{
  "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": 1.287057638168335,
      "learning_rate": 0.00015822784810126583,
      "loss": 7.493,
      "step": 10
    },
    {
      "epoch": 0.002109704641350211,
      "grad_norm": 1.1503698825836182,
      "learning_rate": 0.00031645569620253165,
      "loss": 6.8817,
      "step": 20
    },
    {
      "epoch": 0.0031645569620253164,
      "grad_norm": 0.8247020244598389,
      "learning_rate": 0.00047468354430379745,
      "loss": 6.2305,
      "step": 30
    },
    {
      "epoch": 0.004219409282700422,
      "grad_norm": 1.1445114612579346,
      "learning_rate": 0.0006329113924050633,
      "loss": 5.7539,
      "step": 40
    },
    {
      "epoch": 0.005274261603375527,
      "grad_norm": 1.1456326246261597,
      "learning_rate": 0.0007911392405063291,
      "loss": 5.2901,
      "step": 50
    },
    {
      "epoch": 0.006329113924050633,
      "grad_norm": 1.277427315711975,
      "learning_rate": 0.0009493670886075949,
      "loss": 4.7825,
      "step": 60
    },
    {
      "epoch": 0.007383966244725738,
      "grad_norm": 0.8188556432723999,
      "learning_rate": 0.0011075949367088608,
      "loss": 4.3843,
      "step": 70
    },
    {
      "epoch": 0.008438818565400843,
      "grad_norm": 1.1751933097839355,
      "learning_rate": 0.0012658227848101266,
      "loss": 4.1334,
      "step": 80
    },
    {
      "epoch": 0.00949367088607595,
      "grad_norm": 1.001449465751648,
      "learning_rate": 0.0014240506329113926,
      "loss": 3.9229,
      "step": 90
    },
    {
      "epoch": 0.010548523206751054,
      "grad_norm": 1.0194555521011353,
      "learning_rate": 0.0015,
      "loss": 3.7921,
      "step": 100
    },
    {
      "epoch": 0.011603375527426161,
      "grad_norm": 0.8997014760971069,
      "learning_rate": 0.0015,
      "loss": 3.6375,
      "step": 110
    },
    {
      "epoch": 0.012658227848101266,
      "grad_norm": 0.8892844915390015,
      "learning_rate": 0.0015,
      "loss": 3.5209,
      "step": 120
    },
    {
      "epoch": 0.013713080168776372,
      "grad_norm": 0.8641502261161804,
      "learning_rate": 0.0015,
      "loss": 3.4067,
      "step": 130
    },
    {
      "epoch": 0.014767932489451477,
      "grad_norm": 0.6706660985946655,
      "learning_rate": 0.0015,
      "loss": 3.303,
      "step": 140
    },
    {
      "epoch": 0.015822784810126583,
      "grad_norm": 1.0603797435760498,
      "learning_rate": 0.0015,
      "loss": 3.2231,
      "step": 150
    },
    {
      "epoch": 0.016877637130801686,
      "grad_norm": 0.6161502003669739,
      "learning_rate": 0.0015,
      "loss": 3.1518,
      "step": 160
    },
    {
      "epoch": 0.017932489451476793,
      "grad_norm": 0.8447583913803101,
      "learning_rate": 0.0015,
      "loss": 3.077,
      "step": 170
    },
    {
      "epoch": 0.0189873417721519,
      "grad_norm": 0.9560238122940063,
      "learning_rate": 0.0015,
      "loss": 3.032,
      "step": 180
    },
    {
      "epoch": 0.020042194092827006,
      "grad_norm": 0.7279493808746338,
      "learning_rate": 0.0015,
      "loss": 2.9698,
      "step": 190
    },
    {
      "epoch": 0.02109704641350211,
      "grad_norm": 0.7134253978729248,
      "learning_rate": 0.0015,
      "loss": 2.9251,
      "step": 200
    },
    {
      "epoch": 0.022151898734177215,
      "grad_norm": 0.8752930760383606,
      "learning_rate": 0.0015,
      "loss": 2.8832,
      "step": 210
    },
    {
      "epoch": 0.023206751054852322,
      "grad_norm": 0.9525153636932373,
      "learning_rate": 0.0015,
      "loss": 2.8475,
      "step": 220
    },
    {
      "epoch": 0.024261603375527425,
      "grad_norm": 0.7820525765419006,
      "learning_rate": 0.0015,
      "loss": 2.7967,
      "step": 230
    },
    {
      "epoch": 0.02531645569620253,
      "grad_norm": 1.0114189386367798,
      "learning_rate": 0.0015,
      "loss": 2.7558,
      "step": 240
    },
    {
      "epoch": 0.026371308016877638,
      "grad_norm": 0.8659444451332092,
      "learning_rate": 0.0015,
      "loss": 2.7227,
      "step": 250
    },
    {
      "epoch": 0.027426160337552744,
      "grad_norm": 0.7615576982498169,
      "learning_rate": 0.0015,
      "loss": 2.6843,
      "step": 260
    },
    {
      "epoch": 0.028481012658227847,
      "grad_norm": 0.966295599937439,
      "learning_rate": 0.0015,
      "loss": 2.6542,
      "step": 270
    },
    {
      "epoch": 0.029535864978902954,
      "grad_norm": 0.6935604810714722,
      "learning_rate": 0.0015,
      "loss": 2.6189,
      "step": 280
    },
    {
      "epoch": 0.03059071729957806,
      "grad_norm": 0.9959357976913452,
      "learning_rate": 0.0015,
      "loss": 2.5925,
      "step": 290
    },
    {
      "epoch": 0.03164556962025317,
      "grad_norm": 0.9867107272148132,
      "learning_rate": 0.0015,
      "loss": 2.5665,
      "step": 300
    },
    {
      "epoch": 0.03270042194092827,
      "grad_norm": 1.0502711534500122,
      "learning_rate": 0.0015,
      "loss": 2.543,
      "step": 310
    },
    {
      "epoch": 0.03375527426160337,
      "grad_norm": 1.0446499586105347,
      "learning_rate": 0.0015,
      "loss": 2.5109,
      "step": 320
    },
    {
      "epoch": 0.03481012658227848,
      "grad_norm": 1.0539124011993408,
      "learning_rate": 0.0015,
      "loss": 2.5022,
      "step": 330
    },
    {
      "epoch": 0.035864978902953586,
      "grad_norm": 0.9436084628105164,
      "learning_rate": 0.0015,
      "loss": 2.4776,
      "step": 340
    },
    {
      "epoch": 0.03691983122362869,
      "grad_norm": 1.2166537046432495,
      "learning_rate": 0.0015,
      "loss": 2.4373,
      "step": 350
    },
    {
      "epoch": 0.0379746835443038,
      "grad_norm": 0.8346443772315979,
      "learning_rate": 0.0015,
      "loss": 2.4208,
      "step": 360
    },
    {
      "epoch": 0.039029535864978905,
      "grad_norm": 1.111228346824646,
      "learning_rate": 0.0015,
      "loss": 2.4084,
      "step": 370
    },
    {
      "epoch": 0.04008438818565401,
      "grad_norm": 0.9931886196136475,
      "learning_rate": 0.0015,
      "loss": 2.3971,
      "step": 380
    },
    {
      "epoch": 0.04113924050632911,
      "grad_norm": 0.8816179037094116,
      "learning_rate": 0.0015,
      "loss": 2.3766,
      "step": 390
    },
    {
      "epoch": 0.04219409282700422,
      "grad_norm": 1.3231067657470703,
      "learning_rate": 0.0015,
      "loss": 2.3473,
      "step": 400
    },
    {
      "epoch": 0.043248945147679324,
      "grad_norm": 1.3582069873809814,
      "learning_rate": 0.0015,
      "loss": 2.3382,
      "step": 410
    },
    {
      "epoch": 0.04430379746835443,
      "grad_norm": 0.9265342950820923,
      "learning_rate": 0.0015,
      "loss": 2.3211,
      "step": 420
    },
    {
      "epoch": 0.04535864978902954,
      "grad_norm": 0.8295995593070984,
      "learning_rate": 0.0015,
      "loss": 2.2953,
      "step": 430
    },
    {
      "epoch": 0.046413502109704644,
      "grad_norm": 0.7636935710906982,
      "learning_rate": 0.0015,
      "loss": 2.2797,
      "step": 440
    },
    {
      "epoch": 0.04746835443037975,
      "grad_norm": 0.9759564399719238,
      "learning_rate": 0.0015,
      "loss": 2.2836,
      "step": 450
    },
    {
      "epoch": 0.04852320675105485,
      "grad_norm": 0.954789936542511,
      "learning_rate": 0.0015,
      "loss": 2.2589,
      "step": 460
    },
    {
      "epoch": 0.049578059071729956,
      "grad_norm": 0.9172025322914124,
      "learning_rate": 0.0015,
      "loss": 2.2413,
      "step": 470
    },
    {
      "epoch": 0.05063291139240506,
      "grad_norm": 1.5287584066390991,
      "learning_rate": 0.0015,
      "loss": 2.2391,
      "step": 480
    },
    {
      "epoch": 0.05168776371308017,
      "grad_norm": 0.9626224637031555,
      "learning_rate": 0.0015,
      "loss": 2.2156,
      "step": 490
    },
    {
      "epoch": 0.052742616033755275,
      "grad_norm": 1.1066144704818726,
      "learning_rate": 0.0015,
      "loss": 2.204,
      "step": 500
    },
    {
      "epoch": 0.05379746835443038,
      "grad_norm": 1.016810655593872,
      "learning_rate": 0.0015,
      "loss": 2.1937,
      "step": 510
    },
    {
      "epoch": 0.05485232067510549,
      "grad_norm": 1.048801302909851,
      "learning_rate": 0.0015,
      "loss": 2.1789,
      "step": 520
    },
    {
      "epoch": 0.05590717299578059,
      "grad_norm": 0.7893497347831726,
      "learning_rate": 0.0015,
      "loss": 2.1647,
      "step": 530
    },
    {
      "epoch": 0.056962025316455694,
      "grad_norm": 0.7989654541015625,
      "learning_rate": 0.0015,
      "loss": 2.1446,
      "step": 540
    },
    {
      "epoch": 0.0580168776371308,
      "grad_norm": 0.6843968033790588,
      "learning_rate": 0.0015,
      "loss": 2.1471,
      "step": 550
    },
    {
      "epoch": 0.05907172995780591,
      "grad_norm": 0.6927700042724609,
      "learning_rate": 0.0015,
      "loss": 2.1136,
      "step": 560
    },
    {
      "epoch": 0.060126582278481014,
      "grad_norm": 0.7592052221298218,
      "learning_rate": 0.0015,
      "loss": 2.1168,
      "step": 570
    },
    {
      "epoch": 0.06118143459915612,
      "grad_norm": 0.8444345593452454,
      "learning_rate": 0.0015,
      "loss": 2.1171,
      "step": 580
    },
    {
      "epoch": 0.06223628691983123,
      "grad_norm": 0.8080502152442932,
      "learning_rate": 0.0015,
      "loss": 2.0922,
      "step": 590
    },
    {
      "epoch": 0.06329113924050633,
      "grad_norm": 0.9479168057441711,
      "learning_rate": 0.0015,
      "loss": 2.0748,
      "step": 600
    },
    {
      "epoch": 0.06434599156118144,
      "grad_norm": 0.8750613331794739,
      "learning_rate": 0.0015,
      "loss": 2.0789,
      "step": 610
    },
    {
      "epoch": 0.06540084388185655,
      "grad_norm": 0.7039671540260315,
      "learning_rate": 0.0015,
      "loss": 2.0664,
      "step": 620
    },
    {
      "epoch": 0.06645569620253164,
      "grad_norm": 1.0618560314178467,
      "learning_rate": 0.0015,
      "loss": 2.0536,
      "step": 630
    },
    {
      "epoch": 0.06751054852320675,
      "grad_norm": 0.769457221031189,
      "learning_rate": 0.0015,
      "loss": 2.0597,
      "step": 640
    },
    {
      "epoch": 0.06856540084388185,
      "grad_norm": 1.1503156423568726,
      "learning_rate": 0.0015,
      "loss": 2.0516,
      "step": 650
    },
    {
      "epoch": 0.06962025316455696,
      "grad_norm": 0.7246300578117371,
      "learning_rate": 0.0015,
      "loss": 2.0387,
      "step": 660
    },
    {
      "epoch": 0.07067510548523206,
      "grad_norm": 0.9715001583099365,
      "learning_rate": 0.0015,
      "loss": 2.0169,
      "step": 670
    },
    {
      "epoch": 0.07172995780590717,
      "grad_norm": 0.9007167816162109,
      "learning_rate": 0.0015,
      "loss": 2.014,
      "step": 680
    },
    {
      "epoch": 0.07278481012658228,
      "grad_norm": 0.6881843209266663,
      "learning_rate": 0.0015,
      "loss": 2.0238,
      "step": 690
    },
    {
      "epoch": 0.07383966244725738,
      "grad_norm": 0.91755211353302,
      "learning_rate": 0.0015,
      "loss": 2.0044,
      "step": 700
    },
    {
      "epoch": 0.07489451476793249,
      "grad_norm": 0.9991128444671631,
      "learning_rate": 0.0015,
      "loss": 1.9964,
      "step": 710
    },
    {
      "epoch": 0.0759493670886076,
      "grad_norm": 1.063872218132019,
      "learning_rate": 0.0015,
      "loss": 1.9769,
      "step": 720
    },
    {
      "epoch": 0.0770042194092827,
      "grad_norm": 0.8963314890861511,
      "learning_rate": 0.0015,
      "loss": 1.9727,
      "step": 730
    },
    {
      "epoch": 0.07805907172995781,
      "grad_norm": 0.8157698512077332,
      "learning_rate": 0.0015,
      "loss": 1.9798,
      "step": 740
    },
    {
      "epoch": 0.07911392405063292,
      "grad_norm": 0.9056829214096069,
      "learning_rate": 0.0015,
      "loss": 1.9594,
      "step": 750
    },
    {
      "epoch": 0.08016877637130802,
      "grad_norm": 2.2598671913146973,
      "learning_rate": 0.0015,
      "loss": 1.966,
      "step": 760
    },
    {
      "epoch": 0.08122362869198312,
      "grad_norm": 1.7556381225585938,
      "learning_rate": 0.0015,
      "loss": 1.9694,
      "step": 770
    },
    {
      "epoch": 0.08227848101265822,
      "grad_norm": 0.961376965045929,
      "learning_rate": 0.0015,
      "loss": 1.9446,
      "step": 780
    },
    {
      "epoch": 0.08333333333333333,
      "grad_norm": 0.6460594534873962,
      "learning_rate": 0.0015,
      "loss": 1.9423,
      "step": 790
    },
    {
      "epoch": 0.08438818565400844,
      "grad_norm": 0.685418426990509,
      "learning_rate": 0.0015,
      "loss": 1.9244,
      "step": 800
    },
    {
      "epoch": 0.08544303797468354,
      "grad_norm": 0.7907118797302246,
      "learning_rate": 0.0015,
      "loss": 1.9153,
      "step": 810
    },
    {
      "epoch": 0.08649789029535865,
      "grad_norm": 1.173556923866272,
      "learning_rate": 0.0015,
      "loss": 1.9268,
      "step": 820
    },
    {
      "epoch": 0.08755274261603375,
      "grad_norm": 0.8807255029678345,
      "learning_rate": 0.0015,
      "loss": 1.9277,
      "step": 830
    },
    {
      "epoch": 0.08860759493670886,
      "grad_norm": 0.8901069760322571,
      "learning_rate": 0.0015,
      "loss": 1.906,
      "step": 840
    },
    {
      "epoch": 0.08966244725738397,
      "grad_norm": 0.6625086665153503,
      "learning_rate": 0.0015,
      "loss": 1.9081,
      "step": 850
    },
    {
      "epoch": 0.09071729957805907,
      "grad_norm": 0.8131834268569946,
      "learning_rate": 0.0015,
      "loss": 1.9053,
      "step": 860
    },
    {
      "epoch": 0.09177215189873418,
      "grad_norm": 0.7555605173110962,
      "learning_rate": 0.0015,
      "loss": 1.902,
      "step": 870
    },
    {
      "epoch": 0.09282700421940929,
      "grad_norm": 1.0310981273651123,
      "learning_rate": 0.0015,
      "loss": 1.8961,
      "step": 880
    },
    {
      "epoch": 0.0938818565400844,
      "grad_norm": 0.8803991079330444,
      "learning_rate": 0.0015,
      "loss": 1.8881,
      "step": 890
    },
    {
      "epoch": 0.0949367088607595,
      "grad_norm": 1.0702396631240845,
      "learning_rate": 0.0015,
      "loss": 1.8865,
      "step": 900
    },
    {
      "epoch": 0.09599156118143459,
      "grad_norm": 0.9415758848190308,
      "learning_rate": 0.0015,
      "loss": 1.8874,
      "step": 910
    },
    {
      "epoch": 0.0970464135021097,
      "grad_norm": 0.6488352417945862,
      "learning_rate": 0.0015,
      "loss": 1.8798,
      "step": 920
    },
    {
      "epoch": 0.0981012658227848,
      "grad_norm": 0.7208075523376465,
      "learning_rate": 0.0015,
      "loss": 1.8594,
      "step": 930
    },
    {
      "epoch": 0.09915611814345991,
      "grad_norm": 0.7142037749290466,
      "learning_rate": 0.0015,
      "loss": 1.8703,
      "step": 940
    },
    {
      "epoch": 0.10021097046413502,
      "grad_norm": 0.9566423892974854,
      "learning_rate": 0.0015,
      "loss": 1.8532,
      "step": 950
    },
    {
      "epoch": 0.10126582278481013,
      "grad_norm": 0.7550307512283325,
      "learning_rate": 0.0015,
      "loss": 1.8603,
      "step": 960
    },
    {
      "epoch": 0.10232067510548523,
      "grad_norm": 1.5818613767623901,
      "learning_rate": 0.0015,
      "loss": 1.8627,
      "step": 970
    },
    {
      "epoch": 0.10337552742616034,
      "grad_norm": 0.8488968014717102,
      "learning_rate": 0.0015,
      "loss": 1.8433,
      "step": 980
    },
    {
      "epoch": 0.10443037974683544,
      "grad_norm": 0.6491256356239319,
      "learning_rate": 0.0015,
      "loss": 1.8331,
      "step": 990
    },
    {
      "epoch": 0.10548523206751055,
      "grad_norm": 1.0147662162780762,
      "learning_rate": 0.0015,
      "loss": 1.8486,
      "step": 1000
    },
    {
      "epoch": 0.10654008438818566,
      "grad_norm": 0.7335547804832458,
      "learning_rate": 0.0015,
      "loss": 1.8418,
      "step": 1010
    },
    {
      "epoch": 0.10759493670886076,
      "grad_norm": 0.9607897996902466,
      "learning_rate": 0.0015,
      "loss": 1.8293,
      "step": 1020
    },
    {
      "epoch": 0.10864978902953587,
      "grad_norm": 0.8564672470092773,
      "learning_rate": 0.0015,
      "loss": 1.8439,
      "step": 1030
    },
    {
      "epoch": 0.10970464135021098,
      "grad_norm": 0.733195424079895,
      "learning_rate": 0.0015,
      "loss": 1.8199,
      "step": 1040
    },
    {
      "epoch": 0.11075949367088607,
      "grad_norm": 1.1174650192260742,
      "learning_rate": 0.0015,
      "loss": 1.8218,
      "step": 1050
    },
    {
      "epoch": 0.11181434599156118,
      "grad_norm": 0.6878591179847717,
      "learning_rate": 0.0015,
      "loss": 1.8116,
      "step": 1060
    },
    {
      "epoch": 0.11286919831223628,
      "grad_norm": 0.6646955013275146,
      "learning_rate": 0.0015,
      "loss": 1.8037,
      "step": 1070
    },
    {
      "epoch": 0.11392405063291139,
      "grad_norm": 0.9651567339897156,
      "learning_rate": 0.0015,
      "loss": 1.8073,
      "step": 1080
    },
    {
      "epoch": 0.1149789029535865,
      "grad_norm": 0.684890627861023,
      "learning_rate": 0.0015,
      "loss": 1.816,
      "step": 1090
    },
    {
      "epoch": 0.1160337552742616,
      "grad_norm": 0.7184986472129822,
      "learning_rate": 0.0015,
      "loss": 1.8012,
      "step": 1100
    },
    {
      "epoch": 0.11708860759493671,
      "grad_norm": 0.7160841822624207,
      "learning_rate": 0.0015,
      "loss": 1.8065,
      "step": 1110
    },
    {
      "epoch": 0.11814345991561181,
      "grad_norm": 0.6141655445098877,
      "learning_rate": 0.0015,
      "loss": 1.8002,
      "step": 1120
    },
    {
      "epoch": 0.11919831223628692,
      "grad_norm": 1.2699943780899048,
      "learning_rate": 0.0015,
      "loss": 1.7895,
      "step": 1130
    },
    {
      "epoch": 0.12025316455696203,
      "grad_norm": 0.6769657731056213,
      "learning_rate": 0.0015,
      "loss": 1.8034,
      "step": 1140
    },
    {
      "epoch": 0.12130801687763713,
      "grad_norm": 1.2477600574493408,
      "learning_rate": 0.0015,
      "loss": 1.7845,
      "step": 1150
    },
    {
      "epoch": 0.12236286919831224,
      "grad_norm": 0.64894038438797,
      "learning_rate": 0.0015,
      "loss": 1.7927,
      "step": 1160
    },
    {
      "epoch": 0.12341772151898735,
      "grad_norm": 1.0948996543884277,
      "learning_rate": 0.0015,
      "loss": 1.7806,
      "step": 1170
    },
    {
      "epoch": 0.12447257383966245,
      "grad_norm": 0.7002706527709961,
      "learning_rate": 0.0015,
      "loss": 1.7644,
      "step": 1180
    },
    {
      "epoch": 0.12552742616033755,
      "grad_norm": 0.7476921081542969,
      "learning_rate": 0.0015,
      "loss": 1.7703,
      "step": 1190
    },
    {
      "epoch": 0.12658227848101267,
      "grad_norm": 0.6751740574836731,
      "learning_rate": 0.0015,
      "loss": 1.781,
      "step": 1200
    },
    {
      "epoch": 0.12763713080168776,
      "grad_norm": 0.6483523845672607,
      "learning_rate": 0.0015,
      "loss": 1.7759,
      "step": 1210
    },
    {
      "epoch": 0.12869198312236288,
      "grad_norm": 0.6249744892120361,
      "learning_rate": 0.0015,
      "loss": 1.76,
      "step": 1220
    },
    {
      "epoch": 0.12974683544303797,
      "grad_norm": 0.947671115398407,
      "learning_rate": 0.0015,
      "loss": 1.749,
      "step": 1230
    },
    {
      "epoch": 0.1308016877637131,
      "grad_norm": 0.8034379482269287,
      "learning_rate": 0.0015,
      "loss": 1.77,
      "step": 1240
    },
    {
      "epoch": 0.13185654008438819,
      "grad_norm": 0.7275480628013611,
      "learning_rate": 0.0015,
      "loss": 1.7711,
      "step": 1250
    },
    {
      "epoch": 0.13291139240506328,
      "grad_norm": 0.6855311989784241,
      "learning_rate": 0.0015,
      "loss": 1.7571,
      "step": 1260
    },
    {
      "epoch": 0.1339662447257384,
      "grad_norm": 0.6835574507713318,
      "learning_rate": 0.0015,
      "loss": 1.7524,
      "step": 1270
    },
    {
      "epoch": 0.1350210970464135,
      "grad_norm": 0.7019007802009583,
      "learning_rate": 0.0015,
      "loss": 1.7617,
      "step": 1280
    },
    {
      "epoch": 0.1360759493670886,
      "grad_norm": 0.7281892895698547,
      "learning_rate": 0.0015,
      "loss": 1.7492,
      "step": 1290
    },
    {
      "epoch": 0.1371308016877637,
      "grad_norm": 0.7096613645553589,
      "learning_rate": 0.0015,
      "loss": 1.736,
      "step": 1300
    },
    {
      "epoch": 0.13818565400843882,
      "grad_norm": 0.6526144742965698,
      "learning_rate": 0.0015,
      "loss": 1.7432,
      "step": 1310
    },
    {
      "epoch": 0.13924050632911392,
      "grad_norm": 1.017628788948059,
      "learning_rate": 0.0015,
      "loss": 1.749,
      "step": 1320
    },
    {
      "epoch": 0.14029535864978904,
      "grad_norm": 0.7124719619750977,
      "learning_rate": 0.0015,
      "loss": 1.735,
      "step": 1330
    },
    {
      "epoch": 0.14135021097046413,
      "grad_norm": 0.6463326215744019,
      "learning_rate": 0.0015,
      "loss": 1.7382,
      "step": 1340
    },
    {
      "epoch": 0.14240506329113925,
      "grad_norm": 0.6644458770751953,
      "learning_rate": 0.0015,
      "loss": 1.7433,
      "step": 1350
    },
    {
      "epoch": 0.14345991561181434,
      "grad_norm": 0.6406483054161072,
      "learning_rate": 0.0015,
      "loss": 1.7316,
      "step": 1360
    },
    {
      "epoch": 0.14451476793248946,
      "grad_norm": 0.7143809199333191,
      "learning_rate": 0.0015,
      "loss": 1.729,
      "step": 1370
    },
    {
      "epoch": 0.14556962025316456,
      "grad_norm": 0.7517651319503784,
      "learning_rate": 0.0015,
      "loss": 1.7217,
      "step": 1380
    },
    {
      "epoch": 0.14662447257383968,
      "grad_norm": 0.6045621037483215,
      "learning_rate": 0.0015,
      "loss": 1.7301,
      "step": 1390
    },
    {
      "epoch": 0.14767932489451477,
      "grad_norm": 0.7317776679992676,
      "learning_rate": 0.0015,
      "loss": 1.7126,
      "step": 1400
    },
    {
      "epoch": 0.14873417721518986,
      "grad_norm": 0.6872985363006592,
      "learning_rate": 0.0015,
      "loss": 1.7179,
      "step": 1410
    },
    {
      "epoch": 0.14978902953586498,
      "grad_norm": 1.0588369369506836,
      "learning_rate": 0.0015,
      "loss": 1.726,
      "step": 1420
    },
    {
      "epoch": 0.15084388185654007,
      "grad_norm": 0.721889317035675,
      "learning_rate": 0.0015,
      "loss": 1.7155,
      "step": 1430
    },
    {
      "epoch": 0.1518987341772152,
      "grad_norm": 0.6461079716682434,
      "learning_rate": 0.0015,
      "loss": 1.71,
      "step": 1440
    },
    {
      "epoch": 0.1529535864978903,
      "grad_norm": 0.6328428983688354,
      "learning_rate": 0.0015,
      "loss": 1.7119,
      "step": 1450
    },
    {
      "epoch": 0.1540084388185654,
      "grad_norm": 0.7252086400985718,
      "learning_rate": 0.0015,
      "loss": 1.7071,
      "step": 1460
    },
    {
      "epoch": 0.1550632911392405,
      "grad_norm": 0.762008786201477,
      "learning_rate": 0.0015,
      "loss": 1.7101,
      "step": 1470
    },
    {
      "epoch": 0.15611814345991562,
      "grad_norm": 0.6981770992279053,
      "learning_rate": 0.0015,
      "loss": 1.7072,
      "step": 1480
    },
    {
      "epoch": 0.1571729957805907,
      "grad_norm": 0.6552959680557251,
      "learning_rate": 0.0015,
      "loss": 1.705,
      "step": 1490
    },
    {
      "epoch": 0.15822784810126583,
      "grad_norm": 0.6281026601791382,
      "learning_rate": 0.0015,
      "loss": 1.7046,
      "step": 1500
    },
    {
      "epoch": 0.15928270042194093,
      "grad_norm": 0.7670705914497375,
      "learning_rate": 0.0015,
      "loss": 1.7056,
      "step": 1510
    },
    {
      "epoch": 0.16033755274261605,
      "grad_norm": 0.6418710350990295,
      "learning_rate": 0.0015,
      "loss": 1.708,
      "step": 1520
    },
    {
      "epoch": 0.16139240506329114,
      "grad_norm": 0.6629166603088379,
      "learning_rate": 0.0015,
      "loss": 1.6885,
      "step": 1530
    },
    {
      "epoch": 0.16244725738396623,
      "grad_norm": 0.6065171360969543,
      "learning_rate": 0.0015,
      "loss": 1.6795,
      "step": 1540
    },
    {
      "epoch": 0.16350210970464135,
      "grad_norm": 0.7254469990730286,
      "learning_rate": 0.0015,
      "loss": 1.6805,
      "step": 1550
    },
    {
      "epoch": 0.16455696202531644,
      "grad_norm": 0.7255260944366455,
      "learning_rate": 0.0015,
      "loss": 1.684,
      "step": 1560
    },
    {
      "epoch": 0.16561181434599156,
      "grad_norm": 0.7033355832099915,
      "learning_rate": 0.0015,
      "loss": 1.6746,
      "step": 1570
    },
    {
      "epoch": 0.16666666666666666,
      "grad_norm": 0.717984139919281,
      "learning_rate": 0.0015,
      "loss": 1.6859,
      "step": 1580
    },
    {
      "epoch": 0.16772151898734178,
      "grad_norm": 1.111241102218628,
      "learning_rate": 0.0015,
      "loss": 1.6855,
      "step": 1590
    },
    {
      "epoch": 0.16877637130801687,
      "grad_norm": 0.6471601128578186,
      "learning_rate": 0.0015,
      "loss": 1.6874,
      "step": 1600
    },
    {
      "epoch": 0.169831223628692,
      "grad_norm": 0.7955036163330078,
      "learning_rate": 0.0015,
      "loss": 1.6845,
      "step": 1610
    },
    {
      "epoch": 0.17088607594936708,
      "grad_norm": 0.6496426463127136,
      "learning_rate": 0.0015,
      "loss": 1.6521,
      "step": 1620
    },
    {
      "epoch": 0.1719409282700422,
      "grad_norm": 0.6558811664581299,
      "learning_rate": 0.0015,
      "loss": 1.6693,
      "step": 1630
    },
    {
      "epoch": 0.1729957805907173,
      "grad_norm": 0.6988083720207214,
      "learning_rate": 0.0015,
      "loss": 1.6729,
      "step": 1640
    },
    {
      "epoch": 0.17405063291139242,
      "grad_norm": 1.0070551633834839,
      "learning_rate": 0.0015,
      "loss": 1.6632,
      "step": 1650
    },
    {
      "epoch": 0.1751054852320675,
      "grad_norm": 0.6134887337684631,
      "learning_rate": 0.0015,
      "loss": 1.6628,
      "step": 1660
    },
    {
      "epoch": 0.17616033755274263,
      "grad_norm": 0.9757758378982544,
      "learning_rate": 0.0015,
      "loss": 1.6783,
      "step": 1670
    },
    {
      "epoch": 0.17721518987341772,
      "grad_norm": 1.0612666606903076,
      "learning_rate": 0.0015,
      "loss": 1.6654,
      "step": 1680
    },
    {
      "epoch": 0.17827004219409281,
      "grad_norm": 0.880418062210083,
      "learning_rate": 0.0015,
      "loss": 1.6585,
      "step": 1690
    },
    {
      "epoch": 0.17932489451476794,
      "grad_norm": 0.795171856880188,
      "learning_rate": 0.0015,
      "loss": 1.669,
      "step": 1700
    },
    {
      "epoch": 0.18037974683544303,
      "grad_norm": 0.7361051440238953,
      "learning_rate": 0.0015,
      "loss": 1.6457,
      "step": 1710
    },
    {
      "epoch": 0.18143459915611815,
      "grad_norm": 0.6480020880699158,
      "learning_rate": 0.0015,
      "loss": 1.6483,
      "step": 1720
    },
    {
      "epoch": 0.18248945147679324,
      "grad_norm": 0.6146023869514465,
      "learning_rate": 0.0015,
      "loss": 1.6529,
      "step": 1730
    },
    {
      "epoch": 0.18354430379746836,
      "grad_norm": 0.8712873458862305,
      "learning_rate": 0.0015,
      "loss": 1.6534,
      "step": 1740
    },
    {
      "epoch": 0.18459915611814345,
      "grad_norm": 0.6795451045036316,
      "learning_rate": 0.0015,
      "loss": 1.677,
      "step": 1750
    },
    {
      "epoch": 0.18565400843881857,
      "grad_norm": 0.6434656381607056,
      "learning_rate": 0.0015,
      "loss": 1.6534,
      "step": 1760
    },
    {
      "epoch": 0.18670886075949367,
      "grad_norm": 0.5913978815078735,
      "learning_rate": 0.0015,
      "loss": 1.6559,
      "step": 1770
    },
    {
      "epoch": 0.1877637130801688,
      "grad_norm": 0.7321423888206482,
      "learning_rate": 0.0015,
      "loss": 1.655,
      "step": 1780
    },
    {
      "epoch": 0.18881856540084388,
      "grad_norm": 0.8003367781639099,
      "learning_rate": 0.0015,
      "loss": 1.6598,
      "step": 1790
    },
    {
      "epoch": 0.189873417721519,
      "grad_norm": 0.6629058718681335,
      "learning_rate": 0.0015,
      "loss": 1.6509,
      "step": 1800
    },
    {
      "epoch": 0.1909282700421941,
      "grad_norm": 0.7159475088119507,
      "learning_rate": 0.0015,
      "loss": 1.6404,
      "step": 1810
    },
    {
      "epoch": 0.19198312236286919,
      "grad_norm": 0.7505967020988464,
      "learning_rate": 0.0015,
      "loss": 1.6398,
      "step": 1820
    },
    {
      "epoch": 0.1930379746835443,
      "grad_norm": 0.7928625345230103,
      "learning_rate": 0.0015,
      "loss": 1.6411,
      "step": 1830
    },
    {
      "epoch": 0.1940928270042194,
      "grad_norm": 0.6108897924423218,
      "learning_rate": 0.0015,
      "loss": 1.6375,
      "step": 1840
    },
    {
      "epoch": 0.19514767932489452,
      "grad_norm": 0.5757226943969727,
      "learning_rate": 0.0015,
      "loss": 1.641,
      "step": 1850
    },
    {
      "epoch": 0.1962025316455696,
      "grad_norm": 0.6521726250648499,
      "learning_rate": 0.0015,
      "loss": 1.6436,
      "step": 1860
    },
    {
      "epoch": 0.19725738396624473,
      "grad_norm": 0.6217014193534851,
      "learning_rate": 0.0015,
      "loss": 1.6433,
      "step": 1870
    },
    {
      "epoch": 0.19831223628691982,
      "grad_norm": 0.5712935328483582,
      "learning_rate": 0.0015,
      "loss": 1.6323,
      "step": 1880
    },
    {
      "epoch": 0.19936708860759494,
      "grad_norm": 0.5738839507102966,
      "learning_rate": 0.0015,
      "loss": 1.6389,
      "step": 1890
    },
    {
      "epoch": 0.20042194092827004,
      "grad_norm": 1.0032706260681152,
      "learning_rate": 0.0015,
      "loss": 1.6353,
      "step": 1900
    },
    {
      "epoch": 0.20147679324894516,
      "grad_norm": 0.9717868566513062,
      "learning_rate": 0.0015,
      "loss": 1.6325,
      "step": 1910
    },
    {
      "epoch": 0.20253164556962025,
      "grad_norm": 0.6785035729408264,
      "learning_rate": 0.0015,
      "loss": 1.63,
      "step": 1920
    },
    {
      "epoch": 0.20358649789029537,
      "grad_norm": 0.9020541906356812,
      "learning_rate": 0.0015,
      "loss": 1.6334,
      "step": 1930
    },
    {
      "epoch": 0.20464135021097046,
      "grad_norm": 0.6292390823364258,
      "learning_rate": 0.0015,
      "loss": 1.6229,
      "step": 1940
    },
    {
      "epoch": 0.20569620253164558,
      "grad_norm": 0.5943929553031921,
      "learning_rate": 0.0015,
      "loss": 1.6215,
      "step": 1950
    },
    {
      "epoch": 0.20675105485232068,
      "grad_norm": 0.8829261064529419,
      "learning_rate": 0.0015,
      "loss": 1.6241,
      "step": 1960
    },
    {
      "epoch": 0.20780590717299577,
      "grad_norm": 0.588376522064209,
      "learning_rate": 0.0015,
      "loss": 1.6362,
      "step": 1970
    },
    {
      "epoch": 0.2088607594936709,
      "grad_norm": 0.6464346647262573,
      "learning_rate": 0.0015,
      "loss": 1.6171,
      "step": 1980
    },
    {
      "epoch": 0.20991561181434598,
      "grad_norm": 0.9764966368675232,
      "learning_rate": 0.0015,
      "loss": 1.6159,
      "step": 1990
    },
    {
      "epoch": 0.2109704641350211,
      "grad_norm": 0.6633862853050232,
      "learning_rate": 0.0015,
      "loss": 1.6163,
      "step": 2000
    },
    {
      "epoch": 0.2120253164556962,
      "grad_norm": 0.5747328996658325,
      "learning_rate": 0.0015,
      "loss": 1.6201,
      "step": 2010
    },
    {
      "epoch": 0.21308016877637131,
      "grad_norm": 0.72202467918396,
      "learning_rate": 0.0015,
      "loss": 1.6308,
      "step": 2020
    },
    {
      "epoch": 0.2141350210970464,
      "grad_norm": 0.6208357810974121,
      "learning_rate": 0.0015,
      "loss": 1.6182,
      "step": 2030
    },
    {
      "epoch": 0.21518987341772153,
      "grad_norm": 0.5996553301811218,
      "learning_rate": 0.0015,
      "loss": 1.616,
      "step": 2040
    },
    {
      "epoch": 0.21624472573839662,
      "grad_norm": 0.6539575457572937,
      "learning_rate": 0.0015,
      "loss": 1.6152,
      "step": 2050
    },
    {
      "epoch": 0.21729957805907174,
      "grad_norm": 0.6510769724845886,
      "learning_rate": 0.0015,
      "loss": 1.6067,
      "step": 2060
    },
    {
      "epoch": 0.21835443037974683,
      "grad_norm": 0.9695384502410889,
      "learning_rate": 0.0015,
      "loss": 1.6117,
      "step": 2070
    },
    {
      "epoch": 0.21940928270042195,
      "grad_norm": 0.7847108840942383,
      "learning_rate": 0.0015,
      "loss": 1.6173,
      "step": 2080
    },
    {
      "epoch": 0.22046413502109705,
      "grad_norm": 0.9148688316345215,
      "learning_rate": 0.0015,
      "loss": 1.6027,
      "step": 2090
    },
    {
      "epoch": 0.22151898734177214,
      "grad_norm": 1.5051568746566772,
      "learning_rate": 0.0015,
      "loss": 1.6142,
      "step": 2100
    },
    {
      "epoch": 0.22257383966244726,
      "grad_norm": 1.2215869426727295,
      "learning_rate": 0.0015,
      "loss": 1.6157,
      "step": 2110
    },
    {
      "epoch": 0.22362869198312235,
      "grad_norm": 0.6897009611129761,
      "learning_rate": 0.0015,
      "loss": 1.5938,
      "step": 2120
    },
    {
      "epoch": 0.22468354430379747,
      "grad_norm": 0.6504973769187927,
      "learning_rate": 0.0015,
      "loss": 1.5992,
      "step": 2130
    },
    {
      "epoch": 0.22573839662447256,
      "grad_norm": 0.6026249527931213,
      "learning_rate": 0.0015,
      "loss": 1.5935,
      "step": 2140
    },
    {
      "epoch": 0.22679324894514769,
      "grad_norm": 0.8656939268112183,
      "learning_rate": 0.0015,
      "loss": 1.6027,
      "step": 2150
    },
    {
      "epoch": 0.22784810126582278,
      "grad_norm": 0.7233303189277649,
      "learning_rate": 0.0015,
      "loss": 1.6188,
      "step": 2160
    },
    {
      "epoch": 0.2289029535864979,
      "grad_norm": 0.5548332929611206,
      "learning_rate": 0.0015,
      "loss": 1.6011,
      "step": 2170
    },
    {
      "epoch": 0.229957805907173,
      "grad_norm": 0.5613821148872375,
      "learning_rate": 0.0015,
      "loss": 1.5798,
      "step": 2180
    },
    {
      "epoch": 0.2310126582278481,
      "grad_norm": 0.6372296810150146,
      "learning_rate": 0.0015,
      "loss": 1.6007,
      "step": 2190
    },
    {
      "epoch": 0.2320675105485232,
      "grad_norm": 0.6220502853393555,
      "learning_rate": 0.0015,
      "loss": 1.6027,
      "step": 2200
    },
    {
      "epoch": 0.23312236286919832,
      "grad_norm": 0.6468200087547302,
      "learning_rate": 0.0015,
      "loss": 1.586,
      "step": 2210
    },
    {
      "epoch": 0.23417721518987342,
      "grad_norm": 0.5483685731887817,
      "learning_rate": 0.0015,
      "loss": 1.5976,
      "step": 2220
    },
    {
      "epoch": 0.23523206751054854,
      "grad_norm": 0.6158987283706665,
      "learning_rate": 0.0015,
      "loss": 1.5956,
      "step": 2230
    },
    {
      "epoch": 0.23628691983122363,
      "grad_norm": 0.7173906564712524,
      "learning_rate": 0.0015,
      "loss": 1.591,
      "step": 2240
    },
    {
      "epoch": 0.23734177215189872,
      "grad_norm": 0.8330214023590088,
      "learning_rate": 0.0015,
      "loss": 1.5947,
      "step": 2250
    },
    {
      "epoch": 0.23839662447257384,
      "grad_norm": 0.9982806444168091,
      "learning_rate": 0.0015,
      "loss": 1.5824,
      "step": 2260
    },
    {
      "epoch": 0.23945147679324894,
      "grad_norm": 0.7386844754219055,
      "learning_rate": 0.0015,
      "loss": 1.6135,
      "step": 2270
    },
    {
      "epoch": 0.24050632911392406,
      "grad_norm": 0.7794486284255981,
      "learning_rate": 0.0015,
      "loss": 1.5971,
      "step": 2280
    },
    {
      "epoch": 0.24156118143459915,
      "grad_norm": 0.6920127868652344,
      "learning_rate": 0.0015,
      "loss": 1.5859,
      "step": 2290
    },
    {
      "epoch": 0.24261603375527427,
      "grad_norm": 0.6448893547058105,
      "learning_rate": 0.0015,
      "loss": 1.5838,
      "step": 2300
    },
    {
      "epoch": 0.24367088607594936,
      "grad_norm": 0.6755775809288025,
      "learning_rate": 0.0015,
      "loss": 1.5821,
      "step": 2310
    },
    {
      "epoch": 0.24472573839662448,
      "grad_norm": 0.6389298439025879,
      "learning_rate": 0.0015,
      "loss": 1.594,
      "step": 2320
    },
    {
      "epoch": 0.24578059071729957,
      "grad_norm": 0.7176726460456848,
      "learning_rate": 0.0015,
      "loss": 1.5784,
      "step": 2330
    },
    {
      "epoch": 0.2468354430379747,
      "grad_norm": 0.900411069393158,
      "learning_rate": 0.0015,
      "loss": 1.5744,
      "step": 2340
    },
    {
      "epoch": 0.2478902953586498,
      "grad_norm": 0.6456978917121887,
      "learning_rate": 0.0015,
      "loss": 1.5759,
      "step": 2350
    },
    {
      "epoch": 0.2489451476793249,
      "grad_norm": 0.6599090695381165,
      "learning_rate": 0.0015,
      "loss": 1.5817,
      "step": 2360
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8039332032203674,
      "learning_rate": 0.0015,
      "loss": 1.5822,
      "step": 2370
    },
    {
      "epoch": 0.2510548523206751,
      "grad_norm": 0.5929813385009766,
      "learning_rate": 0.0015,
      "loss": 1.5788,
      "step": 2380
    },
    {
      "epoch": 0.2521097046413502,
      "grad_norm": 0.5901683568954468,
      "learning_rate": 0.0015,
      "loss": 1.5762,
      "step": 2390
    },
    {
      "epoch": 0.25316455696202533,
      "grad_norm": 0.5641725659370422,
      "learning_rate": 0.0015,
      "loss": 1.5796,
      "step": 2400
    },
    {
      "epoch": 0.2542194092827004,
      "grad_norm": 0.8158353567123413,
      "learning_rate": 0.0015,
      "loss": 1.5798,
      "step": 2410
    },
    {
      "epoch": 0.2552742616033755,
      "grad_norm": 0.6062980890274048,
      "learning_rate": 0.0015,
      "loss": 1.5776,
      "step": 2420
    },
    {
      "epoch": 0.2563291139240506,
      "grad_norm": 0.6236847639083862,
      "learning_rate": 0.0015,
      "loss": 1.5721,
      "step": 2430
    },
    {
      "epoch": 0.25738396624472576,
      "grad_norm": 0.8233560919761658,
      "learning_rate": 0.0015,
      "loss": 1.5822,
      "step": 2440
    },
    {
      "epoch": 0.25843881856540085,
      "grad_norm": 0.5583575367927551,
      "learning_rate": 0.0015,
      "loss": 1.5671,
      "step": 2450
    },
    {
      "epoch": 0.25949367088607594,
      "grad_norm": 0.624183177947998,
      "learning_rate": 0.0015,
      "loss": 1.5848,
      "step": 2460
    },
    {
      "epoch": 0.26054852320675104,
      "grad_norm": 0.9696111679077148,
      "learning_rate": 0.0015,
      "loss": 1.5736,
      "step": 2470
    },
    {
      "epoch": 0.2616033755274262,
      "grad_norm": 0.7765077352523804,
      "learning_rate": 0.0015,
      "loss": 1.5799,
      "step": 2480
    },
    {
      "epoch": 0.2626582278481013,
      "grad_norm": 0.5858256816864014,
      "learning_rate": 0.0015,
      "loss": 1.5623,
      "step": 2490
    },
    {
      "epoch": 0.26371308016877637,
      "grad_norm": 0.7113674283027649,
      "learning_rate": 0.0015,
      "loss": 1.5721,
      "step": 2500
    },
    {
      "epoch": 0.26476793248945146,
      "grad_norm": 0.6105078458786011,
      "learning_rate": 0.0015,
      "loss": 1.5703,
      "step": 2510
    },
    {
      "epoch": 0.26582278481012656,
      "grad_norm": 0.5850993990898132,
      "learning_rate": 0.0015,
      "loss": 1.571,
      "step": 2520
    },
    {
      "epoch": 0.2668776371308017,
      "grad_norm": 0.5758876800537109,
      "learning_rate": 0.0015,
      "loss": 1.5656,
      "step": 2530
    },
    {
      "epoch": 0.2679324894514768,
      "grad_norm": 0.937073290348053,
      "learning_rate": 0.0015,
      "loss": 1.5667,
      "step": 2540
    },
    {
      "epoch": 0.2689873417721519,
      "grad_norm": 0.8201426863670349,
      "learning_rate": 0.0015,
      "loss": 1.5633,
      "step": 2550
    },
    {
      "epoch": 0.270042194092827,
      "grad_norm": 0.7046356797218323,
      "learning_rate": 0.0015,
      "loss": 1.5732,
      "step": 2560
    },
    {
      "epoch": 0.27109704641350213,
      "grad_norm": 0.5730412602424622,
      "learning_rate": 0.0015,
      "loss": 1.5643,
      "step": 2570
    },
    {
      "epoch": 0.2721518987341772,
      "grad_norm": 0.6409282684326172,
      "learning_rate": 0.0015,
      "loss": 1.5635,
      "step": 2580
    },
    {
      "epoch": 0.2732067510548523,
      "grad_norm": 0.6453495025634766,
      "learning_rate": 0.0015,
      "loss": 1.5704,
      "step": 2590
    },
    {
      "epoch": 0.2742616033755274,
      "grad_norm": 0.583181619644165,
      "learning_rate": 0.0015,
      "loss": 1.5686,
      "step": 2600
    },
    {
      "epoch": 0.27531645569620256,
      "grad_norm": 0.591073215007782,
      "learning_rate": 0.0015,
      "loss": 1.5657,
      "step": 2610
    },
    {
      "epoch": 0.27637130801687765,
      "grad_norm": 0.7396721243858337,
      "learning_rate": 0.0015,
      "loss": 1.5672,
      "step": 2620
    },
    {
      "epoch": 0.27742616033755274,
      "grad_norm": 0.6038991212844849,
      "learning_rate": 0.0015,
      "loss": 1.5564,
      "step": 2630
    },
    {
      "epoch": 0.27848101265822783,
      "grad_norm": 0.5844075679779053,
      "learning_rate": 0.0015,
      "loss": 1.5652,
      "step": 2640
    },
    {
      "epoch": 0.2795358649789029,
      "grad_norm": 0.608181357383728,
      "learning_rate": 0.0015,
      "loss": 1.5589,
      "step": 2650
    },
    {
      "epoch": 0.2805907172995781,
      "grad_norm": 0.6012187600135803,
      "learning_rate": 0.0015,
      "loss": 1.5528,
      "step": 2660
    },
    {
      "epoch": 0.28164556962025317,
      "grad_norm": 0.5790621638298035,
      "learning_rate": 0.0015,
      "loss": 1.5533,
      "step": 2670
    },
    {
      "epoch": 0.28270042194092826,
      "grad_norm": 0.588222324848175,
      "learning_rate": 0.0015,
      "loss": 1.5533,
      "step": 2680
    },
    {
      "epoch": 0.28375527426160335,
      "grad_norm": 0.6873937249183655,
      "learning_rate": 0.0015,
      "loss": 1.5546,
      "step": 2690
    },
    {
      "epoch": 0.2848101265822785,
      "grad_norm": 0.6786103248596191,
      "learning_rate": 0.0015,
      "loss": 1.5508,
      "step": 2700
    },
    {
      "epoch": 0.2858649789029536,
      "grad_norm": 0.6643319725990295,
      "learning_rate": 0.0015,
      "loss": 1.5603,
      "step": 2710
    },
    {
      "epoch": 0.2869198312236287,
      "grad_norm": 0.6390370726585388,
      "learning_rate": 0.0015,
      "loss": 1.5569,
      "step": 2720
    },
    {
      "epoch": 0.2879746835443038,
      "grad_norm": 0.6126376390457153,
      "learning_rate": 0.0015,
      "loss": 1.5368,
      "step": 2730
    },
    {
      "epoch": 0.2890295358649789,
      "grad_norm": 0.9072438478469849,
      "learning_rate": 0.0015,
      "loss": 1.5517,
      "step": 2740
    },
    {
      "epoch": 0.290084388185654,
      "grad_norm": 0.5998278856277466,
      "learning_rate": 0.0015,
      "loss": 1.5553,
      "step": 2750
    },
    {
      "epoch": 0.2911392405063291,
      "grad_norm": 0.7560575008392334,
      "learning_rate": 0.0015,
      "loss": 1.5626,
      "step": 2760
    },
    {
      "epoch": 0.2921940928270042,
      "grad_norm": 0.6209825277328491,
      "learning_rate": 0.0015,
      "loss": 1.5596,
      "step": 2770
    },
    {
      "epoch": 0.29324894514767935,
      "grad_norm": 0.5998792052268982,
      "learning_rate": 0.0015,
      "loss": 1.5538,
      "step": 2780
    },
    {
      "epoch": 0.29430379746835444,
      "grad_norm": 0.5925680994987488,
      "learning_rate": 0.0015,
      "loss": 1.5511,
      "step": 2790
    },
    {
      "epoch": 0.29535864978902954,
      "grad_norm": 0.6163322925567627,
      "learning_rate": 0.0015,
      "loss": 1.5482,
      "step": 2800
    },
    {
      "epoch": 0.29641350210970463,
      "grad_norm": 0.5704933404922485,
      "learning_rate": 0.0015,
      "loss": 1.5517,
      "step": 2810
    },
    {
      "epoch": 0.2974683544303797,
      "grad_norm": 0.7040290236473083,
      "learning_rate": 0.0015,
      "loss": 1.5578,
      "step": 2820
    },
    {
      "epoch": 0.29852320675105487,
      "grad_norm": 1.134084701538086,
      "learning_rate": 0.0015,
      "loss": 1.5333,
      "step": 2830
    },
    {
      "epoch": 0.29957805907172996,
      "grad_norm": 0.8534347414970398,
      "learning_rate": 0.0015,
      "loss": 1.536,
      "step": 2840
    },
    {
      "epoch": 0.30063291139240506,
      "grad_norm": 0.6510926485061646,
      "learning_rate": 0.0015,
      "loss": 1.5402,
      "step": 2850
    },
    {
      "epoch": 0.30168776371308015,
      "grad_norm": 0.6544208526611328,
      "learning_rate": 0.0015,
      "loss": 1.5485,
      "step": 2860
    },
    {
      "epoch": 0.3027426160337553,
      "grad_norm": 0.8508293628692627,
      "learning_rate": 0.0015,
      "loss": 1.5484,
      "step": 2870
    },
    {
      "epoch": 0.3037974683544304,
      "grad_norm": 0.727138102054596,
      "learning_rate": 0.0015,
      "loss": 1.5327,
      "step": 2880
    },
    {
      "epoch": 0.3048523206751055,
      "grad_norm": 0.6009069085121155,
      "learning_rate": 0.0015,
      "loss": 1.5375,
      "step": 2890
    },
    {
      "epoch": 0.3059071729957806,
      "grad_norm": 0.655423104763031,
      "learning_rate": 0.0015,
      "loss": 1.5445,
      "step": 2900
    },
    {
      "epoch": 0.3069620253164557,
      "grad_norm": 0.6968240141868591,
      "learning_rate": 0.0015,
      "loss": 1.5421,
      "step": 2910
    },
    {
      "epoch": 0.3080168776371308,
      "grad_norm": 0.6441303491592407,
      "learning_rate": 0.0015,
      "loss": 1.538,
      "step": 2920
    },
    {
      "epoch": 0.3090717299578059,
      "grad_norm": 0.5854324102401733,
      "learning_rate": 0.0015,
      "loss": 1.5354,
      "step": 2930
    },
    {
      "epoch": 0.310126582278481,
      "grad_norm": 0.764230489730835,
      "learning_rate": 0.0015,
      "loss": 1.5394,
      "step": 2940
    },
    {
      "epoch": 0.3111814345991561,
      "grad_norm": 0.5712303519248962,
      "learning_rate": 0.0015,
      "loss": 1.5532,
      "step": 2950
    },
    {
      "epoch": 0.31223628691983124,
      "grad_norm": 0.7080296277999878,
      "learning_rate": 0.0015,
      "loss": 1.5318,
      "step": 2960
    },
    {
      "epoch": 0.31329113924050633,
      "grad_norm": 0.8196799159049988,
      "learning_rate": 0.0015,
      "loss": 1.522,
      "step": 2970
    },
    {
      "epoch": 0.3143459915611814,
      "grad_norm": 0.5653908252716064,
      "learning_rate": 0.0015,
      "loss": 1.5286,
      "step": 2980
    },
    {
      "epoch": 0.3154008438818565,
      "grad_norm": 0.5506159067153931,
      "learning_rate": 0.0015,
      "loss": 1.5425,
      "step": 2990
    },
    {
      "epoch": 0.31645569620253167,
      "grad_norm": 0.5746380686759949,
      "learning_rate": 0.0015,
      "loss": 1.5282,
      "step": 3000
    },
    {
      "epoch": 0.31751054852320676,
      "grad_norm": 0.6572897434234619,
      "learning_rate": 0.0015,
      "loss": 1.5325,
      "step": 3010
    },
    {
      "epoch": 0.31856540084388185,
      "grad_norm": 0.8061972856521606,
      "learning_rate": 0.0015,
      "loss": 1.5413,
      "step": 3020
    },
    {
      "epoch": 0.31962025316455694,
      "grad_norm": 0.6429711580276489,
      "learning_rate": 0.0015,
      "loss": 1.5306,
      "step": 3030
    },
    {
      "epoch": 0.3206751054852321,
      "grad_norm": 0.7164819240570068,
      "learning_rate": 0.0015,
      "loss": 1.5265,
      "step": 3040
    },
    {
      "epoch": 0.3217299578059072,
      "grad_norm": 0.6932748556137085,
      "learning_rate": 0.0015,
      "loss": 1.5249,
      "step": 3050
    },
    {
      "epoch": 0.3227848101265823,
      "grad_norm": 0.5477355718612671,
      "learning_rate": 0.0015,
      "loss": 1.537,
      "step": 3060
    },
    {
      "epoch": 0.32383966244725737,
      "grad_norm": 0.7214810848236084,
      "learning_rate": 0.0015,
      "loss": 1.5272,
      "step": 3070
    },
    {
      "epoch": 0.32489451476793246,
      "grad_norm": 0.552919328212738,
      "learning_rate": 0.0015,
      "loss": 1.5181,
      "step": 3080
    },
    {
      "epoch": 0.3259493670886076,
      "grad_norm": 0.5526236295700073,
      "learning_rate": 0.0015,
      "loss": 1.5351,
      "step": 3090
    },
    {
      "epoch": 0.3270042194092827,
      "grad_norm": 0.654093861579895,
      "learning_rate": 0.0015,
      "loss": 1.5216,
      "step": 3100
    },
    {
      "epoch": 0.3280590717299578,
      "grad_norm": 0.5659670233726501,
      "learning_rate": 0.0015,
      "loss": 1.5279,
      "step": 3110
    },
    {
      "epoch": 0.3291139240506329,
      "grad_norm": 0.6855720281600952,
      "learning_rate": 0.0015,
      "loss": 1.5274,
      "step": 3120
    },
    {
      "epoch": 0.33016877637130804,
      "grad_norm": 0.6791499257087708,
      "learning_rate": 0.0015,
      "loss": 1.5403,
      "step": 3130
    },
    {
      "epoch": 0.33122362869198313,
      "grad_norm": 0.5190626382827759,
      "learning_rate": 0.0015,
      "loss": 1.5176,
      "step": 3140
    },
    {
      "epoch": 0.3322784810126582,
      "grad_norm": 0.6249583959579468,
      "learning_rate": 0.0015,
      "loss": 1.5276,
      "step": 3150
    },
    {
      "epoch": 0.3333333333333333,
      "grad_norm": 0.5765541195869446,
      "learning_rate": 0.0015,
      "loss": 1.5267,
      "step": 3160
    },
    {
      "epoch": 0.33438818565400846,
      "grad_norm": 0.5980014801025391,
      "learning_rate": 0.0015,
      "loss": 1.5295,
      "step": 3170
    },
    {
      "epoch": 0.33544303797468356,
      "grad_norm": 0.7137629985809326,
      "learning_rate": 0.0015,
      "loss": 1.5256,
      "step": 3180
    },
    {
      "epoch": 0.33649789029535865,
      "grad_norm": 0.7207750082015991,
      "learning_rate": 0.0015,
      "loss": 1.5235,
      "step": 3190
    },
    {
      "epoch": 0.33755274261603374,
      "grad_norm": 0.6043395400047302,
      "learning_rate": 0.0015,
      "loss": 1.5185,
      "step": 3200
    },
    {
      "epoch": 0.33860759493670883,
      "grad_norm": 0.6252288818359375,
      "learning_rate": 0.0015,
      "loss": 1.5332,
      "step": 3210
    },
    {
      "epoch": 0.339662447257384,
      "grad_norm": 0.671881377696991,
      "learning_rate": 0.0015,
      "loss": 1.5218,
      "step": 3220
    },
    {
      "epoch": 0.3407172995780591,
      "grad_norm": 0.5771369338035583,
      "learning_rate": 0.0015,
      "loss": 1.5155,
      "step": 3230
    },
    {
      "epoch": 0.34177215189873417,
      "grad_norm": 0.5985080003738403,
      "learning_rate": 0.0015,
      "loss": 1.497,
      "step": 3240
    },
    {
      "epoch": 0.34282700421940926,
      "grad_norm": 0.5753167867660522,
      "learning_rate": 0.0015,
      "loss": 1.5201,
      "step": 3250
    },
    {
      "epoch": 0.3438818565400844,
      "grad_norm": 0.6402337551116943,
      "learning_rate": 0.0015,
      "loss": 1.5114,
      "step": 3260
    },
    {
      "epoch": 0.3449367088607595,
      "grad_norm": 0.5531371831893921,
      "learning_rate": 0.0015,
      "loss": 1.5299,
      "step": 3270
    },
    {
      "epoch": 0.3459915611814346,
      "grad_norm": 0.7978616952896118,
      "learning_rate": 0.0015,
      "loss": 1.5292,
      "step": 3280
    },
    {
      "epoch": 0.3470464135021097,
      "grad_norm": 1.0608034133911133,
      "learning_rate": 0.0015,
      "loss": 1.5115,
      "step": 3290
    },
    {
      "epoch": 0.34810126582278483,
      "grad_norm": 0.5867448449134827,
      "learning_rate": 0.0015,
      "loss": 1.5211,
      "step": 3300
    },
    {
      "epoch": 0.3491561181434599,
      "grad_norm": 0.650109052658081,
      "learning_rate": 0.0015,
      "loss": 1.5204,
      "step": 3310
    },
    {
      "epoch": 0.350210970464135,
      "grad_norm": 0.5901459455490112,
      "learning_rate": 0.0015,
      "loss": 1.5275,
      "step": 3320
    },
    {
      "epoch": 0.3512658227848101,
      "grad_norm": 0.564175546169281,
      "learning_rate": 0.0015,
      "loss": 1.5155,
      "step": 3330
    },
    {
      "epoch": 0.35232067510548526,
      "grad_norm": 0.612292468547821,
      "learning_rate": 0.0015,
      "loss": 1.5163,
      "step": 3340
    },
    {
      "epoch": 0.35337552742616035,
      "grad_norm": 0.5642638802528381,
      "learning_rate": 0.0015,
      "loss": 1.5093,
      "step": 3350
    },
    {
      "epoch": 0.35443037974683544,
      "grad_norm": 0.6836768984794617,
      "learning_rate": 0.0015,
      "loss": 1.5109,
      "step": 3360
    },
    {
      "epoch": 0.35548523206751054,
      "grad_norm": 0.5777236819267273,
      "learning_rate": 0.0015,
      "loss": 1.5191,
      "step": 3370
    },
    {
      "epoch": 0.35654008438818563,
      "grad_norm": 0.5549036264419556,
      "learning_rate": 0.0015,
      "loss": 1.5119,
      "step": 3380
    },
    {
      "epoch": 0.3575949367088608,
      "grad_norm": 0.5068067908287048,
      "learning_rate": 0.0015,
      "loss": 1.5091,
      "step": 3390
    },
    {
      "epoch": 0.35864978902953587,
      "grad_norm": 1.0121623277664185,
      "learning_rate": 0.0015,
      "loss": 1.5237,
      "step": 3400
    },
    {
      "epoch": 0.35970464135021096,
      "grad_norm": 0.8498140573501587,
      "learning_rate": 0.0015,
      "loss": 1.5158,
      "step": 3410
    },
    {
      "epoch": 0.36075949367088606,
      "grad_norm": 0.5821650624275208,
      "learning_rate": 0.0015,
      "loss": 1.4927,
      "step": 3420
    },
    {
      "epoch": 0.3618143459915612,
      "grad_norm": 0.5692257285118103,
      "learning_rate": 0.0015,
      "loss": 1.5085,
      "step": 3430
    },
    {
      "epoch": 0.3628691983122363,
      "grad_norm": 0.5395897626876831,
      "learning_rate": 0.0015,
      "loss": 1.5076,
      "step": 3440
    },
    {
      "epoch": 0.3639240506329114,
      "grad_norm": 0.6442540287971497,
      "learning_rate": 0.0015,
      "loss": 1.5048,
      "step": 3450
    },
    {
      "epoch": 0.3649789029535865,
      "grad_norm": 0.757856547832489,
      "learning_rate": 0.0015,
      "loss": 1.4975,
      "step": 3460
    },
    {
      "epoch": 0.36603375527426163,
      "grad_norm": 0.8908534646034241,
      "learning_rate": 0.0015,
      "loss": 1.5149,
      "step": 3470
    },
    {
      "epoch": 0.3670886075949367,
      "grad_norm": 0.7035073041915894,
      "learning_rate": 0.0015,
      "loss": 1.5206,
      "step": 3480
    },
    {
      "epoch": 0.3681434599156118,
      "grad_norm": 0.6767370104789734,
      "learning_rate": 0.0015,
      "loss": 1.5121,
      "step": 3490
    },
    {
      "epoch": 0.3691983122362869,
      "grad_norm": 0.7631784081459045,
      "learning_rate": 0.0015,
      "loss": 1.4937,
      "step": 3500
    },
    {
      "epoch": 0.370253164556962,
      "grad_norm": 0.7743115425109863,
      "learning_rate": 0.0015,
      "loss": 1.5156,
      "step": 3510
    },
    {
      "epoch": 0.37130801687763715,
      "grad_norm": 0.5623247623443604,
      "learning_rate": 0.0015,
      "loss": 1.5102,
      "step": 3520
    },
    {
      "epoch": 0.37236286919831224,
      "grad_norm": 0.5539934635162354,
      "learning_rate": 0.0015,
      "loss": 1.52,
      "step": 3530
    },
    {
      "epoch": 0.37341772151898733,
      "grad_norm": 0.8669719696044922,
      "learning_rate": 0.0015,
      "loss": 1.5121,
      "step": 3540
    },
    {
      "epoch": 0.3744725738396624,
      "grad_norm": 0.7609106302261353,
      "learning_rate": 0.0015,
      "loss": 1.5022,
      "step": 3550
    },
    {
      "epoch": 0.3755274261603376,
      "grad_norm": 0.5223315358161926,
      "learning_rate": 0.0015,
      "loss": 1.5022,
      "step": 3560
    },
    {
      "epoch": 0.37658227848101267,
      "grad_norm": 0.6668152213096619,
      "learning_rate": 0.0015,
      "loss": 1.4971,
      "step": 3570
    },
    {
      "epoch": 0.37763713080168776,
      "grad_norm": 0.6139928698539734,
      "learning_rate": 0.0015,
      "loss": 1.5007,
      "step": 3580
    },
    {
      "epoch": 0.37869198312236285,
      "grad_norm": 0.5609003305435181,
      "learning_rate": 0.0015,
      "loss": 1.5118,
      "step": 3590
    },
    {
      "epoch": 0.379746835443038,
      "grad_norm": 0.6565743088722229,
      "learning_rate": 0.0015,
      "loss": 1.4974,
      "step": 3600
    },
    {
      "epoch": 0.3808016877637131,
      "grad_norm": 0.59303879737854,
      "learning_rate": 0.0015,
      "loss": 1.5003,
      "step": 3610
    },
    {
      "epoch": 0.3818565400843882,
      "grad_norm": 0.5740610361099243,
      "learning_rate": 0.0015,
      "loss": 1.4995,
      "step": 3620
    },
    {
      "epoch": 0.3829113924050633,
      "grad_norm": 0.6683691740036011,
      "learning_rate": 0.0015,
      "loss": 1.5021,
      "step": 3630
    },
    {
      "epoch": 0.38396624472573837,
      "grad_norm": 0.6576066613197327,
      "learning_rate": 0.0015,
      "loss": 1.5002,
      "step": 3640
    },
    {
      "epoch": 0.3850210970464135,
      "grad_norm": 0.6867544651031494,
      "learning_rate": 0.0015,
      "loss": 1.5098,
      "step": 3650
    },
    {
      "epoch": 0.3860759493670886,
      "grad_norm": 0.6396692395210266,
      "learning_rate": 0.0015,
      "loss": 1.5042,
      "step": 3660
    },
    {
      "epoch": 0.3871308016877637,
      "grad_norm": 0.7358971834182739,
      "learning_rate": 0.0015,
      "loss": 1.5005,
      "step": 3670
    },
    {
      "epoch": 0.3881856540084388,
      "grad_norm": 0.7496529221534729,
      "learning_rate": 0.0015,
      "loss": 1.4978,
      "step": 3680
    },
    {
      "epoch": 0.38924050632911394,
      "grad_norm": 0.7263187766075134,
      "learning_rate": 0.0015,
      "loss": 1.5092,
      "step": 3690
    },
    {
      "epoch": 0.39029535864978904,
      "grad_norm": 0.6615262031555176,
      "learning_rate": 0.0015,
      "loss": 1.5096,
      "step": 3700
    },
    {
      "epoch": 0.39135021097046413,
      "grad_norm": 0.5842890739440918,
      "learning_rate": 0.0015,
      "loss": 1.4985,
      "step": 3710
    },
    {
      "epoch": 0.3924050632911392,
      "grad_norm": 0.5979508757591248,
      "learning_rate": 0.0015,
      "loss": 1.4786,
      "step": 3720
    },
    {
      "epoch": 0.39345991561181437,
      "grad_norm": 0.6373541951179504,
      "learning_rate": 0.0015,
      "loss": 1.4962,
      "step": 3730
    },
    {
      "epoch": 0.39451476793248946,
      "grad_norm": 0.5276340246200562,
      "learning_rate": 0.0015,
      "loss": 1.4974,
      "step": 3740
    },
    {
      "epoch": 0.39556962025316456,
      "grad_norm": 0.6625521183013916,
      "learning_rate": 0.0015,
      "loss": 1.5011,
      "step": 3750
    },
    {
      "epoch": 0.39662447257383965,
      "grad_norm": 0.9527451395988464,
      "learning_rate": 0.0015,
      "loss": 1.4941,
      "step": 3760
    },
    {
      "epoch": 0.39767932489451474,
      "grad_norm": 0.577045738697052,
      "learning_rate": 0.0015,
      "loss": 1.5092,
      "step": 3770
    },
    {
      "epoch": 0.3987341772151899,
      "grad_norm": 0.5803396105766296,
      "learning_rate": 0.0015,
      "loss": 1.4919,
      "step": 3780
    },
    {
      "epoch": 0.399789029535865,
      "grad_norm": 0.8232485055923462,
      "learning_rate": 0.0015,
      "loss": 1.4985,
      "step": 3790
    },
    {
      "epoch": 0.4008438818565401,
      "grad_norm": 0.7094146609306335,
      "learning_rate": 0.0015,
      "loss": 1.5032,
      "step": 3800
    },
    {
      "epoch": 0.40189873417721517,
      "grad_norm": 0.5717623233795166,
      "learning_rate": 0.0015,
      "loss": 1.4909,
      "step": 3810
    },
    {
      "epoch": 0.4029535864978903,
      "grad_norm": 0.5873256921768188,
      "learning_rate": 0.0015,
      "loss": 1.4807,
      "step": 3820
    },
    {
      "epoch": 0.4040084388185654,
      "grad_norm": 0.5121378898620605,
      "learning_rate": 0.0015,
      "loss": 1.4857,
      "step": 3830
    },
    {
      "epoch": 0.4050632911392405,
      "grad_norm": 0.6827294826507568,
      "learning_rate": 0.0015,
      "loss": 1.5033,
      "step": 3840
    },
    {
      "epoch": 0.4061181434599156,
      "grad_norm": 0.757856011390686,
      "learning_rate": 0.0015,
      "loss": 1.4858,
      "step": 3850
    },
    {
      "epoch": 0.40717299578059074,
      "grad_norm": 0.7232317924499512,
      "learning_rate": 0.0015,
      "loss": 1.4996,
      "step": 3860
    },
    {
      "epoch": 0.40822784810126583,
      "grad_norm": 0.5782884955406189,
      "learning_rate": 0.0015,
      "loss": 1.4885,
      "step": 3870
    },
    {
      "epoch": 0.4092827004219409,
      "grad_norm": 0.5484322309494019,
      "learning_rate": 0.0015,
      "loss": 1.4819,
      "step": 3880
    },
    {
      "epoch": 0.410337552742616,
      "grad_norm": 0.5790377855300903,
      "learning_rate": 0.0015,
      "loss": 1.4803,
      "step": 3890
    },
    {
      "epoch": 0.41139240506329117,
      "grad_norm": 0.9478213787078857,
      "learning_rate": 0.0015,
      "loss": 1.4817,
      "step": 3900
    },
    {
      "epoch": 0.41244725738396626,
      "grad_norm": 0.6198221445083618,
      "learning_rate": 0.0015,
      "loss": 1.4852,
      "step": 3910
    },
    {
      "epoch": 0.41350210970464135,
      "grad_norm": 0.5510705709457397,
      "learning_rate": 0.0015,
      "loss": 1.4816,
      "step": 3920
    },
    {
      "epoch": 0.41455696202531644,
      "grad_norm": 0.6039211750030518,
      "learning_rate": 0.0015,
      "loss": 1.4861,
      "step": 3930
    },
    {
      "epoch": 0.41561181434599154,
      "grad_norm": 0.6572819352149963,
      "learning_rate": 0.0015,
      "loss": 1.482,
      "step": 3940
    },
    {
      "epoch": 0.4166666666666667,
      "grad_norm": 0.6410369277000427,
      "learning_rate": 0.0015,
      "loss": 1.487,
      "step": 3950
    },
    {
      "epoch": 0.4177215189873418,
      "grad_norm": 0.5450556874275208,
      "learning_rate": 0.0015,
      "loss": 1.4878,
      "step": 3960
    },
    {
      "epoch": 0.41877637130801687,
      "grad_norm": 0.5562349557876587,
      "learning_rate": 0.0015,
      "loss": 1.4784,
      "step": 3970
    },
    {
      "epoch": 0.41983122362869196,
      "grad_norm": 0.5236467719078064,
      "learning_rate": 0.0015,
      "loss": 1.486,
      "step": 3980
    },
    {
      "epoch": 0.4208860759493671,
      "grad_norm": 0.5885394215583801,
      "learning_rate": 0.0015,
      "loss": 1.4787,
      "step": 3990
    },
    {
      "epoch": 0.4219409282700422,
      "grad_norm": 0.5541431903839111,
      "learning_rate": 0.0015,
      "loss": 1.4913,
      "step": 4000
    },
    {
      "epoch": 0.4229957805907173,
      "grad_norm": 0.5044839978218079,
      "learning_rate": 0.0015,
      "loss": 1.4815,
      "step": 4010
    },
    {
      "epoch": 0.4240506329113924,
      "grad_norm": 0.5214411020278931,
      "learning_rate": 0.0015,
      "loss": 1.4853,
      "step": 4020
    },
    {
      "epoch": 0.42510548523206754,
      "grad_norm": 0.5292476415634155,
      "learning_rate": 0.0015,
      "loss": 1.4768,
      "step": 4030
    },
    {
      "epoch": 0.42616033755274263,
      "grad_norm": 0.5479418635368347,
      "learning_rate": 0.0015,
      "loss": 1.4874,
      "step": 4040
    },
    {
      "epoch": 0.4272151898734177,
      "grad_norm": 0.5096611976623535,
      "learning_rate": 0.0015,
      "loss": 1.485,
      "step": 4050
    },
    {
      "epoch": 0.4282700421940928,
      "grad_norm": 0.7310690879821777,
      "learning_rate": 0.0015,
      "loss": 1.4901,
      "step": 4060
    },
    {
      "epoch": 0.4293248945147679,
      "grad_norm": 0.7201768755912781,
      "learning_rate": 0.0015,
      "loss": 1.4856,
      "step": 4070
    },
    {
      "epoch": 0.43037974683544306,
      "grad_norm": 0.6559078097343445,
      "learning_rate": 0.0015,
      "loss": 1.4914,
      "step": 4080
    },
    {
      "epoch": 0.43143459915611815,
      "grad_norm": 0.5437326431274414,
      "learning_rate": 0.0015,
      "loss": 1.4879,
      "step": 4090
    },
    {
      "epoch": 0.43248945147679324,
      "grad_norm": 0.6245301365852356,
      "learning_rate": 0.0015,
      "loss": 1.48,
      "step": 4100
    },
    {
      "epoch": 0.43354430379746833,
      "grad_norm": 0.76190584897995,
      "learning_rate": 0.0015,
      "loss": 1.4751,
      "step": 4110
    },
    {
      "epoch": 0.4345991561181435,
      "grad_norm": 0.6346284747123718,
      "learning_rate": 0.0015,
      "loss": 1.4807,
      "step": 4120
    },
    {
      "epoch": 0.4356540084388186,
      "grad_norm": 0.5197469592094421,
      "learning_rate": 0.0015,
      "loss": 1.4746,
      "step": 4130
    },
    {
      "epoch": 0.43670886075949367,
      "grad_norm": 0.5903440117835999,
      "learning_rate": 0.0015,
      "loss": 1.476,
      "step": 4140
    },
    {
      "epoch": 0.43776371308016876,
      "grad_norm": 0.5719578862190247,
      "learning_rate": 0.0015,
      "loss": 1.4832,
      "step": 4150
    },
    {
      "epoch": 0.4388185654008439,
      "grad_norm": 0.5282468795776367,
      "learning_rate": 0.0015,
      "loss": 1.4761,
      "step": 4160
    },
    {
      "epoch": 0.439873417721519,
      "grad_norm": 0.634348452091217,
      "learning_rate": 0.0015,
      "loss": 1.4761,
      "step": 4170
    },
    {
      "epoch": 0.4409282700421941,
      "grad_norm": 0.5045709609985352,
      "learning_rate": 0.0015,
      "loss": 1.4732,
      "step": 4180
    },
    {
      "epoch": 0.4419831223628692,
      "grad_norm": 0.5601181387901306,
      "learning_rate": 0.0015,
      "loss": 1.4782,
      "step": 4190
    },
    {
      "epoch": 0.4430379746835443,
      "grad_norm": 0.6216740608215332,
      "learning_rate": 0.0015,
      "loss": 1.4799,
      "step": 4200
    },
    {
      "epoch": 0.4440928270042194,
      "grad_norm": 0.5911687612533569,
      "learning_rate": 0.0015,
      "loss": 1.4684,
      "step": 4210
    },
    {
      "epoch": 0.4451476793248945,
      "grad_norm": 0.6351373791694641,
      "learning_rate": 0.0015,
      "loss": 1.4785,
      "step": 4220
    },
    {
      "epoch": 0.4462025316455696,
      "grad_norm": 0.8020367622375488,
      "learning_rate": 0.0015,
      "loss": 1.4766,
      "step": 4230
    },
    {
      "epoch": 0.4472573839662447,
      "grad_norm": 0.7376629114151001,
      "learning_rate": 0.0015,
      "loss": 1.4748,
      "step": 4240
    },
    {
      "epoch": 0.44831223628691985,
      "grad_norm": 0.6472592949867249,
      "learning_rate": 0.0015,
      "loss": 1.4762,
      "step": 4250
    },
    {
      "epoch": 0.44936708860759494,
      "grad_norm": 0.6546268463134766,
      "learning_rate": 0.0015,
      "loss": 1.483,
      "step": 4260
    },
    {
      "epoch": 0.45042194092827004,
      "grad_norm": 0.522511899471283,
      "learning_rate": 0.0015,
      "loss": 1.4714,
      "step": 4270
    },
    {
      "epoch": 0.45147679324894513,
      "grad_norm": 0.8491767048835754,
      "learning_rate": 0.0015,
      "loss": 1.4682,
      "step": 4280
    },
    {
      "epoch": 0.4525316455696203,
      "grad_norm": 0.5159468650817871,
      "learning_rate": 0.0015,
      "loss": 1.4784,
      "step": 4290
    },
    {
      "epoch": 0.45358649789029537,
      "grad_norm": 0.8218620419502258,
      "learning_rate": 0.0015,
      "loss": 1.4759,
      "step": 4300
    },
    {
      "epoch": 0.45464135021097046,
      "grad_norm": 0.5754440426826477,
      "learning_rate": 0.0015,
      "loss": 1.4731,
      "step": 4310
    },
    {
      "epoch": 0.45569620253164556,
      "grad_norm": 0.7513793110847473,
      "learning_rate": 0.0015,
      "loss": 1.4716,
      "step": 4320
    },
    {
      "epoch": 0.45675105485232065,
      "grad_norm": 0.5255351066589355,
      "learning_rate": 0.0015,
      "loss": 1.4759,
      "step": 4330
    },
    {
      "epoch": 0.4578059071729958,
      "grad_norm": 0.6012159585952759,
      "learning_rate": 0.0015,
      "loss": 1.4681,
      "step": 4340
    },
    {
      "epoch": 0.4588607594936709,
      "grad_norm": 0.7910928726196289,
      "learning_rate": 0.0015,
      "loss": 1.4714,
      "step": 4350
    },
    {
      "epoch": 0.459915611814346,
      "grad_norm": 0.5239707827568054,
      "learning_rate": 0.0015,
      "loss": 1.4874,
      "step": 4360
    },
    {
      "epoch": 0.4609704641350211,
      "grad_norm": 0.5172773003578186,
      "learning_rate": 0.0015,
      "loss": 1.4628,
      "step": 4370
    },
    {
      "epoch": 0.4620253164556962,
      "grad_norm": 0.6210146546363831,
      "learning_rate": 0.0015,
      "loss": 1.4697,
      "step": 4380
    },
    {
      "epoch": 0.4630801687763713,
      "grad_norm": 0.5867434740066528,
      "learning_rate": 0.0015,
      "loss": 1.4665,
      "step": 4390
    },
    {
      "epoch": 0.4641350210970464,
      "grad_norm": 0.6506653428077698,
      "learning_rate": 0.0015,
      "loss": 1.4698,
      "step": 4400
    },
    {
      "epoch": 0.4651898734177215,
      "grad_norm": 0.5838228464126587,
      "learning_rate": 0.0015,
      "loss": 1.4849,
      "step": 4410
    },
    {
      "epoch": 0.46624472573839665,
      "grad_norm": 0.7401859164237976,
      "learning_rate": 0.0015,
      "loss": 1.4671,
      "step": 4420
    },
    {
      "epoch": 0.46729957805907174,
      "grad_norm": 0.5894936323165894,
      "learning_rate": 0.0015,
      "loss": 1.4717,
      "step": 4430
    },
    {
      "epoch": 0.46835443037974683,
      "grad_norm": 0.7206613421440125,
      "learning_rate": 0.0015,
      "loss": 1.4685,
      "step": 4440
    },
    {
      "epoch": 0.4694092827004219,
      "grad_norm": 0.672594428062439,
      "learning_rate": 0.0015,
      "loss": 1.4757,
      "step": 4450
    },
    {
      "epoch": 0.4704641350210971,
      "grad_norm": 0.49490922689437866,
      "learning_rate": 0.0015,
      "loss": 1.4603,
      "step": 4460
    },
    {
      "epoch": 0.47151898734177217,
      "grad_norm": 0.5114975571632385,
      "learning_rate": 0.0015,
      "loss": 1.471,
      "step": 4470
    },
    {
      "epoch": 0.47257383966244726,
      "grad_norm": 0.596366286277771,
      "learning_rate": 0.0015,
      "loss": 1.4771,
      "step": 4480
    },
    {
      "epoch": 0.47362869198312235,
      "grad_norm": 0.5820817351341248,
      "learning_rate": 0.0015,
      "loss": 1.4641,
      "step": 4490
    },
    {
      "epoch": 0.47468354430379744,
      "grad_norm": 0.5126115083694458,
      "learning_rate": 0.0015,
      "loss": 1.4645,
      "step": 4500
    },
    {
      "epoch": 0.4757383966244726,
      "grad_norm": 0.5245903134346008,
      "learning_rate": 0.0015,
      "loss": 1.4679,
      "step": 4510
    },
    {
      "epoch": 0.4767932489451477,
      "grad_norm": 0.5689454674720764,
      "learning_rate": 0.0015,
      "loss": 1.4523,
      "step": 4520
    },
    {
      "epoch": 0.4778481012658228,
      "grad_norm": 0.5443896055221558,
      "learning_rate": 0.0015,
      "loss": 1.4732,
      "step": 4530
    },
    {
      "epoch": 0.47890295358649787,
      "grad_norm": 0.5211290717124939,
      "learning_rate": 0.0015,
      "loss": 1.4715,
      "step": 4540
    },
    {
      "epoch": 0.479957805907173,
      "grad_norm": 0.5262769460678101,
      "learning_rate": 0.0015,
      "loss": 1.4684,
      "step": 4550
    },
    {
      "epoch": 0.4810126582278481,
      "grad_norm": 0.5398851037025452,
      "learning_rate": 0.0015,
      "loss": 1.4587,
      "step": 4560
    },
    {
      "epoch": 0.4820675105485232,
      "grad_norm": 0.7674636840820312,
      "learning_rate": 0.0015,
      "loss": 1.4421,
      "step": 4570
    },
    {
      "epoch": 0.4831223628691983,
      "grad_norm": 0.6799883842468262,
      "learning_rate": 0.0015,
      "loss": 1.4659,
      "step": 4580
    },
    {
      "epoch": 0.48417721518987344,
      "grad_norm": 0.6476281881332397,
      "learning_rate": 0.0015,
      "loss": 1.4518,
      "step": 4590
    },
    {
      "epoch": 0.48523206751054854,
      "grad_norm": 0.5203509330749512,
      "learning_rate": 0.0015,
      "loss": 1.473,
      "step": 4600
    },
    {
      "epoch": 0.48628691983122363,
      "grad_norm": 0.8790157437324524,
      "learning_rate": 0.0015,
      "loss": 1.4751,
      "step": 4610
    },
    {
      "epoch": 0.4873417721518987,
      "grad_norm": 0.5221405029296875,
      "learning_rate": 0.0015,
      "loss": 1.4508,
      "step": 4620
    },
    {
      "epoch": 0.4883966244725738,
      "grad_norm": 0.5264320969581604,
      "learning_rate": 0.0015,
      "loss": 1.4686,
      "step": 4630
    },
    {
      "epoch": 0.48945147679324896,
      "grad_norm": 0.5238685607910156,
      "learning_rate": 0.0015,
      "loss": 1.469,
      "step": 4640
    },
    {
      "epoch": 0.49050632911392406,
      "grad_norm": 0.6246622204780579,
      "learning_rate": 0.0015,
      "loss": 1.4701,
      "step": 4650
    },
    {
      "epoch": 0.49156118143459915,
      "grad_norm": 0.5449501872062683,
      "learning_rate": 0.0015,
      "loss": 1.4587,
      "step": 4660
    },
    {
      "epoch": 0.49261603375527424,
      "grad_norm": 0.6455876231193542,
      "learning_rate": 0.0015,
      "loss": 1.4572,
      "step": 4670
    },
    {
      "epoch": 0.4936708860759494,
      "grad_norm": 0.5345314741134644,
      "learning_rate": 0.0015,
      "loss": 1.4561,
      "step": 4680
    },
    {
      "epoch": 0.4947257383966245,
      "grad_norm": 0.682424783706665,
      "learning_rate": 0.0015,
      "loss": 1.4515,
      "step": 4690
    },
    {
      "epoch": 0.4957805907172996,
      "grad_norm": 0.574175238609314,
      "learning_rate": 0.0015,
      "loss": 1.4604,
      "step": 4700
    },
    {
      "epoch": 0.49683544303797467,
      "grad_norm": 0.593390166759491,
      "learning_rate": 0.0015,
      "loss": 1.462,
      "step": 4710
    },
    {
      "epoch": 0.4978902953586498,
      "grad_norm": 0.6850962042808533,
      "learning_rate": 0.0015,
      "loss": 1.4572,
      "step": 4720
    },
    {
      "epoch": 0.4989451476793249,
      "grad_norm": 0.585054874420166,
      "learning_rate": 0.0015,
      "loss": 1.4569,
      "step": 4730
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.5914109349250793,
      "learning_rate": 0.0015,
      "loss": 1.4559,
      "step": 4740
    },
    {
      "epoch": 0.5010548523206751,
      "grad_norm": 0.8646875023841858,
      "learning_rate": 0.0015,
      "loss": 1.4754,
      "step": 4750
    },
    {
      "epoch": 0.5021097046413502,
      "grad_norm": 0.5117290616035461,
      "learning_rate": 0.0015,
      "loss": 1.459,
      "step": 4760
    },
    {
      "epoch": 0.5031645569620253,
      "grad_norm": 0.47725585103034973,
      "learning_rate": 0.0015,
      "loss": 1.4595,
      "step": 4770
    },
    {
      "epoch": 0.5042194092827004,
      "grad_norm": 0.4821884334087372,
      "learning_rate": 0.0015,
      "loss": 1.4676,
      "step": 4780
    },
    {
      "epoch": 0.5052742616033755,
      "grad_norm": 0.7134470343589783,
      "learning_rate": 0.0015,
      "loss": 1.4519,
      "step": 4790
    },
    {
      "epoch": 0.5063291139240507,
      "grad_norm": 0.5524974465370178,
      "learning_rate": 0.0015,
      "loss": 1.4559,
      "step": 4800
    },
    {
      "epoch": 0.5073839662447257,
      "grad_norm": 0.5358974933624268,
      "learning_rate": 0.0015,
      "loss": 1.4546,
      "step": 4810
    },
    {
      "epoch": 0.5084388185654009,
      "grad_norm": 0.7514449954032898,
      "learning_rate": 0.0015,
      "loss": 1.474,
      "step": 4820
    },
    {
      "epoch": 0.509493670886076,
      "grad_norm": 0.5801400542259216,
      "learning_rate": 0.0015,
      "loss": 1.4584,
      "step": 4830
    },
    {
      "epoch": 0.510548523206751,
      "grad_norm": 0.5107520818710327,
      "learning_rate": 0.0015,
      "loss": 1.4498,
      "step": 4840
    },
    {
      "epoch": 0.5116033755274262,
      "grad_norm": 0.503389298915863,
      "learning_rate": 0.0015,
      "loss": 1.4641,
      "step": 4850
    },
    {
      "epoch": 0.5126582278481012,
      "grad_norm": 0.6112860441207886,
      "learning_rate": 0.0015,
      "loss": 1.4503,
      "step": 4860
    },
    {
      "epoch": 0.5137130801687764,
      "grad_norm": 0.7044478058815002,
      "learning_rate": 0.0015,
      "loss": 1.444,
      "step": 4870
    },
    {
      "epoch": 0.5147679324894515,
      "grad_norm": 0.9283525943756104,
      "learning_rate": 0.0015,
      "loss": 1.4676,
      "step": 4880
    },
    {
      "epoch": 0.5158227848101266,
      "grad_norm": 0.5343573093414307,
      "learning_rate": 0.0015,
      "loss": 1.4547,
      "step": 4890
    },
    {
      "epoch": 0.5168776371308017,
      "grad_norm": 0.5741806030273438,
      "learning_rate": 0.0015,
      "loss": 1.4633,
      "step": 4900
    },
    {
      "epoch": 0.5179324894514767,
      "grad_norm": 0.49245160818099976,
      "learning_rate": 0.0015,
      "loss": 1.4589,
      "step": 4910
    },
    {
      "epoch": 0.5189873417721519,
      "grad_norm": 1.0656530857086182,
      "learning_rate": 0.0015,
      "loss": 1.4482,
      "step": 4920
    },
    {
      "epoch": 0.520042194092827,
      "grad_norm": 0.8917819261550903,
      "learning_rate": 0.0015,
      "loss": 1.4562,
      "step": 4930
    },
    {
      "epoch": 0.5210970464135021,
      "grad_norm": 0.5806669592857361,
      "learning_rate": 0.0015,
      "loss": 1.4565,
      "step": 4940
    },
    {
      "epoch": 0.5221518987341772,
      "grad_norm": 0.6887030005455017,
      "learning_rate": 0.0015,
      "loss": 1.4625,
      "step": 4950
    },
    {
      "epoch": 0.5232067510548524,
      "grad_norm": 0.6455850005149841,
      "learning_rate": 0.0015,
      "loss": 1.443,
      "step": 4960
    },
    {
      "epoch": 0.5242616033755274,
      "grad_norm": 0.6216683983802795,
      "learning_rate": 0.0015,
      "loss": 1.454,
      "step": 4970
    },
    {
      "epoch": 0.5253164556962026,
      "grad_norm": 0.5712945461273193,
      "learning_rate": 0.0015,
      "loss": 1.4471,
      "step": 4980
    },
    {
      "epoch": 0.5263713080168776,
      "grad_norm": 0.4948515295982361,
      "learning_rate": 0.0015,
      "loss": 1.4468,
      "step": 4990
    },
    {
      "epoch": 0.5274261603375527,
      "grad_norm": 0.5773718357086182,
      "learning_rate": 0.0015,
      "loss": 1.4607,
      "step": 5000
    },
    {
      "epoch": 0.5284810126582279,
      "grad_norm": 0.5605436563491821,
      "learning_rate": 0.0015,
      "loss": 1.451,
      "step": 5010
    },
    {
      "epoch": 0.5295358649789029,
      "grad_norm": 0.5638588666915894,
      "learning_rate": 0.0015,
      "loss": 1.4471,
      "step": 5020
    },
    {
      "epoch": 0.5305907172995781,
      "grad_norm": 0.4892915189266205,
      "learning_rate": 0.0015,
      "loss": 1.4307,
      "step": 5030
    },
    {
      "epoch": 0.5316455696202531,
      "grad_norm": 0.5163865089416504,
      "learning_rate": 0.0015,
      "loss": 1.4585,
      "step": 5040
    },
    {
      "epoch": 0.5327004219409283,
      "grad_norm": 0.5103743076324463,
      "learning_rate": 0.0015,
      "loss": 1.4489,
      "step": 5050
    },
    {
      "epoch": 0.5337552742616034,
      "grad_norm": 0.6309067606925964,
      "learning_rate": 0.0015,
      "loss": 1.4599,
      "step": 5060
    },
    {
      "epoch": 0.5348101265822784,
      "grad_norm": 0.6969106197357178,
      "learning_rate": 0.0015,
      "loss": 1.4496,
      "step": 5070
    },
    {
      "epoch": 0.5358649789029536,
      "grad_norm": 0.6979928016662598,
      "learning_rate": 0.0015,
      "loss": 1.4631,
      "step": 5080
    },
    {
      "epoch": 0.5369198312236287,
      "grad_norm": 0.5774954557418823,
      "learning_rate": 0.0015,
      "loss": 1.4531,
      "step": 5090
    },
    {
      "epoch": 0.5379746835443038,
      "grad_norm": 0.5764656662940979,
      "learning_rate": 0.0015,
      "loss": 1.4368,
      "step": 5100
    },
    {
      "epoch": 0.5390295358649789,
      "grad_norm": 0.6822845935821533,
      "learning_rate": 0.0015,
      "loss": 1.4515,
      "step": 5110
    },
    {
      "epoch": 0.540084388185654,
      "grad_norm": 0.6386136412620544,
      "learning_rate": 0.0015,
      "loss": 1.4481,
      "step": 5120
    },
    {
      "epoch": 0.5411392405063291,
      "grad_norm": 0.5767738223075867,
      "learning_rate": 0.0015,
      "loss": 1.4451,
      "step": 5130
    },
    {
      "epoch": 0.5421940928270043,
      "grad_norm": 0.6771249771118164,
      "learning_rate": 0.0015,
      "loss": 1.46,
      "step": 5140
    },
    {
      "epoch": 0.5432489451476793,
      "grad_norm": 0.5186454057693481,
      "learning_rate": 0.0015,
      "loss": 1.455,
      "step": 5150
    },
    {
      "epoch": 0.5443037974683544,
      "grad_norm": 0.5029733777046204,
      "learning_rate": 0.0015,
      "loss": 1.4483,
      "step": 5160
    },
    {
      "epoch": 0.5453586497890295,
      "grad_norm": 0.5092059373855591,
      "learning_rate": 0.0015,
      "loss": 1.448,
      "step": 5170
    },
    {
      "epoch": 0.5464135021097046,
      "grad_norm": 0.6321627497673035,
      "learning_rate": 0.0015,
      "loss": 1.4585,
      "step": 5180
    },
    {
      "epoch": 0.5474683544303798,
      "grad_norm": 0.5579228401184082,
      "learning_rate": 0.0015,
      "loss": 1.4437,
      "step": 5190
    },
    {
      "epoch": 0.5485232067510548,
      "grad_norm": 0.6607871651649475,
      "learning_rate": 0.0015,
      "loss": 1.4559,
      "step": 5200
    },
    {
      "epoch": 0.54957805907173,
      "grad_norm": 0.8128629326820374,
      "learning_rate": 0.0015,
      "loss": 1.4422,
      "step": 5210
    },
    {
      "epoch": 0.5506329113924051,
      "grad_norm": 0.5744751691818237,
      "learning_rate": 0.0015,
      "loss": 1.4501,
      "step": 5220
    },
    {
      "epoch": 0.5516877637130801,
      "grad_norm": 0.49478304386138916,
      "learning_rate": 0.0015,
      "loss": 1.4443,
      "step": 5230
    },
    {
      "epoch": 0.5527426160337553,
      "grad_norm": 0.5468266010284424,
      "learning_rate": 0.0015,
      "loss": 1.4515,
      "step": 5240
    },
    {
      "epoch": 0.5537974683544303,
      "grad_norm": 0.8689888715744019,
      "learning_rate": 0.0015,
      "loss": 1.4417,
      "step": 5250
    },
    {
      "epoch": 0.5548523206751055,
      "grad_norm": 0.5586831569671631,
      "learning_rate": 0.0015,
      "loss": 1.449,
      "step": 5260
    },
    {
      "epoch": 0.5559071729957806,
      "grad_norm": 0.49822619557380676,
      "learning_rate": 0.0015,
      "loss": 1.4535,
      "step": 5270
    },
    {
      "epoch": 0.5569620253164557,
      "grad_norm": 0.779325008392334,
      "learning_rate": 0.0015,
      "loss": 1.4429,
      "step": 5280
    },
    {
      "epoch": 0.5580168776371308,
      "grad_norm": 0.525687038898468,
      "learning_rate": 0.0015,
      "loss": 1.445,
      "step": 5290
    },
    {
      "epoch": 0.5590717299578059,
      "grad_norm": 0.7627896070480347,
      "learning_rate": 0.0015,
      "loss": 1.4522,
      "step": 5300
    },
    {
      "epoch": 0.560126582278481,
      "grad_norm": 0.7745131254196167,
      "learning_rate": 0.0015,
      "loss": 1.4402,
      "step": 5310
    },
    {
      "epoch": 0.5611814345991561,
      "grad_norm": 0.7163421511650085,
      "learning_rate": 0.0015,
      "loss": 1.4404,
      "step": 5320
    },
    {
      "epoch": 0.5622362869198312,
      "grad_norm": 0.4869208335876465,
      "learning_rate": 0.0015,
      "loss": 1.4489,
      "step": 5330
    },
    {
      "epoch": 0.5632911392405063,
      "grad_norm": 0.5700405240058899,
      "learning_rate": 0.0015,
      "loss": 1.4406,
      "step": 5340
    },
    {
      "epoch": 0.5643459915611815,
      "grad_norm": 0.4926275908946991,
      "learning_rate": 0.0015,
      "loss": 1.4348,
      "step": 5350
    },
    {
      "epoch": 0.5654008438818565,
      "grad_norm": 0.6980654001235962,
      "learning_rate": 0.0015,
      "loss": 1.4462,
      "step": 5360
    },
    {
      "epoch": 0.5664556962025317,
      "grad_norm": 0.4659419655799866,
      "learning_rate": 0.0015,
      "loss": 1.4437,
      "step": 5370
    },
    {
      "epoch": 0.5675105485232067,
      "grad_norm": 0.5911593437194824,
      "learning_rate": 0.0015,
      "loss": 1.4395,
      "step": 5380
    },
    {
      "epoch": 0.5685654008438819,
      "grad_norm": 0.6820240616798401,
      "learning_rate": 0.0015,
      "loss": 1.4387,
      "step": 5390
    },
    {
      "epoch": 0.569620253164557,
      "grad_norm": 0.5553368926048279,
      "learning_rate": 0.0015,
      "loss": 1.439,
      "step": 5400
    },
    {
      "epoch": 0.570675105485232,
      "grad_norm": 0.5284712910652161,
      "learning_rate": 0.0015,
      "loss": 1.4402,
      "step": 5410
    },
    {
      "epoch": 0.5717299578059072,
      "grad_norm": 0.4866778552532196,
      "learning_rate": 0.0015,
      "loss": 1.4353,
      "step": 5420
    },
    {
      "epoch": 0.5727848101265823,
      "grad_norm": 0.6628943085670471,
      "learning_rate": 0.0015,
      "loss": 1.4335,
      "step": 5430
    },
    {
      "epoch": 0.5738396624472574,
      "grad_norm": 0.5917623043060303,
      "learning_rate": 0.0015,
      "loss": 1.4297,
      "step": 5440
    },
    {
      "epoch": 0.5748945147679325,
      "grad_norm": 0.5535296201705933,
      "learning_rate": 0.0015,
      "loss": 1.4389,
      "step": 5450
    },
    {
      "epoch": 0.5759493670886076,
      "grad_norm": 0.7727319598197937,
      "learning_rate": 0.0015,
      "loss": 1.4402,
      "step": 5460
    },
    {
      "epoch": 0.5770042194092827,
      "grad_norm": 0.7488126754760742,
      "learning_rate": 0.0015,
      "loss": 1.441,
      "step": 5470
    },
    {
      "epoch": 0.5780590717299579,
      "grad_norm": 0.49722662568092346,
      "learning_rate": 0.0015,
      "loss": 1.4451,
      "step": 5480
    },
    {
      "epoch": 0.5791139240506329,
      "grad_norm": 0.5322569012641907,
      "learning_rate": 0.0015,
      "loss": 1.4477,
      "step": 5490
    },
    {
      "epoch": 0.580168776371308,
      "grad_norm": 0.5261563062667847,
      "learning_rate": 0.0015,
      "loss": 1.4374,
      "step": 5500
    },
    {
      "epoch": 0.5812236286919831,
      "grad_norm": 0.7318642735481262,
      "learning_rate": 0.0015,
      "loss": 1.444,
      "step": 5510
    },
    {
      "epoch": 0.5822784810126582,
      "grad_norm": 0.5357314944267273,
      "learning_rate": 0.0015,
      "loss": 1.4344,
      "step": 5520
    },
    {
      "epoch": 0.5833333333333334,
      "grad_norm": 0.8815802931785583,
      "learning_rate": 0.0015,
      "loss": 1.4308,
      "step": 5530
    },
    {
      "epoch": 0.5843881856540084,
      "grad_norm": 0.7622281908988953,
      "learning_rate": 0.0015,
      "loss": 1.4353,
      "step": 5540
    },
    {
      "epoch": 0.5854430379746836,
      "grad_norm": 1.0153924226760864,
      "learning_rate": 0.0015,
      "loss": 1.4541,
      "step": 5550
    },
    {
      "epoch": 0.5864978902953587,
      "grad_norm": 0.4632505178451538,
      "learning_rate": 0.0015,
      "loss": 1.445,
      "step": 5560
    },
    {
      "epoch": 0.5875527426160337,
      "grad_norm": 0.4806533753871918,
      "learning_rate": 0.0015,
      "loss": 1.4308,
      "step": 5570
    },
    {
      "epoch": 0.5886075949367089,
      "grad_norm": 0.5095115900039673,
      "learning_rate": 0.0015,
      "loss": 1.4322,
      "step": 5580
    },
    {
      "epoch": 0.5896624472573839,
      "grad_norm": 0.5210410952568054,
      "learning_rate": 0.0015,
      "loss": 1.4353,
      "step": 5590
    },
    {
      "epoch": 0.5907172995780591,
      "grad_norm": 0.916880190372467,
      "learning_rate": 0.0015,
      "loss": 1.4326,
      "step": 5600
    },
    {
      "epoch": 0.5917721518987342,
      "grad_norm": 0.5186699032783508,
      "learning_rate": 0.0015,
      "loss": 1.4402,
      "step": 5610
    },
    {
      "epoch": 0.5928270042194093,
      "grad_norm": 0.5431554913520813,
      "learning_rate": 0.0015,
      "loss": 1.4485,
      "step": 5620
    },
    {
      "epoch": 0.5938818565400844,
      "grad_norm": 0.795992910861969,
      "learning_rate": 0.0015,
      "loss": 1.4392,
      "step": 5630
    },
    {
      "epoch": 0.5949367088607594,
      "grad_norm": 0.6337339282035828,
      "learning_rate": 0.0015,
      "loss": 1.4493,
      "step": 5640
    },
    {
      "epoch": 0.5959915611814346,
      "grad_norm": 0.5427796244621277,
      "learning_rate": 0.0015,
      "loss": 1.4438,
      "step": 5650
    },
    {
      "epoch": 0.5970464135021097,
      "grad_norm": 0.8991336226463318,
      "learning_rate": 0.0015,
      "loss": 1.4417,
      "step": 5660
    },
    {
      "epoch": 0.5981012658227848,
      "grad_norm": 0.46260586380958557,
      "learning_rate": 0.0015,
      "loss": 1.4432,
      "step": 5670
    },
    {
      "epoch": 0.5991561181434599,
      "grad_norm": 0.5012056827545166,
      "learning_rate": 0.0015,
      "loss": 1.4457,
      "step": 5680
    },
    {
      "epoch": 0.6002109704641351,
      "grad_norm": 0.8590911030769348,
      "learning_rate": 0.0015,
      "loss": 1.4259,
      "step": 5690
    },
    {
      "epoch": 0.6012658227848101,
      "grad_norm": 0.6026596426963806,
      "learning_rate": 0.0015,
      "loss": 1.4315,
      "step": 5700
    },
    {
      "epoch": 0.6023206751054853,
      "grad_norm": 0.500627338886261,
      "learning_rate": 0.0015,
      "loss": 1.4334,
      "step": 5710
    },
    {
      "epoch": 0.6033755274261603,
      "grad_norm": 0.5152232050895691,
      "learning_rate": 0.0015,
      "loss": 1.4378,
      "step": 5720
    },
    {
      "epoch": 0.6044303797468354,
      "grad_norm": 0.640267014503479,
      "learning_rate": 0.0015,
      "loss": 1.4349,
      "step": 5730
    },
    {
      "epoch": 0.6054852320675106,
      "grad_norm": 0.4892360270023346,
      "learning_rate": 0.0015,
      "loss": 1.4318,
      "step": 5740
    },
    {
      "epoch": 0.6065400843881856,
      "grad_norm": 0.5285003781318665,
      "learning_rate": 0.0015,
      "loss": 1.408,
      "step": 5750
    },
    {
      "epoch": 0.6075949367088608,
      "grad_norm": 0.6576545238494873,
      "learning_rate": 0.0015,
      "loss": 1.4222,
      "step": 5760
    },
    {
      "epoch": 0.6086497890295358,
      "grad_norm": 0.5402347445487976,
      "learning_rate": 0.0015,
      "loss": 1.4228,
      "step": 5770
    },
    {
      "epoch": 0.609704641350211,
      "grad_norm": 0.5078423619270325,
      "learning_rate": 0.0015,
      "loss": 1.4292,
      "step": 5780
    },
    {
      "epoch": 0.6107594936708861,
      "grad_norm": 0.5238718390464783,
      "learning_rate": 0.0015,
      "loss": 1.4272,
      "step": 5790
    },
    {
      "epoch": 0.6118143459915611,
      "grad_norm": 0.5361505150794983,
      "learning_rate": 0.0015,
      "loss": 1.4318,
      "step": 5800
    },
    {
      "epoch": 0.6128691983122363,
      "grad_norm": 0.518470823764801,
      "learning_rate": 0.0015,
      "loss": 1.4362,
      "step": 5810
    },
    {
      "epoch": 0.6139240506329114,
      "grad_norm": 0.5448722839355469,
      "learning_rate": 0.0015,
      "loss": 1.4482,
      "step": 5820
    },
    {
      "epoch": 0.6149789029535865,
      "grad_norm": 0.6580650806427002,
      "learning_rate": 0.0015,
      "loss": 1.4313,
      "step": 5830
    },
    {
      "epoch": 0.6160337552742616,
      "grad_norm": 0.5401043891906738,
      "learning_rate": 0.0015,
      "loss": 1.4419,
      "step": 5840
    },
    {
      "epoch": 0.6170886075949367,
      "grad_norm": 0.5187148451805115,
      "learning_rate": 0.0015,
      "loss": 1.4435,
      "step": 5850
    },
    {
      "epoch": 0.6181434599156118,
      "grad_norm": 0.5287416577339172,
      "learning_rate": 0.0015,
      "loss": 1.433,
      "step": 5860
    },
    {
      "epoch": 0.619198312236287,
      "grad_norm": 0.47637939453125,
      "learning_rate": 0.0015,
      "loss": 1.4278,
      "step": 5870
    },
    {
      "epoch": 0.620253164556962,
      "grad_norm": 0.5101406574249268,
      "learning_rate": 0.0015,
      "loss": 1.4351,
      "step": 5880
    },
    {
      "epoch": 0.6213080168776371,
      "grad_norm": 0.5212313532829285,
      "learning_rate": 0.0015,
      "loss": 1.4282,
      "step": 5890
    },
    {
      "epoch": 0.6223628691983122,
      "grad_norm": 0.49211445450782776,
      "learning_rate": 0.0015,
      "loss": 1.4304,
      "step": 5900
    },
    {
      "epoch": 0.6234177215189873,
      "grad_norm": 0.71826171875,
      "learning_rate": 0.0015,
      "loss": 1.4259,
      "step": 5910
    },
    {
      "epoch": 0.6244725738396625,
      "grad_norm": 0.7862614393234253,
      "learning_rate": 0.0015,
      "loss": 1.4205,
      "step": 5920
    },
    {
      "epoch": 0.6255274261603375,
      "grad_norm": 0.6212243437767029,
      "learning_rate": 0.0015,
      "loss": 1.4265,
      "step": 5930
    },
    {
      "epoch": 0.6265822784810127,
      "grad_norm": 0.5889667868614197,
      "learning_rate": 0.0015,
      "loss": 1.4336,
      "step": 5940
    },
    {
      "epoch": 0.6276371308016878,
      "grad_norm": 0.5171526670455933,
      "learning_rate": 0.0015,
      "loss": 1.4217,
      "step": 5950
    },
    {
      "epoch": 0.6286919831223629,
      "grad_norm": 0.4913538098335266,
      "learning_rate": 0.0015,
      "loss": 1.4293,
      "step": 5960
    },
    {
      "epoch": 0.629746835443038,
      "grad_norm": 0.5577138066291809,
      "learning_rate": 0.0015,
      "loss": 1.4321,
      "step": 5970
    },
    {
      "epoch": 0.630801687763713,
      "grad_norm": 0.5826645493507385,
      "learning_rate": 0.0015,
      "loss": 1.417,
      "step": 5980
    },
    {
      "epoch": 0.6318565400843882,
      "grad_norm": 0.7300624251365662,
      "learning_rate": 0.0015,
      "loss": 1.4326,
      "step": 5990
    },
    {
      "epoch": 0.6329113924050633,
      "grad_norm": 0.5512855648994446,
      "learning_rate": 0.0015,
      "loss": 1.4392,
      "step": 6000
    },
    {
      "epoch": 0.6339662447257384,
      "grad_norm": 0.4720965325832367,
      "learning_rate": 0.0015,
      "loss": 1.4295,
      "step": 6010
    },
    {
      "epoch": 0.6350210970464135,
      "grad_norm": 0.7857840657234192,
      "learning_rate": 0.0015,
      "loss": 1.4315,
      "step": 6020
    },
    {
      "epoch": 0.6360759493670886,
      "grad_norm": 0.5015669465065002,
      "learning_rate": 0.0015,
      "loss": 1.4127,
      "step": 6030
    },
    {
      "epoch": 0.6371308016877637,
      "grad_norm": 0.5542659759521484,
      "learning_rate": 0.0015,
      "loss": 1.433,
      "step": 6040
    },
    {
      "epoch": 0.6381856540084389,
      "grad_norm": 0.5255542993545532,
      "learning_rate": 0.0015,
      "loss": 1.4176,
      "step": 6050
    },
    {
      "epoch": 0.6392405063291139,
      "grad_norm": 0.5467193126678467,
      "learning_rate": 0.0015,
      "loss": 1.4318,
      "step": 6060
    },
    {
      "epoch": 0.640295358649789,
      "grad_norm": 0.6633917093276978,
      "learning_rate": 0.0015,
      "loss": 1.4249,
      "step": 6070
    },
    {
      "epoch": 0.6413502109704642,
      "grad_norm": 0.4869944155216217,
      "learning_rate": 0.0015,
      "loss": 1.4248,
      "step": 6080
    },
    {
      "epoch": 0.6424050632911392,
      "grad_norm": 0.6345862150192261,
      "learning_rate": 0.0015,
      "loss": 1.4247,
      "step": 6090
    },
    {
      "epoch": 0.6434599156118144,
      "grad_norm": 0.5846347808837891,
      "learning_rate": 0.0015,
      "loss": 1.4155,
      "step": 6100
    },
    {
      "epoch": 0.6445147679324894,
      "grad_norm": 0.5600190758705139,
      "learning_rate": 0.0015,
      "loss": 1.4285,
      "step": 6110
    },
    {
      "epoch": 0.6455696202531646,
      "grad_norm": 0.5569438338279724,
      "learning_rate": 0.0015,
      "loss": 1.4262,
      "step": 6120
    },
    {
      "epoch": 0.6466244725738397,
      "grad_norm": 0.5091149806976318,
      "learning_rate": 0.0015,
      "loss": 1.4236,
      "step": 6130
    },
    {
      "epoch": 0.6476793248945147,
      "grad_norm": 0.9297945499420166,
      "learning_rate": 0.0015,
      "loss": 1.4242,
      "step": 6140
    },
    {
      "epoch": 0.6487341772151899,
      "grad_norm": 0.5270900130271912,
      "learning_rate": 0.0015,
      "loss": 1.435,
      "step": 6150
    },
    {
      "epoch": 0.6497890295358649,
      "grad_norm": 0.5317861437797546,
      "learning_rate": 0.0015,
      "loss": 1.4305,
      "step": 6160
    },
    {
      "epoch": 0.6508438818565401,
      "grad_norm": 0.5058794021606445,
      "learning_rate": 0.0014834368975312174,
      "loss": 1.4039,
      "step": 6170
    },
    {
      "epoch": 0.6518987341772152,
      "grad_norm": 0.5960896611213684,
      "learning_rate": 0.0014629899726345957,
      "loss": 1.4285,
      "step": 6180
    },
    {
      "epoch": 0.6529535864978903,
      "grad_norm": 0.5103433728218079,
      "learning_rate": 0.0014428248775471316,
      "loss": 1.4316,
      "step": 6190
    },
    {
      "epoch": 0.6540084388185654,
      "grad_norm": 0.7470860481262207,
      "learning_rate": 0.00142293772767289,
      "loss": 1.4222,
      "step": 6200
    },
    {
      "epoch": 0.6550632911392406,
      "grad_norm": 0.5231701731681824,
      "learning_rate": 0.001403324691959192,
      "loss": 1.4167,
      "step": 6210
    },
    {
      "epoch": 0.6561181434599156,
      "grad_norm": 0.47347888350486755,
      "learning_rate": 0.0013839819921586025,
      "loss": 1.4235,
      "step": 6220
    },
    {
      "epoch": 0.6571729957805907,
      "grad_norm": 0.6941648125648499,
      "learning_rate": 0.0013649059021010894,
      "loss": 1.4119,
      "step": 6230
    },
    {
      "epoch": 0.6582278481012658,
      "grad_norm": 0.5365994572639465,
      "learning_rate": 0.0013460927469762154,
      "loss": 1.4132,
      "step": 6240
    },
    {
      "epoch": 0.6592827004219409,
      "grad_norm": 0.506763219833374,
      "learning_rate": 0.0013275389026252255,
      "loss": 1.4223,
      "step": 6250
    },
    {
      "epoch": 0.6603375527426161,
      "grad_norm": 0.511383593082428,
      "learning_rate": 0.0013092407948428887,
      "loss": 1.4082,
      "step": 6260
    },
    {
      "epoch": 0.6613924050632911,
      "grad_norm": 0.7219920754432678,
      "learning_rate": 0.001291194898688966,
      "loss": 1.4188,
      "step": 6270
    },
    {
      "epoch": 0.6624472573839663,
      "grad_norm": 0.5096638798713684,
      "learning_rate": 0.001273397737809166,
      "loss": 1.414,
      "step": 6280
    },
    {
      "epoch": 0.6635021097046413,
      "grad_norm": 0.4544900059700012,
      "learning_rate": 0.001255845883765463,
      "loss": 1.4105,
      "step": 6290
    },
    {
      "epoch": 0.6645569620253164,
      "grad_norm": 0.6520730257034302,
      "learning_rate": 0.001238535955375642,
      "loss": 1.4045,
      "step": 6300
    },
    {
      "epoch": 0.6656118143459916,
      "grad_norm": 0.5283231735229492,
      "learning_rate": 0.0012214646180619506,
      "loss": 1.4023,
      "step": 6310
    },
    {
      "epoch": 0.6666666666666666,
      "grad_norm": 0.48720583319664,
      "learning_rate": 0.001204628583208727,
      "loss": 1.3973,
      "step": 6320
    },
    {
      "epoch": 0.6677215189873418,
      "grad_norm": 0.5155741572380066,
      "learning_rate": 0.0011880246075288827,
      "loss": 1.4046,
      "step": 6330
    },
    {
      "epoch": 0.6687763713080169,
      "grad_norm": 0.5268201231956482,
      "learning_rate": 0.001171649492439115,
      "loss": 1.3973,
      "step": 6340
    },
    {
      "epoch": 0.669831223628692,
      "grad_norm": 0.49016422033309937,
      "learning_rate": 0.0011555000834437364,
      "loss": 1.4011,
      "step": 6350
    },
    {
      "epoch": 0.6708860759493671,
      "grad_norm": 0.5848821401596069,
      "learning_rate": 0.0011395732695269908,
      "loss": 1.4003,
      "step": 6360
    },
    {
      "epoch": 0.6719409282700421,
      "grad_norm": 0.5101801753044128,
      "learning_rate": 0.0011238659825537505,
      "loss": 1.3787,
      "step": 6370
    },
    {
      "epoch": 0.6729957805907173,
      "grad_norm": 0.5121952295303345,
      "learning_rate": 0.0011083751966784717,
      "loss": 1.3821,
      "step": 6380
    },
    {
      "epoch": 0.6740506329113924,
      "grad_norm": 1.0398744344711304,
      "learning_rate": 0.0010930979277622953,
      "loss": 1.3971,
      "step": 6390
    },
    {
      "epoch": 0.6751054852320675,
      "grad_norm": 0.5912628769874573,
      "learning_rate": 0.0010780312327981854,
      "loss": 1.3963,
      "step": 6400
    },
    {
      "epoch": 0.6761603375527426,
      "grad_norm": 0.47290322184562683,
      "learning_rate": 0.0010631722093439888,
      "loss": 1.3885,
      "step": 6410
    },
    {
      "epoch": 0.6772151898734177,
      "grad_norm": 0.4813903272151947,
      "learning_rate": 0.00104851799496331,
      "loss": 1.3794,
      "step": 6420
    },
    {
      "epoch": 0.6782700421940928,
      "grad_norm": 0.4812224209308624,
      "learning_rate": 0.0010340657666740914,
      "loss": 1.3879,
      "step": 6430
    },
    {
      "epoch": 0.679324894514768,
      "grad_norm": 0.6159099340438843,
      "learning_rate": 0.0010198127404047975,
      "loss": 1.3715,
      "step": 6440
    },
    {
      "epoch": 0.680379746835443,
      "grad_norm": 0.528146505355835,
      "learning_rate": 0.0010057561704580897,
      "loss": 1.3755,
      "step": 6450
    },
    {
      "epoch": 0.6814345991561181,
      "grad_norm": 0.5106189846992493,
      "learning_rate": 0.0009918933489818985,
      "loss": 1.3873,
      "step": 6460
    },
    {
      "epoch": 0.6824894514767933,
      "grad_norm": 0.46552208065986633,
      "learning_rate": 0.0009782216054477827,
      "loss": 1.3795,
      "step": 6470
    },
    {
      "epoch": 0.6835443037974683,
      "grad_norm": 0.6433939337730408,
      "learning_rate": 0.0009647383061364801,
      "loss": 1.386,
      "step": 6480
    },
    {
      "epoch": 0.6845991561181435,
      "grad_norm": 0.45110300183296204,
      "learning_rate": 0.0009514408536305495,
      "loss": 1.3803,
      "step": 6490
    },
    {
      "epoch": 0.6856540084388185,
      "grad_norm": 0.47798672318458557,
      "learning_rate": 0.0009383266863140042,
      "loss": 1.3944,
      "step": 6500
    },
    {
      "epoch": 0.6867088607594937,
      "grad_norm": 0.484041690826416,
      "learning_rate": 0.000925393277878844,
      "loss": 1.3933,
      "step": 6510
    },
    {
      "epoch": 0.6877637130801688,
      "grad_norm": 0.4432898163795471,
      "learning_rate": 0.0009126381368383879,
      "loss": 1.3769,
      "step": 6520
    },
    {
      "epoch": 0.6888185654008439,
      "grad_norm": 0.5166557431221008,
      "learning_rate": 0.0009000588060473156,
      "loss": 1.3702,
      "step": 6530
    },
    {
      "epoch": 0.689873417721519,
      "grad_norm": 0.554893434047699,
      "learning_rate": 0.0008876528622283235,
      "loss": 1.3785,
      "step": 6540
    },
    {
      "epoch": 0.6909282700421941,
      "grad_norm": 0.4747559130191803,
      "learning_rate": 0.0008754179155053053,
      "loss": 1.3695,
      "step": 6550
    },
    {
      "epoch": 0.6919831223628692,
      "grad_norm": 0.5023005604743958,
      "learning_rate": 0.0008633516089429683,
      "loss": 1.3726,
      "step": 6560
    },
    {
      "epoch": 0.6930379746835443,
      "grad_norm": 0.4733406901359558,
      "learning_rate": 0.0008514516180927928,
      "loss": 1.3682,
      "step": 6570
    },
    {
      "epoch": 0.6940928270042194,
      "grad_norm": 0.5170013904571533,
      "learning_rate": 0.0008397156505452524,
      "loss": 1.365,
      "step": 6580
    },
    {
      "epoch": 0.6951476793248945,
      "grad_norm": 0.5229912400245667,
      "learning_rate": 0.0008281414454882051,
      "loss": 1.3693,
      "step": 6590
    },
    {
      "epoch": 0.6962025316455697,
      "grad_norm": 0.5185620188713074,
      "learning_rate": 0.0008167267732713704,
      "loss": 1.3725,
      "step": 6600
    },
    {
      "epoch": 0.6972573839662447,
      "grad_norm": 0.44782182574272156,
      "learning_rate": 0.0008054694349768117,
      "loss": 1.3589,
      "step": 6610
    },
    {
      "epoch": 0.6983122362869199,
      "grad_norm": 0.4861155152320862,
      "learning_rate": 0.0007943672619953359,
      "loss": 1.3664,
      "step": 6620
    },
    {
      "epoch": 0.6993670886075949,
      "grad_norm": 0.5309790372848511,
      "learning_rate": 0.0007834181156087356,
      "loss": 1.3607,
      "step": 6630
    },
    {
      "epoch": 0.70042194092827,
      "grad_norm": 0.583339512348175,
      "learning_rate": 0.0007726198865777852,
      "loss": 1.3637,
      "step": 6640
    },
    {
      "epoch": 0.7014767932489452,
      "grad_norm": 0.4820302426815033,
      "learning_rate": 0.0007619704947359191,
      "loss": 1.3567,
      "step": 6650
    },
    {
      "epoch": 0.7025316455696202,
      "grad_norm": 0.4411340653896332,
      "learning_rate": 0.0007514678885885087,
      "loss": 1.3555,
      "step": 6660
    },
    {
      "epoch": 0.7035864978902954,
      "grad_norm": 0.5487621426582336,
      "learning_rate": 0.0007411100449176633,
      "loss": 1.3593,
      "step": 6670
    },
    {
      "epoch": 0.7046413502109705,
      "grad_norm": 0.45803865790367126,
      "learning_rate": 0.0007308949683924791,
      "loss": 1.361,
      "step": 6680
    },
    {
      "epoch": 0.7056962025316456,
      "grad_norm": 0.5200080871582031,
      "learning_rate": 0.000720820691184658,
      "loss": 1.3527,
      "step": 6690
    },
    {
      "epoch": 0.7067510548523207,
      "grad_norm": 0.5422410368919373,
      "learning_rate": 0.0007108852725894269,
      "loss": 1.355,
      "step": 6700
    },
    {
      "epoch": 0.7078059071729957,
      "grad_norm": 0.549124538898468,
      "learning_rate": 0.000701086798651681,
      "loss": 1.3548,
      "step": 6710
    },
    {
      "epoch": 0.7088607594936709,
      "grad_norm": 0.6717361807823181,
      "learning_rate": 0.0006914233817972798,
      "loss": 1.3441,
      "step": 6720
    },
    {
      "epoch": 0.709915611814346,
      "grad_norm": 0.47141385078430176,
      "learning_rate": 0.0006818931604694261,
      "loss": 1.3531,
      "step": 6730
    },
    {
      "epoch": 0.7109704641350211,
      "grad_norm": 0.5401121377944946,
      "learning_rate": 0.0006724942987700563,
      "loss": 1.3554,
      "step": 6740
    },
    {
      "epoch": 0.7120253164556962,
      "grad_norm": 0.45940276980400085,
      "learning_rate": 0.0006632249861061732,
      "loss": 1.3587,
      "step": 6750
    },
    {
      "epoch": 0.7130801687763713,
      "grad_norm": 0.6030657291412354,
      "learning_rate": 0.0006540834368410549,
      "loss": 1.3511,
      "step": 6760
    },
    {
      "epoch": 0.7141350210970464,
      "grad_norm": 0.5427432060241699,
      "learning_rate": 0.0006450678899502701,
      "loss": 1.3525,
      "step": 6770
    },
    {
      "epoch": 0.7151898734177216,
      "grad_norm": 0.47614455223083496,
      "learning_rate": 0.0006361766086824345,
      "loss": 1.3521,
      "step": 6780
    },
    {
      "epoch": 0.7162447257383966,
      "grad_norm": 0.521483302116394,
      "learning_rate": 0.000627407880224645,
      "loss": 1.3564,
      "step": 6790
    },
    {
      "epoch": 0.7172995780590717,
      "grad_norm": 0.4478306174278259,
      "learning_rate": 0.0006187600153725225,
      "loss": 1.3411,
      "step": 6800
    },
    {
      "epoch": 0.7183544303797469,
      "grad_norm": 0.7660099864006042,
      "learning_rate": 0.0006102313482048055,
      "loss": 1.3471,
      "step": 6810
    },
    {
      "epoch": 0.7194092827004219,
      "grad_norm": 0.5013904571533203,
      "learning_rate": 0.0006018202357624274,
      "loss": 1.3473,
      "step": 6820
    },
    {
      "epoch": 0.7204641350210971,
      "grad_norm": 0.5144420266151428,
      "learning_rate": 0.0005935250577320168,
      "loss": 1.3428,
      "step": 6830
    },
    {
      "epoch": 0.7215189873417721,
      "grad_norm": 0.5867617726325989,
      "learning_rate": 0.0005853442161337618,
      "loss": 1.3347,
      "step": 6840
    },
    {
      "epoch": 0.7225738396624473,
      "grad_norm": 0.4562567472457886,
      "learning_rate": 0.0005772761350135759,
      "loss": 1.3429,
      "step": 6850
    },
    {
      "epoch": 0.7236286919831224,
      "grad_norm": 0.5033745765686035,
      "learning_rate": 0.0005693192601395058,
      "loss": 1.3374,
      "step": 6860
    },
    {
      "epoch": 0.7246835443037974,
      "grad_norm": 0.458080530166626,
      "learning_rate": 0.000561472058702326,
      "loss": 1.3334,
      "step": 6870
    },
    {
      "epoch": 0.7257383966244726,
      "grad_norm": 0.5429106950759888,
      "learning_rate": 0.000553733019020258,
      "loss": 1.3422,
      "step": 6880
    },
    {
      "epoch": 0.7267932489451476,
      "grad_norm": 0.4628061056137085,
      "learning_rate": 0.0005461006502477612,
      "loss": 1.3305,
      "step": 6890
    },
    {
      "epoch": 0.7278481012658228,
      "grad_norm": 0.48667672276496887,
      "learning_rate": 0.0005385734820883369,
      "loss": 1.3325,
      "step": 6900
    },
    {
      "epoch": 0.7289029535864979,
      "grad_norm": 0.48949795961380005,
      "learning_rate": 0.0005311500645112907,
      "loss": 1.3571,
      "step": 6910
    },
    {
      "epoch": 0.729957805907173,
      "grad_norm": 0.5208607316017151,
      "learning_rate": 0.0005238289674723993,
      "loss": 1.3346,
      "step": 6920
    },
    {
      "epoch": 0.7310126582278481,
      "grad_norm": 0.44086530804634094,
      "learning_rate": 0.0005166087806384274,
      "loss": 1.3396,
      "step": 6930
    },
    {
      "epoch": 0.7320675105485233,
      "grad_norm": 0.49967527389526367,
      "learning_rate": 0.0005094881131154418,
      "loss": 1.3422,
      "step": 6940
    },
    {
      "epoch": 0.7331223628691983,
      "grad_norm": 0.46255460381507874,
      "learning_rate": 0.0005024655931808696,
      "loss": 1.339,
      "step": 6950
    },
    {
      "epoch": 0.7341772151898734,
      "grad_norm": 0.458722323179245,
      "learning_rate": 0.0004955398680192508,
      "loss": 1.3294,
      "step": 6960
    },
    {
      "epoch": 0.7352320675105485,
      "grad_norm": 0.45329201221466064,
      "learning_rate": 0.000488709603461632,
      "loss": 1.3266,
      "step": 6970
    },
    {
      "epoch": 0.7362869198312236,
      "grad_norm": 0.48035183548927307,
      "learning_rate": 0.000481973483728553,
      "loss": 1.3301,
      "step": 6980
    },
    {
      "epoch": 0.7373417721518988,
      "grad_norm": 0.4558931291103363,
      "learning_rate": 0.0004753302111765748,
      "loss": 1.3267,
      "step": 6990
    },
    {
      "epoch": 0.7383966244725738,
      "grad_norm": 0.48222649097442627,
      "learning_rate": 0.0004687785060483032,
      "loss": 1.3419,
      "step": 7000
    },
    {
      "epoch": 0.739451476793249,
      "grad_norm": 0.4744454324245453,
      "learning_rate": 0.0004623171062258558,
      "loss": 1.3102,
      "step": 7010
    },
    {
      "epoch": 0.740506329113924,
      "grad_norm": 0.6059104204177856,
      "learning_rate": 0.0004559447669877288,
      "loss": 1.3238,
      "step": 7020
    },
    {
      "epoch": 0.7415611814345991,
      "grad_norm": 0.483710914850235,
      "learning_rate": 0.00044966026076901413,
      "loss": 1.3307,
      "step": 7030
    },
    {
      "epoch": 0.7426160337552743,
      "grad_norm": 0.7889571785926819,
      "learning_rate": 0.00044346237692492177,
      "loss": 1.3351,
      "step": 7040
    },
    {
      "epoch": 0.7436708860759493,
      "grad_norm": 0.4923046827316284,
      "learning_rate": 0.0004373499214975615,
      "loss": 1.3228,
      "step": 7050
    },
    {
      "epoch": 0.7447257383966245,
      "grad_norm": 0.5098389387130737,
      "learning_rate": 0.0004313217169859396,
      "loss": 1.3265,
      "step": 7060
    },
    {
      "epoch": 0.7457805907172996,
      "grad_norm": 0.525782585144043,
      "learning_rate": 0.0004253766021191256,
      "loss": 1.3322,
      "step": 7070
    },
    {
      "epoch": 0.7468354430379747,
      "grad_norm": 0.45359766483306885,
      "learning_rate": 0.00041951343163254497,
      "loss": 1.3282,
      "step": 7080
    },
    {
      "epoch": 0.7478902953586498,
      "grad_norm": 0.5485321283340454,
      "learning_rate": 0.00041373107604735626,
      "loss": 1.326,
      "step": 7090
    },
    {
      "epoch": 0.7489451476793249,
      "grad_norm": 0.45252397656440735,
      "learning_rate": 0.0004080284214528687,
      "loss": 1.3231,
      "step": 7100
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.46911129355430603,
      "learning_rate": 0.0004024043692919589,
      "loss": 1.3321,
      "step": 7110
    },
    {
      "epoch": 0.7510548523206751,
      "grad_norm": 0.6181014776229858,
      "learning_rate": 0.0003968578361494449,
      "loss": 1.3265,
      "step": 7120
    },
    {
      "epoch": 0.7521097046413502,
      "grad_norm": 0.4632599353790283,
      "learning_rate": 0.000391387753543378,
      "loss": 1.3333,
      "step": 7130
    },
    {
      "epoch": 0.7531645569620253,
      "grad_norm": 0.47347453236579895,
      "learning_rate": 0.00038599306771921023,
      "loss": 1.3179,
      "step": 7140
    },
    {
      "epoch": 0.7542194092827004,
      "grad_norm": 0.5055296421051025,
      "learning_rate": 0.0003806727394468004,
      "loss": 1.3138,
      "step": 7150
    },
    {
      "epoch": 0.7552742616033755,
      "grad_norm": 0.49436309933662415,
      "learning_rate": 0.0003754257438202162,
      "loss": 1.3251,
      "step": 7160
    },
    {
      "epoch": 0.7563291139240507,
      "grad_norm": 0.5135729312896729,
      "learning_rate": 0.0003702510700602974,
      "loss": 1.3313,
      "step": 7170
    },
    {
      "epoch": 0.7573839662447257,
      "grad_norm": 0.474187433719635,
      "learning_rate": 0.0003651477213199393,
      "loss": 1.3079,
      "step": 7180
    },
    {
      "epoch": 0.7584388185654009,
      "grad_norm": 0.45640918612480164,
      "learning_rate": 0.000360114714492061,
      "loss": 1.3116,
      "step": 7190
    },
    {
      "epoch": 0.759493670886076,
      "grad_norm": 0.48335134983062744,
      "learning_rate": 0.0003551510800202195,
      "loss": 1.3161,
      "step": 7200
    },
    {
      "epoch": 0.760548523206751,
      "grad_norm": 0.6336145401000977,
      "learning_rate": 0.0003502558617118353,
      "loss": 1.3199,
      "step": 7210
    },
    {
      "epoch": 0.7616033755274262,
      "grad_norm": 0.45102179050445557,
      "learning_rate": 0.0003454281165539914,
      "loss": 1.3327,
      "step": 7220
    },
    {
      "epoch": 0.7626582278481012,
      "grad_norm": 0.4698895215988159,
      "learning_rate": 0.00034066691453177176,
      "loss": 1.3291,
      "step": 7230
    },
    {
      "epoch": 0.7637130801687764,
      "grad_norm": 0.48220446705818176,
      "learning_rate": 0.0003359713384491037,
      "loss": 1.3238,
      "step": 7240
    },
    {
      "epoch": 0.7647679324894515,
      "grad_norm": 0.45443829894065857,
      "learning_rate": 0.00033134048375206944,
      "loss": 1.3214,
      "step": 7250
    },
    {
      "epoch": 0.7658227848101266,
      "grad_norm": 0.47124338150024414,
      "learning_rate": 0.0003267734583546536,
      "loss": 1.3166,
      "step": 7260
    },
    {
      "epoch": 0.7668776371308017,
      "grad_norm": 0.451641321182251,
      "learning_rate": 0.00032226938246689157,
      "loss": 1.3144,
      "step": 7270
    },
    {
      "epoch": 0.7679324894514767,
      "grad_norm": 0.45359188318252563,
      "learning_rate": 0.0003178273884253874,
      "loss": 1.3152,
      "step": 7280
    },
    {
      "epoch": 0.7689873417721519,
      "grad_norm": 0.5116612315177917,
      "learning_rate": 0.0003134466205261674,
      "loss": 1.3269,
      "step": 7290
    },
    {
      "epoch": 0.770042194092827,
      "grad_norm": 0.48015034198760986,
      "learning_rate": 0.0003091262348598378,
      "loss": 1.3301,
      "step": 7300
    },
    {
      "epoch": 0.7710970464135021,
      "grad_norm": 0.45035305619239807,
      "learning_rate": 0.0003048653991490141,
      "loss": 1.3112,
      "step": 7310
    },
    {
      "epoch": 0.7721518987341772,
      "grad_norm": 0.4732269048690796,
      "learning_rate": 0.00030066329258799187,
      "loss": 1.3067,
      "step": 7320
    },
    {
      "epoch": 0.7732067510548524,
      "grad_norm": 0.5562229156494141,
      "learning_rate": 0.0002965191056846266,
      "loss": 1.3113,
      "step": 7330
    },
    {
      "epoch": 0.7742616033755274,
      "grad_norm": 0.5415487885475159,
      "learning_rate": 0.000292432040104394,
      "loss": 1.3091,
      "step": 7340
    },
    {
      "epoch": 0.7753164556962026,
      "grad_norm": 0.4745350778102875,
      "learning_rate": 0.00028840130851659853,
      "loss": 1.3042,
      "step": 7350
    },
    {
      "epoch": 0.7763713080168776,
      "grad_norm": 0.4584297239780426,
      "learning_rate": 0.0002844261344427028,
      "loss": 1.3122,
      "step": 7360
    },
    {
      "epoch": 0.7774261603375527,
      "grad_norm": 0.5179589986801147,
      "learning_rate": 0.0002805057521067471,
      "loss": 1.3135,
      "step": 7370
    },
    {
      "epoch": 0.7784810126582279,
      "grad_norm": 0.4719301760196686,
      "learning_rate": 0.00027663940628783017,
      "loss": 1.3013,
      "step": 7380
    },
    {
      "epoch": 0.7795358649789029,
      "grad_norm": 0.45558246970176697,
      "learning_rate": 0.00027282635217462393,
      "loss": 1.3106,
      "step": 7390
    },
    {
      "epoch": 0.7805907172995781,
      "grad_norm": 0.47958579659461975,
      "learning_rate": 0.0002690658552218937,
      "loss": 1.3187,
      "step": 7400
    },
    {
      "epoch": 0.7816455696202531,
      "grad_norm": 0.44435548782348633,
      "learning_rate": 0.00026535719100899516,
      "loss": 1.2973,
      "step": 7410
    },
    {
      "epoch": 0.7827004219409283,
      "grad_norm": 0.4782116115093231,
      "learning_rate": 0.00026169964510032245,
      "loss": 1.3045,
      "step": 7420
    },
    {
      "epoch": 0.7837552742616034,
      "grad_norm": 0.4518148601055145,
      "learning_rate": 0.00025809251290767984,
      "loss": 1.2897,
      "step": 7430
    },
    {
      "epoch": 0.7848101265822784,
      "grad_norm": 0.4496789872646332,
      "learning_rate": 0.00025453509955454957,
      "loss": 1.2964,
      "step": 7440
    },
    {
      "epoch": 0.7858649789029536,
      "grad_norm": 0.44662004709243774,
      "learning_rate": 0.00025102671974223175,
      "loss": 1.2985,
      "step": 7450
    },
    {
      "epoch": 0.7869198312236287,
      "grad_norm": 0.45507729053497314,
      "learning_rate": 0.00024756669761782815,
      "loss": 1.3085,
      "step": 7460
    },
    {
      "epoch": 0.7879746835443038,
      "grad_norm": 0.4831513464450836,
      "learning_rate": 0.0002441543666440464,
      "loss": 1.3018,
      "step": 7470
    },
    {
      "epoch": 0.7890295358649789,
      "grad_norm": 0.4986232817173004,
      "learning_rate": 0.00024078906947079878,
      "loss": 1.3051,
      "step": 7480
    },
    {
      "epoch": 0.790084388185654,
      "grad_norm": 0.4647130072116852,
      "learning_rate": 0.00023747015780857005,
      "loss": 1.3119,
      "step": 7490
    },
    {
      "epoch": 0.7911392405063291,
      "grad_norm": 0.4940994679927826,
      "learning_rate": 0.00023419699230353144,
      "loss": 1.3066,
      "step": 7500
    },
    {
      "epoch": 0.7921940928270043,
      "grad_norm": 0.5053297877311707,
      "learning_rate": 0.00023096894241437586,
      "loss": 1.3132,
      "step": 7510
    },
    {
      "epoch": 0.7932489451476793,
      "grad_norm": 0.43774738907814026,
      "learning_rate": 0.00022778538629085056,
      "loss": 1.2967,
      "step": 7520
    },
    {
      "epoch": 0.7943037974683544,
      "grad_norm": 0.4893394410610199,
      "learning_rate": 0.00022464571065396427,
      "loss": 1.2988,
      "step": 7530
    },
    {
      "epoch": 0.7953586497890295,
      "grad_norm": 0.4648034870624542,
      "learning_rate": 0.00022154931067784521,
      "loss": 1.2975,
      "step": 7540
    },
    {
      "epoch": 0.7964135021097046,
      "grad_norm": 0.4623718857765198,
      "learning_rate": 0.00021849558987322782,
      "loss": 1.2934,
      "step": 7550
    },
    {
      "epoch": 0.7974683544303798,
      "grad_norm": 0.49164512753486633,
      "learning_rate": 0.0002154839599725452,
      "loss": 1.297,
      "step": 7560
    },
    {
      "epoch": 0.7985232067510548,
      "grad_norm": 0.4767402410507202,
      "learning_rate": 0.00021251384081660544,
      "loss": 1.3072,
      "step": 7570
    },
    {
      "epoch": 0.79957805907173,
      "grad_norm": 0.45326516032218933,
      "learning_rate": 0.0002095846602428303,
      "loss": 1.3039,
      "step": 7580
    },
    {
      "epoch": 0.8006329113924051,
      "grad_norm": 0.4626515805721283,
      "learning_rate": 0.00020669585397503358,
      "loss": 1.2943,
      "step": 7590
    },
    {
      "epoch": 0.8016877637130801,
      "grad_norm": 0.5333366990089417,
      "learning_rate": 0.0002038468655147195,
      "loss": 1.2978,
      "step": 7600
    },
    {
      "epoch": 0.8027426160337553,
      "grad_norm": 0.46550512313842773,
      "learning_rate": 0.00020103714603387894,
      "loss": 1.3115,
      "step": 7610
    },
    {
      "epoch": 0.8037974683544303,
      "grad_norm": 0.471002995967865,
      "learning_rate": 0.00019826615426926338,
      "loss": 1.2836,
      "step": 7620
    },
    {
      "epoch": 0.8048523206751055,
      "grad_norm": 0.4654116630554199,
      "learning_rate": 0.00019553335641811625,
      "loss": 1.3087,
      "step": 7630
    },
    {
      "epoch": 0.8059071729957806,
      "grad_norm": 0.4869087040424347,
      "learning_rate": 0.0001928382260353415,
      "loss": 1.2974,
      "step": 7640
    },
    {
      "epoch": 0.8069620253164557,
      "grad_norm": 0.45881882309913635,
      "learning_rate": 0.00019018024393208902,
      "loss": 1.3155,
      "step": 7650
    },
    {
      "epoch": 0.8080168776371308,
      "grad_norm": 0.46567991375923157,
      "learning_rate": 0.00018755889807573872,
      "loss": 1.2918,
      "step": 7660
    },
    {
      "epoch": 0.8090717299578059,
      "grad_norm": 0.45518314838409424,
      "learning_rate": 0.00018497368349126262,
      "loss": 1.2962,
      "step": 7670
    },
    {
      "epoch": 0.810126582278481,
      "grad_norm": 0.5840961933135986,
      "learning_rate": 0.00018242410216394648,
      "loss": 1.3106,
      "step": 7680
    },
    {
      "epoch": 0.8111814345991561,
      "grad_norm": 0.5052498579025269,
      "learning_rate": 0.0001799096629434529,
      "loss": 1.2875,
      "step": 7690
    },
    {
      "epoch": 0.8122362869198312,
      "grad_norm": 0.4697405695915222,
      "learning_rate": 0.00017742988144920578,
      "loss": 1.2954,
      "step": 7700
    },
    {
      "epoch": 0.8132911392405063,
      "grad_norm": 0.6029151678085327,
      "learning_rate": 0.00017498427997707976,
      "loss": 1.2945,
      "step": 7710
    },
    {
      "epoch": 0.8143459915611815,
      "grad_norm": 0.45356860756874084,
      "learning_rate": 0.00017257238740737548,
      "loss": 1.2984,
      "step": 7720
    },
    {
      "epoch": 0.8154008438818565,
      "grad_norm": 0.4896215796470642,
      "learning_rate": 0.00017019373911406307,
      "loss": 1.3037,
      "step": 7730
    },
    {
      "epoch": 0.8164556962025317,
      "grad_norm": 0.4377332329750061,
      "learning_rate": 0.000167847876875277,
      "loss": 1.304,
      "step": 7740
    },
    {
      "epoch": 0.8175105485232067,
      "grad_norm": 0.49951452016830444,
      "learning_rate": 0.00016553434878504428,
      "loss": 1.2869,
      "step": 7750
    },
    {
      "epoch": 0.8185654008438819,
      "grad_norm": 0.5342950820922852,
      "learning_rate": 0.00016325270916622947,
      "loss": 1.2892,
      "step": 7760
    },
    {
      "epoch": 0.819620253164557,
      "grad_norm": 0.47930166125297546,
      "learning_rate": 0.00016100251848467966,
      "loss": 1.2981,
      "step": 7770
    },
    {
      "epoch": 0.820675105485232,
      "grad_norm": 0.5480371117591858,
      "learning_rate": 0.0001587833432645528,
      "loss": 1.2891,
      "step": 7780
    },
    {
      "epoch": 0.8217299578059072,
      "grad_norm": 0.5335478186607361,
      "learning_rate": 0.00015659475600481292,
      "loss": 1.3058,
      "step": 7790
    },
    {
      "epoch": 0.8227848101265823,
      "grad_norm": 0.4627496898174286,
      "learning_rate": 0.00015443633509687688,
      "loss": 1.2934,
      "step": 7800
    },
    {
      "epoch": 0.8238396624472574,
      "grad_norm": 0.4837168753147125,
      "learning_rate": 0.00015230766474339536,
      "loss": 1.2941,
      "step": 7810
    },
    {
      "epoch": 0.8248945147679325,
      "grad_norm": 0.4632161557674408,
      "learning_rate": 0.00015020833487815416,
      "loss": 1.3028,
      "step": 7820
    },
    {
      "epoch": 0.8259493670886076,
      "grad_norm": 0.462186723947525,
      "learning_rate": 0.0001481379410870792,
      "loss": 1.2894,
      "step": 7830
    },
    {
      "epoch": 0.8270042194092827,
      "grad_norm": 0.45506542921066284,
      "learning_rate": 0.00014609608453033013,
      "loss": 1.2765,
      "step": 7840
    },
    {
      "epoch": 0.8280590717299579,
      "grad_norm": 0.46044138073921204,
      "learning_rate": 0.00014408237186546807,
      "loss": 1.2864,
      "step": 7850
    },
    {
      "epoch": 0.8291139240506329,
      "grad_norm": 0.4451313018798828,
      "learning_rate": 0.00014209641517168273,
      "loss": 1.2744,
      "step": 7860
    },
    {
      "epoch": 0.830168776371308,
      "grad_norm": 0.5132520198822021,
      "learning_rate": 0.00014013783187506265,
      "loss": 1.2894,
      "step": 7870
    },
    {
      "epoch": 0.8312236286919831,
      "grad_norm": 0.48073527216911316,
      "learning_rate": 0.00013820624467489697,
      "loss": 1.3076,
      "step": 7880
    },
    {
      "epoch": 0.8322784810126582,
      "grad_norm": 0.4535788297653198,
      "learning_rate": 0.00013630128147099213,
      "loss": 1.303,
      "step": 7890
    },
    {
      "epoch": 0.8333333333333334,
      "grad_norm": 0.4657829701900482,
      "learning_rate": 0.00013442257529199068,
      "loss": 1.2821,
      "step": 7900
    },
    {
      "epoch": 0.8343881856540084,
      "grad_norm": 0.4929972290992737,
      "learning_rate": 0.00013256976422467803,
      "loss": 1.2925,
      "step": 7910
    },
    {
      "epoch": 0.8354430379746836,
      "grad_norm": 0.49591654539108276,
      "learning_rate": 0.00013074249134426366,
      "loss": 1.2919,
      "step": 7920
    },
    {
      "epoch": 0.8364978902953587,
      "grad_norm": 0.47268256545066833,
      "learning_rate": 0.0001289404046456233,
      "loss": 1.3015,
      "step": 7930
    },
    {
      "epoch": 0.8375527426160337,
      "grad_norm": 0.480153888463974,
      "learning_rate": 0.0001271631569754887,
      "loss": 1.2945,
      "step": 7940
    },
    {
      "epoch": 0.8386075949367089,
      "grad_norm": 0.4759581685066223,
      "learning_rate": 0.0001254104059655723,
      "loss": 1.2985,
      "step": 7950
    },
    {
      "epoch": 0.8396624472573839,
      "grad_norm": 0.42623016238212585,
      "learning_rate": 0.00012368181396661337,
      "loss": 1.2832,
      "step": 7960
    },
    {
      "epoch": 0.8407172995780591,
      "grad_norm": 0.46416598558425903,
      "learning_rate": 0.00012197704798333364,
      "loss": 1.2832,
      "step": 7970
    },
    {
      "epoch": 0.8417721518987342,
      "grad_norm": 0.4567941129207611,
      "learning_rate": 0.00012029577961028894,
      "loss": 1.2829,
      "step": 7980
    },
    {
      "epoch": 0.8428270042194093,
      "grad_norm": 0.570604145526886,
      "learning_rate": 0.00011863768496860542,
      "loss": 1.2949,
      "step": 7990
    },
    {
      "epoch": 0.8438818565400844,
      "grad_norm": 0.45113933086395264,
      "learning_rate": 0.00011700244464358777,
      "loss": 1.2881,
      "step": 8000
    },
    {
      "epoch": 0.8449367088607594,
      "grad_norm": 0.49941763281822205,
      "learning_rate": 0.00011538974362318715,
      "loss": 1.2957,
      "step": 8010
    },
    {
      "epoch": 0.8459915611814346,
      "grad_norm": 0.4934632182121277,
      "learning_rate": 0.00011379927123731737,
      "loss": 1.2884,
      "step": 8020
    },
    {
      "epoch": 0.8470464135021097,
      "grad_norm": 0.4868738651275635,
      "learning_rate": 0.0001122307210980077,
      "loss": 1.2876,
      "step": 8030
    },
    {
      "epoch": 0.8481012658227848,
      "grad_norm": 0.4697803258895874,
      "learning_rate": 0.00011068379104038026,
      "loss": 1.2961,
      "step": 8040
    },
    {
      "epoch": 0.8491561181434599,
      "grad_norm": 0.44700032472610474,
      "learning_rate": 0.00010915818306444116,
      "loss": 1.2803,
      "step": 8050
    },
    {
      "epoch": 0.8502109704641351,
      "grad_norm": 0.44790133833885193,
      "learning_rate": 0.00010765360327767384,
      "loss": 1.2904,
      "step": 8060
    },
    {
      "epoch": 0.8512658227848101,
      "grad_norm": 0.4414917528629303,
      "learning_rate": 0.00010616976183842376,
      "loss": 1.2914,
      "step": 8070
    },
    {
      "epoch": 0.8523206751054853,
      "grad_norm": 0.4861648976802826,
      "learning_rate": 0.00010470637290006365,
      "loss": 1.2938,
      "step": 8080
    },
    {
      "epoch": 0.8533755274261603,
      "grad_norm": 0.46201643347740173,
      "learning_rate": 0.00010326315455592764,
      "loss": 1.284,
      "step": 8090
    },
    {
      "epoch": 0.8544303797468354,
      "grad_norm": 0.5016726851463318,
      "learning_rate": 0.0001018398287850053,
      "loss": 1.2758,
      "step": 8100
    },
    {
      "epoch": 0.8554852320675106,
      "grad_norm": 0.4872339963912964,
      "learning_rate": 0.00010043612139838357,
      "loss": 1.2959,
      "step": 8110
    },
    {
      "epoch": 0.8565400843881856,
      "grad_norm": 0.4454995095729828,
      "learning_rate": 9.905176198642719e-05,
      "loss": 1.2849,
      "step": 8120
    },
    {
      "epoch": 0.8575949367088608,
      "grad_norm": 0.45504704117774963,
      "learning_rate": 9.76864838666871e-05,
      "loss": 1.2873,
      "step": 8130
    },
    {
      "epoch": 0.8586497890295358,
      "grad_norm": 0.4604946970939636,
      "learning_rate": 9.634002403252676e-05,
      "loss": 1.2822,
      "step": 8140
    },
    {
      "epoch": 0.859704641350211,
      "grad_norm": 0.43320751190185547,
      "learning_rate": 9.501212310245681e-05,
      "loss": 1.2803,
      "step": 8150
    },
    {
      "epoch": 0.8607594936708861,
      "grad_norm": 0.4747781753540039,
      "learning_rate": 9.370252527016777e-05,
      "loss": 1.292,
      "step": 8160
    },
    {
      "epoch": 0.8618143459915611,
      "grad_norm": 0.4684956669807434,
      "learning_rate": 9.241097825525163e-05,
      "loss": 1.2824,
      "step": 8170
    },
    {
      "epoch": 0.8628691983122363,
      "grad_norm": 0.4697885513305664,
      "learning_rate": 9.113723325460276e-05,
      "loss": 1.289,
      "step": 8180
    },
    {
      "epoch": 0.8639240506329114,
      "grad_norm": 0.57652747631073,
      "learning_rate": 8.988104489448849e-05,
      "loss": 1.2845,
      "step": 8190
    },
    {
      "epoch": 0.8649789029535865,
      "grad_norm": 0.46039921045303345,
      "learning_rate": 8.864217118328042e-05,
      "loss": 1.2937,
      "step": 8200
    },
    {
      "epoch": 0.8660337552742616,
      "grad_norm": 0.45358946919441223,
      "learning_rate": 8.742037346483729e-05,
      "loss": 1.2864,
      "step": 8210
    },
    {
      "epoch": 0.8670886075949367,
      "grad_norm": 0.46776285767555237,
      "learning_rate": 8.62154163725303e-05,
      "loss": 1.2926,
      "step": 8220
    },
    {
      "epoch": 0.8681434599156118,
      "grad_norm": 0.4323977828025818,
      "learning_rate": 8.502706778390219e-05,
      "loss": 1.2901,
      "step": 8230
    },
    {
      "epoch": 0.869198312236287,
      "grad_norm": 0.4356761574745178,
      "learning_rate": 8.38550987759513e-05,
      "loss": 1.2866,
      "step": 8240
    },
    {
      "epoch": 0.870253164556962,
      "grad_norm": 0.45652660727500916,
      "learning_rate": 8.269928358103191e-05,
      "loss": 1.3016,
      "step": 8250
    },
    {
      "epoch": 0.8713080168776371,
      "grad_norm": 0.4988255798816681,
      "learning_rate": 8.155939954336243e-05,
      "loss": 1.2945,
      "step": 8260
    },
    {
      "epoch": 0.8723628691983122,
      "grad_norm": 0.4944089949131012,
      "learning_rate": 8.043522707613312e-05,
      "loss": 1.2897,
      "step": 8270
    },
    {
      "epoch": 0.8734177215189873,
      "grad_norm": 0.44518062472343445,
      "learning_rate": 7.932654961920486e-05,
      "loss": 1.2724,
      "step": 8280
    },
    {
      "epoch": 0.8744725738396625,
      "grad_norm": 0.49174994230270386,
      "learning_rate": 7.823315359739135e-05,
      "loss": 1.2732,
      "step": 8290
    },
    {
      "epoch": 0.8755274261603375,
      "grad_norm": 0.4520399272441864,
      "learning_rate": 7.715482837931577e-05,
      "loss": 1.2987,
      "step": 8300
    },
    {
      "epoch": 0.8765822784810127,
      "grad_norm": 0.4695223271846771,
      "learning_rate": 7.6091366236835e-05,
      "loss": 1.2764,
      "step": 8310
    },
    {
      "epoch": 0.8776371308016878,
      "grad_norm": 0.4494788944721222,
      "learning_rate": 7.504256230502289e-05,
      "loss": 1.2933,
      "step": 8320
    },
    {
      "epoch": 0.8786919831223629,
      "grad_norm": 0.4499630630016327,
      "learning_rate": 7.400821454270524e-05,
      "loss": 1.2884,
      "step": 8330
    },
    {
      "epoch": 0.879746835443038,
      "grad_norm": 0.4443351924419403,
      "learning_rate": 7.29881236935386e-05,
      "loss": 1.2758,
      "step": 8340
    },
    {
      "epoch": 0.880801687763713,
      "grad_norm": 0.45193472504615784,
      "learning_rate": 7.198209324762562e-05,
      "loss": 1.2798,
      "step": 8350
    },
    {
      "epoch": 0.8818565400843882,
      "grad_norm": 0.4622740149497986,
      "learning_rate": 7.098992940365946e-05,
      "loss": 1.2762,
      "step": 8360
    },
    {
      "epoch": 0.8829113924050633,
      "grad_norm": 0.442023366689682,
      "learning_rate": 7.001144103159e-05,
      "loss": 1.2863,
      "step": 8370
    },
    {
      "epoch": 0.8839662447257384,
      "grad_norm": 0.45359861850738525,
      "learning_rate": 6.904643963580461e-05,
      "loss": 1.2924,
      "step": 8380
    },
    {
      "epoch": 0.8850210970464135,
      "grad_norm": 0.44865256547927856,
      "learning_rate": 6.809473931881644e-05,
      "loss": 1.2812,
      "step": 8390
    },
    {
      "epoch": 0.8860759493670886,
      "grad_norm": 0.43512946367263794,
      "learning_rate": 6.71561567454532e-05,
      "loss": 1.2812,
      "step": 8400
    },
    {
      "epoch": 0.8871308016877637,
      "grad_norm": 0.44648268818855286,
      "learning_rate": 6.623051110753948e-05,
      "loss": 1.3014,
      "step": 8410
    },
    {
      "epoch": 0.8881856540084389,
      "grad_norm": 0.4547604024410248,
      "learning_rate": 6.531762408906607e-05,
      "loss": 1.2877,
      "step": 8420
    },
    {
      "epoch": 0.8892405063291139,
      "grad_norm": 0.4438287615776062,
      "learning_rate": 6.441731983183912e-05,
      "loss": 1.2848,
      "step": 8430
    },
    {
      "epoch": 0.890295358649789,
      "grad_norm": 0.44475919008255005,
      "learning_rate": 6.352942490160292e-05,
      "loss": 1.2769,
      "step": 8440
    },
    {
      "epoch": 0.8913502109704642,
      "grad_norm": 0.49799758195877075,
      "learning_rate": 6.265376825462966e-05,
      "loss": 1.2861,
      "step": 8450
    },
    {
      "epoch": 0.8924050632911392,
      "grad_norm": 0.45425400137901306,
      "learning_rate": 6.179018120476945e-05,
      "loss": 1.2815,
      "step": 8460
    },
    {
      "epoch": 0.8934599156118144,
      "grad_norm": 0.4422910809516907,
      "learning_rate": 6.0938497390954946e-05,
      "loss": 1.2835,
      "step": 8470
    },
    {
      "epoch": 0.8945147679324894,
      "grad_norm": 0.4422453045845032,
      "learning_rate": 6.009855274515339e-05,
      "loss": 1.2766,
      "step": 8480
    },
    {
      "epoch": 0.8955696202531646,
      "grad_norm": 0.4605245292186737,
      "learning_rate": 5.9270185460760735e-05,
      "loss": 1.2832,
      "step": 8490
    },
    {
      "epoch": 0.8966244725738397,
      "grad_norm": 0.43407219648361206,
      "learning_rate": 5.8453235961431225e-05,
      "loss": 1.2733,
      "step": 8500
    },
    {
      "epoch": 0.8976793248945147,
      "grad_norm": 0.48641687631607056,
      "learning_rate": 5.764754687033678e-05,
      "loss": 1.2842,
      "step": 8510
    },
    {
      "epoch": 0.8987341772151899,
      "grad_norm": 0.4426994025707245,
      "learning_rate": 5.6852962979849836e-05,
      "loss": 1.2791,
      "step": 8520
    },
    {
      "epoch": 0.8997890295358649,
      "grad_norm": 0.4594253599643707,
      "learning_rate": 5.6069331221644284e-05,
      "loss": 1.2937,
      "step": 8530
    },
    {
      "epoch": 0.9008438818565401,
      "grad_norm": 0.4511406719684601,
      "learning_rate": 5.529650063720842e-05,
      "loss": 1.2897,
      "step": 8540
    },
    {
      "epoch": 0.9018987341772152,
      "grad_norm": 0.48013877868652344,
      "learning_rate": 5.453432234876445e-05,
      "loss": 1.2818,
      "step": 8550
    },
    {
      "epoch": 0.9029535864978903,
      "grad_norm": 0.4629244804382324,
      "learning_rate": 5.37826495305886e-05,
      "loss": 1.2728,
      "step": 8560
    },
    {
      "epoch": 0.9040084388185654,
      "grad_norm": 0.5308079123497009,
      "learning_rate": 5.304133738072674e-05,
      "loss": 1.2912,
      "step": 8570
    },
    {
      "epoch": 0.9050632911392406,
      "grad_norm": 0.4697956144809723,
      "learning_rate": 5.2310243093099814e-05,
      "loss": 1.284,
      "step": 8580
    },
    {
      "epoch": 0.9061181434599156,
      "grad_norm": 0.49656587839126587,
      "learning_rate": 5.158922582999368e-05,
      "loss": 1.2875,
      "step": 8590
    },
    {
      "epoch": 0.9071729957805907,
      "grad_norm": 0.4452250599861145,
      "learning_rate": 5.087814669492819e-05,
      "loss": 1.2792,
      "step": 8600
    },
    {
      "epoch": 0.9082278481012658,
      "grad_norm": 0.46281588077545166,
      "learning_rate": 5.017686870590028e-05,
      "loss": 1.281,
      "step": 8610
    },
    {
      "epoch": 0.9092827004219409,
      "grad_norm": 0.4981115758419037,
      "learning_rate": 4.948525676899577e-05,
      "loss": 1.2771,
      "step": 8620
    },
    {
      "epoch": 0.9103375527426161,
      "grad_norm": 0.4706834852695465,
      "learning_rate": 4.880317765236493e-05,
      "loss": 1.2874,
      "step": 8630
    },
    {
      "epoch": 0.9113924050632911,
      "grad_norm": 0.5250294804573059,
      "learning_rate": 4.8130499960556755e-05,
      "loss": 1.2672,
      "step": 8640
    },
    {
      "epoch": 0.9124472573839663,
      "grad_norm": 0.4542900323867798,
      "learning_rate": 4.746709410920699e-05,
      "loss": 1.2737,
      "step": 8650
    },
    {
      "epoch": 0.9135021097046413,
      "grad_norm": 0.5390377044677734,
      "learning_rate": 4.681283230007507e-05,
      "loss": 1.2745,
      "step": 8660
    },
    {
      "epoch": 0.9145569620253164,
      "grad_norm": 0.4557817578315735,
      "learning_rate": 4.616758849642509e-05,
      "loss": 1.2836,
      "step": 8670
    },
    {
      "epoch": 0.9156118143459916,
      "grad_norm": 0.446895033121109,
      "learning_rate": 4.553123839874615e-05,
      "loss": 1.2922,
      "step": 8680
    },
    {
      "epoch": 0.9166666666666666,
      "grad_norm": 0.4644860625267029,
      "learning_rate": 4.490365942080736e-05,
      "loss": 1.2832,
      "step": 8690
    },
    {
      "epoch": 0.9177215189873418,
      "grad_norm": 0.46043819189071655,
      "learning_rate": 4.428473066604285e-05,
      "loss": 1.2813,
      "step": 8700
    },
    {
      "epoch": 0.9187763713080169,
      "grad_norm": 0.4642040729522705,
      "learning_rate": 4.367433290426233e-05,
      "loss": 1.281,
      "step": 8710
    },
    {
      "epoch": 0.919831223628692,
      "grad_norm": 0.4453197419643402,
      "learning_rate": 4.3072348548682595e-05,
      "loss": 1.2764,
      "step": 8720
    },
    {
      "epoch": 0.9208860759493671,
      "grad_norm": 0.45742228627204895,
      "learning_rate": 4.247866163327575e-05,
      "loss": 1.2878,
      "step": 8730
    },
    {
      "epoch": 0.9219409282700421,
      "grad_norm": 0.44467175006866455,
      "learning_rate": 4.1893157790429404e-05,
      "loss": 1.2782,
      "step": 8740
    },
    {
      "epoch": 0.9229957805907173,
      "grad_norm": 0.44879403710365295,
      "learning_rate": 4.1315724228915066e-05,
      "loss": 1.2744,
      "step": 8750
    },
    {
      "epoch": 0.9240506329113924,
      "grad_norm": 0.5003914833068848,
      "learning_rate": 4.074624971216005e-05,
      "loss": 1.2633,
      "step": 8760
    },
    {
      "epoch": 0.9251054852320675,
      "grad_norm": 0.43093347549438477,
      "learning_rate": 4.018462453681889e-05,
      "loss": 1.2738,
      "step": 8770
    },
    {
      "epoch": 0.9261603375527426,
      "grad_norm": 0.45481476187705994,
      "learning_rate": 3.963074051164014e-05,
      "loss": 1.2806,
      "step": 8780
    },
    {
      "epoch": 0.9272151898734177,
      "grad_norm": 0.44279158115386963,
      "learning_rate": 3.908449093662446e-05,
      "loss": 1.2753,
      "step": 8790
    },
    {
      "epoch": 0.9282700421940928,
      "grad_norm": 0.4584684371948242,
      "learning_rate": 3.854577058246998e-05,
      "loss": 1.272,
      "step": 8800
    },
    {
      "epoch": 0.929324894514768,
      "grad_norm": 0.44226986169815063,
      "learning_rate": 3.801447567030094e-05,
      "loss": 1.2926,
      "step": 8810
    },
    {
      "epoch": 0.930379746835443,
      "grad_norm": 0.43229052424430847,
      "learning_rate": 3.7490503851675777e-05,
      "loss": 1.2785,
      "step": 8820
    },
    {
      "epoch": 0.9314345991561181,
      "grad_norm": 0.44430503249168396,
      "learning_rate": 3.6973754188870806e-05,
      "loss": 1.2925,
      "step": 8830
    },
    {
      "epoch": 0.9324894514767933,
      "grad_norm": 0.439351350069046,
      "learning_rate": 3.6464127135435536e-05,
      "loss": 1.2797,
      "step": 8840
    },
    {
      "epoch": 0.9335443037974683,
      "grad_norm": 0.5136128664016724,
      "learning_rate": 3.596152451701616e-05,
      "loss": 1.2785,
      "step": 8850
    },
    {
      "epoch": 0.9345991561181435,
      "grad_norm": 0.5367308259010315,
      "learning_rate": 3.5465849512443226e-05,
      "loss": 1.279,
      "step": 8860
    },
    {
      "epoch": 0.9356540084388185,
      "grad_norm": 0.4388138949871063,
      "learning_rate": 3.4977006635080086e-05,
      "loss": 1.2818,
      "step": 8870
    },
    {
      "epoch": 0.9367088607594937,
      "grad_norm": 0.46991193294525146,
      "learning_rate": 3.449490171442838e-05,
      "loss": 1.285,
      "step": 8880
    },
    {
      "epoch": 0.9377637130801688,
      "grad_norm": 0.4393314719200134,
      "learning_rate": 3.401944187798702e-05,
      "loss": 1.2844,
      "step": 8890
    },
    {
      "epoch": 0.9388185654008439,
      "grad_norm": 0.4360581934452057,
      "learning_rate": 3.355053553336137e-05,
      "loss": 1.2736,
      "step": 8900
    },
    {
      "epoch": 0.939873417721519,
      "grad_norm": 0.46192970871925354,
      "learning_rate": 3.308809235061882e-05,
      "loss": 1.2738,
      "step": 8910
    },
    {
      "epoch": 0.9409282700421941,
      "grad_norm": 0.45846766233444214,
      "learning_rate": 3.263202324488772e-05,
      "loss": 1.2816,
      "step": 8920
    },
    {
      "epoch": 0.9419831223628692,
      "grad_norm": 0.45226266980171204,
      "learning_rate": 3.218224035919609e-05,
      "loss": 1.276,
      "step": 8930
    },
    {
      "epoch": 0.9430379746835443,
      "grad_norm": 0.44694724678993225,
      "learning_rate": 3.173865704754688e-05,
      "loss": 1.2776,
      "step": 8940
    },
    {
      "epoch": 0.9440928270042194,
      "grad_norm": 0.44097021222114563,
      "learning_rate": 3.130118785822657e-05,
      "loss": 1.2783,
      "step": 8950
    },
    {
      "epoch": 0.9451476793248945,
      "grad_norm": 0.4459550380706787,
      "learning_rate": 3.08697485173437e-05,
      "loss": 1.2839,
      "step": 8960
    },
    {
      "epoch": 0.9462025316455697,
      "grad_norm": 0.5135442614555359,
      "learning_rate": 3.0444255912594442e-05,
      "loss": 1.2796,
      "step": 8970
    },
    {
      "epoch": 0.9472573839662447,
      "grad_norm": 0.5376752018928528,
      "learning_rate": 3.002462807725185e-05,
      "loss": 1.2766,
      "step": 8980
    },
    {
      "epoch": 0.9483122362869199,
      "grad_norm": 0.45783647894859314,
      "learning_rate": 2.9610784174375868e-05,
      "loss": 1.2825,
      "step": 8990
    },
    {
      "epoch": 0.9493670886075949,
      "grad_norm": 0.44302478432655334,
      "learning_rate": 2.920264448124087e-05,
      "loss": 1.2746,
      "step": 9000
    },
    {
      "epoch": 0.95042194092827,
      "grad_norm": 0.45025965571403503,
      "learning_rate": 2.8800130373977936e-05,
      "loss": 1.2745,
      "step": 9010
    },
    {
      "epoch": 0.9514767932489452,
      "grad_norm": 0.4385267496109009,
      "learning_rate": 2.84031643124288e-05,
      "loss": 1.275,
      "step": 9020
    },
    {
      "epoch": 0.9525316455696202,
      "grad_norm": 0.5064539313316345,
      "learning_rate": 2.8011669825208517e-05,
      "loss": 1.2921,
      "step": 9030
    },
    {
      "epoch": 0.9535864978902954,
      "grad_norm": 0.45993825793266296,
      "learning_rate": 2.762557149497405e-05,
      "loss": 1.271,
      "step": 9040
    },
    {
      "epoch": 0.9546413502109705,
      "grad_norm": 0.43950140476226807,
      "learning_rate": 2.724479494389592e-05,
      "loss": 1.2812,
      "step": 9050
    },
    {
      "epoch": 0.9556962025316456,
      "grad_norm": 0.44806602597236633,
      "learning_rate": 2.6869266819330058e-05,
      "loss": 1.2812,
      "step": 9060
    },
    {
      "epoch": 0.9567510548523207,
      "grad_norm": 0.44085240364074707,
      "learning_rate": 2.6498914779687228e-05,
      "loss": 1.2822,
      "step": 9070
    },
    {
      "epoch": 0.9578059071729957,
      "grad_norm": 0.4372364580631256,
      "learning_rate": 2.6133667480497115e-05,
      "loss": 1.2841,
      "step": 9080
    },
    {
      "epoch": 0.9588607594936709,
      "grad_norm": 0.4652843475341797,
      "learning_rate": 2.5773454560664597e-05,
      "loss": 1.2842,
      "step": 9090
    },
    {
      "epoch": 0.959915611814346,
      "grad_norm": 0.43991342186927795,
      "learning_rate": 2.541820662891541e-05,
      "loss": 1.2643,
      "step": 9100
    },
    {
      "epoch": 0.9609704641350211,
      "grad_norm": 0.43201401829719543,
      "learning_rate": 2.5067855250428616e-05,
      "loss": 1.2751,
      "step": 9110
    },
    {
      "epoch": 0.9620253164556962,
      "grad_norm": 0.45959627628326416,
      "learning_rate": 2.472233293365335e-05,
      "loss": 1.2856,
      "step": 9120
    },
    {
      "epoch": 0.9630801687763713,
      "grad_norm": 0.4455472528934479,
      "learning_rate": 2.4381573117307307e-05,
      "loss": 1.2697,
      "step": 9130
    },
    {
      "epoch": 0.9641350210970464,
      "grad_norm": 0.4947901666164398,
      "learning_rate": 2.4045510157554362e-05,
      "loss": 1.2891,
      "step": 9140
    },
    {
      "epoch": 0.9651898734177216,
      "grad_norm": 0.46208250522613525,
      "learning_rate": 2.3714079315358985e-05,
      "loss": 1.2769,
      "step": 9150
    },
    {
      "epoch": 0.9662447257383966,
      "grad_norm": 0.4391043484210968,
      "learning_rate": 2.338721674401494e-05,
      "loss": 1.2841,
      "step": 9160
    },
    {
      "epoch": 0.9672995780590717,
      "grad_norm": 0.44544529914855957,
      "learning_rate": 2.30648594768459e-05,
      "loss": 1.2806,
      "step": 9170
    },
    {
      "epoch": 0.9683544303797469,
      "grad_norm": 0.4403120279312134,
      "learning_rate": 2.2746945415075523e-05,
      "loss": 1.2772,
      "step": 9180
    },
    {
      "epoch": 0.9694092827004219,
      "grad_norm": 0.4416530132293701,
      "learning_rate": 2.2433413315864803e-05,
      "loss": 1.273,
      "step": 9190
    },
    {
      "epoch": 0.9704641350210971,
      "grad_norm": 0.48513439297676086,
      "learning_rate": 2.2124202780514277e-05,
      "loss": 1.2764,
      "step": 9200
    },
    {
      "epoch": 0.9715189873417721,
      "grad_norm": 0.45357707142829895,
      "learning_rate": 2.1819254242828815e-05,
      "loss": 1.2783,
      "step": 9210
    },
    {
      "epoch": 0.9725738396624473,
      "grad_norm": 0.44976767897605896,
      "learning_rate": 2.151850895764285e-05,
      "loss": 1.2753,
      "step": 9220
    },
    {
      "epoch": 0.9736286919831224,
      "grad_norm": 0.42965638637542725,
      "learning_rate": 2.12219089895037e-05,
      "loss": 1.2672,
      "step": 9230
    },
    {
      "epoch": 0.9746835443037974,
      "grad_norm": 0.463337242603302,
      "learning_rate": 2.092939720151092e-05,
      "loss": 1.2605,
      "step": 9240
    },
    {
      "epoch": 0.9757383966244726,
      "grad_norm": 0.43859803676605225,
      "learning_rate": 2.064091724430947e-05,
      "loss": 1.274,
      "step": 9250
    },
    {
      "epoch": 0.9767932489451476,
      "grad_norm": 0.45552858710289,
      "learning_rate": 2.0356413545234603e-05,
      "loss": 1.2699,
      "step": 9260
    },
    {
      "epoch": 0.9778481012658228,
      "grad_norm": 0.44371435046195984,
      "learning_rate": 2.0075831297606357e-05,
      "loss": 1.2815,
      "step": 9270
    },
    {
      "epoch": 0.9789029535864979,
      "grad_norm": 0.44908347725868225,
      "learning_rate": 1.9799116450171627e-05,
      "loss": 1.2692,
      "step": 9280
    },
    {
      "epoch": 0.979957805907173,
      "grad_norm": 0.4544650614261627,
      "learning_rate": 1.952621569669175e-05,
      "loss": 1.2748,
      "step": 9290
    },
    {
      "epoch": 0.9810126582278481,
      "grad_norm": 0.447017639875412,
      "learning_rate": 1.9257076465673605e-05,
      "loss": 1.2767,
      "step": 9300
    },
    {
      "epoch": 0.9820675105485233,
      "grad_norm": 0.5013083815574646,
      "learning_rate": 1.899164691024229e-05,
      "loss": 1.2767,
      "step": 9310
    },
    {
      "epoch": 0.9831223628691983,
      "grad_norm": 0.4764220118522644,
      "learning_rate": 1.872987589815331e-05,
      "loss": 1.2719,
      "step": 9320
    },
    {
      "epoch": 0.9841772151898734,
      "grad_norm": 0.4521596133708954,
      "learning_rate": 1.8471713001942538e-05,
      "loss": 1.2836,
      "step": 9330
    },
    {
      "epoch": 0.9852320675105485,
      "grad_norm": 0.4511817693710327,
      "learning_rate": 1.8217108489211845e-05,
      "loss": 1.2813,
      "step": 9340
    },
    {
      "epoch": 0.9862869198312236,
      "grad_norm": 0.45081964135169983,
      "learning_rate": 1.7966013313048696e-05,
      "loss": 1.2914,
      "step": 9350
    },
    {
      "epoch": 0.9873417721518988,
      "grad_norm": 0.4383713901042938,
      "learning_rate": 1.7718379102577752e-05,
      "loss": 1.2836,
      "step": 9360
    },
    {
      "epoch": 0.9883966244725738,
      "grad_norm": 0.44428008794784546,
      "learning_rate": 1.7474158153642745e-05,
      "loss": 1.283,
      "step": 9370
    },
    {
      "epoch": 0.989451476793249,
      "grad_norm": 0.43199053406715393,
      "learning_rate": 1.7233303419616745e-05,
      "loss": 1.2668,
      "step": 9380
    },
    {
      "epoch": 0.990506329113924,
      "grad_norm": 0.4628349244594574,
      "learning_rate": 1.699576850233916e-05,
      "loss": 1.2725,
      "step": 9390
    },
    {
      "epoch": 0.9915611814345991,
      "grad_norm": 0.45799240469932556,
      "learning_rate": 1.6761507643177553e-05,
      "loss": 1.2869,
      "step": 9400
    },
    {
      "epoch": 0.9926160337552743,
      "grad_norm": 0.454426109790802,
      "learning_rate": 1.6530475714212752e-05,
      "loss": 1.2621,
      "step": 9410
    },
    {
      "epoch": 0.9936708860759493,
      "grad_norm": 0.43662726879119873,
      "learning_rate": 1.6302628209545423e-05,
      "loss": 1.28,
      "step": 9420
    },
    {
      "epoch": 0.9947257383966245,
      "grad_norm": 0.46751806139945984,
      "learning_rate": 1.6077921236722464e-05,
      "loss": 1.2791,
      "step": 9430
    },
    {
      "epoch": 0.9957805907172996,
      "grad_norm": 0.44598811864852905,
      "learning_rate": 1.5856311508281594e-05,
      "loss": 1.2755,
      "step": 9440
    },
    {
      "epoch": 0.9968354430379747,
      "grad_norm": 0.47813931107521057,
      "learning_rate": 1.5637756333412454e-05,
      "loss": 1.2843,
      "step": 9450
    },
    {
      "epoch": 0.9978902953586498,
      "grad_norm": 0.44852492213249207,
      "learning_rate": 1.542221360973268e-05,
      "loss": 1.2692,
      "step": 9460
    },
    {
      "epoch": 0.9989451476793249,
      "grad_norm": 0.44214802980422974,
      "learning_rate": 1.5209641815177312e-05,
      "loss": 1.2839,
      "step": 9470
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.3379602432250977,
      "learning_rate": 1.5e-05,
      "loss": 1.2704,
      "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": 1.8335001376290816e+16,
  "train_batch_size": 1024,
  "trial_name": null,
  "trial_params": null
}