{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.0,
  "eval_steps": 500,
  "global_step": 9480,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0010548523206751054,
      "grad_norm": 1.4955074787139893,
      "learning_rate": 0.00015789473684210527,
      "loss": 7.6087,
      "step": 10
    },
    {
      "epoch": 0.002109704641350211,
      "grad_norm": 1.1852571964263916,
      "learning_rate": 0.00031578947368421053,
      "loss": 7.2343,
      "step": 20
    },
    {
      "epoch": 0.0031645569620253164,
      "grad_norm": 1.4482988119125366,
      "learning_rate": 0.00047368421052631577,
      "loss": 6.8612,
      "step": 30
    },
    {
      "epoch": 0.004219409282700422,
      "grad_norm": 1.4734665155410767,
      "learning_rate": 0.0006315789473684211,
      "loss": 6.4914,
      "step": 40
    },
    {
      "epoch": 0.005274261603375527,
      "grad_norm": 1.0499128103256226,
      "learning_rate": 0.0007894736842105263,
      "loss": 6.2224,
      "step": 50
    },
    {
      "epoch": 0.006329113924050633,
      "grad_norm": 1.2521113157272339,
      "learning_rate": 0.0009473684210526315,
      "loss": 6.1004,
      "step": 60
    },
    {
      "epoch": 0.007383966244725738,
      "grad_norm": 1.7382813692092896,
      "learning_rate": 0.0011052631578947368,
      "loss": 6.2123,
      "step": 70
    },
    {
      "epoch": 0.008438818565400843,
      "grad_norm": 25.527446746826172,
      "learning_rate": 0.0012631578947368421,
      "loss": 6.3898,
      "step": 80
    },
    {
      "epoch": 0.00949367088607595,
      "grad_norm": 7.692363739013672,
      "learning_rate": 0.001405263157894737,
      "loss": 6.5055,
      "step": 90
    },
    {
      "epoch": 0.010548523206751054,
      "grad_norm": 3.504094362258911,
      "learning_rate": 0.0014999993276701756,
      "loss": 6.469,
      "step": 100
    },
    {
      "epoch": 0.011603375527426161,
      "grad_norm": 1.358729362487793,
      "learning_rate": 0.001499991763973493,
      "loss": 6.3882,
      "step": 110
    },
    {
      "epoch": 0.012658227848101266,
      "grad_norm": 27.70503044128418,
      "learning_rate": 0.0014999757962528846,
      "loss": 6.6067,
      "step": 120
    },
    {
      "epoch": 0.013713080168776372,
      "grad_norm": 14.84610366821289,
      "learning_rate": 0.0014999514246872762,
      "loss": 6.353,
      "step": 130
    },
    {
      "epoch": 0.014767932489451477,
      "grad_norm": 1.9949373006820679,
      "learning_rate": 0.0014999186495497636,
      "loss": 6.5599,
      "step": 140
    },
    {
      "epoch": 0.015822784810126583,
      "grad_norm": 16.629016876220703,
      "learning_rate": 0.001499877471207608,
      "loss": 6.1935,
      "step": 150
    },
    {
      "epoch": 0.016877637130801686,
      "grad_norm": 39.98115921020508,
      "learning_rate": 0.0014998332263391192,
      "loss": 6.3391,
      "step": 160
    },
    {
      "epoch": 0.017932489451476793,
      "grad_norm": 4.969245910644531,
      "learning_rate": 0.0014997760832571839,
      "loss": 6.1064,
      "step": 170
    },
    {
      "epoch": 0.0189873417721519,
      "grad_norm": 1.1446038484573364,
      "learning_rate": 0.0014997105385681306,
      "loss": 6.3438,
      "step": 180
    },
    {
      "epoch": 0.020042194092827006,
      "grad_norm": 3.734700918197632,
      "learning_rate": 0.0014996365930064197,
      "loss": 6.4487,
      "step": 190
    },
    {
      "epoch": 0.02109704641350211,
      "grad_norm": 11.851530075073242,
      "learning_rate": 0.001499554247400647,
      "loss": 6.1444,
      "step": 200
    },
    {
      "epoch": 0.022151898734177215,
      "grad_norm": 1.2074213027954102,
      "learning_rate": 0.001499463502673535,
      "loss": 6.2079,
      "step": 210
    },
    {
      "epoch": 0.023206751054852322,
      "grad_norm": 20.981239318847656,
      "learning_rate": 0.0014993643598419234,
      "loss": 6.4194,
      "step": 220
    },
    {
      "epoch": 0.024261603375527425,
      "grad_norm": 7.646417617797852,
      "learning_rate": 0.001499256820016755,
      "loss": 6.0984,
      "step": 230
    },
    {
      "epoch": 0.02531645569620253,
      "grad_norm": 0.9586338996887207,
      "learning_rate": 0.0014991408844030672,
      "loss": 6.2391,
      "step": 240
    },
    {
      "epoch": 0.026371308016877638,
      "grad_norm": 21.232391357421875,
      "learning_rate": 0.0014990165542999746,
      "loss": 6.1045,
      "step": 250
    },
    {
      "epoch": 0.027426160337552744,
      "grad_norm": 47.847408294677734,
      "learning_rate": 0.0014988838311006565,
      "loss": 6.3164,
      "step": 260
    },
    {
      "epoch": 0.028481012658227847,
      "grad_norm": 26.20873260498047,
      "learning_rate": 0.0014987427162923416,
      "loss": 6.2124,
      "step": 270
    },
    {
      "epoch": 0.029535864978902954,
      "grad_norm": 2.2058753967285156,
      "learning_rate": 0.0014985932114562896,
      "loss": 6.4459,
      "step": 280
    },
    {
      "epoch": 0.03059071729957806,
      "grad_norm": 1.8475031852722168,
      "learning_rate": 0.0014984353182677759,
      "loss": 6.1968,
      "step": 290
    },
    {
      "epoch": 0.03164556962025317,
      "grad_norm": Infinity,
      "learning_rate": 0.0014982860438171187,
      "loss": 6.2505,
      "step": 300
    },
    {
      "epoch": 0.03270042194092827,
      "grad_norm": 11.123407363891602,
      "learning_rate": 0.0014981122177109383,
      "loss": 6.3171,
      "step": 310
    },
    {
      "epoch": 0.03375527426160337,
      "grad_norm": 1.9859116077423096,
      "learning_rate": 0.0014979300086420655,
      "loss": 6.0927,
      "step": 320
    },
    {
      "epoch": 0.03481012658227848,
      "grad_norm": 0.975177526473999,
      "learning_rate": 0.001497739418652242,
      "loss": 6.3139,
      "step": 330
    },
    {
      "epoch": 0.035864978902953586,
      "grad_norm": 28.991073608398438,
      "learning_rate": 0.0014975404498771222,
      "loss": 6.1409,
      "step": 340
    },
    {
      "epoch": 0.03691983122362869,
      "grad_norm": 3.932039976119995,
      "learning_rate": 0.0014973331045462475,
      "loss": 6.0396,
      "step": 350
    },
    {
      "epoch": 0.0379746835443038,
      "grad_norm": 9.37112808227539,
      "learning_rate": 0.0014971173849830243,
      "loss": 6.0739,
      "step": 360
    },
    {
      "epoch": 0.039029535864978905,
      "grad_norm": 20.821840286254883,
      "learning_rate": 0.0014968932936046953,
      "loss": 6.2279,
      "step": 370
    },
    {
      "epoch": 0.04008438818565401,
      "grad_norm": 1.634981393814087,
      "learning_rate": 0.0014966608329223137,
      "loss": 6.2517,
      "step": 380
    },
    {
      "epoch": 0.04113924050632911,
      "grad_norm": 0.9171635508537292,
      "learning_rate": 0.0014964200055407153,
      "loss": 6.0175,
      "step": 390
    },
    {
      "epoch": 0.04219409282700422,
      "grad_norm": 24.53272247314453,
      "learning_rate": 0.0014961708141584885,
      "loss": 5.9509,
      "step": 400
    },
    {
      "epoch": 0.043248945147679324,
      "grad_norm": 1.7777763605117798,
      "learning_rate": 0.001495913261567944,
      "loss": 6.0547,
      "step": 410
    },
    {
      "epoch": 0.04430379746835443,
      "grad_norm": 1.304945468902588,
      "learning_rate": 0.0014956473506550845,
      "loss": 5.9514,
      "step": 420
    },
    {
      "epoch": 0.04535864978902954,
      "grad_norm": 1.0723352432250977,
      "learning_rate": 0.001495373084399572,
      "loss": 5.9259,
      "step": 430
    },
    {
      "epoch": 0.046413502109704644,
      "grad_norm": 0.6883535385131836,
      "learning_rate": 0.0014950904658746933,
      "loss": 5.984,
      "step": 440
    },
    {
      "epoch": 0.04746835443037975,
      "grad_norm": 1.6531939506530762,
      "learning_rate": 0.0014947994982473273,
      "loss": 6.0531,
      "step": 450
    },
    {
      "epoch": 0.04852320675105485,
      "grad_norm": 0.6506848931312561,
      "learning_rate": 0.0014945001847779082,
      "loss": 6.2724,
      "step": 460
    },
    {
      "epoch": 0.049578059071729956,
      "grad_norm": 0.9666869640350342,
      "learning_rate": 0.0014941925288203897,
      "loss": 5.8783,
      "step": 470
    },
    {
      "epoch": 0.05063291139240506,
      "grad_norm": 4.7674970626831055,
      "learning_rate": 0.0014938765338222068,
      "loss": 6.0206,
      "step": 480
    },
    {
      "epoch": 0.05168776371308017,
      "grad_norm": 0.7304931282997131,
      "learning_rate": 0.0014935522033242379,
      "loss": 5.904,
      "step": 490
    },
    {
      "epoch": 0.052742616033755275,
      "grad_norm": 11.769916534423828,
      "learning_rate": 0.0014932195409607645,
      "loss": 5.8601,
      "step": 500
    },
    {
      "epoch": 0.05379746835443038,
      "grad_norm": 1.9517804384231567,
      "learning_rate": 0.001492878550459431,
      "loss": 6.1078,
      "step": 510
    },
    {
      "epoch": 0.05485232067510549,
      "grad_norm": 20.521808624267578,
      "learning_rate": 0.0014925292356412025,
      "loss": 5.9196,
      "step": 520
    },
    {
      "epoch": 0.05590717299578059,
      "grad_norm": 26.059223175048828,
      "learning_rate": 0.001492171600420322,
      "loss": 5.9167,
      "step": 530
    },
    {
      "epoch": 0.056962025316455694,
      "grad_norm": 10.85610580444336,
      "learning_rate": 0.0014918056488042665,
      "loss": 5.9687,
      "step": 540
    },
    {
      "epoch": 0.0580168776371308,
      "grad_norm": 2.926618814468384,
      "learning_rate": 0.001491431384893703,
      "loss": 5.8261,
      "step": 550
    },
    {
      "epoch": 0.05907172995780591,
      "grad_norm": 28.76153564453125,
      "learning_rate": 0.0014910488128824409,
      "loss": 5.8701,
      "step": 560
    },
    {
      "epoch": 0.060126582278481014,
      "grad_norm": 0.7511129975318909,
      "learning_rate": 0.0014906579370573868,
      "loss": 5.8192,
      "step": 570
    },
    {
      "epoch": 0.06118143459915612,
      "grad_norm": 1.507708191871643,
      "learning_rate": 0.0014902587617984951,
      "loss": 5.9549,
      "step": 580
    },
    {
      "epoch": 0.06223628691983123,
      "grad_norm": 1.3214479684829712,
      "learning_rate": 0.0014898512915787192,
      "loss": 5.7937,
      "step": 590
    },
    {
      "epoch": 0.06329113924050633,
      "grad_norm": 0.9652126431465149,
      "learning_rate": 0.0014894355309639621,
      "loss": 5.9931,
      "step": 600
    },
    {
      "epoch": 0.06434599156118144,
      "grad_norm": 1.2022321224212646,
      "learning_rate": 0.0014890114846130248,
      "loss": 5.8308,
      "step": 610
    },
    {
      "epoch": 0.06540084388185655,
      "grad_norm": 0.7490571141242981,
      "learning_rate": 0.0014885791572775533,
      "loss": 5.8179,
      "step": 620
    },
    {
      "epoch": 0.06645569620253164,
      "grad_norm": 2.2113516330718994,
      "learning_rate": 0.0014881385538019867,
      "loss": 5.7837,
      "step": 630
    },
    {
      "epoch": 0.06751054852320675,
      "grad_norm": 0.9326190948486328,
      "learning_rate": 0.0014876896791235015,
      "loss": 5.7873,
      "step": 640
    },
    {
      "epoch": 0.06856540084388185,
      "grad_norm": 25.712390899658203,
      "learning_rate": 0.0014872325382719587,
      "loss": 5.8261,
      "step": 650
    },
    {
      "epoch": 0.06962025316455696,
      "grad_norm": 1.984717845916748,
      "learning_rate": 0.001486767136369844,
      "loss": 5.8581,
      "step": 660
    },
    {
      "epoch": 0.07067510548523206,
      "grad_norm": 2.278379440307617,
      "learning_rate": 0.0014862934786322131,
      "loss": 5.8848,
      "step": 670
    },
    {
      "epoch": 0.07172995780590717,
      "grad_norm": 2.4504611492156982,
      "learning_rate": 0.0014858115703666325,
      "loss": 5.7952,
      "step": 680
    },
    {
      "epoch": 0.07278481012658228,
      "grad_norm": 2.039900064468384,
      "learning_rate": 0.001485321416973119,
      "loss": 5.9687,
      "step": 690
    },
    {
      "epoch": 0.07383966244725738,
      "grad_norm": 2.5772364139556885,
      "learning_rate": 0.0014848230239440812,
      "loss": 5.6788,
      "step": 700
    },
    {
      "epoch": 0.07489451476793249,
      "grad_norm": 2.3077950477600098,
      "learning_rate": 0.0014843163968642566,
      "loss": 5.688,
      "step": 710
    },
    {
      "epoch": 0.0759493670886076,
      "grad_norm": 1.388044834136963,
      "learning_rate": 0.0014838015414106486,
      "loss": 5.6627,
      "step": 720
    },
    {
      "epoch": 0.0770042194092827,
      "grad_norm": 1.6177185773849487,
      "learning_rate": 0.0014832784633524638,
      "loss": 5.603,
      "step": 730
    },
    {
      "epoch": 0.07805907172995781,
      "grad_norm": 2.3754689693450928,
      "learning_rate": 0.0014827471685510477,
      "loss": 5.575,
      "step": 740
    },
    {
      "epoch": 0.07911392405063292,
      "grad_norm": 1.5429346561431885,
      "learning_rate": 0.0014822076629598176,
      "loss": 5.5788,
      "step": 750
    },
    {
      "epoch": 0.08016877637130802,
      "grad_norm": 1.5719550848007202,
      "learning_rate": 0.0014816599526241974,
      "loss": 5.5297,
      "step": 760
    },
    {
      "epoch": 0.08122362869198312,
      "grad_norm": 1.4323952198028564,
      "learning_rate": 0.0014811040436815486,
      "loss": 5.4571,
      "step": 770
    },
    {
      "epoch": 0.08227848101265822,
      "grad_norm": 1.3504546880722046,
      "learning_rate": 0.0014805399423611025,
      "loss": 5.4423,
      "step": 780
    },
    {
      "epoch": 0.08333333333333333,
      "grad_norm": 1.519071340560913,
      "learning_rate": 0.0014799676549838898,
      "loss": 5.4328,
      "step": 790
    },
    {
      "epoch": 0.08438818565400844,
      "grad_norm": 1.3911455869674683,
      "learning_rate": 0.00147938718796267,
      "loss": 5.3922,
      "step": 800
    },
    {
      "epoch": 0.08544303797468354,
      "grad_norm": 1.9010822772979736,
      "learning_rate": 0.0014787985478018593,
      "loss": 5.383,
      "step": 810
    },
    {
      "epoch": 0.08649789029535865,
      "grad_norm": 1.662956953048706,
      "learning_rate": 0.0014782017410974583,
      "loss": 5.3926,
      "step": 820
    },
    {
      "epoch": 0.08755274261603375,
      "grad_norm": 1.5445090532302856,
      "learning_rate": 0.0014775967745369778,
      "loss": 5.3807,
      "step": 830
    },
    {
      "epoch": 0.08860759493670886,
      "grad_norm": 1.5067335367202759,
      "learning_rate": 0.0014769836548993631,
      "loss": 5.3539,
      "step": 840
    },
    {
      "epoch": 0.08966244725738397,
      "grad_norm": 1.7864198684692383,
      "learning_rate": 0.0014763623890549193,
      "loss": 5.3637,
      "step": 850
    },
    {
      "epoch": 0.09071729957805907,
      "grad_norm": 5.682808876037598,
      "learning_rate": 0.0014757329839652335,
      "loss": 5.3767,
      "step": 860
    },
    {
      "epoch": 0.09177215189873418,
      "grad_norm": 2.5971896648406982,
      "learning_rate": 0.001475095446683097,
      "loss": 5.3898,
      "step": 870
    },
    {
      "epoch": 0.09282700421940929,
      "grad_norm": 1.902963399887085,
      "learning_rate": 0.0014744497843524266,
      "loss": 5.3684,
      "step": 880
    },
    {
      "epoch": 0.0938818565400844,
      "grad_norm": 1.5862916707992554,
      "learning_rate": 0.0014737960042081836,
      "loss": 5.3468,
      "step": 890
    },
    {
      "epoch": 0.0949367088607595,
      "grad_norm": 2.078481912612915,
      "learning_rate": 0.001473134113576294,
      "loss": 5.3363,
      "step": 900
    },
    {
      "epoch": 0.09599156118143459,
      "grad_norm": 26.264135360717773,
      "learning_rate": 0.0014724641198735659,
      "loss": 5.4566,
      "step": 910
    },
    {
      "epoch": 0.0970464135021097,
      "grad_norm": 17.928512573242188,
      "learning_rate": 0.0014717860306076049,
      "loss": 5.4373,
      "step": 920
    },
    {
      "epoch": 0.0981012658227848,
      "grad_norm": 13.263995170593262,
      "learning_rate": 0.001471099853376733,
      "loss": 5.462,
      "step": 930
    },
    {
      "epoch": 0.09915611814345991,
      "grad_norm": 7.3359456062316895,
      "learning_rate": 0.001470405595869901,
      "loss": 5.4371,
      "step": 940
    },
    {
      "epoch": 0.10021097046413502,
      "grad_norm": 2.4762542247772217,
      "learning_rate": 0.0014697032658666036,
      "loss": 5.46,
      "step": 950
    },
    {
      "epoch": 0.10126582278481013,
      "grad_norm": 4.583887100219727,
      "learning_rate": 0.0014689928712367907,
      "loss": 5.6235,
      "step": 960
    },
    {
      "epoch": 0.10232067510548523,
      "grad_norm": 2.1167221069335938,
      "learning_rate": 0.0014682744199407817,
      "loss": 5.8745,
      "step": 970
    },
    {
      "epoch": 0.10337552742616034,
      "grad_norm": 2.869769334793091,
      "learning_rate": 0.0014675479200291738,
      "loss": 5.6713,
      "step": 980
    },
    {
      "epoch": 0.10443037974683544,
      "grad_norm": 5.200531005859375,
      "learning_rate": 0.0014668133796427532,
      "loss": 5.6873,
      "step": 990
    },
    {
      "epoch": 0.10548523206751055,
      "grad_norm": 3.0656626224517822,
      "learning_rate": 0.0014660708070124038,
      "loss": 5.77,
      "step": 1000
    },
    {
      "epoch": 0.10654008438818566,
      "grad_norm": 1.5934990644454956,
      "learning_rate": 0.0014653202104590146,
      "loss": 5.5021,
      "step": 1010
    },
    {
      "epoch": 0.10759493670886076,
      "grad_norm": 20.099164962768555,
      "learning_rate": 0.001464561598393386,
      "loss": 5.4818,
      "step": 1020
    },
    {
      "epoch": 0.10864978902953587,
      "grad_norm": 8.185047149658203,
      "learning_rate": 0.0014637949793161371,
      "loss": 5.8422,
      "step": 1030
    },
    {
      "epoch": 0.10970464135021098,
      "grad_norm": 2.036337375640869,
      "learning_rate": 0.0014630203618176088,
      "loss": 5.629,
      "step": 1040
    },
    {
      "epoch": 0.11075949367088607,
      "grad_norm": 2.5447497367858887,
      "learning_rate": 0.0014622377545777687,
      "loss": 5.4836,
      "step": 1050
    },
    {
      "epoch": 0.11181434599156118,
      "grad_norm": 5.428991794586182,
      "learning_rate": 0.0014614471663661123,
      "loss": 5.8285,
      "step": 1060
    },
    {
      "epoch": 0.11286919831223628,
      "grad_norm": 3.257345199584961,
      "learning_rate": 0.0014606486060415673,
      "loss": 5.6546,
      "step": 1070
    },
    {
      "epoch": 0.11392405063291139,
      "grad_norm": 6.7466864585876465,
      "learning_rate": 0.0014598420825523918,
      "loss": 5.608,
      "step": 1080
    },
    {
      "epoch": 0.1149789029535865,
      "grad_norm": 3.765772819519043,
      "learning_rate": 0.0014590276049360755,
      "loss": 5.7426,
      "step": 1090
    },
    {
      "epoch": 0.1160337552742616,
      "grad_norm": 2.1478514671325684,
      "learning_rate": 0.0014582051823192374,
      "loss": 5.6029,
      "step": 1100
    },
    {
      "epoch": 0.11708860759493671,
      "grad_norm": 1.7784197330474854,
      "learning_rate": 0.0014573748239175247,
      "loss": 5.4245,
      "step": 1110
    },
    {
      "epoch": 0.11814345991561181,
      "grad_norm": 1.705778956413269,
      "learning_rate": 0.0014565365390355087,
      "loss": 5.5136,
      "step": 1120
    },
    {
      "epoch": 0.11919831223628692,
      "grad_norm": 1.9040672779083252,
      "learning_rate": 0.0014556903370665807,
      "loss": 5.443,
      "step": 1130
    },
    {
      "epoch": 0.12025316455696203,
      "grad_norm": 1.9954731464385986,
      "learning_rate": 0.0014548362274928476,
      "loss": 5.386,
      "step": 1140
    },
    {
      "epoch": 0.12130801687763713,
      "grad_norm": 11.958235740661621,
      "learning_rate": 0.0014539742198850234,
      "loss": 5.3513,
      "step": 1150
    },
    {
      "epoch": 0.12236286919831224,
      "grad_norm": 2.2634782791137695,
      "learning_rate": 0.0014531043239023247,
      "loss": 5.3413,
      "step": 1160
    },
    {
      "epoch": 0.12341772151898735,
      "grad_norm": 2.512418270111084,
      "learning_rate": 0.0014522265492923608,
      "loss": 5.3044,
      "step": 1170
    },
    {
      "epoch": 0.12447257383966245,
      "grad_norm": 4.138009071350098,
      "learning_rate": 0.0014513409058910243,
      "loss": 5.3048,
      "step": 1180
    },
    {
      "epoch": 0.12552742616033755,
      "grad_norm": 2.446187973022461,
      "learning_rate": 0.0014504474036223826,
      "loss": 5.2955,
      "step": 1190
    },
    {
      "epoch": 0.12658227848101267,
      "grad_norm": 2.151524305343628,
      "learning_rate": 0.0014495460524985644,
      "loss": 5.26,
      "step": 1200
    },
    {
      "epoch": 0.12763713080168776,
      "grad_norm": 2.684720993041992,
      "learning_rate": 0.0014486368626196494,
      "loss": 5.2618,
      "step": 1210
    },
    {
      "epoch": 0.12869198312236288,
      "grad_norm": 2.5549938678741455,
      "learning_rate": 0.0014477198441735543,
      "loss": 5.2473,
      "step": 1220
    },
    {
      "epoch": 0.12974683544303797,
      "grad_norm": 2.3496365547180176,
      "learning_rate": 0.001446795007435918,
      "loss": 5.2329,
      "step": 1230
    },
    {
      "epoch": 0.1308016877637131,
      "grad_norm": 1.9384872913360596,
      "learning_rate": 0.0014458623627699883,
      "loss": 5.2268,
      "step": 1240
    },
    {
      "epoch": 0.13185654008438819,
      "grad_norm": 2.716400623321533,
      "learning_rate": 0.0014449219206265041,
      "loss": 5.2578,
      "step": 1250
    },
    {
      "epoch": 0.13291139240506328,
      "grad_norm": 1.674951195716858,
      "learning_rate": 0.0014439736915435786,
      "loss": 5.2682,
      "step": 1260
    },
    {
      "epoch": 0.1339662447257384,
      "grad_norm": 1.987032413482666,
      "learning_rate": 0.0014430176861465812,
      "loss": 5.1841,
      "step": 1270
    },
    {
      "epoch": 0.1350210970464135,
      "grad_norm": 2.017362117767334,
      "learning_rate": 0.0014420539151480199,
      "loss": 5.215,
      "step": 1280
    },
    {
      "epoch": 0.1360759493670886,
      "grad_norm": 1.741590142250061,
      "learning_rate": 0.0014410823893474193,
      "loss": 5.1823,
      "step": 1290
    },
    {
      "epoch": 0.1371308016877637,
      "grad_norm": 2.4790561199188232,
      "learning_rate": 0.0014401031196312,
      "loss": 5.1719,
      "step": 1300
    },
    {
      "epoch": 0.13818565400843882,
      "grad_norm": 2.8530025482177734,
      "learning_rate": 0.0014391161169725573,
      "loss": 5.1794,
      "step": 1310
    },
    {
      "epoch": 0.13924050632911392,
      "grad_norm": 4.470705986022949,
      "learning_rate": 0.0014381213924313386,
      "loss": 5.1886,
      "step": 1320
    },
    {
      "epoch": 0.14029535864978904,
      "grad_norm": 1.5181102752685547,
      "learning_rate": 0.0014371189571539174,
      "loss": 5.2539,
      "step": 1330
    },
    {
      "epoch": 0.14135021097046413,
      "grad_norm": 1.5661035776138306,
      "learning_rate": 0.0014361088223730704,
      "loss": 5.3015,
      "step": 1340
    },
    {
      "epoch": 0.14240506329113925,
      "grad_norm": 2.310349941253662,
      "learning_rate": 0.0014350909994078516,
      "loss": 5.2073,
      "step": 1350
    },
    {
      "epoch": 0.14345991561181434,
      "grad_norm": 4.454345703125,
      "learning_rate": 0.001434065499663464,
      "loss": 5.202,
      "step": 1360
    },
    {
      "epoch": 0.14451476793248946,
      "grad_norm": 1.5452641248703003,
      "learning_rate": 0.001433032334631133,
      "loss": 5.2179,
      "step": 1370
    },
    {
      "epoch": 0.14556962025316456,
      "grad_norm": 2.83355450630188,
      "learning_rate": 0.0014319915158879776,
      "loss": 5.1246,
      "step": 1380
    },
    {
      "epoch": 0.14662447257383968,
      "grad_norm": 1.2328715324401855,
      "learning_rate": 0.0014309430550968794,
      "loss": 5.1222,
      "step": 1390
    },
    {
      "epoch": 0.14767932489451477,
      "grad_norm": 1.7708998918533325,
      "learning_rate": 0.001429886964006354,
      "loss": 5.0799,
      "step": 1400
    },
    {
      "epoch": 0.14873417721518986,
      "grad_norm": 2.349193811416626,
      "learning_rate": 0.0014288232544504174,
      "loss": 5.1315,
      "step": 1410
    },
    {
      "epoch": 0.14978902953586498,
      "grad_norm": 2.6197426319122314,
      "learning_rate": 0.0014277519383484548,
      "loss": 5.1204,
      "step": 1420
    },
    {
      "epoch": 0.15084388185654007,
      "grad_norm": 2.097552537918091,
      "learning_rate": 0.0014266730277050863,
      "loss": 5.1609,
      "step": 1430
    },
    {
      "epoch": 0.1518987341772152,
      "grad_norm": 1.9973235130310059,
      "learning_rate": 0.0014255865346100324,
      "loss": 5.0869,
      "step": 1440
    },
    {
      "epoch": 0.1529535864978903,
      "grad_norm": 2.1963138580322266,
      "learning_rate": 0.0014244924712379786,
      "loss": 5.0731,
      "step": 1450
    },
    {
      "epoch": 0.1540084388185654,
      "grad_norm": 1.9913876056671143,
      "learning_rate": 0.0014233908498484393,
      "loss": 5.076,
      "step": 1460
    },
    {
      "epoch": 0.1550632911392405,
      "grad_norm": 2.04766583442688,
      "learning_rate": 0.0014222816827856202,
      "loss": 5.0631,
      "step": 1470
    },
    {
      "epoch": 0.15611814345991562,
      "grad_norm": 2.4936294555664062,
      "learning_rate": 0.0014211649824782797,
      "loss": 5.0616,
      "step": 1480
    },
    {
      "epoch": 0.1571729957805907,
      "grad_norm": 2.5747439861297607,
      "learning_rate": 0.0014200407614395898,
      "loss": 5.0554,
      "step": 1490
    },
    {
      "epoch": 0.15822784810126583,
      "grad_norm": 2.3981728553771973,
      "learning_rate": 0.0014189090322669967,
      "loss": 5.0566,
      "step": 1500
    },
    {
      "epoch": 0.15928270042194093,
      "grad_norm": 2.201969623565674,
      "learning_rate": 0.001417769807642078,
      "loss": 5.0772,
      "step": 1510
    },
    {
      "epoch": 0.16033755274261605,
      "grad_norm": 6.607595920562744,
      "learning_rate": 0.0014166231003304019,
      "loss": 5.0717,
      "step": 1520
    },
    {
      "epoch": 0.16139240506329114,
      "grad_norm": 2.4264588356018066,
      "learning_rate": 0.0014154689231813838,
      "loss": 5.0602,
      "step": 1530
    },
    {
      "epoch": 0.16244725738396623,
      "grad_norm": 2.041923999786377,
      "learning_rate": 0.0014143072891281425,
      "loss": 5.0269,
      "step": 1540
    },
    {
      "epoch": 0.16350210970464135,
      "grad_norm": 2.711793899536133,
      "learning_rate": 0.0014131382111873543,
      "loss": 5.0404,
      "step": 1550
    },
    {
      "epoch": 0.16455696202531644,
      "grad_norm": 2.552089214324951,
      "learning_rate": 0.0014119617024591089,
      "loss": 5.0436,
      "step": 1560
    },
    {
      "epoch": 0.16561181434599156,
      "grad_norm": 1.8372821807861328,
      "learning_rate": 0.0014107777761267605,
      "loss": 5.1018,
      "step": 1570
    },
    {
      "epoch": 0.16666666666666666,
      "grad_norm": 1.857517957687378,
      "learning_rate": 0.0014095864454567821,
      "loss": 5.0017,
      "step": 1580
    },
    {
      "epoch": 0.16772151898734178,
      "grad_norm": 2.4412038326263428,
      "learning_rate": 0.0014083877237986153,
      "loss": 4.9886,
      "step": 1590
    },
    {
      "epoch": 0.16877637130801687,
      "grad_norm": 1.83058500289917,
      "learning_rate": 0.001407181624584522,
      "loss": 5.098,
      "step": 1600
    },
    {
      "epoch": 0.169831223628692,
      "grad_norm": 3.6769912242889404,
      "learning_rate": 0.0014059681613294327,
      "loss": 5.0198,
      "step": 1610
    },
    {
      "epoch": 0.17088607594936708,
      "grad_norm": 1.8989999294281006,
      "learning_rate": 0.0014047473476307955,
      "loss": 5.051,
      "step": 1620
    },
    {
      "epoch": 0.1719409282700422,
      "grad_norm": 2.89774489402771,
      "learning_rate": 0.0014035191971684242,
      "loss": 5.0427,
      "step": 1630
    },
    {
      "epoch": 0.1729957805907173,
      "grad_norm": 2.3925230503082275,
      "learning_rate": 0.0014022837237043441,
      "loss": 5.0374,
      "step": 1640
    },
    {
      "epoch": 0.17405063291139242,
      "grad_norm": 2.1988370418548584,
      "learning_rate": 0.001401040941082639,
      "loss": 5.0086,
      "step": 1650
    },
    {
      "epoch": 0.1751054852320675,
      "grad_norm": 2.2729415893554688,
      "learning_rate": 0.0013997908632292948,
      "loss": 5.0359,
      "step": 1660
    },
    {
      "epoch": 0.17616033755274263,
      "grad_norm": 2.042452573776245,
      "learning_rate": 0.0013985335041520443,
      "loss": 4.9858,
      "step": 1670
    },
    {
      "epoch": 0.17721518987341772,
      "grad_norm": 2.6199772357940674,
      "learning_rate": 0.00139726887794021,
      "loss": 4.9499,
      "step": 1680
    },
    {
      "epoch": 0.17827004219409281,
      "grad_norm": 2.379549026489258,
      "learning_rate": 0.001395996998764546,
      "loss": 4.9954,
      "step": 1690
    },
    {
      "epoch": 0.17932489451476794,
      "grad_norm": 2.333350419998169,
      "learning_rate": 0.0013947178808770794,
      "loss": 5.0037,
      "step": 1700
    },
    {
      "epoch": 0.18037974683544303,
      "grad_norm": 2.261662244796753,
      "learning_rate": 0.0013934315386109509,
      "loss": 5.0054,
      "step": 1710
    },
    {
      "epoch": 0.18143459915611815,
      "grad_norm": 2.3880715370178223,
      "learning_rate": 0.0013921379863802536,
      "loss": 4.9978,
      "step": 1720
    },
    {
      "epoch": 0.18248945147679324,
      "grad_norm": 2.2738215923309326,
      "learning_rate": 0.0013908372386798717,
      "loss": 5.0145,
      "step": 1730
    },
    {
      "epoch": 0.18354430379746836,
      "grad_norm": 2.1261074542999268,
      "learning_rate": 0.0013895293100853188,
      "loss": 4.9925,
      "step": 1740
    },
    {
      "epoch": 0.18459915611814345,
      "grad_norm": 3.12884783744812,
      "learning_rate": 0.0013882142152525732,
      "loss": 4.9938,
      "step": 1750
    },
    {
      "epoch": 0.18565400843881857,
      "grad_norm": 2.532956838607788,
      "learning_rate": 0.0013868919689179143,
      "loss": 4.9914,
      "step": 1760
    },
    {
      "epoch": 0.18670886075949367,
      "grad_norm": 1.8527002334594727,
      "learning_rate": 0.0013855625858977584,
      "loss": 5.0011,
      "step": 1770
    },
    {
      "epoch": 0.1877637130801688,
      "grad_norm": 2.036776065826416,
      "learning_rate": 0.001384226081088491,
      "loss": 4.954,
      "step": 1780
    },
    {
      "epoch": 0.18881856540084388,
      "grad_norm": 2.065924644470215,
      "learning_rate": 0.0013828824694663013,
      "loss": 4.9374,
      "step": 1790
    },
    {
      "epoch": 0.189873417721519,
      "grad_norm": 2.005577802658081,
      "learning_rate": 0.0013815317660870138,
      "loss": 5.1765,
      "step": 1800
    },
    {
      "epoch": 0.1909282700421941,
      "grad_norm": 2.4041264057159424,
      "learning_rate": 0.0013801739860859188,
      "loss": 5.0036,
      "step": 1810
    },
    {
      "epoch": 0.19198312236286919,
      "grad_norm": 2.1881420612335205,
      "learning_rate": 0.001378809144677605,
      "loss": 4.9521,
      "step": 1820
    },
    {
      "epoch": 0.1930379746835443,
      "grad_norm": 2.115715503692627,
      "learning_rate": 0.0013774372571557856,
      "loss": 4.9124,
      "step": 1830
    },
    {
      "epoch": 0.1940928270042194,
      "grad_norm": 2.3303935527801514,
      "learning_rate": 0.0013760583388931315,
      "loss": 4.9313,
      "step": 1840
    },
    {
      "epoch": 0.19514767932489452,
      "grad_norm": 2.541919708251953,
      "learning_rate": 0.0013746724053410944,
      "loss": 4.9502,
      "step": 1850
    },
    {
      "epoch": 0.1962025316455696,
      "grad_norm": 2.633435010910034,
      "learning_rate": 0.0013732794720297367,
      "loss": 4.9382,
      "step": 1860
    },
    {
      "epoch": 0.19725738396624473,
      "grad_norm": 2.3229167461395264,
      "learning_rate": 0.001371879554567556,
      "loss": 4.943,
      "step": 1870
    },
    {
      "epoch": 0.19831223628691982,
      "grad_norm": 2.1760222911834717,
      "learning_rate": 0.0013704726686413116,
      "loss": 4.9184,
      "step": 1880
    },
    {
      "epoch": 0.19936708860759494,
      "grad_norm": 2.105520009994507,
      "learning_rate": 0.0013690588300158467,
      "loss": 4.9237,
      "step": 1890
    },
    {
      "epoch": 0.20042194092827004,
      "grad_norm": 2.0479912757873535,
      "learning_rate": 0.0013676380545339136,
      "loss": 4.8809,
      "step": 1900
    },
    {
      "epoch": 0.20147679324894516,
      "grad_norm": 2.4494359493255615,
      "learning_rate": 0.0013662103581159955,
      "loss": 4.8583,
      "step": 1910
    },
    {
      "epoch": 0.20253164556962025,
      "grad_norm": 2.7703845500946045,
      "learning_rate": 0.001364775756760128,
      "loss": 4.8918,
      "step": 1920
    },
    {
      "epoch": 0.20358649789029537,
      "grad_norm": 2.984938383102417,
      "learning_rate": 0.0013633342665417192,
      "loss": 4.9412,
      "step": 1930
    },
    {
      "epoch": 0.20464135021097046,
      "grad_norm": 3.19482684135437,
      "learning_rate": 0.0013618859036133714,
      "loss": 4.9358,
      "step": 1940
    },
    {
      "epoch": 0.20569620253164558,
      "grad_norm": 3.1088333129882812,
      "learning_rate": 0.0013604306842046983,
      "loss": 4.9446,
      "step": 1950
    },
    {
      "epoch": 0.20675105485232068,
      "grad_norm": 2.5788025856018066,
      "learning_rate": 0.0013589686246221438,
      "loss": 4.9531,
      "step": 1960
    },
    {
      "epoch": 0.20780590717299577,
      "grad_norm": 2.4632956981658936,
      "learning_rate": 0.0013574997412487996,
      "loss": 4.9389,
      "step": 1970
    },
    {
      "epoch": 0.2088607594936709,
      "grad_norm": 6.852120399475098,
      "learning_rate": 0.001356024050544221,
      "loss": 4.955,
      "step": 1980
    },
    {
      "epoch": 0.20991561181434598,
      "grad_norm": 3.0540056228637695,
      "learning_rate": 0.001354541569044243,
      "loss": 4.9499,
      "step": 1990
    },
    {
      "epoch": 0.2109704641350211,
      "grad_norm": 3.690631151199341,
      "learning_rate": 0.0013530523133607948,
      "loss": 4.9237,
      "step": 2000
    },
    {
      "epoch": 0.2120253164556962,
      "grad_norm": 2.374018907546997,
      "learning_rate": 0.001351556300181713,
      "loss": 4.9658,
      "step": 2010
    },
    {
      "epoch": 0.21308016877637131,
      "grad_norm": 2.495804786682129,
      "learning_rate": 0.0013500535462705565,
      "loss": 4.9405,
      "step": 2020
    },
    {
      "epoch": 0.2141350210970464,
      "grad_norm": 2.2292323112487793,
      "learning_rate": 0.001348544068466416,
      "loss": 4.8988,
      "step": 2030
    },
    {
      "epoch": 0.21518987341772153,
      "grad_norm": 2.3601019382476807,
      "learning_rate": 0.0013470278836837275,
      "loss": 4.9084,
      "step": 2040
    },
    {
      "epoch": 0.21624472573839662,
      "grad_norm": 2.504920482635498,
      "learning_rate": 0.001345505008912082,
      "loss": 4.9056,
      "step": 2050
    },
    {
      "epoch": 0.21729957805907174,
      "grad_norm": 2.249976396560669,
      "learning_rate": 0.0013439754612160353,
      "loss": 4.8952,
      "step": 2060
    },
    {
      "epoch": 0.21835443037974683,
      "grad_norm": 2.449305772781372,
      "learning_rate": 0.0013424392577349152,
      "loss": 4.8808,
      "step": 2070
    },
    {
      "epoch": 0.21940928270042195,
      "grad_norm": 2.094162940979004,
      "learning_rate": 0.0013408964156826327,
      "loss": 4.8791,
      "step": 2080
    },
    {
      "epoch": 0.22046413502109705,
      "grad_norm": 2.3999764919281006,
      "learning_rate": 0.0013393469523474858,
      "loss": 4.8657,
      "step": 2090
    },
    {
      "epoch": 0.22151898734177214,
      "grad_norm": 2.1957786083221436,
      "learning_rate": 0.001337790885091968,
      "loss": 4.8674,
      "step": 2100
    },
    {
      "epoch": 0.22257383966244726,
      "grad_norm": 2.2604613304138184,
      "learning_rate": 0.0013362282313525728,
      "loss": 4.8582,
      "step": 2110
    },
    {
      "epoch": 0.22362869198312235,
      "grad_norm": 2.236294746398926,
      "learning_rate": 0.001334659008639598,
      "loss": 4.8683,
      "step": 2120
    },
    {
      "epoch": 0.22468354430379747,
      "grad_norm": 2.395782709121704,
      "learning_rate": 0.0013330832345369505,
      "loss": 4.8471,
      "step": 2130
    },
    {
      "epoch": 0.22573839662447256,
      "grad_norm": 2.2968459129333496,
      "learning_rate": 0.0013315009267019487,
      "loss": 4.8372,
      "step": 2140
    },
    {
      "epoch": 0.22679324894514769,
      "grad_norm": 2.510758638381958,
      "learning_rate": 0.0013299121028651246,
      "loss": 4.8502,
      "step": 2150
    },
    {
      "epoch": 0.22784810126582278,
      "grad_norm": 3.057630777359009,
      "learning_rate": 0.0013283167808300247,
      "loss": 4.8717,
      "step": 2160
    },
    {
      "epoch": 0.2289029535864979,
      "grad_norm": 2.8189473152160645,
      "learning_rate": 0.0013267149784730117,
      "loss": 4.8901,
      "step": 2170
    },
    {
      "epoch": 0.229957805907173,
      "grad_norm": 3.082125663757324,
      "learning_rate": 0.0013251067137430629,
      "loss": 4.8708,
      "step": 2180
    },
    {
      "epoch": 0.2310126582278481,
      "grad_norm": 2.5968902111053467,
      "learning_rate": 0.00132349200466157,
      "loss": 4.91,
      "step": 2190
    },
    {
      "epoch": 0.2320675105485232,
      "grad_norm": 4.17230224609375,
      "learning_rate": 0.001321870869322137,
      "loss": 4.8878,
      "step": 2200
    },
    {
      "epoch": 0.23312236286919832,
      "grad_norm": 2.535311222076416,
      "learning_rate": 0.0013202433258903761,
      "loss": 4.9012,
      "step": 2210
    },
    {
      "epoch": 0.23417721518987342,
      "grad_norm": 2.5178253650665283,
      "learning_rate": 0.0013186093926037072,
      "loss": 4.8749,
      "step": 2220
    },
    {
      "epoch": 0.23523206751054854,
      "grad_norm": 2.5765836238861084,
      "learning_rate": 0.0013169690877711502,
      "loss": 4.8521,
      "step": 2230
    },
    {
      "epoch": 0.23628691983122363,
      "grad_norm": 3.2549009323120117,
      "learning_rate": 0.0013153224297731215,
      "loss": 4.8736,
      "step": 2240
    },
    {
      "epoch": 0.23734177215189872,
      "grad_norm": 2.315842390060425,
      "learning_rate": 0.001313669437061228,
      "loss": 4.8556,
      "step": 2250
    },
    {
      "epoch": 0.23839662447257384,
      "grad_norm": 2.5131428241729736,
      "learning_rate": 0.0013120101281580605,
      "loss": 4.826,
      "step": 2260
    },
    {
      "epoch": 0.23945147679324894,
      "grad_norm": 2.3892571926116943,
      "learning_rate": 0.001310344521656985,
      "loss": 4.8456,
      "step": 2270
    },
    {
      "epoch": 0.24050632911392406,
      "grad_norm": 2.556605815887451,
      "learning_rate": 0.0013086726362219363,
      "loss": 4.8452,
      "step": 2280
    },
    {
      "epoch": 0.24156118143459915,
      "grad_norm": 2.3933675289154053,
      "learning_rate": 0.0013069944905872064,
      "loss": 4.8838,
      "step": 2290
    },
    {
      "epoch": 0.24261603375527427,
      "grad_norm": 2.5870211124420166,
      "learning_rate": 0.001305310103557237,
      "loss": 4.8431,
      "step": 2300
    },
    {
      "epoch": 0.24367088607594936,
      "grad_norm": 2.5385541915893555,
      "learning_rate": 0.001303619494006407,
      "loss": 4.8257,
      "step": 2310
    },
    {
      "epoch": 0.24472573839662448,
      "grad_norm": 2.5317254066467285,
      "learning_rate": 0.001301922680878822,
      "loss": 4.8613,
      "step": 2320
    },
    {
      "epoch": 0.24578059071729957,
      "grad_norm": 2.903883218765259,
      "learning_rate": 0.0013002196831881014,
      "loss": 4.829,
      "step": 2330
    },
    {
      "epoch": 0.2468354430379747,
      "grad_norm": 2.642857789993286,
      "learning_rate": 0.0012985105200171664,
      "loss": 4.8255,
      "step": 2340
    },
    {
      "epoch": 0.2478902953586498,
      "grad_norm": 3.0150463581085205,
      "learning_rate": 0.0012967952105180243,
      "loss": 4.8294,
      "step": 2350
    },
    {
      "epoch": 0.2489451476793249,
      "grad_norm": 2.6593401432037354,
      "learning_rate": 0.001295073773911556,
      "loss": 4.8073,
      "step": 2360
    },
    {
      "epoch": 0.25,
      "grad_norm": 3.1376888751983643,
      "learning_rate": 0.001293346229487299,
      "loss": 4.8181,
      "step": 2370
    },
    {
      "epoch": 0.2510548523206751,
      "grad_norm": 3.406196117401123,
      "learning_rate": 0.0012916125966032322,
      "loss": 4.8379,
      "step": 2380
    },
    {
      "epoch": 0.2521097046413502,
      "grad_norm": 4.167253017425537,
      "learning_rate": 0.0012898728946855588,
      "loss": 4.8486,
      "step": 2390
    },
    {
      "epoch": 0.25316455696202533,
      "grad_norm": 2.59110426902771,
      "learning_rate": 0.001288127143228488,
      "loss": 4.8456,
      "step": 2400
    },
    {
      "epoch": 0.2542194092827004,
      "grad_norm": 2.7719688415527344,
      "learning_rate": 0.0012863753617940172,
      "loss": 4.8361,
      "step": 2410
    },
    {
      "epoch": 0.2552742616033755,
      "grad_norm": 2.9639158248901367,
      "learning_rate": 0.001284617570011713,
      "loss": 4.8355,
      "step": 2420
    },
    {
      "epoch": 0.2563291139240506,
      "grad_norm": 2.5431289672851562,
      "learning_rate": 0.0012828537875784905,
      "loss": 4.8184,
      "step": 2430
    },
    {
      "epoch": 0.25738396624472576,
      "grad_norm": 3.6541528701782227,
      "learning_rate": 0.001281084034258393,
      "loss": 4.8096,
      "step": 2440
    },
    {
      "epoch": 0.25843881856540085,
      "grad_norm": 2.870486259460449,
      "learning_rate": 0.0012793083298823708,
      "loss": 4.8081,
      "step": 2450
    },
    {
      "epoch": 0.25949367088607594,
      "grad_norm": 2.7524688243865967,
      "learning_rate": 0.0012775266943480582,
      "loss": 4.8195,
      "step": 2460
    },
    {
      "epoch": 0.26054852320675104,
      "grad_norm": 2.76414155960083,
      "learning_rate": 0.0012757391476195517,
      "loss": 4.7903,
      "step": 2470
    },
    {
      "epoch": 0.2616033755274262,
      "grad_norm": 3.576305389404297,
      "learning_rate": 0.0012739457097271849,
      "loss": 4.8077,
      "step": 2480
    },
    {
      "epoch": 0.2626582278481013,
      "grad_norm": 2.469773530960083,
      "learning_rate": 0.0012721464007673055,
      "loss": 4.8419,
      "step": 2490
    },
    {
      "epoch": 0.26371308016877637,
      "grad_norm": 2.934180498123169,
      "learning_rate": 0.0012703412409020484,
      "loss": 4.8055,
      "step": 2500
    },
    {
      "epoch": 0.26476793248945146,
      "grad_norm": 14.20032024383545,
      "learning_rate": 0.0012685302503591118,
      "loss": 4.8917,
      "step": 2510
    },
    {
      "epoch": 0.26582278481012656,
      "grad_norm": 3.133007287979126,
      "learning_rate": 0.001266713449431529,
      "loss": 4.8588,
      "step": 2520
    },
    {
      "epoch": 0.2668776371308017,
      "grad_norm": 2.526608943939209,
      "learning_rate": 0.001264890858477441,
      "loss": 4.7892,
      "step": 2530
    },
    {
      "epoch": 0.2679324894514768,
      "grad_norm": 2.7648675441741943,
      "learning_rate": 0.0012630624979198697,
      "loss": 4.7691,
      "step": 2540
    },
    {
      "epoch": 0.2689873417721519,
      "grad_norm": 2.9659976959228516,
      "learning_rate": 0.0012612283882464882,
      "loss": 4.7657,
      "step": 2550
    },
    {
      "epoch": 0.270042194092827,
      "grad_norm": 2.685804843902588,
      "learning_rate": 0.0012593885500093906,
      "loss": 4.7829,
      "step": 2560
    },
    {
      "epoch": 0.27109704641350213,
      "grad_norm": 2.5587093830108643,
      "learning_rate": 0.0012575430038248628,
      "loss": 4.7709,
      "step": 2570
    },
    {
      "epoch": 0.2721518987341772,
      "grad_norm": 2.9499034881591797,
      "learning_rate": 0.0012556917703731509,
      "loss": 4.765,
      "step": 2580
    },
    {
      "epoch": 0.2732067510548523,
      "grad_norm": 2.5674445629119873,
      "learning_rate": 0.00125383487039823,
      "loss": 4.7795,
      "step": 2590
    },
    {
      "epoch": 0.2742616033755274,
      "grad_norm": 2.9498372077941895,
      "learning_rate": 0.0012519723247075706,
      "loss": 4.7738,
      "step": 2600
    },
    {
      "epoch": 0.27531645569620256,
      "grad_norm": 2.933891773223877,
      "learning_rate": 0.0012501041541719067,
      "loss": 4.7914,
      "step": 2610
    },
    {
      "epoch": 0.27637130801687765,
      "grad_norm": 2.7813920974731445,
      "learning_rate": 0.0012482303797250014,
      "loss": 4.7757,
      "step": 2620
    },
    {
      "epoch": 0.27742616033755274,
      "grad_norm": 2.9547176361083984,
      "learning_rate": 0.0012463510223634125,
      "loss": 4.7734,
      "step": 2630
    },
    {
      "epoch": 0.27848101265822783,
      "grad_norm": 2.472228527069092,
      "learning_rate": 0.0012444661031462566,
      "loss": 4.7739,
      "step": 2640
    },
    {
      "epoch": 0.2795358649789029,
      "grad_norm": 3.028064489364624,
      "learning_rate": 0.0012425756431949742,
      "loss": 4.7681,
      "step": 2650
    },
    {
      "epoch": 0.2805907172995781,
      "grad_norm": 3.141963005065918,
      "learning_rate": 0.0012406796636930918,
      "loss": 4.7624,
      "step": 2660
    },
    {
      "epoch": 0.28164556962025317,
      "grad_norm": 2.920543670654297,
      "learning_rate": 0.0012387781858859857,
      "loss": 4.7818,
      "step": 2670
    },
    {
      "epoch": 0.28270042194092826,
      "grad_norm": 2.7779557704925537,
      "learning_rate": 0.0012368712310806432,
      "loss": 4.794,
      "step": 2680
    },
    {
      "epoch": 0.28375527426160335,
      "grad_norm": 2.517868757247925,
      "learning_rate": 0.001234958820645424,
      "loss": 4.7548,
      "step": 2690
    },
    {
      "epoch": 0.2848101265822785,
      "grad_norm": 3.2706551551818848,
      "learning_rate": 0.0012330409760098208,
      "loss": 4.7749,
      "step": 2700
    },
    {
      "epoch": 0.2858649789029536,
      "grad_norm": 3.8404297828674316,
      "learning_rate": 0.0012311177186642194,
      "loss": 4.7622,
      "step": 2710
    },
    {
      "epoch": 0.2869198312236287,
      "grad_norm": 3.0984554290771484,
      "learning_rate": 0.001229189070159657,
      "loss": 4.7836,
      "step": 2720
    },
    {
      "epoch": 0.2879746835443038,
      "grad_norm": 3.7385094165802,
      "learning_rate": 0.0012272550521075824,
      "loss": 4.7533,
      "step": 2730
    },
    {
      "epoch": 0.2890295358649789,
      "grad_norm": 3.1707301139831543,
      "learning_rate": 0.0012253156861796119,
      "loss": 4.7528,
      "step": 2740
    },
    {
      "epoch": 0.290084388185654,
      "grad_norm": 2.649907350540161,
      "learning_rate": 0.001223370994107288,
      "loss": 4.7289,
      "step": 2750
    },
    {
      "epoch": 0.2911392405063291,
      "grad_norm": 3.0319390296936035,
      "learning_rate": 0.001221420997681834,
      "loss": 4.7231,
      "step": 2760
    },
    {
      "epoch": 0.2921940928270042,
      "grad_norm": 3.859168767929077,
      "learning_rate": 0.001219465718753913,
      "loss": 4.7402,
      "step": 2770
    },
    {
      "epoch": 0.29324894514767935,
      "grad_norm": 2.6841399669647217,
      "learning_rate": 0.00121750517923338,
      "loss": 4.7369,
      "step": 2780
    },
    {
      "epoch": 0.29430379746835444,
      "grad_norm": 2.609192132949829,
      "learning_rate": 0.001215539401089037,
      "loss": 4.7213,
      "step": 2790
    },
    {
      "epoch": 0.29535864978902954,
      "grad_norm": 2.8833703994750977,
      "learning_rate": 0.0012135684063483891,
      "loss": 4.743,
      "step": 2800
    },
    {
      "epoch": 0.29641350210970463,
      "grad_norm": 2.5281152725219727,
      "learning_rate": 0.0012115922170973935,
      "loss": 4.7121,
      "step": 2810
    },
    {
      "epoch": 0.2974683544303797,
      "grad_norm": 3.071457624435425,
      "learning_rate": 0.0012096108554802165,
      "loss": 4.726,
      "step": 2820
    },
    {
      "epoch": 0.29852320675105487,
      "grad_norm": 2.8602724075317383,
      "learning_rate": 0.0012076243436989823,
      "loss": 4.7402,
      "step": 2830
    },
    {
      "epoch": 0.29957805907172996,
      "grad_norm": 2.686028480529785,
      "learning_rate": 0.001205632704013525,
      "loss": 4.7209,
      "step": 2840
    },
    {
      "epoch": 0.30063291139240506,
      "grad_norm": 3.447962760925293,
      "learning_rate": 0.0012036359587411405,
      "loss": 4.7218,
      "step": 2850
    },
    {
      "epoch": 0.30168776371308015,
      "grad_norm": 3.034770965576172,
      "learning_rate": 0.0012016341302563342,
      "loss": 4.7423,
      "step": 2860
    },
    {
      "epoch": 0.3027426160337553,
      "grad_norm": 3.1976609230041504,
      "learning_rate": 0.0011996272409905717,
      "loss": 4.7306,
      "step": 2870
    },
    {
      "epoch": 0.3037974683544304,
      "grad_norm": 3.0060033798217773,
      "learning_rate": 0.001197615313432027,
      "loss": 4.7189,
      "step": 2880
    },
    {
      "epoch": 0.3048523206751055,
      "grad_norm": 3.354975700378418,
      "learning_rate": 0.0011955983701253312,
      "loss": 4.7151,
      "step": 2890
    },
    {
      "epoch": 0.3059071729957806,
      "grad_norm": 2.883009672164917,
      "learning_rate": 0.0011935764336713187,
      "loss": 4.7402,
      "step": 2900
    },
    {
      "epoch": 0.3069620253164557,
      "grad_norm": 2.7966136932373047,
      "learning_rate": 0.0011915495267267745,
      "loss": 4.7447,
      "step": 2910
    },
    {
      "epoch": 0.3080168776371308,
      "grad_norm": 3.373680353164673,
      "learning_rate": 0.001189517672004181,
      "loss": 4.736,
      "step": 2920
    },
    {
      "epoch": 0.3090717299578059,
      "grad_norm": 2.5867300033569336,
      "learning_rate": 0.0011874808922714623,
      "loss": 4.7326,
      "step": 2930
    },
    {
      "epoch": 0.310126582278481,
      "grad_norm": 2.7779064178466797,
      "learning_rate": 0.00118543921035173,
      "loss": 4.7195,
      "step": 2940
    },
    {
      "epoch": 0.3111814345991561,
      "grad_norm": 2.6546518802642822,
      "learning_rate": 0.0011833926491230265,
      "loss": 4.7024,
      "step": 2950
    },
    {
      "epoch": 0.31223628691983124,
      "grad_norm": 2.8331165313720703,
      "learning_rate": 0.0011813412315180704,
      "loss": 4.6846,
      "step": 2960
    },
    {
      "epoch": 0.31329113924050633,
      "grad_norm": 2.4070839881896973,
      "learning_rate": 0.0011792849805239967,
      "loss": 4.7142,
      "step": 2970
    },
    {
      "epoch": 0.3143459915611814,
      "grad_norm": 2.8148386478424072,
      "learning_rate": 0.0011772239191821029,
      "loss": 4.6901,
      "step": 2980
    },
    {
      "epoch": 0.3154008438818565,
      "grad_norm": 3.1715586185455322,
      "learning_rate": 0.001175158070587587,
      "loss": 4.7108,
      "step": 2990
    },
    {
      "epoch": 0.31645569620253167,
      "grad_norm": 3.4174892902374268,
      "learning_rate": 0.0011730874578892913,
      "loss": 4.7169,
      "step": 3000
    },
    {
      "epoch": 0.31751054852320676,
      "grad_norm": 3.138155937194824,
      "learning_rate": 0.0011710121042894425,
      "loss": 4.7171,
      "step": 3010
    },
    {
      "epoch": 0.31856540084388185,
      "grad_norm": 3.323488712310791,
      "learning_rate": 0.0011689320330433904,
      "loss": 4.7121,
      "step": 3020
    },
    {
      "epoch": 0.31962025316455694,
      "grad_norm": 3.087789297103882,
      "learning_rate": 0.0011668472674593497,
      "loss": 4.7117,
      "step": 3030
    },
    {
      "epoch": 0.3206751054852321,
      "grad_norm": 3.9287619590759277,
      "learning_rate": 0.0011647578308981363,
      "loss": 4.7122,
      "step": 3040
    },
    {
      "epoch": 0.3217299578059072,
      "grad_norm": 2.920612335205078,
      "learning_rate": 0.0011626637467729072,
      "loss": 4.6993,
      "step": 3050
    },
    {
      "epoch": 0.3227848101265823,
      "grad_norm": 2.7639284133911133,
      "learning_rate": 0.0011605650385488977,
      "loss": 4.6876,
      "step": 3060
    },
    {
      "epoch": 0.32383966244725737,
      "grad_norm": 2.5909385681152344,
      "learning_rate": 0.0011584617297431578,
      "loss": 4.6858,
      "step": 3070
    },
    {
      "epoch": 0.32489451476793246,
      "grad_norm": 3.622706174850464,
      "learning_rate": 0.0011563538439242902,
      "loss": 4.6875,
      "step": 3080
    },
    {
      "epoch": 0.3259493670886076,
      "grad_norm": 3.129620313644409,
      "learning_rate": 0.0011542414047121842,
      "loss": 4.7045,
      "step": 3090
    },
    {
      "epoch": 0.3270042194092827,
      "grad_norm": 3.420426607131958,
      "learning_rate": 0.0011521244357777533,
      "loss": 4.7081,
      "step": 3100
    },
    {
      "epoch": 0.3280590717299578,
      "grad_norm": 3.2381699085235596,
      "learning_rate": 0.0011500029608426676,
      "loss": 4.7029,
      "step": 3110
    },
    {
      "epoch": 0.3291139240506329,
      "grad_norm": 2.674516439437866,
      "learning_rate": 0.00114787700367909,
      "loss": 4.6829,
      "step": 3120
    },
    {
      "epoch": 0.33016877637130804,
      "grad_norm": 2.992539882659912,
      "learning_rate": 0.001145746588109408,
      "loss": 4.677,
      "step": 3130
    },
    {
      "epoch": 0.33122362869198313,
      "grad_norm": 2.480374813079834,
      "learning_rate": 0.0011436117380059692,
      "loss": 4.6523,
      "step": 3140
    },
    {
      "epoch": 0.3322784810126582,
      "grad_norm": 3.1188433170318604,
      "learning_rate": 0.0011414724772908105,
      "loss": 4.6511,
      "step": 3150
    },
    {
      "epoch": 0.3333333333333333,
      "grad_norm": 3.0314910411834717,
      "learning_rate": 0.0011393288299353934,
      "loss": 4.6664,
      "step": 3160
    },
    {
      "epoch": 0.33438818565400846,
      "grad_norm": 3.0282297134399414,
      "learning_rate": 0.001137180819960333,
      "loss": 4.6642,
      "step": 3170
    },
    {
      "epoch": 0.33544303797468356,
      "grad_norm": 3.9388370513916016,
      "learning_rate": 0.0011350284714351298,
      "loss": 4.6795,
      "step": 3180
    },
    {
      "epoch": 0.33649789029535865,
      "grad_norm": 3.2480647563934326,
      "learning_rate": 0.0011328718084779004,
      "loss": 4.6782,
      "step": 3190
    },
    {
      "epoch": 0.33755274261603374,
      "grad_norm": 3.7007675170898438,
      "learning_rate": 0.001130710855255106,
      "loss": 4.6828,
      "step": 3200
    },
    {
      "epoch": 0.33860759493670883,
      "grad_norm": 2.5426015853881836,
      "learning_rate": 0.0011285456359812825,
      "loss": 4.6866,
      "step": 3210
    },
    {
      "epoch": 0.339662447257384,
      "grad_norm": 2.65291690826416,
      "learning_rate": 0.0011263761749187693,
      "loss": 4.6515,
      "step": 3220
    },
    {
      "epoch": 0.3407172995780591,
      "grad_norm": 2.700204372406006,
      "learning_rate": 0.001124202496377437,
      "loss": 4.6502,
      "step": 3230
    },
    {
      "epoch": 0.34177215189873417,
      "grad_norm": 3.059195041656494,
      "learning_rate": 0.0011220246247144149,
      "loss": 4.622,
      "step": 3240
    },
    {
      "epoch": 0.34282700421940926,
      "grad_norm": 2.9939522743225098,
      "learning_rate": 0.0011198425843338183,
      "loss": 4.6311,
      "step": 3250
    },
    {
      "epoch": 0.3438818565400844,
      "grad_norm": 3.6204516887664795,
      "learning_rate": 0.0011176563996864754,
      "loss": 4.6152,
      "step": 3260
    },
    {
      "epoch": 0.3449367088607595,
      "grad_norm": 2.8647263050079346,
      "learning_rate": 0.0011154660952696525,
      "loss": 4.6962,
      "step": 3270
    },
    {
      "epoch": 0.3459915611814346,
      "grad_norm": 2.8345961570739746,
      "learning_rate": 0.00111327169562678,
      "loss": 4.6507,
      "step": 3280
    },
    {
      "epoch": 0.3470464135021097,
      "grad_norm": 3.119981050491333,
      "learning_rate": 0.0011110732253471777,
      "loss": 4.6423,
      "step": 3290
    },
    {
      "epoch": 0.34810126582278483,
      "grad_norm": 2.9357082843780518,
      "learning_rate": 0.0011088707090657784,
      "loss": 4.636,
      "step": 3300
    },
    {
      "epoch": 0.3491561181434599,
      "grad_norm": 2.935332775115967,
      "learning_rate": 0.0011066641714628522,
      "loss": 4.646,
      "step": 3310
    },
    {
      "epoch": 0.350210970464135,
      "grad_norm": 2.913219690322876,
      "learning_rate": 0.0011044536372637307,
      "loss": 4.6506,
      "step": 3320
    },
    {
      "epoch": 0.3512658227848101,
      "grad_norm": 3.2490315437316895,
      "learning_rate": 0.001102239131238529,
      "loss": 4.6387,
      "step": 3330
    },
    {
      "epoch": 0.35232067510548526,
      "grad_norm": 2.9215688705444336,
      "learning_rate": 0.0011000206782018683,
      "loss": 4.6413,
      "step": 3340
    },
    {
      "epoch": 0.35337552742616035,
      "grad_norm": 2.824744939804077,
      "learning_rate": 0.0010977983030125982,
      "loss": 4.659,
      "step": 3350
    },
    {
      "epoch": 0.35443037974683544,
      "grad_norm": 2.8020405769348145,
      "learning_rate": 0.0010955720305735176,
      "loss": 4.6563,
      "step": 3360
    },
    {
      "epoch": 0.35548523206751054,
      "grad_norm": 2.5569417476654053,
      "learning_rate": 0.0010933418858310965,
      "loss": 4.6612,
      "step": 3370
    },
    {
      "epoch": 0.35654008438818563,
      "grad_norm": 2.829496145248413,
      "learning_rate": 0.0010911078937751954,
      "loss": 4.6706,
      "step": 3380
    },
    {
      "epoch": 0.3575949367088608,
      "grad_norm": 2.9988930225372314,
      "learning_rate": 0.0010888700794387867,
      "loss": 4.6448,
      "step": 3390
    },
    {
      "epoch": 0.35864978902953587,
      "grad_norm": 2.840087652206421,
      "learning_rate": 0.001086628467897672,
      "loss": 4.6458,
      "step": 3400
    },
    {
      "epoch": 0.35970464135021096,
      "grad_norm": 3.391510009765625,
      "learning_rate": 0.0010843830842702036,
      "loss": 4.6318,
      "step": 3410
    },
    {
      "epoch": 0.36075949367088606,
      "grad_norm": 2.902096748352051,
      "learning_rate": 0.0010821339537170015,
      "loss": 4.6336,
      "step": 3420
    },
    {
      "epoch": 0.3618143459915612,
      "grad_norm": 2.794826030731201,
      "learning_rate": 0.0010798811014406716,
      "loss": 4.6141,
      "step": 3430
    },
    {
      "epoch": 0.3628691983122363,
      "grad_norm": 2.7015678882598877,
      "learning_rate": 0.0010776245526855235,
      "loss": 4.6076,
      "step": 3440
    },
    {
      "epoch": 0.3639240506329114,
      "grad_norm": 3.226649522781372,
      "learning_rate": 0.0010753643327372886,
      "loss": 4.6197,
      "step": 3450
    },
    {
      "epoch": 0.3649789029535865,
      "grad_norm": 3.007014274597168,
      "learning_rate": 0.001073100466922834,
      "loss": 4.6341,
      "step": 3460
    },
    {
      "epoch": 0.36603375527426163,
      "grad_norm": 2.9536163806915283,
      "learning_rate": 0.0010708329806098822,
      "loss": 4.6276,
      "step": 3470
    },
    {
      "epoch": 0.3670886075949367,
      "grad_norm": 2.414544105529785,
      "learning_rate": 0.0010685618992067243,
      "loss": 4.622,
      "step": 3480
    },
    {
      "epoch": 0.3681434599156118,
      "grad_norm": 2.5139007568359375,
      "learning_rate": 0.0010662872481619367,
      "loss": 4.6093,
      "step": 3490
    },
    {
      "epoch": 0.3691983122362869,
      "grad_norm": 2.673452377319336,
      "learning_rate": 0.0010640090529640948,
      "loss": 4.5998,
      "step": 3500
    },
    {
      "epoch": 0.370253164556962,
      "grad_norm": 3.1928837299346924,
      "learning_rate": 0.001061727339141488,
      "loss": 4.6124,
      "step": 3510
    },
    {
      "epoch": 0.37130801687763715,
      "grad_norm": 2.9210877418518066,
      "learning_rate": 0.0010594421322618341,
      "loss": 4.6103,
      "step": 3520
    },
    {
      "epoch": 0.37236286919831224,
      "grad_norm": 2.8277082443237305,
      "learning_rate": 0.001057153457931992,
      "loss": 4.6126,
      "step": 3530
    },
    {
      "epoch": 0.37341772151898733,
      "grad_norm": 2.984498977661133,
      "learning_rate": 0.0010548613417976748,
      "loss": 4.6394,
      "step": 3540
    },
    {
      "epoch": 0.3744725738396624,
      "grad_norm": 2.687344789505005,
      "learning_rate": 0.0010525658095431635,
      "loss": 4.6244,
      "step": 3550
    },
    {
      "epoch": 0.3755274261603376,
      "grad_norm": 3.3516616821289062,
      "learning_rate": 0.0010502668868910174,
      "loss": 4.6197,
      "step": 3560
    },
    {
      "epoch": 0.37658227848101267,
      "grad_norm": 4.074513912200928,
      "learning_rate": 0.0010479645996017875,
      "loss": 4.6509,
      "step": 3570
    },
    {
      "epoch": 0.37763713080168776,
      "grad_norm": 4.737973690032959,
      "learning_rate": 0.0010456589734737273,
      "loss": 4.6472,
      "step": 3580
    },
    {
      "epoch": 0.37869198312236285,
      "grad_norm": 2.885319471359253,
      "learning_rate": 0.001043350034342503,
      "loss": 4.6515,
      "step": 3590
    },
    {
      "epoch": 0.379746835443038,
      "grad_norm": 3.344156503677368,
      "learning_rate": 0.0010410378080809052,
      "loss": 4.6053,
      "step": 3600
    },
    {
      "epoch": 0.3808016877637131,
      "grad_norm": 2.7792258262634277,
      "learning_rate": 0.001038722320598558,
      "loss": 4.6058,
      "step": 3610
    },
    {
      "epoch": 0.3818565400843882,
      "grad_norm": 3.3080620765686035,
      "learning_rate": 0.0010364035978416297,
      "loss": 4.6035,
      "step": 3620
    },
    {
      "epoch": 0.3829113924050633,
      "grad_norm": 2.8839192390441895,
      "learning_rate": 0.0010340816657925407,
      "loss": 4.6016,
      "step": 3630
    },
    {
      "epoch": 0.38396624472573837,
      "grad_norm": 2.7843339443206787,
      "learning_rate": 0.0010317565504696733,
      "loss": 4.5949,
      "step": 3640
    },
    {
      "epoch": 0.3850210970464135,
      "grad_norm": 2.5266425609588623,
      "learning_rate": 0.0010294282779270802,
      "loss": 4.5968,
      "step": 3650
    },
    {
      "epoch": 0.3860759493670886,
      "grad_norm": 2.542349338531494,
      "learning_rate": 0.0010270968742541917,
      "loss": 4.5723,
      "step": 3660
    },
    {
      "epoch": 0.3871308016877637,
      "grad_norm": 3.161289691925049,
      "learning_rate": 0.001024762365575525,
      "loss": 4.6007,
      "step": 3670
    },
    {
      "epoch": 0.3881856540084388,
      "grad_norm": 2.4420015811920166,
      "learning_rate": 0.0010224247780503892,
      "loss": 4.5904,
      "step": 3680
    },
    {
      "epoch": 0.38924050632911394,
      "grad_norm": 2.9263546466827393,
      "learning_rate": 0.001020084137872594,
      "loss": 4.5761,
      "step": 3690
    },
    {
      "epoch": 0.39029535864978904,
      "grad_norm": 2.9866251945495605,
      "learning_rate": 0.0010177404712701558,
      "loss": 4.5985,
      "step": 3700
    },
    {
      "epoch": 0.39135021097046413,
      "grad_norm": 3.401718854904175,
      "learning_rate": 0.001015393804505003,
      "loss": 4.5929,
      "step": 3710
    },
    {
      "epoch": 0.3924050632911392,
      "grad_norm": 3.172886371612549,
      "learning_rate": 0.0010130441638726828,
      "loss": 4.5936,
      "step": 3720
    },
    {
      "epoch": 0.39345991561181437,
      "grad_norm": 3.3034322261810303,
      "learning_rate": 0.0010106915757020654,
      "loss": 4.6136,
      "step": 3730
    },
    {
      "epoch": 0.39451476793248946,
      "grad_norm": 2.8849706649780273,
      "learning_rate": 0.0010083360663550502,
      "loss": 4.62,
      "step": 3740
    },
    {
      "epoch": 0.39556962025316456,
      "grad_norm": 3.4315314292907715,
      "learning_rate": 0.0010059776622262698,
      "loss": 4.6161,
      "step": 3750
    },
    {
      "epoch": 0.39662447257383965,
      "grad_norm": 2.871699333190918,
      "learning_rate": 0.0010036163897427937,
      "loss": 4.6007,
      "step": 3760
    },
    {
      "epoch": 0.39767932489451474,
      "grad_norm": 3.350747585296631,
      "learning_rate": 0.001001252275363833,
      "loss": 4.5864,
      "step": 3770
    },
    {
      "epoch": 0.3987341772151899,
      "grad_norm": 6.969882965087891,
      "learning_rate": 0.0009988853455804442,
      "loss": 4.5909,
      "step": 3780
    },
    {
      "epoch": 0.399789029535865,
      "grad_norm": 3.071051836013794,
      "learning_rate": 0.0009965156269152308,
      "loss": 4.6002,
      "step": 3790
    },
    {
      "epoch": 0.4008438818565401,
      "grad_norm": 3.1520445346832275,
      "learning_rate": 0.0009941431459220475,
      "loss": 4.5835,
      "step": 3800
    },
    {
      "epoch": 0.40189873417721517,
      "grad_norm": 3.175671339035034,
      "learning_rate": 0.0009917679291857027,
      "loss": 4.587,
      "step": 3810
    },
    {
      "epoch": 0.4029535864978903,
      "grad_norm": 3.8424770832061768,
      "learning_rate": 0.0009893900033216593,
      "loss": 4.5847,
      "step": 3820
    },
    {
      "epoch": 0.4040084388185654,
      "grad_norm": 2.9847989082336426,
      "learning_rate": 0.000987009394975738,
      "loss": 4.593,
      "step": 3830
    },
    {
      "epoch": 0.4050632911392405,
      "grad_norm": 2.673767566680908,
      "learning_rate": 0.0009846261308238177,
      "loss": 4.6003,
      "step": 3840
    },
    {
      "epoch": 0.4061181434599156,
      "grad_norm": 3.192701578140259,
      "learning_rate": 0.0009822402375715366,
      "loss": 4.5998,
      "step": 3850
    },
    {
      "epoch": 0.40717299578059074,
      "grad_norm": 3.954310655593872,
      "learning_rate": 0.000979851741953994,
      "loss": 4.5992,
      "step": 3860
    },
    {
      "epoch": 0.40822784810126583,
      "grad_norm": 2.981520414352417,
      "learning_rate": 0.0009774606707354493,
      "loss": 4.5949,
      "step": 3870
    },
    {
      "epoch": 0.4092827004219409,
      "grad_norm": 3.151970148086548,
      "learning_rate": 0.0009750670507090233,
      "loss": 4.5648,
      "step": 3880
    },
    {
      "epoch": 0.410337552742616,
      "grad_norm": 2.6016056537628174,
      "learning_rate": 0.0009726709086963967,
      "loss": 4.5472,
      "step": 3890
    },
    {
      "epoch": 0.41139240506329117,
      "grad_norm": 3.549995183944702,
      "learning_rate": 0.0009702722715475113,
      "loss": 4.5533,
      "step": 3900
    },
    {
      "epoch": 0.41244725738396626,
      "grad_norm": 2.8463425636291504,
      "learning_rate": 0.0009678711661402672,
      "loss": 4.5612,
      "step": 3910
    },
    {
      "epoch": 0.41350210970464135,
      "grad_norm": 3.008061647415161,
      "learning_rate": 0.0009654676193802232,
      "loss": 4.5726,
      "step": 3920
    },
    {
      "epoch": 0.41455696202531644,
      "grad_norm": 3.4539527893066406,
      "learning_rate": 0.000963061658200294,
      "loss": 4.5766,
      "step": 3930
    },
    {
      "epoch": 0.41561181434599154,
      "grad_norm": 3.225945234298706,
      "learning_rate": 0.0009606533095604499,
      "loss": 4.5825,
      "step": 3940
    },
    {
      "epoch": 0.4166666666666667,
      "grad_norm": 3.2258594036102295,
      "learning_rate": 0.0009582426004474129,
      "loss": 4.572,
      "step": 3950
    },
    {
      "epoch": 0.4177215189873418,
      "grad_norm": 2.8319954872131348,
      "learning_rate": 0.0009558295578743559,
      "loss": 4.5805,
      "step": 3960
    },
    {
      "epoch": 0.41877637130801687,
      "grad_norm": 2.5683205127716064,
      "learning_rate": 0.0009534142088805994,
      "loss": 4.5558,
      "step": 3970
    },
    {
      "epoch": 0.41983122362869196,
      "grad_norm": 2.871180534362793,
      "learning_rate": 0.000950996580531308,
      "loss": 4.5538,
      "step": 3980
    },
    {
      "epoch": 0.4208860759493671,
      "grad_norm": 3.601771593093872,
      "learning_rate": 0.000948576699917188,
      "loss": 4.5528,
      "step": 3990
    },
    {
      "epoch": 0.4219409282700422,
      "grad_norm": 3.3274970054626465,
      "learning_rate": 0.0009461545941541832,
      "loss": 4.5973,
      "step": 4000
    },
    {
      "epoch": 0.4229957805907173,
      "grad_norm": 3.1300699710845947,
      "learning_rate": 0.0009437302903831712,
      "loss": 4.5732,
      "step": 4010
    },
    {
      "epoch": 0.4240506329113924,
      "grad_norm": 3.4191720485687256,
      "learning_rate": 0.0009413038157696595,
      "loss": 4.5666,
      "step": 4020
    },
    {
      "epoch": 0.42510548523206754,
      "grad_norm": 3.307903289794922,
      "learning_rate": 0.0009388751975034815,
      "loss": 4.5491,
      "step": 4030
    },
    {
      "epoch": 0.42616033755274263,
      "grad_norm": 3.1058967113494873,
      "learning_rate": 0.0009364444627984902,
      "loss": 4.5665,
      "step": 4040
    },
    {
      "epoch": 0.4272151898734177,
      "grad_norm": 3.429597854614258,
      "learning_rate": 0.0009340116388922551,
      "loss": 4.5653,
      "step": 4050
    },
    {
      "epoch": 0.4282700421940928,
      "grad_norm": 3.3910765647888184,
      "learning_rate": 0.0009315767530457556,
      "loss": 4.5783,
      "step": 4060
    },
    {
      "epoch": 0.4293248945147679,
      "grad_norm": 3.5425071716308594,
      "learning_rate": 0.0009291398325430771,
      "loss": 4.5947,
      "step": 4070
    },
    {
      "epoch": 0.43037974683544306,
      "grad_norm": 2.863107442855835,
      "learning_rate": 0.0009267009046911032,
      "loss": 4.5497,
      "step": 4080
    },
    {
      "epoch": 0.43143459915611815,
      "grad_norm": 2.8670520782470703,
      "learning_rate": 0.0009242599968192119,
      "loss": 4.554,
      "step": 4090
    },
    {
      "epoch": 0.43248945147679324,
      "grad_norm": 3.9779365062713623,
      "learning_rate": 0.0009218171362789674,
      "loss": 4.536,
      "step": 4100
    },
    {
      "epoch": 0.43354430379746833,
      "grad_norm": 2.962700128555298,
      "learning_rate": 0.000919372350443815,
      "loss": 4.5446,
      "step": 4110
    },
    {
      "epoch": 0.4345991561181435,
      "grad_norm": 3.4159317016601562,
      "learning_rate": 0.0009169256667087738,
      "loss": 4.5579,
      "step": 4120
    },
    {
      "epoch": 0.4356540084388186,
      "grad_norm": 3.5039987564086914,
      "learning_rate": 0.0009144771124901295,
      "loss": 4.5447,
      "step": 4130
    },
    {
      "epoch": 0.43670886075949367,
      "grad_norm": 3.0800342559814453,
      "learning_rate": 0.0009120267152251281,
      "loss": 4.5624,
      "step": 4140
    },
    {
      "epoch": 0.43776371308016876,
      "grad_norm": 2.8276429176330566,
      "learning_rate": 0.0009095745023716671,
      "loss": 4.5667,
      "step": 4150
    },
    {
      "epoch": 0.4388185654008439,
      "grad_norm": 2.726032018661499,
      "learning_rate": 0.0009071205014079888,
      "loss": 4.535,
      "step": 4160
    },
    {
      "epoch": 0.439873417721519,
      "grad_norm": 3.5510318279266357,
      "learning_rate": 0.0009046647398323728,
      "loss": 4.5347,
      "step": 4170
    },
    {
      "epoch": 0.4409282700421941,
      "grad_norm": 3.399991989135742,
      "learning_rate": 0.0009022072451628263,
      "loss": 4.559,
      "step": 4180
    },
    {
      "epoch": 0.4419831223628692,
      "grad_norm": 3.2381067276000977,
      "learning_rate": 0.0008997480449367771,
      "loss": 4.5635,
      "step": 4190
    },
    {
      "epoch": 0.4430379746835443,
      "grad_norm": 3.290466070175171,
      "learning_rate": 0.0008972871667107643,
      "loss": 4.5389,
      "step": 4200
    },
    {
      "epoch": 0.4440928270042194,
      "grad_norm": 3.3038268089294434,
      "learning_rate": 0.0008948246380601303,
      "loss": 4.5415,
      "step": 4210
    },
    {
      "epoch": 0.4451476793248945,
      "grad_norm": 2.8505473136901855,
      "learning_rate": 0.000892360486578711,
      "loss": 4.5383,
      "step": 4220
    },
    {
      "epoch": 0.4462025316455696,
      "grad_norm": 2.6588008403778076,
      "learning_rate": 0.0008898947398785271,
      "loss": 4.5277,
      "step": 4230
    },
    {
      "epoch": 0.4472573839662447,
      "grad_norm": 3.50152325630188,
      "learning_rate": 0.000887427425589474,
      "loss": 4.531,
      "step": 4240
    },
    {
      "epoch": 0.44831223628691985,
      "grad_norm": 3.214852809906006,
      "learning_rate": 0.0008849585713590134,
      "loss": 4.55,
      "step": 4250
    },
    {
      "epoch": 0.44936708860759494,
      "grad_norm": 3.2937514781951904,
      "learning_rate": 0.0008824882048518622,
      "loss": 4.5575,
      "step": 4260
    },
    {
      "epoch": 0.45042194092827004,
      "grad_norm": 3.690120220184326,
      "learning_rate": 0.0008800163537496837,
      "loss": 4.5584,
      "step": 4270
    },
    {
      "epoch": 0.45147679324894513,
      "grad_norm": 3.255889415740967,
      "learning_rate": 0.0008775430457507759,
      "loss": 4.5667,
      "step": 4280
    },
    {
      "epoch": 0.4525316455696203,
      "grad_norm": 3.4331369400024414,
      "learning_rate": 0.0008750683085697632,
      "loss": 4.5741,
      "step": 4290
    },
    {
      "epoch": 0.45358649789029537,
      "grad_norm": 2.9492852687835693,
      "learning_rate": 0.0008725921699372839,
      "loss": 4.5895,
      "step": 4300
    },
    {
      "epoch": 0.45464135021097046,
      "grad_norm": 2.763516426086426,
      "learning_rate": 0.0008701146575996804,
      "loss": 4.555,
      "step": 4310
    },
    {
      "epoch": 0.45569620253164556,
      "grad_norm": 3.1769704818725586,
      "learning_rate": 0.0008676357993186882,
      "loss": 4.5153,
      "step": 4320
    },
    {
      "epoch": 0.45675105485232065,
      "grad_norm": 3.3468739986419678,
      "learning_rate": 0.0008654036990414308,
      "loss": 4.5427,
      "step": 4330
    },
    {
      "epoch": 0.4578059071729958,
      "grad_norm": 3.2965071201324463,
      "learning_rate": 0.0008629223600053465,
      "loss": 4.5411,
      "step": 4340
    },
    {
      "epoch": 0.4588607594936709,
      "grad_norm": 2.8209431171417236,
      "learning_rate": 0.0008604397556190797,
      "loss": 4.5357,
      "step": 4350
    },
    {
      "epoch": 0.459915611814346,
      "grad_norm": 2.6621696949005127,
      "learning_rate": 0.0008579559137014254,
      "loss": 4.5361,
      "step": 4360
    },
    {
      "epoch": 0.4609704641350211,
      "grad_norm": 3.6703906059265137,
      "learning_rate": 0.0008554708620850445,
      "loss": 4.5243,
      "step": 4370
    },
    {
      "epoch": 0.4620253164556962,
      "grad_norm": 2.6653189659118652,
      "learning_rate": 0.0008529846286161539,
      "loss": 4.5379,
      "step": 4380
    },
    {
      "epoch": 0.4630801687763713,
      "grad_norm": 4.057528018951416,
      "learning_rate": 0.0008504972411542138,
      "loss": 4.5415,
      "step": 4390
    },
    {
      "epoch": 0.4641350210970464,
      "grad_norm": 2.342545509338379,
      "learning_rate": 0.000848008727571615,
      "loss": 4.5662,
      "step": 4400
    },
    {
      "epoch": 0.4651898734177215,
      "grad_norm": 2.9106974601745605,
      "learning_rate": 0.0008455191157533677,
      "loss": 4.5453,
      "step": 4410
    },
    {
      "epoch": 0.46624472573839665,
      "grad_norm": 4.176686763763428,
      "learning_rate": 0.0008430284335967876,
      "loss": 4.5473,
      "step": 4420
    },
    {
      "epoch": 0.46729957805907174,
      "grad_norm": 3.731269598007202,
      "learning_rate": 0.0008405367090111845,
      "loss": 4.5439,
      "step": 4430
    },
    {
      "epoch": 0.46835443037974683,
      "grad_norm": 3.6147356033325195,
      "learning_rate": 0.0008380439699175493,
      "loss": 4.5363,
      "step": 4440
    },
    {
      "epoch": 0.4694092827004219,
      "grad_norm": 3.511557102203369,
      "learning_rate": 0.0008355502442482403,
      "loss": 4.5265,
      "step": 4450
    },
    {
      "epoch": 0.4704641350210971,
      "grad_norm": 3.691112995147705,
      "learning_rate": 0.0008330555599466716,
      "loss": 4.5124,
      "step": 4460
    },
    {
      "epoch": 0.47151898734177217,
      "grad_norm": 3.494528293609619,
      "learning_rate": 0.0008305599449669989,
      "loss": 4.5338,
      "step": 4470
    },
    {
      "epoch": 0.47257383966244726,
      "grad_norm": 3.026498794555664,
      "learning_rate": 0.0008280634272738066,
      "loss": 4.5385,
      "step": 4480
    },
    {
      "epoch": 0.47362869198312235,
      "grad_norm": 3.2435731887817383,
      "learning_rate": 0.0008255660348417944,
      "loss": 4.5209,
      "step": 4490
    },
    {
      "epoch": 0.47468354430379744,
      "grad_norm": 3.309622287750244,
      "learning_rate": 0.0008230677956554637,
      "loss": 4.5224,
      "step": 4500
    },
    {
      "epoch": 0.4757383966244726,
      "grad_norm": 3.1670479774475098,
      "learning_rate": 0.0008205687377088048,
      "loss": 4.5176,
      "step": 4510
    },
    {
      "epoch": 0.4767932489451477,
      "grad_norm": 3.0397000312805176,
      "learning_rate": 0.0008180688890049823,
      "loss": 4.516,
      "step": 4520
    },
    {
      "epoch": 0.4778481012658228,
      "grad_norm": 3.4361259937286377,
      "learning_rate": 0.0008155682775560215,
      "loss": 4.52,
      "step": 4530
    },
    {
      "epoch": 0.47890295358649787,
      "grad_norm": 3.779851198196411,
      "learning_rate": 0.0008130669313824944,
      "loss": 4.517,
      "step": 4540
    },
    {
      "epoch": 0.479957805907173,
      "grad_norm": 3.648378372192383,
      "learning_rate": 0.0008105648785132065,
      "loss": 4.5121,
      "step": 4550
    },
    {
      "epoch": 0.4810126582278481,
      "grad_norm": 2.8818464279174805,
      "learning_rate": 0.0008080621469848817,
      "loss": 4.5188,
      "step": 4560
    },
    {
      "epoch": 0.4820675105485232,
      "grad_norm": 3.1171231269836426,
      "learning_rate": 0.0008055587648418492,
      "loss": 4.4975,
      "step": 4570
    },
    {
      "epoch": 0.4831223628691983,
      "grad_norm": 3.2030348777770996,
      "learning_rate": 0.0008030547601357281,
      "loss": 4.4998,
      "step": 4580
    },
    {
      "epoch": 0.48417721518987344,
      "grad_norm": 3.1776087284088135,
      "learning_rate": 0.0008005501609251136,
      "loss": 4.518,
      "step": 4590
    },
    {
      "epoch": 0.48523206751054854,
      "grad_norm": 3.1814539432525635,
      "learning_rate": 0.0007980449952752633,
      "loss": 4.5233,
      "step": 4600
    },
    {
      "epoch": 0.48628691983122363,
      "grad_norm": 3.009540557861328,
      "learning_rate": 0.000795539291257781,
      "loss": 4.5171,
      "step": 4610
    },
    {
      "epoch": 0.4873417721518987,
      "grad_norm": 4.04673433303833,
      "learning_rate": 0.000793033076950304,
      "loss": 4.5101,
      "step": 4620
    },
    {
      "epoch": 0.4883966244725738,
      "grad_norm": 3.194821834564209,
      "learning_rate": 0.0007905263804361873,
      "loss": 4.5462,
      "step": 4630
    },
    {
      "epoch": 0.48945147679324896,
      "grad_norm": 3.2426836490631104,
      "learning_rate": 0.0007880192298041893,
      "loss": 4.5206,
      "step": 4640
    },
    {
      "epoch": 0.49050632911392406,
      "grad_norm": 3.2996058464050293,
      "learning_rate": 0.0007855116531481572,
      "loss": 4.51,
      "step": 4650
    },
    {
      "epoch": 0.49156118143459915,
      "grad_norm": 3.2834324836730957,
      "learning_rate": 0.0007830036785667116,
      "loss": 4.5004,
      "step": 4660
    },
    {
      "epoch": 0.49261603375527424,
      "grad_norm": 2.7887423038482666,
      "learning_rate": 0.0007804953341629326,
      "loss": 4.5193,
      "step": 4670
    },
    {
      "epoch": 0.4936708860759494,
      "grad_norm": 3.367150068283081,
      "learning_rate": 0.0007779866480440437,
      "loss": 4.5042,
      "step": 4680
    },
    {
      "epoch": 0.4947257383966245,
      "grad_norm": 2.977492570877075,
      "learning_rate": 0.0007754776483210981,
      "loss": 4.488,
      "step": 4690
    },
    {
      "epoch": 0.4957805907172996,
      "grad_norm": 3.519120454788208,
      "learning_rate": 0.0007729683631086627,
      "loss": 4.5012,
      "step": 4700
    },
    {
      "epoch": 0.49683544303797467,
      "grad_norm": 3.321234703063965,
      "learning_rate": 0.0007704588205245034,
      "loss": 4.509,
      "step": 4710
    },
    {
      "epoch": 0.4978902953586498,
      "grad_norm": 3.230801820755005,
      "learning_rate": 0.0007679490486892705,
      "loss": 4.5065,
      "step": 4720
    },
    {
      "epoch": 0.4989451476793249,
      "grad_norm": 3.595907688140869,
      "learning_rate": 0.0007654390757261827,
      "loss": 4.5063,
      "step": 4730
    },
    {
      "epoch": 0.5,
      "grad_norm": 3.3380320072174072,
      "learning_rate": 0.0007629289297607127,
      "loss": 4.5023,
      "step": 4740
    },
    {
      "epoch": 0.5010548523206751,
      "grad_norm": 3.2435503005981445,
      "learning_rate": 0.000760418638920272,
      "loss": 4.495,
      "step": 4750
    },
    {
      "epoch": 0.5021097046413502,
      "grad_norm": 3.4500696659088135,
      "learning_rate": 0.0007579082313338943,
      "loss": 4.4807,
      "step": 4760
    },
    {
      "epoch": 0.5031645569620253,
      "grad_norm": 2.8793153762817383,
      "learning_rate": 0.0007553977351319235,
      "loss": 4.4803,
      "step": 4770
    },
    {
      "epoch": 0.5042194092827004,
      "grad_norm": 3.8507115840911865,
      "learning_rate": 0.0007528871784456948,
      "loss": 4.4794,
      "step": 4780
    },
    {
      "epoch": 0.5052742616033755,
      "grad_norm": 3.6089282035827637,
      "learning_rate": 0.0007503765894072217,
      "loss": 4.5068,
      "step": 4790
    },
    {
      "epoch": 0.5063291139240507,
      "grad_norm": 3.528726100921631,
      "learning_rate": 0.0007478659961488805,
      "loss": 4.5079,
      "step": 4800
    },
    {
      "epoch": 0.5073839662447257,
      "grad_norm": 3.119722366333008,
      "learning_rate": 0.0007453554268030946,
      "loss": 4.5003,
      "step": 4810
    },
    {
      "epoch": 0.5084388185654009,
      "grad_norm": 3.3979053497314453,
      "learning_rate": 0.0007428449095020192,
      "loss": 4.5129,
      "step": 4820
    },
    {
      "epoch": 0.509493670886076,
      "grad_norm": 3.525012493133545,
      "learning_rate": 0.0007403344723772265,
      "loss": 4.4934,
      "step": 4830
    },
    {
      "epoch": 0.510548523206751,
      "grad_norm": 3.0945098400115967,
      "learning_rate": 0.0007378241435593901,
      "loss": 4.4938,
      "step": 4840
    },
    {
      "epoch": 0.5116033755274262,
      "grad_norm": 3.4211156368255615,
      "learning_rate": 0.0007353139511779707,
      "loss": 4.4854,
      "step": 4850
    },
    {
      "epoch": 0.5126582278481012,
      "grad_norm": 3.3163328170776367,
      "learning_rate": 0.0007328039233608993,
      "loss": 4.4702,
      "step": 4860
    },
    {
      "epoch": 0.5137130801687764,
      "grad_norm": 3.465574026107788,
      "learning_rate": 0.0007302940882342634,
      "loss": 4.4925,
      "step": 4870
    },
    {
      "epoch": 0.5147679324894515,
      "grad_norm": 3.7374582290649414,
      "learning_rate": 0.0007277844739219908,
      "loss": 4.5083,
      "step": 4880
    },
    {
      "epoch": 0.5158227848101266,
      "grad_norm": 3.117872953414917,
      "learning_rate": 0.0007252751085455355,
      "loss": 4.5039,
      "step": 4890
    },
    {
      "epoch": 0.5168776371308017,
      "grad_norm": 3.559821367263794,
      "learning_rate": 0.0007227660202235616,
      "loss": 4.4943,
      "step": 4900
    },
    {
      "epoch": 0.5179324894514767,
      "grad_norm": 3.142214775085449,
      "learning_rate": 0.0007202572370716292,
      "loss": 4.496,
      "step": 4910
    },
    {
      "epoch": 0.5189873417721519,
      "grad_norm": 3.048147439956665,
      "learning_rate": 0.0007177487872018784,
      "loss": 4.4815,
      "step": 4920
    },
    {
      "epoch": 0.520042194092827,
      "grad_norm": 2.8286218643188477,
      "learning_rate": 0.0007152406987227149,
      "loss": 4.4776,
      "step": 4930
    },
    {
      "epoch": 0.5210970464135021,
      "grad_norm": 3.049813985824585,
      "learning_rate": 0.0007127329997384946,
      "loss": 4.4658,
      "step": 4940
    },
    {
      "epoch": 0.5221518987341772,
      "grad_norm": 3.621373176574707,
      "learning_rate": 0.0007102257183492092,
      "loss": 4.4758,
      "step": 4950
    },
    {
      "epoch": 0.5232067510548524,
      "grad_norm": 3.5828049182891846,
      "learning_rate": 0.000707718882650171,
      "loss": 4.4793,
      "step": 4960
    },
    {
      "epoch": 0.5242616033755274,
      "grad_norm": 3.276440382003784,
      "learning_rate": 0.0007052125207316975,
      "loss": 4.4842,
      "step": 4970
    },
    {
      "epoch": 0.5253164556962026,
      "grad_norm": 2.9380855560302734,
      "learning_rate": 0.0007027066606787988,
      "loss": 4.4611,
      "step": 4980
    },
    {
      "epoch": 0.5263713080168776,
      "grad_norm": 2.8913345336914062,
      "learning_rate": 0.0007002013305708598,
      "loss": 4.4561,
      "step": 4990
    },
    {
      "epoch": 0.5274261603375527,
      "grad_norm": 2.8837273120880127,
      "learning_rate": 0.0006976965584813277,
      "loss": 4.4543,
      "step": 5000
    },
    {
      "epoch": 0.5284810126582279,
      "grad_norm": 3.460888147354126,
      "learning_rate": 0.000695192372477397,
      "loss": 4.4592,
      "step": 5010
    },
    {
      "epoch": 0.5295358649789029,
      "grad_norm": 3.782970428466797,
      "learning_rate": 0.0006926888006196944,
      "loss": 4.466,
      "step": 5020
    },
    {
      "epoch": 0.5305907172995781,
      "grad_norm": 3.4546279907226562,
      "learning_rate": 0.000690185870961965,
      "loss": 4.4753,
      "step": 5030
    },
    {
      "epoch": 0.5316455696202531,
      "grad_norm": 3.430248737335205,
      "learning_rate": 0.0006876836115507579,
      "loss": 4.4824,
      "step": 5040
    },
    {
      "epoch": 0.5327004219409283,
      "grad_norm": 3.0455105304718018,
      "learning_rate": 0.0006851820504251117,
      "loss": 4.4762,
      "step": 5050
    },
    {
      "epoch": 0.5337552742616034,
      "grad_norm": 3.2662668228149414,
      "learning_rate": 0.0006826812156162401,
      "loss": 4.4696,
      "step": 5060
    },
    {
      "epoch": 0.5348101265822784,
      "grad_norm": 3.1972880363464355,
      "learning_rate": 0.0006801811351472185,
      "loss": 4.4603,
      "step": 5070
    },
    {
      "epoch": 0.5358649789029536,
      "grad_norm": 3.152986764907837,
      "learning_rate": 0.000677681837032669,
      "loss": 4.4692,
      "step": 5080
    },
    {
      "epoch": 0.5369198312236287,
      "grad_norm": 3.7994041442871094,
      "learning_rate": 0.0006751833492784476,
      "loss": 4.4679,
      "step": 5090
    },
    {
      "epoch": 0.5379746835443038,
      "grad_norm": 3.1490654945373535,
      "learning_rate": 0.0006726856998813291,
      "loss": 4.4551,
      "step": 5100
    },
    {
      "epoch": 0.5390295358649789,
      "grad_norm": 3.9151251316070557,
      "learning_rate": 0.0006701889168286953,
      "loss": 4.4552,
      "step": 5110
    },
    {
      "epoch": 0.540084388185654,
      "grad_norm": 2.940554141998291,
      "learning_rate": 0.000667693028098219,
      "loss": 4.4685,
      "step": 5120
    },
    {
      "epoch": 0.5411392405063291,
      "grad_norm": 2.848872423171997,
      "learning_rate": 0.0006651980616575522,
      "loss": 4.4488,
      "step": 5130
    },
    {
      "epoch": 0.5421940928270043,
      "grad_norm": 3.832012891769409,
      "learning_rate": 0.0006627040454640123,
      "loss": 4.4518,
      "step": 5140
    },
    {
      "epoch": 0.5432489451476793,
      "grad_norm": 3.0563342571258545,
      "learning_rate": 0.0006602110074642682,
      "loss": 4.4558,
      "step": 5150
    },
    {
      "epoch": 0.5443037974683544,
      "grad_norm": 3.192211151123047,
      "learning_rate": 0.0006577189755940282,
      "loss": 4.449,
      "step": 5160
    },
    {
      "epoch": 0.5453586497890295,
      "grad_norm": 3.361438512802124,
      "learning_rate": 0.0006552279777777258,
      "loss": 4.4538,
      "step": 5170
    },
    {
      "epoch": 0.5464135021097046,
      "grad_norm": 3.468320846557617,
      "learning_rate": 0.0006527380419282088,
      "loss": 4.4533,
      "step": 5180
    },
    {
      "epoch": 0.5474683544303798,
      "grad_norm": 3.1625330448150635,
      "learning_rate": 0.0006502491959464235,
      "loss": 4.4369,
      "step": 5190
    },
    {
      "epoch": 0.5485232067510548,
      "grad_norm": 3.177511215209961,
      "learning_rate": 0.0006477614677211046,
      "loss": 4.4523,
      "step": 5200
    },
    {
      "epoch": 0.54957805907173,
      "grad_norm": 2.7941603660583496,
      "learning_rate": 0.0006452748851284615,
      "loss": 4.432,
      "step": 5210
    },
    {
      "epoch": 0.5506329113924051,
      "grad_norm": 3.381495714187622,
      "learning_rate": 0.0006427894760318664,
      "loss": 4.4325,
      "step": 5220
    },
    {
      "epoch": 0.5516877637130801,
      "grad_norm": 3.354574203491211,
      "learning_rate": 0.0006403052682815415,
      "loss": 4.4502,
      "step": 5230
    },
    {
      "epoch": 0.5527426160337553,
      "grad_norm": 3.33038592338562,
      "learning_rate": 0.0006378222897142482,
      "loss": 4.4638,
      "step": 5240
    },
    {
      "epoch": 0.5537974683544303,
      "grad_norm": 2.9987823963165283,
      "learning_rate": 0.0006353405681529734,
      "loss": 4.4527,
      "step": 5250
    },
    {
      "epoch": 0.5548523206751055,
      "grad_norm": 3.688260078430176,
      "learning_rate": 0.000632860131406619,
      "loss": 4.4503,
      "step": 5260
    },
    {
      "epoch": 0.5559071729957806,
      "grad_norm": 3.7564122676849365,
      "learning_rate": 0.00063038100726969,
      "loss": 4.4557,
      "step": 5270
    },
    {
      "epoch": 0.5569620253164557,
      "grad_norm": 3.0735015869140625,
      "learning_rate": 0.0006279032235219829,
      "loss": 4.4523,
      "step": 5280
    },
    {
      "epoch": 0.5580168776371308,
      "grad_norm": 2.9314091205596924,
      "learning_rate": 0.0006254268079282743,
      "loss": 4.4371,
      "step": 5290
    },
    {
      "epoch": 0.5590717299578059,
      "grad_norm": 3.121161699295044,
      "learning_rate": 0.00062295178823801,
      "loss": 4.4247,
      "step": 5300
    },
    {
      "epoch": 0.560126582278481,
      "grad_norm": 3.307239055633545,
      "learning_rate": 0.0006204781921849945,
      "loss": 4.439,
      "step": 5310
    },
    {
      "epoch": 0.5611814345991561,
      "grad_norm": 3.380671262741089,
      "learning_rate": 0.0006180060474870793,
      "loss": 4.4486,
      "step": 5320
    },
    {
      "epoch": 0.5622362869198312,
      "grad_norm": 3.1390392780303955,
      "learning_rate": 0.000615535381845853,
      "loss": 4.4462,
      "step": 5330
    },
    {
      "epoch": 0.5632911392405063,
      "grad_norm": 3.6875271797180176,
      "learning_rate": 0.0006130662229463301,
      "loss": 4.4394,
      "step": 5340
    },
    {
      "epoch": 0.5643459915611815,
      "grad_norm": 3.0988991260528564,
      "learning_rate": 0.0006105985984566421,
      "loss": 4.4435,
      "step": 5350
    },
    {
      "epoch": 0.5654008438818565,
      "grad_norm": 3.780428647994995,
      "learning_rate": 0.0006081325360277257,
      "loss": 4.4404,
      "step": 5360
    },
    {
      "epoch": 0.5664556962025317,
      "grad_norm": 3.4755797386169434,
      "learning_rate": 0.0006056680632930154,
      "loss": 4.4523,
      "step": 5370
    },
    {
      "epoch": 0.5675105485232067,
      "grad_norm": 2.8833975791931152,
      "learning_rate": 0.0006032052078681312,
      "loss": 4.4396,
      "step": 5380
    },
    {
      "epoch": 0.5685654008438819,
      "grad_norm": 3.0409247875213623,
      "learning_rate": 0.0006007439973505707,
      "loss": 4.4202,
      "step": 5390
    },
    {
      "epoch": 0.569620253164557,
      "grad_norm": 3.9057488441467285,
      "learning_rate": 0.0005982844593193995,
      "loss": 4.4252,
      "step": 5400
    },
    {
      "epoch": 0.570675105485232,
      "grad_norm": 3.0410540103912354,
      "learning_rate": 0.0005958266213349422,
      "loss": 4.444,
      "step": 5410
    },
    {
      "epoch": 0.5717299578059072,
      "grad_norm": 2.9824821949005127,
      "learning_rate": 0.0005933705109384735,
      "loss": 4.4266,
      "step": 5420
    },
    {
      "epoch": 0.5727848101265823,
      "grad_norm": 3.238953113555908,
      "learning_rate": 0.0005909161556519096,
      "loss": 4.4305,
      "step": 5430
    },
    {
      "epoch": 0.5738396624472574,
      "grad_norm": 3.3729684352874756,
      "learning_rate": 0.0005884635829775002,
      "loss": 4.4307,
      "step": 5440
    },
    {
      "epoch": 0.5748945147679325,
      "grad_norm": 3.900089740753174,
      "learning_rate": 0.0005860128203975196,
      "loss": 4.4492,
      "step": 5450
    },
    {
      "epoch": 0.5759493670886076,
      "grad_norm": 3.2625949382781982,
      "learning_rate": 0.0005835638953739589,
      "loss": 4.4485,
      "step": 5460
    },
    {
      "epoch": 0.5770042194092827,
      "grad_norm": 3.4712352752685547,
      "learning_rate": 0.0005811168353482191,
      "loss": 4.4413,
      "step": 5470
    },
    {
      "epoch": 0.5780590717299579,
      "grad_norm": 3.7630677223205566,
      "learning_rate": 0.0005786716677408025,
      "loss": 4.4323,
      "step": 5480
    },
    {
      "epoch": 0.5791139240506329,
      "grad_norm": 3.911316394805908,
      "learning_rate": 0.0005762284199510059,
      "loss": 4.4242,
      "step": 5490
    },
    {
      "epoch": 0.580168776371308,
      "grad_norm": 3.6942920684814453,
      "learning_rate": 0.0005737871193566141,
      "loss": 4.4474,
      "step": 5500
    },
    {
      "epoch": 0.5812236286919831,
      "grad_norm": 3.516204357147217,
      "learning_rate": 0.0005713477933135923,
      "loss": 4.4505,
      "step": 5510
    },
    {
      "epoch": 0.5822784810126582,
      "grad_norm": 3.2319347858428955,
      "learning_rate": 0.0005689104691557798,
      "loss": 4.4385,
      "step": 5520
    },
    {
      "epoch": 0.5833333333333334,
      "grad_norm": 3.6643333435058594,
      "learning_rate": 0.0005664751741945839,
      "loss": 4.4434,
      "step": 5530
    },
    {
      "epoch": 0.5843881856540084,
      "grad_norm": 3.1461639404296875,
      "learning_rate": 0.0005640419357186738,
      "loss": 4.4358,
      "step": 5540
    },
    {
      "epoch": 0.5854430379746836,
      "grad_norm": 3.308415651321411,
      "learning_rate": 0.0005616107809936746,
      "loss": 4.4299,
      "step": 5550
    },
    {
      "epoch": 0.5864978902953587,
      "grad_norm": 3.5486373901367188,
      "learning_rate": 0.0005591817372618621,
      "loss": 4.4287,
      "step": 5560
    },
    {
      "epoch": 0.5875527426160337,
      "grad_norm": 3.634885549545288,
      "learning_rate": 0.0005567548317418576,
      "loss": 4.4279,
      "step": 5570
    },
    {
      "epoch": 0.5886075949367089,
      "grad_norm": 3.5677859783172607,
      "learning_rate": 0.0005543300916283223,
      "loss": 4.4271,
      "step": 5580
    },
    {
      "epoch": 0.5896624472573839,
      "grad_norm": 3.5759193897247314,
      "learning_rate": 0.0005519075440916534,
      "loss": 4.4384,
      "step": 5590
    },
    {
      "epoch": 0.5907172995780591,
      "grad_norm": 4.042449951171875,
      "learning_rate": 0.0005494872162776786,
      "loss": 4.4376,
      "step": 5600
    },
    {
      "epoch": 0.5917721518987342,
      "grad_norm": 2.929685592651367,
      "learning_rate": 0.0005470691353073531,
      "loss": 4.4359,
      "step": 5610
    },
    {
      "epoch": 0.5928270042194093,
      "grad_norm": 3.289767265319824,
      "learning_rate": 0.0005446533282764543,
      "loss": 4.4237,
      "step": 5620
    },
    {
      "epoch": 0.5938818565400844,
      "grad_norm": 4.240915298461914,
      "learning_rate": 0.0005422398222552806,
      "loss": 4.4282,
      "step": 5630
    },
    {
      "epoch": 0.5949367088607594,
      "grad_norm": 3.4194846153259277,
      "learning_rate": 0.0005398286442883448,
      "loss": 4.4455,
      "step": 5640
    },
    {
      "epoch": 0.5959915611814346,
      "grad_norm": 4.180428981781006,
      "learning_rate": 0.0005374198213940734,
      "loss": 4.4277,
      "step": 5650
    },
    {
      "epoch": 0.5970464135021097,
      "grad_norm": 3.635732889175415,
      "learning_rate": 0.0005350133805645034,
      "loss": 4.4373,
      "step": 5660
    },
    {
      "epoch": 0.5981012658227848,
      "grad_norm": 3.748232841491699,
      "learning_rate": 0.000532609348764979,
      "loss": 4.4384,
      "step": 5670
    },
    {
      "epoch": 0.5991561181434599,
      "grad_norm": 3.5721983909606934,
      "learning_rate": 0.0005302077529338507,
      "loss": 4.44,
      "step": 5680
    },
    {
      "epoch": 0.6002109704641351,
      "grad_norm": 3.570868730545044,
      "learning_rate": 0.0005278086199821718,
      "loss": 4.4338,
      "step": 5690
    },
    {
      "epoch": 0.6012658227848101,
      "grad_norm": 3.287091016769409,
      "learning_rate": 0.0005254119767933992,
      "loss": 4.429,
      "step": 5700
    },
    {
      "epoch": 0.6023206751054853,
      "grad_norm": 3.0535316467285156,
      "learning_rate": 0.00052301785022309,
      "loss": 4.4242,
      "step": 5710
    },
    {
      "epoch": 0.6033755274261603,
      "grad_norm": 3.1510660648345947,
      "learning_rate": 0.0005206262670986012,
      "loss": 4.4157,
      "step": 5720
    },
    {
      "epoch": 0.6044303797468354,
      "grad_norm": 3.5819060802459717,
      "learning_rate": 0.0005182372542187895,
      "loss": 4.4143,
      "step": 5730
    },
    {
      "epoch": 0.6054852320675106,
      "grad_norm": 3.865048408508301,
      "learning_rate": 0.0005158508383537109,
      "loss": 4.4288,
      "step": 5740
    },
    {
      "epoch": 0.6065400843881856,
      "grad_norm": 3.293900966644287,
      "learning_rate": 0.00051346704624432,
      "loss": 4.4172,
      "step": 5750
    },
    {
      "epoch": 0.6075949367088608,
      "grad_norm": 3.0748679637908936,
      "learning_rate": 0.000511085904602172,
      "loss": 4.4249,
      "step": 5760
    },
    {
      "epoch": 0.6086497890295358,
      "grad_norm": 4.089321613311768,
      "learning_rate": 0.0005087074401091212,
      "loss": 4.4221,
      "step": 5770
    },
    {
      "epoch": 0.609704641350211,
      "grad_norm": 3.5329396724700928,
      "learning_rate": 0.0005063316794170239,
      "loss": 4.4356,
      "step": 5780
    },
    {
      "epoch": 0.6107594936708861,
      "grad_norm": 3.693916082382202,
      "learning_rate": 0.0005039586491474386,
      "loss": 4.4157,
      "step": 5790
    },
    {
      "epoch": 0.6118143459915611,
      "grad_norm": 3.4709231853485107,
      "learning_rate": 0.0005015883758913281,
      "loss": 4.4246,
      "step": 5800
    },
    {
      "epoch": 0.6128691983122363,
      "grad_norm": 3.9887397289276123,
      "learning_rate": 0.0004992208862087616,
      "loss": 4.4245,
      "step": 5810
    },
    {
      "epoch": 0.6139240506329114,
      "grad_norm": 3.357384204864502,
      "learning_rate": 0.0004968562066286168,
      "loss": 4.4342,
      "step": 5820
    },
    {
      "epoch": 0.6149789029535865,
      "grad_norm": 3.5773301124572754,
      "learning_rate": 0.0004944943636482836,
      "loss": 4.4197,
      "step": 5830
    },
    {
      "epoch": 0.6160337552742616,
      "grad_norm": 3.454458236694336,
      "learning_rate": 0.0004921353837333657,
      "loss": 4.4113,
      "step": 5840
    },
    {
      "epoch": 0.6170886075949367,
      "grad_norm": 3.6243648529052734,
      "learning_rate": 0.0004897792933173847,
      "loss": 4.4162,
      "step": 5850
    },
    {
      "epoch": 0.6181434599156118,
      "grad_norm": 3.646883249282837,
      "learning_rate": 0.0004874261188014842,
      "loss": 4.4063,
      "step": 5860
    },
    {
      "epoch": 0.619198312236287,
      "grad_norm": 3.113924026489258,
      "learning_rate": 0.00048507588655413367,
      "loss": 4.398,
      "step": 5870
    },
    {
      "epoch": 0.620253164556962,
      "grad_norm": 3.4230501651763916,
      "learning_rate": 0.0004827286229108331,
      "loss": 4.3981,
      "step": 5880
    },
    {
      "epoch": 0.6213080168776371,
      "grad_norm": 3.9238340854644775,
      "learning_rate": 0.0004803843541738173,
      "loss": 4.4069,
      "step": 5890
    },
    {
      "epoch": 0.6223628691983122,
      "grad_norm": 3.303436517715454,
      "learning_rate": 0.0004780431066117629,
      "loss": 4.4218,
      "step": 5900
    },
    {
      "epoch": 0.6234177215189873,
      "grad_norm": 3.3049023151397705,
      "learning_rate": 0.00047570490645949175,
      "loss": 4.4103,
      "step": 5910
    },
    {
      "epoch": 0.6244725738396625,
      "grad_norm": 3.1957497596740723,
      "learning_rate": 0.0004733697799176781,
      "loss": 4.4006,
      "step": 5920
    },
    {
      "epoch": 0.6255274261603375,
      "grad_norm": 3.774670362472534,
      "learning_rate": 0.0004710377531525552,
      "loss": 4.3986,
      "step": 5930
    },
    {
      "epoch": 0.6265822784810127,
      "grad_norm": 3.9058892726898193,
      "learning_rate": 0.00046870885229562153,
      "loss": 4.4136,
      "step": 5940
    },
    {
      "epoch": 0.6276371308016878,
      "grad_norm": 3.088975191116333,
      "learning_rate": 0.00046638310344334835,
      "loss": 4.411,
      "step": 5950
    },
    {
      "epoch": 0.6286919831223629,
      "grad_norm": 3.696850299835205,
      "learning_rate": 0.0004640605326568874,
      "loss": 4.3938,
      "step": 5960
    },
    {
      "epoch": 0.629746835443038,
      "grad_norm": 4.2880539894104,
      "learning_rate": 0.00046174116596177833,
      "loss": 4.4008,
      "step": 5970
    },
    {
      "epoch": 0.630801687763713,
      "grad_norm": 3.434882879257202,
      "learning_rate": 0.00045942502934765735,
      "loss": 4.4128,
      "step": 5980
    },
    {
      "epoch": 0.6318565400843882,
      "grad_norm": 3.839188814163208,
      "learning_rate": 0.00045711214876796623,
      "loss": 4.42,
      "step": 5990
    },
    {
      "epoch": 0.6329113924050633,
      "grad_norm": 4.07228946685791,
      "learning_rate": 0.00045480255013966123,
      "loss": 4.4306,
      "step": 6000
    },
    {
      "epoch": 0.6339662447257384,
      "grad_norm": 3.7824175357818604,
      "learning_rate": 0.0004524962593429227,
      "loss": 4.4249,
      "step": 6010
    },
    {
      "epoch": 0.6350210970464135,
      "grad_norm": 3.6401214599609375,
      "learning_rate": 0.0004501933022208649,
      "loss": 4.4232,
      "step": 6020
    },
    {
      "epoch": 0.6360759493670886,
      "grad_norm": 3.3243372440338135,
      "learning_rate": 0.0004478937045792474,
      "loss": 4.399,
      "step": 6030
    },
    {
      "epoch": 0.6371308016877637,
      "grad_norm": 3.512340545654297,
      "learning_rate": 0.00044559749218618444,
      "loss": 4.4001,
      "step": 6040
    },
    {
      "epoch": 0.6381856540084389,
      "grad_norm": 3.8359873294830322,
      "learning_rate": 0.0004433046907718571,
      "loss": 4.4026,
      "step": 6050
    },
    {
      "epoch": 0.6392405063291139,
      "grad_norm": 4.5959272384643555,
      "learning_rate": 0.0004410153260282246,
      "loss": 4.4087,
      "step": 6060
    },
    {
      "epoch": 0.640295358649789,
      "grad_norm": 3.5270018577575684,
      "learning_rate": 0.0004387294236087368,
      "loss": 4.4123,
      "step": 6070
    },
    {
      "epoch": 0.6413502109704642,
      "grad_norm": 3.62646484375,
      "learning_rate": 0.0004364470091280463,
      "loss": 4.4145,
      "step": 6080
    },
    {
      "epoch": 0.6424050632911392,
      "grad_norm": 3.673903226852417,
      "learning_rate": 0.00043416810816172244,
      "loss": 4.4123,
      "step": 6090
    },
    {
      "epoch": 0.6434599156118144,
      "grad_norm": 3.8226370811462402,
      "learning_rate": 0.0004318927462459629,
      "loss": 4.4055,
      "step": 6100
    },
    {
      "epoch": 0.6445147679324894,
      "grad_norm": 3.7879207134246826,
      "learning_rate": 0.000429620948877309,
      "loss": 4.4165,
      "step": 6110
    },
    {
      "epoch": 0.6455696202531646,
      "grad_norm": 3.618147611618042,
      "learning_rate": 0.00042735274151235953,
      "loss": 4.4136,
      "step": 6120
    },
    {
      "epoch": 0.6466244725738397,
      "grad_norm": 3.8073320388793945,
      "learning_rate": 0.0004250881495674855,
      "loss": 4.4136,
      "step": 6130
    },
    {
      "epoch": 0.6476793248945147,
      "grad_norm": 3.799858331680298,
      "learning_rate": 0.00042282719841854567,
      "loss": 4.4157,
      "step": 6140
    },
    {
      "epoch": 0.6487341772151899,
      "grad_norm": 3.574157953262329,
      "learning_rate": 0.0004205699134006011,
      "loss": 4.401,
      "step": 6150
    },
    {
      "epoch": 0.6497890295358649,
      "grad_norm": 3.354156494140625,
      "learning_rate": 0.00041831631980763324,
      "loss": 4.3985,
      "step": 6160
    },
    {
      "epoch": 0.6508438818565401,
      "grad_norm": 3.283491373062134,
      "learning_rate": 0.0004160664428922586,
      "loss": 4.3842,
      "step": 6170
    },
    {
      "epoch": 0.6518987341772152,
      "grad_norm": 3.2704107761383057,
      "learning_rate": 0.0004138203078654463,
      "loss": 4.3795,
      "step": 6180
    },
    {
      "epoch": 0.6529535864978903,
      "grad_norm": 3.7862582206726074,
      "learning_rate": 0.00041157793989623625,
      "loss": 4.3912,
      "step": 6190
    },
    {
      "epoch": 0.6540084388185654,
      "grad_norm": 3.939448118209839,
      "learning_rate": 0.0004093393641114565,
      "loss": 4.4069,
      "step": 6200
    },
    {
      "epoch": 0.6550632911392406,
      "grad_norm": 3.688568592071533,
      "learning_rate": 0.00040710460559544167,
      "loss": 4.4065,
      "step": 6210
    },
    {
      "epoch": 0.6561181434599156,
      "grad_norm": 3.3631296157836914,
      "learning_rate": 0.00040487368938975214,
      "loss": 4.4016,
      "step": 6220
    },
    {
      "epoch": 0.6571729957805907,
      "grad_norm": 3.3922948837280273,
      "learning_rate": 0.00040264664049289336,
      "loss": 4.3946,
      "step": 6230
    },
    {
      "epoch": 0.6582278481012658,
      "grad_norm": 3.472234010696411,
      "learning_rate": 0.0004004234838600357,
      "loss": 4.3857,
      "step": 6240
    },
    {
      "epoch": 0.6592827004219409,
      "grad_norm": 3.768303394317627,
      "learning_rate": 0.00039820424440273474,
      "loss": 4.3916,
      "step": 6250
    },
    {
      "epoch": 0.6603375527426161,
      "grad_norm": 3.9347000122070312,
      "learning_rate": 0.00039598894698865216,
      "loss": 4.4005,
      "step": 6260
    },
    {
      "epoch": 0.6613924050632911,
      "grad_norm": 3.940767526626587,
      "learning_rate": 0.0003937776164412773,
      "loss": 4.4019,
      "step": 6270
    },
    {
      "epoch": 0.6624472573839663,
      "grad_norm": 3.7503554821014404,
      "learning_rate": 0.0003915702775396483,
      "loss": 4.4121,
      "step": 6280
    },
    {
      "epoch": 0.6635021097046413,
      "grad_norm": 3.3544650077819824,
      "learning_rate": 0.0003893669550180761,
      "loss": 4.4084,
      "step": 6290
    },
    {
      "epoch": 0.6645569620253164,
      "grad_norm": 3.7076947689056396,
      "learning_rate": 0.00038716767356586487,
      "loss": 4.3961,
      "step": 6300
    },
    {
      "epoch": 0.6656118143459916,
      "grad_norm": 3.1956872940063477,
      "learning_rate": 0.0003849724578270374,
      "loss": 4.3892,
      "step": 6310
    },
    {
      "epoch": 0.6666666666666666,
      "grad_norm": 3.7957167625427246,
      "learning_rate": 0.0003827813324000578,
      "loss": 4.3895,
      "step": 6320
    },
    {
      "epoch": 0.6677215189873418,
      "grad_norm": 3.4237401485443115,
      "learning_rate": 0.00038059432183755633,
      "loss": 4.3827,
      "step": 6330
    },
    {
      "epoch": 0.6687763713080169,
      "grad_norm": 3.747195243835449,
      "learning_rate": 0.00037841145064605416,
      "loss": 4.3759,
      "step": 6340
    },
    {
      "epoch": 0.669831223628692,
      "grad_norm": 3.242246150970459,
      "learning_rate": 0.0003762327432856892,
      "loss": 4.3754,
      "step": 6350
    },
    {
      "epoch": 0.6708860759493671,
      "grad_norm": 4.00953483581543,
      "learning_rate": 0.000374058224169941,
      "loss": 4.3676,
      "step": 6360
    },
    {
      "epoch": 0.6719409282700421,
      "grad_norm": 4.017233371734619,
      "learning_rate": 0.00037188791766535825,
      "loss": 4.3769,
      "step": 6370
    },
    {
      "epoch": 0.6729957805907173,
      "grad_norm": 3.5586800575256348,
      "learning_rate": 0.0003697218480912848,
      "loss": 4.3664,
      "step": 6380
    },
    {
      "epoch": 0.6740506329113924,
      "grad_norm": 3.630129098892212,
      "learning_rate": 0.0003675600397195881,
      "loss": 4.3673,
      "step": 6390
    },
    {
      "epoch": 0.6751054852320675,
      "grad_norm": 3.9428844451904297,
      "learning_rate": 0.0003654025167743864,
      "loss": 4.381,
      "step": 6400
    },
    {
      "epoch": 0.6761603375527426,
      "grad_norm": 4.001683712005615,
      "learning_rate": 0.00036324930343177754,
      "loss": 4.3941,
      "step": 6410
    },
    {
      "epoch": 0.6772151898734177,
      "grad_norm": 3.8821346759796143,
      "learning_rate": 0.00036110042381956895,
      "loss": 4.4058,
      "step": 6420
    },
    {
      "epoch": 0.6782700421940928,
      "grad_norm": 3.7010772228240967,
      "learning_rate": 0.0003589559020170058,
      "loss": 4.3982,
      "step": 6430
    },
    {
      "epoch": 0.679324894514768,
      "grad_norm": 3.5186564922332764,
      "learning_rate": 0.0003568157620545019,
      "loss": 4.3919,
      "step": 6440
    },
    {
      "epoch": 0.680379746835443,
      "grad_norm": 3.541464328765869,
      "learning_rate": 0.00035468002791337047,
      "loss": 4.3893,
      "step": 6450
    },
    {
      "epoch": 0.6814345991561181,
      "grad_norm": 3.2619521617889404,
      "learning_rate": 0.0003525487235255556,
      "loss": 4.3887,
      "step": 6460
    },
    {
      "epoch": 0.6824894514767933,
      "grad_norm": 3.504340648651123,
      "learning_rate": 0.00035042187277336325,
      "loss": 4.3722,
      "step": 6470
    },
    {
      "epoch": 0.6835443037974683,
      "grad_norm": 3.4049315452575684,
      "learning_rate": 0.00034829949948919517,
      "loss": 4.3749,
      "step": 6480
    },
    {
      "epoch": 0.6845991561181435,
      "grad_norm": 3.731231927871704,
      "learning_rate": 0.00034618162745528,
      "loss": 4.374,
      "step": 6490
    },
    {
      "epoch": 0.6856540084388185,
      "grad_norm": 3.831712245941162,
      "learning_rate": 0.0003440682804034081,
      "loss": 4.3945,
      "step": 6500
    },
    {
      "epoch": 0.6867088607594937,
      "grad_norm": 3.501157522201538,
      "learning_rate": 0.0003419594820146652,
      "loss": 4.3883,
      "step": 6510
    },
    {
      "epoch": 0.6877637130801688,
      "grad_norm": 3.498037815093994,
      "learning_rate": 0.0003398552559191667,
      "loss": 4.3794,
      "step": 6520
    },
    {
      "epoch": 0.6888185654008439,
      "grad_norm": 3.7124767303466797,
      "learning_rate": 0.0003377556256957936,
      "loss": 4.3822,
      "step": 6530
    },
    {
      "epoch": 0.689873417721519,
      "grad_norm": 3.8737070560455322,
      "learning_rate": 0.0003356606148719277,
      "loss": 4.368,
      "step": 6540
    },
    {
      "epoch": 0.6909282700421941,
      "grad_norm": 3.512211322784424,
      "learning_rate": 0.0003335702469231884,
      "loss": 4.3747,
      "step": 6550
    },
    {
      "epoch": 0.6919831223628692,
      "grad_norm": 3.4354124069213867,
      "learning_rate": 0.0003314845452731691,
      "loss": 4.3743,
      "step": 6560
    },
    {
      "epoch": 0.6930379746835443,
      "grad_norm": 3.535693883895874,
      "learning_rate": 0.00032940353329317533,
      "loss": 4.3701,
      "step": 6570
    },
    {
      "epoch": 0.6940928270042194,
      "grad_norm": 3.44981050491333,
      "learning_rate": 0.00032732723430196236,
      "loss": 4.3728,
      "step": 6580
    },
    {
      "epoch": 0.6951476793248945,
      "grad_norm": 3.921905040740967,
      "learning_rate": 0.0003252556715654743,
      "loss": 4.3698,
      "step": 6590
    },
    {
      "epoch": 0.6962025316455697,
      "grad_norm": 4.131836891174316,
      "learning_rate": 0.00032339533378696424,
      "loss": 4.3871,
      "step": 6600
    },
    {
      "epoch": 0.6972573839662447,
      "grad_norm": 4.10322904586792,
      "learning_rate": 0.00032133283384184173,
      "loss": 4.3915,
      "step": 6610
    },
    {
      "epoch": 0.6983122362869199,
      "grad_norm": 4.003731727600098,
      "learning_rate": 0.00031927513732162856,
      "loss": 4.3889,
      "step": 6620
    },
    {
      "epoch": 0.6993670886075949,
      "grad_norm": 3.930255889892578,
      "learning_rate": 0.00031722226728381854,
      "loss": 4.3926,
      "step": 6630
    },
    {
      "epoch": 0.70042194092827,
      "grad_norm": 3.518800735473633,
      "learning_rate": 0.000315174246731823,
      "loss": 4.3817,
      "step": 6640
    },
    {
      "epoch": 0.7014767932489452,
      "grad_norm": 3.8460512161254883,
      "learning_rate": 0.00031313109861471223,
      "loss": 4.3859,
      "step": 6650
    },
    {
      "epoch": 0.7025316455696202,
      "grad_norm": 4.673398017883301,
      "learning_rate": 0.0003110928458269584,
      "loss": 4.3895,
      "step": 6660
    },
    {
      "epoch": 0.7035864978902954,
      "grad_norm": 4.496396541595459,
      "learning_rate": 0.00030905951120817934,
      "loss": 4.3949,
      "step": 6670
    },
    {
      "epoch": 0.7046413502109705,
      "grad_norm": 3.6391491889953613,
      "learning_rate": 0.00030703111754288204,
      "loss": 4.3955,
      "step": 6680
    },
    {
      "epoch": 0.7056962025316456,
      "grad_norm": 3.6032145023345947,
      "learning_rate": 0.0003050076875602084,
      "loss": 4.3824,
      "step": 6690
    },
    {
      "epoch": 0.7067510548523207,
      "grad_norm": 3.64402437210083,
      "learning_rate": 0.00030298924393367923,
      "loss": 4.3746,
      "step": 6700
    },
    {
      "epoch": 0.7078059071729957,
      "grad_norm": 3.6424930095672607,
      "learning_rate": 0.0003009758092809409,
      "loss": 4.3671,
      "step": 6710
    },
    {
      "epoch": 0.7088607594936709,
      "grad_norm": 3.5588512420654297,
      "learning_rate": 0.00029896740616351187,
      "loss": 4.3713,
      "step": 6720
    },
    {
      "epoch": 0.709915611814346,
      "grad_norm": 4.200392723083496,
      "learning_rate": 0.00029696405708652966,
      "loss": 4.3738,
      "step": 6730
    },
    {
      "epoch": 0.7109704641350211,
      "grad_norm": 4.480619430541992,
      "learning_rate": 0.00029496578449849867,
      "loss": 4.3912,
      "step": 6740
    },
    {
      "epoch": 0.7120253164556962,
      "grad_norm": 4.200009346008301,
      "learning_rate": 0.00029297261079103945,
      "loss": 4.3997,
      "step": 6750
    },
    {
      "epoch": 0.7130801687763713,
      "grad_norm": 4.229198455810547,
      "learning_rate": 0.00029098455829863653,
      "loss": 4.3907,
      "step": 6760
    },
    {
      "epoch": 0.7141350210970464,
      "grad_norm": 3.606245517730713,
      "learning_rate": 0.0002890016492983886,
      "loss": 4.3884,
      "step": 6770
    },
    {
      "epoch": 0.7151898734177216,
      "grad_norm": 3.7194836139678955,
      "learning_rate": 0.00028702390600975937,
      "loss": 4.3705,
      "step": 6780
    },
    {
      "epoch": 0.7162447257383966,
      "grad_norm": 4.486773490905762,
      "learning_rate": 0.0002850513505943281,
      "loss": 4.3755,
      "step": 6790
    },
    {
      "epoch": 0.7172995780590717,
      "grad_norm": 3.8849730491638184,
      "learning_rate": 0.0002830840051555414,
      "loss": 4.3817,
      "step": 6800
    },
    {
      "epoch": 0.7183544303797469,
      "grad_norm": 3.8619751930236816,
      "learning_rate": 0.0002811218917384652,
      "loss": 4.3804,
      "step": 6810
    },
    {
      "epoch": 0.7194092827004219,
      "grad_norm": 3.620556116104126,
      "learning_rate": 0.00027916503232953895,
      "loss": 4.3841,
      "step": 6820
    },
    {
      "epoch": 0.7204641350210971,
      "grad_norm": 3.744481086730957,
      "learning_rate": 0.00027721344885632765,
      "loss": 4.3676,
      "step": 6830
    },
    {
      "epoch": 0.7215189873417721,
      "grad_norm": 3.7527897357940674,
      "learning_rate": 0.000275267163187277,
      "loss": 4.3638,
      "step": 6840
    },
    {
      "epoch": 0.7225738396624473,
      "grad_norm": 4.121515274047852,
      "learning_rate": 0.00027332619713146816,
      "loss": 4.3621,
      "step": 6850
    },
    {
      "epoch": 0.7236286919831224,
      "grad_norm": 4.567071437835693,
      "learning_rate": 0.0002713905724383737,
      "loss": 4.3811,
      "step": 6860
    },
    {
      "epoch": 0.7246835443037974,
      "grad_norm": 4.5510687828063965,
      "learning_rate": 0.00026946031079761346,
      "loss": 4.3894,
      "step": 6870
    },
    {
      "epoch": 0.7257383966244726,
      "grad_norm": 4.033400535583496,
      "learning_rate": 0.00026753543383871143,
      "loss": 4.3857,
      "step": 6880
    },
    {
      "epoch": 0.7267932489451476,
      "grad_norm": 3.6749753952026367,
      "learning_rate": 0.00026561596313085396,
      "loss": 4.3849,
      "step": 6890
    },
    {
      "epoch": 0.7278481012658228,
      "grad_norm": 4.349104404449463,
      "learning_rate": 0.00026370192018264766,
      "loss": 4.3773,
      "step": 6900
    },
    {
      "epoch": 0.7289029535864979,
      "grad_norm": 3.414034843444824,
      "learning_rate": 0.0002617933264418782,
      "loss": 4.393,
      "step": 6910
    },
    {
      "epoch": 0.729957805907173,
      "grad_norm": 3.7386233806610107,
      "learning_rate": 0.00025989020329527057,
      "loss": 4.3893,
      "step": 6920
    },
    {
      "epoch": 0.7310126582278481,
      "grad_norm": 4.024693012237549,
      "learning_rate": 0.0002579925720682487,
      "loss": 4.3858,
      "step": 6930
    },
    {
      "epoch": 0.7320675105485233,
      "grad_norm": 3.7527976036071777,
      "learning_rate": 0.00025610045402469695,
      "loss": 4.3818,
      "step": 6940
    },
    {
      "epoch": 0.7331223628691983,
      "grad_norm": 3.790043354034424,
      "learning_rate": 0.0002542138703667224,
      "loss": 4.3861,
      "step": 6950
    },
    {
      "epoch": 0.7341772151898734,
      "grad_norm": 3.7102062702178955,
      "learning_rate": 0.0002523328422344158,
      "loss": 4.3724,
      "step": 6960
    },
    {
      "epoch": 0.7352320675105485,
      "grad_norm": 3.894064426422119,
      "learning_rate": 0.0002504573907056159,
      "loss": 4.3728,
      "step": 6970
    },
    {
      "epoch": 0.7362869198312236,
      "grad_norm": 3.662856340408325,
      "learning_rate": 0.0002485875367956729,
      "loss": 4.3741,
      "step": 6980
    },
    {
      "epoch": 0.7373417721518988,
      "grad_norm": 3.7241017818450928,
      "learning_rate": 0.0002467233014572127,
      "loss": 4.3703,
      "step": 6990
    },
    {
      "epoch": 0.7383966244725738,
      "grad_norm": 3.7095248699188232,
      "learning_rate": 0.00024486470557990247,
      "loss": 4.3842,
      "step": 7000
    },
    {
      "epoch": 0.739451476793249,
      "grad_norm": 3.674132823944092,
      "learning_rate": 0.00024301176999021702,
      "loss": 4.3703,
      "step": 7010
    },
    {
      "epoch": 0.740506329113924,
      "grad_norm": 4.808440685272217,
      "learning_rate": 0.0002411645154512041,
      "loss": 4.3855,
      "step": 7020
    },
    {
      "epoch": 0.7415611814345991,
      "grad_norm": 4.470676898956299,
      "learning_rate": 0.0002393229626622528,
      "loss": 4.4104,
      "step": 7030
    },
    {
      "epoch": 0.7426160337552743,
      "grad_norm": 4.692394256591797,
      "learning_rate": 0.00023748713225886137,
      "loss": 4.4069,
      "step": 7040
    },
    {
      "epoch": 0.7436708860759493,
      "grad_norm": 4.287346839904785,
      "learning_rate": 0.0002356570448124058,
      "loss": 4.3971,
      "step": 7050
    },
    {
      "epoch": 0.7447257383966245,
      "grad_norm": 4.723076820373535,
      "learning_rate": 0.00023383272082990963,
      "loss": 4.403,
      "step": 7060
    },
    {
      "epoch": 0.7457805907172996,
      "grad_norm": 4.151793956756592,
      "learning_rate": 0.00023201418075381364,
      "loss": 4.4055,
      "step": 7070
    },
    {
      "epoch": 0.7468354430379747,
      "grad_norm": 4.408592224121094,
      "learning_rate": 0.00023020144496174781,
      "loss": 4.4025,
      "step": 7080
    },
    {
      "epoch": 0.7478902953586498,
      "grad_norm": 4.173978328704834,
      "learning_rate": 0.00022839453376630149,
      "loss": 4.4019,
      "step": 7090
    },
    {
      "epoch": 0.7489451476793249,
      "grad_norm": 3.501197338104248,
      "learning_rate": 0.00022659346741479708,
      "loss": 4.3909,
      "step": 7100
    },
    {
      "epoch": 0.75,
      "grad_norm": 3.5134923458099365,
      "learning_rate": 0.0002247982660890623,
      "loss": 4.3881,
      "step": 7110
    },
    {
      "epoch": 0.7510548523206751,
      "grad_norm": 3.7013754844665527,
      "learning_rate": 0.00022300894990520478,
      "loss": 4.3876,
      "step": 7120
    },
    {
      "epoch": 0.7521097046413502,
      "grad_norm": 3.92083477973938,
      "learning_rate": 0.00022122553891338586,
      "loss": 4.3796,
      "step": 7130
    },
    {
      "epoch": 0.7531645569620253,
      "grad_norm": 3.8984177112579346,
      "learning_rate": 0.00021944805309759643,
      "loss": 4.3724,
      "step": 7140
    },
    {
      "epoch": 0.7542194092827004,
      "grad_norm": 3.743947744369507,
      "learning_rate": 0.0002176765123754334,
      "loss": 4.3716,
      "step": 7150
    },
    {
      "epoch": 0.7552742616033755,
      "grad_norm": 5.435453414916992,
      "learning_rate": 0.00021591093659787528,
      "loss": 4.3866,
      "step": 7160
    },
    {
      "epoch": 0.7563291139240507,
      "grad_norm": 4.548297882080078,
      "learning_rate": 0.0002141513455490609,
      "loss": 4.405,
      "step": 7170
    },
    {
      "epoch": 0.7573839662447257,
      "grad_norm": 6.07659387588501,
      "learning_rate": 0.000212397758946067,
      "loss": 4.4025,
      "step": 7180
    },
    {
      "epoch": 0.7584388185654009,
      "grad_norm": 4.572222709655762,
      "learning_rate": 0.00021065019643868785,
      "loss": 4.4128,
      "step": 7190
    },
    {
      "epoch": 0.759493670886076,
      "grad_norm": 4.403155326843262,
      "learning_rate": 0.0002089086776092146,
      "loss": 4.4171,
      "step": 7200
    },
    {
      "epoch": 0.760548523206751,
      "grad_norm": 4.462301731109619,
      "learning_rate": 0.000207173221972216,
      "loss": 4.4115,
      "step": 7210
    },
    {
      "epoch": 0.7616033755274262,
      "grad_norm": 4.3970513343811035,
      "learning_rate": 0.00020544384897431997,
      "loss": 4.4081,
      "step": 7220
    },
    {
      "epoch": 0.7626582278481012,
      "grad_norm": 4.407256603240967,
      "learning_rate": 0.00020372057799399534,
      "loss": 4.3959,
      "step": 7230
    },
    {
      "epoch": 0.7637130801687764,
      "grad_norm": 5.197177886962891,
      "learning_rate": 0.00020200342834133497,
      "loss": 4.3984,
      "step": 7240
    },
    {
      "epoch": 0.7647679324894515,
      "grad_norm": 4.821730613708496,
      "learning_rate": 0.00020029241925783908,
      "loss": 4.4153,
      "step": 7250
    },
    {
      "epoch": 0.7658227848101266,
      "grad_norm": 4.892157077789307,
      "learning_rate": 0.00019858756991619978,
      "loss": 4.4072,
      "step": 7260
    },
    {
      "epoch": 0.7668776371308017,
      "grad_norm": 4.649753570556641,
      "learning_rate": 0.0001968888994200868,
      "loss": 4.4009,
      "step": 7270
    },
    {
      "epoch": 0.7679324894514767,
      "grad_norm": 4.3715715408325195,
      "learning_rate": 0.0001951964268039322,
      "loss": 4.4158,
      "step": 7280
    },
    {
      "epoch": 0.7689873417721519,
      "grad_norm": 4.261277675628662,
      "learning_rate": 0.00019351017103271805,
      "loss": 4.4077,
      "step": 7290
    },
    {
      "epoch": 0.770042194092827,
      "grad_norm": 3.977024555206299,
      "learning_rate": 0.0001918301510017638,
      "loss": 4.4081,
      "step": 7300
    },
    {
      "epoch": 0.7710970464135021,
      "grad_norm": 4.2188849449157715,
      "learning_rate": 0.0001901563855365141,
      "loss": 4.3877,
      "step": 7310
    },
    {
      "epoch": 0.7721518987341772,
      "grad_norm": 3.770799398422241,
      "learning_rate": 0.00018848889339232833,
      "loss": 4.3952,
      "step": 7320
    },
    {
      "epoch": 0.7732067510548524,
      "grad_norm": 3.986297845840454,
      "learning_rate": 0.00018682769325426986,
      "loss": 4.3874,
      "step": 7330
    },
    {
      "epoch": 0.7742616033755274,
      "grad_norm": 4.629383563995361,
      "learning_rate": 0.00018517280373689789,
      "loss": 4.3939,
      "step": 7340
    },
    {
      "epoch": 0.7753164556962026,
      "grad_norm": 4.652378559112549,
      "learning_rate": 0.0001835242433840573,
      "loss": 4.3803,
      "step": 7350
    },
    {
      "epoch": 0.7763713080168776,
      "grad_norm": 4.738195896148682,
      "learning_rate": 0.00018188203066867178,
      "loss": 4.386,
      "step": 7360
    },
    {
      "epoch": 0.7774261603375527,
      "grad_norm": 4.806765079498291,
      "learning_rate": 0.0001802461839925368,
      "loss": 4.3943,
      "step": 7370
    },
    {
      "epoch": 0.7784810126582279,
      "grad_norm": 4.887210845947266,
      "learning_rate": 0.00017861672168611293,
      "loss": 4.4027,
      "step": 7380
    },
    {
      "epoch": 0.7795358649789029,
      "grad_norm": 4.753091335296631,
      "learning_rate": 0.0001769936620083211,
      "loss": 4.4072,
      "step": 7390
    },
    {
      "epoch": 0.7805907172995781,
      "grad_norm": 4.431006908416748,
      "learning_rate": 0.00017537702314633722,
      "loss": 4.4095,
      "step": 7400
    },
    {
      "epoch": 0.7816455696202531,
      "grad_norm": 4.29086446762085,
      "learning_rate": 0.0001737668232153896,
      "loss": 4.3964,
      "step": 7410
    },
    {
      "epoch": 0.7827004219409283,
      "grad_norm": 4.8123087882995605,
      "learning_rate": 0.0001721630802585545,
      "loss": 4.3985,
      "step": 7420
    },
    {
      "epoch": 0.7837552742616034,
      "grad_norm": 5.079682350158691,
      "learning_rate": 0.00017056581224655473,
      "loss": 4.3937,
      "step": 7430
    },
    {
      "epoch": 0.7848101265822784,
      "grad_norm": 4.984195232391357,
      "learning_rate": 0.0001689750370775584,
      "loss": 4.395,
      "step": 7440
    },
    {
      "epoch": 0.7858649789029536,
      "grad_norm": 4.379343032836914,
      "learning_rate": 0.00016739077257697804,
      "loss": 4.3945,
      "step": 7450
    },
    {
      "epoch": 0.7869198312236287,
      "grad_norm": 4.672163486480713,
      "learning_rate": 0.00016581303649727076,
      "loss": 4.3866,
      "step": 7460
    },
    {
      "epoch": 0.7879746835443038,
      "grad_norm": 4.594658374786377,
      "learning_rate": 0.00016424184651773997,
      "loss": 4.3833,
      "step": 7470
    },
    {
      "epoch": 0.7890295358649789,
      "grad_norm": 4.179958820343018,
      "learning_rate": 0.00016267722024433654,
      "loss": 4.3788,
      "step": 7480
    },
    {
      "epoch": 0.790084388185654,
      "grad_norm": 4.41969633102417,
      "learning_rate": 0.00016111917520946175,
      "loss": 4.3801,
      "step": 7490
    },
    {
      "epoch": 0.7911392405063291,
      "grad_norm": 4.30935525894165,
      "learning_rate": 0.00015956772887177115,
      "loss": 4.3853,
      "step": 7500
    },
    {
      "epoch": 0.7921940928270043,
      "grad_norm": 4.808230400085449,
      "learning_rate": 0.0001580228986159783,
      "loss": 4.3948,
      "step": 7510
    },
    {
      "epoch": 0.7932489451476793,
      "grad_norm": 4.508942127227783,
      "learning_rate": 0.00015648470175266057,
      "loss": 4.3888,
      "step": 7520
    },
    {
      "epoch": 0.7943037974683544,
      "grad_norm": 5.142141342163086,
      "learning_rate": 0.00015495315551806486,
      "loss": 4.3909,
      "step": 7530
    },
    {
      "epoch": 0.7953586497890295,
      "grad_norm": 4.58635950088501,
      "learning_rate": 0.00015342827707391475,
      "loss": 4.39,
      "step": 7540
    },
    {
      "epoch": 0.7964135021097046,
      "grad_norm": 4.537664413452148,
      "learning_rate": 0.00015191008350721772,
      "loss": 4.3869,
      "step": 7550
    },
    {
      "epoch": 0.7974683544303798,
      "grad_norm": 4.170709133148193,
      "learning_rate": 0.000150398591830074,
      "loss": 4.3884,
      "step": 7560
    },
    {
      "epoch": 0.7985232067510548,
      "grad_norm": 3.8539907932281494,
      "learning_rate": 0.00014889381897948575,
      "loss": 4.3804,
      "step": 7570
    },
    {
      "epoch": 0.79957805907173,
      "grad_norm": 3.9293534755706787,
      "learning_rate": 0.00014739578181716765,
      "loss": 4.3799,
      "step": 7580
    },
    {
      "epoch": 0.8006329113924051,
      "grad_norm": 3.94140362739563,
      "learning_rate": 0.0001459044971293575,
      "loss": 4.3743,
      "step": 7590
    },
    {
      "epoch": 0.8016877637130801,
      "grad_norm": 3.880321741104126,
      "learning_rate": 0.00014441998162662847,
      "loss": 4.3733,
      "step": 7600
    },
    {
      "epoch": 0.8027426160337553,
      "grad_norm": 4.039881229400635,
      "learning_rate": 0.00014294225194370154,
      "loss": 4.378,
      "step": 7610
    },
    {
      "epoch": 0.8037974683544303,
      "grad_norm": 4.3156561851501465,
      "learning_rate": 0.0001414713246392594,
      "loss": 4.3648,
      "step": 7620
    },
    {
      "epoch": 0.8048523206751055,
      "grad_norm": 5.146883487701416,
      "learning_rate": 0.00014000721619576077,
      "loss": 4.379,
      "step": 7630
    },
    {
      "epoch": 0.8059071729957806,
      "grad_norm": 4.7725043296813965,
      "learning_rate": 0.0001385499430192557,
      "loss": 4.3793,
      "step": 7640
    },
    {
      "epoch": 0.8069620253164557,
      "grad_norm": 4.335951805114746,
      "learning_rate": 0.00013709952143920148,
      "loss": 4.385,
      "step": 7650
    },
    {
      "epoch": 0.8080168776371308,
      "grad_norm": 4.610930919647217,
      "learning_rate": 0.00013565596770828025,
      "loss": 4.3866,
      "step": 7660
    },
    {
      "epoch": 0.8090717299578059,
      "grad_norm": 4.222512245178223,
      "learning_rate": 0.0001342192980022166,
      "loss": 4.3834,
      "step": 7670
    },
    {
      "epoch": 0.810126582278481,
      "grad_norm": 4.0573410987854,
      "learning_rate": 0.000132789528419596,
      "loss": 4.3855,
      "step": 7680
    },
    {
      "epoch": 0.8111814345991561,
      "grad_norm": 3.8743696212768555,
      "learning_rate": 0.00013136667498168464,
      "loss": 4.3764,
      "step": 7690
    },
    {
      "epoch": 0.8122362869198312,
      "grad_norm": 4.4134979248046875,
      "learning_rate": 0.0001299507536322502,
      "loss": 4.3675,
      "step": 7700
    },
    {
      "epoch": 0.8132911392405063,
      "grad_norm": 4.273777008056641,
      "learning_rate": 0.0001285417802373827,
      "loss": 4.3726,
      "step": 7710
    },
    {
      "epoch": 0.8143459915611815,
      "grad_norm": 4.689339637756348,
      "learning_rate": 0.00012713977058531685,
      "loss": 4.3683,
      "step": 7720
    },
    {
      "epoch": 0.8154008438818565,
      "grad_norm": 4.132012844085693,
      "learning_rate": 0.0001257447403862557,
      "loss": 4.3757,
      "step": 7730
    },
    {
      "epoch": 0.8164556962025317,
      "grad_norm": 4.815252780914307,
      "learning_rate": 0.0001243567052721937,
      "loss": 4.3696,
      "step": 7740
    },
    {
      "epoch": 0.8175105485232067,
      "grad_norm": 4.820760250091553,
      "learning_rate": 0.000122975680796742,
      "loss": 4.3695,
      "step": 7750
    },
    {
      "epoch": 0.8185654008438819,
      "grad_norm": 4.230940818786621,
      "learning_rate": 0.0001216016824349542,
      "loss": 4.39,
      "step": 7760
    },
    {
      "epoch": 0.819620253164557,
      "grad_norm": 4.352594375610352,
      "learning_rate": 0.0001202347255831529,
      "loss": 4.3925,
      "step": 7770
    },
    {
      "epoch": 0.820675105485232,
      "grad_norm": 4.459583282470703,
      "learning_rate": 0.00011887482555875695,
      "loss": 4.3852,
      "step": 7780
    },
    {
      "epoch": 0.8217299578059072,
      "grad_norm": 4.590174198150635,
      "learning_rate": 0.00011752199760011017,
      "loss": 4.4029,
      "step": 7790
    },
    {
      "epoch": 0.8227848101265823,
      "grad_norm": 4.487617015838623,
      "learning_rate": 0.00011617625686631056,
      "loss": 4.386,
      "step": 7800
    },
    {
      "epoch": 0.8238396624472574,
      "grad_norm": 3.5979554653167725,
      "learning_rate": 0.0001148376184370401,
      "loss": 4.3906,
      "step": 7810
    },
    {
      "epoch": 0.8248945147679325,
      "grad_norm": 4.116230487823486,
      "learning_rate": 0.00011350609731239597,
      "loss": 4.3833,
      "step": 7820
    },
    {
      "epoch": 0.8259493670886076,
      "grad_norm": 4.587367534637451,
      "learning_rate": 0.00011218170841272254,
      "loss": 4.3826,
      "step": 7830
    },
    {
      "epoch": 0.8270042194092827,
      "grad_norm": 4.124564170837402,
      "learning_rate": 0.00011086446657844412,
      "loss": 4.3691,
      "step": 7840
    },
    {
      "epoch": 0.8280590717299579,
      "grad_norm": 4.274283409118652,
      "learning_rate": 0.00010955438656989849,
      "loss": 4.368,
      "step": 7850
    },
    {
      "epoch": 0.8291139240506329,
      "grad_norm": 4.521188259124756,
      "learning_rate": 0.00010825148306717222,
      "loss": 4.3689,
      "step": 7860
    },
    {
      "epoch": 0.830168776371308,
      "grad_norm": 4.270509719848633,
      "learning_rate": 0.00010695577066993495,
      "loss": 4.371,
      "step": 7870
    },
    {
      "epoch": 0.8312236286919831,
      "grad_norm": 4.4330878257751465,
      "learning_rate": 0.00010566726389727693,
      "loss": 4.3696,
      "step": 7880
    },
    {
      "epoch": 0.8322784810126582,
      "grad_norm": 4.582380771636963,
      "learning_rate": 0.00010438597718754561,
      "loss": 4.3776,
      "step": 7890
    },
    {
      "epoch": 0.8333333333333334,
      "grad_norm": 4.395266532897949,
      "learning_rate": 0.00010311192489818421,
      "loss": 4.3657,
      "step": 7900
    },
    {
      "epoch": 0.8343881856540084,
      "grad_norm": 4.5557050704956055,
      "learning_rate": 0.00010184512130557074,
      "loss": 4.362,
      "step": 7910
    },
    {
      "epoch": 0.8354430379746836,
      "grad_norm": 3.9632482528686523,
      "learning_rate": 0.0001005855806048581,
      "loss": 4.3698,
      "step": 7920
    },
    {
      "epoch": 0.8364978902953587,
      "grad_norm": 4.014522075653076,
      "learning_rate": 9.933331690981473e-05,
      "loss": 4.3691,
      "step": 7930
    },
    {
      "epoch": 0.8375527426160337,
      "grad_norm": 4.5319695472717285,
      "learning_rate": 9.80883442526668e-05,
      "loss": 4.3666,
      "step": 7940
    },
    {
      "epoch": 0.8386075949367089,
      "grad_norm": 4.800873279571533,
      "learning_rate": 9.685067658394095e-05,
      "loss": 4.3616,
      "step": 7950
    },
    {
      "epoch": 0.8396624472573839,
      "grad_norm": 4.71228551864624,
      "learning_rate": 9.562032777230772e-05,
      "loss": 4.3658,
      "step": 7960
    },
    {
      "epoch": 0.8407172995780591,
      "grad_norm": 4.438753128051758,
      "learning_rate": 9.439731160442619e-05,
      "loss": 4.3759,
      "step": 7970
    },
    {
      "epoch": 0.8417721518987342,
      "grad_norm": 4.477622985839844,
      "learning_rate": 9.31816417847898e-05,
      "loss": 4.3685,
      "step": 7980
    },
    {
      "epoch": 0.8428270042194093,
      "grad_norm": 4.731598377227783,
      "learning_rate": 9.197333193557237e-05,
      "loss": 4.3805,
      "step": 7990
    },
    {
      "epoch": 0.8438818565400844,
      "grad_norm": 5.390109062194824,
      "learning_rate": 9.077239559647591e-05,
      "loss": 4.3806,
      "step": 8000
    },
    {
      "epoch": 0.8449367088607594,
      "grad_norm": 4.585193634033203,
      "learning_rate": 8.957884622457854e-05,
      "loss": 4.3886,
      "step": 8010
    },
    {
      "epoch": 0.8459915611814346,
      "grad_norm": 4.763319969177246,
      "learning_rate": 8.839269719418361e-05,
      "loss": 4.3859,
      "step": 8020
    },
    {
      "epoch": 0.8470464135021097,
      "grad_norm": 5.112292766571045,
      "learning_rate": 8.721396179667019e-05,
      "loss": 4.378,
      "step": 8030
    },
    {
      "epoch": 0.8481012658227848,
      "grad_norm": 4.915692329406738,
      "learning_rate": 8.604265324034405e-05,
      "loss": 4.3883,
      "step": 8040
    },
    {
      "epoch": 0.8491561181434599,
      "grad_norm": 4.355308532714844,
      "learning_rate": 8.48787846502893e-05,
      "loss": 4.3828,
      "step": 8050
    },
    {
      "epoch": 0.8502109704641351,
      "grad_norm": 4.728878021240234,
      "learning_rate": 8.372236906822217e-05,
      "loss": 4.3801,
      "step": 8060
    },
    {
      "epoch": 0.8512658227848101,
      "grad_norm": 4.365956783294678,
      "learning_rate": 8.257341945234365e-05,
      "loss": 4.3862,
      "step": 8070
    },
    {
      "epoch": 0.8523206751054853,
      "grad_norm": 4.66141939163208,
      "learning_rate": 8.143194867719534e-05,
      "loss": 4.3852,
      "step": 8080
    },
    {
      "epoch": 0.8533755274261603,
      "grad_norm": 4.6612982749938965,
      "learning_rate": 8.029796953351445e-05,
      "loss": 4.388,
      "step": 8090
    },
    {
      "epoch": 0.8544303797468354,
      "grad_norm": 4.586324214935303,
      "learning_rate": 7.917149472809113e-05,
      "loss": 4.3843,
      "step": 8100
    },
    {
      "epoch": 0.8554852320675106,
      "grad_norm": 4.743220806121826,
      "learning_rate": 7.805253688362557e-05,
      "loss": 4.3925,
      "step": 8110
    },
    {
      "epoch": 0.8565400843881856,
      "grad_norm": 5.358142375946045,
      "learning_rate": 7.694110853858671e-05,
      "loss": 4.3898,
      "step": 8120
    },
    {
      "epoch": 0.8575949367088608,
      "grad_norm": 4.709871768951416,
      "learning_rate": 7.583722214707206e-05,
      "loss": 4.397,
      "step": 8130
    },
    {
      "epoch": 0.8586497890295358,
      "grad_norm": 4.837324142456055,
      "learning_rate": 7.474089007866756e-05,
      "loss": 4.3966,
      "step": 8140
    },
    {
      "epoch": 0.859704641350211,
      "grad_norm": 5.2710957527160645,
      "learning_rate": 7.365212461830933e-05,
      "loss": 4.3974,
      "step": 8150
    },
    {
      "epoch": 0.8607594936708861,
      "grad_norm": 6.114295959472656,
      "learning_rate": 7.257093796614597e-05,
      "loss": 4.3986,
      "step": 8160
    },
    {
      "epoch": 0.8618143459915611,
      "grad_norm": 4.955920219421387,
      "learning_rate": 7.149734223740187e-05,
      "loss": 4.3961,
      "step": 8170
    },
    {
      "epoch": 0.8628691983122363,
      "grad_norm": 4.632691860198975,
      "learning_rate": 7.043134946224123e-05,
      "loss": 4.4014,
      "step": 8180
    },
    {
      "epoch": 0.8639240506329114,
      "grad_norm": 5.305182933807373,
      "learning_rate": 6.937297158563389e-05,
      "loss": 4.4075,
      "step": 8190
    },
    {
      "epoch": 0.8649789029535865,
      "grad_norm": 5.309530735015869,
      "learning_rate": 6.832222046722069e-05,
      "loss": 4.4059,
      "step": 8200
    },
    {
      "epoch": 0.8660337552742616,
      "grad_norm": 5.890382289886475,
      "learning_rate": 6.72791078811808e-05,
      "loss": 4.4044,
      "step": 8210
    },
    {
      "epoch": 0.8670886075949367,
      "grad_norm": 3.940697431564331,
      "learning_rate": 6.62436455161003e-05,
      "loss": 4.4062,
      "step": 8220
    },
    {
      "epoch": 0.8681434599156118,
      "grad_norm": 4.872069835662842,
      "learning_rate": 6.521584497484043e-05,
      "loss": 4.4114,
      "step": 8230
    },
    {
      "epoch": 0.869198312236287,
      "grad_norm": 4.983556747436523,
      "learning_rate": 6.419571777440814e-05,
      "loss": 4.4073,
      "step": 8240
    },
    {
      "epoch": 0.870253164556962,
      "grad_norm": 4.884251117706299,
      "learning_rate": 6.318327534582688e-05,
      "loss": 4.4159,
      "step": 8250
    },
    {
      "epoch": 0.8713080168776371,
      "grad_norm": 5.57700252532959,
      "learning_rate": 6.217852903400841e-05,
      "loss": 4.4143,
      "step": 8260
    },
    {
      "epoch": 0.8723628691983122,
      "grad_norm": 5.079037666320801,
      "learning_rate": 6.118149009762574e-05,
      "loss": 4.4095,
      "step": 8270
    },
    {
      "epoch": 0.8734177215189873,
      "grad_norm": 6.428347587585449,
      "learning_rate": 6.0192169708987026e-05,
      "loss": 4.4103,
      "step": 8280
    },
    {
      "epoch": 0.8744725738396625,
      "grad_norm": 5.536078453063965,
      "learning_rate": 5.921057895391027e-05,
      "loss": 4.408,
      "step": 8290
    },
    {
      "epoch": 0.8755274261603375,
      "grad_norm": 5.461312294006348,
      "learning_rate": 5.823672883159911e-05,
      "loss": 4.4214,
      "step": 8300
    },
    {
      "epoch": 0.8765822784810127,
      "grad_norm": 4.465426445007324,
      "learning_rate": 5.727063025451973e-05,
      "loss": 4.411,
      "step": 8310
    },
    {
      "epoch": 0.8776371308016878,
      "grad_norm": 5.200120449066162,
      "learning_rate": 5.631229404827845e-05,
      "loss": 4.4162,
      "step": 8320
    },
    {
      "epoch": 0.8786919831223629,
      "grad_norm": 5.456932544708252,
      "learning_rate": 5.536173095150043e-05,
      "loss": 4.4167,
      "step": 8330
    },
    {
      "epoch": 0.879746835443038,
      "grad_norm": 5.080636978149414,
      "learning_rate": 5.441895161570934e-05,
      "loss": 4.4091,
      "step": 8340
    },
    {
      "epoch": 0.880801687763713,
      "grad_norm": 5.389331340789795,
      "learning_rate": 5.348396660520785e-05,
      "loss": 4.4121,
      "step": 8350
    },
    {
      "epoch": 0.8818565400843882,
      "grad_norm": 5.144793510437012,
      "learning_rate": 5.25567863969596e-05,
      "loss": 4.4099,
      "step": 8360
    },
    {
      "epoch": 0.8829113924050633,
      "grad_norm": 5.905550003051758,
      "learning_rate": 5.1637421380471586e-05,
      "loss": 4.4185,
      "step": 8370
    },
    {
      "epoch": 0.8839662447257384,
      "grad_norm": 5.687742710113525,
      "learning_rate": 5.072588185767763e-05,
      "loss": 4.4257,
      "step": 8380
    },
    {
      "epoch": 0.8850210970464135,
      "grad_norm": 5.415053844451904,
      "learning_rate": 4.982217804282332e-05,
      "loss": 4.4203,
      "step": 8390
    },
    {
      "epoch": 0.8860759493670886,
      "grad_norm": 5.728246688842773,
      "learning_rate": 4.89263200623512e-05,
      "loss": 4.4184,
      "step": 8400
    },
    {
      "epoch": 0.8871308016877637,
      "grad_norm": 5.92189359664917,
      "learning_rate": 4.803831795478719e-05,
      "loss": 4.4294,
      "step": 8410
    },
    {
      "epoch": 0.8881856540084389,
      "grad_norm": 5.393908500671387,
      "learning_rate": 4.715818167062863e-05,
      "loss": 4.4278,
      "step": 8420
    },
    {
      "epoch": 0.8892405063291139,
      "grad_norm": 5.276932716369629,
      "learning_rate": 4.628592107223229e-05,
      "loss": 4.4276,
      "step": 8430
    },
    {
      "epoch": 0.890295358649789,
      "grad_norm": 4.774796009063721,
      "learning_rate": 4.542154593370401e-05,
      "loss": 4.4209,
      "step": 8440
    },
    {
      "epoch": 0.8913502109704642,
      "grad_norm": 4.472014427185059,
      "learning_rate": 4.4565065940789515e-05,
      "loss": 4.4284,
      "step": 8450
    },
    {
      "epoch": 0.8924050632911392,
      "grad_norm": 5.397584915161133,
      "learning_rate": 4.3716490690765194e-05,
      "loss": 4.4306,
      "step": 8460
    },
    {
      "epoch": 0.8934599156118144,
      "grad_norm": 4.690983772277832,
      "learning_rate": 4.287582969233103e-05,
      "loss": 4.4296,
      "step": 8470
    },
    {
      "epoch": 0.8945147679324894,
      "grad_norm": 5.7475714683532715,
      "learning_rate": 4.204309236550405e-05,
      "loss": 4.428,
      "step": 8480
    },
    {
      "epoch": 0.8955696202531646,
      "grad_norm": 6.488740921020508,
      "learning_rate": 4.1218288041512534e-05,
      "loss": 4.4294,
      "step": 8490
    },
    {
      "epoch": 0.8966244725738397,
      "grad_norm": 5.0946784019470215,
      "learning_rate": 4.0401425962691804e-05,
      "loss": 4.4247,
      "step": 8500
    },
    {
      "epoch": 0.8976793248945147,
      "grad_norm": 5.348075866699219,
      "learning_rate": 3.95925152823802e-05,
      "loss": 4.4288,
      "step": 8510
    },
    {
      "epoch": 0.8987341772151899,
      "grad_norm": 5.505839824676514,
      "learning_rate": 3.879156506481699e-05,
      "loss": 4.4327,
      "step": 8520
    },
    {
      "epoch": 0.8997890295358649,
      "grad_norm": 5.539851665496826,
      "learning_rate": 3.79985842850403e-05,
      "loss": 4.4318,
      "step": 8530
    },
    {
      "epoch": 0.9008438818565401,
      "grad_norm": 4.971736907958984,
      "learning_rate": 3.72135818287872e-05,
      "loss": 4.4376,
      "step": 8540
    },
    {
      "epoch": 0.9018987341772152,
      "grad_norm": 5.057995319366455,
      "learning_rate": 3.643656649239327e-05,
      "loss": 4.4293,
      "step": 8550
    },
    {
      "epoch": 0.9029535864978903,
      "grad_norm": 4.96615743637085,
      "learning_rate": 3.566754698269492e-05,
      "loss": 4.4194,
      "step": 8560
    },
    {
      "epoch": 0.9040084388185654,
      "grad_norm": 5.738513469696045,
      "learning_rate": 3.498227297979198e-05,
      "loss": 4.4248,
      "step": 8570
    },
    {
      "epoch": 0.9050632911392406,
      "grad_norm": 6.041141986846924,
      "learning_rate": 3.422846920717893e-05,
      "loss": 4.4239,
      "step": 8580
    },
    {
      "epoch": 0.9061181434599156,
      "grad_norm": 4.8311567306518555,
      "learning_rate": 3.3482686004069755e-05,
      "loss": 4.422,
      "step": 8590
    },
    {
      "epoch": 0.9071729957805907,
      "grad_norm": 4.712303161621094,
      "learning_rate": 3.274493172732926e-05,
      "loss": 4.4269,
      "step": 8600
    },
    {
      "epoch": 0.9082278481012658,
      "grad_norm": 4.703649997711182,
      "learning_rate": 3.201521464385443e-05,
      "loss": 4.4261,
      "step": 8610
    },
    {
      "epoch": 0.9092827004219409,
      "grad_norm": 5.938491344451904,
      "learning_rate": 3.129354293048148e-05,
      "loss": 4.4227,
      "step": 8620
    },
    {
      "epoch": 0.9103375527426161,
      "grad_norm": 5.341588497161865,
      "learning_rate": 3.057992467389431e-05,
      "loss": 4.4311,
      "step": 8630
    },
    {
      "epoch": 0.9113924050632911,
      "grad_norm": 4.706961154937744,
      "learning_rate": 2.9874367870534018e-05,
      "loss": 4.4195,
      "step": 8640
    },
    {
      "epoch": 0.9124472573839663,
      "grad_norm": 6.839292049407959,
      "learning_rate": 2.9176880426508957e-05,
      "loss": 4.4275,
      "step": 8650
    },
    {
      "epoch": 0.9135021097046413,
      "grad_norm": 5.095970153808594,
      "learning_rate": 2.8487470157506633e-05,
      "loss": 4.4212,
      "step": 8660
    },
    {
      "epoch": 0.9145569620253164,
      "grad_norm": 6.843786239624023,
      "learning_rate": 2.7806144788705718e-05,
      "loss": 4.4259,
      "step": 8670
    },
    {
      "epoch": 0.9156118143459916,
      "grad_norm": 5.412052631378174,
      "learning_rate": 2.7132911954689672e-05,
      "loss": 4.4357,
      "step": 8680
    },
    {
      "epoch": 0.9166666666666666,
      "grad_norm": 4.609084606170654,
      "learning_rate": 2.6467779199361e-05,
      "loss": 4.4354,
      "step": 8690
    },
    {
      "epoch": 0.9177215189873418,
      "grad_norm": 2.8259389400482178,
      "learning_rate": 2.5810753975857136e-05,
      "loss": 4.433,
      "step": 8700
    },
    {
      "epoch": 0.9187763713080169,
      "grad_norm": 5.063839435577393,
      "learning_rate": 2.516184364646637e-05,
      "loss": 4.4264,
      "step": 8710
    },
    {
      "epoch": 0.919831223628692,
      "grad_norm": 4.950676441192627,
      "learning_rate": 2.4521055482546046e-05,
      "loss": 4.4245,
      "step": 8720
    },
    {
      "epoch": 0.9208860759493671,
      "grad_norm": 4.4326982498168945,
      "learning_rate": 2.3888396664440232e-05,
      "loss": 4.4235,
      "step": 8730
    },
    {
      "epoch": 0.9219409282700421,
      "grad_norm": 3.8274829387664795,
      "learning_rate": 2.3263874281400034e-05,
      "loss": 4.411,
      "step": 8740
    },
    {
      "epoch": 0.9229957805907173,
      "grad_norm": 4.54547119140625,
      "learning_rate": 2.2647495331503565e-05,
      "loss": 4.4105,
      "step": 8750
    },
    {
      "epoch": 0.9240506329113924,
      "grad_norm": 4.614604949951172,
      "learning_rate": 2.203926672157802e-05,
      "loss": 4.3998,
      "step": 8760
    },
    {
      "epoch": 0.9251054852320675,
      "grad_norm": 5.983366012573242,
      "learning_rate": 2.1439195267121902e-05,
      "loss": 4.4114,
      "step": 8770
    },
    {
      "epoch": 0.9261603375527426,
      "grad_norm": 5.845203876495361,
      "learning_rate": 2.0847287692228905e-05,
      "loss": 4.4106,
      "step": 8780
    },
    {
      "epoch": 0.9272151898734177,
      "grad_norm": 4.725266933441162,
      "learning_rate": 2.0263550629512406e-05,
      "loss": 4.4053,
      "step": 8790
    },
    {
      "epoch": 0.9282700421940928,
      "grad_norm": 4.612358093261719,
      "learning_rate": 1.9687990620031266e-05,
      "loss": 4.3961,
      "step": 8800
    },
    {
      "epoch": 0.929324894514768,
      "grad_norm": 3.2170205116271973,
      "learning_rate": 1.912061411321639e-05,
      "loss": 4.4024,
      "step": 8810
    },
    {
      "epoch": 0.930379746835443,
      "grad_norm": 5.258325099945068,
      "learning_rate": 1.856142746679862e-05,
      "loss": 4.3941,
      "step": 8820
    },
    {
      "epoch": 0.9314345991561181,
      "grad_norm": 5.477872371673584,
      "learning_rate": 1.8010436946737292e-05,
      "loss": 4.3911,
      "step": 8830
    },
    {
      "epoch": 0.9324894514767933,
      "grad_norm": 3.811497688293457,
      "learning_rate": 1.7467648727150202e-05,
      "loss": 4.382,
      "step": 8840
    },
    {
      "epoch": 0.9335443037974683,
      "grad_norm": 3.602501153945923,
      "learning_rate": 1.6933068890244595e-05,
      "loss": 4.3833,
      "step": 8850
    },
    {
      "epoch": 0.9345991561181435,
      "grad_norm": 2.851118564605713,
      "learning_rate": 1.6406703426248366e-05,
      "loss": 4.3824,
      "step": 8860
    },
    {
      "epoch": 0.9356540084388185,
      "grad_norm": 5.403396129608154,
      "learning_rate": 1.588855823334362e-05,
      "loss": 4.3845,
      "step": 8870
    },
    {
      "epoch": 0.9367088607594937,
      "grad_norm": 3.470393419265747,
      "learning_rate": 1.5378639117600234e-05,
      "loss": 4.3746,
      "step": 8880
    },
    {
      "epoch": 0.9377637130801688,
      "grad_norm": 3.050715446472168,
      "learning_rate": 1.4876951792910987e-05,
      "loss": 4.3727,
      "step": 8890
    },
    {
      "epoch": 0.9388185654008439,
      "grad_norm": 5.984134674072266,
      "learning_rate": 1.4383501880927103e-05,
      "loss": 4.3606,
      "step": 8900
    },
    {
      "epoch": 0.939873417721519,
      "grad_norm": 3.905161142349243,
      "learning_rate": 1.3898294910995979e-05,
      "loss": 4.3594,
      "step": 8910
    },
    {
      "epoch": 0.9409282700421941,
      "grad_norm": 3.751277446746826,
      "learning_rate": 1.3421336320098565e-05,
      "loss": 4.3548,
      "step": 8920
    },
    {
      "epoch": 0.9419831223628692,
      "grad_norm": 3.874727249145508,
      "learning_rate": 1.2952631452788826e-05,
      "loss": 4.3561,
      "step": 8930
    },
    {
      "epoch": 0.9430379746835443,
      "grad_norm": 3.0299079418182373,
      "learning_rate": 1.2492185561133545e-05,
      "loss": 4.353,
      "step": 8940
    },
    {
      "epoch": 0.9440928270042194,
      "grad_norm": 5.066340923309326,
      "learning_rate": 1.2040003804653864e-05,
      "loss": 4.3466,
      "step": 8950
    },
    {
      "epoch": 0.9451476793248945,
      "grad_norm": 3.3527212142944336,
      "learning_rate": 1.1596091250267171e-05,
      "loss": 4.3476,
      "step": 8960
    },
    {
      "epoch": 0.9462025316455697,
      "grad_norm": 3.7896416187286377,
      "learning_rate": 1.1160452872230303e-05,
      "loss": 4.3495,
      "step": 8970
    },
    {
      "epoch": 0.9472573839662447,
      "grad_norm": 4.118512153625488,
      "learning_rate": 1.0733093552084016e-05,
      "loss": 4.3427,
      "step": 8980
    },
    {
      "epoch": 0.9483122362869199,
      "grad_norm": 3.0552775859832764,
      "learning_rate": 1.0314018078598275e-05,
      "loss": 4.3364,
      "step": 8990
    },
    {
      "epoch": 0.9493670886075949,
      "grad_norm": 3.220763921737671,
      "learning_rate": 9.903231147718294e-06,
      "loss": 4.3329,
      "step": 9000
    },
    {
      "epoch": 0.95042194092827,
      "grad_norm": 4.1169114112854,
      "learning_rate": 9.500737362512168e-06,
      "loss": 4.3293,
      "step": 9010
    },
    {
      "epoch": 0.9514767932489452,
      "grad_norm": 4.713864326477051,
      "learning_rate": 9.106541233119409e-06,
      "loss": 4.3245,
      "step": 9020
    },
    {
      "epoch": 0.9525316455696202,
      "grad_norm": 2.800969123840332,
      "learning_rate": 8.72064717670007e-06,
      "loss": 4.3337,
      "step": 9030
    },
    {
      "epoch": 0.9535864978902954,
      "grad_norm": 3.336726665496826,
      "learning_rate": 8.343059517385454e-06,
      "loss": 4.3194,
      "step": 9040
    },
    {
      "epoch": 0.9546413502109705,
      "grad_norm": 4.9808197021484375,
      "learning_rate": 7.973782486229737e-06,
      "loss": 4.3226,
      "step": 9050
    },
    {
      "epoch": 0.9556962025316456,
      "grad_norm": 2.651531457901001,
      "learning_rate": 7.612820221162331e-06,
      "loss": 4.3271,
      "step": 9060
    },
    {
      "epoch": 0.9567510548523207,
      "grad_norm": 2.764997959136963,
      "learning_rate": 7.260176766941601e-06,
      "loss": 4.3214,
      "step": 9070
    },
    {
      "epoch": 0.9578059071729957,
      "grad_norm": 2.8649086952209473,
      "learning_rate": 6.915856075109722e-06,
      "loss": 4.312,
      "step": 9080
    },
    {
      "epoch": 0.9588607594936709,
      "grad_norm": 3.470050573348999,
      "learning_rate": 6.57986200394814e-06,
      "loss": 4.3127,
      "step": 9090
    },
    {
      "epoch": 0.959915611814346,
      "grad_norm": 3.361154556274414,
      "learning_rate": 6.252198318434432e-06,
      "loss": 4.3,
      "step": 9100
    },
    {
      "epoch": 0.9609704641350211,
      "grad_norm": 3.4893271923065186,
      "learning_rate": 5.932868690200266e-06,
      "loss": 4.301,
      "step": 9110
    },
    {
      "epoch": 0.9620253164556962,
      "grad_norm": 2.939012050628662,
      "learning_rate": 5.6218766974900915e-06,
      "loss": 4.2976,
      "step": 9120
    },
    {
      "epoch": 0.9630801687763713,
      "grad_norm": 2.9008474349975586,
      "learning_rate": 5.319225825120927e-06,
      "loss": 4.2939,
      "step": 9130
    },
    {
      "epoch": 0.9641350210970464,
      "grad_norm": 3.1818408966064453,
      "learning_rate": 5.024919464443723e-06,
      "loss": 4.3053,
      "step": 9140
    },
    {
      "epoch": 0.9651898734177216,
      "grad_norm": 2.980567455291748,
      "learning_rate": 4.738960913304724e-06,
      "loss": 4.2963,
      "step": 9150
    },
    {
      "epoch": 0.9662447257383966,
      "grad_norm": 2.5399811267852783,
      "learning_rate": 4.4613533760093365e-06,
      "loss": 4.2908,
      "step": 9160
    },
    {
      "epoch": 0.9672995780590717,
      "grad_norm": 2.5070652961730957,
      "learning_rate": 4.192099963285484e-06,
      "loss": 4.2949,
      "step": 9170
    },
    {
      "epoch": 0.9683544303797469,
      "grad_norm": 2.995500087738037,
      "learning_rate": 3.931203692249141e-06,
      "loss": 4.2937,
      "step": 9180
    },
    {
      "epoch": 0.9694092827004219,
      "grad_norm": 2.7085208892822266,
      "learning_rate": 3.6786674863704406e-06,
      "loss": 4.2907,
      "step": 9190
    },
    {
      "epoch": 0.9704641350210971,
      "grad_norm": 2.682438373565674,
      "learning_rate": 3.4344941754408663e-06,
      "loss": 4.282,
      "step": 9200
    },
    {
      "epoch": 0.9715189873417721,
      "grad_norm": 2.8788564205169678,
      "learning_rate": 3.198686495541531e-06,
      "loss": 4.2877,
      "step": 9210
    },
    {
      "epoch": 0.9725738396624473,
      "grad_norm": 3.1367478370666504,
      "learning_rate": 2.9712470890126962e-06,
      "loss": 4.281,
      "step": 9220
    },
    {
      "epoch": 0.9736286919831224,
      "grad_norm": 2.5653536319732666,
      "learning_rate": 2.752178504423969e-06,
      "loss": 4.2778,
      "step": 9230
    },
    {
      "epoch": 0.9746835443037974,
      "grad_norm": 2.9827747344970703,
      "learning_rate": 2.541483196545735e-06,
      "loss": 4.2774,
      "step": 9240
    },
    {
      "epoch": 0.9757383966244726,
      "grad_norm": 2.6680901050567627,
      "learning_rate": 2.3391635263218526e-06,
      "loss": 4.2771,
      "step": 9250
    },
    {
      "epoch": 0.9767932489451476,
      "grad_norm": 2.926652193069458,
      "learning_rate": 2.1452217608430857e-06,
      "loss": 4.2821,
      "step": 9260
    },
    {
      "epoch": 0.9778481012658228,
      "grad_norm": 2.62249493598938,
      "learning_rate": 1.959660073321545e-06,
      "loss": 4.2846,
      "step": 9270
    },
    {
      "epoch": 0.9789029535864979,
      "grad_norm": 2.454524517059326,
      "learning_rate": 1.782480543066456e-06,
      "loss": 4.2772,
      "step": 9280
    },
    {
      "epoch": 0.979957805907173,
      "grad_norm": 2.571977138519287,
      "learning_rate": 1.6136851554611753e-06,
      "loss": 4.2774,
      "step": 9290
    },
    {
      "epoch": 0.9810126582278481,
      "grad_norm": 2.8542158603668213,
      "learning_rate": 1.4532758019402958e-06,
      "loss": 4.274,
      "step": 9300
    },
    {
      "epoch": 0.9820675105485233,
      "grad_norm": 2.400996208190918,
      "learning_rate": 1.3012542799689108e-06,
      "loss": 4.275,
      "step": 9310
    },
    {
      "epoch": 0.9831223628691983,
      "grad_norm": 2.8169894218444824,
      "learning_rate": 1.1576222930225478e-06,
      "loss": 4.2768,
      "step": 9320
    },
    {
      "epoch": 0.9841772151898734,
      "grad_norm": 2.989379405975342,
      "learning_rate": 1.0223814505676832e-06,
      "loss": 4.2709,
      "step": 9330
    },
    {
      "epoch": 0.9852320675105485,
      "grad_norm": 2.76350474357605,
      "learning_rate": 8.955332680440076e-07,
      "loss": 4.2804,
      "step": 9340
    },
    {
      "epoch": 0.9862869198312236,
      "grad_norm": 2.5476675033569336,
      "learning_rate": 7.770791668472721e-07,
      "loss": 4.2807,
      "step": 9350
    },
    {
      "epoch": 0.9873417721518988,
      "grad_norm": 2.7354023456573486,
      "learning_rate": 6.67020474313551e-07,
      "loss": 4.2751,
      "step": 9360
    },
    {
      "epoch": 0.9883966244725738,
      "grad_norm": 3.1976516246795654,
      "learning_rate": 5.653584237043374e-07,
      "loss": 4.2857,
      "step": 9370
    },
    {
      "epoch": 0.989451476793249,
      "grad_norm": 2.8704121112823486,
      "learning_rate": 4.720941541923873e-07,
      "loss": 4.2701,
      "step": 9380
    },
    {
      "epoch": 0.990506329113924,
      "grad_norm": 2.676304340362549,
      "learning_rate": 3.8722871084956313e-07,
      "loss": 4.27,
      "step": 9390
    },
    {
      "epoch": 0.9915611814345991,
      "grad_norm": 3.100907802581787,
      "learning_rate": 3.107630446344267e-07,
      "loss": 4.2794,
      "step": 9400
    },
    {
      "epoch": 0.9926160337552743,
      "grad_norm": 3.127851724624634,
      "learning_rate": 2.426980123821643e-07,
      "loss": 4.256,
      "step": 9410
    },
    {
      "epoch": 0.9936708860759493,
      "grad_norm": 2.606079339981079,
      "learning_rate": 1.8303437679476065e-07,
      "loss": 4.2649,
      "step": 9420
    },
    {
      "epoch": 0.9947257383966245,
      "grad_norm": 2.7296714782714844,
      "learning_rate": 1.3177280643233979e-07,
      "loss": 4.2745,
      "step": 9430
    },
    {
      "epoch": 0.9957805907172996,
      "grad_norm": 2.6256513595581055,
      "learning_rate": 8.891387570575393e-08,
      "loss": 4.2673,
      "step": 9440
    },
    {
      "epoch": 0.9968354430379747,
      "grad_norm": 2.8849363327026367,
      "learning_rate": 5.4458064870338553e-08,
      "loss": 4.2725,
      "step": 9450
    },
    {
      "epoch": 0.9978902953586498,
      "grad_norm": 2.8232624530792236,
      "learning_rate": 2.8405760020250304e-08,
      "loss": 4.2723,
      "step": 9460
    },
    {
      "epoch": 0.9989451476793249,
      "grad_norm": 3.2149593830108643,
      "learning_rate": 1.075725308438691e-08,
      "loss": 4.2736,
      "step": 9470
    },
    {
      "epoch": 1.0,
      "grad_norm": 8.678837776184082,
      "learning_rate": 1.5127418226401623e-09,
      "loss": 4.2769,
      "step": 9480
    }
  ],
  "logging_steps": 10,
  "max_steps": 9480,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 1000,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 1.832308198648013e+16,
  "train_batch_size": 1024,
  "trial_name": null,
  "trial_params": null
}