887306,
      "grad_norm": 11.93320083618164,
      "learning_rate": 9.814562954438458e-06,
      "loss": 0.7166,
      "step": 30830
    },
    {
      "epoch": 2.4135232430740334,
      "grad_norm": 14.719927787780762,
      "learning_rate": 9.801485588743003e-06,
      "loss": 0.4868,
      "step": 30840
    },
    {
      "epoch": 2.4143058381593363,
      "grad_norm": 18.446847915649414,
      "learning_rate": 9.78840822304755e-06,
      "loss": 0.9531,
      "step": 30850
    },
    {
      "epoch": 2.415088433244639,
      "grad_norm": 9.735895156860352,
      "learning_rate": 9.775330857352095e-06,
      "loss": 0.8718,
      "step": 30860
    },
    {
      "epoch": 2.415871028329942,
      "grad_norm": 7.555359363555908,
      "learning_rate": 9.762253491656642e-06,
      "loss": 0.7098,
      "step": 30870
    },
    {
      "epoch": 2.416653623415245,
      "grad_norm": 12.193242073059082,
      "learning_rate": 9.749176125961188e-06,
      "loss": 0.5474,
      "step": 30880
    },
    {
      "epoch": 2.417436218500548,
      "grad_norm": 11.65748405456543,
      "learning_rate": 9.736098760265732e-06,
      "loss": 0.8635,
      "step": 30890
    },
    {
      "epoch": 2.4182188135858507,
      "grad_norm": 6.3704118728637695,
      "learning_rate": 9.723021394570279e-06,
      "loss": 0.7416,
      "step": 30900
    },
    {
      "epoch": 2.4190014086711535,
      "grad_norm": 10.107367515563965,
      "learning_rate": 9.709944028874823e-06,
      "loss": 0.9468,
      "step": 30910
    },
    {
      "epoch": 2.4197840037564564,
      "grad_norm": 0.5407813191413879,
      "learning_rate": 9.69686666317937e-06,
      "loss": 0.5822,
      "step": 30920
    },
    {
      "epoch": 2.4205665988417593,
      "grad_norm": 6.115610599517822,
      "learning_rate": 9.683789297483916e-06,
      "loss": 0.4464,
      "step": 30930
    },
    {
      "epoch": 2.421349193927062,
      "grad_norm": 11.724448204040527,
      "learning_rate": 9.67071193178846e-06,
      "loss": 0.8373,
      "step": 30940
    },
    {
      "epoch": 2.422131789012365,
      "grad_norm": 3.7409982681274414,
      "learning_rate": 9.657634566093007e-06,
      "loss": 0.8388,
      "step": 30950
    },
    {
      "epoch": 2.422914384097668,
      "grad_norm": 7.386172294616699,
      "learning_rate": 9.644557200397553e-06,
      "loss": 0.5074,
      "step": 30960
    },
    {
      "epoch": 2.423696979182971,
      "grad_norm": 9.878191947937012,
      "learning_rate": 9.631479834702097e-06,
      "loss": 0.4161,
      "step": 30970
    },
    {
      "epoch": 2.4244795742682737,
      "grad_norm": 30.379453659057617,
      "learning_rate": 9.618402469006644e-06,
      "loss": 0.9334,
      "step": 30980
    },
    {
      "epoch": 2.4252621693535765,
      "grad_norm": 6.872690200805664,
      "learning_rate": 9.60532510331119e-06,
      "loss": 1.0268,
      "step": 30990
    },
    {
      "epoch": 2.4260447644388794,
      "grad_norm": 8.643999099731445,
      "learning_rate": 9.592247737615735e-06,
      "loss": 0.6097,
      "step": 31000
    },
    {
      "epoch": 2.4268273595241823,
      "grad_norm": 0.7294657826423645,
      "learning_rate": 9.579170371920281e-06,
      "loss": 0.7454,
      "step": 31010
    },
    {
      "epoch": 2.427609954609485,
      "grad_norm": 11.163689613342285,
      "learning_rate": 9.566093006224827e-06,
      "loss": 0.6096,
      "step": 31020
    },
    {
      "epoch": 2.428392549694788,
      "grad_norm": 6.000378608703613,
      "learning_rate": 9.553015640529372e-06,
      "loss": 0.7654,
      "step": 31030
    },
    {
      "epoch": 2.429175144780091,
      "grad_norm": 11.96697998046875,
      "learning_rate": 9.539938274833918e-06,
      "loss": 0.833,
      "step": 31040
    },
    {
      "epoch": 2.4299577398653938,
      "grad_norm": 25.728086471557617,
      "learning_rate": 9.526860909138463e-06,
      "loss": 1.0321,
      "step": 31050
    },
    {
      "epoch": 2.4307403349506966,
      "grad_norm": 5.082222938537598,
      "learning_rate": 9.51378354344301e-06,
      "loss": 0.7308,
      "step": 31060
    },
    {
      "epoch": 2.4315229300359995,
      "grad_norm": 0.9146837592124939,
      "learning_rate": 9.500706177747555e-06,
      "loss": 0.7195,
      "step": 31070
    },
    {
      "epoch": 2.4323055251213024,
      "grad_norm": 15.493927955627441,
      "learning_rate": 9.4876288120521e-06,
      "loss": 0.8631,
      "step": 31080
    },
    {
      "epoch": 2.4330881202066053,
      "grad_norm": 1.9040849208831787,
      "learning_rate": 9.474551446356648e-06,
      "loss": 0.8389,
      "step": 31090
    },
    {
      "epoch": 2.433870715291908,
      "grad_norm": 7.047844409942627,
      "learning_rate": 9.461474080661192e-06,
      "loss": 0.4491,
      "step": 31100
    },
    {
      "epoch": 2.434653310377211,
      "grad_norm": 5.688796520233154,
      "learning_rate": 9.448396714965738e-06,
      "loss": 0.5629,
      "step": 31110
    },
    {
      "epoch": 2.435435905462514,
      "grad_norm": 6.690713405609131,
      "learning_rate": 9.435319349270283e-06,
      "loss": 0.9569,
      "step": 31120
    },
    {
      "epoch": 2.4362185005478167,
      "grad_norm": 1.197656512260437,
      "learning_rate": 9.42224198357483e-06,
      "loss": 0.5884,
      "step": 31130
    },
    {
      "epoch": 2.4370010956331196,
      "grad_norm": 0.439404159784317,
      "learning_rate": 9.409164617879376e-06,
      "loss": 0.5942,
      "step": 31140
    },
    {
      "epoch": 2.4377836907184225,
      "grad_norm": 16.45393180847168,
      "learning_rate": 9.39608725218392e-06,
      "loss": 0.859,
      "step": 31150
    },
    {
      "epoch": 2.4385662858037254,
      "grad_norm": 17.7769718170166,
      "learning_rate": 9.383009886488466e-06,
      "loss": 0.6381,
      "step": 31160
    },
    {
      "epoch": 2.4393488808890282,
      "grad_norm": 7.870083332061768,
      "learning_rate": 9.369932520793013e-06,
      "loss": 0.5771,
      "step": 31170
    },
    {
      "epoch": 2.440131475974331,
      "grad_norm": 13.4640531539917,
      "learning_rate": 9.356855155097557e-06,
      "loss": 0.6401,
      "step": 31180
    },
    {
      "epoch": 2.440914071059634,
      "grad_norm": 1.9040311574935913,
      "learning_rate": 9.343777789402103e-06,
      "loss": 0.4234,
      "step": 31190
    },
    {
      "epoch": 2.441696666144937,
      "grad_norm": 18.436290740966797,
      "learning_rate": 9.33070042370665e-06,
      "loss": 0.4866,
      "step": 31200
    },
    {
      "epoch": 2.4424792612302393,
      "grad_norm": 3.886659860610962,
      "learning_rate": 9.317623058011194e-06,
      "loss": 0.6424,
      "step": 31210
    },
    {
      "epoch": 2.443261856315542,
      "grad_norm": 0.28430935740470886,
      "learning_rate": 9.30454569231574e-06,
      "loss": 0.6333,
      "step": 31220
    },
    {
      "epoch": 2.444044451400845,
      "grad_norm": 20.098173141479492,
      "learning_rate": 9.291468326620285e-06,
      "loss": 0.863,
      "step": 31230
    },
    {
      "epoch": 2.444827046486148,
      "grad_norm": 5.974355697631836,
      "learning_rate": 9.278390960924831e-06,
      "loss": 0.6398,
      "step": 31240
    },
    {
      "epoch": 2.4456096415714508,
      "grad_norm": 12.299909591674805,
      "learning_rate": 9.265313595229378e-06,
      "loss": 0.7337,
      "step": 31250
    },
    {
      "epoch": 2.4463922366567536,
      "grad_norm": 6.695287704467773,
      "learning_rate": 9.252236229533922e-06,
      "loss": 0.8323,
      "step": 31260
    },
    {
      "epoch": 2.4471748317420565,
      "grad_norm": 25.23521614074707,
      "learning_rate": 9.239158863838469e-06,
      "loss": 0.5254,
      "step": 31270
    },
    {
      "epoch": 2.4479574268273594,
      "grad_norm": 15.541640281677246,
      "learning_rate": 9.226081498143015e-06,
      "loss": 0.8569,
      "step": 31280
    },
    {
      "epoch": 2.4487400219126623,
      "grad_norm": 10.7070894241333,
      "learning_rate": 9.21300413244756e-06,
      "loss": 0.7017,
      "step": 31290
    },
    {
      "epoch": 2.449522616997965,
      "grad_norm": 26.2651424407959,
      "learning_rate": 9.199926766752106e-06,
      "loss": 0.7959,
      "step": 31300
    },
    {
      "epoch": 2.450305212083268,
      "grad_norm": 37.27280044555664,
      "learning_rate": 9.186849401056652e-06,
      "loss": 1.0015,
      "step": 31310
    },
    {
      "epoch": 2.451087807168571,
      "grad_norm": 6.527960300445557,
      "learning_rate": 9.173772035361196e-06,
      "loss": 1.0893,
      "step": 31320
    },
    {
      "epoch": 2.4518704022538738,
      "grad_norm": 14.504951477050781,
      "learning_rate": 9.160694669665743e-06,
      "loss": 0.6028,
      "step": 31330
    },
    {
      "epoch": 2.4526529973391766,
      "grad_norm": 13.266035079956055,
      "learning_rate": 9.147617303970289e-06,
      "loss": 0.585,
      "step": 31340
    },
    {
      "epoch": 2.4534355924244795,
      "grad_norm": 5.2584991455078125,
      "learning_rate": 9.134539938274835e-06,
      "loss": 0.9494,
      "step": 31350
    },
    {
      "epoch": 2.4542181875097824,
      "grad_norm": 20.05630111694336,
      "learning_rate": 9.12146257257938e-06,
      "loss": 0.7172,
      "step": 31360
    },
    {
      "epoch": 2.4550007825950853,
      "grad_norm": 13.848949432373047,
      "learning_rate": 9.108385206883924e-06,
      "loss": 1.1992,
      "step": 31370
    },
    {
      "epoch": 2.455783377680388,
      "grad_norm": 5.7982072830200195,
      "learning_rate": 9.095307841188472e-06,
      "loss": 0.5809,
      "step": 31380
    },
    {
      "epoch": 2.456565972765691,
      "grad_norm": 3.398956060409546,
      "learning_rate": 9.082230475493017e-06,
      "loss": 0.4829,
      "step": 31390
    },
    {
      "epoch": 2.457348567850994,
      "grad_norm": 10.303606033325195,
      "learning_rate": 9.069153109797563e-06,
      "loss": 0.5532,
      "step": 31400
    },
    {
      "epoch": 2.4581311629362967,
      "grad_norm": 21.863584518432617,
      "learning_rate": 9.05607574410211e-06,
      "loss": 0.7791,
      "step": 31410
    },
    {
      "epoch": 2.4589137580215996,
      "grad_norm": 0.7159149646759033,
      "learning_rate": 9.042998378406654e-06,
      "loss": 0.7071,
      "step": 31420
    },
    {
      "epoch": 2.4596963531069025,
      "grad_norm": 19.533344268798828,
      "learning_rate": 9.0299210127112e-06,
      "loss": 0.8805,
      "step": 31430
    },
    {
      "epoch": 2.4604789481922054,
      "grad_norm": 1.2925578355789185,
      "learning_rate": 9.016843647015745e-06,
      "loss": 0.5241,
      "step": 31440
    },
    {
      "epoch": 2.4612615432775082,
      "grad_norm": 5.106130123138428,
      "learning_rate": 9.003766281320291e-06,
      "loss": 0.9155,
      "step": 31450
    },
    {
      "epoch": 2.462044138362811,
      "grad_norm": 3.9629783630371094,
      "learning_rate": 8.990688915624837e-06,
      "loss": 0.5641,
      "step": 31460
    },
    {
      "epoch": 2.462826733448114,
      "grad_norm": 10.881854057312012,
      "learning_rate": 8.977611549929382e-06,
      "loss": 0.6514,
      "step": 31470
    },
    {
      "epoch": 2.463609328533417,
      "grad_norm": 2.3472397327423096,
      "learning_rate": 8.964534184233928e-06,
      "loss": 0.9978,
      "step": 31480
    },
    {
      "epoch": 2.4643919236187197,
      "grad_norm": 3.0084238052368164,
      "learning_rate": 8.951456818538475e-06,
      "loss": 0.3495,
      "step": 31490
    },
    {
      "epoch": 2.4651745187040226,
      "grad_norm": 7.128630638122559,
      "learning_rate": 8.938379452843019e-06,
      "loss": 0.9089,
      "step": 31500
    },
    {
      "epoch": 2.4659571137893255,
      "grad_norm": 0.4740367531776428,
      "learning_rate": 8.925302087147565e-06,
      "loss": 0.588,
      "step": 31510
    },
    {
      "epoch": 2.4667397088746283,
      "grad_norm": 16.99712371826172,
      "learning_rate": 8.912224721452112e-06,
      "loss": 0.6036,
      "step": 31520
    },
    {
      "epoch": 2.467522303959931,
      "grad_norm": 9.081247329711914,
      "learning_rate": 8.899147355756656e-06,
      "loss": 0.7219,
      "step": 31530
    },
    {
      "epoch": 2.468304899045234,
      "grad_norm": 20.40117645263672,
      "learning_rate": 8.886069990061202e-06,
      "loss": 0.8105,
      "step": 31540
    },
    {
      "epoch": 2.469087494130537,
      "grad_norm": 13.856961250305176,
      "learning_rate": 8.872992624365749e-06,
      "loss": 0.9482,
      "step": 31550
    },
    {
      "epoch": 2.46987008921584,
      "grad_norm": 22.798582077026367,
      "learning_rate": 8.859915258670293e-06,
      "loss": 0.8832,
      "step": 31560
    },
    {
      "epoch": 2.4706526843011427,
      "grad_norm": 2.520781993865967,
      "learning_rate": 8.84683789297484e-06,
      "loss": 0.6404,
      "step": 31570
    },
    {
      "epoch": 2.4714352793864456,
      "grad_norm": 20.526485443115234,
      "learning_rate": 8.833760527279384e-06,
      "loss": 0.8438,
      "step": 31580
    },
    {
      "epoch": 2.4722178744717485,
      "grad_norm": 0.5928929448127747,
      "learning_rate": 8.820683161583932e-06,
      "loss": 0.4986,
      "step": 31590
    },
    {
      "epoch": 2.4730004695570513,
      "grad_norm": 16.787538528442383,
      "learning_rate": 8.807605795888477e-06,
      "loss": 0.9049,
      "step": 31600
    },
    {
      "epoch": 2.473783064642354,
      "grad_norm": 1.6677383184432983,
      "learning_rate": 8.794528430193021e-06,
      "loss": 1.0972,
      "step": 31610
    },
    {
      "epoch": 2.474565659727657,
      "grad_norm": 0.40457507967948914,
      "learning_rate": 8.78145106449757e-06,
      "loss": 0.4358,
      "step": 31620
    },
    {
      "epoch": 2.47534825481296,
      "grad_norm": 0.40989387035369873,
      "learning_rate": 8.768373698802114e-06,
      "loss": 0.8244,
      "step": 31630
    },
    {
      "epoch": 2.476130849898263,
      "grad_norm": 28.869569778442383,
      "learning_rate": 8.75529633310666e-06,
      "loss": 0.7695,
      "step": 31640
    },
    {
      "epoch": 2.4769134449835657,
      "grad_norm": 17.014266967773438,
      "learning_rate": 8.742218967411205e-06,
      "loss": 1.1468,
      "step": 31650
    },
    {
      "epoch": 2.477696040068868,
      "grad_norm": 8.686812400817871,
      "learning_rate": 8.729141601715751e-06,
      "loss": 0.576,
      "step": 31660
    },
    {
      "epoch": 2.478478635154171,
      "grad_norm": 29.842525482177734,
      "learning_rate": 8.716064236020297e-06,
      "loss": 0.9254,
      "step": 31670
    },
    {
      "epoch": 2.479261230239474,
      "grad_norm": 45.63397979736328,
      "learning_rate": 8.702986870324842e-06,
      "loss": 0.6956,
      "step": 31680
    },
    {
      "epoch": 2.4800438253247767,
      "grad_norm": 22.00835609436035,
      "learning_rate": 8.689909504629388e-06,
      "loss": 0.7233,
      "step": 31690
    },
    {
      "epoch": 2.4808264204100796,
      "grad_norm": 18.8168888092041,
      "learning_rate": 8.676832138933934e-06,
      "loss": 0.705,
      "step": 31700
    },
    {
      "epoch": 2.4816090154953825,
      "grad_norm": 1.5156739950180054,
      "learning_rate": 8.663754773238479e-06,
      "loss": 0.5535,
      "step": 31710
    },
    {
      "epoch": 2.4823916105806854,
      "grad_norm": 24.591001510620117,
      "learning_rate": 8.650677407543025e-06,
      "loss": 0.6954,
      "step": 31720
    },
    {
      "epoch": 2.4831742056659882,
      "grad_norm": 6.043793201446533,
      "learning_rate": 8.637600041847571e-06,
      "loss": 0.6439,
      "step": 31730
    },
    {
      "epoch": 2.483956800751291,
      "grad_norm": 17.689584732055664,
      "learning_rate": 8.624522676152116e-06,
      "loss": 0.7422,
      "step": 31740
    },
    {
      "epoch": 2.484739395836594,
      "grad_norm": 15.040742874145508,
      "learning_rate": 8.611445310456662e-06,
      "loss": 0.9838,
      "step": 31750
    },
    {
      "epoch": 2.485521990921897,
      "grad_norm": 17.099130630493164,
      "learning_rate": 8.598367944761208e-06,
      "loss": 0.7232,
      "step": 31760
    },
    {
      "epoch": 2.4863045860071997,
      "grad_norm": 17.36056137084961,
      "learning_rate": 8.585290579065753e-06,
      "loss": 0.8732,
      "step": 31770
    },
    {
      "epoch": 2.4870871810925026,
      "grad_norm": 1.98184335231781,
      "learning_rate": 8.5722132133703e-06,
      "loss": 0.4857,
      "step": 31780
    },
    {
      "epoch": 2.4878697761778055,
      "grad_norm": 0.5368244051933289,
      "learning_rate": 8.559135847674844e-06,
      "loss": 0.7545,
      "step": 31790
    },
    {
      "epoch": 2.4886523712631083,
      "grad_norm": 7.159245014190674,
      "learning_rate": 8.54605848197939e-06,
      "loss": 0.7562,
      "step": 31800
    },
    {
      "epoch": 2.489434966348411,
      "grad_norm": 12.431865692138672,
      "learning_rate": 8.532981116283936e-06,
      "loss": 0.3898,
      "step": 31810
    },
    {
      "epoch": 2.490217561433714,
      "grad_norm": 19.239044189453125,
      "learning_rate": 8.519903750588481e-06,
      "loss": 0.3945,
      "step": 31820
    },
    {
      "epoch": 2.491000156519017,
      "grad_norm": 9.894957542419434,
      "learning_rate": 8.506826384893029e-06,
      "loss": 0.5513,
      "step": 31830
    },
    {
      "epoch": 2.49178275160432,
      "grad_norm": 35.12620162963867,
      "learning_rate": 8.493749019197574e-06,
      "loss": 0.8848,
      "step": 31840
    },
    {
      "epoch": 2.4925653466896227,
      "grad_norm": 27.85025405883789,
      "learning_rate": 8.480671653502118e-06,
      "loss": 0.6948,
      "step": 31850
    },
    {
      "epoch": 2.4933479417749256,
      "grad_norm": 7.987176895141602,
      "learning_rate": 8.467594287806664e-06,
      "loss": 0.7377,
      "step": 31860
    },
    {
      "epoch": 2.4941305368602285,
      "grad_norm": 19.510234832763672,
      "learning_rate": 8.45451692211121e-06,
      "loss": 0.8354,
      "step": 31870
    },
    {
      "epoch": 2.4949131319455313,
      "grad_norm": 8.02720832824707,
      "learning_rate": 8.441439556415757e-06,
      "loss": 0.7478,
      "step": 31880
    },
    {
      "epoch": 2.495695727030834,
      "grad_norm": 23.989206314086914,
      "learning_rate": 8.428362190720301e-06,
      "loss": 0.7383,
      "step": 31890
    },
    {
      "epoch": 2.496478322116137,
      "grad_norm": 4.353612899780273,
      "learning_rate": 8.415284825024846e-06,
      "loss": 0.7655,
      "step": 31900
    },
    {
      "epoch": 2.49726091720144,
      "grad_norm": 18.283161163330078,
      "learning_rate": 8.402207459329394e-06,
      "loss": 0.486,
      "step": 31910
    },
    {
      "epoch": 2.498043512286743,
      "grad_norm": 58.5724983215332,
      "learning_rate": 8.389130093633939e-06,
      "loss": 0.8046,
      "step": 31920
    },
    {
      "epoch": 2.4988261073720457,
      "grad_norm": 21.99188232421875,
      "learning_rate": 8.376052727938483e-06,
      "loss": 0.9712,
      "step": 31930
    },
    {
      "epoch": 2.4996087024573486,
      "grad_norm": 12.901755332946777,
      "learning_rate": 8.362975362243031e-06,
      "loss": 0.5255,
      "step": 31940
    },
    {
      "epoch": 2.5003912975426514,
      "grad_norm": 12.740682601928711,
      "learning_rate": 8.349897996547576e-06,
      "loss": 0.7396,
      "step": 31950
    },
    {
      "epoch": 2.5011738926279543,
      "grad_norm": 20.436227798461914,
      "learning_rate": 8.336820630852122e-06,
      "loss": 0.7315,
      "step": 31960
    },
    {
      "epoch": 2.501956487713257,
      "grad_norm": 0.6875084638595581,
      "learning_rate": 8.323743265156667e-06,
      "loss": 0.5318,
      "step": 31970
    },
    {
      "epoch": 2.50273908279856,
      "grad_norm": 1.271857738494873,
      "learning_rate": 8.310665899461213e-06,
      "loss": 0.7382,
      "step": 31980
    },
    {
      "epoch": 2.503521677883863,
      "grad_norm": 8.123979568481445,
      "learning_rate": 8.297588533765759e-06,
      "loss": 1.0065,
      "step": 31990
    },
    {
      "epoch": 2.504304272969166,
      "grad_norm": 9.52841854095459,
      "learning_rate": 8.284511168070304e-06,
      "loss": 1.2005,
      "step": 32000
    },
    {
      "epoch": 2.5050868680544687,
      "grad_norm": 0.5434541702270508,
      "learning_rate": 8.27143380237485e-06,
      "loss": 0.7376,
      "step": 32010
    },
    {
      "epoch": 2.5058694631397715,
      "grad_norm": 7.713812828063965,
      "learning_rate": 8.258356436679396e-06,
      "loss": 0.4976,
      "step": 32020
    },
    {
      "epoch": 2.5066520582250744,
      "grad_norm": 27.72173309326172,
      "learning_rate": 8.24527907098394e-06,
      "loss": 0.5453,
      "step": 32030
    },
    {
      "epoch": 2.5074346533103773,
      "grad_norm": 22.745649337768555,
      "learning_rate": 8.232201705288487e-06,
      "loss": 1.0384,
      "step": 32040
    },
    {
      "epoch": 2.50821724839568,
      "grad_norm": 19.019407272338867,
      "learning_rate": 8.219124339593033e-06,
      "loss": 0.7896,
      "step": 32050
    },
    {
      "epoch": 2.508999843480983,
      "grad_norm": 6.062029838562012,
      "learning_rate": 8.206046973897578e-06,
      "loss": 0.9048,
      "step": 32060
    },
    {
      "epoch": 2.509782438566286,
      "grad_norm": 13.826373100280762,
      "learning_rate": 8.192969608202124e-06,
      "loss": 0.8504,
      "step": 32070
    },
    {
      "epoch": 2.510565033651589,
      "grad_norm": 7.268385887145996,
      "learning_rate": 8.17989224250667e-06,
      "loss": 0.8069,
      "step": 32080
    },
    {
      "epoch": 2.5113476287368917,
      "grad_norm": 3.784433603286743,
      "learning_rate": 8.166814876811215e-06,
      "loss": 0.6561,
      "step": 32090
    },
    {
      "epoch": 2.5121302238221945,
      "grad_norm": 5.333542346954346,
      "learning_rate": 8.153737511115761e-06,
      "loss": 0.6568,
      "step": 32100
    },
    {
      "epoch": 2.5129128189074974,
      "grad_norm": 2.4103894233703613,
      "learning_rate": 8.140660145420306e-06,
      "loss": 0.5268,
      "step": 32110
    },
    {
      "epoch": 2.5136954139928003,
      "grad_norm": 8.810480117797852,
      "learning_rate": 8.127582779724854e-06,
      "loss": 1.0571,
      "step": 32120
    },
    {
      "epoch": 2.514478009078103,
      "grad_norm": 10.47120475769043,
      "learning_rate": 8.114505414029398e-06,
      "loss": 0.6016,
      "step": 32130
    },
    {
      "epoch": 2.515260604163406,
      "grad_norm": 20.964515686035156,
      "learning_rate": 8.101428048333943e-06,
      "loss": 1.047,
      "step": 32140
    },
    {
      "epoch": 2.516043199248709,
      "grad_norm": 2.132219076156616,
      "learning_rate": 8.088350682638491e-06,
      "loss": 0.9117,
      "step": 32150
    },
    {
      "epoch": 2.5168257943340118,
      "grad_norm": 21.46027946472168,
      "learning_rate": 8.075273316943035e-06,
      "loss": 0.9667,
      "step": 32160
    },
    {
      "epoch": 2.5176083894193146,
      "grad_norm": 7.7289886474609375,
      "learning_rate": 8.062195951247582e-06,
      "loss": 1.0474,
      "step": 32170
    },
    {
      "epoch": 2.5183909845046175,
      "grad_norm": 3.17512845993042,
      "learning_rate": 8.049118585552126e-06,
      "loss": 0.7737,
      "step": 32180
    },
    {
      "epoch": 2.5191735795899204,
      "grad_norm": 6.8368120193481445,
      "learning_rate": 8.036041219856673e-06,
      "loss": 0.7254,
      "step": 32190
    },
    {
      "epoch": 2.5199561746752233,
      "grad_norm": 14.615065574645996,
      "learning_rate": 8.022963854161219e-06,
      "loss": 0.6053,
      "step": 32200
    },
    {
      "epoch": 2.520738769760526,
      "grad_norm": 1.6337316036224365,
      "learning_rate": 8.009886488465763e-06,
      "loss": 0.7895,
      "step": 32210
    },
    {
      "epoch": 2.521521364845829,
      "grad_norm": 4.531688213348389,
      "learning_rate": 7.99680912277031e-06,
      "loss": 0.8634,
      "step": 32220
    },
    {
      "epoch": 2.522303959931132,
      "grad_norm": 0.8607191443443298,
      "learning_rate": 7.983731757074856e-06,
      "loss": 0.7609,
      "step": 32230
    },
    {
      "epoch": 2.5230865550164348,
      "grad_norm": 15.034092903137207,
      "learning_rate": 7.9706543913794e-06,
      "loss": 0.8867,
      "step": 32240
    },
    {
      "epoch": 2.5238691501017376,
      "grad_norm": 0.9993868470191956,
      "learning_rate": 7.957577025683947e-06,
      "loss": 0.6265,
      "step": 32250
    },
    {
      "epoch": 2.5246517451870405,
      "grad_norm": 2.419583797454834,
      "learning_rate": 7.944499659988493e-06,
      "loss": 1.117,
      "step": 32260
    },
    {
      "epoch": 2.525434340272343,
      "grad_norm": 6.304348468780518,
      "learning_rate": 7.931422294293038e-06,
      "loss": 0.352,
      "step": 32270
    },
    {
      "epoch": 2.526216935357646,
      "grad_norm": 11.577851295471191,
      "learning_rate": 7.918344928597584e-06,
      "loss": 0.8797,
      "step": 32280
    },
    {
      "epoch": 2.5269995304429487,
      "grad_norm": 2.0015869140625,
      "learning_rate": 7.90526756290213e-06,
      "loss": 0.4684,
      "step": 32290
    },
    {
      "epoch": 2.5277821255282515,
      "grad_norm": 6.767703056335449,
      "learning_rate": 7.892190197206675e-06,
      "loss": 0.5183,
      "step": 32300
    },
    {
      "epoch": 2.5285647206135544,
      "grad_norm": 13.852494239807129,
      "learning_rate": 7.879112831511221e-06,
      "loss": 0.4179,
      "step": 32310
    },
    {
      "epoch": 2.5293473156988573,
      "grad_norm": 6.536184310913086,
      "learning_rate": 7.866035465815766e-06,
      "loss": 0.7293,
      "step": 32320
    },
    {
      "epoch": 2.53012991078416,
      "grad_norm": 16.69721221923828,
      "learning_rate": 7.852958100120312e-06,
      "loss": 0.8453,
      "step": 32330
    },
    {
      "epoch": 2.530912505869463,
      "grad_norm": 16.249847412109375,
      "learning_rate": 7.839880734424858e-06,
      "loss": 0.9403,
      "step": 32340
    },
    {
      "epoch": 2.531695100954766,
      "grad_norm": 6.956534385681152,
      "learning_rate": 7.826803368729403e-06,
      "loss": 0.4261,
      "step": 32350
    },
    {
      "epoch": 2.532477696040069,
      "grad_norm": 2.7835705280303955,
      "learning_rate": 7.81372600303395e-06,
      "loss": 0.6176,
      "step": 32360
    },
    {
      "epoch": 2.5332602911253717,
      "grad_norm": 14.83127212524414,
      "learning_rate": 7.800648637338495e-06,
      "loss": 0.5269,
      "step": 32370
    },
    {
      "epoch": 2.5340428862106745,
      "grad_norm": 5.56640625,
      "learning_rate": 7.78757127164304e-06,
      "loss": 0.5357,
      "step": 32380
    },
    {
      "epoch": 2.5348254812959774,
      "grad_norm": 11.228816032409668,
      "learning_rate": 7.774493905947586e-06,
      "loss": 1.0227,
      "step": 32390
    },
    {
      "epoch": 2.5356080763812803,
      "grad_norm": 1.1954811811447144,
      "learning_rate": 7.761416540252132e-06,
      "loss": 0.5598,
      "step": 32400
    },
    {
      "epoch": 2.536390671466583,
      "grad_norm": 12.06454086303711,
      "learning_rate": 7.748339174556679e-06,
      "loss": 0.4218,
      "step": 32410
    },
    {
      "epoch": 2.537173266551886,
      "grad_norm": 6.061407089233398,
      "learning_rate": 7.735261808861223e-06,
      "loss": 0.2669,
      "step": 32420
    },
    {
      "epoch": 2.537955861637189,
      "grad_norm": 1.9098435640335083,
      "learning_rate": 7.72218444316577e-06,
      "loss": 0.7811,
      "step": 32430
    },
    {
      "epoch": 2.5387384567224918,
      "grad_norm": 6.435410022735596,
      "learning_rate": 7.709107077470316e-06,
      "loss": 0.2838,
      "step": 32440
    },
    {
      "epoch": 2.5395210518077946,
      "grad_norm": 18.934436798095703,
      "learning_rate": 7.69602971177486e-06,
      "loss": 0.8856,
      "step": 32450
    },
    {
      "epoch": 2.5403036468930975,
      "grad_norm": 8.180354118347168,
      "learning_rate": 7.682952346079405e-06,
      "loss": 0.9417,
      "step": 32460
    },
    {
      "epoch": 2.5410862419784004,
      "grad_norm": 8.263339042663574,
      "learning_rate": 7.669874980383953e-06,
      "loss": 0.7872,
      "step": 32470
    },
    {
      "epoch": 2.5418688370637033,
      "grad_norm": 1.190590262413025,
      "learning_rate": 7.656797614688497e-06,
      "loss": 0.7764,
      "step": 32480
    },
    {
      "epoch": 2.542651432149006,
      "grad_norm": 9.019638061523438,
      "learning_rate": 7.643720248993044e-06,
      "loss": 0.718,
      "step": 32490
    },
    {
      "epoch": 2.543434027234309,
      "grad_norm": 22.97794532775879,
      "learning_rate": 7.63064288329759e-06,
      "loss": 0.7455,
      "step": 32500
    },
    {
      "epoch": 2.544216622319612,
      "grad_norm": 18.92841911315918,
      "learning_rate": 7.617565517602135e-06,
      "loss": 0.9705,
      "step": 32510
    },
    {
      "epoch": 2.5449992174049147,
      "grad_norm": 0.2811354100704193,
      "learning_rate": 7.60448815190668e-06,
      "loss": 0.982,
      "step": 32520
    },
    {
      "epoch": 2.5457818124902176,
      "grad_norm": 0.753531813621521,
      "learning_rate": 7.591410786211225e-06,
      "loss": 0.5531,
      "step": 32530
    },
    {
      "epoch": 2.5465644075755205,
      "grad_norm": 8.35390567779541,
      "learning_rate": 7.578333420515772e-06,
      "loss": 0.543,
      "step": 32540
    },
    {
      "epoch": 2.5473470026608234,
      "grad_norm": 5.417882442474365,
      "learning_rate": 7.565256054820318e-06,
      "loss": 0.8531,
      "step": 32550
    },
    {
      "epoch": 2.5481295977461262,
      "grad_norm": 22.54418182373047,
      "learning_rate": 7.552178689124862e-06,
      "loss": 0.484,
      "step": 32560
    },
    {
      "epoch": 2.548912192831429,
      "grad_norm": 15.071898460388184,
      "learning_rate": 7.539101323429408e-06,
      "loss": 0.7586,
      "step": 32570
    },
    {
      "epoch": 2.549694787916732,
      "grad_norm": 3.555537700653076,
      "learning_rate": 7.526023957733955e-06,
      "loss": 0.4381,
      "step": 32580
    },
    {
      "epoch": 2.550477383002035,
      "grad_norm": 16.9280948638916,
      "learning_rate": 7.5129465920385e-06,
      "loss": 0.4558,
      "step": 32590
    },
    {
      "epoch": 2.5512599780873377,
      "grad_norm": 10.393722534179688,
      "learning_rate": 7.499869226343046e-06,
      "loss": 1.056,
      "step": 32600
    },
    {
      "epoch": 2.5520425731726406,
      "grad_norm": 14.328028678894043,
      "learning_rate": 7.486791860647592e-06,
      "loss": 0.7675,
      "step": 32610
    },
    {
      "epoch": 2.5528251682579435,
      "grad_norm": 15.673120498657227,
      "learning_rate": 7.473714494952137e-06,
      "loss": 0.7308,
      "step": 32620
    },
    {
      "epoch": 2.5536077633432464,
      "grad_norm": 0.4299066662788391,
      "learning_rate": 7.460637129256683e-06,
      "loss": 0.6365,
      "step": 32630
    },
    {
      "epoch": 2.5543903584285492,
      "grad_norm": 8.647504806518555,
      "learning_rate": 7.447559763561228e-06,
      "loss": 0.8,
      "step": 32640
    },
    {
      "epoch": 2.555172953513852,
      "grad_norm": 17.805086135864258,
      "learning_rate": 7.4344823978657745e-06,
      "loss": 0.9398,
      "step": 32650
    },
    {
      "epoch": 2.5559555485991545,
      "grad_norm": 5.845183372497559,
      "learning_rate": 7.42140503217032e-06,
      "loss": 0.43,
      "step": 32660
    },
    {
      "epoch": 2.5567381436844574,
      "grad_norm": 2.4258081912994385,
      "learning_rate": 7.408327666474865e-06,
      "loss": 0.6607,
      "step": 32670
    },
    {
      "epoch": 2.5575207387697603,
      "grad_norm": 17.046628952026367,
      "learning_rate": 7.395250300779412e-06,
      "loss": 0.4973,
      "step": 32680
    },
    {
      "epoch": 2.558303333855063,
      "grad_norm": 15.116846084594727,
      "learning_rate": 7.382172935083957e-06,
      "loss": 0.83,
      "step": 32690
    },
    {
      "epoch": 2.559085928940366,
      "grad_norm": 9.062068939208984,
      "learning_rate": 7.3690955693885025e-06,
      "loss": 0.6329,
      "step": 32700
    },
    {
      "epoch": 2.559868524025669,
      "grad_norm": 0.9695742130279541,
      "learning_rate": 7.356018203693048e-06,
      "loss": 0.2283,
      "step": 32710
    },
    {
      "epoch": 2.5606511191109718,
      "grad_norm": 17.0100040435791,
      "learning_rate": 7.342940837997594e-06,
      "loss": 1.0916,
      "step": 32720
    },
    {
      "epoch": 2.5614337141962746,
      "grad_norm": 8.559564590454102,
      "learning_rate": 7.32986347230214e-06,
      "loss": 0.6703,
      "step": 32730
    },
    {
      "epoch": 2.5622163092815775,
      "grad_norm": 31.80432891845703,
      "learning_rate": 7.316786106606685e-06,
      "loss": 0.9495,
      "step": 32740
    },
    {
      "epoch": 2.5629989043668804,
      "grad_norm": 6.002424716949463,
      "learning_rate": 7.303708740911232e-06,
      "loss": 0.579,
      "step": 32750
    },
    {
      "epoch": 2.5637814994521833,
      "grad_norm": 21.18274688720703,
      "learning_rate": 7.290631375215777e-06,
      "loss": 0.7938,
      "step": 32760
    },
    {
      "epoch": 2.564564094537486,
      "grad_norm": 8.272347450256348,
      "learning_rate": 7.277554009520322e-06,
      "loss": 0.7874,
      "step": 32770
    },
    {
      "epoch": 2.565346689622789,
      "grad_norm": 10.985671043395996,
      "learning_rate": 7.2644766438248675e-06,
      "loss": 0.8887,
      "step": 32780
    },
    {
      "epoch": 2.566129284708092,
      "grad_norm": 16.500871658325195,
      "learning_rate": 7.251399278129415e-06,
      "loss": 0.8229,
      "step": 32790
    },
    {
      "epoch": 2.5669118797933947,
      "grad_norm": 5.596132755279541,
      "learning_rate": 7.238321912433959e-06,
      "loss": 1.0384,
      "step": 32800
    },
    {
      "epoch": 2.5676944748786976,
      "grad_norm": 33.69523239135742,
      "learning_rate": 7.225244546738505e-06,
      "loss": 0.893,
      "step": 32810
    },
    {
      "epoch": 2.5684770699640005,
      "grad_norm": 7.226828575134277,
      "learning_rate": 7.212167181043052e-06,
      "loss": 0.7263,
      "step": 32820
    },
    {
      "epoch": 2.5692596650493034,
      "grad_norm": 4.991841793060303,
      "learning_rate": 7.199089815347597e-06,
      "loss": 0.6373,
      "step": 32830
    },
    {
      "epoch": 2.5700422601346062,
      "grad_norm": 1.6649580001831055,
      "learning_rate": 7.186012449652143e-06,
      "loss": 0.6425,
      "step": 32840
    },
    {
      "epoch": 2.570824855219909,
      "grad_norm": 17.293880462646484,
      "learning_rate": 7.172935083956687e-06,
      "loss": 0.8801,
      "step": 32850
    },
    {
      "epoch": 2.571607450305212,
      "grad_norm": 9.285714149475098,
      "learning_rate": 7.159857718261234e-06,
      "loss": 0.6029,
      "step": 32860
    },
    {
      "epoch": 2.572390045390515,
      "grad_norm": 17.2375431060791,
      "learning_rate": 7.14678035256578e-06,
      "loss": 0.9215,
      "step": 32870
    },
    {
      "epoch": 2.5731726404758177,
      "grad_norm": 2.195262908935547,
      "learning_rate": 7.133702986870325e-06,
      "loss": 0.7886,
      "step": 32880
    },
    {
      "epoch": 2.5739552355611206,
      "grad_norm": 0.8458954691886902,
      "learning_rate": 7.120625621174871e-06,
      "loss": 0.6741,
      "step": 32890
    },
    {
      "epoch": 2.5747378306464235,
      "grad_norm": 13.181777954101562,
      "learning_rate": 7.107548255479417e-06,
      "loss": 0.8866,
      "step": 32900
    },
    {
      "epoch": 2.5755204257317263,
      "grad_norm": 16.30411148071289,
      "learning_rate": 7.094470889783962e-06,
      "loss": 0.8054,
      "step": 32910
    },
    {
      "epoch": 2.576303020817029,
      "grad_norm": 2.1934893131256104,
      "learning_rate": 7.081393524088508e-06,
      "loss": 0.7414,
      "step": 32920
    },
    {
      "epoch": 2.577085615902332,
      "grad_norm": 7.233014106750488,
      "learning_rate": 7.068316158393054e-06,
      "loss": 0.6945,
      "step": 32930
    },
    {
      "epoch": 2.577868210987635,
      "grad_norm": 2.213125228881836,
      "learning_rate": 7.055238792697599e-06,
      "loss": 0.722,
      "step": 32940
    },
    {
      "epoch": 2.578650806072938,
      "grad_norm": 34.82026290893555,
      "learning_rate": 7.042161427002145e-06,
      "loss": 0.7554,
      "step": 32950
    },
    {
      "epoch": 2.5794334011582407,
      "grad_norm": 6.025775909423828,
      "learning_rate": 7.029084061306691e-06,
      "loss": 0.6454,
      "step": 32960
    },
    {
      "epoch": 2.5802159962435436,
      "grad_norm": 5.201601982116699,
      "learning_rate": 7.016006695611236e-06,
      "loss": 0.759,
      "step": 32970
    },
    {
      "epoch": 2.5809985913288465,
      "grad_norm": 2.070659637451172,
      "learning_rate": 7.002929329915782e-06,
      "loss": 0.306,
      "step": 32980
    },
    {
      "epoch": 2.5817811864141493,
      "grad_norm": 17.710857391357422,
      "learning_rate": 6.989851964220327e-06,
      "loss": 1.0944,
      "step": 32990
    },
    {
      "epoch": 2.582563781499452,
      "grad_norm": 6.109461307525635,
      "learning_rate": 6.9767745985248735e-06,
      "loss": 1.0581,
      "step": 33000
    },
    {
      "epoch": 2.583346376584755,
      "grad_norm": 8.126009941101074,
      "learning_rate": 6.963697232829419e-06,
      "loss": 0.7241,
      "step": 33010
    },
    {
      "epoch": 2.584128971670058,
      "grad_norm": 13.292947769165039,
      "learning_rate": 6.950619867133964e-06,
      "loss": 0.5445,
      "step": 33020
    },
    {
      "epoch": 2.584911566755361,
      "grad_norm": 0.7024896740913391,
      "learning_rate": 6.9375425014385115e-06,
      "loss": 0.5562,
      "step": 33030
    },
    {
      "epoch": 2.5856941618406637,
      "grad_norm": 12.53141975402832,
      "learning_rate": 6.924465135743057e-06,
      "loss": 0.6498,
      "step": 33040
    },
    {
      "epoch": 2.5864767569259666,
      "grad_norm": 9.889642715454102,
      "learning_rate": 6.9113877700476015e-06,
      "loss": 0.6969,
      "step": 33050
    },
    {
      "epoch": 2.5872593520112694,
      "grad_norm": 1.4434373378753662,
      "learning_rate": 6.898310404352147e-06,
      "loss": 0.8302,
      "step": 33060
    },
    {
      "epoch": 2.5880419470965723,
      "grad_norm": 21.145610809326172,
      "learning_rate": 6.885233038656694e-06,
      "loss": 0.5532,
      "step": 33070
    },
    {
      "epoch": 2.588824542181875,
      "grad_norm": 43.53620529174805,
      "learning_rate": 6.872155672961239e-06,
      "loss": 0.9685,
      "step": 33080
    },
    {
      "epoch": 2.589607137267178,
      "grad_norm": 9.543582916259766,
      "learning_rate": 6.859078307265784e-06,
      "loss": 1.0409,
      "step": 33090
    },
    {
      "epoch": 2.590389732352481,
      "grad_norm": 16.745101928710938,
      "learning_rate": 6.846000941570331e-06,
      "loss": 0.7434,
      "step": 33100
    },
    {
      "epoch": 2.591172327437784,
      "grad_norm": 12.927112579345703,
      "learning_rate": 6.8329235758748765e-06,
      "loss": 0.5435,
      "step": 33110
    },
    {
      "epoch": 2.5919549225230867,
      "grad_norm": 12.728775024414062,
      "learning_rate": 6.819846210179422e-06,
      "loss": 0.9032,
      "step": 33120
    },
    {
      "epoch": 2.5927375176083896,
      "grad_norm": 3.379873037338257,
      "learning_rate": 6.806768844483967e-06,
      "loss": 0.6527,
      "step": 33130
    },
    {
      "epoch": 2.5935201126936924,
      "grad_norm": 6.401591777801514,
      "learning_rate": 6.793691478788514e-06,
      "loss": 0.7561,
      "step": 33140
    },
    {
      "epoch": 2.5943027077789953,
      "grad_norm": 19.384206771850586,
      "learning_rate": 6.780614113093059e-06,
      "loss": 0.649,
      "step": 33150
    },
    {
      "epoch": 2.595085302864298,
      "grad_norm": 3.1531736850738525,
      "learning_rate": 6.7675367473976045e-06,
      "loss": 0.6772,
      "step": 33160
    },
    {
      "epoch": 2.595867897949601,
      "grad_norm": 6.712210655212402,
      "learning_rate": 6.754459381702151e-06,
      "loss": 0.7753,
      "step": 33170
    },
    {
      "epoch": 2.596650493034904,
      "grad_norm": 9.33231258392334,
      "learning_rate": 6.741382016006696e-06,
      "loss": 0.562,
      "step": 33180
    },
    {
      "epoch": 2.597433088120207,
      "grad_norm": 15.960693359375,
      "learning_rate": 6.7283046503112416e-06,
      "loss": 0.6847,
      "step": 33190
    },
    {
      "epoch": 2.5982156832055097,
      "grad_norm": 9.782405853271484,
      "learning_rate": 6.715227284615787e-06,
      "loss": 0.7373,
      "step": 33200
    },
    {
      "epoch": 2.5989982782908125,
      "grad_norm": 8.069889068603516,
      "learning_rate": 6.702149918920333e-06,
      "loss": 0.4836,
      "step": 33210
    },
    {
      "epoch": 2.5997808733761154,
      "grad_norm": 1.057631254196167,
      "learning_rate": 6.689072553224879e-06,
      "loss": 0.587,
      "step": 33220
    },
    {
      "epoch": 2.6005634684614183,
      "grad_norm": 8.793396949768066,
      "learning_rate": 6.675995187529424e-06,
      "loss": 0.6884,
      "step": 33230
    },
    {
      "epoch": 2.601346063546721,
      "grad_norm": 20.934450149536133,
      "learning_rate": 6.6629178218339695e-06,
      "loss": 1.0039,
      "step": 33240
    },
    {
      "epoch": 2.602128658632024,
      "grad_norm": 25.35379409790039,
      "learning_rate": 6.649840456138516e-06,
      "loss": 0.7759,
      "step": 33250
    },
    {
      "epoch": 2.602911253717327,
      "grad_norm": 0.614309549331665,
      "learning_rate": 6.636763090443061e-06,
      "loss": 0.6053,
      "step": 33260
    },
    {
      "epoch": 2.6036938488026298,
      "grad_norm": 1.059214472770691,
      "learning_rate": 6.623685724747607e-06,
      "loss": 0.4109,
      "step": 33270
    },
    {
      "epoch": 2.6044764438879326,
      "grad_norm": 8.553698539733887,
      "learning_rate": 6.610608359052154e-06,
      "loss": 1.0171,
      "step": 33280
    },
    {
      "epoch": 2.6052590389732355,
      "grad_norm": 10.337935447692871,
      "learning_rate": 6.597530993356698e-06,
      "loss": 0.5207,
      "step": 33290
    },
    {
      "epoch": 2.606041634058538,
      "grad_norm": 13.166960716247559,
      "learning_rate": 6.584453627661244e-06,
      "loss": 1.2185,
      "step": 33300
    },
    {
      "epoch": 2.606824229143841,
      "grad_norm": 8.349058151245117,
      "learning_rate": 6.571376261965789e-06,
      "loss": 0.8213,
      "step": 33310
    },
    {
      "epoch": 2.6076068242291437,
      "grad_norm": 13.141828536987305,
      "learning_rate": 6.558298896270336e-06,
      "loss": 0.6619,
      "step": 33320
    },
    {
      "epoch": 2.6083894193144466,
      "grad_norm": 22.99647331237793,
      "learning_rate": 6.545221530574881e-06,
      "loss": 0.6055,
      "step": 33330
    },
    {
      "epoch": 2.6091720143997494,
      "grad_norm": 13.278194427490234,
      "learning_rate": 6.532144164879426e-06,
      "loss": 1.0619,
      "step": 33340
    },
    {
      "epoch": 2.6099546094850523,
      "grad_norm": 15.94066333770752,
      "learning_rate": 6.519066799183973e-06,
      "loss": 1.1748,
      "step": 33350
    },
    {
      "epoch": 2.610737204570355,
      "grad_norm": 1.990920901298523,
      "learning_rate": 6.505989433488519e-06,
      "loss": 0.6998,
      "step": 33360
    },
    {
      "epoch": 2.611519799655658,
      "grad_norm": 1.1447454690933228,
      "learning_rate": 6.492912067793064e-06,
      "loss": 0.723,
      "step": 33370
    },
    {
      "epoch": 2.612302394740961,
      "grad_norm": 0.6840011477470398,
      "learning_rate": 6.479834702097609e-06,
      "loss": 0.4613,
      "step": 33380
    },
    {
      "epoch": 2.613084989826264,
      "grad_norm": 6.298853874206543,
      "learning_rate": 6.466757336402156e-06,
      "loss": 0.7485,
      "step": 33390
    },
    {
      "epoch": 2.6138675849115667,
      "grad_norm": 11.26960277557373,
      "learning_rate": 6.453679970706701e-06,
      "loss": 0.8145,
      "step": 33400
    },
    {
      "epoch": 2.6146501799968696,
      "grad_norm": 7.319981098175049,
      "learning_rate": 6.440602605011247e-06,
      "loss": 0.8478,
      "step": 33410
    },
    {
      "epoch": 2.6154327750821724,
      "grad_norm": 13.057417869567871,
      "learning_rate": 6.427525239315793e-06,
      "loss": 1.0856,
      "step": 33420
    },
    {
      "epoch": 2.6162153701674753,
      "grad_norm": 0.8143628239631653,
      "learning_rate": 6.414447873620338e-06,
      "loss": 0.5544,
      "step": 33430
    },
    {
      "epoch": 2.616997965252778,
      "grad_norm": 19.406457901000977,
      "learning_rate": 6.401370507924884e-06,
      "loss": 0.7189,
      "step": 33440
    },
    {
      "epoch": 2.617780560338081,
      "grad_norm": 6.702972888946533,
      "learning_rate": 6.388293142229429e-06,
      "loss": 0.8618,
      "step": 33450
    },
    {
      "epoch": 2.618563155423384,
      "grad_norm": 7.997715473175049,
      "learning_rate": 6.3752157765339755e-06,
      "loss": 0.6755,
      "step": 33460
    },
    {
      "epoch": 2.619345750508687,
      "grad_norm": 2.0654754638671875,
      "learning_rate": 6.362138410838521e-06,
      "loss": 0.7757,
      "step": 33470
    },
    {
      "epoch": 2.6201283455939897,
      "grad_norm": 4.009860992431641,
      "learning_rate": 6.349061045143066e-06,
      "loss": 0.8582,
      "step": 33480
    },
    {
      "epoch": 2.6209109406792925,
      "grad_norm": 4.828362941741943,
      "learning_rate": 6.335983679447613e-06,
      "loss": 0.5741,
      "step": 33490
    },
    {
      "epoch": 2.6216935357645954,
      "grad_norm": 6.387131690979004,
      "learning_rate": 6.322906313752158e-06,
      "loss": 0.7347,
      "step": 33500
    },
    {
      "epoch": 2.6224761308498983,
      "grad_norm": 7.132858753204346,
      "learning_rate": 6.3098289480567035e-06,
      "loss": 0.7475,
      "step": 33510
    },
    {
      "epoch": 2.623258725935201,
      "grad_norm": 7.3209919929504395,
      "learning_rate": 6.296751582361249e-06,
      "loss": 0.5061,
      "step": 33520
    },
    {
      "epoch": 2.624041321020504,
      "grad_norm": 19.37627410888672,
      "learning_rate": 6.283674216665795e-06,
      "loss": 1.029,
      "step": 33530
    },
    {
      "epoch": 2.624823916105807,
      "grad_norm": 4.118471145629883,
      "learning_rate": 6.2705968509703406e-06,
      "loss": 0.7152,
      "step": 33540
    },
    {
      "epoch": 2.6256065111911098,
      "grad_norm": 11.544184684753418,
      "learning_rate": 6.257519485274886e-06,
      "loss": 0.6421,
      "step": 33550
    },
    {
      "epoch": 2.6263891062764126,
      "grad_norm": 74.74986267089844,
      "learning_rate": 6.244442119579432e-06,
      "loss": 0.7352,
      "step": 33560
    },
    {
      "epoch": 2.6271717013617155,
      "grad_norm": 9.823049545288086,
      "learning_rate": 6.2313647538839785e-06,
      "loss": 0.622,
      "step": 33570
    },
    {
      "epoch": 2.6279542964470184,
      "grad_norm": 8.065079689025879,
      "learning_rate": 6.218287388188523e-06,
      "loss": 0.5249,
      "step": 33580
    },
    {
      "epoch": 2.6287368915323213,
      "grad_norm": 3.6866910457611084,
      "learning_rate": 6.205210022493069e-06,
      "loss": 0.5482,
      "step": 33590
    },
    {
      "epoch": 2.629519486617624,
      "grad_norm": 15.242494583129883,
      "learning_rate": 6.192132656797615e-06,
      "loss": 0.8049,
      "step": 33600
    },
    {
      "epoch": 2.630302081702927,
      "grad_norm": 2.978766918182373,
      "learning_rate": 6.179055291102161e-06,
      "loss": 0.5018,
      "step": 33610
    },
    {
      "epoch": 2.63108467678823,
      "grad_norm": 13.925713539123535,
      "learning_rate": 6.1659779254067065e-06,
      "loss": 1.0916,
      "step": 33620
    },
    {
      "epoch": 2.6318672718735328,
      "grad_norm": 27.877046585083008,
      "learning_rate": 6.152900559711252e-06,
      "loss": 0.7439,
      "step": 33630
    },
    {
      "epoch": 2.6326498669588356,
      "grad_norm": 11.179394721984863,
      "learning_rate": 6.139823194015798e-06,
      "loss": 0.8607,
      "step": 33640
    },
    {
      "epoch": 2.6334324620441385,
      "grad_norm": 0.8525254726409912,
      "learning_rate": 6.1267458283203436e-06,
      "loss": 0.9659,
      "step": 33650
    },
    {
      "epoch": 2.6342150571294414,
      "grad_norm": 11.008577346801758,
      "learning_rate": 6.113668462624889e-06,
      "loss": 0.6964,
      "step": 33660
    },
    {
      "epoch": 2.6349976522147442,
      "grad_norm": 0.46373146772384644,
      "learning_rate": 6.100591096929434e-06,
      "loss": 0.3651,
      "step": 33670
    },
    {
      "epoch": 2.635780247300047,
      "grad_norm": 7.124873161315918,
      "learning_rate": 6.087513731233981e-06,
      "loss": 0.399,
      "step": 33680
    },
    {
      "epoch": 2.6365628423853495,
      "grad_norm": 0.667665958404541,
      "learning_rate": 6.074436365538527e-06,
      "loss": 0.5046,
      "step": 33690
    },
    {
      "epoch": 2.6373454374706524,
      "grad_norm": 3.6071505546569824,
      "learning_rate": 6.0613589998430715e-06,
      "loss": 0.7373,
      "step": 33700
    },
    {
      "epoch": 2.6381280325559553,
      "grad_norm": 19.26127052307129,
      "learning_rate": 6.048281634147618e-06,
      "loss": 0.6664,
      "step": 33710
    },
    {
      "epoch": 2.638910627641258,
      "grad_norm": 6.235515594482422,
      "learning_rate": 6.035204268452163e-06,
      "loss": 0.6536,
      "step": 33720
    },
    {
      "epoch": 2.639693222726561,
      "grad_norm": 7.750848293304443,
      "learning_rate": 6.0221269027567095e-06,
      "loss": 0.6668,
      "step": 33730
    },
    {
      "epoch": 2.640475817811864,
      "grad_norm": 12.478826522827148,
      "learning_rate": 6.009049537061254e-06,
      "loss": 0.7115,
      "step": 33740
    },
    {
      "epoch": 2.641258412897167,
      "grad_norm": 12.31501293182373,
      "learning_rate": 5.9959721713658e-06,
      "loss": 0.6568,
      "step": 33750
    },
    {
      "epoch": 2.6420410079824697,
      "grad_norm": 1.3561756610870361,
      "learning_rate": 5.982894805670346e-06,
      "loss": 0.5751,
      "step": 33760
    },
    {
      "epoch": 2.6428236030677725,
      "grad_norm": 21.62635612487793,
      "learning_rate": 5.969817439974892e-06,
      "loss": 0.9142,
      "step": 33770
    },
    {
      "epoch": 2.6436061981530754,
      "grad_norm": 6.1137237548828125,
      "learning_rate": 5.956740074279437e-06,
      "loss": 0.8858,
      "step": 33780
    },
    {
      "epoch": 2.6443887932383783,
      "grad_norm": 11.21961784362793,
      "learning_rate": 5.943662708583983e-06,
      "loss": 0.7184,
      "step": 33790
    },
    {
      "epoch": 2.645171388323681,
      "grad_norm": 7.075743675231934,
      "learning_rate": 5.930585342888529e-06,
      "loss": 0.4949,
      "step": 33800
    },
    {
      "epoch": 2.645953983408984,
      "grad_norm": 5.290485382080078,
      "learning_rate": 5.9175079771930745e-06,
      "loss": 0.5477,
      "step": 33810
    },
    {
      "epoch": 2.646736578494287,
      "grad_norm": 8.952984809875488,
      "learning_rate": 5.90443061149762e-06,
      "loss": 0.7323,
      "step": 33820
    },
    {
      "epoch": 2.6475191735795898,
      "grad_norm": 7.082942485809326,
      "learning_rate": 5.891353245802165e-06,
      "loss": 0.9095,
      "step": 33830
    },
    {
      "epoch": 2.6483017686648926,
      "grad_norm": 24.46170425415039,
      "learning_rate": 5.878275880106712e-06,
      "loss": 1.0019,
      "step": 33840
    },
    {
      "epoch": 2.6490843637501955,
      "grad_norm": 14.414262771606445,
      "learning_rate": 5.865198514411258e-06,
      "loss": 0.6799,
      "step": 33850
    },
    {
      "epoch": 2.6498669588354984,
      "grad_norm": 11.708986282348633,
      "learning_rate": 5.8521211487158025e-06,
      "loss": 1.0637,
      "step": 33860
    },
    {
      "epoch": 2.6506495539208013,
      "grad_norm": 6.6502814292907715,
      "learning_rate": 5.839043783020349e-06,
      "loss": 0.5916,
      "step": 33870
    },
    {
      "epoch": 2.651432149006104,
      "grad_norm": 12.423233985900879,
      "learning_rate": 5.825966417324894e-06,
      "loss": 0.7347,
      "step": 33880
    },
    {
      "epoch": 2.652214744091407,
      "grad_norm": 14.423246383666992,
      "learning_rate": 5.81288905162944e-06,
      "loss": 0.4385,
      "step": 33890
    },
    {
      "epoch": 2.65299733917671,
      "grad_norm": 0.2683638036251068,
      "learning_rate": 5.799811685933986e-06,
      "loss": 0.762,
      "step": 33900
    },
    {
      "epoch": 2.6537799342620128,
      "grad_norm": 6.031275749206543,
      "learning_rate": 5.786734320238531e-06,
      "loss": 0.8408,
      "step": 33910
    },
    {
      "epoch": 2.6545625293473156,
      "grad_norm": 7.939637660980225,
      "learning_rate": 5.7736569545430775e-06,
      "loss": 0.6321,
      "step": 33920
    },
    {
      "epoch": 2.6553451244326185,
      "grad_norm": 2.987109661102295,
      "learning_rate": 5.760579588847623e-06,
      "loss": 0.8298,
      "step": 33930
    },
    {
      "epoch": 2.6561277195179214,
      "grad_norm": 36.416709899902344,
      "learning_rate": 5.747502223152168e-06,
      "loss": 0.8916,
      "step": 33940
    },
    {
      "epoch": 2.6569103146032242,
      "grad_norm": 15.124258995056152,
      "learning_rate": 5.734424857456714e-06,
      "loss": 0.6673,
      "step": 33950
    },
    {
      "epoch": 2.657692909688527,
      "grad_norm": 6.831678867340088,
      "learning_rate": 5.72134749176126e-06,
      "loss": 0.6492,
      "step": 33960
    },
    {
      "epoch": 2.65847550477383,
      "grad_norm": 15.634554862976074,
      "learning_rate": 5.7082701260658055e-06,
      "loss": 0.5834,
      "step": 33970
    },
    {
      "epoch": 2.659258099859133,
      "grad_norm": 15.23625659942627,
      "learning_rate": 5.695192760370351e-06,
      "loss": 0.7711,
      "step": 33980
    },
    {
      "epoch": 2.6600406949444357,
      "grad_norm": 22.983205795288086,
      "learning_rate": 5.682115394674897e-06,
      "loss": 0.6485,
      "step": 33990
    },
    {
      "epoch": 2.6608232900297386,
      "grad_norm": 7.566847801208496,
      "learning_rate": 5.6690380289794426e-06,
      "loss": 0.346,
      "step": 34000
    },
    {
      "epoch": 2.6616058851150415,
      "grad_norm": 45.72237014770508,
      "learning_rate": 5.655960663283989e-06,
      "loss": 0.643,
      "step": 34010
    },
    {
      "epoch": 2.6623884802003444,
      "grad_norm": 16.55185317993164,
      "learning_rate": 5.642883297588534e-06,
      "loss": 0.9506,
      "step": 34020
    },
    {
      "epoch": 2.6631710752856472,
      "grad_norm": 10.80521297454834,
      "learning_rate": 5.62980593189308e-06,
      "loss": 0.8396,
      "step": 34030
    },
    {
      "epoch": 2.66395367037095,
      "grad_norm": 8.830947875976562,
      "learning_rate": 5.616728566197625e-06,
      "loss": 0.5419,
      "step": 34040
    },
    {
      "epoch": 2.664736265456253,
      "grad_norm": 8.175249099731445,
      "learning_rate": 5.603651200502171e-06,
      "loss": 0.6172,
      "step": 34050
    },
    {
      "epoch": 2.665518860541556,
      "grad_norm": 6.788190841674805,
      "learning_rate": 5.590573834806717e-06,
      "loss": 0.4864,
      "step": 34060
    },
    {
      "epoch": 2.6663014556268587,
      "grad_norm": 18.44725227355957,
      "learning_rate": 5.577496469111262e-06,
      "loss": 0.7376,
      "step": 34070
    },
    {
      "epoch": 2.6670840507121616,
      "grad_norm": 2.609499931335449,
      "learning_rate": 5.5644191034158085e-06,
      "loss": 0.909,
      "step": 34080
    },
    {
      "epoch": 2.6678666457974645,
      "grad_norm": 15.377727508544922,
      "learning_rate": 5.551341737720354e-06,
      "loss": 0.6296,
      "step": 34090
    },
    {
      "epoch": 2.6686492408827673,
      "grad_norm": 7.088497161865234,
      "learning_rate": 5.5382643720249e-06,
      "loss": 0.6463,
      "step": 34100
    },
    {
      "epoch": 2.66943183596807,
      "grad_norm": 24.792640686035156,
      "learning_rate": 5.525187006329445e-06,
      "loss": 0.7344,
      "step": 34110
    },
    {
      "epoch": 2.670214431053373,
      "grad_norm": 13.153467178344727,
      "learning_rate": 5.512109640633991e-06,
      "loss": 0.8543,
      "step": 34120
    },
    {
      "epoch": 2.670997026138676,
      "grad_norm": 5.756148338317871,
      "learning_rate": 5.499032274938536e-06,
      "loss": 0.6529,
      "step": 34130
    },
    {
      "epoch": 2.671779621223979,
      "grad_norm": 14.590121269226074,
      "learning_rate": 5.485954909243083e-06,
      "loss": 0.7017,
      "step": 34140
    },
    {
      "epoch": 2.6725622163092817,
      "grad_norm": 1.1708576679229736,
      "learning_rate": 5.472877543547628e-06,
      "loss": 0.9092,
      "step": 34150
    },
    {
      "epoch": 2.6733448113945846,
      "grad_norm": 16.442161560058594,
      "learning_rate": 5.4598001778521735e-06,
      "loss": 0.62,
      "step": 34160
    },
    {
      "epoch": 2.6741274064798874,
      "grad_norm": 28.343055725097656,
      "learning_rate": 5.44672281215672e-06,
      "loss": 0.7205,
      "step": 34170
    },
    {
      "epoch": 2.6749100015651903,
      "grad_norm": 43.30875778198242,
      "learning_rate": 5.433645446461265e-06,
      "loss": 1.0019,
      "step": 34180
    },
    {
      "epoch": 2.675692596650493,
      "grad_norm": 1.268323302268982,
      "learning_rate": 5.420568080765811e-06,
      "loss": 0.4579,
      "step": 34190
    },
    {
      "epoch": 2.676475191735796,
      "grad_norm": 6.778244972229004,
      "learning_rate": 5.407490715070356e-06,
      "loss": 0.7304,
      "step": 34200
    },
    {
      "epoch": 2.677257786821099,
      "grad_norm": 12.512145042419434,
      "learning_rate": 5.394413349374902e-06,
      "loss": 0.9279,
      "step": 34210
    },
    {
      "epoch": 2.678040381906402,
      "grad_norm": 6.986786365509033,
      "learning_rate": 5.3813359836794486e-06,
      "loss": 0.5206,
      "step": 34220
    },
    {
      "epoch": 2.6788229769917047,
      "grad_norm": 28.168231964111328,
      "learning_rate": 5.368258617983993e-06,
      "loss": 0.9873,
      "step": 34230
    },
    {
      "epoch": 2.6796055720770076,
      "grad_norm": 1.377869963645935,
      "learning_rate": 5.355181252288539e-06,
      "loss": 0.6612,
      "step": 34240
    },
    {
      "epoch": 2.6803881671623104,
      "grad_norm": 11.23703384399414,
      "learning_rate": 5.342103886593085e-06,
      "loss": 0.4155,
      "step": 34250
    },
    {
      "epoch": 2.6811707622476133,
      "grad_norm": 0.875393807888031,
      "learning_rate": 5.329026520897631e-06,
      "loss": 0.4966,
      "step": 34260
    },
    {
      "epoch": 2.681953357332916,
      "grad_norm": 10.555171012878418,
      "learning_rate": 5.315949155202176e-06,
      "loss": 0.6641,
      "step": 34270
    },
    {
      "epoch": 2.682735952418219,
      "grad_norm": 7.8737664222717285,
      "learning_rate": 5.302871789506722e-06,
      "loss": 0.9789,
      "step": 34280
    },
    {
      "epoch": 2.683518547503522,
      "grad_norm": 18.99407386779785,
      "learning_rate": 5.289794423811268e-06,
      "loss": 0.7772,
      "step": 34290
    },
    {
      "epoch": 2.684301142588825,
      "grad_norm": 9.166439056396484,
      "learning_rate": 5.276717058115814e-06,
      "loss": 0.639,
      "step": 34300
    },
    {
      "epoch": 2.6850837376741277,
      "grad_norm": 21.10564422607422,
      "learning_rate": 5.263639692420359e-06,
      "loss": 1.0321,
      "step": 34310
    },
    {
      "epoch": 2.6858663327594305,
      "grad_norm": 7.730587959289551,
      "learning_rate": 5.2505623267249044e-06,
      "loss": 0.4976,
      "step": 34320
    },
    {
      "epoch": 2.686648927844733,
      "grad_norm": 12.398092269897461,
      "learning_rate": 5.237484961029451e-06,
      "loss": 0.6725,
      "step": 34330
    },
    {
      "epoch": 2.687431522930036,
      "grad_norm": 24.600297927856445,
      "learning_rate": 5.224407595333996e-06,
      "loss": 0.8147,
      "step": 34340
    },
    {
      "epoch": 2.6882141180153387,
      "grad_norm": 19.34317398071289,
      "learning_rate": 5.2113302296385416e-06,
      "loss": 0.6742,
      "step": 34350
    },
    {
      "epoch": 2.6889967131006416,
      "grad_norm": 11.833714485168457,
      "learning_rate": 5.198252863943088e-06,
      "loss": 1.2217,
      "step": 34360
    },
    {
      "epoch": 2.6897793081859445,
      "grad_norm": 35.539730072021484,
      "learning_rate": 5.185175498247633e-06,
      "loss": 0.6265,
      "step": 34370
    },
    {
      "epoch": 2.6905619032712473,
      "grad_norm": 26.768753051757812,
      "learning_rate": 5.1720981325521795e-06,
      "loss": 0.6183,
      "step": 34380
    },
    {
      "epoch": 2.69134449835655,
      "grad_norm": 0.6185401678085327,
      "learning_rate": 5.159020766856724e-06,
      "loss": 0.681,
      "step": 34390
    },
    {
      "epoch": 2.692127093441853,
      "grad_norm": 14.489583015441895,
      "learning_rate": 5.14594340116127e-06,
      "loss": 0.8851,
      "step": 34400
    },
    {
      "epoch": 2.692909688527156,
      "grad_norm": 9.272436141967773,
      "learning_rate": 5.132866035465816e-06,
      "loss": 0.9092,
      "step": 34410
    },
    {
      "epoch": 2.693692283612459,
      "grad_norm": 17.54269027709961,
      "learning_rate": 5.119788669770362e-06,
      "loss": 0.9811,
      "step": 34420
    },
    {
      "epoch": 2.6944748786977617,
      "grad_norm": 12.028277397155762,
      "learning_rate": 5.1067113040749074e-06,
      "loss": 0.6914,
      "step": 34430
    },
    {
      "epoch": 2.6952574737830646,
      "grad_norm": 6.046919822692871,
      "learning_rate": 5.093633938379453e-06,
      "loss": 0.8079,
      "step": 34440
    },
    {
      "epoch": 2.6960400688683674,
      "grad_norm": 13.82658576965332,
      "learning_rate": 5.080556572683999e-06,
      "loss": 0.7301,
      "step": 34450
    },
    {
      "epoch": 2.6968226639536703,
      "grad_norm": 16.785480499267578,
      "learning_rate": 5.0674792069885446e-06,
      "loss": 0.9773,
      "step": 34460
    },
    {
      "epoch": 2.697605259038973,
      "grad_norm": 9.914671897888184,
      "learning_rate": 5.05440184129309e-06,
      "loss": 0.5696,
      "step": 34470
    },
    {
      "epoch": 2.698387854124276,
      "grad_norm": 10.302706718444824,
      "learning_rate": 5.041324475597635e-06,
      "loss": 0.7207,
      "step": 34480
    },
    {
      "epoch": 2.699170449209579,
      "grad_norm": 9.022631645202637,
      "learning_rate": 5.028247109902182e-06,
      "loss": 0.8718,
      "step": 34490
    },
    {
      "epoch": 2.699953044294882,
      "grad_norm": 7.581448554992676,
      "learning_rate": 5.015169744206727e-06,
      "loss": 0.6932,
      "step": 34500
    },
    {
      "epoch": 2.7007356393801847,
      "grad_norm": 10.060355186462402,
      "learning_rate": 5.0020923785112725e-06,
      "loss": 1.0029,
      "step": 34510
    },
    {
      "epoch": 2.7015182344654876,
      "grad_norm": 8.174872398376465,
      "learning_rate": 4.989015012815819e-06,
      "loss": 0.6202,
      "step": 34520
    },
    {
      "epoch": 2.7023008295507904,
      "grad_norm": 3.0044562816619873,
      "learning_rate": 4.975937647120364e-06,
      "loss": 0.9389,
      "step": 34530
    },
    {
      "epoch": 2.7030834246360933,
      "grad_norm": 5.078964710235596,
      "learning_rate": 4.9628602814249105e-06,
      "loss": 0.6225,
      "step": 34540
    },
    {
      "epoch": 2.703866019721396,
      "grad_norm": 0.9534396529197693,
      "learning_rate": 4.949782915729456e-06,
      "loss": 0.591,
      "step": 34550
    },
    {
      "epoch": 2.704648614806699,
      "grad_norm": 4.204588413238525,
      "learning_rate": 4.936705550034001e-06,
      "loss": 0.5551,
      "step": 34560
    },
    {
      "epoch": 2.705431209892002,
      "grad_norm": 2.048945426940918,
      "learning_rate": 4.923628184338547e-06,
      "loss": 0.4717,
      "step": 34570
    },
    {
      "epoch": 2.706213804977305,
      "grad_norm": 12.158958435058594,
      "learning_rate": 4.910550818643093e-06,
      "loss": 0.6116,
      "step": 34580
    },
    {
      "epoch": 2.7069964000626077,
      "grad_norm": 4.644761085510254,
      "learning_rate": 4.897473452947638e-06,
      "loss": 0.7262,
      "step": 34590
    },
    {
      "epoch": 2.7077789951479105,
      "grad_norm": 29.796836853027344,
      "learning_rate": 4.884396087252184e-06,
      "loss": 0.7144,
      "step": 34600
    },
    {
      "epoch": 2.7085615902332134,
      "grad_norm": 11.79688835144043,
      "learning_rate": 4.87131872155673e-06,
      "loss": 0.6751,
      "step": 34610
    },
    {
      "epoch": 2.7093441853185163,
      "grad_norm": 20.006114959716797,
      "learning_rate": 4.8582413558612755e-06,
      "loss": 0.7275,
      "step": 34620
    },
    {
      "epoch": 2.710126780403819,
      "grad_norm": 7.088937282562256,
      "learning_rate": 4.845163990165822e-06,
      "loss": 0.8559,
      "step": 34630
    },
    {
      "epoch": 2.710909375489122,
      "grad_norm": 15.858793258666992,
      "learning_rate": 4.832086624470366e-06,
      "loss": 0.9437,
      "step": 34640
    },
    {
      "epoch": 2.711691970574425,
      "grad_norm": 2.0190792083740234,
      "learning_rate": 4.819009258774913e-06,
      "loss": 0.8724,
      "step": 34650
    },
    {
      "epoch": 2.7124745656597278,
      "grad_norm": 3.1430978775024414,
      "learning_rate": 4.805931893079459e-06,
      "loss": 0.8119,
      "step": 34660
    },
    {
      "epoch": 2.7132571607450306,
      "grad_norm": 20.656370162963867,
      "learning_rate": 4.792854527384004e-06,
      "loss": 0.5238,
      "step": 34670
    },
    {
      "epoch": 2.7140397558303335,
      "grad_norm": 11.926289558410645,
      "learning_rate": 4.77977716168855e-06,
      "loss": 1.0732,
      "step": 34680
    },
    {
      "epoch": 2.7148223509156364,
      "grad_norm": 1.3583849668502808,
      "learning_rate": 4.766699795993095e-06,
      "loss": 0.927,
      "step": 34690
    },
    {
      "epoch": 2.7156049460009393,
      "grad_norm": 6.2370500564575195,
      "learning_rate": 4.753622430297641e-06,
      "loss": 0.6055,
      "step": 34700
    },
    {
      "epoch": 2.716387541086242,
      "grad_norm": 1.2735414505004883,
      "learning_rate": 4.740545064602187e-06,
      "loss": 0.671,
      "step": 34710
    },
    {
      "epoch": 2.7171701361715446,
      "grad_norm": 10.584949493408203,
      "learning_rate": 4.727467698906732e-06,
      "loss": 0.7229,
      "step": 34720
    },
    {
      "epoch": 2.7179527312568474,
      "grad_norm": 13.788246154785156,
      "learning_rate": 4.7143903332112785e-06,
      "loss": 0.783,
      "step": 34730
    },
    {
      "epoch": 2.7187353263421503,
      "grad_norm": 17.501121520996094,
      "learning_rate": 4.701312967515824e-06,
      "loss": 0.6692,
      "step": 34740
    },
    {
      "epoch": 2.719517921427453,
      "grad_norm": 8.13898754119873,
      "learning_rate": 4.68823560182037e-06,
      "loss": 0.4818,
      "step": 34750
    },
    {
      "epoch": 2.720300516512756,
      "grad_norm": 12.910299301147461,
      "learning_rate": 4.675158236124915e-06,
      "loss": 1.0346,
      "step": 34760
    },
    {
      "epoch": 2.721083111598059,
      "grad_norm": 13.354480743408203,
      "learning_rate": 4.662080870429461e-06,
      "loss": 0.9233,
      "step": 34770
    },
    {
      "epoch": 2.721865706683362,
      "grad_norm": 11.301316261291504,
      "learning_rate": 4.6490035047340064e-06,
      "loss": 0.6732,
      "step": 34780
    },
    {
      "epoch": 2.7226483017686647,
      "grad_norm": 14.115667343139648,
      "learning_rate": 4.635926139038553e-06,
      "loss": 0.8613,
      "step": 34790
    },
    {
      "epoch": 2.7234308968539676,
      "grad_norm": 8.87197494506836,
      "learning_rate": 4.622848773343097e-06,
      "loss": 0.626,
      "step": 34800
    },
    {
      "epoch": 2.7242134919392704,
      "grad_norm": 8.950675010681152,
      "learning_rate": 4.6097714076476435e-06,
      "loss": 0.5094,
      "step": 34810
    },
    {
      "epoch": 2.7249960870245733,
      "grad_norm": 7.019244194030762,
      "learning_rate": 4.59669404195219e-06,
      "loss": 0.5868,
      "step": 34820
    },
    {
      "epoch": 2.725778682109876,
      "grad_norm": 7.725409984588623,
      "learning_rate": 4.583616676256735e-06,
      "loss": 0.9983,
      "step": 34830
    },
    {
      "epoch": 2.726561277195179,
      "grad_norm": 5.8486223220825195,
      "learning_rate": 4.570539310561281e-06,
      "loss": 0.4991,
      "step": 34840
    },
    {
      "epoch": 2.727343872280482,
      "grad_norm": 13.258395195007324,
      "learning_rate": 4.557461944865826e-06,
      "loss": 0.7146,
      "step": 34850
    },
    {
      "epoch": 2.728126467365785,
      "grad_norm": 1.7685010433197021,
      "learning_rate": 4.544384579170372e-06,
      "loss": 0.6173,
      "step": 34860
    },
    {
      "epoch": 2.7289090624510877,
      "grad_norm": 14.958503723144531,
      "learning_rate": 4.531307213474918e-06,
      "loss": 0.864,
      "step": 34870
    },
    {
      "epoch": 2.7296916575363905,
      "grad_norm": 8.130288124084473,
      "learning_rate": 4.518229847779463e-06,
      "loss": 0.7194,
      "step": 34880
    },
    {
      "epoch": 2.7304742526216934,
      "grad_norm": 23.750200271606445,
      "learning_rate": 4.5051524820840094e-06,
      "loss": 0.7092,
      "step": 34890
    },
    {
      "epoch": 2.7312568477069963,
      "grad_norm": 14.77503490447998,
      "learning_rate": 4.492075116388555e-06,
      "loss": 0.6992,
      "step": 34900
    },
    {
      "epoch": 2.732039442792299,
      "grad_norm": 5.572540283203125,
      "learning_rate": 4.478997750693101e-06,
      "loss": 0.5618,
      "step": 34910
    },
    {
      "epoch": 2.732822037877602,
      "grad_norm": 4.521900653839111,
      "learning_rate": 4.465920384997646e-06,
      "loss": 0.649,
      "step": 34920
    },
    {
      "epoch": 2.733604632962905,
      "grad_norm": 26.446434020996094,
      "learning_rate": 4.452843019302192e-06,
      "loss": 0.7229,
      "step": 34930
    },
    {
      "epoch": 2.7343872280482078,
      "grad_norm": 32.12471008300781,
      "learning_rate": 4.439765653606737e-06,
      "loss": 0.5192,
      "step": 34940
    },
    {
      "epoch": 2.7351698231335106,
      "grad_norm": 7.7058820724487305,
      "learning_rate": 4.426688287911284e-06,
      "loss": 0.602,
      "step": 34950
    },
    {
      "epoch": 2.7359524182188135,
      "grad_norm": 6.055981159210205,
      "learning_rate": 4.413610922215829e-06,
      "loss": 0.7587,
      "step": 34960
    },
    {
      "epoch": 2.7367350133041164,
      "grad_norm": 8.089569091796875,
      "learning_rate": 4.4005335565203745e-06,
      "loss": 0.638,
      "step": 34970
    },
    {
      "epoch": 2.7375176083894193,
      "grad_norm": 1.6886560916900635,
      "learning_rate": 4.387456190824921e-06,
      "loss": 0.4666,
      "step": 34980
    },
    {
      "epoch": 2.738300203474722,
      "grad_norm": 8.782814025878906,
      "learning_rate": 4.374378825129466e-06,
      "loss": 0.7289,
      "step": 34990
    },
    {
      "epoch": 2.739082798560025,
      "grad_norm": 0.4579627215862274,
      "learning_rate": 4.361301459434012e-06,
      "loss": 0.89,
      "step": 35000
    },
    {
      "epoch": 2.739865393645328,
      "grad_norm": 13.968287467956543,
      "learning_rate": 4.348224093738557e-06,
      "loss": 0.5987,
      "step": 35010
    },
    {
      "epoch": 2.7406479887306308,
      "grad_norm": 2.027644634246826,
      "learning_rate": 4.335146728043103e-06,
      "loss": 0.9144,
      "step": 35020
    },
    {
      "epoch": 2.7414305838159336,
      "grad_norm": 13.512250900268555,
      "learning_rate": 4.3220693623476495e-06,
      "loss": 1.0479,
      "step": 35030
    },
    {
      "epoch": 2.7422131789012365,
      "grad_norm": 9.369832992553711,
      "learning_rate": 4.308991996652194e-06,
      "loss": 0.7181,
      "step": 35040
    },
    {
      "epoch": 2.7429957739865394,
      "grad_norm": 5.756157875061035,
      "learning_rate": 4.29591463095674e-06,
      "loss": 0.6558,
      "step": 35050
    },
    {
      "epoch": 2.7437783690718422,
      "grad_norm": 9.397490501403809,
      "learning_rate": 4.282837265261286e-06,
      "loss": 0.4959,
      "step": 35060
    },
    {
      "epoch": 2.744560964157145,
      "grad_norm": 18.522729873657227,
      "learning_rate": 4.269759899565832e-06,
      "loss": 1.2553,
      "step": 35070
    },
    {
      "epoch": 2.745343559242448,
      "grad_norm": 13.220213890075684,
      "learning_rate": 4.2566825338703775e-06,
      "loss": 0.8869,
      "step": 35080
    },
    {
      "epoch": 2.746126154327751,
      "grad_norm": 9.779932975769043,
      "learning_rate": 4.243605168174923e-06,
      "loss": 0.5194,
      "step": 35090
    },
    {
      "epoch": 2.7469087494130537,
      "grad_norm": 30.80905532836914,
      "learning_rate": 4.230527802479468e-06,
      "loss": 0.7553,
      "step": 35100
    },
    {
      "epoch": 2.7476913444983566,
      "grad_norm": 72.88187408447266,
      "learning_rate": 4.217450436784015e-06,
      "loss": 0.754,
      "step": 35110
    },
    {
      "epoch": 2.7484739395836595,
      "grad_norm": 8.525201797485352,
      "learning_rate": 4.20437307108856e-06,
      "loss": 0.7512,
      "step": 35120
    },
    {
      "epoch": 2.7492565346689624,
      "grad_norm": 5.3757500648498535,
      "learning_rate": 4.1912957053931054e-06,
      "loss": 0.8443,
      "step": 35130
    },
    {
      "epoch": 2.7500391297542652,
      "grad_norm": 15.878448486328125,
      "learning_rate": 4.178218339697652e-06,
      "loss": 0.8536,
      "step": 35140
    },
    {
      "epoch": 2.750821724839568,
      "grad_norm": 17.76711082458496,
      "learning_rate": 4.165140974002197e-06,
      "loss": 1.0795,
      "step": 35150
    },
    {
      "epoch": 2.751604319924871,
      "grad_norm": 6.131062030792236,
      "learning_rate": 4.1520636083067425e-06,
      "loss": 0.8182,
      "step": 35160
    },
    {
      "epoch": 2.752386915010174,
      "grad_norm": 2.649897575378418,
      "learning_rate": 4.138986242611288e-06,
      "loss": 0.6159,
      "step": 35170
    },
    {
      "epoch": 2.7531695100954767,
      "grad_norm": 4.551820278167725,
      "learning_rate": 4.125908876915834e-06,
      "loss": 0.6202,
      "step": 35180
    },
    {
      "epoch": 2.7539521051807796,
      "grad_norm": 9.420653343200684,
      "learning_rate": 4.1128315112203805e-06,
      "loss": 0.8613,
      "step": 35190
    },
    {
      "epoch": 2.7547347002660825,
      "grad_norm": 4.523326396942139,
      "learning_rate": 4.099754145524926e-06,
      "loss": 0.6184,
      "step": 35200
    },
    {
      "epoch": 2.7555172953513853,
      "grad_norm": 48.80735778808594,
      "learning_rate": 4.086676779829471e-06,
      "loss": 0.9542,
      "step": 35210
    },
    {
      "epoch": 2.756299890436688,
      "grad_norm": 1.4872734546661377,
      "learning_rate": 4.073599414134017e-06,
      "loss": 0.7406,
      "step": 35220
    },
    {
      "epoch": 2.757082485521991,
      "grad_norm": 1.026899814605713,
      "learning_rate": 4.060522048438563e-06,
      "loss": 0.4309,
      "step": 35230
    },
    {
      "epoch": 2.757865080607294,
      "grad_norm": 5.423003673553467,
      "learning_rate": 4.0474446827431084e-06,
      "loss": 0.6586,
      "step": 35240
    },
    {
      "epoch": 2.758647675692597,
      "grad_norm": 3.191208839416504,
      "learning_rate": 4.034367317047654e-06,
      "loss": 0.5515,
      "step": 35250
    },
    {
      "epoch": 2.7594302707778997,
      "grad_norm": 8.172831535339355,
      "learning_rate": 4.0212899513522e-06,
      "loss": 0.4475,
      "step": 35260
    },
    {
      "epoch": 2.7602128658632026,
      "grad_norm": 5.6994428634643555,
      "learning_rate": 4.0082125856567455e-06,
      "loss": 0.7024,
      "step": 35270
    },
    {
      "epoch": 2.7609954609485055,
      "grad_norm": 1.4904232025146484,
      "learning_rate": 3.995135219961292e-06,
      "loss": 0.5241,
      "step": 35280
    },
    {
      "epoch": 2.7617780560338083,
      "grad_norm": 2.0913145542144775,
      "learning_rate": 3.982057854265836e-06,
      "loss": 0.5424,
      "step": 35290
    },
    {
      "epoch": 2.762560651119111,
      "grad_norm": 13.603065490722656,
      "learning_rate": 3.968980488570383e-06,
      "loss": 1.1144,
      "step": 35300
    },
    {
      "epoch": 2.763343246204414,
      "grad_norm": 8.125349998474121,
      "learning_rate": 3.955903122874928e-06,
      "loss": 1.056,
      "step": 35310
    },
    {
      "epoch": 2.764125841289717,
      "grad_norm": 27.10942840576172,
      "learning_rate": 3.942825757179474e-06,
      "loss": 1.0157,
      "step": 35320
    },
    {
      "epoch": 2.76490843637502,
      "grad_norm": 3.647597074508667,
      "learning_rate": 3.92974839148402e-06,
      "loss": 0.6747,
      "step": 35330
    },
    {
      "epoch": 2.7656910314603227,
      "grad_norm": 5.075794219970703,
      "learning_rate": 3.916671025788565e-06,
      "loss": 0.648,
      "step": 35340
    },
    {
      "epoch": 2.7664736265456256,
      "grad_norm": 3.1272873878479004,
      "learning_rate": 3.9035936600931114e-06,
      "loss": 0.3447,
      "step": 35350
    },
    {
      "epoch": 2.767256221630928,
      "grad_norm": 19.273727416992188,
      "learning_rate": 3.890516294397657e-06,
      "loss": 0.7445,
      "step": 35360
    },
    {
      "epoch": 2.768038816716231,
      "grad_norm": 26.986860275268555,
      "learning_rate": 3.877438928702202e-06,
      "loss": 1.0362,
      "step": 35370
    },
    {
      "epoch": 2.7688214118015337,
      "grad_norm": 11.771247863769531,
      "learning_rate": 3.864361563006748e-06,
      "loss": 0.3712,
      "step": 35380
    },
    {
      "epoch": 2.7696040068868366,
      "grad_norm": 3.548382520675659,
      "learning_rate": 3.851284197311294e-06,
      "loss": 0.6166,
      "step": 35390
    },
    {
      "epoch": 2.7703866019721395,
      "grad_norm": 6.529423236846924,
      "learning_rate": 3.83820683161584e-06,
      "loss": 0.8554,
      "step": 35400
    },
    {
      "epoch": 2.7711691970574424,
      "grad_norm": 10.425653457641602,
      "learning_rate": 3.825129465920385e-06,
      "loss": 0.6061,
      "step": 35410
    },
    {
      "epoch": 2.7719517921427452,
      "grad_norm": 10.77440357208252,
      "learning_rate": 3.812052100224931e-06,
      "loss": 0.3863,
      "step": 35420
    },
    {
      "epoch": 2.772734387228048,
      "grad_norm": 11.853697776794434,
      "learning_rate": 3.7989747345294765e-06,
      "loss": 0.844,
      "step": 35430
    },
    {
      "epoch": 2.773516982313351,
      "grad_norm": 4.1795878410339355,
      "learning_rate": 3.7858973688340223e-06,
      "loss": 0.8743,
      "step": 35440
    },
    {
      "epoch": 2.774299577398654,
      "grad_norm": 14.35256576538086,
      "learning_rate": 3.7728200031385677e-06,
      "loss": 0.4112,
      "step": 35450
    },
    {
      "epoch": 2.7750821724839567,
      "grad_norm": 12.832934379577637,
      "learning_rate": 3.7597426374431136e-06,
      "loss": 1.1942,
      "step": 35460
    },
    {
      "epoch": 2.7758647675692596,
      "grad_norm": 8.471647262573242,
      "learning_rate": 3.746665271747659e-06,
      "loss": 0.4911,
      "step": 35470
    },
    {
      "epoch": 2.7766473626545625,
      "grad_norm": 14.994590759277344,
      "learning_rate": 3.733587906052205e-06,
      "loss": 0.8603,
      "step": 35480
    },
    {
      "epoch": 2.7774299577398653,
      "grad_norm": 7.713823318481445,
      "learning_rate": 3.720510540356751e-06,
      "loss": 0.8574,
      "step": 35490
    },
    {
      "epoch": 2.778212552825168,
      "grad_norm": 11.794962882995605,
      "learning_rate": 3.707433174661296e-06,
      "loss": 0.6983,
      "step": 35500
    },
    {
      "epoch": 2.778995147910471,
      "grad_norm": 0.7283281087875366,
      "learning_rate": 3.6943558089658424e-06,
      "loss": 0.3396,
      "step": 35510
    },
    {
      "epoch": 2.779777742995774,
      "grad_norm": 16.414894104003906,
      "learning_rate": 3.6812784432703874e-06,
      "loss": 0.5901,
      "step": 35520
    },
    {
      "epoch": 2.780560338081077,
      "grad_norm": 23.793254852294922,
      "learning_rate": 3.6682010775749336e-06,
      "loss": 0.5931,
      "step": 35530
    },
    {
      "epoch": 2.7813429331663797,
      "grad_norm": 5.559248447418213,
      "learning_rate": 3.655123711879479e-06,
      "loss": 0.6766,
      "step": 35540
    },
    {
      "epoch": 2.7821255282516826,
      "grad_norm": 9.585288047790527,
      "learning_rate": 3.642046346184025e-06,
      "loss": 0.7482,
      "step": 35550
    },
    {
      "epoch": 2.7829081233369855,
      "grad_norm": 4.927308082580566,
      "learning_rate": 3.6289689804885708e-06,
      "loss": 0.7849,
      "step": 35560
    },
    {
      "epoch": 2.7836907184222883,
      "grad_norm": 7.554004669189453,
      "learning_rate": 3.615891614793116e-06,
      "loss": 0.6278,
      "step": 35570
    },
    {
      "epoch": 2.784473313507591,
      "grad_norm": 38.3396110534668,
      "learning_rate": 3.602814249097662e-06,
      "loss": 0.561,
      "step": 35580
    },
    {
      "epoch": 2.785255908592894,
      "grad_norm": 24.549968719482422,
      "learning_rate": 3.5897368834022074e-06,
      "loss": 0.6927,
      "step": 35590
    },
    {
      "epoch": 2.786038503678197,
      "grad_norm": 5.812121868133545,
      "learning_rate": 3.5766595177067533e-06,
      "loss": 0.8288,
      "step": 35600
    },
    {
      "epoch": 2.7868210987635,
      "grad_norm": 13.674505233764648,
      "learning_rate": 3.5635821520112987e-06,
      "loss": 0.6758,
      "step": 35610
    },
    {
      "epoch": 2.7876036938488027,
      "grad_norm": 7.57142448425293,
      "learning_rate": 3.5505047863158445e-06,
      "loss": 0.8874,
      "step": 35620
    },
    {
      "epoch": 2.7883862889341056,
      "grad_norm": 17.34562110900879,
      "learning_rate": 3.537427420620391e-06,
      "loss": 0.5998,
      "step": 35630
    },
    {
      "epoch": 2.7891688840194084,
      "grad_norm": 6.191683292388916,
      "learning_rate": 3.524350054924936e-06,
      "loss": 0.739,
      "step": 35640
    },
    {
      "epoch": 2.7899514791047113,
      "grad_norm": 9.148412704467773,
      "learning_rate": 3.511272689229482e-06,
      "loss": 1.0123,
      "step": 35650
    },
    {
      "epoch": 2.790734074190014,
      "grad_norm": 4.867461204528809,
      "learning_rate": 3.4981953235340275e-06,
      "loss": 0.5047,
      "step": 35660
    },
    {
      "epoch": 2.791516669275317,
      "grad_norm": 12.698532104492188,
      "learning_rate": 3.4851179578385733e-06,
      "loss": 1.2636,
      "step": 35670
    },
    {
      "epoch": 2.79229926436062,
      "grad_norm": 6.185513973236084,
      "learning_rate": 3.4720405921431187e-06,
      "loss": 0.8122,
      "step": 35680
    },
    {
      "epoch": 2.793081859445923,
      "grad_norm": 8.304533958435059,
      "learning_rate": 3.4589632264476646e-06,
      "loss": 0.5489,
      "step": 35690
    },
    {
      "epoch": 2.7938644545312257,
      "grad_norm": 11.662370681762695,
      "learning_rate": 3.4458858607522104e-06,
      "loss": 1.1972,
      "step": 35700
    },
    {
      "epoch": 2.7946470496165285,
      "grad_norm": 12.861279487609863,
      "learning_rate": 3.432808495056756e-06,
      "loss": 0.5832,
      "step": 35710
    },
    {
      "epoch": 2.7954296447018314,
      "grad_norm": 17.551504135131836,
      "learning_rate": 3.4197311293613017e-06,
      "loss": 0.8253,
      "step": 35720
    },
    {
      "epoch": 2.7962122397871343,
      "grad_norm": 13.047377586364746,
      "learning_rate": 3.406653763665847e-06,
      "loss": 0.861,
      "step": 35730
    },
    {
      "epoch": 2.796994834872437,
      "grad_norm": 11.32228946685791,
      "learning_rate": 3.393576397970393e-06,
      "loss": 0.8661,
      "step": 35740
    },
    {
      "epoch": 2.7977774299577396,
      "grad_norm": 29.681936264038086,
      "learning_rate": 3.3804990322749384e-06,
      "loss": 0.8164,
      "step": 35750
    },
    {
      "epoch": 2.7985600250430425,
      "grad_norm": 24.955476760864258,
      "learning_rate": 3.3674216665794846e-06,
      "loss": 0.9472,
      "step": 35760
    },
    {
      "epoch": 2.7993426201283453,
      "grad_norm": 6.509047031402588,
      "learning_rate": 3.3543443008840305e-06,
      "loss": 0.3114,
      "step": 35770
    },
    {
      "epoch": 2.800125215213648,
      "grad_norm": 9.1921968460083,
      "learning_rate": 3.341266935188576e-06,
      "loss": 0.4626,
      "step": 35780
    },
    {
      "epoch": 2.800907810298951,
      "grad_norm": 6.350562572479248,
      "learning_rate": 3.3281895694931217e-06,
      "loss": 0.8596,
      "step": 35790
    },
    {
      "epoch": 2.801690405384254,
      "grad_norm": 0.9379338026046753,
      "learning_rate": 3.315112203797667e-06,
      "loss": 0.8676,
      "step": 35800
    },
    {
      "epoch": 2.802473000469557,
      "grad_norm": 16.597753524780273,
      "learning_rate": 3.302034838102213e-06,
      "loss": 0.7003,
      "step": 35810
    },
    {
      "epoch": 2.8032555955548597,
      "grad_norm": 6.674816608428955,
      "learning_rate": 3.2889574724067584e-06,
      "loss": 0.7664,
      "step": 35820
    },
    {
      "epoch": 2.8040381906401626,
      "grad_norm": 11.619166374206543,
      "learning_rate": 3.2758801067113043e-06,
      "loss": 0.5036,
      "step": 35830
    },
    {
      "epoch": 2.8048207857254654,
      "grad_norm": 3.024773597717285,
      "learning_rate": 3.2628027410158497e-06,
      "loss": 0.398,
      "step": 35840
    },
    {
      "epoch": 2.8056033808107683,
      "grad_norm": 7.8417277336120605,
      "learning_rate": 3.2497253753203955e-06,
      "loss": 0.6036,
      "step": 35850
    },
    {
      "epoch": 2.806385975896071,
      "grad_norm": 5.337311744689941,
      "learning_rate": 3.2366480096249414e-06,
      "loss": 0.5669,
      "step": 35860
    },
    {
      "epoch": 2.807168570981374,
      "grad_norm": 6.854114055633545,
      "learning_rate": 3.223570643929487e-06,
      "loss": 0.922,
      "step": 35870
    },
    {
      "epoch": 2.807951166066677,
      "grad_norm": 11.540621757507324,
      "learning_rate": 3.210493278234033e-06,
      "loss": 0.6771,
      "step": 35880
    },
    {
      "epoch": 2.80873376115198,
      "grad_norm": 1.2320688962936401,
      "learning_rate": 3.197415912538578e-06,
      "loss": 0.4556,
      "step": 35890
    },
    {
      "epoch": 2.8095163562372827,
      "grad_norm": 8.700519561767578,
      "learning_rate": 3.1843385468431243e-06,
      "loss": 0.7882,
      "step": 35900
    },
    {
      "epoch": 2.8102989513225856,
      "grad_norm": 9.386844635009766,
      "learning_rate": 3.1712611811476693e-06,
      "loss": 0.6028,
      "step": 35910
    },
    {
      "epoch": 2.8110815464078884,
      "grad_norm": 0.9431166648864746,
      "learning_rate": 3.1581838154522156e-06,
      "loss": 0.6691,
      "step": 35920
    },
    {
      "epoch": 2.8118641414931913,
      "grad_norm": 2.6966357231140137,
      "learning_rate": 3.1451064497567614e-06,
      "loss": 0.5709,
      "step": 35930
    },
    {
      "epoch": 2.812646736578494,
      "grad_norm": 1.9803627729415894,
      "learning_rate": 3.132029084061307e-06,
      "loss": 0.6031,
      "step": 35940
    },
    {
      "epoch": 2.813429331663797,
      "grad_norm": 8.394257545471191,
      "learning_rate": 3.1189517183658523e-06,
      "loss": 0.7571,
      "step": 35950
    },
    {
      "epoch": 2.8142119267491,
      "grad_norm": 12.328886985778809,
      "learning_rate": 3.1058743526703985e-06,
      "loss": 0.9348,
      "step": 35960
    },
    {
      "epoch": 2.814994521834403,
      "grad_norm": 2.8134593963623047,
      "learning_rate": 3.092796986974944e-06,
      "loss": 0.6703,
      "step": 35970
    },
    {
      "epoch": 2.8157771169197057,
      "grad_norm": 4.61700963973999,
      "learning_rate": 3.07971962127949e-06,
      "loss": 0.637,
      "step": 35980
    },
    {
      "epoch": 2.8165597120050085,
      "grad_norm": 6.629024982452393,
      "learning_rate": 3.0666422555840352e-06,
      "loss": 0.7191,
      "step": 35990
    },
    {
      "epoch": 2.8173423070903114,
      "grad_norm": 10.618043899536133,
      "learning_rate": 3.053564889888581e-06,
      "loss": 0.6064,
      "step": 36000
    },
    {
      "epoch": 2.8181249021756143,
      "grad_norm": 7.86555290222168,
      "learning_rate": 3.0404875241931265e-06,
      "loss": 0.7048,
      "step": 36010
    },
    {
      "epoch": 2.818907497260917,
      "grad_norm": 8.720849990844727,
      "learning_rate": 3.0274101584976723e-06,
      "loss": 0.6159,
      "step": 36020
    },
    {
      "epoch": 2.81969009234622,
      "grad_norm": 16.808483123779297,
      "learning_rate": 3.014332792802218e-06,
      "loss": 0.9026,
      "step": 36030
    },
    {
      "epoch": 2.820472687431523,
      "grad_norm": 0.9032634496688843,
      "learning_rate": 3.001255427106764e-06,
      "loss": 0.755,
      "step": 36040
    },
    {
      "epoch": 2.821255282516826,
      "grad_norm": 6.52615213394165,
      "learning_rate": 2.9881780614113094e-06,
      "loss": 1.0453,
      "step": 36050
    },
    {
      "epoch": 2.8220378776021287,
      "grad_norm": 13.291709899902344,
      "learning_rate": 2.9751006957158553e-06,
      "loss": 0.4998,
      "step": 36060
    },
    {
      "epoch": 2.8228204726874315,
      "grad_norm": 2.1286838054656982,
      "learning_rate": 2.9620233300204007e-06,
      "loss": 0.4986,
      "step": 36070
    },
    {
      "epoch": 2.8236030677727344,
      "grad_norm": 4.156805038452148,
      "learning_rate": 2.9489459643249465e-06,
      "loss": 0.789,
      "step": 36080
    },
    {
      "epoch": 2.8243856628580373,
      "grad_norm": 18.90001106262207,
      "learning_rate": 2.935868598629492e-06,
      "loss": 1.1047,
      "step": 36090
    },
    {
      "epoch": 2.82516825794334,
      "grad_norm": 10.534180641174316,
      "learning_rate": 2.922791232934038e-06,
      "loss": 0.5624,
      "step": 36100
    },
    {
      "epoch": 2.825950853028643,
      "grad_norm": 1.1399853229522705,
      "learning_rate": 2.9097138672385836e-06,
      "loss": 0.5094,
      "step": 36110
    },
    {
      "epoch": 2.826733448113946,
      "grad_norm": 12.93802547454834,
      "learning_rate": 2.8966365015431295e-06,
      "loss": 0.7302,
      "step": 36120
    },
    {
      "epoch": 2.8275160431992488,
      "grad_norm": 19.900774002075195,
      "learning_rate": 2.883559135847675e-06,
      "loss": 0.6775,
      "step": 36130
    },
    {
      "epoch": 2.8282986382845516,
      "grad_norm": 0.6332023739814758,
      "learning_rate": 2.8704817701522207e-06,
      "loss": 0.3651,
      "step": 36140
    },
    {
      "epoch": 2.8290812333698545,
      "grad_norm": 11.84322452545166,
      "learning_rate": 2.857404404456766e-06,
      "loss": 0.8455,
      "step": 36150
    },
    {
      "epoch": 2.8298638284551574,
      "grad_norm": 20.553375244140625,
      "learning_rate": 2.844327038761312e-06,
      "loss": 0.7464,
      "step": 36160
    },
    {
      "epoch": 2.8306464235404603,
      "grad_norm": 10.160680770874023,
      "learning_rate": 2.8312496730658574e-06,
      "loss": 0.7613,
      "step": 36170
    },
    {
      "epoch": 2.831429018625763,
      "grad_norm": 0.9732387661933899,
      "learning_rate": 2.8181723073704037e-06,
      "loss": 0.479,
      "step": 36180
    },
    {
      "epoch": 2.832211613711066,
      "grad_norm": 10.3092679977417,
      "learning_rate": 2.805094941674949e-06,
      "loss": 0.9746,
      "step": 36190
    },
    {
      "epoch": 2.832994208796369,
      "grad_norm": 8.308418273925781,
      "learning_rate": 2.792017575979495e-06,
      "loss": 0.4973,
      "step": 36200
    },
    {
      "epoch": 2.8337768038816717,
      "grad_norm": 6.229053497314453,
      "learning_rate": 2.7789402102840404e-06,
      "loss": 0.9205,
      "step": 36210
    },
    {
      "epoch": 2.8345593989669746,
      "grad_norm": 12.39807415008545,
      "learning_rate": 2.7658628445885862e-06,
      "loss": 0.7354,
      "step": 36220
    },
    {
      "epoch": 2.8353419940522775,
      "grad_norm": 4.701016902923584,
      "learning_rate": 2.7527854788931316e-06,
      "loss": 0.7653,
      "step": 36230
    },
    {
      "epoch": 2.8361245891375804,
      "grad_norm": 3.694126605987549,
      "learning_rate": 2.7397081131976775e-06,
      "loss": 0.6514,
      "step": 36240
    },
    {
      "epoch": 2.8369071842228832,
      "grad_norm": 5.573387622833252,
      "learning_rate": 2.7266307475022233e-06,
      "loss": 0.5515,
      "step": 36250
    },
    {
      "epoch": 2.837689779308186,
      "grad_norm": 10.240132331848145,
      "learning_rate": 2.713553381806769e-06,
      "loss": 0.9128,
      "step": 36260
    },
    {
      "epoch": 2.838472374393489,
      "grad_norm": 6.564284324645996,
      "learning_rate": 2.7004760161113146e-06,
      "loss": 0.8964,
      "step": 36270
    },
    {
      "epoch": 2.839254969478792,
      "grad_norm": 4.337865829467773,
      "learning_rate": 2.6873986504158604e-06,
      "loss": 0.69,
      "step": 36280
    },
    {
      "epoch": 2.8400375645640947,
      "grad_norm": 9.228569984436035,
      "learning_rate": 2.674321284720406e-06,
      "loss": 0.6923,
      "step": 36290
    },
    {
      "epoch": 2.8408201596493976,
      "grad_norm": 1.382559061050415,
      "learning_rate": 2.6612439190249517e-06,
      "loss": 0.8544,
      "step": 36300
    },
    {
      "epoch": 2.8416027547347005,
      "grad_norm": 22.1043758392334,
      "learning_rate": 2.6481665533294975e-06,
      "loss": 1.1101,
      "step": 36310
    },
    {
      "epoch": 2.8423853498200033,
      "grad_norm": 7.525698661804199,
      "learning_rate": 2.635089187634043e-06,
      "loss": 0.5159,
      "step": 36320
    },
    {
      "epoch": 2.843167944905306,
      "grad_norm": 1.8359074592590332,
      "learning_rate": 2.622011821938589e-06,
      "loss": 0.5553,
      "step": 36330
    },
    {
      "epoch": 2.843950539990609,
      "grad_norm": 5.866932392120361,
      "learning_rate": 2.6089344562431346e-06,
      "loss": 0.6956,
      "step": 36340
    },
    {
      "epoch": 2.844733135075912,
      "grad_norm": 9.528510093688965,
      "learning_rate": 2.5958570905476805e-06,
      "loss": 0.8587,
      "step": 36350
    },
    {
      "epoch": 2.845515730161215,
      "grad_norm": 11.096677780151367,
      "learning_rate": 2.582779724852226e-06,
      "loss": 0.7129,
      "step": 36360
    },
    {
      "epoch": 2.8462983252465177,
      "grad_norm": 0.6579805612564087,
      "learning_rate": 2.5697023591567717e-06,
      "loss": 0.3382,
      "step": 36370
    },
    {
      "epoch": 2.8470809203318206,
      "grad_norm": 10.94510269165039,
      "learning_rate": 2.556624993461317e-06,
      "loss": 0.7215,
      "step": 36380
    },
    {
      "epoch": 2.847863515417123,
      "grad_norm": 6.422784805297852,
      "learning_rate": 2.543547627765863e-06,
      "loss": 0.7032,
      "step": 36390
    },
    {
      "epoch": 2.848646110502426,
      "grad_norm": 6.459847450256348,
      "learning_rate": 2.5304702620704084e-06,
      "loss": 0.9663,
      "step": 36400
    },
    {
      "epoch": 2.8494287055877288,
      "grad_norm": 43.70629119873047,
      "learning_rate": 2.5173928963749547e-06,
      "loss": 0.3453,
      "step": 36410
    },
    {
      "epoch": 2.8502113006730316,
      "grad_norm": 0.5185939073562622,
      "learning_rate": 2.5043155306795e-06,
      "loss": 0.7403,
      "step": 36420
    },
    {
      "epoch": 2.8509938957583345,
      "grad_norm": 10.393421173095703,
      "learning_rate": 2.491238164984046e-06,
      "loss": 0.6828,
      "step": 36430
    },
    {
      "epoch": 2.8517764908436374,
      "grad_norm": 7.958818435668945,
      "learning_rate": 2.4781607992885914e-06,
      "loss": 0.5557,
      "step": 36440
    },
    {
      "epoch": 2.8525590859289403,
      "grad_norm": 3.913233995437622,
      "learning_rate": 2.465083433593137e-06,
      "loss": 0.7947,
      "step": 36450
    },
    {
      "epoch": 2.853341681014243,
      "grad_norm": 10.085615158081055,
      "learning_rate": 2.4520060678976826e-06,
      "loss": 0.7166,
      "step": 36460
    },
    {
      "epoch": 2.854124276099546,
      "grad_norm": 8.217813491821289,
      "learning_rate": 2.4389287022022285e-06,
      "loss": 0.8008,
      "step": 36470
    },
    {
      "epoch": 2.854906871184849,
      "grad_norm": 11.552096366882324,
      "learning_rate": 2.4258513365067743e-06,
      "loss": 0.7586,
      "step": 36480
    },
    {
      "epoch": 2.8556894662701517,
      "grad_norm": 32.847232818603516,
      "learning_rate": 2.41277397081132e-06,
      "loss": 0.6063,
      "step": 36490
    },
    {
      "epoch": 2.8564720613554546,
      "grad_norm": 6.513487815856934,
      "learning_rate": 2.3996966051158656e-06,
      "loss": 1.0029,
      "step": 36500
    },
    {
      "epoch": 2.8572546564407575,
      "grad_norm": 11.472500801086426,
      "learning_rate": 2.3866192394204114e-06,
      "loss": 0.6903,
      "step": 36510
    },
    {
      "epoch": 2.8580372515260604,
      "grad_norm": 18.477832794189453,
      "learning_rate": 2.373541873724957e-06,
      "loss": 1.0837,
      "step": 36520
    },
    {
      "epoch": 2.8588198466113632,
      "grad_norm": 1.7773563861846924,
      "learning_rate": 2.3604645080295027e-06,
      "loss": 0.5069,
      "step": 36530
    },
    {
      "epoch": 2.859602441696666,
      "grad_norm": 8.952263832092285,
      "learning_rate": 2.347387142334048e-06,
      "loss": 0.6497,
      "step": 36540
    },
    {
      "epoch": 2.860385036781969,
      "grad_norm": 3.7958877086639404,
      "learning_rate": 2.334309776638594e-06,
      "loss": 0.5308,
      "step": 36550
    },
    {
      "epoch": 2.861167631867272,
      "grad_norm": 1.2016311883926392,
      "learning_rate": 2.32123241094314e-06,
      "loss": 0.4707,
      "step": 36560
    },
    {
      "epoch": 2.8619502269525747,
      "grad_norm": 5.848893165588379,
      "learning_rate": 2.3081550452476856e-06,
      "loss": 0.7327,
      "step": 36570
    },
    {
      "epoch": 2.8627328220378776,
      "grad_norm": 8.186148643493652,
      "learning_rate": 2.295077679552231e-06,
      "loss": 1.1191,
      "step": 36580
    },
    {
      "epoch": 2.8635154171231805,
      "grad_norm": 6.806766033172607,
      "learning_rate": 2.282000313856777e-06,
      "loss": 0.5435,
      "step": 36590
    },
    {
      "epoch": 2.8642980122084833,
      "grad_norm": 22.549304962158203,
      "learning_rate": 2.2689229481613223e-06,
      "loss": 0.7414,
      "step": 36600
    },
    {
      "epoch": 2.865080607293786,
      "grad_norm": 8.620624542236328,
      "learning_rate": 2.255845582465868e-06,
      "loss": 0.6361,
      "step": 36610
    },
    {
      "epoch": 2.865863202379089,
      "grad_norm": 19.13719367980957,
      "learning_rate": 2.2427682167704136e-06,
      "loss": 0.6215,
      "step": 36620
    },
    {
      "epoch": 2.866645797464392,
      "grad_norm": 7.055408477783203,
      "learning_rate": 2.22969085107496e-06,
      "loss": 0.5934,
      "step": 36630
    },
    {
      "epoch": 2.867428392549695,
      "grad_norm": 2.395890474319458,
      "learning_rate": 2.2166134853795053e-06,
      "loss": 0.7807,
      "step": 36640
    },
    {
      "epoch": 2.8682109876349977,
      "grad_norm": 12.845955848693848,
      "learning_rate": 2.203536119684051e-06,
      "loss": 0.9455,
      "step": 36650
    },
    {
      "epoch": 2.8689935827203006,
      "grad_norm": 2.4037046432495117,
      "learning_rate": 2.1904587539885965e-06,
      "loss": 0.5779,
      "step": 36660
    },
    {
      "epoch": 2.8697761778056035,
      "grad_norm": 11.436544418334961,
      "learning_rate": 2.1773813882931424e-06,
      "loss": 0.4622,
      "step": 36670
    },
    {
      "epoch": 2.8705587728909063,
      "grad_norm": 0.7707127928733826,
      "learning_rate": 2.1643040225976878e-06,
      "loss": 0.6885,
      "step": 36680
    },
    {
      "epoch": 2.871341367976209,
      "grad_norm": 11.740891456604004,
      "learning_rate": 2.1512266569022336e-06,
      "loss": 0.6598,
      "step": 36690
    },
    {
      "epoch": 2.872123963061512,
      "grad_norm": 10.085000991821289,
      "learning_rate": 2.1381492912067795e-06,
      "loss": 0.6518,
      "step": 36700
    },
    {
      "epoch": 2.872906558146815,
      "grad_norm": 2.417854070663452,
      "learning_rate": 2.1250719255113253e-06,
      "loss": 0.7228,
      "step": 36710
    },
    {
      "epoch": 2.873689153232118,
      "grad_norm": 14.682175636291504,
      "learning_rate": 2.1119945598158707e-06,
      "loss": 0.8838,
      "step": 36720
    },
    {
      "epoch": 2.8744717483174207,
      "grad_norm": 16.492204666137695,
      "learning_rate": 2.0989171941204166e-06,
      "loss": 0.6851,
      "step": 36730
    },
    {
      "epoch": 2.8752543434027236,
      "grad_norm": 4.876648426055908,
      "learning_rate": 2.085839828424962e-06,
      "loss": 0.3679,
      "step": 36740
    },
    {
      "epoch": 2.8760369384880264,
      "grad_norm": 10.147693634033203,
      "learning_rate": 2.072762462729508e-06,
      "loss": 1.2205,
      "step": 36750
    },
    {
      "epoch": 2.8768195335733293,
      "grad_norm": 3.4383957386016846,
      "learning_rate": 2.0596850970340533e-06,
      "loss": 0.7035,
      "step": 36760
    },
    {
      "epoch": 2.877602128658632,
      "grad_norm": 0.6684929728507996,
      "learning_rate": 2.046607731338599e-06,
      "loss": 0.8609,
      "step": 36770
    },
    {
      "epoch": 2.8783847237439346,
      "grad_norm": 10.68365478515625,
      "learning_rate": 2.033530365643145e-06,
      "loss": 0.4038,
      "step": 36780
    },
    {
      "epoch": 2.8791673188292375,
      "grad_norm": 16.39832305908203,
      "learning_rate": 2.0204529999476908e-06,
      "loss": 0.5823,
      "step": 36790
    },
    {
      "epoch": 2.8799499139145404,
      "grad_norm": 7.605412006378174,
      "learning_rate": 2.007375634252236e-06,
      "loss": 1.0939,
      "step": 36800
    },
    {
      "epoch": 2.8807325089998432,
      "grad_norm": 22.588748931884766,
      "learning_rate": 1.994298268556782e-06,
      "loss": 0.8249,
      "step": 36810
    },
    {
      "epoch": 2.881515104085146,
      "grad_norm": 8.232193946838379,
      "learning_rate": 1.9812209028613275e-06,
      "loss": 0.8518,
      "step": 36820
    },
    {
      "epoch": 2.882297699170449,
      "grad_norm": 30.50523567199707,
      "learning_rate": 1.9681435371658733e-06,
      "loss": 0.5423,
      "step": 36830
    },
    {
      "epoch": 2.883080294255752,
      "grad_norm": 24.951337814331055,
      "learning_rate": 1.955066171470419e-06,
      "loss": 0.5785,
      "step": 36840
    },
    {
      "epoch": 2.8838628893410547,
      "grad_norm": 55.29884338378906,
      "learning_rate": 1.941988805774965e-06,
      "loss": 0.4887,
      "step": 36850
    },
    {
      "epoch": 2.8846454844263576,
      "grad_norm": 10.484580993652344,
      "learning_rate": 1.9289114400795104e-06,
      "loss": 0.8066,
      "step": 36860
    },
    {
      "epoch": 2.8854280795116605,
      "grad_norm": 2.5673046112060547,
      "learning_rate": 1.9158340743840563e-06,
      "loss": 0.7534,
      "step": 36870
    },
    {
      "epoch": 2.8862106745969633,
      "grad_norm": 15.462202072143555,
      "learning_rate": 1.9027567086886019e-06,
      "loss": 0.5282,
      "step": 36880
    },
    {
      "epoch": 2.886993269682266,
      "grad_norm": 11.608144760131836,
      "learning_rate": 1.8896793429931475e-06,
      "loss": 0.6862,
      "step": 36890
    },
    {
      "epoch": 2.887775864767569,
      "grad_norm": 1.8110872507095337,
      "learning_rate": 1.8766019772976932e-06,
      "loss": 0.5707,
      "step": 36900
    },
    {
      "epoch": 2.888558459852872,
      "grad_norm": 7.837405204772949,
      "learning_rate": 1.8635246116022388e-06,
      "loss": 0.6571,
      "step": 36910
    },
    {
      "epoch": 2.889341054938175,
      "grad_norm": 2.4520745277404785,
      "learning_rate": 1.8504472459067844e-06,
      "loss": 0.5447,
      "step": 36920
    },
    {
      "epoch": 2.8901236500234777,
      "grad_norm": 11.997979164123535,
      "learning_rate": 1.8373698802113305e-06,
      "loss": 1.2384,
      "step": 36930
    },
    {
      "epoch": 2.8909062451087806,
      "grad_norm": 2.346130609512329,
      "learning_rate": 1.824292514515876e-06,
      "loss": 0.7394,
      "step": 36940
    },
    {
      "epoch": 2.8916888401940835,
      "grad_norm": 11.07087516784668,
      "learning_rate": 1.8112151488204217e-06,
      "loss": 0.6234,
      "step": 36950
    },
    {
      "epoch": 2.8924714352793863,
      "grad_norm": 8.48650074005127,
      "learning_rate": 1.7981377831249674e-06,
      "loss": 0.7645,
      "step": 36960
    },
    {
      "epoch": 2.893254030364689,
      "grad_norm": 10.063311576843262,
      "learning_rate": 1.785060417429513e-06,
      "loss": 0.6843,
      "step": 36970
    },
    {
      "epoch": 2.894036625449992,
      "grad_norm": 7.3756279945373535,
      "learning_rate": 1.7719830517340586e-06,
      "loss": 0.6582,
      "step": 36980
    },
    {
      "epoch": 2.894819220535295,
      "grad_norm": 23.006959915161133,
      "learning_rate": 1.7589056860386043e-06,
      "loss": 0.6322,
      "step": 36990
    },
    {
      "epoch": 2.895601815620598,
      "grad_norm": 1.1195987462997437,
      "learning_rate": 1.7458283203431503e-06,
      "loss": 0.3504,
      "step": 37000
    },
    {
      "epoch": 2.8963844107059007,
      "grad_norm": 7.376308441162109,
      "learning_rate": 1.732750954647696e-06,
      "loss": 0.6234,
      "step": 37010
    },
    {
      "epoch": 2.8971670057912036,
      "grad_norm": 7.504573345184326,
      "learning_rate": 1.7196735889522416e-06,
      "loss": 1.2236,
      "step": 37020
    },
    {
      "epoch": 2.8979496008765064,
      "grad_norm": 19.947460174560547,
      "learning_rate": 1.7065962232567872e-06,
      "loss": 0.8845,
      "step": 37030
    },
    {
      "epoch": 2.8987321959618093,
      "grad_norm": 5.4673237800598145,
      "learning_rate": 1.6935188575613328e-06,
      "loss": 0.5592,
      "step": 37040
    },
    {
      "epoch": 2.899514791047112,
      "grad_norm": 3.2176737785339355,
      "learning_rate": 1.6804414918658785e-06,
      "loss": 0.6401,
      "step": 37050
    },
    {
      "epoch": 2.900297386132415,
      "grad_norm": 10.98465633392334,
      "learning_rate": 1.667364126170424e-06,
      "loss": 0.9713,
      "step": 37060
    },
    {
      "epoch": 2.901079981217718,
      "grad_norm": 8.362898826599121,
      "learning_rate": 1.6542867604749697e-06,
      "loss": 0.8393,
      "step": 37070
    },
    {
      "epoch": 2.901862576303021,
      "grad_norm": 29.148468017578125,
      "learning_rate": 1.6412093947795158e-06,
      "loss": 0.7698,
      "step": 37080
    },
    {
      "epoch": 2.9026451713883237,
      "grad_norm": 9.009899139404297,
      "learning_rate": 1.6281320290840614e-06,
      "loss": 0.3981,
      "step": 37090
    },
    {
      "epoch": 2.9034277664736265,
      "grad_norm": 1.626450538635254,
      "learning_rate": 1.615054663388607e-06,
      "loss": 0.4814,
      "step": 37100
    },
    {
      "epoch": 2.9042103615589294,
      "grad_norm": 8.096089363098145,
      "learning_rate": 1.6019772976931527e-06,
      "loss": 0.7965,
      "step": 37110
    },
    {
      "epoch": 2.9049929566442323,
      "grad_norm": 46.247920989990234,
      "learning_rate": 1.5888999319976983e-06,
      "loss": 0.6648,
      "step": 37120
    },
    {
      "epoch": 2.905775551729535,
      "grad_norm": 4.5413713455200195,
      "learning_rate": 1.5758225663022441e-06,
      "loss": 0.5054,
      "step": 37130
    },
    {
      "epoch": 2.906558146814838,
      "grad_norm": 24.517478942871094,
      "learning_rate": 1.5627452006067898e-06,
      "loss": 0.665,
      "step": 37140
    },
    {
      "epoch": 2.907340741900141,
      "grad_norm": 1.6872478723526,
      "learning_rate": 1.5496678349113354e-06,
      "loss": 0.794,
      "step": 37150
    },
    {
      "epoch": 2.908123336985444,
      "grad_norm": 2.3681888580322266,
      "learning_rate": 1.5365904692158813e-06,
      "loss": 0.7217,
      "step": 37160
    },
    {
      "epoch": 2.9089059320707467,
      "grad_norm": 19.591007232666016,
      "learning_rate": 1.5235131035204269e-06,
      "loss": 0.7224,
      "step": 37170
    },
    {
      "epoch": 2.9096885271560495,
      "grad_norm": 10.025452613830566,
      "learning_rate": 1.5104357378249727e-06,
      "loss": 0.5765,
      "step": 37180
    },
    {
      "epoch": 2.9104711222413524,
      "grad_norm": 7.313584327697754,
      "learning_rate": 1.4973583721295184e-06,
      "loss": 0.7884,
      "step": 37190
    },
    {
      "epoch": 2.9112537173266553,
      "grad_norm": 6.8786773681640625,
      "learning_rate": 1.484281006434064e-06,
      "loss": 0.622,
      "step": 37200
    },
    {
      "epoch": 2.912036312411958,
      "grad_norm": 6.045290470123291,
      "learning_rate": 1.4712036407386098e-06,
      "loss": 0.5787,
      "step": 37210
    },
    {
      "epoch": 2.912818907497261,
      "grad_norm": 11.056214332580566,
      "learning_rate": 1.4581262750431555e-06,
      "loss": 0.8838,
      "step": 37220
    },
    {
      "epoch": 2.913601502582564,
      "grad_norm": 11.080055236816406,
      "learning_rate": 1.445048909347701e-06,
      "loss": 0.2428,
      "step": 37230
    },
    {
      "epoch": 2.9143840976678668,
      "grad_norm": 0.8766770958900452,
      "learning_rate": 1.431971543652247e-06,
      "loss": 0.5648,
      "step": 37240
    },
    {
      "epoch": 2.9151666927531696,
      "grad_norm": 9.81221866607666,
      "learning_rate": 1.4188941779567926e-06,
      "loss": 0.55,
      "step": 37250
    },
    {
      "epoch": 2.9159492878384725,
      "grad_norm": 9.341150283813477,
      "learning_rate": 1.4058168122613382e-06,
      "loss": 0.4677,
      "step": 37260
    },
    {
      "epoch": 2.9167318829237754,
      "grad_norm": 24.369522094726562,
      "learning_rate": 1.3927394465658838e-06,
      "loss": 0.6264,
      "step": 37270
    },
    {
      "epoch": 2.9175144780090783,
      "grad_norm": 6.924059867858887,
      "learning_rate": 1.3796620808704297e-06,
      "loss": 0.9533,
      "step": 37280
    },
    {
      "epoch": 2.918297073094381,
      "grad_norm": 28.886695861816406,
      "learning_rate": 1.3665847151749753e-06,
      "loss": 1.0318,
      "step": 37290
    },
    {
      "epoch": 2.919079668179684,
      "grad_norm": 4.513106346130371,
      "learning_rate": 1.353507349479521e-06,
      "loss": 0.454,
      "step": 37300
    },
    {
      "epoch": 2.919862263264987,
      "grad_norm": 2.744166374206543,
      "learning_rate": 1.3404299837840666e-06,
      "loss": 0.553,
      "step": 37310
    },
    {
      "epoch": 2.9206448583502898,
      "grad_norm": 33.084415435791016,
      "learning_rate": 1.3273526180886124e-06,
      "loss": 0.9027,
      "step": 37320
    },
    {
      "epoch": 2.9214274534355926,
      "grad_norm": 11.021342277526855,
      "learning_rate": 1.314275252393158e-06,
      "loss": 0.4138,
      "step": 37330
    },
    {
      "epoch": 2.9222100485208955,
      "grad_norm": 3.305356979370117,
      "learning_rate": 1.3011978866977037e-06,
      "loss": 0.5039,
      "step": 37340
    },
    {
      "epoch": 2.9229926436061984,
      "grad_norm": 0.6287873983383179,
      "learning_rate": 1.2881205210022495e-06,
      "loss": 0.801,
      "step": 37350
    },
    {
      "epoch": 2.9237752386915012,
      "grad_norm": 10.81692123413086,
      "learning_rate": 1.2750431553067951e-06,
      "loss": 0.7862,
      "step": 37360
    },
    {
      "epoch": 2.924557833776804,
      "grad_norm": 29.80510139465332,
      "learning_rate": 1.2619657896113408e-06,
      "loss": 0.588,
      "step": 37370
    },
    {
      "epoch": 2.925340428862107,
      "grad_norm": 11.335555076599121,
      "learning_rate": 1.2488884239158864e-06,
      "loss": 0.9256,
      "step": 37380
    },
    {
      "epoch": 2.92612302394741,
      "grad_norm": 6.229099750518799,
      "learning_rate": 1.2358110582204323e-06,
      "loss": 0.4196,
      "step": 37390
    },
    {
      "epoch": 2.9269056190327127,
      "grad_norm": 13.83948040008545,
      "learning_rate": 1.2227336925249779e-06,
      "loss": 0.7652,
      "step": 37400
    },
    {
      "epoch": 2.9276882141180156,
      "grad_norm": 1.8511555194854736,
      "learning_rate": 1.2096563268295235e-06,
      "loss": 0.5196,
      "step": 37410
    },
    {
      "epoch": 2.928470809203318,
      "grad_norm": 0.9737424254417419,
      "learning_rate": 1.1965789611340691e-06,
      "loss": 0.4701,
      "step": 37420
    },
    {
      "epoch": 2.929253404288621,
      "grad_norm": 6.3484673500061035,
      "learning_rate": 1.183501595438615e-06,
      "loss": 0.6345,
      "step": 37430
    },
    {
      "epoch": 2.930035999373924,
      "grad_norm": 13.76927375793457,
      "learning_rate": 1.1704242297431606e-06,
      "loss": 0.9296,
      "step": 37440
    },
    {
      "epoch": 2.9308185944592267,
      "grad_norm": 4.543705463409424,
      "learning_rate": 1.1573468640477062e-06,
      "loss": 0.8371,
      "step": 37450
    },
    {
      "epoch": 2.9316011895445295,
      "grad_norm": 2.433224678039551,
      "learning_rate": 1.1442694983522519e-06,
      "loss": 0.651,
      "step": 37460
    },
    {
      "epoch": 2.9323837846298324,
      "grad_norm": 14.282033920288086,
      "learning_rate": 1.1311921326567977e-06,
      "loss": 1.2171,
      "step": 37470
    },
    {
      "epoch": 2.9331663797151353,
      "grad_norm": 9.552964210510254,
      "learning_rate": 1.1181147669613434e-06,
      "loss": 0.6731,
      "step": 37480
    },
    {
      "epoch": 2.933948974800438,
      "grad_norm": 12.107645988464355,
      "learning_rate": 1.105037401265889e-06,
      "loss": 0.6799,
      "step": 37490
    },
    {
      "epoch": 2.934731569885741,
      "grad_norm": 9.431314468383789,
      "learning_rate": 1.0919600355704348e-06,
      "loss": 0.6832,
      "step": 37500
    },
    {
      "epoch": 2.935514164971044,
      "grad_norm": 20.994592666625977,
      "learning_rate": 1.0788826698749805e-06,
      "loss": 0.8142,
      "step": 37510
    },
    {
      "epoch": 2.9362967600563468,
      "grad_norm": 3.879408121109009,
      "learning_rate": 1.065805304179526e-06,
      "loss": 0.7382,
      "step": 37520
    },
    {
      "epoch": 2.9370793551416496,
      "grad_norm": 11.126795768737793,
      "learning_rate": 1.0527279384840717e-06,
      "loss": 0.8417,
      "step": 37530
    },
    {
      "epoch": 2.9378619502269525,
      "grad_norm": 1.0176334381103516,
      "learning_rate": 1.0396505727886176e-06,
      "loss": 0.5563,
      "step": 37540
    },
    {
      "epoch": 2.9386445453122554,
      "grad_norm": 10.815564155578613,
      "learning_rate": 1.0265732070931632e-06,
      "loss": 1.0177,
      "step": 37550
    },
    {
      "epoch": 2.9394271403975583,
      "grad_norm": 8.465948104858398,
      "learning_rate": 1.0134958413977088e-06,
      "loss": 0.8493,
      "step": 37560
    },
    {
      "epoch": 2.940209735482861,
      "grad_norm": 2.53523588180542,
      "learning_rate": 1.0004184757022545e-06,
      "loss": 0.6103,
      "step": 37570
    },
    {
      "epoch": 2.940992330568164,
      "grad_norm": 17.025314331054688,
      "learning_rate": 9.873411100068003e-07,
      "loss": 0.6844,
      "step": 37580
    },
    {
      "epoch": 2.941774925653467,
      "grad_norm": 5.340662479400635,
      "learning_rate": 9.74263744311346e-07,
      "loss": 0.5071,
      "step": 37590
    },
    {
      "epoch": 2.9425575207387697,
      "grad_norm": 10.797042846679688,
      "learning_rate": 9.611863786158916e-07,
      "loss": 0.8308,
      "step": 37600
    },
    {
      "epoch": 2.9433401158240726,
      "grad_norm": 4.851880073547363,
      "learning_rate": 9.481090129204373e-07,
      "loss": 0.4165,
      "step": 37610
    },
    {
      "epoch": 2.9441227109093755,
      "grad_norm": 11.61478042602539,
      "learning_rate": 9.35031647224983e-07,
      "loss": 0.903,
      "step": 37620
    },
    {
      "epoch": 2.9449053059946784,
      "grad_norm": 14.066832542419434,
      "learning_rate": 9.219542815295288e-07,
      "loss": 0.9362,
      "step": 37630
    },
    {
      "epoch": 2.9456879010799812,
      "grad_norm": 7.411101818084717,
      "learning_rate": 9.088769158340744e-07,
      "loss": 0.3218,
      "step": 37640
    },
    {
      "epoch": 2.946470496165284,
      "grad_norm": 11.522933959960938,
      "learning_rate": 8.957995501386201e-07,
      "loss": 0.9958,
      "step": 37650
    },
    {
      "epoch": 2.947253091250587,
      "grad_norm": 0.8049660921096802,
      "learning_rate": 8.827221844431659e-07,
      "loss": 0.729,
      "step": 37660
    },
    {
      "epoch": 2.94803568633589,
      "grad_norm": 7.179473876953125,
      "learning_rate": 8.696448187477115e-07,
      "loss": 0.4898,
      "step": 37670
    },
    {
      "epoch": 2.9488182814211927,
      "grad_norm": 4.486979961395264,
      "learning_rate": 8.565674530522571e-07,
      "loss": 0.8551,
      "step": 37680
    },
    {
      "epoch": 2.9496008765064956,
      "grad_norm": 16.19339942932129,
      "learning_rate": 8.43490087356803e-07,
      "loss": 0.8054,
      "step": 37690
    },
    {
      "epoch": 2.9503834715917985,
      "grad_norm": 25.61888313293457,
      "learning_rate": 8.304127216613486e-07,
      "loss": 0.5929,
      "step": 37700
    },
    {
      "epoch": 2.9511660666771014,
      "grad_norm": 25.413965225219727,
      "learning_rate": 8.173353559658942e-07,
      "loss": 0.5678,
      "step": 37710
    },
    {
      "epoch": 2.9519486617624042,
      "grad_norm": 19.199832916259766,
      "learning_rate": 8.042579902704399e-07,
      "loss": 0.5259,
      "step": 37720
    },
    {
      "epoch": 2.952731256847707,
      "grad_norm": 8.515106201171875,
      "learning_rate": 7.911806245749857e-07,
      "loss": 0.5383,
      "step": 37730
    },
    {
      "epoch": 2.95351385193301,
      "grad_norm": 12.9583740234375,
      "learning_rate": 7.781032588795314e-07,
      "loss": 0.3814,
      "step": 37740
    },
    {
      "epoch": 2.954296447018313,
      "grad_norm": 7.751794338226318,
      "learning_rate": 7.650258931840771e-07,
      "loss": 0.8422,
      "step": 37750
    },
    {
      "epoch": 2.9550790421036157,
      "grad_norm": 15.403846740722656,
      "learning_rate": 7.519485274886227e-07,
      "loss": 0.6741,
      "step": 37760
    },
    {
      "epoch": 2.9558616371889186,
      "grad_norm": 13.252533912658691,
      "learning_rate": 7.388711617931685e-07,
      "loss": 0.4346,
      "step": 37770
    },
    {
      "epoch": 2.9566442322742215,
      "grad_norm": 10.873077392578125,
      "learning_rate": 7.257937960977141e-07,
      "loss": 0.7645,
      "step": 37780
    },
    {
      "epoch": 2.9574268273595243,
      "grad_norm": 1.1702799797058105,
      "learning_rate": 7.127164304022598e-07,
      "loss": 0.5728,
      "step": 37790
    },
    {
      "epoch": 2.958209422444827,
      "grad_norm": 23.273025512695312,
      "learning_rate": 6.996390647068055e-07,
      "loss": 0.7844,
      "step": 37800
    },
    {
      "epoch": 2.9589920175301296,
      "grad_norm": 1.877732515335083,
      "learning_rate": 6.865616990113512e-07,
      "loss": 0.6773,
      "step": 37810
    },
    {
      "epoch": 2.9597746126154325,
      "grad_norm": 21.530946731567383,
      "learning_rate": 6.734843333158968e-07,
      "loss": 0.8026,
      "step": 37820
    },
    {
      "epoch": 2.9605572077007354,
      "grad_norm": 10.222853660583496,
      "learning_rate": 6.604069676204426e-07,
      "loss": 0.809,
      "step": 37830
    },
    {
      "epoch": 2.9613398027860383,
      "grad_norm": 12.760886192321777,
      "learning_rate": 6.473296019249882e-07,
      "loss": 0.7933,
      "step": 37840
    },
    {
      "epoch": 2.962122397871341,
      "grad_norm": 1.4226796627044678,
      "learning_rate": 6.342522362295339e-07,
      "loss": 0.7135,
      "step": 37850
    },
    {
      "epoch": 2.962904992956644,
      "grad_norm": 17.69437026977539,
      "learning_rate": 6.211748705340796e-07,
      "loss": 0.8218,
      "step": 37860
    },
    {
      "epoch": 2.963687588041947,
      "grad_norm": 10.599156379699707,
      "learning_rate": 6.080975048386253e-07,
      "loss": 0.9576,
      "step": 37870
    },
    {
      "epoch": 2.9644701831272497,
      "grad_norm": 13.199945449829102,
      "learning_rate": 5.95020139143171e-07,
      "loss": 0.9726,
      "step": 37880
    },
    {
      "epoch": 2.9652527782125526,
      "grad_norm": 2.0103464126586914,
      "learning_rate": 5.819427734477167e-07,
      "loss": 0.8316,
      "step": 37890
    },
    {
      "epoch": 2.9660353732978555,
      "grad_norm": 17.143587112426758,
      "learning_rate": 5.688654077522624e-07,
      "loss": 0.6332,
      "step": 37900
    },
    {
      "epoch": 2.9668179683831584,
      "grad_norm": 29.060230255126953,
      "learning_rate": 5.557880420568081e-07,
      "loss": 0.8476,
      "step": 37910
    },
    {
      "epoch": 2.9676005634684612,
      "grad_norm": 25.74798011779785,
      "learning_rate": 5.427106763613539e-07,
      "loss": 1.0869,
      "step": 37920
    },
    {
      "epoch": 2.968383158553764,
      "grad_norm": 1.0243486166000366,
      "learning_rate": 5.296333106658995e-07,
      "loss": 0.6742,
      "step": 37930
    },
    {
      "epoch": 2.969165753639067,
      "grad_norm": 17.20146942138672,
      "learning_rate": 5.165559449704452e-07,
      "loss": 0.7117,
      "step": 37940
    },
    {
      "epoch": 2.96994834872437,
      "grad_norm": 18.029983520507812,
      "learning_rate": 5.034785792749909e-07,
      "loss": 0.6885,
      "step": 37950
    },
    {
      "epoch": 2.9707309438096727,
      "grad_norm": 2.302759885787964,
      "learning_rate": 4.904012135795366e-07,
      "loss": 0.6716,
      "step": 37960
    },
    {
      "epoch": 2.9715135388949756,
      "grad_norm": 10.953829765319824,
      "learning_rate": 4.773238478840822e-07,
      "loss": 0.5748,
      "step": 37970
    },
    {
      "epoch": 2.9722961339802785,
      "grad_norm": 17.09203338623047,
      "learning_rate": 4.64246482188628e-07,
      "loss": 0.9346,
      "step": 37980
    },
    {
      "epoch": 2.9730787290655813,
      "grad_norm": 20.471723556518555,
      "learning_rate": 4.511691164931736e-07,
      "loss": 0.9338,
      "step": 37990
    },
    {
      "epoch": 2.9738613241508842,
      "grad_norm": 5.635737895965576,
      "learning_rate": 4.3809175079771935e-07,
      "loss": 0.6251,
      "step": 38000
    },
    {
      "epoch": 2.974643919236187,
      "grad_norm": 1.0423531532287598,
      "learning_rate": 4.250143851022651e-07,
      "loss": 0.8691,
      "step": 38010
    },
    {
      "epoch": 2.97542651432149,
      "grad_norm": 10.607963562011719,
      "learning_rate": 4.119370194068107e-07,
      "loss": 0.5708,
      "step": 38020
    },
    {
      "epoch": 2.976209109406793,
      "grad_norm": 7.223966121673584,
      "learning_rate": 3.9885965371135645e-07,
      "loss": 0.7103,
      "step": 38030
    },
    {
      "epoch": 2.9769917044920957,
      "grad_norm": 4.935091972351074,
      "learning_rate": 3.857822880159021e-07,
      "loss": 0.6837,
      "step": 38040
    },
    {
      "epoch": 2.9777742995773986,
      "grad_norm": 3.386044502258301,
      "learning_rate": 3.7270492232044777e-07,
      "loss": 0.5959,
      "step": 38050
    },
    {
      "epoch": 2.9785568946627015,
      "grad_norm": 8.097821235656738,
      "learning_rate": 3.5962755662499345e-07,
      "loss": 1.0133,
      "step": 38060
    },
    {
      "epoch": 2.9793394897480043,
      "grad_norm": 3.4774656295776367,
      "learning_rate": 3.4655019092953914e-07,
      "loss": 0.4002,
      "step": 38070
    },
    {
      "epoch": 2.980122084833307,
      "grad_norm": 12.345613479614258,
      "learning_rate": 3.3347282523408487e-07,
      "loss": 0.9552,
      "step": 38080
    },
    {
      "epoch": 2.98090467991861,
      "grad_norm": 10.337151527404785,
      "learning_rate": 3.2039545953863056e-07,
      "loss": 0.7397,
      "step": 38090
    },
    {
      "epoch": 2.981687275003913,
      "grad_norm": 1.5771514177322388,
      "learning_rate": 3.0731809384317624e-07,
      "loss": 0.4655,
      "step": 38100
    },
    {
      "epoch": 2.982469870089216,
      "grad_norm": 44.95768737792969,
      "learning_rate": 2.942407281477219e-07,
      "loss": 0.6655,
      "step": 38110
    },
    {
      "epoch": 2.9832524651745187,
      "grad_norm": 10.899580001831055,
      "learning_rate": 2.811633624522676e-07,
      "loss": 0.7744,
      "step": 38120
    },
    {
      "epoch": 2.9840350602598216,
      "grad_norm": 3.731318712234497,
      "learning_rate": 2.6808599675681335e-07,
      "loss": 0.291,
      "step": 38130
    },
    {
      "epoch": 2.9848176553451244,
      "grad_norm": 16.743480682373047,
      "learning_rate": 2.5500863106135903e-07,
      "loss": 0.5029,
      "step": 38140
    },
    {
      "epoch": 2.9856002504304273,
      "grad_norm": 9.426115989685059,
      "learning_rate": 2.419312653659047e-07,
      "loss": 0.6364,
      "step": 38150
    },
    {
      "epoch": 2.98638284551573,
      "grad_norm": 0.5100630521774292,
      "learning_rate": 2.288538996704504e-07,
      "loss": 0.8278,
      "step": 38160
    },
    {
      "epoch": 2.987165440601033,
      "grad_norm": 15.931503295898438,
      "learning_rate": 2.1577653397499608e-07,
      "loss": 1.0194,
      "step": 38170
    },
    {
      "epoch": 2.987948035686336,
      "grad_norm": 5.2312211990356445,
      "learning_rate": 2.0269916827954177e-07,
      "loss": 0.592,
      "step": 38180
    },
    {
      "epoch": 2.988730630771639,
      "grad_norm": 16.656484603881836,
      "learning_rate": 1.8962180258408748e-07,
      "loss": 0.4707,
      "step": 38190
    },
    {
      "epoch": 2.9895132258569417,
      "grad_norm": 7.3249335289001465,
      "learning_rate": 1.7654443688863316e-07,
      "loss": 0.245,
      "step": 38200
    },
    {
      "epoch": 2.9902958209422446,
      "grad_norm": 12.254904747009277,
      "learning_rate": 1.6346707119317884e-07,
      "loss": 0.68,
      "step": 38210
    },
    {
      "epoch": 2.9910784160275474,
      "grad_norm": 13.061176300048828,
      "learning_rate": 1.5038970549772453e-07,
      "loss": 0.9209,
      "step": 38220
    },
    {
      "epoch": 2.9918610111128503,
      "grad_norm": 1.4836596250534058,
      "learning_rate": 1.3731233980227024e-07,
      "loss": 0.3928,
      "step": 38230
    },
    {
      "epoch": 2.992643606198153,
      "grad_norm": 4.143230438232422,
      "learning_rate": 1.2423497410681595e-07,
      "loss": 0.8556,
      "step": 38240
    },
    {
      "epoch": 2.993426201283456,
      "grad_norm": 1.0740455389022827,
      "learning_rate": 1.1115760841136163e-07,
      "loss": 0.4114,
      "step": 38250
    },
    {
      "epoch": 2.994208796368759,
      "grad_norm": 0.39830392599105835,
      "learning_rate": 9.808024271590732e-08,
      "loss": 0.9712,
      "step": 38260
    },
    {
      "epoch": 2.994991391454062,
      "grad_norm": 14.544798851013184,
      "learning_rate": 8.5002877020453e-08,
      "loss": 0.6454,
      "step": 38270
    },
    {
      "epoch": 2.9957739865393647,
      "grad_norm": 4.2745561599731445,
      "learning_rate": 7.19255113249987e-08,
      "loss": 0.7762,
      "step": 38280
    },
    {
      "epoch": 2.9965565816246675,
      "grad_norm": 2.329939126968384,
      "learning_rate": 5.884814562954439e-08,
      "loss": 0.3357,
      "step": 38290
    },
    {
      "epoch": 2.9973391767099704,
      "grad_norm": 9.065643310546875,
      "learning_rate": 4.577077993409008e-08,
      "loss": 0.8252,
      "step": 38300
    },
    {
      "epoch": 2.9981217717952733,
      "grad_norm": 17.068172454833984,
      "learning_rate": 3.269341423863577e-08,
      "loss": 0.829,
      "step": 38310
    },
    {
      "epoch": 2.998904366880576,
      "grad_norm": 18.842267990112305,
      "learning_rate": 1.961604854318146e-08,
      "loss": 0.4368,
      "step": 38320
    },
    {
      "epoch": 2.999686961965879,
      "grad_norm": 10.072205543518066,
      "learning_rate": 6.538682847727155e-09,
      "loss": 0.7626,
      "step": 38330
    }
  ],
  "logging_steps": 10,
  "max_steps": 38334,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 5162113160439312.0,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}
